Lecture 14: M/O/F/ for
Engineering Applications -
Part 2
BJ Furman
28NOV2011
The Plan for Today
Matlab/Octave/FreeMat (M/O/F) for
engineering applications – part 2
Recap M/O/F key concepts
Element-by-element operations (dot operator)
Function examples
2D graphs
File IO
Resources for more information
Learning Objectives
Distinguish between matrix and array
arithmetic, and use them appropriately
Explain the differences between script
files and functions
Explain the basic elements of creating a
2D graph
Explain how to read from and write to
files
Last Lecture
Overview of Matlab/Octave
Useful commands
The array as a fundamental element
creating arrays
indexing elements
vectors
colon operator
linspace
logspace
extracting sub-arrays using indexing and the colon operator
special matrices
ones(), zeros(), diag(), eye()
Introduction to plotting in M/O/F via script file
comparison to Ch and Excel
Arrays, Vectors, and Matrices
Array :
A collection of data elements
arranged in rows and columns
Vector:
A one-dimensional array >> A=[1:9] row or column?
(single row or single column
of data elements) >> B=[1:9]’ column vector
Matrix :
A two-dimensional array >> C=[1:3; 4:6; 7:9]
(more than one row and/or
C=
column of data elements) 1 2 3
4 5 6
7 8 9
Colon Operator vs. linspace()
What will the following do?
D=0 : 3 : 10 base : increment : limit
F=linspace(0, 10, 11) start : end : n
Observations about the two methods:
both methods produce vectors with equally spaced
elements
colon operator method allows you to specify the first
element and the interval spacing between elements,
but not the number of elements
If just start : end, then spacing is 1
linspace method allows you to specify the first and
last elements of the vector, but not the spacing
between elements
Review of length() and size()
What will the following do?
length(ones(1,3))
length returns the number of elements of the vector
size(zeros(2,3))
size returns the size of the dimensions of its
argument
Review of Array Manipulation
Given G=[2 4 6; 8 10 12], what is:
G(2,3)
G(: , 2)
G(4)
G(1,1)=0
G(1, :)=0
Observations:
Array indexing begins with 1 (contrast with C)
: means “all of” the elements in that dimension
Extract elements by indexing
Extract sub-arrays using vectors as the indexing
arguments
More Array Manipulation
Suppose H=1:9 (what will this produce?)
How could we form into a 3x3 matrix?
Reshape function
I=reshape(H,3,3)
How to transpose the rows to be the columns? I = I’
Reverse the order of the rows of I
J=I(3:-1:1,:)
Reverse the order of the columns of I (how?)
K=I(:, 3:-1:1)
Reverse the order of all the elements of I (how?)
Pseudocode:
Index I in reverse order
L = reshape(I(9:-1:1),3,3)
Reshape
Matrix and Array Arithmetic
Arithmetic operators:
+ - * / \ ^ ’
add. sub. mult. right div. left div. expon. algebr. transpose
addition and subtraction are done element-by-element (same for
matrix and array arithmetic)
Unless one is a scalar, the operands must be of the same size
scalar (matrix or array) --> ?
(matrix or array) (matrix or array) --> ?
(matrix or arrays) must be of the same size
For the other operators, need to distinguish between matrix and array
operation
n columns B
Matrix arithmetic operations
mn
per rules of linear algebra
rows and columns must conform A n rows n p
For example, A x B: must have column and row agreement
Array arithmetic operations
element-by-element
Denote with dot operator: .* ./ .\ .^ .’ (array transpose)
Matrix and Array Arithmetic Examples
Scalar and matrix operands
If L=ones(1,5) and M=ones(1,4)
N = 2*L --> ?
N–1 --> ?
Non-scalar operations
If O = [ 1:5 ]
O + M --> ? ??? Error using ==> +
Matrix dimensions must agree.
??? Error using ==> * (1x5 * 1x5 does not work!)
L * O --> ? Inner matrix dimensions must agree.
L* O’ --> ? 15 Same as sum(L .* O)
(1x5 * 5x1 works! Inner matrix dimensions agree. Results in a 1x1)
Array Operations
Element-by-element array operation
Ex: Given a set of distances and times, calculate
average speeds and maximum of averages
How would you do this in C?
Pseudocode:
Calculate avg. speeds: speed[i] = distance[i] / time[i], for i=1 to 4
Determine maximum speed
M/O/F (vectorize!):
distances=[120, 213, 87, 35] (in miles)
times=[ 2, 3.8, 0.9, 0.6] (in hours)
speeds=distances ./ times %( note: ‘dot /’ divide element-by-element)
max_speed=max(speeds)
To get the maximum speed and its index:
[max_speed, i] = max(speeds)
Circuit Analysis Equations i2
i1
R2
i1 i2 i3 0
Matrix operations +V 0i1 0i2 i3 R1 V
R3
i3
R1
Matrix division 0i1 R2 R3 i2 0i3 V
Recall the circuit analysis
R1=10k
1 1 1 i1 0
R2=R3=5k 0 0 R1 i2 V
V=10V
Matrix solution 0 R2 R3 0 i3 V
1
Ri V R 1Ri R 1V i R V R i V
use ' left' division to solve for i
iR\V Think of it like inverting R and multiplying on the left side of V
If we had iR = V instead, we’d use ‘right’ division to solve for i: ( i = R / V )
Think of it like inverting R and multiplying on the right side of V: i = VR-1
Circuit Analysis Solution
Circuit analysis solution: •R1=10k
Build R, build V, solve for i •R2=R3=5k
Build R •V=10V
all at once 1 1 i1 0
1
R=[1 -1 1; 0 0 10e3; 0 10e3 0] 0
or 0 R1 i2 V
build by rows and combine 0 R2 R3 0 i3 V
eq1 = [ 1 -1 1]
eq2 = [0 0 10e3]
eq3 = [0 10e3 0]
R i V
R = [eq1; eq2; eq3]
Build V
V = [0 10 10]’ (note: transposed)
Solve I = R \ V
I=R\V
Dot Product Example
Another example of element-by-element
operations
v1
dot product of two vectors v1 v2 v1 v2 cos( )
v2
v1 3iˆ 2 ˆj 5kˆ v2 2iˆ 4 ˆj 10kˆ
iˆ, ˆj , kˆ are unit vectors for a cartesian coordinate system
their coefficients are called ' measure numbers'
what is v1 v2?
v1 v2 (3)(2) (2)(4) (5)(10) 6 8 50 52
Dot Product Function Development
Define the problem
Create a function that will take two vectors as arguments and
will return their vector dot product
Inputs
v1, v2 (three-element row vectors)
Outputs
z (the dot product)
Algorithm
Multiply v1 and v2 element-by-element
Sum the element-by-element products
Return the sum Try it in Matlab/Octave :
v1 [3 2 5] v 2 [2 4 10]
sum(v1 . * v 2)
Dot Product Function in M/O/F
Write the function
function [z] = dot_prod(v1, v2)
% dot_prod(v1,v2) computes the vector dot product between vectors v1 and v2
% Function dot_prod(v1,v2) computes and returns the vector dot product between
vectors v1 and v2
z = sum(v1.*v2);
Test it out
A = [ 1 2 3 ];
B = [ 4 5 6 ]; % what should A dot B result in?
A_dot_B = dot_prod(A,B)
Review of Functions
Functions
Like script M-files, but several differences:
first line (function declaration) must be of the form:
function [output args] = function_name(input args)
variables generated in the function are local to the
function, whereas for script files, variables are global
must be named, ‘function_name.m’ (same as file name)
Make sure you add comments at the start that
describe what the function does (see example code)
Example: root-mean-square function, N
rms.m x2
i
Given, x [ x1 , x2 ,..., x N ] RMS i 1
N
Root Mean-Square Function Development
Functions, cont. N
Example: root-mean-
square function, cont.
i 1
xi2
Given, x [ x1 , x2 ,..., x N ] RMS
Pseudocode: N
square each element of
x
sum the squares Take the square root
divide by N
take the square root
rms = sqrt(ms)
Square each element Before you write the
xs = x .^2 function, make sure the
Sum the squares name you propose is
sums = sum(xs) not already used!
Divide by N Use: which name
N = length(x)
to check
ms = sums/N
Root Mean-Square Function Implementation
Functions, cont.
Example: root-mean-square
function, cont.
function [y] = rms(v)
H1 comment line
% RMS(v) root mean square of the elements of the column vector v
(used in lookfor) % Function rms(v) returns the root mean square of the elements
Comments that % of the column vector, v
will be displayed
by help command
vs = v.^2; % what does this line do? Also note semicolon.
s = length(v);
y = sqrt(sum(vs)/s);
Let v=sin([0: 0.01*pi: 2*pi]’), one period of a sine wave. The RMS value
of a sine wave is its amplitude*1/sqrt(2)
Does rms() work with a row vector? How about a matrix?
More Robust Root Mean-Square Function
Functions,
function [rmsout] = rms2(v)
cont. %RMS2(v) Root mean square of v
% Function rms2(v) returns a row vector, where
Make rms % each element is the rms value of values in each
function more % column of v
robust vs = v.^2;
s = size(v);
to work with rmsout = sqrt(sum(vs,1)/s(1));
row or column
vector or
matrix with
column vectors
of data
File I/O with M/O/F
Data Input - simplest method
load command
Ex: load(‘data_file.txt’)
reads on a row-by-row basis
data values separated by spaces or commas and
rows terminated by new line
columns must have the same number of elements
data is stored in workspace in an array with same
name as the argument used in the load function
Ex. Portland International Airport monthly rainfall
load (‘[Link]’) % must be in search path!
File I/O with M/O/F, cont.
Data Output - simplest method
Save command
Ex: save(‘data_file_name’)
Saves all the variables into a .mat file named
‘data_file_name’
Many other commands are available for special
purpose file I/O
File I/O and Plotting Example
% read data into PDXprecip matrix
load('[Link]');
% copy first column of PDXprecip into month
month = PDXprecip(:,1);
% and second column into precip
precip = PDXprecip(:,2);
% plot precip vs. month with circles
plot(month,precip,'o');
% add axis labels and plot title
xlabel('month of the year');
ylabel('mean precipitation (inches)');
title('Mean monthly precipitation at Portland International Airport');
file_io_example.m
Adapted from: [Link]
visited 15NOV2009
More on Plotting General Format:
plot (x, y, fmt, ...)
Add a red line through the
% plot precip vs. month with circles
data
plot(month,precip,'o',month,precip,'-r');
Plot multiple sets of data
on a single graph and add % copy first column of PDXtemperature into month
month = PDXtemperature(:,1);
a legend
% and second column into high_temp
grid on high_temp = PDXtemperature(:,2);
Sub-plots % and third column into low temp
Format: low_temp = PDXtemperature(:,3);
subplot (m,n,p) % and fourth column into avg temp
Figure window divided avg = PDXtemperature(:,4);
into m x n matrix of % generate the plot
plotting areas plot(month,high_temp,'ko',month,low_temp,'k+',month,avg,‘r-');
Procedure:
% add axis labels and plot title
Pick the sub-plot xlabel('Month');
window ylabel('temperature (degrees F)');
Execute plot title('Monthly average temperature for PDX');
commands for that
sub-plot % add a plot legend using labels read from the file
legend('High','Low','Avg');
multi_plot.m
Vector Dot Product Example
Find the X and Y components of the vector, V
that is, find v x and v y , so that v x v y v
Y v
vy
v
ĵ vx
iˆ X
ˆ v iˆ cos( ) v (1) cos( ) v cos( )
vx v i x x x
v y v ˆj v y ˆj cos(90 ) v y (1) cos(90 ) v y sin( )
Back
Review
References
Matlab. (2009, November 6). In Wikipedia, the free encyclopedia.
Retrieved November 6, 2009, from
[Link]
Matlab tutorials:
[Link]
GNU Octave. (2009, October 31). In Wikipedia, the free
encyclopedia. Retrieved November 6, 2009, from
[Link]
Octave main page: [Link]
([Link] access to pre-built installers)
Octave tutorials: [Link]
[Link]
FreeMat. [Link]
[Link]
ChabotEngineeringCourses/[Link]
