Tensor Calculus

Made Simple
δil δim δin
δ jl δ jm δ jn 3 1
δkl δkm δkn +1
δ jl
m
∂i δ jm
− δ i 2

jk
ε lm
k
= δ il
Σ ∇ f
εi

Taha Sochi
Preface
This book is prepared from personal notes and tutorials about tensor calculus at an intro-
ductory level. The language and method used in presenting the ideas and techniques of
tensor calculus make it very suitable for learning this subject by the beginners who have
not been exposed previously to this elegant discipline of mathematics. Yes, some gen-
eral background in arithmetic, elementary algebra, calculus and linear algebra is needed
to understand the book and follow the development of ideas and techniques of tensors.
However, we made considerable efforts to reduce this dependency on foreign literature by
summarizing the main items needed in this regard to make the book self-contained. The
book also contains a number of graphic illustrations to aid the readers and students in
their effort to visualize the ideas and understand the abstract concepts. In addition to
the graphic illustrations, we used illustrative techniques such as highlighting key terms by
boldface fonts.
The book also contains sets of clearly explained exercises which cover most of the ma-
terials presented in the book where each set is given in an orderly manner in the end of
each chapter. These exercises are designed to provide thorough revisions of the supplied
materials and hence they make an essential component of the book and its learning objec-
tives. Therefore, they should not be considered as a decorative accessory to the book. We
also populated the text with hyperlinks, for the ebook users, to facilitate referencing and
connecting related objects so that the reader can go forth and back with minimum effort
and time and without compromising the continuity of reading by interrupting the chain
of thoughts.
The book is also furnished with a rather detailed index section which provides access to
many keywords and concepts throughout the book. Although this index, like any other
index, is not comprehensive, it is supposed to provide access to all the essential terms with

1
particular emphasis on the positions where the enlisted terms have a particular significance
in their context such as where they are defined or linked to key ideas. Despite the fact
that this index may be of little use to the ebook users who can conduct thorough searches
in the book for any term electronically by using the more convenient way of finding words
through “Find” function or similar functions in the document viewer, the index is composed
with the vision to be like a glossary and hence it serves two purposes. For that reason,
the index is used in the exercises as a reference for key terms and concepts of common use
in tensor calculus and related keywords from other disciplines of mathematics.
In view of all the above factors, the present text can be used as a textbook or as a
reference for an introductory course on tensor algebra and calculus or as a guide for self-
studying and learning. I tried to be as clear as possible and to highlight the key issues of
the subject at an introductory level in a concise form. I hope I have achieved some success
in reaching these objectives for the majority of my target audience.
Finally, I should make a short statement about credits in making this book following the
tradition in writing book prefaces. In fact everything in the book is made by the author
including all the graphic illustrations, front and back covers, indexing, typesetting, and
overall design. However, I should acknowledge the use of the LATEX typesetting package
and the LATEX based document preparation package LYX for facilitating many things in
typesetting and design which cannot be done easily or at all without their versatile and
powerful capabilities. I also used the Ipe extensible drawing editor program for making
all the graphic illustrations in the book as well as the front and back covers.
Taha Sochi
London, November 2016

2
Contents
Preface 1

Table of Contents 3

List of Figures 6

Nomenclature 7

1 Preliminaries 8
1.1 Historical Overview of Development & Use of Tensor Calculus . . . . . . . 8
1.2 General Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3 General Mathematical Background . . . . . . . . . . . . . . . . . . . . . . 15
1.3.1 Coordinate Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3.2 Vector Algebra and Calculus . . . . . . . . . . . . . . . . . . . . . . 23
1.3.3 Matrix Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2 Tensors 47
2.1 General Background about Tensors . . . . . . . . . . . . . . . . . . . . . . 47
2.2 General Terms and Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.3 General Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.4 Examples of Tensors of Different Ranks . . . . . . . . . . . . . . . . . . . . 57
2.5 Applications of Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.6 Types of Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.6.1 Covariant and Contravariant Tensors . . . . . . . . . . . . . . . . . 59
2.6.2 True and Pseudo Tensors . . . . . . . . . . . . . . . . . . . . . . . . 65

3
 2.6.3 Absolute and Relative Tensors . . . . . . . . . . . . . . . . . . . . . 68
2.6.4 Isotropic and Anisotropic Tensors . . . . . . . . . . . . . . . . . . . 70
2.6.5 Symmetric and Anti-symmetric Tensors . . . . . . . . . . . . . . . . 70
2.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

3 Tensor Operations 83
3.1 Addition and Subtraction . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.2 Multiplication of Tensor by Scalar . . . . . . . . . . . . . . . . . . . . . . . 84
3.3 Tensor Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.4 Contraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.5 Inner Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.6 Permutation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.7 Tensor Test: Quotient Rule . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

4 δ and  Tensors 96
4.1 Kronecker δ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.2 Permutation  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.3 Useful Identities Involving δ or/and  . . . . . . . . . . . . . . . . . . . . . 101
4.3.1 Identities Involving δ . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.3.2 Identities Involving  . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.3.3 Identities Involving δ and  . . . . . . . . . . . . . . . . . . . . . . . 106
4.4 Generalized Kronecker δ . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

5 Applications of Tensor Notation and Techniques 116

5.1 Common Definitions in Tensor Notation . . . . . . . . . . . . . . . . . . . 116
5.2 Scalar Invariants of Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . 119

4
 5.3 Common Differential Operations in Tensor Notation . . . . . . . . . . . . . 120
5.3.1 Cartesian Coordinate System . . . . . . . . . . . . . . . . . . . . . 121
5.3.2 Cylindrical Coordinate System . . . . . . . . . . . . . . . . . . . . . 124
5.3.3 Spherical Coordinate System . . . . . . . . . . . . . . . . . . . . . . 125
5.3.4 General Orthogonal Coordinate System . . . . . . . . . . . . . . . . 126
5.4 Common Identities in Vector and Tensor Notation . . . . . . . . . . . . . . 127
5.5 Integral Theorems in Tensor Notation . . . . . . . . . . . . . . . . . . . . . 132
5.6 Examples of Using Tensor Techniques to Prove Identities . . . . . . . . . . 134
5.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

6 Metric Tensor 149

6.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

7 Covariant Differentiation 155

7.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

References 162

Index 163

5
List of Figures
1 Cartesian coordinate system and its basis vectors with components of a vector 17
2 Cylindrical coordinate system and its basis vectors . . . . . . . . . . . . . . . 18
3 Plane polar coordinate system and its basis vectors . . . . . . . . . . . . . . . 20
4 Spherical coordinate system and its basis vectors . . . . . . . . . . . . . . . . 21
5 General curvilinear coordinate system and its covariant basis vectors . . . . . 22
6 Dot product of two vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
7 Cross product of two vectors and right hand rule . . . . . . . . . . . . . . . . 25
8 Scalar triple product of three vectors . . . . . . . . . . . . . . . . . . . . . . . 27
9 Gradient of a scalar field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
10 Illustration of Stokes integral theorem with the right hand twist rule . . . . . 34
11 Illustration of a rank-3 tensor in a 3D space . . . . . . . . . . . . . . . . . . . 48
12 Proper and improper transformations of a Cartesian coordinate system . . . . 52
13 Unit dyads associated with the double directions of rank-2 tensors in a 3D space 59
14 Reciprocity relation between the covariant and contravariant basis vectors . . 65
15 Representation of a vector in covariant and contravariant basis vector sets . . 66
16 The behavior of true and pseudo vectors on reflecting the coordinate system . 67
17 Illustration of rank-3 permutation tensor ijk . . . . . . . . . . . . . . . . . . . 98
18 The cyclic nature of the even and odd permutations of the indices of ijk . . . 99
19 Mnemonic device for the sequence of indices in the epsilon-delta identity . . . 108

6
Nomenclature
In the following table, we define some of the common symbols, notations and abbreviations
which are used in the book to avoid ambiguity and confusion.

∇ nabla differential operator

∇f gradient of scalar f
∇·A divergence of vector A
∇×A curl of vector A
∇2 or ∆ Laplacian operator
⊥ perpendicular to
2D, 3D, nD two-dimensional, three-dimensional, n-dimensional
det determinant of matrix
Ei ith covariant basis vector
Ei ith contravariant basis vector
Eq./Eqs. Equation/Equations
hi scale factor for ith coordinate in general orthogonal system
iff if and only if
r, θ, φ coordinates of spherical system in 3D space
tr trace of matrix
u1 , u2 , u3 coordinates of general orthogonal system in 3D space
x1 , x2 , x3 labels of coordinate axes of Cartesian system in 3D space
X1 , X2 , X3 same as the previous entry
x1 , x2 , x3 coordinates of general curvilinear system in 3D space
x, y, z coordinates of points in Cartesian system in 3D space
ρ, φ, z coordinates of cylindrical system in 3D space

7
Chapter 1

Preliminaries

In this introductory chapter, we provide the reader with a general overview about the
historical development of tensor calculus and its role in modern mathematics, science
and engineering. We also provide a general set of notes about the notations and conven-
tions which are generally followed in the writing of this book. A general mathematical
background about coordinate systems, vector algebra and calculus and matrix algebra
is also presented to make the book, to some extent, self-sufficient. Although the general
mathematical background section is not comprehensive, it contains essential mathematical
terminology and concepts which are needed in the development of the ideas and methods
of tensor calculus in the subsequent chapters of the book.

1.1 Historical Overview of Development & Use of Tensor Calculus

First, we should remark that some of the following historical statements represent approx-
imate rather than exact historical facts due to the reality that many of the first hand
historical records are missing or not available to the author. Moreover, many ideas, termi-
nology, notation and techniques of tensor calculus, like any other field of knowledge and
practice, have been developed gradually over long periods of time although the credit is
usually attributed to a few individuals due to their prominence and fame or because they
played crucial and distinctive roles in the creation and development of the subject.
It is believed that the word “tensor” was coined by Hamilton but he used it in a
rather different meaning to what is being used for in modern mathematics and science.

8
1.1 Historical Overview of Development & Use of Tensor Calculus 9

The credit for attaching this term to its modern technical meaning, approximately in
the late nineteenth century, is usually given to Voigt. Apparently, the term “tensor”
was originally derived from the Latin word “tensus” which means tension or stress since
one of the first uses of tensors (in whatever meaning) was related to the mathematical
description of mechanical stress.
The names “tensor calculus” or “tensor analysis” have been used to label this sub-
ject in its modern form rather recently, probably in the second quarter of the twenties
century. The early forms of this subject have been called “Ricci calculus” or “absolute
differential calculus”. The latter names may still be found in the modern literature of
tensor calculus.
Many mathematicians and scientists have contributed to the development of tensor cal-
culus directly or indirectly. However, numerous components of the modern tensor calculus
were not developed as such and for the purpose of the theory of tensors but as parts or
byproducts of other disciplines, notably the differential geometry of curves and surfaces.
This generally applies prior to the official launch of tensor calculus as an independent
branch of mathematics by Ricci and Levi-Civita who are commonly recognized as the
founding fathers of this discipline.
Several concepts and techniques of tensor calculus have been developed by Gauss and
Riemann in the nineteenth century, mainly as part of their efforts to develop and for-
mulate the theory of differential geometry of curves and surfaces. Their contributions are
highlighted by the fact that many concepts, methods and equations, which are related
directly to tensor calculus or to subjects with a close affinity to tensor calculus, bear their
names, e.g. Gaussian coordinates, Gauss curvature tensor, Gauss-Codazzi equations, Rie-
mannian metric tensor, Riemannian manifold, and Riemann-Christoffel curvature tensor.
A major player in the development of tensor calculus is Christoffel whose contribution
is documented, for instance, in the Christoffel symbols of the first and second kind which
1.1 Historical Overview of Development & Use of Tensor Calculus 10

infuse throughout the whole subject of tensor calculus and play very essential roles. Also,
the above mentioned Riemann-Christoffel curvature tensor is another acknowledgment of
his achievements in this respect. Bianchi has also contributed a number of important
ideas and techniques, such as Bianchi identities, which played an important role in the
subsequent development of this subject.
As indicated above, the major role in the creation and development of tensor calculus
in its modern style is attributed to Ricci and Levi-Civita in the end of the nineteenth
century and the beginning of the twentieth century where these mathematicians extracted
and merged the previous collection of tensor calculus notations and techniques and intro-
duced a number of new ideas, notations and methods and hence they gave the subject
its modern face and contemporary style. For this reason, they are usually accredited for
the creation of tensor calculus as a standalone mathematical discipline although major
components of this subject have been invented by their predecessors as indicated above.
The role of Ricci and Levi-Civita in the creation and renovation of tensor calculus is
documented in terms like Ricci calculus, Ricci curvature tensor, Levi-Civita symbol, and
Levi-Civita identity.
Many other mathematicians and scientists have made valuable indirect contributions to
the subject of tensor calculus by the creation of mathematical ideas which subsequently
played important roles in the development of tensor calculus although they are not related
directly to tensors. A prominent example of this category is Kronecker and his famous
Kronecker delta symbol which is embraced in tensor calculus and adapted for extensive
use in tensor notation and techniques.
The widespread use of tensor calculus in science has begun with the rise of the general
theory of relativity which is formulated in the language, methods and techniques of
tensor calculus. Tensor calculus has also found essential roles to play in a number of
other disciplines with a particular significance in differential geometry, continuum
1.2 General Conventions 11

mechanics and fluid dynamics.

Nowadays, tensor algebra and calculus propagate throughout many branches of mathe-
matics, physical sciences and engineering. The success of tensor calculus is credited to its
elegance, power and concision as well as the clarity and eloquence of its notation. Indeed,
tensor calculus is not just a collection of mathematical methods and techniques but it
is a compact and effective language for communicating ideas and expressing concepts of
varying degrees of complexity with rigor and reasonable simplicity.

1.2 General Conventions

In this book, we largely follow certain conventions and general notations; most of which
are commonly used in the mathematical literature although they may not be universally
approved. In this section, we outline the most important of these conventions and nota-
tions. We also give some initial definitions of the most basic terms and concepts in tensor
calculus; more thorough technical definitions will follow, if needed, in the forthcoming sec-
tions and chapters. It should be remarked that tensor calculus is riddled with conflicting
conventions and terminology. In this book we use what we believe to be the most common,
clear and useful of all of these.
Scalars are algebraic objects which are uniquely identified by their magnitude (abso-
lute value) and sign (±), while vectors are broadly geometric objects which are uniquely
identified by their magnitude (length) and direction in a presumed underlying space. At
this early stage in this book, we generically define tensor as an organized array of math-
ematical objects such as numbers or functions.
In generic terms, the rank of a tensor signifies the complexity of its structure. Rank-0
tensors are called scalars while rank-1 tensors are called vectors. Rank-2 tensors may
be called dyads although dyad, in common use, may be designated to the outer product
(see § 3.3) of two vectors and hence it is a special case of rank-2 tensors assuming it meets
1.2 General Conventions 12

the requirements of a tensor and hence transforms as a tensor. Like rank-2 tensors, rank-3
tensors may be called triads. Similar labels, which are much less common in use, may be
attached to higher rank tensors; however, none of these will be used in the present book.
More generic names for higher rank tensors, such as polyad, are also in use.
We may use the term “tensor” to mean tensors of all ranks including scalars (rank-0)
and vectors (rank-1). We may also use this term as opposite to scalar and vector, i.e.
tensor of rank-n where n > 1. In almost all cases, the meaning should be obvious from the
context. It should be remarked that in the present book all tensors of all ranks and types
are assumed to be real quantities, i.e. they have real rather than complex components.
Due to its introductory nature, the present book is largely based on assuming an under-
lying orthonormal Cartesian coordinate system. However, certain parts of the book are
based on other types of coordinate system; in these cases this is stated explicitly or made
clear by the notation and context. As will be outlined later (see § Orthonormal Cartesian
Coordinate System), we mean by “orthonormal” a system with mutually perpendicular
and uniformly scaled axes with a unit basis vector set. We may also use “rectangular” for
a similar meaning or to exclude oblique coordinate systems[1] which may also be labeled
by some as Cartesian. Some of the statements in this book in which these terms are used
may not strictly require these conditions but we add these terms to focus the attention on
the type of the coordinate system which these statements are made about.
To label scalars, we use non-indexed lower case light face italic Latin letters (e.g. f and
h), while for vectors, we use non-indexed lower or upper case bold face non-italic Latin
letters (e.g. a and A). The exception to this is the basis vectors where indexed bold face
lower or upper case non-italic symbols (e.g. e1 and Ei ) are used. However, there should be
no confusion or ambiguity about the meaning of any one of these symbols. As for tensors
of rank > 1, non-indexed upper case bold face non-italic Latin letters (e.g. A and B) are
[1]
Oblique coordinate systems are usually characterized by similar features as orthonormal Cartesian
systems but with non-perpendicular axes.
1.2 General Conventions 13

used.[2]
Indexed light face italic Latin symbols (e.g. ai and Bijk ) are used in this book to denote
tensors of rank > 0 in their explicit tensor form, i.e. index notation. Such symbols
may also be used to denote the components of these tensors. The meaning is usually
transparent and can be identified from the context if it is not declared explicitly. Tensor
indices in this book are lower case Latin letters, usually taken from the middle of the Latin
alphabet like (i, j, k). We also use numbered indices like (i1 , i2 , . . . , ik ) when the number
of tensor indices is variable. Numbers are also used as indices (e.g. 12 ) in some occasions
for obvious purposes such as making statements about particular components.
Mathematical identities and definitions are generally denoted in the mathematical
literature by using the symbol “≡”. However, for simplicity we use the equality sign “=”
to mark identities and mathematical definitions as well as normal equalities.
We use vertical bars (i.e. |·|) to symbolize determinants and square brackets (i.e.
[·]) to symbolize matrices. This applies where these symbols contain arrays of objects;
otherwise they have their normal meaning according to the context, e.g. bars embracing
vectors mean modulus of the vector. We normally use indexed square brackets (e.g.
[A]i and [∇f ]i ) to denote the ith component of vectors in their symbolic or vector notation.
For tensors of higher rank, more than one index is needed to denote their components,
e.g. [A]ij for the ij th component of a rank-2 tensor.
Partial derivative symbol with a subscript index (e.g. ∂i ) is frequently used to denote
the ith component of the gradient operator nabla ∇:[3]

∂
∂i = ∇ i = (1)
∂xi

A comma preceding a subscript index (e.g. , i) is also used to denote partial differen-
[2]
Since matrices in this book are supposed to represent rank-2 tensors, they also follow the rules of
labeling tensors symbolically by using non-indexed upper case bold face non-italic Latin letters.
[3]
This mainly applies in this book to Cartesian coordinate systems.
1.2 General Conventions 14

tiation with respect to the ith spatial coordinate, mainly in Cartesian systems, e.g.

∂A
A,i = (2)
∂xi

Partial derivative symbol with a spatial subscript, rather than an index, is used to denote
partial differentiation with respect to that spatial variable. For instance:

∂
∂r = ∇r = (3)
∂r

is used for the partial derivative with respect to the radial coordinate r in spherical coor-
dinate systems identified by the spatial variables (r, θ, φ).[4]
Partial derivative symbol with repeated double index is used to denote the Laplacian
operator:
∂ii = ∂i ∂i = ∇2 = ∆ (4)

The notation is not affected by using repeated double index other than i (e.g. ∂jj or ∂kk ).
The following notations:

∂ii2 ∂2 ∂i ∂ i (5)

are also used in the literature of tensor calculus to symbolize the Laplacian operator.
However, these notations will not be used in the present book.
We follow the common convention of using a subscript semicolon preceding a subscript
index (e.g. Akl;i ) to symbolize the operation of covariant differentiation with respect
to the ith coordinate (see § 7). The semicolon notation may also be attached to the
normal differential operators for the same purpose, e.g. ∇;i or ∂;i to indicate covariant
differentiation with respect to the variable indexed by i.

[4]
As indicated, in notations like ∂r the subscript is used as a label rather than an index.
1.3 General Mathematical Background 15

Finally, all transformation equations in the present book are assumed to be continuous
and real, and all derivatives are continuous in their domain of variables. Based on the
continuity condition of the differentiable quantities, the individual differential operators
in the second (and higher) order partial derivatives with respect to different indices are
commutative, that is:
∂i ∂j = ∂j ∂i (6)

1.3 General Mathematical Background

In this section, we provide a general mathematical background which is largely required to

understand the forthcoming materials about tensors and their algebra and calculus. This
mathematical background is intended to provide the essential amount needed for the book
to be self-contained, to some degree, but it is not comprehensive. Also some concepts and
techniques discussed in this section require more elementary materials which should be
obtained from textbooks at more basic levels.

1.3.1 Coordinate Systems

In generic terms, a coordinate system is a mathematical device used to identify the location
of points in a given space. In tensor calculus, a coordinate system is needed to define non-
scalar tensors in a specific form and identify their components in reference to the basis
set of the system. Hence, non-scalar tensors require a predefined coordinate system to be
fully identified.[5]
There are many types of coordinate system, the most common ones are: the orthonormal
Cartesian, the cylindrical and the spherical. A 2D version of the cylindrical system is the

[5]
There are special tensors of numerical nature, such as the Kronecker and permutation tensors, which
do not require a particular coordinate system for their full and unambiguous identification since their
components are invariant under coordinate transformations (the reader is referred to Chapter 4 for
details). However, this issue may be debated.
1.3.1 Coordinate Systems 16

plane polar system. The most general type of coordinate system is the general curvilinear.
A subset of the latter is the orthogonal curvilinear. These types of coordinate system are
briefly investigated in the following subsections.

A. Orthonormal Cartesian Coordinate System

This is the simplest and the most commonly used coordinate system. It consists, in its
simplest form, of three mutually orthogonal straight axes that meet at a common point
called the origin of coordinates O. The three axes, assuming a 3D space, are scaled
uniformly and hence they all have the same unit length. Each axis has a unit vector
oriented along the positive direction of that axis.[6] These three unit vectors are called the
basis vectors or the bases of the system. These basis vectors are constant in magnitude
and direction throughout the system.[7] This system with its basis vectors (e1 , e2 and e3 )
is depicted in Figure 1.
The three axes, as well as the basis vectors, are usually labeled according to the right
hand rule, that is if the index finger of the right hand is pointing in the positive direction
of the first axis and its middle finger is pointing in the positive direction of the second axis
then the thumb will be pointing in the positive direction of the third axis.

B. Cylindrical Coordinate System

The cylindrical coordinate system is defined by three parameters: ρ, φ and z which range
over: 0 ≤ ρ < ∞, 0 ≤ φ < 2π and −∞ < z < ∞. These parameters identify the
coordinates of a point P in a 3D space where ρ represents the perpendicular distance
from the point to the x3 -axis of a corresponding rectangular Cartesian system, φ represents
the angle between the x1 -axis and the line connecting the origin of coordinates O to the

[6]
As indicated before, these features are what qualify a Cartesian system to be described as orthonormal.
[7]
In fact, the basis vectors are constant only in rectangular Cartesian and oblique Cartesian coordinate
systems. As indicated before, the oblique Cartesian systems are the same as the rectangular Cartesian
but with the exception that their axes are not mutually orthogonal. Also, labeling the oblique as
Cartesian may be controversial.
1.3.1 Coordinate Systems 17

x3 x3

v3
v

e3
v2 x2
O O
e1 e2
v1

x1 x2 x1
Figure 1: Orthonormal right-handed Cartesian coordinate system and its basis vectors
e1 , e2 and e3 in a 3D space (left frame) with the components of a vector v in this system
(right frame).

perpendicular projection of the point on the x1 -x2 plane of the corresponding Cartesian
system, and z is the same as the third coordinate of the point in the reference Cartesian
system. The sense of the angle φ is given by the right hand twist rule, that is if the
fingers of the right hand curl in the sense of rotation from the x1 -axis towards the line of
projection, then the thumb will be pointing in the positive direction of the x3 -axis.
The basis vectors of the cylindrical system (eρ , eφ and ez ) are pointing in the direction
of increasing ρ, φ and z respectively. Hence, while ez is constant in magnitude and
direction throughout the system, eρ and eφ are coordinate-dependent as they vary in
direction from point to point. All these basis vectors are mutually perpendicular and
they are defined to be of unit length. Figure 2 is a graphical illustration of the cylindrical
coordinate system and its basis vectors with a corresponding reference Cartesian system
in a standard position.
The transformation from the Cartesian coordinates (x, y, z) of a particular point in
1.3.1 Coordinate Systems 18

x3

z
ez

eφ
P eρ

O
φ x2

x1
Figure 2: Cylindrical coordinate system, superimposed on a rectangular Cartesian system
in a standard position, and its basis vectors eρ , eφ and ez in a 3D space. The point P
in the figure is identified simultaneously by (x, y, z) coordinates in the Cartesian system
and by (ρ, φ, z) coordinates in the cylindrical system where these coordinates are related
through the two sets of Eqs. 7 and 8.

the space to the cylindrical coordinates (ρ, φ, z) of that point, where the two systems are
in a standard position, is performed through the following equations:[8]

p y
ρ= x2 + y 2 φ = arctan z=z (7)
x

while the opposite transformation from the cylindrical to the Cartesian coordinates is

[8] y


In the second equation, arctan x should be selected consistent with the signs of x and y to be in the
right quadrant.
1.3.1 Coordinate Systems 19

performed by the following equations:

x = ρ cos φ y = ρ sin φ z=z (8)

C. Plane Polar Coordinate System

The plane polar coordinate system is the same as the cylindrical coordinate system with
the absence of the z coordinate, and hence all the equations of the polar system are
obtained by setting z = 0 in the equations of the cylindrical coordinate system, that is:

p y
ρ = x2 + y 2 φ = arctan (9)
x

and
x = ρ cos φ y = ρ sin φ (10)

This system is illustrated graphically in Figure 3 where the point P is located by (ρ, φ) in
the polar system and by (x, y) in the corresponding 2D Cartesian system.

D. Spherical Coordinate System

The spherical coordinate system is defined by three parameters: r, θ and φ which range
over: 0 ≤ r < ∞, 0 ≤ θ ≤ π and 0 ≤ φ < 2π. These parameters identify the coordinates
of a point P in a 3D space where r represents the distance from the origin of coordinates
O to P , θ is the angle from the positive x3 -axis of the corresponding Cartesian system to
the line connecting the origin of coordinates O to P , and φ is the same as in the cylindrical
coordinate system.
The basis vectors of the spherical system (er , eθ and eφ ) are pointing in the direction
of increasing r, θ and φ respectively. Hence, all these basis vectors are coordinate-
dependent as they vary in direction from point to point. All these basis vectors are
1.3.1 Coordinate Systems 20

x2
e2
eφ
eρ
e1
P
y = ρ sin φ

φ
O x1
x = ρ cos φ

Figure 3: Plane polar coordinate system, superimposed on a 2D rectangular Cartesian

system in a standard position, and their basis vectors eρ and eφ for the polar system and
e1 and e2 for the Cartesian system. The point P in the figure is identified simultaneously
by (x, y) coordinates in the Cartesian system and by (ρ, φ) coordinates in the polar system
where these coordinates are related through the two sets of Eqs. 9 and 10.

mutually perpendicular and they are defined to be of unit length. Figure 4 is a

graphical illustration of the spherical coordinate system and its basis vectors with a cor-
responding reference rectangular Cartesian system in a standard position.
The transformation from the Cartesian coordinates (x, y, z) of a particular point in
the space to the spherical coordinates (r, θ, φ) of that point, where the two systems are in
a standard position, is performed by the following equations:[9]
!
p z y
r= x2 + y 2 + z 2 θ = arccos p φ = arctan (11)
x2 + y 2 + z 2 x

[9] y


Again, arctan x in the third equation should be selected consistent with the signs of x and y.
1.3.1 Coordinate Systems 21

x3

er
P eφ

r eθ
θ
O
φ x2

x1

Figure 4: Spherical coordinate system, superimposed on a rectangular Cartesian system

in a standard position, and its basis vectors er , eθ and eφ in a 3D space. The point P
in the figure is identified simultaneously by (x, y, z) coordinates in the Cartesian system
and by (r, θ, φ) coordinates in the spherical system where these coordinates are related
through the two sets of Eqs. 11 and 12.

while the opposite transformation from the spherical to the Cartesian coordinates is
performed by the following equations:

x = r sin θ cos φ y = r sin θ sin φ z = r cos θ (12)

E. General Curvilinear Coordinate System

The general curvilinear coordinate system is characterized by having coordinate axes which
1.3.1 Coordinate Systems 22

are curved in general. Also, its basis vectors are generally position-dependent and
hence they are variable in magnitude and direction throughout the system. Consequently,
the basis vectors are not necessarily of unit length or mutually orthogonal. A graphic
demonstration of the general curvilinear coordinate system at a particular point in the
space with its covariant basis vectors (see § 2.6.1) is shown in Figure 5.

x3

E3

x2
E2
E1

x1

Figure 5: General curvilinear coordinate system and its covariant basis vectors E1 , E2 and
E3 (see § 2.6.1) in a 3D space, where x1 , x2 and x3 are the labels of the coordinates.

F. General Orthogonal Curvilinear Coordinate System

The general orthogonal curvilinear coordinate system is a subset of the general curvilinear
coordinate system as described above. It is distinguished from the other subsets of the
general curvilinear system by having coordinate axes and basis vectors which are
mutually orthogonal throughout the system. The cylindrical and spherical systems are
examples of orthogonal curvilinear coordinate systems.
1.3.2 Vector Algebra and Calculus 23

1.3.2 Vector Algebra and Calculus

This subsection provides a short introduction to vector algebra and calculus, a subject that
is closely related to tensor calculus. In fact many ideas and methods of tensor calculus
find their precursors and roots in vector algebra and calculus and hence many concepts,
techniques and notations of the former can be viewed as extensions and generalizations of
their counterparts in the latter.

A. Dot Product of Vectors

The dot product, or scalar product, of two vectors is a scalar quantity which has two
interpretations: geometric and algebraic. Geometrically, the dot product of two vectors
a and b can be interpreted as the projection of a onto b times the length of b, or as
the projection of b onto a times the length of a, as demonstrated in Figure 6. In both
cases, the dot product is obtained by taking the product of the length of the two vectors
involved times the cosine of the angle between them when their tails are made to coincide,
that is:
a · b = |a| |b| cos θ (13)

where the dot between a and b on the left hand side of the equation stands for the dot
product operation, 0 ≤ θ ≤ π is the angle between the two vectors and the bars notation
means the modulus or the length of the vector.
Algebraically, the dot product is the sum of the products of the corresponding com-
ponents of the two vectors, that is:

a · b = a1 b1 + a2 b2 + a3 b3 (14)

where ai and bj (i, j = 1, 2, 3) are the components of a and b respectively. Here, we

are assuming an orthonormal Cartesian system in a 3D space; the formula can be easily
1.3.2 Vector Algebra and Calculus 24

|b| cos θ
|a|
a
θ θ θ
b |a| cos θ
|b|
Figure 6: Demonstration of the geometric interpretation of the dot product of two vectors
a and b (left frame) as the projection of a onto b times the length of b (middle frame) or
as the projection of b onto a times the length of a (right frame).

extended to an nD space, that is:

X
n
a·b= ai bi (15)
i=1

π
From Eq. 13, it is obvious that the dot product is positive when 0 ≤ θ < 2
, zero
π π
when θ = 2
(i.e. the two vectors are orthogonal), and negative when 2
< θ ≤ π. The
magnitude of the dot product is equal to the product of the lengths of the two vectors
when they have the same orientation (i.e. parallel or anti-parallel). Based on the above
given facts, the dot product is commutative, that is:

a·b=b·a (16)

B. Cross Product of Vectors

Geometrically, the cross product, or vector product, of two vectors, a and b, is a

vector whose length is equal to the area of the parallelogram defined by the two vectors
as its two main sides when their tails coincide and whose orientation is perpendicular
to the plane of the parallelogram with a direction defined by the right hand rule as
1.3.2 Vector Algebra and Calculus 25

demonstrated in Figure 7. Hence the cross product of two vectors a and b is given by:

a × b = (|a| |b| sin θ) n (17)

where 0 ≤ θ ≤ π is the angle between the two vectors when their tails coincide and n is
a unit vector perpendicular to the plane containing a and b and is directed according to
the right hand rule.
From Eq. 17, it can be seen that the cross product vector is zero when θ = 0 or θ = π,
i.e. when the two vectors are parallel or anti-parallel respectively. Also, the length of the
cross product vector is equal to the product of the lengths of the two vectors when the
two vectors are orthogonal (i.e. θ = π2 ).

|a × b| = a×b
|a| |b| sin θ a×b

a
b
a θ b

Figure 7: Graphical demonstration of the cross product of two vectors a and b (left frame)
with the right hand rule (right frame).

Algebraically, the cross product of two vectors a and b is expressed by the following
determinant (see § Determinant of Matrix) where the determinant is expanded along its
1.3.2 Vector Algebra and Calculus 26

first row, that is:

e1 e2 e3
a×b= a1 a2 a3 = (a2 b3 − a3 b2 ) e1 + (a3 b1 − a1 b3 ) e2 + (a1 b2 − a2 b1 ) e3 (18)

b1 b2 b3

Here, we are assuming an orthonormal Cartesian coordinate system in a 3D space with a

basis vector set e1 , e2 and e3 .
Based on the above given facts, since the direction is determined by the right hand rule
the cross product is anti-commutative, that is:

a × b = −b × a (19)

C. Scalar Triple Product of Vectors

The scalar triple product of three vectors (a, b and c) is a scalar quantity defined by the
expression:
a · (b × c) (20)

where the dot and multiplication symbols stand respectively for the dot and cross product
operations of two vectors as defined above. Hence, the scalar triple product is defined
geometrically by:
a · (b × c) = |a| |b| |c| sin φ cos θ (21)

where φ is the angle between b and c while θ is the angle between a and b × c. The scalar
triple product is illustrated graphically in Figure 8. As there is no meaning of a cross
product operation between a scalar and a vector, the parentheses in the above equation
are redundant although they provide a clearer and more clean notation.
Now, since |b × c| (= |b| |c| sin φ) is equal to the area of the parallelogram whose two
1.3.2 Vector Algebra and Calculus 27

b×c

θ c

φ b
Figure 8: Graphic illustration of the scalar triple product of three vectors a, b and c. The
magnitude of this product is equal to the volume of the seen parallelepiped.

main sides are b and c, while |a| cos θ represents the projection of a onto the orientation
of b × c and hence it is equal to the height of the parallelepiped (refer to Figure 8),
the magnitude of the scalar triple product is equal to the volume of the parallelepiped
whose three main sides are a, b and c while its sign is positive or negative depending,
respectively, on whether the vectors a, b and c form a right-handed or left-handed system.
The scalar triple product is invariant to a cyclic permutation of the symbols of the
three vectors involved, that is:

a · (b × c) = c · (a × b) = b · (c × a) (22)

It is also invariant to an exchange of the dot and cross product symbols, that is:

a · (b × c) = (a × b) · c (23)
1.3.2 Vector Algebra and Calculus 28

Hence, from the three possibilities of the first invariance with the two possibilities of the
second invariance, we have six equal expressions for the scalar triple product of three
vectors.[10] The other six possibilities of the scalar triple product of three vectors, which
are obtained from the first six possibilities with the opposite cyclic permutations, are also
equal to each other for the same reason. However, they are equal in magnitude to the first
six possibilities but are different in sign.[11]
From the above interpretation of the scalar triple product as the signed volume of the
parallelepiped formed by the three vectors, it is obvious that this product is zero when the
three vectors are coplanar. This, of course, includes the possibility of being collinear.
The scalar triple product of three vectors is also defined algebraically as the determi-
nant (refer to § Determinant of Matrix) of the matrix formed by the components of the
three vectors as its rows or columns in the given order, that is:

a1 a2 a3 a1 b 1 c 1
a · (b × c) = det(a, b, c) = b1 b2 b 3 = a2 b 2 c 2 (24)

c1 c2 c3 a3 b 3 c 3

D. Vector Triple Product of Vectors

The vector triple product of three vectors (a, b and c) is a vector quantity defined by
the following expressions:

a × (b × c) or (a × b) × c (25)

where the multiplication symbols stand for the cross product operation of two vectors as

[10]
This equality can be explained by the fact that the magnitude of the scalar triple product is equal
to the volume of the parallelepiped whereas the cyclic permutation preserves the sign of these six
possibilities.
[11]
The stated facts about the other six possibilities can be explained by the two invariances (as explained
in the previous footnote) plus the fact that the cross product operation is anti-commutative.
1.3.2 Vector Algebra and Calculus 29

defined previously. Vector triple product is not associative, so in general we have:[12]

a × (b × c) 6= (a × b) × c (26)

E. Differential Operations of nabla Operator[13]

There are many differential operations that can be performed on scalar and vector fields.[14]
Here, we define the most important and widely used of these operations that involve the
nabla ∇ differential operator, that is the gradient, divergence and curl. We also define
the Laplacian differential operator which is based on a combination of the gradient and
divergence operations. The definitions here are given in Cartesian coordinates only. More
definitions of these operations and operators will be given in § 5.3 in terms of tensor
notation for the Cartesian, as well as the equivalent definitions for some non-Cartesian
coordinate systems.
The differential vector operator nabla ∇ is defined in rectangular Cartesian coordinate
systems by the following expression:

∂ ∂ ∂
∇= i+ j+ k (27)
∂x ∂y ∂z

where i, j and k are the unit vectors in the x, y and z directions respectively.
The gradient of a scalar field f (x, y, z) is a vector defined, in Cartesian coordinate

[12]
Associativity here should be understood in its context as stated by the inequality.
[13]
Although we generally use (x, y, z) for the Cartesian coordinates of a particular point in the space
while we use (x1 , x2 , x3 ) to label the axes and the coordinates of the Cartesian system in general, in
this subsection we use (x, y, z), instead of (x1 , x2 , x3 ), because it is more commonly used in vector
algebra and calculus and is notationally clearer, especially at this level and at this stage in the book.
We also label the basis vectors of the Cartesian system with (i, j, k), instead of (e1 , e2 , e3 ), for similar
reasons.
[14]
As we will see later in the book, similar differential operations can also be defined and performed on
higher rank tensor fields.
1.3.2 Vector Algebra and Calculus 30

systems, by:
∂f ∂f ∂f
∇f = i+ j+ k (28)
∂x ∂y ∂z

Geometrically, the gradient of a scalar field f (x, y, z), at any point in the space where
the field is defined, is a vector normal to the surface f (x, y, z) = constant (refer to Figure
9) pointing in the direction of the fastest increase in the field at that point.

x3 ∇f

r dr

f = constant

x1 x2
Figure 9: The gradient of a scalar field f (x, y, z) as a vector normal to the surface
f (x, y, z) = constant pointing in the direction of the fastest increase in the field at that
point.

The gradient operation is distributive but not commutative or associative, that is:

∇ (f + h) = ∇f + ∇h (29)

∇f 6= f ∇ (30)

(∇f ) h 6= ∇ (f h) (31)

where f and h are differentiable scalar functions of position.

1.3.2 Vector Algebra and Calculus 31

The divergence of a vector field v(x, y, z) is a scalar quantity defined as the dot
product of the nabla operator with the vector. Hence, in Cartesian coordinate systems it
is given by:
∂vx ∂vy ∂vz
∇·v = + + (32)
∂x ∂y ∂z

where vx , vy and vz are the components of v in the x, y and z directions respectively.

In broad terms, the physical significance of the divergence of a vector field is that it is a
measure of how much the field diverges or converges at a particular point in the space
where the field is defined. When the divergence of a vector field is identically zero, the
field is called solenoidal.
The divergence operation is distributive but not commutative or associative, that
is:

∇ · (A + B) = ∇ · A + ∇ · B (33)

∇ · A 6= A · ∇ (34)

∇ · (f A) 6= ∇f · A (35)

where A and B are differentiable vector functions of position.

The curl of a vector field v(x, y, z) is a vector defined as the cross product of the
nabla operator with the vector. Hence, in Cartesian coordinate systems it is given by
(refer to § Cross Product of Vectors):

i j k      
∂vz ∂vy ∂vx ∂vz ∂vy ∂vx
∇×v = ∂ ∂ ∂ = − i+ − j+ − k (36)
∂x ∂y ∂z ∂y ∂z ∂z ∂x ∂x ∂y
vx vy vz

Broadly speaking, the curl of a vector field is a quantitative measure of the circulation
or rotation of the field at a given point in the space where the field is defined. When the
1.3.2 Vector Algebra and Calculus 32

curl of a vector field vanishes identically, the field is called irrotational.

The curl operation is distributive but not commutative or associative, that is:

∇ × (A + B) = ∇ × A + ∇ × B (37)

∇ × A 6= A × ∇ (38)

∇ × (A × B) 6= (∇ × A) × B (39)

where A and B are differentiable vector functions of position.

The Laplacian[15] scalar operator ∇2 is defined as the divergence of the gradient
operator and hence it is given, in Cartesian coordinates, by:

   
2 ∂ ∂ ∂ ∂ ∂ ∂ ∂2 ∂2 ∂2
∇ =∇·∇= i+ j+ k · i+ j+ k = + + (40)
∂x ∂y ∂z ∂x ∂y ∂z ∂x2 ∂y 2 ∂z 2

The Laplacian can act on scalar, vector and tensor fields of higher rank. When the
Laplacian operates on a tensor (in its general sense which includes scalar and vector) it
produces a tensor of the same rank; hence the Laplacian of a scalar is a scalar, the
Laplacian of a vector is a vector, the Laplacian of a rank-2 tensor is a rank-2 tensor, and
so on.

F. Divergence Theorem

The divergence theorem, which is also known as Gauss theorem, is a mathematical

statement of the intuitive idea that the integral of the divergence of a vector field over
a given volume is equal to the total flux of the vector field out of the surface enclosing
the volume. Symbolically, the divergence theorem states that:

˚ ¨
∇ · A dτ = A · n dσ (41)
V S

[15]
This operator is also known as the harmonic operator.
1.3.2 Vector Algebra and Calculus 33

where A is a differentiable vector field, V is a bounded volume in an nD space enclosed by

a surface S, dτ and dσ are volume and surface elements respectively, and n is a variable
unit vector normal to the surface.
The divergence theorem is useful for converting volume integrals into surface in-
tegrals and vice versa. In many cases, this can result in a considerable simplification of
the required mathematical work when one of these integrals is easier to manipulate and
evaluate than the other, or even overcoming a mathematical hurdle when one of the inte-
grals cannot be evaluated analytically. Moreover, the divergence theorem plays a crucial
role in many mathematical proofs and theoretical arguments in mathematical and physical
theories.

G. Stokes Theorem

Stokes theorem is a mathematical statement that the integral of the curl of a vector
field over an open surface is equal to the line integral of the field around the perimeter
surrounding the surface, that is:

¨ ˆ
(∇ × A) · n dσ = A · dr (42)
S C

where A is a differentiable vector field, C symbolizes the perimeter of the surface S, dr

is a vector element tangent to the perimeter, and the other symbols are as defined in the
divergence theorem. The perimeter should be traversed in a sense related to the direction
of the normal vector n by the right hand twist rule, that is when the fingers of the right
hand twist in the sense of traversing the perimeter the thumb will point approximately in
the direction of n, as seen in Figure 10.
Similar to the divergence theorem, Stokes theorem is useful for converting surface
integrals into line integrals and vice versa, which is useful in many cases for reducing
the amount of mathematical work or overcoming technical and mathematical difficulties.
1.3.3 Matrix Algebra 34

x3 n

S dσ

C dr

x2

x1
Figure 10: Illustration of Stokes integral theorem (left frame) with the right hand twist
rule (right frame).

Stokes theorem is also crucial in the development of many proofs and theoretical arguments
in mathematics and science.

1.3.3 Matrix Algebra

There is a close relation between rank-2 tensors and square matrices where the latter
usually represent the former. Hence, there are many ideas, techniques and notations
which are common or similar between the two subjects. We therefore provide in this
subsection a set of short introductory notes about matrix algebra to supply the reader
with the essential terminology and methods of matrix algebra which are needed in the
development of the forthcoming chapters about tensor calculus.

A. Definition of Matrix

A matrix is a rectangular array of mathematical objects (mainly numbers or functions)

which is subject to certain rules in its manipulation and over which certain mathematical
1.3.3 Matrix Algebra 35

operations are defined. Hence, two indices are needed to define a matrix unambiguously
where the first index labels the rows while the second index labels the columns. A matrix
which consists of m rows and n columns is said to be an m × n matrix.
The elements or entries of a matrix A are usually labeled with light-face symbols similar
to the symbol used to label the matrix where each element is suffixed with two indices:
the first refers to the row number of the entry and the second refers to its column number.
Hence, for a matrix A the entry in its second row and fifth column is labeled A25 .
The two indices of a matrix are not required to have the same range since a matrix can
have different number of rows and columns. When the two indices have the same range,
the matrix is described as square matrix. Examples of matrices are:
 
   
 B11 
 A11 A12 A13     C11 C12 
   B    (43)
 21 
A21 A22 A23   C21 C22
B31

When the range of the row/column index has only one value (i.e. 1) while the range of the
other index has multiple values the matrix is described as row/column matrix. Vectors
may be represented by row or column matrices.[16] Scalars may be regarded as a trivial
case of matrices.
For a square matrix, the entries with equal values of row and column indices are called
the main diagonal of the matrix. For example, the entries of the main diagonal of the
third matrix in Eq. 43 are C11 and C22 . The elements of the other diagonal running from
the top right corner to the bottom left corner form the trailing or anti-diagonal, i.e.
C12 and C21 in the previous example.

B. Special Matrices

[16]
In this book, only square matrices are of primary interest as they are qualified to represent uniformly
dimensioned rank-2 tensors. Row and column matrices are also qualified to represent employed vectors.
1.3.3 Matrix Algebra 36

The zero matrix is a matrix whose all entries are 0. The identity or unit or unity
matrix is a square matrix whose all entries are 0 except those on its main diagonal which
are 1. A matrix is described as singular iff its determinant is zero (see § Determinant of
Matrix). A singular matrix has no inverse (see § Inverse of Matrix). A square matrix is
called diagonal if all of its elements which are not on the main diagonal are zero.
The transpose of a matrix is a matrix obtained by exchanging the rows and columns of
the original matrix. For example, if A is a 3 × 3 matrix and AT is its transpose then:[17]
   
 A11 A12 A13   A11 A21 A31 
   
A=
 A21 A22 A23

 ⇒ AT = 
 A12 A22 A32

 (44)
   
A31 A32 A33 A13 A23 A33

The transposition operation is defined even for non-square matrices. For square matrices,
transposition represents a reflection of the matrix elements in the main diagonal of the
matrix.

C. Matrix Multiplication

The multiplication of two matrices, A of m × k dimensions and B of k × n dimensions, is

defined as an operation that produces a matrix C of m × n dimensions whose Cij entry
is the dot product of the ith row of the first matrix A and the j th column of the second
matrix B. Hence, if A is a 3 × 2 matrix and B is a 2 × 2 matrix, then their product AB
is a 3 × 2 matrix which is given by:
   
 
 A11 A12   A11 B11 + A12 B21 A11 B12 + A12 B22 
   B11 B12   
AB = 
 A21 A22

 =
 A21 B11 + A22 B21 A21 B12 + A22 B22 
 (45)
  B21 B22  
A31 A32 A31 B11 + A32 B21 A31 B12 + A32 B22

[17]
The indexing of the entries of AT is not standard; the purpose of this is to demonstrate the exchange
of rows and columns.
1.3.3 Matrix Algebra 37

From the above, it can be seen that matrix multiplication is defined only when the
number of columns of the first matrix is equal to the number of rows of the second matrix.
Matrix multiplication is associative and distributive over a sum of compatible matrices,
but it is not commutative in general even if both forms of the product are defined, that
is:

(AB) C = A (BC) (46)

A (B + C) = AB + AC (47)

AB 6= BA (48)

As seen above, no symbol is used to indicate the operation of matrix multiplication

according to the notation of matrix algebra, i.e. the two matrices are put side by side
with no symbol in between. However, in tensor symbolic notation such an operation is
usually represented by a dot between the symbols of the two matrices, as will be discussed
later in the book.[18]

D. Trace of Matrix

The trace of a matrix is the sum of its diagonal elements, therefore if a matrix A is
given by:  
 A11 A12 A13 
 
A=
 A21 A22 A23

 (49)
 
A31 A32 A33

then its trace is given by:

tr (A) = A11 + A22 + A33 (50)

[18]
In brief, AB represents an inner product of A and B according to the matrix notation, and an outer
product of A and B according to the symbolic notation of tensors, while A · B represents an inner
product of A and B according to the symbolic notation of tensors. Hence, AB in the matrix notation
is equivalent to A · B in the symbolic notation of tensors.
1.3.3 Matrix Algebra 38

From its definition, it is obvious that the trace of a matrix is a scalar and it is defined
only for square matrices.

E. Determinant of Matrix

The determinant is a scalar quantity associated with a square matrix. There are several
definitions for the determinant of a matrix; the most direct one is that the determinant of
a 2 × 2 matrix is the product of the elements of its main diagonal minus the product of
the elements of its trailing diagonal, that is:
 
 A11 A12  A11 A12
A=  ⇒ det (A) = = A11 A22 − A12 A21 (51)
A21 A22 A21 A22

The determinant of an n × n (n > 2) matrix is then defined, recursively, as the sum of the
products of each entry of any one of its rows or columns times the cofactor of that entry
where the cofactor of an entry is defined as the determinant obtained from eliminating the
row and column of that entry from the parent matrix with a sign given by (−1)i+j with i
and j being the indices of the row and column of that entry.[19] For example:
 
 A11 A12 A13 
 
A = 
 A21 A22 A23


 
A31 A32 A33
⇓ (52)

A22 A23 A21 A23 A21 A22

det (A) = A11 − A12 + A13
A32 A33 A31 A33 A31 A32

where the determinant is evaluated along the first row. It should be remarked that the

[19]
In fact, this rule for the determinant of an n × n matrix applies even to the 2 × 2 matrix if the cofactor
of an entry in this case is taken as a single entry with the designated sign. However, we separated the
2 × 2 matrix case in the definition to be more clear and to avoid possible confusion.
1.3.3 Matrix Algebra 39

determinant of a matrix and the determinant of its transpose are equal, that is:

det (A) = det AT (53)

where A is a square matrix and T stands for the transposition operation. Another remark
is that the determinant of a diagonal matrix is the product of its main diagonal elements.

F. Inverse of Matrix

The inverse of a square matrix A is a square matrix A−1 where:

AA−1 = A−1 A = I (54)

with I being the identity matrix (see § Special Matrices) of the same dimensions as A.
The inverse of a square matrix is formed by transposing the matrix of cofactors of the
original matrix with dividing each element of the transposed matrix of cofactors by the
determinant of the original matrix.[20] From this definition, it is obvious that a matrix
possesses an inverse only if its determinant is not zero, i.e. it must be non-singular.
It should be remarked that this definition includes the 2 × 2 matrices where the cofactor
of an entry is a single entry with the designated sign, that is:
   
 A11 A12  1  A22 −A12 
A=  ⇒ A−1 =   (55)
A21 A22 A11 A22 − A12 A21 −A A11
21

Another remark is that the inverse of an invertible diagonal matrix is a diagonal matrix
obtained by taking the reciprocal of the corresponding diagonal elements of the original
matrix.

[20]
The matrix of cofactors (or cofactor matrix) is made of the cofactors of its elements taking the same
positions as the positions of these elements. The transposed matrix of cofactors may be called the
adjugate or adjoint matrix although the terminology may differ between the authors.
1.4 Exercises 40

1.4 Exercises

1.1 Name three mathematicians accredited for the development of tensor calculus. For
each one of these mathematicians, give a mathematical technical term that bears his
name.

1.2 What are the main scientific disciplines that employ the language and techniques of
tensor calculus?

1.3 Mention one cause for the widespread use of tensor calculus in science.

1.4 Describe some of the distinctive features of tensor calculus which contributed to its
success and extensive use in mathematics, science and engineering.

1.5 Give preliminary definitions of the following terms: scalar, vector, tensor, rank of
tensor, and dyad.

1.6 What is the meaning of the following mathematical symbols?

∂i ∂ii ∇ A,i ∆ Ai;k

1.7 Is the following equality correct? If so, is there any condition for this to hold?

∂k ∂l = ∂l ∂k

1.8 Pick five terms from the Index about partial derivative operators and their symbols
as used in tensor calculus and explain each one of these terms giving some examples
for their use.

1.9 Describe, briefly, the following six coordinate systems outlining their main features:
1.4 Exercises 41

orthonormal Cartesian, cylindrical, plane polar, spherical, general curvilinear, and

general orthogonal.

1.10 Which of the six coordinate systems in the previous exercise are orthogonal?

1.11 What “basis vectors” of a coordinate system means and what purpose they serve?

1.12 Which of the six coordinate systems mentioned in the previous exercises have constant
basis vectors (i.e. some or all of their basis vectors are constant both in magnitude
and in direction)?

1.13 Which of the above six coordinate systems have unit basis vectors by definition or
convention?

1.14 Explain the meaning of the coordinates in the cylindrical and spherical systems (i.e.
ρ, φ and z for the cylindrical, and r, θ and φ for the spherical).

1.15 What is the relation between the cylindrical and plane polar coordinate systems?

1.16 Is there any common coordinates between the above six coordinate systems? If so,
what? Investigate this thoroughly by comparing each pair of these systems.

1.17 Write the transformation equations between the following coordinate systems in both
directions: Cartesian and cylindrical, and Cartesian and spherical.

1.18 Make a sketch representing a spherical coordinate system, with its basis vectors,
superimposed on a rectangular Cartesian system in a standard position.

1.19 What are the geometric and algebraic definitions of the dot product of two vectors?
What is the interpretation of the geometric definition?

1.20 What are the geometric and algebraic definitions of the cross product of two vectors?
What is the interpretation of the geometric definition?
1.4 Exercises 42

1.21 What is the dot product of the vectors A and B if A = (1.9, −6.3, 0) and B =
(−4, −0.34, 11.9)?

1.22 What is the cross product of the vectors in the previous exercise? Write this cross
product in its determinantal form and expand it.

1.23 Define the scalar triple product operation of three vectors geometrically and alge-
braically.

1.24 What is the geometric interpretation of the scalar triple product? What is the condi-
tion for this product to be zero?

1.25 Is it necessary to use parentheses in the writing of scalar triple products and why? Is
it possible to interchange the dot and cross symbols in the product?

1.26 Calculate the following scalar triple products:

a · (b × c) a · (d × c) d · (c × b) a · (c × b) (a × b) · c

where a = (7, −0.4, 9.5), b = (−12.9, −11.7, 3.1), c = (2.4, 22.7, −6.9) and d =
(−56.4, 29.5, 33.8). Note that some of these products may be found directly from
other products with no need for detailed calculations.

1.27 Write the twelve possibilities of the scalar triple product a · (b × c) and divide them
into two sets where the entries in each set are equal. What is the relation between
the two sets?

1.28 What is the vector triple product of three vectors in mathematical terms? Is it scalar
or vector? Is it associative?

1.29 Give the mathematical expression for the nabla ∇ differential operator in Cartesian
systems.
1.4 Exercises 43

1.30 State the mathematical definition of the gradient of a scalar field f in Cartesian
coordinates. Is it scalar or vector?

1.31 Is the gradient operation commutative, associative or distributive? Express these

properties mathematically.

1.32 What is the relation between the gradient of a scalar field f and the surfaces of
constant f ? Make a simple sketch to illustrate this relation.

1.33 Define, mathematically, the divergence of a vector field V in Cartesian coordinates.

Is it scalar or vector?

1.34 Define “solenoidal” vector field descriptively and mathematically.

1.35 What is the physical significance of the divergence of a vector field?

1.36 Is the divergence operation commutative, associative or distributive? Give your an-
swer in words and in mathematical forms.

1.37 Define the curl of a vector field V in Cartesian coordinates using the determinantal
and the expanded forms with full explanation of all the symbols involved. Is the curl
scalar or vector?

1.38 What is the physical significance of the curl of a vector field?

1.39 What is the technical term used to describe a vector field whose curl vanishes identi-
cally?

1.40 Is the curl operation commutative, associative or distributive? Express these proper-
ties symbolically.

1.41 Describe, in words, the Laplacian operator ∇2 and how it is obtained. What are the
other symbols used to denote it?
1.4 Exercises 44

1.42 Give the mathematical expression of the Laplacian operator in Cartesian systems.
Using this mathematical expression, explain why the Laplacian is a scalar rather than
a vector operator?

1.43 Can the Laplacian operator act on rank-0, rank-1 and rank-n (n > 1) tensor fields?
If so, what is the rank of the resulting field in each case?

1.44 Collect from the Index all the terms related to the nabla based differential operations
and classify them according to each one of these operations.

1.45 Write down the mathematical expression for the divergence theorem, defining all the
symbols involved, and explain the meaning of this theorem in words.

1.46 What are the main uses of the divergence theorem in mathematics and science? Ex-
plain why this theorem is very useful theoretically and practically.

1.47 If a vector field is given in the Cartesian coordinates by A = (−0.5, 9.3, 6.5), verify
the divergence theorem for a cube defined by the plane surfaces x1 = −1, x2 = 1,
y1 = −1, y2 = 1, z1 = −1, and z2 = 1.

1.48 Write down the mathematical expression for Stokes theorem with the definition of all
the symbols involved and explain its meaning in words. What is this theorem useful
for? Why it is very useful?

1.49 If a vector field is given in the Cartesian coordinates by A = (2y, −3x, 1.5z), verify
Stokes theorem for a hemispherical surface x2 + y 2 + z 2 = 9 for z ≥ 0.

1.50 Make a simple sketch to demonstrate Stokes theorem with sufficient explanations and
definitions of the symbols involved.

1.51 Give concise definitions for the following terms related to matrices: matrix, square
matrix, main diagonal, trailing diagonal, transpose, identity matrix, unit matrix,
1.4 Exercises 45

singular, trace, determinant, cofactor, and inverse.

1.52 Explain the way by which matrices are indexed.

1.53 How many indices are needed in indexing a 2 × 3 matrix, an n × n matrix, and an
m × k matrix? Explain, in each case, why.

1.54 Does the order of the matrix indices matter? If so, what is the meaning of changing
this order?

1.55 Is it possible to write a vector as a matrix? If so, what is the condition that should
be imposed on the indices and how many forms a vector can have when it is written
as a matrix?

1.56 Write down the following matrices in a standard rectangular array form (similar to
the examples in Eq. 43) using conventional symbols for their entries with a proper
indexing: 3 × 4 matrix A, 1 × 5 matrix B, 2 × 2 matrix C, and 3 × 1 matrix D.

1.57 Give detailed mathematical definitions of the determinant, trace and inverse of matrix,
explaining any symbol or technical term involved in these definitions.

1.58 Find the following matrix multiplications: AB, BC, and CB where:
   
 
 9.6 6.3 −22   −3.8 −2.0 
     3 8.4 61.3 
A= 
 −3.8 2.5 2.9  B=
 4.6 11.6 
 C= 
    −5 −33 5.9
−6 3.2 7.5 12.0 25.9

1.59 Referring to the matrices A, B and C in the previous exercise, find all the permuta-
tions (repetitive and non-repetitive) involving two of these three matrices, and classify
them into two groups: those which do represent possible matrix multiplication and
those which do not.
1.4 Exercises 46

1.60 Is matrix multiplication associative? commutative? distributive over matrix addition?

1.61 Calculate the trace, the determinant, and the inverse (if the inverse exists) of the
following matrices:
   
 3.2 2.6 1.6   5.2 2.7 3.6 
   
D=
 12.9 −1.9 2.4

 E=
 −10.4 −5.4 −7.2


   
−11.9 33.2 −22.5 −31.9 13.2 −23.7

1.62 Which, if any, of the matrices D and E in the previous exercise is singular?

1.63 Select from the Index six terms connected to the special matrices which are defined
in § Special Matrices.
Chapter 2

Tensors

In this chapter, we present the essential terms and definitions related to tensors, the
conventions and notations which are used in their representation, the general rules that
govern their manipulation, and their main types and classifications. We also provide some
illuminating examples of tensors of various complexity as well as an overview of their use
in mathematics, science and engineering.

2.1 General Background about Tensors

A tensor is an array of mathematical objects (usually numbers or functions) which

transforms according to certain rules under coordinates change. In a d-dimensional space,
a tensor of rank-n has dn components which may be specified with reference to a given
coordinate system. Accordingly, a scalar, such as temperature, is a rank-0 tensor with
(assuming a 3D space) 30 = 1 component, a vector, such as force, is a rank-1 tensor with
31 = 3 components, and stress is a rank-2 tensor with 32 = 9 components. Figure 11
graphically illustrates the structure of a rank-3 tensor in a 3D space.
The dn components of a tensor are identified by n distinct integer indices (e.g. i, j, k)
which are attached, according to the commonly-employed tensor notation, as super-
scripts or subscripts or a mix of these to the right side of the symbol utilized to label the
tensor, e.g. Aijk , Aijk and Ajk
i . Each tensor index takes all the values over a predefined

range of dimensions such as 1 to d in the above example of a d-dimensional space. In gen-

47
2.1 General Background about Tensors 48

A331 A332 A333

A321 A322 A323

A311 A312 A313

A231 A232 A233

A221 A222 A223

A211 A212 A213

A131 A132 A133

A121 A122 A123

A111 A112 A113

Figure 11: Graphical illustration of a rank-3 tensor Aijk in a 3D space, i.e. each one of
i, j, k ranges over 1, 2, 3.

eral, all tensor indices have the same range, i.e. they are uniformly dimensioned.[21] When
the range of tensor indices is not stated explicitly, it is usually assumed to have the values
1, 2, 3. However, the range must be stated explicitly or implicitly to avoid ambiguity.
The characteristic property of tensors is that they satisfy the principle of invari-
ance under certain coordinate transformations. Therefore, formulating the fundamental
physical laws in a tensor form ensures that they are form-invariant; hence they are

[21]
This assertion, in fact, applies to the common cases of tensor applications; however, there are instances,
for example in the differential geometry of curves and surfaces, of tensors which are not uniformly
dimensioned because the tensor is related to two spaces with different dimensions.
2.2 General Terms and Concepts 49

objectively-representing the physical reality and do not depend on the observer. Having
the same form in different coordinate systems may be labeled as being covariant although
this word is also used for a different meaning in tensor calculus, as will be fully explained
in § 2.6.1.
While tensors of rank-0 are generally represented in a common form of light face non-
indexed italic symbols like f and g, tensors of rank ≥ 1 are represented in several forms
and notations, the main ones are the index-free notation, which may also be called the
direct or symbolic or Gibbs notation, and the indicial notation which is also called
the index or component or tensor notation. The first is a geometrically oriented
notation with no reference to a particular reference frame and hence it is intrinsically
invariant to the choice of coordinate systems, whereas the second takes an algebraic
form based on components identified by indices and hence the notation is suggestive of
an underlying coordinate system, although being a tensor makes it form-invariant under
certain coordinate transformations and therefore it possesses certain invariant properties.
The index-free notation is usually identified by using bold face non-italic symbols, like a
and B, while the indicial notation is identified by using light face indexed italic symbols
such as ai and Bij .
It is noteworthy that although rank-0 and rank-1 tensors are, respectively, scalars and
vectors, not all scalars and vectors (in their generic sense) are tensors of these ranks.
Similarly, rank-2 tensors are normally represented by square matrices but not all square
matrices represent rank-2 tensors.

2.2 General Terms and Concepts

In the following, we introduce and define a number of essential concepts and terms which
form a principal part of the technical and conceptual structure of tensor calculus. These
concepts and terms are needed in the development of the forthcoming sections and chap-
2.2 General Terms and Concepts 50

ters.
Tensor term is a product of tensors including scalars and vectors. Tensor expression
is an algebraic sum (or more generally a linear combination) of tensor terms which may
be a trivial sum in the case of a single term. Tensor equality (which is symbolized by
“=”) is an equality of two tensor terms and/or expressions. A special case of this is tensor
identity which is an equality of general validity.[22]
An index that occurs once in a tensor term is a free index while an index that occurs
twice in a tensor term is a dummy or bound index. The order of a tensor is identified
by the number of its indices (e.g. Aijk is a tensor of order 3) which normally identifies
the tensor rank as well. However, when contraction (see § 3.4) of indices operation takes
place once or more, the order of the tensor is not affected but its rank is reduced by two
for each contraction operation.[23] Hence, the order of a tensor is equal to the number of
all of its indices including the dummy indices, while the rank is equal to the number of its
free indices only.
Tensors with subscript indices, like Aij , are called covariant, while tensors with su-
perscript indices, like Ak , are called contravariant. Tensors with both types of indices,
like Almn
lk , are called mixed type. More details about this classification will follow in §

2.6.1. Subscript indices, rather than subscripted tensors, are also dubbed covariant and
superscript indices are dubbed contravariant.
The Zero tensor is a tensor whose all components are zero. The Unit tensor or unity
tensor, which is usually defined for rank-2 tensors, is a tensor whose all elements are zero
except the ones with identical values of all indices which are assigned the value 1.
In general terms, a transformation from an nD space to another nD space is a cor-

[22]
As indicated previously, the symbol “≡” may be used for identity as well as for definition.
[23]
In the literature of tensor calculus, rank and order of tensors are generally used interchangeably;
however some authors differentiate between the two as they assign order to the total number of indices,
including contracted indices, while they reserve rank to the number of free indices. We think the latter
is better and hence in the present book we embrace this terminology.
2.2 General Terms and Concepts 51

relation that maps a point from the first space (original) to a point in the second space
(transformed) where each point in the original and transformed spaces is identified by n
independent variables or coordinates. To distinguish between the two sets of coordinates
in the two spaces, the coordinates of the points in the transformed space may be no-
tated with barred symbols like (x̄1 , x̄2 , . . . , x̄n ) or (x̄1 , x̄2 , . . . , x̄n ) where the superscripts
and subscripts are indices, while the coordinates of the points in the original space are
notated with unbarred symbols like (x1 , x2 , . . . , xn ) or (x1 , x2 , . . . , xn ). Under certain
conditions, such a transformation is unique and hence an inverse transformation from
the transformed to the original space is also defined.
Mathematically, each one of the direct and inverse transformations can be regarded
as a mathematical correlation expressed by a set of equations in which each coordinate in
one space is considered as a function of the coordinates in the other space. Hence, the
transformations between the two sets of coordinates in the two spaces can be expressed
mathematically by the following two sets of independent relations:

x̄i = x̄i (x1 , x2 , . . . , xn ) (56)

xi = xi (x̄1 , x̄2 , . . . , x̄n ) (57)

where i = 1, 2, . . . , n.
An alternative to the latter view of considering the transformation as a mapping be-
tween two different spaces is to view it as a correlation relating the same point in the same
space but observed from two different coordinate systems which are subject to a similar
transformation. The following will be largely based on the latter view.
Coordinate transformations are described as proper when they preserve the handed-
ness (right- or left-handed) of the coordinate system and improper when they reverse
the handedness. Improper transformations involve an odd number of coordinate axes in-
2.2 General Terms and Concepts 52

versions in the origin of coordinates. Inversion of axes may be called improper rotation
while ordinary rotation is described as proper rotation. Figure 12 illustrates proper and
improper coordinate transformations of a rectangular Cartesian system.

X3 x3 X3 x3
X1

O O
X2 X2

x1 x2 x1 x2
X1 Proper Improper
Figure 12: Proper and improper transformations of a rectangular Cartesian coordinate
system where the former is achieved by a rotation of the coordinate system while the
latter is achieved by a rotation followed by a reflection of the first axis in the origin of
coordinates. The transformed systems are shown as dashed while the original system is
shown as solid.

Transformations can be active, when they change the state of the observed object (e.g.
translating the object in space), or passive when they are based on keeping the state
of the object and changing the state of the coordinate system from which the object is
observed. Such distinction is based on an implicit assumption of a more general frame of
reference in the background.
A permutation of a set of objects, which are normally numbers like (1, 2, . . . , n) or
symbols like (i, j, k), is a particular ordering or arrangement of these objects. An even
permutation is a permutation resulting from an even number of single-step exchanges
(also known as transpositions) of neighboring objects starting from a presumed original
permutation of these objects. Similarly, an odd permutation is a permutation resulting
2.3 General Rules 53

from an odd number of such exchanges. It has been shown that when a transformation from
one permutation to another can be done in different ways, possibly with different numbers
of exchanges, the parity of all these possible transformations is the same, i.e. all are even
or all are odd, and hence there is no ambiguity in characterizing the transformation from
one permutation to another by the parity alone.

2.3 General Rules

In the following, we present some very general rules that apply to the mathematical ex-
pressions and relations in tensor calculus. No index is allowed to occur more than twice
in a legitimate tensor term.[24] A free index should be understood to vary over its range
(e.g. 1, . . . , n) and hence it can be interpreted as saying “for all components represented
by the index”. Therefore, a free index represents a number of terms or expressions or
equalities equal to the number of allowed values of its range. For example, when i and j
can vary over the range 1, . . . , n the following expression:

Ai + Bi (58)

represents n separate expressions while the following equation:

Aji = Bij (59)

represents n × n separate equations.

[24]
We adopt this assertion, which is common in the literature of tensor calculus, as we think it is suitable
for this level. However, there are many instances in the literature of tensor calculus where indices are
repeated more than twice in a single term. The bottom line is that as long as the tensor expression
makes sense and the intention is clear, such repetitions should be allowed with no need in our view to
take special precaution like using parentheses. In particular, the forthcoming summation convention
will not apply automatically in such cases although summation
P on such indices, if needed, can be
carried out explicitly, by using the summation symbol or by a special declaration of such intention
similar to the summation convention. Anyway, in the present book we will not use indices repeated
more than twice in a single term.
2.3 General Rules 54

According to the summation convention, which is widely used in the literature of

tensor calculus including in the present book, dummy indices imply summation over
their range, e.g. for an nD space we have:

X
n
Ai Bi = Ai Bi = A1 B1 + A2 B2 + . . . + An Bn (60)
i=1
X
n X
n
δij Aij = δij Aij (61)
i=1 j=1
Xn X n Xn
ijk Aij B k = ijk Aij B k (62)
i=1 j=1 k=1

When dummy indices do not imply summation, the situation must be clarified by enclosing
such indices in parentheses or by underscoring or by using upper case letters (with
declaration of these conventions) or by adding a clarifying comment like “no summation
on repeated indices”.[25]
Each tensor index should conform to one of the forthcoming variance transformation
rules as given by Eqs. 70 and 71, i.e. it is either covariant or contravariant. For
orthonormal Cartesian coordinate systems, the two variance types (i.e. covariant and
contravariant) do not differ because the metric tensor is given by the Kronecker delta
(refer to § 4.1 and 6) and hence any index can be upper or lower although it is common
to use lower indices in such cases.
For tensor invariance, a pair of dummy indices should in general be complementary
in their variance type, i.e. one covariant and the other contravariant. However, for or-
thonormal Cartesian systems the two are the same and hence when both dummy indices
are covariant or both are contravariant it should be understood as an indication that the
underlying coordinate system is orthonormal Cartesian if the possibility of an error is

[25]
These precautions are obviously needed if the summation convention is adopted in general but it does
not apply in some exceptional cases where repeated indices are needed in the notation with no intention
of summation.
2.3 General Rules 55

excluded.
As indicated earlier, tensor order is equal to the number of its indices while tensor
rank is equal to the number of its free indices; hence vectors (terms, expressions and
equalities) are represented by a single free index and rank-2 tensors are represented by
two free indices. The dimension of a tensor is determined by the range taken by its
indices.
The rank of all terms in legitimate tensor expressions and equalities must be the same.
Moreover, each term in valid tensor expressions and equalities must have the same set
of free indices (e.g. i, j, k). Also, a free index should keep its variance type in every
term in valid tensor expressions and equations, i.e. it must be covariant in all terms or
contravariant in all terms.
While free indices should be named uniformly in all terms of tensor expressions and
equalities, dummy indices can be named in each term independently, e.g.

j
Aiik + Bjk lm
+ Clmk jk
Dij = Eik jm
+ Fim (63)

A free index in an expression or equality can be renamed uniformly using a different

symbol, as long as this symbol is not already in use, assuming that both symbols vary
over the same range, i.e. have the same dimension.
Examples of legitimate tensor terms, expressions and equalities are:

Aij
ij Aim ink
m + Bnk Cij = Aij − Bij a = Bjj (64)

while examples of illegitimate tensor terms, expressions and equalities are:

Biii Ai + Bij Ai + B j Ai − B i Aii = Bi (65)

2.3 General Rules 56

Indexing is generally distributive over the terms of tensor expressions and equalities,
for example:
[A + B]i = [A]i + [B]i (66)

and
[A = B]i ⇐⇒ [A]i = [B]i (67)

Unlike scalars and tensor components, which are essentially scalars in a generic sense,
operators cannot in general be freely reordered in tensor terms, therefore we have:

f h = hf Ai B i = B i Ai (68)

but:
∂i Ai 6= Ai ∂i (69)

It should be remarked that the order of the indices[26] of a given tensor is important
and hence it should be observed and clarified, because two tensors with the same set of
indices and with the same indicial structure but with different indicial order are not equal
in general. For example, Aijk is not equal to Ajik unless A is symmetric (refer to § 2.6.5)
with respect to the indices i and j. Similarly, B mln is not equal to B lmn unless B is
symmetric in its indices l and m.
The confusion about the order of indices occurs, in particular, in the case of mixed
type tensors such as Aijk . Spaces are usually used in this case to indicate the order, e.g.
i
the latter tensor is symbolized as Aj k if the order of the indices is j, i, k while it is
symbolized as Ai jk if the order of the indices is i, j, k.[27] Dots may also be used in the
case of mixed type tensors to indicate, more explicitly, the order of the indices and remove

[26]
This should not be confused with the order of tensor as defined above in the same context as tensor
rank.
[27]
We exaggerate the spacing here for clarity.
2.4 Examples of Tensors of Different Ranks 57

any ambiguity. For example, if the indices i, j, k of the tensor A, which is covariant in i
and k and contravariant in j, are of that order, then A may be symbolized as Ai j. k where
the dot between i and k indicates that j is in the middle.[28]
Finally, many of the identities in the present book which are given in a covariant or a
contravariant or a mixed form are similarly valid for the other forms and hence they
can be obtained with a minimal effort. The objective of reporting in only one form is
conciseness and to avoid unnecessary repetition. Moreover, in the case of orthonormal
Cartesian systems the variance type of indices is irrelevant.

2.4 Examples of Tensors of Different Ranks

Examples of rank-0 tensors (scalars) are energy, mass, temperature, volume and density.
These are totally identified by a single number regardless of any coordinate system and
hence they are invariant under coordinate transformations.[29]
Examples of rank-1 tensors (vectors) are displacement, force, electric field, velocity and
acceleration. These require for their complete identification a number, representing their
magnitude, and a direction representing their geometric orientation within their space.
Alternatively, they can be uniquely identified by a set of numbers, equal to the number
of dimensions of the underlying space, in reference to a particular coordinate system and
hence this identification is system-dependent although they still have system-invariant
properties such as length.
Examples of rank-2 tensors are the Kronecker delta (see § 4.1), stress, strain, rate of

[28]
In many places in this book (like other books) and for the convenience in typesetting, the order of the
indices is not clarified by spacing or inserting dots in the case of mixed type tensors. This commonly
occurs where the order of the indices is irrelevant in the given context or the order is clear. Sometimes,
the order of the indices may be indicated implicitly by the alphabetical order of the selected indices,
e.g. Ajk
i means Ai
jk
.
[29]
The focus of this section is on providing examples of tensors of different ranks. As we will see later in
this chapter (refer to § 2.6.2), there are true and pseudo scalars, vectors and tensors, and hence some
of the statements and examples given here may qualify for certain restrictions and conditions.
2.5 Applications of Tensors 58

strain and inertia tensors. These require for their full identification a set of numbers each
of which is associated with two directions. These double directions are usually identified
by a set of unit dyads. Figure 13 is a graphic illustration of the nine unit dyads which are
associated with the double directions of rank-2 tensors in a 3D space with a rectangular
Cartesian coordinate system.
Examples of rank-3 tensors are the Levi-Civita tensor (see § 4.2) in 3D spaces and
the tensor of piezoelectric moduli. Examples of rank-4 tensors are the elasticity or stiff-
ness tensor, the compliance tensor and the fourth-order moment of inertia tensor. It is
noteworthy that tensors of high ranks are relatively rare in science and engineering.

2.5 Applications of Tensors

Tensor calculus is a very powerful mathematical tool; hence tensors are commonplace in
science and engineering where they are used to represent physical and synthetic objects
and ideas in mathematical invariant forms. Tensor notation and techniques are used in
many branches of mathematics, science and engineering such as differential geometry,
fluid mechanics, continuum mechanics, general relativity and structural engi-
neering. Tensor calculus is used for elegant and compact formulation and presentation
of equations and identities in mathematics, science and engineering. It is also used for
algebraic manipulation of mathematical expressions and proving identities in a neat and
succinct way (refer to § 5.6). As indicated earlier, the invariance of tensor forms serves a
theoretically and practically important role by allowing the formulation of physical laws
in coordinate-free forms.

2.6 Types of Tensors

In the following subsections we introduce a number of tensor types and categories and
highlight their main characteristics and differences. These types and categories are not
2.6.1 Covariant and Contravariant Tensors 59

x3 x3 x3

e1 e1 e1 e2 e1 e3
x1 x2 x1 x2 x1 x2
x3 x3 x3

e2 e1 e2 e2 e2 e3
x1 x2 x1 x2 x1 x2
x3 x3 x3

e3 e1 e3 e2 e3 e3
x1 x2 x1 x2 x1 x2
Figure 13: The nine unit dyads associated with the double directions of rank-2 tensors
in a 3D space with a rectangular Cartesian coordinate system. The vectors ei and ej
(i, j = 1, 2, 3) are unit vectors in the directions of coordinate axes where the first indexed
e (blue) represents the first vector of the dyad while the second indexed e (red) represents
the second vector of the dyad. In these nine frames, the first vector is fixed along each
row while the second vector is fixed along each column.

mutually exclusive and hence they overlap in general; moreover they may not be ex-
haustive in their classes as some tensors may not instantiate any one of a complementary
set of types such as being symmetric or anti-symmetric.

2.6.1 Covariant and Contravariant Tensors

These are the main types of tensor with regard to the rules of their transformation between
different coordinate systems. Covariant tensors are notated with subscript indices (e.g.
2.6.1 Covariant and Contravariant Tensors 60

Ai ) while contravariant tensors are notated with superscript indices (e.g. Aij ). A
covariant tensor is transformed according to the following rule:

∂xj
Āi = Aj (70)
∂ x̄i

while a contravariant tensor is transformed according to the following rule:

∂ x̄i j
Āi = A (71)
∂xj

where the barred and unbarred symbols represent the same mathematical object (tensor
or coordinate) in the transformed and original coordinate systems respectively.
An example of covariant tensors is the gradient of a scalar field while an example of
contravariant tensors is the displacement vector. Some tensors of rank > 1 have mixed
variance type, i.e. they are covariant in some indices and contravariant in others. In this
case the covariant variables are indexed with subscripts while the contravariant variables
are indexed with superscripts, e.g. Aji which is covariant in i and contravariant in j. A
mixed type tensor transforms covariantly in its covariant indices and contravariantly in its
contravariant indices, e.g.
∂ x̄l ∂xj ∂ x̄n i k
Ālmn = A (72)
∂xi ∂ x̄m ∂xk j

To clarify the pattern of mathematical transformation of tensors, we explain step-

by-step the practical rules to follow in writing tensor transformation equations between
two coordinate systems, unbarred and barred, where for clarity we color the symbols of
the tensor and the coordinates belonging to the unbarred system with blue while we use
red to mark the symbols belonging to the barred system. Since there are three types of
tensors: covariant, contravariant and mixed, we use three equations in each step.
In this demonstration we use rank-4 tensors as examples since this is sufficiently general
2.6.1 Covariant and Contravariant Tensors 61

and hence adequate to elucidate the rules for transforming tensors of any rank. The
demonstration is based on the assumption that the transformation is taking place from
the unbarred system to the barred system; the same rules should apply for the opposite
transformation from the barred system to the unbarred system. We use the sign “$” for
the equality in the transitional steps to indicate that the equalities are under construction
and are not complete.
We start with the very generic equations between the barred tensor Ā and the unbarred
tensor A for the three types:

Ā $ A (covariant)

Ā $ A (contravariant) (73)

Ā $ A (mixed)

We assume that the barred tensor and its coordinates are indexed with ijkl and the
unbarred are indexed with npqr, so we add these indices in their presumed order and
position (lower or upper) paying particular attention to the order in the mixed type:

Āijkl $ Anpqr

Āijkl $ Anpqr (74)

Āijkl $ Anpqr

Since the barred and unbarred tensors are of the same type, as they represent the same
tensor in two coordinate systems,[30] the indices on the two sides of the equalities should
match in their position and order. We then insert a number of partial differential operators
on the right hand side of the equations equal to the rank of these tensors, which is 4 in our

[30]
Similar basis vectors are assumed.
2.6.1 Covariant and Contravariant Tensors 62

example. These operators represent the transformation rules for each pair of corresponding
coordinates, one from the barred and one from the unbarred:

∂ ∂ ∂ ∂
Āijkl $ ∂ ∂ ∂ ∂
Anpqr

Āijkl $ ∂ ∂ ∂ ∂
∂ ∂ ∂ ∂
Anpqr (75)

Āijkl $ ∂ ∂ ∂ ∂
∂ ∂ ∂ ∂
Anpqr

Now we insert the coordinates of the barred system into the partial differential operators
noting that (i) the positions of any index on the two sides should match, i.e. both upper
or both lower, since they are free indices in different terms of tensor equalities, (ii) a
superscript index in the denominator of a partial derivative is in lieu of a covariant
index in the numerator,[31] and (iii) the order of the coordinates should match the order
of the indices in the tensor, that is:

∂ ∂ ∂ ∂
Āijkl $ ∂xi ∂xj ∂xk ∂xl
Anpqr
∂xi ∂xj ∂xk ∂xl
Āijkl $ ∂ ∂ ∂ ∂
Anpqr (76)
∂xi ∂xj ∂
Āijkl $ ∂
∂
∂ ∂xk ∂xl
Anpqr

For consistency, these coordinates should be barred as they belong to the barred tensor;
hence we add bars:

∂ ∂ ∂ ∂
Āijkl $ ∂ x̄i ∂ x̄j ∂ x̄k ∂ x̄l
Anpqr
∂ x̄i ∂ x̄j ∂ x̄k ∂ x̄l
Āijkl $ ∂ ∂ ∂ ∂
Anpqr (77)
∂ x̄i ∂ x̄j ∂
Āijkl $ ∂
∂
∂ ∂ x̄k ∂ x̄l
Anpqr

[31]
The use of upper indices in the symbols of general coordinates is to indicate the fact that the coordinates
and their differentials transform contravariantly.
2.6.1 Covariant and Contravariant Tensors 63

Finally, we insert the coordinates of the unbarred system into the partial differential
operators noting that (i) the positions of the repeated indices on the same side should
be opposite, i.e. one upper and one lower, since they are dummy indices and hence the
position of the index of the unbarred coordinate should be opposite to its position in the
unbarred tensor, (ii) an upper index in the denominator is in lieu of a lower index in the
numerator, and (iii) the order of the coordinates should match the order of the indices in
the tensor:

∂xn ∂xp ∂xq ∂xr

Āijkl = ∂ x̄i ∂ x̄j ∂ x̄k ∂ x̄l
Anpqr
∂ x̄i ∂ x̄j ∂ x̄k ∂ x̄l
Āijkl = ∂xn ∂xp ∂xq ∂xr
Anpqr (78)
∂ x̄i ∂ x̄j ∂xq ∂xr
Āijkl = ∂xn ∂xp ∂ x̄k ∂ x̄l
Anpqr

We also replaced the “$” sign in the final set of equations with the strict equality sign “=”
as the equations now are complete.
The covariant and contravariant types of a tensor are linked through the metric tensor,
as will be detailed later in the book (refer to § 6). As indicated before, for orthonormal
Cartesian systems there is no difference between covariant and contravariant tensors,
and hence the indices can be upper or lower although it is common to use lower indices in
this case.
A tensor of m contravariant indices and n covariant indices may be called type (m, n)
tensor. When one or both variance types are absent, zero is used to refer to the absent
variance type in this notation. Accordingly, Akij is a type (1, 2) tensor, B ik is a type (2, 0)
ts
tensor, Cm is a type (0, 1) tensor, and Dpqr is a type (2, 3) tensor.
The vectors providing the basis set for a coordinate system are of covariant type when
they are tangent to the coordinate axes, and they are of contravariant type when they
are perpendicular to the local surfaces of constant coordinates. These two sets, like the
2.6.1 Covariant and Contravariant Tensors 64

tensors themselves, are identical for orthonormal Cartesian systems.

Formally, the covariant and contravariant basis vectors are given respectively by:

∂r
Ei = Ei = ∇xi (79)
∂xi

where r = xi ei is the position vector in Cartesian coordinates and xi is a general curvi-

linear coordinate. As before, a superscript in the denominator of partial derivatives is
equivalent to a subscript in the numerator. It should be remarked that in general the
basis vectors (whether covariant or contravariant ) are not necessarily of unit length
and/or mutually orthogonal although they may be so.[32]
The two sets of covariant and contravariant basis vectors are reciprocal systems and
hence they satisfy the following reciprocity relation:

Ei · Ej = δij (80)

where δij is the Kronecker delta (refer to § 4.1) which can be represented by the unity
matrix (see § Special Matrices). The reciprocity of these two sets of basis vectors is
illustrated schematically in Figure 14 for the case of a 2D space.
A vector can be represented either by covariant components with contravariant coor-
dinate basis vectors or by contravariant components with covariant coordinate basis
vectors. For example, a vector A can be expressed as:

A = Ai Ei or A = A i Ei (81)

where Ei and Ei are the contravariant and covariant basis vectors respectively. This is
illustrated graphically in Figure 15 for a vector A in a 2D space. The use of the covariant
[32]
In fact there are standard mathematical procedures to orthonormalize the basis set if it is not and if
this is needed.
2.6.2 True and Pseudo Tensors 65

E2

E2
φ
E1
φ E1

Figure 14: The reciprocity relation between the covariant and contravariant basis vectors
in a 2D space where E1 ⊥ E2 , E1 ⊥ E2 , and |E1 | |E1 | cos φ = |E2 | |E2 | cos φ = 1.

or contravariant form of the vector representation is a matter of choice and convenience

since these two representations are equivalent as they represent and correctly describe the
same object.
More generally, a tensor of any rank (≥ 1) can be represented covariantly using con-
travariant basis tensors of that rank, or contravariantly using covariant basis tensors, or
in a mixed form using a mixed basis of opposite type. For example, a rank-2 tensor A
can be written as:

A = Aij Ei Ej = Aij Ei Ej = Aij Ei Ej = Aij Ei Ej (82)

where Ei Ej , Ei Ej , Ei Ej and Ei Ej are dyadic products of the basis vectors of the presumed
system (refer to § 2.4 and 3.3).

2.6.2 True and Pseudo Tensors

These are also called polar and axial tensors respectively although it is more common to
use these terms for vectors. Pseudo tensors may also be called tensor densities.[33] True
[33]
The terminology in this part, like many other parts, is not universal.
2.6.2 True and Pseudo Tensors 66

A2 |E2| A
E2
A2 E2

E2
A1 E1
E1
E1 A1 |E1|
Figure 15: The representation of a vector A in covariant and contravariant basis vector
sets in a 2D space where the components shown at the four points are with reference to
unit vectors in the given directions, e.g. A1 |E1 | is a component with reference to a unit
vector in the direction of E1 .

tensors are proper or ordinary tensors and hence they are invariant under coordinate
transformations, while pseudo tensors are not proper tensors since they do not transform
invariantly as they acquire a minus sign under improper orthogonal transformations which
involve inversion of coordinate axes through the origin of coordinates with a change of
system handedness.
Figure 16 demonstrates the behavior of a true vector v and a pseudo vector p where the
former keeps its direction following a reflection of the coordinate system through the
origin of coordinates while the latter reverses its direction following this operation.
Because true and pseudo tensors have different mathematical properties and represent
different types of physical entities, the terms of consistent tensor expressions and equa-
tions should be uniform in their true and pseudo type, i.e. all terms are true or all are
pseudo.
The direct product (refer to § 3.3) of true tensors is a true tensor. The direct product
of even number of pseudo tensors is a true tensor, while the direct product of odd
2.6.2 True and Pseudo Tensors 67

x3 V
X2 X1
v
p

O
O
P

x1 x2 X3
Figure 16: The behavior of a true vector (v and V) and a pseudo vector (p and P) on
reflecting the coordinate system in the origin of coordinates. The lower case symbols stand
for the objects in the original system while the upper case symbols stand for the same
objects in the reflected system.

number of pseudo tensors is a pseudo tensor. The direct product of a mix of true and
pseudo tensors is a true or pseudo tensor depending on the number of pseudo tensors
involved in the product as being even or odd respectively.
Similar rules to those of the direct product apply to the cross product, including
the curl operation, involving tensors (which are usually of rank-1) with the addition
of a pseudo factor for each cross product operation. This factor is contributed by the
permutation tensor  which is implicit in the definition of the cross product (see Eqs.
173 and 192). As we will see in § 4.2, the permutation tensor is a pseudo tensor.
In summary, what determines the tensor type (true or pseudo) of the tensor terms involv-
ing direct[34] and cross products is the parity of the multiplicative factors of pseudo type
plus the number of cross product operations involved since each cross product operation
contributes an  factor.
Examples of true scalars are temperature, mass and the dot product of two polar or
[34]
Inner product (see § 3.5) is the result of a direct product (see § 3.3) operation followed by a contraction
(see § 3.4) and hence it is like a direct product in this context.
2.6.3 Absolute and Relative Tensors 68

two axial vectors, while examples of pseudo scalars are the dot product of an axial
vector and a polar vector and the scalar triple product of polar vectors. Examples of
polar vectors are displacement and acceleration, while examples of axial vectors are
angular velocity and cross product of polar vectors in general, including the curl operation
on polar vectors, due to the involvement of the permutation symbol  which is a pseudo
tensor as stated already. As indicated before, the essence of the distinction between true
(i.e. polar) and pseudo (i.e. axial) vectors is that the direction of a pseudo vector depends
on the observer choice of the handedness of the coordinate system whereas the direction
of a true vector is independent of such a choice.
Examples of true tensors of rank-2 are stress and rate of strain tensors, while examples
of pseudo tensors of rank-2 are direct products of two vectors: one polar and one axial.
Examples of true tensors of higher ranks are piezoelectric moduli tensor (rank-3) and
elasticity tensor (rank-4), while examples of pseudo tensors of higher ranks are the
permutation tensor of these ranks.

2.6.3 Absolute and Relative Tensors

Considering an arbitrary transformation from a general coordinate system to another, a

tensor of weight w is defined by the following general tensor transformation:

w
∂x ∂ x̄i ∂ x̄j ∂ x̄k ∂xd ∂xe ∂xf ab...c
Āij...klm...n = · · · · · · A de...f (83)
∂ x̄ ∂xa ∂xb ∂xc ∂ x̄l ∂ x̄m ∂ x̄n
2.6.3 Absolute and Relative Tensors 69

∂x
where ∂ x̄
is the Jacobian of the transformation between the two systems.[35] When
w = 0 the tensor is described as an absolute or true tensor, and when w 6= 0 the tensor
is described as a relative tensor. When w = −1 the tensor may be described as a
pseudo tensor, while when w = 1 the tensor may be described as a tensor density.[36]
As indicated earlier, a tensor of m contravariant indices and n covariant indices may be
described as a tensor of type (m, n). This may be extended to include the weight w as a
third entry and hence the type of the tensor is identified by (m, n, w).
Relative tensors can be added and subtracted (see § 3.1) if they are of the same variance
type and have the same weight;[37] the result is a tensor of the same type and weight. Also,
relative tensors can be equated if they are of the same type and weight. Multiplication
of relative tensors produces a relative tensor whose weight is the sum of the weights of
the original tensors. Hence, if the weights are added up to a non-zero value the result is
a relative tensor of that weight; otherwise it is an absolute tensor.

[35]
The Jacobian J is the determinant of the Jacobian matrix J of the transformation between the unbarred
and barred systems, that is:
∂x1 ∂x1 ∂x1
∂ x̄1 ∂ x̄22 ··· ∂ x̄n
∂x2 ∂x ∂x2
∂ x̄1 ∂ x̄2 ··· ∂ x̄n
J = det (J) = .. .. .. .. (84)
. . . .
∂xn ∂xn ∂xn
∂ x̄1 ∂ x̄2 ··· ∂ x̄n

For more details, the reader is advised to consult more advanced textbooks on this subject.
[36]
Some of these labels are used differently by different authors as the terminology of tensor calculus is
not universally approved and hence the conventions of each author should be checked. Also, there is an
obvious overlap between this classification (i.e. absolute and relative) and the previous classification
(i.e. true and pseudo) at least according to some conventions.
[37]
This statement should be generalized by including w = 0 which corresponds to absolute tensors and
hence “relative” in this statement is more general than being opposite to “absolute”. Accordingly, and
from the perspective of relative tensors (i.e. assuming that other qualifications such as matching in
the indicial structure are met), two absolute tensors can be added/subtracted but an absolute and a
relative tensor (i.e. with w 6= 0) cannot since they are “relative” tensors with different weights.
2.6.4 Isotropic and Anisotropic Tensors 70

2.6.4 Isotropic and Anisotropic Tensors

Isotropic tensors are characterized by the property that the values of their components
are invariant under coordinate transformation by proper rotation of axes. In contrast,
the values of the components of anisotropic tensors are dependent on the orientation of the
coordinate axes. Notable examples of isotropic tensors are scalars (rank-0), the vector 0
(rank-1), Kronecker delta δij (rank-2) and Levi-Civita tensor ijk (rank-3). Many tensors
describing physical properties of materials, such as stress and magnetic susceptibility, are
anisotropic.
Direct and inner products (see § 3.3 and 3.5) of isotropic tensors are isotropic tensors.
The zero tensor of any rank is isotropic; therefore if the components of a tensor vanish in
a particular coordinate system they will vanish in all properly and improperly rotated
coordinate systems.[38] Consequently, if the components of two tensors are identical in a
particular coordinate system they are identical in all transformed coordinate systems.
This means that tensor equalities and identities are invariant under coordinate transfor-
mations, which is one of the main motivations for the use of tensors in mathematics and
science. As indicated, all rank-0 tensors (scalars) are isotropic. Also, the zero vector, 0,
of any dimension is isotropic; in fact it is the only rank-1 isotropic tensor.

2.6.5 Symmetric and Anti-symmetric Tensors

These types of tensor apply to high ranks only (rank ≥ 2).[39] Moreover, these types are
not exhaustive, even for tensors of rank ≥ 2, as there are high-rank tensors which are
neither symmetric nor anti-symmetric. A rank-2 tensor Aij is symmetric iff for all i

[38]
For improper rotation, this is more general than being isotropic.
[39]
Symmetry and anti-symmetry of tensors require in their definition two free indices at least; hence a
scalar with no index and a vector with a single index do not qualify to be symmetric or anti-symmetric.
2.6.5 Symmetric and Anti-symmetric Tensors 71

and j the following condition is satisfied:

Aji = Aij (85)

and anti-symmetric or skew-symmetric iff for all i and j the following condition is
satisfied:
Aji = −Aij (86)

Similar conditions apply to contravariant type tensors (refer also to the following).
A rank-n tensor Ai1 ...in is symmetric in its two indices ij and il iff the following
condition applies identically:

Ai1 ...il ...ij ...in = Ai1 ...ij ...il ...in (87)

and anti-symmetric in its two indices ij and il iff the following condition applies iden-
tically:
Ai1 ...il ...ij ...in = −Ai1 ...ij ...il ...in (88)

Any rank-2 tensor Aij can be synthesized from (or decomposed into) a symmetric part
A(ij) , which is marked with round brackets enclosing the indices, and an anti-symmetric
part A[ij] , which is marked with square brackets, where the following relations apply:

1 1
Aij = A(ij) + A[ij] A(ij) = (Aij + Aji ) A[ij] = (Aij − Aji ) (89)
2 2

Similarly, a rank-3 tensor Aijk can be symmetrized by the following relation:

1
A(ijk) = (Aijk + Akij + Ajki + Aikj + Ajik + Akji ) (90)
3!
2.6.5 Symmetric and Anti-symmetric Tensors 72

and anti-symmetrized by the following relation:

1
A[ijk] = (Aijk + Akij + Ajki − Aikj − Ajik − Akji ) (91)
3!

More generally, a rank-n tensor Ai1 ...in can be symmetrized by:

1
A(i1 ...in ) = (sum of all even & odd permutations of indices i’s) (92)
n!

and anti-symmetrized by:

1
A[i1 ...in ] = (sum of all even permutations minus sum of all odd permutations) (93)
n!

A tensor of high rank (> 2) may be symmetrized or anti-symmetrized with respect to

only some of its indices instead of all of its indices. For example, in the following the
tensor A is symmetrized and anti-symmetrized only with respect to its first two indices:

1 1
A(ij)k = (Aijk + Ajik ) A[ij]k = (Aijk − Ajik ) (94)
2 2

A tensor is described as totally symmetric iff it is symmetric with respect to all of its
indices, that is:
Ai1 ...in = A(i1 ...in ) (95)

and totally anti-symmetric iff it is anti-symmetric in all of its indices, that is:

Ai1 ...in = A[i1 ...in ] (96)

For a totally anti-symmetric tensor, non-zero entries can occur only when all the indices
are different.
2.7 Exercises 73

It should be remarked that the indices whose exchange defines the symmetry and anti-
symmetry relations should be of the same variance type, i.e. both upper or both lower.
Another important remark is that the symmetry and anti-symmetry characteristic of a ten-
sor is invariant under coordinate transformations. Hence, a symmetric/anti-symmetric
tensor in one coordinate system is symmetric/anti-symmetric in all other coordinate sys-
tems. Similarly, a tensor which is neither symmetric nor anti-symmetric in one coordinate
system remains so in all other coordinate systems.[40]
Finally, for a symmetric tensor Aij and an anti-symmetric tensor B ij (or the other way
around) we have the following useful and widely used identity:

Aij B ij = 0 (97)

This is because an exchange of indices will change the sign of one tensor only and this will
change the sign of the term in the summation resulting in having a sum of terms which is
identically zero due to the fact that each term in the sum has its own negation.

2.7 Exercises

2.1 Make a sketch of a rank-2 tensor Aij in a 4D space similar to Figure 11. What this
tensor looks like?

2.2 What are the two main types of notation used for labeling tensors? State two names
for each.

2.3 Make a detailed comparison between the two types of notation in the previous question
stating any advantages or disadvantages in using one of these notations or the other.

[40]
In this context, like many other contexts in this book, there are certain restrictions on the type and
conditions of the coordinate transformations under which such statements are valid. However, these
details cannot be discussed here due to the elementary level of this book.
2.7 Exercises 74

In which context each one of these notations is more appropriate to use than the
other?

2.4 What is the principle of invariance of tensors and why it is one of the main reasons
for the use of tensors in science?

2.5 What are the two different meanings of the term “covariant” in tensor calculus?

2.6 State the type of each one of the following tensors considering the number and position
of indices (i.e. covariant, contravariant, rank, scalar, vector, etc.):

ai Bijk f bk C ji

2.7 Define the following technical terms which are related to tensors: term, expression,
equality, order, rank, zero tensor, unit tensor, free index, dummy index, covariant,
contravariant, and mixed.

2.8 Which of the following is a scalar, vector or rank-2 tensor: temperature, stress, cross
product of two vectors, dot product of two vectors, and rate of strain?

2.9 What is the number of entries of a rank-0 tensor in a 2D space and in a 5D space?
What is the number of entries of a rank-1 tensor in these spaces?

2.10 What is the difference between the order and rank of a tensor considering the different
conventions in this regard?

2.11 What is the number of entries of a rank-3 tensor in a 4D space? What is the number
of entries of a rank-4 tensor in a 3D space?

2.12 Describe direct and inverse coordinate transformations between spaces and write the
generic equations for these transformations.
2.7 Exercises 75

2.13 What are proper and improper transformations? Draw a simple sketch to demonstrate
them.

2.14 Define the following terms related to permutation of indices: permutation, even, odd,
parity, and transposition.

2.15 Find all the permutations of the following four letters assuming no repetition: (i, j, k, l).

2.16 Give three even permutations and three odd permutations of the symbols (α, β, γ, δ)
in the stated order.

2.17 Discuss all the similarities and differences between free and dummy indices.

2.18 What is the maximum number of repetitive indices that can occur in each term of a
legitimate tensor expression?

2.19 How many components are represented by each one of the following assuming a 4D
space?

Ajk
i f +g C mn − Dnm 5Dk + 4Ak = Bk

2.20 What is the “summation convention”? To what type of indices this convention applies?

2.21 Is it always the case that the summation convention applies when an index is repeated?
If not, what precaution should be taken to avoid ambiguity and confusion?

2.22 In which cases a pair of dummy indices should be of different variance type (i.e. one
upper and one lower)? In what type of coordinate systems these repeated indices can
be of the same variance type and why?

2.23 What are the rules that the free indices should obey when they occur in the terms of
tensor expressions and equalities?
2.7 Exercises 76

2.24 What is illegitimate about the following tensor expressions and equalities considering
in your answer all the possible violations?

Aij + Bijk C n − Dn = Bm Aij = Aji Aj = f

2.25 Which of the following tensor expressions and equalities is legitimate and which is
illegitimate?

B i + Cjij Ai − Bik C m + Dm = Bmm

m
Bki = Aik

State in each illegitimate case all the reasons for illegitimacy.

2.26 Which is right and which is wrong of the following tensor equalities?

∂n An = An ∂n [B]k + [D]k = [B + D]k ab = ba Aij Mkl = Mkl Aji

Explain in each case why the equality is right or wrong.

2.27 Choose from the Index six entries related to the general rules that apply to the math-
ematical expressions and equalities in tensor calculus.

2.28 Give at least two examples of tensors used in mathematics, science and engineering
for each one of the following ranks: 0, 1, 2 and 3.

2.29 State the special names given to the rank-0 and rank-1 tensors.

2.30 What is the difference, if any, between rank-2 tensors and matrices?

2.31 Is the following statement correct? If not, re-write it correctly: “all rank-0 tensors are
vectors and vice versa, and all rank-1 tensors are scalars and vice versa”.
2.7 Exercises 77

2.32 Give clear and detailed definitions of scalars and vectors and compare them. What is
common and what is different between the two?

2.33 Make a simple sketch of the nine unit dyads associated with the double directions of
rank-2 tensors in a 3D space.

2.34 Name three of the scientific disciplines that heavily rely on tensor calculus notation
and techniques.

2.35 What are the main features of tensor calculus that make it very useful and successful
in mathematical, scientific and engineering applications.

2.36 Why tensor calculus is used in the formulation and presentation of the laws of physics?

2.37 Give concise definitions for the covariant and contravariant types of tensor.

2.38 Describe how the covariant and contravariant types are notated and how they differ
in their transformation between coordinate systems.

2.39 Give examples of tensors used in mathematics and science which are covariant and
other examples which are contravariant.

2.40 Write the mathematical transformation rules of the following tensors: Aijk to Ārst
and B mn to B̄ pq .

2.41 Explain how mixed type tensors are defined and notated in tensor calculus.

2.42 Write the mathematical rule for transforming the mixed type tensor Dijklm to D̄pqrst .

2.43 From the Index, find all the terms that start with the word “Mixed” and are related
specifically to tensors of rank 2.
2.7 Exercises 78

2.44 Express the following tensors in indicial notation: a rank-3 covariant tensor A, a
rank-4 contravariant tensor B, a rank-5 mixed type tensor C which is covariant in ij
indices and contravariant in kmn indices where the indices are ordered as ikmnj.

2.45 Write step-by-step, similar to the detailed example given in § 2.6.1, the mathematical
transformations of the following tensors: Aij to Ārs , B lmn to B̄ pqr , C ijmn to C̄ pqrs and
Dmkl to D̄r st .

2.46 What is the relation between the rank and the (m, n) type of a tensor?

2.47 Write, in indicial notation, the following tensors: A of type (0, 4), B of type (3, 1), C
of type (0, 0), D of type (3, 4), E of type (2, 0) and F of type (1, 1).

2.48 What is the rank of each one of the tensors in the previous question? Are there tensors
among them which may not have been notated properly?

2.49 Which tensor provides the link between the covariant and contravariant types of a
given tensor D?

2.50 What coordinate system(s) in which the covariant and contravariant types of a tensor
do not differ? What is the usual tensor notation used in this case?

2.51 Define in detail, qualitatively and mathematically, the covariant and contravariant
types of the basis vectors of a general coordinate system explaining all the symbols
used in your definition.

2.52 Is it necessary that the basis vectors of the previous exercise are mutually orthogonal
and/or of unit length?

2.53 Is the following statement correct? “A superscript in the denominator of partial

derivatives is equivalent to a superscript in the numerator”. Explain why.
2.7 Exercises 79

2.54 What is the reciprocity relation that links the covariant and contravariant basis vec-
tors? Express this relation mathematically.

2.55 What is the interpretation of the reciprocity relation (refer to Figure 14 in your
explanation)?

2.56 Are the covariant and contravariant forms of a specific tensor A represent the same
mathematical object? If so, in what sense they are equal from the perspective of
different coordinate systems?

2.57 Correct, if necessary, the following statement: “A tensor of any rank (≥ 1) can be rep-
resented covariantly using contravariant basis tensors of that rank, or contravariantly
using contravariant basis tensors, or in a mixed form using a mixed basis of the same
type”.

2.58 Make corrections, if needed, to the following equations assuming a general curvilinear
coordinate system where, in each case, all the possible ways of correction should be
considered:

B = B i Ei M = Mij Ei D = D i Ei Ej C = C i Ej F = F n En T = T rs Es Er

2.59 What is the technical term used to label the following objects: Ei Ej , Ei Ej , Ei Ej and
Ei Ej ? What they mean?

2.60 What sort of tensor components that the objects in the previous question should be
associated with?

2.61 What is the difference between true and pseudo vectors? Which of these is called
axial and which is called polar?

2.62 Make a sketch demonstrating the behavior of true and pseudo vectors.
2.7 Exercises 80

2.63 Is the following statement correct? “The terms of tensor expressions and equations
should be uniform in their true and pseudo type”. Explain why.

2.64 There are four possibilities for the direct product of two tensors of true and pseudo
types. Discuss all these possibilities with respect to the type of the tensor produced
by this operation and if it is true or pseudo. Also discuss in detail the cross product
and curl operations from this perspective.

2.65 Give examples for the true and pseudo types of scalars, vectors and rank-2 tensors.

2.66 Explain, in words and equations, the meaning of absolute and relative tensors. Do
these intersect in some cases with true and pseudo tensors (at least according to some
conventions)?

2.67 What “Jacobian” and “weight” mean in the context of absolute and relative tensors?

2.68 Someone stated: “A is a tensor of type (2, 4, −1)”. What these three numbers refer
to?

2.69 What is the type of the tensor in the previous exercise from the perspectives of lower
and upper indices and absolute and relative tensors? What is the rank of this tensor?

2.70 What is the weight of a tensor A produced from multiplying a tensor of weight −1
by a tensor of weight 2? Is A relative or absolute? Is it true or not?

2.71 Define isotropic and anisotropic tensors and give examples for each using tensors of
different ranks.

2.72 What is the state of the inner and outer products of two isotropic tensors?

2.73 Why if a tensor equation is valid in a particular coordinate system it should also be
valid in all other coordinate systems under admissible coordinate transformations?
Use the isotropy of the zero tensor in your explanation.
2.7 Exercises 81

2.74 Define “symmetric” and “anti-symmetric” tensors and write the mathematical condi-
tion that applies to each assuming a rank-2 tensor.

2.75 Do we have symmetric/anti-symmetric scalars or vectors? If not, why?

2.76 Is it the case that any tensor of rank > 1 should be either symmetric or anti-
symmetric?

2.77 Give an example, writing all the components in numbers or symbols, of a symmetric
tensor of rank-2 in a 3D space. Do the same for an anti-symmetric tensor of the same
rank.

2.78 Give, if possible, an example of a rank-2 tensor which is neither symmetric nor anti-
symmetric assuming a 4D space.

2.79 Using the Index in the back of the book, gather all the terms related to the symmetric
and anti-symmetric tensors including the symbols used in their notations.

2.80 Is it true that any rank-2 tensor can be decomposed into a symmetric part and an
anti-symmetric part? If so, write down the mathematical expressions representing
these parts in terms of the original tensor. Is this also true for a general rank-n
tensor?

2.81 What is the meaning of the round and square brackets which are used to contain
indices in the indexed symbol of a tensor (e.g. A(ij) and B [km]n )?

2.82 Can the indices of symmetry/anti-symmetry be of different variance type?

2.83 Is it possible that a rank-n (n > 2) tensor is symmetric/anti-symmetric with respect

to some, but not all, of its indices? If so, give an example of a rank-3 tensor which is
symmetric or anti-symmetric with respect to only two of its indices.
2.7 Exercises 82

2.84 For a rank-3 covariant tensor Aijk , how many possibilities of symmetry and anti-
symmetry do we have? Consider in your answer total, as well as partial, symmetry
and anti-symmetry. Is there another possibility (i.e. the tensor is neither symmetric
nor anti-symmetric with respect to any pair of its indices)?

2.85 Can a tensor be symmetric with respect to some combinations of its indices and
anti-symmetric with respect to the other combinations? If so, can you give a simple
example of such a tensor?

2.86 Repeat the previous exercise considering the additional possibility that the tensor is
neither symmetric nor anti-symmetric with respect to another set of indices, i.e. it is
symmetric, anti-symmetric and neither with respect to different sets of indices.[41]

2.87 A is a rank-3 totally symmetric tensor and B is a rank-3 totally anti-symmetric tensor.
Write all the mathematical conditions that these tensors satisfy.

2.88 Justify the following statement: “For a totally anti-symmetric tensor, non-zero entries
can occur only when all the indices are different”. Use mathematical, as well as
descriptive, language in your answer.

2.89 For a totally anti-symmetric tensor Bijk in a 3D space, write all the elements of this
tensor which are identically zero. Consider the possibility that it may be easier to
find first the elements which are not identically zero, then exclude the rest.[42]

[41]
The best way to tackle this sort of exercises is to build a table or array of appropriate dimensions
where the indexed components in symbolic or numeric formats are considered and a trial and error
approach is used to investigate the possibility of creating such a tensor.
[42]
The concepts of repetitive and non-repetitive permutations may be useful in tackling this question.
Chapter 3

Tensor Operations

There are various operations that can be performed on tensors to produce other tensors in
general. Examples of these operations are addition/subtraction, multiplication by a scalar
(rank-0 tensor), multiplication of tensors (each of rank > 0), contraction and permutation.
Some of these operations, such as addition and multiplication, involve more than one
tensor while others, such as contraction and permutation, are performed on a single
tensor. In this chapter we provide a glimpse on the main elementary tensor operations
of algebraic nature that permeate tensor algebra and calculus.
First, we should remark that the last section of this chapter, which is about the quotient
rule for tensor test, is added to this chapter because it is the most appropriate place for
it in the present book considering the dependency of the definition of this rule on other
tensor operations; otherwise the section is not about a tensor operation in the same sense
as the operations presented in the other sections of this chapter. Another remark is that in
tensor algebra division is allowed only for scalars, hence if the components of an indexed
tensor should appear in a denominator, the tensor should be redefined to avoid this, e.g.
1
Bi = Ai
.

83
3.1 Addition and Subtraction 84

3.1 Addition and Subtraction

Tensors of the same rank and type[43] can be added algebraically to produce a tensor
of the same rank and type, e.g.

a=b+c Ai = Bi − Ci Aij = Bji + Cji (98)

The added/subtracted terms should have the same indicial structure with regard to their
j
free indices, as explained in § 2.3; hence Aijk and Bik cannot be added or subtracted al-
though they are of the same rank and type, but Ami i
mjk and Bjk can be added and subtracted.

Addition of tensors is associative and commutative, that is:[44]

(A + B) + C = A + (B + C) (99)

A+B = B+A (100)

3.2 Multiplication of Tensor by Scalar

A tensor can be multiplied by a scalar, which generally should not be zero, to produce
a tensor of the same variance type, rank and indicial structure, e.g.

Ajik = aBik
j
(101)

where a is a non-zero scalar. As indicated by the equation, multiplying a tensor by a

scalar means multiplying each component of the tensor by that scalar. Multiplication

[43]
Here, “type” refers to variance type (covariant/contravariant/mixed) and true/pseudo type as well as
other qualifications to which the tensors participating in an addition or subtraction operation should
match such as having the same weight if they are relative tensors, as outlined previously (refer for
example to § 2.6.3).
[44]
Associativity and commutativity can include subtraction if the minus sign is absorbed in the subtracted
tensor; in which case the operation is converted to addition.
3.3 Tensor Multiplication 85

by a scalar is commutative, and associative when more than two factors are involved.

3.3 Tensor Multiplication

This may also be called outer or exterior or direct or dyadic multiplication, although
some of these names may be reserved for operations on vectors. On multiplying each
component of a tensor of rank r by each component of a tensor of rank k, both of dimension
m, a tensor of rank (r + k) with mr+k components is obtained where the variance type
of each index (covariant or contravariant) is preserved.
For example, if A and B are covariant tensors of rank-1, then on multiplying A by B
we obtain a covariant tensor C of rank-2 where the components of C are given by:

Cij = Ai Bj (102)

while on multiplying B by A we obtain a covariant tensor D of rank-2 where the compo-

nents of D are given by:
Dij = Bi Aj (103)

Similarly, if A is a contravariant tensor of rank-2 and B is a covariant tensor of rank-2,

then on multiplying A by B we obtain a mixed tensor C of rank-4 where the components
of C are given by:

C ijkl = Aij Bkl (104)

while on multiplying B by A we obtain a mixed tensor D of rank-4 where the components

of D are given by:

Dijkl = Bij Akl (105)

3.3 Tensor Multiplication 86

In the outer product operation, it is generally understood that all the indices of the involved
tensors have the same range although this may not always be the case.[45]
In general, the outer product of tensors yields a tensor. The outer product of a tensor
of type (m, n) by a tensor of type (p, q) results in a tensor of type (m + p, n + q). This
means that the tensor rank in the outer product operation is additive and the operation
conserves the variance type of each index of the tensors involved.
The direct multiplication of tensors may be marked by the symbol , mostly when using
symbolic notation for tensors, e.g. A  B. However, in the present book no symbol is
being used for the operation of direct multiplication and hence the operation is symbolized
by putting the symbols of the tensors side by side, e.g. AB where A and B are non-
scalar tensors. In this regard, the reader should be vigilant to avoid confusion with the
operation of matrix multiplication which, according to the notation of matrix algebra,
is also symbolized as AB where A and B are matrices of compatible dimensions, since
matrix multiplication is an inner product, rather than an outer product, operation.
The direct multiplication of tensors is not commutative in general as indicated above;
however it is distributive with respect to the algebraic sum of tensors, that is:[46]

AB 6= BA (106)

A (B ± C) = AB ± AC (B ± C) A = BA ± CA (107)

As indicated before, the rank-2 tensor constructed by the direct multiplication of two
vectors is commonly called dyad. Tensors may be expressed as an outer product of vectors
where the rank of the resultant product is equal to the number of the vectors involved, e.g.
2 for dyads and 3 for triads. However, not every tensor can be synthesized as a product of
[45]
As indicated before, there are cases of tensors which are not uniformly dimensioned, and in some cases
these tensors can be regarded as the result of an outer product of lower rank tensors.
[46]
Regarding the associativity of direct multiplication, there seems to be cases in which this operation is
not associative. The interested reader is advised to refer to the research literature on this subject.
3.4 Contraction 87

lower rank tensors. Multiplication of a tensor by a scalar (refer to § 3.2) may be regarded
as a special case of direct multiplication.

3.4 Contraction

The contraction operation of a tensor of rank > 1 is to make two free indices identical,
by unifying their symbols, and perform summation over these repeated indices, e.g.

Aji contraction
−−−−−−−−→ Aii (108)

Ajk
il contraction on jl Amk
im (109)
−−−−−−−−−−−−→

Contraction results in a reduction of the rank by 2 since it implies the annihilation of

two free indices. Therefore, the contraction of a rank-2 tensor is a scalar, the contraction
of a rank-3 tensor is a vector, the contraction of a rank-4 tensor is a rank-2 tensor, and so
on.
For general non-Cartesian coordinate systems, the pair of contracted indices should
be different in their variance type, i.e. one upper and one lower. Hence, contraction of a
mixed tensor of type (m, n) will, in general, produce a tensor of type (m − 1, n − 1). A
tensor of type (p, q) can, therefore, have p × q possible contractions, i.e. one contraction
for each combination of lower and upper indices.
A common example of contraction is the dot product operation on vectors (see § Dot
Product) which can be regarded as a direct multiplication (refer to § 3.3) of the two vectors,
which results in a rank-2 tensor, followed by a contraction. Also, in matrix algebra, taking
the trace of a square matrix, by summing its diagonal elements, can be considered as a
contraction operation on the rank-2 tensor represented by the matrix, and hence it yields
the trace which is a scalar.
Conducting a contraction operation on a tensor results into a tensor. Similarly, the
3.5 Inner Product 88

application of a contraction operation on a relative tensor (see § 2.6.3) produces a relative

tensor of the same weight as the original tensor.

3.5 Inner Product

On taking the outer product (refer to § 3.3) of two tensors of rank ≥ 1 followed by a
contraction (refer to § 3.4) on two indices of the product, an inner product of the two
tensors is formed. Hence, if one of the original tensors is of rank-m and the other is of
rank-n, the inner product will be of rank-(m + n − 2). In the symbolic notation of tensor
calculus, the inner product operation is usually symbolized by a single dot between the
two tensors, e.g. A · B, to indicate the contraction operation which follows the outer
multiplication.
In general, the inner product is not commutative. When one[47] or both of the tensors
involved in the inner product are of rank > 1 then the order of the multiplicands does
matter in general, that is:[48]
A · B 6= B · A (110)

However, the inner product operation is distributive with respect to the algebraic sum
of tensors, that is:

A · (B ± C) = A · B ± A · C (B ± C) · A = B · A ± C · A (111)
[47]
The non-commutativity of the inner product in the case of only one of the involved tensors is of rank
> 1 may not be obvious; however, a simple example is multiplying Ai and Bjkl with a contraction of
j and k or j and l.
[48]
In fact, this statement is rather vague and rudimentary and may not apply in some cases. There
are many details related to the issue of commutativity of inner product of tensors which cannot be
discussed here due to the level of this book. In general, several issues should be considered in this
regard such as the order of the indices in the outer product of the two tensors involved in the inner
product, the (m, n) type of the outer product and whether the contracted indices are contributed by
the same tensor or the two tensors involved in the product assuming that the first case is conventionally
an inner product operation. Another important issue to be considered is that the contracted indices
must, in general, be of opposite variance type. Many of these details can be worked out rather easily
by the vigilant reader from first principles if they cannot be obtained from the textbooks of tensor
calculus.
3.5 Inner Product 89

As indicated before (see § 3.4), the dot product of two vectors is an example of the
inner product of tensors, i.e. it is an inner product of two rank-1 tensors to produce a
rank-0 tensor. For example, if a is a covariant vector and b is a contravariant vector, then
their dot product can be depicted as follow:

[ab]ij = ai bj contraction
−−−−−−−−→ a · b = ai b i (112)

Another common example, from linear algebra, of inner product is the multiplication
of a matrix representing a rank-2 tensor, by a vector, which is a rank-1 tensor, to produce
a vector. For example, if A is a rank-2 covariant tensor and b is a contravariant vector,
then their inner product can be depicted, according to tensor calculus, as follow:[49]

[Ab]ijk = Aij bk contraction on jk [A · b]i = Aij bj (113)

−−−−−−−−−−−−−→

This operation is equivalent to the above mentioned operation of multiplying a matrix by

a vector as defined in linear algebra.
The multiplication of two n × n matrices, as defined in linear algebra, to produce
another n × n matrix is another example of inner product (see Eq. 171). In this operation,
each one of the matrices involved in the multiplication, as well as the product itself, can
represent a rank-2 tensor.
For tensors whose outer product produces a tensor of rank > 2 and type (m, n) where
m, n > 0, various contraction operations between different pairs of indices of opposite
variance type can occur and hence more than one inner product, which are different in
general, can be defined. Moreover, when the outer product produces a tensor of rank > 3

[49]
It should be emphasized that we are using the symbolic notation of tensor calculus, rather than the
matrix notation, in writing Ab and A · b to represent, respectively, the outer and inner products. In
matrix notation, Ab is used to represent the product of a matrix by a vector which is an inner product
according to the terminology of tensor calculus.
3.5 Inner Product 90

and type (m, n) where m, n > 1, more than one contraction can take place simultane-
ously.[50]
There are more specialized types of inner product; some of these may be defined
differently by different authors. For example, a double inner product of two rank-2 tensors,
A and B, may be defined and denoted by double vertically- or horizontally-aligned dots
(e.g. A : B or A · · B) to indicate double contraction taking place between different
pairs of indices. An instance of these types is the inner product with double contraction
of two dyads which is commonly defined by:[51]

ab : cd = (a · c) (b · d) (114)

The single dots in the right hand side of this equation symbolize the conventional dot
product of two vectors. The result of this operation is obviously a scalar since it is the
product of two scalars, as seen from the right hand side of the equation.
Some authors may define a different type of double contraction inner product of two
dyads, symbolized by two horizontally-aligned dots, which may be called a transposed
contraction, and is given by:

ab · · cd = ab : dc = (a · d) (b · c) (115)

where the result is also a scalar. However, different authors may have different conventions
and hence one should be on the lookout for such differences.
For two rank-2 tensors, the aforementioned double contraction inner products are sim-

[50]
In these statements, we assume that the contracted indices can be contributed by the same tensor in
the product as well as by the two tensors (i.e. one index from each tensor); otherwise more details are
required. We are also assuming a general coordinate system where the contracted indices should be
opposite in their variance type.
[51]
This is also defined differently by some authors.
3.6 Permutation 91

ilarly defined as in the case of two dyads, that is:

A : B = Aij Bij A · · B = Aij Bji (116)

Inner products with higher multiplicities of contraction are similarly defined, and hence
they can be regarded as trivial extensions of the inner products with lower contraction
multiplicities.
The inner product of tensors produces a tensor because the inner product is an outer
product operation followed by a contraction operation and both of these operations on
tensors produce tensors, as stated before.

3.6 Permutation

A tensor may be obtained by exchanging the indices of another tensor. For example, Aikj
is a permutation of the tensor Aijk . A common example of the permutation operation of
tensors is the transposition of a matrix (refer to § Special Matrices) representing a rank-2
tensor since the first and second indices, which represent the rows and columns of the
matrix, are exchanged in this operation.
It is obvious that tensor permutation applies only to tensors of rank > 1 since no
exchange of indices can occur on a scalar with no index or on a vector with a single index.
The collection of tensors obtained by permuting the indices of a reference tensor may be
called isomers.

3.7 Tensor Test: Quotient Rule

Sometimes a tensor-like object may be suspected for being a tensor; in such cases a test
based on what is called the “quotient rule” [52] can be used to clarify the situation. Accord-

[52]
This should not be confused with the quotient rule of differentiation.
3.8 Exercises 92

ing to this rule, if the inner product of a suspected tensor with a known tensor is a tensor
then the suspect is a tensor. In more formal terms, if it is not known if A is a tensor but
it is known that B and C are tensors; moreover it is known that the following relation
holds true in all rotated (i.e. properly-transformed) coordinate frames:

Apq...k...m Bij...k...n = Cpq...mij...n (117)

then A is a tensor. Here, A, B and C are respectively of ranks m, n and (m + n − 2),

where the rank of C is reduced by 2 due to the contraction on k which can be any index
of A and B independently.[53]
Testing for being a tensor can also be done by applying the first principles through
direct substitution in the transformation equations of tensors to see if the alleged tensor
satisfies the transformation rules or not. However, using the quotient rule is generally
more convenient and requires less work. It is noteworthy that the quotient rule may be
considered by some authors as a replacement for the division operation which is not defined
for tensors.

3.8 Exercises

3.1 Give preliminary definitions of the following tensor operations: addition, multiplica-
tion by a scalar, tensor multiplication, contraction, inner product and permutation.
Which of these operations involve a single tensor?

3.2 Give typical examples of addition/subtraction for rank-n (0 ≤ n ≤ 3) tensors.

3.3 Is it possible to add two tensors of different ranks or different variance types? Is
addition of tensors associative or commutative?

[53]
We assume, of course, that the rules of contraction of indices, such as being of opposite variance type
in the case of non-Cartesian coordinates, are satisfied in this operation.
3.8 Exercises 93

3.4 Discuss, in detail, the operation of multiplication of a tensor by a scalar and compare
it to the operation of tensor multiplication. Can we regard multiplying two scalars as
an example of multiplying a tensor by a scalar?

3.5 What is the meaning of the term “outer product” and what are the other terms used
to label this operation?

3.6 C is a tensor of rank-3 and D is a tensor of rank-2, what is the rank of their outer
product CD? What is the rank of CD if it is subjected subsequently to a double
contraction operation?

3.7 A is a tensor of type (m, n) and B is a tensor of type (s, t), what is the type of their
direct product AB?

3.8 Discuss the operations of dot and cross product of two vectors (see § Dot Product
and Cross Product) from the perspective of the outer product operation of tensors.

3.9 Collect from the Index all the terms related to the tensor operations of addition
and permutation and classify these terms according to each operation giving a short
definition of each.

3.10 Are the following two statements correct (make corrections if necessary)? “The outer
multiplication of tensors is commutative but not distributive over sum of tensors” and
“The outer multiplication of two tensors may produce a scalar”.

3.11 What is contraction of tensor? How many free indices are consumed in a single
contraction operation?

3.12 Is it possible that the contracted indices are of the same variance type? If so, what is
the condition that should be satisfied for this to happen?
3.8 Exercises 94

3.13 A is a tensor of type (m, n) where m, n > 1, what is its type after two contraction
operations assuming a general coordinate system?

3.14 Does the contraction operation change the weight of a relative tensor?

3.15 Explain how the operation of multiplication of two matrices, as defined in linear
algebra, involves a contraction operation. What is the rank of each matrix and what
is the rank of the product? Is this consistent with the rule of reduction of rank by
contraction?

3.16 Explain, in detail, the operation of inner product of two tensors and how it is related
to the operations of contraction and outer product of tensors.

3.17 What is the rank and type of a tensor resulting from an inner product operation
of a tensor of type (m, n) with a tensor of type (s, t)? How many possibilities do
we have for this inner product considering the different possibilities of the embedded
contraction operation?

3.18 Give an example of a commutative inner product of two tensors and another example
of a non-commutative inner product.

3.19 Is the inner product operation distributive over algebraic addition of tensors?

3.20 Give an example from matrix algebra of inner product of tensors explaining in detail
how the two are related.

3.21 Discuss specialized types of inner product operations that involve more than one
contraction operation focusing in particular on the operations A : B and A·· B where
A and B are two tensors of rank > 1.

3.22 A double inner product operation is conducted on a tensor of type (1, 1) with a tensor
of type (1, 2). How many possibilities do we have for this operation? What is the
3.8 Exercises 95

rank and type of the resulting tensor? Is it covariant, contravariant or mixed?

3.23 Gather from the Index all the terms that refer to notations used in the operations of
inner and outer product of tensors.

3.24 Assess the following statement considering the two meanings of the word “tensor”
related to the rank: “Inner product operation of two tensors does not necessarily
produce a tensor”. Can this statement be correct in a sense and wrong in another?

3.25 What is the operation of tensor permutation and how it is related to the operation of
transposition of matrices?

3.26 Is it possible to permute scalars or vectors and why?

3.27 What is the meaning of the term “isomers”?

3.28 Describe in detail the quotient rule and how it is used as a test for tensors.

3.29 Why the quotient rule is used instead of the standard transformation equations of
tensors?
Chapter 4

δ and  Tensors

In this chapter, we conduct a preliminary investigation about the δ and  tensors and
their properties and functions as well as the relation between them. These tensors are
of particular importance in tensor calculus due to their distinctive properties and unique
transformation attributes. They are numerical tensors with fixed components in all
coordinate systems. The first is called Kronecker delta or unit tensor, while the second
is called Levi-Civita,[54] permutation, anti-symmetric and alternating tensor.

4.1 Kronecker δ

The Kronecker δ is a rank-2 tensor in all dimensions. It is defined as:




1 (i = j)
δij = (i, j = 1, 2, . . . n) (118)


0 (i 6= j)

where n is the space dimension, and hence it can be considered as the identity matrix.
For example, in a 3D space the Kronecker δ tensor is given by:
   
 δ11 δ12 δ13   1 0 0 
   
[δij ] = 
 δ21 δ22 δ23
= 0
  1 0 
 (119)
   
δ31 δ32 δ33 0 0 1
[54]
This name is usually used for the rank-3 tensor. Also some authors distinguish between the permu-
tation tensor and the Levi-Civita tensor even for rank-3. Moreover, some of the common labels and
descriptions of  are more specific to rank-3.

96
4.2 Permutation  97

The components of the covariant, contravariant and mixed types of this tensor are the
same, that is:
δij = δ ij = δ ij = δij (120)

The Kronecker δ tensor is symmetric, that is:

δij = δji δ ij = δ ji (121)

where i, j = 1, 2, . . . , n. Moreover, it is conserved[55] under all proper and improper

coordinate transformations. Since it is conserved under proper transformations, it is an
isotropic tensor.[56]

4.2 Permutation 

The permutation tensor  has a rank equal to the number of dimensions, and hence a
rank-n permutation tensor has nn components. The rank-2 permutation tensor ij is
defined by:
12 = 1 21 = −1 11 = 22 = 0 (122)

Similarly, the rank-3 permutation tensor ijk is defined by:





 1 (i, j, k is even permutation of 1,2,3)



ijk = −1 (i, j, k is odd permutation of 1,2,3) (123)






 0 (repeated index)

Figure 17 is a graphical illustration of the rank-3 permutation tensor ijk while Figure

[55]
“Conserved” means that the tensor keeps the values of its components following a coordinate transfor-
mation.
[56]
Here, being conserved under all transformations is stronger than being isotropic as the former applies
even under improper coordinate transformations while isotropy is restricted to proper transformations.
4.2 Permutation  98

18, which may be used as a mnemonic device, demonstrates the cyclic nature of the
three even permutations of the indices of the rank-3 permutation tensor and the three odd
permutations of these indices assuming no repetition in indices. The three permutations
in each case are obtained by starting from a given number in the cycle and rotating in the
given direction to obtain the other two numbers in the permutation.

331 332 333

321 322 323

311 313
312
231 232 233

221 222 223

211 213
212
131 132
133

121 122 123

111 112 113

Figure 17: Graphical illustration of the rank-3 permutation tensor ijk where circular nodes
represent 0, square nodes represent 1 and triangular nodes represent −1.

The definition of the rank-n permutation tensor (i.e. i1 i2 ...in ) is similar to the definition
of the rank-3 permutation tensor with regard to the repetition in its indices (i1 , i2 , · · · , in )
4.2 Permutation  99

3 1 3 1
+1 −1

2 2
Even Odd
Figure 18: Graphical demonstration of the cyclic nature of the even and odd permutations
of the indices of the rank-3 permutation tensor assuming no repetition in indices.

and being even or odd permutations in their correspondence to (1, 2, · · · , n), that is:




 1 (i1 , i2 , . . . , in is even permutation of 1, 2, . . . , n)



i1 i2 ...in = −1 (i1 , i2 , . . . , in is odd permutation of 1, 2, . . . , n) (124)






 0 (repeated index)

As well as the inductive definition of the permutation tensor (as given by Eqs. 122, 123
and 124), the permutation tensor of any rank can also be defined analytically where the
entries of the tensor are calculated from closed form formulae. The entries of the rank-2
permutation tensor can be calculated from the following closed form equation:

ij = (j − i) (125)

Similarly, for the rank-3 permutation tensor we have:

1
ijk = (j − i) (k − i) (k − j) (126)
2
4.2 Permutation  100

while for the rank-4 permutation tensor we have:

1
ijkl = (j − i) (k − i) (l − i) (k − j) (l − j) (l − k) (127)
12

More generally, the entries of the rank-n permutation tensor can be obtained from the
following identity:
" #
Y
n−1
1 Y
n
1 Y
a1 a2 ···an = (aj − ai ) = (aj − ai ) (128)
i=1
i! j=i+1 S(n − 1) 1≤i<j≤n

where S(n − 1) is the super factorial function of the argument (n − 1) which is defined
by:
Y
k
S(k) = i! = 1! · 2! · . . . · k! (129)
i=1

A simpler formula for calculating the entries of the rank-n permutation tensor can be
obtained from the previous one by dropping the magnitude of the multiplication factors
and taking their signs only, that is:
!
Y Y
a1 a2 ···an = sgn (aj − ai ) = sgn (aj − ai ) (130)
1≤i<j≤n 1≤i<j≤n

where sgn(k) is the sign function of the argument k which is defined by:



+1
 (k > 0)



sgn(k) = −1 (k < 0) (131)






 0 (k = 0)

The permutation tensor is totally anti-symmetric (see § 2.6.5) in each pair of its
4.3 Useful Identities Involving δ or/and  101

indices, i.e. it changes sign on swapping any two of its indices, that is:

i1 ...ik ...il ...in = −i1 ...il ...ik ...in (132)

The reason is that any exchange of two indices requires an even/odd number of single-
step shifts to the right of the first index plus an odd/even number of single-step shifts to
the left of the second index, so the total number of shifts is odd and hence it is an odd
permutation of the original arrangement.
The permutation tensor is a pseudo tensor since it acquires a minus sign under an
improper orthogonal transformation of coordinates, i.e. inversion of axes with possible
superposition of rotation (see § 2.2). However, it is an isotropic tensor since it is conserved
under proper coordinate transformations.
The permutation tensor may be considered as a contravariant relative tensor of weight
+1 or a covariant relative tensor of weight −1. Hence, in 2D, 3D and nD spaces we
have the following identities for the components of the permutation tensor:[57]

ij = ij ijk = ijk i1 i2 ...in = i1 i2 ...in (133)

4.3 Useful Identities Involving δ or/and 

In the following subsections we introduce and discuss a number of common identities which
involve the Kronecker and permutations tensors. Some of these identities involve only one
of these tensors while others involve both.

[57]
We note that in equalities like this we are equating the components, as indicated above.
4.3.1 Identities Involving δ 102

4.3.1 Identities Involving δ

When an index of the Kronecker delta is involved in a contraction operation by repeating

an index in another tensor in its own term, the effect of this is to replace the shared index
in the other tensor by the other index of the Kronecker delta, that is:

δij Aj = Ai (134)

In such cases the Kronecker delta is described as an index replacement or substitution

operator. Hence, we have:
δij δjk = δik (135)

Similarly:
δij δjk δki = δik δki = δii = n (136)

where n is the space dimension. The last part of this equation (i.e. δii = n) can be easily
justified by the fact that δii is the trace of the identity tensor considering the summation
convention.
Due to the fact that the coordinates are independent of each other (see § 2.2), we also
have the following identity:[58]

∂xi
= ∂j xi = xi,j = δij (137)
∂xj

Hence, in an nD space we obtain the following identity from the last two identities:

∂i xi = δii = n (138)

[58]
This identity, like many other identities in this chapter and in the book in general, is valid even for
general coordinate systems although we use Cartesian notation to avoid unnecessary distraction at this
level. The alert reader should be able to notate such identities in their general forms.
4.3.1 Identities Involving δ 103

Based on the above identities and facts, the following identity can be shown to apply in
orthonormal Cartesian coordinate systems:

∂xi ∂xj
= = δij = δ ij (139)
∂xj ∂xi

This identity is based on the two facts that the coordinates are independent, and the
covariant and contravariant types are the same in orthonormal Cartesian coordinate sys-
tems.
Similarly, for a coordinate system with a set of orthonormal[59] basis vectors, such as the
orthonormal Cartesian system, the following identity can be easily proved:

ei · ej = δij (140)

where the indexed e are the basis vectors. This identity is no more than a mathematical
statement of the fact that the basis vectors in orthonormal systems are mutually orthogonal
and of unit length.
Finally, the double inner product of two dyads (see § 3.5) formed by an orthonormal set
of basis vectors of a given coordinate system satisfies the following identity:

ei ej : ek el = δik δjl (141)

which is a combination of Eq. 114 and Eq. 140.

[59]
As explained previously, “orthonormal” means that the vectors in the set are mutually orthogonal and
each one of them is of unit length.
4.3.2 Identities Involving  104

4.3.2 Identities Involving 

From the definition of the rank-3 permutation tensor, we have the following identity which
demonstrates the sense of cyclic order of the non-repetitive permutations of this tensor:

ijk = kij = jki = −ikj = −jik = −kji (142)

This identity is also a demonstration of the fact that the rank-3 permutation tensor is
totally anti-symmetric in all of its indices since a shift of any two indices reverses its
sign.[60] Moreover, it reflects the fact that this tensor has only one independent non-
zero component since any one of the non-zero entries, all of which are given by Eq. 142,
can be obtained from any other one of these entries.
We also have the following identity for the rank-n permutation tensor:[61]

i1 i2 ···in i1 i2 ···in = n! (143)

This identity is based on the fact that the left hand side is actually the sum of the
squares of i1 i2 ···in over all the n! non-repetitive permutations of n different indices where
the value of  of each one of these permutations is either +1 or −1 and hence in both cases
their square is 1.
The double inner product of the rank-3 permutation tensor and a symmetric tensor
Ajk is given by the following identity:

ijk Ajk = 0 (144)

[60]
This may also be seen to apply to the zero entries of this tensor which correspond to the permutations
with repetitive indices.
[61]
This is a product of the rank-n permutation tensor by itself entry-by-entry with the application of the
summation convention and hence it can be seen as a multi-contraction inner product of the permutation
tensor by itself.
4.3.2 Identities Involving  105

This is because an exchange of the two indices of Ajk does not affect its value due to
the symmetry of Ajk whereas a similar exchange in these indices in ijk results in a sign
change; hence each term in the sum has its own negative and therefore the total sum is
identically zero.
Another identity with a trivial outcome that involves the rank-3 permutation tensor
and a vector A is the following:

ijk Ai Aj = ijk Ai Ak = ijk Aj Ak = 0 (145)

This can be explained by the fact that, due to the commutativity of ordinary multiplica-
tion, an exchange of the indices in A’s will not affect the value but a similar exchange in
the corresponding indices of ijk will cause a change in sign; hence each term in the sum
has its own negative and therefore the total sum will be zero.
Finally, for a set of orthonormal basis vectors in a 3D space with a right-handed
coordinate system, the following identities are satisfied:

ei × ej = ijk ek (146)

ei · (ej × ek ) = ijk (147)

These identities are based, respectively, on the forthcoming definitions of the cross product
(see Eq. 173) and the scalar triple product (see Eq. 174) in tensor notation plus the fact
that these vectors are unit vectors.
4.3.3 Identities Involving δ and  106

4.3.3 Identities Involving δ and 

For the rank-2 permutation tensor, we have the following identity which involves the
Kronecker delta in 2D:
δik δil
ij kl = = δik δjl − δil δjk (148)
δjk δjl

This identity can simply be proved inductively by building a table for the values on the
left and right hand sides as the indices are varied. The pattern of the indices in the
determinant of this identity is simple, that is the indices of the first  provide the indices
for the rows while the indices of the second  provide the indices for the columns.[62]
Another useful identity involving the rank-2 permutation tensor with the Kronecker
delta in 2D is the following:
il kl = δik (149)

This can be obtained from the previous identity by replacing j with l followed by a minimal
algebraic manipulation using tensor calculus rules.[63]
Similarly, we have the following identity which correlates the rank-3 permutation tensor
to the Kronecker delta in 3D:

δil δim δin

ijk lmn = δjl δjm δjn = (δil δjm δkn + δim δjn δkl + δin δjl δkm ) − (150)

δkl δkm δkn

(δil δjn δkm + δim δjl δkn + δin δjm δkl )
[62]
The role of these indices in indexing the rows and columns can be shifted. This can be explained
by the fact that the positions of the two epsilons can be exchanged, since ordinary multiplication is
commutative, and hence the role of the epsilons in providing the indices for the rows and columns
will be shifted. This can also be done by taking the transposition of the array
of the determinant,
which does not change the value of the determinant since det (A) = det AT , with an exchange of
the indices of the Kronecker symbols since the Kronecker symbol is symmetric in its two indices.
[63]
That is:
il kl = δik δll − δil δlk = 2δik − δil δlk = 2δik − δik = δik
4.3.3 Identities Involving δ and  107

Again, the indices in the determinant of this identity follow the same pattern as that of
Eq. 148.
Another useful identity in this category is the following:

δil δim
ijk lmk = = δil δjm − δim δjl (151)
δjl δjm

This identity can be obtained from the identity of Eq. 150 by replacing n with k.[64] The
pattern of the indices in this identity is as before if we exclude the repetitive indices.
More generally, the determinantal form of Eqs. 148 and 150, which link the rank-2
and rank-3 permutation tensors to the Kronecker tensors in 2D and 3D spaces, can be
extended to link the rank-n permutation tensor to the Kronecker tensor in an nD space,
that is:
δi 1 j 1 δi 1 j 2 · · · δi 1 j n
δi 2 j 1 δi 2 j 2 · · · δi 2 j n
i1 i2 ···in j1 j2 ···jn = .. .. .. (152)
...
. . .
δin j1 δin j2 · · · δi n j n

Again, the pattern of the indices in the determinant of this identity in their relation to
the indices of the two epsilons follow the same rules as those of Eqs. 148 and 150.
The identity of Eq. 151, which may be called the epsilon-delta identity, the con-
tracted epsilon identity or the Levi-Civita identity, is very useful in manipulating
and simplifying tensor expressions and proving vector and tensor identities; examples of
which will be seen in § 5.6. The sequence of indices of the δ’s in the expanded form on the

[64]
That is:

ijk lmk = δil δjm δkk + δim δjk δkl + δik δjl δkm − δil δjk δkm − δim δjl δkk − δik δjm δkl
= 3δil δjm + δim δjl + δim δjl − δil δjm − 3δim δjl − δil δjm
= δil δjm − δim δjl
4.3.3 Identities Involving δ and  108

right hand side of this identity can be easily memorized using the following mnemonic
expression:[65]
(F F × SS) − (F S × SF ) (153)

where the first and second F stand respectively for the first index in the first and second 
while the first and second S stand respectively for the second index in the first and second
, as illustrated graphically in Figure 19. The mnemonic device of Eq. 153 can also be
used to memorize the sequence of indices in Eq. 148.

FS SF
 i j k  l m k = δ il δ jm − δ im δ jl
FF SS
Figure 19: Graphical illustration of the mnemonic device of Eq. 153 which is used to
remember the sequence of indices in the epsilon-delta identity of Eq. 151.

Other common identities in this category are:

ijk ljk = 2δil (154)

ijk ijk = 2δii = 2 × 3 = 3! = 6 (155)

The first of these identities can be obtained from Eq. 151 with the replacement of m with
j followed by some basic tensor manipulation,[66] while the second can be obtained from
[65]
In fact, the determinantal form given by Eq. 151 can also be considered as a mnemonic device where
the first and second indices of the first  index the rows while the first and second indices of the second
 index the columns, as given above. However, the mnemonic device of Eq. 153 is more economic
in terms of the required work and more convenient in writing. It should also be remarked that the
determinantal form in all the above equations is in fact a mnemonic device for these equations where
the expanded form, if needed, can be easily obtained from the determinant which can be easily built
following the simple pattern of indices, as explained above.
[66]
That is:
ijk ljk = δil δjj − δij δjl = 3δil − δil = 2δil
4.4 Generalized Kronecker δ 109

the first by replacing l with i and applying the summation convention in 3D. The second
identity is, in fact, an instance of Eq. 143 for a 3D space. Its restriction to the 3D space is
justified by the fact that the rank and dimension of the permutation tensor are the same,
which is 3 in this case. As indicated previously, δii is the trace of the identity tensor, and
hence in a 3D space it is equal to 3.
Another one of the common identities involving the rank-3 permutation tensor with the
Kronecker delta in 3D is the following:

ijk δ1i δ2j δ3k = 123 = 1 (156)

This identity is based on the use of the Kronecker delta as an index replacement operator
where each one of the deltas replaces an index in the permutation tensor.
Finally, the following identity can be obtained from the definition of the rank-3 permuta-
tion tensor (Eq. 123) and the use of the Kronecker delta as an index replacement operator
(Eq. 134):
ijk δij = ijk δik = ijk δjk = 0 (157)

4.4 Generalized Kronecker δ

The generalized Kronecker delta is defined inductively by:





 1 (j1 . . . jn is even permutation of i1 . . . in )



δji11 ...i
...jn =
n
−1 (j1 . . . jn is odd permutation of i1 . . . in ) (158)






 0 (repeated i’s or j’s)
4.4 Generalized Kronecker δ 110

It can also be defined analytically by the following n × n determinant:

δji11 δji12 · · · δji1n

δji21 δji22 · · · δji2n
δji11 ...i
...jn
n
= .. .. . . . (159)
. . . ..

δjin1 δjin2 · · · δjinn

where the δji entries in the determinant are the ordinary Kronecker deltas as defined
previously. In this equation, the pattern of the indices in the generalized Kronecker delta
symbol δji11 ...i n
...jn in connection to the indices in the determinant is similar to the previous

patterns, that is the upper indices in δji11 ...i n

...jn provide the upper indices in the ordinary

deltas by indexing the rows of the determinant, while the lower indices in δji11 ...i n
...jn provide

the lower indices in the ordinary deltas by indexing the columns of the determinant.
From the above given identities, it can be shown that:

δji11 δji12 · · · δji1n

i1 ...in
δji21 δji22 · · · δji2n
 j1 ...jn = .. .. . . . (160)
. . . ..

δjin1 δjin2 · · · δjinn

Now, on comparing the last equation with the definition of the generalized Kronecker
delta, i.e. Eq. 159, we conclude that:

i1 ...in j1 ...jn = δji11 ...i

...jn
n
(161)

As an instance of Eq. 161, the relation between the rank-n permutation tensor in its
covariant and contravariant forms and the generalized Kronecker delta in an nD space is
4.4 Generalized Kronecker δ 111

given by:
i1 ...in = δi11... n
...in i1 ...in = δ i11...
...in
n (162)

where the first of these equations can be obtained from Eq. 161 by substituting (1 . . . n)
for (i1 . . . in ) in the two sides with relabeling j with i and noting that 1 ... n = 1, while the
second equation can be obtained from Eq. 161 by substituting (1 . . . n) for (j1 . . . jn ) and
noting that 1 ... n = 1.
Hence, the permutation tensor  can be considered as an instance of the generalized
Kronecker delta. Consequently, the rank-n permutation tensor can be written as an n × n
determinant consisting of the ordinary Kronecker deltas. Moreover, Eq. 162 can provide
another definition for the permutation tensor in its covariant and contravariant forms, in
addition to the previous inductive and analytic definitions of this tensor as given by Eqs.
124 and 128.
Returning to the widely used epsilon-delta identity of Eq. 151, if we define:[67]

ij ijk
δlm = δlmk (163)

and consider the above identities which correlate the permutation tensor, the generalized
Kronecker tensor and the ordinary Kronecker tensor, then an identity equivalent to Eq.
151 that involves only the generalized and ordinary Kronecker deltas can be obtained,
that is:
ij i j
δlm = δli δm
j
− δm δl (164)

The mnemonic device of Eq. 153 can also be used with this form of the identity with
minimal adjustments to the meaning of the symbols involved.
Other identities involving the permutation tensor and the ordinary Kronecker delta can
also be formulated in terms of the generalized Kronecker delta.
[67] ijk
In fact this can be obtained from the determinantal form of δlmn by relabeling n with k.
4.5 Exercises 112

4.5 Exercises

4.1 What “numerical tensor” means in connection with the Kronecker δ and the permu-
tation  tensors?

4.2 State all the names used to label the Kronecker and permutation tensors.

4.3 What is the meaning of “conserved under coordinate transformations” in relation to

the Kronecker and permutation tensors?

4.4 State the mathematical definition of the Kronecker δ tensor.

4.5 What is the rank of the Kronecker δ tensor in an nD space?

4.6 Write down the matrix representing the Kronecker δ tensor in a 3D space.

4.7 Is there any difference between the components of the covariant, contravariant and
mixed types of the Kronecker δ tensor?

4.8 Explain how the Kronecker δ acts as an index replacement operator giving an example
in a mathematical form.

4.9 How many mathematical definitions of the rank-n permutation tensor we have? State
one of these definitions explaining all the symbols involved.

4.10 What is the rank of the permutation tensor in an nD space?

4.11 Make a graphical illustration of the array representing the rank-2 and rank-3 permu-
tation tensors.

4.12 Is there any difference between the components of the covariant and contravariant
types of the permutation tensor?
4.5 Exercises 113

4.13 How the covariant and contravariant types of the permutation tensor are related to
the concept of relative tensor?

4.14 State the distinctive properties of the permutation tensor.

4.15 How many entries the rank-3 permutation tensor has? How many non-zero entries it
has? How many independent entries it has?

4.16 Is the permutation tensor true or pseudo and why?

4.17 State, in words, the cyclic property of the even and odd non-repetitive permutations
of the rank-3 permutation tensor with a simple sketch to illustrate this property.

4.18 From the Index, find all the terms which are common to the Kronecker and permu-
tation tensors.

4.19 Correct the following equations:

δij Aj = Aj δij δjk = δjk δij δjk δki = n! xi,j = δii

4.20 In what type of coordinate system the following equation applies?

∂i xj = ∂j xi

4.21 Complete the following equation assuming a 4D space:

∂ i xi = ?

4.22 Complete the following equations where the indexed e are orthonormal basis vectors
4.5 Exercises 114

of a particular coordinate system:

ei · ej = ? ei ej : ek el = ?

4.23 Write down the equations representing the cyclic order of the rank-3 permutation
tensor. What is the conclusion from these equations with regard to the symmetry or
anti-symmetry of this tensor and the number of its independent non-zero components?

4.24 Write the analytical expressions of the rank-3 and rank-4 permutation tensors.

4.25 Collect from the Index all the terms from matrix algebra which have connection to
the Kronecker δ.

4.26 Correct, if necessary, the following equations:

i1 ···in i1 ···in = n ijk Cj Ck = 0 ijk Djk = 0 ei × ej = ijk ej (ei × ej ) · ek = ijk

where C is a vector, D is a symmetric rank-2 tensor, and the indexed e are orthonormal
basis vectors in a 3D space with a right-handed coordinate system.

4.27 What is wrong with the following equations?

ijk δ1i δ2j δ3k = −1 ij kl = δik δjl + δil δjk il kl = δil ij ij = 3!

4.28 Write the following in their determinantal form describing the general pattern of the
relation between the indices of  and δ and the indices of the rows and columns of the
determinant:

ijk lmk ijk lmn i1 ···in j1 ···jn δji11 ...i
...jn
n
4.5 Exercises 115

4.29 Give two mnemonic devices used to memorize the widely used epsilon-delta identity
and make a simple graphic illustration for one of these.

4.30 Correct, if necessary, the following equations:

rst rst = 3! pst qst = 2δpq rst δrt = rst δst

4.31 State the mathematical definition of the generalized Kronecker delta δji11 ...i n
...jn .

4.32 Write each one of i1 ...in and i1 ...in in terms of the generalized Kronecker δ.

4.33 Make a survey, based on the Index, about the general mathematical terms used in the
operations conducted by using the Kronecker and permutation tensors.

4.34 Write the mathematical relation that links the covariant permutation tensor, the
contravariant permutation tensor, and the generalized Kronecker delta.

4.35 State the widely used epsilon-delta identity in terms of the generalized and ordinary
Kronecker deltas.
Chapter 5

Applications of Tensor Notation and Tech-

niques

In this chapter, we provide common definitions in the language of tensor calculus for
some basic concepts and operations from matrix and vector algebra. We also provide a
preliminary investigation of the nabla based differential operators and operations using, in
part, tensor notation. Common identities in vector calculus as well as the integral theorems
of Gauss and Stokes are also presented from this perspective. Finally, we provide a rather
extensive set of detailed examples about the use of tensor language and techniques in
proving common mathematical identities from vector calculus.

5.1 Common Definitions in Tensor Notation

In this section, we give some common definitions of concepts and operations from vector
and matrix algebra in tensor notation.
The trace of a matrix A representing a rank-2 tensor in an nD space is given by:

tr (A) = Aii (i = 1, . . . , n) (165)

For a 3 × 3 matrix representing a rank-2 tensor in a 3D space, the determinant is given

116
5.1 Common Definitions in Tensor Notation 117

by:
A11 A12 A13
det (A) = A21 A22 A23 = ijk A1i A2j A3k = ijk Ai1 Aj2 Ak3 (166)

A31 A32 A33

where the last two equalities represent the expansion of the determinant by row and by
column. Alternatively, the determinant of a 3 × 3 matrix can be given by:

1
det (A) = ijk lmn Ail Ajm Akn (167)
3!

More generally, for an n × n matrix representing a rank-2 tensor in an nD space, the

determinant is given by:

det (A) = i1 ···in A1i1 . . . Anin

= i1 ···in Ai1 1 . . . Ain n

1
= i ···i j ···j Ai j . . . Ain jn (168)
n! 1 n 1 n 1 1

The inverse of a 3 × 3 matrix A representing a rank-2 tensor is given by:

 −1  1
A ij = jmn ipq Amp Anq (169)
2 det (A)

The multiplication of a matrix A by a vector b, as defined in linear algebra, is given

by:
[Ab]i = Aij bj (170)

It should be remarked that in the writing of Ab we are using matrix notation. According
to the symbolic notation of tensors, the multiplication operation should be denoted by
a dot between the symbols of the tensor and the vector, i.e. A·b.[68]
[68]
Matrix multiplication in matrix algebra is equivalent to inner product in tensor algebra.
5.1 Common Definitions in Tensor Notation 118

The multiplication of two compatible matrices A and B, as defined in linear algebra,

is given by:
[AB]ik = Aij Bjk (171)

Again, we are using here matrix notation in the writing of AB; otherwise a dot should be
inserted between the symbols of the two matrices.
The dot product of two vectors of the same dimension is given by:

A · B = δij Ai Bj = Ai Bi (172)

The readers are referred to § 3.5 for a more general definition of this type of product that
includes higher rank tensors. Similarly, the cross product of two vectors in a 3D space
is given by:
[A × B]i = ijk Aj Bk (173)

The scalar triple product of three vectors in a 3D space is given by:

A1 A2 A3
A · (B × C) = B1 B2 B3 = ijk Ai Bj Ck (174)

C1 C2 C3

while the vector triple product of three vectors in a 3D space is given by:

[A × (B × C)]i = ijk klm Aj Bl Cm (175)

The expression of the other principal form of the vector triple product [i.e. (A × B) × C]
can be obtained from the above form by changing the order of the factors in the external
cross product and reversing the sign; other operations, like relabeling the indices and
exchanging some of the indices of the epsilons with a shift in sign, can then follow to
5.2 Scalar Invariants of Tensors 119

obtain a more organized form. The expressions of the subsidiary forms of the vector triple
product [e.g. B×(A × C) or (A × C)×B] can be obtained from the above with relabeling
the vectors in the indicial form according to their order in the symbolic form.

5.2 Scalar Invariants of Tensors

In the following, we list and write in tensor notation a number of invariants of low rank
tensors which have special importance due to their widespread applications in vector and
tensor calculus. All these invariants are scalars.
The value of a scalar (rank-0 tensor), which consists of a magnitude and a sign, is in-
variant under coordinate transformations. An invariant of a vector (rank-1 tensor) under
coordinate transformations is its magnitude, i.e. length.[69] The main three independent
scalar invariants of a rank-2 tensor A under a change of basis are:

I = tr (A) = Aii (176)

II = tr A2 = Aij Aji (177)


III = tr A3 = Aij Ajk Aki (178)

Different forms of the three invariants of a rank-2 tensor A, which are also widely used,
are the following (noting that some of these definitions may belong to 3D specifically):

I1 = I = Aii (179)
1 2
1
I2 = I − II = (Aii Ajj − Aij Aji ) (180)
2 2
1 3
1
I3 = det (A) = I − 3I II + 2III = ijk pqr Aip Ajq Akr (181)
3! 3!

where I, II and III are as defined in the previous set of equations.

[69]
The direction is also invariant but it is not a scalar! In fact the magnitude alone is invariant under
coordinate transformations even for pseudo vectors because it is a true scalar.
5.3 Common Differential Operations in Tensor Notation 120

The invariants I, II and III can similarly be defined in terms of the invariants I1 , I2
and I3 as follow:

I = I1 (182)

II = I12 − 2I2 (183)

III = I13 − 3I1 I2 + 3I3 (184)

Since the determinant of a matrix representing a rank-2 tensor is invariant, as given

above, then if the determinant vanishes in one coordinate system, it will vanish in all
transformed coordinate systems, and if not it will not. Consequently, if a rank-2 tensor is
invertible in a particular coordinate system, it will be invertible in all coordinate systems,
and if not it will not.
Ten joint invariants between two rank-2 tensors, A and B, can be formed; these
are: tr (A), tr (B), tr (A2 ), tr (B2 ), tr (A3 ), tr (B3 ), tr (A · B), tr (A2 · B), tr (A · B2 ) and
tr (A2 · B2 ).

5.3 Common Differential Operations in Tensor Notation

In the first part of this section, we present the most common nabla based differential
operations in Cartesian coordinates as defined by tensor notation. These operations are
based on the various types of interaction between the vector differential operator nabla
∇ with tensors of different ranks where some of these interactions involve the dot and
cross product operations. We then define in the subsequent parts of this section some of
these differential operations in the most commonly used non-Cartesian coordinate systems,
namely cylindrical and spherical, as well as in the general orthogonal coordinate
systems.
Regarding the cylindrical and spherical systems, we could have used indexed generalized
5.3.1 Cartesian Coordinate System 121

coordinates like x1 , x2 and x3 to represent the cylindrical coordinates (ρ, φ, z) and the
spherical coordinates (r, θ, φ) and hence we express these operations in tensor notation as
we did for the Cartesian system. However, for more clarity at this level and to follow the
more conventional practice, we use the coordinates of these systems as suffixes in place
of the usual indices used in the tensor notation.[70] Also, for clarity and simplicity of
some formulae, we use numbers to index the coordinates of the general orthogonal system,
where a 3D space is assumed, although this is not a compact way of representation since
the formulae are given in their expanded form.

5.3.1 Cartesian Coordinate System

The nabla operator ∇ is a spatial partial differential operator which is defined in Cartesian
coordinate systems by:
∂
∇i = (185)
∂xi

The gradient of a differentiable scalar function of position f is a vector obtained by

applying the nabla operator on f and hence it is defined by:

∂f
[∇f ]i = ∇i f = = ∂i f = f,i (186)
∂xi

Similarly, the gradient of a differentiable vector function of position A is the outer product
(refer to § 3.3) between the ∇ operator and the vector and hence it is a rank-2 tensor given
by:
[∇A]ij = ∂i Aj (187)

The divergence of a differentiable vector A is the dot product of the nabla operator

[70]
There is another reason that is the components given in the cylindrical and spherical coordinates are
physical, not covariant or contravariant, and hence suffixing with coordinates looks more appropriate.
The interested reader should consult on this issue more advanced textbooks of tensor calculus.
5.3.1 Cartesian Coordinate System 122

and the vector A and hence it is a scalar given by:

∂Ai ∂Ai
∇ · A = δij = = ∇i Ai = ∂i Ai = Ai,i (188)
∂xj ∂xi

The divergence operation can also be viewed as taking the gradient of the vector followed
by a contraction. Hence, the divergence of a vector is invariant because it is the trace of
a rank-2 tensor (see § 5.2).[71]
Similarly, the divergence of a differentiable rank-2 tensor A is a vector defined in one
of its forms by:
[∇ · A]i = ∂j Aji (189)

and in another form by:

[∇ · A]j = ∂i Aji (190)

These two different forms can be given, respectively, in symbolic notation by:

∇·A ∇ · AT (191)

where AT is the transpose of A.

More generally, the divergence of a tensor of rank n ≥ 2, which is a tensor of rank-
(n − 1), can be defined in several forms, which are different in general, depending on the
combination of the contracted indices.
The curl of a differentiable vector A is the cross product of the nabla operator and
the vector A and hence it is a vector defined by:

∂Ak
[∇ × A]i = ijk = ijk ∇j Ak = ijk ∂j Ak = ijk Ak,j (192)
∂xj

[71]
It may also be argued more simply that the divergence of a vector is a scalar and hence it is invariant.
5.3.1 Cartesian Coordinate System 123

The curl operation may be generalized to tensors of rank > 1, and hence the curl of a
differentiable rank-2 tensor A can be defined as a rank-2 tensor given by:

[∇ × A]ij = imn ∂m Anj (193)

The Laplacian scalar operator acting on a differentiable scalar f is given by:

∂ 2f ∂ 2f
∆f = ∇2 f = δij = = ∇ii f = ∂ii f = f,ii (194)
∂xi ∂xj ∂xi ∂xi

The Laplacian operator acting on a differentiable vector A is defined for each component
of the vector in a similar manner to the definition of the Laplacian acting on a scalar, that
is:
 2 
∇ A i = ∇2 [A]i = ∂jj Ai (195)

The following scalar differential operator is commonly used in science (e.g. in fluid
dynamics):
∂
A · ∇ = Ai ∇i = Ai = Ai ∂i (196)
∂xi

where A is a vector. As indicated earlier, the order of Ai and ∂i should be respected. The
following vector differential operator also has common applications in science:

[A × ∇]i = ijk Aj ∂k (197)

where, again, the order should be respected.

It should be remarked that the differentiation of a tensor increases its rank by one,
by introducing an extra covariant index, unless it implies a contraction in which case it
reduces the rank by one. Therefore the gradient of a scalar is a vector and the gradient
of a vector is a rank-2 tensor (∂i Aj ), while the divergence of a vector is a scalar and the
5.3.2 Cylindrical Coordinate System 124

divergence of a rank-2 tensor is a vector (∂j Aji or ∂i Aji ). This may be justified by the fact
that nabla ∇ is a vector operator. On the other hand the Laplacian operator does not
change the rank since it is a scalar operator; hence the Laplacian of a scalar is a scalar
and the Laplacian of a vector is a vector.

5.3.2 Cylindrical Coordinate System

For the cylindrical system identified by the coordinates (ρ, φ, z) with a set of orthonormal
basis vectors eρ , eφ and ez we have the following definitions for the nabla based operators
and operations.[72]
The nabla operator ∇ is given by:

1
∇ = eρ ∂ρ + eφ ∂φ + ez ∂z (198)
ρ

The Laplacian operator is given by:

1 1
∇2 = ∂ρρ + ∂ρ + 2 ∂φφ + ∂zz (199)
ρ ρ

The gradient of a differentiable scalar f is given by:

∂f 1 ∂f ∂f
∇f = eρ + eφ + ez (200)
∂ρ ρ ∂φ ∂z

The divergence of a differentiable vector A is given by:

 
1 ∂ (ρAρ ) ∂Aφ ∂ (ρAz )
∇·A= + + (201)
ρ ∂ρ ∂φ ∂z

[72]
It should be obvious that since ρ, φ and z are labels for specific coordinates and not variable indices,
the summation convention does not apply.
5.3.3 Spherical Coordinate System 125

The curl of a differentiable vector A is given by:

eρ ρeφ ez
1
∇×A= ∂ ∂ ∂ (202)
ρ ∂ρ ∂φ ∂z

Aρ ρAφ Az

For the plane polar coordinate system, these operators and operations can be obtained
by dropping the z components or terms from the cylindrical form of the above operators
and operations.

5.3.3 Spherical Coordinate System

For the spherical system identified by the coordinates (r, θ, φ) with a set of orthonormal
basis vectors er , eθ and eφ we have the following definitions for the nabla based operators
and operations.[73]
The nabla operator ∇ is given by:

1 1
∇ = er ∂r + eθ ∂θ + eφ ∂φ (203)
r r sin θ

The Laplacian operator is given by:

2 1 cos θ 1
∇2 = ∂rr + ∂r + 2 ∂θθ + 2 ∂θ + 2 2 ∂φφ (204)
r r r sin θ r sin θ

The gradient of a differentiable scalar f is given by:

∂f 1 ∂f 1 ∂f
∇f = er + eθ + eφ (205)
∂r r ∂θ r sin θ ∂φ

[73]
Again, the summation convention does not apply to r, θ and φ.
5.3.4 General Orthogonal Coordinate System 126

The divergence of a differentiable vector A is given by:

 
1 ∂ (r2 Ar ) ∂ (sin θAθ ) ∂Aφ
∇·A= 2 sin θ +r +r (206)
r sin θ ∂r ∂θ ∂φ

The curl of a differentiable vector A is given by:

er reθ r sin θeφ

1
∇×A= ∂ ∂ ∂ (207)
r2 sin θ ∂r ∂θ ∂φ

Ar rAθ r sin θAφ

5.3.4 General Orthogonal Coordinate System

For general orthogonal systems in a 3D space identified by the coordinates (u1 , u2 , u3 )

with a set of unit basis vectors u1 , u2 and u3 and scale factors h1 , h2 and h3 where
∂r
hi = ∂ui
and r = xi ei is the position vector, we have the following definitions for the
nabla based operators and operations.
The nabla operator ∇ is given by:

u1 ∂ u2 ∂ u3 ∂
∇= 1
+ 2
+ (208)
h1 ∂u h2 ∂u h3 ∂u3

The Laplacian operator is given by:

      
2 1 ∂ h2 h3 ∂ ∂ h1 h3 ∂ ∂ h1 h2 ∂
∇ = + 2 + 3 (209)
h1 h2 h3 ∂u1 h1 ∂u1 ∂u h2 ∂u2 ∂u h3 ∂u3

The gradient of a differentiable scalar f is given by:

u1 ∂f u2 ∂f u3 ∂f
∇f = 1
+ 2
+ (210)
h1 ∂u h2 ∂u h3 ∂u3
5.4 Common Identities in Vector and Tensor Notation 127

The divergence of a differentiable vector A is given by:

 
1 ∂ ∂ ∂
∇·A= (h2 h3 A1 ) + 2 (h1 h3 A2 ) + 3 (h1 h2 A3 ) (211)
h1 h2 h3 ∂u1 ∂u ∂u

The curl of a differentiable vector A is given by:

h1 u1 h2 u2 h3 u3
1
∇×A= ∂ ∂ ∂ (212)
h1 h2 h3 ∂u1 ∂u2 ∂u3

h1 A1 h2 A2 h3 A3

5.4 Common Identities in Vector and Tensor Notation

In this section, we present some of the widely used identities of vector calculus using the
traditional vector notation as well as its equivalent tensor notation. In the following
bullet points, f and h are differentiable scalar fields; A, B, C and D are differentiable
vector fields; and r = xi ei is the position vector.
•

∇·r = n

m (213)

∂ i xi = n

where n is the space dimension.

5.4 Common Identities in Vector and Tensor Notation 128

∇×r = 0

m (214)

ijk ∂j xk = 0

∇ (a · r) = a

m (215)

∂i (aj xj ) = ai

where a is a constant vector.

•

∇ · (∇f ) = ∇2 f

m (216)

∂i (∂i f ) = ∂ii f

∇ · (∇ × A) = 0

m (217)

ijk ∂i ∂j Ak = 0
5.4 Common Identities in Vector and Tensor Notation 129

∇ × (∇f ) = 0

m (218)

ijk ∂j ∂k f = 0

∇ (f h) = f ∇h + h∇f

m (219)

∂i (f h) = f ∂i h + h∂i f

∇ · (f A) = f ∇ · A + A · ∇f

m (220)

∂i (f Ai ) = f ∂i Ai + Ai ∂i f

∇ × (f A) = f ∇ × A + ∇f × A

m (221)

ijk ∂j (f Ak ) = f ijk ∂j Ak + ijk (∂j f ) Ak

5.4 Common Identities in Vector and Tensor Notation 130

A · (B × C) = C · (A × B) = B · (C × A)

m m (222)

ijk Ai Bj Ck = kij Ck Ai Bj = jki Bj Ck Ai

A × (B × C) = B (A · C) − C (A · B)

m (223)

ijk Aj klm Bl Cm = Bi (Am Cm ) − Ci (Al Bl )

A × (∇ × B) = (∇B) · A − A · ∇B

m (224)

ijk klm Aj ∂l Bm = (∂i Bm ) Am − Al (∂l Bi )

∇ × (∇ × A) = ∇ (∇ · A) − ∇2 A

m (225)

ijk klm ∂j ∂l Am = ∂i (∂m Am ) − ∂ll Ai

5.4 Common Identities in Vector and Tensor Notation 131

∇ (A · B) = A × (∇ × B) + B × (∇ × A) + (A · ∇) B + (B · ∇) A

m (226)

∂i (Am Bm ) = ijk Aj (klm ∂l Bm ) + ijk Bj (klm ∂l Am ) + (Al ∂l ) Bi + (Bl ∂l ) Ai

∇ · (A × B) = B · (∇ × A) − A · (∇ × B)

m (227)

∂i (ijk Aj Bk ) = Bk (kij ∂i Aj ) − Aj (jik ∂i Bk )

∇ × (A × B) = (B · ∇) A + (∇ · B) A − (∇ · A) B − (A · ∇) B

m (228)

ijk klm ∂j (Al Bm ) = (Bm ∂m ) Ai + (∂m Bm ) Ai − (∂j Aj ) Bi − (Aj ∂j ) Bi

A·C A·D
(A × B) · (C × D) =
B·C B·D
m (229)

ijk Aj Bk ilm Cl Dm = (Al Cl ) (Bm Dm ) − (Am Dm ) (Bl Cl )

5.5 Integral Theorems in Tensor Notation 132

(A × B) × (C × D) = [D · (A × B)] C − [C · (A × B)] D

m (230)

ijk jmn Am Bn kpq Cp Dq = (qmn Dq Am Bn ) Ci − (pmn Cp Am Bn ) Di

• In vector and tensor notations, the condition for a vector field A to be solenoidal is:

∇·A = 0

m (231)

∂i Ai = 0

• In vector and tensor notations, the condition for a vector field A to be irrotational is:

∇×A = 0

m (232)

ijk ∂j Ak = 0

5.5 Integral Theorems in Tensor Notation

The divergence theorem for a differentiable vector field A in vector and tensor nota-
tions is given by:

˚ ¨
∇ · A dτ = A · n dσ
V S

m (233)
ˆ ˆ
∂i Ai dτ = Ai ni dσ
V S
5.5 Integral Theorems in Tensor Notation 133

where V is a bounded region in an nD space enclosed by a generalized surface S, dτ and dσ

are generalized volume and surface elements respectively, n and ni are unit vector normal
to the surface and its ith component respectively, and the index i ranges over 1, . . . , n.
Similarly, the divergence theorem for a differentiable rank-2 tensor field A in tensor
notation for the first index is given by:

ˆ ˆ
∂i Ail dτ = Ail ni dσ (234)
V S

while the divergence theorem for differentiable tensor fields of higher rank A in tensor
notation for the index k is given by:

ˆ ˆ
∂k Aij...k...m dτ = Aij...k...m nk dσ (235)
V S

Stokes theorem for a differentiable vector field A in vector and tensor notations is
given by:

¨ ˆ
(∇ × A) · n dσ = A · dr
S C

m (236)
ˆ ˆ
ijk ∂j Ak ni dσ = Ai dxi
S C

where C stands for the perimeter of the surface S, and dr is a vector element tangent to
the perimeter while the other symbols are as defined before.
Similarly, Stokes theorem for a differentiable rank-2 tensor field A in tensor notation
for the first index is given by:

ˆ ˆ
ijk ∂j Akl ni dσ = Ail dxi (237)
S C
5.6 Examples of Using Tensor Techniques to Prove Identities 134

while Stokes theorem for differentiable tensor fields of higher rank A in tensor notation
for the index k is given by:

ˆ ˆ
ijk ∂j Alm...k...n ni dσ = Alm...k...n dxk (238)
S C

5.6 Examples of Using Tensor Techniques to Prove Identities

In this section we provide some examples for using tensor techniques to prove vector
and tensor identities. These examples, which are based on the identities given in § 5.4,
demonstrate the elegance, efficiency and clarity of the methods and notation of tensor
calculus.
• ∇ · r = n:

∇ · r = ∂i xi (Eq. 188)

= δii (Eq. 138)

=n (Eq. 138)

• ∇ × r = 0:

[∇ × r]i = ijk ∂j xk (Eq. 192)

= ijk δkj (Eq. 137)

= ijj (Eq. 134)

=0 (Eq. 123)

Since i is a free index, the identity is proved for all components.

• ∇ (a · r) = a:

[∇ (a · r)]i = ∂i (aj xj ) (Eqs. 186 & 172)

= aj ∂i xj + xj ∂i aj (product rule)
5.6 Examples of Using Tensor Techniques to Prove Identities 135

= aj ∂i xj (aj is constant)

= aj δji (Eq. 137)

= ai (Eq. 134)

= [a]i (definition of index)

Since i is a free index, the identity is proved for all components.

• ∇ · (∇f ) = ∇2 f :

∇ · (∇f ) = ∂i [∇f ]i (Eq. 188)

= ∂i (∂i f ) (Eq. 186)

= ∂i ∂i f (rules of differentiation)

= ∂ii f (definition of 2nd derivative)

= ∇2 f (Eq. 194)

• ∇ · (∇ × A) = 0:

∇ · (∇ × A) = ∂i [∇ × A]i (Eq. 188)

= ∂i (ijk ∂j Ak ) (Eq. 192)

= ijk ∂i ∂j Ak (∂ not acting on )

= ijk ∂j ∂i Ak (continuity condition)

= −jik ∂j ∂i Ak (Eq. 142)

= −ijk ∂i ∂j Ak (relabeling dummy indices i and j)

=0 (since ijk ∂i ∂j Ak = −ijk ∂i ∂j Ak )

This can also be concluded from line three by arguing that: since by the continuity
condition ∂i and ∂j can change their order with no change in the value of the term while
5.6 Examples of Using Tensor Techniques to Prove Identities 136

a corresponding change of the order of i and j in ijk results in a sign change, we see that
each term in the sum has its own negative and hence the terms add up to zero (see Eq.
145).
• ∇ × (∇f ) = 0:

[∇ × (∇f )]i = ijk ∂j [∇f ]k (Eq. 192)

= ijk ∂j (∂k f ) (Eq. 186)

= ijk ∂j ∂k f (rules of differentiation)

= ijk ∂k ∂j f (continuity condition)

= −ikj ∂k ∂j f (Eq. 142)

= −ijk ∂j ∂k f (relabeling dummy indices j and k)

=0 (since ijk ∂j ∂k f = −ijk ∂j ∂k f )

This can also be concluded from line three by a similar argument to the one given in the
previous point. Because [∇ × (∇f )]i is an arbitrary component, then each component is
zero.
• ∇ (f h) = f ∇h + h∇f :

[∇ (f h)]i = ∂i (f h) (Eq. 186)

= f ∂i h + h∂i f (product rule)

= [f ∇h]i + [h∇f ]i (Eq. 186)

= [f ∇h + h∇f ]i (Eq. 66)

Because i is a free index, the identity is proved for all components.

• ∇ · (f A) = f ∇ · A + A · ∇f :

∇ · (f A) = ∂i [f A]i (Eq. 188)

5.6 Examples of Using Tensor Techniques to Prove Identities 137

= ∂i (f Ai ) (definition of index)

= f ∂i Ai + Ai ∂i f (product rule)

= f ∇ · A + A · ∇f (Eqs. 188 & 196)

• ∇ × (f A) = f ∇ × A + ∇f × A:

[∇ × (f A)]i = ijk ∂j [f A]k (Eq. 192)

= ijk ∂j (f Ak ) (definition of index)

= f ijk ∂j Ak + ijk (∂j f ) Ak (product rule & commutativity)

= f ijk ∂j Ak + ijk [∇f ]j Ak (Eq. 186)

= [f ∇ × A]i + [∇f × A]i (Eqs. 192 & 173)

= [f ∇ × A + ∇f × A]i (Eq. 66)

Because i is a free index, the identity is proved for all components.

• A · (B × C) = C · (A × B) = B · (C × A):

A · (B × C) = ijk Ai Bj Ck (Eq. 174)

= kij Ai Bj Ck (Eq. 142)

= kij Ck Ai Bj (commutativity)

= C · (A × B) (Eq. 174)

= jki Ai Bj Ck (Eq. 142)

= jki Bj Ck Ai (commutativity)

= B · (C × A) (Eq. 174)

The negative permutations of this identity can be similarly obtained and proved by
changing the order of the vectors in the cross products which results in a sign change.
5.6 Examples of Using Tensor Techniques to Prove Identities 138

• A × (B × C) = B (A · C) − C (A · B):

[A × (B × C)]i = ijk Aj [B × C]k (Eq. 173)

= ijk Aj klm Bl Cm (Eq. 173)

= ijk klm Aj Bl Cm (commutativity)

= ijk lmk Aj Bl Cm (Eq. 142)

= (δil δjm − δim δjl ) Aj Bl Cm (Eq. 151)

= δil δjm Aj Bl Cm − δim δjl Aj Bl Cm (distributivity)

= (δil Bl ) (δjm Aj Cm ) − (δim Cm ) (δjl Aj Bl ) (commutativity & grouping)

= Bi (Am Cm ) − Ci (Al Bl ) (Eq. 134)

= Bi (A · C) − Ci (A · B) (Eq. 172)

= [B (A · C)]i − [C (A · B)]i (definition of index)

= [B (A · C) − C (A · B)]i (Eq. 66)

Because i is a free index, the identity is proved for all components.

Other variants of this identity [e.g. (A × B) × C] can be obtained and proved similarly
by changing the order of the factors in the external cross product with adding a minus
sign.
• A × (∇ × B) = (∇B) · A − A · ∇B:

[A × (∇ × B)]i = ijk Aj [∇ × B]k (Eq. 173)

= ijk Aj klm ∂l Bm (Eq. 192)

= ijk klm Aj ∂l Bm (commutativity)

= ijk lmk Aj ∂l Bm (Eq. 142)

= (δil δjm − δim δjl ) Aj ∂l Bm (Eq. 151)

= δil δjm Aj ∂l Bm − δim δjl Aj ∂l Bm (distributivity)

5.6 Examples of Using Tensor Techniques to Prove Identities 139

= Am ∂i Bm − Al ∂l Bi (Eq. 134)

= (∂i Bm ) Am − Al (∂l Bi ) (commutativity & grouping)

= [(∇B) · A]i − [A · ∇B]i (Eq. 187 & § 3.5)

= [(∇B) · A − A · ∇B]i (Eq. 66)

Because i is a free index, the identity is proved for all components.

• ∇ × (∇ × A) = ∇ (∇ · A) − ∇2 A:

[∇ × (∇ × A)]i = ijk ∂j [∇ × A]k (Eq. 192)

= ijk ∂j (klm ∂l Am ) (Eq. 192)

= ijk klm ∂j (∂l Am ) (∂ not acting on )

= ijk lmk ∂j ∂l Am (Eq. 142 & definition of derivative)

= (δil δjm − δim δjl ) ∂j ∂l Am (Eq. 151)

= δil δjm ∂j ∂l Am − δim δjl ∂j ∂l Am (distributivity)

= ∂m ∂i Am − ∂l ∂l Ai (Eq. 134)

= ∂i (∂m Am ) − ∂ll Ai (∂ shift, grouping & Eq. 4)

 
= [∇ (∇ · A)]i − ∇2 A i (Eqs. 188, 186 & 195)
 
= ∇ (∇ · A) − ∇2 A i (Eq. 66)

Because i is a free index, the identity is proved for all components.

This identity can also be considered as an instance of the identity before the last one,
observing that in the second term on the right hand side the Laplacian should precede the
vector, and hence no independent proof is required.
• ∇ (A · B) = A × (∇ × B) + B × (∇ × A) + (A · ∇) B + (B · ∇) A:
We start from the right hand side and end with the left hand side:
5.6 Examples of Using Tensor Techniques to Prove Identities 140

[A × (∇ × B) + B × (∇ × A) + (A · ∇) B + (B · ∇) A]i =

[A × (∇ × B)]i + [B × (∇ × A)]i + [(A · ∇) B]i + [(B · ∇) A]i =

(Eq. 66)

ijk Aj [∇ × B]k + ijk Bj [∇ × A]k + (Al ∂l ) Bi + (Bl ∂l ) Ai =

(Eqs. 173, 188 & indexing)

ijk Aj (klm ∂l Bm ) + ijk Bj (klm ∂l Am ) + (Al ∂l ) Bi + (Bl ∂l ) Ai =

(Eq. 192)

ijk klm Aj ∂l Bm + ijk klm Bj ∂l Am + (Al ∂l ) Bi + (Bl ∂l ) Ai =

(commutativity)

ijk lmk Aj ∂l Bm + ijk lmk Bj ∂l Am + (Al ∂l ) Bi + (Bl ∂l ) Ai =

(Eq. 142)

(δil δjm − δim δjl ) Aj ∂l Bm + (δil δjm − δim δjl ) Bj ∂l Am + (Al ∂l ) Bi + (Bl ∂l ) Ai =

(Eq. 151)

(δil δjm Aj ∂l Bm − δim δjl Aj ∂l Bm ) + (δil δjm Bj ∂l Am − δim δjl Bj ∂l Am )

+ (Al ∂l ) Bi + (Bl ∂l ) Ai =

(distributivity)

δil δjm Aj ∂l Bm − Al ∂l Bi + δil δjm Bj ∂l Am − Bl ∂l Ai + (Al ∂l ) Bi + (Bl ∂l ) Ai =

(Eq. 134)

δil δjm Aj ∂l Bm − (Al ∂l ) Bi + δil δjm Bj ∂l Am − (Bl ∂l ) Ai + (Al ∂l ) Bi + (Bl ∂l ) Ai =

(grouping)

δil δjm Aj ∂l Bm + δil δjm Bj ∂l Am =

(cancellation)

Am ∂i Bm + Bm ∂i Am =
5.6 Examples of Using Tensor Techniques to Prove Identities 141

(Eq. 134)

∂i (Am Bm ) =

(product rule)

[∇ (A · B)]i

(Eqs. 186 & 188)

Because i is a free index, the identity is proved for all components.

• ∇ · (A × B) = B · (∇ × A) − A · (∇ × B):

∇ · (A × B) = ∂i [A × B]i (Eq. 188)

= ∂i (ijk Aj Bk ) (Eq. 173)

= ijk ∂i (Aj Bk ) (∂ not acting on )

= ijk (Bk ∂i Aj + Aj ∂i Bk ) (product rule)

= ijk Bk ∂i Aj + ijk Aj ∂i Bk (distributivity)

= kij Bk ∂i Aj − jik Aj ∂i Bk (Eq. 142)

= Bk (kij ∂i Aj ) − Aj (jik ∂i Bk ) (commutativity & grouping)

= Bk [∇ × A]k − Aj [∇ × B]j (Eq. 192)

= B · (∇ × A) − A · (∇ × B) (Eq. 172)

• ∇ × (A × B) = (B · ∇) A + (∇ · B) A − (∇ · A) B − (A · ∇) B:

[∇ × (A × B)]i = ijk ∂j [A × B]k (Eq. 192)

= ijk ∂j (klm Al Bm ) (Eq. 173)

= ijk klm ∂j (Al Bm ) (∂ not acting on )

= ijk klm (Bm ∂j Al + Al ∂j Bm ) (product rule)

= ijk lmk (Bm ∂j Al + Al ∂j Bm ) (Eq. 142)

5.6 Examples of Using Tensor Techniques to Prove Identities 142

= (δil δjm − δim δjl ) (Bm ∂j Al + Al ∂j Bm ) (Eq. 151)

= δil δjm Bm ∂j Al + δil δjm Al ∂j Bm −

δim δjl Bm ∂j Al − δim δjl Al ∂j Bm (distributivity)

= Bm ∂m Ai + Ai ∂m Bm − Bi ∂j Aj − Aj ∂j Bi (Eq. 134)

= (Bm ∂m ) Ai + (∂m Bm ) Ai − (∂j Aj ) Bi − (Aj ∂j ) Bi (grouping)

= [(B · ∇) A]i + [(∇ · B) A]i −

[(∇ · A) B]i − [(A · ∇) B]i (Eqs. 196 & 188)

= [(B · ∇) A + (∇ · B) A − (∇ · A) B − (A · ∇) B]i (Eq. 66)

Because i is a free index, the identity is proved for all components.

A·C A·D
• (A × B) · (C × D) = :
B·C B·D

(A × B) · (C × D) = [A × B]i [C × D]i (Eq. 172)

= ijk Aj Bk ilm Cl Dm (Eq. 173)

= ijk ilm Aj Bk Cl Dm (commutativity)

= (δjl δkm − δjm δkl ) Aj Bk Cl Dm (Eqs. 142 & 151)

= δjl δkm Aj Bk Cl Dm − δjm δkl Aj Bk Cl Dm (distributivity)

= (δjl Aj Cl ) (δkm Bk Dm ) − (δjm Aj Dm ) (δkl Bk Cl ) (commutativity)

= (Al Cl ) (Bm Dm ) − (Am Dm ) (Bl Cl ) (Eq. 134)

= (A · C) (B · D) − (A · D) (B · C) (Eq. 172)

A·C A·D
= (definition)
B·C B·D

• (A × B) × (C × D) = [D · (A × B)] C − [C · (A × B)] D:
5.7 Exercises 143

[(A × B) × (C × D)]i = ijk [A × B]j [C × D]k (Eq. 173)

= ijk jmn Am Bn kpq Cp Dq (Eq. 173)

= ijk kpq jmn Am Bn Cp Dq (commutativity)

= ijk pqk jmn Am Bn Cp Dq (Eq. 142)

= (δip δjq − δiq δjp ) jmn Am Bn Cp Dq (Eq. 151)

= (δip δjq jmn − δiq δjp jmn ) Am Bn Cp Dq (distributivity)

= (δip qmn − δiq pmn ) Am Bn Cp Dq (Eq. 134)

= δip qmn Am Bn Cp Dq − δiq pmn Am Bn Cp Dq (distributivity)

= qmn Am Bn Ci Dq − pmn Am Bn Cp Di (Eq. 134)

= qmn Dq Am Bn Ci − pmn Cp Am Bn Di (commutativity)

= (qmn Dq Am Bn ) Ci − (pmn Cp Am Bn ) Di (grouping)

= [D · (A × B)] Ci − [C · (A × B)] Di (Eq. 174)

= [[D · (A × B)] C]i − [[C · (A × B)] D]i (index definition)

= [[D · (A × B)] C − [C · (A × B)] D]i (Eq. 66)

Because i is a free index, the identity is proved for all components.

5.7 Exercises

5.1 Write, in tensor notation, the mathematical expression for the trace, determinant and
inverse of an n × n matrix.

5.2 Repeat the previous exercise for the multiplication of a matrix by a vector and the
multiplication of two n × n matrices.

5.3 Define mathematically the dot and cross product operations of two vectors using
tensor notation.
5.7 Exercises 144

5.4 Repeat the previous exercise for the scalar triple product and vector triple product
operations of three vectors.

5.5 Define mathematically, using tensor notation, the three scalar invariants of a rank-2
tensor: I, II and III.

5.6 Express the three scalar invariants I, II and III in terms of the other three invariants
I1 , I2 and I3 and vice versa.

5.7 Explain why the three invariants I1 , I2 and I3 are scalars using in your argument the
fact that the three main invariants I, II and III are traces?

5.8 Gather six terms from the Index about the scalar invariants of tensors.

5.9 Justify, giving a detailed explanation, the following statement: “If a rank-2 tensor
is invertible in a particular coordinate system it is invertible in all other coordinate
systems, and if it is singular in a particular coordinate system it is singular in all
other coordinate systems”. Use in your explanation the fact that the determinant is
invariant under admissible coordinate transformations.

5.10 What are the ten joint invariants between two rank-2 tensors?

5.11 Provide a concise mathematical definition of the nabla differential operator ∇ in

Cartesian coordinate systems using tensor notation.

5.12 What is the rank and variance type of the gradient of a differentiable scalar field in
general curvilinear coordinate systems?

5.13 State, in tensor notation, the mathematical expression for the gradient of a differen-
tiable scalar field in a Cartesian system.
5.7 Exercises 145

5.14 What is the gradient of the following scalar functions of position f, g and h where
x1 , x2 and x3 are the Cartesian coordinates and a, b and c are constants?

f = 1.3x1 − 2.6ex2 + 19.8x3 g = ax3 + bex2 h = a(x1 )3 − sin x3 + c(x3 )2

5.15 State, in tensor notation, the mathematical expression for the gradient of a differen-
tiable vector field in a Cartesian system.

5.16 What is the gradient of the following vector where x1 , x2 and x3 are the Cartesian
coordinates?
V = (2x1 − 1.2x2 , x1 + x3 , x2 x3 )

What is the rank of this gradient?

5.17 Explain, in detail, why the divergence of a vector is invariant.

5.18 What is the rank of the divergence of a rank-n (n > 0) tensor and why?

5.19 State, using vector and tensor notations, the mathematical definition of the divergence
operation of a vector in a Cartesian coordinate system.

5.20 Discuss in detail the following statement: “The divergence of a vector is a gradient
operation followed by a contraction”. How this is related to the trace of a rank-2
tensor?

5.21 Write down the mathematical expression of the two forms of the divergence of a rank-2
tensor.

5.22 How many forms do we have for the divergence of a rank-n (n > 0) tensor and why?
Assume in your answer that the divergence operation can be conducted with respect
to any one of the tensor indices.
5.7 Exercises 146

5.23 Find the divergence of the following vectors U and V where x1 , x2 and x3 are the
Cartesian coordinates:

U = 9.3x1 , 6.3 cos x2 , 3.6x1 e−1.2x3 V = x2 sin x1 , 5(x2 )3 , 16.3x3

5.24 State, in tensor notation, the mathematical expression for the curl of a vector and of
a rank-2 tensor assuming a Cartesian coordinate system.

5.25 By using the Index, make a list of terms and notations related to matrix algebra which
are used in this chapter.

5.26 Define, in tensor notation, the Laplacian operator acting on a differentiable scalar
field in a Cartesian coordinate system.

5.27 Is the Laplacian a scalar or a vector operator?

5.28 What is the meaning of the Laplacian operator acting on a differentiable vector field?

5.29 What is the rank of a rank-n tensor acted upon by the Laplacian operator?

5.30 Define mathematically the following operators assuming a Cartesian coordinate sys-
tem:
A·∇ A×∇ (239)

What is the rank of each one of these operators?

5.31 Make a general statement about how differentiation of tensors affects their rank dis-
cussing in detail from this perspective the gradient and divergence operations.

5.32 State the mathematical expressions for the following operators and operations assum-
ing a cylindrical coordinate system: nabla operator, Laplacian operator, gradient of
a scalar, divergence of a vector, and curl of a vector.
5.7 Exercises 147

5.33 Explain how the expressions for the operators and operations in the previous exercise
can be obtained for the plane polar coordinate system from the expressions of the
cylindrical system.

5.34 State the mathematical expressions for the following operators and operations assum-
ing a spherical coordinate system: nabla operator, Laplacian operator, gradient of a
scalar, divergence of a vector, and curl of a vector.

5.35 Repeat the previous exercise for the general orthogonal coordinate system.

5.36 Express, in tensor notation, the mathematical condition for a vector field to be
solenoidal.

5.37 Express, in tensor notation, the mathematical condition for a vector field to be irro-
tational.

5.38 Express, in tensor notation, the divergence theorem for a differentiable vector field
explaining all the symbols involved. Repeat the exercise for a differentiable tensor
field of an arbitrary rank (> 0).

5.39 Express, in tensor notation, Stokes theorem for a differentiable vector field explaining
all the symbols involved. Repeat the exercise for a differentiable tensor field of an
arbitrary rank (> 0).

5.40 Express the following identities in tensor notation:

∇·r = n

∇ (a · r) = a

∇ · (∇ × A) = 0

∇ (f h) = f ∇h + h∇f
5.7 Exercises 148

∇ × (f A) = f ∇ × A + ∇f × A

A × (B × C) = B (A · C) − C (A · B)

∇ × (∇ × A) = ∇ (∇ · A) − ∇2 A

∇ · (A × B) = B · (∇ × A) − A · (∇ × B)

A·C A·D
(A × B) · (C × D) =
B·C B·D

5.41 Prove the following identities using the language and techniques of tensor calculus:

∇×r = 0

∇ · (∇f ) = ∇2 f

∇ × (∇f ) = 0

∇ · (f A) = f ∇ · A + A · ∇f

A · (B × C) = C · (A × B) = B · (C × A)

A × (∇ × B) = (∇B) · A − A · ∇B

∇ (A · B) = A × (∇ × B) + B × (∇ × A) + (A · ∇) B + (B · ∇) A

∇ × (A × B) = (B · ∇) A + (∇ · B) A − (∇ · A) B − (A · ∇) B

(A × B) × (C × D) = [D · (A × B)] C − [C · (A × B)] D
Chapter 6

Metric Tensor

The subject of the present chapter is the metric tensor which is one of the most important
special tensors, if not the most important of all, in tensor calculus. Its versatile usage and
functionalities permeate the whole discipline of tensor calculus.
The metric tensor is a rank-2 tensor which may also be called the fundamental ten-
sor. The main purpose of the metric tensor is to generalize the concept of distance to
general curvilinear coordinate frames and maintain the invariance of distance in different
coordinate systems.
In orthonormal Cartesian coordinate systems the distance element squared, (ds)2 ,
between two infinitesimally neighboring points in space, one with coordinates xi and the
other with coordinates xi + dxi , is given by:

(ds)2 = dxi dxi = δij dxi dxj (240)

This definition of distance is the key to introducing a rank-2 tensor, gij , called the metric
tensor which, for a general coordinate system, is defined by:

(ds)2 = gij dxi dxj (241)

The above defined metric tensor is of covariant type. The metric tensor has also a
contravariant type which is usually notated with g ij .

149
6 METRIC TENSOR 150

The components of the covariant and contravariant metric tensor are given by:

gij = Ei · Ej g ij = Ei · Ej (242)

where the indexed E are the covariant and contravariant basis vectors as defined in §
2.6.1.
The metric tensor has also a mixed type which is given by:

g ij = Ei · Ej = δ ij gi j = Ei · Ej = δi j (243)

and hence it is the same as the unity tensor.

For a coordinate system in which the metric tensor can be cast in a diagonal form where
the diagonal elements are ±1 the metric is called flat. For Cartesian coordinate systems,
which are orthonormal flat-space systems, we have:

g ij = δ ij = gij = δij (244)

The metric tensor is symmetric in its two indices, that is:

gij = gji g ij = g ji (245)

This can be easily explained by the commutativity of the dot product of vectors in reference
to the above equations involving the dot product of the basis vectors.
The contravariant metric tensor is used for raising covariant indices of covariant and
mixed tensors, e.g.
Ai = g ik Ak Aij = g ik Akj (246)

Similarly, the covariant metric tensor is used for lowering contravariant indices of con-
6 METRIC TENSOR 151

travariant and mixed tensors, e.g.

Ai = gik Ak Aij = gik Akj (247)

In these raising and lowering operations the metric tensor acts, like a Kronecker delta, as
an index replacement operator as well as shifting the position of the index.
Because it is possible to shift the index position of a tensor by using the covariant and
contravariant types of the metric tensor, a given tensor can be cast into a covariant or a
contravariant form, as well as a mixed form in the case of tensors of rank > 1. However,
it should be emphasized that the order of the indices must be respected in this process,
because two tensors with the same indicial structure but with different indicial order are
not equal in general, as stated before. For example:

Aij = gjk Aik 6= Aj i = gjk Aki (248)

Some authors insert dots (e.g. Ai· j and A·ji ) to remove any ambiguity about the order of
the indices.
The covariant and contravariant metric tensors are inverses of each other, that is:

 −1  
[gij ] = g ij g ij = [gij ]−1 (249)

Hence:
g ik gkj = δ ij gik g kj = δi j (250)

It is common to reserve the term metric tensor to the covariant form and call the con-
travariant form, which is its inverse, the associate or conjugate or reciprocal metric
tensor. As a tensor, the metric has a significance regardless of any coordinate system
although it requires a coordinate system to be represented in a specific form. For orthog-
6 METRIC TENSOR 152

onal coordinate systems the metric tensor is diagonal, i.e. gij = g ij = 0 for i 6= j.
As indicated before, for orthonormal Cartesian coordinate systems in a 3D space, the
metric tensor is given in its covariant and contravariant forms by the 3 × 3 unit matrix,
that is:  
 1 0 0 
   ij   ij 
[gij ] = [δij ] = 
 0 1 0 
= δ = g (251)
 
0 0 1

For cylindrical coordinate systems with coordinates (ρ, φ, z), the metric tensor is given
in its covariant and contravariant forms by:
   
 1 0 0   1 0 0 
     
[gij ] = 
 0 ρ2 0 
 g ij
=
 0
1
ρ2
0 
 (252)
   
0 0 1 0 0 1

while for spherical coordinate systems with coordinates (r, θ, φ), the metric tensor is given
in its covariant and contravariant forms by:
   
 1 0 0   1 0 0 
     
[gij ] = 
 0 r2 0 
 g ij
=
 0
1
r2
0 
 (253)
   
0 0 r2 sin2 θ 0 0 1
r2 sin2 θ

As seen, all these metric tensors are diagonal since all these coordinate systems are
orthogonal. We also notice that all the corresponding diagonal elements of the covariant
and contravariant types are reciprocals of each other. This can be easily explained by
the fact that these two types are inverses of each other, plus the fact that the inverse of
an invertible diagonal matrix is a diagonal matrix obtained by taking the reciprocal of the
corresponding diagonal elements of the original matrix, as stated in § Inverse of Matrix.
6.1 Exercises 153

6.1 Exercises

6.1 Describe in details, using mathematical tensor language when necessary, the metric
tensor discussing its rank, purpose, designations, variance types, symmetry, its role
in the definition of distance, and its relation to the covariant and contravariant basis
vectors.

6.2 What is the relation between the covariant and contravariant types of the metric
tensor? Express this relation mathematically. Also define mathematically the mixed
type metric tensor.

6.3 Correct, if necessary, the following equations:

g ij = δ ii g ij = Ei · Ej (ds) = gij dxi dxj gij = Ei · Ej Ei · Ej = δ ji

6.4 What “flat metric” means? Give an example of a coordinate system with a flat metric.

6.5 Describe the index-shifting (raising/lowering) operators and their relation to the met-
ric tensor. How these operators facilitate the transformation between the covariant,
contravariant and mixed types of a given tensor?

6.6 Find from the Index all the names and labels of the metric tensor in its covariant,
contravariant and mixed types.

6.7 What is wrong with the following equations?

Ci = g ij Cj Di = g ij Dj Ai = δij Aj

Make the necessary corrections considering all the possibilities in each case.
6.1 Exercises 154

6.8 Is it necessary to keep the order of the indices which are shifted by the index-shifting
operators and why?

6.9 How and why dots may be inserted to avoid confusion about the order of the indices
following an index-shifting operation?

6.10 Express, mathematically, the fact that the contravariant and covariant metric tensors
are inverses of each other.

6.11 Collect from the Index a list of operators and operations which have particular links
to the metric tensor.

6.12 Correct, if necessary, the following statement: “The term metric tensor is usually used
to label the covariant form of the metric, while the contravariant form of the metric
is called the conjugate or associate or reciprocal metric tensor”.

6.13 Write, in matrix form, the covariant and contravariant types of the metric tensor of
the Cartesian, cylindrical and spherical coordinate systems of a 3D flat space.

6.14 Regarding the previous question, what do you notice about the corresponding diag-
onal elements of the covariant and contravariant types of the metric tensor in these
systems? Does this relate to the fact that these types are inverses of each other?
Chapter 7

Covariant Differentiation

The focus of this chapter is the operation of covariant differentiation of tensors which, in
a sense, is a generalization of the ordinary differentiation. The ordinary derivative of
a tensor is not a tensor in general. The objective of covariant differentiation is to ensure
the invariance of derivative (i.e. being a tensor) in general coordinate systems, and
this results in applying more sophisticated rules using Christoffel symbols where different
differentiation rules for covariant and contravariant indices apply. The resulting covariant
derivative is a tensor which is one rank higher than the differentiated tensor.
The Christoffel symbol of the second kind is defined by:

 
k g kl ∂gil ∂gjl ∂gij
ij = + − (254)
2 ∂xj ∂xi ∂xl

where the indexed g is the metric tensor in its contravariant and covariant forms with
implied summation over l. It is noteworthy that Christoffel symbols are not tensors.
The Christoffel symbols of the second kind are symmetric in their two lower indices,
that is:
k k
ij = ji (255)

For Cartesian coordinate systems, the Christoffel symbols are zero for all values of the
indices. For cylindrical coordinate systems, marked with the coordinates (ρ, φ, z), the

155
7 COVARIANT DIFFERENTIATION 156

Christoffel symbols are zero for all values of the indices except:

1
22 = −ρ (256)
2 2 1
12 = 21 = (257)
ρ

where (1, 2, 3) stand for (ρ, φ, z).

For spherical coordinate systems, marked with the coordinates (r, θ, φ), the Christoffel
symbols are zero for all values of the indices except:

1
22 = −r (258)
1
33 = −r sin2 θ (259)
2 2 1
12 = 21 = (260)
r
2
33 = − sin θ cos θ (261)
3 3 1
13 = 31 = (262)
r
3 3
23 = 32 = cot θ (263)

where (1, 2, 3) stand for (r, θ, φ).

For a differentiable scalar f , the covariant derivative is the same as the ordinary partial
derivative, that is:
f;i = f,i = ∂i f (264)

This is justified by the fact that the covariant derivative is different from the ordinary
partial derivative because the basis vectors in general coordinate systems are dependent
on their spatial position, and since a scalar is independent of the basis vectors the covariant
and partial derivatives are identical.
For a differentiable vector A, the covariant derivative of the covariant and contravariant
7 COVARIANT DIFFERENTIATION 157

forms of the vector is given by:

k
Aj;i = ∂i Aj − ji Ak (covariant) (265)
j
Aj;i = ∂i Aj + ki Ak (contravariant) (266)

For a differentiable rank-2 tensor A, the covariant derivative of the covariant, con-
travariant and mixed forms of the tensor is given by:

l l
Ajk;i = ∂i Ajk − ji Alk − ki Ajl (covariant) (267)
j k
Ajk;i = ∂i Ajk + li Alk + li Ajl (contravariant) (268)
k l
Akj;i = ∂i Akj + li Alj − ji Akl (mixed) (269)

More generally, for a differentiable rank-n tensor A, the covariant derivative is given
by:

i j k
Aij...k ij...k
lm...p;q = ∂q Alm...p + aq Aaj...k
lm...p + aq Aia...k
lm...p + · · · + aq Aij...a
lm...p
a a a
− lq Aij...k
am...p − mq Aij...k
la...p − · · · − pq Aij...k
lm...a (270)

From the last equations, a pattern for the covariant differentiation operation emerges,
that is it starts with an ordinary partial derivative term, then for each tensor index an
extra Christoffel symbol term is added, positive for superscripts and negative for subscripts,
where the differentiation index is the second of the lower indices in the Christoffel symbol.
Since the Christoffel symbols are identically zero in Cartesian coordinate systems, the
covariant derivative is the same as the ordinary partial derivative for all tensor ranks.
Another important fact about covariant differentiation is that the covariant derivative of
the metric tensor in its covariant, contravariant and mixed forms is zero in all coordinate
systems and hence it is treated like a constant in covariant differentiation.
7 COVARIANT DIFFERENTIATION 158

Several rules of ordinary differentiation similarly apply to covariant differentiation.

For example, covariant differentiation is a linear operation with respect to algebraic sums
of tensor terms, that is:
∂;i (aA ± bB) = a ∂;i A ± b ∂;i B (271)

where a and b are scalar constants and A and B are differentiable tensor fields.
The product rule of ordinary differentiation also applies to covariant differentiation of
tensor multiplication, that is:

∂;i (AB) = (∂;i A) B + A∂;i B (272)

However, as seen in this equation, the order of the tensors should be observed since tensor
multiplication, unlike ordinary algebraic multiplication, is not commutative. The product
rule is also valid for the inner product of tensors because the inner product is an outer
product operation followed by a contraction of indices, and covariant differentiation and
contraction of indices do commute.
Since the covariant derivative of the metric tensor is identically zero, as stated above,
the covariant derivative operator bypasses the index raising/lowering operator, that is:

∂;m gij Aj = gij ∂;m Aj (273)

and hence the metric tensor behaves like a constant with respect to the covariant differ-
ential operator.
A principal difference between the partial differentiation and the covariant differen-
tiation is that for successive differential operations with respect to different indices the
partial derivative operators do commute with each other, assuming certain continuity
7.1 Exercises 159

conditions, but the covariant differential operators do not commute, that is:

∂i ∂j = ∂j ∂i but ∂;i ∂;j 6= ∂;j ∂;i (274)

Higher order covariant derivatives are similarly defined as derivatives of derivatives;

however the order of differentiation, in the case of differentiating with respect to different
indices, should be respected, as explained above.

7.1 Exercises

7.1 Is the ordinary derivative of a tensor necessarily a tensor or not? Can the ordinary
derivative of a tensor be a tensor? If so, give an example.

7.2 Explain the purpose of the covariant derivative and how it is related to the invariance
property of tensors.

7.3 Is the covariant derivative of a tensor necessarily a tensor? If so, what is the rank of
the covariant derivative of a rank-n tensor?

7.4 How is the Christoffel symbol of the second kind symbolized? Describe the arrange-
ment of its indices.

7.5 State the mathematical definition of the Christoffel symbol of the second kind in terms
of the metric tensor defining all the symbols involved.

7.6 The Christoffel symbols of the second kind are symmetric in which of their indices?

7.7 Why the Christoffel symbols of the second kind are identically zero in the Carte-
sian coordinate systems? Use in your explanation the mathematical definition of the
Christoffel symbols.
7.1 Exercises 160

7.8 Give the Christoffel symbols of the second kind for the cylindrical and spherical co-
ordinate systems explaining the meaning of the indices used.

7.9 From the Index, collect five terms used in the definition and description of the Christof-
fel symbols of the second kind.

7.10 What is the meaning, within the context of tensor differentiation, of the comma “,”
and semicolon “;” when used as subscripts preceding a tensor index?

7.11 Why the covariant derivative of a differentiable scalar is the same as the ordinary
partial derivative and how is this related to the basis vectors of coordinate systems?

7.12 Differentiate the following tensors covariantly:

As Bt Cij Dpq E mn Alm...p

ij...k

7.13 Explain the mathematical pattern followed in the operation of covariant differentiation
of tensors. Does this pattern also apply to rank-0 tensors?

7.14 The covariant derivative in Cartesian coordinate systems is the same as the ordinary
partial derivative for all tensor ranks. Explain why.

7.15 What is the covariant derivative of the covariant and contravariant forms of the metric
tensor for an arbitrary type of coordinate system? How is this related to the fact that
the covariant derivative operator bypasses the index-shifting operator?

7.16 Which rules of ordinary differentiation apply equally to covariant differentiation and
which do not? Make mathematical statements about all these rules with sufficient
explanation of the symbols and operations involved.

7.17 Make corrections, where necessary, in the following equations explaining in each case
why the equation should or should not be amended:
7.1 Exercises 161

(C ± D);i = ∂;i C ± ∂;i D

∂;i (AB) = B (∂;i A) + A∂;i B

gij Aj ;m = gij Aj ;m

∂;i ∂;j = ∂;j ∂;i

∂i ∂j = ∂j ∂i

7.18 How do you define the second and higher order covariant derivatives of tensors? Do
these derivatives follow the same rules as the ordinary partial derivatives of the same
order in the case of different differentiation indices?

7.19 From the Index, find all the terms that refer to symbols used in the notation of
ordinary partial derivatives and covariant derivatives.
References
G.B. Arfken; H.J. Weber; F.E. Harris. Mathematical Methods for Physicists A Compre-
hensive Guide. Elsevier Academic Press, seventh edition, 2013.

R.B. Bird; R.C. Armstrong; O. Hassager. Dynamics of Polymeric Liquids, volume 1.

John Wiley & Sons, second edition, 1987.

R.B. Bird; W.E. Stewart; E.N. Lightfoot. Transport Phenomena. John Wiley & Sons,
second edition, 2002.

M.L. Boas. Mathematical Methods in the Physical Sciences. John Wiley & Sons Inc.,
third edition, 2006.

C.F. Chan Man Fong; D. De Kee; P.N. Kaloni. Advanced Mathematics for Engineering
and Science. World Scientific Publishing Co. Pte. Ltd., first edition, 2003.

T.L. Chow. Mathematical Methods for Physicists: A concise introduction. Cambridge

University Press, first edition, 2003.

J.H. Heinbockel. Introduction to Tensor Calculus and Continuum Mechanics. 1996.

D.C. Kay. Schaum’s Outline of Theory and Problems of Tensor Calculus. McGraw-Hill,
first edition, 1988.

K.F. Riley; M.P. Hobson; S.J. Bence. Mathematical Methods for Physics and Engineering.
Cambridge University Press, third edition, 2006.

D. Zwillinger, editor. CRC Standard Mathematical Tables and Formulae. CRC Press,
32nd edition, 2012.

162
Index
Absolute identities, 10
differential calculus, 9 Bound index, 50
tensor, 68, 69, 80
Calculus, 1
Active transformation, 52
Cartesian coordinate system, 7, 12, 16, 17,
Addition/Subtraction of tensors, 69, 83, 84,
41, 121, 152, 154, 155
92, 94
Christoffel, 9
Adjoint matrix, 39
symbol, 9, 155–157, 159
Adjugate matrix, 39
Circulation of field, 31
Algebraic sum, 50, 86, 88, 158
Cofactor, 38, 39, 45
Alternating tensor, 96
Column of matrix, 28, 35–38, 108, 114, 117
Anisotropic tensor, 70, 80
Comma notation, 13, 160
Anti-
Commutative, 15, 30–32, 37, 43, 46, 84–86,
commutative, 26, 28
88, 92, 93
diagonal, 35
Component
symmetric tensor, 59, 70–73, 81, 82, 96
notation, 49
symmetry of tensor, 73, 81, 114
of metric tensor, 150
Arithmetic, 1
of tensor, 12, 13, 47–49, 53, 56, 64, 70,
Associate metric tensor, 151, 154
75, 81, 83–85, 96, 97, 104, 114
Associative, 29–32, 37, 42, 43, 46, 84, 85, 92
of vector, 13, 17, 23, 28, 31, 66, 123, 133
Axial
Conjugate metric tensor, 151, 154
tensor, 65
Conserved, 97, 101, 112
vector, 68, 79
Continuum mechanics, 11, 58
Axis of coordinates, 16, 19, 51, 52, 59, 63,
Contracted epsilon identity, 107
66, 70, 101
Contraction, 50, 67, 83, 87–94, 102, 122,
Bar notation, 13, 23 123, 145, 158
Barred symbol, 51, 60–63, 69 Contravariant
Basis basis tensor, 65, 79
set, 26, 63, 103, 105 basis vector, 7, 64–66, 79, 150, 153
vector, 12, 16–22, 26, 41, 61, 156, 160 component, 64
Bianchi, 10 index, 50, 54, 55, 60, 63, 69, 155

163
 Kronecker δ, 112 Differential
metric tensor, 149–155, 157, 160 geometry, 9, 10, 48, 58
permutation tensor, 101, 110–112, 115 operation, 29, 120, 121, 158
tensor, 50, 57, 59, 60, 63, 74, 77, 78, 151 operator, 7, 14, 15, 29, 42, 61–63, 120,
Convergence of field, 31 121, 123, 158, 159
Coordinate system, 15, 16, 51 Dimension
Covariant of matrix, 36, 39, 45
basis tensor, 65, 79 of space, 7, 47, 48, 57, 102, 127
basis vector, 7, 22, 64–66, 79, 150, 153 of tensor, 47, 48, 55, 57, 85, 109
component, 64 Direct
derivative, 155–161 multiplication, 85–87
differential operator, 158, 159 multiplication symbol (), 86
differentiation, 14, 155, 157, 158, 160 notation, 49
index, 50, 54, 55, 60, 62, 63, 69, 123 product, 66–68, 70, 80
Kronecker δ, 112 transformation, 51, 74
metric tensor, 149–155, 157, 160 Direction of vector, 11, 16, 17, 19, 24, 29–31,
permutation tensor, 101, 110–112, 115 33, 49, 57–59, 66, 68, 77, 119
tensor, 50, 57, 59, 60, 63, 74, 77, 78, 150
Distance, 16, 19, 149, 153
Cross product, 24–28, 31, 41, 42, 67, 68, 74, element, 149
80, 93, 105, 118, 120, 137, 138, 143Distributive, 30–32, 37, 43, 46, 56, 86, 88,
Curl operation, 7, 29, 31–33, 43, 67, 68, 80, 93, 94
122, 123, 125–127, 146, 147 Divergence
Cyclic permutation, 27, 28 of field, 31
Cylindrical coordinate system, 7, 16–19, 41, operation, 7, 29, 31, 32, 43, 121–124,
124, 146, 147, 152, 154, 155 126, 127, 145–147
theorem, 32, 33, 44, 132, 133, 147
Determinant, 7, 13, 25, 28, 36, 38, 39, 43,
Division, 83, 92
45, 46, 116, 117, 120, 143, 144
Dot
Determinantal form, 42, 107, 108, 111, 114
notation, 23, 26, 27, 42, 56, 88, 90, 117,
Diagonal
118, 151
matrix, 36
product, 23, 24, 31, 36, 41, 42, 67, 74,
metric tensor, 150, 152
87, 89, 90, 93, 118
Differentiable
Double
scalar field, 127, 144, 146
contraction, 90, 93
vector field, 33, 127, 132, 133, 145–147
direction, 58, 59, 77

164
 dot notation ( : and ··), 90 Gradient operation, 7, 29, 30, 43, 60, 121,
inner product, 90, 94, 103 123–126, 144–147
Dummy index, 50, 54, 55, 63, 74, 75
Hamilton, 8
Dyad, 11, 40, 59, 86, 90, 91, 103
Handedness, 51, 66, 68
Dyadic
Harmonic operator, 32
multiplication, 85
product, 65 Identity
matrix, 36, 39, 44, 96
epsilon-delta identity, 107, 108, 111, 115
tensor, 102, 109
Even permutation, 52, 75, 98, 99, 113
Illegitimate tensor notation, 55, 76
Exterior multiplication, 85
Improper
Flat rotation, 52, 70
metric, 150, 153 transformation, 51, 52, 66, 75, 97, 101
space, 150, 154 Independent component of tensor, 104, 114
Fluid Index
dynamics, 11, 123 -free notation, 49
mechanics, 58 lowering operator, 150, 153, 158
Form-invariant, 48, 49 notation, 13, 49
Frame of reference, 49, 52, 92, 149 of matrix, 35, 45
Free index, 50, 53, 55, 62, 74, 75, 84, 87 of tensor, 13, 47–51, 53–55, 60, 74, 75,
Fundamental tensor, 149 83
raising operator, 150, 153, 158
Gauss, 9, 116
replacement operator, 102, 109, 112, 151
-Codazzi equations, 9
shifting operator, 153, 154, 160
curvature tensor, 9
Indexed square bracket, 13
theorem, 32
Indicial
Gaussian coordinates, 9
notation, 49, 78
General
structure, 56, 69, 84, 151
curvilinear coordinate system, 7, 16, 21,
Inductive definition, 99, 109, 111
22, 41, 62, 64, 144, 149
Inner product, 37, 67, 70, 88–92, 94, 95, 103,
orthogonal coordinate system, 7, 16, 22,
117, 158
41, 120, 121, 126, 147, 152
Integral, 32, 33
relativity, 10, 58
theorem, 32–34, 116, 132
Generalized Kronecker δ, 109–111, 115
Invariance
Gibbs notation, 49
of derivative, 155

165
 of distance, 149 algebra, 34, 87, 94, 116, 117
of tensor, 48, 54, 58, 74, 159 notation, 37, 86, 117, 118
Invariant, 27, 49, 57, 66, 70, 73, 119, 120, of cofactors, 39
122, 144, 145 Metric tensor, 54, 63, 149–154, 157, 158
Inverse Mixed
of matrix, 36, 39, 45, 46, 117, 143, 144, basis tensor, 65, 79
154 Kronecker δ, 112
transformation, 51, 74 metric tensor, 150, 153, 157
Inversion of axes, 52, 66, 101 tensor, 50, 57, 60, 61, 65, 74, 77, 79, 87,
Invertible matrix, 39, 120, 144, 152 150, 151
Irrotational, 32, 132, 147 Modulus, 13, 23
Isotropic tensor, 70, 80, 97, 101 Multiplication
of matrices, 36, 37, 45, 46, 86, 89, 94,
Jacobian, 69, 80
117, 118, 143
matrix, 69
of matrix by vector, 89, 117, 143
Kronecker δ, 10, 54, 57, 64, 70, 96, 97, 101, of tensor by scalar, 83–85, 87, 92, 93
102, 109–112, 115, 151 of tensors, 69, 83, 85–88, 92, 93, 158
Mutually
Laplacian operator, 7, 14, 29, 32, 43, 44,
orthogonal, 16, 22, 64, 78, 103
123–126, 139, 146, 147
perpendicular, 12, 17, 20
Legitimate tensor notation, 53, 55, 75, 76
Length of vector, 11, 16, 17, 20, 22–25, 57, nabla operator, 7, 13, 29, 31, 42, 120, 121,
64, 78, 119 124–126, 144, 146, 147
Levi-Civita, 9, 10 Non-
identity, 10, 107 repetitive permutation, 45, 82, 104, 113
symbol, 10 scalar tensor, 15, 86
tensor, 58, 70, 96 singular matrix, 39
Linear Normal vector, 30, 33
algebra, 1, 89, 117, 118 Numerical tensor, 96, 112
combination, 50
Oblique coordinate system, 12, 16
operation, 158
Odd permutation, 52, 75, 98, 99, 101, 113
Lower index, 54, 61–63, 73, 75, 87
Opposite
Main diagonal, 35, 36, 38, 44 cyclic permutation, 28
Mapping of spaces, 51 transformation, 18, 21, 61
Matrix, 34, 35, 49 Order of tensor, 50, 55, 74

166
Origin of coordinates, 16, 19, 52, 66, 67 Principle of invariance, 48, 74
Original space, 51 Product rule of differentiation, 158
Orthogonal transformation, 66, 101 Projection, 17, 23, 24, 27
Orthonormal Proper
basis vectors, 103, 105, 113, 114, 124, rotation, 52, 70
125 tensor, 66, 68
Cartesian system, 12, 15–17, 23, 41, 54, transformation, 51, 52, 75, 97
55, 63, 64, 103, 149, 152 vector, 68
Outer product, 11, 37, 80, 86, 88–91, 93, 94, Pseudo
121, 158 scalar, 68, 80
tensor, 65–69, 80, 84, 101, 113
Parallelepiped, 27, 28
vector, 66–68, 79, 80, 119
Parallelogram, 24, 26
Parenthesis notation, 26, 42, 53, 54 Quotient rule, 91, 92, 95
Parity, 53, 67, 75
Range
Partial
of coordinate, 16, 19
anti-symmetry, 82
of index, 35, 47, 48, 53–55, 86, 133
derivative, 13–15, 62, 64, 78, 156–158,
Rank of tensor, 11–13, 40, 47–50, 55, 57, 58,
160, 161
74, 85
derivative operator, 158
Reciprocal
differentiation, 14, 158
metric tensor, 151, 154
symmetry, 82
system, 64
Passive transformation, 52
Reciprocity relation, 64, 65, 79
Permutation, 52, 53, 75, 104, 137
Rectangular Cartesian system, 16, 20, 29,
of tensor, 83, 91, 92, 95
52, 58, 59
symbol, 68
Relative tensor, 68, 69, 80, 88, 94, 101
tensor, 67, 68, 96–101, 104–106, 109, 111–
Repetitive permutation, 45, 82, 104
114
Ricci, 9, 10
Physical component, 121
calculus, 9, 10
Plane polar coordinate system, 19, 20, 41,
curvature tensor, 10
125, 147
Riemann, 9
Polar
-Christoffel curvature tensor, 9
tensor, 65
Riemannian
vector, 67, 68, 79
manifold, 9
Polyad, 12
metric tensor, 9

167
Right hand Superscript index, 47, 50, 51, 60, 62, 64, 78,
rule, 16, 24, 25 157
twist rule, 33, 34 Surface element, 33, 133
Rotation, 31, 52, 70, 101 Symbolic notation, 13, 37, 49, 86, 88, 117,
Round bracket, 71, 81 122
Row of matrix, 26, 28, 35–38, 108, 114, 117 Symmetric tensor, 59, 70–73, 81, 82, 97, 104,
Rules 114, 150
of covariant differentiation, 158–161 Symmetry of tensor, 73, 81, 105, 114, 153
of ordinary differentiation, 158–161
Tensor, 8, 11–13, 49
Scalar, 11, 12, 35, 40, 47, 49, 50, 56, 57, 70, algebra, 2, 11, 15, 83, 117
74, 76, 77, 156 analysis, 9
invariant of tensor, 119, 144 calculus, 1, 2, 8–11, 14, 15, 23, 40, 49,
operator, 32, 123, 124 50, 53, 54, 58, 69, 74, 77, 96, 119,
product, 23 134, 148
triple product, 26–28, 42, 68, 105, 118, density, 65, 69
144 equality, 50, 53, 55, 56, 62, 66, 74, 76,
Semicolon notation, 14, 160 80
Sign function, 100 expression, 50, 53, 55, 56, 66, 74–76, 80
Singular matrix, 36, 39, 45, 46, 144 identity, 50
Skew-symmetric tensor, 71 notation, 29, 47, 49, 58, 78, 105, 116,
Solenoidal, 31, 43, 132, 147 119–121, 127, 132–134, 143–147
Spherical coordinate system, 7, 19–21, 41, operation, 83, 92
125, 147, 152, 154, 156 term, 50, 53, 55, 56, 66, 67, 74, 76, 80,
Square 158
bracket, 13, 71, 81 Totally
matrix, 35, 36, 38, 39, 44 anti-symmetric, 72, 82, 100, 104
Stokes, 116 symmetric, 72, 82
theorem, 33, 34, 44, 133, 134, 147 Trace, 7, 37, 38, 45, 46, 87, 116, 122, 143–
Subscript index, 13, 14, 47, 50, 51, 59, 60, 145
64, 157 Trailing diagonal, 35, 38, 44
Substitution operator, 102 Transformation of coordinates, 15, 17, 20,
Summation convention, 53, 54, 75, 102, 124, 41, 48–54, 57, 59, 60, 62, 66, 68, 70,
125 73, 77, 78, 80, 92, 95, 119, 144, 153
Super factorial function, 100 Transformed space, 51

168
Transpose of matrix, 36, 44, 122 Weight of tensor, 68, 69, 80, 88, 94, 101
Transposed contraction, 90
Zero
Transposition
matrix, 36
of indices, 52, 75
tensor, 50, 70, 74
of matrix, 36, 91, 95
Triad, 12, 86
True
scalar, 67, 80
tensor, 65–67, 69, 80, 84, 113
vector, 66, 67, 79, 80

Unbarred symbol, 51, 60–63, 69

Uniformly
dimensioned, 35, 48, 86
scaled, 12, 16
Unit
basis vector, 16, 41, 126
dyad, 58, 59, 77
matrix, 36, 44, 152
tensor, 50, 74, 96
vector, 16, 17, 20, 25, 29, 33, 59, 66, 133
Unity
matrix, 36
tensor, 50, 150
Upper index, 54, 61–63, 73, 75, 87

Variance type, 54, 55, 60, 63, 69, 73, 75, 81,
84, 85, 87, 92, 93, 144, 153
Vector, 11–13, 17, 49, 50, 55, 76, 77
algebra, 23
calculus, 23, 119, 127
notation, 13, 127, 132, 133, 145
operator, 29, 124
triple product, 28, 29, 42, 118, 144
Voigt, 9
Volume element, 33, 133

169
About the Book
This book is prepared from personal notes and tutorials about
tensor calculus at an introductory level. The language and
method used in presenting the ideas and techniques of tensor
calculus make it very suitable for learning this subject by the
beginners who have not been exposed previously to this elegant
branch of mathematics. Considerable efforts have been made to
reduce the dependency on foreign texts by summarizing the main
concepts needed to make the book self-contained. The book also
contains a number of good-quality graphic illustrations to aid
the readers and students in their effort to visualize the ideas and
understand the abstract concepts. Furthermore, illustrative tech-
niques, such as highlighting key terms by boldface fonts, have
been employed. The book also contains sets of clearly explained
exercises which cover most of the given materials. These ex-
ercises are designed to provide thorough revisions of the sup-
plied materials and hence they make an essential component of
the book and its learning objectives. The book is also furnished
with a rather detailed index and populated with cross references,
which are hyperlinked for the ebook users, to facilitate connect-
ing related subjects and ideas.

About the Author

The author of the book possesses a diverse academic and re-
search background. He holds a BEng in electronics engineer-
ing, a BSc in physics, a PhD in petroleum engineering, a PhD
in crystallography, and a PhD in atomic physics and astronomy.
He also worked as a postdoctoral research associate for several
years. He published dozens of scientific research papers in many
refereed journals and produced numerous academically oriented
documents which are freely available on the world wide web.