Mathematica Beyond

Mathematics
Although many books have been written about Mathematica, very few of them cover the new
functionality added to the most recent versions of the program. This thoroughly revised second
edition of Mathematica Beyond Mathematics: The Wolfram Language in the Real World
introduces the new features using real-world examples based on the experience of the author as
a consultant and Wolfram certified instructor. The examples strike a balance between relevance
and difficulty in terms of Mathematica syntax, allowing readers to incrementally build up their
Mathematica skills as they go through the chapters While reading this book, you will also learn
more about the Wolfram Language and how to use it to solve a wide variety of problems.

The author raises questions from a wide range of topics and answers them by taking full advan-
tage of Mathematica’s latest features. For example: What sources of energy does the world really
use? Are our cities getting warmer? Is the novel El Quixote written in Pi? Is it possible to reliably
date the Earth using radioactive isotopes? How can we find planets outside our solar system?
How can we model epidemics, earthquakes and other natural phenomena? What is the best way
to compare organisms genetically?

This new edition introduces the new capabilities added to the latest version of Mathematica
(version 13), and discusses new topics related to machine learning, big data, finance economics,
and physics.

New to the Second Edition

• Separate sections containing carefully selected additional resources that can be accessed
from either Mathematica or online
• Online Supplementary materials including code snippets used in the book and additional
examples.
• Updated commands to take full advantage of Mathematica 13.

José Guillermo Sánchez León (http://diarium.usal.es/guillermo) is an engineer, physicist and

mathematics PhD holder. He is currently a mathematics and statistical consultant and teaches
Mathematical modeling at the University of Salamanca. He has worked in the energy industry
and conducted research in a variety of fields: Modeling, optimization, medical physics, astron-
omy, finance and others. In 1999, he was awarded a research grant at Wolfram Research Inc.
headquarters in Champaign (Illinois, USA) after his statistical applications with Mathematica
project won a competition sponsored by the company. Since then, he has been an active Math-
ematica and webMathematica alpha and beta tester.

He is also a Wolfram certified instructor and has extensive experience in teaching and develop-
ing programs with both applications. Among his more than 100 articles, there are several where
Mathematica and web Mathematica have been used extensively.
Taylor & Francis
Taylor & Francis Group
http://taylorandfrancis.com
 Mathematica Beyond
Mathematics
The Wolfram Language in the Real World
Second Edition

José Guillermo Sánchez León

University of Salamanca, Spain

Translated by
Rubén García Berasategui
Second edition published 2023

by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742

and by CRC Press

4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN

© 2023 José Guillermo Sánchez León

CRC Press is an imprint of Taylor & Francis Group, LLC

Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright
holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowl-
edged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or
utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including pho-
tocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission
from the publishers.

For permission to photocopy or use material electronically from this work, access www.copyright.com or contact the
Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. For works that are
not available on CCC please contact [email protected]

Trademark notice: Product or corporate names may be trademarks or registered trademarks and are used only for
identification and explanation without intent to infringe.

Library of Congress Cataloging‑in‑Publication Data

Names: Sánchez León, José Guillermo, author.

Title: Mathematica beyond mathematics : the Wolfram language in the real
world / José Guillermo Sánchez León, University of Salamanca, Spain ;
translated by Rubén García Berasategui.
Description: Second edition. | Boca Raton, FL : Chapman & Hall/CRC Press,
2023. | Includes bibliographical references and index.
Identifiers: LCCN 2022030387 (print) | LCCN 2022030388 (ebook) | ISBN
9781032004839 (hbk) | ISBN 9781032010236 (pbk) | ISBN 9781003176800
(ebk)
Subjects: LCSH: Mathematica (Computer file) | Mathematics--Data processing.
Classification: LCC QA76.95 .S26 2023 (print) | LCC QA76.95 (ebook) | DDC
510.285/536--dc23/eng/20221104
LC record available at https://lccn.loc.gov/2022030387
LC ebook record available at https://lccn.loc.gov/2022030388

ISBN: 978-1-032-00483-9 (hbk)

ISBN: 978-1-032-01023-6 (pbk)
ISBN: 978-1-003-17680-0 (ebk)

DOI: 10.1201/ 9781003176800

Typeset in Times New Roman

by KnowledgeWorks Global Ltd.

Publisher’s note: This book has been prepared from camera-ready copy provided by the authors.
Contents

Preface IX

1. Getting Started 1
1.1 Mathematica, an Integrated Technical Computing System 1
1.2 First Steps 3
1.3 Editing Notebooks 12
1.4 Basic Ideas 19
1.5 From Graphics to Machine Learning 29
1.6 Additional Resources and Supplementary Materials 40

2. Programming: The Beauty and Power of the Wolfram Language 41

2.1 Mathematica’s Programming Language: The Wolfram Language 41
2.2 Lists Operations 46
2.3 Association and Dataset 50
2.4 Matrix Operations 52
2.5 Set, SetDelayed, and Dynamic Variables 55
2.6 Functional vs. Procedural Programming 57
2.7 Apply, Map, and Other Related Functions 60
2.8 Iterative Functions 63
2.9 Pure Functions 64
2.10 Global and Local Variables 70
2.11 Conditional Expressions and Conditions 72
2.12 Accuracy and Precision 79
2.13 Choosing the Method of Computation 82
2.14 Optimizing the Computation Time 84
2.15 Cloud Deployment 86
2.16 Package Development 87
2.17 Additional Resources 90

3. Interactive Applications, Image Processing, and More 91

3.1 The Manipulate Function 91
vi Mathematica beyond mathematics

3.2 Creating Demonstrations 97

3.3 Image Processing 100
3.4 Image Manipulation 105
3.5 Graphs and Networks 110
3.6 Application: Finding the Period of a Pendulum 113
3.7 Additional Resources 116

4. Accessing Scientific and Technical Information 117

4.1 The Wolfram Data Framework: Introducing Entities 117
4.2 Computable Data Functions 123
4.3 The Wolfram Data Repository 127
4.4 Weather Data in Real Time 129
4.5 Chemical and Physical Properties of Elements and Compounds 132
4.6 Life Sciences and Medicine 136
4.7 Earth Sciences and Geographic Data 142
4.8 Additional Resources 151

5. Data Analysis and Manipulation 153

5.1 Importing/Exporting 153
5.2 Statistical Analysis 162
5.3 Probability Distributions 169
5.4 Exploratory Data Analysis 181
5.5 Bootstrapping and Confidence Estimates 189
5.6 Curve Fitting 194
5.7 Time Series Analysis 203
5.8 Spatial Statistics 204
5.9 Additional Resources 208

6. Machine Learning and Neural Networks 207

6.1 What is Machine Learning 207
6.2 Classification 212
6.3 Prediction 221
6.4 Working with Neural Networks 225
6.5 Additional Resources 230

7. Calculating
and Other Mathematical Tales 231

7.1 The Origins of
231
7.2 Archimedes’ Approximation 232
7.3
with More Than One Billion Decimals 235
7.4 Buffon’s Method 240
7.5 Application: Are the Decimal Digits of
Random? 242
Contents vii

7.6 The Strange Connection 246

7.7 The Riemann Hypothesis 248
7.8 Looking for the Magic Prime Formula 252
7.9 Additional Resources 254

8. Looking at the Sky 255

8.1 A Short Astronomical Walk 255
8.2 Solar Analemma 259
8.3 Stargazing 260
8.4 Application: Determining the Color of the Stars 279
8.5 The Measurement of Distances Across the Universe 283
8.6 Application: Binary Systems and the Search for Exoplanets 287
8.7 Light Curves 291
8.8 Additional Resources 300

9. Nuclei and Radiations 301

9.1 Nuclear and Particle Physics 301
9.2 What are Isotopes? 302
9.3 Decay Constants, Decay Periods and Half-Lives 304
9.4 Decay Chains 308
9.5 Application: Modeling the Evolution of a Chain of Isotopes Over Time 313
9.6 Application: Dating the History of Humankind 316
9.7 Application: Calculating Binding Energies 321
9.8 Radiation Attenuation 328
9.9 Additional Resources 330

10. Modeling: Applications in Biokinetics, Epidemiology and More 331

10.1 Compartmental Modeling 331
10.2 Epidemiological Models 342
10.3 Physiological Modeling 346
10.4 Fitting a Model 351
10.5 Optimal Experimental Designs (OED) 355
10.6 BIOKMOD: The New Iodine OIR Model (ICRP 137) 360
10.7 Additional Modeling Examples 364
10.8 Modeling Using PDEs 366
10.9 System Modeler 369
10.10 Additional Resources 370

11. Economic, Financial and Optimization Applications 371

11.1 Accessing Economic Information 371
11.2 Financial Information 374
11.3 Financial Functions 381
viii Mathematica beyond mathematics

11.4 Optimization 392

11.5 The Shortest Path Problem 403
11.6 Optimum Flows 407
11.7 Blockchains 410
11.8 Additional Resources 412

12. Faster, Further 415

12.1 Parallel Computing 413
12.2 Parallel Programming 414
12.3 The Mandelbrot Set 422
12.4 Comparing Organisms Genetically 427
12.5 Software Development with Wolfram Workbench 431
12.6 Compute Unified Device Architecture (CUDA) 434
12.7 Connecting with Other Programs and Devices 434
12.8 Additional Resources 435

Index 437
Preface

Mathematica and the Wolfram Language

If you have used Mathematica occasionally or heard of it, you may have the false
impression that it is a program for performing complicated calculations, usually for
academic purposes. However, this idea is far from the truth. Actually, Mathematica
is much more than that. By putting together the computational power and ease of
use of the Wolfram Language, Mathematica’s high-level general-purpose
programming language, the program can be used in any scientific or technical field:
aerospace engineering, environmental sciences, financial risk management, medical
imaging and many others.
Mathematica can be considered a tool that empowers non-professional
programmers to develop applications, although if you are a professional
programmer, you will see that the software provides a development environment
similar to the ones available for C++ or Python. You can even use the program to
control external devices.

About the Book

Although many books have been written about Mathematica, very few of them
cover the new functionality added to the most recent versions of the program. This
text introduces the new features using real-world examples, based on the
experience of the author as a consultant. In the process, you will also learn more
about the Wolfram Language and how you can use it to solve a wide variety of
problems. Both are the most important objectives of the book. To accomplish that,
the author raises questions from a wide range of topics and answers them by taking
full advantage of Mathematica’s latest features. For example: What is the hidden
image in “The Ambassadors” painting by Holbein? What sources of energy does
the world really use? How can we calculate tolerance limits in manufacturing
processes? Are our cities getting warmer? Is the novel “El Quijote” written in Pi?
How do we know how old our planet is? How can we find planets outside our solar
system? How big is our galaxy? And the universe? How do we know it? How can
x Mathematica beyond mathematics

we model the distribution of radioactive isotopes in the human body? And a

tsunami? What are and how can we create Mandelbrot fractals? How can we
measure the genetic distance between species? How can we perform financial
calculations in real time? How do we value financial derivatives? How can we
make entertaining simulations for teaching mathematics, physics, statistics, ... ?
Why are there no free quarks?
The answers to the previous questions will not only help you master Mathematica,
but also make you more familiar with the corresponding topics themselves.

Book Objectives
This book will not only be useful to newcomers but also to those familiar with the
program and interested in learning more about the new functionality included in the
latest versions.
Those readers with minimal or no knowledge of Mathematica are strongly advised
to read Chapter 1 along with Stephen Wolfram’s book An Elementary Introduction
to the Wolfram Language available from within the program documentation under
the resources section: Help
Wolfram Documentation
Intro Book & Course


This text will also make it easy to start programming using the Wolfram Language
and to learn how to take full advantage of its capabilities.
The final objective of Mathematica Beyond Mathematics is to help you avoid
feeling overwhelmed by the software’s vast capabilities. The author has explored a
significant part of them choosing the most relevant parts and illustrating them with
examples from many different sources including the program documentation. Links
to additional resources are also provided. The main aim of all this is to reduce
significantly the amount of time required to master the tool. The Mathematica
functions used will be explained using short sentences and simple examples.

Use of the Book

Although this book is not a manual, inexperienced readers should at least read the
first four chapters in consecutive order to gain a solid understanding of how
Mathematica works. Regarding the rest of the chapters, you should keep in mind
the following: a) Chapter 5 deals with relatively advanced probability and statistics
topics; c) Chapter 6 is a short introduction to machine learning and neural
networks; d) Chapters 7 to 11 explore topics related to a single area of knowledge
(mathematics, astronomy, nuclear physics, modeling, and economics and finance).
You can read them according to your preferences; and d) Chapter 12 is for those
readers facing problems requiring big computational resources (parallel computing,
grid-enabled calculations, etc).
The principal theme of each chapter is used as the motivation to illustrate certain
features of the program that you may find useful for solving a great variety of
Preface xi

problems. For example: Chapter 8, related to astronomy, explains how to create

dynamic images; Chapter 10, covering the modeling of biological systems,
discusses the resolution of differential equations in Mathematica using concrete
examples.

Second Edition and the Supplementary Materials

Mathematica Beyond Mathematics 2nd Edition improves and updates the first
edition. In this endeavor, Ruben Garcia Berasategui, a lecturer at Jakarta
International College, has played a fundamental role, not only by translating the
text but also by making insightful comments and suggestions.
Everything shown in the text has been done using Mathematica, including the
access to information sources.
At the end of each chapter, there is a section containing additional resources
carefully selected that can be accessed most of the time from either Mathematica or
online.
In the book’s website: http://www.mathematicabeyondmathematics.com/ you can
find supplementary materials including code and files used in the examples.
Although most of the contents of the first edition are still relevant, this second
edition also covers functionality added in versions 12 and 13 and includes many
more examples. To keep the number of pages similar to the first edition, some of
the previous contents have been moved online (on the book's website).
This edition has been written and edited using Mathematica 13, although it should
still be useful with future releases of the program as well. At least until 2025, if
there are any changes to the program’s functionality as described in the book, they
will be covered as supplementary material in the website.
The author would appreciate any comments or suggestions. Please send them to:
[email protected], with the subject: ‘Mathematica Beyond Mathematics’.
If any errors are found, an errata list will be created and made available online (on
the book’s website).

Salamanca (Spain), November 2022

Author
J. Guillermo Sánchez León (http://diarium.usal.es/guillermo) is an engineer,
physicist, and mathematics PhD holder. He is currently a mathematics and
statistical consultant and teaches mathematical modeling at the University of
Salamanca. He has worked in the energy industry and conducted research in a
variety of fields: Modeling, optimization, medical physics, astronomy, finance, and
others. In 1999, he was awarded a research grant at Wolfram Research Inc.
headquarters in Champaign, Illinois, US after his statistical applications with
Mathematica project won a competition sponsored by the company. Since then, he
xii Mathematica beyond mathematics

has been an active Mathematica and webMathematica alpha and beta tester. He is
also a Wolfram-certified instructor and has extensive experience in teaching and
developing programs with both applications. Among his more than 100 articles,
there are several where Mathematica and webMathematica have been used
extensively.

Reviewer
Rubén García Berasategui is a math, finance, and statistics lecturer at Jakarta
International College. He holds a bachelor’s degree in business administration and
an MBA. He’s been using Mathematica since 1997 and in 2012 became the first
Wolfram certified-instructor in Southeast Asia. He has been training hundreds of
newcomers to Mathematica and sharing his passion for the program with them ever
since.

Acknowledgments
The author would like to express his gratitude to Ruben Garcia Berasategui for his
help in writing both editions. Most likely, I would not have written the book
without his participation.
The author would also like to thank Callum Fraser (Editor of CRC Press, Taylor &
Francis Group), for his vision in realizing the need in the market for a book about
Mathematica like this one, and Mansi Kabra ( Senior Editorial Assistant of CRC
Press , Taylor & Francis Group) for his encouragement and support during the
production process.
The author would also like to say thank you to Addlink Software Científico
(http://www.addlink.es), sponsor of the first Spanish edition of this book; Antonio
Molina and Juan Antonio Rubio for their editorial guidance; and J. M. Cascón
(USAL), J. López-Fidalgo (UCLM), S. Miranda (Solventis), R. Pappalardo (US), J.
M. Rodríguez (USAL) and C. Tejero (USAL), for their comments and corrections
to the first edition.
1
Getting Started

This chapter introduces Wolfram Mathematica, a software program that

goes beyond numeric and symbolic calculations. It is a modern technical
computing system based on a high-level general-purpose programming
language, the Wolfram Language (WL). You will be able to start using the
application after a few minutes with it but, if you want to take full
advantage of its capabilities, you will have to spend time to master its
language although probably less than with other programming languages
such as C++ or Java, given its gentler learning curve.
The sections in this chapter provide an overview of the program’s
capabilities, show you how to get started, interact with Mathematica’s
front end and explain the main ideas behind WL through examples
ranging from basic plotting to image manipulation and machine learning.

1.1 Mathematica, an Integrated Technical Computing System

Mathematica took a revolutionary step forward with the introduction of the free-
form linguistic input, which consists of writing in plain English a request that the
program will try to answer. Using this format ensures that newcomers can get
started very quickly. For experienced users, it means help in finding the most
appropriate functions to perform a desired calculation. To get an idea of
Mathematica’s capabilities before beginning to use the program, let’s take a look at
the following examples (for now, you just need to read what follows, later we will
show you how to write inputs and run commands):

We can even ask very specific questions: To learn more about the temperature in a location,
just write down the name of the place followed by “temperature.” The most important
advantage of this approach is that the information from the answer can be used inside the
Mathematica environment.
 2      


  Salamanca temperature

AirTemperatureData
Salamanca CITY 

 
 28.
C

The AirTemperatureData function is shown in the output. It is one of the WL functions

for accessing scientific and technical databases from many different fields over the Internet.
The data have been already formatted by Wolfram Research for their direct manipulation
inside the application.

Here is another example of this kind of functions. We can choose a chemical compound and
get many of its properties. In this case, we get the plot of the caffeine molecule.


 Caffeine
"MoleculePlot"

 


With very brief syntax we can build functions that in other programming languages
would require several lines of code.

In less than half a line we can write the necessary instructions to make an interactive model
of the Julia fractal set.


 Manipulate
JuliaSetPlot
0.365  k , PlotLegends  Automatic,
k, 0.385, 0.3, 0.7

The code below, generates a map with a 10-km-radius around where you are (this is usually
based on your IP address):


 GeoGraphics
GeoDisk
Here, Quantity
10, "km"
1. Getting Started 3

Apart from the examples given above showcasing some of Mathematica’s

functionality, it is also worth mentioning:
(i) The Wolfram Predictive Interface, which makes suggestions about how to proceed after
executing a command, (ii) The context-sensitive input assistant to help us choose the correct
function, (iii) The possibility of running the program from the browser (WolframCloud.com
or https://www.wolfram.com/wolfram-one/), including specific cloud-computing-related
commands, (iv) The machine learning capabilities that, for example, enable us to analyze a
handwritten text and keep on improving its translation through repetition, (v) Very powerful
functions for graphics and image processing. We will refer to these and other new
capabilities later on.
Visit https://www.wolfram.com/language/gallery/, the Wolfram Language Code
Gallery, to find additional examples covering a wide variety of fields.

1.2 First Steps

1.2.1 Starting
The program can be run locally with Mathematica or Wolfram|One or over the
Internet using Wolfram Cloud or Mathematica Online.
Mathematica: https://www.wolfram.com/mathematica
Wolfram|One: https://www.wolfram.com/wolfram-one
Wolfram Cloud: https://www.wolframcloud.com
Mathematica Online: https://www.wolfram.com/mathematica/online
From here on, unless mentioned otherwise, we will assume that you have installed
Mathematica locally and activated its license. However, if you are using any of the
other two alternatives, the process is quite similar. When installing it, an assistant
will guide you through the activation process. If this is the first time, it will take
you to the user portal (https://user.wolfram.com), where you will have to sign up.
The website will ask you to register using an email address and create a Wolfram
ID and a password. It will also automatically generate an activation key. Remember
4 Mathematica beyond mathematics

your Wolfram ID and password because they will be useful later on if, for example,
you would like to install the program in a different computer or have an expired
license. In both cases, you will need a new activation key. The Wolfram ID and
password will also be necessary to access the Wolfram Cloud or Wolfram|One, and
we will give a specific example at the end of the chapter.
To start the program in Windows, under the programs menu in your system, click
on the Wolfram Mathematica icon.
Under OS X, the program is located in the applications folder, and you can launch
it either locating it with Finder, using the Dock or from the launchpad (OS X Lion
or more recent versions).
By default, a welcome screen (Figure 1.1) similar to this one will appear:

Figure 1.1 The Welcome Screen

This screen contains several external links (hyperlinks). For example, by clicking
on Resources, you will be able to access a broad collection of reference materials
and tutorials. In the upper left-hand corner you will see this icon: . If
you are a registered user, chances are you already have a Wolfram ID. In that case,
click on the icon and sign in. Once the process has been completed, the icon will be
replaced with the name that you used when you signed up, as shown in the previous
image. The advantage of this is to gain access from within Mathematica to your
files in the cloud (WolframCloud) but for now, you can skip this step if desired. If
you click on New Document, below the icon shown at the beginning of the
paragraph, a new blank Notebook will be created. Notebooks are the fundamental
way to interact with Mathematica. You can also create them by clicking on File 
New  Notebook.
A notebook is initially a blank page. We can write on it the text and instructions
that we want. It is always a good idea to save the file before we start working on it;
as in most programs, this can be done in the menu bar: File  Save As  “Name of
the File” (Mathematica by default will assign it the extension .nb).
It is usually convenient to have access to the Formatting toolbar, select in the
menu bar Window and check Toolbar  Formatting. Please note the drop-down
    

box to the left of the toolbar (Figure 1.2), that we’ll refer to as: style box. It will be
useful later on.

Figure 1.2   ’s Formatting and Default Toolbar.

Once you begin to type, a new cell will be created. It will be identified by a blue
right-bracket square (]) (“cell marker”) that appears on the right side of the
notebook. Each cell constitutes a unit with its own properties. To see the cell type,
place the cursor inside or on the cell marker. The type will appear in the style box
(Figure, Input, Text, Title, Program...). We can also change the format directly
inside the box. Type in a text cell and try to change its style.
This is a cell formatted using the Program style.

When a blank notebook is created, it has a style (cell types, fonts, etc.) that assigns
to each type of cell certain properties (evaluatable, editable, etc.). The first time a
notebook is created, Mathematica assigns a style named Default (by default). Later
on we will see how to choose different styles. In the Default style, when a new cell
is created, it will be of the Input type, which is the one used normally for
calculations. When we evaluate an Input type cell, “In[n]:= our calculation” will be
shown and a new cell of type Output will be generated where you will see the
calculation result in the form of “Out[n]:= result” (n showing the evaluation order).
However, in this book we sometimes use an option to omit the symbols “In[n]:=”
and “Out[n]:=”. These and other options to customize the program can be accessed
through Edit  Preferences ....

To execute an Input cell select the cell (placing the cursor on the cell marker)
and press 
 or  on the numeric keypad. You can also access the
contextual menu by right-clicking with the mouse and selecting “Evaluate Cell”.


 22

 
 4

Figure 1.3 The Suggestions Bar.

When a cell is executed, a toolbar appears below its output as shown above (Figure
1.3). This bar, named the Suggestions Bar, provides immediate access to possible
next steps optimized for your results. For example, If you click on “range” you
will see that a new input is being generated and a new suggestions bar will appear
right below the result:

  Range
4

 
 1, 2, 3, 4
6      

The Suggestions Bar is part of the predictive interface. It tries to guide you with
suggestions to simplify entries and ideas for further calculations. It is one more step
to reduce the time it takes to learn how to use the program.
All written instructions inside a cell will be executed sequentially.
To create a new cell we just need to place the cursor below the cell where we are
and then start writing. We can also create a new cell where the cursor is located by
pressing +.
To facilitate writing, Mathematica provides Palettes. These can be loaded by
clicking on Palettes in the menu bar. There are several palettes to make it easier to
type mathematical symbols such as the Basic Math Assistant or the Other  Basic
Math Input palettes. It would be useful from now on to keep one of them open.

Using a palette or the keyboard shortcuts write and execute the following expression. Please
remember: an empty space is the equivalent of a multiplication symbol: that is, instead of
45 you can write 4 5.
45


 a  4 b  32 b
5
 
 4a7b

Alternatively, you can use keyboard shortcuts to write special symbols, subscripts,
superscripts, etc. The most useful ones are: subscript  + _ , superscript  + ^
or  + 6, fraction  + / and the square root symbol  + 2. Depending on your
keyboard configuration, to use some of the previous symbols you may have to
press first +  (key CAPS) for example: superscript  +  + ^.
The previous Input cell was written in the standard format (StandardForm). It is the
one used by default in Input cells. To check it, position the cursor on the cell
marker and in the menu bar choose: Cell  Convert To . InputForm will appear
with a check mark in front of it.


 4 5 5 a  Sqrt
4 b  3 ^ 2 b;

The InputForm is very similar to the one used by other programming languages
such as C, FORTRAN or Python. For instance: multiplication = " * ", division = " /
", addition = " + ", subtraction = " - " , exponentiation = " ^ ", and square root of a
= "Sqrt
a". We can replace the multiplication symbol by an empty space. It is
very important to keep that in mind since Mathematica interprets differently "a2"
and "a 2": "a2" is a unique symbol and "a 2" is a 2. We have also added
semicolon “;”, thus the instruction will be executed, but its output will not be
shown. This is useful to hide intermediate results.
Initially, it was its symbolic calculation capabilities that made Mathematica popular
in the academic world.

As an example, write and execute the following in an input cell (use a palette to help you
enter the content).
     


 x,y Sinx y

cosx y 1
 
  x sinx y
2 y 2

A customized style, to which we will refer later, has been defined in this notebook
so that the outputs are shown in a format that corresponds to traditional
mathematical notation. In Mathematica it is called TraditionalForm. We could
have defined a style so that the inputs also use the traditional notation as well, but
in practice it is not recommended. However, you can convert a cell to the
TraditionalForm style anytime you want.

Copy the input cell from above to a new cell and in the menu bar select Cell  Convert To
 TraditionalForm. Notice that the cell bracket appears with a small vertical dashed line.
You will be able to use the same menu to convert to other formats.

2 sin









cosx y 1
 
  x sinx y
2 y 2

The natural language or free-form format enables us to write in plain English the
operation we want to execute or the problem we want to solve. To start using it,
press = at the beginning of the cell (input type) and you will notice that (“Free-
form Input”) will appear. After that, write your desired operation. Your instruction
will automatically be converted into standard Mathematica notation, and then the
answer will be displayed. Let’s take a look at some examples.

To add the numbers from 1 to 10 you can write “sum 1 to 10” or other similar English
expressions such as “sum from 1 to 10”.


  sum 1 to 10

Sum
n, n, 1, 10

   55

If you follow the previous steps, the entry proposed may be different from the one
shown here. The reason behind this is that when using the free-form format, the
program interprets your input by connecting to Wolfram|Alpha online, and the
knowledge engine is constantly evolving.
Move the cursor over Sum
i,i, 1, 10and the message “Replace cell with
this input” will appear. If you click on the symbol + located in the upper right-hand
corner, additional information about the executed operation will be shown.
If you have received any message related to connection problems, it is because the
free-form format and the use of data collections that we will refer to later on,
requires that your computer is connected to the Internet. If it is and you still have
connectivity problems it may be related to the firewall configuration in your
network. In that case you may have to configure the proxy or ask your network
 8      

administrator for assistance.

How do I test internet connectivity?: http://support.wolfram.com/kb/12420

To solve an equation such as 3 x2  2 x  4  0, you can literally type what you want
Mathematica to do. The output shows the correct Mathematica input syntax and the answer,
which in this case includes both, the exact solution and the decimal approximation.


  solve 3 x^2
2 x  4  0

Solve
3 x ^ 2  2 x  4 0, x

1 1
 
 x 1  13  , x  13  1
3 3

When you don’t know the correct function syntax, you can use the free-form
format to write what you want to do. Mathematica probably will show you the
correct syntax immediately; click on it and only the input in the correct notation
will remain with the free-form format entry disappearing.

The free-form input format generally works well when it’s dealing with brief
instructions in English. In practice it is very useful when one doesn’t know the
specific syntax to perform certain operations in Mathematica.
It’s also very useful to use Inline Free-form Input.

In this example, we assume we don’t know the syntax to define an integral. In an input cell
we click on Insert  Inline Free-form Input and will appear. Inside this box we type
integrate x^2.


 Integrate x^2

x3
 

3

The entry is automatically converted to standard WL (Wolfram Language) or Mathematica
syntax or by clicking over the symbol, the cell will go back to its free-form format.


 Integrate
x ^ 2, x

x3
 

3

By clicking in the input is automatically converted to the standard syntax in
Mathematica. It can be used as a template.


 Integrate
x ^ 2, x;

You can ask practically anything that can be computed. The below value is changing. The
distance from the Earth to the Moon varies depending on the position of the Moon in its
orbit around the Earth.


 Moon PLANETARY MOON
   !  

 
 370 017. km
    "

Here we combine the Inline Free-form Input with QuantityMagnitude , a Mathematica

function that gives the magnitude value of a quantity:


 QuantityMagnitude
earth gravity 

 
 9.80

The Mathematica free-form input is integrated with Wolfram|Alpha,

http://www.wolframalpha.com, a type of search engine that defines itself as a
computational knowledge engine and that instead of returning links like Google or
Bing, provides an answer to a given question in English. If you want, you can
access Wolfram|Alpha directly from within Mathematica defining a cell in the
NaturalLanguageInput style (or Shift +Alt+ 9) or writing “ ==” ( ) at the
beginning of the cell (input type).

major


 Wolfram|Alpha pod

 
 C4 D4 E4 F4 G4 A4 B4 C5

From the output, not shown, we selected the part that we were interested in (“Music
notation”) by clicking on the symbol and choosing the desired option (“Formatted pod”).
A new entry was generated displaying only the musical scale.
WolframAlpha
"Do major",
IncludePods  "MusicNotation", AppearanceElements  "Pods",
TimeConstraint  30, Automatic, Automatic, Automatic

If you click on Play sound you will hear the sound.

The most important difference between using the free-form input notation and
the Wolfram|Alpha one, is that in the first case the entry that we write is sent by
the program to a Wolfram server that will return the syntax in Mathematica and
it will be executed locally, while by using he the Wolfram|Alpha notation, all
the processing is done in the server, and our computer only displays the output.
If we have the program locally installed, it is better to use the free-form notation
to get the Mathematica syntax. In this way, we will be able to use the output
easily in subsequent calculations.

Wolfram|Alpha is specially useful when utilizing a browser since we don’t need to

have any software installed. Additionally, a new program named Wolfram|Alpha
Notebook Edition that uses the NaturalLanguageInput style by default, is available:
https://www.wolfram.com/wolfram-alpha-notebook-edition. Both applications can
10 Mathematica beyond mathematics

be used in many kind of devices (desktops, tablets, mobiles, and more.)

Besides all the previously described ways to define the cell type, there is another
one that consists of placing the cursor immediately next to the cell to which you
want to add another one. You will see “+” below the cell to its left. Click on it and
choose the cell type that you want to define: Input, , (Alpha Input), Text,
etc.

1.2.2 The Help System

If you have doubts about what function to employ, the best approach in many cases
is to use the free-form input format and see the proposed syntax. However, if you
would like to deepen your knowledge about the program you will need to use its
help system.
The help system consists of different modules or guides (Help Wolfram
Documentation) from where you can access an specific topic with its
corresponding instructions (guide/...). Additionally, you will find tutorials and
external links. Clicking on any topic will show its contents.

Figure 1.4 The Wolfram Language Documentation Center (partial screen).

In the screen-shot shown above (Figure 1.4), after clicking on Data Manipulation
& Analysis  Importing & Exporting, Mathematica displays the window below
(Figure 1.5). Notice that besides giving us the functions related to the chosen topic,
we can also access tutorials and even external resources such as video
presentations.

Figure 1.5 How to import and export using Mathematica.

If we don’t have a clear idea of the function we want to use or we remember it

1. Getting Started 11

vaguely, we can use the search toolbar (Figure 1.6) with two or three words related
to what we are looking for.

Figure 1.6 The Search Toolbar.

Another type of help is the one given when a mistake is made. In this case we
should pay attention to the text or sound messages that we have received. For
example, if we try to evaluate the following text cell, we will receive a beep.
“This is a text entry”
With Help  Why the Beep?... we will find out the reason behind the beep.
An extra source of help is the Input Assistant with its context-sensitive
autocompletion and function templates features: when you start writing a function,
Mathematica helps you automatically complete the code (similar to the system used
by some mobile phones for writing messages). For example: If you start typing Pl
Mathematica will display all the functions that start with those two letters right
below them (Figure 1.7). Once you find the appropriate entry, click on it and a
template will be added.

Figure 1.7 The Input Assistant in action.

Sometimes, a down arrow will be shown at the end of the function offering us
several templates (Figure 1.8). Choose the desired one.

Figure 1.8 Function templates for the Plot function.

Additionally, a color code enables syntax error detection. If after typing a function
its color is blue, probably it has not been entered correctly. All Mathematica and
12 Mathematica beyond mathematics

user-defined functions will be shown in black. For example: If you write in an input
cell Nmaximize, it will be displayed with a blue font. However, once you change it
to NMaximize, the function is now shown in black and a template is displayed.

1.3 Editing Notebooks

We’ve seen that the interaction with Mathematica is done through the Front End
(what we usually see when using Mathematica), and this generates files named
"Mathematica notebooks". What you are currently reading is one such notebook.
Each Mathematica notebook (*.nb) is a complete and interactive document
containing text, tables, graphics, calculations, and other elements. The program also
enables high-quality editing so notebooks can be ready for publication. You can
save notebooks in many different formats (*.pdf, *.tex, *.ps, *.cdf, ...).
It is important to keep in mind the task at hand when using Mathematica. For example, you may
want to prepare a report, create a tutorial or write a book chapter (this book has been written and
edited in Mathematica and saved as a PDF file). Therefore, the first thing to do before starting
using notebooks is to choose the appropriate style although you can always change it afterwards.

1.3.1 Notebook Structure

Each cell has a style. For example, the previous cell has the style Subsection
associated to it. This was done by selecting its cell marker (]) and choosing
Subsection in the style box.
Notebooks are organized automatically in a hierarchy of different cell types (title,
section, input, output, figures, etc). If you click on the exterior cell marker, the cells
will be grouped. For example, if you click on a section, all the cells associated to it
will collapse and only the section title will be visible. Click once more to see the
entire section again.
This is an ordinary text with the color formatted Format  Text Color.

Click on a text cell and modify its format.

With Insert  Hyperlink we can create hyperlinks to jump from one location to
another inside the same notebook, to another notebook or to an external link.

The Format menu contains options to change the font, face, size, color, etc.,
of the contents of the cells. We can even use special characters such as .
This notebook, like any other, has a predefined style. This style tells Mathematica
how to display its contents. There are many such styles. To choose one you can use
Format  Stylesheet. To define the style of a particular cell use Format  Style.
Additionally, inside a style, we can choose the appropriate working environment:
Format  Screen Environment. Try changing the appearance of a notebook by
modifying the working environment.
1. Getting Started 13

1.3.2 Choosing Your Style

To create or edit an article, book, or manual, the Palettes  Writing Assistant
palette is of great help. Load it. Note that is divided into three sections: Writing and
Formatting, Typesetting, and Help and Settings. See the help included in the palette.
If you click inside the palette on the button Stylesheet Chooser... (alternatively you
can go to Format  Stylesheet  Stylesheet Chooser...) a separate window with
icons representing different styles will appear (Figure 1.9). If you press on the
upper part of any of the icons (New) a new sample notebook formatted according to
that style will be generated. You will be able to use it and modify its contents to
suit your needs. If you click on the bottom part of the icons (Apply), the style will
be applied to the currently active notebook. For example, in the next screenshot
(Figure 1.9), by clicking on the upper part of the Textbook icon, a notebook with a
Textbook template style will be opened.

Figure 1.9 The Stylesheet Chooser.

When a new cell is created, depending on the stylesheet, by default it will be of a

certain format type, for example with the Default stylesheet new cells are of type
Input.

When typing in a new notebook choose the stylesheet from the beginning with
Format  Stylesheet  Stylesheet Chooser... . After clicking on the upper side
of the icon representing the desired style (New), you will be able to use and
modify the template according to your needs.

1.3.3 Typesetting Advice

If you are interested in emphasizing text containing formulas using classical
mathematical notation, begin by typing in a text cell. Select the formula and define
14      

it as Inline Math Cell by pressing  + 9 or, in the Writing Assistant palette,

choose Math Cells  Inline Math Cell. Finally, in the same palette in Cell
Properties click on Frame and select the appropriate one . That’s how the following
example was done:

What is the domain of the function f x  x2  1 ?

If you’d like to type a sequence of formulas and align them at the equal signs, in
Writing Assistant, choose Math Cells  Equal Symbol Aligned Math Cell.

ab  c
a  cb
Figure 1.10 Aligning equations at the equal sign.

Sometimes it may be useful to type both, the formula and its result, as part of the
text. This can be done by writing and executing the function in the same text cell.
For example: using a palette write x3  x = x3  x. Now select the second term of
the equation and press ++, or, in the menu bar, choose Evaluation 
4
Evaluate in place, and the previous expression will become x3  x = x4 .
When creating a document, you may be interested in seeing only the result. In that
case you can hide the Input cell.

To hide the input just click directly on the output marker cell. The following example was
created by executing Plot
Sin
x, x,2 Pi, 2 Pi and then hiding the
command to display only the graph.


 Plot
Sin
x, x, 2 Pi, 2 Pi
1.0

0.5

 

6 4 2 2 4 6
0.5

1.0

All Mathematica functions are in English, but the menus and palettes can also be
displayed in any of the additional 12 languages offered since version 11. The
alternatives include Chinese, French, German, Italian, Japanese, Korean and
Spanish among others. Users can take advantage of this functionality by going to
Edit  Preferences  Interfaces. Once an option is selected, every function will
be automatically annotated with a “code caption” in the chosen language. Since
Mathematica 11, the program includes a real-time spell checker that works with
any of the supported languages. Alternatively, users of previous versions can
review the English contents of a notebook with Edit  Check Spelling... .
With Edit  Preferences and with Format  Option Inspector (see
tutorial/OptionInspector) you will be able to customize many features of
Mathematica that can be applied globally, to a notebook or to a particular cell. For
    

example: with Format  Option Inspector  Global Options  File Locations

you can change Mathematica default directories.

1.3.4 The Traditional Style

We’ve seen that Mathematica by default uses the style: StandardForm in input
cells (Input). However, if we’d like to create professional looking documents, it’s
recommended to use the TraditionalForm style that is similar to the one used in
classical mathematical notation.
If the main purpose of a document is to be read by others, it may be convenient to
present the inputs and outputs using the traditional notation. It’s likely that the
reader will not realize that it was created with Mathematica. This is specially true
when the document is saved using the cdf format to which we will refer in the
following section.

You can write your inputs in StandardForm. Once you verify that they work,
you can convert them to the traditional style.

Example: Let’s type:
2
x  x  x2


 Limit
, x  0
x
 
 2x

We convert the cell to the traditional form by selecting it and clicking on:
Cell  Convert to  TraditionalForm. Then the cell above will be shown as follows:


 x2 
2


 lim
x0 x
 
 2x

If an entry is in the traditional style, to see how it was originally typed convert it
to the standard style by selecting the cell and choosing from the menu bar
Cell  Convert to  StandardForm.

We can make an output to be shown in the traditional form by adding

TraditionalForm at the end of each input. There are several ways to apply
this style to all the outputs in a notebook.

Go to Format  Option Inspector (by category) Cell Options  New Cell Defaults 
CommonDefaultFormatTypes and specify for “Output” TraditionalForm instead of
StandardForm. The form for “Input” can also changed to TraditionalForm, but it is
not usually recommendable.

Another easy way is to type the following:


 SetOptions
EvaluationNotebook
,
CommonDefaultFormatTypes  "Output"  TraditionalForm
It’s recommended to include the function above in a cell at the end of the notebook and
define it as an initialization cell (click on the cell marker and check Cell  Cell Properties 
Initialization Cell). This way, the cell will be the first one to be executed every time we
open the notebook.
16 Mathematica beyond mathematics

In many of the styles available in Stylesheet Chooser... outputs are displayed using
the standard form. In this chapter, we have used a customized stylesheet that shows
outputs in the traditional format. In the rest of the book, we will almost always use
the standard form since what we are trying to highlight is how to create functions
with Mathematica and for that purpose, the classical format is not adequate.

1.3.5 Automatic Numbering of Equations and Reference Creation

Textbooks and scientific articles frequently use numbered equations. This can be
done as follows:

First open a new notebook (File  New  Notebook) and select one of the stylesheets
available in the format menu. For instance Textbook: Format  Stylesheet  Book 
Textbook.

Then create a new cell and choose EquationNumbered as its style (Format  Style 
EquationNumbered). The cell will be automatically numbered, then you can write the
equation, for example: a + b = 1.
a+b=1 (1.1)

If you have opened a new notebook using the Textbook style, you will probably see
(0.1) instead of (1.1). It’s possible to modify the numbering scheme, but it is
beyond the scope of this chapter. When you need to number equations, sections,
and so on, remember that it is a good idea to select the stylesheet directly from the
Writing Assistant palette: Palettes  Writing Assistant  Stylesheet Chooser... .
After selecting Textbook (New), a new notebook will be generated with many
different numbered options such as equations, sections, etc. You can then use that
notebook as a template.
If you wish to insert an automatic reference to the equation proceed as follows:

Write an equation in a cell with the EquationNumbered style. Then go to Cell  Cell Tags
 Add/Remove... and add a tag to the formula. Let’s give it the tag: “par” (we write “par”,
from parabola, although we could have used any other name). A good idea would be also to
go to Cell  Cell Tags and check  Show Cell Tags, keeping it checked while creating the
document, and only unchecking it after we are done. This will enable us to see the cell tags
at all times. The first equation done this way is written below:
par
x2 = 1 (1.2)

If you wish to refer to this equation, type the following: (1. ), put the cursor after the dot and
go to Insert  Automatic Numbering. In the opening dialog (Figure 1.11) scroll down to
EquationNumbered, choose the tag of the equation, in our case: “par” and then press OK
(see the screen below). After this has been done, the number of the equation appears
between the parentheses: (1.2). If you click on the “2” the cursor will go immediately to the
actual equation.
1. Getting Started 17

Figure 1.11 Creating an automatically numbered object.

1.3.6 Cell Labels Display

As we have mentioned previously, any time we execute an input cell, a label in the
form In[n] is shown. However, we can choose not to display it. This can be done
from Edit  Preferences or typing SetOptions[SelectedNotebook[],
ShowCellLabel  False]. You can also reset the counter (set In[n] to 1) in the
middle of a session by typing the following in an input cell: $Line=0.

1.3.7 Save as...

Documents created with Mathematica are saved by default with extension “nb”. If
you share any of these documents with someone, that person will require access to
the program.
However, Mathematica can also save documents in other formats. To do that, in the
menu bar, choose File  Save As  . Then, from the list of options, select the
desired one.
Very often, we may be interested in saving a document in pdf format to reach a
broad audience. We can also use TeX (widely used in professional journals in
mathematics) or HTML/XML, if we want to use it in a website. Probably, it would
be better to save the web files in XML since everything can be included in a single
file. HTML, in contrast, contains a main file and numerous additional ones
associated to it. However, if we’d like to keep Mathematica’s interactive
capabilities, the most appropriate format would be cdf (Computable Document
Format).

1.3.8 The Computable Document Format (*.cdf)

As previously discussed, if you’d like other people without access to Mathematica
to read your documents without losing their interactivity, you should save them in
the cdf format: File  CDF Export  Standalone or File  Save As 
Computable Document (*.cdf).
18 Mathematica beyond mathematics

The main advantage of this format is that it keeps the interactivity of the objects
created with Manipulate. In this case, your documents can be read with the
Wolfram CDF Player (available for download for free from
http://www.wolfram.com/cdf-player). We recommend the inclusion of the player
web link when sending the file to ensure that the intended reader can open it .
The Wolfram CDF Player is not just a reader of Mathematica documents for users
without access to the program. It can also avoid displaying the functions used in the
inputs, showing only the outputs (remember the trick that we have seen previously
to hide the inputs and only show the outputs). Furthermore, it can include dynamic
objects created with Manipulate so that readers can experiment with changing
parameters and visualizing the corresponding effects in real time. Additionally,
readers will not even know that the original document was generated in
Mathematica. To see all these capabilities in action take a look at the following cdf
document:
http://www.wolfram.com/cdf/uses-examples/BriggsCochraneCalculus/BriggsCochraneCalculus.cdf
Many of the Wolfram Research websites, such as Wolfram|Alpha
(http://www.wolframalpha.com) , allow visitors to generate and download
documents in cdf format. The same happens with the Wolfram Demonstrations
Project (http://demonstrations.wolfram.com) where the files are available in cdf
format and you can download and run them locally.

1.3.9 The Wolfram Cloud

As stated at the beginning of the chapter, Mathematica can also be ran in the cloud,
in a special online platform named Wolfram Cloud:
http://www.wolframcloud.com/. This platform can also be used to store files
remotely and, as a matter of fact, all the files generated locally are interchangeable
with the cloud-generated ones. If you don’t have access to Mathematica locally but
you would like to start familiarizing yourself with the program, you can do so
directly using Wolfram Cloud’s free basic plan. If you are planning to use this
option heavily, the cost will depend on your requirements and whether you already
own a Mathematica license or not. In any case, you must register as a user using a
Wolfram ID.
After signing up, you can access it by visiting http://www.wolframcloud.com. If
you already registered in the user portal, you can use the same Wolfram ID.
Once you enter your Wolfram ID and password, the screenshot shown in Figure
1.12 will appear although you might see an initial screen that is somewhat different
since the online platform is constantly evolving.
1. Getting Started 19

Figure 1.12 Welcome Screen in the Wolfram Development Platform.

You can now create a new notebook by clicking on: Create a New Notebook, and
reproduce most of the examples discussed in the chapter. You can also save the
notebook online and when accessing the file again, no matter from what location,
continue working on it. Additionally, you have the possibility of accessing cloud
files from a local Mathematica installation: File  Open from Cloud... or File 
Save to Cloud.... Mathematica has specific functions related to the Wolfram
Cloud. In later chapters, we will refer to them.

1.4 Basic Ideas

Users with prior knowledge of other programming languages very often try to
apply the same style (usually called procedural) when using Mathematica. The
program itself makes it easy to do so since it has many similar instructions such as:
For, Do, or Print. Although this programming style works, we should try to avoid it
if we want to take full advantage of the Wolfram Language’s capabilities. For that
purpose we should use a functional style, which basically consists of building
operations in the same way as classical mathematics.
This section is for both, those unfamiliar with the WL (or Mathematica) and users
that still use procedural language routines. Because of that, if you have ever
programmed in other languages please try not to use the same programming
approach.
Although we use the formal Wolfram Language syntax in this section, many of the
examples shown below can be replicated using natural language.
In this brief summary, we will refer to some of the most frequently used functions
and ideas. We will often not explain the syntax or we’ll do it in a very concise way.
For additional details, just type the function, select it with the cursor, and press
<F1>. If you don’t understand everything that you read, don’t worry, you will be
given more information in later chapters.
20      

When you don’t know the syntax of a function, type it, select it with the cursor,
and press the <F1> key.

1.4.1 Some Initial Concepts

Let’s begin by writing the expression below using the Basic Math Assistant palette;
7
Cos
3  Sin
1


 expr1 5 x2 
3
2 y

 sin 1
 
  60  22 3 x2
7 y
This notebook uses a style where the outputs are shown in the traditional form.
We have assigned it the name expr1, to be able to use the result later on. To write
the transcendental numbers  and , we have taken advantage of special symbols (
is not the same as e; e is a letter without any other meaning) located in the Basic
Math Assistant palette. Alternatively, we could have used the keyboard and write
Pi for  and E^ or Exp[] for . Notice the use of a single equal sign "=" to indicate
sameness or equivalence in equations. For comparisons, we will type a double
equal sign "==".

The same operations can be written using the typical InputForm notation:


 5 Cos
3 Pi 2 ^ 1 3 x ^ 2  Pi Sin
1 E ^ 7 y
 sin 1
 
  60  22 3 x2
7 y

Here x is replaced with 3 and y with 5. To do that, you apply the following syntax: “expr .
rule” that will replace expr with the contents of rule.


 expr1 . x  3, y  5
 sin 1
 
  540  22 3
5 7
There is something that should draw our attention in this output. There are terms
that have not been evaluated: , sin(1), and 7 . This is because Mathematica does
not simplify or make approximations if that means losing precision. It literally
works with infinite precision. Nevertheless, we can force it to compute the decimal
approximation, the usual approach in other programs. To do that, we will use the
function N as shown in the next example. A similar result can be also obtained by
including decimals in some of the numbers.

In this example we use the % symbol to recall the last output. N or N[expr] use machine
precision in their calculations even though the output may show fewer decimals. By default,
5 decimals are shown, but with N

, n we can use a precision n as large as desired.


  N

 
 857.196

If you apply several replacement rules to the same expression in succession, the
replacement takes places consecutively:
    $

We first apply (/.) the rule (ab) to the expression a x  b y and then the second rule (bc)
is applied to the previous result.

  axby .a b . b c

 
 cxcy

When an assignment is no longer going to be used, it may be convenient to delete it

using Clear.

The following input removes the assignment previously associated to sol.


 Clear
expr1

Now, when we type sol we see that it has no assignment.


 expr1

 
 expr1

1.4.2 Functions
Mathematica contains thousands of prebuilt functions (Functions) that are always
capitalized and whose arguments are always written inside square brackets [arg].
It’s important to keep in mind that the program differentiates between lower and
upper cases. For example: NMinimize is not the same as Nminimize. However,
users can define a function using lower case.
Lists are a fundamental Mathematica concept, frequently used in many contexts. In
a list, elements are inside curly brackets {}. Later on we will refer to them.

Remember these five rules: All functions in the WL are capitalized: Log;
arguments are written inside square brackets: Sin
2 x  1; list elements are
inside curly brackets: {a1 , ..., an }; ranges are written as lists: Plot
x, x,
0, 2 Pi; and parentheses are used to group operations: Log
a x 
Sin
b x a  b .

Use Mathematica’s help files (Help  Wolfram Documentation) to investigate

functions and their arguments. Remember that you can select a function with the
cursor and press <F1> to get information about it. Besides its syntax, examples,
tutorials, and other related functions are displayed.

Write Pl in an input cell. The context-sensitive input assistant will show you all the
functions that start with Pl. Select Plot. Next to Plot a new down arrow will appear. Click on
it and a template like the one shown below will be created.
Plot
,  ,  ,  
Figure 1.13 Template for the Plot function.

Figure 1.13 shows an input type cell. If you’d like to avoid its evaluation, go to:
Cell  Cell Properties and uncheck Evaluable. That is what we did in this case in
the original document since the idea is to see the cell content but not to execute it.

In the previous template, all the basic function arguments are displayed. Now you can fill
them in using the options as the next example shows. Execute the Mathematica expression
below.
22      


 Plot
Cos
x, x, 0, 30

1.0

0.5

 

5 10 15 20 25 30
0.5

1.0

Figure 1.14 The Suggestions Bar for plots.

The resulting output is a plot with the suggestion bar (Figure 1.14) appearing right below it.
Thanks to the WolframPredictiveInterface, the Image Assistant and Drawing Tools provide
point-and-click image processing and graphics editing. Click on theme... and a menu with
several choices to customize the plot will unfold (Figure 1.15).

Figure 1.15 Customizing a plot using the “theme...” option in the Suggestions Bar.

Besides arguments, WL functions also have options. One way of seeing those
options directly is Options[func]:

Solve options:


 Options
Solve

 
 Assumptions  $Assumptions, Cubics Automatic, GeneratedParameters C,

InverseFunctions Automatic, MaxExtraConditions 0, Method Automatic,
Modulus 0, Quartics Automatic, VerifySolutions Automatic, WorkingPrecision 

Next, we’ll show some examples. Replicate them using the Basic Math Assistant
palette that includes templates for hundreds of commands.

Definite integral example:
4 3

   Sin
x Cos
x  x

3 sin x 1 1 1
 
  sin 3 x  sin 5 x  sin 7 x
64 64 320 448

Use Simplify anytime you need to simplify a complex expression.


 Simplifyx8  4 x6 y2  6 x4 y4  4 x2 y6  y8 
4
 
 x2  y2 
    $

There are also other functions to manipulate expressions. You can find them in the
palette Other  Algebraic Manipulation. Note that you can manipulate an entire
expression or just part of it.

Try to use it with this example simplifying the Sin and Cos arguments separately.
Sinx8  4 x6 y2  6 x4 y4  4 x2 y6  y8   Cosx4  2 x2 y2  y4 

To get:
4 2
Sinx2  y2    Cosx2  y2  

In this example a random distribution of spheres in a three-dimensional space is shown.

  Graphics3D
Yellow, Sphere
RandomInteger
5 , 5, 10, 3

 


You can rotate the image to see it from different angles by clicking with the mouse
cursor inside of it.
If you haven’t used Mathematica previously, the last command may seem strange.
Let’s analyze it step by step:

Highlight RandomInteger and press <F1>. The help page for the command will be
displayed. In this case with RandomInteger
5, 5, 10, 3 we are generating 10
sublists with 3 elements, each element being a random number between 5 and 5. We are
going to use them to simulate the coordinates {x, y, z}.

The previous function is inside Sphere (do the same, highlight Sphere and press <F1>) a
command to define a sphere or a collection of spheres with the coordinates

x1 , y1 , z1 , x2 , y2 , z2 , , r, where r is the radii of the spheres. If omitted, as
in this example, it’s assumed that r  1.


 Sphere
RandomInteger
5, 5, 10, 3
4 1 4
2 3 2
4 4 1
1 4 4
2 1 0
 
 Sphere 
5 4 2
2 5 5
1 5 2
4 5 4
2 3 1
Note: In this particular notebook, list outputs are sometimes shown as matrices.
24      

We use the command Graphics3D to display the graphs. We have added Yellow
to indicate that the graphs, the spheres in this case, should be yellow and the option
ImageSize to fix the graph size and make its proportions more adequate for this
example. Finally we arrive at the instruction:


 Graphics3D
Yellow, Sphere

 

 


We have iconized the argument inside Sphere (highlight the part of expression to be
iconized, right-click on it and then select Un/Iconize selection).

When you don’t understand an instruction consisting of several different

commands, decompose it into its individual instructions, and analyze each one
separately using the help system.

In the next cell, we use Rotate to show a rotated output, in this case by 1 radian, compared
to the usual display.


 Rotate
1 Sqrt
1  x, 1
 Type inside the output and you'll see that it retains its properties.
1

 

x

In the above evaluation cell (Input) we can include a comment in the following format: (*
comment *). Nevertheless, it is recommended to include the comments in separate cells in
text format. Remember that you can do that by choosing Text or other format in the style box.

Practically anything can be used as a symbol. For example, let’s first generate a small red
sphere:


 Graphics3D
Red, Sphere
, ImageSize  30

 


Then, substitute the symbol thus obtained, , and paste it below replacing b. Finally,
evaluate the cell.
3


 Expand1  b 

 
 b3  3 b2  3 b  1
    $

3


 Expand
1  

3 2
 
 3 3 1

1.4.3 List and Association

Lists (List) are a fundamental concept in the WL. Lists or ranges are written
inside curly brackets. The list elements are separated by commas: {a, b, ...}. For
a description of list operations see: guide/ListManipulation. We’ll describe them
in more detail in later chapters.

Let’s create a lists of lists of random numbers.


 mat RandomInteger
10, 5, 5;

We add InputForm to see the output displayed in the internal format used by the WL.
Matrices are represented as lists of lists.


 mat InputForm
 
%%
# &
9, 5, 3, 7, 3, 6, 1, 10, 10, 10, 6, 8, 8, 8, 6, 9, 7, 8, 0, 3, 10,

If desired, matrices can also be displayed in standard mathematical notation.


 MatrixForm
mat
 
%% '&
9 5 3 7 3
6 1 10 10 10
6 8 8 8 6
9 7 8 0 3
10 4 7 7 6

Since they are lists, matrices can be handled as such. For example, here we extract [[...]] (or
Part[...]) the 4th element of the 2nd sublist 2nd row, 4th column):


 mat2, 4

 
 10

In the expression below we assign the symbol sol to the solution of a x2  b x  c  0.


 sol Solvea x2  b x  c 0, x;

Remember that for assignments, a single equal sign "=" is used, while to
indicate an equality in an equation or a comparison, we will use a double equal
sign "==" that when typed will be automatically converted to “”.

The output is a list.


 sol

 b2  4 a c  b b2  4 a c  b
 
 x , x
2a 2a

Next, we extract the second solution (note that both solutions are inside a list):
26      


 sol2

b2  4 a c  b
 
 x
2a

To verify that the previous result is correct, we use "expr . rule", replacing the x in the
equation with its values from the solution.


 a x2  b x  c 0 . sol
2
 b2  4 a c  b b  b2  4 a c  b
 
   c  0,
4a 2a
2
 b2  4 a c  b  b  b2  4 a c  b
 c0
4a 2a
Although we can see that the substitution has taken place, it’s often convenient to
simplify it. For that we use Simplify[, where % enables us to call the result (Out)
of the last entry and evaluate the cell. This method is very useful in complex
situations where we want to check that Mathematica actually returns the correct
answer.

We can check that the equality holds and therefore the solution is correct.


 Simplify


 
 True, True

For further details about matrix operations you can consult the documentation:
guide/MatrixOperations.

Association
key1
val1 , key2
val2 ,  is another important structure in the

WL. It represents an association between keys and values.


 assoc "Height"  175, "Weight"  75, "Sex"  "Female"

 
 Height 175, Weight 75, Sex Female

Keys extracts keys from an association and Values extracts values:


 Keys
assoc

 
 Height, Weight, Sex


 Values
assoc

 
 175, 75, Female

We can query the value associated with the key "Weight":


 assoc"Weight"

 
 75
    $

1.4.4 Defining Your Own Functions

Mathematica makes easy define functions in a similar way to mathematics. The

usual syntax in the case of a function of one variable is: 
 
.
The blank ("_") next to the variable is used to specify the independent function
variable.

Let’s see an example:


 f
x : 0.2 Cos0.3 x2 

Now we can assign a value to the independent variable and we’ll get the function value.


 f
3

 
 0.180814

We could have also typed the previous two operations in the same cell (don’t forget in this
case to enter “;” at the end of each function). However, it’s better to use one cell for each
function until you have enough practice. It will help you find mistakes.


 f
x : 0.2 Cos0.3 x2 ;
f
3
   0.180814

Now we are ready to visualize the Derivative (') of f
x in a specific interval.


 Plot
f '
x, x, 0, 10

1.0

0.5
 

2 4 6 8 10
0.5

1.0

The same approach can be extended to several variables. In this case, we use the previously
defined f x function.


 g
x, y : f
x 2.3 Exp 0.3 y2 

Now we can present the result in a 3D graph with Plot3D. We use the option PlotRange
All to display all the points in the range for which the function is calculated. Remove the
option to see what happens.


 Plot3D
g
x, y, x, 2 , 2 , y, 2 , 2 , PlotRange  All

 

28      

1.4.5 Pure Functions

Pure functions are often used when programming in the WL, and they will be
described in detail in Chapter 2. To create them, we can either use Function or a
combination of & and # symbols.

Here is an example of a pure function (f x  x^2  3):


 f  ^ 2  3 &

 
 12  3 &


 f
3

 
 12

1.4.6 Example Data

The installation of the program includes a set of files with data from different fields
that can be used to test certain functions. We will use some of the data in this
book. To download them, use ExampleData.
To find out all the areas for which there are examples available, just type
ExampleData[].

The command below downloads the original text of Charles Darwin’s book On the Origin of
Species:


 txt ExampleData
"Text", "OriginOfSpecies";

To find out how many times the word “evolution” appears compared to the word “selection”
we use the function StringCount, specifying the chosen word or text.


 StringCount
txt, "evolution"

 
 4


 StringCount
txt, "selection"

 
 351

A fundamental feature of Mathematica is its capability for importing and exporting

files in different formats. The functions that we need for that are Import and
Export. They enable us to import or export many different types: XLS, CSV, TSV,
DBF, MDB, ...

In this case we import the first sheet of an XLSX file from the ExampleData folder:


 Import
"ExampleData cities.xlsx", "Data", 1
City Country Population
Tokyo Japan 8.3  106
 
 Chicago United States 2.8  106
London United Kingdom 7.4  106
Berlin Germany 3.4  106

1.4.7 Additional Functions and Packages

The latest releases of Mathematica contain more than 7,000 built-in functions.
However, it’s always possible to add extra functionality by creating your own.
    $"

Apart from that, you can also use functions created by other Mathematica users
located in the Wolfram Function Repository:
https://resources.wolframcloud.com/FunctionRepository. They can be accessed
using ResourceFunction:

We look for the functions related to astronomy (the output is not shown).


 ResourceSearch
"Astronomy"

“AstroDistance” computes the current distance between Mars and Jupiter:


 ResourceFunction
"AstroDistance"

Entity
"Planet", "Mars", Entity
"Planet", "Jupiter"
 
 3.91542 au

Several functions can be combined to develop applications for specific purposes

(packages). The packages included in Mathematica can be seen in Documentation
Center  Add-ons and Packages.
Additional packages (add-ons) developed by users for specific purposes are
available in http://packagedata.net/.
Normally, all these packages will be located in Addons/Applications, a subfolder in
the Mathematica installation directory.
To load a package you can type Needs[“package name`”] or <<package name`. For
example, the following package includes various functions related to the properties
of the radiation emitted by a black body.


 Needs
"BlackBodyRadiation`"


 ? "BlackBodyRadiation` "

BlackBodyRadiation`
 

BlackBodyProfile MaxPower PeakWavelength TotalPower

The following package function calculates the peak wavelength of a black body for a given
temperature.


 PeakWavelength
5000 Kelvin

 
 5.79554  107 Meter

Chapter 2 provides a brief introduction to package development with actual

examples interspersed throughout the rest of the book.

1.5 From Graphics to Machine Learning

1.5.1 Graphics
One of the most remarkable aspects of Mathematica is its graphical capabilities.
The program’s tool for 2D graphics can be accessed by pressing +D or by
selecting in the menu bar Graphics  Drawing Tool.
The most commonly used functions to represent graphics are: Plot, Plot3D, and
30      

ListPlot. However, there are many more.

When typing Plot the context-sensitive input assistant will show you all the words that
include the word Plot, browse them. Alternatively type the following command and execute
it (its output is omitted):
? Plot

The names that are shown can give you a hint about the type of plot they generate.
Click on the chosen name to get detailed information. In some cases, you will be
referred to the plot options.

The graphical functions don’t end there. The ones that include the word Chart are usually
related to statistical graphs. Similarly to the previous case, you can use the expression:
? Chart

Here we use a figure known as the “Utah Teapot”, included in ExampleData, to make a 3D
bar chart.


 g ExampleData
"Geometry3D", "UtahTeapot";


 BarChart3D
1, 2, 3, ChartElements  g, BoxRatios  4, 1, 1

 


In the following example we use the options AxesLabel to add labels to the axes and
PlotLegends  “Expressions” to show the represented functions:


 Plot
Sin
x, Cos
x, Tan
x, x, 0, 2 Pi,
AxesLabel  "x", "fx ", PlotLegends  "Expressions"
fx
4

2 sin'
 
 cos'
x
1 2 3 4 5 6
tan'
2

4

Since Mathematica 13.0, we can create plots with multiple axes:

    


 ListLinePlot
Table
Table
x ^ n, x, 40, n, 3, MultiaxisArrangement  All

40
1500 60 000

50 000
30

1000 40 000

 
 20
30 000

500 20 000
10
10 000

0 0 0
0 10 20 30 40

We can even find the intersection points of a circumference and a parabola and graphically
represent them. To write special symbols such as “” , the best thing to do is to use
Mathematica’s palettes, which we will discuss later on.
3


 pts  Solve
2  2  1   2
2   0, 
, 
2
1 1 1 1 1 1
 
 x  5  5, y 1  5 , x 5  5, y 1  5 ,
2 2 4 2 2 4
1 1 1 1 1 1
x  5  5, y  5  1 , x 5  5, y  5  1
2 2 4 2 2 4

Show

3
ContourPlot
x2  y2 1, y  2 x2  0, x, 1.5, 1.5, y, 1.5, 1.5,
2
Graphics
Red, PointSize
Medium, Point
x, y . pts

1.5

1.0

0.5

0.0
 


0.5

1.0

1.5
1.5 1.0 0.5 0.0 0.5 1.0 1.5

Graphics-related functions include numerous options (output omitted):

Options
Plot3D

The next cell displays the surface cos x sin y bounded by the ring 1  x2  y2  4 (To write
“” you can type “<=”). Note that to set the region where the function exists, we use the
RegionFunction option.
32      


 Plot3D
Cos
x  Sin
y, x, 2, 2, y, 2, 2,
RegionFunction  Function
x, y, z, 1  x ^ 2  y ^ 2  4

 


RegionPlot3D enables us to plot multiple regions at a time:


 RegionPlot3D
1  x ^ 2  y ^ 2  2, 1  x ^ 2  z ^ 2  2, 1  y ^ 2  z ^ 2  2,
x, 2, 2, y, 2, 2, z, 2, 2, Mesh  None

 


You can interact with 3D graphics: Click on the graph to rotate it, with  + click
(or -click) you can zoom, and with +click you can move it around.

The code below generates the famous Lissajous curves. Using Grid we can show a two-
dimensional mesh of objects, in this case graphs. We also use the Tooltip function to
display labels related to objects (not only graphics) that will be shown when the mouse
pointer hovers over the objects. Here, when the mouse pointer is over the chosen curve,
you’ll see the function that generated it.
    


 Grid
Table
Tooltip
ParametricPlot
Sin
n t, Sin
m t, t, 0, 2 Pi,

ImageSize  70, Frame  True, FrameTicks  None, Axes  False,
Sin
n t, Sin
m t, m, 3, n, 3

 


Other functions that do not include Plot or Chart but that are also useful for
creating graphs are Graphics, Graphics3D, and ColorData, as the following
example shows.

Representation of the water molecule.


 Graphics3D
Specularity
White, 50, ColorData
"Atoms", "H",
Sphere
0, 0, 0, .7, Sphere
1.4, 0, 0, .7,
ColorData
"Atoms", "O", Sphere
.7, 0, .7, Lighting  "Neutral"

 


Using Canvas, we can annotate existing graphs (click inside):


 Canvas
Plot
Sin
x, x, 0, 2 
34      

CSGRegion a new function very useful for CAD Modeling


 CSGRegion
"Union", Cuboid
30, 30, 1, 30, 30, 1,
Cylinder
30, 0, 0, 30, 0, 0, 10,
Cylinder
0, 30, 0, 0, 30, 0, 10

If you want to learn more about graphics, you can select the relevant text with the
cursor and press <F1>. For example, type: tutorial/GraphicsAndSoundOverview,
select it and press <F1>.

1.5.2 Images
Image processing has experienced a significant improvement. When clicking on an
image you will see a menu with numerous options (Figure 1.16).

The following command loads an image.
ExampleData
"TestImage", "Lena"

Figure 1.16 The Image Assistant.

The Image Assistant appears below the image. Click on any of available options. For
example, after clicking on more... we see the following menu (Figure 1.17):

Figure 1.17 Additional capabilities available from the “more... option” in the Image Assistant.
    

After selecting one of the Actions & Tools, a miniature version of the image appears where
you can check the effect on the original image. For example: When selecting invert colors,
you’ll see Figure 1.18.

Figure 1.18 Inverting the colors using the Image Assistant.

After getting the desired effect, press Apply to modify the image. An alternative
way to do that is to select the function directly from the menu:
ColorNegate[image]. Copy it to an input cell and execute it. This last procedure
is the appropriate one if you want to create a program or explain how the image has
been modified. We will be using this method when we refer to image processing in
later chapters.

1.5.3 Manipulate
Mathematica has functionality to enable dynamic and interactive operations. One
of the most significant examples of this is the Manipulate function.

Type the Mathematica expression shown below. Notice how sliders are created for each of
the parameters being defined. Click on the + symbol included in the output to show the
buttons created by Manipulate.


 Manipulate
Plot
Sin
frequency x  alpha, x, 2 , 2 ,
frequency, 1, 5, alpha, 0, Pi 2

In http://demonstrations.wolfram.com, you can find thousands of small interactive

applications of great educational value. Throughout this book, we’ll refer to them
using the word “demonstrations”. If you don’t know how to build them just
36      

download and open the files; later on you’ll be able to modify their code to suit
your needs or even make new ones. In Chapter 4, we’ll learn more details.

This example explores Coulomb’s law. This law describes the interaction between two
charges (based on ‘Potential Field of Two Charges’ created by Herbert W. Franke:
http://demonstrations.wolfram.com/PotentialFieldOfTwoCharges/).


 Manipulate
ContourPlot
q1 Norm
x, y  p1  q2 Norm
x, y  p2,
x, 2, 2, y, 2, 2, Contours  10,
q1, 1, 3, 3, q2, 1, 3, 3,
p, 1, 0, 1, 0, 1, 1, 1, 1, Locator, Deployed  True

Move the locators around with the mouse to see how the potential fields change.

1.5.4 Gauges
Mathematica includes different types of customizable gauges that can behave
dynamically:


 Through
AngularGauge, VerticalGauge, ThermometerGauge, HorizontalGauge

42, 0, 100, ImageSize  Tiny

100
40 60 80
60
 
 20 80 , , ,
40 0 20 40 60 80 100
0 100 20
0

With just a few lines of code, we can generate two dynamic clocks synchronized with both,
our local time and a different time zone.


 Dynamic
Refresh
Row

ClockGauge
AbsoluteTime
, PlotLabel  Style
"Local", Large, Bold,
ClockGauge
AbsoluteTime
TimeZone  9,
PlotLabel  Style
"Tokyo", Large, Bold, UpdateInterval  1 
    

If you find the previous example difficult to understand, remember that you can always
access the documentation to familiarize yourself with the functions used to generate the
clocks.

1.5.5 Things to Remember

It’s important to keep in mind that inputs are executed based on the order of
evaluation and not the order in which they are shown on the screen. This means
that if a function calls previously defined functions, those functions should have
been evaluated in advance. It’s not enough to see them already typed. For example,
if you modify an assignment that is used later on, you will have to execute again all
the related cells.

Evaluate sequentially the following three cells:

  a 3  first

 
 3


 b xa  second

 
 x3


 a b third

 
 3 x3

Now assign a the value of 2 in the first cell. What happens when you re-execute the
third instruction (a b)?

  retains information about all the variables defined in a

session, even after starting a new  . Because of that, when starting a
new section that will not be using the variables defined previously, it is a good
idea to clear them first.

One way to remove all the values and definitions is as follows:


 ClearAll
"Global` "

There’s another way to limit the use of variables to a specific context:

Define this line as a subsubsection.

Select all the cells corresponding to this subsubsection and in the main menu choose:
Evaluation  Notebook's Default Context  Unique to Each Cell Group.

Define the function below:


 f
x : 3x


 f
3

 
 9
38      

Define this line as a new subsubsection.

If you execute the same function f, you’ll see that is not defined.


 f
3

 
 f 3

If you really want to remove all the information and not only the variables, the best
way is to quit the session. This requires that you exit Mathematica (actually it
requires that you exit the program Kernel). You can do that in several ways: With
File  Exit or if you don’t want to exit the notebook use either Evaluation  Quit
Kernel or in an input cell type Quit[]. Once you have exited, the program removes
from memory all the assignments and definitions. When you execute the next
command the kernel will be loaded again.
Although Mathematica tries to maintain compatibility with notebooks created in
previous versions, it is not always complete so you may have to make some
modifications. When you open for the first time a notebook created in a previous
Mathematica version, the program will offer you the possibility of checking the file
automatically; Accept it and read the comments that you may receive carefully. In
many cases, Mathematica will modify expressions directly and let you know about
the changes made giving you the option to accept or reject them. In other cases it
will offer suggestions.

Remember: If you don’t understand a function use the Documentation Center.

1.5.6 Entities
The Wolfram Language (WL) contains real-world data accessible through the
entity framework, describe in detail in Chapter 4.

For example, to find out the population of Japan, we would just ask Mathematica for the
population value corresponding to the “Japan” entity. Using (Insert  Inline Free-
form Input) we write “Japan” and “Population” inside EntityValue:
EntityValue Japan , Population 


 EntityValue
Japan COUNTRY , ##(  

 
 126 476 458 people

1.5.7 Wikipedia

We can even obtain information from Wikipedia using the function WikipediaData.
    "


 WikipediaData
"Alhambra", "ImageList"3

 


1.5.8 Machine Learning

Mathematica also includes machine learning functions. Let’s see an example.

Handwrite different numbers on a white piece of paper, scan them or type them in a touch
screen (in MS Windows you can use the Crop application) and label them:


 labeledData   2,  5,  8,  0,  2,  7,  5,  1,  3,
 0,  3,  9,  6,  2,  8,  2,  0,  6,  6,
 1,  1,  7,  8,  5,  0,  4,  7,  6,  0,  2,
 5,  3,  1,  5,  6,  7,  5,  4,  1,  9,
 3,  6,  8,  0,  9,  3,  0,  3,  7,  4,

 4,  3,  8,  0,  4,  1,  3,  7,  6,  4,

 7,  2,  7,  2,  5,  2,  0,  9,  8,  9,

 8,  1,  6,  4,  8,  5,  8,  0,  6,  7,
 4,  5,  8,  4,  3,  1,  5,  1,  9,  9,

 9,  2,  4,  7,  3,  1,  9,  2,  9,  6;

Use the command Classify to teach the program to associate the handwriting to a certain
digit based on the examples. This function uses statistical criteria to assign a weight to each
potential match.


 digits Classify
labeledData

 

 
 ClassifierFunction 
     

Copy from above the numbers 0 to 10 in a list like the following:

  digits , , , , , , , , , 

 
 0, 1, 2, 3, 4, 5, 6, 7, 6, 4

Note that the classification has been done correctly except in the case of 6 that has been
mistaken with the digit 4. The program assigns probabilities based on the stored data. You
can check that the probability assigned to the symbol is higher for 4 than for 6. Even a
person may not be sure whether the handwriting represents a 6 or a 4.
40      


 digits , "TopProbabilities"

 
 6 0.82

If 6 is written in a less equivocal way, the probability of a correct identification improves

substantially.


 digits , "TopProbabilities"

 
 6 0.82

1.6 Additional Resources and Supplementary Materials

Supplementary materials are available on the book’s website:
http://www.mathematicabeyondmathematics.com/
The Documentation Center is the primary source about Mathematica and the
Wolfram Language: Help  Documentation Center or
http://reference.wolfram.com/language/
You may also find the following links useful:
An Elementary Introduction to the Wolfram Language by Stephen Wolfram:
http://www.wolfram.com/language/elementary-introduction/ (there is also a print version).
Introductory tutorials:
http://reference.wolfram.com/language/tutorial/IntroductionOverview.html
Explanatory video guides: http://www.wolfram.com/broadcast
Open courses: https://www.wolfram.com/wolfram-u/
News and ideas: http://blog.wolfram.com , including a short description of the additional
capabilities of new releases: https://blog.wolfram.com/version-releases/
For an overview of all the free resources that Wolfram Research makes available,
visit:
https://wolfram.com/open-materials/
 2
Programming: The Beauty and Power of
the Wolfram Language

This chapter covers some of the concepts and functions that we need to
program in Mathematica using WL. It also explains some useful notions
for generating efficient code. In particular, we will learn how to code
using pure functions, very important to create compact programs that take
full advantage of the power of Mathematica. The upcoming sections also
include examples of some of the most commonly used programming
functions and describe the fundamental ideas behind package
development, essential to build new applications that incorporate user-
defined functionality. Finally, guidance will be provided on tools
available for large projects or to develop programs that can be run in the
cloud. This chapter may be the least fun, but it is important to have a
basic understanding of the concepts discussed in its pages to be able to
follow the rest of the book. If you are an advanced user you may skip it,
although we recommend that you at least skim through its contents in case
you find something useful.

2.1 Mathematica’s Programming Language: The Wolfram Language

As we have seen already in previous chapters, Mathematica is not only a program
to perform symbolic and numeric calculations. It is a complete technical system
that can be used to develop anything: from traffic control applications to
sophisticated image manipulation interfaces. All of this is possible thanks to the
power of the Wolfram Language, the high-level general-purpose programming
language used by Mathematica. To start using it we need to know its fundamentals,
and that’s what we are going to learn in this chapter. This section will cover the
basic components of the language.

2.1.1 Atoms and Expressions

Everything in the Wolfram Language is an expression, including notebooks.
Expressions are written in the form: Head[..., {}] and are ultimately built from a

      

small number of distinct types of atomic elements: Integer, Real, Rational,

Complex, String, and Symbol. Do not get confused with the domains: Integers (
),
Reals (), Rationals () and Complexes ().
 AtomQ can be used to test whether an expression is an atom:


 10
3 ;

 % gives the last result. %% gives the result before last.



( times) gives the  th
previous result.

  AtomQ

 
 True


 Head

 
 Rational

Head is always capitalized and followed by a pair of enclosing square brackets

containing elements separated by commas. Those elements sometimes go inside
braces. Here is one example:


 Heada  b  c

 
 Plus

 Instead of x + y + z we write:


 Plusx, y, z

 
 #

 With FullForm we can see the basic form of expressions. One head, Plus in this case, and
elements separated by commas inside brackets:


 FullFormx  y  z
 
  
Plus, , #

Cells are also expressions. Shown below is a copy of this cell displaying its
contents. This was done by duplicating this cell, selecting the copy and clicking on:

   .
Cell[TextData[{“Cells are also expressions. Shown below is a copy of this cell displaying its contents. This was done by selecting the
copied cell and clicking on: “,
StyleBox[“Cell”, FontWeight->”Bold”], “  “, StyleBox[“Show Expression”, FontWeight->”Bold”], “. “}], “Text AM”,
CellChangeTimes->{{3.6*^9, 3.6*^9}, {3.6*^9, 3.6396977*^9}, {3.6*^9, 3.6396978*^9}, {3.6*^9, 3.6*^9}, {3.6*^9, 3.6*^9}}]

All objects in     are one of the following types:

 Numbers: (i) Integer: Head used for integers, (ii) Real: Head of number with a decimal
separator (floating-point), (iii) Rational: Head of a rational number (ratio between two
integers such as 3/4 or 27/8), (iv) Complex: Head of a complex (sum of a real and an
imaginary number: 3 + 2 ).
 Symbols: A sequence of letters, digits, or symbols not starting with a number (we will
describe them later).
 Strings: Any expression between double quotation marks: “this is a  ”.
 SparseArrays: Matrices or lists in which most of the elements are 0:
  !  " # $ #%



 MatrixFormSparseArrayi, i
RandomInteger8, 4, 4

     
7 0 0 0
0 0 0 0
0 0 8 0
0 0 0 6


 AtomQ

 
 True


 Head

 
 SparseArray

Graphs: A collection of objects, joined by vertices in pairs.


 Graph1  2, 2  3, 3  1

 



 AtomQ

 
 True


 Head

 
 Graph

    internally always separates expressions first into their atomic

components before computing them.
 As mentioned before, we will often use as a precaution the command ClearAll
[“Global`*”] at the end or at the beginning of a section to avoid problems with variables
previously defined:


 ClearAll"Global` "

2.1.2 Levels in Expressions

We’ve seen that     manipulates basic objects located at the head of each
expression (Head).
 Remember that with FullForm we can see the basic form of expressions. The example
below shows that    internally is  multiplied by 1 .


 FullForm a
b
 
  
Times, Power, 1

 Here we apply FullForm to a more complicated expression.


 FullForm1.5 Sin2 Pi
4
 
  
1.5`


     

 We get the result of the operation but not the structure of the initial expression as we
wanted. This is because the calculation is done first and then FullForm is applied to the
result, just a number. To avoid the execution of the calculation we use HoldForm.


 FullForm HoldForm1.5 Sin2 Pi
4
 
  
HoldFormTimes1.5`, SinTimes2, TimesPi, Power4, 1

 The structure of any expression adopts a tree-like format. Here we represent the previous
expression in such a format using TreeForm.


 expr HoldForm1.5 Sin2 Pi
4
2
 
 1.5 sin
4


 TreeForm expr
 
  

HoldForm

Times

1.5 Sin

Times

2 Times

Power

4
1

 Depth gives the total number of levels:


 Depthexpr

 
 7

 LeafCount gives the total number of individual atomic elements:


 LeafCountexpr

 
 11

 Accessing individual levels:


 Levelexpr, 4

 
 2,
4
If we analyze the previous tree we can see that the first expression is 1/4. Internally
it is represented as 41 . Next, that expression is multiplied by Pi, the result
multiplied by 2, and so on until reaching the top of the tree.
 Trace enables us to see the order in which operations are executed:
  !  " # $ #%



 Trace 1.5 Sin2 Pi
4

1 1   2 2 2  
 
 , , , , , , , , sin , 1 , 1.5  1, 1.5
4 4 4 4 4 4 4 2 2
 The previous output seems to duplicate certain steps. However, with FullForm we can see
that it’s actually the basic structure that is changing. So when we see  1 , 1
, internally
4 4
1 1
there has been a structural change: the expression has changed from (4 ) to ( ).
4


 FullForm
 
  
ListListListListListHoldFormPower4, 1, HoldFormRational1, 4,
HoldFormTimesPi, Rational1, 4, HoldFormTimesRational1, 4, Pi,
HoldFormTimes2, TimesRational1, 4, Pi, HoldFormTimes2, Rational1, 4, Pi,
HoldFormTimesRational1, 4, 2, Pi, HoldFormTimesRational1, 2, Pi,
HoldFormSinTimesRational1, 2, Pi, HoldForm1,
HoldFormTimes1.5`, 1, HoldForm1.5`
 With a simple transformation rule we can modify the previous expression so that
multiplication becomes addition.


 HoldForm1.5 Sin2 Pi
4
. Times  Plus
1
 
 1.5  sin 2   
4
 The next function shows a red circumference of radius 1.

  gr GraphicsRed, Circle0, 0, 1

 


 Its internal form is as follows:


 FullFormgr
 
  
GraphicsListRGBColor1, 0, 0, CircleList0, 0, 1

  TreeFormgr

 
  

Graphics

List

RGBColor Circle

1 0 0 List 1

0 0

If you move the cursor over an individual frame you will notice that it displays the
result of the operations until that frame.

      

 A small transformation will allow us to modify the expression so that instead of a red
circumference (Circle), we get a red circle (Disk).


 gr
. Circle  Disk

 


Once we know the structure of an expression, we will be able to modify it.

Additionally, when encountering outputs from operations that we don’t understand,
their internal form can give us clues about how the program works.

2.1.3 The Wolfram Language Paradigm

At the core of the Wolfram Language (or    ) is the foundational idea
that everything—data, programs, formulas, graphics, documents—can be
represented as symbolic expressions. And it is this unifying concept that
underlies the Wolfram Language symbolic programming paradigm.

$  are the ultimate atoms of symbolic data letters, digits, combinations of
letters and digits not starting with a digit, and special symbols such as: $, , , ,
, . Upper- and lower-case letters are distinguished. Once a symbol has been
created during a session, it will remain in memory unless explicitly removed. An
alternative is to use functions that automatically delete a symbol once it’s been
used.
 If we have a list and we’d like to know the type of objects that it contains, we apply Head to
each of the list elements. This can be done using Map..


 MapHead, 3, 3
2, 1.3, 3  4 I, I, 2 , Pi, a, b, a, , 

 
 Integer, Rational, Real, Complex, Complex, Power, Symbol, List, Symbol, Graphics, Image


 ClearAll"Global` "

2.2 Lists Operations

A basic structure in     is the list. The elements of a list can be numbers,
variables, functions, strings, other lists, or any combination thereof; they are written
inside braces  and are separated by commas. If we feel comfortable with the
classical terminology of mathematics we can interpret a simple list as a vector, a
two-dimensional list (a list of sublists) as a matrix, and a list of three or more
dimensions as a tensor, keeping in mind that the tensor elements can be sublists of
any dimension.
 Below we define a list consisting of three sublists.


 list 2, 3, 6, 3, 1, 5, 4, 3, 7;
  !  " # $ #%



 list

TreeForm

 
  

List

List List List

2 3 6 3 1 5 4 3 7

 The previous list can be shown in traditional notation as a matrix, but we should not forget
that internally     treats it as a list.


 list

TraditionalForm
 
    
2 3 6
3 1 5
4 3 7
If the notebook has been configured to display outputs using the traditional format (as in this
case), it’s not necessary to use //MatrixForm.

  MatrixForm1, 2, 3, 4 
 
   
1 2
3 4
 Length gives the number of elements in the list and for list with sublists Dimensions can be
used to determine the length at each level:


 Lengthlist, Dimensionslist

 
 3, 3, 3

It’s important to distinguish operations between lists and note that they don’t
behave in the same way as operations between matrices, especially with respect to
multiplication.
 This is an example of list multiplication. We write * to emphasize that it’s a multiplication,
but if we leave an empty space instead, the result would be the same.


 a, b, c, d x, y
 
 

 
 Here we multiply matrices, or if you prefer, a matrix by a vector. To indicate that this is a
matrix multiplication we use Dot
or “.” which is the same). Note that “*” and “.” are
different operations.

      


 a, b, c, d.x, y

 
     ,     

 In the mathematical rule for multiplying matrices a row vector is multiplied by a column
vector.     is less strict and allows the multiplication of two lists that do not meet
the mathematical multiplication rule. It chooses the most appropriate way to perform the
operation.

  x, y, z . a, b, c

 
 #


 x, y, z . a, b, c

 
        #

 In general, the sum of lists of equal dimension is another list of the same dimension whose
elements are the sum of the elements of the initial lists based on their position.


 1, 2, 3 a, b, c  x, y, z

 
     1,     2,   #  3

 If we multiply a list by a constant, each element of the list will be multiplied by it. The same
happens if we add a constant to a list.


 1, 2, 3, 4 k
 2
 

3 4


 1, 2, 3, 4  c
1 2
 

3 4

 It is easy to get the First or the Last element of a list:


 list2 a, b, c;


 Firstlist2, Lastlist2

 
 , 

 Or


 list21 , list2 1 

 
 , 

To extract a list element based on its position  in the list we use Part[ ] or, in
compact form, [[]].
 In this example, we show two equivalent ways of extracting the second element of a list:

  Partlist2, 2, a, b, c2 

 
 , 

In the case of a list of sublists, first we identify the sublist and then the element
within the sublist.
 In the following example, we extract element 3 from sublist 2:


 a1, a2, b1, b2, b3, c12, 3

 
 b3
  !  " # $ #%


 In the example below, we extract elements 2 to 4. Note that the syntax 
;; means to
extract from  to ".

  list3 a11, 42, x2  y2 , Pi, , "house", ;


 list32 ;; 4 , list34 ;; , list3 ;; 3 

 
 42, 2  2 , , , , house, , a11, 42, 2  2 

 Here we extract elements 2 and 3 from sublists 1 and 2.


 a1, a2, a3, b1, b2, b3, c1, c2, c31, 2, 2, 3
a2 a3
 

b2 b3
 The code below, shows how to get the first column and the first row.


 a1, a2, a3, b1, b2, b3, c1, c2, c3All, 1

 
 a1, b1, c1


 a1, a2, a3, b1, b2, b3, c1, c2, c31, All

 
 a1, a2, a3

 In this case, we use ReplacePart to replace a list element, specifically from sublist 2 we
replace element 2 that is b2 with c2.


 ReplaceParta1, a2, a3, b1, b2, b3, 2, 2  c2
a1 a2 a3
 

b1 c2 b3
 Next, we use  to rearrange the elements in a list.


 list a, b, c, d, e, f;


 list1, 4, 2, 5, 3, 6

 
 , , , , , 

Lists can also be used to specify intervals for graphical plots and in other functions that require
ranges as inputs.


 RandomReal3, 3, 2, 3
2.95474 1.10704 2.41065
 

1.09333 2.03312 1.45064

The solutions (  ) are often displayed in list format. If we need a specific part of
the solution, we can use the previous command to extract it.
 To extract the second solution of this second-degree equation:


 sol Solve1.2 x ^ 2  0.27 x 0.3 0, x

 
   0.625 ,   0.4

 We can proceed as follows:


 x
. sol2

 
 0.4

Other functions very useful for list manipulation are Flatten, Take, and Drop.
      

 With Flatten[ ] we flatten out nested lists. With Flatten[ ] we flatten the list to level .


 Flattena, b, c, d, e

 
 , , , ,

 With Take[  ] we extract the first  elements of the list, and with  the last . With
Take[ {"}] we can extract elements  to ". In the example below, we flatten out the list
before taking the first 5 elements:


 TakeFlattena, b, c, d, e, f, g, 5

 
 , , , ,

 With Drop[ ] we remove the first  elements of the list, and with  the last . Drop[ 
%"&] removes elements  to ".

  Dropa, b, c, d, e, f, g, 3

 
 , , , 

 There are commands particularly useful when operating on lists. It’s easy to identify them
since they generally contain List as part of their names. You can check using the help
system:
? List

We omit the output due to space constraints.

There are many more functions used to manipulate lists. We recommend you
consult the documentation:guide/ListManipulation.

2.3 Association and Dataset

Association and Dataset can be considered extensions to the concept of lists. With
these new additions, Mathematica has a very powerful way to deal with structured
data.
 Association ( <|...|>) is a dictionary data structure containing key-value pairs. An
association in     represents labeled data as key  value pairs:


 a  1, b  2, c  3, d  4

 
   1,   2,   3,   4

 The value is directly associated with the key:

  a  1, b  2, c  3, d  4 c

 
 3

 This data structure can be broken down into individual lists containing just its Keys or
Values. Keys must be unique.


 assoc a  1, 2, 3  100, 4  200, "sphere" 

 
   1, 2, 3  100, 4  200, sphere  

  !  " # $ #% 


 Keysassoc, Valuesassoc

 2, 3 4 sphere
 

1 100 200

 To extract the value of an element, we can either use a single square bracket with the
element’s key ([key]) or Part with the position of the element in the list ([[position]]) :


 assoc"sphere", assoc4 

 
 ,

An alternative and more general method to extract values from association objects is to use
Lookup:


 Lookupassoc, a, "sphere"

 
 1,

 With Normal, the association is converted to a normal expression (in this case a list).


 Normalassoc

 
   1, 2, 3  100, 4  200, sphere 

 AssociationMap and AssociationThread can construct associations from lists:


 AssociationMapf, a, b, c

 
     ,     ,     


 AssociationThreada, b, c, 1, 2, 3

 
   1,   2,   3


 AssociationThread"a", "b", "c"  1, 2, 3

 
 a  1, b  2, c  3

 Common functions for list operations work directly on the keys in an association if “Key” is
added to the name:


 assoc 3  c, 4  d, 1  a, 2  b

 
 3  , 4  , 1  , 2  


 KeySortassoc

 
 1  , 2  , 3  , 4   


 KeySelectassoc, OddQ, KeyTakeassoc, 2, 3, KeyDropassoc, 1, 2 

   3  , 1  , 2  , 3  , 3  , 4   

A dataset is formed from a list of associations:

 The example below shows a simple dataset with 2 rows and 3 columns. It consists of an
association of associations. Most datasets are displayed in tabular form:
      


 data Dataset "1"  "A"  3, "B"  4, "C"  2 ,

"2"  "A"  4, "B"  1, "C"  3 

A B C

 

1 3 4 2
2 4 1 3

 You can extract parts from datasets in a similar way to the one used with lists. To extract the
element from “row 2” and “column C”:


 data"2", "C"

 
 3

 Get the elements from an entire row and from all the rows for a particular column.


 data"1", dataAll, "B"

A 3
1 4
B 4 ,
2 1
 


C 2

Wolfram documentation gives examples showing how to use a Dataset. If you want
to dig deeper into Mathematica’s capabilities for handling structured data, there is a
vast collection of computable datasets in the Wolfram Data Repository
(https://datarepository.wolframcloud.com/ ) that we will use later.

2.4 Matrix Operations

Matrix operations appear very often when performing algebraic calculations. The
use of SparseArray speeds up computations significantly.
 Here we generate 5 sublists with 5 random integers, ranging from 0 to 10, in each one.
(Note: SeedRandom is used to make the example replicable.)


 SeedRandom3142;


 mat RandomInteger10, 5, 5

6 1 3 7 9
3 3 9 4 9
 
 8 9 4 0 6
10 7 10 0 0
3 1 1 9 10
 The next function displays the previous output in matrix format and evaluates its transpose,
inverse, determinant, and eigenvalues. We use TabView to place the outputs under different
labels visible after clicking on the corresponding tab.
  !  " # $ #% 


 TabView

"Matrix A"  MatrixFormmat, "Transpose"  MatrixFormTransposemat,
"Inverse"  MatrixFormInversemat, "Determinant"  Detmat,
"Eigenvalues"  NEigenvaluesmat

Matrix A Transpose Inverse Determinant Eigenvalues

6 1 3 7 9
3 3 9 4 9
 
 8 9 4 0 6
10 7 10 0 0
3 1 1 9 10

Next, we create an equation of the form   = , where  corresponds to the

matrix mat. We will solve it and check that the solution is correct.
 We generate one dimension matrix  randomly (a VectorQ in Mathematica notation).


 b RandomInteger10, 5

 
 5, 1, 6, 9, 1

 We solve the equation using LinearSolve.


 xvec LinearSolvemat, b
20 191 3121 403 3719 337
 
 , , , ,
16 684 8342 8342 16 684 16 684
 We verify that the solution is correct ("." is used for matrix multiplication).


 mat.xvec

 
 5, 1, 6, 9, 1

Matrix operations very frequently have matrices where only few elements have non-
zero values. These kinds of matrices are called sparse arrays. In these cases, it is
recommended to use SparseArray.
 We build a tridiagonal matrix 55 whose main diagonal elements are 9s and its adjacent 1s.


 s SparseArrayi, i  9, i, j
; Absi j 1  1, 5, 5

Specified elements: 13
 
 SparseArray 
Dimensions: 5, 5

 We show it numerically and represent it graphically using MatrixPlot.


 MatrixFormNormals
 
   
9 1 0 0 0
1 9 1 0 0
0 1 9 1 0
0 0 1 9 1
0 0 0 1 9

     


 MatrixPlots

1 2 3 4 5

1 1

2 2

 
 3 3

4 4

5 5

1 2 3 4 5

We build a linear system of 5 equations of the form   = , with and  the
matrices previously defined.
 We solve the equation and check the calculation time:


 TimingLinearSolves, b
869 7 22 487 11
 
 0., , , , ,
1560 520 39 520 1560
 We repeat the same process but for a system of 50,000 variables and 50,000 equations, of
the form   =, where and  are called sLarge and bLarge, respectively. We use “;”
to avoid displaying the output since it would take too much space.


 TimingbLarge RandomReal1, 50 000;

 
 0., Null


 TimingsLarge SparseArray

i, i  9, i, j
; Absi j 1  1, 50 000, 50 000;
 
 0.59375, Null


 MatrixPlotsLarge, ColorFunction  "Rainbow"

1 10 000 20 000 30 000 40 000 50 000
1 1

10 000 10 000

20 000 20 000
 


30 000 30 000

40 000 40 000

50 000 50 000
1 10 000 20 000 30 000 40 000 50 000

 We solve the system   =  and see that the calculation time is extremely short.


 Timingxvec LinearSolvesLarge, bLarge;

 
 0.03125, Null

 We don’t show the result, but we can check that the answer is correct by verifying that
    0. With Chop we eliminate the insignificant terms, unavoidable in numerical
calculations involving decimal numbers.
  !  " # $ #% 


 NormsLarge.xvec bLarge

Chop

 
 0

2.5 Set, SetDelayed, and Dynamic Variables


 ClearAll"Global` "

A frequently asked question is how to differentiate between "=" (Set) and ":="
(SetDelayed).


 x RandomReal
y : RandomReal
 
 0.921297


 x x

 
 0.


 y y

 
 0.129974

 Find out about a symbol’s definitions with Information (?):


 ?x

Symbol

Global`x

Assignment
 

 0.921297

Full Name Global`x


 ?y

Symbol

Global`y

Assignment
 

 : RandomReal

Full Name Global`y

 If we want to remove the definition assigned to x from the global context we can use Clear.


 Clearx

 Now x and  doesn’t have any assignments:


 x

 
 
      

What has happened? When "=" is used in f exp, the result of evaluating expis
assigned tof immediately. From then on, whenever f is called, the program will
always replace f with the value of exp. In the previous example, the value of x was
stored right away, while in the case of y, what gets stored is the function that will
be executed every time it’s called. Because of that, when calculating y y,
RandomRealis executed twice, returning a different result each time.
 The following expression generates a list of 5 random integers between 0 and 5. We call it
data1. These numbers have been stored and they will always be used anytime data1 is
called.


 data1 RandomInteger5, 5

 
 3, 1, 1, 0, 3

 Now we type the same above expression but using “:=”. We name this new list data2.
From now on, whenever data2 is called, the expression that generates the 5 random
numbers will be executed anew. What is stored is the function RandomInteger5,5 not
the actual numbers it generates.


 data2 : RandomInteger5, 5

 The difference can be seen by calling and executing the above defined expressions.


 data1

 
 3, 1, 1, 0, 3


 data2

 
 5, 3, 3, 1, 4

 In the following example we combine “:=” with “=”. What happens?


 data3a : data3b RandomInteger5, 5


 data3a, data3b 

0 1 4 0 1
 

0 1 4 0 1


 data3b, data3a
0 1 4 0 1
 

1 2 4 2 1


 data3a, data3b
2 1 3 3 5
 

2 1 3 3 5

Remember: When using ":=" (SetDelayed) in f: exp the expression exp is not
evaluated until f is used in a posterior calculation, and it will be executed each
time f is called. However, if you use “=”, the function f exp will only be
evaluated the first time it is called; after that, every time f is evaluated it will be
replaced by exp.

The same role that “ =” y “:=” play when defining functions, “” and “: (or :>)”
play when making replacements:
 Immediate replacement (Rule).
  !  " # $ #% 


 x, x, x
. x  RandomReal

 
 0.186714, 0.186714, 0.186714

 Delayed replacement (RuleDelayed).


 x, x, x
. x
RandomReal

 
 0.613595, 0.114433, 0.0048334


 Clearx

We can make dynamic assignments in which the symbol (make sure that in the
menu bar         is checked) returns an entry
that changes dynamically as we make new assignments to it.
 Let’s create a variable.


 Dynamicdvar

 
 dvar

 If we now assign to dvar different values or expressions, you’ll see that the output above
keeps on changing.


 dvar 25
 
 5


 dvar Integratex ^ 2, x

3
 

3
 We can also create a box with InputField and write any  in it.


 InputFieldDynamicdvar 

 
 dvar

 Try to use any fx in the box above, for example Sinx, and you will see how the next
graph is updated. The process will repeat itself anytime you write a new function in the box.


 DynamicPlotdvar, x, 0, 5, Filling  Axis


 ClearAll"Global` "

2.6 Functional vs. Procedural Programming

The power of the Wolfram Language can be seen specially when we use it to create
programs. Readers with experience in procedural programming using languages
such as C or FORTRAN, sometimes use the same approach with    even
      

though it may not be the most appropriate one. Because of this, if you are a
programmer we would like to encourage you not to replicate the procedural
paradigm when programming with    . It will be worth the effort.
 The usual method to define a function with a single argument in the Wolfram Language is:


 fx : x2

 Apply the function:


 f2

 
 4


 f 2

 
 4

 Function   has added new syntax since Mathematica 12.2 .


 f x x2

 
   2

Let’s see an example for creating a function to calculate the factorial of a number
using different programming styles without using the built-in factorial symbol “!”:
 Using a procedural style, we could do as follows:


 factorial1n :
temp 1; Dotemp temp counter, counter, 1, n; temp


 factorial15

 
 120

 With     we can create a function that virtually transcribes the definition of
factorial (note the use of “ ” in the first definition and of “: ” in the second one; the reason
will be explained later):


 factorial20 1;
factorial2n : n factorial2n 1

  factorial25

 
 120
Notice that we combined “=” and “=:” to calculate the factorial. The use of “=” takes
precedence over “:=”.
It’s normal to use “:=”when defining functions. However, in many cases it is also
possible to use " =" or ":=" without affecting the final result.
 Trace shows each step of the evaluation:


 Tracefactorial22

 
 factorial2 2 , 2 factorial2 2  1 ,
2  1, 1 , factorial2 1 , 1 factorial2 1  1 , 1  1, 0 , factorial2 0 , 1 , 1  1, 1 , 2  1, 2

We can use this example to show the difference between Unset (.) and Clear:


 Definitionfactorial2

 
 factorial2 0 1

factorial2 n :  factorial2   1
  !  " # $ #% 

 Unsetremoves a particular definition of factorial2:


 factorial20 .


 Definitionfactorial2

 
 factorial2 n :  factorial2   1

 Clear removes all definitions of factorial2:

  Clearfactorial2


 Definitionfactorial2

 
 Null

The next example consists of building, using different programming styles, a

command to add the elements of a list of consecutive integer numbers and compare
execution times (we use the function AbsoluteTiming to show the absolute number
of seconds in real time that have elapsed, together with the result obtained). We
avoid directly applying the formula for calculating the sum of an arithmetic series,
which would be the easiest and most effective way.
 We start with a list consisting of the first one million integers.


 list Range106 ;

 A procedural method (traditional) to add the elements of this list is:


 AbsoluteTimingsum 0; Dosum sum  listi , i, Lengthlist; sum

 
 0.87435, 500 000 500 000

The functional method (using different instructions in each case) is much more
efficient.
 We perform the same operation in three different ways and display the execution time. In
the first case we use Total, a built-in function. If available, these functions are usually the
fastest way to get things done in    . Later on in the chapter, we will explain
Apply (or @@) in more detail.


 AbsoluteTimingTotallist

 
 0.000703, 500 000 500 000


 AbsoluteTimingApplyPlus, list

 
 0.205671, 500 000 500 000

In this example, we build a function to calculate the sum of the square roots of 
consecutive integers :  1 + ...+  }, and apply it with  50.
 In the functional style, we transcribe almost literally the traditional notation in    
This approach is not only simpler but also more effective.
n


 rootsumn :  i
i 1


 rootsum50

N

 
 239.036
Why do we use “//N”? To find out, see what happens after removing it from the previous
function definition.
Another way of programming where     displays its power and flexibility
      

is rule-based programming.
 For example: To convert {{1 , 1 , 1 },..., { ,  ,  },...} into {1  1  1 , ....,
{     }, ...} we just create the corresponding replacement rule.


 a1, b1, c1, a2, b2, c2, a3, b3, c3, a4, b4, c4
.
a, b, c  a
b  c
a1 a2 a3 a4
 
 , , ,
b1  c1 b2  c2 b3  c3 b4  c4
This rule works fine except when the list and its sublists are both of the same size. For
example, the list consists of three sublists and each of the sublists contains three elements:
{{a1, b1, c1},{a2, b2, c2},{a3, b3, c3}.


 a1, b1, c1, a2, b2, c2, a3, b3, c3
. a, b, c  a
b  c
a1 b1 c1
 
 , ,
a2  a3 b2  b3 c2  c3
There are different ways to avoid this problem. One is to use Replace at level 1.


 Replacea1, b1, c1, a2, b2, c2, a3, b3, c3,
a, b, c  a
b  c, 1
a1 a2 a3
 
 , ,
b1  c1 b2  c2 b3  c3
The effective use of pattern matching is very important for good programming. A pattern object
may consist of blanks: a single Blank (_) with a single underscore character, a BlankSequence
(__) with two underscores or a BlankNullSequence (___) with three underscores.


 Clearf, g


 1, f, fa, fa, b, gc1, c2
. f  "replaced"

 
 1,  , replaced,  ,  ,  c1, c2

  1, f, fa, fa, b, gc1, c2
. f  "replaced"

 
 1,  , replaced, replaced,  c1, c2


 1, f, fa, fa, b, gc1, c2
. f  "replaced"

 
 1, replaced, replaced, replaced,  c1, c2


 ClearAll"Global` "

2.7 Apply, Map, and Other Related Functions

In this section, we have selected some of the most commonly used functions for
programming.
 Map[   ] or f/@ applies f to each of the elements of  . This is probably the
most frequently used function in programming.


 Mapg, a, b, c

 
   ,   ,  


 g
a, b, c

 
   ,   ,  
  !  " # $ #% 


 MapSin, funcx, y, 0, 
2, 90 

 
 func sin  , sin  , 0, 1, 1

Many build-up functions are applied to different levels. Here Mapis used to show
how it works at different levels.
 We have the below list:


 data a, b, 1, 2, 3, 4, 5;
TreeFormdata
    

List

List List

 List List

1 2 3 4 5

 By default, Map works at level 1:


 Mapf, data

 
  ,  ,  1, 2, 3 , 4, 5

 Applying Map at other levels:


 Mapf, data, 2 applied at level 2 
   
 

 1, 2, 3  4, 5


 Mapf, data, 0 applied at level 0 
 
 
 
1, 2, 3 4, 5

 Apply[ ] or f@@list applies f to list.


 Applyf, a, b, c

 
  , , 


 f a, b, c

 
  , , 


 Applyf, a, b, c

 
  ,  , 

 We use FullForm to see how Apply works:


 FullFormf x, y, z
 
  
 , , #

 Here Apply is used to multiply the elements of a list:

      


 Times a, b, c, d

 
 

 Have you realized that the previous pattern can be used to compute the factorial of a
number? For example, 5!:


 ApplyTimes, Range5

 
 120

 Apply (@@@) at level 1:


 Applyf, a, b, c, d, e, 1

 
  , ,  ,  ,

  f a, b, c, d, e

 
  , ,  ,  ,

An alternative way of defining functions is to use “ ” (available since Mathematica 12.2).

Note that parentheses are required when entering the function definition.


 x, y x^2  y  1, 2, 3, 4, a, b
2
 
 3, 13,   

 Using @ in succession enables the creation of composite functions as shown in this example:


 f g h x

 
  


 fx : d x ^ 2 ; gy : c Expy; hz : a Cosb z;


 f g h x

 
 2  2  cos 


 Clearf, g, h

 Notice that f with Map is applied to each element of the expression while with Apply
actually changes the entire expression by replacing its head.


 Mapf, a, b, c, Applyf, a, b, c

 
   ,   ,   ,  , , 

 MapAll or f//@ applies f to each of the expressions.


 MapAllf, a, b, c, d

 
     ,   ,    ,    


 f

a, b, c, d

 
     ,   ,    ,    

Some functions, when applied to lists, operate on the individual list elements, e.g.
f[{, , ...]  {f[], f[], ...}. In this case, the function is said to be Listable.
 In the case of listable functions, Map is implicitly included in them as this example shows.


 Sina, b, c

 
 sin  , sin  , sin 

  !  " # $ #% 


 Sin a, b, c

 
 sin  , sin  , sin 

 Using Attributes, we can check that Sin is listable.

  AttributesSin

 
 Listable, NumericFunction, Protected

 See also: MapAt, MapThread, Outer, Inner, Through, Convolve, and ListConvolve.


 MapAtf, a, b, c, d, 2

 
 ,   , , 


 MapThreadf, a, b, c, x, y, z

 
  ,  ,  ,  ,  , #


 Outerf, a, b, x, y, z

 ,   ,   , #
 

 ,   ,   , #

  Innerf, a, b, c, d, x, y, g

     ,  ,  ,  ,   ,  ,  , 


 InnerPower, a, b, c, x, y, z, Times

 
   #


 Throughf, g, hx

 
   ,   ,  


 ListConvolvex, y, a, b, c

 
     ,     

 In the next example, we apply MapIndexed together with Labeled to identify the position of
each term.

  MapIndexedLabeled, DSinx ^ 4, x, 4

 
 40 sin4   24 cos4   192 cos2  sin2 

3 1 2

 A related function is Thread.


 Thread a, b, c  x, y, z

 
   ,   ,   #

2.8 Iterative Functions

The most commonly used iterative functions in     are Nest, FixedPoint,
and Fold along with their extensions NestList, NestWhile, FixedPointList, and
FoldList.
People with experience in other programming languages tend to make iterations
using For and Do (also available in    ), but normally we should avoid
them and use the commands in the previous paragraph. In the examples that follow
and throughout the rest of the book we will do just that.

     

 NestList is specially useful for iterative calculations.


 fx : 1  x^2


 NestListf, x, 4

2 2 2 2 2 2
 
 , 2  1, 2  1  1, 2  1  1  1, 2  1  1  1  1

 In this example we use NestList to build an iterative formula for computing the future value
of an initial investment of  at an annual interest rate  for " years and show its growth year
by year. We assume an initial capital of 100 and an annual interest rate of 3% for 10 years.


 interestn : 1  0.03 n


 NestListinterest, 100, 10

 
 100, 103., 106.09, 109.273, 112.551, 115.927, 119.405, 122.987, 126.677, 130.477, 134.392

Anytime we want to do something in    , it’s possible that the program

already has a built-in function to perform the operation.
 In this example, we sum consecutively all the elements in a list.


 Accumulatea, b, c, d, e

 
 ,   ,     ,       ,        

As mentioned previously, we will not always find the specific function for what we
want to do. However, there are versatile functions useful to know that we can use to
perform a specific computation. When operating with lists, FoldList is such a
function.
 Here we use FoldList as an alternative to Accumulate.


 RestFoldListPlus, 0, a, b, c, d, e

 
 ,   ,     ,       ,        


 ClearAll"Global` "

2.9 Pure Functions

          

. They are functions without explicit names. Pure functions behave
like operators telling arguments what to do. Initially they may appear cryptic but
their use will end up showing us how powerful they are. Therefore, it’s convenient
to start using them as soon as the opportunity arises.
In the Wolfram Language, you can have functions without explicit names, also
known as “anonymous” functions.
 After removing all variables from memory, we are going to demonstrate several ways of
defining the same pure function:


 Functionu, 3  ux

 
 3


 Function3  x

 
 3
  !  " # $ #% 


 3   &x

 
 3

In this last case, instead of Function we have used the # symbol and to finish
defining the function we have added &at the end. Since the combination of # and
& represent dummy variables we don’t need to store them in memory, reducing the
computation time. Next we are going to see several examples using this notation.
 This function squares its argument and adds 3 to it.

  2  3 &x

 
 2  3

 We define an operator, name it f, and see its behavior.


 f 3   &

 
 1  3 &


 fa, fb

 
   3,   3

 The same result can be obtained with Map ( /@ ):


 f
a, b

 
   3,   3


 ^2  1 & 
funcx, y, 1, 4

 
 func2  1, 2  1, 2, 17

 Here is combined HoldForm and a pure function. (Remember that HoldForm holds an
expression unevaluated):
10

   12 &i
i 1

 
 385
10


  HoldForm12  &i
i 1

   12  22  32  42  52  62  72  82  92  102

 Other examples of usage of pure functions:


 Nest 1   ^ 2 &, 1, 3

 
 676


 Select1, 1, 2, 2, 3,   0 &

 
 1, 2, 3
      


 Plot3Dx ^ 2  y ^ 2, x, 1, 1, y, 1, 1, ColorFunction  Hue3 &

 


 Examples of a pure function with multiples arguments (note the use of

):


 1 ^ 2  2 ^ 4 &x, y

 
   4
2


 Plot3Dx ^ 2  y ^ 2, x, 1, 1, y, 1, 1,

ColorFunction  RGBColor1, 2, 3 & 3& 

 


 #n, ## represent, respectively, the nth and all the arguments supplied to the pure function:


 Clearf


 f &a, b, c

 
  


 f2 &a, b, c

 
  


 f2 &a, b, c

 
  , 


 f &a, b, c

 
  , , 

 There are many ways to define a function in     as shown below for the function
  2  sin   3.


 f1x : x ^ 2  Sinx 3; f11

   sin 1  2


 f2 Functionx, x ^ 2  Sinx 3; f21

 
 sin 1  2


 f3 Function ^ 2  Sin 3; f31

 
 sin 1  2
  !  " # $ #% 


 f4  ^ 2  Sin 3 &; f41

 
 sin 1  2


  ^ 2  Sin 3 &1

 
 sin 1  2

 The combination of a pure function and Apply at level 1 (the short form for Apply[
 ,{1}] is ) gives a powerful method to operate with a list.


 f 1  k 22  &;


 f a1, b1, a2, b2, a3, b3

 
 a1  b12 , a2  b22 , a3  b32  

 In this example, we want to solve the equation below step-by-step expressing the result as a
function of x.


 yx 2;

 Let’s see the internal form of the previous expression.


 FullFormy  x 2
 
  
EqualPlus, , 2

 We want to solve for  by moving  to the right side. That is equivalent to subtracting  from
both sides of the equal sign (note that is necessary to use parenthesis to apply the pure
function to the entire equation).


  x &
y  x 2

 
 2

 Another way of doing it is by using the built-in function AddSides:


 AddSidesy  x 2, x

 
 2

 Next, we generate a list of random values first.


 data RandomReal10, 20;

 Then, we draw three lines, red, green, and blue, each one representing the median, the mean,
and the geometric mean of the data, respectively. We use MapThread to define the
beginning and the end of each line.


 lines
MapThread1, Line0, 2, 20, 2 &, Red, Green, Blue,
Mediandata, Meandata, GeometricMeandata;
 Finally, we combine the data with the generated lines.
      


 ListPlotdata, Epilog  lines

10

6
 

4

5 10 15 20

 Here we calculate ni1

i for  = 50,000 using two methods: a combination of existing
    commands and a pure function. We compare calculation times and see that in
this case they are similar.
50 000


 TimingTotal  i 

N
i 1

 
 0.34375, 7.45367  106 


 rootsum2n : Total  &
Rangen

N


 Timingrootsum250 000

 
 0.390625, 7.45367  106 

 In the next example, we use NestList to compute a continuous fraction.


 NestList1  1
1   &, a, 4
1 1 1 1
 
 ,  1,  1,  1, 1
1 1 1 1
2 1
2 1
2
1 2 2
1 1
2
 1

 The previous method, with  1, can be used to calculate 2 . With Nest only the last term
is shown.


 Nest1  1
1   &, 1, 10

N

 
 1.41421

 FixedPointList is similar to NestList except that it will execute until the solution converges.


 FixedPointList1  1
1   &, 0.1

 
 0.1, 1.90909, 1.34375, 1.42667, 1.41209, 1.41458, 1.41415,

1.41422, 1.41421, 1.41421, 1.41421, 1.41421, 1.41421, 1.41421, 1.41421,
1.41421, 1.41421, 1.41421, 1.41421, 1.41421, 1.41421, 1.41421, 1.41421
 In this example, we use FixedPoint to implement the famous Newton’s method, where x0 is
the initial point.
http://mathworld.wolfram.com/NewtonsMethod.html


 newtonf, x0 :
FixedPoint f
f ' &, Nx0, 10
 We apply the previous function to solve Log  5   0 (remember that Log[x]
represents the natural logarithm of x).
  !  " # $ #% 


 newton Log 5
 &, 0.2

 
 3.76868

In the examples below, we use different methods to simulate a random walk, also
known as the drunk man’s walk. Let’s suppose that a drunk man tries to cross the
street and each step he takes can randomly go in any direction. Many physical
phenomena, such as Brownian motion, are variations of random walks (The
    built-in function is RandomWalkProcess.)
 The function below is a very simple case of a random walk along a line. Each step has a
maximum length of 1 and can go toward both ends of the line with equal probability. To
simulate it, we generate random numbers between 1 and 1 (if it goes in one direction we
use + and in the opposite direction .) What will be the distance covered after  steps?


 RandomWalksteps : AccumulateRandomReal 1, 1, steps

 We create a function to graphically represent the distance covered after  steps. In this case,
we assume  500.


 ListPlotRandomWalk500, Filling  Axis

 

2

100 200 300 400 500

2

4

 Random walks can also be created with a combination of NestList and RandomChoice. The
individual steps can be forward or backward with equal probability. We randomly generate
1s and -1s with 1 meaning a step forward and -1 indicating a step backward. Starting at an
arbitrary point, we want to know the position after  steps. In this case, we have chosen
ListLinePlot instead of ListPlot.


 ListLinePlotNestList  RandomChoice1, 1 &, 0, 500
5

100 200 300 400 500

5

 
 10

15

20

25
      

2.10 Global and Local Variables

A common problem in programming languages is how to avoid conflicts among
variables. Be careful and do not confuse the mathematical concept of variable with
its concept in programming, the one we use here. In this case, when we mention
variables we refer to assignments or definitions made for later use. If we create a
variable and later on use the same name to define a new one, conflicts may arise
between them.
Don’t forget that unless you exit the session, even though you create a new
notebook the functions previously defined during the session will still be active.
 If you have executed all the cells until here, you’ll have the following list of used names:


 Names"Global` "

 
 a, a1, a11, a2, a3, a4, assoc, b, b1, b2, b3, b4, bLarge, c, c1, c2, c3, c4, counter, d,
data, data1, data2, data3a, data3b, dvar, e, expr, f , f1, f2, f3, f4, factorial1, factorial2,
func, g, gr, h, i, interest, j, k, lines, list, list2, list3, mat, n, newton, RandomWalk,
rootsum, rootsum2, s, sLarge, sol, steps, sum, temp, u, x, x0, xvec, x$, y, z

 Symbols starting with $ that you haven’t defined may appear. They correspond to internal
assignments done by the program. In addition, if you used Clear[“Global`*”] earlier during
the session, you will have deleted the previous assignments but not their names. For
example:  will not have any value assigned to it, but RandomWalk will. You still have not
cleared that assignment. Furthermore, both symbols still exist in the global context.


 ?a

Symbol

Global`a
 


Full Name Global`a

  ? RandomWalk

Symbol

Global`RandomWalk

Definitions
 

RandomWalksteps  : AccumulateRandomReal1, 1 , steps 

Full Name Global`RandomWalk

 As usual, if we want to eliminate all the previously defined assignments we can use:

  ClearAll"Global` "

Remember that there’s another way to limit the use of variables to a specific
context:    !  
   "    

#  .
  !  " # $ #% 

 With ClearAll[“Global`*”] all the assignments made will be deleted but not the variable
names themselves. They can be removed as follows:


 Remove"Global` "


 Names"Global` "

 
 


 ?a

 
 MissingUnknownSymbol, a


 ? RandomWalk

 
 MissingUnknownSymbol, RandomWalk

However, a more logical way to proceed is to avoid global variables and use local
variables instead, that is, variables that only have meaning inside an expression.
 Let’s define the following variable in a global context.


 y 3

 
 3

 Let’s use y within a local context. To create local variables we will use Module or With. For
example, in the next expression, y appears, but in this case it’s a local variable, defined
using With, and its use is restricted to operations inside the defined expression.


 f1x : Withy x  1, 1  y  y ^ 2

 If we apply the previous function when  3, since    1 then  4 (local variable),

that is replaced by 1 +  + ^2.


 f13

 
 21


 f2x : Moduley, y x  1; 1  y  y ^ 2


 f23

 
 21

 Nevertheless, that doesn’t affect the global variable  that will still be  3.


 y

 
 3

The main difference between both functions is that with Module we explicitly
indicate what variables are local while when using With, replacement rules are
assigned to the local symbols, that is: With  0 ,   0 ,
,  . In
many cases, you can use them interchangeably.
Another related function is Block, very similar to Module. The main difference is
that while the latter creates temporary assignments for the specified variables
during the evaluation, the former suspends any existing assignments for those
variables and restores the assignments once the evaluation is done.


 Moduley, Expand 1  y ^ 2

 
 y$1786902  2 y$178690  1
      


 Blocky, Expand 1  y ^ 2

 
 16

The behaviour of both is similar, but in some occasions With is faster than Module:


 TimingDoModulex 5, x;, 10 ^ 5

   0.140625, Null


 TimingDoWithx 5, x;, 10 ^ 5

   0.046875, Null

You can find an interesting comparison between them using Inactive on:
http://www.wolfram.com/mathematica/new-in-10/inactive-objects/optimize-
code.html
 Catch returns the argument of Throw:


 Catcha 1; b 2; c 3; Throwc; d 4; e 5

 
 3


 a, b, c, d, e the evaluation stopped after Throwc 

 
 1, 2, 3, ,

 Calculate the smallest prime number greater than 200:


 CatchFori 1, i  100, i , IfPrimei  200, ThrowPrimei

 
 211

2.11 Conditional Expressions and Conditions


 ClearAll"Global` "

A predicate is a function that can take two values: {True, False}. This type of
function is called Boolean. Figure 2.1 shows all the built-in symbols in
    used to represent Boolean functions and their corresponding Wolfram
Language function.
The Wolfram Language provides various ways to set up    , which
specify that particular expressions should be evaluated only if certain conditions
hold.
  !  " # $ #% 

Figure 2.1        !

 Predicates are used when defining conditions. These can appear under very different
situations. They can be used to logically combine multiple predicates:


 22 4

 
 True


 32 9 && 2  Pi

 
 True


 x 3;
IntegerQx && EvenQx
 
 False

 To check whether two expressions are equivalent “===” can be used. Note that 3 y 3.0 in
    are different kind of numbers.


 1 3
3

 
 True


 1 3.0
3

 
 False

If and Which use predicates for tests. A predicate can return either True or False:
 The conditional statement     can be realized in the Wolfram Language by If:


 x 5;
IfPrimeQx, "Prime", "Composite", Whichx 1, a, x 2, b, x 5, c
 
 Prime, c

Which and Switch are useful when dealing with more than two branches.
 Which uses predicates, while Switch uses pattern matching:

     


 x 10;

Whichx  0, Print"Positive",
x 0, Print"Zero", x  0, Print"Negative"
Positive


 x 10;

Switchx, t
; t  0, Print"Positive", t
; t 0, Print"Zero",
t
; t  0, Print"Negative", , Print"Not a valid input"
Positive

 Let’s use f to define a function whose behavior will be different depending on whether we
are using 1 or 2 arguments:


 fx : 0.2 x ^ 2


 fx, y : 0.2 x ^ 2 y ^ 3


 f3

 
 1.8


 f3, 5

 
 225.

 A frequently used format when defining conditions is “_h” to indicate that the replacement
will happen for any expression whose head is h as the following examples show:


 Clearf, g


 fx  gx
. xf  1
x

1
 
  10 
10
 Here is an example of finding the square roots of only the even integers in a list:


 4, 4, Pi, 1.41, Sqrt2, 2
3
. x
; x  0 && EvenQx
Sqrtx

4
2
 
 2, 4, , 1.41, 2,
3
 Notice that Sqrt[2] is interpreted as a integer because the condition is applied to the atomic
expression (the integer inside of Sqrt):


 MapHead, 4, 4, Pi, 1.41, Sqrt2, 2
3

 
 Integer, Integer, Symbol, Real, Power, Rational


 MapAtomQ, 4, 4, Pi, 1.41, Sqrt2, 2
3

 
 True, True, True, True, False, True

 Here is an example of finding the square roots of the only the positive integers in a list:


 4, 2, 3.14, 9, 25, 16
. x
; x  0 && IntegerQx
Sqrtx

 
 2, 2, 3.14, 3, 25, 4

In a function, the easiest way to limit a parameter n so that the function will not be
executed unless the parameter meets the condition   is by typing n:
 In this example n is required to be an integer.


 Clearf
  !  " # $ #% 


 fx, nInteger : x^n


 f3.2, 2

 
 0.0976562


 f3.2, 2.1

 
  3.2, 2.1

An alternative way to establish a condition on a parameter n is with n?Test

(guide/TestingExpressions). As a matter of fact, this is best way to do it since it’s
the fastest one.
 In this example, n must be a non-negative integer (  0).


 gx, nInteger ? NonNegative : x^n


 g3.2, 2

 
 10.24


 g3.2, 2

 
  3.2, 2

A condition can also be defined using “
;”, the function f: exp
;cond is

evaluated only if the condition is met. Other equivalent forms are:
f
;cond: exp and f:exp
;cond.
 In the next example the function takes different forms depending on the selected region (the
same can be done with Piecewise, as we will see later on):


 Clearf


 fx : x ^ 2
; x  0


 fx : x
; 2 x  0


 fx : "x must be bigger than or equal to 2"
; x  2

 We apply the function to different values of x.


 f0.2, f3, f 3

 
 0.04, 9, x must be bigger than or equal to 2

 MatchQ can be used to check what classes of expressions match a pattern:


 MatchQ4.18, x
; x  4

 
 True

In the next example, we combine several conditions that must be met to apply the

binomial coefficients formula . There is a specific function for this calculation:
Binomial, but here we want to create it ourselves.
 The function below calculates the binomial coefficients if  is an integer, is an integer 0,
and   (implying that n0).

  binomialnInteger, rInteger ? NonNegative
; n  r :
n
r n r 
 We add a message that will be displayed if the conditions are not met.
      


 binomialn, r :

"Check that n is an integer, r is a non negative integer and n  r"
 We test the function for several cases.


 binomial3, 2

 
 3


 binomial2, 3

 
 Check that n is an integer, r is a nonnegative integer and n  r

 Another way to provide help and error message is the following:


 binomial::usage "This function gives the binomial coefficient n, r"

 
 This function gives the binomial coefficient n, r


 ? binomial

Symbol

 
 This function gives the binomial coefficient n, r

  binomial::badarg

"Check that n is an integer, r is a non negative integer and n  r";


 binomialn, r : Messagebinomial::badarg


 binomial 3, 2

binomial: Check that n is an integer, r is a nonnegative integer and n r

 In the following case, we establish the condition that  has to be either integer, rational, or
complex to return 2 . The function will be returned unevaluated in any other case.


 squareofnumbernInteger nRational nComplex : n^2

  Mapsquareofnumber, 3, 3
2, 1.3, 3  4 I, I, 2 , Pi, a, b, a

9
   9, , squareofnumber 1.3 , 7  24 , 1, squareofnumber 2 ,
4
squareofnumber  , squareofnumber ,  , squareofnumber 

We can also define conditions with the following functions: If, Which, Switch, and
Piecewise.
 In this example, the function is defined using intervals.


 Plot
Piecewisex ^ 3  6, x  2, x, 2  x  0, x ^ 2, x  0, x, 5, 5

20

4 2 2 4
20
 

40

60

80
  !  " # $ #% 

 The use of /; combined with FreeQ is particularly useful. For example, we create a function
that will consider everything that is not  as a constant.


 fc x, x : c fx, x
; FreeQc, x


 f3 x, x, fa x, x, f 1  x x, x

    30, 10 ,   10, 10 ,  110, 10


 Clearf, g

 A double underscore “__” or BlankSequence
indicates a sequence of one or more

expressions as shown in this example:


 fxSymbol, pInteger : ApplyPlus, x ^ p

  fy, 1, 2, 3

 
 3  2  


 fy

 
  

 A triple underscore “___” or BlankNullSequence
enables the use of any number of

arguments. This is especially useful when defining optional arguments as we’ll see later on.


 gxSymbol, pInteger : ApplyPlus, x ^ p

  gy, 1, 2, 3

 
 3  2  


 gy

 
 0

 A double underscore here is used to plot vectors and labels


 Vectorplot1pts :  ? NumberQ,  ? NumberQ,  :
GraphicsArrow0, 0,  &
pts,
InsetToString,   0.1 &
pts, Axes  True


 Vectorplot1 1, 1, 2, 1, 2, 1
1, 1 2, 1
1.0

0.5

 

1.0 0.5 0.5 1.0 1.5 2.0

0.5

2, 1
1.0

 Here it is an elegant example (from '  '      (
 )   by

D.B. Wagner) of sorting the elements of a list using the triple underscore pattern.


 sortera, b, c, d
; b  c : sortera, c, b, d


 sorterx : x

      


 sorter5, 2, 7, 1, 4

 
 1, 2, 4, 5, 7

 Certainly, we could have used Sort instead of sorter.


 Sort5, 2, 7, 1, 4

 
 1, 2, 4, 5, 7

 In this example, we use Cases to select the list elements of the form  .


 Cases1, x, x , x ^ y, x ^ y

 
  10 , 10 

 The pattern “_.” makes it possible to define an argument that can be omitted. By using
x^y. (note the dot after the underscore),  is also selected since the argument of can
be omitted.

  Cases1, x, x , x ^ y, x ^ y.

 
 10, 10 , 10 

We use DeleteCases to eliminate those elements of a list that meet certain criteria.
 We remove the elements that are decimals (we use Real, different from Reals that refers to
the real numbers).

  DeleteCases1, 1, x, 2, 3, y, 3
5, 2.7, Pi, 9, y, Real
3
 
 1, 1, 10, 2, 3, , , , 9, 
5
 We eliminate from the list those sublists containing 0 as their last term.


 DeleteCases
1, 1, 0, 2, 3, 1, y, 3
5, .7,  Pi, 9, z, a, 0
2, 3 1
3
 
 , 0.7
5
, 9 #
 Alternatives (|) allows you to specify a number of different choices so that an expression is
declared a match for the pattern when it matches any one of those choices:


 Casesa, b, 2, c, 4, 2.3, R, "Hello", Integer Symbol

 
 , , 2, , 4, *

 If you want a match where a particular kind of object is repeated a number of times in this
case Repeated (..) can be used


 , a, a, a, b, a, a, a, a
. a ..  x

 
  , 10, ,  , 10, 10

 A list containing at most 5 integers:


 Cases1, 3, 5, 1, 3, 5, 4, 6, r, s, Range9,
Repeated Integer, 5
 
 1, 3, 5 , 1, 3, 5, 4, 6

 Except[] can be used with a pattern object, which represents any expression except one
that matches .
  !  " # $ #% 


 Cases0, 5, "Hello", Plot, x : ExceptString Symbol
x2 

   0, 25

 Including a second argument in Except[] allows you to represent any expression that
matches  but not . Here, all integers except 0 are returned:


 Cases0, 5, "Hello", Plot, 7.5, 8, Except0, Integer

 
 5, 8

The functions to find patterns within strings are very similar to the ones we’ve
seen so far. They usually have “String” at the beginning of them. The main
difference is the use of “~~” to represent a sequence of string expressions.
 StringCases[“ ”,  ] will match substrings in   that match  . The next example
selects all the substrings that start with “ab”:


 StringCases"aby mno abc abd bac xyz", "ab"  

 
 aby, abc, abd

 StringFreeQ[“ ”,  ] returns True if no substrings in   contain  . Otherwise, it

returns False.


 StringFreeQ"aby mno abc abd bac xyz", "ab"  

 
 False

 StringMatchQ[“ ”,  ] checks whether   contains the pattern  . The string
below contains the sequence “ATG” so the function returns True:

  StringMatchQ"AAGCTGCGTCGTAAATGCGAGT",   "ATG"  

 
 True

2.12 Accuracy and Precision

The number of significant digits of a number  used in calculations is called
precision. To know the precision used in a calculation used Precision.
Accuracy (Accuracy) is the number of significant digits to the right of the
decimal point. Precision is a relative error measure, while accuracy measures the
absolute error.
 By default N uses machine precision.


 ClearAll"Global` "


 PrecisionN

 
 MachinePrecision


 PrecisionN, 40

 
 40.

 However, if we don’t use a decimal approximation, the precision of Pi is infinite.


 Precision

 


WorkingPrecision controls the precision level during internal calculations while

PrecisionGoal determines the precision in the final result. Other options related to
      

accuracy and precision are: AccuracyGoal and Tolerance.

Besides knowing how to use the options mentioned above, there’s an important
issue that you should know:     has different rules for working with
accuracy and precision. An approximate quantity is identified by the presence of a
decimal point. Therefore, 0.14 and 14.0 are approximate numbers, and we will refer
to them as decimals. If we rationalize 0.14 we get 14/100, the ratio of two integers.
When doing operations with integers or combinations of them (rationals, integer
roots, etc.),     generates exact results.
 For example:
5 7


 2  3
12
110 893
 

16 384
 is different from:
5.0 7


 2  3
12
 
 6.76837

There are several functions to convert a decimal number into an exact one.
 For example: Round returns the closest integer to  Floor the closest integer to 
that is less than or equal to , and Ceiling the closest integer to  that is greater than or
equal to .


 Round3.3, Floor3.3, Ceiling3.3

 
 3, 3, 4

In the next example, we are going to see the difference between operating with
integers (or expressions composed of integers, such as fractions) and operating with
decimal approximations.
 We create the following function:


 fx : 11 x a

 We iterate it and check that we always get the same result:


 NestListf, a
10, 20
                    
 
 , , , , , , , , , , , , , , , , , , , ,
10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10
 However, when we replace 1/10 with its decimal approximation 0.1, we can clearly see the
errors associated with machine precision.


 NestListf, 0.1 a, 20

 
 0.1 , 0.1 , 0.1 , 0.1 , 0.1 , 0.1 , 0.1 , 0.1 , 0.1 , 0.1 , 0.1 , 0.100002 , 0.100025 ,
0.100279 , 0.103066 , 0.133729 , 0.471014 , 4.18116 , 44.9927 , 493.92 , 5432.12 
 This effect is not associated to the way     operates. It will happen with any
program performing the same operation using machine precision. For example, try to
replicate the previous calculation, replacing  with a positive integer (for example 3), in a
program such as Microsoft Excel, and you will see that the errors multiply.


 fx : 11 x 3
  !  " # $ #% 


 NestListf, 0.3, 20

 
 0.3, 0.3, 0.3, 0.3, 0.3, 0.3, 0.3, 0.3, 0.3, 0.3, 0.3, 0.299995, 0.299949, 0.299443,
0.293868, 0.232543, 0.442028, 7.86231, 89.4854, 987.34, 10 863.7
 Rationalize converts a decimal number to the nearest rational with the smallest denominator.

  Rationalize0.3
3
 

10
 When used in the previous iteration we eliminate the numerical errors since    
uses infinite precision for calculations without decimal numbers.


 NestListf, Rationalize0.3, 20
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
 
 , , , , , , , , , , , , , , , , , , , ,
10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10
Let’s show another more complicated example, where we will see as well the effect
of using decimal approximations instead of exact calculations.
Consider the following system of differential equations:
1 ' 1  2.5 1  2
2 ' 2 1  2
Initial Conditions : 1 0 1 0 0
 We solve it using DSolve. Depending on the machine precision of your computer, you may
find numerical problems when visualizing the solutions. It can usually solved rationalizing
the decimal coefficients, then     computes with integers and rationals using
infinite precision.


 ClearAllx1, x2


 x1t, x2t x1t, x2t
. DSolve

x1 't 1 Rationalize2.5  x1t  x2t,
x2 't 2 x1t x2t, x10 0, x20 0,
x1t, x2t, t

ExpandAll
41 7 41 7 41 7 41 7
41 7   41 7  41 7   41 7 
  5 4 4  5 4 4   14 4 4  14 4 4
 
  4 4   4 4   2 2 4 4  2 4 4 
41 41 41 41

  Plotx1t, x2t, t, 0, 100

4

 
 2

20 40 60 80 100

A situation where it is common to experience difficulties is when performing

matrix operations (matrix inversion, resolution of systems of differential equations
      

with decimal coefficients, etc.).

When problems associated to operations with decimals arise, it is always a good

idea to rationalize the terms containing those decimals; although you may have
to consider the increase in computation time.

2.13 Choosing the Method of Computation

The power of     becomes particularly evident in the precision with
which you can do calculations and the ability to choose the most appropriate
method or algorithm given the required computation.
 NIntegrate is used for numerical integrations and like almost all other calculation-related
functions, unless indicated otherwise, will choose by default a computation method.
However, you can select a different method or, for a given method, choose the most
appropriate parameters. In the example that follows we compare three methods after
showing graphically the method of quadrature. Note the use of Defer.


 fx : Exp x


 ShowPlotfx, x, 3, 3,

DiscretePlotfx, x, 3, 3, ExtentSize  Left ,
3
PlotLabel  StyleDefer Exp x  x, Small
3
3
 exp  
3

20

15
 

10

3 2 1 1 2 3


 NIntegratefx, x, 3, 3, Method  , PrecisionGoal  2 &

Automatic, "RiemannRule", "Type"  "Left", "TrapezoidalRule"
 
 20.0357, 20.2215, 20.036

 Here we use EvaluationMonitor to display the value of x used in each iteration.


 Print"x ", x

x 


 NIntegratefx, x, 1, 1, Method  "TrapezoidalRule",

PrecisionGoal  2, EvaluationMonitor
Print"x ", x
  !  " # $ #% 

x 1.

x 0.75

x 0.5

x 0.25

x 0.

x 0.25

x 0.5

x 0.75

x 1.

 
 2.35045

 The next function is a piecewise function, that is, it takes different values in each of the
intervals.
ClearAll"Global` "
Sin10 x
x0
x
1
0x1


 fun x ;
Sin2000 x x2 1  x  2
Cosx
2x
x

 To represent it graphically     uses different calculation methods depending on

the interval.


 Plotfun, x, 10, 10, ImageSize  300
1.5

1.0

0.5

 

10 5 5 10

0.5

1.0

 We can control the level of precision as follows:

  NIntegrateSin2000 x x ^ 2, x, 1, 2, 10

 
 0.001275015530


 NNIntegrateSin2000 x x ^ 2, x, 1, 2, 10

 
 0.00127502


 NIntegrateSin2000 x x ^ 2, x, 1, 2, WorkingPrecision  10

 
 0.001275015530

     

 Next, we present an example of a differential equation in which the solving method

(tutorial/NDSolveOverview) depends on the value of !. NDSolve chooses the best algorithm
given !. We use Manipulate, a function that we will cover in more detail in the next chapter.


 Manipulate
res NDSolvey 't yt2 yt3 , y0 , y, t, 0, 2
;
p1 Plot y
. Firstrest, t, 0, 2
, Mesh  All,
, 0.01, 0.0001, 0.01

We should not worry most of the time about choosing a specific computation
method.    + default usually chooses the most appropriate one:
https://www.wolfram.com/algorithmbase/
However, we can always check if that is the case for a given problem by comparing
the results of different methods.

2.14 Optimizing the Computation Time

Another interesting aspect of     is its use of previously calculated
results. Before showing it with an example, it would be better to quit the session to
remove any previous information stored in memory.


 Quit

Whenever possible, built-in functions should be used instead of recreating the

function from other constructs. Built-in functions use compiled C code directly,
thus are faster.
 In the next example we show the calculation time of a complicated definite integral for
different integration limits.


 Timing f1x1 : IntegrateSinx ^ 3, x, 0, x1;
Tablef1i, i, 1, 10;
 
 3.48438, Null

 We repeat the same exercise using “=” instead of “:=” .


 Timing f2x1 IntegrateSinx ^ 3, x, 0, x1;
Tablef2i, i, 1, 10;
 
 1.17188, Null
  !  " # $ #% 

The calculation time is faster in this last case. It seems logical, since in the first
example the integral f1 is calculated 10 times, each time applying a different
integration limit, while in the second case f2 provides the solution to the integral as
a function of the integration limit x1, and then replaces the value of x1 directly in
the solution 10 times.
 We repeat the same pattern using again “=” and the calculation is even faster.


 Timing f3x1 IntegrateSinx ^ 3, x, 0, x1;
Tablef3i, i, 1, 10;
 
 0.15625, Null

The reason behind this behavior is that for functions such as: Integrate, Simplify
and others,     keeps a cache of previous results that it will use later if
necessary to make calculations more efficient:
http://library.wolfram.com/infocenter/Conferences/5832/
 If we now delete the cache, the calculation takes longer:


 ClearSystemCache;


 Timing f3x1 IntegrateSinx ^ 3, x, 0, x1;

Tablef3i, i, 1, 10;
 
 0.53125, Null

 The code below shows a traditional definition, each time f5, 5, 5, 5, 5 is computed,
the consumed time is the same:


 ClearAllfOrd
fOrda1, a2, a3, a4, a5 : NIntegrate1  x1  x2  x3  x4  x5 ,
x1, 0, a1, x2, 0, a2, x3, 0, a3, x4, 0, a4, x5, 0, a5


 fOrd5, 5, 5, 5, 5

AbsoluteTiming

 
 0.197992, 910.019

 Alternatively, we can combine =: and = for a faster computation. The function definition
below incorporates memoization (a technique by which previously computed results are
stored and reused to speed up the overall evaluation). Subsequent evaluations are much
faster, since f5, 5, 5, 5, 5 is stored in the definition itself (as a downvalue):


 fMemoa1, a2, a3, a4, a5 : fMemoa1, a2, a3, a4, a5
NIntegrate1  x1  x2  x3  x4  x5 , x1, 0, a1,
x2, 0, a2, x3, 0, a3, x4, 0, a4, x5, 0, a5


 fMemo5, 5, 5, 5, 5

AbsoluteTiming

 
 0.225405, 910.019


 fMemo5, 5, 5, 5, 5

AbsoluteTiming

 
 1.2  106 , 910.019

In many situations we can significantly reduce the computation time by using

Compile. Normally it’s used with functions that only need to evaluate numerical
arguments. This approach enables the program to save time by avoiding checks
related to symbolic operations. The function generates a compiled function that
applies specific algorithms to numerical calculations.
      

 Example: We use Compile to calculate the following function that only accepts real values
as arguments:


 cf Compilex, Real, x ^ 2 1
1  x Sinx

Argument count: 1
 
 CompiledFunction 
Argument types:  Real

 CompiledFunction evaluates the expression using the compiled code:


 cfPi

 
 9.8696

 We can now represent graphically the compiled version of the function:


 Plotcfx, x, 1, 3

"
"
"
 

"
"

  

2.15 Cloud Deployment

The function CloudDeploy deploys objects to the Wolfram Cloud. Let’s take a look
at some examples.
 The function below creates a web page with a 100-point font message:


 CloudDeployStyle"My first cloud program", 100

 
 CloudObject
https:www.wolframcloud.comobj4b9408f2
6e92
4419
b55d
b959e5514c0d

 The first time we execute it,     will ask for our Wolfram ID and password as
shown in Figure 2.2:

Figure 2.2 #$#   % "

 Clicking on the hyperlink just created will automatically take us to the web page shown in
Figure 2.3:
  !  " # $ #% 

Figure 2.3 &'% ( ) "

 We can deploy any    function to the cloud. Figure 2.4 shows the deployment of
a dynamic interface:

  CloudDeployManipulatePlotSinx 1  k x, x, 0, 6, k, 0, 2

 
 CloudObject
https:www.wolframcloud.comobjbaeed005
91cd
4131
86bc
bc06e4e7e75e

Figure 2.4 *  "

For further information:

http://reference.wolfram.com/language/guide/CloudFunctionsAndDeployment.html


 ClearAll"Global` "

2.16 Package Development

   ’s capabilities can be extended substantially with the use of packages.
Packages are conceptually similar to what other programming languages call
subroutines or subprograms. They basically involve the creation of functions, often
complex, that can be called when needed without having to define them again.

2.16.1 The Structure of a Package

A package has the following structure:
BeginPackage ["name of the package`"]
 ::usage  "f brief description" , (here we would include the headings of the functions
that we are going to use with a brief description, that will be seen by the user later on).
Begin["`Private`"]
     , (here we would have the option to place functions that will be used by f but that
will remain private. That is, the user will not be able to access them).
     , (here we would program the package functions accessible to the user).
End
      

EndPackage[]

2.16.2 A Simple Package

We’d like to build the function below and make sure that it’s going to be available
for later use in any notebook.

 , ,   1    ,   1,   , 0 "  " 1

 Write the following in a code cell ($   
)


 BeginPackage"BinomialPackage`"

Binomialdistribution::usage
"Binomialdistributionn, r, p calculates the binomial probability distribu
under the conditions that n must be an integer, n  1, n  r and 0  p 

Begin"`Private`"

 
 BinomialPackage`

 
 BinomialPackage`Private`


 BinomialdistributionnInteger, rInteger?NonNegative,

p?NonNegative
; n r && 0p1 : n
r n r p^r 1 p^

Binomialdistributionn, r, p: "Check that n is an integer,

r is a non negative integer,n  r and 0  p  1"

End

EndPackage

 
 BinomialPackage`Private`

 Next, from the main menu bar select $   % . A new window in the package
environment will appear (Figure 2.5). After that, copy the previous cell to this new
notebook, turn it into an initialization cell (

%     &')and, in
this case, save it as a package ($   ( % ) in the Data
folder located in our working directory. Use as the name the one defined in BeginPackage:
“Binomialpackage.wl” (remember to choose the extension wl or m).
  !  " # $ #% 

Figure 2.5 (+ ,  

 In this environment we can test, modify, and even debug the package. Once we see it is
working correctly we will save it. A package is an ASCII file. It can also be modified using
a text editor such as Notebook (in Windows).
 As the last step, we exit the session (to be precise: we exit the kernel) to start a new one.
This can be done without closing the notebook by choosing in the menu bar  
)*  or typing and evaluating:


 Quit

2.16.3 Installing and Loading Packages

Best practice is to put packages (.m/.wl files and dependencies) in the directory:


 FileNameJoin$InstallationDirectory, "Applications"

 
 C:\Program Files\Wolfram Research\Mathematica\13.0\Applications

Then the package will be available for all users that run Mathematica installed in this directory.
However, in this case we will define a specific directory.
 Let’s make the folder Data our working directory.


 SetDirectoryFileNameJoinNotebookDirectory, "Data"

 
 F:\Mi unidad\Mathematica\MBN13\Data

 To use the package we need to load it first (or, for testing purposes, execute it inside a
notebook). We can use either << or Needs, but in this case we choose the latter since it has
the advantage of only loading the package if it hasn’t been loaded previously.


 Needs"BinomialPackage`"

 One way to know what functions are available inside the package is as follows:
      


 ? BinomialPackage`

Symbol

Binomialdistributionn, r, p calculates the binomial probability distribution under

 


the conditions that n must be an integer, n
1, n r and 0 p 1

 At the time of loading the package, all its functions are executed, in this case just one:
Binomialdistribution.


 Binomialdistribution5, 3, 0.1

 
 0.0081


 Binomialdistribution5, 3, 1.4

 
 Check that n is an integer, r is a nonnegative integer,n  r and 0 " p " 1

2.17 Additional Resources

As in most cases, the best available information comes from either    ’s
own documentation or Wolfram’s website:
!     '   a short introduction to the Wolfram Language for
programmers, is available in the Documentation Center and also on the web:
(http://www.wolfram.com/language/fast-introduction-for-programmers/)
Wolfram’s reference page for the Wolfram Language: http://www.wolfram.com/language/
Functions and references:
http://reference.wolfram.com/mathematica/tutorial/FunctionsAndProgramsOverview.html
Video courses: https://www.wolfram.com/wolfram-u/catalog/language/
There are many books to learn how to program with    and even though
some of them were written more than 20 years ago, their ideas and explanations of
core concepts are still useful. Here are some of them:
'  '      (
 )   by David B. Wagner (available as a
free download [2022-01-05]:
https://www.dropbox.com/s/j2dsyvptnxjd369/Wagner%20All%20Parts-RC.pdf. Even
though it was written for     3, its exposition of functional programming is
excellent.
   ,'   by Roman E. Maeder:
http://library.wolfram.com/infocenter/Conferences/183/

    - . collection written in     5 by Michael Trott is the
“Bible” of Mathematica:
http://www.mathematicaguidebooks.org
http://packagedata.net/ is a global repository of     packages that makes
it easy for users to find additional functionality in a wide variety of scientific and
technical fields.
3
Interactive Applications, Image
Processing, and More

This chapter explores the use of dynamic functions for creating interactive
simulations (demonstrations) using the Manipulate and DynamicModule
functions. It also shows how to manipulate and process 2D and 3D
images to detect edges, remove backgrounds or identify specific image
elements. The creation of graphs and networks is also covered by giving
an overview of the Graph function. The last section puts some of the
previously discussed Mathematica’s capabilities into action by explaining
in detail the calculation of the period of a pendulum from a series of
images. All the functionality discussed in the chapter is illustrated with
relevant examples.

3.1 The Manipulate Function

One of the most interesting functions in the Wolfram Language is Manipulate (
https://reference.wolfram.com/language/tutorial/IntroductionToManipulate.html ).
We can easily build dynamic applications with it. This function has numerous
options, and in this section we show some of them. Normally we will use more than
one option at a time; if you prefer you can try them separately and then combine
them. Remember that if you don’t know the syntax of a function you can always
use the free-form input first and later modify the Mathematica output to suit your
needs.

We type in an input cell: manipulate sin x from 0 to 10. An input and an output are
generated.

       

In the input cell above we click on the floating label “Replace cell with this input” that
appears right below our command. This way we obtain the function below that we can use
as a basic Manipulate template. Notice that a is a parameter that can take values between 0
and 10 with a default value of 1. Try moving the slide bar with the cursor.

  Manipulate
Plot
Sin
a x, x, 0, 10, a, 1, 0, 10

In some of the examples that follow we have added the options ContentSize and
ImageSize to reduce the output size.

The next example shows some of the many options of Manipulate. You can include a drop-
down button, but it will be automatically added when a parameter contains five or more
choices as shown below. In this case, we also have the possibility to choose several filling
alternatives for displaying a graph. We use PlotLabel as well to show the parameter value
when labeling the graph. By adding the option Appearance “Labeled”, we can display to
the right of the sliding bars the actual values of the parameters. If you’d like them to be
shown by default, you should add the option Appearance  “Open”. We can also label the
parameters using the syntax: {{a, initial a value, “a label”}, minimum a value, maximum a
value}. Another interesting option is ControlType. For example, we can use Checkbox to
include check boxes. The syntax is: {a,{a1,a2}}, where {a1,a2} refers to the alternatives
available, a1 is the default option. When there are more than two choices, you can use
RadioButtonBar.


 Manipulate

Plot
Exp
a x Sin
b x, x, 0, 10,
PlotLabel
Style
StringForm
"Exp
a x Sin
b x ", 10,
PlotStyle
Directive
color, AbsoluteThickness
thickness,
Axes
axes, a, 0.1, "Amplitude a", 0, 0.5, Appearance
"Open",
b, 3, "Amplitude b", 1, 10, Appearance
"Labeled",
axes, True, False,
color, Orange, Orange, Red, Yellow, Green, Blue,
thickness, 1, 3, 5, ControlType
RadioButtonBar

          


The function “ManipulatePlot” is available in the repository:

https://resources.wolframcloud.com/FunctionRepository/resources/ManipulatePlot/
We can use it to generate a plot in which both, the plot ranges and function parameters can
be manipulated dynamically.


 ResourceFunction
"ManipulatePlot"

 


 ManipulatePlot




 ManipulatePlot
Sin
a  x  b, Cos
a  x  b,

x, 2  Pi, 2  Pi, a, 1, 0, 2, b, 1, 0, 2

If we use Locator with Manipulate we will be able to move objects in the output
with the mouse. Additionally, all the information associated to those objects will be
dynamically updated. The syntax of Locator is: {{parameter, {x, y}},
Locator}.

In this example, we can move the three vertices of a polygon and see how the triangle area is
filled automatically.


 Manipulate

Graphics
Pink, Polygon
pts, PlotRange
1.1, ImageSize
Tiny,
pts, 0, 0, 1, 0, 0, 1, Locator

       

We are going to extend the functionality of the previous command to calculate and
display the area of the triangle, but first let’s introduce a way to call a function from
within Manipulate.
Any local variables that are used by a function should be passed into that function as an
argument. Because of that, the first example doesn’t work:


 ClearAll
f;
f
x : Sin
a x;
Manipulate
Plot
f
x, x, 0, 10, a, 1, 5


 ClearAll
f;

f
a, x : Sin
a x;
Manipulate
Plot
f
a, x, x, 0, 10, a, 1, 5

Use the option SaveDefinitions or we can use Initialization with the function that we would
like to evaluate first when Manipulate is executed. The SaveDefinitions option determines
whether external definitions required to display the content area are automatically included
in the output:


 ClearAll
f;
f
x : x ^ 2;
Manipulate
f
x, x, 0, 100, SaveDefinitions
True

          



 Manipulate
h
y, y, 0, 100, 1, Initialization  h
x : x ^ 2 

Going back to the example of the triangle, to explicitly include the value of its area we have
to define a function that calculates it from the Cartesian coordinates of its three vertices: v 1 ,
v2 , and v3 .


 areaTriang
v1, v2, v3 :
Abs
Det
Join
Transpose
v1, v2, v3, 1, 1, 1 2


 areaTriang
3, 2, 4, 1, 7, 3
1
 

2

Now we include areaTriang in the previously typed command. Move the vertices with the
mouse to see how the area is dynamically updated.


 Manipulate
Graphics
Pink, Polygon
pts, PlotRange
1,
PlotLabel
StringForm
"Triangle Area `1`", areaTriang
pts,
ImageSize
Tiny,
pts, 0, 0, 1, 0, 0, 1 2, Locator,
Initialization  areaTriang
v1, v2, v3 :
Abs
Det
Join
Transpose
v1, v2, v3, 1, 1, 1 2 

In this example, we use BSplineFunction to generate a B-Spline curve associated with all
the displayed points. If we use LocatorAutoCreate  True, we will be able to add or
remove points using +click (+click in OS X).


 Manipulate

Show
Graphics
Point
p, Green, Line
p, ImageSize
Small,
ParametricPlot
BSplineFunction
p
t, t, 0, 1,
p, pts, Locator, LocatorAutoCreate
True,
Initialization  pts 1, 5, 3, 2, 6, 4 

       

Dynamic and DynamicModule are other functions used for dynamic applications.

Add Dynamic around bits that can be updated independently of the rest:


 Manipulate
Show
ContourPlot
Cos
a x  Cos
a y, x, 0, 1, y, 0, 1,
Graphics
Disk
Dynamic
p, 0.02, a, 10, 0, 20, p, 0, 0, 1, 1

DynamicModule exhibits many of the same scoping behaviors as Module, plus a few extra
features:


 DynamicModule
x 0.5, Slider
Dynamic
x, Dynamic
x

 
  , 0.5

DynamicModule stores all the definitions of its local variables by default.


 DynamicModule
x 3, f, f
t : t ^ 2; f
x

   9

The example that follows shows how to use it to display some of the different control types
available in Mathematica.

          



 DynamicModule
pua, rbba, cbba, tgba, Grid


Style
"Control Type", FontWeight
"Bold",
Style
"Example", FontWeight
"Bold",
"Slider",
Manipulate
x, x, 0, 1, ControlType
Slider, LabelStyle
16,
"2DSlider",
Manipulate
x, x, 0, 1, ControlType
Slider2D, LabelStyle
16,
"PopupMenu", Row
PopupMenu
Dynamic
pua, a, b, c, d,
" ", InputField
Dynamic
pua, FieldSize
5,
"RadioButtonBar", Row
RadioButtonBar
Dynamic
rbba, Range
5,
" ", InputField
Dynamic
rbba, FieldSize
5,
"CheckBoxBar", Row
CheckboxBar
Dynamic
cbba, 1, 2, 3,
" ", InputField
Dynamic
cbba, FieldSize
5,
"TogglerBar", Row
TogglerBar
Dynamic
tgba, Range
5,
" ", InputField
Dynamic
tgba, FieldSize
7,
"InputField", Manipulate
input,
input, 3.14, ControlType
InputField, LabelStyle
16,
"ColorSlider", ColorSlider
Pink
, BaseStyle
"TraditionalForm", FontFamily
"Helvetica",
Alignment
Left, Background
LightYellow,
Frame
All, Spacings
2, 0.62


 Quit


3.2 Creating Demonstrations

The main use of Manipulate is to create eye-catching interactive visualizations with
just a few lines of code. In the Wolfram Demonstrations Project site
(http://demonstrations.wolfram.com/), you can find more than 12,000 of them. We
will be showing here some easy examples.

       

3.2.1 A Random Walk

We use the function RandomWalkProcess to represent a random walk on a line with the
probability of a positive unit step p and the probability of a negative unit step 1
p. By
default we assume n steps = 500 and set the interval from 100 to 1,000 and p
0.5. Each
time we execute the command, even though the number of steps is the same, the distance
covered will be different since we are assuming random walks.


 Manipulate

ListPlot
RandomFunction
RandomWalkProcess
p, 1, n, Filling
Axis,
n, 500, "Steps", 100, 1000 , 1, Appearance
"Labeled",
p, 0.5, "Probability", 0, 1, Appearance
"Labeled"

3.2.2 Zooming and the Rosetta Stone

In this example, we show how we can zoom on an image and scroll through it. In
this case the image is a picture of the Rosetta Stone but any other picture would do
as well.

We can import it directly with:


 ClearAll
"Global`"  Remove all the variables from memory 


 rosettaimage Import

"https: upload.wikimedia.org wikipedia commons c ca RosettaStoneBW.
jpeg";

          


 rosetta Image
rosettaimage, ImageSize
Small

 


Now we can copy it, as shown below, or type rosetta instead.


 Manipulate

ImageResizeImageCrop , 300, 300 zoom, pts,

300, 300, Resampling
ControlActive
"Nearest", "Lanczos",

pts, 0, 0, "pts", 1, 1, 1, 1,
zoom, 1, "Zoom", 1, 10

We can generate a similar effect using DynamicImage .

       


 DynamicImage , ZoomFactor
2, ImageSize
300, 300

3.2.3 Derivatives in Action


 ClearAll
"Global`"

In this example, we use derivatives to show the tangent at the selected point for different
curves. This type of visualization is useful to represent the tangent to a curve and introduce
the concept of the derivative.


 Manipulate
Plot
f
x, f
x1  f '
x1  x  x1 , x, 2 Pi, 2 Pi,
PlotRange
2, 2, Epilog
PointSize
0.025, Point
x1, f
x1,
x1, 2 Pi, 2 Pi,  f, Sin, Cos, Tan, Sec, Csc, Cot

3.3 Image Processing

The image processing functionality of Mathematica has greatly improved in the
latest versions of the program. We are going to give a brief overview of it. You can
extend your knowledge by going through this tutorial: tutorial/ImageProcessing. At
the end of this chapter, there are several links in case you may be interested in
learning more about the topic.
In the first chapter, we saw that when you click on an image, the Image Assistant,
a toolbar providing several options to modify the image will appear right below it.
As a matter of fact, many of the operations that we will show here can be executed
directly using this toolbar. However, for clarity purposes we will show the
Mathematica commands instead of describing how to use the toolbar. Besides, if
we were interested in creating a program we would have to type the required
instructions.

          

3.3.1 Playing with The Giralda, Machu Picchu and More

We import a JPG image (in this example, The Giralda of Seville, although with a search
engine you can find many images similar to the one below or even use other images as
well). The resolution of the image was preserved when downloaded. You can also cut and
paste from a website.


 Clear
"Global`" Clear all symbols in a given context


 giralda ;

We can check the image size in pixels. It consists of a n  m grid of triplets, each one
representing a pixel.


 ImageDimensions
giralda

 
 635, 635

We can see the RGB value of the individual pixels:


 ImageData
giralda

0.607843, 0.733333, 0.886275, 0.607843, 0.733333, 0.886275,

0.607843, 0.733333, 0.886275,  629  ,
0.458824, 0.572549, 0.792157, 0.458824, 0.572549, 0.792157,
 

0.458824, 0.572549, 0.792157,  633  ,
 1  

  
 


 
 

  

We adjust the contrast and the brightness (if you prefer, instead of copying and pasting the
image, just type “giralda”).


 ManipulateImageAdjust , a, b,

a, 1.5, "Contrast", 2, 2, b, 0.5, "Brightness", 0, 1

       

The following command detects the edges of an image.


 edges EdgeDetect
giralda

 


Here we compare both images.

GraphicsRow
giralda, edges

 


ImagePartition splits an image into a collection of subimages of a certain size. In this case,
we apply the function to a picture of Machu Picchu:


 GridImagePartition , 300, Frame
All


           

ImageStich enables us to stitch together a list of images by merging overlapping images into
a single one. This function is one of the latest additions to Mathematica 13. The example
below is included in the function documentation:


 ImageStitch , , 

3.3.2 Background Removal and Image Composition

In this example, we remove the background of an image and create a new one by combining
the result with an image of the Andromeda Galaxy.


 foreground RemoveBackground , "Background", Green;

        


 ImageCompose , foreground

 


This example, available in the documentation, shows how to identify and highlight
sickle cells using a combination of the Binarize, ComponentMeasurements and
HighlightImage functions in Mathematica.


 i ;

First, binarize the image to get the foreground:


 b Binarize
i, 0, .7

 


Then, compute centroids and the radii of disks with the same area as those centroids only for
the largest components of the image:

  cells ComponentMeasurements
b, "Centroid", "EquivalentDiskRadius",
AdjacentBorderCount  0 && 50  Area  1200 &
 
 7  196.016, 146.391, 17.0101,
8  141.599, 146.897, 16.982, 11  88.4614, 123.396, 15.7367,
12  218.991, 111.964, 18.412, 13  51.7714, 105.796, 16.9163,
14  141.656, 108.643, 16.6412, 16  178.551, 88.5622, 16.9257,
17  91.3829, 86.464, 16.8125, 18  223.054, 74.4934, 15.5843,
21  115.626, 54.4298, 14.9057, 22  78.1822, 51.4477, 15.0121,
23  197.218, 45.1391, 17.7697, 25  163.209, 31.0179, 15.4818

Finally, highlight the selected components:

           


 HighlightImage
i, Circle  cellsAll, 2 

 


Can you identify the location of the common points in these two images? (Images
available in the ImageCorrespondingPoints documentation.)


 i1, i2  , ;

The output displays the matching positions between the two images.


 matches ImageCorrespondingPoints  i1, i2;
MapThread

Show
1, Graphics
Yellow, MapIndexed
Inset
21 , 1 & , 2 &,
i1, i2, matches

3
14 1 2 14
20 8
8 37
16 3
42 35 21 42
1830 1
20 5 26 4 2
371621 19 11 24 6
19 518
7 30
 
  35 , 40 25 13 7 11 
22
26 4 39381023 34 24
13 3332 15 27 28 31
3141
6 25 34 9 413332 22
40 10 23
28 17
291236
39 38279
15 17
2912 36

The rotation angle between both images can also be found:


 matches ImageCorrespondingPoints
i1, i2;
a1 ArcTan   1 2  ImageDimensions  i1 &  matches1 ;
a2 ArcTan   1 2  ImageDimensions  i2 &  matches2 ;
RootMeanSquare
Mod
a1  a2, 2 Pi Degree
 
 39.6143

3.4 Image Manipulation

3.4.1 3D Images
There are many functions (guide/3DImages) for 3D image processing, as this
example shows:
        


 ctHead ExampleData
"TestImage3D", "CThead";


 head Image3D
ctHead, BoxRatios
1, 1, 1;


 Manipulate

Image3D
head, ClipRange
x, 256, 0, y, z, 100,
x, 100, 0, 256, 1, y, 150, 0, 256, 1, z, 50, 0, 100, 1

3.4.2 Eyesight Recovery

Next, we show how to improve image resolution.

Here’s an example of an image with fuzzy text (you can find the picture in the
ImageDeconvolve help file):


 fuzzy ;

With a deconvolution we can substantially improve the image quality.


 ImageDeconvolve
fuzzy, BoxMatrix
3 49, Method
"TotalVariation"

 


Here is a simulation of a signal with noise.


           


 data Table
Sin
i ^ 2  i  RandomReal
.3, .3, i, 0, Pi, 0.001;

ListLinePlot
data

1.0

0.5

 

500 1000 1500 2000 2500 3000
0.5

1.0

The next function uses a filter to remove the noise component of the signal.


 Manipulate
With
res LowpassFilter
data, ,
GraphicsColumn
Spectrogram
res, ListLinePlot
res,
 , Pi, .01, Appearance
"Labeled"

3.4.3 Edge Detection

Mathematica has several functions for detecting edges and borders that are very
useful for image processing.

In the next example, we capture the image of the author while writing this book and apply
the FindFaces command to detect his face.


 picture CurrentImage



 picture ;


 faces FindFaces
picture

 
 Rectangle154.5, 38.5, 237.5, 150.5

        


 Show
picture,

Graphics
EdgeForm
Red, Thick, Opacity
0, Rectangle  faces

  

3.4.4 Barcode Recognition

This example shows the use of BarcodeRecognize to identify barcodes.


 boardingPass ;





 BarcodeRecognize
boardingPass, "Data", "PDF417"

 
 M2VANALMSICKMARKUS ECSEZLJ HNDDXBEK 0313 271Y018K0254

13620B1WW3271BEK 251762174509010 1 EK 116054035 CSEZLJ
DXBDUSEK 0055 271Y019K0249 127251762174509011 1 EK 116054035

3.4.5 Morphological Thinning

The function Thinning looks for the skeleton of an image through morphological
analysis.

In this example, the image is a labyrinth. Thinning replaces the background with lines
showing possible paths.


 labyrinth

PruningThinning , Padding
1, Padding
1

  

Now we create the graph of the image:


 MorphologicalGraph
labyrinth

          

3.4.6 A Painting with a Surprise

The next example shows how to visualize an image from a certain perspective and
how to select a part of it. The image was imported from a Wolfram blog:
http://blog.wolfram.com/2010/10/27/the-ambassadors


 ambassadors
Image
Import
"https: content.wolfram.com uploads sites 39 2010 10
TheAmbassadorsBlogPost01.jpg", ImageSize
150

 


The image above corresponds to the painting “The Ambassadors” by Holbein. There’s a
skull in it. Can you see it?

With ImagePerspectiveTransformation, we can apply a transformation to see the image from
a different perspective and, combined with TransformationFunction, display the part of the
image that we want to visualize.


 ImagePerspectiveTransformationambassadors,
0.5 0.87 0.05
TransformationFunction 1.34 2.7 0.43 
0.62 0.2 0.525

  
       

3.4.7 Video
Mathematica has added numerous capabilities for video processing.

Generate a video of a 3D object rotating around the z axis:


 g ExampleData
"Geometry3D", "SpaceShuttle"1, 2 ;


 VideoGenerator
Graphics3D
EdgeForm
,

GeometricTransformation
g, RotationTransform
2 2 , 0, 0, 1,
SphericalRegion
True, Boxed
False &,
2, RasterSize
300, FrameRate
10

A basic introduction to video and audio using the Wolfram Language can be found
at tutorial/VideoBasics and tutorial/AudioBasics:
https://reference.wolfram.com/language/tutorial/VideoBasics.html
https://reference.wolfram.com/language/tutorial/AudioBasics.html

3.5 Graphs and Networks

Graphs and networks is another area where the latest Mathematica versions
incorporate new functionality: guide/GraphsAndNetworks. A graph can be
represented using Graph, a function with many options. Let’s see a basic example.

We begin by defining the connections between vertices: i  j or i j.


 graph1 "a1" "a2", "a2" "a3", "a2" "a1", "a3" "a4";

Next we define the capacity of each edge and vertex. Finally we add labels to identify the
vertices.

          


 Graph
graph1, EdgeCapacity
2, 3, 4, 5,

VertexCapacity
2, 3, 4, 6, VertexLabels
"Name"

  
  

a4 a3 a2 a1

We can use special symbols as labels.


 g Graphgraph1, EdgeCapacity
2, 3, 4, 5, VertexCapacity
2, 3, 4, 6,

VertexLabels
"Name", EdgeLabels
"a1" "a2"
,

"a2" "a3"
, "a2" "a1"
, "a3" "a4"


 


We may have a graph representation in a certain format from which we would like to extract
information. We can do it as follows:


 EdgeRules
g

 
 a1  a2, a2  a3, a2  a1, a3  a4


 PropertyValue
g, EdgeCapacity

 
 2, 3, 4, 5


 PropertyValue
g, VertexCapacity

 
 2, 3, 4, 6


 PropertyValue
g, EdgeLabels

 
 a2  a3  , a3  a4  , a2  a1  , a1  a2  

In the next example, we find the shortest way to link all the locations in Figure 3.1 with
fiber optic cable:
       

Maywood

14 28 10
Downers Oakbrook
27 17
Grove 25 Addison
23 11
20
10 21
Stickney
15
Wheaton
20 McCook
10
28

Bellwood

Figure 3.2    

After creating the network graph with Mathematica (Figure 3.2, code not shown) and
naming it “g”, we find the minimum spanning tree containing all the villages (Figure 3.3):


 fibernetwork FindSpanningTree
g;


 HighlightGraph
g, fibernetwork, GraphHighlightStyle
"Thick"

 


Maywood

14 28 10
Downers Oakbrook
27 17
Grove 25
Addison
23 11
20
10 21
Stickney
15
Wheaton
McCook
20
10
28

Bellwood

Figure 3.3  ! " # 

In this example, we generate random graphs and search for cliques (Figure 3.4):


 Grid  Table
HighlightGraph
g RandomGraph
8, 16,
Subgraph
g, First
FindClique
g, 2, 4

           

Figure 3.4 $%!  &  

The command below returns the edges and vertices of a dodecahedron in graph format:


 Graphics3D

Opacity
.8, PolyhedronData
"Dodecahedron", "Faces", "Polygon",
g PolyhedronData
"Dodecahedron", "SkeletonGraph"

 
  , 

Here we are trying to find the shortest path passing once through each vertex (this is the
famous Hamilton’s Icosian game):


 h PathGraph  First
FindHamiltonianCycle
g;


 HighlightGraph
g, h, GraphHighlightStyle
"Thick", Graphics3D

Opacity
.8, PolyhedronData
"Dodecahedron", "Faces", "Polygon",
Red, Tube
PolyhedronData
"Dodecahedron", "VertexCoordinates"
Append
VertexList
h, VertexList
h1  , 0.1

 
  , 
        

3.6 Application: Finding the Period of a Pendulum


 ClearAll
"Global`"

The example that follows is significantly more difficult than the previous ones. Our
objective is to find out the period of a pendulum by analyzing a sequence of video
images. The images were obtained with CurrentImage (if your computer has a
webcam you can use this function). The function adds a time stamp to the image,
information that we will use later on in the example:

ims  , , , ,

, , , , , ,

, , , , ;

Source: Wolfram seminars for image processing, although currently, the seminar is not
available.

Here is the video sequence:


 ListAnimate
ims

We begin by highlighting the most prominent lines to identify the ones corresponding to the
pendulum.


 edges EdgeDetect  ims;

In the images above, the pendulum lines are the ones close to the vertical. We can identify
them using Radon.


 radons ImageAdjust  Radon
, Automatic, Pi 4, Pi 4 &  edges ;

Binarize detects the brightest points and extracts their x-coordinates to obtain the angle.


 xs ComponentMeasurements
, "Centroid"1, 2, 1 & 
Binarize
, .9 &  radons
 
 89.5, 96., 111.5, 124., 125.5, 122.5, 112.5, 100.5, 87.5, 89., 95.5, 106., 121.5, 127.5, 124.

In the next step, we obtain the relationship between the positions and the time by converting
the x-coordinates to radians and combining them with the time stamps of the images.

           


 thetas xs 215  .5  Pi 4;


 times Mod
Options
, "MetaInformation"1, 2, 1 , 100 &  ims

   36.0593128, 36.3133382, 36.6473716, 36.9003969, 37.1534222,

37.5664635, 37.8204889, 38.1605229, 38.4995568, 38.9105979,
39.1666235, 39.4196488, 39.8376906, 40.1757244, 40.4267495


 datapoints Transpose
times, thetas;
ListPlot
datapoints

 

 

  


 

We can now fit the data to a sinusoid curve using NonlinearModelFit.


 nlm NonlinearModelFit
datapoints,
a Sin
b t  c   d, a, .07, b, 3, c, 2, d, 0, t

 
 FittedModel  
  

  
 

  Show
ListPlot
datapoints,

Plot
Normal
nlm, t, 36, 41, PlotRange
All

 

 

  
 

 


 nlm
"BestFitParameters"

 
 a  0.0746392, b  2.16036, c  10.0063, d  0.00305312

We estimate the period, in seconds:


 period 2 Pi b .

 
 2.90839

Next, we use FormulaData to find the formula to calculate the length of the pendulum:
        

After initially typing “ FormulaData
“Pend ” then you will have the option to choose the
full function:


 FormulaData
"PendulumSmallOscillations", "Standard"

l 1
 
 T 2 ,f , vmax 2 g l 1  cos0  
g T


 pendulumeq FormulaData
"PendulumSmallOscillations", "Standard"1

l
 
 T 2
g


 pendulumeq InputForm

 

  
QuantityVariable"T", "Period" 
2 Pi SqrtQuantityVariable"l", "Length"QuantityVariable"g",
"GravitationalAcceleration"

Finally, we apply the formula to compute the length, in meters, of our pendulum:


 Solvependulumeq . QuantityVariable
"T", "Period"
period,
QuantityVariable
"l", "Length"
l, QuantityVariable

"g", "GravitationalAcceleration"
 ''    , l

 
 l  2.1012 m s2 

3.7 Additional Resources

To access the following resources, if they refer to the documentation files, write
them down in a notebook, select them, and press <F1>. In the case of external
links, copy them in a browser.
About Demonstrations
Manipulate: tutorial/IntroductionToManipulate
Advanced Manipulate functionality tutorial: tutorial/AdvancedManipulateFunctionality
Introduction to Dynamic tutorial: tutorial/IntroductionToDynamic
The Wolfram Demonstrations Project: http://demonstrations.wolfram.com
About Images and Videos
https://www.wolfram.com/wolfram-u/introduction-to-image-processing
Image processing tutorial: tutorial/ImageProcessing
Image Processing & Analysis guide page: guide/ImageProcessing;
Video processing: https://reference.wolfram.com/language/guide/VideoProcessing.html
A collection of video presentations and seminars is available on:
https://www.wolfram.com/wolfram-u/catalog/image-signal-processing
 4
Accessing Scientific and Technical
Information

This chapter introduces different ways to access real-world computable

data with the help of Entities and computable data functions such as
ChemicalData, CountryData, FinancialData, or WeatherData. Using
these Mathematica’s capabilities, the chapter shows how to access
databases containing up-to-date information from many different fields
such as: Chemistry, Climatology, Economics, Geography, History, Life
Sciences and Medicine, and many more. It also provides an overview of
the Wolfram data repository. The examples in the chapter cover topics as
varied as how to visualize the evolution of COVID-19 in selected
countries, explore molecules or generate stream plots using real-time
wind data in Spain.
Later in the book, we will make extensive use of this knowledge when
discussing examples.

4.1 The Wolfram Data Framework: Introducing Entities

The Wolfram Language and the Wolfram Knowledgebase through the Wolfram
Data Framework™ (WDF) provide a standardized representation of real-world
computable data accessible through entities. These entities are grouped into
hundreds of entity types. The corresponding Mathematica function is EntityType,
and it can return values from a wide variety of entity types such as: Chemistry,
City, Country, Financial, Movie, Planet, and many more. In fact, the Wolfram
Knowledgebase is continuously expanding to include new fields. When using them,
Mathematica will connect to a server at Wolfram Research, which in turn will
access a specialized database containing all the relevant data. The execution time
may vary depending on the speed of your Internet connection and that of the
Wolfram Research server itself.
Wolfram Knowledgebase: https://www.wolfram.com/knowledgebase/
118 Mathematica beyond mathematics

WDF: https://www.wolfram.com/data-framework/

Using the Inline free-form input format (Ctrl+= in a US keyboard) we can know the type of
entity of the word written inside the box:
France

In[
]:= France COUNTRY

Out[
]= France
So, France is an “Entity” of the type “Country”.
If you get an error message when evaluating the previous expression it may be due to
problems with your firewall or proxy configuration for accessing the Internet. Check your
Mathematica settings in:
         . If the issue is proxy-related,
access the help system for information about how to troubleshoot it:
tutorial/TroubleshootingInternetConnectivity.

A list of entities within a given type can be found using EntityList:
In[
]:= EntityList
"Country"  Shallow
Out[
]//Shallow=

 Afghanistan , Aland Islands , Albania , Algeria , American Samoa ,

Andorra , Angola , Anguilla , Antarctica , Antigua and Barbuda , 240

We will use Shallow or Short to reduce the size of the output. Remove it to see the
complete output.
The entity type “Country” not only includes all the individual countries of the world. It also
contains classes (EntityClass ), groups of entities with a shared characteristic. For instance,
the entity class “Europe” represents all of the countries in Europe.
In[
]:= EntityClass
"Country", "Europe"

Out[
]= Europe

In[
]:= EntityList
  Shallow

Out[
]//Shallow=

 Albania , Andorra , Austria , Belarus , Belgium ,

Bosnia and Herzegovina , Bulgaria , Croatia , Cyprus , Czech Republic , 41

Each entity has properties associated to it. Here are the ones associated to “Country”:
In[
]:= Entity
"Country"
"Properties"  Shallow
Out[
]//Shallow=

 adjusted net national income , seasonal bank borrowings from Fed, plus adjustments , regions ,

adult population , obese adults , number of aggravated assaults , rate of aggravated assault ,
aggregate home value , aggregate home value, householder 15 to 24 years ,
aggregate home value, householder 25 to 34 years , 741

If we want to know the     expression for a certain property (e.g., the adult
population) we place the mouse over “adult population” in the previous output and we see
that the exact syntax is:
In[
]:= EntityProperty
"Country", "AdultPopulation"

Out[
]= adult population

4. Accessing Scientific and Technical Information 119

The actual value of a property is given by EntityValue:

In[
]:= EntityValue
Entity
"Country", "Japan",
EntityProperty
"Country", "AdultPopulation"
Out[
]= 7.44373  107 people

Since the above expression is very cumbersome to type every time we want to know the value of
a property,     offers us several approaches to simplify our queries:

For example, combining just Entity and EntityProperty, we can obtain the same result:
In[
]:= Entity
"Country", "Japan"
EntityProperty
"Country", "AdultPopulation"

Out[
]= 7.44373  107 people

This expression can be further simplified:
In[
]:= Entity
"Country", "Japan"
"AdultPopulation"

Out[
]= 7.44373  107 people

However, it’s often more convenient to just use the Inline free-form input format (Ctrl+= in
a US keyboard):

In[
]:= Japan COUNTRY
adult population 

Out[
]= 7.44373  107 people

If you click on the symbol to accept the interpretation and then convert the cell to
standard form, the result will be a familiar one:
In[ ]:= Entity
"Country", "Japan"
EntityProperty
"Country", "AdultPopulation"

Out[
]= 7.44373  107 people

WolframAlpha can help us find out the age interval of the adult population property. (You
can just write == Adult population):
In[
]:= WolframAlpha
"Adult population",
IncludePods  "Input", AppearanceElements  "Pods"

Input interpretation:
Out[
]=
all countries, dependencies, and territories adult population

 

The following example uses EntityValue to find out the emissions of greenhouse gases
(GreenhouseEmissions) in Europe:
In[
]:= EntityValue
Europe COUNTRIES ,

greenhouse gas emissions , "EntityAssociation"  Short

Out[
]//Short=

 Albania  —, Andorra  —, 47,

8
United Kingdom  6.57396  10 metric tons of carbon dioxide equivalent yr,
Vatican City  —

In summary, the workflow for obtaining data (e.g., the diameter of Earth) is as follows:
 Create the entity, using Entity:
120 Mathematica beyond mathematics

In[
]:= Entity
"Planet", "Earth"

Out[
]= Earth
 Create the property, using EntityProperty:
In[
]:= EntityProperty
"Planet", "Diameter"

Out[
]= average diameter

 Obtain the property value for the entity, using EntityValue:
In[
]:= EntityValue
Entity
"Planet", "Earth",
EntityProperty
"Planet", "Diameter"
Out[
]= 12 742.018 km

or using the shorter notation Entity   :

In[
]:= Entity
"Planet", "Earth"
"Diameter"

Out[
]= 12 742.018 km

Summarizing :

Of course, we can also use the free-form input:

In[
]:= Earth PLANET
average diameter 

Out[
]= 12 742.018 km

Quite often, the last approach (with Ctrl+=) is the most convenient one because we don’t need
to worry about the exact syntax. However, for programing purposes, the standard form is
recommended to improve speed of execution and avoid ambiguity issues.
The price that we have to pay for such convenience is that     sometimes may not be
able to “read our minds”.

Example: Let’s say we want to learn more about the special powers of Mercury, the Roman
god of trade, merchants, thieves, and travelers. Since mercury is an entity present in more
than one entity type (planets, chemical elements, periodicals, ...), if we just write “mercury”
in free form,    ’s guess will be most probably that you are referring to a chemical
element. You can use EntityTypeName to verify it:

In[
]:= EntityTypeName
mercury ELEMENT 

Out[
]= Element
4. Accessing Scientific and Technical Information 121

To tell the program that we refer to the mythological figure, we need to press and
choose the appropriate option after clicking on “more”:

In[
]:= Mercury MYTHOLOGICAL FIGURE

Out[
]= Mercury
Now we can find out the actual name for that property:
In[
]:= Mercury MYTHOLOGICAL FIGURE
"Properties"

Out[
]=  alternate names , place of birth , body form , culture , place of death ,

entity classes , gender , image , locations , major temples , name , native names ,
patronages , special powers , subtype , symbols , type , literary references 
And finally learn that the ancient god was famous because he could fly and move very fast:
In[
]:= Mercury MYTHOLOGICAL FIGURE
"SpecialPowers"

Out[
]= flight, swiftness

Sometimes, even after going through all the options, we may not see what we are looking
for. The reason for this is that not all of the potential interpretations are included. To find
them use:
In[
]:= SemanticInterpretation
"Mercury", AmbiguityFunction  All  InputForm
Out[
]//InputForm=
AmbiguityList[{Entity["Element", "Mercury"], Entity["Planet", "Mercury"], Entity["Chemical",
"Mercury"],
EntityClass["MannedSpaceMission", "ProjectMercuryMannedMission"], Entity["Mineral", "Mercury"],
Entity["Mythology", "Mercury"], Entity["Periodical", "TheMercuryAUSHobart158944"],
Entity["Periodical", "TheMercuryZAFGreyvilleDurban156816"], Entity["Periodical",
"TheMercuryPAPottstown14607"]}]

Note that one of the interpretations refers to an EntityClass. A     function that
represents a class of entities identified by a certain characteristic. In this case, the class is
“ProjectMercuryMannedMission” included in the “MannedSpaceMission” entity type.
“MannedSpaceMission” contains the following entity classes:
In[
]:= EntityClassList
"MannedSpaceMission"  Shallow
Out[
]//Shallow=

 manned space missions , Apollo landings , Apollo crewed mission ,

Chinese crewed mission , Earth orbit crewed mission ,

International Space Station mission , lunar crewed mission , manned space mission ,
Mercury Atlas crewed mission , Mercury Redstone crewed mission , 13

The properties of the class “ApolloLandings” are:

In[
]:= EntityClass
"MannedSpaceMission", "ApolloLandings"
"Properties" 
Shallow
Out[
]//Shallow=

 alternate names , callsign , command and service module callsign , operator countries , crew ,

description , disciplines , distance traveled , Earth landing date , entity classes , 20

The entities included in the class “ApolloLandings” and the graphical representation
corresponding to the “LunarLandingDate” property are:
122 Mathematica beyond mathematics

In[
]:= EntityList
EntityClass
"MannedSpaceMission", "ApolloLandings"

Out[
]=  Apollo 11 , Apollo 12 , Apollo 14 , Apollo 15 , Apollo 16 , Apollo 17 

In[
]:= TimelinePlot

EntityList
EntityClass
"MannedSpaceMission", "ApolloLandings" 
"LunarLandingDate"

Apollo 14

Apollo 15

Out[
]=
Apollo 11 Apollo 16

Apollo 12 Apollo 17

1970 1971 1972

As we can see, Apollo 11 crew:

In[
]:= Apollo 11


Out[
]=  Neil Armstrong , Michael Collins , Buzz Aldrin 

This mission was led by the astronaut Neil Armstrong born on:

In[
]:= Neil Armstrong PERSON
"BirthDate"

Out[
]= Tue 5 Aug 1930

Additional possibility: the Entity framework can also be used to construct and execute queries
for relational databases (https://reference.wolfram.com/language/tutorial/RelationalDataba-
sesQuickStart.html .)
For additional information about entity types visit: https://reference.wolfram.com/lan-
guage/guide/EntityTypes.html.
If you are interested in knowing the sources of Wolfram Knowledgebase (https://www.wolfram-
.com/knowledgebase/ ) there are several ways:
 You can find them in the help for the type of entity on which you want to obtain the information. If the
information is available for that entity you will find it at the end where it says “Related links”. For example,
if you consult the Planet entity type help “Related links”, it will direct you to:
https://www.wolfram.com/knowledgebase/source-information/?page=PlanetData. However, not all entity
types include this information. If not, try the following:
 Ask Wolfram Alpha (https://www.wolfram.com/knowledgebase/source-information/ ). For example: If you
want to know where the data for the MannedSpaceMission entity type comes from, write in Wolfram
Alpha: https://www.wolframalpha.com/input/?i=Manned+Space+Mission. You will notice that the sources
are indicated at the end. If you then click on the icon, you will see:
4. Accessing Scientific and Technical Information 123

 The last option would be to contact Wolfram directly: https://www.wolfram.com/contact-us/

4.2 Computable Data Functions

The syntax described in the previous section to access entities was first introduced
in version 10. Previous versions of     used specific functions to access
computable data. They’re still useful.

Most of those functions finished with the suffix Data. You can use ?Data to explore some
of them (output not shown). Not all functions ending in “Data” are about computable data.
In[
]:= ? Data

Here are some of the most well-known:

Physics and Chemistry: ElementData, ChemicalData, IsotopeData, ParticleData ...
Earth Sciences: WeatherData, GeodesyData, GeoDistance ...
Astronomy: PlanetData, StarData ...
Life Sciences and Medicine: GenomeData, GenomeLookup, ProteinData ...
Economics and Finance: FinancialData, CountryData ...
Mathematics: FiniteGroupData, GraphData, KnotData, LatticeData, PolyhedronData ...
Linguistics: WordData, WordFrequencyData ...
Engineering: AircraftData, BridgeData, NuclearReactorData ...
People and History: PersonData, SocialMediaData, MovieData, Historical PeriodData ...
The sources used (if available) can be found in the help of the corresponding computable data
type under the “Related links” section. For instance, the link below will take you to all the
sources of information about WordData:
https://reference.wolfram.com/language/note/WordDataSourceInformation.html
In general, any computable data function is of the form: Function
/  
  0 /  0, or Function
/  0  /  0, {/  0, ...}, where
Entity is an entity identified by its name: Country, Element, Isotope ... and   
is information related to such entity. Currently, we can use both syntaxes
interchangeably most of the time.
Let’s see with examples how these functions operate.

To find out all the properties available for a specific function we can access the
documentation or just type:
124 Mathematica beyond mathematics

In[
]:= CityData
"Properties"

Out[
]= AlternateNames, Coordinates, Country, Elevation, FullName, Latitude, LocationLink,

Longitude, Name, Population, Region, RegionName, StandardName, TimeZone

If you type the entity, in this case “Salamanca”,     will return all the cities in the
world with that name:
In[
]:= CityData
"Salamanca"

Out[
]=  Salamanca , Salamanca , Salamanca , Salamanca , Salamanca , Salamanca 

By hovering the mouse over the list elements, you will see the details for each of the cities.
If you prefer to see them explicitly, just type the following:
In[
]:= salamanca 
SemanticInterpretation
"Salamanca",
Entity
"City", , AmbiguityFunction  All
Out[
]= AmbiguityList
 Salamanca , Salamanca , Salamanca , Salamanca , Salamanca , Salamanca , Salamanca

In[
]:= Take
First
salamanca, 6  InputForm

TableForm[{Entity["City", {"Salamanca", "Salamanca", "Spain"}], Entity["City", {"Salamanca",
"Guanajuato", "Mexico"}], Entity["City", {"Salamanca", "Coquimbo", "Chile"}],
Entity["City",{"Salamanca", "NewYork","UnitedStates"}], Entity["City",
{"Salamanca","NegrosOccidental", "Philippines"}], Entity["City", {"Salamanca", "Colon", "Panama"}]}]

If we intend to refer to a particular one, we need to be more precise and tell    
exactly what we want (if we let the program select the city for us, it may not make the right
choice). For example: if we’d like to know the population of the Spanish city of Salamanca,
we can use Entity with the output returned from the previous command.
In[
]:= CityData
Entity
"City", "Salamanca", "Salamanca", "Spain", "Population"

Out[
]= 159 754 people

Alternatively, we can use Interpreter (this function can be used any time there is ambiguity
risk).
In[
]:= Interpreter
"City"
"Salamanca, Spain"

Out[
]= Salamanca

Copy the output and paste it below:

In[
]:= CityData
Salamanca CITY , "Population"

Out[
]= 159 754 people

Notice that the previous output displays the units explicitly, in this case   . To see the
output without units, use QuantityMagnitude.
In[
]:= QuantityMagnitude
CityData

Entity
"City", "Salamanca", "Salamanca", "Spain" , "Population"
Out[
]= 159 754

If you wish to know all the information (properties) for a given entity of a computable data
function (DF) you can use the syntax:
Transpose[{DF[“Properties”], DF[entities, #]& /@ DF[“Properties”]}]

In the example below, we find all the information available in CityData for the city of
Salamanca, Salamanca province, Spain: {"Salamanca","Salamanca","Spain"}.
4. Accessing Scientific and Technical Information 125

In[
]:= Transpose
CityData
"Properties",

CityData
Entity
"City", "Salamanca", "Salamanca", "Spain" , &
CityData
"Properties"
AlternateNames Salamanque
Coordinates 40.97, 5.67
Country Spain
Elevation 818 m
FullName Salamanca, Salamanca, Spain
Latitude 40.97
Out[
]=
LocationLink http:maps.google.commaps?q 40.97,5.67&z 12&t h
Longitude 5.67
Name Salamanca
Population 159 754 people
Region Salamanca
RegionName Salamanca
StandardName Salamanca, Salamanca, Spain
TimeZone 1h
If we now select “http://maps.google.com/...” and press <F1> we will see the location of
Salamanca in GoogleMaps.

Moving from cities to stars, the example below shows all the different entity classes
available for the star entity.We add Shallow to show only the first few lines of the result.
Remove it to see the complete output.
In[
]:= StarData
"Classes"  Shallow
Out[
]//Shallow=

 stars , Algol variable stars , 2 Canum Venaticorum variable stars ,

Centauri , AM Herculis variable stars , Am stars , Ap stars ,
AR Lacertae variable stars , barium stars , Bayer stars , 178

In[
]:= StarData

Centauri STARS 

Out[
]=  Proxima Centauri , Rigil Kentaurus , Toliman 

The following function displays a table with the entities, properties, and classes available for
the specified collection:
In[ ]:= DataTable
fun Symbol : Module
data,
data  
Style
fun, "Section",
FontFamily  "Lucida Grande", FontSize  14, SpanFromLeft,
"", "
", "  ",
"Entities", Length
fun
, Append
RandomChoice
fun
, 2, " ",
"Properties", Length
fun
"Properties",
Append
RandomChoice
fun
"Properties", 2, " ",
"Classes", Length
fun
"Classes",
Append
RandomChoice
fun
"Classes", 2, " "
;
Grid
data, Frame  All
 126 Mathematica beyond mathematics

In[
]:= DataTable
CountryData

 
"  ' 
Entities 249  Barbados , Ireland , 
Out[
]=
Properties 223 ForeignRegisteredShips, HIVAIDSPopulation, 
Classes 337  australia extended ,

International Telecommunications Satellite Organization , 

Apart from asking     for information about many different entities, you can also
query the program about formulas:

For instance, if you need the formula for relativistic kinetic energy, you can type:
In[
]:= FormulaLookup
"Relativistic Kinetic Energy"

Out[
]= KineticEnergyRelativistic

Now we can use FormulaData to obtain the actual formula:

In[
]:= FormulaData
"KineticEnergyRelativistic"
1
Out[
]= 1  2     1 ,   
1 2  %2 1

Of course, you can always use Free-form input by typing “=” at the beginning of an input
cell followed by the relevant words in plain English.

In[
]:=  Kinetic Energy Relativistic formula

FormulaData
"KineticEnergyRelativistic"

1
Out[
]= 1  2     1 ,   
1 2  %2 1

    also has the capability to transcribe a given expression using SpokenString:
In[
]:= transcribe  SpokenString
Sqrt
x  y  z

Out[
]= square root of the quantity x over the quantity y plus z

In the next example, we use DictionaryLookup to search for English (other languages are
also available) words starting with “sh” and ending in “y”:
In[
]:= DictionaryLookup
"English", "sh"   "y"

Out[
]= shabbily, shabby, shadily, shadowy, shady, shaggy, shakily, shaky, shallowly, shamefacedly,
shamefully, shamelessly, shandy, shanty, shapelessly, shapely, sharply, shatteringly, shay,
sheeny, sheepishly, sherry, shiftily, shiftlessly, shifty, shimmery, shimmy, shinny, shiny,
shirty, shiveringly, shivery, shockingly, shoddily, shoddy, shortly, shortsightedly, shorty,
showery, showily, showy, shrewdly, shrilly, shrinkingly, shrubbery, shrubby, shyly

Using the program, we can also identify a language, transliterate expressions to different
alphabets, translate between languages, and even create a button that when pressed will
return a spoken representation of the argument:
In[
]:= text  "
,   ?";

In[
]:= origen  LanguageIdentify
text

Out[
]= Russian
4. Accessing Scientific and Technical Information 127

In[
]:= Transliterate
text, origen  "Spanish"

Out[
]= Privet, kak dela?

In[
]:= transcribe  TextTranslation
"
,   ?", "Spanish"

Out[
]= Hola cómo estás?

In[
]:= Button
"Press", Speak
transcribe

Out[
]= Press

Next, we are going to obtain the phonetic transcription (in American English) of the first
part of this sentence. We use SpokenString to split the phrase into words and the property
"PhoneticForm", from WordData, to get the phonetic transcription of each word. Since
the output is a list containing the words’ phonetic transcriptions, we apply Row[list, “ ”] to
join the elements substituting the commas with blank spaces.
In[
]:= Row
WordData
, "PhoneticForm" & 
StringSplit
"we are going to obtain the phonetic transcription", " "
Out[
]=


 
 
 
  
   
 

4.3 The Wolfram Data Repository

The Wolfram Data Repository contains an expanding collection of publicly
available computable datasets ready to be used for visualization, analysis and more.
The datasets can be accessed directly from within     or downloaded
using a web browser. Individual users can contribute datasets to the repository. An
overview of the repository is available at:
https://datarepository.wolframcloud.com/
The most common way to access resources available in the repository is to use
ResourceSearch and ResourceObject to find and retrieve a data resource and then
import the actual contents using ResourceData to proceed with the required
computation(s):
In[
]:= ResourceSearch
"magna carta"

In[
]:= ResourceObject
"Magna Carta"
"Properties"

Out[
]= AllVersions, Categories, ContentElementLocations, ContentElements, ContentSize,

ContentTypes, ContributorInformation, DefaultContentElement, Description,
Documentation, DocumentationLink, DOI, DownloadedVersion, ExampleNotebook,
ExampleNotebookObject, Format, InformationElements, Keywords, LatestUpdate, Name,
Originator, Properties, ReleaseDate, RepositoryLocation, ResourceLocations, ResourceType,
ShortName, SourceMetadata, UUID, Version, VersionInformation, VersionsAvailable
128 Mathematica beyond mathematics

In[
]:= ResourceObject
"Magna Carta"
"SourceMetadata"

Out[
]= Creator  King John of England, his barons, and Stephen Langton,

Date  Year: 1215 , Language  English, Source  English translation of Magna Carta 

In[ ]:= TextCases
ResourceData
"Magna Carta", "Sentence"1 ;; 2

Out[
]= Preamble:,
John, by the grace of God, king of England, lord of Ireland, duke of Normandy and Aquitaine,
and count of Anjou, to the archbishop, bishops, abbots, earls, barons, justiciaries,
foresters, sheriffs, stewards, servants, and to all his bailiffs and liege subjects, greetings.

We can perform many different type of analyses using the Wolfram Data
Repository. The example below shows the visualization of the evolution of
COVID-19 related cases and deaths in selected countries in the world.

During the pandemic, the “Epidemic Data for Novel Coronavirus COVID-19” was updated
daily with detailed information for provinces in Australia, Canada, and China and counties
and states in the US. For the rest of the world, the information was available at the national
level:
In[ ]:= ResourceUpdate
"Epidemic Data for Novel Coronavirus COVID19"

Out[
]= ResourceObject

Name: Epidemic Data for Novel Coronavirus COVID19 »

Type: DataResource 
Description: Estimated cases of novel coronavirus
COVID19, formerly known as 2019nCoV inf

In[
]:= ResourceObject
"Epidemic Data for Novel Coronavirus COVID19"

"ContentElements"
Out[
]= AustraliaProvinces, CanadaProvinces, ChinaProvinces, Dataset,
PuertoRicoProvinces, RawData, USCounties, USStates, WorldCountries

Here is the data for all the world countries:
In[
]:= covid19  ResourceData

"Epidemic Data for Novel Coronavirus COVID19", "WorldCountries";

The evolution of confirmed cases and deaths are shown for Spain:
In[
]:= DateListLogPlot

Values
Normal covid19
Select
"Country"  Entity
"Country", "Spain" &,
"ConfirmedCases", "Deaths",
PlotLabel  "Evolution of Cases and Deaths in Spain",
PlotLegends  "Confirmed Cases", "Deaths"
4. Accessing Scientific and Technical Information 129

We can also visualize the estimated deaths per 100,000 people by country (choosing the 30
countries with most deaths due to the pandemic):
In[
]:= DateListLogPlot

Callout
Tooltip
10 ^ 5 2  1
"Population", CommonName
1, 1 &
Normal ResourceData

"Epidemic Data for Novel Coronavirus COVID19", "WorldCountries"

TakeLargestBy
Deaths
"LastValue" &, 30

All,  Country, Deaths &, Sequence

PlotRange  All, GridLines  Automatic, AspectRatio  1.5,
ImageSize  500, PlotLabel  "Estimated Deaths per 100,000 People log \
scale"

4.4 Weather Data in Real Time

ClearAll
"Global`"
Clears values and definitions for all the previously defined symbols

With computable data we can have access to historical as well as real-time weather
information: WeatherData, WeatherForecastData, AirTemperatureData,
AirPressureData, WindSpeedData, WindDirectionData, WindVectorData.
The
[“   0, “  0,{      }] returns the value of the
variable specified in “  ” for a given location and period (by default it
returns the most recent value). Of course you can always use the Free-form input
 130 Mathematica beyond mathematics

first and then the actual function returned by    .

For example, if you want to know the temperature in your location, just type:

In[
]:=  Temperature

AirTemperatureData


Out[
]= 17. C

Notice that the function used by     to get the temperature is AirTemperatureData.
By default, it will provide you with the temperature of your location (actually your IP
address location). This information is obtained from weather stations affiliated with the
World Meteorological Organization (WMO). You can use the same function to find out
from which weather station you got the air temperature reading.
In[
]:= AirTemperatureData
"StationName"

Out[
]= LESA

Alternatively, GeoNearest can also be used to obtain the same information:

In[
]:= GeoNearest
"WeatherStation", Here 

Out[
]=  LESA 

The weather station position can be found as follows:

In[
]:= Entity
"WeatherStation", "LESA"
"Position"

Out[
]= GeoPosition40.959, 5.498

In this example, we get the average daily temperature in Salamanca (Spain) from January 1,
2016 to January 1, 2022 and represent it graphically. Notice the seasonality.
In[
]:= DateListPlot
AirTemperatureData

Salamanca CITY , 2016, 1, 1, 2022, 1, 1, "Day"

25

20

15

Out[
]= 10

5

2016 2018 2020 2022

The graph below displays the range of daily temperatures over the past 180 days that
corresponds to that station code.
4. Accessing Scientific and Technical Information 131

In[
]:= DateListPlot
WeatherData
"LESA", 1,

DateString
DatePlus
180, DateString
DatePlus
0, "Day" & 
"MeanTemperature", "MaxTemperature", "MinTemperature",
Joined  True, False, False, Filling  2  3, Mesh  All

30

20

Out[
]= 10

Jan Mar May

The following function generates a series of weather-related forecasts for the week after:
In[
]:= Manipulate
data  WeatherForecastData

Entity
"City", "Salamanca", "Salamanca", "Spain";
DateListPlot
Normal
data
g . Interval
x   Mean
x,
FrameLabel  Automatic, g, "Temperature", "WindSpeed",
"WindDirection", "PrecipitationAmount", SaveDefinitions  True

The final example of this section generates 100 pseudo-random geo position points on the
surface of Spain, obtains their most recent win vector readings from the nearest weather
station, and finally generates a stream plot:

In[
]:= pts  RandomGeoPosition
Spain COUNTRY , 100

Number of points: 100

Out[ ]= GeoPosition Lat bounds: 36.7, 43.4 
Lon bounds: 8.64, 3.07

In[
]:= wind  WindVectorData
pts, Now, "DownwindGeoVector";

In[
]:= GeoStreamPlot
wind

132 Mathematica beyond mathematics

4.5 Chemical and Physical Properties of Elements and Compounds

4.5.1 Element Data

With ElementData or Entity[“Element”] and ChemicalData or Entity[“Chemical”],
we can access a wide variety of information about chemical elements and chemical
compounds. WolframAlpha’s website has additional details about what exactly can
be accessed:
https://www.wolframalpha.com/examples/science-and-technology/chemistry/

Both functions use the following color scheme when representing atoms and molecules:
In[
]:= ClearAll
"Global`"

In[
]:= ColorData
"Atoms", "Panel"

H He

Li Be B C N O F Ne

Na Mg Al Si P S Cl Ar

K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr

Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe
Out[
]=
Cs Ba Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po At Rn

Fr Ra Rf Db Sg Bh Hs Mt Ds Rg Cn Nh Fl Mc Lv Ts Og

La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Ac Th Pa U Np Pu Am Cm Bk Cf Es Fm Md No Lr

If you click on an element symbol, you will see the color code assigned by the function.

Let’s take a look at the properties available in ElementData:
4. Accessing Scientific and Technical Information 133

In[
]:= ElementData
"Properties"  Shallow

Out[
]//Shallow=

 adiabatic index , allotropes , allotropic multiplicities , alternate names , atomic mass ,

atomic number , atomic radius , atomic symbol , block , boiling point , 110

  
       displays the requested property

information for a given element.

The next function shows the crystalline structure of an element in its solid phase:

In[
]:= ElementData
"Uranium", "CrystalStructureImage"

We can use Dataset to display all the information associated with an entity and
DeleteMissing to eliminate those properties without information. We can also include the
option MaxItems to limit the maximum number of items displayed simultaneously.

In[
]:= Dataset
DeleteMissing
carbon ELEMENT
"Dataset", MaxItems  10

The following function tells us the abundance (as a fraction of the mass) of the different
elements in the human body:
In[
]:= humanabundance 
Table
QuantityMagnitude
ElementData
z, "HumanAbundance",
ElementData
z, z, 92;
We keep only those whose presence in the body is greater than 0.1% of the body mass. We
order them by comparing the first element of each sublist using a combination of Sort and a
pure function.
134 Mathematica beyond mathematics

In[
]:= Sort
Cases
humanabundance, x , y  ; x  0.001, 11  21 &

61. oxygen
23. carbon
10. hydrogen
2.6 nitrogen
1.4 calcium
1.1 phosphorus
0.20 potassium
Out[
]= 0.20 sulfur
0.14 sodium
0.12 chlorine
0.027 magnesium
0.026 silicon
0.0060 iron
0.0037 fluorine
0.0033 zinc

4.5.2 Molecule Data

With Molecule, we can represent a molecule with atoms and bonds. We can also generate molecules from
chemical entities.
In[
]:= Molecule
"O", "H", "H", Bond
1, 2, "Single", Bond
1, 3, "Single"

Formula: H2 O
Out[
]= Molecule 
Atoms: 3 Bonds: 2

In the example that follows, we show the caffeine molecular plot. We can move the plot
with the mouse to visualize the molecule from different perspectives:
In[ ]:= m  Molecule
caffeine CHEMICAL 

Formula: C8 H10 N4 O2
Out[
]= Molecule O
N

N
N

N Atoms: 24 Bonds: 25 

In[
]:= MoleculePlot3D
m

Out[
]=

With ChemicalReaction you can enter a reaction as a string, and then ReactionBalance is
used to balance it. Finally reaction balance equation is represented in the traditional form:
4. Accessing Scientific and Technical Information 135

In[
]:= ReactionBalance
ChemicalReaction
"C8H10N4O2  O2  CO2  H2O  N4"

"EquationDisplay"  TraditionalForm
Out[ ]//TraditionalForm=
2 C8 H10 N4 O2 19 O2  16 CO2 10 H2 O 2 N4

4.5.3 Chemical Data

We can obtain chemical data by either using ChemicalData[“name”,”property”] or

Entity[“name”][“property”]. Both functions will show the information specified in
“property” for a given compound.
In[
]:= ChemicalData
"Butane", "MoleculePlot"

Out[
]=

A graph application below is shown. First the graph structure for a molecule is represented
and then a 6-cycle can find:

In[
]:= ChemicalData
caffeine CHEMICAL , "StructureGraph"

H
O H
C
H H
H
C C
N
H N C

C H
Out[ ]=
C C
N
O N

H C H

In[
]:= FindIsomorphicSubgraph
, CycleGraph
6

C
N C
Out[
]= 

C C
N

We can also compare the properties of two biomolecules:
In[
]:=

EntityValue
cytosine CHEMICAL , glycine CHEMICAL ,

structure diagram , melting point , "MolarMass" , "Dataset"

136 Mathematica beyond mathematics

You can now even get step-by-step answers to problems in Chemistry:

https://www.wolframalpha.com/examples/pro-features/step-by-step-solutions/step-by-step-
chemistry/

4.6 Life Sciences and Medicine

The Wolfram Language provides access to an extensive collection of data related to
life sciences and medicine. Computations related to genomics, food and nutrition,
neuroscience, molecular biology and many other fields can be readily perform. This
section covers examples about the anatomy of humans and animals, DNA and
proteins.

4.6.1 Anatomic Data

Human and animal anatomical structure data are available in Mathematica. Their
representation is based on the UMLS (Unified Medical Language System):

Example: Simple hierarchical network of UMLS concept

Here is the list of the properties associated with the anatomical structure data:
4. Accessing Scientific and Technical Information 137

In[
]:= Entity
"AnatomicalStructure"
"Properties"  Shallow

Out[
]//Shallow=

 abbreviation , adjacent joint count , adjacent joints ,

adjacent structures , afferent vessel , alternate names , anatomical groups ,

anatomical regions , anatomical spaces , arterial supply , 127

The function below returns the parts that form the lung:

In[
]:= lung ANATOMICAL STRUCTURE
constitutional parts   Shallow

Out[
]//Shallow=

 lobe of lung , visceral pleura , anterior pulmonary nerve plexus ,

posterior pulmonary nerve plexus , root of lung ,

intrapulmonary part of tracheobronchial tree , lung parenchyma ,
intrapulmonary part of bronchial artery , intrapulmonary part of pulmonary arterial tree ,
intrapulmonary part of pulmonary venous tree organ , 2

We can also visualize its anatomical structure:

In[
]:= EntityValue
lung ANATOMICAL STRUCTURE ,

EntityProperty
"AnatomicalStructure",
"RegionalLocationHighlightedImage", "View"  "All"

We generate a 3-D plot of the alimentary system:

In[
]:= AnatomyPlot3D
alimentary system ANATOMICAL STRUCTURE 

138 Mathematica beyond mathematics

This instruction returns all the available properties for dogs’ muscles:

In[
]:= muscle (dog) ANIMAL ANATOMICAL STRUCTURE
"Properties"  Shallow

Out[
]//Shallow=

 adjacent joint count , adjacent joints , anatomical groups , articulating bones ,

associated species , bilateral components , body location image ,

body side , bone articulation , bone articulation count , 36

We can even answer questions such as: “What does the dog's femur bone connect to?”

In[
]:= femur (dog) ANIMAL ANATOMICAL STRUCTURE
bone articulation 

Out[
]=  hip bone (dog) , patella (dog) , tibia (dog) 

Or find out the body location image of horses’ organs:

In[
]:= organ (domestic horse) ANIMAL ANATOMICAL STRUCTURE
"BodyLocationImage"

Out[
]=

4.6.2 Genetic and Protein Data

The following examples refer to typical functions applicable to genetic and protein
data.
4. Accessing Scientific and Technical Information 139

We can inspect the properties of the DNA entity and visualize its structure:

In[
]:= DNA BIOMOLECULAR SEQUENCE TYPE
"Properties"  Shallow

Out[
]//Shallow=

 additional bond types , allowed bonding cardinality rules ,

allowed bonding rules , alphabet , alphabet rules , alternate encoding rules ,

all associated bond types , caption , circular , complement letter rules , 5

In[
]:= DNA BIOMOLECULAR SEQUENCE TYPE
alphabet rules 

Out[
]= A  adenine , C  cytosine , G  guanine , T  thymine 

In[
]:= MoleculePlot3D
adenine CHEMICAL 

Out[
]=

We can obtain even more complex representations using the new functions BioSequence
and BioSequenceComplement:
In[
]:= MoleculePlot3D
Molecule
BioSequence
"DNA", "CTGA"

Out[
]=

In[
]:= BioSequenceComplement
BioSequence
"DNA", "CTGA"

Type: DNA Sequence

Out[
]= BioSequence 
Content: GACT
4 letters

GenomeLookup shows the chromosomes in which a specified DNA sequence appears along
with its position inside of them.
ClearAll
"Global`"
140 Mathematica beyond mathematics

In[
]:= GenomeLookup
"AATTTCCGTTAAAT"

Chromosome2, 1 213 319 362, 213 319 375
Chromosome3, 1 83 249 225, 83 249 238
Chromosome7, 1 78 235 739, 78 235 752
Chromosome9, 1 5 746 733, 5 746 746
Out[
]=
Chromosome11, 1 41 489 916, 41 489 929
Chromosome11, 1 124 909 525, 124 909 538
Chromosome22, 1 11 132 978, 11 132 991
ChromosomeX, 1 144 568 117, 144 568 130

GenomeData[“gene”,”property”] returns the DNA sequence for a specified gene on the

reference human genome. With “property” we can specify a characteristic of the chosen
gene. In this example, we show the biological processes where the gene “ACAA2” plays a
role:
In[ ]:= GenomeData
"ACAA2", "BiologicalProcesses"

Out[
]= CholesterolBiosyntheticProcess,
FattyAcidMetabolicProcess, LipidMetabolicProcess, MetabolicProcess

ProteinData[“protein”] returns the amino acids of a given protein. In this case, we display
the sequence of the protein A2M. We use Short to avoid displaying the complete output.
In[
]:= ProteinData
"A2M"  Short
Out[
]//Short=
MGKNKLLHPSLVLLLLVLLPTDASVSGKPQYMVLVPSLLHTETTEKGCVLLSYLN 
LDKVSNQTLSLFFTVLQDVPVRDLKPAIVKVYDYYETDEFAIAEYNAPCSKDLGNA

ProteinData[“protein”,”property”] gives us the option to choose the property of the protein
we want to know about. As an illustration, to visualize the A2M protein we can use
“MoleculePlot” (this property is not yet available for all proteins at the time of writing).
In[
]:= ProteinData
"A2M", "MoleculePlot"

WolframAlpha (e.g., starting a input cell ==) or the Free-form input allows us to see an
overview of available results for the specified gene. (Cassel, B. J. (2018, October 16–19).
2        &   "  [Conference Presentation]. Wolfram Technology
Conference 2018)
4. Accessing Scientific and Technical Information 141

mus81 gene

In[
]:= MUS81 structure-specific endonuclease subunit GENE

Out[
]= MUS81 structure-specific endonuclease subunit

We can retrieve the chromosomal context elided above:

In[
]:= EntityValue
, "Context"
chromosome 11

Out[
]=

Metabolic pathways are the key cycles by which food and fuel are converted into structures
and energy for biological systems. Here is the famous TCA cycle:

TCA cycle

In[
]:= Wolfram|Alpha pod

TCA  citric acid cycle 

Out[
]=
tricarboxylic acid cycle  TCA cycle  aerobic respiration

Click “WolframAlpha results” choosing “PathwayTopology” and click in the upper right
corner PathwayTopolog(+) “Formatted Pod”
In[ ]:= WolframAlpha
"TCA cycle",
IncludePods  "PathwayTopology:MetabolicPathwayData",
AppearanceElements  "Pods",
TimeConstraint  20, Automatic, Automatic, Automatic

The following function is considerably more complicated. It have been adapted

142 Mathematica beyond mathematics

from examples provided by Wolfram Research.

The function below identifies the chromosomes and the position where a DNA sequence is
located inside of them :
In[
]:= FindSequence
sequence  :
chrpos 
SortBy

Module
chr  , len, scale  N
GenomeData
, "SequenceLength" 
GenomeData
"Chromosome1", "SequenceLength",
rls,
rls  Flatten
 1, 1, 1  All, 2, 1  scale  len & 
SplitBy
Select
GenomeLookup
sequence,
1, 2  1 &, 1, 1 &,  ;
len  GenomeData
chr, "SequenceLength";
chr,
Extract
GenomeData
chr, "GBandScaledPositions",
Position
GenomeData
chr, "GBandStainingCodes",
"acen", 1, 11, 1  scale,
scale, GenomeData
, "AlternateStandardNames"1,
GenomeData
chr, "AlternateStandardNames"2, . rls & 
Most GenomeData
"Chromosomes", 5 &;

Here we use the previously defined function (
  ) to find, for a given
sequence, the chromosomes and the position in which they appear.
In[
]:= FindSequence
"AACCTTGGGAAATT"  Shallow
Out[
]//Shallow=
Chromosome1, 0.471457, 1., 1, 1, 3, Chromosome2, 0.362134, 0.982614, 2, 2, 5,
Chromosome3, 0.354132, 0.806884, 3, 3, 3,
Chromosome4, 0.199253, 0.773603, 4, 4, 8,
Chromosome5, 0.166101, 0.731479, 5, 5, 3,
Chromosome6, 0.23412, 0.691204, 6, 6, 1,
Chromosome7, 0.227125, 0.642352, 7, 7, 3,
Chromosome8, 0.168017, 0.591608, 8, 8, 1,
Chromosome9, 0.180048, 0.567334, 9, 9, ,
Chromosome10, 0.163645, 0.547522, 10, 10, 1, 14

4.7 Earth Sciences and Geographic Data

In[
]:= ClearAll
"Global`"

There are many functions available for the computation of data related to earth
sciences and geography:
https://reference.wolfram.com/language/guide/EarthSciencesDataAndComputation.html
http://reference.wolfram.com/language/guide/GeographicData.html

4.7.1 Geodata and Geographics

We can start getting an idea by asking     to list all the functions that start with
the letter Geo (output omitted).
In[
]:= ? Geo
4. Accessing Scientific and Technical Information 143

The names of the functions enable us to identify those related to geography, almost all of
them except the ones starting with Geometric. Additionally, we can always access the help
documentation to see their syntax. Let’s take a look at some of them.

Using FindGeoLocation[] you can find your current geolocation of your computer (you can
find information about the criteria used to find your computer location in the Help) :
In[
]:= FindGeoLocation


Out[
]= GeoPosition40.97, 5.66

Next we use GeoGraphics to display a map of Salamanca (a City of Spain) and its
surroundings. Remember that in many cases you will find the Free-form input very useful as
well ( 
     and type Salamanca).
In[
]:= GeoGraphics
GeoDisk
Salamanca CITY , Quantity
50, "km"

An easy way of represents a dynamic, interactive, two-dimensional geographical image is

with     .

In[
]:= DynamicGeoGraphics
Salamanca CITY , GeoRange  Quantity
10, "km"

144 Mathematica beyond mathematics

Here we use GeoNearest to see the closest locations to Salamanca:

In[
]:= GeoNearest
"City", Salamanca , All, 10 km

Out[
]=  Salamanca , Villamayor , Santa Marta de Tormes ,

Villares de la Reina , Carbajosa de la Sagrada , Aldeatejada , Cabrerizos ,

Carrascal de Barregas , Doninos de Salamanca , Monterrubio de Armuna ,
San Cristobal de la Cuesta , Moriscos , Arapiles , Florida de Liebana ,
Castellanos de Villiquera , Castellanos de Moriscos , Pelabravo 

GeoElevationData returns the elevation of a geographic site above sea level in meters,
informs us about the minimum and maximum elevations, and creates a relief plot:
In[ ]:= elevation 
GeoElevationData
GeoDisk
Salamanca CITY , Quantity
100, "km"

Dimensions: 361, 477

Out[ ]= QuantityArray 
Unit: Meters

Data not in notebook. Store now

In[
]:= MinMax
elevation

Out[
]= 140. m, 2426. m

In[
]:= ReliefPlot
elevation, DataReversed  True,

ImageSize  500, ColorFunctionScaling  False,
ColorFunction  ColorData
"HypsometricTints",
PlotLegends  Placed
Automatic, Right
4. Accessing Scientific and Technical Information 145

In[
]:= GeoRegionValuePlot
CountryData
  "LiteracyRate",

ColorFunctionBinning  "Quantile", 20,
GeoBackground  "VectorMarketing", GeoProjection  "Mollweide"

99.886
99.002
97.850
Out[
]= 95.635
89.080
76.800
45.000

We can use CloudDeploy in combination with GeoGraphics to build a web

application.

The function below generates a map of a chosen location (we will normally enter the name
of a city) and its surroundings for a given radius:
In[
]:= CloudDeploy

FormFunction
"Location"  "Location", "Radius"  "Quantity",
GeoGraphics
GeoDisk
Location, Radius &, "PNG"

Out[
]= CloudObjecthttps:www.wolframcloud.comobj047bd1e9759948e581817f69dd1561a2 

We can now copy this newly created link to our website to make it accessible from any
device with web browsing capabilities:
< a href = "https://www.wolframcloud.com/objects/..." > City Map (html) < /a >

The application is now available to any user, although the first time we access it, we may
have to enter our Wolfram ID and password. However, we don't need to have    
installed in our computers. Once we enter the required data (Figure 4.1) we get the
corresponding map (Figure 4.2).

Figure 4.1 Choosing the location and the radius.

146 Mathematica beyond mathematics

Figure 4.2 WolframCloud generated map.

The Earth, although not flat, is not really a sphere, it is a spheroid with the polar
radius being shorter than the equatorial
Terrestrial measurements are usually expressed as established by the The International Terres-
trial Reference Frame (ITRF):
https://www.iers.org/IERS/EN/DataProducts/ITRF/itrf.html

International Terrestrial Reference Frame

In[
]:=

Out[
]= ITRF00

Using the ITRF data the difference between the equatorial and polar radius is:
In[
]:= GeodesyData
"ITRF00", "Semiaxes"

Out[
]= 6.37814  106 m, 6.35675  106 m

In[
]:= Subtract 

Out[
]= 21 384.7 m

In this example, we show how to calculate the distance between two locations on
the Earth: Where I am (Here) to Sydney

In[
]:= loc  GeoPosition
Sydney CITY 

Out[ ]= GeoPosition33.87, 151.21

In[
]:= GeoDistance
Here, loc

Out[
]= 17 836.4 km
4. Accessing Scientific and Technical Information 147

This it is the the shortest distance (geodesic) that can be represented as follows:
In[
]:= GeoGraphics

PointSize
Large, Point
Here, Point
loc, Arrow GeoPath
Here, loc,
GeoGridLines  Quantity
10, "AngularDegrees",
GeoProjection  "Orthographic"

Out[
]=

4.7.2 On Earth and Beyond

This example returns the location of all the earthquakes that happened in Chile between
2011-1-1 and 2021-12-31 with magnitudes between 6 and 10:
In[
]:= data  EarthquakeData
Entity
"Country", "Chile",
6, 10, 2011, 1, 1, 2020, 12, 31, "Position"

Time: 02 Jan 2011 to 14 Dec 2020

Out[
]= EventSeries 
Data points: 36
148 Mathematica beyond mathematics

In[
]:= GeoListPlot
data

Out[
]=

GeoListPlot can also generate maps for locations outside the Earth. In the example below,
we visualize the landing of Apollo 11, the first manned lunar landing mission:
In[ ]:= GeoListPlot
Apollo 11 MANNED SPACE MISSION ,
GeoRange  All, GeoProjection  "Orthographic",
GeoLabels  True, LabelStyle  White, ImageSize  Small

4.7.3 Travelling Around the World

In this example, we combine several functions to show some travel routes
originating in Madrid.
4. Accessing Scientific and Technical Information 149

We first define the cities that we want to travel to. It is better to add the country when typing
the city to avoid ambiguities in case there are several cities with the same name.
{"London, England", "Nuuk, Greenland", "Buenos Aires, Argentina", "Los Angeles, USA",
"Malmo, Sweden", "Ulan Bator, Mongolia", "Sidney, Australia"}
In[
]:= cities 

London CITY , Nuuk CITY ,

Buenos Aires CITY , Los Angeles CITY ,

Stockholm CITY , Ulaanbaatar CITY , Sydney CITY

Out[
]=  London , Nuuk , Buenos Aires , Los Angeles , Stockholm , Ulaanbaatar , Sydney 

In[
]:= GeoGraphPlot
Thread
Madrid CITY  cities, GeoCenter  Madrid CITY ,

GraphLayout  "Geodesic", VertexLabels  "Name",
GeoBackground  "VectorDark", GeoProjection  "WinkelTripel"

Out[
]=

4.7.4 What Happens in the Valparaiso Geomagnetic Field?

During a visit to Universidad Técnica Federico Santa María (UFSM), a Chilean
university located in Valparaíso, I observed a strange behavior in the compass of
my smartphone. The magnetic field sensor indicated a value (around 20 T) that
seemed unusually low to me. Maybe it wasn’t working properly?
150 Mathematica beyond mathematics

I decided to first compare the geomagnetic information available for Valparaiso

(Chile) with the one available for my place of residence (Salamanca, Spain) using
GeomagneticModelData.

Here is the resulting comparison table:

In[
]:= GeomagneticModelData
Salamanca CITY ,

GeomagneticModelData
Valparaiso CITY   Dataset

Then I examined the magnitude of the geomagnetic field across the entire planet. The
magnitude at Valparaíso was very low indicating that there was actually no problem with
the magnetic sensor of my mobile phone.
In[
]:= data  GeomagneticModelData

90, 180., 90, 180., "Magnitude", GeoZoomLevel  2;
In[
]:= GeoContourPlot

Flatten Table
GeoPosition
 16  i  45  8, 32  j  45  8 
data33  i, 65  j, i, 1, 32, j, 1, 64, ContourShading  True,
ColorFunction  "TemperatureMap", PlotLegends  Automatic

nT

62 631

53 641
Out[
]= 44 650

35 660

26 670

The next series of commands generate a plot displaying the streamlines based on magnetic
direction fields from the South Pole to Valparaiso:

In[
]:= GeoPosition
Valparaiso CITY 

Out[
]= GeoPosition33.04, 71.64

In[
]:= coords  GeoPosition

Flatten
Table
lat, lon, lat, 10, 80, 10, lon, 50, 90, 10, 1

Number of points: 40
Out[
]= GeoPosition Lat bounds: 80, 10 
Lon bounds: 90, 50
4. Accessing Scientific and Technical Information 151

gm  GeomagneticModelData
coords;
norms  QuantityMagnitude
gmAll, "Magnitude";
declinations  gmAll, "Declination";
vectors  Transpose
norms, declinations;
In[
]:= GeoStreamPlot
GeoVector
coords  vectors,
GeoBackground  "ReliefMap", StreamStyle  White

Out[
]=

The intensity and position of the magnetic field changes slightly over time (see
GeomagneticModelData >Applications). Even the positions of the magnetic poles
themselves have interchanged several times in the past:
In[
]:= TextSentences
WikipediaData
"Geomagnetic Reversal"1 ;; 5

Out[
]= A geomagnetic reversal is a change in a planet's magnetic

field such that the positions of magnetic north and magnetic south are
interchanged not to be confused with geographic north and geographic south .,
The Earth's field has alternated between periods of normal polarity, in which
the predominant direction of the field was the same as the present
direction, and reverse polarity, in which it was the opposite.,
These periods are called chrons., Reversal occurrences are statistically random.,
There have been 183 reversals over the last
83 million years on average once every 450,000 years .

4.8 Additional Resources

To access the following resources, if they refer to the help files, write their
locations in a notebook (e.g., guide/EarthSciencesDataAndComputation), select
them and press <F1>. In the case of external links, copy the web addresses in a
browser:
Astronomical Data: guide/AstronomicalComputationAndData
Biomolecular Sequences: guide/BiomolecularSequences
Earth Sciences: guide/EarthSciencesDataAndComputation
EngineeringData: guide/EngineeringData
152 Mathematica beyond mathematics

Financial and Economic Data: guide/FinancialAndEconomicData

Geodesy: guide/Geodesy
GeoGraphics:tutorial/GeoGraphics
Life Sciences and Medicine: guide/LifeSciencesAndMedicineDataAndComputation
Socioeconomic Data: guide/SocioeconomicAndDemographicData
Physics and Chemistry: guide/PhysicsAndChemistryDataAndComputation
Transportation Data: guide/TransportationData
WordData: ref/WordData
5
Data Analysis and Manipulation

This chapter gives an overview of the software’s capabilities for data

analysis. The WL probably includes as many probability and statistical
functions, if not more, than any other specialized computer programs for
analyzing data. The downside is that, very often, users must have a deep
understanding of what they are trying to accomplish to take full
advantage of the program’s capabilities in this area. Topics such as
importing and exporting data, exploring data, curve fitting, analysis of
time series, and even how to perform spatial statistics calculations are all
covered. After reading this chapter, the reader will feel comfortable
performing typical data analysis tasks with Mathematica.

5.1 Importing/Exporting
A fundamental feature of Mathematica is its capability for importing and exporting
files in different formats. The commands we need for that are Import and
Export. They enable us to import or export many different types: XLS, CSV, TSV,
DBF, MDB, ...
A file can be imported or exported by typing the complete source or destination
path.

In Windows, to type the file location for import or export purposes use the following syntax:
“...\\directory\\file.ext” , as shown in the following example:
Import
"...\\Mathbeyond\\Data\\file.xls"
If we want to use the working directory in which the files are located for importing
or exporting purposes, normally the easiest way will be to create a subdirectory
inside the directory where the notebook is.

The following function checks the directory of the active notebook.
In[
]:= NotebookDirectory


Out[
]= F:\Mi unidad\Mathematica\MBN13\

154 Mathematica beyond mathematics

In the directory above, create a subdirectory named Data.

With the following function we set the subdirectory Data as our default working directory,
the one that the Import and Export functions will use during this session.
In[
]:= SetDirectory
FileNameJoin
NotebookDirectory
, "Data"

Out[
]= G:\Mi unidad\MBN2Ed\Data

We make the previous cell an Initialization Cell to ensure its evaluation before any
other cell in the notebook by clicking on the cell marker and choosing from the
menu bar: Cell
Properties
Initialization Cell). This means that all the data will
be imported/exported from the Data subdirectory. We may even consider “hiding”
the initialization cell by moving it to the end of the notebook. Remember that the
order of operations is not the same as it appears on screen. It is the order in which
the cells are evaluated. Therefore, even if an initialization cell is located at the end,
it will still be the first one to be executed.

The total number of available formats can be seen using $ImportFormats and
$ExportFormats.
In[
]:=
$ImportFormats  Length, $ExportFormats  Length

Out[
]=
235, 192

In this section, we are going to use, among others, the files available as examples in
a subdirectory named ExampleData created during the program installation. Some
of these examples are downloaded from www.wolfram.com (Internet connection
required).

We are interested in the ones related to statistics. The following function displays the name
of the example, a brief description, and the dimensions of its data matrix.
In[
]:= Shallow

2, ExampleData
, "Description", Dimensions
ExampleData
 & 
ExampleData
"Statistics", 5, 3
Out[
]//Shallow=


AirlinePassengerMiles, Revenue passenger miles flown by commercial airlines.,
24,

AirplaneGlass, Time to failure for airplane glass.,
31,

AnimalWeights, Brain and body weights for 28 animal species.,
28, 3,

AnorexiaTreatment, Anorexia data on weight change.,
72, 3,

AnscombeRegressionLines, Anscombe's 4 regression line data.,
11, 8, 106

5.1.1 Import

To import files use Import "pathfile.ext", "Elements" , where file is the name
of the file you want to import. If the file is in the Data directory previously
defined, you only need to type the file name. If it is located somewhere else you
will need to enter the complete path (remember that you can use
Insert  File Path...). You can also import files directly from the
Internet. To do that, specify the desired URL in Import[“http://url”]. As
an option you can import only certain file elements, the ones specified by
Elements.
5. Data Analysis and Manipulation 155

If your computer uses a non-American configuration remember that Mathematica

uses the dot “.” as a decimal separator and the comma “,” to distinguish between
list elements. If you have another configuration you may have problems when
importing files. The program enables the configuration of the format in Edit

Preferences...
Appearance to suit the user needs. Nevertheless, it is better to
modify the operating system global settings so that all programs use the same
setup. In Windows that can be done by selecting in the Control Panel “.” as
decimal separator and the comma “,” as list separator. In OS X, the same can be
done by going to System Preferences...
Language & Region
Advanced...

Next, we import an xls file containing data about the atomic elements. Note that the program
can figure out the type of file from its extension. In this case, the program identifies it as a
Microsoft Excel file.
In[
]:= Import
"ExampleDataelements.xls"

Out[
]= 
AtomicNumber, Abbreviation, Name, AtomicWeight
1., H, Hydrogen, 1.00793
2., He, Helium

Notice that the previous output has the following structure {{{ ...}}}. The reason is that
Mathematica assumes the file is an Excel spreadsheet that may contain several worksheets.
Using “Data”, 1, we tell the program to import the data from worksheet 1 (do not
confuse this Data with the working subdirectory Data that we have created earlier):
In[
]:= examplexls  Import
"ExampleDataelements.xls", "Data", 1
AtomicNumber Abbreviation Name AtomicWeight
1. H Hydrogen 1.00793
2. He Helium 4.00259
3. Li Lithium 6.94141
4. Be Beryllium 9.01218
Out[
]=
5. B Boron 10.8086
6. C Carbon 12.0107
7. N Nitrogen 14.0067
8. O Oxygen 15.9961
9. F Fluorine 18.9984

We show below how to extract the first 5 rows of the worksheet and display them in a table
format.
In[
]:= TableForm
Take
examplexls, 5
Out[
]//TableForm=
AtomicNumber Abbreviation Name AtomicWeight
1. H Hydrogen 1.00793
2. He Helium 4.00259
3. Li Lithium 6.94141
4. Be Beryllium 9.01218

The next function imports a single column. In this case, we get the fourth column containing
the atomic weights.
In[
]:= Import
"ExampleDataelements.xls", "Data", 1, All, 4

Out[
]=
AtomicWeight, 1.00793, 4.00259, 6.94141,

9.01218, 10.8086, 12.0107, 14.0067, 15.9961, 18.9984

To remove the first sublist containing the headings (the same could have been done using
Drop
data, 1), we use Rest.
In[
]:= data  Rest
Import
"ExampleDataelements.xls" , "Data", 1;
156 Mathematica beyond mathematics

We can also visualize the atomic weights of the imported data. Tooltip is included so that
when the cursor is over a point in the graphic, its corresponding atomic weight can be seen.
In[
]:= ListPlot
Tooltip
dataAll, 4,
Filling  Axis, PlotLabel  "Atomic Weight"
Atomic Weight
20

15

Out[ ]= 10

2 4 6 8

Try to copy and paste links to any of the numerous images available in:
https://earthobservatory.nasa.gov/.
Note: The examples using files imported from the Internet may not work if their web addresses have been
changed or the files have been deleted.
In[
]:= Import
"https:eoimages.gsfc.nasa.govimagesimagerecords7800078349
arctic vir 2012147 lrg.jpg"

Out[
]=

With Import[..., “Elements”], we can find out what elements of the file are available to
import (e.g., ImageSize).
In[
]:= Import
"https:eoimages.gsfc.nasa.govimagesimagerecords7800078349
arctic vir 2012147 lrg.jpg", "Elements"
Out[
]=
BitDepth, CameraTopOrientation, Channels, ColorMap, ColorProfileData, ColorSpace,
Comments, Data, DateTime, Exif , FlashUsed, GeoPosition, GPSDateTime,
Graphics, Image, ImageSize, IPTC, MakerNote, MetaInformation, RawData,
RawExif , RawIPTC, RawXMP, RedEyeCorrection, Summary, Thumbnail, XMP
In[
]:= Import
"https:eoimages.gsfc.nasa.govimagesimagerecords7800078349
arctic vir 2012147 lrg.jpg", "ImageSize"
Out[
]=
1500, 1500

From the URL address of a website we can limit the information selected to only
what we are interested in. A good practice is to first open the URL in a browser and
see what information suits our needs. After importing the web page, you will notice
that it will appear in Mathematica as a list of lists. Analyze how the information is
5. Data Analysis and Manipulation 157

structured. You need to find out what lists contain the information you want and
then use

... to extract it. Alternatively, you can also use the functions:
URLExecute or URLDownload.
Importing from databases is a frequent data import operation. For example:
importing from an Access mdb file.

We import the complete structure of the database:
In[ ]:= Import
"ExampleDatabuildings.mdb", "Datasets", "Buildings", "Labels" 

Out[
]=
Rank, Name, City, Country, Year, Stories, Height

We use Short or Shallow to display just a few data elements (this is practical if there’s
only one table and it’s not very big).
In[
]:= Import
"ExampleDatabuildings.mdb"   Short
Out[
]//Short=
1

We import the records associated to several fields and display them in a table format:
In[
]:= Text
Style
Grid
Transpose
Import

"ExampleDatabuildings.mdb", "Datasets", "Buildings", "LabeledData",
"Rank", "Name", "City", "Height"  , Frame  All , Small

1 Taipei 101 Taipei 508

2 Petronas Tower 1 Kuala Lumpur 452
3 Petronas Tower 2 Kuala Lumpur 452
4 Sears Tower Chicago 442
5 Jin Mao Building Shanghai 421
6 Two International Finance Centre Hong Kong 415
7 CITIC Plaza Guangzhou 391
8 Shun Hing Square Shenzhen 384
9 Empire State Building New York 381
10 Central Plaza Hong Kong 374
Out[
]=
11 Bank of China Hong Kong 367
12 Emirates Tower One Dubai 355
13 Tuntex Sky Tower Kaohsiung 348
14 Aon Centre Chicago 346
15 The Center Hong Kong 346
16 John Hancock Center Chicago 344
17 Shimao International Plaza Shanghai 333
18 Minsheng Bank Building Wuhan 331
19 Ryugyong Hotel Pyongyang 330
20 Q1 Gold Coast 323

For big databases (ORACLE, SQLServer, etc.), with a large number of tables and
thousands of records, it’s better to access them with an ODBC connection as we
will show later.

5.1.2 Semantic Import

The function SemanticImport detects and interprets the type of imported objects.

Here is an example of importing CSV data as a DataSet using SemanticImport. The original
source is the United Nations Sustainable Development Goals (UN SDG) Database available
at https://unstats.un.org/sdgs/indicators/database. The data were originally stored in the
Wolfram Cloud by Maureen Baehr and Ben Kicker.
158 Mathematica beyond mathematics

In[
]:= poverty  SemanticImport

"https:www.wolframcloud.comobjben.kickertsdgdata.csv";

Since the dataset is quite large, we just display its first entry:
In[ ]:= poverty
1

We extract the element in the third position. Notice that with Dataset we use [...] instead of
[[...]].
In[
]:= poverty
1, 3

Out[
]= World

If you hover the mouse over World, the following message will appear on the screen:
Entity
“Country”, “World”. Remember that in the WL the object type entities
(Entity ) have associated additional information that you can access directly, e.g., the
world population (Remember that % refers to the last output).
In[
]:=
"Population"

Out[
]= 7 794 798 729 people

The function below shows the fields {“SeriesDescription”, “Value”, “Time_Detail”} for
the World entity. Notice that Select[#”GeoAreaName” “World”&] would not work since
we are referring to the entity “World” and not to the word “World”.
In[
]:= povertyworld 
poverty
Select
"GeoAreaName"  Entity
"Country", "World" &,
"SeriesDescription", "Value", "Time Detail"

We now group the selected data by years (“Time_Detail”) and display only the information
of the related year by clicking on it:
In[
]:= GroupBy
povertyworld, "Time Detail"
5. Data Analysis and Manipulation 159

5.1.3 Export

To export a file use the function Export
"pathfile.ext", expr, format where

file is the name of the file that will be created from the expression expr.
Normally the file will be exported in the format specified by *.ext, for example:
file.xls will create an Excel file. Additionally, the format can also be specified
with the third argument of the function (format). You can even export only parts
of the data using the optional fourth argument (elems):
Export
"pathfile.ext", expr, format, elems.

The variety of export formats is quite extensive: $ExportFormats. We can even

export in compression formats such as zip.
As it is the case when importing, if we don’t specify the export path, the files will
be copied to our working directory (Data in this case).

Generate a 3D graph:
In[
]:= gr  BarChart3D
RandomInteger
7, 4, RandomInteger
 7, 1, 4,
ChartElementFunction  "Cone"

Out[
]=

Export the graph as a file named “cones.tif” in TIFF format.

160 Mathematica beyond mathematics

In[
]:= Export
"cones.tif", gr, "TIFF";

You can then import the image previously exported with: Import
“cones.tif”.
We are going to show next how to export an animation.

First, we create an animation using Manipulate. In later chapters, we will see this
interesting function in more detail.
In[
]:= example3DAnimation  Manipulate

Plot3D
Cos
k x y, Sin
k x y, x, 0, Pi,
y, 0, Pi, Filling  Automatic, PlotRange  1, k, 0, 2

Then, we export it to a video format, in this case MP4. (Normally this operation may last a
few seconds).
In[
]:= Export
"example3DAnimation.mp4", example3DAnimation

Out[
]= example3DAnimation.mp4

Now we can either manually locate the exported file and open it or call it directly from
Mathematica and have the system open it using the default application for MP4 files.
In[
]:= SystemOpen
"example3DAnimation.mp4"

5.1.4 Metadata
As we have previously seen, data from image files usually carry additional
information about the images themselves. In this case, we download a file in the
ARC Grid format, used in topography for representing terrain elevation.

We import the following ARCGrid file.
5. Data Analysis and Manipulation 161

In[
]:= Import
"http:exampledata.wolfram.comArcGRID.zip", "ARCGrid"

Out[
]=

Notice that this is a compressed file. We can decompress it to find out what files are
included.
In[ ]:= Import
"http:exampledata.wolfram.comArcGRID.zip", "ZIP"

Out[
]=
NED 93751926\.DS Store, NED 93751926\info\arc.dir, NED 93751926\info\arc0000.dat,

NED 93751926\info\arc0000.nit, NED 93751926\info\arc0001.dat,
NED 93751926\info\arc0001.nit, NED 93751926\meta1.html,
NED 93751926\Metadata.xml, NED 93751926\ned 93751926\dblbnd.adf ,
NED 93751926\ned 93751926\hdr.adf , NED 93751926\ned 93751926\prj.adf ,
NED 93751926\ned 93751926\sta.adf , NED 93751926\ned 93751926\w001001.adf ,
NED 93751926\ned 93751926\w001001x.adf , NED 93751926\NED 93751926.aux

Next, we represent the terrain elevations in 3D (we use MaxPlotPoints100 to speed up the
computations).
In[
]:= ListPlot3D
Import
"http:exampledata.wolfram.comArcGRID.zip",
"ArcGRID", "Data", MaxPlotPoints  100,
ColorFunction  "SouthwestColors", Mesh  None

Out[
]=

Let’s take a look at the metadata inside.

162 Mathematica beyond mathematics

In[
]:= Import
"http:exampledata.wolfram.comArcGRID.zip",

"ArcGRID", "Elements"
Out[
]=
Centering, CentralScaleFactor, CoordinateSystem, CoordinateSystemInformation, Data,
DataFormat, Datum, ElevationRange, Graphics, GridOrigin, Image, InverseFlattening,
LinearUnits, Projection, ProjectionName, RasterSize, ReferenceModel, ReliefImage,
SemimajorAxis, SemiminorAxis, SpatialRange, SpatialResolution, StandardParallels 

Now, we extract the information related to the coordinates system.
In[
]:= Import
"http:exampledata.wolfram.comArcGRID.zip",
"ArcGRID", "CoordinateSystemInformation"
Out[
]= GEOGCS 
NAD83, DATUM 
North American Datum 1983,
SPHEROID 
GRS 1980, 6 378 137, 298.257, AUTHORITY 
EPSG, 7019,
TOWGS84 
0, 0, 0, 0, 0, 0, 0, AUTHORITY 
EPSG, 6269,
PRIMEM 
Greenwich, 0, AUTHORITY 
EPSG, 8901,
UNIT 
degree, 0.0174533, AUTHORITY 
EPSG, 9108,
AXIS 
Lat, NORTH, AXIS 
Long, EAST, AUTHORITY 
EPSG, 4269

The following function enables us to create a link to http://maps.google.com using the
latitude and the longitude.
In[
]:= CoordinatesToMapLink
lat , long  : Hyperlink

StringJoin
"http:maps.google.commaps?q ",
ToString  lat, " ", ToString  long, "&z12&th"


We can now get the coordinates of the previous file and set up a link to Google Maps.
In[
]:= coords  Import
"http:exampledata.wolfram.comArcGRID.zip",
"ArcGRID", "SpatialRange"
40.0617 40.1447
Out[
]=
88.3158 88.1619

Since we are establishing the link with the center of the imported grid, we need to calculate
the average of the latitude and longitude.
In[
]:= CoordinatesToMapLink
QuantityMagnitude
Map
Mean, coords

Out[
]= http:maps.google.commaps?q 40.1032 88.2389&z 12&t h

5.2 Statistical Analysis

Currently, given the ease with which we can access very large amounts of data,
managing and extracting insights from them can become quite challenging. Maybe
we as humans are not well equipped to handle naturally more than 7 or 8 elements
at once; in a famous experiment, people didn’t notice when an extra person was
added to a group of 8 individuals acting on a stage. However, humankind has
invented statistical tools that enable us to reduce a large number of data points to a
much smaller number of values (mean, median, standard deviation, and others) to
make it easier to learn the most relevant information from the original data.
According to The Oxford Dictionary of Statistical Terms, statistics is the study of
the collection, analysis, interpretation, presentation, and organization of data.
Therefore, without data there can be no statistics. In statistical studies, we will
frequently reduce the original data, frequently called population, to a reduced
5. Data Analysis and Manipulation 163

number of selected representative values, also known as samples. Their

visualization will be very helpful.

5.2.1 Basic Statistical Functions

When working with big volumes of data, it’s very common to perform a
preliminary exploratory data analysis. In this section, we are going to see how to do
it in Mathematica:
Basic Statistics:
https://reference.wolfram.com/language/tutorial/NumericalOperationsOnData.html

We use CityData to obtain the distribution of the US population by city. We include
DeleteMissing to eliminate those cases where the information is not complete.

  is used to show only the population figure:
In[
]:= usacities  QuantityMagnitude
DeleteMissing

CityData
, "Population" &  CityData
All, "USA"  N;
The previous function is equivalent to Map[funcarg,  &, {city1 , ..., cityn }], where the list
of cities is generated with: CityData[
All, "USA"]. We use Shallow to show a small
sample of the total output.
In[
]:= usacities  Shallow
Out[
]//Shallow=
8.55041  106 , 3.97188  106 , 2.72055  106 , 2.29622  106 , 1.56744  106 ,
1.56303  106 , 1.46985  106 , 1.39493  106 , 1.30009  106 , 1.02691  106 , 32 722

We can find out the number of cities in the list:

In[
]:= Length
usacities

Out[
]= 32 732

We build a simple frequency table using BinCounts
x1 , x2 , , b1 , b2 ,  to
count the number of xi in the intervals
b1 , b2 ,
b2 , b3 , ..
In[
]:= TableForm

"0 102 ", "102 103 ", "103 104 ",
"104 105 ", "105 106 ", "106 107 ", BinCounts
usacities,
 0, 100, 1000, 10 000, 100 000, 1 000 000, 10 000 000
Out[
]//TableForm=

0 102 102 103 103 104 104 105 105 106 106 107
2310 13 274 12 795 4039 304 10

We calculate the most common descriptive statistics:
In[
]:= Mean
, StandardDeviation
, Skewness
, Kurtosis
,
Quantile
, .6, InterquartileRange
 &
usacities
Out[
]=
7773.91, 65 287.1, 81.8769, 9560.86, 1775., 3718.5

Generate a histogram for a list of values using a log scale.

164 Mathematica beyond mathematics

In[
]:= Histogram
usacities, "Log", "LogCount", Frame  True

104

1000

100
Out[
]=

10

1
100 105

5.2.2 Graphical Representations

Graphic representations are very important in statistic analysis. We can create
visualizations from custom data.

In the following example, the word size increases based on its frequency of appearance in a
text. In this case, we use the first 300 words of the Declaration of Independence included in
ExampleData .
In[
]:= Text
Row
With
data  ExampleData

"Text", "DeclarationOfIndependence", "Words"1 ;; 200,
MapIndexed
Style
, 2 Count
Take
data, First
2,  &, data, " "

Out[ ]= When in the Course of human events, it becomes necessary for one people to dissolve the political bands which have connected them with another, and to assume, among the Powers of the earth, the separate and equal station to which the Laws of Nature and of Nature's God entitle them, a decent respect to the
opinions of mankind requires that they should declare the causes which impel them to the separation. We hold these truths to be self
evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these

are Life, Liberty, and the pursuit of Happiness.

That to secure these rights, Governments are instituted among Men, deriving their just powers from the consent of the governed,

That whenever any Form of Government

becomes destructive of these ends, it is the Right of the People to to alter or abolish it, and to institute new Government, laying its foundation on such principles and organizing its powers in

such form, as to them shall seem most likely to effect their Safety and Happiness. Prudence, indeed, will dictate that Governments long established should not be changed for light and transient causes; and accordingly all experience

The functions below show how to create a bar chart of the frequency of characters:
In[
]:= letters  LetterCounts

ExampleData
"Text", "DeclarationOfIndependence", IgnoreCase  True
Out[
]= e  928, t  686, o  578, a  540, n  537, s  519, i  495, r  489,
h  393, l  278, d  266, u  218, c  204, f  190, m  178, p  151,
g  150, w  117, b  105, y  91, v  76, j  33, k  21, x  10, q  6, z  4
5. Data Analysis and Manipulation 165

In[
]:= BarChart
letters, ChartLabels  Keys
letters

800

600

Out[
]=
400

200

0
e t o a n s i r h l d u c f m p g w b y v j k x q z

5.2.3 Global Warming

You can access a huge collection of computable datasets from the Wolfram Data
Repository:
https://datarepository.wolframcloud.com/
An introduction to the resource with examples of how it can be used is available at:
https://blog.wolfram.com/2017/04/20/launching-the-wolfram-data-repository-data-publishing-
that-really-works/

To retrieve a data collection from the data repository you can use    name.
Among the available curated data is the evolution of the global mean temperature compared
to the mean of the period 1951 — 1980. In the example below, we download the Global-
mean temperatures dataset (click on + or » to obtain additional information). A detailed
description, including information about all the references used, can be found at:
https://data.giss.nasa.gov/gistemp/.
In[ ]:= ResourceObject
"NASA GISTEMP Global Means dTs"

Out[
]= ResourceObject

Name: NASA GISTEMP Global Means dTs »

Type: DataResource
Description: Global
mean monthly, seasonal, and annual temperatures dTs since 1880

The function below imports the dataset using   and sets the number of items to
be displayed at once to 10. You can use the mouse to scroll through the data.
In[ ]:= Dataset
ResourceData
"NASA GISTEMP Global Means dTs", MaxItems  10
166 Mathematica beyond mathematics

To extract the contents of specific columns, we use the following syntax: data[All,
{Column Names}], in this case we choose “Year” and “JanToDec”.
In[
]:= data  ResourceData
"NASA GISTEMP Global Means dTs"

All, "Year", "JanToDec";

The values are the deviations with respect to the “normal” , where “normal” always means
the average over the 30-year period 1951 — 1980 for that place and time of the year. The
years 1951 and 1980 correspond to rows 72 and 101, respectively. The main statistics for
that period are shown below. Remember that the use of the syntax f1
, ...,
fn
&
data1 is equivalent to f1
data1, ..., fn
data1
In[ ]:= Mean
, Median
, StandardDeviation
, Max
, Min
  &

data
72 ;; 101, "JanToDec"
Out[ ]=
0.000333333 C, 0.01 C, 0.109686 C, 0.27 C, 0.2 C
The mean should be zero; the small difference is due to rounding errors. The median is very
close to zero, and the maximum and minimum figures indicate that during those 30 years,
there was no clear trend (their absolute values are quite similar). However, when we analyze
all the available data, we can see that starting in the 1970’s the mean global temperature
displays a clear upward trend.
In[
]:= DateListPlot
data, PlotTheme  "Web", AxesOrigin  Automatic, 0

0.75

0.50

0.25
Out[
]=

0.25

0.50
1900 1925 1950 1975 2000

To insert the label Min in the graph, copy Min to the clipboard and double click inside the
plot. Then, right-click and choose “Paste into Graphic”. Finally, move the label to the
appropriate location.
5. Data Analysis and Manipulation 167

5.2.4 Analyzing the Weather Data of Your Location

If you want to know your local temperature, just type in an Inline Free-format Input (or Ctrl
+ = in a US keyboard) : . Then, notice that the function used to get the
temperature is AirTemperatureData .
In[
]:= AirTemperatureData


Out[
]= 18. C

In[
]:= IconData
"AirTemperature", 

Out[
]=
18C

By default, AirTemperatureData returns the most recent measurement for air

temperature near your location (actually near the location associated with the IP address
through which your computer connects to the Internet, which can sometimes differ greatly
from your real location). You can use the same function to find out from what weather
station you got the measurement:
In[
]:= AirTemperatureData
"StationName"

Out[
]= LESA

We can then analyze the temperature measurements for this weather station. The function
below returns all the data available for December 14, 2020 measured at 30-minutes intervals.
In[
]:= AirTemperatureData
"LESA",
DateObject
2020, 12, 14, DateObject
2020, 12, 15

Time: 14 Dec 2020 00:00:00 to 15 Dec 2020 00:00:00

Out[
]= TimeSeries
Data points: 49

The output is given as a time series, a format that is very useful if you want to perform
additional statistical calculations. However, we can also display the actual values using
Normal .   is used in this case to show only 3 random temperature readings.
In[ ]:= RandomChoice
Normal
, 3

      ! 9. C

Out[
]=      " ! 9. C

      ! 12.6 C

To compute the mean temperature in a circle with a 200-km radius centered around the
weather station, we first obtain LESA’s coordinates and then use GeoDisk.
In[
]:= AirTemperatureData
"LESA", "Coordinates"1

Out[
]=
40.959, 5.498

In[
]:= AirTemperatureData
GeoDisk
, Quantity
200, "km", DateObject
, Mean

Out[
]= 18.4 C
168 Mathematica beyond mathematics

GeoNearest can be used to find out the name of the nearest weather station.

In[
]:= GeoNearest "WeatherStation", Here , 2

Out[
]=  LESA , LEVD 

In[
]:= GeoListPlot , GeoRange  Spain COUNTRY , GeoLabels  True

Out[
]=

Let’s compare now the temperatures in both weather stations.

We analyze the mean monthly temperatures for 120 months starting from January 2010.
QuantityMagnitude is used to show only the amount not the unit (ºC o ºF). The computation
may take a few minutes because the data are imported from an external source.
In[
]:= data1, data2  QuantityMagnitude

AirTemperatureData
"LESA", DateObject
2010, , Mean & 
Range
120 , AirTemperatureData
"LEVD",
DateObject
2010, , Mean &  Range
120 ;

We calculate the covariance and correlation between the two lists created previously.
In[ ]:= Covariance
data1, data2, Correlation
data1, data2

Out[
]=
46.2391, 0.990725

We visualize data1 and data2 using Histogram. If you position the cursor inside any of
the bars, its value will be displayed.
In[
]:= Histogram
data1, data2, ChartLegends  "LESA", "LEVD"
35
30
25
20 LESA
Out[
]=
15
LEVD
10
5
0
5 10 15 20 25 30

We can also visualize the data using BoxWhiskerChart . When placing the cursor on any
of the boxes, you will see the statistical information associated with it.
5. Data Analysis and Manipulation 169

In[
]:= BoxWhiskerChart
data1, data2, ChartLabels  "LESA", "LEVD"

25

20

15
Out[
]=
10

0
LESA LEVD

5.3 Probability Distributions

5.3.1 Data and Probability Distributions

Many processes in biology, economics, engineering, finance, and many other
scientific and technical fields that involve randomness can be modeled using
probability distributions. Two important concepts to master are the probability
density function (pdf) and the cumulative density function (cdf).

The Wikipedia definition:
In[
]:= TextSentences
WikipediaData
"probability density function"1 ;; 3

Out[
]=
In probability theory, a probability density function PDF, or density of a continuous random
variable, is a function whose value at any given sample or point in the sample space
the set of possible values taken by the random variable can be interpreted as providing
a relative likelihood that the value of the random variable would be close to that sample.,
In other words, while the absolute likelihood for a continuous random variable to take
on any particular value is 0 since there is an infinite set of possible values to
begin with, the value of the PDF at two different samples can be used to infer,
in any particular draw of the random variable, how much more likely it is that
the random variable would be close to one sample compared to the other sample.,
In a more precise sense, the PDF is used to specify the probability of the random variable
falling within a particular range of values, as opposed to taking on any one value.

The formal definition:

In[
]:= probability density function MATHWORLD TERM definition

The probability density function PDF P x of a continuous distribution is defined as the derivative
of the cumulative distribution function D x,
D x P xx
P x P 
Out[
]= P x,
so
D x P X  x
x
  P  d .

Different kinds of probability distribution can be used depending on the random processes
involved. The number of probability distributions (guide/ParametricStatisticalDistributions)
170 Mathematica beyond mathematics

available in Mathematica is substantially higher than the one available in other well-known
statistical programs. Additionally, Mathematica’s symbolic capabilities enable us to build our
own distributions using variations of the available ones.

This example shows the PDF and CDF of a Poisson distribution.
In[
]:= pdf  PDF
PoissonDistribution
, x
 
x
x0
Out[
]= x
0 True

In[
]:= cdf  CDF
PoissonDistribution
, x

Qx 1,  x  0
Out[
]=
0 True

Here several statistical parameters (mean, variance, skewness, and kurtosis) are computed.
In[ ]:= Mean
, Variance
, StandardDeviation
, Skewness
, Kurtosis
 &

PoissonDistribution

1 1
Out[ ]= , , , , 3
 

Looking at the output above, we can see that the Poisson distribution depends on only one
parameter: . This means that the variance is equal to the mean and therefore, the standard
deviation  =  .
The Poisson distribution is used to model the number of events x occurring within a given time
interval with the parameter  indicating the average number of events per interval. A typical
Poisson process is the radioactive disintegration.
Example: After measuring an environmental sample of uranium over a few hours, its parameter has been
determined to be  = 10 disintegrations/minute.

The PDF and CDF of measure x = {0, 1, 2, ...} disintegrations in one minute are represented
below. An option in PlotLegends is included to place the legend in the desired position.
In[
]:= DiscretePlot
Evaluate
pdf, cdf .  10, x, 0, 15, PlotRange  All,
PlotLegends  Placed
"Probability", "Cumulative Probability", Below

Out[
]=

Probability Cumulative Probability

#$ expr, x  dist (to write  you can use the palettes or the shortcut: Esc
key dist Esc key) calculates the probability of expr under the assumption that x
5. Data Analysis and Manipulation 171

follows the probability distribution dist.

The probability that there are 15 or more disintegrations in the uranium sample in a given
minute is:
In[
]:= Probability
x 15, x  PoissonDistribution
10  N

Out[
]= 0.0834585

Expectation expr, x  dist calculates the mathematical expectation of expr

assuming that x follows the probability distribution dist.

The moment of order 2 of a Poisson distribution is:
In[
]:= Expectation
x ^ 2, x  PoissonDistribution


Out[
]= 2 

The moment of order n of a Poisson distribution is:

In[
]:= Expectation
x ^ n, x  PoissonDistribution
,
Assumptions  n  Integers && n 0
Out[ ]= Bn 

BellB is-the Bell polynomial Bnx . (e.g., n = 4)

In[
]:= BellB
4, 

Out[
]= 4 6 3 7 2 

5.3.2 The Normal or Gaussian Distribution

The best-known and probably the most commonly occurring probability
distribution in nature is the Normal or Gaussian (although it was not Carl F. Gauss
who discovered it).

Using Inline Free-form Input ( ) we can get information about it.

In[
]:= normal distribution MATHWORLD TERM definition

A normal distribution in a variate X with mean  and variance 2 is a statistic distribution with
probability density function
1 x 2 2 2 
P x  e
 2
Out[
]=
on the domain x  , . While statisticians and mathematicians uniformly use the term “normal
distribution” for this distribution, physicists sometimes call it a Gaussian distribution and, because of
its curved flaring shape, social scientists refer to it as the “bell curve.” Feller uses the symbol ! x for
P x in the above equation, but then switches to " x in Feller.

One of the main features of probability distributions is that if certain experimental

data follow a specific distribution, the difference between the theoretical
distribution and the experimental one will get smaller and smaller as the number of
experimental data points gets larger and larger.

In this example, we use Histogram to display a histogram of experimental data (in this
case simulated data) following a normal probability distribution with mean 0 and standard
deviation 1, commonly written as N(0,1), and compare it with its theoretical distribution.
The experimental data are simulated using  % [dist], a Mathematica function
that generates random numbers that follow a distribution dist . Strictly speaking, we should
172 Mathematica beyond mathematics

be talking about pseudorandom numbers since they are generated by algorithms. However,
for most practical applications they can be considered random numbers. We place the
function inside Manipulate using n, the number of data points, as the parameter. We can
clearly see that the shape of the histogram gets closer to the probability density function
(PDF) of the theoretical N(0,1) as the number of simulated data points increases.

In[
]:= Manipulate

Histogram
RandomVariate
NormalDistribution
0, 1, n, Automatic, "PDF",
PlotRange   4, 4, 0, 0.45, AxesOrigin   4, 0, Epilog  First 
Plot
PDF
NormalDistribution
0, 1, x, x, 4, 4, PlotStyle  Red,
n, 1000, "Number of data points", 50, 10 000

5.3.3 Automatically Finding Probability Distributions

You can use FindDistribution to find the distribution of your data.
Example: A pacemaker has a redundant system consisting of independent units so that when a single unit
fails, the pacemaker does not need to be replaced.

The data below represents the failing times of 100 pacemakers.
In[
]:= pacemaker  13.6, 9.3, 6.5, 31., 4.2, 12.5, 15.3, 19.1, 6.5, 30.5, 77.2,
21.7, 8.1, 19.7, 8.3, 24.9, 23., 22.3, 24.8, 18.5, 24., 13.1, 2.9,
10.4, 19.1, 25.3, 12.2, 3.9, 7.3, 14., 8.2, 10.2, 8.4, 6.4, 5.9,
13.6, 25.1, 19., 7.1, 7.3, 3.3, 6.3, 24.1, 14.9, 16.1, 28.5, 17.1,
8.6, 5.3, 23.2, 27.9, 17.1, 15.8, 12.4, 5.9, 26.4, 3.4, 29.5,
28.4, 10.7, 7.5, 15.1, 11., 7.5, 9.7, 7., 59.1, 2.8, 7.4, 21.7,
10.9, 15.9, 6.4, 16., 0.3, 11.4, 20.1, 28.1, 12.7, 27.6, 10.,
5.9, 4.4, 2.7, 6.5, 30., 19.2, 7.9, 16.5, 18.4, 15.9, 6.6, 6.5;

We can now proceed to find the underlying distribution. By default, FindDistribution will
return the one that fits the data the best. However, we can change that behaviour by adding a
second argument to the function as we do in the example below. In this case, we ask the
program to show us the best two distributions:
In[
]:= pacedistribution  FindDistribution
pacemaker, 2

Out[
]=
GammaDistribution1.79873, 8.73275, WeibullDistribution1.34491, 17.1987

In[
]:= gamma, weibull  pacedistribution;

5. Data Analysis and Manipulation 173

In fact, the most commonly used distributions for modeling equipment failure are the
Gamma distribution (GammaDistribution) and the Weibull distribution
(WeibullDistribution). We can even use the function with a third argument to generate a
Dataset containing the values of all possible properties associated to them:
In[ ]:= FindDistribution
pacemaker, 2, All

The function below displays the graphs of their probability density functions along with the
histogram of the data:
In[
]:= Show
Histogram
pacemaker, Automatic, "PDF",
Plot
PDF
gamma, x, PDF
weibull, x, x, 0, 50,
Filling  Axis, PlotLegends  "Gamma PDF", "Weibull PDF"

Gamma PDF
Out[
]=
Weibull PDF

The probability that the pacemaker will still be working after 20 years is:
In[
]:= Probability
x 20, x  gamma  N

Out[
]= 0.279733

5.3.4 Mixture Distributions

In this subsection, we’re going to analyze the “OldFaithful” dataset related to
geyser emissions. Let’s see a detailed description of the file.
In[
]:= ExampleData
"Statistics", "OldFaithful", "Properties"

Out[
]=
ApplicationAreas, ColumnDescriptions, ColumnHeadings,

ColumnTypes, DataElements, DataType, Description, Dimensions, EventData,
EventSeries, LongDescription, Name, ObservationCount, Source, TimeSeries
In[
]:= ExampleData
"Statistics", "OldFaithful", "LongDescription"

Out[
]= Data on eruptions of Old Faithful geyser, October 1980. Variables are the
duration in seconds of the current eruption, and the time in minutes to the next
eruption. Collected by volunteers, and supplied by the Yellowstone National Park
Geologist. Data was not collected between approximately midnight and 6 AM.
174 Mathematica beyond mathematics

In[
]:= Transpose
ExampleData
"Statistics", "OldFaithful", "ColumnHeadings",

ExampleData
"Statistics", "OldFaithful",
"ColumnDescriptions"  TableForm
Out[ ]//TableForm=
Duration Eruption time in minutes
WaitingTime Waiting time to next eruption in minutes

We assign names to the column data and create histograms for each column:
In[
]:= OldFaithfuldata  ExampleData
"Statistics", "OldFaithful";

In[
]:= duration, waitingtime  Transpose
OldFaithfuldata;

In[
]:= GraphicsRow

Histogram
duration, 12, "PDF", ChartElementFunction  "FadingRectangle",
PlotLabel  Style
"Duration", 10, Blue, Histogram
waitingtime,
12, "PDF", ChartElementFunction  "GradientRectangle",
PlotLabel  Style
"Waiting Time", 10, Blue, ImageSize  Medium

Out[
]=

We can measure the correlation between the two variables.

In[
]:= Correlation
duration, waitingtime

Out[
]= 0.900811

They are highly correlated. To visualize it we can use PairedHistogram after rescaling the
data.
In[
]:= sclData1 
With
dat1  MinMax
 &
duration, dat2  MinMax
 &
waitingtime,
Rescale
, dat1, dat2 &  duration;
In[
]:= PairedHistogram
sclData1, waitingtime, 12, "PDF",
ChartElementFunction  "GlassRectangle", ChartLabels 
Style
"Scaled Duration", "Text", Style
"Waiting Time", "Text"

Out[ ]=

If we have two probability distributions, we can obtain a mixture distribution with

5. Data Analysis and Manipulation 175

the following function:

MixtureDistribution
p1, p2, N
m1, s1 , N
m2, s2]
where p1, p2 represent the proportion of Nm1, s1 and Nm2, s2, respectively.

Let’s build a mixture distribution using the “OldFaithful” dataset again.
In[
]:= mixDist1  EstimatedDistribution
waitingtime, MixtureDistribution
p1, p2,
NormalDistribution
m1, s1, NormalDistribution
m2, s2
Out[
]= MixtureDistribution
0.639116, 0.360884,

NormalDistribution80.091, 5.86777, NormalDistribution54.6148, 5.87117

Now let’s plot together both, the original data and the mixture distribution, approximating it:
In[ ]:= Show
Histogram
waitingtime, 25, "PDF",
Plot
PDF
mixDist1, x, x, 41, 100, PlotStyle  Red
0.05

0.04

0.03

Out[
]=

0.02

0.01

0.00
50 60 70 80 90

The new distribution can be used for statistical computations such as:

The probability that the time between two emissions will be 80 minutes or more:
In[
]:= Probability
x 80, x  mixDist1

Out[
]= 0.323516

5.3.5 Empirical and Adjusted Distributions

In this example, based on a case study from a Wolfram virtual seminar,
WeatherData is used to investigate the feasibility of using wind to generate
electricity in Salamanca (Spain).

Let’s find the name of the weather station near Salamanca first:
In[
]:= wsSalamanca  WeatherData
"Salamanca", "Spain", "StationName"

Out[
]= LESA

We can now get the average daily wind speed, in km/h, for the area beginning in 2000 and
eliminate incomplete records. We use DeleteMissing[list, 1, 2] to remove any sublists
containing missing data within the first 2 levels. We also use “Path” to extract Time-value
pairs:
176 Mathematica beyond mathematics

In[
]:= wsData 

WeatherData
wsSalamanca, "MeanWindSpeed", 2000, 1, 1, Date
, "Day"

Time: 01 Jan 2000 to 19 May 2022

Out[
]= TimeSeries
Data points: 8121

Data not in notebook. Store now

In[
]:= wData  DeleteMissing
wsData
"Path", 1, 2;

Let’s display the first 5 elements of our list:

In[
]:= Take
wData, 5
3 155 673 600 2.78 kmh
3 155 760 000 2.78 kmh
Out[
]= 3 155 846 400 2.04 kmh
3 155 932 800 2.04 kmh
3 156 019 200 2.22 kmh

The entries are ordered pairs containing the date and the corresponding average wind speed.
The figures assigned to the dates correspond to the number of seconds elapsed since January
1, 1900. The reason behind this way of representing time is that Mathematica often uses
AbsoluteTime. By default, it gives us the absolute time at the moment the function is
executed.
In[
]:= AbsoluteTime


Out[
]= 3.8622060040812979  109

We can show it in a date format:

In[
]:= DateList
AbsoluteTime


Out[
]=
2022, 5, 22, 11, 0, 4.0962878

We need to extract the wind speeds from the data. We use the Part function to get a list of
the daily average speeds, convert them to new units (meters/second), and visualize the result
with histograms.
In[
]:= wndData  1000 wDataAll, 2  602 ;
GraphicsRow
histplotpdf, histplotcdf  Histogram
wndData, 45,
"PDF", PlotLabel  Style
"Daily average speed", "Title", 12,
Histogram
wndData, 45, "CDF", PlotLabel  Style

"Accumulated daily average speed", "Title", 12, ImageSize  Large

Daily average speed Accumulated daily average

0.30 1.0

0.25
0.8
0.20
Out[
]= 0.6
0.15
0.4
0.10

0.05 0.2

0.00 0.0
2 4 6 8 2 4 6
5. Data Analysis and Manipulation 177

We can now fit the previous data to an existing distribution. However, we can also create a
new one based on the data. This last operation can be done automatically with the function
SmoothKernelDistribution.
In[
]:= e  SmoothKernelDistribution
QuantityMagnitude  wndData

Type: SmoothKernel
Out[
]= DataDistribution
Data points: 8072

While the ability to estimate a distribution is a valuable tool in statistical analysis, one must
be careful when actually using it. It is possible for estimated distributions to have very small
negative values in the tails. Estimated distributions of this sort behave in a similar manner to
interpolating functions and can show unexpected behavior.
In[
]:= Show

histplotpdf,
estplot  Plot
PDF
e , x, x, 0, 13, PlotStyle  Thick, Red

Daily average speed
0.30

0.25

0.20

Out[
]=
0.15

0.10

0.05

0.00
0 2 4 6 8

Assuming that the generator we are using for our analysis requires at least a wind speed of
3.5 m/s to generate electricity, we can use the Probability function to estimate the proportion
of time the generator will be contributing to the grid.
In[
]:= Probability
x 3.5, x  e 

Out[
]= 0.298209

An alternative method would be to estimate the distribution parameters and construct the
distribution. The graphical representation of the data suggests that the wind speed could be
approximated using a LogNormalDistribution. Zeros are eliminated because they are not
compatible with such distribution.
In[
]:= params  FindDistributionParameters

DeleteCases
QuantityMagnitude
wndData, 0, LogNormalDistribution
, 
Out[
]=
  0.836653,   0.7039

In[
]:= p  LogNormalDistribution
,  . params

Out[
]= LogNormalDistribution0.836653, 0.7039

Visualize our distribution estimate:

178 Mathematica beyond mathematics

In[
]:= GraphicsRow
Show

histplotpdf,
estplot  PlotPDFp , x , x, 0, 14, PlotStyle  Thick, Red
 , Show

histplotcdf,
estplot  PlotCDFp , x , x, 0, 14, PlotStyle  Thick, Red 


Daily average speed Accumulated daily average speed

1.0
0.30
0.25 0.8

Out[
]= 0.20 0.6

0.15
0.4
0.10
0.2
0.05
0.00 0.0
2 4 6 8 2 4 6 8

We perform a test to assess the quality of the fit:

In[
]:= tstData 
DistributionFitTestDrop
Sort
wndData, 1,  p , "HypothesisTestData" ;

In[
]:= Column
Style
tstData
"AutomaticTest", "Text",

Style
tstData
"TestConclusion", "Text", Frame  True

CramerVonMises
The null hypothesis that the data is distributed according to the
Out[
]=
QuantityDistributionLogNormalDistribution0.836653, 0.7039, kmh
is rejected at the 5 percent level based on the Cramér-von Mises test.

Get the test statistic and p-value:
In[
]:= tstData
"TestDataTable"
Statistic P-Value
Out[
]=
Cramér-von Mises 2.48879 1.03184  10
6

The test tells us that the fit is not good, but when there is a large number of data points, the
best possible test is the graphical representation of the data alongside the density function of
the estimated distribution. In that case, the fit is quite reasonable. We can recalculate the
probability of the wind speed being at least 3.5 m/s and see that the result is very close to the
one previously obtained.
In[
]:= Probabilityx 3.5, x  p

Out[
]= 0.27721

5.3.6 Automatic Probability Calculations

In the next couple of examples, we use the following functions related to
probability calculations: NProbability, Expectation, and OrderDistribution:
A person is standing on a road counting the number of cars that pass before seeing
a black one. Assuming that 10% of the cars are black, find the expected number of
5. Data Analysis and Manipulation 179

cars that will pass before a black one appears.

We use the geometric distribution (GeometricDistribution[p]) to model the number of trials
until success, given a probability of success of p. The expected number of cars will be the
mean of the GeometricDistribution with p = 0.1. It can be computed in two different ways:
In[ ]:= Mean
GeometricDistribution
0.1,
Expectation
x, x  GeometricDistribution
0.1
Out[ ]=
9., 9.

Find the probability of counting 8 or fewer cars before seeing a black one.
In[
]:= NProbability
x  9, x  GeometricDistribution
0.1

Out[
]= 0.651322

Three friends are standing on a road in different places counting the number of cars
until a black one appears. The winner will be the first one to spot a black car.
Assuming that 10% of the cars are black, find the expected number of cars that will
pass before one of the friends sees a black car.

We use OrderDistribution[
dist, n, k], where k represents the k th -order statistics
distribution for n observations from the distribution dist.
In[
]:= Expectation
x, x  OrderDistribution
GeometricDistribution
0.1, 3, 3

Out[
]= 16.9006
The output indicates the average total number of cars that the friends will count before a
black one appears. Since it’s an average, the number doesn’t have to be an integer.

What is the probability that the winner will count less than 10 cars and at least 4?
In[ ]:= Probability
Conditioned
x  10, x 4,
x  OrderDistribution
GeometricDistribution
0.1, 3, 3
Out[
]= 0.245621

There are many more statistical functions available. You may be interested in
exploring some of the latest ones, such as: CopulaDistribution ,
MarginalDistribution, and TransformedDistribution.

5.3.7 Application: An Election Tie

The example in this section refers to a real scenario that happened in Spain during
an extraordinary party congress discussing a highly controversial proposal. The
party decided to put the proposal to a secret ballot and the participants had to either
accept it or reject it. The result was a tie: 1,515 people voted Yes and 1,515 voted
No. The outcome was so unexpected that the media speculated whether the result
had been rigged, and the blogosphere fueled a debate that included the use of
statistical software about how likely a tie was in a voting process involving two
options and 3,030 participants. In our analysis, we make the assumption that all the
participants vote for either option so there are no null or blank votes even though
those kinds of votes would not affect the result.
Next, we answer the following question: What is the probability that a Yes/No vote
with 3,030 people results in a tie?
These types of processes with only two possible outcomes are known as Bernoulli
processes and can be modeled using a binomial distribution: X ~ Binomial(n, p)
180 Mathematica beyond mathematics

with n in this case being the number of voters and p the probability of a Yes vote.
The answer will depend on our previous knowledge regarding the value of p.

If we don’t know anything about p in advance, then the probability can take any value
between 0 and 1. This is equivalent to the assumption that p follows a uniform distribution
between 0 and 1. For any value of n, the solution is:
In[
]:= sol1
n  : Probability
x  n  2,
x  ParameterMixtureDistribution

BinomialDistribution
n, p, p  UniformDistribution
0, 1

With n = 3030, the probability of a tie is:
In[
]:= sol1
3030
1
Out[
]=
3031
A more direct way of calculating p is to enumerate all the different possibilities: {{3030
Yes, 0 No}, {3029 Yes, 1 No},...., {1 Yes, 3029 No},{ 0 Yes, 3030 No}}. There are a total
of 3,031 possible cases, and since we know nothing about p, we assume that all of them have
the same probability of happening. Therefore, p = 1/3,031 for any result and a tie is one of
the potential outcomes.

In the actual case, there was some previous knowledge. There had been earlier votes with
#
outcomes very close to a tie. Let’s assume that p = 0.49 with a standard deviation of 0.02.
Then we can calculate the probability assuming that p follows a Normal probability
distribution: X ~ Normal(0.49, 0.02):
In[
]:= NIntegrate
PDF
NormalDistribution
0.49, 0.02, p 
PDF
BinomialDistribution
3030, p, 1515,
p, 0, 1, WorkingPrecision  10
Out[
]= 0.005402504702

If we’d like to perform a Bayesian analysis, it’s convenient to assume that p follows a Beta
distribution with parameters $ and %, whose actual values depend on our prior knowledge of
p.
In[
]:= sol2
n ,  ,   : Probability
x  n  2,
x  ParameterMixtureDistribution

BinomialDistribution
n, p, p  BetaDistribution
, 

Without previous knowledge about p, then $ = 1 and % = 1 and the result is the same as the
one previously obtained.
In[
]:= sol2
3030, 1, 1
1
Out[
]=
3031

For the secret ballot scenario, we have some prior knowledge since the results of the
previous 3 or 4 votes had been very close to a tie. We can therefore assume that p is
approximately 1/2. This is equivalent to $ and % taking very large values. An extreme case:
In[
]:= sol2
3030, 100 000, 100 000  N

Out[
]= 0.0143853

If we knew for sure that p was exactly 1/2, this problem would be equivalent to computing
the probability that after tossing a fair coin 3,030 times, we would get the same number of
heads and tails.
5. Data Analysis and Manipulation 181

In[
]:= PDF
BinomialDistribution
3030, 1  2, 1515  N

Out[
]= 0.0144938

In summary, the probability that a tie could happen was not that small (keeping in
mind that earlier votes were close affairs). Highly unlikely events happen all the
time. An extreme example occurred shortly after the vote, also in Spain. The same
person won the biggest prize in the national lottery for two consecutive weeks. In
this game, the probability of winning the highest prize (“El Gordo”) is 1/100,000.
This means that a priori, the probability that a person purchasing just one ticket
each time will win twice in a row is 1/100,000  1/100,000 = 10 10 quite a small
likelihood. Actually, the odds of dying from lightning are higher. It’s not unusual
that with so much going on around us at any given moment, the small probability
events are the ones drawing our attention.

5.4 Exploratory Data Analysis

5.4.1 Customized Graphical Representations

This section will present several types of graphs for displaying multivariate data.
For additional information see: guide/StatisticalVisualization.

To build statistical graphs load the palette: Palettes & Chart Elements Schemes.

Many empirical data are represented by more than one random variable; we will
refer to them as multivariate data.

The function below plots the PDF Student’s t distribution for two variables.
In[
]:= Plot3D
PDF
MultivariateTDistribution
1, 1  2, 1  2, 1, 20, x, y,
x, 3, 3, y, 3, 3

Out[
]=

Some statistical studies may contain simulated instead of empirical data. In those
cases, random number generators are used. Mathematica has the functions
RandomReal and RandomInteger to generate random numbers. Since these
functions use algorithms, strictly speaking they don’t provide truly random data.
However, in most situations we can assume that they do.
182 Mathematica beyond mathematics

The next function generates random numbers from a multivariate Poisson distribution. If
you remove the “;” you will notice that the result is a list of sublists, that can be interpreted
as a matrix with three columns:
In[
]:= multidata 
RandomVariate
MultivariatePoissonDistribution
1, 2, 3, 4, 125;

We represent each sublist with a point. We include a red point to show the mean.
In[
]:= Show
ListPointPlot3D
multidata,
Graphics3D
Red, PointSize
Large, Point
Mean
multidata

Out[
]=

The next graph shows a histogram of the first two dimensions of the random data previously
generated (multidata) along with a histogram of 5,000 pairs of normally distributed random
numbers. (Remember that to extract list elements we use: Part or

...).
In[
]:= GraphicsRow
Histogram3D
multidataAll, 1 ;; 2, 5, Histogram3D

RandomReal
NormalDistribution
, 5000, 2, ImageSize  Medium

Out[
]=

An alternative way of representing the same data with the edges smoothed out is by using
SmoothHistogram3D and SmoothDensityHistogram (the change in color is associated to the
actual data value).
5. Data Analysis and Manipulation 183

In[
]:= Row
SmoothHistogram3D
multidataAll, 1 ;; 2, ImageSize  Medium,

SmoothDensityHistogram
multidataAll, 1 ;; 2, ImageSize  Medium

Out[
]=

5.4.2 Life Expectancy

Here we obtain the life expectancy in Spain, US, France, and Mexico using CountryData.
The result is displayed in which each country is represented by its flag. Use Outer to get
several properties for a set of entities.
In[
]:= Outer
f, a, b, x, y, z
f a, x f a, y f a, z
Out[
]=
f b, x f b, y f b, z
In[ ]:= data  Outer
CountryData, "Spain", "UnitedStates", "France", "Mexico",
"LifeExpectancy", "FemaleLifeExpectancy", "MaleLifeExpectancy";
In[
]:= flags  
ImageResize
CountryData
"Spain", "Flag", 50,
ImageResize
CountryData
"UnitedStates", "Flag", 50,
ImageResize
CountryData
"France", "Flag", 50,
ImageResize
CountryData
"Mexico", "Flag", 50;
184 Mathematica beyond mathematics

In[
]:= BarChart3D
data,

ChartElementFunction  "Cylinder", ChartStyle  "Pastel",
ChartLegends  "Life Expectancy\\\Cell
\" \"\",
"Female ", "Male  ",
ChartLabels  Placed
flags, .5, 0, 1.15,
Placed
"\\\Cell
\" \"\", "", "", Center,
BarSpacing  .1, .6, PlotLabel  Style
"Life expectancy",
16, FontFamily  "Times", ImageSize  Large

Life Expectancy

Out[
]= Female 

Male 

With CountryData we get not only the Life Expectancy but also the Gross Domestic
Product (GDP) per Capita and Population for all the countries in South America:
In[
]:= data1  Table
Tooltip
CountryData
c, 1 & 
"LifeExpectancy", "GDPPerCapita", "Population",
Column
CountryData
c, "Name", CountryData
c, "Population",
c, CountryData
"SouthAmerica";

The countries that don’t have available information are deleted:
In[ ]:= data2  DeleteMissing
data1, 1, 3

Out[
]=
 78.07 yr $8441.92 per person per year 45 195 777 people ,
 70.7 yr $3143.05 per person per year 11 673 029 people ,
 74.98 yr $6796.84 per person per year 212 559 409 people ,
 79.57 yr $13 231.7 per person per year 19 116 209 people ,
 76.91 yr $5332.77 per person per year 50 882 884 people ,
 77.76 yr $5600.39 per person per year 17 643 060 people ,
 77.9 yr $35 400. per person per year 3483 people ,
 77.121 yr $8300. per person per year 298 682 people ,
 71.59 yr $6955.94 per person per year 786 559 people ,
 78.13 yr $4949.75 per person per year 7 132 530 people ,
 74.96 yr $6126.87 per person per year 32 971 846 people ,
 73.57 yr $6491.14 per person per year 586 634 people ,
 78.19 yr $15 438.4 per person per year 3 473 727 people ,
 72.22 yr $16 055.6 per person per year 28 435 943 people 
5. Data Analysis and Manipulation 185

The result is displayed as a bubble chart. The diameters of the bubbles are proportional to
the population size. You can see the country name and its corresponding population by
putting the mouse over each bubble:
In[
]:= BubbleChart
data2, FrameLabel  "Life expectancy", "GDPcapita"

35 000

30 000

25 000
GDP capita

20 000

Out[
]=
15 000

10 000

5000

0
70 72 74 76 78 80
Life expectancy

5.4.3 Fisher’s Iris Data

Ronald Aylmer Fisher (1890 –1962) was one of the leading scientists of the 20th
century making major contributions to fields such as statistics, evolutionary
biology, and genetics. The next example is a multivariate analysis case study that
analyses four characteristics of three iris species. The data come from one of
Fisher’s classic articles, and they are available in the Wolfram Data Repository.
In[
]:= fisheriris  ResourceObject
"Sample Data: Fisher's Irises"

Name: Sample Data: Fisher's Irises »

Out[
]= ResourceObject
Type: DataResource 
Description: Fisher's iris data

In 1936 he introduced the Iris flower data set as an example of discriminant analysis. Please
read carefully the following sentences describing the experiment:
186 Mathematica beyond mathematics

In[
]:= TextSentences
WikipediaData
"Iris flower data set" ;; 6

Out[
]=
The Iris flower data set or Fisher's Iris data set is a multivariate data set introduced by the British
statistician and biologist Ronald Fisher in his 1936 paper The use of multiple
measurements in taxonomic problems as an example of linear discriminant analysis.,
It is sometimes called Anderson's Iris data set because Edgar Anderson collected the
data to quantify the morphologic variation of Iris flowers of three related species.,
Two of the three species were collected in the Gaspé Peninsula "all from the
same pasture, and picked on the same day and measured at
the same time by the same person with the same apparatus".,
The data set consists of 50 samples from each of three species of Iris
Iris setosa, Iris virginica and Iris versicolor.,
Four features were measured from each sample: the length and the
width of the sepals and petals, in centimeters.,
Based on the combination of these four features, Fisher developed a linear
discriminant model to distinguish the species from each other.
Fisher's paper was published in the Annals of Eugenics and includes
discussion of the contained techniques' applications to the field of Phrenology.

We use ResourceData to find out the source of the Iris data set:
In[
]:= ResourceData
fisheriris, "Source"

Out[
]= Fisher,R.A. "The use of multiple measurements in taxonomic problems" Annual Eugenics, 7, Part II,
179 188 1936; also in "Contributions to Mathematical Statistics" John Wiley, NY, 1950.
This paper is freely available online in the following site:
https://onlinelibrary.wiley.com/doi/epdf/10.1111/j.1469-1809.1936.tb02137.x

Here we show the three species of Iris. (At the time of this writing, the images were located
in positions 3 to 5. However, as Wikipedia is subject to frequent changes, if the function
below doesn’t return the correct images modify the numbers accordingly.)
In[
]:= WikipediaData
"Iris flower data set", "ImageList"3 ;; 5

In this case, we are going to do a simple exploratory data analysis (EDA). Later on
in the book, we will use the same data to show how to perform some machine
learning tasks.

Here we can see the elements in the file:
5. Data Analysis and Manipulation 187

In[
]:= fisheriris
"ContentElements"

Out[
]=
ColumnDescriptions, ColumnTypes, Content, DataType,

Dimensions, ObservationCount, RawData, Source, TrainingData, TestData

The function below retrieves the description of the columns:
In[
]:= ResourceData
fisheriris, "ColumnDescriptions"

Out[
]=
Sepal length in cm., Sepal width in cm., Petal length in cm., Petal width in cm., Species of iris 

Now, we display the data of the 50 samples from each of three species of Iris (Iris setosa,
Iris virginica, and Iris versicolor):
In[
]:= FisherIrisData  ResourceData
"Sample Data: Fisher's Irises"

To select the data only for the setosa species we proceed as follows:
In[
]:= setosadata  FisherIrisData
Select
"Species"  "setosa" &

The following table contains information about some of the basic statistics used in EDA
applied to the setosa dataset.
In[
]:= sepalpetal  "SepalLength", "SepalWidth", "PetalLength", "PetalWidth";

In[
]:= TableForm

Map
setosadata
Max, , setosadata
Min, , setosadata
Mean, ,
setosadata
Median, , setosadata
InterquartileRange, ,
setosadata
StandardDeviation,   &, sepalpetal,
TableHeadings  sepalpetal, "Max", "Min", "Mean", "Median",
"Interquartile Range", "StandardDeviation"

The same statistical information for all the species based on their sepal lengths:
In[
]:= FisherIrisData
GroupBy
"Species",
Max
 , Min
, Mean
, Median
, InterquartileRange
,
StandardDeviation
 &, "SepalLength"
188 Mathematica beyond mathematics

Here we compare the sepal length histograms of the three species:

In[
]:= FisherIrisData
GroupBy
"Species",
Histogram
, 3, 8, 0.5 &, "SepalLength"

Next, we calculate the covariances and correlations to explore the relationships between the
variables.
In[
]:= sepalpetaldata  Transpose
Normal
FisherIrisData
All,  &  sepalpetal;

In[
]:= Through
Covariance, Correlation
sepalpetaldata

Out[
]=  0.685694 cm2 , 0.042434 cm2 , 1.27432 cm2 , 0.516271 cm2 ,

 0.042434 cm2 , 0.189979 cm2 , 0.329656 cm2 , 0.121639 cm2 ,

 1.27432 cm2 , 0.329656 cm2 , 3.11628 cm2 , 1.29561 cm2 ,

 0.516271 cm2 , 0.121639 cm2 , 1.29561 cm2 , 0.581006 cm2 ,

1., 0.11757, 0.871754, 0.817941, 0.11757, 1., 0.42844, 0.366126,
0.871754, 0.42844, 1., 0.962865, 0.817941, 0.366126, 0.962865, 1.

The results indicate that sepal length is strongly correlated with both, petal length and petal
width. They also show that petal length is strongly associated to petal width. However, we
can see that the linear association between sepal length and sepal width is quite weak.
5. Data Analysis and Manipulation 189

5.5 Bootstrapping and Confidence Estimates

The following data are the total daily precipitation amounts in cm during 2020 from
the weather station LESA:
In[
]:= precipitationData  DeleteMissing

Transpose
Normal
WeatherData
"LESA", "TotalPrecipitation",
2020, 1, 1, 2020, 12, 31, "Day"2;

We can assume that these data represent the entire population. The average precipitation, in
cm/day, during that year in that meteorological station is:
In[
]:= Mean
precipitationData

Out[
]= 0.0947443

Let’s supposed we don’t know all the data for that year and instead we only have
precipitation information for just 30 random days.
We can use SeedRandom to get repeatable random values:
In[ ]:= sample1  SeedRandom
2514; RandomChoice
precipitationData, 30

Out[
]=
0.2, 0., 0., 0., 0.02, 0., 0., 0., 0., 0., 0., 0., 0.05, 0., 0.,
0.28, 0., 0.02, 0., 0.43, 0.02, 0., 0.28, 0., 0., 0., 0., 1.45, 0., 0.05

We want to estimate the daily mean precipitation during that year along with its standard
deviation based on the sample:
StandardDeviation
sample1
In[
]:= ms1, ss1  Mean
sample1, 
Length
sample1

Out[
]=
0.0933333, 0.0506176

A 68% confidence interval for the population mean is:

In[
]:= ms1 ss1, ms1 ss1

Out[
]=
0.0427157, 0.143951

We can also get directly a confidence interval around the mean as follows:
In[
]:= MeanAround
sample1

Out[
]= 0.09  0.05

This confidence interval indicates that 68.27% of the time the true population mean will be
within its bounds. We can make such a statement because according to the central limit
theorem, the mean statistic of the sample follows a normal distribution regardless of the
underlying population distribution.
In[
]:= Probability
x , x  NormalDistribution
,   N

Out[
]= 0.682689

If instead of a single weather station we have precipitation information from several stations
located in the area, even though we may not have the corresponding dates for such data, we
can still estimate the mean daily precipitation for the year. We use again SeedRandom to
make the example reproducible:
In[
]:= samples  SeedRandom
1234; RandomChoice
precipitationData, 34,
SeedRandom
7341; RandomChoice
precipitationData, 27,
SeedRandom
3211; RandomChoice
precipitationData, 27,
SeedRandom
9247; RandomChoice
precipitationData, 25,
SeedRandom
3782; RandomChoice
precipitationData, 34;
190 Mathematica beyond mathematics

We group the data to simulate a bigger single sample and find out that the mean confidence
interval gets smaller:
In[
]:= MeanAround
Flatten
samples

Out[
]= 0.077  0.026

We can obtain a very similar result if we calculate the mean of the sample means:
In[
]:= MeanAround
Mean
 &  samples

Out[
]= 0.077  0.027

In a real-world situation, we may only have a single sample from a population. In

this case, it’s very likely that we will still be able to improve our confidence
interval estimate for the population mean by using bootstrapping. This statistical
technique is similar to the one used in the example above where we had several
samples but with a fundamental difference: the successive samples are generated
from the original one and not from the actual population. Using the original sample,
we generate first a new one using sampling with replacement so that some of the
data points may not be sampled and others may be sampled more than once. We
calculate its mean and use it to approximate the mean of the population. Since this
value would not be very precise, we repeat the previous two steps many times, in
the process obtaining a new mean estimate each time. We can then construct a
distribution with all those values, a bootstrap distribution, representing a proxy for
the actual distribution of the population mean. Finally, using the bootstrap
distribution we can calculate its mean (a point estimate) and confidence intervals
around it the same way as before.

Now we put our newly acquired knowledge about bootstrapping into practice. The function
below creates a collection of 100 datasets from the data using sampling with replacement. In
theory, we could have created our own bootstrapping function, but since that would have
been time-consuming, we decided to see if someone had already created it by searching the
Wolfram Function Repository. In this case, using the keyword “bootstrap”, we got a link to
the “BootstrapStatistics” function page:
https://resources.wolframcloud.com/FunctionRepository/resources/BootstrapStatistics/
That page also contains information about the syntax of the function, and a link to the source
notebook where we can see the actual Mathematica code. Remember that to use the
functions available in the Wolfram Function Repository you need an Internet connection
since those functions are not part of Mathematica.
In[
]:= MeanAround

Flatten
ResourceFunction
"BootstrapStatistics"
sample1 , 100, Mean
Out[
]= 0.099  0.004
The accuracy of this technique depends on how representative of the population the original
sample is. The less representative it is, the less accurate the computed approximation will be.
For example, if our original sample had contained an unusually high number of rainy days,
we would have obtained an inaccurate result.

Let’s see another example. This one is related to the calculation of the probability of
catastrophic flooding by analyzing a river flow. This type of analyses often use data related
to the maximum flows registered over a certain period of time. As in previous examples, we
are going to use real data. In this case, the maximum instantaneous annual flows in the
Tormes river in Salamanca (Spain) from 1965 — 2018 (the data is not available for all the
years):
5. Data Analysis and Manipulation 191

In[
]:= flowTormes   68.2, 604.8, 139.4, 95.7, 327., 357.7, 134.3, 99.6, 300.2,
208.7, 48.6, 43.1, 128.5, 553.9, 45.4, 119.9, 63.5, 243.8,
274.8, 117., 134.4, 179.1, 29.6, 168.9, 23.3, 13.7, 510.2,
48.5, 420.6, 330.9, 576.9, 22.3, 73., 750.2, 33.6, 471.5, 138.2,
53.4, 67.6, 250.9, 216.7, 242.5, 188.8, 66.2, 43.1, 322.4;

Using bootstrapping, we can get a larger set to give us a better idea of how much this
measure varies. We also calculate its 95% confidence interval:
In[
]:= flowTormes1 
ResourceFunction
"BootstrapStatistics"
flowTormes , 10 000, Mean;
Histogram
flowTormes1

1500

1000

Out[
]=

500

0
150 200 250 300

In[
]:= Quantile
flowTormes1, 0.025, 0.975

Out[
]=
153.887, 256.615

The function to compute the maximum and minimum flows during that period is shown
below:
In[
]:= Through
 Min, Max
flowTormes1

Out[
]=
112.543, 312.302

Next, we approximate the data using a smooth kernel distribution. In statistics, kernel
density estimation (KDE) is a non-parametric way to estimate the probability density
function of a random variable. KDE is a fundamental data-smoothing problem where
inferences about the population are made, based on a finite data sample.
In[
]:=   SmoothKernelDistribution
flowTormes ;

In[
]:= bSamp  ResourceFunction
"BootstrapStatistics"
flowTormes, 250;

We now smooth over each bootstrapped sample and obtain the confidence estimates:
In[
]:= B  SmoothKernelDistribution
 &  bSamp;

Next, we calculate the confidence band from the bootstrapped examples (we choose a range
close to the minimum and maximum flows of the river):
In[
]:= rng  Range
10., 1000;
pdf  Table
PDF
i, rng, i, B ;
In[
]:= High  Table
Quantile
i, .975, i, Transpose
pdf;
Low  Table
Quantile
i, .025, i, Transpose
pdf;

Finally, we visualize the estimate of the PDF with 95% confidence bands:
192 Mathematica beyond mathematics

In[
]:= Show
Plot
PDF
, x, x, 0, 1000, PlotRange  All, AxesLabel  "m3s",
ListLinePlot
Transpose
rng, High, Transpose
rng, Low,
PlotStyle  Dashed, Gray, Filling  1  2

0.004

0.003

Out[
]= 0.002

0.001

m3 s
200 400 600 800 1000

When designing big engineering projects such as damns, bridges, or nuclear power
stations, is necessary to take into consideration events with very low probability of
happening but high negative impact. Massive floods, big earthquakes, and tsunamis
are instances of such events, and their probabilities can be calculated using extreme
value theory.
Going back to our example, studying Tormes’ annual maximum flows since 1965
would very useful when planning and designing infrastructure (e.g., road location)
along the river path to minimize the risk of catastrophic flooding. In cases like this
one, the most frequently used distributions are generalized extreme value
distributions such as the Gumbel distribution. [Gumbel, E. J. (1958). Statistics of
Extremes. Irvington, New York: Columbia University Press.]
Before showing how to analyze the Tormes’ flow using the Gumbel distribution,
there are a couple of concepts from extreme value theory that we need to explain
first: exceedance and return period.
The concept of exceedance refers to the number of times in N future observations
that we will observe a more extreme observation m than a certain threshold

(determined from n historical observations). That number can be modelled as a
continuous random variate following a distribution w n, m, N .
Regarding the second concept, one of the most important objectives of frequency
analysis is to calculate the recurrence interval of events, a.k.a the return period. If
the variable (x), equal to or greater than an event of magnitude
 , occurs once in T
years, then the probability of occurrence P(X  x) in a given year of the variable is:
1
Px  xT 
T
1
T
1 Px  xT 

Based on the previous two equations, we can now defined the return period as the
average length of time between events of the same magnitude or greater. This can
also be interpreted as the number of observations such that, on average, there is one
5. Data Analysis and Manipulation 193

observation equalling or exceeding x. For example, we may be interested in

estimating the maximum annual rainfall that can be expected to occur every 100
years.
Now that we have defined the concepts of exceedance and return period, we can
see how it can be applied in our example, example that has actually been possible
thanks to Jim Baldwin’s answer in https://community.wolfram.com/groups/-
/m/t/2187431
The calculation of the mean. the standard deviation and other properties of a Gumbel distribu-
tion: Gumbel, E. J. (1954). Statistical Theory of Extreme Values and Some Practical Applica-
tions. A Series of Lectures. Bureau of Standards, Applied Mathematics Division.
To calculate the maximum expected flow x(T) for a return period T we use the
following equation:
T
LogLog' ( yn
# T 1
xT  x X 5.2
n

where: x# = mean and X = standard deviation of the variate X, in our case the river
flow series.

That is:
In[
]:= xmean, sd, size 
Mean
, StandardDeviation
, Length
 &
flowTormes
Out[
]=
203.274, 181.765, 46

The values for yn and n are estimated as follows:

In[
]:= Y
n  : Mean
Table
Log
Log
m  n 1, m, 1, n

In[
]:= S
n  : StandardDeviation

Table
Log
Log
m  n 1, m, 1, n Sqrt
n 1  n

The final step is the application of (5.1) to calculate the expected maximum flows for
different return periods:
T sd
In[ ]:= TableForm Table T, xmean Log Log Y
size ,
T 1 S
size
T, 100, 1000, 100 , TableHeadings 

None, "Return Period\nyears", "Maximun Flow\nm3s"  N

Out[
]//TableForm=

Return Period Maximun Flow

years m3s
100. 1014.15
200. 1123.75
300. 1187.76
400. 1233.15
500. 1268.34
600. 1297.09
700. 1321.4
800. 1342.45
900. 1361.02
1000. 1377.62
194 Mathematica beyond mathematics

5.6 Curve Fitting

When performing experimental studies, it’s common to fit the results to a model.
Mathematica has several functions for doing that: FindFit, FindFormula, Fit,
FittedModel, GeneralizedLinearModelFit, LinearModelFit, NonlinearModelFit,
and TimeSeriesModelFit. For more details, please see:
https://reference.wolfram.com/language/guide/StatisticalModelAnalysis.html.
In[
]:= ClearAll
"Global`"Clear all symbols as usual 

In this section, we are going to see some of those functions through basic examples.
We use two files containing the data we are going to fit: noisydata.xls and
pulsar1257.dat. These two files must be in a subdirectory, named “Data”, inside the
directory where the notebook is located.

Remember that in the first section of this chapter, we discussed how to create such a folder
and make it the working directory. In case you haven’t done so yet, just execute the
following function after creating the “Data” subdirectory in your computer:
In[
]:= SetDirectory
FileNameJoin
NotebookDirectory
, "Data"

Out[
]= F:\Mi unidad\Mathematica\MBN13\Data

5.6.1 A Linear Regression Model

We are going to fit the data in noisydata.xls to a third-degree polynomial using
LinearModelFit, the Mathematica function most commonly used when fitting
linear regression models and nonlinear ones such as polynomials. This function
will create a linear model of the form y# %0 %1 f1 %2 f2  (even though the
model is linear, the fi don’t need to be linear functions; they can be, for example,
polynomials.)
The assumptions required for this type of fitting to be valid are that the original
data yi must be independent and normally distributed, with mean y# i and a constant
standard deviation s.

We import the data from the xls file.
In[ ]:= data  Import
"noisydata.xls", "Data", 1;
dataplot  ListPlot
data
150

100

50

Out[
]=

5 5 10 15

50

100

150

We fit a third-degree polynomial.

In[
]:= fit  LinearModelFitdata,
1, x, x2 , x3 , x ;
5. Data Analysis and Manipulation 195

This is the ANOVA table:

In[
]:= fit
"ANOVATable"
DF SS MS F-Statistic P-Value

1 165 320. 165 320. 1626.9 3.47061  10
32

2
Out[
]=

1 58 481.7 58 481.7 575.515 3.78263  10
24
3

1 114 956. 114 956. 1131.27 2.41932  10
29
Error 37 3759.8 101.616
Total 40 342 517.

Now we graphically show both the data and the fitted function.
In[
]:= Show
dataplot,
Plot
fit
x, x, 5, 15, PlotStyle  Red

150

100

50

Out[
]=

5 5 10 15

50

100

150

5.6.2 A Nonlinear Regression Model

In this example, we fit the experimental light measurement data collected over a
three-year period from pulsar PSR1257+12 (originally obtained by Alex Wolszczan
and available in the file pulsar1257.dat) to the following function:
f t = ) + % cos(t *) + + cos(t ,) + $ sin(t *) + - sin(t ,).
For nonlinear fit we use NonlinearModelFit.

We import the file:
In[
]:= data  Import
"pulsar1257.dat";

Represent it graphically:
1
In[
]:= dataplot  ListPlot data, AspectRatio  , PlotStyle  Red
4
3
2
1
Out[
]=
1.0 1.5 2.0 2.5 3.0 3.5

1

2

We fit the function f t previously mentioned:

196 Mathematica beyond mathematics

In[
]:= pulsarfit 

NonlinearModelFit
data,  Sin
 t  Cos
 t  Sin
 t  Cos
 t ,
, 1, , 1, , 1, , 0, , 23.31, , 34.64, , 0, t

Out[
]= FittedModel 1.33261 cos23.3869    0.209533 cos34.5111    19 sin18 
1.29803 sin34.5111    0.0769581

We put the data and the function on the same graph to check the quality of the fit:
In[ ]:= Show
Plot
pulsarfit
t, t, 0.68, 3.76,
dataplot
3

Out[
]=
1.0 1.5 2.0 2.5 3.0 3.5

1

2

Once the model has been created, we can perform all kinds of statistical analyses. However,
it would be useful to list all the properties available for the model:
In[
]:= pulsarfit
"Properties"  Shallow
Out[
]//Shallow=

AdjustedRSquared, AIC, AICc, ANOVATable,
ANOVATableDegreesOfFreedom, ANOVATableEntries, ANOVATableMeanSquares,
ANOVATableSumsOfSquares, BestFit, BestFitParameters, 40

For example, here we display the residuals:
In[
]:= ListPlot
pulsarfit
"FitResiduals", GridLines  None, Automatic
0.06

0.04

0.02

Out[
]=

50 100 150

0.02

0.04
5. Data Analysis and Manipulation 197

5.6.3 Analysis of the Evolution of Two Cell Populations

In this subsection, we are going to analyze the evolution over time of two cell
populations {time (in hours), density (thousands of cells per square millimeter) of
cells type 1, and type 2}
ClearAll
"Global`"

cells  "time", "Population 1", "Population 2",

0.5`, 0.7`, 0.1`, 1.`, 1.3`, 0.2`, 1.5`, 1.9`, 0.3`,
2.`, 2.3`, 0.4`, 2.5`, 2.7`, 0.5`, 3.`, 3.`, 0.6`,
3.5`, 3.3`, 0.7`, 4.`, 3.6`, 0.7`, 4.5`, 3.8`, 0.8`,
5.`, 4.`, 0.8`, 5.5`, 4.2`, 0.9`, 6.`, 4.3`, 1.`,
6.5`, 4.4`, 1.`, 7.`, 4.5`, 1.1`, 7.5`, 4.6`, 1.1`,
8.`, 4.6`, 1.2`, 8.5`, 4.7`, 1.2`, 9.`, 4.8`, 1.3`;

We remove the headings and reorganize the data to get two separate lists representing the
evolution of the two cell populations.
In[
]:= data  Drop
cells, 1;

In[
]:= ti, p1, p2  Transpose
data;

In[
]:= population1, population2  Transpose
ti, p1, Transpose
ti, p2

0.5, 0.7
1., 1.3
1.5, 1.9
2., 2.3
2.5, 2.7
3., 3.
3.5, 3.3
4., 3.6
4.5, 3.8
5., 4.
Out[
]=

0.5, 0.1
1., 0.2
1.5, 0.3
2., 0.4
2.5, 0.5
3., 0.6
3.5, 0.7
4., 0.7
4.5, 0.8
5., 0.8

We plot the evolution of both populations.

In[
]:= ListPlot
population1, population2, Joined  True,
PlotLabel  Style
"Cell Populations Evolution", Bold
Cell Populations Evolution
5

Out[
]=

2 4 6 8

The shape of the curves suggests that they could be approximated by functions of the form:
a b Exp k t. We use the function FindFit to find the best-fit parameters for the fitting
functions.
In[
]:= fitpopulation1  FindFit
population1, a1 b1 Exp
k1  t, a1, b1, k1, t

Out[
]=
a1  5.11361, b1  5.14107, k1  0.303326

198 Mathematica beyond mathematics

In[
]:= fitpopulation2  FindFit
population2, a2 b2 Exp
k2  t, a2, b2, k2, t

Out[
]=
a2  1.9681, b2  1.9703, k2  0.114289

The next example, showing how to customize a graph, includes options that we
haven’t covered yet (we will explain them later in the book). As usual, if any of the
instructions are not clear, consult Mathematica help by selecting the function and
pressing F1. You can use this code as a template for similar cases. Alternatively,
you might find it easier to create the labels and plot legends with the drawing tools
(.+D).

We display the fitted curves along with the original data after having replaced the symbolic
parameters in the curves with the values obtained from FindFit. We also customize the
size of the points representing the experimental data.
In[
]:= fitsol1  a1 b1 Exp
k1  t . fitpopulation1;
fitsol2  a2 b2 Exp
k2  t . fitpopulation2;
Plot
fitsol1, fitsol2, t, 0, 10, AxesLabel  "thours",
"Concentration", PlotLegends  Placed
"Population 1", "Population 2",
Center, Epilog  Hue
0.3, PointSize
0.02, Map
Point,
population1  , Hue
0, PointSize
0.02, Map
Point, population2,
PlotLabel  Style
Framed
"Evolution of 2 cell populations",
12, Blue, Background  Lighter
Cyan

Evolution of 2 cell populations

Concentration
5

Out[
]= 3
Population 1

Population 2
2

thours
2 4 6 8 10

5.6.4 Global Primary Energy Consumption by Source

In the next example, we are going to use what we have learned so far to analyze the
evolution of energy consumption globally.
In[
]:= ClearAll
"Global`"

The following function will take you to a website that contains the file with historical
production data for different types of energy by country and region.
In[
]:= SystemOpen

"http:www.bp.comenglobalcorporateenergy economicsstatistical
review of world energy.html"
Download it by clicking on “Statistical Review of World Energy - all data, 1965-2021” and
save it without changing its name in the subdirectory Data. The file that we are going to use
corresponds to the 2016 (the last one available when this was written) report containing data
5. Data Analysis and Manipulation 199

until the end 2021 (Report 2022).

We are going to conduct several studies using this XLSX file. In principle, we could
modify directly the data using Microsoft Excel and create a new workbook with
only the required information for our analyses. This method would probably be the
most suitable one, but in this case, for educational purposes, we will perform all the
required transformations in Mathematica. This will help us to have a better
understanding of the file manipulation capabilities of the program, although it will
require that we pay careful attention to the Mathematica syntax.
The file contains many sheets. Before importing it, open the spreadsheet in Excel or
any other similar program and examine its contents. Note that each worksheet has a
label describing its contents. This is normally very useful.

Before importing an Excel file containing several worksheets, it’s better to

rename each sheet using a descriptive label replacing the ones used by default:
Sheet1, Sheet2, ... .

With the help of the following functions, we are going to import part of the file:

We can see the labels for the worksheets by using the argument "Sheets".
In[
]:= Import

"https:www.bp.comcontentdambpbusiness sitesenglobalcorporate
xlsxenergy economicsstatistical reviewbp stats review 2022 all
data.xlsx", "Sheets"   Short
Out[
]//Short=
Contents, Primary Energy Consumption,
Primary Energy  Cons by fuel, 78, Definitions, Methodology

We analyze the global consumption of primary energy (including all types:

electricity, transportation, etc.) in 2021, from different sources (oil, gas,
hydropower, nuclear, etc.)

From the sheet labels we conclude that the information we want is in the sheet “Primary
Energy - Cons by fuel”.
In[
]:= primaryenergyDatabyFuel  Import

"https:www.bp.comcontentdambpbusiness sitesenglobalcorporate
xlsxenergy economicsstatistical reviewbp stats review 2022 all
data.xlsx",
"Sheets", "Primary Energy Cons by fuel";

We use TableView to see the imported data. You can either increase the number of visible
rows with ImageSize or explore the data like in a spreadsheet, using the slide bars on the
right and at the bottom to move around the sheet and clicking on specific cells.
In[
]:= TableView
primaryenergyDatabyFuel
The output is not shown

Notice that the information we want is in row 3 (energy types) and row 91 (the total global
consumption, in Exajoules).
In[
]:= energytype1, totalcons  
 primaryenergyDatabyFuel3, primaryenergyDatabyFuel91
200 Mathematica beyond mathematics

Out[
]= Exajoules, Oil, Natural Gas, Coal, Nuclear energy,

Hydro electric, Renew ables, Total, Oil, Natural Gas,
Coal, Nuclear energy, Hydro electric, Renew ables, Total,
Total World, 174.171, 138.441, 151.07, 24.4415, 41.0898, 34.7991,
564.012, 184.214, 145.349, 160.104, 25.3128, 40.2596, 39.9126, 595.151
In[
]:= energytype, totalcons2021  energytype19 ;; 14, totalcons9 ;; 14

Out[
]= Oil, Natural Gas, Coal, Nuclear energy, Hydro electric, Renew ables,
184.214, 145.349, 160.104, 25.3128, 40.2596, 39.9126

We now visualize the data using a pie chart showing the percentage share of the total for
each energy source. An interesting observation is the fact that fossil fuels represent the vast
majority of the world primary energy consumption with renewables (excluding
hydroelectric) playing a very small role (wind, sun, biomass, etc.).
In[ ]:= PieChart
totalcons2021  Total
totalcons2021,
ChartStyle  "DarkRainbow", ChartLabels  energytype, LabelingFunction 
Placed
Row
NumberForm
100 , 3, 1, " ", "RadialCallout" &,
PlotLabel  Style
"Sources of Energy  ", Bold
Sources of Energy


31.0

Oil
24.4

Natural Gas

Out[
]=

Renew ables
6.7

Hydro electric

Nuclear energy
6.8 Coal

4.3
26.9

To highlight an individual source of energy just click on its wedge.

If you’d like to customize a graph, use the information available in the help
system. For example, go to the documentation page for PieChart, and look for
an example that suits your needs.

Now we are going to analyze the evolution of the global primary energy
consumption and look for a function that will enable us to extrapolate energy
consumption over the coming years.
5. Data Analysis and Manipulation 201

We import the sheet “Primary Energy - Consumption”:

In[
]:= primaryenergyDataYear  Import

"https:www.bp.comcontentdambpbusiness sitesenglobalcorporate
xlsxenergy economicsstatistical reviewbp stats review 2022 all
data.xlsx", "Sheets", "Primary Energy Consumption" ;

With TableView the imported data can be shown:
In[
]:= TableView
primaryenergyDataYear

Notice that the desired information is in row 3 (year) and row 111 (Total World, in
Exajoules). We also eliminate not relevant rows and columns.
In[
]:= primaryevolution 
Drop
Drop
Transpose
primaryenergyDataYear3, 112, 1, 4;
Shallow
primaryevolution 
Out[
]//Shallow=
1965., 110.01, 1966., 115.411, 1967., 119.987,
1968., 127.923, 1969., 136.364, 1970., 144., 1971., 147.92,
1972., 155.41, 1973., 164.257, 1974., 162.18, 46

We plot the data. Using Tooltip, if you click on a curve point, its value will be shown:
202 Mathematica beyond mathematics

In[
]:= dataplot  ListLinePlot
Tooltip
primaryevolution ,

PlotStyle  PointSize
.01, Mesh  All, AxesLabel  "Year", "Exajoules",
PlotLabel  Style
"Energy Consumption Evolution", Bold
Energy Consumption Evolution
Exajoules
250

200

150
Out[
]=

100

50

Year
1970 1980 1990 2000 2010 2020

Here, FindFormula is used to find a function that approximates the data:

In[
]:= fit  FindFormula
primaryevolution, t

1.12669  107  861.742 t  1.70908  106 Log
t 1965.  t 1980.64

6393.34  3.31167 t 1980.64  t 1995.29
Out[
]= 3317.39  1.77334 t 1995.29  t 2008.22
1038.06  0.631693 t 2008.22  t 2020.
0 True

After that, we extract the fit function for the last period to extrapolate beyond of 2021:
In[
]:= model  fit1, 4, 1

Out[
]= 1038.06  0.631693 t

In[
]:= Show
Plot
model, t, 2006, 2025, AxesOrigin  1960, 100,

PlotStyle  Pink, Thick, PlotRange  All, dataplot

240

220

200

180
Out[
]=
160

140

120

1970 1980 1990 2000 2010 2020

5. Data Analysis and Manipulation 203

5.7 Time Series Analysis

In this section we are going to see an example of how to use time series to forecast
the weather.

We start collecting the daily average temperatures in the LESA weather station from 1980
until the end of 2021:
In[
]:= averagetempSalamanca 
WeatherData
"LESA", "MeanTemperature", 1980, 2021, "Day"

Time: 01 Jan 1980 to 31 Dec 2021

Out[
]= TimeSeries
Data points: 15 215

The next function fits the data to a typical time series model:
In[
]:= tsm  TimeSeriesModelFit
averagetempSalamanca

Family: ARMA
Out[
]= TimeSeriesModel
Order: 2, 7

We can now use the fitted model to forecast for example the average daily temperature
today and one month from now:
In[ ]:= today  DateValue
Today, "Year", "Month"

Out[
]=
2022, 5

In[
]:= tsm
today

Out[
]= 10.7181 C

In[
]:= tsm
DatePlus
today, 1, "Month"

Out[
]= 11.0916 C

Next, we estimate average monthly temperatures using only the data from the last 10 years:
In[
]:= start  DateValue
DatePlus
Today,  10, "Year", "Year", "Month";
tspec  start, today, "Month";
In[
]:= temp  TimeSeries
WeatherData
"Salamanca", "Temperature", tspec, "Value",
start, Automatic, "Month";

Then display them graphically. Notice the seasonality.
In[
]:= DateListPlot
temp, Joined  True

20

15

Out[
]=
10

0
2014 2016 2018 2020 2022

We fit the data to a time series model:

204 Mathematica beyond mathematics

In[
]:= tsm  TimeSeriesModelFit
temp

Family: SARIMA
Out[
]= TimeSeriesModel
Order: 0, 0, 0 , 1, 1, 1 12

We visualize next year’s forecast for the average monthly temperature using 95%
confidence bands:
In[
]:= fdates  DateRange
today, DatePlus
today, 1, "Year", "Month";

In[
]:= forecast  tsm  fdates;

bands  tsm
"PredictionLimits"
 &  fdates;
In[
]:= DateListPlot
bandsAll, 1, forecast, bandsAll, 2,
today, Automatic, "Month", Joined  True, PlotStyle  Gray, Automatic

25

20

15

Out[
]=
10

Jul Oct Jan Apr

The data records are highly correlated (The p-values are small when the data is
autocorrelated):
In[
]:= ListPlot
Table
AutocorrelationTest
temp, i, i, 1, 6, Filling  Axis

8.  10
24

6.  10
24

24
Out[ ]= 4.  10

2.  10
24

1 2 3 4 5 6

5.8 Spatial Statistics

The newest version of Mathematica has the capability to perform spatial statistics
calculations:
https://reference.wolfram.com/language/guide/SpatialStatistics.html
5. Data Analysis and Manipulation 205

We can now compute continuous values like temperature or elevation associated to

spatial data points. Additionally, we can also get statistical measures such as center,
density, and homogeneity when analyzing locations of things such as trees,
earthquakes, or crimes.
In this example, we show how to estimate the temperature of points in a certain
geographical area (Italy in this case) starting from a single point.

First, we get a list of all the Italian weather stations accessible through Wolfram Resources
and eliminate those for which there’s no information. We extract the last recorded
temperature for each one. Alternatively, we could have chosen data from a specific period of
time.
In[
]:= stations  GeoEntities
Entity
"Country", "France", "WeatherStation";

In[
]:= temp1, locs1  DeleteMissing
AirTemperatureData  stations,

DeleteMissing
Through
stations
"Coordinates";
In[
]:= sf  SpatialEstimate
locs1  temp1

Dimension: 2
Out[
]= SpatialEstimatorFunction
SpatialTrendFunction: 1 
SpatialNoiseLevel: Automatic

Now we can display the temperature at a specified location:

In[
]:= UnitConvert
Quantity
sf
45, 10, "Kelvins", "DegreeCelsius"

Out[
]= 14.3398 C

We can even generate a temperature map for the entire country:

In[
]:= GeoDensityPlot
locs1  temp1,
ColorFunction  "Rainbow",
RegionFunction  Entity
"Country", "France", GeoBackground  None,
PlotLegends  Automatic

C
20

15

Out[
]= 10

0
206 Mathematica beyond mathematics

5.9 Additional Resources

To access the following resources, if they refer to the help files, write their
locations in a notebook (e.g., guide/ListManipulation), select them, and press
<F1>. In the case of external links, copy the web addresses in a browser:
Importing and Exporting: guide/ImportingAndExporting
List Manipulation: guide/ListManipulation
How to Do Statistical Analysis: howto/DoStatisticalAnalysis
Descriptive Statistics: guide/DescriptiveStatistics
Parametric Statistical Distributions: guide/Parametric Statistical Distributions
Curve Fitting: guide/CurveFittingAndApproximateFunctions
Time Series: guide/TimeSeries
DatabaseLink Tutorial: DatabaseLink/tutorial/Overview
Spatial Statistics: guide/SpatialStatistics
Statistics Demonstrations: http://demonstrations.wolfram.com/topic.html?topic=Statistics
Probability Demonstrations: http://demonstrations.wolfram.com/topic.html?topic=Probability
6

        

The number of applications in the real world that use machine learning
continues to increase unabated. The Wolfram Language has both, general
and domain-specific functions, that implement machine learning
algorithms. After briefly discussing the concept of machine learning and
showing some examples of Mathematica’s capabilities in this area for
identifying languages, estimating probability distributions and
highlighting faces, this chapter focuses on the two most important
functions for performing machine learning tasks with WL:
 and
  . Using data from the Titanic and Fisher’s iris datasets among
others, the reader will learn how to classify data into different clusters
and predict numerical outcomes. The last section contains a short
introduction to neural networks.

6.1 What Is Machine Learning?

Nowadays, we know that many of our interactions with our mobile phones, tablets,
computers, and other digital appliances are mediated by machine learning
algorithms, but what is machine learning? Let’s ask the relevant Wikipedia entry
and retrieve its first five sentences:


 TextSentences
WikipediaData
"Machine Learning"
;; 5

 

Machine learning ML is the study of computer algorithms that can improve automatically through
experience and by the use of data., It is seen as a part of artificial intelligence.,
Machine learning algorithms build a model based on sample data, known as training data,
in order to make predictions or decisions without being explicitly programmed to do so.,
Machine learning algorithms are used in a wide variety of applications, such as in
medicine, email filtering, speech recognition, and computer vision, where it is
difficult or unfeasible to develop conventional algorithms to perform the needed tasks.,
A subset of machine learning is closely related to computational statistics, which focuses
on making predictions using computers; but not all machine learning is statistical learning.

In this chapter, we focus on showing Mathematica functions for developing

machine learning (ML) applications or functions that use ML algorithms for
208      

solving domain-specific problems. However, we are not going to explain how those
algorithms actually work.
The two most important functions in WL for ML are Classify and Predict. They are
used for either classifying inputs into categories or class labels or predicting
numeric outputs. The next two sections will cover these functions in more detail.
Next, we show some toy examples of domain-specific functions that use ML:

The function below, identifies the language used to write text:


 LanguageIdentify
 "thank you", "merci", "dar las gracias", "
", "
 "

 
  English , French , Spanish , Chinese , Russian 

In this case, we analyze data from patients suffering from kidney failure to find the
probability distribution that best approximates the time to infection after having the
catheter replaced:

We get the description of the data and store the time to kidney infection (months):


 ExampleData
"Statistics", "KidneyInfection", "LongDescription"

 
 Time to first exitsite infection in months

in patients with renal insufficiency. The data are right censored.


 KidneyInfection  ExampleData
"Statistics", "KidneyInfection"
All, 1;


 KidneyInfection  Short

 
 

1.5, 3.5, 4.5, 4.5, 5.5, 8.5, 8.5, 9.5, 10.5,
101, 19.5, 19.5, 20.5, 22.5, 24.5, 25.5, 26.5, 26.5, 28.5

Next, we use FindDistribution to find the best distribution that fits the data. This function
uses ML algorithms when computing the answer.


 e  FindDistribution
KidneyInfection

 
 ExponentialDistribution
0.108901

Finally, we compare the histogram of the data to the PDF of the estimated distribution:


 Show
Histogram
KidneyInfection, 0, 28, 2, "ProbabilityDensity",
Plot
PDF
e, x, x, 0, 28, PlotStyle  Thick, PlotRange  All

0.10

0.08

0.06
 


0.04

0.02

0.00
0 5 10 15 20 25

The following examples use images available in Mathematica help to show some
  ! 209

functions related to image analysis.

The image below is from the 5th Solvay International Conference on Electrons and Photons
held in 1927, that included the world’s most famous physicists at that time. Can you identify
Niels Bohr? Albert Einstein?


 i ;


 HighlightImage
i, FindFaces
, "BoundingBox" &

 


We can use ImageCases to identify certain subimages:


 ImageCases , dog SPECIES SPECIFICATION

 



The function ImageContents tries to identify the contents of an input image and returns a
dataset of identified entities:
210      


 ImageContents

Image Concept BoundingBox Probability

elephant Rectangle
434.504, 101 0.837853

elephant Rectangle
334.952, 155 0.780911

zebra Rectangle
35.9143, 18. 0.779942

 


elephant Rectangle
133.816, 108 0.694424

zebra Rectangle
3.92214, 152 0.631345

zebra Rectangle
133.058, 96. 0.603996

We can also create a dataset containing information from a given text:

First, we create the text and ask Mathematica to identify all possible entities:


 str  "Japan is an island country in East Asia. It is situated in the
northwest Pacific Ocean, and is bordered on the west by the
Sea of Japan, while extending from the Sea of Okhotsk in the
north toward the East China Sea and Taiwan in the south.";
content  TextContents
str
  ! 211

Then, we make use of the power of Interpreter to transform those entities into a dataset:


 TextCases
str,
Union  Lookup
Normal
content, "Type"  "Interpretation"  Dataset

The ML algorithms implemented in Mathematica can also be applied to video

processing as the example below shows:

Find the number of faces in each video frame:


 v  Import
"http:exampledata.wolfram.comoffice.ts";


 VideoFrameList
v, 3

 

, , 


 ts  VideoMapTimeSeries
Length
FindFaces
Image &, v;

Plot the result:

212      


 ListLinePlot
ts

2.0

1.5

 
 1.0

0.5

2 4 6 8 10 12 14

6.2 Classification
In ML there are two approaches for teaching computers to classify inputs:
unsupervised learning and supervised learning.
Under the first one, an algorithm will try to identify patterns from inputs that are
neither classified nor labeled. Functions such as ClusterClassify and FindClusters
implement the cluster analysis unsupervised algorithm to classify data into clusters,
groups sharing similar characteristics. For additional information please refer to the
“Partitioning Data into Clusters” technical note available at:
https://reference.wolfram.com/language/tutorial/NumericalOperationsOnData.html#2948.

6.2.1 Classify
The supervised learning approach uses inputs that have been already labeled to
teach algorithms how to classify new data. One of the most important functions in
Mathematica for supervised learning is Classify. Let’s see it in action:

In this example, we use a set of images captured either during the day or at night to train an
algorithm to recognize when the image was taken:


 daynight  Classify

 "Night",  "Day",  "Night",

 "Night",  "Day",  "Night",

 "Day",  "Day",  "Night",  "Night",

 "Day",  "Night",  "Night",  "Day",

 "Night",  "Night",  "Day",  "Day",

 "Day",  "Day",  "Night",  "Night",

 "Day",  "Night",  "Night",  "Day",

 "Day",  "Day",  "Night",  "Day"

Input type: Image

 
 ClassifierFunction 
Classes: Day, Night
  ! 213

After generating a classifier function based on the training set, we can now ask the
algorithm to classify new images:


 daynight , , ,

, ,

 
 Day, Night, Day, Night, Night, Night

This function has numerous options that allow us to get additional information such
as the accuracy of the classifier or the confusion matrix:

We first generate a training list and train a classifier with it:


 traininglist  1  "A", 2  "A", 3.5  "B", 3  "A", 5  "B", 6  "B";

  c  Classify
traininglist

Input type: Numerical

 
 ClassifierFunction 
Classes: A, B

Now, we can use test data to assess how good the classifier function is by measuring its
accuracy and uncertainty:


 testset  1.2  "A", 2.1  "A", 3.  "B", 6.4  "B";


 cm  ClassifierMeasurements
c, testset;


 cm
"Accuracy", ComputeUncertainty  True

 
 0.75  0.25

Finally, we can also visualize the confusion matrix to find out how many new inputs were
misclassified:


 ClassifierMeasurements
c, testset, "ConfusionMatrixPlot"
A

A 2 0 2
actual class

 


B 1 1 2
3

predicted class

Let’s imagine we have a wide variety of colors obtained by mixing red and blue
214      

and we would like to classify them into just two categories: Red or Blue.

The first thing we need to do is to train a supervised learning algorithm with a labeled data
set. In principle, the bigger the training set, the better the resulting classifier will be. For
this step, we generate a list of colors mixing red and blue, with 1-x and x being the
proportion of red and blue, respectively. That is 100% red is x = 0, and 100% blue is x = 1.


 Table
x, x, 0, 1, 1  16
1 1 3 1 5
5 11 3 13 7 15 3 7 1 9
 
 0, , , , , ,
, , , , , ,1 , , , ,
16 8 16 4 16 8 16 2 16 8 16 4 16 8 16


 colorsblend  Table
Blend
Red, Blue, x, x, 0, 1, 1  16

    , , , , , , , , , , , , , , , , 

Next, we tell Classify which colors are red and which ones are blue . If the mix is 50%/50%,
we don’t assign any label to that color:


 reddish  colorsblend
1 ;; 8

 
  , , , , , , , 


 blueddish  colorsblend
10 ;; 17

 
  , , , , , , , 


 c  Classify
Red  reddish, Blue  blueddish

Input type: Color

 
 ClassifierFunction 
Classes: ,

We can now test the classifier with mixtures close to 50%/50%:


 GatherBy
Table
x, x, 1  2 1  7, 1  2 1  7, 1  15,  1  2 &  N

 
 0.357143, 0.42381, 0.490476, 0.557143, 0.62381


 colorsblendtest1 

Table
Blend
Red, Blue, x, x, 1  2 1  7, 1  2 1  7, 1  15
 
  , , , , 

Just like people, the classifier may have difficulties assigning a color to mixtures close to
50%/50%. However, in this case it’s doing a great job by correctly assigning the red color to
the third element of the list (49.048% blue):


 c
colorsblendtest1

 
  , , , , 

Finally, we can check the associated probabilities for each possible class:


 c
colorsblendtest1, "Probabilities"

 
   0.000068972,  0.999931,
  0.0060001,  0.994,   0.345658,  0.654342,
  0.978826,  0.0211736,   0.999753,  0.000247122

In this next example, we use
 , the list of colors we have just created, as
an input to LearnDistribution to generate a new classifier in the form of a
probability distribution.

We create the learned distribution first:
  ! 215


 ld  LearnDistribution

 , , , , , , , , , , , , , , , , 

Input type: Color

   LearnedDistribution 
Method: Multinormal

Then, we generate 100 random colors using it:


 RandomVariate
ld, 100

 
  , , , , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , , , 

Approximately half are red, half are blue, and none are green. This is the expected result
given the initial mixtures.


 PDF
ld,  , , 

 
 4371.41983563893, 4370.57360271242, 1.7340803597  10 52 699

Let’s see an example where multiple inputs are associated to just one output.

We are going to mix colors again using, in this case, the three basic colors: Red (R), Green
(G), and Blue (B). We can do this in the WL with the function RGBColor
r, g, b


 Map
Graphics
RGBColor
, Disk
 &, 1, 0, 0, 0, 1, 0,
0, 0, 1, 0.5, 0.5, 0, 0.5, 0, 0.5, 0, 0.5, 0.5

We generate 100 random lists of the form {r,g,b} to reproduce the colors.

  SeedRandom
1234

Method: ExtendedCA
 
 RandomGeneratorState 
State hash: 3 259 750 555 964 806 719


 list  RandomReal
1, 100, 3;


 RGBColor
 &  list

 
  , , , , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , , , 

We can now label each list either Red, Green, or Blue depending on the position of its
largest element. For example, if the biggest value is the second element, we assign the color
Green to the list.


 listcolor 
TakeLargest
  "Index", 1 &  list . 1  Red, 2  Green, 3  Blue ;


 training  Thread
list  Flatten
listcolor;
216      


 clcolor  Classify
training

Input type: NumericalVector length: 3

 
 ClassifierFunction 
Classes: , ,

Let’s see how well the new classifier works.

We generate a 5  5 image first:


 img  Image
RandomReal
1, 5, 5, 3, ColorSpace  "RGB"

 


Next, using the new classifier, we convert that image into a different one containing only the
three basic colors:


 Image
Map
clcolor
 &, Table
ImageData
img
i, 1 ;; 5, i, 5

 



 imgexample  ExampleData
"TestImage", "House"

 



 ImageDimensions
imgexample

 
 256, 256


 Image

Map
clcolor
 &, Table
ImageData
imgexample
i, 1 ;; 256, i, 256

 


In the previous example, Classify chose automatically the learning method.

However, we can also tell the function what algorithm to use (for further details
refer to the documentation pages).

In this case, we use “GradientBoostedTrees” instead of “LogisticRegression”, the one that
Classify automatically chose:
  ! 217


 clcolor1  Classify
training, Method  "GradientBoostedTrees"

Input type: NumericalVector length: 3

 
 ClassifierFunction 
Classes: , ,


 Image

Map
clcolor1
 &, Table
ImageData
imgexample
i, 1 ;; 256, i, 256

 


As mentioned previously, ClusterClassify can be very useful when our data is not
labeled. Let’s see an example.

First, we randomly generate 90 colors:


 colors  RandomColor
90

 
  , , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , ,
, , , , , , , , , , , , , , , , , 

Next, we classify them into 3 clusters:


 c  ClusterClassify
colors, 3

Input type: Color

 
 ClassifierFunction 
Classes: 1, 2, 3

And, finally, gather them by their class number:


 GatherBy
colors, c  TableForm
 
 "# 

The example below shows how to find the dominant color in an image.

We choose an image of a sunflower in this case:
218      


 image  ;

Although we often represent colors using the RGB specification, sometimes it is more
convenient to use a different color space such as LAB. The command below uses
ColorConvert to change the color specification of the image .


 ColorConvert
image, "LAB"

 


To process it, we first extract the data:


 imageData  Flatten
ImageData
ColorConvert
image, "LAB", 1;

Then, we cluster the array of pixels just obtained:


 c  ClusterClassify
imageData, 4, Method  "KMedoids";

Next, we use the classifier to assign a cluster to each pixel and find four dominant colors:


 decision  c
imageData;


 LABColor  Mean  Pick
imageData, decision,  &  Range
4

 

, , , 

Finally, we get binary masks for each dominant color:


 Image 
ComponentMeasurements
image, Partition
decision, 426, "Mask"
All, 2

 
  , , ,

6.2.2 Titanic Survival

In this example, we use the Titanic dataset that includes the class, age, gender, and
survivor status for 1,309 passengers that were aboard the ship.
This dataset contains a mixture of numerical and textual data.


 Short
data  ExampleData
"MachineLearning", "Titanic", "Data"
 
 
1st, 29., female  survived, 1st, 0.9167, male  survived,
1305 , 3rd, 27., male  died, 3rd, 29., male  died
  ! 219


 titanic  Classify
data

Input type: Nominal, Numerical, Nominal

 
 ClassifierFunction 
Classes: died, survived

Now that we have our classifier, we can ask questions, such as the probability that a 50-year
old male would have either died or survived:


 titanic
"1st", 50, "male", "Probabilities"

 
 died  0.551655, survived  0.448345

We can also visualize the probability of survival for first-class males based on their ages:


 Plot
titanic
"1st", age, "male",
"Probability", "survived", age, 0, 70
0.60

0.55

0.50

0.45
 

0.40

0.35

0.30

10 20 30 40 50 60 70

We can even predict the age of a passenger using the rest of the features. For example, the
predicted age of a female in 3rd class that survived is:

  titanicAge  Predict
data . HoldPattern
class, age, sex  outcome 
class, sex, outcome  age;
titanicAge
"3rd", "female", "survived"
 
 21.5452

For our last example in this section, we use Fisher’s Iris dataset. The dataset
consists of 150 samples from three species of Iris (setosa, versicolor, and virginica).
Each sample contains four measurements: the length and the width of the sepals
and petals. We want to be able to correctly classify new Iris flowers based on those
four measurements:

Let’s download the data first:


 irisData  ExampleData
"MachineLearning", "FisherIris", "Data";


 irisData  Short

 
 


5.1, 3.5, 1.4, 0.2 setosa,
4.9, 3., 1.4, 0.2 setosa, 147,
5.9, 3., 5.1, 1.8 virginica

Next, we train the algorithm using 100 randomly selected samples:

220      


 SeedRandom
1234;

irisTrainingData  RandomSample
irisData, 100;
irisCf  Classify
irisTrainingData

Input type: NumericalVector length: 4

 
 ClassifierFunction 
Classes: setosa, versicolor, virginica

The remaining samples can then be used as test data:


 irisTestData  Complement
irisData, irisTrainingData;

Finally, using ClassifierMeasurements on the test dataset, we calculate the accuracy of the
algorithm, visualize the confusion matrix (2 samples of versicolor were misclassified as
virginica) and see what examples had the highest actual-class probabilities:


 irisCm  ClassifierMeasurements
irisCf, irisTestData;


 irisCm
"Accuracy"

 
 0.96


 irisCm
"ConfusionMatrixPlot"

versicolor

virginica
setosa

setosa 15 0 0 15
actual class

 

versicolor 0 14 2 16

virginica 0 0 19 19
15

14

21

predicted class


 irisCm
"BestClassifiedExamples"

 
 7.7, 2.6, 6.9, 2.3  virginica, 7.7, 2.8, 6.7, 2.  virginica,
7.7, 3.8, 6.7, 2.2  virginica, 7.2, 3.6, 6.1, 2.5  virginica,
6.7, 3.3, 5.7, 2.5  virginica, 6.3, 3.4, 5.6, 2.4  virginica,
4.4, 3.2, 1.3, 0.2  setosa, 6.8, 3.2, 5.9, 2.3  virginica,
5.5, 4.2, 1.4, 0.2  setosa, 4.3, 3., 1.1, 0.1  setosa

Evaluating the probabilities for the first element of the previous output, we can see the
classifier is 100% certain that those measurements correspond to an Iris virginica flower:


 irisCf
7.7, 2.6, 6.9, 2.3, "Probabilities"

 
 setosa  2.24715  1021 , versicolor  3.51601  108 , virginica  1.

  ! 221

6.3 Prediction

6.3.1 The Predict function

Apart from Classify, the other main function for implementing supervised learning
in Mathematica is Predict. The example below shows how to use it for fitting
functions to data.

We first simulate input-output data of the form: x y


 trainingdata  Table
x  Sin
x RandomVariate
NormalDistribution
0, .3,
x, RandomReal
 5, 5, 50 ;

Then, we predict any intermediate values using the “GaussianProcess” method. For further
information about all the methods available for prediction, read the function documentation.


 p  Predict
trainingdata, Method  "GaussianProcess" 

Input type: Numerical

 
 PredictorFunction 
Method: GaussianProcess

Finally, we compare visually the training data with values obtained with the predictor
function:


 Show
ListPlot
List  trainingdata, Plot
p
x, x, 10, 10
1.5

1.0

0.5

 
 4 2 2 4

0.5

1.0

1.5

Let’s suppose that a person was drinking water heavily contaminated with lead for
100 days. To find out the lead concentration after the exposure, urine samples were
taken during the next 200 days. Our objective is to create a function to accurately
predict the evolution of lead concentration after the first 200 days.
222      

sampleslead  1, 2.58, 2, 2.95, 3, 2.66, 4, 3.83, 5, 4.18,
6, 4.18, 7, 15.1, 8, 12.65, 9, 12.53, 10, 11.57,
11, 11.69, 12, 10.97, 13, 11.97, 14, 12.18, 15, 16.56,
16, 16.2, 17, 15.9, 18, 21.02, 19, 19.52, 20, 18.63,
21, 18.2, 22, 17.72, 23, 18.63, 24, 18.24, 25, 19.18,
26, 18.37, 27, 18.81, 28, 18.41, 29, 18.36, 30, 18.31,
31, 19.87, 32, 20.08, 33, 32.34, 34, 30.29, 35, 30.77,
36, 29.19, 37, 28.35, 38, 30.07, 39, 29.58, 40, 30.7,
41, 29.19, 42, 28.17, 43, 32.65, 44, 30.96, 45, 30.6,
46, 31.97, 47, 32.95, 48, 32.04, 49, 33.9, 50, 32.59,
51, 31.74, 52, 31.18, 53, 33.22, 54, 32.06, 55, 31.64,
56, 33.04, 57, 35.26, 58, 34.64, 59, 33.78, 60, 36.14,
61, 34.41, 62, 33.87, 63, 33.02, 64, 32.61, 65, 32.58,
66, 31.76, 67, 32.17, 68, 35.32, 69, 35.04,
70, 33.88, 71, 33.14, 72, 34.5, 73, 34.54, 74, 33.24,
75, 33.58, 76, 33.75, 77, 33.13, 78, 45.23, 79, 41.97,
80, 39.77, 81, 42.11, 82, 42.49, 83, 41.58, 84, 39.78,
85, 43.89, 86, 42.41, 87, 41.36, 88, 40.3, 89, 39.76,
90, 39.84, 91, 39.77, 92, 42.76, 93, 41.29, 94, 41.15,
95, 40.71, 96, 39.84, 97, 39.62, 98, 39.89, 99, 39.24,
100, 38.8, 101, 37.56, 102, 36.53, 103, 35.64,
104, 34.83, 105, 34.07, 106, 33.36, 107, 32.68,
108, 32.01, 109, 31.37, 110, 30.75, 111, 30.13, 112, 29.53,
113, 28.95, 114, 28.37, 115, 27.81, 116, 27.26,
117, 26.72, 118, 26.19, 119, 25.67, 120, 25.17,
121, 24.67, 122, 24.18, 123, 23.7, 124, 23.23, 125, 22.77,
126, 22.32, 127, 21.88, 128, 21.44, 129, 21.02, 130, 20.6,
131, 20.2, 132, 19.8, 133, 19.4, 134, 19.02, 135, 18.64,
136, 18.27, 137, 17.91, 138, 17.56, 139, 17.21,
140, 16.87, 141, 16.53, 142, 16.21, 143, 15.89,
144, 15.57, 145, 15.26, 146, 14.96, 147, 14.66,
148, 14.37, 149, 14.09, 150, 13.81, 151, 13.54, 152, 13.27,
153, 13.01, 154, 12.75, 155, 12.5, 156, 12.25, 157, 12.01,
158, 11.77, 159, 11.54, 160, 11.31, 161, 11.08,
162, 10.86, 163, 10.65, 164, 10.44, 165, 10.23,
166, 10.03, 167, 9.83, 168, 9.64, 169, 9.44, 170, 9.26,
171, 9.07, 172, 8.89, 173, 8.72, 174, 8.55, 175, 8.38,
176, 8.21, 177, 8.05, 178, 7.89, 179, 7.73, 180, 7.58,
181, 7.43, 182, 7.28, 183, 7.14, 184, 7., 185, 6.86,
186, 6.72, 187, 6.59, 188, 6.46, 189, 6.33, 190, 6.21,
191, 6.08, 192, 5.96, 193, 5.84, 194, 5.73, 195, 5.62,
196, 5.5, 197, 5.39, 198, 5.29, 199, 5.18, 200, 5.08;


 inp, out  Transpose
sampleslead;


 p  Predict
inp  out

Input type: Numerical

 
 PredictorFunction 
Method: NearestNeighbors
  ! 223

Once the predictor function has been created, we can plot the predicted values against the
training sample:


 Show
Plot
p
x, x, 1, 250, Exclusions  None,
ListPlot
sampleslead, PlotStyle  Green

40

30

 

20

10

50 100 150 200 250

6.3.2 Wine Quality

In this example (available in the documentation), we want to predict subjective
opinions regarding the quality of wine based on certain physical measurements.

Learn additional details about the example:


 ExampleData
"MachineLearning", "WineQuality", "LongDescription"

 
 Predict the subjectively reported quality of a white wine

on a scale of 110, given 11 physical features of the wine.
These features include properties like the pH of
the wine and its alcohol content.There are 4898 examples.

Load the training dataset:


 trainingset 
ExampleData
"MachineLearning", "WineQuality", "TrainingData";

Visualize a few data points:


 RandomSample
trainingset, 3  TableForm
 
 "# 
7.5, 0.3, 0.32, 1.4, 0.032, 31., 161., 0.99154, 2.95, 0.42, 10.5  5.
6.8, 0.19, 0.4, 9.85, 0.055, 41., 103., 0.99532, 2.98, 0.56, 10.5  6.
9.4, 0.28, 0.3, 1.6, 0.045, 36., 139., 0.99534, 3.11, 0.49, 9.3  5.

Get a description of the variables:

  ExampleData
"MachineLearning", "WineQuality", "VariableDescriptions"

 
 fixed acidity, volatile acidity, citric acid, residual sugar,

chlorides, free sulfur dioxide, total sulfur dioxide, density,
pH, sulphates, alcohol  wine quality score between 110

Visualize the distribution of the "alcohol" and "pH" variables:

224      


 Histogram
trainingset
All, 1, 11, PlotLabel  "alcohol",

Histogram
trainingset
All, 1, 9, PlotLabel  "pH"
alcohol pH
350 500
300
400
250
300
 

200 , 
150 200
100
50 100
0 0
9 10 11 12 13 14 2.8 3.0 3.2 3.4 3.6 3.8

Train a predictor on the training set:


 p  Predict
trainingset

Input type: NumericalVector length: 11

 
 PredictorFunction 
Method: RandomForest

Predict the quality of an unknown wine:


 unknownwine  7.6`, 0.48`, 0.31`, 9.4`,
0.046`, 6.`, 194.`, 0.99714`, 3.07`, 0.61`, 9.4`;


 p
unknownwine

 
 4.96642

After using the “GradientBoostedTrees” method to create a new predictor function, we can
create “SHAPPlots” to measure the impact of specific features on the final prediction. Note
that in this case we download the data set using ResourceData.


 wine  ResourceData
"Sample Data: Wine Quality";
p  Predict
wine  "WineQuality",
Method  "GradientBoostedTrees", PerformanceGoal  "Speed";


 pm  PredictorMeasurements
p, wine;


 pm
"SHAPPlots"

 


The three features that seem to have a significant influence in the predicted quality
according to their SHAP (Shapley additive explanations) values are: Alcohol,
FreeSulfurDioxide, and VolatileAcidity.
  ! 225

6.4 Working with Neural Networks

Many machine learning algorithms use neural networks. At their core, neural
networks are a collection of regression layers that feed into each other in a way that
mimics the architecture of the brain. These collection of layers can be represented
as a symbolic graph where each layer holds fixed information about its type, inputs
and outputs, and weights. The following video from the 3Blue1Brown YouTube
channel provides an excellent overview of the topic:
https://www.youtube.com/watch?v=aircAruvnKk
WL features a wide selection of functions to build and train them. For example,
most of the Mathematica commands that contain the word “Layer” or start with
“Net” can be used for working with neural networks:


 ? Layer

In Mathematica, we can manipulate the neural network expressions created by the

language like any other symbolic object and, for example, optimize their weights
using training data. A more detailed overview of neural networks in the Wolfram
Language is available at:
https://reference.wolfram.com/language/tutorial/NeuralNetworksOverview.html
The Wolfram Neural Net Repository contains a vast collection of trained and
untrained models to build, train an deploy neural networks:
https://resources.wolframcloud.com/NeuralNetRepository/
The following examples show how to get started using them in Mathematica.

6.4.1 A Basic Introductory Example

Given a list of data points {xi , yi }:


 data  0, 3.1, 1, 3.4, 2, 4.2, 3, 4.4, 4, 4.9, 5, 5;

if we want to find the best linear fit, we can just use Fit to obtain the desired equation of the
line:


 model  Fit
data, 1, x, x

 
 0.405714 x  3.15238

However, let’s use this toy example to show how easy it would be to plot the line
of best fit and visualize the errors using LeastSquaresPlot, one of the functions
available in the Wolfram Function Repository:
https://resources.wolframcloud.com/FunctionRepository/resources/LeastSquaresPlot/
226      




 LeastSquaresPlot
data

    
  

5.0

4.5

 

4.0

3.5

1 2 3 4 5

6.4.2 Using a Neural Network

An alternative approach would be to use a neural network to find the coefficients for the
slope and the y-intercept:


 net  LinearLayer
1, "Input"  "Scalar", "Output"  "Scalar"

Input : scalar
 
 LinearLayer    
Output : scalar


 inputOutputData  Rule  data

 

0 3.1, 1 3.4, 2 4.2, 3 4.4, 4 4.9, 5 5


 trained  NetTrain
net, inputOutputData

Input : scalar
 
 LinearLayer 
Output : scalar

We can now extract the values of the slope and the intercept by using the options “Weights”
and “Biases”, respectively:


 NetExtract
trained, "Weights"  Normal

 
  0.407973 


 NetExtract
trained, "Biases"  Normal

 

3.14473


 NetChain
trained, trained
1

 
 4.59413

The resulting graph is very similar to the one obtained previously:

  ! 227


 Show

ListPlot
data,
Plot
trained
x, x, 0, 5


5.0

4.5

 
 4.0

3.5

1 2 3 4 5

6.4.3 Revisiting the Titanic Dataset

In this example we build a neural network classifier to assign the labels “survived”
or “died” to passengers, based on information about their ticket class, age, and sex:

As usual, we download the dataset first:


 ResourceData
"Sample Data: Titanic Survival"
228      

We then modify the data to follow this pattern: {{Class, Age, Sex} survived or {Class,
Age, Sex} died}. Missing data is represented by “—”:


 dataset 
Rule
Lookup
, "Class", "Age", "Sex", Lookup
, "SurvivalStatus" & 
ResourceData
"Sample Data: Titanic Survival";
train, test  Normal  TakeDrop
dataset, 1000;
Short
train
 
 

3rd, Missing
Not Available, male  survived,

307 ,
3rd, 29. yr , male  died

Next, we define an neural network using NetChain. In this case, we are creating a multilayer
perceptron (MLP) classifier with a "FeatureExtractor" encoder to deal automatically
with mixed and missing data:


 net  NetChain

DropoutLayer
, LinearLayer
12, ElementwiseLayer
LogisticSigmoid,
LinearLayer
4, LinearLayer
2, SoftmaxLayer
,
"Output"  NetDecoder
"Class", "survived", "died",
"Input"  NetEncoder
"FeatureExtractor"
;

We now train the feature extractor and the classifier jointly, using NetTrain:


 trainedNet  NetTrain
net, train

Input port: feature extractor

 
 NetChain 
Output port: class

Finally, we test the classifier on unseen data:


 cm  ClassifierMeasurements
trainedNet, test;
AssociationMap
cm,
"ConfusionMatrixPlot", "Accuracy", "Precision"  Dataset

49 28
actual class

ConfusionMatrixPlot
46 186

 

predicted class

Accuracy 0.760518
Precision
survived  0.515789, died  0.869159

We leave it as an exercise for the reader to compare this MLP with the classifier we
created previously using Classify. Which one generates better results?
  ! 229

6.4.4 Handwritten Numbers

This example shows how to download and train a neural network to recognize
handwritten figures:
For this exercise NVIDIA GPU with supported compute capability should be installed, in other case skip
it.

Let’s download the model first using NetModel:


 net  NetModel
"LeNet Trained on MNIST Data"

Input port: image

 
 NetChain Output port: class 
Number of layers: 11

Next, we test the net:

  net  , , , ,

 
 8, 0, 4, 1, 6

We can also train the model with new cases using NetTrain and test it:


 newnet  NetTrain net,  1,  2,  3,

 4,  5,  6,  7,  8, 9

Input port: image

 
 NetChain Output port: class 
Number of layers: 11


 newnet

 
 7

Based on the existing net, it’s very easy to initialize a new model (NetInitialize) and train it
(NetTrain) to classify handwritten inputs as either “Arabic” or “Latin”:


 languageNet  NetInitialize

NetJoin

NetTake
net, 9,
NetChain
2, SoftmaxLayer
,
"Output"  NetDecoder
"Class", "Arabic", "Latin",
All

Input port: image

 
 NetChain Output port: class 
Number of layers: 11
230      


 trainedLanguageNet  NetTrain languageNet,

 "Arabic",  "Arabic",  "Arabic",  "Arabic",

 "Arabic",  "Arabic",  "Latin",  "Latin",

 "Latin",  "Latin",  "Latin" , TargetDevice  "GPU"

Input port: image

 
 NetChain Output port: class 
Number of layers: 11


 trainedLanguageNet 

, , , , , , , , , ,

 
 Arabic, Arabic, Arabic, Arabic,

Arabic, Arabic, Latin, Latin, Latin, Latin, Latin


 trainedLanguageNet , "Probabilities"

 
 Arabic  1., Latin  1.9582  107

6.5 Additional Resources

The machine learning functionality of Mathematica is under active development so
it would be a good idea to visit the following pages after each new version release:
https://reference.wolfram.com/language/guide/MachineLearning.html
https://reference.wolfram.com/language/workflowguide/MachineLearning.html
https://reference.wolfram.com/language/guide/NeuralNetworks.html
https://reference.wolfram.com/language/tutorial/NeuralNetworksOverview.html
Introduction to Machine Learning by Etienne Bernard is a very nice introduction to
machine learning using Mathematica that can be read online at:
https://wolfr.am/iml. A printed version is also available.
7
Calculating
and Other Mathematical
Tales

The first six sections of this chapter are dedicated to
, one of the most
famous mathematical constants. What looks like a simple geometric ratio
between the length of a circle’s circumference and its diameter has
surprising properties and unexpected connections to seemingly unrelated
fields. Many famous mathematicians throughout history have created
their own formulas to calculate the largest number of decimals of
in the
most efficient possible way, something that they currently still do. The last
two sections discuss the Riemann hypothesis, one of the most important
unsolved problems in mathematics, and its connection to the prime
numbers.

7.1 The Origins of

Approximately 4,000 years ago, Egyptians and Babylonians already knew that the
ratio between the circle’s circumference and its diameter, what we call
(pi)
nowadays, had a constant value slightly higher than 3.
The first estimates were most likely based on experimental measurements. After
drawing a circle and measuring its perimeter and diameter, you will notice that the
perimeter is a bit more than three times the diameter.
The Old Testament includes an estimation of
in the First Book of Kings 7:23,
referring to the construction of Solomon’s temple. We can read: “And he made a
molten sea, ten cubits from the one brim to the other: it was round all about, and his
height was five cubits: and a line of thirty cubits did compass it round about”; that
is, its circumference was 30 cubits and its diameter was 10 cubits, so the
relationship between the circumference and the diameter is 3.
In the Great Pyramid of Giza there are ratios clearly showing that the Egyptians
had obtained a good approximation to the value of . The Rhind Papyrus (circa
1650 B.C.), one of the best sources about mathematics in Ancient Egypt, contains
232      

256
the following ratio as the value of : 81
=3.16049.
The Babylonians noticed that the length of a circle was slightly bigger than the one
of an hexagon inscribed inside it getting a  value of 3 18 = 3.125, quite close to the
correct value of 3.14159....
The following example describes a function to inscribe a n-sided polygon inside a
circle.
 The display of a polygon inscribed in a circle can be done using the
 and  
functions inside the   command.

 without arguments indicates a circle
of radius r
1 and center at x, y
0, 0.  
x1 , y1 , ...,xn , yn  represents a
polygon whose vertices are given by the coordinates x1 , y1 , ...,xn , yn . We use   to
place text inside the graphical object. In the function definition we include ?  and
? to execute the function only if n is an even positive integer.


 polig1
n ? EvenQ ? Positive : Graphics
Circle
,
LightGray, Polygon
Table
Cos
2
k  n, Sin
2
k  n , k, n  ,
Inset
"Inscribed Polygon" , 0, 0  , ImageSize  130
 We can add a warning message that will be displayed if the requirements for n are not met.
This function must be defined immediately after the previous one. This way,
 will
check first whether n is a positive even number and if not, it will display the warning
message. If we reverse the order in which we define the functions, it will not work.


 polig1
n : "n must be an even positive number."


 polig1
6

 
 Inscribed Polygon


 polig1
5

 
 n must be an even positive number.


 polig1
2

 
 n must be an even positive number.

An alternative way to display a 2D figure without using Mathematica functions is

to use the Drawing Tools, available in the 
 menu or by pressing +D in
Windows or +T in OS X.

7.2 Archimedes’ Approximation

Archimedes (287–212 B.C.) is one of the most famous individuals in the history of
science. Even today some of his discoveries are still being studied in schools and
his famous palimpsest, the oldest surviving copy of his works, has been the subject
          233

of extensive research since 1999. He lived at the end of the Ancient Greece era, the
same period as other famous mathematicians such as Euclid (ca. 325–ca. 270 B.C.)
and Eratosthenes (ca. 284–ca. 192 B.C.). They were all located in a small
geographical region consisting of several cities around the Aegean sea, a lucky
coincidence that has not happened again in the history of humankind. Archimedes’
life is the subject of many legends of which the best-known one is probably the
story of how he was able to set on fire a large fleet of Roman galleys that were
trying to attack Syracuse. His extraordinary intelligence was acknowledged even by
his own enemies. Plutarch tells us that during the siege of Syracuse a Roman
soldier found the old man so absorbed in his geometric drawings that he ignored his
orders and was killed as a result. Marcellus, the person in charge of the Roman
troops, had given express orders to spare his life and when he heard the news he
punished the soldier, mourned Archimedes’ death, and buried him with honors.
About the The Archimedes Palimpsest: http://www.archimedespalimpsest.org/
In Archimedes’ tomb there is a cylinder enclosing a sphere, a reference to one of
his greatest findings, that the sphere has 2/3 the volume of its circumscribing
cylinder. The same 2/3 ratio applies to all surfaces.
Next we are going to show how to inscribe a sphere inside a cylinder.
 To show a cylinder enclosing a sphere, we use the syntax Graphics3D[{{properties,
Cylinder}, {properties, Sphere}}, options]. We use Opacity to define the degree of
transparency in the cylinder. We can also choose the color and other relevant details using
options. Since we are not using any arguments for either Cylinder
 or Sphere
, the
commands assume by default that r
1, x, y, z
0, 0, 0 and, in the case of the cylinder,
h
1.


 Graphics3D
Opacity
0.2, Blue, Cylinder
 ,  Green, Sphere
 ,
Boxed  False, ImageSize  100

 


 The same graph can be done with ContourPlot3D, a Mathematica command used for any 3D
function for which we know its analytic expression. PlotLabel, a general graphics option,
adds a legend to the graph. In this example we use the equations of both geometric figures
as legends and display them on separate lines with TableForm.
234      


 ContourPlot3D
x ^ 2 y ^ 2 z ^ 2  1, x ^ 2 y ^ 2  1 , x, 1, 1 , y, 1, 1 ,

z, 1, 1 , ContourStyle  Green, Blue, Opacity
0.2 , Mesh  None,
Axes  False, Boxed  False, ImageSize  200, PlotLabel  TableForm

"Sphere:" x ^ 2 y ^ 2 z ^ 2  1, "Cylinder:" x ^ 2 y ^ 2  1  0  z  1 

 


The method that follows enables us to verify that the ratio between the volume of a
sphere and that of its circumscribing cylinder is 2/3 (without using the formulas for
their respective volumes).
We first compute the volume of the sphere, Vsphe , and of the cylinder ,Vcyl , with
r
1 by integrating their analytical expressions and then calculate Vsphe /Vcil .
Archimedes obviously did not use this method since he didn’t know integral
calculus.
 We use Boole to indicate that the integral of the points that do not satisfy the condition is 0.


 volsphe  Integrate
Boole
x ^ 2 y^2 z ^ 2  1,
x, 1, 1 , y, 1, 1 , z, 1, 1 
4

 

3


 volcyl 
Integrate
Boole
x ^ 2 y ^ 2  1, x, 1, 1 , y, 1, 1 , z, 1, 1 
 
 2


 volsphe  volcyl

2
 

3
One of Archimedes’ biggest discoveries was a new method for computing the value
of
. He realized that by dividing a circle into identical sectors, the sum of their
areas would resemble that of a sum of triangles whose total area would
approximate the area of the circle. Geometrically, the previous method is
equivalent to placing the sectors together inverting every other one. If each sector is
replaced by its inscribed triangle, that Archimedes knew how to calculate, the
resulting area would be very close to that of the circle. Furthermore, when r
1,
          235

since the area of the circle is Ac

r2 , then Ac

, and from that formula we can
deduce that the value of
is going to be between the value area of the inscribed
polygon, Ains and that of the circumscribed polygon Acir , that is, Ains  Ac


Acir .
 The previous procedure can be easily modeled in Mathematica. We use GraphicsRow to
display all the figures in the same row. The higher the number of triangles, the closer the
resemblance of the resulting graphical object to a rectangle.


 Manipulate

GraphicsRow
Table
Graphics
EdgeForm
Opacity
1, Table
Hue
i  n, .5,
Disk
i Sin

 n ,  1 ^ i 1 1 Cos

 n  2 , 1,

 1 ^ i  2 1  n,  1 ^ i  2 1  n  , i, n  , n, 4, m ,
m, 6, "number of sectors" , 4, 20, 2, Appearance  "Labeled" 

7.3
with More Than One Trillion Decimals

A 16-digit approximation of
, like the one available in any calculator, is more than
enough for practical computations.
For example, the calculation error when measuring the earth circumference with
such a number would be much smaller than 1 mm.
 The average distance from the earth to the sun is:


 sunDist  Average Sun distance in meters

 
 
 

 The difference between using the exact value of
and its approximation with a 
precision is insignificant.
16


 2
10 sunDist

 
 0.0000939951143117 m

Despite this fact, the number of functions created to compute
with the largest
number of decimals in the fastest possible way since Archimedes is remarkable.
Some of the best mathematicians in history have developed their own approaches.
In the following paragraphs, we’ll show some of them.
Although Archimedes’ method is inefficient, little progress was made until Newton
in 1666 described a radically new method.
(Ve)The Discovery That Transformed Pi: https://www.youtube.com/watch?v=gMlf1ELvRzc
236      

He found a pi approximation by evaluating the first 22 terms of this sum:

     

      
   ! 
  ! 

Newton generated this expression for
by considering the shape below. His
approach to calculate the famous constant is based on computing the area of a
semicircle using an integral.


 Graphics
LightBlue, Disk
0.5, 0 , 0.5, 2 Pi  3., Pi  ,
LightGreen, Polygon
0.25, 0 , 0.5, 0 , 0.25, Sqrt
3  4.  ,
Circle
0.5, 0 , 0.5, 0, Pi , Line
0, 0 , 0, 0.6 ,
Line
0.5, 0 , 0.25, 0.5 Sqrt
3  2 , 0.25, 0 , ,
Text
Style
"A", 16, 0., 0. , 1, 1 ,
Text
Style
"B", 16, 0.26, 0 , 1, 1 ,
Text
Style
"C", 16, 0.5, 0 , 1, 1 ,
Text
Style
"D", 16, 0.25, 0.5 Sqrt
3.  2 , 1, 1  ,
Axes  True, False , Ticks  1  4, 1  2, 1 , 1  2 ,
PlotRange   0.1, 1.1 ,  .1, 0.6 

Here is a more detailed explanation of his solution:

122

 The area  under the arc AD (light blue) is equal to the area of the sector ADC (
),
6 24
1 1 1 1 1 3
minus the area of triangle BCD (light green) ( 2 ( BC  BD) = 2 ( 4  4
 )= ). That
42 32

3
is, 24

32
.


 sol1  Solve
  pi  24 Sqrt
3  32, pi
3
 
 pi 32   3
4
 The circle has center (1/2, 0) and radius = 1/2. Therefore:
1 2 1


 sol2  SolveValues x y2  , y
2 4

 
  x  x2 , x  x2

 Now the integration can be computed directly

  Integrate
sol2 2 , x, 0, 1  4 
1
 
 4
3 3
96
          237

3


 Simplify 3 32 
4
 


Obviously, in the 17th century, these calculations took much longer.

 Since y
x  x2
x 1  x , we can now find the value of x 1  x using the
binomial expansion:


 bexpansion  x Series 1 x , x, 0, 4 

x32 x52 x72

 
 x     O x92
2 8 16
x32 x52 x72


 Integrate x, , , , x, 0, 1  4 
2 8 16
1 1 1 1
 
 , , ,
12 160 3584 36 864



! 
    
"   
     
  
    
         
   
    
  ! 
  ! 
 Therefore:


 sol1 1, 1
3
 
 pi 32   3
4
1 1 1 1


 .    N
12 160 3584 36 864
 
 pi 3.14169

Contrary to expectations given the original definition of
as the ratio between the
perimeter of a circumference and its diameter, many of its later definitions do not
imply a geometric relationships. The original one was of a practical nature. Other
non-geometrical definitions are also valid. As a matter of fact, we could take any of
them as the definition of
. Next we present some examples.
 The formula below is probably the easiest way (but not the most effective one) to calculate

. Notice that Mathematica is able to obtain an exact result for infinite series.
  1j


 4
j0
2j 1

 


 The next one shows
as the result of an integral computation. It’s due to Leibniz, whose
integration methods won over those from his contemporary fellow mathematician Newton.
 1


  t
0 t2 1


 

2
238      

 There are very curious methods such as the ones that compute
using a series of continuous
fractions. Here’s one example:


 4
 1 1  3 22  5 32  7 42  9 52  11 62  13 

4 1
 
  1

4
9
3
5
16
7
25
9
36
11
13

One of the most surprising formulas to calculate
comes from Ramanujan

(1887–1920), the famous Indian mathematician.
Born to a poor family, he didn’t have a formal education, learning mathematics
completely on his own (The movie The Man Who Knew Infinity is based in his
life). When he was in his teens, he read a book containing only theorems and a list
of problems with their respective solutions. Since he could mentally arrived at the
answer without the need to write down intermediate calculations, he thought that
would be the ideal way to proceed: to figure out mentally the solution and how to
get it and just write down the final result. Accordingly, he made a list of several
identities that according to him were self-evident and sent them to several
mathematicians. One of them, Hardy, quickly realized that the author of the list was
a genius, and he invited him to Oxford. From his time in England, we have the
following famous anecdote: One day Hardy was telling Ramanujan that he had
taken a cab with a very boring license plate number 1729, to which Ramanujan
replied that on the contrary that number was very interesting since it was the
smallest one that could be decomposed into the sum of two different cubes.
 In fact:


 93 103  13 123  1729.

 
 True

 It’s not by chance then that in 1914 Ramanujan found the following surprising formula:
2 2 n 4 j  26 390 j 1103


 ramapi
n : 
9801 j0 44 j j 4  994 j


 ramapi


1
 



 To check how fast the above formula converges we proceed as follows: Since
=
1/ramapi[], we increase the number of digits in the expression
- 1/ramapi[n], using #[] to
force Mathematica to display the numeric approximation. We can see that with just the first
3 terms, the approximation is accurate to 30 decimal places:
          239

1


 TableFormTablen, N
, 10, n, 1, 5 
ramapi
n
 
  
1 6.395362624 1016
2 5.682423256 1024
3 5.238896280 1032
4 4.944187579 1040
5 4.741011769 1048

 If we add a 6th term we’ll get an error message. The reason is that the series convergence is
so fast that we need to substantially increase the computation precision. To do that we
change the value of the variable $MaxExtraPrecision and limit the scope of the new
assignment with Block to the previous expression only; otherwise, the new value would be
applied to all calculations done after the latest definition.


 Block$MaxExtraPrecision  10 000,
1
TableFormTablen, N
, 10, n, 1, 10 
ramapi
n
 
  
1 6.395362624 1016
2 5.682423256 1024
3 5.238896280 1032
4 4.944187579 1040
5 4.741011769 1048
6 4.598865016 1056
7 4.499979208 1064
8 4.433276097 1072
9 4.391477319 1080
10 4.369600809 1088

By March 14, 2019 (the Pi day) the approximation reached 31.4 trillions digits
(31.4 1012 ) using the Chudnovsky algorithm (it is the used by Mathematica). To
put that figure in perspective, if we wanted to print that many digits we would need
a library of 30 million volumes (500 km worth of shelves) .
n  1k  6  k   13 591 409 545 140 134  k


 chudnovsky
n : 12 
k0 3  k  k 3  640 3203k 32


 chudnovsky


1
 


To verify that a sequence is correct, we compare part of it with the ones obtained
by different methods. For example:
 We can combine RealDigits and [[...]] to show the digits located in positions n to m + n after
the decimal point. We extract the digit n+1 since 3, the first digit, is not decimal. We then
use n  m  5 to increase the precision with 5 more digits than the required ones to
calculate m  n.


 pinth
n, m : RealDigits
N

3, n m 5 1, n ;; n m
240      

 For example, we can check that after position 762, the digit 9 appears six consecutive times.


 pinth
760, 10

 
 3, 4, 9, 9, 9, 9, 9, 9, 8, 3, 7

But, how can a record be tested to see if it is still valid? Usually the Bailey–
Borwein–Plouffe’s formula (1996) is used.
n 1 4 2 1 1


 bayley1
n : 
k0 16k 8k 1 8k 4 8k 5 8k 6

“On the rapid computation of various polylogarithm constants” (Mathematics of

Computation 66–218, 1997) the authors describe a method in which by using the
previous formula it can calculate directly in base 16 the nth digit of
. We can use
this fact to verify whether a record is valid.
 The formulas that we have seen so far represent just a small fraction of the existing ones.
We can take a look at some of those formulas by using Mathematica’s free-form input.
Press = and type Pi. To see the results, expand the output by clicking on the + symbol in the
top right corner.

7.4 Buffon’s Method

The Frenchman Georges-Louis Leclerc (1707–1788), Comte de Buffon, was a
naturalist whose encyclopedia Histoire Naturelle had a great influence on his 19th
century colleagues. Charles Darwin even cites him in his book On the Origin of
Species. In mathematics, he’s famous for his experimental method to calculate
.
The method, as shown in Figure 7.1, consists of randomly throwing a needle of
length l onto a plane containing parallel lines at a fixed distance d l from each
other.

Figure 7.1 Visualizing Buffon’s method for calculating
.

When throwing the needle, there are two possibilities: either the needle crosses one
of the lines or it doesn’t. The probabilities associated to each of those outcomes are,
          241

respectively, p0 = 1 - 2 r  , and p1 = 2 r , with r
l  d and 
1 
.

Let N1 be the number of times that the needle crosses a line out of n throws. Then
we can estimate p1 using p 1
N1  n and consequently obtain . Therefore,
=1/.

  #
  
 

The previous experiment has been performed on several occasions, and it has
always required thousands of throws to achieve a precision better than two digits.
Nowadays we use simulations where the throws are replaced by computer-
generated (pseudo)random numbers.
Simulation-based procedures, usually known as Monte Carlo methods, enable us to
calculate
probabilistically in many different ways.
A very simple example to approximate the value of
is to first draw a circle of
radius 1 circumscribed by a square with sides of length 2 and then to drop a ball m
times onto the square counting how many times (n) the ball falls inside the circle.
Assuming that the ball can land with equal probability on any point in the square,
the probability that it will land inside the circle is given by the ratio between the
area of the circle and that of the square, that is, mn 
r2  l2 , since l
d
2, then
4 n/m
.
 The above experiment can be simulated with Mathematica. We just need to generate m pairs
of real numbers between 1 and 1 and count how many of them “land” inside a circle with
radius r
1 (we can use
 to check that the pairs verify x2  y2 <1 and count with
$  how many they are).


 pib
m : Module
data, n ,
data  RandomReal
 1, 1 , m, 2 ;
n  Length
Cases
data, x, y ; x ^ 2 y ^ 2  1;
4. n  m
 The method converges very slowly.


 pib
100 000

 
 3.14408

 The previous function can be extended to display the results graphically. Notice that text
needs to go between quotation marks and that inside a string of text we can use \t or \n to
insert a tab or a new line, respectively.


 Manipulate
Module
data, n ,
data  RandomReal
 1, 1 , m, 2 ;
n  Length
Cases
data, x, y ; x ^ 2 y ^ 2  1;
Text  Style
Column
Graphics

PointSize
0.006, Orange, Disk
0, 0 , 1, Black, Point
data ,
Row
"inside: ", n, "\toutside: ", m n, "\ntotal: ", m,
"\t
 4 ", n, "", m, "  ", 4. n  m  , "Label",
m, 100, "Random pairs generated" , 1, 10 000,
1, Appearance  "Labeled" 
242      

To find out more about Buffon’s method you can visit:

http://demonstrations.wolfram.com/BuffonsNeedleProblem
Throwing Buffon’s Needle with Mathematica by Enis Siniksaran: https://www.mathematica-
journal.com/2009/01/12/throwing-buffons-needle-with-mathematica/

7.5 Application: Are the Decimal Digits of
Random?

It doesn’t matter what method we use, if it’s correct, we will always get the same
sequence of
decimals even though such a sequence will never exhibit any
periodicity since
is an irrational number (actually a transcendental one) and,
therefore, it cannot be expressed as the ratio of two integers. However, if we
explore a long enough sequence of its decimals, we will discover an interesting
behavior: the frequency in which the digits appear is similar, that is, the numbers 0,
1, 2, ..., 9 will appear approximately with the same probability (this characteristic
doesn’t depend on expressing
in base 10. The same behavior happens regardless
of the chosen base). Since this property is the same that random numbers originated
from a discrete uniform distribution have, unless we knew in advance that certain
digits were
decimals, we wouldn’t be able to differentiate them from random
numbers generated by a roulette with 10 positions, all equally likely. Numbers that
have this property are called normal numbers. In the case of
, at least considering
only the digits computed until now, it’s been checked that it is normal but it has not
been proved that it would be normal for an infinite number of digits although
everything suggests that it would.
 To convert the digits of
to list elements, we use the function RealDigits to decompose the
decimal approximation of
into a list of numbers. For example: 3.14159... becomes:


 RealDigits
N

 

 
 3, 1, 4, 1, 5, 9, 2, 6, 5, 3, 5, 8, 9, 7, 9, 3, 1

 The 1 at the end indicates that the first element corresponds to the integer part. If we just
want the decimal part we can subtract 3 from 3.14159... : 3.14159 ...  3 =0.14159. For
example, the first 10 decimal digits of
are:
          243


 RealDigits
N

3, 10 1

 
 1, 4, 1, 5, 9, 2, 6, 5, 3, 5

 The following expression enables us to find a specific
digit in a short time. In the example
below we find the 10,000,000th digit.


 Timing
First
RealDigits
Pi, 10, 1, 10 ^ 7

 
 9.875, 7

 This short expression tells us how often numbers 0–9 appear in the first 100,000 decimals of

.


 Sort
Tally
RealDigits
N

3, 100 000 1 
0 9999
1 10 137
2 9908
3 10 025
4 9971
  
5 10 026
6 10 029
7 10 025
8 9978
9 9902

The previous output seems to confirm that in base 10 all digits are present with
approximately the same probability. This indicates that they are distributed
following a random discrete uniform distribution.
 To test this hypothesis for the first 10,000
decimals we use a combination of the functions
DistributionFitTest and DiscreteUniformDistribution.


 vals  RealDigits
N

3, 10 000 1 ;


 h  DistributionFitTest
vals,

DiscreteUniformDistribution
,  , "HypothesisTestData";
h
"TestDataTable", All 

  
 

      

P-Values above 0.05 (as in this case) suggest that the data come from a discrete uniform
distribution. If this were true then
would be normal.
 The command below returns the frequency of numbers 0–9 as a function of the total number
of digits used in the calculation.


 Manipulate

BarChart
BinCounts
RealDigits
N

3, 10 000 1, 1 ;; u , 0, 10, 1 ,
ChartLabels  Range
0, 9,
u, 100, "Total number of digits" ,
1, 10 000, 1, Appearance  "Labeled" 
244      

In a discrete uniform random distribution, all number combinations are possible.

However, the probability of some of them happening is very small. For example:
the probability that the same number, in this case any digit between 0 and 9, will
repeat itself is very small, although sometimes it happens.
Let’s create a command to identify when that happens with the digits of
.
 Split is very useful to divide lists into sublists consisting of identical elements.


 Split
RealDigits
Pi, 10, 100 1 

 
 3, 1, 4, 1, 5, 9, 2, 6, 5, 3, 5, 8, 9, 7, 9, 3, 2, 3, 8,
4, 6, 2, 6, 4, 3, 3, 8, 3, 2, 7, 9, 5, 0, 2, 8, 8, 4, 1, 9,
7, 1, 6, 9, 3, 9, 9, 3, 7, 5, 1, 0, 5, 8, 2, 0, 9, 7, 4, 9,
4, 4, 5, 9, 2, 3, 0, 7, 8, 1, 6, 4, 0, 6, 2, 8, 6, 2, 0, 8,
9, 9, 8, 6, 2, 8, 0, 3, 4, 8, 2, 5, 3, 4, 2, 1, 1, 7, 0, 6, 7

 Notice that some sublists contain more than one element. This happens when a digit repeats
itself once or more consecutively. With the next function we can count the size of each
sublist:


 Length  Split
RealDigits
Pi, 10, 100 1 

 
 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1,
1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1
We see that in 6 cases a digit repeated itself once, and there were no instances of a number
appearing consecutively 3 or more times.
 The next function counts the number of sublists with 1, 2, ..., n terms. Using Drop we
eliminate the first element since it corresponds to the number of empty sublists. The
function is then applied to the first 106 digits of
.


 data1 
Drop
BinCounts
Length  Split
RealDigits
Pi, 10, 1 000 000 1 , 1;
 We present the results in table format: 1 indicates that a digit does not appear in consecutive
positions, 2 that the same digit appears in two consecutive positions and so on.

  TableForm
Table
i, i, Length
data1 , data1 ,
TableHeadings  "repetitions", "frequency" , None 
 
  
repetitions 1 2 3 4 5 6 7
frequency 809 776 81 256 7976 830 77 12 1
          245

 By adding Position we can identify the order in which a certain repetition happens. For
example, to check the positions in which the sequence {9,9} appears within a 100 decimal
digits approximation of
, we would type:


 Position
Split
First  RealDigits
Pi 3, 10, 100, 9, 9 
42
  
75
 To relate the sublist position to the number of digits, we proceed as follows:


 Part
Accumulate
Length  Split
RealDigits
Pi 3, 10, 100 1 ,
42, 75 1 1
 
 44, 79

 This result indicates that the sequence 99 appears first in the 44th and 45th digits and the
next one in the 79th and 80th ones. We can check it using the table below. In it,
decimals
are grouped by rows containing 10 digits each.


 TableForm
Partition
First  RealDigits
Pi 3, 10, 100, 10,
TableHeadings  Range
10, Range
10 
 
  
1 2 3 4 5 6 7 8 9 10
1 1 4 1 5 9 2 6 5 3 5
2 8 9 7 9 3 2 3 8 4 6
3 2 6 4 3 3 8 3 2 7 9
4 5 0 2 8 8 4 1 9 7 1
5 6 9 3 9 9 3 7 5 1 0
6 5 8 2 0 9 7 4 9 4 4
7 5 9 2 3 0 7 8 1 6 4
8 0 6 2 8 6 2 0 8 9 9
9 8 6 2 8 0 3 4 8 2 5
10 3 4 2 1 1 7 0 6 7 9
 The function SequencePosition can be used to find the position where the digit n will be
repeated r consecutive times.
In the example below, using the first 10,000,000 digits of
, we find the cases in which the
number 9 appears consecutively exactly 7 times.

  digitsPi  First
RealDigits
N
Pi, 10 ^ 7;


 SequencePosition
digitsPi, PadRight
9 , 7, 9

1 722 777 1 722 783
3 389 381 3 389 387
 

4 313 728 4 313 734
5 466 170 5 466 176
According to the output in the first 10,000,000 digits of
in 4 occasions the number 9
appears consecutively 7 times. The first occurrence starts at the 1,722,777-th digit (or
1,722,776-th digit after the decimal point) and finishes at position 1,722,783-th.
We may be tempted to believe that any possible combination of numbers, no matter
how unlikely, will eventually happen as long as we use enough digits. For example,
the most recent
approximations have shown that there’s a sequence in which 0 is
repeated 12 consecutive times and the same happens with the numbers 1–9. There’s
even a position in which 8 appears repeated 13 consecutive times. The frequency of
occurrence is consistent with a discrete uniform distribution: the probabilities, in
base 10, that 0 will appear consecutively once, twice, or 12 times are: 1/10, 1/102
246      

and 1  1012 , respectively. This means that, using a
approximation of 1012 digits,
we’d most likely see one sequence of 12 consecutive 0s. The same could be said
for other sequences.
 In the example below, we look for the pattern 3131313 in the first then million digits of Pi:


 SequencePosition
digitsPi, 3, 1, 3, 1, 3, 1, 3 

 
  3 662 426 3 662 432 

These types of patterns can lead us to a variant of what Borges proposed in his
famous work The Library of Babel. The text describes a library containing all the
books ever written and the ones that may be written in the future (actually, Borges
mentions the number of pages and the number of lines per page but this is
irrelevant for the purpose of our argument). Let’s suppose that we could write a
book where all its letters were a, followed by one containing only b’s and so on
until writing a final book with all z’s. This way we would cover all the possible
combinations of letters. The resulting number would be huge, immensely bigger
than the number of atoms in the universe. Nevertheless it would be a finite number.
A variant of the previous story consists of replacing the order of the letters with
probabilities. Let’s supposed that we have a roulette with 35 positions representing
the 26 letters of the English alphabet plus certain symbols such as punctuation
marks, whitespace characters, etc. If we played such roulette enough times, we
would end up not only with every possible word, sentence, and paragraph but also
with every possible book! The only problem would be the time and the space to
store the results. We can speculate that the same applies to
. For example, if we
express it in base 35 and assign the digit 1 to the letter a, the digit 2 to b and so on,
with a number of digits big enough we will find the library of Babel (it doesn’t
matter how big the number needs to be since
has an infinite number of digits).
This means that all books: past, present, and future, have already been written and
they are in the decimals of
.
Look at: https://community.wolfram.com/groups/-/m/t/1623981
Surprisingly, in contemporary physics when we speculate about the existence of
parallel universes, we arrive at what we might consider an extreme version of the
previous speculation. If there is an infinite number of universes, all the possibilities
that don’t violate the laws of physics occur. Therefore, right now there’s someone
identical to you doing exactly the same as you’re doing. According to this theory
all the variants of your life will happen although in different universes.

7.6 The Strange Connection

is a transcendental number. This means that a finite polynomial of the form a0 +
a1 x + a2 x2 + ... + an xn = 0, with the coefficients a0 ... an being rational numbers (or
integers), in which
is one of its roots (solutions) doesn’t exist. As a result, it can
be proven that constructing a square with the same area of a given circle by a finite
number of steps only using a compass and a straightedge, also known as squaring
the circle, is not possible. There are other transcendental numbers with  being
probably the second best-known one after
.
          247


 N
, 100

 
 2.7182818284590452353602874713526624977572470936999595749669676277240766303535475
94571382178525166427
 As with
, the number  can be computed using several different methods (Click on + in the
next cell).


 When is expressed in decimal form, something funny happens (we use
NumberForm combined with ExponentFunction to avoid the output being displayed in
scientific notation).


 NumberFormNPi 163
, 30, ExponentFunction  Null &
 
!  
262537412640768743.999999999999
We can see that the first 12 decimal digits are all 9s. This could give the false impression
that all the decimal digits are 9, but that is not the case. It would be enough to increase the
precision by one more digit to check that it is only a lucky coincidence.
In a separate category we include imaginary numbers, those whose square is a
negative number as in the solution to the equation x 2  1
0.
 Notice that to express imaginary numbers, Mathematica uses the symbol  and not the Latin
letter i.


 Solve
x ^ 2 1  0, x

 
 x , x 

Numbers consisting of a real part and an imaginary one are called complex
numbers. As a matter of fact, all real numbers can be considered a special case of
complex numbers where the imaginary element is 0. Sometimes strange
connections exist among different numbers as in the case of the following equation
discovered by Euler:
 x  cosx   sinx  x  

This connection between different fundamental constants cannot be random. If we

assume x

, the previous expression becomes:
 
1

We can see that in just one expression there are two transcendental constants and
the imaginary unit connected to each other and, the most surprising thing, the result
is the integer  1. It’s hard to believe that all these strange connections between
numbers are random.
1
Pi is also connected to one of Riemann’s Zeta functions: (s) = 
i
1 ks
, which we
will discuss in further detail in the next section. We can see that % [2] converges
to:
248      


 Zeta
2

2
 

6

also appears in fundamental notions in physics such as Heisenberg’s uncertainty
principle or Einstein’s general relativity theory. It seems that
is much more than
the ratio between the perimeter of a circumference and its diameter!

7.7 The Riemann Hypothesis

We’ve seen in the previous section that
, at least according to its latest
approximations using billions of digits, is probably a normal number. However,
there’s no proof that this is the case if we consider an infinite number of digits. This
is an open problem in mathematics although not the only or most important one.
David Hilbert, one of the most influential mathematicians of the 19th and 20th
centuries, at the Paris conference of the International Congress of Mathematicians
in 1900 gave a task to mathematicians for the 20th century: the resolution of 23
problems (ten given in the conference and the rest published later). Although most
of them have been already solved, the most famous unsolved one is the Riemann’s
hypothesis about which Bernhard Riemann himself stated that if he were to
resuscitate after 500 years, the first question he would ask would be whether the
“damn hypothesis” had been proven or not. This problem remains unsolved and is
one of the seven “Millennium Problems”, a list of problems named by the Clay
Mathematics Institute that carry a prize of 1 million dollars for their complete
mathematical solution.
The Riemann’s Zeta function definition is:
1
Let s
a  b i   , for Re s 1, s  
k
1 ks
.
http://mathworld.wolfram.com/RiemannZetaFunction.html


 ClearAll
"Global`"

 Given that s is a complex variable, (s) is also complex. We can visualize its real component
Re s and its imaginary one Im s with the following Mathematica command:


 Plot3D
Re
Zeta
a b I , a, 10, 10 , b, 10, 10 , Mesh  None,
Plot3D
Im
Zeta
a b I , a, 10, 10 , b, 10, 10 , Mesh  None

 
 ,
          249

 It can be shown (using expansions of analytic functions, a complex analysis concept), that
for negative even values, that is when s
 2 n ,  s
0. We can check it for some cases
(in the example below when n ranges from 1 to 100 in steps of 10).


 Table
Zeta
2 n , n, 1, 100, 10 

 
 0, 0, 0, 0, 0, 0, 0, 0, 0, 0

 Graphically we can see that by plotting s
a  b i with b
0, s
0 for

a
2, ..., 2 n:


 Plot
Re
Zeta
a , Im
Zeta
a  , a, 30, 2 , PlotLegends  "Expressions"

0.03

0.02

0.01
Re


 

30 25 20 15 10 5 Im


0.01

0.02

0.03

 Another way to visualize this is by plotting 1/ (s), with s
x  y . Remember that
x+y = x2  y2 . Using the inverse expression, the zeros become asymptotes that look like
columns. In the graph below, we can see these “columns” appearing when s
 2 n.
(This function and several others that follow are based on the ones described by Stan Wagon
in his book Mathematica in Action).

  Plot3D
1  Abs
Zeta
x  y, x, 20, 2 , y, 1, 1 ,
MeshFunctions  3 &, MeshStyle  Blue, Mesh  10,
PlotPoints  60, MaxRecursion  3 ,
Boxed  False, BoxRatios  5, 2, 2 , AxesLabel  "x", "y", None ,
Ticks  Automatic, Range
22, 2, Range
0, 4 

 

250      

The  2 n
0 zeros, are named     .

 Besides the trivial zeros, there are other complex values of s, with 0 Re s 1, for
which the zeta function becomes also zero. These zeros are called      . The
Riemann’s hypothesis states that all these non-trivial zeroes are located in the line
1
s
2  b  (known as the critical line) or put it in other words, that the real part of every non-
trivial zero in the Riemann’s Zeta function is 1/2. The graph below shows that this is true for
the chosen interval.


 Plot
Re
Zeta
1  2 b I , Im
Zeta
1  2 b I  ,
b, 30, 30 , PlotLegends  "Expressions"
3

1
1 Re    
2
 

1
Im    
2
30 20 10 10 20 30

1

 In Mathematica the function ZetaZero
k represents the k-th zero of s in the critical
1
line s  2
 i b with the smallest imaginary part. For example the first zero corresponds to:


 s  N
ZetaZero
1, 15

 
 0.5000000000000  14.1347251417347 

 We check that for that value  s
0.


 Zeta
s  Chop

 
 0

 We’ve seen that the first zero is approximately at 14.1347251417347 . If we would like to
visualize all the existing zeros for a given interval, we can use the following function in
terms of the imaginary component (remember that the real component is always 1/2).


 Manipulate
Plot
RiemannSiegelZ
t, t, 0, k , Epilog  PointSize
.02, Red,
Point
Table
Im
ZetaZero
n, 0 , n, k  , k, 50, 100 
          251

 Next, we plot 1/ (s) below. We can see that all the zeros (columns) are clearly aligned.


 Plot3D
1  Abs
Zeta
x  y, x, 4, 5 , y, 10, 40 ,
MeshFunctions  3 &, MeshStyle  Blue, Mesh  10,
PlotPoints  60, MaxRecursion  3 , ViewPoint  8, 1, 3 ,
Boxed  False, BoxRatios  5, 10, 2 , AxesLabel  "x", "y", None ,
AxesEdge  1, 1 , Automatic, Automatic ,
Ticks  Automatic, Range
0, 30, 10, Range
0, 4 ,
ClippingStyle  None, PlotRange  0, 5 

 


As of the end of 2021, all the non-trivial 1013 zeroes that have been found lie on the
critical line. However, in mathematics, this doesn’t count as a proof. Of all the
approaches that have been proposed to prove the Riemann hypothesis, one of the
most curious ones comes from D. Hilbert and G. Pólya (proposed independently).
They suggested that the zeros of the Riemann Zeta correspond to the eigenvalues of
a linear operator given by an infinite matrix. In 1972 the mathematician Hugh
Montgomery and the physicist Freeman Dyson went even further and proposed a
connection between the zeroes and heavy nuclei energy levels.
We can model these energy levels using what is known as the Spectrum of
Riemannium:
https://www.americanscientist.org/article/the-spectrum-of-riemannium
In Mathematica, we can explore these two theories using random Hermitian
252      

matrices as described in:

https://www.wolfram.com/language/11/random-matrices/zeros-of-the-riemann-zeta-
function.html
However, few mathematicians and scientists believe that there’s a connection.

7.8 Looking for the Magic Prime Formula

One of the main characteristics of the Zeta function is its connection with prime
numbers. Remember that a prime number p is an integer greater than 1 that has no
positive divisors other than 1 and itself. The set of prime numbers (that since Euclid
we know is infinite) is called .
1
It can be shown that s  
k
1 ks
, can also be computed using Euler’s product.
Euler' s product


 WolframAlpha
"Euler's product",
"NamedMathematicalFormulas", 1 , "ComputableData" 
 1  1
 
 Hold !1  pk s   ; Res 1", Hold s  ! ; Res 1  pk  pk "
k
1 s k
1 1  pk s

 That is:
1 1 1 1
s
1  1 1 
!
s s s
2 3 5 p 1  ps

 Prime numbers seem to be randomly distributed. For a given prime number, there’s no
general formula that will give us the next one.


 Table
Prime
n, n, 45, 60 

 
 197, 199, 211, 223, 227, 229, 233, 239, 241, 251, 257, 263, 269, 271, 277, 281

However, there are algorithms to obtain good approximations of the prime

counting function. The prime counting function is the function, notated
(n), that
returns the number of primes less than or equal to a natural number n. In
Mathematica, we refer to it as PrimePi[n].
Prime Counting Function: https://mathworld.wolfram.com/PrimeCountingFunction.html
 The graph below shows how many prime numbers are less than or equal to an arbitrary
number n. We denote this function
(n) (in the example n
200).
          253


 Plot
PrimePi
n, n, 1, 200 , AxesLabel  "n", "
n" 

n

40

30

 


20

10

n
50 100 150 200



  gives the option to choose an specific algorithms. The time of computation for
some of then are shown below


 BarChart
Table
First
AbsoluteTiming
PrimePi
10 ^ n,
First
AbsoluteTiming
PrimePi
10 ^ n, Method  "LMO", First

AbsoluteTiming
PrimePi
10 ^ n, Method  "Meissel" , n, 7, 11, .5 ,
ChartLegends  "Automatic", "Lagarias Miller Odlyzko", "Meissel" 

0.20

0.15

Automatic
 
 0.10 LagariasMillerOdlyz
Meissel

0.05

0.00

The trend is that the frequency of prime numbers decreases as n gets bigger.
There are several approximations to calculate
(n), such as: t/Log(t), or
z
lin
#0 d t  log t (in Mathematica LogIntegralz), but the best one is done by
using Riemann’s Zeta function. Specifically, the program uses the function R n =
RiemannRx .
1n
For x 0, Rx  
n $n li x  n.
 The t/Log(t) approximation can be found using the asymptotic term for PrimePi at Infinity:


 Asymptotic
PrimePi
t, t  Infinity
t
 

logt
 In the following plot, we compare the real value of
(n) with all these approximations.
254      


 Manipulate
Plot
PrimePi
n, n  Log
n, LogIntegral
n, RiemannR
n ,

n, 1.5, p , PlotLegends  "Expressions", p, 10 000 , 100, 10 ^ 6 

 We can visualize the comparison in an even better way by representing the difference
between each of the alternative approximations and
(n). In the case of R(n), the
approximation is very accurate since, at least in the chosen interval, the error never gets too
big (actually it never exceeds the real value by more than 3).


 Plot
n  Log
n PrimePi
n, LogIntegral
n PrimePi
n,
RiemannR
n PrimePi
n , n, 200, 1000 , PlotLabels  "Expressions"

10 lin 
n
5
Rn 
n
400 600 800 1000
5
  
10

15
n
20 
n
logn
25

7.9 Additional Resources

A compilation of different methods used to compute
, their Mathematica
implementations:
http://mathworld.wolfram.com/Pi.html and http://functions.wolfram.com/Constants/Pi
The Wikipedia article https://en.wikipedia.org/wiki/Pi is an excellent resource in
terms of both its contents and its links.
Regarding the Riemann hypothesis see for example: (3Blue1Brown)
https://www.youtube.com/watch?v=sD0NjbwqlYw
8
Looking at the Sky

This chapter gives a brief introduction to astronomy using some of the

functions available in Mathematica. It shows how to get information
about planets using entities, generate solar analemmas, and even create
sky charts for a given location among other things. It also explains the
differences between the Julian and the Gregorian calendars and why,
contrary to popular belief, Miguel de Cervantes and William Shakespeare
did not die on the same day. Applications related to determining the color
of the stars and measuring distances across the universe are also covered.
The last section discusses how to model the orbits of eclipsing binary
stars by creating and analyzing light curves, very useful when searching
for exoplanets.

8.1 A Short Astronomical Walk

Contemplating the sky on a starry night, away from light pollution, is one of the
most fascinating shows that we can enjoy. It’s not strange then that for millennia
humans have looked up at the sky and realized that the sun and the moon follow
regular cycles. The majority of ancient cultures came up with mythological or
religious explanations about celestial objects, but they also used those objects for
practical purposes such as the creation of calendars to schedule harvests. For
example, the Egyptians associated the appearance of Sirius, the brightest star in the
norther hemisphere (now we know it’s a binary system), with the annual flooding
of the Nile. Despite the reduced number of celestial bodies that can be seen with
the naked eye, it’s surprising how much knowledge of the sky that astronomers
already had before the invention of the telescope.
The arrival of the telescope about 400 years ago (1608) heralded a revolution: the
number of known stars went from thousands to millions, the moon, far from being
a perfect sphere, was found to be covered with mountains and craters, the
wandering stars (the planets) were more than mere bright dots, they actually had
satellites, and Saturn had rings (it’s not the only planet). Later on we learned to
256      

analyze the composition of the light coming from space. Many of the objects that
looked like blurry spots under the telescope became galaxies with myriads of stars.
It was discovered that those galaxies were moving apart from each other: the
universe was expanding, an expansion that started around 14 billion years ago. It
turned out that the visible stars and galaxies contain less than 5% of the mass of the
Universe. The rest is dark energy and matter, we know it exists (at least the dark
matter) but we don’t really know what it is. The idea of the existence of dark
energy (“a riddle wrapped inside a mystery”) started with an amazing discovery
(1998): The expansion of the universe is accelerating, a discovery that was worth a
Nobel Prize in 2011. And the discovery race continues, promising new surprises:
the important thing is still the journey.
For astronomical calculations, Mathematica has several functions: Comet (or
CometData) , Exoplanet (or ExoplanetData), Galaxy (or GalaxyData), MinorPlanet
(or MinorPlanetData), MoonPosition, Sunset, Nebula (or NebulaData), Planet (or
PlanetData), Pulsar (or PulsarData), SiderealTime, StarCluster (or
StarClusterData), Star (or StarData), SunPosition, Supernova (or SupernovaData),
Sunrise and many more. These functions replace and extend AstronomicalData
although you can still use it.
The syntax of all these functions is similar to that of other computable functions.
Example: Function
"name", "property" where "name" or entity (Entity) is an
object and "property" is a characteristic or parameter of the object.

For example: to refer to the solar system planets we can type either PlanetData[] or:


 EntityValue
"Planet", "Entities"

 
  Mercury , Venus , Earth , Mars , Jupiter , Saturn , Uranus , Neptune 

If we place the cursor on one of the list elements in the output, for example "Mercury",
Mathematica displays Entity
"Planet", "Mercury". This tells us that "Mercury" belongs to
the "Planet" entity.

We can get the available properties for planets using the free-form input: Insert  Inline
Free-format input and then type “planet” or “planets”. Also, in this example and the
following one, we use Shallow or Short to avoid getting the complete output. If you want to
see all the properties just remove Shallow when executing the command:


 planets PLANETS
"Properties"  Shallow

 
 

 absolute magnitude H , age , albedo , alphanumeric name ,

alternate names , altitude , next maximum altitude , angular diameter ,

angular radius , largest distance from the Sun , 125

The next output informs us about the property "OrbitPeriod" for each one of the planets.
Notice the use of "EntityAssociation" to link names with properties.
  !   ! 257


 planets PLANETS
"OrbitPeriod", "EntityAssociation"

 
  Mercury  87.96926 days, Venus  224.70080 days,

Earth  365.25636 days, Mars  1.8808476 a, Jupiter  11.862615 a,

Saturn  29.447498 a, Uranus  84.016846 a, Neptune  164.79132 a

Here we show Mercury’s average orbit velocity writing Mercury average velocity or
EntityValue[Entity[“Planet”,”Mercury”],”AverageOrbitVelocity”]


 Mercury PLANET
     

 
 47.4 kms

Sometimes, we may want to get the value without its corresponding unit. In those cases, we
can use QuantityMagnitude.


 QuantityMagnitude
Mercury PLANET
     

 
 47.4
The result is given in light-years (ly), a common unit to express distances between stars.

The parsec (pc), and its multiples (kiloparsec or kpc, and megaparsec or Mpc), are used to
measure very long distances such as distances between galaxies. Let's see the relationship
between a parsec and a light year.

   convert parsec to light year

Result

3.262 ly

 
 3.262 ly

The table showing the equivalence to 1 light-year in different units:


 Pane
TableForm
Map
, N  UnitConvert
Quantity
"LightYears",  &,
"Meters", "Parsecs", "SIBase", "SI", "Imperial", "Metric" , 200

Meters 9.46073  1015 m

Parsecs 0.306601 pc
SIBase 9.46073  1015 m
 

SI 63 241.1 au
Imperial 5.87863  1012 mi
Metric 63 241.1 au

For some properties there may be no information available. Remember that with
DeleteMissing we can remove those properties whose headings contain the word “Missing”.


 DeleteMissing
Plutoids
"Density", "EntityAssociation"

 
  136199 Eris (2003 UB313)  2.53 gcm3 , Pluto  1.860 gcm3

Some of the functions related to astronomy need the precise location for which the
information is desired. By default they will use our current time and position.

Example: The following function returns the sunrise time on February 28, 2023, at a place
located 40º N and 10º W, a location that is one hour ahead of GMT as we can see.
258      


 Sunrise
GeoPosition
40, 10, DateObject
2023, 03, 01, 0, 0

 
 Minute: Wed 1 Mar 2023 07:54 GMT
2

Dates and hours are of crucial importance in astronomy. The existence throughout
history, even nowadays, of different calendars makes it difficult to compare dates
of astronomical and historical events. To avoid this problem we use Julian dates. A
Julian date (not to be confused with the Julian calendar: Julian year) indicates the
days that have gone by since January 1, 4713 B.C. at 12:00 PM UT (November 24,
4714 B.C., in the proleptic Gregorian calendar). The name of the Julian date comes
from Julius Scaliger (not Julius Caesar) who proposed it in 1583 after realizing that
such a day was the least common multiple of 3 calendar cycles used by the Julian
calendar. Using 12:00 PM as the starting time instead of 12:00 AM has the
advantage that astronomical observations, usually done at night, are recorded on the
same day. Starting with Mathematica 10.2 the function JulianDate is available. The
function converts dates from our calendar, the Gregorian one, to Julian dates. It
takes into account the modifications to the Julian calendar done by Gregorio XIII in
1582 in most Christian countries that improved the handling of leap years, but in
which the year 0 doesn’t exist.

In England and other countries, the Gregorian calendar was introduced at a later date. This
fact makes statements such as that Shakespeare and Miguel de Cervantes died on the same
day incorrect.


 dateSh  William Shakespeare PERSON
  "  

 
 Tue 23 Apr 1616 Julian calendar


 dateCe  Miguel de Cervantes PERSON
  "  

 
 Fri 22 Apr 1616

Actually, it is not clear when Cervantes died. Although April 22 is the most frequently used
date, the parish register shows April 23. In any case, Cervantes’s and Shakespeare’s deaths
happened 10 days apart. The reason for the confusion is that in England they were still using
the Julian calendar. To compare both dates, we need to convert Shakespeare’s death date to
the Gregorian calendar.


 CalendarConvert
dateSh, "Gregorian"

 
 Tue 3 May 1616

We can alternatively find out the Spanish writer’s death date using the Julian calendar and
calculate the difference in days between both deaths.


 JulianDate
dateSh JulianDate
dateCe

 
 11.

We can now use the above mentioned approach to convert the current date to a Julian date.
We have added Dynamic to dynamically update the calculation with an interval defined with
UpdateInterval, in this case 1 second.
 !   ! 259


 Dynamic
AccountingForm
JulianDate
Date
, 12, UpdateInterval  1

 
 2459870.19238

We can stop the dynamic updating by unchecking it in the Evaluation menu:

Evaluation  Dynamic Updating Enabled.
Star positions in the sky don’t follow an annual cycle exactly. They are subject to
perturbations due to their own motions and other celestial movements, such as the
precession of equinoxes. Because of this, it’s necessary to regularly set up dates to
use when referring to celestial coordinates. These standard reference dates are
known as epochs. The last one was set on January 1, 2000 at 11:58:55.816 AM.
When referring to this epoch we use J2000.0. StarData and other astronomical
functions refer to the position of stars using this latest epoch without taking into
account either the precession of equinoxes or the stars’ own motions. For practical
purposes, these adjustments are irrelevant, but we need to consider them when
making historical projections.

8.2 Solar Analemma

The curve describing the sun and the visible planets in the sky every day of the year
always observed at the same location and time of the day is called analemma.
A solar analemma can be obtained with a camera and patience, taking pictures of
the horizon every day or almost every day of the year from the same position, with
the same exposure and at the same solar time. The result will be a composition
similar to the one below. It shows an analemma taken by Jack Fishburn between
1998—1999 from Bell Laboratories in Murray Hill, New Jersey:


 WikipediaData
"Analemma", "ImageList"3

 

260      

You can also make analemma for other celestial objects such as Venus or Mercury.
With less patience and without having to wait for a year, we can simulate an
analemma with Mathematica from our own computer.

We need to define the location (latitude and longitude).


 salamanca  GeoPosition
Salamanca CITY 

 
 GeoPosition
40.97, 5.67
SunPosition returns the sun position from our position (azimuth and latitude) at 9:00 AM in
GMT (choosing TimeZone0) on January 1, 2016.


 SunPosition
salamanca, DateObject
2016, 1, 1, 9, TimeZone  0

 
 132.99, 10.15

We apply the same function to get the position every 5 days, at the same time, to complete a
year:


 Graphics
Orange, PointSize
Large,
Point  Map
QuantityMagnitude, SunPosition
salamanca,
DateRange
DateObject
2016, 1, 1, 9, 0, TimeZone  0,
DateObject
2016, 12, 31, 9, 0, TimeZone  0, 5

"Values", 2, Frame  True
45

40

35

30

 

25

20

15

10
100 110 120 130

8.3 Stargazing


 Clear
"Global` "

An incredible experience is to look at the sky away from light-polluted cities,

trying to imagine the hidden celestial mechanism generating the incredible view.
We will soon realize the greatness of those night owl geniuses (Hipparchus,
Ptolemy, Copernicus, Brahe, Kepler, and others) that established the foundations of
modern astronomy without any other means than their plain sight and basic
instruments such as the astrolabe to measure the stars’ positions. Everything
changed in 1609 when Galileo used the telescope for astronomical purposes for the
first time. Another equally important invention was the pendulum clock that
8. Looking at the Sky 261

enabled the measurement of time with great precision. Centuries later, telescopes
mounted on satellites and atomic clocks heralded a new revolution.

8.3.1 Naked-Eye Astronomy

When looking at the sky, almost all of the objects that can be seen with the naked
eye are stars. Throughout history, different cultures imagined that some stars
resembled figures, named constellations. The same phenomenon occurred in
civilizations that were not connected in any way. For example: The seven brightest
stars from the Ursa Major were interpreted the same way in very different places.
The inhabitants of the British islands said that those stars were the legendary King
Arthur’s cart. For the Germans, they represented a wagon pulled by three horses.
The Greeks came up with a more imaginative story: In a Greek legend, the god
Zeus and the mortal Callisto had a son named Arcas. Hera, Zeus’s jealous wife,
turned Callisto into a female bear. Arcas, unaware that the bear was actually her
mother almost killed her. Afterward, Zeus also turned Arcas into another female
bear. Callisto is Ursa Major and Arcas is Ursa Minor. Greek mythology is full of
such stories to explain almost all the stars that appear near each other in the sky.
Many of those myths have actually given names to the constellations. Over the
centuries, many constellations were added until they eventually covered the entire
visible sky from any place on Earth. There are 88 constellations in total. In 1930,
the International Astronomical Union established the limits of each of them.
Obviously, they are imaginary lines covering the entire firmament, including both
hemispheres. As in maps, we don’t need to keep all the constellations’ details. They
and the names of their most important stars are represented in a planisphere, a
celestial map that shows us the sky each night for a given date and time.

The function below returns the sky chart for our location at the time given in the input. We
can get Mathematica to create it for us by typing “Sky chart at 21 GMT 2022-02-15” in free-
form input.
262      




 SkyChart

DateObject
2022, 2, 15, TimeObject
21, 0, TimeZone  "GMT"

Moon
 
 E W

S

The program uses “SkyChart”, a function from the repository. To find out more details
about its functionality, we type:


 ResourceObject
"SkyChart"

Name: SkyChart »
 
 ResourceObject Type: Function
Description: Visualize the sky for a specified location, time and date

With the instruments and methods available to astronomers nowadays, the information that
we can obtain about stars is very extensive. In the next example, we show the properties
most commonly used when describing stars:


 properties  "AbsoluteMagnitude", "ApparentMagnitude",
"RightAscension", "Declination", "Altitude", "Azimuth",
"Constellation", "DistanceFromSun", "SpectralClass";

Let’s find those properties for Polaris, the North Star:


 Polaris STAR
properties, "PropertyAssociation"

 
 AbsoluteMagnitude  3.64, ApparentMagnitude  1.97, RightAscension  2h 31m 47.06s ,

Declination  89  15 ' 51. ", Altitude  40  23 ' 43. ", Azimuth  359  23 ' 2.  101 ",
Constellation  Ursa Minor , DistanceFromSun  431.217 ly, SpectralClass  F7:IbIIvSB

The apparent magnitude tells us how bright an astronomical object appears to an

observer regardless of its intrinsic brightness:
http://scienceworld.wolfram.com/astronomy/ApparentMagnitude.html
The smaller the magnitude, the bigger its luminosity. For the most luminous stars,
we use negative values. Most of the people looking at the sky with the naked eye
and with little light pollution just see stars with apparent magnitude of less than 5.
 !   ! 263

Here are the 10 closest stars to Earth (the Sun is excluded), many of them not visible to the
naked eye:


 Through
EntityList
EntityClass
"Star", "StarNearest10"

"Name", "DistanceFromEarth", "ApparentMagnitude"
Barnard's Star 5.93796 ly 9.54
Lalande 21185 8.30785 ly 7.49
Luyten 7268 A 8.56318 ly 12.57
Luyten 7268 B 8.56318 ly 12.7
Proxima Centauri 4.2465 ly 11.01
 

Rigil Kentaurus 4.39282 ly 0.01
Toliman 4.40311 ly 1.35
Sirius 8.59682 ly 1.44
Sun 1.01234 au 26.72
Wolf 359 7.79346 ly 13.45
We can immediately notice that most of them are not visible since their apparent magnitude
is bigger than 5. With the exception of the Sun, the star with the biggest apparent magnitude
is Sirius, a star that we can see during winter nights (in the Northern hemisphere) next to the
Orion constellation. The closest one is Proxima Centauri that is only visible in the Southern
hemisphere (Proxima Centauri rotates around the double system Centaurus A and B in a
500,000-year period. By chance, it’s currently in its orbital position closest to the solar
system).

The number of stars visible without the use of technology is very small compared to the
number of stars in our galaxy. The command below generates a plot representing the naked-
eye stars and their distances to Earth in light years.


 ListPlot
Sort
EntityClass
"Star", "NakedEyeStar"
"DistanceFromSun",
AxesLabel  "stars", "ly"
ly

1500

   1000

500

stars
2000 4000 6000 8000

From the graph above we can see that almost all of the visible stars are less than 1,500 ly
away. Our galaxy, the Milky Way, has a width of approximately 100,000 ly. Therefore, with
our bare eyes we only see a very small fraction of the stars in our own galaxy, most of them
the ones closest to Earth. The previous figure refers to the number of potentially visible stars
without light pollution or moonlight. In practice, the actual number is 2,000 or 3,000, since
we only see part of the sky depending on our position.
Besides stars with apparent magnitude 5 or less, the only other celestial bodies
visible using only our eyesight are: the Moon, the 5 nearest planets and,
264      

sporadically, some comets. In the Northern hemisphere, the only object visible that
doesn’t belong to our galaxy is the Andromeda galaxy. In short, we see an
insignificant fraction of our own galaxy, which in turn is just one among billions of
galaxies.
We’ve seen that among the properties of Star (or StarData) and other astronomical
functions are: “RightAscension”, “Declination”, “Altitude”, and ”Azimuth”. These
properties are commonly used to indicate the position of a celestial body. In
equatorial coordinates, “RightAscension” and “Declination” are similar to latitude
and longitude, but they refer to the celestial sphere so they are independent of the
observer . Another type of coordinates widely used are the alt-azimuthal (alt/az)
that depend on the observer: The “Altitude” (alt) is the height of the star over the
horizon. It goes from 0º to 90º and has a positive sign for stars located above the
horizon and a negative sign for the ones located below it. The ”Azimuth” (az) is the
arc in the horizon measured counterclockwise from the South point until the
object’s vertical. Its value ranges from 0º to 360º.
For further details see: The Celestial Sphere, http://demonstrations.wolfram.com/TheCelestial-
Sphere, by Jeff Bryant.

The next example shows Sirius’s displacement on the first of each month at 22:00 h using
alt/az coordinates in local time.
Planet


 sirius  Table
QuantityMagnitude
EntityValue

Dated
Entity
"Star", "Sirius", DateObject
2022, i, 1, 22, 0, 0,
"Azimuth", "Altitude", "AngularDegrees", i, 1, 12
118.713 6.38653
143.138 23.8375
171.697 31.9130
205.412 28.4294
231.945 14.4112
253.045 6.00471
 

271.772 28.0331
296.751 50.3112
346.221 65.1931
46.9011 58.3531
78.9981 37.7519
98.8455 15.5399
 !   ! 265


 Graphics
Orange, Point
sirius, Frame  True,

FrameLabel  "azimuth", "altitude", AspectRatio  1  2

20

0
altitude

20
 

40

60

50 100 150 200 250 300 350

azimuth

As we can see, there are negative values that correspond to the period of the year in which
Sirius falls below the horizon and as a consequence is not visible. As a matter of fact, the
Egyptians considered that the appearance of the star marked the beginning of a new year.

In the previous example, we didn’t specify the observer’s location and the local time zone.
They can be shown:


 FindGeoLocation
, $TimeZone

 
 GeoPosition
40.97, 5.66, 2.

We can display the time in zone 0, using the Coordinated Universal Time (UTC)


 DateObject
Now, TimeSystem  "UTC", TimeZone  0

 
 Sun 22 May 2022 20:15:48 UTC

However, it’s important to keep in mind that the function actually gives us the position of
the server that we are using to connect to the Internet based on the IP address. Sometimes it
may be far away from our real position. This would be the case if, for example, we were
using a phone to surf the net. We can check it by showing the obtained position in a map:


 GeoGraphics

GeoVisibleRegionBoundary
FindGeoLocation
, GeoZoomLevel  12

 

266      

Naked-eye stars display an amazing regularity. It’s possible to forecast the location
of each star without prior knowledge of celestial mechanics. If we observe a star at
a certain hour, for example at 24:00 h, ignoring the displacement due to the
inclination of the Earth’s axis with respect to the ecliptic, the next day it will be on
the same spot 4 minutes earlier, that is at 23:56 h, completing a cycle in a year. The
explanation that the Greek civilization and others provided for this fact was to
assume that all the stars (which they called fixed, to distinguish them from the
planets or wandering stars that did not exhibit such behavior) were glued to a dome
that rotated with a daily cycle of 23 h and 56 min, the sidereal day (or stellar day).


 sidereal days

 
 1 sidereal day

We can use UnitConvert to find out the actual duration. When executing the command
above, the predictive interface will do the conversion automatically.


 UnitConvert
Quantity
1, "SiderealDays",
MixedUnit
"Hours", "Minutes", "Seconds"
 
 23 h 56 min 4.09054 s

However, over very long periods we can notice discrepancies that at first sight may
seem insignificant during a person’s lifetime but become quite obvious after several
generations. Hipparchus of Nice (c. 190 BC–c. 120 BC) compared the stellar maps
available at the time (celestial cartography is over 2,500 years old!) and realized
that there was a relative movement of the stars with respect to the ecliptic. This
movement is known today as the precession of the equinoxes. This is probably his
most famous discovery. The displacement happens when the Earth’s axis, moving
along a circumference with respect to the ecliptic, rotates with a period of 25,771
years.

Figure 8.1 Precession of the Equinoxes demonstration.

http://demonstrations.wolfram.com/PrecessionOfTheEquinoxes

As a result, the North celestial pole keeps on moving. Today it is close to the North
Star but 4,800 years ago it was pointing toward Thuban (Alpha Draconis). William
Shakespeare did not realize that, when in his play Julius Caesar stated: “But I am
constant as the northern star, Of whose true-fix’d and resting quality There is no
 !   ! 267

fellow in the firmament”.

In reality, all the stars seen from the Earth have a slow displacement motion as a
result of the precession and their own orbits in the galaxy. This last type of
movement is known as proper motion.

We can find the proper motion of the North Star as follows:


 Polaris STAR
##   

 
 11.75 masyr

 means milliarc seconds per year. It’s a value that cannot be observed during a
lifetime. We need to keep in mind that a telescope on the ground can rarely see details
smaller than 1 arc second. However, during long periods even the shapes of the
constellations change.
Astrology (a pseudoscience) attributed predictive powers to the constellations. For
example, they were supposed to determine the future of people born under them.
The firmament was divided into 12 signs corresponding approximately to the
number of lunar cycles in a year. Each division was assigned a symbol called a
zodiac sign. The names of the zodiac are associated with the first constellations
named by the Greeks during the 5th century BC, even though the Babylonians were
the first ones to mention them 4,000 years ago. The starting point was the
constellation pointing toward Aries (the moment when the Sun moves from the
south celestial hemisphere to the north one coinciding with the spring equinox) in
the 5th century BC. Today, we still use the same division although the
constellations are now in a different location to where they were 2,500 years ago.
The real astronomical zodiac dates correspond to the constellation located behind
the solar disk during the spring equinox, in the direction opposite to the Earth’s
direction (the Sun actually passes through 13 constellations and not 12, the 13th
one is Ophiuchus). The zodiac signs must be adjusted by a month to adapt to the
current astronomical reality, so chances are you may have to revise your zodiac
sign (although it wouldn’t be very useful).

The natural language input below returns the location of the Ursa Minor stars in the year
100,000 compared to their current position (dashed line). The rest of the stars experience
similar changes.
268      

The solar day (24 hours) is defined as the time it takes for the Sun to return to its
culmination point (its highest point in the sky). That is, the Earth in 24 hours spins more
than 360 º. In one average tropical year, the number of rotations is:

  1 QuantityMagnitude
UnitConvert
Quantity
1, "TropicalYears", "Days"

 
 366.2421887920

However, the periodicity of the Earth’s rotation is not constant, but it’s slowing
down. Because of that, the rotation is no longer associated to the definition of
seconds, as the base time unit. Now, the official definition is
(https://www.bipm.org/en/si-base-units/second ):
"The unperturbed groundstate hyperfine transition
frequency of the caesium133 atom, to be 9 192 631
770 when expressed in the unit Hz, which is equal to s1 ".
To keep time standards such as UT1, based on the Earth’s rotation, accurate it’s
necessary to add or subtract discrete leap seconds (normally once or twice a year).

Even though the UTC system, which is the standard “civil time”, is not based on tracking
the precise rotation speed of the Earth, there was still a discrepancy between it and the
International Atomic Time (TAI) standard on 2021-12-31:


 Wed 1 Dec 2021 00:00:00 UTC Wed 1 Dec 2021 00:00:00 TAI

 
 37. s

GeoOrientationData takes into account the measured rotation speed of the Earth. The
following code shows the deviation of the duration of the day over 86,400 seconds (24 h) for
the last 20 years:
 !   ! 269


 DateListPlot
GeoOrientationData
DateInterval
Now 20 yr , Now ,

"DayDurationExcess", TargetUnits  "Seconds"

0.002

0.001

 


0.000

0.001

2005 2010 2015 2020

8.3.2 Wandering Planets

Since antiquity, people have realized that there are a small number of stars
exhibiting a behavior different from the one of the fixed stars and their annual
cycles. These stars were once called wandering stars. Nowadays we know that they
correspond to the 5 visible planets: Mercury, Venus, Mars, Jupiter, and Saturn
(Uranus was not added to the list of planets until 1783 given how difficult it is to
observe it with the naked eye). To get information about planets, we use the Planet
entity (or the function PlanetData).

Here all available properties for Jupiter are shown (Click in “atmospheric composition”.)
270      


 Jupiter PLANET
"Dataset"

absolute magnitude H 9.4

age 4.5  109 yr

albedo 0.52
alphanumeric name Jupiter
alternate names the giant planet

altitude 49  3 ' 28. "

next maximum altitude 48  52.1247 '

angular diameter 35.69"

angular radius 17.85"
largest distance from the Sun 8.1608146  108 km

 

largest distance from Sun 8.1608146  108 km
next apoapsis time Sat 30 Dec 2028
last apoapsis time Sat 18 Feb 2017

apparent altitude 49  4 ' 21. "

apparent magnitude 2.21

longitude of ascending node
100.55615

atmospheric composition  20 

atmospheric pressure 0.3 bar

atmospheric scale height 27. km
authalic diameter 1.399  105 km
rows 1–20 of


This function shows the mass and density for each planet:


 Text
Grid
Prepend
EntityValue
"Planet", "Name", "Mass", "Radius",
"Name", "Mass", "Radius", Frame  All, Background  None,
Lighter
Yellow, .9, White, Lighter
Blend
Blue, Green, .8
 !   ! 271

Name Mass Radius

23
Mercury 3.301  10 kg 2.44  103 km
Venus 4.867  1024 kg 6.05  103 km
Earth 5.97  1024 kg 6371.009 km
23
 
 Mars 6.417  10 kg 3390. km
27
Jupiter 1.898  10 kg 6.995  104 km
Saturn 5.683  1026 kg 5.830  104 km
Uranus 8.681  1025 kg 2.536  104 km
Neptune 1.0243  1026 kg 2.46  104 km

The next function displays the number of moons for each planet:


 Length  EntityValue
"Planet", "Satellites", "EntityAssociation"

 
  Mercury  0, Venus  0, Earth  1, Mars  2,

Jupiter  80, Saturn  83, Uranus  27, Neptune  14

As we mentioned previously, if we periodically observe a planet with the naked eye

at the same time and from the same place and we record (sometimes we can take
pictures, as we saw with the example of the Sun) its location with respect to the
horizon, the resulting figure is called an analemma. In Chapter 5, we built the one
corresponding to the Sun. Here we are going to do the same for Mars and Venus.

We need first to define our position (latitude and longitude).


 zone  "Location"  GeoPosition
40.96, 5.66;

Next, we display the Mars and Venus positions at 20:00 h each 5 days for the first 90 days
of 2023.


 analemmamars  Table

QuantityMagnitude
PlanetData
"Mars", EntityProperty
"Planet", "Azimuth",
"Date"  DateObject
DateList
2023, 1, i, 20, zone,
"AngularDegrees", QuantityMagnitude

PlanetData
"Mars", EntityProperty
"Planet", "Altitude",
"Date"  DateObject
DateList
2023, 1, i, 20, zone,
"AngularDegrees", i, 1, 90, 5;


 analemmavenus  Table

QuantityMagnitude
PlanetData
"Venus", EntityProperty
"Planet", "Azimuth"
"Date"  DateObject
DateList
2023, 1, i, 20, zone,
"AngularDegrees", QuantityMagnitude

PlanetData
"Venus", EntityProperty
"Planet", "Altitude",
"Date"  DateObject
DateList
2023, 1, i, 20, zone,
"AngularDegrees", i, 1, 90, 5;

With GraphicsRow we can display both graphs in the same row. The graph for Mars has
been modified using AspectRatio to make its visualization easier.
272      


 GraphicsRow


Graphics
Orange, Point
analemmamars,
Frame  True, FrameLabel  "azimuth", "altitude",
AspectRatio  1  2, PlotLabel  Style
"Mars Analemma", Bold,
Graphics
Orange, Point
analemmavenus, Frame  True,
FrameLabel  "azimuth", "altitude",
PlotLabel  Style
"Venus Analemma", Bold, Frame  All

Mars Analemma Venus Analem

40
70 35
30
60
altitude

altitude
25
 

20
50
15
10
40
5
100 120 140 160 240245250
azimuth azimuth

The points with negative altitude correspond to those times when the planet is below the
horizon and therefore invisible. We cannot see the planet when it’s behind the Sun either.
We can also see that in the case of Venus, the planet is only visible a few degrees
over the horizon. This is because Venus is an interior planet and, as such, when
observed from the Earth, it will never be very high above the horizon. This means
that we will never be able to see it next to the zenith since the sunlight would blind
us. Obviously, this doesn’t happen with exterior planets such as Mars.

The graph below shows the orbits of the terrestrial planets or inner planets: Mercury, Venus,
the Earth and Mars, with the Sun (not in scale) in the center.


 Graphics3D
Yellow, Sphere
0, 0, 0, 0.02, EntityValue

EntityClass
"Planet", "InnerPlanet", "OrbitPath", Boxed  False

 

 !   ! 273

The next four functions, based on examples included in the Wolfram Documentation, enable
us to calculate and display graphically the distance from Earth and the apparent magnitude
of the visible planets over a year:


 visibleplanets
planet , d1 , d2 , property  :
PlanetData
planet, Table
EntityProperty
"Planet", property,
"Date"  DateObject
date, date, DateRange
d1, d2, "Week";
ts
planet , d1 , d2 , property  :
TimeSeries
Transpose
DateRange
d1, d2, "Week",
visibleplanets
planet, d1, d2, property;

  DateListPlot

AssociationMap
ts
, 2021, 1, 1, 2022, 12, 31, "DistanceFromEarth" &,
"Mercury", "Venus", "Mars", "Jupiter", "Saturn",
PlotLegends  "Expressions", FrameLabel  Automatic

10

8 Mercury
Venus
6
au

 
 Mars

4 Jupiter
Saturn
2

0
2021 2022 2023

  DateListPlot

AssociationMap
ts
, 2021, 1, 1, 2022, 12, 31, "ApparentMagnitude" &,
"Mercury", "Venus", "Mars", "Jupiter", "Saturn",
PlotLegends  "Expressions", FrameLabel  Automatic

Mercury
2
Venus
 

0
Mars
Jupiter
2 Saturn

4

2021 2022 2023

Even though naked-eye observation and the use of angle-measuring instruments

(sextant, astrolabe, etc.) advanced the science of astronomy, the arrival of the
telescope and other instruments such as precision clocks was no less important.
They opened new possibilities for exploring the firmament. Later on, in the 20th
274      

century, radio telescopes and spacial astronomy would arrive taking the science to a
whole new level.

8.3.3. Dwarf Planets

We’ve seen that PlanetData doesn’t include Pluto, demoted to the “Dwarf Planet”
category in a controversial meeting of the International Astronomical Union on
August 24, 2006. This action was motivated mainly by the discovery of planets
beyond Pluto, in an area known as the Kuiper Belt that probably includes thousands
of planetoids. Pluto is considered to be part of this belt. All the planets that didn’t
fit the new definition were categorized as dwarf planets. There are reasons to
justify that Pluto doesn’t belong to the same category as the classical planets but
the dwarf label doesn’t seem the most adequate one since there are dwarf planets
that are most likely bigger than the classical planet Mercury.

The celestial bodies that orbit around the Sun but are not planets have been included in the
MinorPlanet entity (or MinorPlanetData).


 EntityValue
"MinorPlanet", "EntityClasses"

 
  minor planets , Amor asteroids , Apollo asteroids ,

asteroid belt , Aten asteroids , dwarf planets , Kuiper Belt objects ,

scattered disk objects , near-Earth asteroids , Plutoids , trans-Neptunian objects ,
Jupiter Trojan asteroids , centaur asteroids , inner main belt asteroids ,
main belt asteroids , Mars crossing asteroids , outer main belt asteroids 

Next, we use a function to know how far away in astronomical units (au) it would be from
the Earth each year starting in 1970-01-01 until 2300


 DateListPlot
, 01, 01 , EntityValue

Entity
"MinorPlanet", "Pluto", EntityProperty
"MinorPlanet",
"DistanceFromEarth", "Date"   &  Range
1970, 2300, 10

45

40
 


35

30

2000 2100 2200 2300

Notice that during our lifetimes Pluto will be further away each year. In July 2015, a probe,
New Horizons, visited it for the first time.

One of the most interesting characteristics of Pluto and other celestial bodies in the Kuiper
Belt is that their orbits are normally slanted with respect to the ecliptic plane, next to the
ones from the classical planets, as we can see with the following function:
 !   ! 275


 Plutoids
"Inclination", "EntityAssociation"

 
  136199 Eris (2003 UB313)  44.143, 136108 Haumea (2003 EL61)  28.1950,

136472 Makemake (2005 FY9)  28.99899, Pluto  17.1

In the coming years, the number of newly discovered dwarf planets will most likely increase
substantially.

We dynamically simulate the movement of the dwarf planets starting on January 1, 2020
using MinorPlanetData. Notice the use of Tooltip to see the name when placing the mouse
over a planet. The function is slow.

  minorplanets  MinorPlanetData
EntityClass
"MinorPlanet", "Plutoid";
Manipulate

Graphics3D
 Tooltip
Sphere
2, 1, 1 & 
Transpose
minorplanets,
QuantityMagnitude

MinorPlanetData
EntityClass
"MinorPlanet", "Plutoid",
EntityProperty
"MinorPlanet", "HelioCoordinates", "Date" 
DateObject
2017, 1, date, ColorData
1, 1,
MinorPlanetData
, "OrbitPath" &  minorplanets,
PlotLabel  DateString
DateList
2017, 1, date,
date, 1, 300  365.25, SaveDefinitions  True

8.3.4 Exoplanets
To obtain information about planets outside of the solar system we can use the
Exoplanet entity (or ExoplanetData).
276      

According to a paper published in Nature on August 25, 2016, there’s an Earth-sized planet
orbiting our nearest neighboring star other than the sun, Proxima Centauri.


 EntityClass
"Exoplanet",
"DistanceFromEarth"  TakeSmallest
1  EntityList

 
  Proxima Centauri b 

Next all the available information regarding this planet:


 DeleteMissing Proxima Centauri b
"Dataset"

Although the planet orbits (0.051 au) closer to its star than Mercury does to the Sun, the star
itself is far fainter than the Sun. The surface temperature would allow the presence of liquid
water. However, the conditions on the surface may be strongly affected by the ultraviolet
and X-ray flares from the star, far more intense than the Earth experiences from the Sun.

8.3.5 Galaxies and Nebulae

Today, we know that the Sun is just one of the stars in a galaxy known as the Milky
Way. However, the notion of a galaxy was not introduced until the 1920s. Before
then, the difference between galaxies and nebulae was not clear. Seen through the
telescope both appeared like blurry objects with shapes resembling clouds. That’s
why the term nebulae was used to refer to both. Galaxies were considered a type of
nebula, and our galaxy was the entire Universe.
  !   ! 277

Let’s see the definition of galaxy:


  galaxy definition

Result

 a splendid assemblage especially of famous people

  tufted evergreen perennial herb having spikes of tiny white flowers and
glossy green round to heart-shaped leaves that become coppery to
 

maroon or purplish in fall

  astronomy a collection of star systems; any of the billions of systems

each having many stars and nebulae and dust

Based on the 3rd definition, nebulae are just part of galaxies. With telescopes we
can only see the ones located in our own galaxy. The generic name nebulae actually
includes two types of very different structures: the planetary nebulae
(“PlanetaryNebula”) generated by the accumulation of dust left after a supernova
explosion, and nebulae (“Nebula”) star nurseries, such as the Orion Nebula. As we
can see the name is confusing as planetary nebulae have nothing to do with the
formation of planets. Other very important galaxy elements are clusters, or globular
clusters (groups of stars). Information about galaxies, clusters, and nebulae can be
found using the Galaxy entity (or GalaxyData), the StarCluster entity (or
StarClusterData), and the Nebula entity (or NebulaData).

Here is the Milky Way diameter in light-year


 Milky Way GALAXY
     

 
 1.0  105 ly

A very interesting nebula is Orion (M42), visible specially during winter in the northern
hemisphere. It is an enormous gas cloud where new stars are constantly being born.


 Orion Nebula

 
 M42


 Entity
"Nebula", "M42"
"Dataset"

278      

We’ve seen that big distances, such as the ones between galaxies, are usually measured in
parsec (pc), the equivalent of the distance corresponding to one arc second. The closest
galaxies to our own are shown below, note the use of Dataset. They all form a group of
galaxies known as the local group. Many of them are quite small.


 lggaxies  Dataset
Local Group of galaxies GALAXIES

"DistanceFromSun", "EntityAssociation";

We select the galaxies closer than 100 kpc:

 !   ! 279


 lggaxies
Select
  100 kpc &

Boötes Dwarf Galaxy 0.0604 Mpc

Canis Major Dwarf Galaxy 2.5  104 ly
Draco Dwarf Galaxy 0.082 Mpc
Large Magellanic Cloud 0.0500 Mpc
Milky Way 2.6  104 ly

 

Sagittarius Dwarf Elliptical Galaxy 0.020 Mpc
Sculptor Dwarf Galaxy 0.079 Mpc
Sextans Dwarf Galaxy 0.086 Mpc
Small Magellanic Cloud 0.0606 Mpc
Ursa Minor Dwarf Galaxy 0.066 Mpc
Willman I 0.037 Mpc

To check which galaxies are visible to the naked eye (Apparent Magnitude < 5), we sort the
apparent magnitudes of the local group members (remember that the smaller the magnitude
the higher the luminosity).


 Dataset
Local Group of galaxies GALAXIES

"ApparentMagnitude", "EntityAssociation"
Select
  5 &

Large Magellanic Cloud 0.9

M31 3.5
Sagittarius Dwarf Elliptical Galaxy 4.5
 


Small Magellanic Cloud 2.2

As mentioned earlier, with our own eyes it is very difficult to see magnitudes greater than 5.
This means that in the northern hemisphere the only visible galaxy is M31 (Andromeda), the
rest of the galaxies with high luminosity (and correspondingly low apparent magnitude) are
only visible from the southern hemisphere.

8.4 Application: Determining the Color of the Stars


 ClearAll
"Global` "

The French philosopher Auguste Comte (1798–1857) believed that the composition
of stars was going to be beyond human knowledge. In his Cours de Philosophie he
used to say that stars would only be known as specks of light in the sky since they
were very far away. However, the analysis of stellar light has enabled us to know
reasonably well stars’ composition, temperature, and evolution.
 280      

A star is what in physics is known as a black body and therefore has a light
spectrum whose characteristics are related to its surface temperature following
Planck’s law.

Using the free-from input we can access Planck’s law formula. First we type “Planck law”
using natural language, press + once Mathematica has understood our input, select the
equation’s pod, and after right-clicking on the subpod and choosing “Copy subpod content”,
paste the information on the cell below.


  Planck law

Equation

2 h c2
L 
5 h c k T  1

L spectral radiance as function of wavelength

 wavelength
 

T temperature

h Planck constant  6.626  1034 J s

c speed of light  2.998  108 m s

k Boltzmann constant  1.381  1023 J K

The spectrum for a specific temperature, and other properties, can be represented using the
PlanckRadiationLaw.


 PlanckRadiationLaw
Quantity
8000, "Kelvins", "SpectralPlot"
1.4  1014

1.2  1014

1.0  1014

8.0  1013
W m3 sr

 
 6.0  1013

4.0  1013

2.0  1013

0
0 5.  107 1.  106 1.5  106 2.  106 2.5  106 3.  106 3.5  106
m

An alternative way is to use FormulaData.


 plancklaw1  FormulaData
"PlanckRadiationLaw", "Wavelength"

2 h c2
 
 L 
h ck
5 exp   1
T
 !   ! 281

Let’s have a look at the inner formula composition:


 InputForm
plancklaw12
 

# $
Quantity
2, "PlanckConstant""SpeedOfLight"^2
1  E^Quantity
None, "PlanckConstant""SpeedOfLight""BoltzmannConsta
QuantityVariable
"T", "Temperature"QuantityVariable
"
", "Wavelength
QuantityVariable
"
", "Wavelength"^5

To get the constant values in the International System (SI):


 plancklaw  UnitConvert
plancklaw12, "SI"

2 h c2
 
 UnitConvert , SI
h ck
5 exp   1
T

Next, we build a function where the wavelength is computed as function of the temperature.


 blackbody
T  : plancklaw .
QuantityVariable
" ", "Wavelength"  Quantity
k , "Meters",
QuantityVariable
"T", "Temperature"  Quantity
T, "Kelvins"


 Plot
Callout
blackbody
8000, "8000 K", Above,
Callout
blackbody
5800, "5800 K", Above,
Callout
blackbody
4000, "4000 K", Above, k, 10 ^ 7, 3  10 ^ 6 ,
AxesOrigin  0, 0, PlotRange  All, MaxRecursion  0

4000 K

3000

2000
 


1000
5800 K 8000 K

5.  107 1.  106 1.5  106 2.  106 2.5  106 3.  106

We are going to build a dynamic representation of Planck’s law that shows the
relationship between color and temperature.

The next function associates a black body spectrum to its temperature (in degrees Kelvin).

  BlackBody
T  :
34
With h  6.62607  10 , c  2.99792458 10 ^ 8, k  1.38065  10 ^ 23,
Plot
2 h c ^ 2  ^ 5   Exp
h c  k T 1,
 , 10 ^ 9, 1000  10 ^ 9, MaxRecursion  0, ColorFunction 
ColorData
"VisibleSpectrum"
10 ^ 9  &, ColorFunctionScaling  False,
Filling  Axis, Ticks  None, PlotLabel  Row
T, " K"

With Manipulate we create the dynamic representation:

Before executing the function, we need to turn off the message indicating machine underflows:


 Off
General::munfl
282      


 Manipulate

BlackBody
T, T, 5800, "temperature", 3000, 10 000,
SaveDefinitions  True

A more complete modelBlackbody Spectrum by Jeff Bryant can be found in:

http://demonstrations.wolfram.com/BlackbodySpectrum

The next function simulates the color of any given star. We use “EffectiveTemperature” to
get the superficial temperature and then relate it to its color:

  starColorPlot
star  :
Graphics3D
ColorData
"BlackBodySpectrum"
QuantityMagnitude

star
"EffectiveTemperature", Sphere
, Boxed  False, Lighting 
"Ambient", Gray, "Directional", LightGray, ImageScaled
0, 0, 1,
PlotLabel  star
"Name"

Let’s apply it to several known stars:


 stars  Entity
"Star",  & 
"Rigel", "Sirius", "Sun", "Capella", "Betelgeuse"

 
  Rigel , Sirius , Sun , Capella , Betelgeuse 

  Through
stars
"Name", "EffectiveTemperature"

Rigel 1.21  104 K

Sirius 9940. K
 
 Sun 5772. K
Capella 5.1  103 K
Betelgeuse 3.6  103 K

Blue corresponds to hotter stars and orange to colder ones:
 !   ! 283


 starColorPlot  stars

 
  , , , , 

8.5 The Measurement of Distances Across the Universe


 ClearAll
"Global` "

Historically, the measurement of distances across the Universe was one of the
fundamental problems in astronomy. The first step toward solving it was to
measure the Earth’s size but before even that, someone had to realize that our
planet is a sphere. It is believed that it was the Pythagoreans, mysterious people
that didn’t leave any written works behind, who for the first time, circa 430 B.C.,
came up with the idea that the Earth was round, based probably on two
observations: a) Sailors traveling from Greece to Northern Africa had noticed that
several constellations, such as Ursa Major, would appear higher in the horizon in
Egypt than in Athens for a given date and time (although there are other geometric
shapes that may explain this fact, the sphere is the one that provides the simplest
explanation), b) The way the Moon hides during lunar eclipses can be easily
explained if one assumes that it’s a shadow projected by a sphere, the Earth, upon
the Moon.
Once it was clear that the Earth was round, the next step was its measurement. It
was Eratosthenes, who eventually became the head of the library of Alexandria, the
first person to measure the Earth’s size. The method he used is legendary and well-
known among astronomy enthusiasts, but it’s worth describing it in the next few
lines. He started measuring the distance between Alexandria and Syene (nowadays
Aswan), that were located approximately on the same meridian (they really differ
by 3º). He also assumed that the Sun was distant enough that its rays would hit the
Earth in a parallel manner. He knew that during the summer solstice in Syene the
light would illuminate the bottom of the wells in the city. At that moment, the
sunbeams would hit the ground perpendicularly while at the same time in
Alexandria they would form an angle of approximately 1/50th of the length of a
circumference (number deduced by the projection of the shadows). He then used a
basic trigonometric identity to figure out that the distance between Alexandria and
Syene was 1/50th of the Earth’s circumference. Since that distance was 5,000
stadia, the Earth’s circumference was 250,000 stadia. There’s no agreement about
the exact conversion rate to meters. Depending on the reference, if we take a value
ranging from 185 m to 157.2 m, the error is between 17% and 1% of the actual
size, quite an extraordinary approximation in any case. In 1492, Christopher
Columbus assumed a considerably smaller figure while trying to convince Queen
Isabella of Spain to finance his trip against the opinion of her advisors, who knew
Eratosthenes’s estimate. He persisted in his error and ended up discovering
America.
284 Mathematica beyond mathematics

Shortly before Eratosthenes’s death, Hipparchus of Nice was born (c. 190 BC).
He’s not only famous due to his discovery of the precession of equinoxes, as we
have already mentioned, but also because of his quite accurate measurement of the
distance between the Earth and the Moon using the shadow projected by the former
onto the latter during Moon eclipses. He also initiated an empirical study of the
apparent magnitude of the stars and their positions. For that purpose, he invented
the ecliptic coordinates (different from the elliptic ones) and used the astrolabe
(invention attributed to him). Many centuries later (from 1989 to 1993) a satellite,
carrying its name, repeated his work with technology from the late-20th century. Its
successor: Gaia
(http://www.esa.int/Our_Activities/Space_Science/Gaia_overview), currently in
mission, will measure the position of the objects in our galaxy with extreme
precision.
We can compute the distance to the nearest stars using geometric methods (Figure
8.2). We measure the position of the star seen from a certain location, and 6 months
later we will observe its apparent displacement. This phenomenon is known as
parallax, and its explanation is similar to what happens when we place our thumb
between and object and our eyes. When closing either eye, we will notice the
apparent movement of the object. In the case of a stellar parallax, each eye
corresponds to the vertices of the base of an isosceles triangle created when joining
the Earth, position in two vertices A and B, separated by 6 months along the orbit
of the star, with vertex C, the position of the star, considered fixed. The parallax is
half the angle ACB.

Figure 8.2 Calculating the distance to the nearest stars.

The distance, d, from the Sun to the star with parallax, p, in radians, can be
computed as follows:
1 AU
d= tan p
= {given that the angle is very small, tan p  p } 1p AU.

Since the values of p are very small, it is convenient to express them in arcseconds, p s .The
parallax of Procyon is:
 !   ! 285


 1  QuantityMagnitude
Procyon parallax  "pc"

 
 3.50 pc

Or directly in parsecs:


 UnitConvert
Procyon STAR
  " %  , "Parsecs"

   3.49568 pc

A big leap forward in the measurement of astronomical distances happened when

Henrietta Swan Leavitt (1868–1921) noticed that certain stars, now known as
Cepheid variables, exhibited regular changes in their luminosity from which their
absolute magnitude could be calculated.
The relationship period/luminosity for the Cepheid variables has been revised
several times since Henrietta Leavitt’s original measurements. Currently, the
following function is usually applied:
M 2.78 log P  1.35

where M is the absolute magnitude of the star (the luminosity of the star if it was 1
pc away) and P is the average period in days.

The previous expression can be written in Mathematica as:


 abm
pd   2.78 Log
10, pd 1.35;

Applying a similar criterion to the one used to determine the distance between a
bulb with known brightness and the brightness experienced by an observer, we can
deduct the following relationship between M, the apparent magnitude m
(magnitude seen from the Earth, calculated as the average between the maximum
and minimum observed magnitudes), and the distance D (in pc):
mM 5 log D  10 5 log D  5

From the previous formulas, we can establish the following equation to compute the
distance using the period (pd) and the apparent magnitude (apm), both of which can be
measured empirically:


 distceph
pd , apm  : Solve
apm abm
pd 5 Log
10, dist 5  0, dist

This method can be applied to stars in our galaxy with the help of big ground
telescopes or space ones. We can also use it to measure the distances to our nearest
galaxies assuming we can find Cepheid variable stars in them.
In https://www.eso.org/public/products/education/ we can find a set of basic
astronomy exercises using real data. In one of them, we are given the results of the
calculations done by Hubble in several Cepheid variables located in the M100
galaxy. According to the data, one star has a period of 53.5 days and an apparent
magnitude of 24.5. Stars of such magnitude can only be observed with large
telescopes.

The absolute magnitude and distance, in pc, are:


 abm
53.5, dist . distceph
53.5, 24.90

 
 6.15482, 1.62542  107 

 286      

It’s one of the most distant objects ever measured using this method.
For remote galaxies the method used is based on measuring the apparent luminosity
of a type Ia supernovae and relating it to the real magnitude (that can be deduced
using theory).
Another key parameter related to the previous one is the redshift experienced by
light. This is an example of the Doppler effect caused by galaxy movements,
mostly associated with the expansion of the universe.

You can find information about redshift typing in a free-form input cell “Cosmological
redshift”). After expanding the cell and telling Mathematica to interpret it as a formula,
you can even do some interactive calculations (click +) .


  Cosmological redshift

Basic dimensions


 UniverseModelData
Quantity
5  10 ^ 9, "LightYears", "Redshift"

 
 0.485872


 QuantityMagnitude 

UniverseModelData
Quantity
5  10 ^ 9, "LightYears", "Redshift";

Astronomer Edwin Powell Hubble (1889–1953) (the famous telescope was named
after him) realized that galaxies frequently presented redshift. The amount changed
depending on the galaxy. Later on, this astronomical phenomenon became known
as Hubble’s law.
cz H0 D

with:
z (redshift) = ob  e  e , where ob corresponds to the observed wavelength and e to the
emitted one; z is therefore a dimensionless number.
c is the speed of light.
D is the actual distance to the galaxy (in Mpc).
H0 is the Hubble constant at the moment of observation (during our lifetime and for the
duration of our civilization, this value will be practically constant).
Once z is known, we can calculate D using basic algebra:
cz
D
H0

Hubble’s law is considered a fundamental relation between recessional velocity and

distance. However, the relation between recessional velocity and redshift depends
on the cosmological model adopted, and is not established except for small
redshifts.
A modified version of Hubble’s law and other properties of the universe are now
included in UniverseModelData.

The Hubble constant, H0 (HubbleH0) is regularly expressed in m/s/Mpc.


 UniverseModelData
"LambdaCDM"

 
 OmegaLambda  68.89 , OmegaMatter  31.11 , OmegaRadiation  0.07 , HubbleH0  H 0 

 !   ! 287

Determine the redshift of a comoving object 5  109 light years away:


 UniverseModelData
Quantity
5  10 ^ 9, "LightYears", "Redshift"

 
 0.485872

Determine the distance of a comoving object, in light years, for a redshift of 0.075:


 f
t ? NumericQ : QuantityMagnitude 
UniverseModelData
Quantity
t, "LightYears", "Redshift";


 FindRoot
0.075  f
dist, dist, 10 ^ 9

FindRoot: The line search decreased the step size to within tolerance specified by AccuracyGoal
and PrecisionGoal but was unable to find a sufficient decrease in the merit function. You may
need more than MachinePrecision digits of working precision to meet these tolerances.

 
 dist  1.00056  109 

Let’s apply this method to the galaxy IC1783.


 zIC1783  GalaxyData
"IC1783", "Redshift"

 
 0.011


 FindRoot
zIC1783  f
dist, dist, 10 ^ 8

 
 dist  1.55537  108 

We compare this value with the best estimate available in the Galaxy entity.


 UnitConvert
IC 1783 GALAXY
  " %  , "ly"

 
 1.256  108 ly

We can see that there are some discrepancies. This is because galaxies are also
affected by the gravitational forces of their galactic neighbors, e.g., Andromeda
influencing the Milky Way. In fact, Andromeda and our galaxy are approaching
each other and will collide in approximately 4 billion years from now.
The possibilities of the astronomical functions in Mathematica don’t end here. Feel
free to continue exploring.

8.6 Application: Binary Systems and the Search for Exoplanets


 Quit


In this section, we will show how we can use Mathematica for other Astronomy
related matters. We will describe, in particular, several features developed by the
author for the study of binary systems and, by extension, the search for exoplanets.
Since 1995, when the first extrasolar planet was found, more than 400 have been
detected. NASA’s Kepler spacecraft launched on March 6, 2009, was specifically
designed to search for extrasolar planets. Probably, by the time you read these
lines, their total number will have increased significantly.
Kepler site: http://www.nasa.gov/kepler
To find exoplanets, we mainly use two methods:
288 Mathematica beyond mathematics

1) Radial velocity: A star with a planet will move in its own small orbit in response
to the planet’s gravity leading to variations in the radial velocity of the star with
respect to Earth. These movements are only detectable for giant planets next to
stars.
2) Transit photometry: When a planet crosses in front of its parent star’s disk the
observed brightness of the star drops by a small amount. This last method is
becoming the most effective one. The principle is similar to the one used when
studying binary star systems.
More than half of the stars that we see are actually systems of two or more stars.
Normally it’s not possible to distinguish between different stars in multiple systems
(not to confuse them with naked-eye binaries that are usually stars that seem close
to each other when seen from Earth but are actually located in different systems
altogether). The detection is usually done with the transit photometry method. This
approach is very effective when dealing with eclipsing binaries, systems of two
stars with coplanar orbits when observed from Earth.
Next we are going to describe how to build this kind of system using Mathematica
and study its brightness. In the example that follows, we use as reference Eclipsing
Binary Stars by Dan Bruton:
http://www.physics.sfasu.edu/astro/ebstar/ebstar.html
Figure 8.3 represents a binary system (although we discuss the case of two stars,
the method is similar if we consider a star and a planet). The coordinates of the
centers are (x1, y1, z1) for star 1 and (x1, y1, z1) for star 2 (assuming
homogeneous spheres, the centers of mass coincide with the geometric centers).
We consider {0, 0, 0} as the coordinates for the center of mass of the system. These
coordinates have been chosen so that an observer, located on Earth, is on the axis
OZ. That is: the one that goes from the center of mass to Earth. For the OY axis, we
take the perpendicular to OZ in the star 1 plane when it forms an angle  = 0. This
happens when the star is closest to the observer.

Figure 8.3 A binary system.

With:
i = orbital inclination: angle, in radians, between the line of vision and the orbital plane axis.
 !   ! 289

R = distance between the centers of both stars. It’s expressed in arbitrary units, usually the
average distance between the stars. In the case of a constant orbit, R = 1.
m1, m2 = stars’ masses (we can choose arbitrary units, but it’s important that they keep the
equivalence q = m2/m1).
r1, r2 = radii. For convenient purposes we choose r1 > r2. In practice, we can choose stars 1
and 2 so that the condition is always met. It’s convenient to express the radii in fractions of
the major semi-axis. In the case of a circular orbit, obviously the major semi-axis is greater
than or equal to the minor one and identical to the radius R.
 = Phase, corresponding to the angle between the star and OZ, using  = 0, the angle where
the center of star 2 is the closest to the observer. In a binary system with circular orbit,  is a
variable parameter (it’s time-dependent) while the previous parameters are constant
(although that is not always the case, for example: There might be some intrinsic changes to
the brightness or even the mass could change as in the case of a mass transfer between stars.
However, in the example that follows we’ll use the simplest case, which happens to be the
2  tt0
most common one).  is known as the azimuthal angle and follows  P
, with P being
the orbital period.
In a binary system, the coordinates {x1, y1, z1} and {x2, y2, z2} for the location of
the stars are given by:
x y z
x1 , y1 , z1 ;
1  1  q 1  1  q 1  1  q
x y z
x2 , y2 , z2 ;
1q 1q 1q

We can convert them to spherical coordinates using the formulas:


 x  R Sin
;
y  R Cos
i Cos
;
z  R Sin
i Cos
;

Then, the position of each star as a function of R, i, , q with q = m2/m1 can be calculated
with the following functions:
x y z


 star1
R , i ,  , q   , , ;
1 1  q 1  q1 1 1  q
x y z
star2
R , i ,  , q   , , ;
1 q 1 q 1 q

We have used the previous equations to define a Mathematica function to visually display
the movement of both stars as a function of R, i, , m1, and m2:
eclbin
R, i, , m2m1, r1, r2 (The condition m2 < m1 is necessary).
290      


 eclbin
rr , ii , rho , qq , rr1 , rr2  :

Module
R, i, , q, r1, r2,

Manipulate
Graphics3D
ControlActive
Opacity
1, Opacity
.6,
Sphere
st1
R, i, , q, r1, Sphere
st2
R, i, , q , r2,
Opacity
1, RGBColor
.6, .74, .36, PointSize
.02,
Point
0, 0, 0, Black, PlotRange   2, 2,  2, 2,  2, 2,
SphericalRegion  True, Boxed  False, Background  Black,
ImageSize  500, 340, ViewPoint  0, 0, 20,
R, rr, "R", 0.1, 5, Appearance  "Labeled",
i, ii, "i", 0, , Appearance  "Labeled",
q, qq, "mass 2  mass 1", 0.1, 1, Appearance  "Labeled",
r1, rr1, "radium 1", 0.1, .5, Appearance  "Labeled",
r2, rr2, "radium 2", 0.1, .5, Appearance  "Labeled",
, rho, " rotation", 0, 4 , Appearance  "Labeled",
R Sin

Initialization  st1
R , i ,  , q  : ,
1
1
q

R Cos
i Cos
 R Cos
 Sin
i

, , st2
R , i ,  , q  :
1 1
1 1
q q

R Sin
 R Cos
i Cos
 R Cos
 Sin
i

, , 
1 q 1 q 1 q

The initial moment  = 0 is the time when both stars are aligned and can be seen
from the OZ axis (where the observer is located), that is when the interposition of
the stars is at its maximum. In a real system, all the parameters would be fixed and
only  would change as a function of time.

Let’s consider the binary system defined by the following parameters R = 2, i = 2 Pi 85
/360, m1 = 0.6, m2 = 0.1, R1 = 0.4, R2 = 0.3.


 eclbin
2, 2 Pi 85  360, 0, 0.1  0.6, 0.4, 0.3
 !   ! 291

8.7 Light Curves

8.7.1 Light Curves in Binary Systems

  Quit


The transit photometry method discussed in the previous section is based on the
study of the changes in brightness when a star or a planet interposes itself between
another star and the observer. As mentioned before, since stars are not visually
distinguishable, even less a star and a planet, what we will see is a variation in the
apparent brightness.
The method described here, with the arrival of astronomical CCD cameras, is
within the reach of Astronomy fans. There have been instances where fans have
even been able to detect exoplanets using this technique. This is a great example of
how science can be done without the need for major resources.
The luminosity l of a star is defined as the amount of energy emitted by the star per
unit of time. The flow F is the energy emitted per unit of surface and per unit of
time, that is F1 = l1 2 and F2 = l2 2 where l1 , l2 = luminosities (in arbitrary
4  R1 4  R2
units, as long as they keep the ratio l2 /l1 ).
Then an observer will see the system with a luminosity given by:
l K F1 A1 F2 A2

where A1 and A2 represent each of the areas of the stars’ disks as seen by the observer, and
K can be determined based on the observer’s detector area, on the OZ axis, and the distance
between the Earth and the binary system.
To find these areas, we need to know the apparent distance  as seen by the
292      

observer (located on the OZ axis). That distance can be calculated as follows:

2 2
 x1  x2 y1  y2

The next function computes (R, i, , q):


 rho
R , i ,  , q   Module
x, y, x1, x2, y1, y2, x  R Sin
;
y  R Cos
i Cos
;
x y
x1, y1  , ;
1 1  q 1 1  q
x y
x2, y2  , ;
1 q 1 q
Sqrt
x2 x1  ^ 2 y2 y1  ^ 2;


 rho
2, 2 Pi 85  360,  , 0.1  0.6 

 
 0.0303845 Cos
2  4. Sin
2

With the example data (R = 2, i = 2 Pi 85 /360, m1 = 0.6, m2 = 0.1), the apparent distance 
changes according to the graph below. Based on our problem formulation, the lowest
apparent size corresponds to the maximum interposition that happens when  = 0, Pi, 2 Pi,
..., n Pi.


 Plot
rho
2, 2 Pi 85  360,  , 0.1  0.6 ,  , 0, 4 Pi, AxesLabel  , 


2.0

1.5

 

1.0

0.5


2 4 6 8 10 12

To calculate A1 and A2, it is necessary to take into account the current stage in the
cycle.

The two figures below show the star with the bigger radius (r1 ) in red, and the one with the
smaller radius (r2 ) in blue.


 Graphics
Red, Disk
0, 0, 2, Blue, Disk
4, 0, 1

 


Without eclipse: Corresponds to the period in which all the light emitted by both
stars in the direction of the observer reaches him. Therefore, this is when the
luminosity is at its maximum. In this stage, the following relationship holds:
  r1 r2 with:
8. Looking at the Sky 293

A1 =  r12 and A2 =  r22

Partial eclipse: This happens when one star partially covers the other one. During
this stage, r1 + r2 >  > r1 - r2 .

Figure 8.4 was created using the following function:
Graphics[{{Red, Circle[{0, 0}, 1]}, {Blue, Circle[{1.15, 0}, 0.5,
{-2 Pi/3, 2 Pi/3}]}, {Dashed, Blue, Circle[{1.15, 0}, 0.5]},
Line[{{0, 0}, {0.9, 0.425}, {0.9, -0.425}, {0, 0}}], Line[{{0.9,
0.425}, {1.15, 0} , {0.9, -0.425} }] }]
We added legends to the output of the function using Mathematica’s 2D Drawing Tools:
+D in Windows or +T in OS X.

r2
r1

!2 "A1 !1

"A2

Figure 8.4 Partial eclipse diagram.

To calculate the amount of light that arrives from the eclipsed star, we have to
subtract the eclipsed area from the total area. This means we have to compute "A1
and "A2 corresponding to the star 1 and star 2 segments, respectively, using the
following formulas:
1 1
"A1 = R12 ( !1 - Sin[!1]); "A2 = R22 ( !2 - Sin[!2])
2 2

The following expressions enable us to compute !1 and !2:

!1 !2
R22 = R12 + 2 - 2 R1  Cos , R12 = R22 + 2 - 2 R2  Cos
2 2

The output below represents the different stages of a partial eclipse:
294      


 Graphics
Blue, Disk
2, 0, 1, Red, Disk
0, 0, 2,
Graphics
Blue, Disk
 2, 0, 1, Red, Disk
0, 0, 2,
Graphics
Red, Disk
0, 0, 2, Blue, Disk
 2, 0, 1,
Graphics
Red, Disk
0, 0, 2, Blue, Disk
2.5, 0, 1 

 
  , , , 

The two figures on the left correspond to the scenario where star 1 (red) is the
closest one, that is z1  z2. In this case, the light that reaches the observer will be:
A1  r12 and A2  r22  "A1  "A2

A1  r12 and A2  r22  "A1  "A2

The rest of the figures corresponds to the case where star 1 (red) is the most distant
one, that is z1 < z2 then:
A1  r12  "A1  "A2 and A2  r22

A1  r12  "A1  "A2 and A2  r22

Total or annular: Corresponds to the period when a star is shaded by the other one
completely. In this stage:  
1 
2 with: A1 = 
21 and A2 0 for z1>z2 and A1 =

21  
22 and A2 =
22 ;

All these conditions are summarized in the table below:

  z1 z2 z1 z2
Stage Condition A1 A1 A1 A1
Without eclipse 
1 
2 
21 
22 
21 
22

Partial eclipse
1 
2 
1 
2 
21 
22  A1  A2 
21  A1  A2 
22

Total or annular  
1 
2 
21 0 
21  
22 
22

Now we are going to include all the criteria described earlier in the function
area
R, i, m2m1, r1, r2, l1, l2, , where: m2  m1 and r1  r2.
The function returns the area that the observer will measure.
Without eclipse: When  # r1 r2 :


 area
R , i , q , r1 , r2 , l1 , l2 ,   :
Module
A1, A2, A1, A2  Pi r1 ^ 2 , Pi r2 ^ 2 ;
l1 A1  r1 ^ 2 l2 A2  r2 ^ 2 ; rho
R, i, , q  r1 r2

Partial eclipse: When r1 r2    r1  r2 :

We first calculate "A1 and "A2
 !   ! 295


 A1
R , i , q , r1 , r2 ,   :

Module
1, , phi, phi1, phi1  Last
Solve

phi
r22  r12 2 2 r1  Cos
, phi .   rho
R, i, , q1, 2;
2
1
r12 1 Sin
1 . 1  phi1 ;
2

A2
R , i , q , r1 , r2 ,   :
Module
2, , phi, phi1, phi1  Last
Solve

phi
r12  r22 2 2 r2  Cos
, phi  .   rho
R, i, , q1, 2;
2
1
r22 2 Sin
2 . 2  phi1;
2

In this type of eclipse, for z1  z2: A1  r12 and A2  r22  "A1  "A2 while for
z1  z2: A1  r12  "A1  "A2 and A2  r22 .


 area
R , i , q , r1 , r2 , l1 , l2 ,   :
R Cos
 Sin
i
Module
z1, z2, a1, a2, A1, A2 , z1  ;
1
1
q

R Cos
 Sin
i
z2  ;
1 q
a1  A1
R, i, q, r1, r2, ;
a2  A2
R, i, q, r1, r2, ;
A1, A2 
If
z1  z2, Pi r1 ^ 2 , Pi r2 ^ 2 a1 a2, Pi r1 ^ 2 a1 a2, Pi r2 ^ 2;
l1 A1  r1 ^ 2 l2 A2  r2 ^ 2  ; r1 r2  rho
R, i, , q  r1 r2

Total or annular eclipse : When  $ r1  r2 :

When r1 > r2 one of the stars will be directly in front of the other one, interposing it. Then a
total eclipse will happen if z1  z2 with A1  r12 and A2 0, or an annular one will occur if
z1  z2 with A1' A1  A2 and A2' A2.


 area
R , i , q , r1 , r2 , l1 , l2 ,   :
R Cos
 Sin
i
Module
z1, z2, A1, A2, z1  ;
1
1
q

R Cos
 Sin
i
z2  ;
1 q
A1, A2  If
z1  z2, Pi r1 ^ 2 , 0, Pi r1 ^ 2 Pi r2 ^ 2 , Pi r2 ^ 2;
l1 A1  r1 ^ 2 l2 A2  r2 ^ 2 ; rho
R, i, , q  r1 r2

Let’s simulate the ideal case of an eclipsing binary. We can distinguish the
repeating phases of an eclipse: one corresponds to the period without eclipse (t1 to
t2), in t2 the eclipse starts and reaches its plenitude between t3 and t4, from t4 to t5
we are in the partial eclipse phase, from t6 to t9 is the transit phase, and when t9 =
296      

t1 the cycle starts again.

We use Plot
area
2, 90 Degree, 0.10.6, 0.6, 0.3, 0.6, 0.2, 2 Pi
t8,  t, 0, 12, AxesLabel  "t", "I", AxesOrigin  0, 0
and include the legends using the Drawing Tools option in the Graphics menu: Graphics 
Drawing Tools.
I
2.5 t2 t9
t1 t5 t6

2.0 t7 t8
t3 t4

1.5

1.0

0.5

t
2 4 6 8 10 12
Figure 8.5 Light curve.

Example: With R = 2, i = 85º, m1 = 0.6, m2 = 0.1, R1 = 0.4, R2 = 0.3, l1 = 0.6, l2

= 0.2, we’d like to show I/Imax.

We first calculate Imax.


 max  NMaximize
area
2, 2 Pi 85  360, 0.1  0.6, 0.4, 0.3, 0.6, 0.2, ,
, 0, 2 Pi1  Quiet
 
 2.51327

Then we display I/Imax.


 Plot
area
2, 85 Degree, 0.1  0.6, 0.4, 0.3, 0.6, 0.2, 2 Pi   max,
 , 0, 1.5, AxesLabel  "Phase", "IImax",
AxesOrigin  0, 0  Quiet
IImax

1.0

0.8

0.6
 


0.4

0.2

Phase
0.2 0.4 0.6 0.8 1.0 1.2 1.4
 !   ! 297

8.7.2 Application: Using Light Curves to Determine the Period-Luminosity Relation

Besides eclipsing variables or planets interposition, periodic changes in luminosity
can happen for other reasons. Probably the best-known example is the change in
luminosity in variable stars, such as Cepheids. The usual method for building an
experimental light curve consists of measuring the luminosity at different times t
and deduce the period from them.

The following expression simulates magnitude as a function of time (in practice, since we
often don’t know this function we try to deduce it from the experimental data). The chosen
function is very simple, and we only use it for didactic purposes.


 mag
t   5 0.14 Cos
0.3 t 0.1;

Let’s assume that we have already taken measurements at different times t i .


 data  Table
t, Round
mag
t, 0.01, t, 0, 200, 5;

We represent graphically the previous data. If we only had the previous data points it would
be very difficult to figure out the underlying pattern (remove Joined  True), joining the
dots makes it easier to observe the cyclical nature of the data. It’s normal for many
intermediate points to be missing, making the analysis even more challenging.


 ListPlot
data, Joined  True
5.15

5.10

5.05

5.00
 


4.95

4.90

50 100 150 200

The analysis method frequently used is to represent the previous phenomenon in a

phase diagram.
To calculate the phase, we need to apply this equation:
t  t0
Decimal part of
P

where:
! represents the phase in cycles.
t0 (normally called epoch) represents the origin of the time variable expressed as a Julian
Date (DJ).
t is the moment when the measurement is taken (normally we will measure the apparent
magnitude, m).
P is the period, the time it takes for the cycle to repeat itself.
298      

An equivalent expression to the previous one and the one that we will be using is:
t  t0
Mod
P

With the function below, Mathematica returns the phase given the initial data points and the
period.

  phi
list , p  : Module
data, data  list;
Transpose
Mod
dataAll, 1  p, 1, dataAll, 2;

In this example we are assuming that the period is known, but in practice that’s
what we are trying to find out. We can check that by using an approximate value of
the period the representation is more diffuse.

With real observations, the period has to be determined, and that is not always easy. The
next function will help us to compute it. We keep in mind that for period-magnitude curves,
the order of representation in the axis OY is inverted. That is, the magnitudes are sorted in
descending order not in the usual ascending one: 3  2  0  1  2.


 curveL
data , per , kmax  :
Module
mag, a, b, k, i, mag  Round
Transpose
data2, 0.1;
Manipulate
ListPlot
Join
phi
data, per k . a , b   a, b,
phi
data, per k . a , b    1 a, b, AxesLabel 
"Period", "Magnitude", AxesOrigin   1, Max
mag 0.1,
Ticks  Automatic, Table
i, i, i, Max
mag 0.1,
Min
mag 0.1, Max
mag Min
mag  10 ,
k, 0, "Deviation from the period in days", 0, kmax

Let’s assume a complete cycle (2). We use the previous function to simulate the
experimental measures and include a random component. We make the assumption that the
period is 20 days (the duration of a complete orbit).


 simulation  Table

, area
2, 2 Pi 85  360, 0.1  0.6, 0.4, 0.3, 0.6, 0.2, 2 Pi   20
RandomReal
NormalDistribution
0, 0.005,
, 0, 60, 0.05;  Quiet

We represent the data in a phase diagram:


 simulation1  Sort
phi
simulation, 20;


 ListPlot
simulation1, Joined  True

2.50

2.45

2.40
 


2.35

2.30

2.25

0.2 0.4 0.6 0.8 1.0

 !   ! 299

The previous luminosity data have to be multiplied by a parameter k to convert

them into a magnitude measure. Remember that higher luminosity numerically
implies lower magnitude. For that reason, to represent the data in a
phase/luminosity diagram we assume k  1.

If you execute the following function and use the slider to choose a 5-day period, you’ll see
that the curve is very similar to the previous figure.


 curveL
Map
Times
1, 1,  &, simulation, 15, 20

8.7.3 Application: Finding the Light Curve of a Known Eclipsing Binary

Let’s apply the newly created function to the luminosity data from NSV 03199
obtained by Garcia-Melendo, E., Henden, A., 1998, IBVS, No. 4546. The data are
in the file: NSV03199.xls in our data directory.

Let’s import the luminosity measures:


 vsdata 
Drop
Import
FileNameJoin
NotebookDirectory
, "Data", "NSV03199.xls",
"XLS", "Data", 1, 1;

The data have Julian dates, so we use as reference JHD=2450510.46542.


 vsdata1  vsdata . a , b   a 2 450 510.46542, b;

We cannot really see the cycle pattern:

300      


 ListPlot
vsdata1

0.05

20 15 10 5 5 10

0.05

0.10
 


0.15

0.20

0.25

0.30

However, when using a period-magnitude curve, the periodic behavior

corresponding to an eclipsing binary can be seen clearly.

We find that the optimum period is 1.046400 days:


 curveL
vsdata1, 1.046400, 0.5

All the previous functions along with other examples in the book have been put
together in a package that can be downloaded from diarium.usal.es/guillermo.

8.8. Additional Resources

Astronomy demonstrations: http://demonstrations.wolfram.com/topic.html?topic=Astronomy
Eric Weisstein’s World of Astronomy: http://scienceworld.wolfram.com/astronomy
To follow objects with variable luminosity (variable stars, supernovae and many more), the
best place is the website of the American Association of Variable Star Observers (AAVSO):
http://www.aavso.org
Author’s website astronomy section (in Spanish): http://diarium.usal.es/guillermo/astronomia
 9

     

Radioactivity is a natural phenomenon that allows us to peek into our

planet’s past and the history of humankind. Its effects when absorbed by
human beings can also be scientifically measured. This chapter covers
radioactivity and other topics related to nuclear and particle physics with
examples such as the modeling of the evolution of isotopes over time or
the dating of the history of humankind. The last two sections discuss the
calculation of binding energies and how to design products that protect
from radiation respectively. Throughout the chapter, the Mathematica
functions
 and   are used extensively.

9.1 Nuclear and Particle Physics

The Entity[“Isotope”, name] (or IsotopeData) and ParticleData functions deal with
nuclear and particle physics applications.

Writing the isotope name inside an Inline Free-form input or


isotope property  returns the value of the specified property for a
given isotope.


 iodine 131
"Properties"  Shallow

 
 

 atomic mass , atomic number , atomic symbol ,

binding energy per nucleon , biological half -life , biological lifetime ,

branching ratios , critical diameter , critical mass , critical organs , 45

The next example shows some of properties of Iodine-131.


 IsotopeData
"Iodine131",  & 
"FullSymbol", "AtomicMass", "BranchingRatios", "DecayModes",
"DaughterNuclides", "HalfLife", "IsotopeAbundance"

 
  131I , 130.906124609 u,
1.00,
BetaDecay,  xenon-131 , 6.9338  105 s, 0.

53 78

The output tells us that this isotope, whose complete symbol is 131
53 I78 , has an atomic mass of
130.09061 atomic mass units (amu), with a single type of radioactive decay, a beta decay,
302      

that disintegrates into Xenon-131 with a disintegration period of    and that it
doesn’t exist in natural form (
).

The example below generates a dataset with information about all the properties available
for Iodine-131.


 DeleteMissing
iodine 131
"Dataset"

9.2 What Are Isotopes?

Matter consists of atoms that can be described in a simple way as a central part
named the nucleus, containing neutrons and protons, surrounded by an electron
cloud. Atomic elements have the same number of protons, but the number of
neutrons may change. Nuclei with the same number of protons, Z, and a different
one for neutrons, N, are the isotopes of that element. Normally, an isotope is
written as A X , where X is the symbol of the element and A is the mass number
given by A = Z + N. For example: atmospheric carbon consists of 3 isotopes:
Carbon-12 (or 12 C) with 6 protons and 6 neutrons, Carbon-13 (or 13 C) with 6
protons and 7 neutrons and Carbon-14 (or 14 C) that has 6 protons and 8 neutrons.
All isotopes of the same element possess the same chemical properties but different
physical ones.
In nature there are 92 elements (we can also find traces of elements with Z > 92,
such as plutonium, but these are artificial elements), with an average of 2 or 3
isotopes each, although there are some elements with only one stable isotope and
    303

others with as many as 8. Isotopes can be stable or unstable. Stable nuclei, the
majority on earth, last “forever”, that is, they always keep the same number of
neutrons and protons. Some theories predict the disintegration of protons, and
therefore of any kind of atom, but they have not been proven yet. Furthermore,
even if they could be validated, the disintegration speed would be so slow that for
all practical purposes we could consider that stable nuclei have an almost eternal
duration. Unstable nuclei (called radioisotopes) are those that over time (depending
on the radioisotope the period can range from a fraction of a second to billions of
years) are transmuted into other elements, called daughter isotopes. The new
nucleus can in turn be stable or unstable. If it’s unstable, the process continues until
the daughter isotope is a stable one.

With
 [“element”, “property”] we can obtain all the known isotopes for a
given element. The element can be written as a complete word or, preferably, as a symbol
using standard notation. If we type “Symbol” in “property”, we’ll get the isotopes in A X
notation. We use Short to display only one line, but the complete output includes more
isotopes.


 IsotopeData
"C", "Symbol"   Short
 
  
8 C, 9 C, 10
C, 11
C, 12
C, 13
C, 3, 17
C, 18
C, 19
C, 20
C, 21
C, 22
C

The decay of a type of nucleus into another one usually happens due to the
emission of particles  or and radiation . In some radionuclides, there are also
emissions of neutrons and protons and even other types of reactions such as
spontaneous fission (SF), common in some heavy nuclei. Fission is the division of
a nucleus into two smaller ones, each one with approximately half the mass of the
original.
Using “DecayModes” or “DecayModeSymbols” as properties we can get the type
of decay for a certain isotope. One isotope can display more than one decay mode
although normally one of them would be the most common one. Using
“BranchingRatios” and “DecayEnergies” we can see, respectively, the probability
associated to each decay mode and its energy measured in keV (kiloelectron-volt).

The command below shows the different decay modes of Uranium–238, and the symbols,
probabilities and energies associated to them.


 Grid
IsotopeData
"Uranium238",   & 
"DecayModes", "DecayModeSymbols", "BranchingRatios", "DecayEnergies"
AlphaEmission SpontaneousFission DoubleBetaDecay
 SF 2 
 

1.00 5.45  107 2.2  1012
4269.75 keV — 1144.2 keV
Notice that the majority of the radioactive decays for U238 correspond to alpha emissions
(the value 1.00 indicates that the probability is close to 100%). However, spontaneous
fissions (SF), even though they have a small probability of happening, play a central role in
chain reactions since they emit neutrons, essential to start a chain reaction.

In short, we can say that a radioactive isotope is like a father who has a
daughter, who in turn may have a granddaughter and so on until eventually
there’s a descendant without offspring, the stable nucleus. The disintegration
 304      

happens at a constant pace, and its characteristics depend on the particular

isotope. In this process, particles (  , , ...) with specific energies are emitted.

You can use the free-form input to get the definition of chain reaction. Probably when you
type “chain reaction definition” you will see several definitions. Just choose the most
appropriate one depending on your needs.


  chain reaction definition



We can also get information by classes. We use Shallow to avoid displaying the complete
output, which is quite large. If you want to see all the classes just remove the command from
the next input.


 IsotopeData
"Classes"  Shallow
 
 

 alpha emission , beta decay , beta delayed alpha emission ,

beta delayed deuteron emission , beta delayed fission , beta delayed four neutron emission ,
beta delayed neutron alpha emission , beta delayed neutron emission ,
beta delayed three neutron emission , beta delayed triton emission , 37

Each class includes isotopes sharing a common characteristic. For example, if we want to
know the isotopes with a beta emission associated to spontaneous fission (a very rare
occurrence, only observed in 3 isotopes out of the more the 3,000 available), the command
below gives us the answer:


 IsotopeData
beta delayed fission ISOTOPES 

 
  actinium-230 , protactinium-236 , protactinium-238 

9.3 Decay Constants, Decay Periods, and Half-Lives

The easiest case of radioactive decay is that of a radioactive isotope A that decays
into a stable nucleus B, that is, A B, where B is not radioactive. This process can
be represented by a very basic compartmental model, such as the one shown in
Figure 9.1 with x t representing the quantity of a certain radioactive isotope A
present at time t in a sample and being the decay constant, specific to each
isotope. The quantity of atoms (measured in the most appropriate way: units,
grams, becquerels, etc.) that decay per unit of time is proportional to the quantity in
the sample at each moment, that is: xt.
    305

xt B

Figure 9.1 Radioactive decay of isotope A.

Let’s assume that at t  0 the sample contains x0 . This process can be

mathematically modeled by a simple first-order differential equation:
x
  x t, with initial condition : x0 at t  0 9.1
t

The solution to the previous differential equation can be obtained using DSolve. Notice that
we use “=” and not “:=” Why? (normally is better to use “:=”, but in this case we choose
“=” since the next time we call the function it will already know the solution without having
to calculate it again. In this case, the time saving is negligible).


 x1
t, x0,  x
t . DSolve
x
t   x
t, x
0  x0, x
t, t1
t
 
 x0 
We type it again using the typical textbook format
t
x t  x0  9.2

In the next example, we use and the total time t as our changing parameters. We also use
the previously defined function within Manipulate using the command Initialization.
Alternatively, we could have used SavedDefinitions.


 Manipulate Plot
y
t, , 1 , t, 0, t1, Filling Bottom,
, 1, 0, 5, t1, 3, 1, 10,
 t
Initialization y
t, , x0 : x0 
306      

In many situations instead of the
  , , it’s better to use the semi-decay period
or    , T12 , defined as the time that it would take for the amount x0 present in the
sample at time t  0 to become half, x0  2.

The relationship between and T12 can be calculated as follows:

  cteDes  . Solve
x1
T, x0,   x0  2 , , Reals1
log2
 

T
Another concept is the    , the life expectation (duration) of the isotopes present
in a sample. It’s the same as calculating the arithmetic mean of the expected duration of the
individual isotopes present in a sample.
 

To calculate the mean of a continuous variable t we compute t = 0 t f t t  0 f t x,
and associate the result to the symbol . We specify that  >0, otherwise we’ll receive a
message stating that the solution is only valid if  >0.


  Integrate
t x0 E ^ t  , t, 0, Infinity,
Assumptions   Reals &&  0  Integrate
x0 E ^ t  ,
t, 0, Infinity, Assumptions   Reals &&  0
1
 


In summary: the radioactive decay constant , the half-life T12 , and the mean
lifetime , are connected as follows: T12  Log 2    Log 2 ,  = 1/ .

With IsotopeData we can find out the half-life, mean lifetime or decay constant of
Iodine-131.


 IsotopeData
"Iodine131",  &  "HalfLife", "Lifetime", "DecayConstant" 

 
 6.9338  105 s, 1.0003  106 s, 9.9967  107 per second

By default the function returns the time in seconds, but we can convert it to other unit using
UnitConvert.
UnitConvert

IsotopeData
"Iodine131",  &  "HalfLife", "Lifetime" , "days"

In the example below we can choose the isotope among several options. We use
QuantityMagnitude to get only the value (in this case, the decay constant), without the unit.


 Manipulate Plot
y
t, 3600  , 1 , t, 0, t1, Filling Bottom,
t1, 3, "Hours", 1, 100,
, QuantityMagnitude
IsotopeData
"Iodine131", "DecayConstant"
"Iodine 131", QuantityMagnitude

IsotopeData
"Iodine125", "DecayConstant" "Iodine 125",
Initialization y
t, , x0 : x0  t 
    307

The next example uses IsotopeData to show all the isotopes for iodine. We also include the
option of selecting the unit for displaying the evolution over time for the chosen isotope.
Since the decay rate for some of the isotopes is very fast we use a logarithmic scale.


 Manipulate LogPlot
y
t, k   , 1, t, 1, t1, Filling Bottom,
t1, 3, 1, 100 , k, 1 "Seconds", 3600 "Hours", 3600  24 "Days",
, "", "Isotope",
Thread
QuantityMagnitude
IsotopeData
"Iodine", "DecayConstant" 
IsotopeData
"Iodine", Initialization y
t, , x0 : x0  t 

For a given radioisotope, activity A, the decay rate in 1 second (1 nuclear

transformation = 1 Becquerel or Bq), can be obtained from the decay constant, (in
s1 ). The activity of nat atoms will be A  nat . Since what is known usually is the
mass, m, we apply the conversion nat = m NA  ma , where NA is the Avogadro’s
number and ma the atomic mass of the element expressed in the same units as m,
normally in grams (g).
  

Iodine-131 emissions from the Fukushima Daiichi nuclear reactor, due to the
accident caused by the tsunami in March 2011, were estimated to be 1.51017 Bq as
308      

of the end of April 2011. Given this information, let’s explore how much mass was
emitted and how the emissions evolved over time.
Activity can be turned into mass using the formula A  nat = m N A  ma
m  ma A   NA .

Iodine-131 decays into Xenon-131, a stable isotope that at room temperature is gas. We use
LayeredGraphPlot to represent the decay chain.


 LayeredGraphPlot
IsotopeData
"I131", "Symbol"
IsotopeData
IsotopeData
"I131", "DaughterNuclides"1, "Symbol",
Left, VertexLabels "Name"
131
I
 

131
Xe

We need to know the decay constant, 131, and the atomic mass. IsotopeData will give us
the latter (although we could have used 131 as an excellent approximation) and also return
the mean lifetime  (in seconds). Afterward we can calculate =1/.


  mat131, 131 IsotopeData
"Iodine131", "AtomicMass",
1  IsotopeData
"Iodine131", "Lifetime" 
   130.906124609 u, 9.9967  107 per second

Now we use the formula m  ma A   NA , where NA (Avogadro’s number) is given by

Mathematica using its natural language capabilities ( “Avogadro’s Number”) or with the
function Quantity. UnitConvert shows the SI units, in this case just the number.


 Quantity
"Avogadro's Number"

 
 N0


 NA UnitConvert
1 NA 

 
 602 214 076 000 000 000 000 000


 1.5  1017 QuantityMagnitude
mat131  QuantityMagnitude
131  NA "grams"

 
 32.617 grams
It may seem strange at first sight that such a big radioactive activity would only result in a
very small mass. What’s actually happening is that the Bq is an extremely small unit. As a
matter of fact, for many years instead of using the Becquerel to measure radioactivity,
people used the Curie, Ci (1 Ci = 3.7 1010 Bq), unit that’s still in use in some places.

To find out in how many days the radioactivity of the Iodine-131 would become just 0.1%
of its initial figure, we use the following command:


 UnitConvert
t . Solve
x1
t, 100, 131  0.1, t, "days"

                        

       

 

79.9774 days

9.4 Decay Chains

  Clear
"Global`"

As we’ve seen, a radioactive isotope is like a father who keeps on having

descendants until the offspring is a stable nucleus. Sometimes, the first descendant
    309

is stable right away, as with Iodine-131 Xenon-131, but other isotopes, like
Radium-226, shown at the end of this section, have more complicated decay chains.
One of Radium-226’s daughter isotopes, Radon-222 (Rn222), is responsible for a
large share of the natural radiation that people are exposed to. Each time we
breathe, a portion of the air includes radon gas. The content of Rn222 changes
substantially from one area to another. For example: granitic houses with poor
ventilation usually have a high content level of radon. The reason is that granite
contains Radium-226 that becomes radon gas when it disintegrates.

An isotope can decay following different branches. For example: Radium-226 decays with a
half-life period of     s into 3 radioisotopes: 222 Rn, 212 Pb and 226 Th. Almost all of
it decays into 222 Rn and an insignificant part into 212 Pb. Occasionally, there can also be
traces of Thorium-226.


 ra226 IsotopeData
"Ra226",  & 
"DaughterNuclides", "BranchingRatios", "HalfLife";
TableForm
ra226, TableHeadings "Isotope", "Fraction", "T s ", None
 
 !

Isótopo radon-222 lead-212 thorium-226

fracción 1.00 2.6  1011 —
T s 5.0  1010 s

In different contexts it is likely that when using IsotopeData you may find that the output
contains the word “Missing”. If you want to eliminate it, you can use DeleteMissing or
DeleteCases[list, _h], which eliminate all the elements in a list whose head starts with h, as
shown in the example below:


 is1 IsotopeData
"U238", "DaughterNuclides"

 
  thorium-234 , —, plutonium-238 


 is2 DeleteCases
is1, Missing

 
  thorium-234 , plutonium-238 

The set of all branches produced by the decay of a parent isotope (like Uranium-238) is
known as the radioactive decay chain. However, the majority of those disintegrations happen
along only one branch. IsotopeData["isotope", "BranchingRatios"] shows the results after
sorting the branches from bigger to smaller in terms of their likelihood of occurrence. When
an isotope decays into one or several isotopes, we can choose to keep just one: the one with
the highest probability. If we apply this criterion to each daughter isotope, we will end up
with a single branch named the main branch. From a practical point of view, in many decay
chains it is enough to study the main branch.

The following function lets us choose the main branch, and within it, the number of
daughter nuclides n that we are interested in. Notice the use of iso_String, and n_Integer to
limit the input type to: “isotope name” (between double quotation marks), and n integer
(number of daughter nuclides that we would like to display); in any other case, the function
will not be executed.


 mainbranch
isoString, nInteger :
NestList
First
IsotopeData
, "DaughterNuclides" &, iso, n

We apply it to Uranium-238 and indicate 14 daughter nuclides, the entire decay chain (one
of the longest ones).
310      


 mainbranch
"U238", 14

 
 U238, thorium-234 , protactinium-234 , uranium-234 , thorium-230 ,

radium-226 , radon-222 , polonium-218 , lead-214 , bismuth-214 ,

polonium-214 , lead-210 , bismuth-210 , polonium-210 , lead-206 

This problem was posted in “comp.soft-sys.math.mathematica”. The function below shows

the elegant solution proposed by one of the list members (Bob Hanlon) that returns a
complete branch without the need to specify n. However, if we are interested in limiting the
number of offsprings that we wish to see, it would be better to use mainbranch.


 mainbranch1
xString :
NestWhileList
IsotopeData
, "DaughterNuclides"1 &,
x, UnsameQ, 2, 100, 2  Quiet;

Let’s apply it to Radon-226


 ra226 mainbranch1
"Ra226"

 
 Ra226, radon-222 , polonium-218 , lead-214 , bismuth-214 ,

polonium-214 , lead-210 , bismuth-210 , polonium-210 , lead-206 

The function below generates a list of isotopes showing: their disintegration period,
daughter nuclides and branching ratios.


 isopro
isolist :
Module
vals, vals Table
IsotopeData
, prop, prop,
"Symbol", "HalfLife", "DaughterNuclides", "BranchingRatios" & 
isolist;
Text
Grid
Prepend
vals, "", "Disintegration\nperiod sec ",
"Daughter nuclides", "Branching ratios", Frame All,
Background None, LightBlue, White, 1 LightYellow 

We use it to show the information related to the principal branch of Radon-226 obtained in
the previous calculation.


 isopro
ra226

Disintegration Daughter nuclides Branching ratios

period
sec
226
Ra 5.0  1010 s
radon-222 , lead-212 , thorium-226 
1.00, 2.6  1011 , —
222
Rn 330 350. s
polonium-218  1.00
218
Po 185.9 s
lead-214 , astatine-218  1.00, 0.00020
214
Pb 1610. s
bismuth-214  1.00

 

214
Bi 1.19  103 s
polonium-214 , 1.0, 0.00021, 0.000030
thallium-210 , lead-210 
214
Po 0.0001643 s
lead-210  1.00
210
Pb 7.01  108 s
bismuth-210 , mercury-206 
1.00, 1.9  108 
210
Bi 4.33  105 s
polonium-210 , thallium-206 
1.00, 1.32  106 
210
Po 1.19557  107 s
lead-206  1.00
206
Pb   
    311

The functions below are part of the documentation related to LayeredGraphPlot. Sometimes
we may get lucky and find examples in the documentation that perfectly match what we
want to do. This is one of those occasions.

The following functions (from ref/LayeredGraphPlot) let us input the parent isotope (the
first in the chain) and get as a result all the daughter nuclides from all the branches except
those that have an insignificant contribution (“Missing”).

  DaughterNuclides
sList :
DeleteCases
Union
Apply
Join, Map
IsotopeData
, "DaughterNuclides" &,
DeleteCases
s, Missing, Missing


 ReachableNuclides
sList :
FixedPoint
Union
Join
, DaughterNuclides
 &, s


 DecayNetwork
isoList :
Apply
Join,
Map
Thread
 DaughterNuclides
 &, ReachableNuclides
iso


 DecayNetworkPlot
s :
LayeredGraphPlot
Map
IsotopeData
, "Symbol" &, DecayNetwork
s, 2,
PerformanceGoal "Quality", VertexShapeFunction
Text
Framed
Style
2, 8, Black, Background LightYellow, 1 & 

We use the last one to show the decay chain of Rn222, the immediate decay product of
Ra226. You can also apply the same function to see the complete chain of the latter. You
may have to adjust the output image size to see it properly. To do that just place the cursor
in one of the corners of the image and resize it accordingly.
312      


 DecayNetworkPlot
"Rn222"

222
Rn

218
Po

214 218
Pb At

214 218
Bi Rn

210 214
 
 Tl Po

209 210
Pb Pb

209 210 206

Bi Bi Hg

205 210 206

Tl Po Tl

206
Pb

The following expression returns all the isotopes that have the same number of neutrons:


 isoneu
neut :
Select
IsotopeData
, IsotopeData
, "NeutronNumber"  neut &

  isoneu
14

 
  boron-19 , carbon-20 , nitrogen-21 , oxygen-22 , fluorine-23 , neon-24 ,

sodium-25 , magnesium-26 , aluminum-27 , silicon-28 , phosphorus-29 ,

sulfur-30 , chlorine-31 , argon-32 , potassium-33 , calcium-34 

Using the previous function we display the decay chains of all the isotopes with 4 neutrons.
    313


 LayeredGraphPlot
Table
IsotopeData
, "Symbol" &  i  & 

IsotopeData
i, "DaughterNuclides", i, isoneu
4  Flatten,
DirectedEdges True, VertexLabeling True
9 12 11 5
B O N H

8 6 12 10 3
Be He N C H
 


4 6 10
He Li B

9.5 Application: Modeling the Evolution of a Chain of Isotopes over

Time
A radioactive decay chain can be considered as a system of compartments in which
the flow is one-directional, as shown in the figure below. This is an example of
what is known in compartment modeling as a catenary model.
Graphics
Circle
0, 0, 0.5 , Text
"1", 0, 0, 0, 0,
Text
"1 ", 1.75, .4, 0, 0, Arrow
0.5, 0, 3, 0,
Circle
3.5, 0, 0.5 , Text
"2", 3.5, 0, 0, 0,
Circle
7, 0, 0.5 , Text
"3", 7, 0, 0, 0,
Arrow
4, 0, 6.5, 0, Text
"2 ", 5.25, .4, 0, 0,
Dashed, Arrow
7.5, 0, 10.5, 0, Text
"i ", 9, .4, 0, 0,
AspectRatio Automatic, Axes None

1 2 i
1 2 3

The first circle represents the first isotope in the chain, with a decay constant
 ,
the second circle refers to the next isotope, with a decay constant
 , and so on.
Let’s consider the case of a chain of 3 isotopes, with q1 t, q2 t, and q3 t the
quantities present of each one at time t, satisfying the following set of differential
equations:
q1 ' t   1 q1 t
q2 ' t  1 q1 t  2 q2 t
q3 ' t  2 q2 t  3 q3 t

As an initial condition let’s assume that at t  0 there’s a quantity (atoms, gm, Bq) of isotope
1 in compartment 1 and nothing in the rest. This means: q1 0  b1, q2 0  q3 0  0 then:
314      


 eq DSolve
q1 '
t  1 q1
t,

q2 '
t  1 q1
t 2 q2
t, q3 '
t  2 q2
t 3 q3
t,
q1
0  b1, q2
0  0, q3
0  0, q1
t, q2
t, q3
t, t;

The retention in compartments 1, 2, and 3 is given by:


 q1
t, q2
t, q3
t q1
t, q2
t, q3
t . eq1  Simplify
1 t 2 t
b1 1   
1 t
 
 b1  , ,
1 2
b1 1 2 t  1 2 3
 1  2 t  1 2
 3  1 t  1 3
 2  3 t  2 3


 1  2  1  3  2  3 

Grouping identical exponential terms we get:
b1 -t 1
1 2 b1 -t 3
1 2 b1 -t 2
1 2
q3 (t) = + + =
( 1 - 2) ( 1 - 3) ( 1 - 3) ( 2 - 3) (- 1 + 2) ( 2 - 3)
-t 1
-t 3
-t 2
b1 1 2 + + =
( 1 - 2) ( 1 - 3) ( 1 - 3) ( 2 - 3) (- 1 + 2) ( 2 - 3)
3-1 3 - j t
= b1  j 
j=1 j=1 np=1,pj ( p - j)

Following the same approach, the previous equation can be extended for n daughter
isotopes:

t
n1 n  j

qn t  b1  j  , with initial conditions : q1 t  b1 , q2 0 . ..  qn 0  0 9.3

j 1 j1 pn1,pj  p  j

This equation is known as Bateman’s equation, in honor of the mathematician who

first published it.

We can type the equation as follows:
n kj t n 1


 q
n : b   kj
j 1
j 1 p 1 kp kj  pn j1 kp kj  j 1

Here we apply it to the parent and the first 3 daughter isotopes (of the main branch) of
Uranium-238.


 u3 mainbranch
"U238", 3

 
 U238, thorium-234 , protactinium-234 , uranium-234 

We need the decay constants . Remember that  = 1/ , with  in seconds. However, since
we prefer to measure t in days we multiply it by 24  3600.


 s Thread
k1 , k2 , k3 , k4  Evaluate

24 3600  QuantityMagnitude
IsotopeData
, "Lifetime" &  u3
 
 k1 4.248  1013 , k2 0.029, k3 2.5, k4 7.735  109 
    315

Next we’re going to calculate the evolution of the radioactivity, Ai , in the previous isotopes.
We assume that at time t  0 there’s 1 gm of U238 in the sample, which is expressed as the
number of its atoms, NA (Avogadro’s number)/238. Now we can compute Ai , using the
general formula A = NA .


 A1
t, A2
t, A3
t, A4
t
k1 q
1, k2 q
2, k3 q
3, k4 q
4 . s .
b UnitConvert
1 NA   238  ExpandAll
13 13
   1.075  109 4.24810 t
, 1.1  109 4.24810 t
 1.1  109 0.029 t ,
13
1.3  107 2.5 t  1.1  109 0.029 t 1.1  109 4.24810 t
,
2.5 t 2 0.029 t 9 7.735109 t 13
0.04  3.  10   1.1  10  1.1  109 4.24810 t


We subtract 108 Bq from the activity of Pa234 with the only purpose of differentiating it
from Th234. Otherwise both graphs would be superimposed since they are in equilibrium as
we will see shortly.


 Plot
A1
t, A2
t 10 ^ 8, A3
t 10, t, 0, 180,
PlotRange All, AxesLabel "t days ", "A t in Bq",
PlotLegends Placed
"U238", "Th234", "Pa234", Center
At in Bq

1  109

8  108

6  108 U238
 

Th234
4  108
Pa234
8
2  10

tdays
50 100 150

In the graph above we can see that after 180 days, A1A2A3. This is known as secular
equilibrium. In this example, it’s enough to calculate the radioactivity of the isotope parent
to know the daughter isotopes’ radioactivity. This happens when 1 < 2 .

The next example shows that given enough time, in this case thousands of years, U238 and
U234 will reach secular equilibrium. Normally we consider that the secular equilibrium has
been reached after a period of time longer than 8 half-life periods of the daughter isotope.
316      


 Plot
A1
t, A4
t, t, 0, 10 ^ 9,

PlotRange All, AxesLabel "t days ", "A t in Bq",
PlotLegends Placed
"U238", "U234", Center
At in Bq

1  109

8  108

8
 
 6  10 U238
U234
4  108

2  108

tdays
2  108 4  108 6  108 8  108 1  109

9.6 Application: Dating the History of Humankind

The fact that radioactive isotopes decay following a well-known physics law even
under extreme conditions, makes them the ideal candidates for dating different type
of samples. The actual isotopes used for that purpose will depend, among other
factors, on the estimated age of the object that we want to date. We are going to
discuss two methods: Carbon-14 (14 C) dating is suitable for organic samples less
than 50,000 years old and the lead-lead method is used when dating rocks
thousands or even millions of years old.

Radiocarbon Dating
The 14 C dating method was discovered in 1947 by Willard Libby. It can be applied
to samples that are no more than 50,000 years old. This method has become the
most frequently used one for measuring the age of organic objects, especially
archeological ones.

The table that follows displays the isotopic abundance (proportion of isotopes) of the
different carbon isotopes and their half-lives (If “HalfLife”< 1 s is not shown). An isotopic
abundance of 0 indicates that the isotope doesn’t exist in nature or is present only as traces.
For example: the proportion of 14 C is very small although, as we will see, it plays a crucial
role in dating.


 vals DeleteCases
Table
IsotopeData
, prop,
prop, "Symbol", "HalfLife", "IsotopeAbundance" & 
IsotopeData
"C", a, b ; b  1 s , c;
    317

Text

Grid
Prepend
vals, "", "Half lives", "Isotope abundance", Frame All,
Background None, LightBlue, White, 1 LightYellow

Halflives Isotope abundance

10
C 19.29 s 0.
11
C 1220.0 s 0.
12
 
 C  0.9889
13
C  0.0111
14 11
C 1.8  10 s 0.
15
C 2.449 s 0.

This method is based on the following principle: 98.9% of the carbon in the Earth’s
atmosphere is Carbon-12, 1.1% Carbon-13, and the rest, 1.18 1014 % is Carbon-14. This
last isotope is the key to the method.
14

The command below returns the decay scheme of C. It decays into 14 N with a beta
emission (1 electron) and has a half-life period of     s (approximately 5,700 years)


 c14 IsotopeData
"C14",  & 
"DaughterNuclides", "BranchingRatios", "DecayModes", "HalfLife"

 
  nitrogen-14 ,
1.00,
BetaDecay, 1.8  1011 s

The nitrogen present in the atmosphere is subject to a continuous neutron

bombardment from cosmic radiation. Some of these neutrons strike the nitrogen
atoms with enough energy to transform them into 14 C that will eventually decay
into 14 N.
If we assume that the cosmic radiation has not changed significantly during the last
50,000 years, the production rate of 14 C atoms will be constant. Since its
disintegration period is relatively short (approximately 5,700 years), the quantity of
atoms that are produced will be similar to the ones that disintegrate, so the
percentage of this isotope out of the total amount of carbon in the atmosphere will
not have changed over the last millennia. Living organisms continuously absorb
carbon when breathing or performing photosynthesis but when they die, this
absorption stops. Two (Carbon-12 and Carbon-13) out of the three carbon isotopes
are stable and will remain so; the other one, Carbon-14, will decay due to its
radioactivity and its percentage of the total amount of carbon will get smaller.
Measuring the content of this latter isotope in an organic sample and comparing it
with the total amount of carbon in the entire sample will allow us to know its age.
In reality, the method is not that simple since the proportion of 14 C in the
atmosphere is not constant, but fluctuates with solar activity, introducing the need
to make adjustments. Furthermore, we have to take into account 14 C originated
from non-organic sources, such as nuclear explosions, mainly during the fifties and
beginning of the sixties in the 20th century.
The evolution of 14 C can be inferred by measuring its content in the trunks of very
old trees. Tree trunks consist of layers or rings, with each one corresponding to one
year in the life of the tree. By counting the number of rings starting from the bark
318      

we can know the age of a tree. This dating technique is known as

dendrochronology. Using this method, scientists have been able to “calibrate”
the 14 C method for samples up to 9,000 years old. This calibration has recently
been extended to 40,000 years. This has been possible thanks to the comparison
between this method and the measurement of U234 and Th230 in corals. When
corals are born, their U234 content is known. From that moment on Th230 starts to
accumulate. Comparing the content of C14 with Th230, we can determine the age
of the coral. Since corals can be dated using the 14 C method, it’s possible to
compare both techniques.
Example: There are 2 grams of carbon in a piece of wood found in an archeological
site and we discover that it has an activity of 10 nuclear transformations per minute
per gm. How old is the piece of wood? For simplification purposes we will assume
that the C14 content in the wood at the time it was cut is the same as in wood
recently cut: 15 nuclear transformations per minute per gm. However, in real cases,
as explained before, we may have to make some adjustments regarding this
assumption.

This problem is a case of a decay chain with only one daughter isotope: A = A0   ,
with  being the decay constant of C14. Therefore, the age of the sample would be:
t


 Solve  A  A0 , t, Reals .
 1  UnitConvert
IsotopeData
"C14", "Lifetime", "year",
A 10, A0 15
 
 t 3.3  103 yr

The Age of the Earth

To estimate the age of the Earth, we need to know its origins. The mainstream
theory of the formation of the solar system and probably, of other planetary
systems similar to ours, is as follows: there are certain areas in galaxies where
clouds of dust and gas accumulate, and once these clouds reach certain mass and
density they may condense as a result of gravitational attraction forces. This
phenomenon can benefit from, or depends on?, the explosion of a star (supernova)
in its proximity. During the condensation, the clouds may break into smaller units
that become protostars. These protostars continue contracting rapidly until their
centers reach very high temperatures and densities causing the fusion (union) of the
lightest atomic nuclei. This fusion emits enough energy to stop the gravitational
collapse. At this moment, stars are born. Surrounding them there are still many
fragments. Some of these fragments start clustering during successive collisions
until they reach a certain size and become planets. This is probably the origin of the
planets closest to the Sun, such as the Earth. The fragments that didn’t become part
of a bigger mass were left wandering around the solar system, originating some of
the currently existing meteorites.
Over long periods of time, orogenic and sedimentary processes have destroyed the
remains of the rocks that originally formed the Earth (primordial), making it
impossible for us to date the Earth directly from them. Fortunately, we’ve been able
to calculate its age using visitors from outer space: meteorites. Their analyses
    319

supports the idea that after the formation of the solar system, with some exceptions,
the distribution of isotopes was homogeneous.
Among meteorites, the carbonaceous chondrite type is particularly interesting.
Chondrites are made of chondrules, molten droplets created during the collisions
that gave birth to planets, that have the isotopic composition of the planet-forming
period. Furthermore, they didn’t experience any further heating intense enough to
melt them again. It’s been possible to determine their age using dating techniques
based on radioactive isotopes with long half-lives. The most commonly used
methods are: the lead-lead method (Pb-Pb) and the rubidium-strontium one (Rb-
Sr). Next, we’re going to discuss the Pb-Pb method.

The decay chains of Uranium-238 and Uranium-235 are of particular interest among natural
206 207
isotopes. Their final nuclides, are respectively: Pb and Pb.


 Row
IsotopeData
, "Symbol" &  mainbranch1
"U238", " "
238 234 234 234
   U Th Pa U
230 226 222 218 214 214 214 210 210 210 206
Th Ra Rn Po Pb Bi Po Pb Bi Po Pb


 Row
IsotopeData
, "Symbol" &  mainbranch1
"U235", " "
235 231 231 227 227 223 219 215 211 211 207 207
   U Th Pa Ac Th Ra Rn Po Pb Bi Tl Pb
206 207

In nature, apart from Pb and Pb, we also have other stable lead isotopes in the
following proportions:


 pbestable DeleteCases
Transpose
IsotopeData
"Pb",
IsotopeData
, "Stable" &  IsotopeData
"Pb" , , False;


 vals
Table
IsotopeData
, prop, prop, "Symbol", "IsotopeAbundance" & 
First
Transpose
pbestable;


 Text
Grid
Prepend
vals, "Isotope", "Isotope Abundance", Frame All,
Background None, LightBlue, White, 1 LightYellow

Isotope Isotope Abundance

204
Pb 0.014
206
 
 Pb 0.241
207
Pb 0.221
208
Pb 0.524

These proportions may vary widely depending on the origin of the sample.

208 Pb is a stable daughter isotope of the decay chain of Th232, and doesn’t play any role in
the dating method we are about to describe. 204 Pb, however, is a primordial isotope. This
means that the amount of this nuclide present on Earth has not changed since the formation
of our planet. Therefore it’s useful for estimating the fraction of the other lead isotopes in a
given sample that are also primordial since their relative fractions are constant everywhere.


 Row
IsotopeData
, "Symbol" &  mainbranch1
"Th232", " "
232 228 228 228 224 220 216 212 212 212 208
   Th Ra Ac Th Ra Rn Po Pb Bi Po Pb

At the moment of the formation of the solar system, U238 and U235, after going
through an homogenization process, started to decay, respectively, into Pb-206 and
Pb-207 thus adding to the existing amount in the rocks.
320      

The quantity from the parent isotope that decays into a daughter isotope is:
Nh t  1  Np t, that is:
t
Nh  Np 

As components of rocks or meteorites, the total number of atoms in the systems

U238 + Pb206 and U235 + Pb207 remains constant.
207 PbP  207 PbI 235 U  235 t
 1

206 PbP  206 PbI 238 U  238 t

 1

with the P and I subscripts indicating, respectively, the present and initial
quantities, 235 and 238 the decay constants of U235 and U238, and t the elapsed
time.
There will also be a certain amount of Pb204 that will not change. Under these
hypotheses we obtain the following relationships:
207 207 235
Pb Pb U
235 t
   1
204 204 204
Pb P Pb I Pb
206 206 238
Pb Pb U
238 t
   1
204 204 204
Pb P Pb I Pb

Rearranging the previous equations we get:

207 207 235
Pb Pb U
235 t
    1
204 204 204
Pb P Pb I Pb
206 206 238
Pb Pb U
238 t
    1
204 204 204
Pb P Pb I Pb

After dividing the first one by the second one:

207 207 206 206 235 t
Pb Pb Pb Pb 1   1
  !    K
204 204 204 204 238 235 238 t
Pb P Pb I Pb P Pb I U U   1
238

The ratio U/ 235 U is:


 IsotopeData
"U238", "IsotopeAbundance" 
IsotopeData
"U235", "IsotopeAbundance"
 
 137.80
238 235
However, in the scientific literature: U/ U =137.88, and this is the value we are going
to use.
 
  
Since   
  
is constant for samples from the same period, the relation
207
Pb 206 Pb
204 /
Pb 204 Pb
is a straight line, as Figure 9.2 shows:
     321

Figure 9.2 Paterson’s lead isotope isochron.

Source: http://en.wikipedia.org/wiki/File:Paterson_isochron _animation.gif

The different lines obtained depend on the original U/Pb ratio. For samples coming from the
oldest meteorites (see the graph ), K = 0.61, meaning that t, in eons (billions of years) is:
235 t
1 1


 K .
137.88 238 t
1
235 1  QuantityMagnitude
UnitConvert
IsotopeData
"U235",
"Lifetime", "eons", 238 1  QuantityMagnitude

UnitConvert
IsotopeData
"U238", "Lifetime", "eons";


 FindRoot
K  0.61 , t, 0.5

 

t 4.54709

We can compare this result with the one given by Mathematica using the free-form input
and see that is quite similar.


  Age of Earth

PlanetData
Entity
"Planet", "Earth", "Age"

 
 4.54  109 yr

Orogenic phenomena have left us without remains of rocks from the time of the
formation of our planet. The oldest ones have been found in zircon crystals from
Mount Narryer, in Western Australia. They are estimated to be more than 4,000
years old, proving that at that time the Earth already had a continental crust.

9.7 Application: Calculating Binding Energies

The mass of an atom is always smaller than the sum of its component particles.
This difference in mass is known as binding energy and is responsible for keeping
the particles together to form the element. To obtain some property for a subatomic
particle we use IsotopeData.

For example, the iron isotope Fe56 consists of N neutrons, and Z protons, whose actual
numbers can be calculated as follows:
 322      


 Nfe56, Zfe56

IsotopeData
"Fe56",  &  "NeutronNumber", "AtomicNumber"
 

30, 26

The sum of the particle masses that form the iron isotope (expressed in MeV  c2 ): mA = N
mn + Z (mp +me ):


 fe56particlesmass Nfe56 ParticleData
"Neutron", "Mass" 
Zfe56 ParticleData
"Proton", "Mass" 
Zfe56 ParticleData
"Electron", "Mass"
 
 52 595.320 MeVc2

We calculate below the conversion factor of 1 amu (atomic mass unit) to 1 MeV:


  AtomicMassUnit in MeV



 
 931.494102 MeV2

We find the mass of the iron isotope Fe56 and express it in MeV:


 fe56mass

UnitConvert
IsotopeData
"Fe56", "AtomicMass", 931.494 c 

 
 52 103.0644 MeVc2

Next we calculate the binding energy per nucleon B, in MeV as well:


 fe56particlesmass fe56mass  Nfe56  Zfe56
2
 
 8.79027 MeVc

We can get approximately the same value directly with (sometimes in MeV  c2 we omit c2
or in natural units we consider c  1):

  IsotopeData
"Fe56", "BindingEnergy"

 
 8.790323 MeV

Another commonly used concept is “excess of mass” calculated as:


 QuantityMagnitude
IsotopeData
"Fe56", "AtomicMass"
IsotopeData
"Fe56", "MassNumber" 931.494
 
 60.6054

We can also calculate it directly:


 IsotopeData
"Fe56", "MassExcess"

 
 60.605352 MeV

The function below displays the binding energies as a function of the mass number. With
Tooltip we can see the information related to a point when placing the cursor on it:


 bindingenergy Tooltip
IsotopeData
, "MassNumber",
IsotopeData
, "BindingEnergy" &  IsotopeData
;
    323


 ListPlot
bindingenergy, Frame True,

FrameLabel "Mass number A", "Binding energy per nucleon B , in MeV",
Axes False, ImageSize 300, 250
Binding energy per nucleon B, in MeV

8.5

8.0

7.5
 

7.0

6.5

6.0
0 50 100 150 200 250
Mass number A

The following command shows the binding energy per nucleon, as a function of the atomic
number and the number of neutrons until Z = 92 (uranium):


 bindingenergyZNB Flatten

Table
 z, a z, IsotopeData
z, a, "BindingEnergy", z, 1, 92,
a, IsotopeData
, "MassNumber" &  IsotopeData
z, 1;

Here we visualize it in 3D:


 ListPlot3D
bindingenergyZNB, AxesLabel "A", "N", "B"

 


Next we show an example with 224 nuclides with less than 15 protons each (Z < 15) and
plot it using the function ColorData["Atoms"] that associates each element to one color. We
can find the value of Z and N by placing the cursor inside the graph, right clicking with the
mouse button and selecting “Get Coordinates”.


 lessthan15 Take
bindingenergyZNB, 224;
324      


 ListDensityPlot
lessthan15, FrameLabel "N", "Z",

ColorFunction Map
ColorData
"Atoms", ElementData
, "Abbreviation" &,
Transpose
lessthan151

 


In this case, we present the same information but using a color code that associates redder
colors to higher binding energies:


 ListDensityPlot
lessthan15, FrameLabel "N", "Z",
InterpolationOrder 1, ColorFunction "Rainbow"

 


Decays and nuclear transformations are related to binding energies. There are two
types of nuclear transformations that are particularly useful: fusion and fission.
Fusion is the process of joining nuclei. This process is the most predominant one in
elements with low atomic numbers, specifically, until iron. The lower the atomic
number the lower the binding energy associated to it. Fusion is the normal method
by which stars generate energy.
Fission is the opposite process of fusion. It consists of splitting apart nuclei to turn
them into lighter ones.
Some isotopes of uranium and plutonium represent a special case where the process
accelerates when interacting with neutrons.

Some very heavy nuclei may experience spontaneous fission. The next function displays the
first five such elements:
    325


 Take
Transpose
Cases
Table
IsotopeData
, prop,

prop, "Symbol", "SpontaneousFission" & 
IsotopeData
, Except
, False1, 5
 
 230 Th, 232
Th, 231
Pa, 234
Pa, 230
U

Since IsotopeData includes “Spontaneous Fission” as a property, we can also get those
elements as follows (we don’t show the complete output):


 IsotopeData
spontaneous fission ISOTOPES   Shallow
 
 

 thorium-230 , thorium-232 , protactinium-231 , protactinium-234 , uranium-230 ,

uranium-232 , uranium-233 , uranium-234 , uranium-235 , uranium-236 , 193

Many of them are transuranic elements obtained by artificial means, although there are also
some isotopes in nature that exhibit this property. They are largely responsible for the
presence of free neutrons that have lifetimes of just a few minutes:


 ParticleData
"Neutron", "Lifetime"

 
 885.6 s

This can be explained by the fact that both, neutrons and protons, are not really elementary
particles but are made of two types of quarks:


 ParticleData
, "QuarkContent" &  "Proton", "Neutron"

 down quark , up quark , up quark 

 

 down quark , down quark , up quark 

Let’s see some of these quarks’ characteristics:


 values
Table
ParticleData
, prop, prop, "Symbol", "Charge", "Mass" & 
"DownQuark", "UpQuark"
1
d  5.0 MeVc2
3
 

2
u 2.2 MeVc2
3


 Text Grid Prepend values, "", "Charge", "Mass MeVc 2 ", Frame All,
Background None, LightBlue, White, 1 LightYellow

Charge Mass
MeVc2 
1
 
 d  5.0 MeV2
3
2
u 2.2 MeV2
3

Notice how small the mass of the quarks is compared to that of a neutron (two quarks down
and one quark up) or a proton (two quarks up and one quark down):


 qd, qu ParticleData
, "Mass" &  "DownQuark", "UpQuark"

 
 5.0 MeVc2 , 2.2 MeVc2 

326      


 quarksproton, quarkneutron

100  2 qu  2  1 qd  ParticleData
"Proton", "Mass",
2 qu  1 qd  ParticleData
"Neutron", "Mass" ""
 

1.5 , 1.0 
This means that the quarks only represent between 1% and 1.5% of the atomic nuclei mass
with the rest coming from their interactions. This is the reason why quarks are not present as
free particles in nature.
The mass of quarks and other particles such as electrons is attributed to the Higgs
field (more accurately to the Brout–Engler–Higgs mechanism), whose existence
was confirmed by the detection of the Higgs boson, officially announce by CERN
on July 4, 2012. In that case, what about our mass? Does it come as well from the
Higgs field? Only a small portion of it.
As we’ve seen, quarks only represent between 1% and 1.5% of the nuclei mass and
that mass is explained by the Higgs field. Therefore, for a person weighing 80 kg,
the Higgs field explains approximately 1 kg. What about the rest? It comes from
gluons, g, that keep the photons and quarks joined. In this case, we should refer to
mass equivalence, m = E/c2 . These particles don’t interact with the Higgs field or
BEH, but they interact with the gravitational field. However, without the Higgs
field, electrons would move at the speed of light, atoms would not have formed,
and therefore we wouldn’t exist.
The Standard Model of Particle Physics, apart from quark d and quark q, also
includes 10 additional types of quarks and antiquarks, the leptons, the gauge
bosons, and the Higgs boson. We will use the EntityClass function, which we saw
in Chapter 5, to represent some properties of these particles.


 TextSentences
WikipediaData
"StandardModel" ;; 2

 

The Standard Model of particle physics is the theory describing three of the four
known fundamental forces electromagnetic, weak and strong interactions,
omitting gravity in the universe and classifying all known elementary particles.,
It was developed in stages throughout the latter half of the 20th century, through the
work of many scientists worldwide, with the current formulation being finalized
in the mid1970s upon experimental confirmation of the existence of quarks.

A few properties of the gauge bosons, quarks, and leptons are summed up in the next tables.

  Dataset
EntityValue
EntityClass
"Particle", "GaugeBoson",
EntityProperty
"Particle", "Lifetime", EntityProperty
"Particle",
"Width", "EntityPropertyAssociation"
    327


 Dataset
EntityValue
EntityClass
"Particle", "Lepton",

"Symbol", "Charge", "Mass", "EntityPropertyAssociation"

  Dataset
EntityValue
EntityClass
"Particle", "Quark",

"Symbol", "Charge", "Mass", "EntityPropertyAssociation"
328      

9.8 Radiation Attenuation

Other function with applications in nuclear physics and radiological protection is:
StoppingPowerData. Let’s take a look at a few examples based on the
documentation pages.

One way to reduce radiation exposure is through the use of shielding. For example, the
thickness of a water layer needed to reduce photon radiation by a tenth is:


 StoppingPowerData
"Water", "Particle" "Photon",
"Energy" Quantity
0.1, "Megaelectronvolts", "HalfValueLayer"
 
 4.06062 cm

In the design of radiation protection products it is important to take into account the mass
attenuation coefficients of the materials used. This attenuation depends on the incident
particles (protons, alpha, protons, electrons, neutrons) and their energy. Here we compare
the mass attenuation coefficient for various substances and energies:
    329


 materials "Water", lead ELEMENT , iron ELEMENT , aluminum ELEMENT ;
ListLogLogPlot

Table
Quantity
x, "Megaelectronvolts", StoppingPowerData
,
"Particle" "Photon", "Energy" Quantity
x, "Megaelectronvolts",
"MassAttenuationCoefficient", x, 0.01, 100, .5 &  materials,
Joined True, PlotLegends materials, AxesLabel Automatic
cm2 g

100

Water
10
lead
 

1 iron

aluminum
0.10

MeV
0.01 0.10 1 10 100

To accurately estimate absorbed doses, we measure the amount of the energy deposited in
human tissues. The example below shows the stopping power changes, dE/dx, for an alpha
particle emitted from U234 as it travels through biological tissue:


 initialKE QuantityMagnitude 
IsotopeData
Entity
"Isotope", "Uranium234", "BindingEnergy"
 
 7.600708


 dEByDxVal
kReal ? Positive :

QuantityMagnitude  StoppingPowerData
"A150TissueEquivalentPlastic",
"Particle" Entity
"Particle", "AlphaParticle", "Energy"
Quantity
k, "Megaelectronvolts", "LinearStoppingPower"  10 000;
dEByDxVal
kReal : 0

  sol NDSolveValue
KiM '
x  dEByDxVal
KiM
x, KiM
0  initialKE,
WhenEvent
KiM
x  0, "StopIntegration", KiM, x, 0, 60,
PrecisionGoal 3;
330      


 Plot
dEByDxVal
sol
x, x, 0, 60,

Frame True, GridLines Automatic, FrameLabel
Row
"Distance Travelled", Quantity
None, "Micrometers", " ", Row

"dEdx", Quantity
None, "Megaelectronvolts"  "Micrometers", " "
0.30

0.25
dEdx MeV"m

0.20

 

0.15

0.10

0 10 20 30 40 50 60
Distance Travelled "m

9.9 Additional Resources

On the Wolfram Demonstrations website you can find quite sophisticated examples
about nuclear physics and isotopes:
Nuclear physics demonstrations:
http://demonstrations.wolfram.com/search.html?query=Nuclear
Isotopes demonstrations: http://demonstrations.wolfram.com/search.html?query=isotopes
Individual demonstrations worth mentioning:
Table of Nuclides: http://demonstrations.wolfram.com/TableOfNuclides by Enrique Zeleny
Isotope Browser: http://demonstrations.wolfram.com/IsotopeBrowser by Theodore Gray and
Yifan Hu
Binding Energies of Isotopes:
http://demonstrations.wolfram.com/BindingEnergiesOfIsotopes by Stephen Wolfram and
Jamie Williams
 10
Modeling: Applications in Biokinetics,
Epidemiology, and more

This chapter shows how to use Mathematica for biokinetic,

pharmacokinetic, physiological and epidemiological modeling, mainly in
the context of building multi-compartment and nonlinear models using
differential equations (DEs). These models also have many applications in
a wide variety of fields such as chemical kinetics, clinical and nuclear
medicine, ecology, and internal dosimetry, among others. There is also a
section covering the statistical analysis of experimental data with a
special emphasis on how to estimate parameters and perform optimal
experimental designs. A short section about partial differential equations
(PDEs) is also included. The chapter concludes with a very brief
introduction to System Modeler, Wolfram’s program for modeling and
simulating physical systems. The reader may find the contents of this
chapter useful for learning about modeling and how to fit nonlinear data
in systems of differential equations.

10.1 Compartmental Modeling

10.1.1 An Introduction to Compartmental Analysis

Compartmental analysis has applications in chemical kinetics, clinical medicine,
ecology, internal dosimetry, nuclear medicine, and pharmacokinetics. It can be
described as the analysis of a system when it is separated into a finite number of
component parts, which are called compartments. Compartments interact through
the exchange of materials or energies. These materials or energies can be chemical
substances, hormones, individuals in a population and so on. A compartmental
system is usually represented by a flow or a block diagram.

The following code generates a diagram representing a basic compartmental system:
332 Mathematica beyond mathematics

In[
]:= Show
Graphics
Arrow


0, 1.5, 0, 0.5,
Text
"Input", 0.4, 1, 0, 1, Circle
0, 0, 0.5 ,
Text
"Compartment 1", 0, 0, 0, 0, Arrow
0.5, 0., 2.5, 0.,
Text
"Flow from 1 to 2", 1.5, .4, 0, 0,
Circle
3, 0, 0.5 , Text
"Compartment 2", 3, 0, 0, 0,
Arrow
3, 0.5, 3, 1.5, Text
"Output", 2.4, 1, 0, 1,
AspectRatio
Automatic, Axes
None

Input

Flow from 1 to 2

Out[
]= Compartment 1 Compartment 2

Output

We adopt the convention of representing compartments with circles or rectangles.

The flow in or out of the compartments is represented by arrows. The ith
compartment of a system of n compartments is labeled i and the size (amount or
content) of the component in compartment i as xi
t. The exchange between
compartments, or between a compartment and the environment is labeled ki j , where
ij represents the flow from i to j. If there is no ambiguity, ki will be used in place of kij .
Additionally, if i  10 and j  10, kij will be written in place of ki j . The environment refers
to the processes that are outside the system and is usually represented by zero, so
ki 0 is the fractional excretion coefficient from the ith compartment to the
environment. If we assume that the substance introduced into the system is a radioactive
isotope, we must consider the radioactive decay, which is given by a constant rate represented
by  (this constant is specific for each isotope). The decay constant can be interpreted as
an equal flow going out of the system in each compartment. The input from the
environment into the jth compartment is called bj
t. With regards to the
environment, we only need to know the flow, bj
t, into the system from the
outside. The ki j s are called fractional transfer rate coefficients. They are usually
assumed to be constants, but in some cases they can be functions of time (that is
ki j
t).
Based on the definitions in the previous paragraph, we can describe the two-
compartment model in Figure 10.1 as follows: b1
t is the input from the
environment to compartment 1, k12 the transfer rate coefficient from compartment 1
10. Modeling: Application in Biokinetics 333

to compartment 2 and k21 from compartment 2 to 1, k20 the transfer rate coefficient
from compartment 1 to the environment (output), and x1
t and x2
t represent the
quantities in compartments 1 and 2 at time t:

b1
t

k12

1 2
x1
t x2
t

k21

k10

Figure 10.1 A two-compartment model.

A Two-compartment Model with Input and Output from Compartment 1

The variables x1
t and x2
t are called the state variables of the system, and their
evolution over time is described by a system of ordinary differential equations
(SODE). In this case:

x1
t  k10 x1
t  k12 x1
t  k21 x2
t  b1
t

10.1
x2
t  k12 x1
t  k21 x2
t

or in matrix notation:

x1
t k10  k12 k21 b1
t
  
x2
t k12 k21 0
We also need to know the initial conditions. In this case, the contents of x1
t and
x2
t at t  0.
x1
0  x10 ; x2
0  x20

This SODE can be solved using DSolve or DSolveValue.

For instance, in day1 , for k10 = 0.1, k12 = 0.2, k21 = 0.3, b1
t = 0.2 t Exp2.1 t, with
x1
0  x2
0  0. The solution is:
In[
]:= x1
t, x2
t 
DSolveValue
x1 '
t  0.1 x1
t  0.2 x1
t  0.3 x2
t  0.2 t Exp
2.1 t ,
x2 '
t  0.2 x1
t  0.3 x2
t, x1
0  0, x2
0  0,
 x1
t, x2
t, t ExpandAll Chop
Out[
]= 0.113208 2.1 t t  0.0652664 2.1 t  0.0413534 0.544949 t  0.0239131 0.055051 t ,
0.0125786 2.1 t t  0.0142399 2.1 t  0.0337649 0.544949 t  0.0195249 0.055051 t 
We have used “ExpandAll // Chop” to eliminate insignificant terms.

The evolution of x1
t and x2
t over time can be visualized as follows (we show two graphs
using different ways to include the legends). In the graphic on the right we use Callout:
334 Mathematica beyond mathematics

In[
]:= GraphicsRow

Plot
x1
t, x2
t, t, 0, 30, AxesOrigin
0, 0, AxesLabel

"Days", "x", PlotLegends
Placed
"x1 t ", "x2 t ", Center,
PlotLabel
Style
"Using PlotLegends", Bold,
Plot
Callout
x1
t, "x1 t ", Above, Callout
x2
t, "x2 t ", 4,
t, 0, 30, AxesOrigin
0, 0, AxesLabel
"Days", "x",
PlotLabel
Style
"Using Callout", Bold, ImageSize
400 

Using PlotLegends
x Using Callout
0.030
0.025 xx1
t
0.04
Out[
]= 0.020 x1
t 0.03 x
t
0.015 2
x2
t 0.02
0.010
0.01
0.005
Days
Days 5 10 15 20 25 30
5 10 15 20 25 30

10.1.2 The Iodine ICRP 78 Biokinetic Model

In[
]:= ClearAll
"Global` "

The flow diagram in Figure 10.2 represents the iodine model (obtained from ICRP
78) where compartment 1 is the blood, compartment 2 is the thyroid, compartment
3 is the rest of the body, and compartment 4 is the bladder. Additionally, 3  0,
i.e., a transfer from compartment 3 to the environment, represents the output to the
gastrointestinal tract (GIT) and 4  0 represents the output, via urine excretion, to
the environment.

b1
t

k12

1 2

k14
k31 k23

k30
4 3

k40

Figure 10.2 The Iodine ICRP model.

We build the compartmental matrix A for iodine given by:

10. Modeling: Application in Biokinetics 335

In[
]:= A  k12  k14 , 0, k31 , 0,

k12 , k23 , 0, 0, 0, k23 , k30  k31 , 0, k14 , 0, 0, k40 ;
In[
]:= A
k12  k14 0 k31 0
k12 k23 0 0
Out[
]=
0 k23 k30  k31 0
k14 0 0 k40

We need to know the transfer constants kij values. According to the ICRP 78, the transfer
rates for iodine, in days1 , are k12  0.3 Log
2 0.25, k14  0.7 Log
2 0.25,
k23  Log
2 80, k30  0.2 Log
2 12, k31  0.8 Log
2 12 and k40  12 ;
0.3 Log
2 0.7 Log
2 Log
2
In[
]:= kidine  k12
, k14
, k23
,
0.25 0.25 80
1 1
k30
0.2 Log
2, k31
0.8 Log
2, k40
12;
12 12
In[
]:= Aiodine  A . kidine
2.77259 0 0.0462098 0
log
2
0.831777  0 0
80
Out[
]=
log
2
0 0.0577623 0
80
1.94081 0 0 12

In[
]:= X
t  x1
t, x2
t, x3
t, x4
t;

We assumed a single input b1 = 1, in t= 0. Then the initial conditions are: x1(t) = 1,

x2(t)=x3(t), x4(t)=0
In[
]:= system  X '
t  Aiodine.X
t;

In[
]:= ic  Thread
X
0  1, 0, 0, 0;

In[
]:= sol 

DSolveValue
system, ic, x1
t, x2
t, x3
t, x4
t, t ExpandAll
2.77254 t 0.0601471 t 0.00632391 t
Out[
]= 1.00003   0.000840995   0.000808587  ,
0.300955 2.77254 t  0.0135875 0.0601471 t  0.287368 0.00632391 t ,
0.00096051 2.77254 t  0.0493651 0.0601471 t  0.0484045 0.00632391 t ,
0.210331 12. t  0.210337 2.77254 t  0.000136703 0.0601471 t  0.000130845 0.00632391 t 

The evolution of the iodine retention in the blood (compartment 1) and in the thyroid
(compartment 2) is plotted as follows:
336 Mathematica beyond mathematics

In[
]:= Plot
sol1 , sol2, t, 0, 10, AxesLabel
"Days", "r",

PlotRange
All, PlotLegends
Placed
"Blood", "Thyroid", Center
r

1.0

0.8

0.6
Out[
]= Blood
Thyroid
0.4

0.2

Days
2 4 6 8 10

10.1.3 Developing Functions for Solving Compartmental Systems

Instead of using DSolve or DSolveValue directly, we are going to develop some
functions for solving multi-compartment models using a friendly notation.
In[
]:= ClearAll
"Global` "

Figure 10.3 shows a more complex compartmental model: The general multi-input
multi-output (MIMO) three-compartment model. We have assumed a radioactive
decay with a disintegration constant  (it can be considered as an output from each
compartment):

b3
t


3

k30
k13 k32

k31 k23
b1
t b2
t

k12
 
1 2

k21

k10 k20

Figure 10.3 The generic three-compartment model.

10. Modeling: Application in Biokinetics 337


x1 (t) = - ( + k10 + k12 + k13 ) x1 (t) + k21 x2 (t) + k31 x3 (t) + b1 (t)

x2 (t) = k12 x1 (t) - ( + k20 + k21 + k23 ) x2 (t) + k32 x3 (t) + b2 (t) (10.2)

x3 (t) = k13 x1 (t) + k23 x2 (t) - ( + k30 + k31 + k32 ) x3 (t) + b3 (t)

The model can be expressed in matrix notation as follows:

x(t) = A x + b(t), t  0
where x(t) is a column vector representing the contents of the compartments:
x(t) = {x1 (t), x2 (t), x3 (t)}T

b(t) is a column vector representing the inputs to compartments 1, 2, and 3:

b(t) = {b1 (t), b2 (t), b3 (t)}T

This pattern can be expanded to systems of n compartments or n state variables (in

the case of physiological models). The equation for any compartment i is given by:


x(t) = A x + b(t), t  0
(10.3)
x (0) = x0
where :
x(t) = {x1 (t), x2 (t), ..., xn (t)}T is a column vector and xi (t) denotes the amount or
content of materials or energies in compartment i at time t.
A is a n n is usually known as the compartmental matrix or system matrix.
b(t) = {b1 (t), b2 (t), ..., bn (t)}T is a column vector where {bi (t)} is the input rate into
compartment i from an outside system.
x(0) = {x1 (0), x2 (0), ..., xn (0)} T are the initial conditions, so xi (0) represents the
amount or content of materials or energies in compartment i at time t = 0.
These models are known as Systems of Ordinary Differential Equations with
Constant Coefficients (SODECC).
In the rest of this section, we’re going to develop several functions that will enable
users, even those with minimal knowledge of differential equations, to solve
SODEs related to compartmental and physiological systems. If you’re interested in
programming, you may want to pay close attention to how the functions have been
created. Otherwise, you just need to follow the examples to see how to use the
commands properly.
We develop the function CompartMatrix1[n, matrixA, ] that returns the matrix of
coefficients of a compartmental system, where n is the number of compartments of
the system, matrixA represents the transfer rates, and  is the radioactive decay
constant (by default  = 0, which means that it is not a radioactive isotope.).The
transfer rates kij  from compartment i to compartment j are written: {{1, 2, k12 }, ...,
{i, j, kij , ...}, by default kij = 0.
338 Mathematica beyond mathematics

In[
]:= CompartMatrix1
n ? IntegerQ, matrixA ? MatrixQ, lambda : 0 :

Module
k, A,
Apply
Set, k  Take
1, 2, Last
1 &  matrixA, 1;
k
i, j : 0;
A  DiagonalMatrix
Table
Sum
k
i, j, j, 0, n, i, 1, n 
Table
k
j, i, i, 1, n, j, 1, n  lambda IdentityMatrix
n;
Notice the use of: ?IntegerQ, matrixA_?MatrixQ to check that the input contains an integer
and a matrix as the first two arguments. Otherwise, the function will not be executed. By
using the structure: lambda_:0, we give lambda (the disintegration constant) a default value
of 0 in case the user doesn’t enter any value for it.

Using this function we can construct the Iodine model (Figure 10.2).
In[
]:= iodine131matrix  CompartMatrix1
4, 1, 2, k12 , 1, 4, k14 ,
4, 0, k40 , 2, 3, k23  , 3, 0, k30  , 3, 1, k31 , ;
In[ ]:=  MatrixForm
Out[
]//MatrixForm=






 
 





  







 

  





We also create the function SystemDSolve1 to solve the model where matrixA is the
coefficients matrix; initcond are the initial conditions: c1 , , cn ; input are the inputs
b1
t, , bn
t ; x is the symbol that represents the retention variables. The function
returns the analytical solution xi (t). We can use t1  t to get the solution as a function of t
instead of as a numerical value. We have used Chop and ExpandAll to eliminate the terms
with values very close to 0.
In[
]:= SystemDSolve1
matrixA ? MatrixQ,
incondList, inputList, t, t1, x, opts ? OptionQ :
Module
A, n, var, i, system, ic, X, X0, sol , A  matrixA;
n  Dimensions
A1;
X  Table
xi
t, i, 1, n; X0  Table
xi
0, i, 1, n;
system  D
X, t  A.X  input;
ic  Thread
X0  incond;
sol  DSolve
system, ic, X, t, opts;
Chop
ExpandAll
sol . t
t1;

To apply the previous function, to the iodine model (Fig 10.2), we also assume as the
initial condition: {1, 0, 0, 0} and as the input function: 1  0.5 Cos
0.3 t, 0, 0, 0 . Also,
in the example we refer to Iodine–131, then we need the decay constant  of the I-131 (in
days1 because kij are in days1 ). It can be obtained as follow:
In[
]:= 131  QuantityMagnitude

UnitConvert iodine-131 ISOTOPE  decay constant , 1 "Days"

Out[
]=   
10. Modeling: Application in Biokinetics 339

In[
]:= iodine131matrix1 

0.3 Log
2 0.7 Log
2 Log
2
iodine131matrix . k12
, k14
, k23
,
0.25 0.25 80
1 1
k30
0.2 Log
2, k31
0.8 Log
2, k40
12, 
131 
12 12
Out[
]= 
       
   


     
  

In[
]:= x1
t1, x2
t1, x3
t1, x4
t1 
x1
t1, x2
t1, x3
t1, x4
t1 . SystemDSolve1
iodine131matrix1,
1, 0, 0, 0, 1  0.5 Cos
0.3 t, 0, 0, 0, t, t1, x1
Out[ ]=       
   
    
  



   
            

  
   
  
  

   
              
     
      
  


   
 
        
 
 
     
    
  
  
 
    
       

    

   

         

     
    
    
 
    

The evolution of the iodine retention in the blood (compartment 1) and in the thyroid
(compartment 2) is plotted below:
In[
]:= Plot
x1
t, x2
t, t, 1, 200, PlotStyle

RGBColor
1, 0, 0, AbsoluteDashing
4, 4, RGBColor
0, 0, 1,
AxesLabel
"Days", "r", PlotLegends

Placed
"Blood", "Thyroid", Center, Framed
, RoundingRadius
5 & ,
PlotLabel
Style
Framed
"b1 t  1  0.5 Cos
0.3 t",
12, Blue, Background
Lighter
Yellow

b1
t 1  0.5 Cos0.3 t

r

3.5

3.0

2.5
Out[ ]=
2.0
Blood

1.5 Thyroid

1.0

0.5

Days
50 100 150 200

In[
]:= Clear
x1, x2, x3, x4, iodine131matrix1

340 Mathematica beyond mathematics

The following function computes numerically the SODE using NDSolve and returns the
solution in the interval {t, tmin , tmax }:
In[ ]:= SystemNDSolve1
matrixA ? MatrixQ, incondList,
inputList, t, tmin, tmax, t1, x, opts ? OptionQ :
Module
A, n, var, i, system, ic, X, X0, sol , A  matrixA;
n  Dimensions
A1;
X  Table
xi
t, i, 1, n; X0  Table
xi
0, i, 1, n;
system  D
X, t  A.X  input;
ic  Thread
X0  incond;
sol  NDSolve
system, ic, X, t, tmin, tmax, opts;
Chop
ExpandAll
sol . t
t1;

This function can be used when SystemDSolve1 does not find a solution, for instance, if
some transfer coefficients are variables:
0.3 Log
2
In[ ]:= iodine131matrix2  iodine131matrix . k12
1  Cos
0.2 t ,
0.25
0.7 Log
2 Log
2
k14
, k23
1  Cos
0.3 t ,
0.25 80
1 1
k30
0.2 Log
2, k31
0.8 Log
2, k40
12, 
131;
12 12
In[
]:= q1
t1, q2
t1, q3
t1, q4
t1 
q1
t1, q2
t1, q3
t1, q4
t1 . SystemNDSolve1
iodine131matrix2,
1, 0, 0, 0, 1  0.5 Cos
0.3 t, 0, 0, 0, t, 0, 100 , t1, q1;
In[
]:= Plot
Callout
q1
t, "Blood", Above, Callout
q2
t, "Thyroid", Above,
t, 1, 100, AxesOrigin
0, 0, AxesLabel
"Days", "r"
r

5 Thyroid

3
Out[
]=

Blood
1

Days
20 40 60 80 100

In[
]:= Clear
q1, q2, q3, q4, iodine131matrix2

10.1.4 Laplace Transforms and Identifiability Analysis

The transfer rates ki j are usually estimated using experimental data as shown in the
next section. The problem that often arises is that there is no unique value of ki j that
satisfies the model but a finite number of values. This issue is addressed by a group
of mathematical methods named identifiability analysis. The Laplace transforms
are very useful in this kind of analysis.
10. Modeling: Application in Biokinetics 341

The solution of a SODE with constant coefficients (eqn. 10(3)) can be found as
follows:
t
x
t  x0 A t  A
t  b

0

Applying Laplace transforms we get:

X
s 
s I  A 1 x0 
s I  A 1 B
s

where X
s and B
s are the Laplace transforms of x
t and b
t respectively.

This equation can be written in Mathematica as:
In[
]:= SystemLTSolve1
matrixAList, initcondList, inputList, t, s, X :
Module
B, A1, R, R1, P, r, n, A1  matrixA; n  Dimensions
A11;
B  s IdentityMatrix
n  A1; R  Inverse
B; P  Apart
R;
r  LaplaceTransform
input, t, s;
R1  Dot
P, initcond  Dot
P, r;
Thread
Subscript
X, 1
s &  Range
n
R1;

Example: In Figure 10.4 (Godfrey, 1983, Example 6.5) the response of the central
compartment 1 to a single input of 1 unit at t  0 has been fitted (no noise is
assumed) to the function x1 exp
t  0.7exp
 5 t  0.2exp
 t  0.1exp
 0.1 t.
The problem consists of determining the unknown transfer rates,
k10 , k12 , k13 , k21 , k31 , assumed to be constants.

b1

k21 k13
2 1 3
k12 k31

k10

Figure 10.4 Catenary three-compartment model.

The Laplace transform X
s of an impulsive input of 1 (we can use any mass unit, so if 1 is
in mg the solution will be in mg) in compartment 1 at t  0 can be modeled using
SystemLTSolve1 with the initial conditions 1, 0, 0 and b
t  0, 0, 0 . To solve the
problem, we will also need to define the compartmental matrix of the system.
In[
]:= tricompartment  CompartMatrix1
3,
1, 2, k12 , 2, 1, k21  , 1, 3, k13 , 3, 1, k31 , 1, 0, k10 ;
In[
]:= X1
s, X2
s, X3
s  X1
s, X2
s, X3
s .
SystemLTSolve1
tricompartment , 1, 0, 0, 0, 0, 0, t, s, X Simplify

k21  s
k31  s
Out[
]= ,
k12 s
k31  s  k10
k21  s
k31  s  s
k21  s
k13  k31  s
k12
k31  s
,
k12 s
k31  s  k10
k21  s
k31  s  s
k21  s
k13  k31  s
k13
k21  s

k12 s
k31  s  k10
k21  s
k31  s  s
k21  s
k13  k31  s
342 Mathematica beyond mathematics

The Laplace transform X1 exp
s of x1 exp
t  0.7exp
5 t  0.2exp
t  0.1exp
0.1 t is:
In[
]:= X1exp  LaplaceTransform
0.7 Exp
5 t  0.2 Exp
t  0.1 Exp
0.1 t, t, s
0.2 0.7 0.1
Out[
]=  
s1 s5 s  0.1

Now, the transfer rate k10 , k12 , k13 , k21 , k31 can be obtained solving X1 exp
s  X1
s. The
following procedure is applied:
In[ ]:= SolveAlways
X1exp  X1
s, s

Out[
]= k10  0.746269, k12  1.25488, k13  1.70885, k21  0.324354, k31  2.06565 ,
k10  0.746269, k12  1.70885, k13  1.25488, k21  2.06565, k31  0.324354
In this case, there are two sets of possible solutions. Notice that neither set contains any
negative rate constants, which would permit the rejection of such set.
All the functions described so far are part of Sysmodel, a package included in
BIOKMOD that we will use in the next section.
In[
]:= Clear
X1, X2, X3, X1exp

10.2 Epidemiological Models

In[ ]:= ClearAll
"Global` "

Compartmental models can also be used to model pandemics. The SIR model is the
most basic one. It consists of three compartments representing three situations: (S)
Susceptible, (I) Infectious, and (R) Recovered. The model can be visualized as
follows:

  
t
 
A very simple version of this model consists of replacing S, I, and R with the
variables s, i, and r, each one representing fractions of the entire population (values
between 0 and 1).
Initially, we assume that almost all of the population is susceptible of getting
infected (i.e., s(0)  1). Over time, that fraction will go down as the infection
spreads. We also assume that the infectious population at t = 0 will be almost 0
(i.e., i(0)  0) and that at the beginning, there will be no people yet who have
recovered (i.e., r(0) = 0). At each point in time, the following equation must hold:
s  i  r  1. Each infectious individual is also assumed to transmit the disease to
b new people daily so the higher the value of b, the faster the epidemic will spread.
Finally, k represents the fraction of the infected population that recovers every day.
For example, if the disease lasts 7 days, k is 1/7.

The equation below describes how the percentage of the population susceptible of getting
infected changes over time. We can assume that the rate of change is going to be negative
since we start with a value close to 1 that goes down over time. Additionally, the rate of
change is proportional to b, the number of contacts, (the higher the number of contacts, the
higher the number of people becoming infectious and therefore the faster s goes down) and
to s itself, the fraction of the population that is susceptible of becoming infectious at any
time t.
10. Modeling: Application in Biokinetics 343

In[
]:= equationS 

s '
t  b s
t i
t;

The second equation, equationI, describes how i changes over time On the right side of the
equation there are two terms. The first one refers to right side of equationS but has the
opposite sign, representing the increase in the infectious population, (a flow from S to I) and
the second term corresponds to the decrease in infectious cases (recovered people):
In[
]:= equationI 
i '
t  b s
t i
t  k i
t;

The last equation models people (as a fraction of the total population) that have recovered.
Expressed in compartmental terms, it indicates a flow from the I compartment to the R
compartment.
In[ ]:= equationR 
r '
t  k i
t;

Let’s solve now this system of equations in terms of b and k by assuming some initial
conditions regarding the variables s, i, and r at t = 0.
In[
]:= sirmodel2   equationS, equationI, equationR;

In[
]:= sols2, solr2, soli2  s, r, i . ParametricNDSolve

Join
sirmodel2, s
0  0.9999, i
0  0.0001, r
0  0.0000,
s, i, r, t, 0, 100,  b, k

 
Out[
]=  
   

 
 
   

 
 
   

Imagine that we got some data {t,i} with t representing the number of days since
the beginning of the epidemic and i the fraction of infectious people, and we would
like to calculate the values for b and k.

The data has been generated assuming b = 0.5 and k = 1/14 and adding a random component
to both parameters: (SeedRandom[31415]; Evaluate[Table[{t, soli2[0.5 +
RandomReal[{-0.1, 0.1}], 1/14 + RandomReal[{-0.01, 0.01}]][t]}, {t, 10, 100, 5}]] /. {a_,
b_} -> {a, Round[b, 0.001]}
In[
]:= sample  10, 0.007, 15, 0.108, 20, 0.188, 25, 0.471,
30, 0.516, 35, 0.456, 40, 0.395, 45, 0.153, 50, 0.168,
55, 0.158, 60, 0.071, 65, 0.048, 70, 0.036, 75, 0.015,
80, 0.011, 85, 0.012, 90, 0.015, 95, 0.006, 100, 0.002;

We use FindFit to estimate b and k.
In[
]:= fit  FindFit
sample, soli2
b, k
t,
b, 0.5, k, 0.01, t, PrecisionGoal
10 
Out[
]=    
  
According to the solution, each infectious person transmits the disease to another person
every two days (b = 0.46 1/2), and an infected person will recover in 14 days (k =
344 Mathematica beyond mathematics

0.071466  1/14).

Let’s plot the fitted model against the actual data:
In[
]:= Plot
Evaluate
sols2
b, k
t, soli2
b, k
t, solr2
b, k
t . fit,
t, 0, 100, PlotRange
0, 1.01, PlotStyle
Blue, Red, Green,
PlotLegends
"Susceptible", "Infectious", "Recovered",
Epilog
Point  sample
1.0

0.8

0.6 Susceptible
Out[
]= Infectious
0.4 Recovered

0.2

0 20 40 60 80 100

We can now make the plot interactive to see how the model changes when we
change b and k. Their default values are the ones from the initial model. This
method is quite useful since we can visually adjust the model to improve the fit.
In[
]:= Manipulate
Plot
sols2
b, k
t, solr2
b, k
t, soli2
b, k
t,
t, 0, 100, PlotStyle
Blue, Red, Green,
PlotLegends
"Susceptible", "Infectious", "Recovered",
Epilog
Point  sample, b, 0.46, 0.1, 1, k, 1 14, 0.01, 1 

We can also analyze the sensitivity of the parameters by evaluating the derivative with
respect to them. The higher the sensitivity, the higher the impact on the model when the
parameter values change. For visualization purposes we add and subtract the derivative of
the model multiplied by 0.1 before plotting it.
10. Modeling: Application in Biokinetics 345

In[
]:= Manipulate

GraphicsColumn
Plot
sols2
b, k
t, solr2
b, k
t, soli2
b, k
t,
t, 0, 100, PlotStyle
Blue, Red, Green, PlotLegends

Placed
"Susceptible", "Infectious", "Recovered", Above, Plot

Evaluate
soli2
b1, k1
t  0, .1, .1 D
soli2
b1, k1
t, b1 .
b1
b, k1
k, t, 0, 100, PlotRange
All,
Filling
2
3, PlotLegends
 "Infectious, b", Plot

Evaluate
soli2
b1, k1
t  0, .1, .1 D
soli2
b1, k1
t, k1 .
b1
b, k1
k, t, 0, 100, PlotRange
All,
Filling
2
3, PlotLegends
 "Infectious, k" ,
b, 0.5, 0.1, 1, k, 1 14, 0.01, 1 
346 Mathematica beyond mathematics

10.3 Physiological Modeling

10.3.1 Michaelis–Menten (M–M) Models

Most drugs are metabolized by enzymes. The enzyme levels will change under
different physiological cases (e.g., when many drugs are administered at the same
time, they might have affinity for the same enzymes and thus compete for the same
enzyme sites) or pathological states (e.g., in case of disease). This change in
enzyme levels produces fluctuations in the concentration levels of drugs in the
blood. For example, when given low-concentration/tolerable doses of Phenytoin, an
anti-epileptic drug, its degradation by the corresponding metabolizing enzymes will
follow first order or linear kinetics, that is, the rate of degradation will be directly
proportional to the drug or substrate concentration. But when given a higher
concentration of Phenytoin, as in the case of drug overdose or accidental poisoning,
the metabolism will initially follow first-order kinetics for some time until all
enzyme sites get occupied/saturated, and then it will change to zero-order kinetics,
that is, the rate of change will become constant or independent of the drug
concentration.
This mixed-order reaction (combining linear with nonlinear kinetics when there are
no more sites available for drug binding) can be depicted by the Michaelis–Menten
(M–M) equation. A example of the M–M model is shown in Figure 10.5. It
describes the disposition in the animal or human body of a drug distributed
according to a mammillary model, consisting of a central component with
peripheral components connecting to it. This is a convenient model for many drugs
for which the equilibrium between drug concentrations in different tissues is not
achieved rapidly. It tries to describe mathematically the concentration data after the
administration of a drug by depicting the body as a two-compartment open model
where the drug is both introduced into, and exits from, the central compartment,
named compartment 1, which can be sampled through the blood (plasma or serum).
It may consist of organs or tissues which, being highly perfused with blood, reach a
rapid equilibrium distribution with the blood. The peripheral compartment,
compartment 2, usually cannot be sampled. It may consist of organs or tissues
which, being poorly perfused with blood, reach equilibrium distribution with the
blood only slowly. The concentration in compartment i is denoted xi .

b0

k12

x1 x2

k21

Vm km

Figure 10.5 A Michaelis–Menten (M–M) model.

10. Modeling: Application in Biokinetics 347

In this case, the drug movement between compartments will be considered as a

linear kinetic process described by the transfer coefficients k12 and k21 . However,
the elimination process will be nonlinear, as it happens for instance in the hepatic
metabolism, and the elimination rate of the drug can be mathematically expressed
by the M–M equation with parameters Vm = maximum transformation speed and km
= M–M constant. The drug administration will be assumed to be an impulsive input
(bolus) b0 . Thus:
 Vm x1
t
x1
t   k21 x1
t  k12 x2
t 
km  x1
t

10.4
x2
t  k21 x1
t  k12 x2
t
x1
0  b0 , x2
0  0

In our example: b0 = 1, Vm = 0.1, km = 0.3, and k12 and k21 are the unknown
parameters. They can be estimated using experimental data.

We define a model x1
t, k12 , k21  and then use the  !"# function to
obtain a numerical solution as a function of k12 and k21:
In[
]:= ClearAll
"Global` "

Vm  0.1; km  0.3;
Vm x1
t
eq1  x1 '
t  k12 x1
t  k21 x2
t  ;
km  x1
t
eq2  x2 '
t  k12 x1
t  k21 x2
t;
In[
]:= sol  ParametricNDSolve

eq1, eq2, x1
0  1, x2
0  0, x1, x2, t, 0, 100, k12 , k21 

 
Out[
]= $  
     

 
$  
     

x1
t, k12 , k21  and x2
t, k12 , k21  can be represented as functions of k12 and k21 .
In[
]:= Manipulate
Plot
Evaluate
x1
k12, k21
t, x2
k12, k21
t . sol,
t, 0, 30, PlotRange
All, PlotLegends

Placed
"x1", "x2", Center, Framed
, RoundingRadius
5 &  ,
k12, 0.03, "k12", 0.01, 0.1, k21, 0.02, "k21", 0.01, 0.1
348 Mathematica beyond mathematics

10.3.2 A Nonlinear Physiological Model

In this section, we present an example of a physiological nonlinear model (M. S.
Roberts, M. Rowland, A dispersion model of hepatic elimination, Journal of
Pharmacokim. Biopharm 14: pp 227-260, 1986.):
An hepatic drug dispersion model in the liver can be represented as a cylindrical
vessel (Figure 10.6). Let’s consider an element of differential thickness dz at
distance z along its length L. Inside this element there is blood, with a volume VB,z
flowing at rate Q through a cross-sectional area A and a velocity v = Q/A,
containing a drug concentration CB (z, t), part of which is unbounded at a
concentration Cu in the plasma.

Figure 10.6 Schematic cross section of the liver:(1) Blood, (2) Space of Disse, (3) The hepatocyte.

The model is represented by a partial differential equation, where the axial

dispersion flux is given by term (1) and the convective flux by term (2).
fuB = Cu / CB , CLint is a clearance constant,  is a permeability constant, and VH is
10. Modeling: Application in Biokinetics 349

the apparent volume of the distribution of the drug.

1
2
 CB
z, t  CB
z, t  CB
z, t
VB D  VB v  fub CLint  CB
z, t  VH
2 z t
z

To scale the equations and make the variables dimensionless we make the
following replacement:
z D DA fub CLint 
Z  ; dn   ; rn  ; c
z, t  CB
z, t
L vL QL Q

After which we obtain:

2 c 1 c rn 1 c
  c 
10.5
2 dn  z dn dn  t
z

with the following the initial conditions (IC):

t  0, c
z, 0  0, 0  z  1

In Mathematica notation, the above result can be written as:

1 r 1
In[
]:= roweq  D
c
z, t, z, 2  D
c
z, t, z  c
z, t  D
c
z, t, t
d d d
 %   
%   
% 
Out[
]=

  
% 
& & &

We can solve it using the Laplace transform method turning the partial differential equation
into an ordinary differential equation:
In[
]:= rowlandeqLT  LaplaceTransform
roweq, t, s
 '"()%   
Out[
]=


&
'"()  
%   
 '"()  
%   
&

%    '"()%   
&

In the above output, the replacement LaplaceTransform[f(z,t), t,s ]  LT[z], takes also
into account the initial condition (c[z,0]=0):
1,0
In[
]:= rowlandeqLT1  rowlandeqLT . LaplaceTransform c
z, t, t, s
cT '
z,
2,0
LaplaceTransform c
z, t, t, s
cT ''
z,
LaplaceTransform
c
z, t, t, s
cT
z, c
z, 0
0

 (% ( %  (%

Out[ ]=

 ( %
& & &

The above equation is an ordinary differential equation that can be solved using !"# :
In[
]:= sol1  cT
z . DSolve
rowlandeqLT1 , cT
z, z
   &  &     &  & 

%   %
Out[
]=   & &
    & &
 

The coefficients C1 and C2 can be determined using the boundary conditions, which in this
case are: at z  0, c  
z, and at z  , c  0, that is, c
0, t  
z and c
, t  0.
350 Mathematica beyond mathematics

For convenience we replace C1 by c1 and C2 by c2.

In[
]:= sol2  sol11 . C
1
c1, C
2
c2
   &  &     &  & 

%   %
Out[
]=    & &
    & &

c1 can be computed using the boundary condition c
0, t  
z:

In[
]:= solc1  Solve

Evaluate
sol2 . z
0   LaplaceTransform
DiracDelta
z, z, s , c11
Out[
]=  


c2 can be computed using the boundary condition c
, t  0:

In[ ]:= solc2  Solve
Evaluate
sol2 . solc1  0, c21 Simplify

Out[
]= 

 &  %

   &

In[
]:= Limit
solc21, 2, z
Infinity, Assumptions
s  0 && r  0 && d  0

Out[
]= 

Hence:
In[
]:= sol3  Evaluate
sol2 . c1
1  c2 . c2
0
   &  & 

%
Out[
]=  & &

In[
]:= InverseLaplaceTransform
sol3, s, t Simplify

  &  
  %%


 & %
Out[
]=
 & % 
 
& 
&

4 d r t2 t2 2 t zz2

z 4dt

In[
]:= csol
z, t  ;

2 t3
2  d
d3

The plot below displays c
z, t for different values of d and r  0:

10. Modeling: Application in Biokinetics 351

In[
]:= Plot
Evaluate
csol
1, t . d
0.01, 0.1, 0.2, 1, 2, r
0,

t, 0.01, 4, PlotRange
0, 4, PlotLegends
0.01, 0.1, 0.2, 1, 2
4

3
0.01
0.1
Out[
]= 2 0.2
1
2
1

0 1 2 3 4

In[
]:= Manipulate ContourPlot Evaluate
csol
z, t . d
d1, r
r1,

z, 0, 30 , t, 0, 30, FrameLabel
"z", "t",
PlotLabel
Style c z,t , Blue, 20 , PlotLegends
Automatic ,
d1, 0.2, "d", 0.01, 1, r1, 0.2, "r", 0, 0.5

10.4 Fitting a Model

In this section, with the help of examples, we describe how to fit biokinetic
parameters using experimental data. For more details about the model see A.
Sánchez-Navarro, C. Casquero, and M. Weiss, “Distribution of Ciprofloxacin and
Ofloxacin in the isolated hindlimb of the rat”, Pharmaceutical Research, 16:
587-591, 1999. We will use the functions developed in Section 10.1.
352 Mathematica beyond mathematics

In[
]:= ClearAll
"Global` "

Let’s suppose that we have a physiological model represented by the following

SODE:
 Q  PS PS Q
x1
t   x1
t  x2
t  b1
t
Vp Vp Vp

 PS PS VTb
x2
t  x1
t   kon x2
t  koff x3
t
10.6
VTu VTu VTu
 VTu
x3
t  kon x2
t  koff x3
t
VTb

where the drug is transported from Vp (vascular volume), by perfusate flow Q, to

VTu (tissue water space) across a permeability barrier (permeability-surface
product PS), and its binding is described by binding/unbinding constants kon and
koff
When the coefficients aij of a SODE are associated with physiologically
meaningful values corresponding to measured physiological parameters or a
function of them, we should use a physiological model instead of a compartmental
one. In those cases, we will directly enter the values for aij to the A matrix, instead
of the kij values.

We build the function CoeffMatrix1[n, matrixA, ] to return the matrix of coefficients for n
retention variables, where the individual elements are the coefficients of the matrix: {{1, 2,
a12 , ..., {i, j, aij , ...} for the ith row and the jth column. By default, aij = 0.
In[
]:= CoeffMatrix1
n ? IntegerQ, matrixA ? MatrixQ :
Module
i, j, k,
Normal
SparseArray
matrixA . i, j, k
i, j
k, n, n;

The coefficients matrix using the function CoeffMatrix1 is:
In[
]:= physiomodel1  CoeffMatrix13,

Q PS PS PS PS
1, 1,   , 1, 2, , 2, 1, , 2, 2,   kon ,
Vp Vp Vp VTu VTu
VTb VTu
2, 3, koff , 3, 2, kon  , 3, 3, koff ;
VTu VTb
In[
]:= physiomodel1
 +   
)) *(
 *(
Out[
]= 

 


 

))
*' *' *' *( *( *( *(

We have a system of ODE, where some parameters par: {kon , koff } are constants
whose values are unknown:
X
t, par  A
t, par X
t  B
t
10.7

Here is developed a function to solve (10.7) using ParametricNDSolve, where the inputs are
given in an easy way
10. Modeling: Application in Biokinetics 353

In[
]:= ParametricSystemNDSolve1
matrixA, incond, input, t, tmin, tmax,

x, par, opts : Module
B, A1, R, P, n, i, A1  matrixA;
n  Dimensions
A11;
R  Join
Thread
Table
xi '
t, i, 1, n 
Table
xi
t, i, 1, n.Transpose
matrixA  input,
Thread
Table
xi
0, i, 1, n  incond;
B  Table
xi , i, 1, n;
Flatten
ParametricNDSolve
R, B, t, tmin, tmax, par, opts;
where:
matrixA is coeffsmatrix A; incond (the initial conditions):{x1
0,.., xn
0}; input: B(t) ={b1 (t),..., bn (t)} x
is the symbol that represents xi ; par: {p1 , ...,pn } are parameters, options are the options of
ParametricNDSolve.

The following parameters values of the model are known:
In[
]:= Q  3; VTb  1; VTu  6.411; Vp  0.973; PS  2.714;

The mathematical expression for b1
t was unknown initially, but an experiment was made
where b1
t was given by the best fit of the input function to experimental data
t1 , b1 , , tn , bn obtained via sampling from an arterial catheter. Here are the data:
In[
]:= dataCateter 
0., 0., 0.05, 402.7, 0.1, 430.3, 0.15, 375.4, 0.2, 292.4,
0.25, 202.2, 0.3, 148.4, 0.35, 96.4, 0.4, 64.9, 0.45, 41.7,
0.5, 25.3, 0.55, 17.8, 0.6, 8.8, 0.65, 6.6, 0.7, 3.2,
0.75, 2.5, 0.8, 1.4, 0.85, 0.9, 0.9, 0.5, 1., 0.2, 1.1, 0.07,
1.2, 0.03, 1.3, 0.01, 1.4, 0.003, 1.45, 0.001, 1.5, 0.001;
In[
]:= ListPlot
dataCateter, Joined
True, PlotRange
All, ImageSize
Small

400

300

Out[
]= 200

100

0.2 0.4 0.6 0.8 1.0 1.2 1.4

The shape of the graph suggests that the function below could provide a good fit:
In[ ]:= b
t  a t Exp
c t . FindFit
dataCateter, a t Exp
c t , a, c, t

 
Out[
]=    

Here are the experimental data and the fitted function:

In[
]:= Plot
b
t, t, 0, 1, Epilog

Hue
0, PointSize
0.02`, Point  dataCateter, ImageSize
Small

400

300

Out[
]= 200

100

0.2 0.4 0.6 0.8 1.0

We also measured x1
t by sampling from the following experimental data

t1 , x1
t1  , , tn , x1
tn  .
354 Mathematica beyond mathematics

In[
]:= X1sample  0.03, 27.5, 0.08, 48.61, 0.28, 133.39, 0.33, 106.18,
0.38, 92.11, 0.48, 60.47, 0.55, 56.54, 0.65, 27.73,
0.75, 23.25, 0.85, 15.83, 0.95, 13.82, 1.05, 10.58, 1.15, 7.83,
1.25, 7.35, 1.35, 6.08, 1.45, 5.21, 1.6, 4.18, 1.8, 3.48,
2., 2.86, 2.2, 2.42, 2.4, 2.11, 2.6, 2.01, 2.8, 1.7, 3., 1.58,
3.4, 1.58, 3.75, 1.44, 4.25, 1.35, 4.75, 1.3, 5.25, 1.29,
6.75, 0.81, 7.25, 0.78, 7.75, 0.73, 8.25, 0.75, 11., 0.68;

We build the model:

The input function and the vector of the initial conditions are 0, 0, 0 :
Q
In[
]:= inputcatheter1   b
t , 0, 0;
Vp

In[
]:= initialcondition  0, 0, 0;

Then, the solution for x1 where the measurements are taken is:
In[
]:= X1a  x1 . ParametricSystemNDSolve1
physiomodel1,
0, 0, 0, inputcatheter1, t, 0.01, 12, x,  koff, kon

 
Out[
]=  
    

Now we want obtained koff and kon for fitting x1
t samples:
In[
]:= FindFit
X1sample , X1a
koff, kon
t, koff, 0.1, 1, kon, 0.1, 1, t

Out[
]= 
)) 
   

Also we can used
   

In[
]:= f
koff ? NumericQ, kon ? NumericQ, t  X1a
koff, kon
t;

In[
]:= nlm  NonlinearModelFit
X1sample ,

f
koff, kon, t, koff, 0.1, 1, kon, 0.1, 1, t;
In[
]:= nlm
"ParameterTable"

    
    
Out[
]=     !"" !"#
  #!! "# "!!  !"! $

Once again, we quit the session before proceeding to the next section.
In[
]:= Quit


10.5 Optimal Experimental Designs (OED)

10.5.1 OED Introduction

Many scientific processes, specially in pharmacokinetics (PK) and
pharmacodynamics (PD) studies, as we have seen before, are defined by systems of
ordinary differential equations (ODEs). If there are unknown parameters that need
to be estimated, the optimal experimental design (OED) approach offers a way to
obtain optimum bias-free and minimum-variance estimators for those parameters
based on specific criteria.
The unknown parameters are usually estimated by fitting experimental data. A
typical model has a solution given by a function r
t, , where   1 , ..., p  are
10. Modeling: Application in Biokinetics 355

the unknown parameters that need to be fitted. The objective is to choose the best
moments{t0 , ..., ti , ..., tn } based on the data, and this can be done using optimal
experimental designs (OED). Here, “best” depends on the objectives of the
practitioner, which are usually linked to a particular optimality criterion, such as
obtaining accurate estimators of the parameters, minimizing the variance of the
predicted response, or optimizing a different characteristic of the model. Many
optimality criteria are based on the Fisher information matrix (FIM). The FIM has a
simple expression for linear models and observations that come from exponential-
family distributions. For more complex models, some formulas can be obtained
from the general information matrix. The usual procedure for a nonlinear model is
to linearize it and apply the well-known toolbox for linear models. The most
widely-used criterion is D-optimality, which focuses on the determinant of the FIM.
In the following pages, we’re going to describe one example using the D-optimality
criterion to obtain the {t0 , ..., ti , ..., tn } values. For further information we refer the
interested reader to: J. Lopez-Fidalgo, J. Rodriguez-Diaz, JM. Sanchez and G.
Santos-Martin, “Optimal designs for compartmental models with correlated
observations”, Journal of Applied Statistics: 32(10), pp.1075-1088, 2005. A D-
optimal design will be a design that maximizes the determinant of the FIM.

We first define r
t,  . In our example: r(t,1 ,2 ) =
 
 
   
 
   being   1 , 2 the unknown parameters.
In[
]:= r
1, 2  1  2.0 t 0.09 2   0.001 t 0.2 2 ;   1, 2;
r
t r
t

Then, we compute 
r
t, 1 , 2  = { ,  }:
1 2

In[
]:= g
t   r
1, 2

Out[
]= 
 
   
 
 


 
 
 
 
 
 
 

Next, we specify the number of points to be used in the optimal design. We assume a 3-
point design with the first one, t0 , defined by the user.
In[
]:= n  3;

For computational purposes, we prefer to use the distance di = ti  ti1 , instead of ti , then ti =
i di with d0  t0 defined by the user. That is, the points to be estimated are:
In[
]:= dd  Table
di , i, n

Out[
]= & & & 

In[
]:= tm  Accumulate
Flatten
d0 , dd

Out[
]= & &  & &  &  & &  &  &  & 

We evaluate (r(t), ) at points t:{t0 , ..., tn }, obtaining X  X1 , ..., Xp  with X1 =

dr
t0  dr
tn  dr
t0  dr
tn 
{ d1
, ..., d1
}, ..., Xp = { dp
, ..., dp
}:

In[
]:= X  g
tm;

A typical choice for the covariance matrix is  = {lij } with lij = exp {|tj tj |} (meaning the
relationship between samples decays exponentially as the time-distance between them
increases), that is:
j1 i1
In[
]:= ff
i, j : Whichi  j, 1, i  j,  ki dk , i  j,  kj dk ;
356 Mathematica beyond mathematics

In[
]:=   Array
ff, n  1, n  1

Out[
]=  
 & 
 & &  
 & & &   
 &  
 & 
 & &  


 & &  
 &  
 &  
 & & &  
 & &  
 & 

We assume that all the measures have approximately the same uncertainty level, that is
2  2i , then covariance matrix is  = 2 . In our example, we assume  = 1 and  = 2.
In[
]:=   2;   1;   2 ;

We also need to give the initial values to 1 and 2 . The ultimate purpose is to find their real
values empirically. However, the method requires an initial guess, (we can use previous
experiments, information derived from similar experiments, etc.) and to find out how
sensible our choice of values has been we can perform a sensitivity analysis. We assume the
following values:
In[
]:= 1  100; 2  5;

The moment d0  t0 for taking the first measurement is defined by the user. In this case, we
take the measurement at t0 = 0.5.
In[
]:= d0  0.5;

Then we can obtain the information matrix (FIM): M = X T 1 X .

In[
]:= m  X . Inverse
. Transpose
X;

A D-optimal design will be a design that maximizes det|M|, the determinant of the
information matrix. This can be done using ,$% as follows (in the next example, we
show that usually it is better to use &,$ or ,$% ):
In[
]:= deter1
x : Det
m . Thread
dd
x;

In[ ]:= sol  Maximize
deter1
dd, Thread
dd  1, dd 

Out[ ]=   &   &  &   

This indicates that the samples should be taken in {ti }. The same solution can be directly
obtained using Biokmod (Section 10.4.4):
In[
]:= optimumt  tm . sol2

Out[
]=        

In practice, since it’s very difficult to take samples at exactly the precise time, we need to
measure the robustness of our calculations: how sensitive they are to small changes. For that
purpose, we calculate their efficiency using the following formula:
!&
(N is the total number of estimated points).
!$ 
For example, to measure the efficiency of d3 :
In[
]:= d3  Table
i, i, 3  d3 . sol2, 3, 3  d3 . sol2, 3, 0.1 ;
10. Modeling: Application in Biokinetics 357

Det
m . Thread
sol2, 1, sol2, 2 , d3


In[
]:= ListPlot n1 &  d3,
Det
m . sol2

Filling
Axis, ColorFunctionScaling
True

1.000

0.999

Out[
]= 0.998

0.997

0.996 10 20 30 40 50 60

The optimum point is the one with efficiency 1. In our case, this corresponds to
Det[m[max]], obtained when t:
In[
]:= optimumt4

Out[
]=  

As usual, to avoid any problems in the next section with previously defined variables we
quit the session.

10.5.2 Michaelis–Menten Models (M–M) and OED

We will now apply OED to the model described in Section 10.1.4 to choose the
best moments {t1 , ..., ti } to take the samples. In this case, the problem is that we
do not have an exact solution of the model so we cannot obtain an analytical
expression for {x1
t, k12 , k21 , x2
t, k12 , k21 }. Because of that, we apply the D-
optimal method described in the previous section. We want to design an experiment
to take the samples in compartment 1.

We repeat the process discussed in 10.1.4 to obtain x1
t, k12 , k21  :
In[
]:= Vm  0.1; km  0.3;
Vm x1
t
eq1  x1 '
t  k12 x1
t  k21 x2
t  ;
km  x1
t
eq2  x2 '
t  k12 x1
t  k21 x2
t;
In[
]:= sol  ParametricNDSolve

eq1, eq2, x1
0  1, x2
0  0, x1, x2, t, 0, 100, k12 , k21  ;
x1
t x1
t

We then proceed to compute numerically  x1
t, k12 , k21  = { , }. For convenience,
k12 k21
we call k12  a and k21  b:
In[
]:= fa
a1 ? NumberQ, b ? NumberQ, t ? NumberQ : D
x1
a, b, a
t . a
a1 . sol

In[
]:= fb
a ? NumberQ, b1 ? NumberQ, t ? NumberQ : D
x1
a, b, b
t . b
b1 . sol

In[
]:= X1
a, b, ti :  fa
a, b, ti, fb
a, b, ti

358 Mathematica beyond mathematics

Next, we define the number of points to be used in the optimal design. We assume a 5-point
design:
In[
]:= n  5  1;

After that, we compute  = {lij } with lij = exp {|tj tj |} (remember that di = ti  ti1 is used
instead of ti :
In[
]:= dd  Table
di , i, n; tt  FoldList
Plus, Subscript
d, 0, dd

Out[
]= & &  & &  &  & &  &  &  & &  &  &  &  & 
j1 i1
In[
]:= ff
i, j : Whichi  j, 1, i  j,  ki dk , i  j,  kj dk ;

In[
]:=   Array
ff, n  1, n  1;

Once we know , we compute the covariance matrix  = 2 . We assume  = 1,  = 1:

In[
]:=   1;   1;   2 ;

The next to last step is the calculation of the information matrix M = X T 1 X . Initially, we
assign arbitrary values to k12 and k21 . As mentioned in the previous section, their real values
are unknown, but we have to make an initial estimation based on the information that we
have. In this case we assume k12 = 0.03 and k21 = 0.02:
In[ ]:= m1
ti : Transpose
Map
X1
0.03, 0.02,  &, ti .
Inverse
.Map
X1
0.03, 0.02,  &, ti

,$% tries to find a global maximum using analytical methods that may take too long
to compute. For this reason, it may be more convenient to use ,$% or
&,$ , Mathematica functions that take a vector of numbers as their inputs. In our
example, with n = 4.
In[
]:= obj
d0 ? NumericQ, d1 ? NumericQ,
d2 ? NumericQ, d3 ? NumericQ, d4 ? NumericQ :
Det
m1
d0, d1  d0, d0  d1  d2, d0  d1  d2  d3, d0  d1  d2  d3  d4

Finally, we find the solution. Remember that a D-optimal design is a design that maximizes
det|M|, the determinant of the information matrix. We can show the intermediate
calculations with ', :
In[
]:= sol1  FindMaximum
obj
d0 , d1 , d2 , d3 , d4 ,
0  d0  10, 0  d1  10, 0  d2  10, 0  d3  10, 0  d4  10,
d0 , 3, d1 , 3, d2 , 3, d3 , 3, d4 , 3 , StepMonitor 
Print
"d0:", d0 , "d1:", d1 , "d2:", d2 , "d3:", d3 , "d4:", d4 
&-  &-  &-  &-  &- 

&-    &-   &-   &-   &-   

&-   &-  &-   &-   &-   

&-   &-  &-   &-   &-   

&-  &-  &-   &-  &-   

&-  &-   &-   &-  &-   

&-  &-   &-   &-  &-   

Out[
]=  &  &   &   &  &   

This indicates that the samples should be taken at {ti }:

10. Modeling: Application in Biokinetics 359

In[
]:= bestsampling  tt . sol12

Out[
]=       

Sensitivity Analysis of k12 , k21

As we have explained before, to solve D-optimal designs we need to give the initial
values to the starting points k12 and k21 . To learn more about how different values
for those points would impact the design, we need to perform a sensitivity analysis.
One way to proceed would be to calculate the efficiency, as we did in the example
of Section 10.3.1, but in this example we will do it using derivatives.

To analyze the sensibility of x1 with respect to k12 , we keep k21 constant at 0.02 and then
x1
k12 , 0.02
calculate k12
, that we call s
t, k12 , to see how it changes over time. The closer the
derivative is to zero, the less sensitive x1 will be to changes in initial values of k12 . We can
x1
0.03,k21 
do the same for k21 , calculating in this case s
t, k21   k
(Notice that Mathematica
21

actually calculates the numerical derivatives of x1[k12, 0.02][t] and x1[0.03, k21][t], since
both are numerical expressions).
In[
]:= Plot D
x1
k12, 0.02
t, k12 . k12
0.03 . sol,
D
x1
0.03, k21
t, k21 . k21
0.02 . sol,
t, 0, 20, AxesLabel
"t", "s t ",
 x1 k12 , 0.02  x1 0.03, k21
PlotLegends
" ", " "
 k12  k21
s
t

t
5 10 15 20

x1 k12 ,0.02

1
k12
Out[
]=
x1 0.03,k21
k21

2

3

The derivative gives an idea of the absolute sensibility, but we are more interested in the
relative sensitivity (rs). This sensibility can be analyzed using the expressions rs
t, k12  
x1
0.03, 0.02
t  s
t, k12  and rs
t, k12   x1
0.03, 0.02
t  s
t, k21 . A good choice
for  is 0.1.
360 Mathematica beyond mathematics

In[
]:= GraphicsRow
 Plot

Evaluate
x1
0.03, 0.02
t  0, .1, .1 D
x1
k12, 0.02, k12
t .
k12
0.03 . sol, t, 0, 20,
Filling
2
3, AxesLabel
"t", "rs t, k12 ", Plot

Evaluate
x1
0.03, 0.02
t  0, .1, .1 D
x1
k21, 0.02, k21
t .
k21
0.02 . sol, t, 0, 20, Filling
2
3,
AxesLabel
"t", "rs t, k21 ", ImageSize
500, 200

rs
t, k12  rs
t, k21 

1.0 1.0

0.8 0.8

0.6 0.6
Out[
]=

0.4 0.4

0.2 0.2

t t
5 10 15 20 5 10 15 20

The left graph indicates that the sensitivity of the solution with respect to k12 increases with t
until t = 10. We can see the pattern by looking at the difference in the measurement of
rs
t, k12  between the 3 potential values of k12 around t = 10. After that, it decreases.
Basically, this means that measurements taken at the beginning or at the end of the interval
{0, 20} would not change significantly for different values of k12 . However, measurements
taken around t  10 may show large discrepancies if our initial estimate for the constant was
not the correct one. For k21 , we obtain a similar conclusion (right graph).

10.6 BIOKMOD: The New Iodine OIR Model (ICRP 137)

In[
]:= ClearAll
"Global` "

In this section, we are going to use the Mathematica package SysModel2 included
in the application Biokmod.
Biokmod can be downloaded at: http://diarium.usal.es/guillermo/biokmod/
To install it, extract the file "biokmodXX.zip" and copy the folder named Biokmod
to the AddOns\Applications directory in the Mathematica installation folder
($InstallationDirectory).

Load the package SysModel2 as usual:
In[
]:= Needs
"Biokmod`SysModel2`"

SysModel, version 2.0.b7 20201130

Here are the package functions:

10. Modeling: Application in Biokinetics 361

In[
]:= ? "Biokmod`SysModel2` "

One of the most important applications of biokinetic models is related to internal

dosimetry. The most commonly used dosimetry calculation methods are the MIRD
(Medical Internal Radiation Dose) and the ICRP ( International Commission on
Radiological Protection). Although both are similar, they use different terminology
and symbols for fundamental quantities, such as the absorbed fraction, specific
absorbed fraction, and various dose coefficients. The MIRD is devoted to the
application of internal dosimetry in diagnostic and therapeutic nuclear medicine.
The International Commission on Radiological Protection (ICRP) is an
organization that also supplies dosimetric models and technical data for use in
providing recommendations regarding limits on ionizing radiation exposure to
workers and members of the general public. MIRD Pamphlet No. 21 describes a
integrated method for MIRD and ICRP.
MIRD Pamphlet No. 21: A Generalized Schema for Radiopharmaceutical Dosimetry—
Standardization of Nomenclature. Wesley E. Bolch, Keith F. Eckerman, George Sgouros and
Stephen R. Thomas. Journal of Nuclear Medicine March 2009, 50 (3) 477-484;
DOI: https://doi.org/10.2967/jnumed.108.056036
Both schema usually require knowledge of the disintegration in a region (or
compartment). The isotope contents xi
t in a compartment i can be calculated
following the methods that we have described previously. Therefore, the number of
transformations, in Bqs, at time in a source region S following an acute intake is
given by the following integral:
Us
  xi
t t
10.8
0

The ICRP has updated the biokinetic models ICRP 130, ICRP 134, ICRP 137,
ICRP 141:
https://journals.sagepub.com/doi/suppl/10.1177/ANIB_48_2-3/suppl_file/OIR_Data
_Viewer_for_P134-P137-P141.zip
The flow diagram for both, ingestion and injection, is presented in the figure below:
362 Mathematica beyond mathematics

Figure 10.7 New iodine model ICRP 137.

The transfer coefficients are:

In[
]:= transfercoeffI  1, 4, 7.26, 1, 16, 11.84, 1, 3, 5.16,
1, 10, 8.6, 1, 12, 600, 1, 8, 25, 1, 6, 15,
3, 22, 50, 10, 21, 50, 4, 5, 95, 4, 1, 36,
5, 2, 0.0077, 5, 1, 0, 12, 1, 330,
12, 13, 35, 13, 12, 56, 8, 1, 100, 6, 1, 100,
2, 14, 15, 14, 2, 21, 14, 15, 1.2, 15, 14, 0.62,
15, 1, 0.14, 2, 9, 3.6, 9, 2, 21, 9, 1, 0.14,
2, 7, 21, 7, 2, 21, 7, 1, 0.14, 7, 11, 0.08,
22, 19, 7200, 19, 21, 2160, 21, 20, 20.57,
20, 11, 6, 20, 1, 594, 16, 18, 12, 11, 23, 2,
23, 24, 2.01, 24, 17, 2.02, 25, 26, 2160, 26, 27, 20.57,
27, 1, 6 fA 1  fA , 27, 11, 6, 28, 25, 7200;
where:
fA is the absorption factor from the alimentary tract. Its value depends on the chemical form of the iodine
intake.
I refers to the decay constant. Its value depends on the isotope.

Then the compartmental matrix as function of fA and I is:
In[
]:= iodinematrix
fA, I  CompartMatrix
28, transfercoeffI , I ;
10. Modeling: Application in Biokinetics 363

The package function AcuteInput computes the content in each compartment as a

function of the time for a single intake at t  0. Alternatively, SystemDSolve can also be
used.
AcuteInput
A, IC,t1, xgives the solution X(t) = {x1(t), ..., xn(t) at time t1 of an
ODE system X'(t) = A X(t) , with initial conditions X(0) = IC
In[
]:= qIodine
fa, lambda, initialcondition, t :
AcuteInput
iodinematrix
fa, lambda, initialcondition, t, x;

The response function for each compartment in case of ingestion at t = 0 (bi = 0 with i  28
and b28 = 1) is created as follows:
In[
]:= initcondition  PadLeft
1, 28

Out[
]= 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1

If we were dealing with injection instead, the b1 coefficient would have been 1 and the rest 0.

Let’s consider the case of I131 ingestion with fA = 0.99 (the value recommended by default
for ICRP 134)
In[
]:= lambdaIodine131 

QuantityMagnitude iodine-131 ISOTOPE
"DecayConstant", "Per day"

Out[
]= 0.086371

In[
]:= iodinecontent 

qIodine
0.99, lambdaIodine131, initcondition, t ExpandAll Chop;

The Us
 (number of transformations, in Bqs, at time ) for 1 Bq in t = 0 ( b28 =1) in each
compartment is given by the expression below:
In[
]:= UsI131
t1  86 400 qContinuous
1, 1, t, t1 &  Evaluate

Evaluate
Table
xi
t, i , 1, 28 . iodinecontent ExpandAll;

We can use the same approach to compute the accumulated disintegrations Us (rounded to
the nearest integer) in all compartments for different values of (in days).
In[
]:= tm  0.1, 0.5, 1.0, 5, 10, 100, 1000;

In[
]:= nround
x : MapAll
If
NumberQ
, Round
,  &, x;

In[
]:= DisintegrationLabel
dis, tafter :

TableForm
Map
Flatten
 &, nround
dis,
TableHeadings
None, Flatten
"Compartment", tafter "day";
In[
]:= dis  Transpose
Join
Range
28, UsI131
 &  tm Quiet;
364 Mathematica beyond mathematics

In[
]:= DisintegrationLabel
dis, tm

Out[
]//TableForm=
Compartment 0.1 day 0.5 day 1. day 5 day 10 day 100 day 1000 day
1 931 3695 4579 4816 4827 4861 4861
2 0 3 16 404 1081 2939 2940
3 73 373 470 496 497 501 501
4 47 203 253 267 267 269 269
5 173 5409 16 050 95 151 160 639 271 834 271 862
6 123 548 685 722 723 728 728
7 0 2 14 392 1060 2897 2897
8 204 914 1142 1203 1206 1214 1214
9 0 0 2 67 182 498 499
10 122 622 784 827 829 835 835
11 46 335 563 741 767 842 842
12 1592 6682 8315 8752 8773 8834 8834
13 770 4094 5173 5462 5474 5513 5513
14 0 1 10 275 749 2059 2060
15 0 0 3 260 904 2920 2920
16 340 3242 4406 4718 4729 4762 4762
17 0 8 98 3889 8121 17 918 17 928
18 127 9269 32 207 213 297 368 634 661 425 661 562
19 2 9 11 11 12 12 12
20 7 77 103 110 110 111 111
21 220 2265 3005 3202 3210 3232 3232
22 1 3 3 3 3 3 3
23 3 117 340 704 729 803 803
24 0 31 165 668 694 767 767
25 40 40 40 40 40 40 40
26 3646 4182 4183 4183 4183 4183 4183
27 124 143 143 143 143 143 143
28 12 12 12 12 12 12 12

The graph that follows displays the disintegration in the thyroids (5). Note that most of the
disintegrations happen during the first few days.
In[
]:= LogLogPlot
Evaluate
UsI131
t15, t1, 0, 100

1  105
5  104

1  104
Out[ ]=
5000

1000
500

0.5 1 5 10 50 100

10.7 Additional Modeling Examples

Mathematica includes many more capabilities for modeling using differential
10. Modeling: Application in Biokinetics 365

equations. In this section, we show some of those capabilities using basic examples.
In[
]:= ClearAll
"Global` "

10.7.1 DEs Using WhenEvent

We can use WhenEvent to solve DEs with variable conditions:
https://reference.wolfram.com/language/ref/WhenEvent.html

The instructions below simulate and visualize the first 20 seconds of a bouncing ball
dropped from a height of 10 m that retains 90% of its velocity in each bounce.
In[
]:= sol1  NDSolve
y ''
t  9.81, y
0  10, y '
0  0,
WhenEvent
y
t  0, y '
t
0.9 y '
t, y, t, 0, 20;
In[
]:= Plot
y
t . sol1, t, 0, 20

10

Out[
]=

5 10 15 20

This next example models a ball falling down a 1-meter-wide staircase from a height of 11
m and an initial velocity of 3 m/s. The first stair is located at h = 10 m and the ball retains
80% of its velocity in each bounce.
In[
]:= sol2 
Block
c  .8, NDSolve
y ''
t  9.81, y
0  11, y '
0  3, a
0  10,
WhenEvent
y
t  a
t  0, y '
t
c y '
t, WhenEvent
Mod
t, 1,
a
t
a
t  1, y, a, t, 0, 8, DiscreteVariables
a ;
Plot
Evaluate
y
t, a
t . sol2, t, 0, 8, Filling
2
0

10

Out[
]=

2 4 6 8

10.7.2 Delay Differential Equations (DDEs)

366 Mathematica beyond mathematics

We can also model systems in Mathematica using delay differential equations

(DDEs), a type of differential equation in which the derivative of the unknown
function at a certain time is given in terms of the values of the function at previous
times. DDEs are also called time-delay systems, systems with after-effect or dead-
time, hereditary systems, equations with deviating argument, or differential-
difference equations.
https://reference.wolfram.com/language/tutorial/NDSolveDelayDifferentialEquations.html
A typical example of these type of models are the Lotka–Volterra systems. These
systems model the growth and interaction of animal species assuming that the
effect of one species on another is delayed by introducing time lags in the
interaction terms.

The code below solves and visualizes the system with and without delay:
In[ ]:= lvsystem
1, 2 : Y1 
t  Y1
t Y2
t  1  1 ,
Y1
0  1, Y2 
t  Y2
t 2  Y1
t  2 , Y2
0  1;
In[
]:= lv  First
NDSolve
lvsystem
0, 0, Y1 , Y2 , t, 0, 25;

In[
]:= lvd  Quiet
First
NDSolve
lvsystem
.01, 0, Y1 , Y2 , t, 0, 25;

In[
]:= ParametricPlot
Evaluate
Y1
t, Y2
t . lv, lvd, t, 0, 25

2.5

2.0

1.5

Out[
]=

1.0

0.5

1.0 1.5 2.0 2.5 3.0 3.5 4.0

10.8 Modeling Using PDEs

In this section, we explore some examples of modeling using PDEs. Although in
certain cases, we may be able to obtain analytical solutions using DESolve, in most
situations we will need to use numerical methods.
In[ ]:= ClearAll
"Global` "
10. Modeling: Application in Biokinetics 367

10.8.1 The Transport Equation

.
The transport equation describes the transportation of a scalar quantity in a space.
In general, it applies to the transportation of a scalar field (e.g., temperature,
material properties or chemical concentration) inside an incompressible fluid. In
mathematics, this equation is also known as the convection-diffusion equation, a
first-order PDE. This is the equation used in the most common transportation
models.



  ./0..
 

Here is an example of a first-order linear partial differential equation modeling a wave:

In[
]:= u1
x, t  DSolveValue
D
u
x, t, t  D
u
x, t, x  0,
u
x, 0  Exp
x ^ 2 , u
x, t, x, t
2
Out[
]= 
xt

The command below simulates its movement:
In[
]:= Animate
Plot
u1
x, t, x, 0, 10,
PlotRange
All, Ticks
False, Filling
Axis,
t, 0, 10, DefaultDuration
15, AnimationRepetitions
1

10.8.2 Tsunami Modeling

We can also model tsunamis with the following PDE:
2 u
t, x 2 u
t, x
  a u
t, x3  b u
t, x2  c u
t, x  d,
 t2  x2
2
 u
t, x
con u
0, x  x ,  0 u
t, x0   u
t, x0 
t

The example below solves and visualizes the previous equation for a given set of parameters:
368 Mathematica beyond mathematics

In[
]:= Manipulate

Plot3D
Evaluate
u
t, x . Quiet
NDSolve
Evaluate
D
u
t, x, t, t 
D
u
t, x, x, x  a u
t, x ^ 3  b u
t, x ^ 2  c u
t, x  d,
u
0, x  E ^ x ^ 2, D
u
t, x, t  0 . t
0,
u
t, x0  u
t, x0, u, t, 0, x0, x, x0, x0,
x, x0, x0, t, 0, x0, MeshFunctions
3 &,
Mesh
None,
ColorFunction
"Rainbow",
PlotPoints
20,
a, 0.3, 4, 0,
b, 0.6, 4, 0,
c, 0.8, 4, 1,
d, 0, 1, 1,
x0, 16, "Solution Range", 5, 20,
ControlPlacement
Left

10.8.3 Burgers’ Equation

Burgers’ equation is a PDE commonly used in applied mathematics when solving
problems in areas such as fluid mechanics, nonlinear acoustics, or gas dynamics.
In[
]:= ClearAll
"Global` "

u 2 u u
 u 0
t  x2 x

Here’s an example of the equation:
In[
]:= ufun  NDSolveValue

D
u
t, x , t  1 100 D
u
t, x, x, x  u
t, x D
u
t, x, x,
u
t, 0  1, u
t, 1  1, u
0, x  Cos
2 Pi x, u, t, 0, 2,
x, 0, 1, Method
"MethodOfLines", "SpatialDiscretization"

"FiniteElement", "MeshOptions"
"MaxCellMeasure"
0.01;
In[
]:= list  Table
Plot
ufun
t, x, x, 0, 1,
PlotRange
1.1, 1.1, ImageSize
Large, t, 0, 2, 1 20;
ListAnimate
list, SaveDefinitions
True
10. Modeling: Application in Biokinetics 369

Notice the use of the finite element method (FEM) to solve the equation. To learn more
about it, the “Finite Element User Guide” provides an overview of how to solve PDEs using
this method:
https://reference.wolfram.com/language/FEMDocumentation/tutorial/FiniteElementOvervie
w.html

10.9 System Modeler

For more advanced modeling needs, we can use System Modeler, a program for
modeling and simulating physical systems: https://www.wolfram.com/system-
modeler/.
It is a full graphical modeling and simulation environment where by using drag and
drop, you can build realistic, multi-domain models in areas as diverse as aerospace,
industrial manufacturing, and life sciences.
Fortunately, we can explore some of the software’s capabilities from within
Mathematica using functions such as SystemModelExamples and
SystemModelPlot:

We first select the desired example and then plot its behavior:
In[
]:= SystemModelExamples


In[
]:= SystemModelPlot
"IntroductoryExamples.Systems.InvertedPendulum"

Cart position
m

1.0
Position
0.5

t Reference
5 10 15 20
0.5

1.0
Out[
]=
Pendulum angle


5
pendulum.angle
t
5 10 15 20
5

10
370 Mathematica beyond mathematics

10.10 Additional Resources and references

About Compartmental and Physiological Modeling
J.A. Jacquez, Compartmental Analysis in Biology and Medicine, The University of Michigan
Press, 1985.
K. Godfrey, Compartmental Models and Their Application, Academic Press, London, 1983.
About Epidemic Modeling
https://community.wolfram.com/groups/-/m/t/1872608
About Biokmod and Optimal Experimental Design (OED)
Visit the author’s website:
https://diarium.usal.es/guillermo/publicaciones/especializadas/
Biokmod can be downloaded from: http://diarium.usal.es/guillermo/biokmod. Extensive
documentation is included.
Some of its features can be run directly over the Internet with the help of webMathematica:
http://oed.usal.es/webMathematica/Biokmod/index.html.
J. Lopez-Fidalgo, J.M. Rodriguez-Diaz, J. M. Sanchez, G. Santos-Martin, MT, Optimal
designs for compartmental models with correlated observations, Journal of Applied
Statistics: 32(10) pp. 1075-1088, 2005.
J. M. Sanchez, J. M. Rodríguez-Díaz, Optimal design and mathematical model applied to
establish bioassay programs, Radiation Protection Dosimetry, doi:10.1093/rpd/ncl499, 2007.
J. M. Rodríguez-Díaz, G. Sánchez-León, Design optimality for models defined by a system
of ordinary differential equations, Biometrical Journal 56 (5), pp. 886-900, 2014.
About Differential Equations
The general guide for DifferentialEquations:
https://reference.wolfram.com/language/guide/DifferentialEquations.html
Working with DSolve: A User's Guide:
https://reference.wolframcloud.com/language/tutorial/DSolveWorkingWithDSolve.html
Symbolic Differential Equation Solving:
https://reference.wolframcloud.com/language/tutorial/DSolveOverview.html
Introduction to Advanced Numerical Differential Equation Solving in the Wolfram
Language:
https://reference.wolfram.com/language/tutorial/NDSolveIntroductoryTutorial.html
Advanced Numerical Differential Equation Solving in the Wolfram Language:
https://reference.wolfram.com/language/tutorial/NDSolveOverview.html
Partial Differential Equations (PDEs):
https://reference.wolframcloud.com/language/tutorial/DSolvePartialDifferentialEquations.ht
ml
Numerical Solutions of Partial Differential Equations:
http://reference.wolfram.com/language/tutorial/NDSolvePDE.html
 11
Economic, Financial, and Optimization
Applications

This chapter explains Mathematica’s functionality related to economics

and finance. It covers how to access economic information, price
financial instruments, such as bonds and derivates, and access both,
historical and real-time financial data. It also discusses constrained
optimization using mainly examples related to operations research and
investment portfolio management.
  , a new function to
solve problems like the Traveling Salesman Problem in a very efficient
way, is shown in the context of creating a tour of South American cities.
The last section provides a very short introduction to blockchains and
cryptocurrencies, including a brief overview of the program’s capabilities
in these two areas.

11.1 Accessing Economic Information

One of the most useful features available in Mathematica is the free-form input,
enabling us to directly access the functionality of Wolfram|Alpha). Here are some
economic-related examples:

We can use the free-form input to add different currencies using exchanges rates that are at
most a few minutes old. For real-time calculations, we would need access to the Wolfram
Finance platform.

In[
]:=  35 dollars
24 euros
300 yens in dollars

UnitConvert
$35.00
24.00
300 , "USDollars"

Out[
]= $60.39

Type Japan vs. Germany using the Wolfram|Alpha Query style:

In[
]:=  Japon vs Germany

372 Mathematica beyond mathematics

Output not shown

Several frames (pods) will be displayed, each of them containing information about an
aspect of the question. In the right corner of each frame, the symbol + appears. Click on one
of them and you will see a menu with several options. In this example, we chose the
Economic Properties frame and clicked on the Subpod content. The following entry was
generated afterward:
In[ ]:= WolframAlpha"Japon vs Germany",
"EconomicProperties:WorldDevelopmentData", 1, "Content"

Japan Germany

GDP $5.058
    $3.846
   

     

    
 


GDP at parity $5.334
    $4.561
   

    
 

    
 


real GDP $4.381
    $3.436
   

     

    
 


Out[
]= GDP per capita $40 193 $46 253

         

    
 

    
 


GDP real growth
4.586   
4.57   

     

     


Gini index 0.329
    
  0.317
    
 




consumer price inflation
0.23    3.14   

     

    
 


We repeat the same steps for “China vs US” but in this case instead of Subpod content, we
choose ComputableData. The output is shown below. The advantage of this choice is that
the output is being generated as a list allowing easy subsequent manipulation.
11. Economic and Financial Applications 373

In[
]:= WolframAlpha"China vs US",

"EconomicProperties:WorldDevelopmentData", 1, "ComputableData"

Out[
]=

, China, United States,

GDP, $1.47227  1013 per year , $2.0953  1013 per year ,

GDP at parity, $2.42832  1013 per year , $2.0953  1013 per year ,

real GDP, $1.46318  1013 per year , $1.92945  1013 per year ,

GDP per capita, $10 430. per person per year ,

$63 210. per person per year ,

GDP real growth, 2.348 per year , 3.405 per year ,

Gini index, 0.382, 0.415,

consumer price inflation, 0.98 per year , 4.7 per year 

In the WolframAlpha style you can select the information you are interested in
by choosing from the menu that appears after clicking on +  Subpod content
or ComputableData in the desired frame. An input cell will be generated to get
only the desired information.

The graph below shows the evolution of the gross domestic product, GDP, of China from
1980 to 2021.
In[
]:= DateListPlotCountryData"China", "GDP", 1980, 2021

1.4  1013

1.2  1013

1.0  1013

8.0  1012
Out[
]=
6.0  1012

4.0  1012

2.0  1012

0
1980 1990 2000 2010 2020

We can also use Wolfram|Alpha to access data related to demographics, and make
predictions.

Here the free-form input is used to import data. In this example, we use it to forecast
China’s population.
 374 Mathematica beyond mathematics

In[
]:=  China population since 1980

Population history

China COUNTRY

population Interval

  ,       

Time: 01 Jan 1980 to 01 Jan 2022

Out[
]= TimeSeries 
Data points: 43

Time series data can be fitted using TimeSeriesModelFit.

In[
]:= ts  TimeSeriesModelFit 

Family: ARIMA
Out[
]= TimeSeriesModel 
Order: 2, 3, 0

Now, the evolution of China’s population over the next 10 years can be estimated using
TimeSeriesForecast.
In[
]:= DateListPlotts"TemporalData", TimeSeriesForecastts, 10,
PlotLegends  "Historical", "Forecast",
PlotLabel  Style"Chinese Population", Bold

Chinese Population

1.4  109

1.3  109
Historical
Out[
]=
9
1.2  10 Forecast

1.1  109

1.0  109

1980 1990 2000 2010 2020 2030

11.2 Financial Information

11.2.1 FinancialData
The FinancialData function in many occasions is the best way to access
finance-related information. Although in theory the program can display
thousands of indicators, in reality the accessible ones are mostly those related to
the US markets. In any case, regardless of whether the information may or may not
be directly accessible from FinancialData, nowadays it would be easy to access
it. For example, many bank platforms enable their clients to download data in an
Excel format that could then be imported into Mathematica and analyzed using any
11. Economic and Financial Applications 375

of the multiple functions related to finance:

Those readers who may need reliable access to other markets or real-time financial
data would probably be interested in taking a look at the Wolfram Finance
Platform, a product from Wolfram Research (the developers of Mathematica) that
focuses on finance:
http://www.wolfram.com/finance-platform/
FinancialData[“name”,”property”,{start,end,interval}] gives the value of the
specified property for the financial entity “name” (it could be an index, a stock, a
commodity, etc.) for a given period of time (by default it returns the most recent
value). We recommend reading the documentation pages to know more about this
function’s capabilities.

The symbols (“name”) used are the ones available in http://finance.yahoo.com/ since Yahoo
Finance supplies the information to the function. In the website, we can find that the symbol
for the Standard & Poor’s 500 index, that includes the 500 largest public companies in
terms of market capitalization that trade in either NYSE or NASDAQ is: ^GSPC (we can
also use SP500). We can check it as follows:
In[
]:= FinancialData"^GSPC", "Name" 

Out[
]= S&P 500 Index

If we want to see the index value (with a few minutes’ delay) we can type:
In[
]:= FinancialData"SP500"

Out[
]= 3677.95

We can also access historical data. In this example, we show the trajectory of the S&P 500
index.
In[
]:= DateListPlotFinancialData"^GSPC",
"January 1 2000", "September 30 2022", Joined  True
5000

4000

3000

Out[
]=
2000

1000

0
2000 2005 2010 2015 2020

If the financial entity is trading at the time of the query, the price that we get is the
one from a few minutes earlier (usually the delay is 15 minutes). The access to
information in “real time” requires additional payments to 3rd-party providers such
as Bloomberg or Reuters. There are even people willing to pay very large amounts
of money to gain an advantage of a few milliseconds when trading:
376 Mathematica beyond mathematics

http://adtmag.com/articles/2011/07/29/why-hft-programmers-earn-top-salaries.aspx.

The following function will dynamically update the S&P500 index (as long as the
command is executed during trading hours):
In[
]:= sp500 : FinancialData"SP500"

In[
]:= FinancialData"SP500"

Out[
]= 3677.95

The output changes every two seconds:

In[
]:= Dynamicsp500, UpdateInterval  2

Out[
]= sp500

We can also represent it in a self-updating graph:

data  ;
DynamicLastAppendTodata, sp500,
UpdateInterval  2, TrackedSymbols  
DynamicDateListPlotdata, Joined  True, SynchronousUpdating  False

As mentioned earlier, if the entity is not available in FinancialData we may still be able to
import the information from specialized websites. An interesting one is Nasdaq Data Link
(formerly known as Quandl), https://data.nasdaq.com/, that provides access to numerous
free and premium economic and financial databases. We can first select the data and
download it in our chosen format: Excel, CSV or any other. Then we can use Import to
analyze them using Mathematica. Alternatively, we can access the data directly using
Import or URLFetch.

11.2.2 Visualization
The number of specialized graphs for economic and financial visualization has
increased significantly in the latest versions of Mathematica.

All the functions below end in Chart. We have already seen some of them in Chapter 6. In
this section, we’re going to cover the ones normally used for analyzing stock prices.
In[
]:= ? Chart

Choose any of them and click on it to access its documentation.

11. Economic and Financial Applications 377

These graphs complement the capabilities of FinancialData but they can also be
used with other functions as shown below:

Let’s visualize the historical closing prices for the Google using different graphs.
In[
]:= data  FinancialData"GOOGL", "Close", 2021, 1, 1, 2022, 9, 30;

If we click on any point inside the plot, we’ll see the information associated to it.
In[
]:= GraphicsColumn
KagiChartdata, PointFigureChartdata, LineBreakChartdata

140.

120.

100.

80.
2021 2022

140.

120.

Out[
]=
100.

80.
2021 2022

140.

120.

100.

80.
2021 2022

The next graph uses the function RenkoChart with EventLabels and Placed to add symbols
(“” y “”) indicating in what position they must be placed.
378 Mathematica beyond mathematics

In[
]:= RenkoChart"NYSE:GE", 2020, 01, 1, 2021, 12, 31,

EventLabels  2020, 01, 30  Placed"", Before,
2020, 05, 31  Placed"", Above

100.

80.

Out[
]=


60.

40.

2020 2021

The graph below displays the trajectory of GE in the NYSE from January 1, 2021 to
December 31, 2021. For each trading day, we can see the open, high, low, and close prices.
In[ ]:= CandlestickChart"NYSE:GE", 2021, 1, 1, 2021, 12, 31

110.

100.

Out[
]=

90.

80.
Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan

ChartElementFunction is an option that when added to BarChart or PieChart3D, improves

the presentation of the data.
In[
]:= GraphicsRowBarChart1, 5, 2, 4, 5, 7,
ChartElementFunction  "GradientScaleRectangle",
PieChart3D3, 2, 2, 4, 1, ChartElementFunction  "TorusSector3D",
SectorOrigin  Automatic, 1, ChartStyle  "Rainbow", Frame  All
11. Economic and Financial Applications 379

In the world of professional stock market investors, there’s a group that uses chart
analysis as its main tool for making investment decisions. Its members are known
as chartists and their techniques are known as technical analysis. They closely
study the past performance of stocks or indices to find trends and support and
resistance levels. These people associate certain graph shapes to future markets ups
and downs. Although there’s no sound mathematical theory justifying their
predictions, all economic newspapers have a section dedicated to this type of
analysis.

The powerful function InteractiveTradingChart, includes practically all the functionality
that a chartist may require. In this example, we display GE’s share price history for a given
period. Notice that in the upper area of the graph, we can see a trimester at a time and using
the bottom part we can move through the entire period. We can also choose the time
interval (days, weeks or months) and even what chart type and indicators to show.
In[
]:= InteractiveTradingChart"NYSE:GE", 2020, 01, 1, 2022, 05, 31
380 Mathematica beyond mathematics

11.2.3 Automatic Adjustments

With Mathematica we can automatically fit our financial data to a given
distribution (by default the program will use the maximum likelihood method) with
the function EstimatedDistribution.

Let’s fit Google’s share price to a lognormal distribution after downloading the data.
In[
]:= googleStock 
FinancialData"GOOG", 2015, 1, 1, 2022, 09, 30, "Day", "Value";
In[ ]:= dist  QuantityMagnitude
EstimatedDistributiongoogleStock, LogNormalDistribution, 
Out[
]= LogNormalDistribution4.08435, 0.483913

The fit is poor on both tails. As a matter of fact, all the efforts that have been made trying to
predict financial markets have produced unsatisfactory results. These markets exhibit what
Mandelbrot (the father of fractal geometry) called “wild randomness” (single observations
can impact the total in a very disproportionate way).
11. Economic and Financial Applications 381

In[
]:= QuantilePlotgoogleStock, dist, Filling  Automatic, FillingStyle  Red

140

120

100

80
Out[
]=

60

40

20

0
0 50 100 150

11.3 Financial Functions

The functions related to financial calculations can be found in guide/Finance. We
will give an overview of some of them in this section.

11.3.1 Time Value Computations

ClearAll"Global` "

The following functions are included under the Time value Computations category:
TimeValue, EffectiveInterest, Annuity, AnnuityDue and Cashflow. Let’s see some
examples.
TimeValue ["s", "i", "t"] calculates the time value of a security s at time t for an
interest specified by i.

If we invest an amount c during t years at an annual interest rate of i, to find the future value
Mathematica will apply the following formula:
In[
]:= TimeValuec, i, t

Out[
]= c i  1t

How much do we need to invest at 3.5% annual interest rate to get 100,000 in 10 years?
In[
]:= SolveTimeValuequantity, .035, 10 100 000, quantity

Out[
]= quantity  70 891.9

We have two options to invest 60,000 for 10 years. Option A offers a fixed annual interest
rate of 4% during the 10 years. Option B offers an interest rate of 3% during years 1 and 2,
4 % in years 3 to 5, and 4.5% until the end of the investment period. Which option is the
best one?
In[
]:= TimeValue60 000, 0.04, 10 product A ,
TimeValue60 000, 0, 0.03, 2, 0.04, 5, 0.045, 10 product B 
Out[
]= 88 814.7, 89 229.2

The function EffectiveInterest"r", "q"returns the effective interest rate

corresponding to an interest specification r, compounded at time intervals q. It is
382 Mathematica beyond mathematics

very useful to avoid making mistakes when comparing nominal and effective
interest rates. Annuity"p", "t"represents an annuity of fixed payments p
made over t periods. Below we show examples using both functions.

What is the effective rate of an annual nominal interest rate of 4% compounded quarterly?
In[ ]:= EffectiveInterest.04, 1 4

Out[
]= 0.040604

We’d like to deposit 60,000 in an account with an annual interest of 4% compounded

quarterly. How much money will we be able to withdraw every quarter so the money lasts
for 10 years?
In[
]:= SolveTimeValueAnnuitywd, 60 000, 10, 1 4,
EffectiveInterest.04, 1 4, 10 0, wd
Out[
]= wd  1827.34

The next set of functions creates a loan amortization table for a given period and interest
rate. In this case, we borrow 150,062 and make monthly payments of 1,666 during 10
years at 6% nominal annual rate. The same approach can be used with different numbers:
In[
]:= years  10; monthlypayments  1666; monthlyinterest  0.06 12;

In[
]:= loanValuemonth  TimeValue

Annuitymonthlypayments, years 12  month, monthlyinterest, 0;
In[ ]:= amortizationList 
Tablei
1, loanValuei, monthlypayments, loanValuei
monthlyinterest, monthlypayments  loanValuei monthlyinterest,
loanValuei
1, i, 0, years 12  1 Round;
In[
]:= amortizationTablemonth1, month2 :
TableFormamortizationList month1 ;; month2 , TableHeadings 
None, "Month", "Beginning\nPrincipal", "Monthly\nPayment",
"Interest\nPayment", "Principal\nPayment", "Ending\nPrincipal"

Let’s see the amortization table during the first 12 months. Try to display the entire period
of the loan.
In[
]:= amortizationTable1, 12
Out[
]//TableForm=

Month Beginning Monthly Interest Principal Ending

Principal Payment Payment Payment Principal
1 150 062 1666 750 916 149 147
2 149 147 1666 746 920 148 226
3 148 226 1666 741 925 147 302
4 147 302 1666 737 929 146 372
5 146 372 1666 732 934 145 438
6 145 438 1666 727 939 144 499
7 144 499 1666 722 944 143 556
8 143 556 1666 718 948 142 607
9 142 607 1666 713 953 141 654
10 141 654 1666 708 958 140 697
11 140 697 1666 703 963 139 734
12 139 734 1666 699 967 138 767

We can check graphically how the remaining principal goes down over time until it
becomes 0 at the end of the loan:
11. Economic and Financial Applications 383

In[
]:= ListLinePlotTablei
1, loanValuei, i, 0, years 12  1,

AxesLabel  "Month", "Remaining principal", ImageSize  300
Remaining principal

150 000

100 000
Out[
]=

50 000

Month
20 40 60 80 100 120

11.3.2 Bonds
ClearAll"Global` "

Bonds are financial assets that we can also analyze with Mathematica. When an
organization needs capital, instead of borrowing from banks, they can issue bonds.
Bonds can be considered a type of loan in which the lenders are the ones
purchasing them. Those lenders can usually be anyone. There are different types of
bonds. Normally they pay a fixed amount of interest at predetermined time
intervals: annually, half-annually, quarterly or even monthly. Some bonds issued
for short periods of time (less than one year), such as Treasury Bills, don’t pay
interest explicitly, they are sold at a discount from their par value.
For analyzing bonds, Mathematica has the function FinancialBond. It takes the
following arguments (some of them are optional):
“FaceValue” face value, par value
“Coupon” coupon rate, payment function
“Maturity” maturity or call/put date
“CouponInterval” coupon payment interval
“RedemptionValue” redemption value
“InterestRate” yield to maturity or yield rate
“Settlement” settlement date
“DayCountBasis” day count convention
Let’s see some examples.

The yield to maturity of a 1,000 30-year bond maturing on June 31, 2018, with a coupon
of 6% paid quarterly that was purchased on January 6, 2012 for 900 would be:
In[
]:= FindRoot
FinancialBond
"FaceValue"  1000, "Coupon"  0.05,
1
"Maturity"  2018, 6, 31, "CouponInterval"  ,
4
"InterestRate"  y, "Settlement"  2012, 1, 6 900, y, .1

Out[
]= y  0.069269

A zero-coupon bond with a redemption value of 1,000 after 10 years is currently sold for
400. Find the implied yield rate compounded semiannually:
384 Mathematica beyond mathematics

In[
]:= FindRoot
FinancialBond

1
"FaceValue"  1000, "Coupon"  0, "Maturity"  10, "CouponInterval"  ,
2
"InterestRate"  y, "Settlement"  0 400, y, .1

Out[
]= y  0.0937605

11.3.3 Derivatives
In[
]:= Clear"Global` "

In Chapter 41 of the book of Genesis, the Egyptian Pharaoh asks Joseph to

interpret a dream he’s had about 7 sleek and fat cows followed by 7 ugly and gaunt
ones. Joseph interprets the dream as periods of plenty followed by periods of
famine and gives the Pharaoh the following piece of advice: “... Let Pharaoh
appoint commissioners over the land to take a fifth of the harvest of Egypt during
the seven years of abundance. They should collect all the food of these good years
that are coming and store up the grain under the authority of Pharaoh, to be kept in
the cities for food. This food should be held in reserve for the country, to be used
during the seven years of famine that will come upon Egypt, so that the country
may not be ruined by the famine”. Some people consider this story as the earliest
mention of the derivative concept based on the idea of fixing in advance the future
price of a certain item. Closer to our current era, on April 3, 1848, the Chicago
Board of Trade was established to create a marketplace for the buying and selling
of several commodities (grains, cattle, hogs, etc.) at fixed prices and following
certain standards. Normally, the buyer would pay in advance a small portion of the
total amount to the seller. The purpose was clear: the buyer could set in advance
how much things were going to cost in the future and the seller could secure a
certain selling price; it could be considered as an insurance against uncertainty.
Since then, this idea has been applied to many other areas: raw materials, shares,
stock indexes, etc.
In finance, a derivative product is one whose price depends on the price of another
product (the underlying asset). We say that the price of the former derives from the
price of the latter. For example, we can create a derivative based on the price of
gold without the need to physically own the gold. In principle, these instruments
should play a beneficial economic role. Very often, however, people associate
them with speculative transactions more similar to gambling than to uncertainty
reduction. There’s a kernel of truth is this opinion since investments in these
products can generate either large losses or large gains. This is the reason why
there have always been investors trying to (and sometimes succeeding) influence
the price of the underlying for speculative purposes. Warren Buffett, the most
successful investor of the 20th century and one of world’s richest men, even
described them as “weapons of mass destruction” although it’s known that some of
his investments take advantage of them, and that if used properly they help to
reduce uncertainty.
When considering derivatives as an alternative to purchasing stocks, the two most
important things to keep in mind are: i) For the same number of shares, derivatives
11. Economic and Financial Applications 385

require a much smaller investment. As a consequence, gains and losses are

significantly higher; ii) when the share price goes up, everybody wins (more
precisely, the latent value is the same for everybody). With derivatives, when
someone wins somebody else loses and the other way around (it’s a zero-sum
game).
An example of a financial derivative is a stock option, or a well-known variant of it
more commonly available to regular investors, a warrant. An option or warrant
gives the owner the right to buy (Call) or sell (Put) the underlying asset at a given
price usually determined at the beginning of the contract. Both have prices
(premiums) that are normally much smaller than the stock associated to them. For
example, we could buy a warrant to purchase Telefónica shares in a year for €15
each and pay €0.5 for it. At expiration, if the shares were trading above €15, let’s
say €16.5, we would make a profit of €1 per share, after subtracting the €0.5
initially paid for it (it’s common to just settle the differences, without actual
delivery of the underlying asset). However, if at expiration Telefónica was trading
below €15, we would just lose the premium paid.
An informative overview of financial derivatives and their statistical measurement
can be found in:
https://www.imf.org/external/pubs/ft/wp/wp9824.pdf.
The list below contains some of the most commonly used terms when describing
derivatives transactions:
Underlying asset: the financial instrument on which a derivative’s price is based.
Call option. An agreement that gives the buyer the right (but not the obligation) to buy the
underlying asset at a certain time and for a certain price from the seller.
Put option. An agreement that gives the buyer the right (but not the obligation) to sell the
underlying asset at a certain time and for a certain price to the seller.
Premium. The amount paid for the right to buy or sell the underlying asset.
Market price. The current price at which the underlying asset can be bought or sold.
Strike. The price at which the agreement can be exercised.
The Greek letters delta, gamma, theta, and vega are used to describe the risk of
changes in the underlying asset.
Delta: It’s the option sensitivity to changes in the underlying asset. It indicates how the
price of the option changes when the underlying price goes up or down. For an option at the
money (strike price same as market price), delta is usually around 50%. This means that if
the price of the underlying asset changes by €2, the price of the option will change by €1.
Gamma: It measures the sensitivity of delta to price changes in the underlying asset. Since
delta is not a linear function, gamma is not linear either. Gamma decreases as the
uncertainty increases.
Vega: This metric indicates how sensitive the option price is to volatility changes in the
underlying asset. It represents the change in the option price if the volatility of the
underlying asset changes by 1%. When volatility goes up, the price of the option also goes
up. Vega may vary depending on whether the option is at the money, in the money or out of
the money.
Theta: A measure of the decrease in the value of an option as a result of the passage of time
(easy to remember since both start with t). The longer the time period until exercise, the
higher Theta is: There’s more time for the underlying asset’s price to move in a direction
favorable for the option holder.
386 Mathematica beyond mathematics

Mathematica gives us a very wide array of options when using

FinancialDerivative. This function returns either the value of the specified financial
instrument (e.g., an American call) , or certain specified properties, such as the
implied volatility or the Greeks.
To make the valuation examples in this section as realistic as possible we will need
certain market data.
Often, to make the purchase of an option cheaper, we can sell another one in such a
way that we limit our potential gains but we also reduce the premium paid. This is
the idea behind zero cost strategies.
An example of this strategy would be as follows: We purchase an option with a
strike price equal to the exercise price (an at the money option) and sell another
one with a higher strike price, let’s suppose 18% higher. Then, this value will be
our maximum return. Obviously, the higher the potential return, the higher the
potential risk and the other way around. This strategy is known as a Call Spread.

In this case we use Telefónica shares trading in the NYSE (the data and functions used in
the example come from Sebastián Miranda from Solventis).
In[
]:= FinancialData"NYSE:TEF", "Name"

Out[
]= Telefónica

We’re going to need the 12-month EURIBOR rate. This type of interest rate is published by
Euribor EBF (http://euribor-rates.eu). Let’s import the 10 most recent rates (remember that
in Chapter 2 we explained how to extract specific parts of a web page) as follows:
In[
]:= Import"http: www.euriborrates.eu euriborrate12months.asp", "Data"
3, 2, 1
4 14 2022, 0.003  4 13 2022,
0.014  4 12 2022, 0.005  4 11 2022,
0.030  4 8 2
Out[
]= 4 1 2022,
0.086  3 1 2022,
0.363  2 1 2022,
0.431  1 3 2022,
0.499  12 1 2
1 3 2022,
0.499  1 4 2021,
0.502  1 2 2020,
0.248  1 2 2019,
0.121  1 2 2

However, unless we’re creating a program to automate the process, probably the easiest
thing to do is to copy the values directly. Besides, it’s unusual for websites to keep their
structure for a long time so, by the time you read these lines, the instruction above may not
work. We’ll store in memory the first rate that we will use for our calculations later on.
In[
]:= euribor12m  0.003 100

Out[
]= 0.00003

For our computations, we need to know several indicators associated to the stock, starting
with its price (Figure 11.1). The values returned are time sensitive and depend on the time
of execution of the query. We show the commands below so readers can replicate the
example.
In[
]:= FinancialData"NYSE:TEF", "Price" stock price,
in USD, in NYSE at the time of the query

Out[
]= $3.26

Figure 11.1 Predictive interface for financial data.

11. Economic and Financial Applications 387

In[
]:= spotPrice  3.36;

Stock volatility during the past 50 days. (Volatility is a measure of the variability of the
stock price over a specific time horizon. It’s usually the standard deviation of the closing
stock price during the last 50 trading sessions, or an alternative chosen time horizon):
In[
]:= FinancialData"NYSE:TEF", "Volatility50Day"

Out[
]= 28.1187

In[
]:= vol  0.3069;

Last year’s dividend yield, computed as the ratio between the dividends paid and the current
stock price.
In[
]:= FinancialData"NYSE:TEF", "DividendYield"

Out[
]= MissingNotAvailable

In[
]:= div  0.0;

Now we can evaluate the price of a call option given its expiration date (in this case, in one
year), and using the current stock price as the strike price:
In[
]:= FinancialDerivative"American", "Call",
"StrikePrice"  spotPrice, "Expiration"  1,
"InterestRate"  euribor12m , "Volatility"  vol,
"CurrentPrice"  spotPrice, "Dividend"  div
Out[
]= 0.399738

In the next graph, we can see the evolution of the value of our call spread as a function of
the underlying asset price and the remaining time. To make it easier to visualize, the solid
green line indicates the current price, our strike price; and the dashed red line, the price of
the underlying asset at which we maximize our profit. We need to keep in mind that our
starting point in the time scale is 1. Remember that our goal is to have a maximum gain of
18%.
In[ ]:= data  , 1  2, FinancialDerivative"American", "Call",
"StrikePrice"  spotPrice, "Expiration"  1,
"InterestRate"  euribor12m , "Volatility"  0.5,
"CurrentPrice"  , "Dividend"  div, "ReferenceTime"  2 
FinancialDerivative"American", "Call",
"StrikePrice"  1.18 spotPrice, "Expiration"  1,
"InterestRate"  euribor12m , "Volatility"  0.5,
"CurrentPrice"  , "Dividend"  div, "ReferenceTime"  2 & 
FlattenTableprice, referencetime, referencetime, 0, 1, 0.05,
price, 0.8 spotPrice, 1.4 spotPrice, 1, 1;
388 Mathematica beyond mathematics

In[
]:= ListPlot3Ddata,

Contours
Mesh  spotPrice, Green, Thick,
1.18 spotPrice, Red, Thick, Dashed,
Formats
AxesLabel 
Style, 12 &  "Underlying Price", "Remaining Time", "Profit" ,
ColorFunction  ColorData"LightTemperatureMap",
ColorFunctionScaling  True

Out[
]=

At expiration, the profile of our profit as a function of the underlying asset price is:
In[
]:= ListPlotCasesdata, x, 0., y  x, y,
Joined  True, PlotStyle  DirectiveThickness0.02, Red,
AxesLabel  "Underlying Price", "Profit"
Profit

0.6

0.5

0.4

Out[
]=
0.3

0.2

0.1

Underlying Price
3.0 3.5 4.0 4.5

In this example, the warrants that we’ve used are of the American type, meaning
that they can be exercised at any time until the expiration date, although it is not
usually optimum to do so.
However, Mathematica, in addition to financial derivatives price calculations, can
also compute their sensitivities, known as Greeks. These sensitivities are the partial
derivatives of the option price with respect to different valuation parameters. For
example, the delta of an option is the partial derivative of its price with respect to
the price of the underlying asset. This measure not only gives us an idea of how
11. Economic and Financial Applications 389

sensitive the option value is to small changes in the price of the underlying asset
but also, for standard options (also known as vanilla), it gives us the probability
that the option will be exercised. Gamma, the second derivative with respect to the
underlying asset price, tells us how sensitive the delta is to changes to it. This
metric is very important for hedging purposes. Rho is the partial derivative with
respect to the interest rate and in the case of stock options is the least sensitive
Greek. Theta is the derivative with respect to time to expiration. Initially, since
options are not usually bought in the money i.e. if a ticker trades at $12, purchasing
an option for $10 is not very common, they don’t have intrinsic value. They only
get value from their time component. The final one, Vega, is the partial derivative
with respect to the volatility of the underlying asset. This is such an important
parameter in option valuation that for certain options, their prices are quoted in
terms of the volatility of their respective underlying assets.

The graph below shows how the Greeks change with respect to the remaining time to
expiration:
In[
]:= ticksmin, max : Table
j, ToStringRound12  j 12  " Months", .03, 0, j, min, max, 0.2
In[
]:= TabViewRule  Transpose"Delta", "Gamma", "Rho", "Theta", "Vega",
ListPlot, Ticks  ticks, Automatic &  TransposeTablei,  & 
FinancialDerivative"American", "Call", "StrikePrice"  55.00,
"Expiration"  1, "InterestRate"  0.1, "Volatility"  0.5,
"CurrentPrice"  50, "Dividend"  0.05, "ReferenceTime"  i,
"Greeks" All, 2 , i, 0, 0.95, 0.05

Delta Gamma Rho Theta Vega

0.5

0.4

Out[
]=
0.3

0.2

0.1

10 Months 7 Months 5 Months 3 Months

The same Greek letters are used to indicate the risks of portfolio of options due to
changes in the price of the underlying assets or market conditions.
Delta: the delta of a portfolio of options with the same underlying assets is the
weighted average of the delta of the individual options.
The delta factor = the change in the price of the option as a result of changes in the
price of the underlying assets. For an at-the-money option, the delta factor is
390 Mathematica beyond mathematics

usually close to 50%, so if the underlying price changes by 1, the price of the
option will change by 0.5.

11.3.4 The Black–Scholes Equation

In this example, we first compute the price of a derivative using Mathematica’s
existing function and then compare it with the result obtained by solving the Black–
Scholes equation directly
http://www.wolfram.com/language/11/partial-differential-equations/find-the-value-of-a-euro-
pean-call-option.html

Let’s start by using FinancialDerivative to compute the price of a European vanilla call
option with the data from the previous section:
In[
]:= spotPrice  3.26; euribor12m  0.00003; vol  0.3069;

In[
]:= FinancialDerivative"European", "Call", "StrikePrice"  spotPrice,

"Expiration"  1, "InterestRate"  euribor12m, "Volatility"  vol,
"CurrentPrice"  spotPrice
Out[
]= 0.397621

Now, let’s do the same computation using the Black–Scholes equation (To create more
visually appealing documents you can convert the input cell to the traditional format:
Cell  Convert to  TraditionalForm):
In[
]:= BlackScholesModel

  ,  1 2  ,    , 
2

  ,     2   0,  ,   max  , 0;
 2 2 


After solving the boundary value problem we get:
In[
]:= dsol  ct, s .
DSolveBlackScholesModel, ct, s, t, s
1 TraditionalForm
Out[
]//TraditionalForm=

1 2 logk   2 r  2  t
T 
2 logs

r T s r T erfc

2 2 2 T
t
2 logk   2 r
2  t
T 
2 logs
k r t erfc
2 2 T
t

Finally, we calculate the price of the derivative and compare it with the one previously
obtained. We can see it’s the same:
In[
]:= dsol . t  0, s  spotPrice, k  spotPrice,   vol, T  1,
r  euribor12m
Out[
]= 0.397621

11.3.5. Basel II Capital Adequacy: Internal Ratings-Based (IRB) Approach

The key cornerstone of prudential regulation of banks (and financial institutions in
general) is to ensure that each bank holds sufficient equity capital to absorb
unexpected losses, that is, the materialization of financial risks, mainly market
(price), credit, and operational risks.
The example below, Basel II Capital Adequacy: Internal Ratings-Based (IRB)
Approach, by Poomjai Nacaskul, shows how to calculate, using the Internal
11. Economic and Financial Applications 391

Ratings-Based approach, the minimum amount of capital that a bank should keep
following the standards defined by the Basel II regulatory framework on bank
capital adequacy.
http://demonstrations.wolfram.com/BaselIICapitalAdequacyInternalRatingsBasedI
RBApproach

To facilitate the understanding of the example, we also include the code:
In[
]:= Setting defaults
DefaultEAD  1 000 000;
MinimumPD  0.0001; MaximumPD  0.25;
MinimumLGD  0.01; DefaultLGD  0.7; MaximumLGD  1.0;
MinimumMaturity  1 250;
DefaultMaturity  1;
MaximumMaturity  30;
In[
]:= Create shorthands for Normal CDF & CDF inverse
Phix : CDFNormalDistribution0, 1, x;
PhiInversep : InverseCDFNormalDistribution0, 1, p;
In[
]:= Correlation Function, Basel Committee's way of parameterising a 2
parameter formula PD, rho with 1 parameter PD .
RPD : 0.12 1  Exp50 PD 1  Exp50

0.24 1  1  Exp50 PD 1  Exp50 ;
ConditionalPercentageExpectedLossPD, LGD : Phi
PhiInversePD
SqrtRPD PhiInverse0.999 Sqrt1  RPD LGD
UnconditionalPercentageExpectedLossPD, LGD : PD LGD;
In[
]:= Maturity Adjustment, with a 1year maturity as a base case
bPD : 0.11852  0.05478 LogPD ^ 2;
MaturityAdjustmentMaturity, PD :
1
Maturity  2.5 bPD 1  1.5 bPD ;
In[
]:= Capital Requirement Calculation: note as PD  1,
this quantity  0, as credit risk is taken care of via provisioning,
i.e. ELoss  exposure x PD x LGD.
CapitalRequirementPD, LGD, Maturity :
ConditionalPercentageExpectedLossPD, LGD 
UnconditionalPercentageExpectedLossPD, LGD
MaturityAdjustmentMaturity, PD;
392 Mathematica beyond mathematics

In[
]:= Manipulate

PaneColumnText  Style"Basel II Capital Adequacy", Bold, 16,
Text  "Regulatory minimal capital for financial institutions
per $ exposure,\nas per Internal Ratings-Based
Approach IRB Risk Weight Function RWF \n",
GridPlotCapitalRequirementPD, LGD, Maturity, PD, 0.0001, 0.25,
PlotLabel  "as a function of Probability of Default PD \nat
a given Maturity M ", ImageSize  320,
Plot3DCapitalRequirementPD, LGD, Maturity, PD, 0.0001, 0.25,
Maturity, 1 250, 30, PlotLabel  "as a function of PD and M\n",
ImageSize  280, Mesh  None, ColorFunction 
Functionx, y, z, ColorData"GrayTones"z,
Alignment  Center, ImageSize  610, 350,
LGD, 0.7, "Loss Given Default LGD ",
0.01, 1, Appearance  "Labeled",
Maturity, 1, "Maturity M in years left graph only ",
1 250, 1 50, 1 12, 1 4, 1, 2, 5, 10, 15, 30,
ControlType  RadioButton, SaveDefinitions  True

To learn more about financial modeling and risk management using Mathematica,
Igor Hlivka has created several excellent example notebooks. You can find them
in:
http://community.wolfram.com/web/ihcomm.

11.4 Optimization

11.4.1 What is Constrained Optimization?

The optimization problems that we discuss in this section are related to finding an
optimum (maximum or minimum, depending on the case) for an objective function
11. Economic and Financial Applications 393

of n variables, given r constraints.

This type of problems is common in several fields such as economics, finance, and
engineering as we will show using different examples. You can find an excellent
tutorial in the documentation pages: “Introduction to Constrained Optimization in
the Wolfram Language” (tutorial/ConstrainedOptimizationIntroduction).
To solve them, Mathematica has the following functions (Figure 11.2):
Function Application Description

FindMinimum  FindMaximum Local Linear programming methods, non  linear,

Numerical optimization interior points, and the use of second derivatives

NMinimize  NMaximize Global Linear programming methods, Nelder  Mead,

Numerical optimization
   ,  
 ,
random search

Minimize  Maximize Exact Linear programming methods, algebraic

Global optimization cylindrical decomposition, Lagrange
multipliers and other analytical methods,
integer linear programming.

LinearProgramming Linear Simplex, modified simplex, interior point, ...

LinearOptimization optimization

Figure 11.2 Mathematica functions for optimization.

The following tree can help us decide the most appropriate function to find the
minimum. (To find the maximum we would use the functions Maximize,
NMaximize or FindMaximum instead). Note: Since Mathematica 12,
LinearProgramming has been superseded by LinearOptimization, but the command
can still be used.
Is your problem linear?

Yes No

LinearProgramming or
L Do you want a global optimum?

Out[
]=
Yes No

Do you want an exact solution? Is your problem small?

Yes No Yes No

Minimize NMinimize FindMinimum

In many cases, the problem can be solved using several functions. With the
exception of LinearProgramming and LinearOptimization, the syntax for the rest of
the functions is very similar: f[{objective function, constraints}, {variables}].
In[
]:= Clear"`Global` "
394 Mathematica beyond mathematics

Let’s see some examples.

Given the objective function (ob) with constraints c1 and c2:
In[
]:= var  x, y; ob  x
y; c1  0  x  1; c2  0  y  2;

The maximum can be found using any of the previously mentioned commands:
In[
]:= NMaximizeob,c1,c2, var, Maximizeob,c1,c2, var,
FindMaximumob,c1,c2, var
3. x  1., y  2.
Out[
]= 3 x  1, y  2
3. x  1., y  2.

We can see that the maximum occurs when {x  1, y  2} and is 3. We can verify it
(remember that “/.” is used for substitutions).
In[
]:= x
y . x  1, y  2

Out[
]= 3

The minimum can be computed as follows:

In[
]:= Minimizeob,c1,c2, var

Out[
]= 0, x  0, y  0

We can interpret the problem graphically by drawing the plane x  y only in the region:
{0 x 1, 0 y 2}.
In[
]:= Plot3Dx
y, x, 1, 2, y, 1, 2,
RegionFunction  Functionx, y, 0  x  1 && 0  y  2,
AxesLabel  Automatic, Mesh  None

Out[
]=

The graph clearly shows that the maximum is in (1,2) and the minimum in (0,0). Use the
mouse to move the graph around to see it from different perspectives.
In certain situations, we’d like the variables to take only integer values. To add this extra
requirement in Mathematica, we use the built-in symbol Integers.

Let’s consider the following model in which we want to minimize the objective function
(ob) (notice that equality when typing constraints is "=="):
In[
]:= Clear"`Global` "
11. Economic and Financial Applications 395

In[
]:= var  x, y;

ob  x
2 y;
c1   5 x
y 7;
c2  x
y  26;
c3  x  3;
c4  y  3;
In[
]:= NMinimizeob, c1, c2, c3, c4, var  Integers

Out[
]= 58., x  4, y  27

Since this is a linear problem, we can also use either LinearOptimization (since
Mathematica 12) or LinearProgramming. As a matter of fact, these functions are
the most appropriate ones for linear problems, especially if the number of variables
is large.
In[
]:= LinearOptimizationob, c1, c2, c3, c4, x  Integers, y  Integers

Out[
]= x  4, y  27

The syntax is:
   [c, m, b] . This function finds the vector x
that minimizes the quantity c.x subject to the constraints m.x b and x 0. We
can limit the values that the variables (some or all of them) can take to just integers
with
   [..., Integers].

For comparison purposes, let’s solve the same problem using
   .
The syntax is:
   [c, m, b] . This function finds the vector x that
minimizes the quantity c.x subject to the constraints m.x b and x 0. We can limit the
values that the variables (some or all of them) can take to just integers with

   [..., Integers].
x  2y  c : 1, 2

5 x  y 7  m1 : 
5, 1 b1 : 7, 0
x  y 26  m2 : 1, 1 b2 : 26, 1
Notice that the syntax to indicate the type of constraint is as follows: {bi , 0} if mi .x  bi ;
{bi , 1} if mi .x bi and {bi ,
1} if mi .x bi .
In[
]:= LinearProgramming1, 2, 5, 1, 1, 1, 7, 0, 26, 1,
3, Infinity, 4, Infinity, Integers Quiet
Out[
]= 4, 27

Next we’ll show some examples for nonlinear optimization.

In[
]:= Clear"`Global` "

In this problem, both, the objective function that we want to minimize and the constraints
are nonlinear.
In[
]:= var  x, y;
ob  x ^ 2
y  1 ^ 2;
c1  x ^ 2
y ^ 2  4;

Let’s calculate the global minimum with Minimize:
In[
]:= Minimizeob, c1, var

Out[
]= 0, x  0, y  1

396 Mathematica beyond mathematics

NMinimize can solve this type of problem faster (although in this example we wouldn’t be
able notice any difference). We add Chop to remove, when the solution is 0, the spurious
values that may appear in the output when using numerical approximations. Execute the
command below without it and see what happens.
In[
]:= NMinimizeob, c1, var Chop

Out[
]=
3.45349  10
8 , x  0, y  0.999999

If we are looking for a local minimum we can use FindMinimum. Although it’s not
necessary, this function is much faster if we enter an initial point from which to start the
search. It’s also a good idea to add Chop.
In[
]:= FindMinimumob, c1, x, 0, y, 0 Chop

8
Out[
]=
3.3699  10 , x  0, y  0.999999

We present the previous problem graphically with Plot3D and ContourPlot, very
convenient functions when trying to find optimum points visually. (The option
Contours specifies how many contours we want to display; if not specified, Mathematica
will choose the optimum number for the given function).
In[
]:= GraphicsRowPlot3D x ^ 2
y  1 ^ 2, x, 3, 3,
y, 3, 3, RegionFunction  Functionx, y, x ^ 2
y ^ 2  4,
AxesLabel  Automatic, Mesh  All, ContourPlotx ^ 2
y  1 ^ 2,
x, 3, 3, y, 3, 3, RegionFunction  Functionx, y, x ^ 2
y ^ 2  4,
Contours  100, FrameLabel  Automatic

Click inside the output from Plot3D and move the graph around to see the minimum from
different angles. In the contour plot, we can see that the minimum corresponds to the
coordinates (0, 1). By default, dark colors are associated to small values and light hues are
linked to bigger ones. If you place your cursor over the contour limits, you’ll see the value
that the function takes at that location. (All the points connected by the same line have the
same function value.)

Here is another example of nonlinear optimization, in this case for finding the global
minimum (Minimize). The objective function is Exp
x y with the constraint that x, y  a
Circle centered in {0, 0} and with a radius r 1. We show the result, both numerically and
graphically.
In[
]:= m  MinimizeExpx y, x, y  Disk, x, y
1 1 1
Out[
]=  , x 
, y

 2 2
11. Economic and Financial Applications 397

In[
]:= g  Plot3DExpx y, x, y  Disk, Axes  True,

Boxed  False, PlotStyle  Opacity  0.6, AxesLabel  Automatic;
In[
]:= Showg, Graphics3DPointSize  Large, Red, Pointx, y, m 1  . m 2 

Out[
]=

The following example shows how the optimization algorithm approaches the optimum
(minimum) point for f(x,y) in an iterative way.
In[ ]:= Clear"Global` "

In[
]:= fx, y : x  1 ^2
2 y  x^2 ^2

In[
]:= Modulepts, pts  LastLastReapSow1.2, 1, f1.2, 1;

FindMinimumfx, y,
x, 1.2, y, 1, StepMonitor  Sowx, y, fx, y;
ShowPlot3Dfx, y, x, 1.4, 1.4, y, 0.6, 2, Mesh  5,
PlotStyle  Automatic, PlotLabel  StyleRow"Finding the minimum of ",
fx, y, " using iteration", 12, Bold, AxesLabel  Automatic,
Graphics3DOrange, PointSize0.025, PointRestMostpts,
Thick, Linepts, Darker  Green, PointSize0.05,
PointLastpts, Red, PointFirstpts

Out[
]=

11.4.2 Application: Supplying Construction Sites

We’d like to optimize the supply of bricks from warehouses to construction sites.
There are two brick warehouses j 1,2}, where we have in stock 20 and 25
398 Mathematica beyond mathematics

tonnes, respectively. From them we supply 6 sites i 1, 2, 3, 4, 5, 6 with a daily

demand in tonnes of: {3, 5, 4, 7, 6, 11}. We assume that the location (latitude and
longitude) of the warehouses and sites is:
In[
]:= warehouses  7, 1, 2, 9;

In[
]:= sites  1.25, 1.25, 8.75, 0.75,

0.5, 4.75, 5.75, 5, 3, 6.5, 7.25, 7.75;

We’d like to minimize the total amount of bricks transported per unit of distance,
i,j xi,j di,j , meeting the demand without exceeding the stock available in the
warehouses.
For that purpose, we have to calculate the minimum of i,j xi,j di,j , where xi,j
corresponds to the quantities transported from the warehouses i:{1, 2} to the sites
j:{1, ..., 6} with di,j representing the distance between warehouse i and site j.

To calculate the distances between the warehouses and the sites we use the function
EuclideanDistance (in the bi-dimensional case, it’s the equivalent to the Pythagorean
theorem):
In[
]:= EuclideanDistancex1, y1, x2, y2

Out[
]= x1
x22  y1
y22

The command below returns triplets, {i, j, di,j }, indicating the distance in km between the
warehouse i and site j. We use Roundvalue, 0.1 to round the distance to the nearest
first decimal.
In[
]:= Tablei, j, RoundEuclideanDistancewarehouses i , sites j , 0.1,
i, 1, 2, j, 1, 6
1, 1, 5.8 1, 2, 1.8 1, 3, 7.5 1, 4, 4.2 1, 5, 6.8 1, 6, 6.8
Out[
]=
2, 1, 7.8 2, 2, 10.7 2, 3, 4.5 2, 4, 5.5 2, 5, 2.7 2, 6, 5.4

Therefore, the variables and the objective function, i,j xi,j di,j , are:
In[
]:= var  x11 , x12 , x13 , x14 , x15 , x16 , x21 , x22 , x23 , x24 , x25 , x26 ;

In[
]:= of  5.8 x11
1.8 x12
7.5 x13
4.2 x14
6.8 x15

6.8 x16
7.8 x21
10.7 x22
4.5 x23
5.5 x24
2.7 x25
5.4 x26 ;

The constraints are: i xi,j = demandj , j xi,j = stocki and xi,j 0
In[
]:= constraints  x11
x21 3 && x12
x22 5 && x13
x23 4 && x14
x24 7 &&
x15
x25 6 && x16
x26 11 && x11
x12
x13
x14
x15
x16  20 &&
x21
x22
x23
x24
x25
x26  25 && x11  0 && x12  0 && x13  0 && x14  0,
x15  0 && x16  0 && x21  0 && x22  0 && x23  0 && x24  0 && x25  0 && x26  0;

The solution is:
In[
]:= NMinimizeof, constraints, var 

Out[
]= 149.4, x11  3., x12  5., x13  0., x14  7., x15  0.,
x16  0., x21  0., x22  0., x23  4., x24  0., x25  6., x26  11.
This result indicates that the optimum solution, 149.4 tonnes  km, is obtained by
transporting 3 tonnes from warehouse 1 to site 1, 5 tonnes from warehouse 1 to site 2 and so
on.
In “Optimal Transport Scheduling”, a demonstration by Yifan Hu, you can see a dynamic
visualization of the resolution of this problem:
11. Economic and Financial Applications 399

http://demonstrations.wolfram.com/OptimalTransportScheduling
If in the demonstration you use 20 and 25 tonnes, you’ll see that the solution is the same
(there’s actually a small difference due to rounding).

11.4.3 Application: Portfolio Optimization

A bank commissions a consulting firm to analyze the most profitable way to invest
10,000,000 for two years given the options in Figure 11.3. The table includes the
expected rate of return (bi-annual) and associated risk.
Product i Return , bi Risk ri
Mortgages 9 3
Mutual funds 12 6
Personal loans 15 8
Commercial loans 8 2
Certificates  Bonds 6 1
Figure 11.3 Investment options.

The capital not invested in any of the products will be placed in government bonds
(assumed to be riskless) with a bi-annual rate of return of 3%. The objective of the
consulting firm is to allocate the capital to each of the products to meet the
following goals:
(a) Maximize the return per  invested.
(b) Keep the possibility of loss to a maximum of 5% of the total amount invested.
(c) Invest at least 20% in commercial loans.
(d) Allocate to mutual funds and personal loans an amount no larger than the one
invested in mortgages.

Variables: xi is the percentage of capital invested in product i. The amount placed in
government bonds can be considered as a new product with an expected return, in
percentage, b6 = 3 and risk r6 = 0. Therefore:
In[ ]:= var  x1 Mortgages , x2 Mutual funds ,
x3 Personal loans , x4 Commercial loans , x5
Certificates Bonds , x6 Government debt ;

Objective function to maximize: 6i 1 bi xi .
In[
]:= of  9 x1
12 x2
15 x3
8 x4
6 x5
3 x6 ;

Constraint 1: All the money is invested: 6i 1 xi = 1.

In[
]:= c1  x1
x2
x3
x4
x5
x6 1;

Constraint 2: The average risk is R = 6i 1 ri xi / 6i 1 xi 5 as r6 = 0 then:

3 x 1
6 x 2
8 x 3
2 x 4
x5
In[
]:= c2   5
x1
x2
x3
x4
x5
3 x1  6 x2  8 x3  2 x4  x5
Out[
]= 5
x1  x2  x3  x4  x5

Constraint 3: At least 20% of the capital has to be invested in commercial loans (x 4 0.2).
In[
]:= c3  x4  0.2;
400 Mathematica beyond mathematics

Constraint 4: The percentage invested in mutual funds x2 ) and personal loans (x3 ) cannot
be bigger than the one invested in mortgages x1 ).
In[
]:= c4  x2
x3  x1 ;

Constraint 5: No percentage invested can be negative (xi 0).

In[
]:= c5  Map  0 &, var;

Using NMaximize, Mathematica tells us that we’d get a return of 11.2% investing 40% in
mortgages, 40% in personal loans, and 20% in commercial loans.
In[
]:= sol  NMaximizeof, c1, c2, c3, c4, c5, var

Out[
]= 11.2, x1  0.4, x2  0., x3  0.4, x4  0.2, x5  0., x6  0.

In[
]:= ""  ToString  AccountingForm10 000 000 sol 1 100, DigitBlock  3

Out[ ]= 1,120,000.

11.4.4 Application: Calculating Cod Consumption

In[
]:= Clear"Global` "

Many problems in economics involve Cobb–Douglas functions: F x, y = A xa yb

with x, y 0 and A, a, b constants. Let’s see one example:
The cod consumption in certain region follows the function cp, i = 1.3 p
1.5 i0.01 ,
where p is the price per kg and i is the daily income in /day. The previous
expression is valid for daily incomes between 20 and 200 per day and the price
per kg varies between 4 and 12. What would be the maximum and minimum
consumption?

We can proceed as usual defining the variables, objective function, and constraints:
In[
]:= cp, i  1.3 p1.05 i0.01 ;

In[
]:= NMinimizecp, i, 4  p  12 && 20  i  200, p, i

Out[
]= 0.0985856, p  12., i  20.

In[
]:= NMaximizecp, i, 0.4  p  1.2 && 20  i  200, p, i

Out[
]= 3.58749, p  0.4, i  200.

We can verify the solution visually using Plot3D and ContourPlot. If we place the cursor on
the second graph and move it around, we can see the different values of the function c r, t.
This way we can check that the upper-left corner corresponds to the largest value
(maximum), and the lower-right corner to the smallest one (minimum). The boundaries of
the curve correspond to the constraints.
In[
]:= GraphicsRowPlot3Dcp, i, p, 4, 12, i, 0, 200,
RegionFunction  Functionp, i, 4  p  12 && 20  i  200,
AxesLabel  "Price", "Income", "c", Mesh  None,
ContourPlotcp, i, p, 4, 12, i, 0, 200, RegionFunction 
Functionp, i, 4  p  12 && 20  i  200, Mesh  None,
Contours  100, FrameLabel  Automatic, PlotLegends  Automatic,
Frame  All, AspectRatio  1 3, ImageSize  600, 300
11. Economic and Financial Applications 401

11.4.5 Transportation Optimization

The department store chain Carrefa would like to outsource the supply of milk
cartons for its centers in Barcelona, Madrid, Seville, and Valladolid. For that
purpose, it receives offers from three suppliers: Pascualo, Simone, and Pulova. The
prices per carton are as follows: Pascualo 3.00, Simone 2.80, and Pulova 2.70.
The maximum number of cartons that each supplier can send daily is shown in the
table below, under the column “Maximum Supply”. The daily demand for each
center is in the “Demand” row. The rest of the information in Figure 11.4 relates to
the transportation costs per carton, in .
Maximum Supply Barcelona Madrid Seville Valladolid
Pascualo 2500 0.5 0.6 0.7 0.5
Simone 1700 0.5 0.6 0.8 0.5
Pulova 1500 0.4 0.5 0.9 0.7
Demand 800 1100 500 300
Figure 11.4 Transportation optimization input data.

Carrefa would like to know how much to order from each supplier to minimize the
total cost.
In[ ]:= Clear"Global` "

This is the type of problem that can be solved with LinearProgramming or the
newest function LinearOptimization. With LinearProgramming, matrix notation is
required. For didactic purposes we’re going to compare both functions using
algebraic notation for LinearOptimization although matrix notation can also be
used.
With LinearOptimization the variables can be defined as follows: xij = number of
cartons from supplier i , i:{1, 2, 3}, with {1 = Pascualo, 2 = Simone, 3 = Pulova},
to center j: {1, 2, 3, 4} , with {1 = Barcelona, 2 = Madrid, 3 = Seville, 4 =
Valladolid}.
In[
]:= var  x11 , x12 , x13 , x14 , x21 , x22 , x23 , x24 , x31 , x32 , x33 , x34 ;

To define the objective function, we have to group terms by supplier. For example: the total
cost of sending x units from Pascualo to Barcelona (x11 ) would be the transportation cost
plus the purchasing price: (0.5 + pa). We use the same approach for calculating the rest of
the variables:
In[
]:= pa, sim, pul  3, 2.8, 2.7;
402 Mathematica beyond mathematics

In[
]:= of   0.5
pa , 0.6
pa , 0.7
pa , 0.5
pa ,

0.5
sim , 0.6
sim , 0.8
sim , 0.5
sim ,
0.4
pul , 0.5
pul , 0.9
pul , 0.7
pul ;

Or in algebraic notation:
In[
]:= of1  of .var

Out[
]= 3.5 x11  3.6 x12  3.7 x13  3.5 x14  3.3 x21 
3.4 x22  3.6 x23  3.3 x24  3.1 x31  3.2 x32  3.6 x33  3.4 x34

The constraint c1 representing the maximum number of cartons of milk that we can receive
from Pascualo is written as (the values of those variables not related to the constraint should
be 0), m1 and bi:
In[
]:= c1  x11
x12
x13
x14  2500;

In[
]:= m1  1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0; b1  2500, 1;

where
1 is equivalent to " "; 0 to "=="; and 1 to " ".

The constraint c2 or m2, related to the maximum capacity of Simone is:
In[
]:= c2  x21
x22
x23
x24  1700;

In[
]:= m2  0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0; b2  1700, 1;

The constraint c3 or m3, dealing with the supply coming from Pulova:
In[
]:= c3  x31
x32
x33
x34  1500;

In[
]:= m3  0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1; b3  1500, 1;

The constraint c4, x11 x21 x31  800, indicating the demand from Barcelona:
In[
]:= c4  x11
x21
x31 800;

In[
]:= m4  1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0; b4  800, 0;

The constraint c5, x12 x22 x32  1100, referring to the demand from Madrid:
In[
]:= m5  0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0; b5  1100, 0;

In[
]:= c5  x12
x22
x32 1100;

The constraint c6, x13 x23 x33  500, about Seville:

In[
]:= c6  x13
x23
x33 500;

In[
]:= m6  0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0; b6  500, 0;

And finally the constraint c7, or m3 to make sure that the needs of Valladolid are being
satisfied:
In[
]:= c7  x14
x24
x34 300;

In[
]:= m7  0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1; b7  300, 0;

Now we can solve the problem (LinearProgramming assumes that all the variables are
nonnegative so we don’t have to include those constraints).
In[
]:= sol1  LinearProgrammingof,
m1, m2, m3, m4, m5, m6, m7, b1, b2, b3, b4, b5, b6, b7
Out[
]= 0., 0., 0., 0., 400., 0., 500., 300., 400., 1100., 0., 0.

Using LinearOptimization the constraint that all the variables are nonnegative should be
included.
In[
]:= c8  Threadvar  0
11. Economic and Financial Applications 403

Out[
]= x11 0, x12 0, x13 0, x14 0, x21 0, x22 0, x23 0, x24 0, x31 0, x32 0, x33 0, x34 0

In[
]:= sol2  LinearOptimizationof1, c1, c2, c3, c4, c5, c6, c7, c8, var

Out[
]= x11  0., x12  0., x13  0., x14  0., x21  0., x22  400.,
x23  500., x24  300., x31  800., x32  700., x33  0., x34  0.

Both commands return the optimal value for each variable. If we’d like to calculate the
minimum total transportation cost, we just need to multiply the objective function by the
solution (we use "." for matrix multiplication). Although sol1 and sol2 are different , the
optimization values are the same; this happens when the problem has more than one
solution.
In[
]:= of.sol1

Out[
]= 8870.

In[
]:= of1 . sol2

Out[
]= 8870.

11.5 The Shortest Path Problem

11.5.1 The Traveling Salesman Problem

There are problems in mathematics that look very specialized and easy to solve at
first sight, but in reality they are very complicated and their resolution can have
applications in a wide variety of fields. One of the most famous examples is the
Traveling Salesman Problem (TSP). This problem can be defined as follows: A
traveler would like to visit N cities starting and finishing in the same arbitrarily
chosen one, minimizing the total distance traveled and passing through each city
only once. If you think about it, there are many real-world problems that are
actually equivalent to the TSP: route optimization (transportation, logistics),
optimum network layout (trains, roads, electricity), even the connections inside
microprocessors to minimize calculation time.
A graphical demonstration of the problem created by Jon McLoone can be found
in:
http://demonstrations.wolfram.com/TravelingSalesmanProblem/
Mathematically, the problem consists of finding a permutation P = {c0 , c1 ,..., cn
1 }
such that dP = ni
11 d ci , ci1 would be a minimum, with dij representing the
distance from city i to city j.
The seemingly simplest solution would be to enumerate all the possible routes,
calculate the distance traveled for each route and select the shortest one. The
problem with this approach is that computing time increases dramatically when the
number of cities gets bigger. For example, a computer would take 3 seconds to
solve the problem if there were 10 cities, a bit more than half a minute for 11 cities,
and almost 80,000 years! to solve the problem for 20 cities.
Conclusion: This solving method is not feasible. Therefore, several alternative
algorithms have been developed that use a variety of strategies to simplify the
problem (something similar to computer chess programs that instead of calculating
404 Mathematica beyond mathematics

all possible moves, use different criteria to choose the most appropriate ones).
For this type of problem, Mathematica has the function:
FindShortestTour[{e1, e2, ...}]. This function takes a list of coordinates: {e1,
e2, ...} and attempts to find the optimum ordering so that the total distance traveled
is minimized and all the points have been visited just once. The output returns
both, the total distance and the order of traveling.
Example: Given a set of points p = {1, 2, 3, 4, 5, 6} each with
coordinates{x,y}:{{4, 3},{1, 1}, {2, 3},{3, -5},{-1, 2},{3, 4}}, find the shortest
path that will visit all the points only once.
In[
]:= d  4, 3, 1, 1, 2, 3, 3, 5, 1, 2, 3, 4;

In[
]:= dist, order  FindShortestTourd

Out[
]= 2 2  5 3 10  65 , 1, 4, 2, 5, 3, 6, 1

We sort the points following the order from the output above and represent them
graphically:
In[
]:= d order
4 3
3
5
1 1
Out[
]=
1 2
2 3
3 4
4 3
In[
]:= GraphicsLine 

Out[
]=

11.5.2 A Tour Around South American Cities

The next example is about organizing a trip to several cities in South America. In
what order should we visit them to minimize the total distance traveled?

The cities we’re visiting are the following (we indicate both, the cities and their respective
regions and countries to avoid ambiguity):
11. Economic and Financial Applications 405

In[
]:= citiesEntity"City", "Asuncion", "Asuncion", "Paraguay",

Entity"City", "Bogota", "DistritoCapital", "Colombia",
Entity"City", "RioDeJaneiro", "RioDeJaneiro", "Brazil",
Entity"City", "BuenosAires", "BuenosAires", "Argentina",
Entity"City", "Caracas", "DistritoCapital", "Venezuela",
Entity"City", "LaPaz", "LaPaz", "Bolivia",
Entity"City", "Lima", "Lima", "Peru",
Entity"City", "Montevideo", "Montevideo", "Uruguay",
Entity"City", "Quito", "Pichincha", "Ecuador",
Entity"City", "Santiago", "Metropolitana", "Chile";

Now we can calculate the best visiting order:
In[
]:= order  LastFindShortestTourGeoPositioncities

Out[
]= 1, 8, 4, 10, 6, 7, 9, 2, 5, 3, 1

Finally, we can display the route on a map:

In[
]:= GeoListPlotcities order , Joined  True

11.5.3 Moon Tourism

It’s the year 2035, and we are planning to organize sightseeing tours around the
Moon. We want to find the shortest route for visiting all the Apollo landing sites.

“Apollo landings” is one of the available entity classes that have been included in the
function MannedSpaceMissionData .
406 Mathematica beyond mathematics

In[
]:= MannedSpaceMissionData"Classes"

Out[
]= manned space missions , Apollo landings , Apollo crewed mission ,

Chinese crewed mission , Earth orbit crewed mission , lunar crewed mission ,
manned space mission , Mercury Atlas crewed mission , Mercury Redstone crewed mission ,
pre-flight test crewed mission , project Gemini crewed mission ,
project Mercury crewed mission , Russian crewed mission ,
Shenzhou crewed mission , Soyuz crewed mission , SpaceX Dragon crewed mission ,
SpaceX crewed mission , STS crewed mission , sub-orbital flight crewed mission ,
United States crewed mission , Voskhod crewed mission , Vostok crewed mission

In[
]:= MannedSpaceMissionData
Apollo manned mission MANNED SPACE MISSIONS 

Out[
]= Apollo 1 , Apollo 7 , Apollo 8 , Apollo 9 , Apollo 10 , Apollo 11 ,

Apollo 12 , Apollo 13 , Apollo 14 , Apollo 15 , Apollo 16 , Apollo 17

In[ ]:= apollolandings 

MannedSpaceMissionData
Apollo landings MANNED SPACE MISSIONS ,

"Position", "EntityAssociation"

Out[
]=  Apollo 11  GeoPosition0.6741, 23.47, Moon  ,

Apollo 12  GeoPosition
3.012,
23.42, Moon  ,

Apollo 14  GeoPosition
3.645,
17.47, Moon  ,
Apollo 15  GeoPosition26.13, 3.634, Moon  ,
Apollo 16  GeoPosition
8.973, 15.50, Moon  ,
Apollo 17  GeoPosition20.19, 30.77, Moon  

The names of the manned landing missions are:

In[
]:= missionnames  Keysapollolandings

Out[
]= Apollo 11 , Apollo 12 , Apollo 14 , Apollo 15 , Apollo 16 , Apollo 17

Notice that Apollo 13 is missing. Let’s find out why:

In[
]:= MannedSpaceMissionData
Apollo 13 MANNED SPACE MISSION , "Description"

Out[
]= intended to be the third manned mission to land on the moon,

mid
mission onboard explosion forced landing to be aborted, returned safely to Earth

Create a map showing the locations of the landings:
In[
]:= GeoListPlotmissionnames, GeoRange  All,
GeoProjection  "Orthographic", GeoLabels  True, LabelStyle  White
11. Economic and Financial Applications 407

To extract the coordinates of those landings:

In[
]:= landingcoords  Valuesapollolandings Flatten

Out[
]= GeoPosition0.6741, 23.47, Moon , GeoPosition
3.012,
23.42, Moon ,

GeoPosition
3.645,
17.47, Moon , GeoPosition26.13, 3.634, Moon ,
GeoPosition
8.973, 15.50, Moon , GeoPosition20.19, 30.77, Moon 

Next, we find out that the shortest route starts where Apollo 11 landed:
In[
]:= order  LastFindShortestTourlandingcoords

Out[
]= 1, 6, 4, 2, 3, 5, 1

Finally, we create a plot showing the path of the tour:

In[
]:= GeoListPlotlandingcoords order , Joined  True, GeoLabels  Automatic

11.6 Optimum Flows

In this example, we want to maximize the total number of freight cars that can be
sent from Vancouver to Winnipeg given the following network characteristics:
The code generating graph g is not shown.
408 Mathematica beyond mathematics

Edmonton

12
16 4 Saskatoon
10
9
In[
]:= g  ;
7 20
Vancouver 13 Calgary
14
Winnipeg
Regina
4

The previous illustration has been created using the Graph function. For simplifying
purposes, let’s just focus on 3 cities: { (1) Vancouver, (2) Calgary, (3) Edmonton}. The
flow directions (ij) and capacities (EdgeCapacity) are: {From (1) Vancouver to (2)
Calgary, 13, from (1) Vancouver to (3) Edmonton, 4, from (3) Edmonton to (2) Calgary, 10
and from (2) Edmonton to (3) Calgary, 4}. We display that information using
DirectedEdge and with the help of VertexLabels and Placed position the labels in the
desired locations. The style is defined using Style [#, “Style”]&. You can use the
documentation to learn more about the syntax for each function.
In[
]:= g1  Graph
"Vancouver", "Calgary", "Edmonton",
 1, 2, 1, 3, 2, 3, 3, 2, Null,  EdgeCapacity  13, 16, 4, 10,
EdgeLabels   DirectedEdge"Vancouver", "Calgary"  Placed13, 0.5, 2, 1,
DirectedEdge"Vancouver", "Edmonton"  Placed16, 0.5, 1.5, 2 ,
DirectedEdge"Edmonton", "Calgary"  Placed10, 0.5, 0.2, 1.5,
DirectedEdge"Calgary", "Edmonton"  Placed4, 0.5, 0.7, 1.5 ,
GraphStyle  "BasicBlack",
VertexLabels   "Edmonton"  Placed" 3 Ed.", Above, Style, Red & ,
"Vancouver"  Placed" 1 Vanc.", After, Style, Blue & ,
"Calgary"  Placed" 2 Cal.", Above, Style, FontFamily  "Helvetica" & ,
VertexSize  Small, ImagePadding  All, ImageSize  Small

3Ed.
2Cal.
10

Out[
]=
13
16

1Vanc.

In the next example, we add VertexCoordinates to specify the coordinates for the center of
the vertices. We also modify several options to make it easier to overlay the graph on a map
of Canada later.
11. Economic and Financial Applications 409

Graph
"Vancouver", "Calgary", "Edmonton",
 1, 2, 1, 3, 2, 3, 3, 2, Null,  EdgeCapacity  13, 16, 4, 10,
EdgeLabels   DirectedEdge"Edmonton", "Calgary"  Placed10, 0.5, 0.2, 1,
DirectedEdge"Vancouver", "Edmonton"  Placed16, 0.5, 1, 0,
DirectedEdge"Vancouver", "Calgary"  Placed13, 0.6, 1, 1.5,
GraphStyle  "BasicBlack", ImagePadding  0, 40, 0, 10,
VertexCoordinates  0.253, 0.112, 0.1419, 0.1239, 0.122, 0.1735,
VertexLabels   "Edmonton" 
Placed"Edmonton", Above, Style, FontFamily  "Helvetica" & ,
"Vancouver"  Placed"Vancouver",
1.5, 1, 0, 0, Style, FontFamily  "Helvetica" & ,
"Calgary"  Placed"Calgary", Below, Style, FontFamily  "Helvetica" & ,
VertexSize  Small

Edmonton

16 10
Out[
]=

Calgary
13
Vancouver

To find the optimum flow we use the function FindMaximumFlow.

In[
]:= F  FindMaximumFlowg, "Vancouver", "Winnipeg", "OptimumFlowData"

Flow value: 23
Out[
]= OptimumFlowData 

In[
]:= F"FlowValue"

Out[
]= 23

We can see the optimum flow per edge as follows:

In[
]:= Text  GridPrepend, F &  F"EdgeList",
Style"Railroad", Bold, Style "Edge Flow", Bold

Railroad Edge Flow

Vancouver  Calgary 7
Vancouver  Edmonton 16
Calgary  Regina 11
Out[
]= Edmonton  Calgary 4
Edmonton  Saskatoon 12
Regina  Saskatoon 7
Regina  Winnipeg 4
Saskatoon  Winnipeg 19
410 Mathematica beyond mathematics

Here we display the result on top of the map of Canada, downloaded using CountryData
(for further details see Chapter 5).
In[
]:= PanelShowCountryData"Canada", "Shape", F"FlowGraph",
PlotRange  0.31, 0.083, 0.04, 0.25

Edmonton

12
16 4 Saskatoon
Out[
]= 10
9

7 20
13 Calgary
Vancouver 14
Winnipeg
Regina 4

11.7 BlockChain
Let’s assume that I’d like to transfer $500 from one of my bank accounts to
another one. This kind of transaction doesn’t actually require the movement of
dollar bills from one bank to another. It just needs a simple change in the banks’
electronic balances. Similar transactions happen every day affecting millions of
accounts without involving any physical money, just modification of electronic bits.
Until recently, this type of financial activity was mostly done using a network of
banks. However, in 2008 someone (or maybe a group of people) under the
pseudonym of Satoshi Nakamoto, sent a message to a cryptography forum
containing a 9-pages document describing a method for creating a digital coin that
could be used as a unit of account and medium of exchange. The following year,
the ideas behind that paper were implemented using a virtual coin named bitcoin:
https://bitcoin.org/bitcoin.pdf .
Although Nakamoto was talking about a concrete coin, he was really describing a
revolutionary technology named blockchain with a multitude of potential
applications.
This technology eliminates intermediaries by decentralizing the network. Users,
instead of banks or other centralized organizations, take control, and it’s them that
keep track and verify all the transactions being recorded on the blockchain. One of
the most recent applications is the recording of the virtual ownership of digital files
through something called non-fungible tokens or NFTs.
11. Economic and Financial Applications 411

Blockchains are basically decentralized digital ledgers in which the entries (blocks)
are interconnected and encrypted to protect the privacy and safety of transactions.
They are, in other words, distributed and secured databases that can contain any
kind of records, not just economic ones. These chains of blocks have a very
important requirement: There have to be users (nodes) in charge of verifying and
approving the transactions so that the blocks containing them can be added to the
ledger.
Although a detailed explanation of blockchains is beyond the scope of this book,
it’s important to mention that to design one, we would need to consider first a list
of basic requirements known as the “five pillars of blockchain”:
1. Permissioned/Private. Writing records is exclusive to members; third parties can be granted read access,
with the general public excluded. The permissions architecture goes beyond “access = everything” and
allows third-party access to specific raw data, as deemed appropriate, for interoperability and application
requirements.
2. Decentralized/P2P. Allowing for equal control over the shared database among all permissioned
participants and of equal importance; distributing the number of full copies of the ledger to maximize the
probability that there will always be a complete record in existence and available for those with permissions
to access.
3. Immutability & Data Integrity. Records are guaranteed to be cryptographically secure, with no possibility
for bad actors to threaten data integrity.
4. Scalability. The ability to secure trillions of transactions or records without compromising the networks
synchronization, security, accessibility, or data integrity.
5. Security. Support for data encryption and the management and enforcement of complex permission settings
for participants and third parties.
Mathematica includes functionality to design and manage blockchains and there
are many videos covering these topics at:
https://www.wolfram.com/broadcast/s?sx=blockchain
The functions available related to blockchains can be used for:
Configuration ($BlockchainBase)
Reading (BlockchainData, BlockchainBlockData, BlockchainTransactionData and
BlockchainAddressData)
Writing purposes (BlockchainTransaction, BlockchainTransactionSign and
BlockchainTransactionSubmit):
Here are some examples:

We can first specify the default blockchain to be used for blockchain computations with
$BlockchainBase:
In[
]:= $BlockchainBase  "Bitcoin"

Out[
]= Bitcoin

Now we can read the information about the specified default blockchain using
BlockchainData:
In[
]:= BlockchainData Dataset
412 Mathematica beyond mathematics

The function below shows how to create a Pay-to-Public-Key-Hash (P2PKH) transaction,

the most common form of transaction on the Bitcoin network using BlockchainTransaction:
In[
]:= BlockchainTransaction
"Inputs"  "TransactionID" 
"9bef194418607318673caaf8f45d52fb60e8c8586e6ce8414c377a279ca14399",
"Index"  0, "Outputs"  "Amount" 
0.0002` , "Address" 

"munDTMqa9V9Uhi3P21FpkY8UfYzvQqpmoQ", "Amount" 
0.16223` ,

"Address"  "mo9QWLSJ1g1ENrTkhK9SSyw7cYJfJLU8QH" ,

"BlockchainBase"  "Bitcoin", "Testnet"

blockchain base: Bitcoin, Testnet

Out[
]= BlockchainTransaction signed: False 

  

As of May 2022, Mathematica supports three virtual coins: Bitcoin, Ethereum, and
Ark (https://ark.io) . This last coin could be very useful if you were interested in
developing blockchain applications since there is a written agreement between
ARK Ecosystem and Wolfram Blockchain Labs (WBL) that integrates the ARK
public API into the Wolfram Language.
WBL: https://www.wolframblockchainlabs.com/
Notice the appearance of the word “Hash” in both, one of the properties of the
BlockchainData output and in the name of the blockchain transaction. This term
refers to a type of mathematical function that converts an input of arbitrary length
into an output of fixed length consisting of a series of unique characters. This
function is crucial to validate transactions, safely store passwords, and sign
electronic documents.

11.8 Additional Resources

The Wolfram Finance Platform: http://www.wolfram.com/finance-platform
Mathematica's finance related functions:
http://reference.wolfram.com/mathematica/guide/Finance.html
Constrained optimization tutorial: tutorial/ConstrainedOptimizationOverview
Financial, Economics and more: https://www.wolfram.com/wolfram-
u/courses/catalog/?topic=finance/ and https://www.wolfram.com/solutions/
 12
Faster, Further

This chapter discusses how to make computationally intensive tasks more

efficient by distributing the calculations among several cores using
parallel computing. In addition, if our computer has a NVidia graphics
card, we also have the option of using the card processors through the
Compute Unified Device Architecture (CUDA) programming interface,
integrated in Mathematica, for extra processing power. The chapter also
describes complementary tools such as Workbench, an excellent
Integrated Development Environment (IDE) for large Mathematica-based
project. Finally, some indications to connexion with other programs and
devices are given.

12.1 Parallel Computing

Computer programs are usually run sequentially: A problem is broken down into a
series of instructions performed one by one in order. Only one instruction is carried
out at any time, with its output being passed to the next one until the entire
program has been executed.

Parallel computing consists of breaking down programs into discrete components

which can be executed simultaneously. In Mathematica, this is done by distributing
the individual program parts among multiple kernels. This is possible when the
computer we’re using has several processors, if we distribute the computation
among several machines or a combination of both. (B. Blais, Introduction to
Parallel Computing, Lawrence Livermore National Laboratory:
https://hpc.llnl.gov/documentation/tutorials/introduction-parallel-computing-tutorial
To do parallel computing with several processors, one of them must act as the
master distributing and synchronizing the operations with the rest of the processors
(located in the same computer and/or in other computers connected to the grid).
This task implies an extra amount of time (communication, synchronization, etc.)
that must be taken into account. The objective is to reduce the total computation


      

time compared to how long it would take to perform the calculation sequentially.
There are many scientific and technical fields where parallel computing is very
useful or even the only realistic way to perform certain types of calculations. For
example: in meteorology, there used to be predictive models that took longer to
evaluate than the actual prediction window.
In the past, parallel programming was a niche area only available for expensive
hardware. However, today even PCs have more than one processor. If you have an
NVidia graphics card, you can also use CUDA (included in Mathematica) for
parallel computing. This computation power gets even bigger if we connect
computers to a grid. You can even participate in collaborative projects: One of the
best examples of this approach was the well-known SETI program, another
example of a collaborative participation project is the study of protein folding:
https://fold.it/.
In summary, Mathematica has the capability to perform calculations in parallel
right out of the box. In addition, it’s possible to set up a grid to distribute
computations to several computers using either gridMathematica or Wolfram
Lightweight Grid Manager
gridMathematica: https://www.wolfram.com/gridmathematica
WLGM: https://www.wolfram.com/lightweight-grid-manager

12.2 Parallel Programming

12.2.1 How to Use More than One Kernel

Usually, computer programs are executed sequentially in a single CPU. Although
not exactly the same, we can say that in Mathematica a sequential program will be
executed in a single kernel, while a program that includes parallel calculations will
be distributed among all the available kernels and its instructions executed
simultaneously.

Let’s start with a list of numbers that we’d like to factorize (transform them into a product
of prime numbers).


 numbers
342 895, 423 822, 1546,
168 972, 123 555, 23 440, 1378, 957, 348 300, 2987;

Factorization plays a crucial role in cryptography. As a matter of fact, the

transmission of encrypted information is often done using numbers obtained from
multiplying two prime numbers. This technique is the most commonly used one
when doing secure commercial transactions over the Internet.

For factorization purposes, Mathematica has the FactorInteger function that when used
along with Map (or
 ) can be applied to more than one number at once:
  



 list
MapFactorInteger, numbers

 
 5, 1, 7, 1, 97, 1, 101, 1, 2, 1, 3, 1, 7, 1, 10 091, 1,
2, 1, 773, 1, 2, 2, 3, 1, 14 081, 1,
3, 1, 5, 1, 8237, 1, 2, 4, 5, 1, 293, 1,
2, 1, 13, 1, 53, 1, 3, 1, 11, 1, 29, 1,
2, 2, 3, 4, 5, 2, 43, 1, 29, 1, 103, 1

The previous result is formatted as a list. Next, we display a table where we can see each
number and its corresponding prime factors.


 GridTransposenumbers, CenterDot  Superscript   &  list,
Frame
All

342 895 51  71  971  1011

423 822 21  31  71  10 0911
1546 21  7731
168 972 22  31  14 0811
123 555 31  51  82371
  
23 440 24  51  2931
1378 21  131  531
957 31  111  291
348 300 2  34  52  431
2

2987 291  1031

Operationally, the kernel or logical core of Mathematica (not to be mistaken with

the physical core of the computer) factorized each number sequentially. That is, it
first factorized 342,895, then 423,822 and so on. In this case, the factorization time
was very small and so wouldn’t justify the use of parallel computations. However,
when the factorization involves big numbers, the calculation time increases
dramatically so we’d be very interested in using methods to reduce it, such as
parallel computing.
For parallel computing purposes, Mathematica has certain functions that are added
to the usual ones to parallelize calculations, but before we can take advantage of
them, we need to set up Mathematica properly.

The standard installation of Mathematica includes the possibility of using more than one
kernel in the host machine with the actual number depending on the license. We can access
the computer configuration by going to: Edit  Preferences  Parallel. We will see a
screen similar to the one in Figure 12.1 (for further details you can watch the video
http://www.wolfram.com/broadcast/screencasts/parallelcomputing).

       

Figure 12.1      

In this case, the number of kernels that will be launched automatically is 4, the same as the
number of computer cores. Notice that the kernels are set by default to low priority, giving
preference to other computer processes.

To launch the available kernels we click on the Parallel Kernel Status... button. We can
also do the same by going the menu bar and choosing: Evaluation  Parallel Kernel
Status. In both cases, a window similar to the one in Figure 12.2 will appear:

Figure 12.2     

An alternative way to launch the kernels directly is to write down the following command
(for this to work, the kernels should not have been launched previously using any of the
methods described above):

  LaunchKernels
  


If you launch more cores than the ones available in the computer, you’ll get a $Failed
message for each core not found (this can happen, for example, if you have configured your
machine for WLGM, to which we will refer later, but the application has not been activated
yet).

We can always find out our available kernels using the following command:


 Kernels

To execute a program in parallel, in addition to having more than one kernel

available, the instructions have to explicitly state that their execution should be
done in parallel. Usually this is done using Parallelize [exp], with exp being the
regular Mathematica command. Parallelize will launch the kernels first, if
they haven’t been launched already using any of the methods described in the
previous paragraphs.

You can also use the commands that directly enable parallel evaluation:


 ? Parallel

System`
Parallel
Parallel
Parallel
Parallel

Parallel
Paralleli
Parallel
Parallel

Comb
epipe
ogra
Subm

Array zation Map Table
 
 ine d m it
Parallel
Parallel
Parallel

Parallel
Paralleli
Parallel
Parallel
Parallel

AxisP
Evalu
Produ

Do ze Needs Sum Try
lot ate ct

To factorize the numbers previously defined, we put the command we used earlier inside
Parallelize (Instead of Parallelize[Map[f,expr]] we can also use the equivalent
expression ParallelMap [f,expr]).


 ParallelizeMapFactorInteger, numbers

 
 5, 1, 7, 1, 97, 1, 101, 1, 2, 1, 3, 1, 7, 1, 10 091, 1,
2, 1, 773, 1, 2, 2, 3, 1, 14 081, 1,
3, 1, 5, 1, 8237, 1, 2, 4, 5, 1, 293, 1,
2, 1, 13, 1, 53, 1, 3, 1, 11, 1, 29, 1,
2, 2, 3, 4, 5, 2, 43, 1, 29, 1, 103, 1

In this case, we have used all the available kernels to simultaneously factorize the
numbers.

       

We can try to use this command with other functions, although as mentioned
before, the execution may not be faster. That will depend on many factors, such as
the type of calculation, the number of kernels available, etc. If the computation
time is short (a few seconds or less) we will not notice any difference or the
parallel calculation may even take longer.

In the example below, we show the computation time with and without parallelization.
When measuring computation times for comparison purposes, it’s recommended to start a
new session (you can use Quit [] to exit the current one) since, as we have seen in previous
chapters, Mathematica will automatically reuse previous calculations reducing the
execution time significantly. We use AbsoluteTiming instead of Timing because
AbsoluteTiming takes into account the entire calculation time while Timing only
considers the CPU time. The “;” tells Mathematica not to print the output of the command
since we’re only interested in its computation time.
10 000


  i ; AbsoluteTiming
i
1

 
 0.928053, Null
10 000


 Parallelize  i ; AbsoluteTiming
i
1

 
 0.415421, Null

Sometimes is not possible to parallelize a computation:

1 x^2


 Parallelize  x
x




            

x2
 
  Logx
2

12.2.2 Defining Functions

When doing parallel programming, it’s very important to remember that if we
define a function and we’d like it to be used by all the kernels, we have to
explicitly state it, as shown in the next example.

We define a function that generates a random number between 1 and 10.


 rand :
RandomInteger10

Next, we use DistributeDefinitions to parallelize the previous function so it can be

used by all the available kernels.


 DistributeDefinitionsrand;

Finally, we execute the function in all the different kernels.


 ParallelEvaluaterand

 
 10, 7, 0, 6, 6, 6

Once a function has been parallelized, it will remain so even after closing the kernels. In
this example, we close the kernels and open them again.
  



 CloseKernels; LaunchKernels;

The definition still works with the new kernels:


 ParallelEvaluaterand

 
 6, 2, 6, 4, 9, 2

Commands in Mathematica that start with Parallel , such as ParallelTable

(equivalent to Parallelize[Table[
]]), will automatically parallelize the definitions,
without the need to use DistributeDefinitions , as the following example shows:


 fln :
2 ^ n


 ParallelTableflk, k, 1, 6

 
 2, 4, 8, 16, 32, 64

An alternative way to execute the previous command is by using With to give values to the
local variables:


 WaitAllTableWithi
i, ParallelSubmit2 ^ i, i, 6

 
 2, 4, 8, 16, 32, 64

Sometimes, we may be interested in performing the calculation in a specific kernel. In this

example, we choose kernel 2 to execute the computations:


 CloseKernels; LaunchKernels;


 kernel2
Kernels2

    
 
 KernelObject

 !" #

Remember that if we don’t parallelize variables, we will find unexpected behavior.

In this example, the equality a


2 (remember that SameQ (===) checks whether two
expressions are identical) returns false even though a
2, because a has not been defined
for all the kernels:


 a
2;
ParallelEvaluatea


2, kernel2
 
 True

We can fix the problem using With:


 Witha
2, ParallelEvaluatea


2, kernel2 

 
 True


 CloseKernels;

12.2.3 ParallelTry


 LaunchKernels;

With ParallelTry we can evaluate a command in different kernels and get the
result from the kernel that executes it first. This can be useful, for example, when
comparing different computation methods.

Here we compare different methods for evaluating the minimum of a function. The output
shows the fastest one:

       


 AbsoluteTimingParallelTry

Sin50 x
1
, NMinimize  Sin10 x y  x2 y2 , y 0 && x 0,
4
x, y, Method
 &, "DifferentialEvolution",

"NelderMead", "RandomSearch", "SimulatedAnnealing"

 
 0.141646, NelderMead,  0.628918, x  0.0942305, y  0.0625364

12.2.4 ParallelCombine
ParallelCombine distributes part of a computation among all the available
kernels combining the partial results. It’s normally used with list elements. Its
syntax is: ParallelCombine[f, {e1 , ..., en }] that applies f to the elements of a list
and groups the partial results.

The function below calculates the first 9 primes. It’s the same as Prime[ {1, ..., 9}]
except that the calculation in this case is done in parallel:


 ParallelCombinePrime, 1, 2, 3, 4, 5, 6, 7, 8, 9

 
 2, 3, 5, 7, 11, 13, 17, 19, 23

12.2.5 ParallelEvaluate
The function ParallelEvaluate evaluates an expression in different kernels
and returns the result from each one of them. By default it uses all the available
kernels but we can specify which ones to use. This function is commonly
employed in calculations involving Monte Carlo simulation. The next example
shows how to calculate  through simulation (an illustrative but quite inefficient
method).

First, we generate m pairs of real numbers {x, y} between 0 and 1. Then we find out which
ones fall inside the circle with radius r  1 with Norm[x y] < 1 equivalent to x2  y2 < 1.
Finally, we multiply by 4 since we’re only generating positive {x, y}, simulating just 1/4 of
the circle:


 Withn
10 ^ 5, ParallelEvaluate
4. CountRandomReal0, 1, n, 2, xy ; Normxy  1 n
 
 3.13396, 3.14492, 3.13684, 3.13684, 3.14632, 3.14472

The result is an estimate from each kernel. We calculate the mean to get a final value:


 Mean

 
 3.1406

12.2.6 ParallelMap
ParallelMap is another commonly used function that, as previously mentioned,
is equivalent to Map for sequential computations.
  


We are going to solve again the previous example of approximating . We generate pairs of
points {x, y} with values between -1/2 and 1/2. Next we select those points inside a circle
with radius 1/2 centered at (0,0) using again Norm[r] < r. We then proceed to calculate 
using the output given by the command below:


 Solvepi r ^ 2 1  interiorPoints total, pi
interiorPoints
 
 pi  
r2 total


 r
1 2;


 DistributeDefinitionsr

 
 r


 pts
ParallelTableRandomRealr, r, 25 000, 2, LengthKernels;


 interiorpts
ParallelMapSelect, Norm  r &  &, pts;

By combining the results, we get the following approximation:

  DistributeDefinitionspts, interiorpts;
NMeanParallelTableLengthinteriorptsi Lengthptsi r ^ 2 ,
i, LengthKernels
 
 3.14755

12.2.7 Finding the Roots of an Equation

We’d like to find the solutions to the equation cost2   t  0.3 with FindRoot, a
function that uses numerical algorithms requiring an initial guess.

The graph below displays the multiple solutions (actually an infinite number of them) that
correspond to the intersection points of both curves:


 PlotCost ^ 2, Expt .3, t, 0, 10

1.0

0.5

  

2 4 6 8 10

0.5

1.0

If we only give the function one initial guess, we will usually get a solution close to it:


 FindRootCost ^ 2  Expt .3, t, 3

   t  2.26454

We are going to find the solution with and without parallelization. We just want to
compare the computation times so we’ll not show the results.

       

Using multiple initial guesses, we can find all the equation roots for a given domain. We
have used Map since what we’re really doing is solving the equation for different starting
points:


 initval
Tablex, x, 0, 4 Pi, .1;


 vals
MapFindRootCost ^ 2  Expt .3, t,  &, initval;
AbsoluteTiming
 
 0.0277301, Null

The command below solves the problem replacing Map with ParallelMap.


 pvals
ParallelMapFindRootCost ^ 2  Expt .3, t,  &,
initval; AbsoluteTiming
 
 0.0536711, Null

In this case, the difference is negligible since, as mentioned before, there are many
factors that may influence the computation time.

We visually represent the solutions:


 ShowPlotCost ^ 2, Expt .3, t, 0, 12, Graphics
Red, PointMapThreadt . 1, 2 &, pvals, Cost ^ 2 . pvals

1.0

0.5

 


2 4 6 8 10 12

0.5

1.0

12.2.8 GridMathematica and Wolfram Lightweight Grid Manager

As mentioned earlier in the chapter, Wolfram Research offers two additional
programs for performing parallel calculations.
GridMathematica extends Mathematica's built-in parallelization capabilities to run
more tasks in parallel over more GPUs and CPUs for faster computations.
Alternatively, Wolfram Lightweight Grid Manager makes it easy to discover and
connect to Wolfram engines installed in other computers in a local network.

12.3 The Mandelbrot Set

12.3.1 Defining the Mandelbrot Set

The Mandelbrot set is arguably the most famous fractal set. It can be created by
recursion as follows :
  


z0  C

zn1  zn2  C

where C is any complex number.

http://mathworld.wolfram.com/MandelbrotSet.html
If we take C such that zn1 doesn’t tend to infinity, the points generated by this
recursion belong to the Mandelbrot set.

The previous series can be easily computed in Mathematica with FixedPointList.


 FixedPointList ^ 2 c &, c, 5
2 2 2 2
 
 c, c  c2 , c  c  c  , c  c  c  c2   ,
2 2 2 2 2 2 2
c  c  c  c  c2   , c  c  c  c  c  c2   

With C = 1, the series diverges (in the example we display only the first 6 terms). It tends to
infinity in absolute terms, so it doesn’t belong to it.


 FixedPointList ^ 2 1 &, 0, 5

 
 0, 1, 2, 5, 26, 677

However if C  1, the series is bounded and therefore belongs to the set.


 FixedPointList ^ 2  1 &, 0, 5

 
 0, 1, 0, 1, 0, 1

12.3.2 Visualizing the Mandelbrot Set

Often, the Mandelbrot set is represented using the escape-time coloring algorithm
based on applying different colors depending on the number of iterations required
to prove if a series belongs to the set. Darker colors (in our example bright red)
indicate that only a few iterations are needed while lighter ones mean that many
iterations were required. There’s always a limit for the maximum number of
iterations allowed. Since Mathematica 10, the build-up function
MandelbrotSetPlot is available, however, we will build a function step by step
to represent the Mandelbrot set.

FixedPointList with the option SameTest enables us to set a logical test to stop the
iteration as soon as it returns true starting with the first two values of the series. The
command below returns the number of iterations needed to get an absolute value greater
than 2 (the limit traditionally used) and if after 50 iterations, the absolute value still doesn’t
meet the condition, it returns the number of iterations +1. The higher the number of
iterations, the higher the certainty that the series belongs to the Mandelbrot set:

  Length
FixedPointList ^ 2  0.2 0.1 I &, 0, 50, SameTest
Abs  2 & 
 
 51

We define a function that implements the previous approach for a given complex number.
In this case we use a maximum of 20 iterations:


 mf1z :

LengthFixedPointList ^ 2 z &, z, 20, SameTest
Abs  2 & 


      

We apply it to find out what points in the complex plane x   y belong to it. We add
AbsoluteTiming to compare the computation times using different methods.


 pts1
Tablemf1x I y, x, 2, 1, 0.002, y, 1, 1, 0.002;
AbsoluteTiming
 
 22.5427, Null

We use ArrayPlot to visualize the previous points.


 ArrayPlotpts1, ColorFunction
Hue

 





12.3.3 Faster
In this section, we show several techniques, apart from parallelization, to reduce
computation time.
These techniques require the presence of a C compiler in the computer.
(CCompilerDriver/tutorial/Overview). For 32-bit systems, an option is to install
Visual C++ Express, a Microsoft application that can be downloaded for free from:

http://www.microsoft.com/visualstudio
In case of 64-bit systems, one option is to install the 64-bit version of MinGW . In
our case, we have downloaded mingw-w64-bin_i686-mingw_20111220 and
copied it to the folder C:\mingw.
Mingw-w64: http://mingw-w64.sourceforge.net

To run the C compiler we need to load the following package first:


 Needs"CCompilerDriver`"

The command below will display all the C compilers available in your computer:


 CCompilers

 
 

If the function above returns an empty list it may be that either you don’t have any compiler
installed or, as in our case, you have installed MinGW in folder: C:/MinGW. To let
Mathematica know that you have the open-source software in your computer, execute the
following commands:


 Needs"CCompilerDriver`GenericCCompiler`"
  



 $CCompiler

"Compiler"
GenericCCompiler, "CompilerInstallation"
"C: MinGW",
"CompilerName"
"x8664w64mingw32gcc.exe";

Now we can launch the available kernels as explained before. To use kernels over
a grid we need to make sure that we have access to them through WLGM.

Here we repeat the calculation from the previous section but in parallel:


 mf2z :

LengthFixedPointList ^ 2 z &, z, 20, SameTest
Abs  2 & 


 DistributeDefinitionsmf2;


 pts2
ParallelTablemf2x I y,

x, 2, 1, 0.002, y, 1, 1, 0.002; AbsoluteTiming
 
 5.33571, Null

We can still significantly reduce the computation time with the command Compile (not
limited to parallel calculations). We use it to indicate that we’re going to evaluate a function
only for numeric inputs. This way the program saves time by not performing the checks
associated to symbolic calculations and can apply its own routines optimized for numerical
calculations.


 mf3
Compilec, Complex,
LengthFixedPointList2 c &, c, 50, SameTest
Abs2  2.0 & ;


 DistributeDefinitionsmf3;

pts3
ParallelTablemf3x y ,
x, 2, 1, 0.002, y, 1, 1, 0.002; AbsoluteTiming
 
 1.05738, Null

We can perform the computation even faster by taking advantage of some of the options of
Compile . Using this function we can convert the Mathematica-defined function into C
code that is compiled and dynamically linked:


 cfun
Compile x, x Sinx ^ 2,
CompilationTarget
"C", RuntimeAttributes
Listable;
cfun3, 4
 
 1.23636, 1.15161

Now, we can define the fastest function so far:


 mf4
Compilec, Complex, Modulenum
1, FixedPoint num ;
2 2 2
1 c &, 0, 99, SameTest
Re1 Im1 4 & ;
num, CompilationTarget
"C",
RuntimeAttributes
Listable, Parallelization
True;

  pts4
mf4Tablex  y, x, 2, 1, 0.002, y, 1, 1, 0.002;
AbsoluteTiming
 
 1.46059, Null

We display the results from the three functions using different color templates and compare
them with the built-in function MandelbrotSetPlot :

       


 GraphicsRowArrayPlotpts2, ColorFunction
"GreenPinkTones",

ArrayPlotpts3, ColorFunction
"BlueGreenYellow",
ArrayPlotpts4 , ColorFunction
"Rainbow"


 GraphicsRowMandelbrotSetPlotColorFunction
"GreenPinkTones",

MandelbrotSetPlotColorFunction
"BlueGreenYellow",
MandelbrotSetPlotColorFunction
"Rainbow"

If you are a programmer you can even generate C code that can be converted into
an executable file (CCodeGenerator/tutorial/Overview) or you can call external
programs (guide/CallingExternalPrograms).

To generate C code we need to load the CCodeGenerator package. In the example below,
we load it after clearing all the variables from memory:

  Clear"Global` "


 Needs"CCodeGenerator`"

The next command generates optimized C code for a given expression:


 s
ExportStringCompilex, Sinx ^ 2, "C";

We use Panel in the following example to show the output of the previous instruction
inside a frame displaying the generated code along with all the required headings and
libraries necessary to compile an executable file.
  



 PanelzStringReplaces, "\r\n"
"\n",

Full, 300, Scrollbars
Automatic, True,
BaseStyle
FontFamily
"Courier", FontSize
8

z include "math.h"

include "WolframRTL.h"

static WolframCompileLibrary Functions funStructCompile;

static mbool initialize 1;

include "m 4a3e4639 9687 41a0 8159 98ba6a7d8dc2.h"

DLLEXPORT int Initialize m 4a3e4639 9687 41a0 8159 98ba6a7d8dc2 WolframLibraryData libData

if initialize

funStructCompile libData compileLibraryFunctions;
initialize 0;

return 0;


DLLEXPORT void Uninitialize m 4a3e4639 9687 41a0 8159 98ba6a7d8dc2 WolframLibraryData libData
 
 
if initialize

initialize 1;



DLLEXPORT int
m 4a3e4639 9687 41a0 8159 98ba6a7d8dc2 WolframLibraryData libData, mreal A1, mreal Res

mreal R0 0;
mreal R0 1;
mreal R0 2;
R0 0 A1;
R0 1 R0 0  R0 0;
R0 2 sin R0 1;
Res R0 2;
funStructCompile WolframLibraryData cleanUp libData, 1;
return 0;


, Full, 300, Scrollbars  Automatic, True

12.4 Comparing Organisms Genetically

Let’s see how parallel computing is used in the real world. In this case, we are
trying to determine the genetic proximity of organisms using a string metric known
as the Levenshtein distance.
The Levenshtein distance is used in information theory to measure the similarity
between two sequences. We can define it as the minimum number of single
character edits necessary to convert one sequence into the other one.
For example: To go from kitten to sitting, we proceed as follows:
kitten  sitten (we substitute 'k' with 's')
sitten  sittin (we substitute 'e' with 'i')
sittin  sitting (we insert 'g' at the end).
This means that the Levenshtein distance is 3 since we have made 3 edits to turn

       

kitten into sitting.

This metric can be calculated in Mathematica with EditDistance:


 EditDistance"kitten", "sitting"

 
 3

Let’s apply this metric to compare the genomes of three organisms using the files
located in the Data directory that, along with other examples in this book, can be
downloaded from http://diarium.usal.es/guillermo.

We first set the working directory to the one where the files are located (Data). This way we
can use Import directly without the need to state the entire path:


 SetDirectoryFileNameJoinNotebookDirectory, "Data"

 
 G:\Mi unidad\Mathematica\MBN13\Data


 picrogenes
Import"picrophilustorridusgene.fna";

acidogenes
Import"thermoplasmaacidophilumgene.fna";
volcaniumgenes
Import"thermoplasmavolcaniumgene.fna";
In the previous example, the computation time was very short, but, when the number of
characters to compare increases, the time required for the calculations increases drastically.

Let’s see the partial contents of one of the files containing the genome of the picrogenes.
The command below shows some of their DNA bases (A,C, G, T):


 picrogenes

ATGAATTCGGATAACTTTCAATCCTATTTAGGTTATTATTTAACCTGGATTTCCTTGGCCATATTTGC
CGAGATACTATACTTTCAATCCTATTTAGGTTATTATTTAACATATATCGGATTTCAAACCCT
TATAGATAGACTAGACTTTCAATCCTATTTAGGTTATTATTTAACGCCAGAGGCATTTCTCCG
TGCGTGGTCCTTTAATGACTTTCAATCCTATTTAGGTTATTATTTAACACAACAGGCAAAAAC
 
 AGAATTAAGACTGGCCATTGAGCTTTCAATCCTATTTAGGTTATTATTTAACCCGTTTCTAAT
AAACAATATAATAAATGATTTTTAA, 1596 ,
ATGTTCACATCAATAGATCATAATAATAGCG  CAATCTTATAGGTGCCTTAAGATGGATATAA

 
 


 
 

  

These three organisms are very simple and their genome size (total number of genes) is,
respectively:


 Length  picrogenes, acidogenes, volcaniumgenes

 
 1598, 1575, 1610

Let’s load the imported data to analyze them using parallelization:


 With picrogenes
picrogenes, acidogenes
acidogenes,
volcaniumgenes
volcaniumgenes, ParallelEvaluatepicro
picrogenes;
acido
acidogenes; volcanium
volcaniumgenes;;;
  


The command below shows the data for one gene that will be used as an input for the first
kernel:


 ParallelEvaluatepicro1, Kernels1

 
 ATGAATTCGGATAACTTTCAATCCTATTTAGGTTATTATTTAACCTGGATTTCCTTGGCCATATTTGCCGAGA
TACTATACTTTCAATCCTATTTAGGTTATTATTTAACATATATCGGATTTCAAACCCTTATAGATAGACT
AGACTTTCAATCCTATTTAGGTTATTATTTAACGCCAGAGGCATTTCTCCGTGCGTGGTCCTTTAATGAC
TTTCAATCCTATTTAGGTTATTATTTAACACAACAGGCAAAAACAGAATTAAGACTGGCCATTGAGCTTT
CAATCCTATTTAGGTTATTATTTAACCCGTTTCTAATAAACAATATAATAAATGATTTTTAA

For the next example, we want to compare the first eight genes of thermoplasma
acidophilum with those of thermoplasma volcanium. However, before doing it,
let’s discuss a simpler case first. In the example below we compare the initial 4
bases of the first 2 genes of volcaniumgenes, with their respective counterparts in
thermoplasma acidophilum.

We start by extracting the relevant bases and genes from both organisms:


 vo1, vo2, aci1, aci2

StringTakevolcaniumgenes1 , 4, StringTakevolcaniumgenes2, 4,
StringTakeacidogenes1, 4, StringTakeacidogenes2, 4 
   ATGA, ATGA, ATGT, ATGG
Notice how the first two bases of thermoplasma volcanium are actually the same.

To make the comparison we’ll use a combination of the EditDistance and Min
functions. However, in the command below we use f1 and f2 first to gain a better
understanding of how the actual instruction works. We show the logic in four steps:

  f2Mapf1x,  & , aci1, aci2

 
 f2f1x, ATGT, f1x, ATGG


 MapFunctionx, f2Mapf1x,  & , aci1, aci2, vo1, vo2

 
 f2f1ATGA, ATGT, f1ATGA, ATGG,

f2f1ATGA, ATGT, f1ATGA, ATGG

Here f1 and f2 are replaced with EditDistance and Min , respectively:


 MapFunctionx, f2MapEditDistancex,  & , aci1, aci2,
vo1, vo2
 
 f21, 1, f21, 1


 MapFunctionx, MinMapEditDistancex,  & , aci1, aci2,

vo1, vo2
 
 1, 1

Finally we apply the previous pattern using ParallelMap to compare the first eight genes
of thermoplasma acidophilum with those of thermoplasma volcanium:


 acidovolcanium

ParallelMapFunctionx, MinMapEditDistancex,  & , acido,
volcaniumgenes1 ;; 8; AbsoluteTiming
 
 20.9208, Null
Depending on the computer being used, the calculation time may take up to several minutes.

       


 acidovolcanium

 
 304, 700, 265, 227, 497, 243, 122, 104

We proceed the same way to compare the genes between picrophilus torridus and
thermoplasma volcanium:


 picrovolcanium

ParallelMapFunctionx, MinMapEditDistancex,  & , picro,
volcaniumgenes1 ;; 8; AbsoluteTiming
 
 22.301, Null


 picrovolcanium

 
 399, 968, 442, 353, 550, 240, 180, 124

The graph below shows the genetic proximity of the first eight genes in both cases:
thermoplasma acidophilum vs. thermoplasma volcanium and picrophilus torridus vs.
thermoplasma volcanium:

  ListLinePlotacidovolcanium, picrovolcanium,
PlotLegends
"acido vs. volcanium", "picro vs. volcanium"
1000

800

600
     
 


400
     

200

2 4 6 8

Comparing thermoplasma acidophilum vs. thermoplasma volcanium and picrophilus

torridus vs. thermoplasma volcanium in terms of their mean distances we get:


 Meanacidovolcanium, Meanpicrovolcanium N

 
 307.75, 407.

A more efficient alternative calculation method using WaitAll is:


 acidovolcanium
MapFunctiongene, ParallelSubmit
MinEditDistancegene,  &  acido, volcaniumgenes1 ;; 8;
picrovolcanium
MapFunctiongene, ParallelSubmit
MinEditDistancegene,  &  picro, volcaniumgenes1 ;; 8;
acidovolcanium;
picrovolcanium;
acidovolcaniumresult
WaitAllacidovolcanium;
picrovolcaniumresult
WaitAllpicrovolcanium; AbsoluteTiming
 
 23.511, Null
The calculation time is shorter than the sum of the times of the previous two computations.

The results are identical to the previously obtained ones:
  



 Meanacidovolcaniumresult N

 
 307.75


 Meanpicrovolcaniumresult N

 
 407.

12.5 Software Development with Wolfram Workbench

If you are going to build large software systems in Mathematica
(tutorial/BuildingLargeSoftwareSystemsInTheWolframLanguage) you may be
interested in using Wolfram Workbench
(http://www.wolfram.com/products/workbench ), an Integrated Development
Environment (IDE) based on software from the Eclipse Foundation
(http://www.eclipse.org). Workbench can help you edit, test and debug code. It can
also generate help files than can be integrated within the Mathematica
documentation. It’s highly recommended if you’d like to develop a high-quality
enterprise application.
A large project will require two things: Code (consisting normally of several
packages) and extensive documentation. Workbench will help you get both.
Wolfram Workbench requires the following software:
1. Mathematica
2. Java Development Kit (JDK) 8 or later
3. Eclipse 4.6 (Neon) or later
Note that the Wolfram Workbench plugin is compatible with the various standard
Eclipse IDEs.
Wolfram recommends having the latest version of these programs installed before
setting up the Wolfram Workbench plugin.

To set up Wolfram Workbench, install the plugin inside Eclipse and configure it to point to
your Mathematica installation as it is shown in the following:
http://support.wolfram.com/kb/27221 instructions.

Ones the Wolfram Workbench plugin is installed, to verify it, choose Help
About
Eclipse and identify the Wolfram Workbench logo toward the bottom of the window.

Figure 12.3   !" !#   $ 

432 Mathematica beyond mathematics

Configuring the Wolfram Workbench Plugin: Open the Eclipse Preferences

(Windows/Linux: Window
Preferences, Mac: Eclipse
Preferences) and select the
Wolfram category on left-hand side of the Preferences window.

Figure 12.4 Workbench Configuration Page.

To create a new application project, go to File
New
Project
Application Project

and type the name of your project.

Figure 12.5 Creating a new project.

After clicking on Next, we can assign a name to the application (in this example: Project1).
12. Faster, Further 433

Figure 12.6 Creating an application project.

Click on Finish. You’ll see in the Package Explorer tab the projects that you have. Select
Project1 and press on PacletInfo.m. Figure 12.6 will appear with basic details about your
project: Author, project version, Mathematica version, and a brief description.

Figure 12.7 Filling in the application details for the new project.

To learn more you can always visit: http://www.wolfram.com/workbench/

At the bottom of the webpage (see Figure 12.7) you will find a link to many useful
videos. Additionally, you can also go through examples of completed projects by
downloading some of them. It will help you to build your own projects and their
associated documentation.


      

12.6 Compute Unified Device Architecture (CUDA)

We start a new session


 Quit

NVidia graphics cards, present in many computers, include multiple processors

that can be used for parallel computations using a programming language named
Compute Unified Device Architecture (CUDA). CUDA takes advantage of the
power of the GPU (Graphical Processing Unit) to allocate computer instructions
among the kernels in the card, boosting the overall performance of the system.
Mathematica includes the possibility of using CUDA. CUDA toolkit for your
system should be installed.Then you can follow the CUDALink/tutorial/Overview
procedure as follow.
CUDA toolkit: https://developer.nvidia.com/cuda-toolkit

To program in CUDA we must load the following package first:

  Needs"CUDALink`"


 CUDAQ

 
 True

If CUDAQ returns False , then CUDALink will not work. For more information, read
"CUDALink Setup".

For CUDALink to work, you need to have an up-to-date driver. You can check this for your
system by running CUDADriverVersion .


 CUDAInformation

Now we can use the CUDA functions. For further information, please see the
tutorial: CUDALink/tutorial/Overview.

12.7 Connexion with Other Programs and Devices

12.7.1 Connexion with Other Programs

Another tool for developers is the Wolfram Engine for Developers. It has the same
core engine as Mathematica but with a different interface and different licensing.
While the complete program is used primarily for interactive computing with the
Wolfram Notebook interface, the Wolfram Engine for Developers is intended to be
called by other programs or web servers, using a variety of program
communication interfaces. This tool is licensed for pre-production use in
developing software and unlike Mathematica, it is not licensed for generating
outputs for commercial or organizational use.
According to the Wolfram website: “The Wolfram Engine is the heart of Wolfram
products such as Mathematica, Wolfram|One and Wolfram|Alpha”.
“Wolfram Engine Community Edition is freely available for pre-production
software development”: https://www.wolfram.com/engine/
If you are a Julia, NodeJS, Python, R, or Ruby user, you can perform computations
using theses languages within Mathematica. Please refer to “ref/ExternalEvaluate”
  


and “workflow/ConfigurePythonForExternalEvaluate” in the Documentation

Center for further details.

For instance, if you have installed Python, use  at the beginning of a input cell and after
selecting it as the language, you can start writing Python code (Figure 12.8):

Figure 12.8 (  )  $   

12.7.2 External Devices

Mathematica’s connectivity capabilities have been greatly enhanced with the
availability of a Wolfram Language framework for connecting external devices.
“The Wolfram Device Framework creates symbolic objects that represent external
devices, streamlines interaction with devices, and facilitates the authoring of
device drivers. A "device" in the framework can represent both an actual device,
such as a temperature sensor, or encapsulate a port, such as a serial port”
The tutorial Developing Device Drivers:
https://reference.wolfram.com/language/tutorial/DevelopingDeviceDrivers.html
explains the internals of the framework for advanced users and developers of
device drivers.
For details of the interaction with devices see: "Using Connected Devices":
https://reference.wolfram.com/language/guide/UsingConnectedDevices.html
As example, if you happen to be an amateur astronomer, you may enjoy the
following blog entry:
http://blog.wolfram.com/2014/12/29/serial-interface-control-of-astronomical-telescopes/
It covers examples of how to use the Wolfram Language to control telescopes
using the LX200 or Celestron NexStar protocols, very popular among astronomy
fans.

12.7.3 WolframAlpha application, APIs and Widgets

Another area that you may be interested in, is the development of applications and
widgets for mobile devices such as smartphones and tablets that make using
WolframAlpha easier. You can find tools to develop them in:
http://products.wolframalpha.com/developers/
An API function with one parameter named x applied to an Association :


 func
APIFunction"x"
"Integer", x 3 &;

       


 func"x"
5

 
 8

https://reference.wolfram.com/language/ref/APIFunction.html

12.8 Additional Resources

The Wolfram Language and the Wolfram Resources are continuously updating. A
summary of the new functionality included in the latest Mathematica versions can
be found in: the “New Features” section of the Documentation Center (Help

Wolfram Documentation) or, alternatively, online:
https://reference.wolfram.com/language/guide/RecentlyAddedFeatures.html
To access the following resources, if they refer to the help files, write their
locations in a notebook (e.g., CUDALink/guide/CUDALink), select them and press
<F1>. In the case of external links, copy the web addresses in a browser:
Parallelization tutorial:
reference.wolfram.com/mathematica/ParallelTools/tutorial/Overview.html.
Wolfram Research products product to parallelization: gridMathematica and Wolfram
Lightweight Grid Manager (WLGM).
CUDALink: CUDALink/guide/CUDALink
Information about Workbench: http://www.wolfram.com/products/workbench
Web applications can be developed with webMathematica:
http://www.wolfram.com/products/webmathematica
Wolfram Engine: https://www.wolfram.com/engine/
Index

! _ Annuity
See Factorial See Blank 382
# __ Apart
See Slot See BlankSequence 341
## ___ Append
See SlotSequence See BlankNullSequence 125
$InstallationDirectory ` Apply(@, @@, @@@)
89 177, 122 61, 62
% (Out[n]) <= or
ArrayPlot
42 See Condition 424, 426
& = Arrow
28 See Set 77, 147
&& (or ) == Association
See Condition See Condition 50, 51
* === AssociationMap
See Times See SameQ 51
> AssociationThread
See Dot See Condition 51
/ -> AtomQ
See Replace See Rule 42
// >= Or  Attributes
15, 20 See Condition 63
/;  AutocorrelationTest
See Condition See Ceiling 204
/@ AbsoluteTiming BarChart
See Map 59 165
:= Accumulate BarcodeRecognize
See SetDelayed 64 108
:-> or :> AirTemperatureData BetaDistribution
See RuleDelayed 129, 130, 167 180
@ Anatomic Data Binarize
See Apply 136 104
@@ AnatomyPlot3D BinCounts
See Apply 137 163
@@@ And (&& o ") Binomial
See Apply See Condition 75
[[…]] AngularGauge Biokmod
See Part 36 360
438 Mathematica beyond mathematics

BioSequence Circle DateList

139 45 176
BlackBody CityData DateListPlot
281, 282 124 130, 166, 203
Blank (_) ClassifierMeasurements DateObject
27, 60 213, 220 167, 258
BlankNullSequence (___) Classify DatePlus
60, 77 212 131, 203
BlankSequence (__) CloseKernels DateRange
60,77 419 204
Block CloudDeploy DateValue
72 86, 87 203
Blockchain ColorData DaughterNuclides
410, 411 33, 132 301, 308
Boole Compartmental Modeling Defer
234 331 82
Bootstrapping Compile DeleteCases
189 86, 425 78
BoxWhiskerChart ComponentMeasurements DeleteMissing
169 104 133, 163
BSplineFunction Condition derivative
95 73, 75 384
BubbleChart ContourPlot Det
185 31, 36, 351 53, 357, 358
C, C++ (Language) ContourPlot3D DiagonalMatrix
424 234 338
Callout CoordinatesToMapLink DictionaryLookup
129, 281, 324 162 126
Canvas Correlation Dimensions
33 168 47
Cases CountryData DinamicModule
78, 79 126, 145, 183 96
CCompilers Covariance DiracDelta
424 168, 188 350
CDF CSGRegion DiscretePlot
169, 170 34 170
Cell Cumulative Density Function DiscreteUniformDist
1-40 See CDF 243
Chart CurrentImage Disk
33 107 46, 96, 235
ChemicalData Cylinder DistributeDefinitions
135 233, 234 418, 425
ChemicalReaction D (or f') DistributionFitTest
135 27, 68, 100 178, 243
Chop Dataset Do
54, 55 50 63,64
Index 439

Dot () Epidemiological Models FinancialDerivative

47 342 386
Drawing Tools Equal (==) FindClique
22 67, 73 112
Drop EstimatedDistribution FindClusters
50 175, 380 212
DSolve EuclideanDistance FindDistribution
81, 84, 305 398 172
DSolveValue Evaluate FindDistributionParameters
333, 335 5, 170, 314 177
Dynamic Evaluate in place FindFaces
96 14 107, 209
DynamicGeoGraphics EvenQ FindFit
143 73, 74 198, 343
DynamicImage ExampleData FindGeoLocation
99 28 143
DynamicModule Except FindHamiltonianCycle
96 79 113
EarthquakeData Exoplanet FindMaximum
147 275 358,393, 394
EdgeDetect Expand FindMaximumFlow
102, 114 24, 25 409
EdgeRules ExpandAll FindMinimum
111 81, 315 393, 396, 397
EditDistance Expectation FindRoot
428, 429 171 287, 321, 387
EffectiveInterest Exploratory Data Analysis FindSequence
381 186 142
Eigenvalues ExponentialDistribution FindShortestTour
53 208 404
Element Export Fisher’s Iris Data
120 45, 154 185
ElementData Extract Fit
132 51, 52 194
Entity Factorial(!) FittedModel
29, 118, 119 58, 59 115, 196
EntityClass FactorInteger FixedPoint
118 414, 415 68
EntityList FieldSize FixedPointList
118 97 68, 423
EntityProperty FileNameJoin Flatten
118 154 50
EntityValue Financial Functions Floor
119 381 80
Epidemic Data FinancialBond FoldList
128 383 64
440 Mathematica beyond mathematics

FormFunction Get (<<) ImageAdjust

145 29 101
FormulaData Global Warming ImageCompose
126 165 104
FormulaLookup Graph ImageContents
126 110, 111 209, 210
Free-form Graphics ImageCorrespondingPoints
7 29 105
FreeQ Graphics3D ImageCrop
77 33 99
FullForm GraphicsColumn ImageData
44, 61 107 101
Function GraphicsRow ImageDeconvolve
27, 58, 60 102 106
GalaxyData Greater (>) ImageDimensions
256, 287 See Condition 101, 105, 216
GammaDistribution GreaterEqual(>=, ) ImagePartition
172 See Condition 102
GenericCCompiler Grid ImagePerspective
424 270, 317 109
GenomeData Head ImageResize
140 41-46 99, 183
GeoDensityPlot HighlightGraph ImageStitch
205 112 103
GeoElevationData HighlightImage Import
144 105 28, 98, 153
GeoEntities Histogram Initialization
205 164 95
GeoGraphics Histogram3D Initialization Cell
2, 265 182 15, 89, 154
GeoListPlot HoldForm InputField
148, 168 44 57
GeometricDistribution Hyperlink InputForm
179 162 6, 25
GeometricMean HypothesisTestData Inset
67 178, 243 77, 105, 232
GeometricTransformation IdentityMatrix Integer
110 338 46, 74
GeoNearest If IntegerQ
130 73 73
GeoPath Image Integers
147 34, 46, 91 171, 395
GeoPosition Image Assistant Integrate
143, 146 34, 100 8
GeoRegionValuePlot Image3D InteractiveTradingChart
145 106 379
Index 441

Interpreter Limit Machine Learning

124 15 186, 207
InterquartileRange Line MandelbrotSetPlot
163, 187 17, 67 423
Interval Linear Regression Model Manipulate
131 194 2, 35, 91
Inverse (Matrix) LinearLayer MannedSpaceMission
53 226 406
InverseCDF LinearModelFit Map (/@)
391 194 60, 61
InverseLaplaceTransform LinearOptimization MapAll
350 395 62
Isotope LinearProgramming MapAt
302 395 63
IsotopeData LinearSolve MapIndexed
123, 301 53, 54 63
JulianDate LineBreakChart MapThread
258 377 63
JuliaSetPlot List MatrixForm
2 25 25, 43, 47
KagiChart Listable MatrixPlot
377 62 54
Kernels ListAnimate MatrixQ
38, 414, 416 114 338, 340, 352
Keys ListDensityPlot Maturity
50, 51 324 383
Kurtosis ListLinePlot Max
163, 170 31, 69 166, 187, 191
Labeled ListLogLogPlot Maximize
63, 92 329 356, 358, 393
LaplaceTransform ListPlot Mean
341, 342, 349 30, 68 67, 131, 162
LaunchKernels ListPlot3D MeanAround
416 161, 323 189, 190
LayeredGraphPlot ListPointPlot3D Median
308 182 67, 166, 187
Length Location Message
47 129 75, 76
Less(<) LogIntegral Metadata
See Condition 254 514
LessEqual(<=) LogNormalDistribution Method
See Condition 177, 380 82, 83, 106
LetterCounts LogPlot Min
154 307 166, 187, 191
Light Curves LowpassFilter Minimize
291 107 393, 394
442 Mathematica beyond mathematics

MinMax NMinimize Pane

144, 174 393, 395, 396 257, 392
MinorPlanetData Nonlinear Regression Model Panel
274 195 132, 410
Missing NonlinearModelFit Parallel Computing
309, 311 115, 194, 195 413-422
MixtureDistribution NonNegative ParallelCombine
175 75, 88 420
Mod Norm ParallelEvaluate
105, 115 36, 420, 421 420
Modeling Normal Parallelization
331-435 51, 53, 115 425
Module NormalDistribution Parallelize
71, 72 175, 180, 182 417
Molecule NotebookDirectory ParallelMap
134 194, 299, 428 417
MoleculePlot3D NProbability ParallelSubmit
139 179 419, 430
MorphologicalGraph NumberForm ParallelTable
108 200, 247 419, 421
Most NumberQ ParallelTry
142 77, 357 419
Names NumericQ ParameterMixtureDist
70, 71 287, 354, 358 180
NDSolve Optimal Experimental Designs ParametricNDSolve
84, 340, 365 354-360 343
NDSolveValue Option Inspector ParametricPlot
329, 368 14, 15 33, 95, 366
Needs OptionQ Part ([[]])
29, 89, 360 338 25, 48
Nest Options ParticleData
65, 68 15, 22, 31 301
NestList Or Partition
64, 68, 69 See Condition 218, 245
NestWhileList OrderDistribution PathGraph
310 178, 179 113
NetChain Outer PDE
228, 229 63 366
NetExtract Package PDF
226 87 169, 170, 172
Neural Network Padding PeakWavelength
226 108 29
NIntegrate PadRight PerformanceGoal
82 245 224, 311
NMaximize PairedHistogram Physiological Modeling
296, 356, 393 174 346
Index 443

Pi Position Radiation
231-248 130, 142, 147 328
Piecewise Positive RadioButton
76 74, 232 392
PieChart Power RadioButtonBar
200 43, 44, 45 92, 96, 97
PieChart3D Precision Radon
378 79 114
Placed PrecisionGoal Radon (gas)
144, 170, 184 79, 82 309
PlanckRadiationLaw Predict Random
280 208, 219, 221 23
PlanetData PredictorMeasurements RandomChoice
256, 271, 273 224 69, 125, 167
Plot Prepend RandomFunction
11, 14, 21 270, 310,317 98
Plot3D Prime RandomGeoPosition
11, 27, 31 72, 73, 252 131
PlotLabel PrimePi RandomGraph
36, 82, 125 253, 254 112
PlotLegends Print RandomInteger
128, 144, 150 74, 83 23, 25, 52
PlotPoints Probability RandomReal
249, 368 98, 169, 170 49, 54, 55
PlotRange Probability density function RandomSample
27, 93, 100 See PDF 220, 223
PlotStyle PropertyValue RandomVariate
92, 172, 175 111 171, 172, 182
PlotTheme ProteinData RandomWalk
166 140 69
Plus (+) Pruning RandomWalkProcess
42, 45, 59 108 98
Point Pure function Range
31, 92, 100 28, 64 5, 59, 62
PointFigureChart Quantile Rational
377 163, 191 42, 45, 46
PointSize QuantilePlot Rationalize
31, 100, 147 381 81
PoissonDistribution Quantity Rationals
170 2, 143, 257 42
Polygon QuantityMagnitude ReactionBalance
232, 236 9, 124, 133 135
PolyhedronData QuantityVariable Real
113 116, 281 42, 46, 78
PopupMenu Quiet RealDigits
97 296, 298, 366 239, 242, 243
444 Mathematica beyond mathematics

Reals Round Slider

42, 78, 306 80, 297 96, 97
Reap RoundingRadius Slider2D
397 339, 347 97
Rectangle Row Slot
108, 210 6,97, 127 28, 64-66
Refresh Rule SlotSequence
36 56 66
RegionFunction RuleDelayed Small
32, 205, 394 57 82, 95, 99
RegionPlot3D RuntimeAttributes SmoothDensityHistogram
32 425 182, 183
ReliefPlot SameQ SmoothHistogram3D
144 419 183
Remove SameTest SmoothKernelDistribution
16, 71 423 177, 191
RemoveBackground SaveDefinitions Solve
103 94, 131 8, 22, 31
RenkoChart Scrollbars SolveAlways
11 427 342
Replace SeedRandom Sort
20 52, 189, 215 78, 134, 178
ReplacePart Select SortBy
49 65, 128, 279 142
Rescale SemanticImport Sow
174 190, 490 397
ResourceData SemanticInterpretation Spacings
127, 128, 165 121, 124 97
ResourceFunction SequencePosition SpanFromLeft
29, 93, 129 245 121
ResourceObject Set (=) SparseArray
127, 128, 165 55 43
ResourceSearch SetDelayed (:=) Spatial Statistics
29 55 204
ResourceUpdate Shallow Speak
128 118, 125, 163 127
Rest Short Spectrogram
64, 155 118 107
RGBColor Show Specularity
66, 215 31, 82, 95 33
RiemannR Show Expression Sphere
253, 254 42 33, 233, 234
RiemannSiegelZ Simplify SphericalRegion
250 22, 26, 237 110, 290
Rotate Skewness Split
24 163, 170 244, 245
Index 445

SplitBy SynchronousUpdating TimeZone

142 376 36, 260
SpokenString SystemOpen Timing
126 160, 198 54
StandardDeviation Table Tiny
163, 170, 187 31, 84, 107 36, 93, 95
StandardForm TableForm Tooltip
6, 15 155, 163, 174 33, 129, 156
StarData TableHeadings ToString
125 187, 193, 244 77, 162, 389
StepMonitor TableView Total
358, 397 199, 201 59, 68
StoppingPowerData TabView Trace
328, 329 53, 389 45, 58
String (" …") Take TrackedSymbols
42, 309 50, 124, 155 379
StringCount TakeSmallest TraditionalForm
28 277 7, 15, 47
StringForm Tally TransformationFunction
92, 95 244 109
StringJoin TextSentences TransformedDistribution
162 151, 169, 186 179
StringReplace ThermometerGauge Transpose
427 36 55, 95, 115
StringSplit Thickness TreeForm
127 388 44, 45, 47
StringTake Thinning True
429 108 72-79
style Thread Tube
5, 6 51, 63, 149 113
Style Through UniformDistribution
36, 82, 86 63, 188, 191 180
StyleBox Ticks Union
42 236, 249, 251 34, 211, 311
Stylesheet TimelinePlot UnitConvert
12, 13 122 205, 257, 266
Suggestions Bar Times UniverseModelData
5, 6, 22 62, 63 286, 287
SunPosition TimeSeries UnsameQ
260 167, 203 310
Sunrise TimeSeriesForecast UpdateInterval
258 374 36, 258, 259
Symbol TimeSeriesModelFit URLFetch
42, 55, 77 203, 204, 374 379
Symbol (element) TimeValue Values
303, 308, 310 381 26, 50, 51
446 Mathematica beyond mathematics

Variance
170
VerticalGauge
36
ViewPoint
251, 290
WaitAll
419, 430
WeatherData
131, 175, 176
WeatherForecastData
131
WhenEvent
329, 365
Which
73, 74, 355
WikipediaData
38, 151, 169
WindVectorData
131
With
71, 72, 107
Wolfram Cloud
18, 86
Wolfram Data Framework
117
Wolfram Data Repository
52, 127, 165
WolframAlpha
9, 119, 114
WordData
127
WorkingPrecision
84, 180
Zeta
247-253
ZetaZero
250
ZoomFactor
100