AN INTRODUCTION TO

TENSOR ANALYSIS
RIVER PUBLISHERS SERIES IN MATHEMATICAL
AND ENGINEERING SCIENCES

Series Editors

MANGEY RAM ALIAKBAR MONTAZER HAGHIGHI

Graphic Era University Prairie View Texas A&M University
India USA

TADASHI DOHI
Hiroshima University
Japan

Indexing: All books published in this series are submitted to the Web of Science Book Citation Index
(BkCI), to SCOPUS, to CrossRef and to Google Scholar for evaluation and indexing.
Mathematics is the basis of all disciplines in science and engineering. Especially applied
mathematics has become complementary to every branch of engineering sciences. The purpose of this
book series is to present novel results in emerging research topics on engineering sciences, as well
as to summarize existing research. It engrosses mathematicians, statisticians, scientists and engineers
in a comprehensive range of research fields with different objectives and skills, such as differential
equations, finite element method, algorithms, discrete mathematics, numerical simulation, machine
leaning, probability and statistics, fuzzy theory, etc.
Books published in the series include professional research monographs, edited volumes, conference
proceedings, handbooks and textbooks, which provide new insights for researchers, specialists in industry,
and graduate students.
Topics covered in the series include, but are not limited to:
• Advanced mechatronics and robotics
• Artificial intelligence
• Automotive systems
• Discrete mathematics and computation
• Fault diagnosis and fault tolerance
• Finite element methods
• Fuzzy and possibility theory
• Industrial automation, process control and networked control systems
• Intelligent control systems
• Neural computing and machine learning
• Operations research and management science
• Optimization and algorithms
• Queueing systems
• Reliability, maintenance and safety for complex systems
• Resilience
• Stochastic modelling and statistical inference
• Supply chain management
• System engineering, control and monitoring
• Tele robotics, human computer interaction, human-robot interaction
For a list of other books in this series, visit www.riverpublishers.com
The NEC and You Perfect Together:
AN INTRODUCTION TO
A Comprehensive Study of the
TENSOR ANALYSIS
National Electrical Code

Dr. Bipin Singh Koranga

M.Sc., Ph.D. (IITB), MIPA,
Gregory P. Bierals
Department
Electrical Design Institute, USA of Physics,

Kirori Mal College, Delhi

Dr. Sanjay Kumar Padaliya

M.Sc., Ph.D. (K.U),
Department of Mathematics,
SGRR (PG) College, Dehradun, India

River Publishers
 Published 2020 by River Publishers
River Publishers
Alsbjergvej 10, 9260 Gistrup, Denmark
www.riverpublishers.com

Distributed exclusively by Routledge

4 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
605 Third Avenue, New York, NY 10017, USA

AN INTRODUCTION TO TENSOR ANALYSIS / by Bipin Singh Koranga,

Sanjay Kumar Padaliya.

© 2020 River Publishers. All rights reserved. No part of this publication may
be reproduced, stored in a retrieval systems, or transmitted in any form or by
any means, mechanical, photocopying, recording or otherwise, without prior
written permission of the publishers.

Routledge is an imprint of the Taylor & Francis Group, an informa

business

ISBN 978-87-7022-581-6 (print)

While every effort is made to provide dependable information, the

publisher, authors, and editors cannot be held responsible for any errors
or omissions.
 Preface

I feel great pleasure in bringing out the first edition of the book
“An Introduction to Tensor Analysis”. This book has been written
especially in accordance with the latest and modified syllabus
framed for under-graduate and postgraduate students. A reasonably
wide coverage in sufficient depth has been attempted. The book
contains sufficient number of problems. We hope that if a student
goes through all these, he or she would appreciate and enjoy the
subject. The work is dedicated to our past students whose inspiration
motivated us to do this work without any stress and strain. The
authors wish to acknowledge his indebtedness to numerous authors
of those books, which were consulates during the preparation of
the matter. We feel great pleasure to express deepest sense of
gratitude, respect and honor to Prof. Uma Sankar, IIT Bombay and
Prof. Mohan Narayan, Mumbai University for conferring valuable
guidance and their patronly behavior. We would like to thank several
of our colleagues in Kirori Mal College and SGRR(PG) College,
Dehradun. We would like to thank Dr. Vinod Kumar, University of
Lucknow, Dr. Imran Khan, Ramjas College, Dr. Shushil Kumar, Hindu
College, bearing for making useful suggestions during the preparation
of manuscript.

v
vi Preface

Errors might have crept in here and there inspite of the care to
avoid them. We will be very grateful for bringing any errors to our
notice. Suggestions or criticisms towards the further improvement
of the book shall be gratefully acknowledged. We shall appreciate
receiving comments and suggestions, which can be sent to the below
emails: [email protected], [email protected]

Thanks to Prof. Angelo Galanty, University of L’Aquila and Dr.

Rakesh Joshi, Australia’s Global University for encouragement. We
are deeply grateful to our family members, who have always been a
source of inspiration for us. In the last, but not the least, we are thankful
to the publisher of this book.
Dr. Bipin Singh Koranga
Dr. Sanjay Kumar Padaliya
 Syllabus

Cartesian Tensors-Transformation of Coordinates. Einstein’s

Summation Convention. Relation between Direction Cosines. Tensors.
Algebra of Tensors. Sum, Differences and Product of Two Tensors.
Contraction. Quotient Law of Tensors. Symmetric and Anti-symmetric
Tensors. Pseudo tensors. Invariant Tensors: Kronecker and Alternating
Tensors. Association of Anti-symmetric Tensor of Order two and
Vectors. Vector Algebra and Calculus using Cartesian Tensors:
Scalar and Vector Products, Scalar and Vector Triple Products.
Differentiation. Gradient, Divergence and Curl of Tensor Fields.
Vector Identities. Tensorial Formulation of Analytical Solid Geometry:
Equation of a Line. Angle Between Lines. Projection of a Line
on another Line. Condition for Two Lines to be Coplanar. Foot
of the Perpendicular from a Point on a Line. Rotation Tensor
(No Derivation). Isotropic Tensors. Tensorial Character of Physical
Quantities. Moment of Inertia Tensor. Stress and Strain Tensors :
Symmetric Nature. Elasticity Tensor. Generalized Hooke’s Law

General Tensors-Transformation of Coordinates. Contravariant and

Covariant Vectors. Contravariant, Covariant and Mixed Tensors.
Kronecker Delta and Permutation Tensors. Algebra of Tensors. Sum,

vii
viii Syllabus

Difference and Product of Two Tensors. Contraction. Quotient Law

of Tensors. Symmetric and Anti-symmetric Tensors. Metric Tensor.
Reciprocal Tensors. Associated Tensors. Christoffel Symbols of
First and Second Kind and their Transformation Laws. Covariant
Derivative. Tensor Form of Gradient, Divergence and Curl.
 Contents

Preface v

Syllabus vii

1 Introduction 1
1.1 Symbols Multi-Suffix . . . . . . . . . . . . . . . . . . 2
1.2 Summation Convention . . . . . . . . . . . . . . . . . 3

2 Cartesian Tensor 7
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Transformation of Coordinates . . . . . . . . . . . . . 8
2.3 Relations Between the Direction Cosines . . . . . . . . 10
2.4 Transformation of Velocity Components . . . . . . . . 11
2.5 First-Order Tensors . . . . . . . . . . . . . . . . . . . 12
2.6 Second-Order Tensors . . . . . . . . . . . . . . . . . 13
2.7 Notation for Tensors . . . . . . . . . . . . . . . . . . 14
2.8 Algebraic Operations on Tensors . . . . . . . . . . . . 14
2.8.1 Sum and Difference of Tensors . . . . . . . . . 15
2.8.2 Product of Tensors . . . . . . . . . . . . . . . 16
2.9 Quotient Law of Tensors. . . . . . . . . . . . . . . . . 17

ix
x Contents

2.10 Contraction Theorem . . . . . . . . . . . . . . . . . . 19

2.11 Symmetric and Skew-Symmetric Tensor . . . . . . . . 21
2.12 Alternate Tensor . . . . . . . . . . . . . . . . . . . . . 23
2.13 Kronecker Tensor . . . . . . . . . . . . . . . . . . . . 24
2.14 Relation Between Alternate and Kronecker Tensors . . 25
2.15 Matrices and Tensors of First and Second Orders . . . 26
2.16 Product of Two Matrices . . . . . . . . . . . . . . . . 28
2.17 Scalar and Vector Inner Product . . . . . . . . . . . . 31
2.17.1 Two Vectors . . . . . . . . . . . . . . . . . . . 31
2.17.2 Scalar Product . . . . . . . . . . . . . . . . . 31
2.17.3 Vector Product . . . . . . . . . . . . . . . . . 31
2.18 Tensor Fields . . . . . . . . . . . . . . . . . . . . . . 32
2.18.1 Gradient of Tensor Field . . . . . . . . . . . . 32
2.18.2 Divergence of Vector Point Function . . . . . . 34
2.18.3 Curl of Vector Point Function . . . . . . . . . 34
2.19 Tensorial Formulation of Gauss’s Theorem . . . . . . . 35
2.20 Tensorial Formulation of Stoke’s Theorem . . . . . . . 35
2.21 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . 36

3 Tensor in Physics 39
3.1 Kinematics of Single Particle . . . . . . . . . . . . . . 39
3.1.1 Momentum . . . . . . . . . . . . . . . . . . . 40
3.1.2 Acceleration . . . . . . . . . . . . . . . . . . 40
3.1.3 Force . . . . . . . . . . . . . . . . . . . . . . 40
3.2 Kinetic Energy and Potential Energy . . . . . . . . . . 41
3.3 Work Function and Potential Energy . . . . . . . . . . 41
 Contents xi

3.4 Momentum and Angular Momentum . . . . . . . . . . 43

3.5 Moment of Inertia . . . . . . . . . . . . . . . . . . . . 44
3.6 Strain Tensor at Any Point . . . . . . . . . . . . . . . 46
3.7 Stress Tensor at any Point P . . . . . . . . . . . . . . . 49
3.7.1 Normal Stress . . . . . . . . . . . . . . . . . . 50
3.7.2 Simple Stress . . . . . . . . . . . . . . . . . . 50
3.7.3 Shearing Stress . . . . . . . . . . . . . . . . . 50
3.8 Generalised Hooke’s Law . . . . . . . . . . . . . . . . 50
3.9 Isotropic Tensor . . . . . . . . . . . . . . . . . . . . . 51
3.10 Exercises . . . . . . . . . . . . . . . . . . . . . . . . 52

4 Tensor in Analytic Solid Geometry 55

4.1 Vector as Directed Line Segments . . . . . . . . . . . 55
4.2 Geometrical Interpretation of the Sum of two Vectors . 57
4.3 Length and Angle between Two Vectors . . . . . . . . 57
4.4 Geometrical Interpretation of Scalar and Vector Products 58
4.4.1 Scalar Triple Product . . . . . . . . . . . . . . 60
4.4.2 Vector Triple Products . . . . . . . . . . . . . 60
4.5 Tensor Formulation of Analytical Solid Geometry . . . 61
4.5.1 Distance Between Two Points P(xi ) and Q(yi ) 61
4.5.2 Angle Between Two Lines with Direction
Cosines . . . . . . . . . . . . . . . . . . . . . 61
4.5.3 The Equation of Plane . . . . . . . . . . . . . 62
4.5.4 Condition for Two Line Coplanar . . . . . . . 63
4.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . 64
xii Contents

5 General Tensor 67
5.1 Curvilinear Coordinates . . . . . . . . . . . . . . . . . 68
5.2 Coordinate Transformation Equation . . . . . . . . . . 68
5.3 Contravariant and Covariant Tensor . . . . . . . . . . 69
5.4 Contravariant Vector or Contravariant Tensor of
Order-One . . . . . . . . . . . . . . . . . . . . . . . . 71
5.5 Covariant Vector or Covariant Tensor of Order-One . . 71
5.6 Mixed Second-Order Tensor . . . . . . . . . . . . . . 72
5.7 General Tensor of Any Order . . . . . . . . . . . . . . 72
5.8 Metric Tensor . . . . . . . . . . . . . . . . . . . . . . 73
5.9 Associate Contravariant Metric Tensor . . . . . . . . . 74
5.10 Associate Metric Tensor . . . . . . . . . . . . . . . . 75
5.11 Christoffel Symbols of the First and Second -Kind . . . 76
5.12 Covariant Derivative of a Covariant Vector . . . . . . . 79
5.13 Covariant Derivative of a Contravariant Vector . . . . . 80
5.14 Exercises . . . . . . . . . . . . . . . . . . . . . . . . 81

6 Tensor in Relativity 85
6.1 Special Theory of Relativity . . . . . . . . . . . . . . 85
6.2 Four-Vectors in Relativity . . . . . . . . . . . . . . . . 88
6.3 Maxwell’s Equations . . . . . . . . . . . . . . . . . . 91
6.4 General Theory of Relativity . . . . . . . . . . . . . . 94
6.5 Spherically Symmetrical Metric . . . . . . . . . . . . 95
6.6 Planetary Motion . . . . . . . . . . . . . . . . . . . . 96
6.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . 97
 Contents xiii

7 Geodesics and Its Coordinate 99

7.1 Families of Curves . . . . . . . . . . . . . . . . . . . 99
7.2 Euler’s Form . . . . . . . . . . . . . . . . . . . . . . 100
7.3 Geodesics . . . . . . . . . . . . . . . . . . . . . . . . 101
7.4 Geodesic Form of the Line Elements . . . . . . . . . . 103
7.5 Geodesic Coordinate . . . . . . . . . . . . . . . . . . 105
7.6 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . 107

Index 109

About the Authors 111

 1
Introduction

The concept of tensor has its origin in the development of differential

geometry by Gauss, Riemann and Christoffel. The investigation of
relation, which remains valid, when we change from one coordinate
system to another coordinate system is the main aim of Tensor
Analysis. The law of physics can not be dependent on the frame of
reference, that the physicist chooses for the purpose of explanining
Tensor Analysis as the mathematical background in which such a law
can be formulated. In particular Einstein found it as excellent tool for
the presentation of the general theory of relativity. Its application to
most branches of Theoretical Physics.
Any natural law giving as it does a relation between different
physical entities is mathematically formulated as relation between the
set of number representation the entities natural law is that. It does not
depend upon the reference frame chosen for the reperesenation of the
entities and the mathmatical formulation of the law should desirably

1
2 Introduction

be such that it remains in the same form when the reference frame is
changed in any manner.
Tensor formulation is very compact and a good deal of clarity is
achieved. This subject had been originally formulated by G. Ricci
which come out as a very natural tool for the description of his
general theory of relativity. It has also since found useful for studies in
differential geometry, mechanics, electromagnitism and elasticity. The
orthogonal rectilinear co-ordinate system only relevent in relation to
the definition of Cartesian Tensor. General Tensor which are related to
the consideration of general curvilinear system of coordinates. It will
be seen that every general tensor is a cartesian tensor but the converse
is not true. Tensor analysis deals with entities and properties of the
choice of reference frames. Tensor is an ideal tool for the study of
natural law.

1.1 Symbols Multi-Suffix

A set of the numbers are denoted by a one suffix symbol such as ai ,

bi and ci , where i takes as its value the three numbers 1, 2, 3. Again
a set of nine numbers considered as a trial of triads is denoted by a
two-suffix symbol such as

aij , bij and cij etci = 1, 2, 3; j = 1, 2, 3.

The set of numbers given by any two suffix symbol aij can be exhibited
in the form of a square array or a matrix.
 1.2 Summation Convention 3
 
a11 a12 a13
 
 a21 a22 a23 
 
a31 a32 a33

where i and j respectively denote row and column suffixes.

A suffix may not take up three and instead may assume any number
of values of n. This will be the case, when we consider of n dimensions.
Thus, in the n dimension Euclidean space, the one-suffix symbol ai
will denote a set of n numbers.

ai ....a3...... an

and the two-suffix symbol aij will denote a set of n2 numbers,

which may be arranged in the form of a square array as
follows
 
a11 a12 a13
 
 a21 a22 a23 
 
a31 a32 a33

The number of suffixes in a multi-suffix symbol is called the order

and the number of values assumed by each suffix is the dimension of
the set.

1.2 Summation Convention

We have to consider the sum of numbers, which constitute certain sub-

set of the sets of numbers given by multi-suffix symbols, when these
4 Introduction

sub-sets arise on giving all possible equal values to some two of the
suffixes. Thus, we may have the sums.

i=n
X
aij = a11 + a22 + a33 + ann
i=1

i=n
X
aiji = a1j1 + a2j2 + a3j3 + anjn
i=1

in the relation to the sets aij and aijk , respectively. It is found

P
convenient to drop the sign of summation and regard the presence
of two equal suffixes itself as denoting summation. Thus, we shall
write,

aij = a11 + a22 + a33 + ann

aiji = a1j1 + a2j2 + a3j3 + anjn

aijij = a1j1j + a2j2j + a3j3 j + anjnj

It will be seen that aij denotes a single number and aiji are one-
suffix and two-suffix symbols respectively. Thus, a symbol having
two identical suffixes stand for the sum obtained by giving all
possible values to the identical suffixes and without altering the
remaining suffixes. The subject of tensor deals with the problem of
the formulation of relations between varous entites in form which
remain unchanged, when we pass from the on system of coordinate
to another. As the facts of invariance of an equation is essentally
 References 5

related to the possible types of system of co-ordinate which reference

to which the equation remains invariant, we must necessary define the
various system of coordinate system to be taken into account relative to
which the invariance is to be occured. We now state that the our basic
purpose of this book is the study of the invariance of equation relative
to the rectangular coordinate system in the three-dimension Euclidean
system. The book can be considered to be in two-parts- cartesian tensor
and general tensor.

References

[1] Harold Jeeffreys (1931), Cartesian Tensors, PP(1-16), Cambridge

University Press(New York)
[2] David C. Kay, Theory and Problem of Tensors Calculus, PP(1-3),
McGraw Hill, Washinton, D.C.
[3] Shanti Narayan (1961), Cartesian Tensors, PP(1-12), S chand,
New Delhi.
[4] DE Bourine and PC Kendell (1967), Vector Analysis and
Cartesian Tensor, PP(245-257), Chapman &Hall.
[5] Barry Spain (1960), Tensor Calculus, PP(1-55), Dover
Publication, Newyork.
[6] A.J. McConnell (1960), Application of Tensor Analysis, PP(1-9),
Khosla Publication, New Delhi.
[7] Zefer Ahson (2000), Tensor Analysis with Applications,
Anamaya Publisher, New Delhi.
[8] U.C. De (2008), Tensor Calculus, PP(1-9), Narosa Publishing
House, New Delhi.
6

[9] J.D Anand (2013), Mathmatical Physics V, PP(151-161), Hai

Anand Publication, New Delhi.
[10] J.K Goyal (1998), Tensor Calculus and Riemanninan Geometry,
Pragati Prakashan, Merrut.
[11] Charlie Harper, (1970), Introduction to Mathmatical Physics,
PP(255-275), Prentice Hall India Learning Private Limited,
Delhi.
[12] M.L Boas (1966), Mathmatical Methods in the Physical Sciences,
PP(496-526), Wiley.
[13] A.W Joshi, (1975), Matrices and Tensor In Physics, PP(187-219),
New Age International, Publishers, New Delhi.
 2
Cartesian Tensor

2.1 Introduction

The subject of tensor deals with the problem of the formulation of

relation between various entities in form which remain invariant, when
we pass from one system of coordinate to another. Invariant of equation
is necessary related to the possible types of system of coordinates
with reference to which the equation remains invariant. We now state
that the primary purpose of this chapter is the study of the invariant
of equation relative to the total of rectangular coordinate system
in the three-dimension Euclidean space. We have to start with the
consideration of manner in which the sets representing various entities
are transformation when we pass from one system of rectangular
coordinates to another. A tensor may be a physical entity that can
be described as a tensor only with respect to some manner of its
representation by means of multi-suffix sets associated with different
systems of axes such that the sets associated with different systems of
coordinate obey the transformation law for tensor. We have employed

7
8 Cartesian Tensor

suffix notation for tensors of any order, we could also employ single
letters such as A, B to denote tensors.

2.2 Transformation of Coordinates

If we have two systems of rectangular coordinate axes OX, OY, OZ;

0 0 0
OX , OY , OZ having the same origin O such that the direction
0 0 0
of cosines of the lines OX , OY , OZ relative to the system
OXYZ are

l1 , m1 , n1 ; l2 , m2 , n2 ; l3 , m3 , n3 .

0 0 0
The following transformation equations express x , y , z in terms
of x, y, z and vice versa:

0
x = l1 x + m1 y + n1 z,

0
y = l2 x + m2 y + n2 z,

0
z = l3 x + m3 y + n3 z,

0 0 0
x = l1 x + l2 y + l3 z ,

0 0 0
y = m1 x + m2 y + m3 z ,

0 0 0
z = n1 x + n2 y + n3 z ,

0 0 0
where x , y , z and x, y, z are the coordinates of the same point relative
to the two systems of coordinate axes. Now we write these equations
 2.2 Transformation of Coordinates 9

in a compact form in terms of multiple suffix sets and summation

convention. We say that xi , x̄i are the coordinates of a point P relative
to the two systems of axes, the range of each suffix being 1,2,3. Let lij
denote the cosine of the angle between OXI and OX¯J . The direction
cosines of OX¯1 ,OX¯2 ,OX̄3 relative to the system OX1 X2 X3 are given
by the columns and those of OX1 , OX2 , OX3 relative to OX̄1 X̄2 X3
are given by the row of the square matrix
 
l11 l12 l13
 
 
 l21 l22 l23 ,
 
 
l31 l32 l33

The equation of coordinate transformation can, be written in terms of

suffix notation as follows;

x¯1 = l11 x1 + l21 x2 + l31 x3 , x1 = l11 x¯1 + l12 x¯2 + l13 x¯3 ,

x¯2 = l12 x1 + l22 x2 + l32 x3 , x1 = l21 x¯1 + l22 x¯2 + l23 x¯3 ,

x¯1 = l13 x1 + l23 x2 + l33 x3 , x1 = l31 x¯1 + l32 x¯2 + l33 x¯3 ,

The summation convention equations are written as;

x¯1 = li1 xi , x1 = l1j x¯j ,

x¯2 = li2 xi , x2 = l2j x¯j ,

x¯3 = li3 xi , x2 = l3j x¯j ,

10 Cartesian Tensor

We can re-write these compactly as a single equations in the form

x¯j = li1 xi , xi = lij x¯j ,

which are complete equivalents of the equation of coordinate

transformation from one system to the another.

2.3 Relations Between the Direction Cosines of Three

Mutually Perpendicular Straight Lines

The direction cosines of the three mutually perpendicular straight lines

OX̄1 , OX̄2 , OX̄3 relative to the system OX1 X2 X3 are

l11 , l21 , l31 ; l12 , l22 , l32 ; l13 , l23 , l33 ,

They are connected by the following relations

l11 l11 + l21 l21 + l31 l31 = 1, l11 l12 + l21 l22 + l31 l32 = 0,

l12 l12 + l21 l22 + l32 l32 = 1, l11 l13 + l22 l23 + l32 l33 = 0,

l13 l13 + l23 l23 + l33 l33 = 1, l13 l11 + l23 l21 + l33 l31 = 0,

In summation convention, we can write these relations as

li1 li1 = 1, li1 li2 = 0,

li2 li2 = 1, li2 li3 = 0,

li3 li3 = 1, li3 li1 = 0.

 2.4 Transformation of Velocity Components 11

Finally all above equations can write as single equation

lij lik = δjk ,

where δjk is the Kroneocker delta.

2.4 Transformation of Velocity Components

It is known that any given velocity can be represented by means

of its three components along three mutually perpendicular lines
and the three components characterize the velocity components.
Clearly the component changes as we pass from one system of
mutually perpendicular lines to another. We consider two rectangular
systems

OX1 X2 X3 , OX̄1 X̄2 X̄3 ,

li, l¯j are the direction cosines of the line of action of the velocity
and v denotes the magnitude of the velocity. If vi , vi denote the
components of the velocity relative to the two systems of axes,
we have

vi = vli , v¯j = v l¯j . (2.1)

If now draw a line through the origin O parallel to the line of action
of the velocity, then the coordinate of the point P on this line at unit
distance from O relative to the two systems of axes are,

li , lj .
12 Cartesian Tensor

The equation of coordinate transformation, we have

l¯j = lij li , li = lij l¯j . (2.2)

From Eq.(2.1) and Eq.(2.2), we have

v¯j = lij vi , vi = lij v¯j . (2.3)

Thus, we see that the equation (2.3) for the transformation of

velocity.

2.5 First-Order Tensors

Any entity represented by a set of three numbers (component)

relative to a system of rectangular axes is called a first-order
tensor, if its components ai , a¯j relative to any two systems
of rectangular axes OX1 X2 X3 , OX̄1 X̄2 X̄3 are connected by the
relation

a¯j = lij ai ; (2.4)

with lij being the cosine of the angle between OX̄I

and OX¯J .
It can be seen that the components of first-order tensor obey the
same transformation law as the coordinate of a point. A tensor of first
order is also called a Vector. Any entity represented by a single number
 2.6 Second-Order Tensors 13

such that the same number represents the entity irrespective of any
underlying system of axes is called a tensor of zero order. A tensor of
order zero is also called a scalar.

2.6 Second-Order Tensors

Considering any two tensors of first order and letting

ai , bj ; a¯p b¯q ,

be the components of the same relative to two different systems of

axes, we have
a¯p = lip ai ,

b¯q = ljp bj ,

where lip , ljq have usual meanings. We have

a¯p b¯q = lip ai ljq bj

= lip ljq ai bj . (2.5)

The right-hand side of Eq.(2.5) denotes the sum 9 term, obtained

by giving the dummy suffixes i, j all possible pairs of value. Any entity
by two-suffix set relatively to a system of rectangular axes is called a
second order, if the sets aij , a¯pq representing the entity relative to any
two systems of rectangular axes
14 Cartesian Tensor

OX1 X2 X3 , OX̄1 X̄2 X̄3 are connected by the relation

a¯pq = lip ljq aij ,

Any entity consider by a set with, m suffixes relative to a system

of rectangular coordinate axes is called a tensor of order m, if the
sets aijkl............., āpqrs........... represented by the entity relative to any two
systems of rectangular axes OX1 X2 X3 , OX̄1 X̄2 X̄3 are connected by
the relation

āpqrs............. = lip ljq lrs lls .......aijkl............... ,

we say that aijkl...... are the components of the tensor relatively to the
rectangular system of axes OX1 X2 X3 .

2.7 Notation for Tensors

In elementery geometrical treatment, where a vector is defined as

a directed line segment. It is usual to denote a vector by a single
faced letter, such as a and b. In the present analytical method
approach

2.8 Algebraic Operations on Tensors

We now define operation on tensor such that by the means of different

operations, we can construct new tensors.
 2.8 Algebraic Operations on Tensors 15

2.8.1 Sum and Difference of Tensors

If aijkl....... , bijkl........ are two tensors of the same order, then

cijkl....... = aijkl..... + bijkl...........

is also a tensor of the same order.

Let
aijkl....... , bijkl........ and āpqrs....... , b̄pqrs........

be the components of the given tensors relative to two systems

OX1 X2 X3 , OX̄1 X̄2 X̄3 . We write

cijkl....... = aijkl..... + bijkl...........

c̄pqrs....... = āpqrs..... + b̄pqrs...........

Let, Iij denote the cosine of the angle between OXi and OX̄j . The
statement proved now showing that

c̄pqrs...... = Iip Ijq Ikr Ils aijkl..... (2.6)

As aijkl....... , bijkl........ are tensors, we have

āpqrs...... = Iip Ijq Ikr Ils aijkl..... (2.7)

b̄pqrs...... = Iip Ijq Ikr Ils bijkl..... (2.8)

16 Cartesian Tensor

Adding 2.6 and 2.7, we obtained 2.8. Hence the new tensor is said
to be the sum of the given tensors. The case of the diffference of tensors
can be obtain in a similar way.

2.8.2 Product of Tensors

If aijkl....... , bijkl........ are two tensors of the orders α and β, respectively,

then

cijkl........pqrs...... = aijkl....... bpqrs........

is a tensor of order α + β.
Let
aijkl......... , bpqrs.......,āi1 j1 k1 l1 ...... ,b̄p1 q1 r1 s1

be the components of the tensor relative to the two systems

OX1 X2 X3 , OX̄1 OX̄2 X¯3 .
We write

cijkl.....pqrs..... = aijkl..... bpqrs........

c̄i1 j1 k1 l1..... p1 q1 r1 s1........... = āi1 j1 k1 l1 b̄p1 q1 r1 s1 ...........

The new tensor obtained is called the product of the tensor. The
product of two tensor is a tensors whose order is the sum of the orders
of the given tensors.
 2.9 Quotient Law of Tensors. 17

2.9 Quotient Law of Tensors.

This law is useful in establishing tensorial character of given entities. If

there be an entity representable by a multi-suffix set aij , relative to any
given system of rectangular axes and if aij , bi is a vector whatsoever,
then aij is a tensor of order two.
We write

aij bi = cj (2.9)

so that cj is a vector.
Let

aij , bi , cj

and

āpq b̄p , c̄q

be components of the given entity and the two vectors relative to two
of axes OX1 X2 X3 , OX̄1 X̄2 X̄3
As given, we have

aij bi = cj ................. (2.10)

apq b¯p = cq ................. (2.11)

18 Cartesian Tensor

Also bi , cj being vectors, we have

cq =¯ljq cj − − − − (2.12)

bi = liq b¯p − − − − − − (2.13)

From these equation, we obtain

āpq b̄p = c̄q

= ljq cj

= ljq aij bi

= ljq aij lip bp

= lip ljq aij b̄p

(āpq − lip ljq aij )b̄p = 0. (2.14)

As the vector b̄p is arbitrary, we consider three vectors, whose

components relative to OX̄1 X̄2 X̄3 .

1, 0, 0; 0, 1, 0; 0, 0, 1.

For these vectors, we have from (2.14),

ā1q − li1 ljq aij = 0, ā2q − li2 ljq aij = 0, ā3q − li3 ljq aij = 0,
 2.10 Contraction Theorem 19

These are equivalent to

āpq − lip ljq aij = 0,

āpq = lip ljq aij ,

So that the components of the given entity obey the tensorial

transformation law.

2.10 Contraction Theorem

If
aijk.................

is a tensor of order m, then the set obtained on identifying any two

suffixes is a tensor of order (m-2).
Let

aijkl....................... apqrs.......................,

be the components of the given tensors relative to two

systems

OX1 X2 X3 , OX̄1 X̄2 X̄3

so that, we have

apqrs....................... = lip ljp lrs lls .................aijkl . (2.15)

20 Cartesian Tensor

Let us consider the second and fourth suffixes and we then

write

aijkl...................... = cik ..............., apqrs.......................... = cpr .

The theorem now amounts to show that

cpr............... = lip lkr.......................... cik .

Identifying q with s in Eq.(2.15), we get

apqrs....................... = lip ljs lkr lls .................aijkl . (2.16)

Now we have

ijs ils = δjl (2.17)

and
δjl aijkl.... = aijkl.... . (2.18)

From (2.16),(2.17) and (2.18), we have

apqrs = lip lkr...... aijkl ,

cpr............... = lip lkr.......................... cik .

Hence the theorem.

 2.11 Symmetric and Skew-Symmetric Tensor 21

2.11 Symmetric and Skew-Symmetric Tensor

Let
aijkl......, apqrs......

be the components of a tensor relative to two systems OX1 X2 X3 ,

OX̄1 X̄2 X̄3 . We shall prove that if aijkl..... is symmetric (skew-
symmetric) in any two suffixes, then so is also apqrs..... in the same
two suffixes.
We have

apqrs........ = lip ljq lkr lls........ aijkl . (2.19)

Now suppose that

aijkl.........

is a symmetric tensor in the second and fourth suffixes. Interchanging

q and s on the two-side of (2.19). We obtain

apqrs = lip ljs lkr lls....... aijkl........... (2.20)

As j and l are dummies, we can interchange them. Then

interchanging j and lon the right of (2.19). We obtained

apqrs = lip ljs lkr lls....... aijkl........... ,

= lip ljq lkr lls....... aijkl........... , (2.21)

22 Cartesian Tensor

with the set aijkl.... being symmetric in the second and fourth suffixes
from (2.19) and (2.21), we have

apqrs........ = apsrq.......

A tensor is said to be symmetric (skew-symmetric) in any two

suffixes, if its Components relative to every coordinate system are
symmetric (skew-symmetric) in the two suffixes in question. For
example, we may see that if ui , vj be any two vectors, then the two
second-order tensors.

ui vj + uj vi , ui vj − uj vi ,

are respectively, symmetric and skew-symmetric. If ui , vj

and wk are any three vectors then the three second-order
tensors

ui vj wk + uj vk wi + uk vi wj + ui vk wj + uj vi wk + uk vj wi

and

ui vj wk + uj vk wi + uk vi wj − ui vk wj − uj vi wk − uk vj wi

are respectively symmetric and skew-symmetric.

We shall now define two special tensor, alternate tensor and
kronecker tensor, of order-three and order-two respectively.
 2.12 Alternate Tensor 23

2.12 Alternate Tensor

Consider an abstract entity of order 3 and dimension 3 such that its

components relative to every system of coordinate axes are the same
and given by ijkl , where, any two of i, j, k are equal
ijkl = 1, if i, j, k is a cyclic permutation of 1,2,3
ijkl = −1, if i, j, k is a anti-cyclic permutation of 1,2,3.
Thus, for unequal values of the suffixes, we have

123 = 231 = 312 = 1, 132 = 213 = 321 = −1.

It will be shown that the entity is a tensor of order three. Let

OX1 X2 X3 , OX̄1 X̄2 X̄3 be two Systems of rectangular axes. Consider,
now the expression

lip liq lkr ijk . (2.22)

For any given system of values of p, q, r, s, the expression 1 consists

of a sum of 33 = 27 terms of which only 6 are non-zero for the other
21 terms corresponding to case when at least two of i, j, k are equal. It
may now be seen that the expression 1 is the same as the determinant.
 
i1p i2p i3p
 
 i1q i2q i3q  .
 
i1r i2r i3r

From the elementary properties of determinants, we see that this

determinant
24 Cartesian Tensor

D = 0, if any two of p, q, r are equal.

D = 1, if p, q, r is a cyclic permutation of 1, 2, 3.
D = −1, if p, q, r is a anti-cyclic permutation of 1, 2, 3.

Thus, we see that the components of the given entity in any

two-systems of rectangular coordinate axes satisfy the tensorial
transformation equation so that the entity is a tensor. This tensor is
known as the Alternate tensor. Clearly, alternate tensor is a skew-
symmetric tensor. The alternate tensor will always be denoted by the
symbol ijk.

2.13 Kronecker Tensor

Consider an entity of order two such that its components relativel to

every coordinate system of axes are the same and given by δij , where

δij = 0, if i 6= j

δij = 1, if i = j.

It will now be shown that this entity is a tensor. In the usual

notation, consider the expression

lip ljq δij ,

This is equal to

l1p l1q + l2p l2q + l3p l3q = 0, if p 6= q

 2.14 Relation Between Alternate and Kronecker Tensors 25

l1p l1q + l2p l2q + l3p l3q = 1, if p = q.

Thus, we see that the given tensor is Kronecker Tensor. The

tensorial character of δij can also be seen to follow easily from the
quotient law of tensors as follows. We have, for any arbitrary vector
ai ,

δij ai = ai ,

so that for any arbitray vector ai , δij ai is also a Vector. Hence, by

Quotient law δij is a tensor of order two. A tensor has the same set
of components relative to every system of coordinate axes is called
an Isotropic tensor. Alternate tensor and kronecker tensor are both
isotropic tensors of orders three and two respectively.

2.14 Relation Between Alternate and Kronecker Tensors

We shall now prove the following important relation between the

alternate tensor and kronecker tensor;

ijm klm = δik δjl − δil δjk.

Here each side is a tensor of order 4 so that the tensor equalities

requried to be proved is equivalent to a set of 81 scalar equalities,
We have to prove that

ij1 kl1 + ij2 kl2 + ij3 kl3 = δik δjl − δil δjk .
26 Cartesian Tensor

We may easily verify that when i, j are equal or when k, l are equal.
In the case of unequal values of i, j and unequal values of k, l, we
may again easily verify that if the pair of unequal values of i, k, j is
different from the pair of unequal values of k, l, Then

ijm klm = δik δjl − δil δjk .

Thus, we are left to consider the possibilities when i, j and k, l take

the pairs of values

1, 2; 1, 3; 2, 3, 2; 1, 3, 1; 3, 2.

Consider the first case so that we may have

i = 1, j = 2, k = 1, l = 2; i = 1, j = 2, k = 2, l = 1

i = 2, j = 1, k = 1, l = 2; i = 2, j = 1, k = 2, l = 1.

In the next case, we have

ijm klm = 1, δik δjl − δil δjk = 1; ijm klm = −1, δik δjl − δil δjk = −1

ijm klm = −1, δik δjl − δil δjk = −1; ijm klm = 1, δik δjl − δil δjk = 1.

The result may no be easily seen to be true in the other case.

2.15 Matrices and Tensors of First and Second Orders

We shall now make some observations and some facts to show how the
manipulation with tesnor of first and second orders can be brought into
 2.15 Matrices and Tensors of First and Second Orders 27

relationship with the algebra of matrices. Consider first any vector. Its
components ai relative to any system of axes may be written in the
form of a row or a column matrix as
 
  a1
 
a1 a2 a3 or  a2  .
 
a3

We shall be writing
 
h i a1
 
[ai ] = a1 a2 a3 or [ai ] = 
 2 .
a 
a3

Consider now any second-order tensor. Its components aij relative

to any system of axes can be written in the form of a matrix such that
aij occurs at the intersection of the ith row and the jth column; the first
elements to the left denoting row and the second denoting column.
Thus, we shall write
 
a11 a12 a13
 
[aij ] = 
 a 21 a 22 a 23
.

a31 a32 a33

However, it should be clearly understood that different matrices

may correspond to the same tensor depending on the system
of coordinate axes to which the tensor. A matrix obtained by
interchanging the row and columns of a given matrix is called the
transpose of the same. The transpose of [aij ] will be denoted by
28 Cartesian Tensor
0
[aij ] . Sum of two matrices of the same type is the matrix whose
elements are the sums of the corresponding elements of the two
matrices.

2.16 Product of Two Matrices

If [aij ] and [bij ] are two m × n and n × p matrices, then their product
is the m × p matrix [cik ] , where

cik = ai1 b1k + ai2 b2k + ai3 b3k + ..................ain bnk

= aij bjk

Considering the summation convention.

We shall now consider the different points. Consider a vector
whose components relative to the systems OX1 X2 X3 and OX̄1 X̄2 X̄3
are ai and aj . Then, lij having the usual meaning, we have

aj = lij ai = ai lij .

In the matrix form, this equality is equivalent to

 
h i h i l11 l12 l13

a1 a2 a3 = a1 a2 a3 
 l21 l22 l23 

l31 l32 l33

or

[aj ] = [ai ] [lij ] .

 2.16 Product of Two Matrices 29

Consider now a tensor of order 2 whose components relative to

OX1 X2 X3 and OX̄1 X̄2 X̄3 are aij and apq . Then, we have

apq = lip ljq aij

In matrix form this is equivalent to

h0 i
apq = lpi [aij ] [ljq ]

where

0
lpi = lip

 
a11 a12 a13
 
 a21 a22 a23 
 
a31 a32 a33
   
l11 l21 l13 a11 a12 a13 l11 l12 l13
   
=  l21   a21 a22 a23   l21 l22 l23  .
l22 l23    
l31 l23 l33 a31 a32 a33 l31 l32 l33

Consider now two tensors aij , bpq . The matrix [aij bjq ] of this inner
product is given by

[aij bjq ] = [aij ] [bjq ] .

This matrices of other inner products of the two given tensors

can also be written down similarly. Consider for example the inner
30 Cartesian Tensor

product
aij bkj

0
We write bjk = bkj . Thus,

h 0
i h 0 i
[aij bkj ] = aij bjk = [aij ] bjk

  
a11 a12 a13 b11 b12 b13
  
=
 a21 a22 a23   b21 b22 b23  .
 
a31 a32 a33 a31 a32 a33

Consider now two vectors ui , vj . We have

   
u1 u1 v1 u1 v2 u1 v3
 h i  
[ui , vj ] =  u2  v1 v2 v3 = 
 
 u 2 v1 u 2 v2 u 2 v3


u3 u3 v1 u3 v2 u3 v3

 
h i v1
 h i
[ui vj ] = u1 u2 u3  v2  = u1 v1 + u2 v2 + u3 v3 .
 
v3

It is usual to note that

 
1 0 0
 
[δij ] = 
 0 1 0 .

0 0 1
 2.17 Scalar and Vector Inner Product 31

2.17 Scalar and Vector Inner Product

2.17.1 Two Vectors

We should remember that associated with any tensor or system of

tensors, we may have other tensors, which arise through the operations
of multiplication and contraction as performed on the given tensors
and other known tensors, especially isotropic tensors. Infacts the tensor
notation supplies a very direct method for setting up invariant which
are entities independent of coordinate system, and tensor are such
entities. We shall now define scalar and vector products of two vectors
which are infact tensors of zeros-order and order-one associated with
the given vector in the manner referred to above. It will be seen the the
two types of products arises in tensor notation. In this connection, we
observe that in seeting from two vectors ui , vj . We are here interested
in setting up other tensors which are either scalars or vectors, tensors
of order zero or one.

2.17.2 Scalar Product

The scalar ui vi is called the scalar product of the two vectors ui , vi .

Thus, the scalar product

ui vj = u1 v1 + u2 v2 + u3 v3 . (2.23)

2.17.3 Vector Product

The vector ijkl ui vj is called the product of the vectors ui , vj taken

in this order. It may be easily seen that the components of this vector
32 Cartesian Tensor

product are

u2 v3 − u3 v1 , u3 v1 − u1 v3 , u1 v2 − u2 v1 . (2.24)

2.18 Tensor Fields

A tensor field or a tensor point function is said to be defined

when there is given a law which is associated to each point region
of space tensor of the same order. Thus a tensor field aij......... of
any order is defined if the components aij..... are a function of
x1 , x2 , x3 .

2.18.1 Gradient of Tensor Field

Let u be a scalar point function so that there is a value associated

with each point of a given region of space. Thus, if OX1 X2 X3 and
OX̄1 X̄2 X̄3 be any two system, then we can look upon u indifferently
as a function of x1 , x2 , x3 and of x1 , x2 , x3 or we can say x1 and xp
which are the coordinates of any point P relative to the system of axes.
For any point P x1 and xp are different but the values of ui are the
same.
au au
Considering now the sets of first-order ,
ax1 ax2
, we have

au auax1 auax2 auax3

= + +
axp ax1 axp ax2 axp ax3 axp

auaxi
= . (2.25)
axi axp
 2.18 Tensor Fields 33

Also, we have

xp = lip xi ,

or
xi = lip xp .

By differentiation, we have

axi
= lip . (2.26)
axp

From Eq.(2.25) and Eq.(2.23), we obtained

au au
= lip .
axp axi

We write

au au
= ai , = a.
axi axp

We have

ap = lip ai .

au
Thus we see that ai , axi
is a tensor of order one a vector.
au au
Components axi
and axp
relative to two systems of axes OX1 X2 X3 ,
OX̄1 X̄2 X̄3 obey the tensorial transformation law. This vector is called
the gradient of the scalar u.
34 Cartesian Tensor

2.18.2 Divergence of Vector Point Function

The scalar of the gradient of a vector point function is called the

divergence of the point function. Thus, if ui is a vector function so
that

aui
uij = ,
axj

is its gradient, then

aui au1 au2 au3

uij = = + + .
axj ax1 ax2 ax3

is called the divergence of ui denoted by the symbol Divui . The

divergence of any tensor is defined as gradient of the contracted on
the first and the last indices. Thus

Div.uijkl = uijkl.................i .

2.18.3 Curl of Vector Point Function

The vector of the gradient of vector point function is called the curl of
the point function. Thus, if ui is a vector function so that

aui
uij = ,
axj

is its gradient, then the vector of the tensor, the vector

ijkl uij
 2.21 Exercise 35

is called the curl or rotor of ui denoted by the symbol curl ui . Clearly,

the components of curl ui are

au3 au2 au1 au3 au2 au1

− , − , − .
ax2 ax3 ax3 ax1 ax1 ax2

The Skew-symmetric second-order tensor associated with the

vector curl ui is

uij − uji .

2.19 Tensorial Formulation of Gauss’s Theorem

If F is a continuously differentiable vector point function and S is a

closed surface enclosing a region V, then
Z Z
S Fi ni dS = V Fi,i dV, (2.27)

where n is the unit outward normal vector.

2.20 Tensorial Formulation of Stoke’s Theorem

If F is a continuously differentiable vector point function and S is a

closed surface enclosing by a curve C, then
Z Z
C Fi ti ds = S ijkl Fk,j ni ds, (2.28)

where the unit vector n at any point S is drawn in the sense in which a
right handed screw would rotated in the sense of description of C.
36

2.21 Exercise

Prove that
(1) δij = 3.
(2) δij ijk = 0.
(3) ijk ijk = 6.
(4) pjk kpj = 2δpq .
(5) Show that
 
δil δim δin
 
ijk kpj =  δ δ δ
 jl jm jn 

δkl δkm δkn

and deduce the relation between alternate tensor and Kronecker

tensor.
(6) If lij are the direction cosines of a transformation of axes prove
that

ikm lmn = jln lij lkl ,

1
lij = ikm jln lkl lmn .
2

(7) if aij is a skew-symmetric second-order tensor, prove that

(δij δik + δil δjk )aik = 0

(8) if aijk is a tensor, prove that ajki is also a tensor.

 References 37

References

[1] Harold Jeeffreys (1931), Cartesian Tensor, PP(1-66), Combridge

University Press (New York).
[2] David C. Kay, Theory and Problem of Tensor Calculus, PP(1-3),
McGraw Hill, Washinton, D.C.
[3] Shanti Narayan (1961), Cartesian Tensor, PP(37-51), S. Chand,
New Delhi.
[4] Barry Spain (1960), Tensor Calculus, PP(1-55), Dover Publication,
Newyork.
[5] Zefer Ahson(2000). Tensor Analysis with Application, Anamaya
Publisher, New Delhi.
 3
Tensor in Physics

3.1 Kinematics of Single Particle

Velocity, momentum and acceleration; consider any particle of mass m

and let at time t, the particle be at point P, where coordinates relative
to a system of axes OXYZ are xi , then the three components of the
dxi
velocity dt
are said to be the components of the velocity of the moving
0 0 0
particle at time t relative to the system OXYZ. Let now OX Y Z
be another system of axes fixed relative to OXYZ so that with usual
notation, we have

xj = lij xi , (3.1)

where lij are constants, differentating Eq.(3.1), with respect to time

we have
0
dxi dxi
= lij .
dt dt
dx;i dx;i
This transformation law, for the components dt
and dt
. Show that
the velocity of a particle is a vector.

39
40 Tensor in Physics

3.1.1 Momentum

The mass m of a particle being scalar

dxi
m ,
dt

called momentum is also a vector.

3.1.2 Acceleration

The three components of acceleration

dx2i
, i = 1 − 3,
dt2

Now component of transformation equation

d 2 xi dx2i
= lij ,
dt2 dt2
dx2i d2 xi
where, dt2
and dt2
satisfied the transformation law for tensor.

3.1.3 Force

Experimentally it was found that one of the reference frames for

which the Newtons’s second law of motion holds to a good degree of
approximation is the reference frame attached to the so called “Fixed
Stars”. In any case, we shall say that any reference frame for which
the law is true is called an inertial frame and we shall assume that the
underlying reference frame is always an inertial frame.
 3.3 Work Function and Potential Energy 41

If Xi is the components of the force along the axes acting on a

particle of mass m, where position at time t is xi , then by Newton’s
second law of motion with appropriate choice of units, we have
0
d 2 xi
m = Xi .
dt2
dx2i
As m is a scalar and dt2
is a vector, we deduce that Xi is also a
vector.

3.2 Kinetic Energy and Potential Energy

If Xi be the force acting upon a particle whose position at any time

t is xi then the work done by the force is given by the line intergal
R
C
Xi dxi , the integral being taken along the path C of integration.
Clearly the work as scalar. The scalar 12 mvi vi is the kinetic energy
of a particle whose mass is m and velocity, vi .

3.3 Work Function and Potential Energy

The work done by a force as a particle moves from some given point
ai to another point xi which depends in general the path followed
from ai to xi and is not as such a point function. We shall however
now consider a case where, the work function is necessary a scalar
function.
Let now Xi be any field of force so that Xi is a vector point
function. We say that this field is conservative if there exists a scalar
42 Tensor in Physics

point function U such that

aU
= Ui − Xi
axi

or in other words,

gradU = −Xi .

Thus, as a particle moves from ai to another point xi along any

path C in the conservative field of force Xi the work done.
Z
= Xi dxi
C

Z
=− Ui dxi
C

Z
aU
=− dxi
C axi

− |U |xaii = U (ai ) − U (xi ) (3.2)

which is depends on the initial and final points and not on the path. We
have

d2 xi
m = Xi ,
dt2

dxi d2 xi dxi
m 2
= Xi .
dt dt dt
 3.4 Momentum and Angular Momentum 43

Integrating, we get

Zt Zt
dxi d2 xi dxi
m dt = Xi dt
dt dt2 dt
to to

Zt
1 dxi dxi
|m |to = Xi dxi . (3.3)
2 dt dt
to

Writing T = m dx i dxi
dt dt
, we obtain from Eq.(3.2) and Eq.(3.3)

T (xi ) − T (ai ) = U (ai ) − U (xi )

T (xi ) + U (xi ) = T (ai ) + U (ai ).

Thus, we see that for a conservative fields of force, the sum of the
kinetic and potential energies remains constant.

3.4 Momentum and Angular Momentum

The momentum generally denoted by Mi of a system is the sum

of the momenta of the particles of the system. Momentum of a
system is also known as linear momentum. The angular momentum
denoted by Hi (O) or simply by Hi about any point O is the sum
of the vector products of the position vectors relative to O and the
particles of the sum of the moments of the momenta of the particle
about O.
44 Tensor in Physics

Thus, we have
X
Mi = mvi
X
Hi (O) = ikl xk mvi ,

where vi is the velocity of the particle of mass m at xi and the

summation extends to all the particles of the system. The definition of
moment as a skew-symmetric second-order tensor, in that, we would
write

Hij (O) = xi vj − xj vi .

The following discussion could also be carried out with Hij (O)
instead of H(O).

3.5 Moment of Inertia

We shall now introduce the concept of moment of inertia tensor of

a rigid body relative to any point. It plays an important part in all
discussions relative to the dynamics of rigid body. By the definition,
of the moment of inertia of a material system about any line OL is

X
mp2

where p is the perpendicular distance of any particle of mass m of the

system from the line OL, and the summation extends to all the particles
of the system.
 3.5 Moment of Inertia 45

We take O as origin and any system of coordinate axes OX1 X2 X3 .

Let li be the direction of cosines of any given line OL and let P (xi )
be any particle of the system having mass m. We now need the length
MP. Now the magnitude of the vector.

ipq lp lq

M P 2 = (ipq lp xq )(irs lr xs )

= ipq irs lp lr xq xs

= lr (xq xq δrs − xs xr )ls . (3.4)

Thus, the moment of inertia of the system about the line OL

X
I= mlr (xq xq δrs − xs xr )ls
X X
= lr (mxq xq δrs − mxs xr )ls . (3.5)

We write

X
lrs = m(xq xq δrs − xs xr ) (3.6)

So that the moment of inertia about the given with direction

cosines lr is

lr lrs ls .
46 Tensor in Physics

It is shown that lrs is a symmetric tensor of second order

independent of the direction cosines of the lines OL and depending
only on the configuration of the given material system relative to
the given material system to the point O. It is known as inertia
tensor. It may easily be seen that the matrix of the components of the
tensor lrs is

 P 
m(x22 + x23 ) − mx1 x2
P P
− mx1 x3
 
 − P mx1 x2 P m(x2 + x2 ) − P mx2 x3 .
 3 1 
m(x21 + x22 )
P P P
− mx3 x1 − mx3 x2

In the case of a continuous system, sums have to be replaced

by integrals. It may easily, seen that I11 , I22 , I33 are the moments of
inertia about the axes OX1 , OX2 , OX3 respectively. The product of
P
inertia about the lines OX1 , OX2 is mx1 x2 and the components
I23 , I31 , I12 of the inertia tensor are the product of inertia with signs
changed about the axes taken in pair.

3.6 Strain Tensor at Any Point

When some forces are applied to a body, the particle of the body
undergoes relative displacement so that the body is deformed. We can
say that the body has experienced a strain or that it is strained. We
shall here be concerned about only small deformation. The rigid body
displacement of translation and rotation do not produce any relative
displacement of the particle so that such displacement do not constitute
strain. We shall now proceed to the analysis strain produced in the
 3.6 Strain Tensor at Any Point 47

neighbourhood of any given point of the body. Let the position of

any particle of the body be determined by its coordinate xi referred
to any set of rectangular coordinate axes consider at any point P (xi )
of the body and suppose that as a result of deformation it is displaced
0
to another point P (xi + si ) so that, we have

−−→0
P P = Si .

clearly, Si is a function of x1 , x2 , x3 and so we may write

S1 = f1 (x1 , x2 , x3 ) (3.7)

here we assume that the three-functions of f1 possess continuous first-

order partical derivatives.
We shall now compute the change in any vector

−−→0
P Q = hi (3.8)

which takes place as a result of the deformation of the body. Let Q be

0
−−→0
displaced to Q so that the vector P Q becomes after displacement the
−−0−→ −−→0
vector P Q” . As the point Q is (xi + hi ), the displacement QQ of the
point Q is

f1 (x1 + h1 , x2 + h2 , x3 + h3 )
af1 af2 af3
= f1 (x1 , x2 , x3 ) + h1 + h2 + h3
ax1 ax2 ax3
af1
= f1 (x1 , x2 , x3 ) + h1
ax1
48 Tensor in Physics

where, we have neglected square and higher power of h1 , the point Q

being in the neighbourhood of P point Q is

af1 af1
x1 + h1 + f1 (x1 , x2 , x3 ) + h1 = x1 + h1 + h1
ax1 ax1

Thus,

−−0−→ afi
P Q” = (x1 + h1 + s1 + hj ) − (x1 + s1 )
axj

afi
s1 + hj
axj

and

−→
P Q = hi .

−→ −−0−→
thus we see that a vector P Q = hi is displaced to P Q” = s1 +hj axafi
j
so
−→ afi
that the change in P Q is hj axj = hj aij , say. By Quotient law, we see
that aij is a tensor of order two. Its components being function of xi ,
we now break up the deformation tendor aij as the sum of a symmetric
and skew-symmetric tensor, we write

1 1
aij = (aij + aji ) + (aij − aji )
2 2

= eij + wij , say

where,eij is symmetric and wij is skew symmetric hi is displaced

to hi + wij hj as a result of rigid body motion of rotation about
 3.7 Stress Tensor at any Point P 49

P. The part of the displacement given by eij hi is called pure strain

and the tensor eij is called the strain tensor at P. The pure strain
gives the relative displacement of the particle as a result of the
deformation.

3.7 Stress Tensor at any Point P

The force acting on a body are either external or internal. The force
consists either of body force such as gravity that is acting on every
particle. If F1 denotes the body force vector per unit volume then the
force acting on an element of volume ∆V is F1 ∆V. Tj denotes the
surface force vector per unit area, then the force acting on an element
of surface ∆S is Tj ∆S. In order to discuss the internal force, we
assume a small element of area ∆S inside the body and denotes the
direction cosines of the normal to this element, which is approximately
planer by nj . We call one side of the element ∆S positive and other
is negative. Then the action of the positive side on the negative side is
the internal surface force Tj ∆S, where Tj is the force per unit area on
the element ∆S. It is called the stress tensor and in general. A function
of the coordinate of the point, which determines the position of the
element ∆S and of the direction cosines nj of the normal to ∆S.
Consider a small rectangular parallelepiped with vertex at the point
P, whose edges are parallel to the coordinate axes. We consider three
stress vectors T(1)i , T(2)j , and T(3)k corresponding to the small elements
of area through P, which are parallel to the coordinate planes. The
stress vectors T(i)j will be called positive it act in the positive direction
of the Y axes. If, however the external normal is co-directional with
50 Tensor in Physics

the negative Y axis. In other words, a stress which tends to stretch will
be regarded as positive. we define nine quantities Eij by the equation

Eij = T(i)j (3.9)

We shall show that Eij is a cartesian tensor called the stress Tensor.
It follow the quotient law that Eij is a cartesian tensor, we now cite
several important case of stress.

3.7.1 Normal Stress

The vector Tij is co-directional with ±nj . Eij = Cδij , where C is a

constant hydrostatic pressure which, is an example of normal stress for
which C is negative.

3.7.2 Simple Stress

Consider the stress tensor Eij = Cli lj , where C is constant and li is a

unit vector. Then the stress vector in the direction li is Tj = Clj . If C
is negative, the stress is called a simple stress.

3.7.3 Shearing Stress

This is specified by the stress tensor Eij = C(li mj + lj mi ), where C

is a constant and li and mi are unit vectors.

3.8 Generalised Hooke’s Law

In the elementary theory of elasticity Hooke’s law states that the tenson
of a string is proportional to strain. The corresponding assumption
 3.9 Isotropic Tensor 51

in the general theory of elasticity is that the stress tensor is a linear

homogeneous function of the strain tensor. That is

Eij = ijkl ekl . (3.10)

It is follows from the quotient law that ijkl is a cartesian tensor

of the fourth order and it is called the elasticity tensor. Further, from
the symmetry of Eij and ekl , we find that eijkl is symmetric not only
with respect to the indices i and j but also with respect to k and l. A
body is said to be homogenous. If the elastic properties of the body are
independent of the point under consideration. We call a body isotropic
if the elastic properties at a point are the same in all directions at
that point. This means that the elasticity tensor, ijkl itself under any
rotation axes.

3.9 Isotropic Tensor

A cartesian tensor which transforms into itself under a rotation of axes

is called isotropic tensor, we know two isotropic tensors namely δij and
εijk . We shall now search for the most general isotropic tensor ijkl of
the fourth-order. Its transformation law becomes

ijkl = air ajs akl ale rstv (3.11)

We can collect the given relation

iiii = jjjj

iijj = iikk = lljj = iikk.

52 Tensor in Physics

ijij = ikik = lllj = ikik.

ijji = ikki = ljjl = lkkl.

Where i, j, k, l are unequal and the summation convention does not

apply. All other components are zero. The most general solution of the
equation is then

ijkl = λδij δkl + µδik δjl + vδjl δjk + kδij δkl (3.12)

where λ, µ, v and k are cartesian invariants and δijkl = 1. If all four

indices are equal and otherwise zero. Finally we carry out a small
rotation, which is represented by

aik = δik + sik (3.13)

where sik is skew-symmetric and of the first order in small quantities.

3.10 Exercises

(1) Show that the strain tensor satisfies the identical relation

ijkl + klij = ikjk + jikl

known as the equation of compatibility

(2) Show that the kinetic energy of a rigid body is

1
T = M ui vi + Tij
2
 References 53

where vi is the velocity of the centre of a gravity. Tij is the kinetic

energy in regard to the motion of the body relative to the centre
of gravity and M is the mass of the body.
(3) If the points of a rigid body have all two simultaneous velocity
due to two angular velocity ωr1 , ωr2 . Show that the resultant
velocity are given by an angular velocity vector

ωr = ωr1 + ωr2 .

References

[1] Harold Jeeffreys (1931), Cartesian Tensor, PP(1-16), Combridge

University Press (New York).
[2] David C. Kay, Theory and Problem of Tensor Calculus, PP(1-3),
McGraw Hill, Washinton, D.C.
[3] Shanti Narayan (1961), Cartesian Tensor, PP(106-127), S. Chand,
New Delhi.
[4] Barry Spain (1960), Tensor Calculus, PP(1-55), Dover Publication,
Newyork.
[5] Zefer Ahson(2000). Tensor Analysis with Application, Anamaya
Publisher, New Delhi.
 4
Tensor in Analytic Solid Geometry

We have defined a vector analytically as a one suffix with prescribed

transformation law, and we shall obtain a geometrical interpretation
of the vector. This will bring our analytical treatment in line with
the usual geometrical treatment. We shall also develop vector algebra
in suffix notation corresponding to the usual algebra in term of
one letter notation for vectors defined geometrically as directed line
segments.

4.1 Vector as Directed Line Segments

Let ai be the component of a vector relative to a system of axes

OX1 X2 X3 . Take points A1 , A2 , A3 in the three axes such that with
some choice of scale,

OA1 = a1 , OA2 = a2 , OA3 = a3 .

Let OA be the diagonal through O of the parallelopiped with

OA1 , OA2 , OA3 as coterminous edges. Then we say that the directed

55
56 Tensor in Analytic Solid Geometry
−→
line segment select OA represents the vector ai . Suppose that ai and
aj are the components of a vector relative to two-system of axes
OX1 X2 X3 , OX̄1 X̄2 X̄3 so that with the usual notation

a¯j = lij ai . (4.1)

−→
Let OA be the directed line segment whose projection on the axes
of the system OX̄1 X̄2 X̄3 is ai . It will now be shown that the projection
−→
of OA on the axes of the system OX̄1 X̄2 X̄3 is a¯j .
Let li , l¯j be the direction cosine of the line OA of the two systems
of axes, so if r be the length of the line.
We have

ai = rli , (4.2)

and

l¯j = lij li ,

or equivalently

li = lij l¯j ,

now

a¯j = lij ai

= lij rli
 4.3 Length and Angle between Two Vectors 57

= rlij li

= rlij lik l¯k

= rδjk l¯k

= r.l¯j .

−→
Thus, the projection of OA on OX̄j is a¯j .

4.2 Geometrical Interpretation of the Sum of two Vectors

Let the given vectors ai , aj be represented by two directed line

−→ −−→
segments OA,OB and let OP be a diagonal of the parallelogram
with OA, OB as a pair of adjacent sides. Then, we shall show
−→
that OP represents the vector ai + bi , this is equivalent to show
−→
that the projection of OP on any line is equal to the sum of the
−→
projections of OA and this is known to be true from elementary
geometry.

4.3 Length and Angle between Two Vectors

The length of a vector is meant, to be the length of the directed line

segment representing the vector. The angle between, the two vectors
is meant the angle between the directed line segments representing the
vectors. Let any vector ai be represented by a directed line segment
−→
OP . By the consideration of the rectangular parallelopiped with one
diagonal OP and edges parallel to the coordinate axes, we may easily
58 Tensor in Analytic Solid Geometry

see that
√
q
OP = a21 + a22 + a23 = a1 a1 .
−→ −−→
Let θ be the angle between the directed line segments OA OB
representing the vector a1 , b1 from the ∆OAB, we have

OA2 + OB 2 − AB 2
cosθ =
2OAOB

[a1 a1 + b1 b1 − (b1 − a1 )(b1 − a1 )]

= √ √
2 a1 a1 b 1 b 1

4.4 Geometrical Interpretation of Scalar and Vector

Products
−−→ −−→
If OA, OB are two directed lie segments representing vectors ai , aj ,
then we have

ai b = OA.OB.cosθ,

so that the scalar product of two vectors denotes the product of the
length of the vectors and the cosine of the angle between the vectors.
We now come to the case of the vector product

ck = ijk ai bi ,

of the vector ai , bj .
We first find the length of ck , and we have

ck ck = (ijk ai bj )(pqk ap bp )
 4.4 Geometrical Interpretation of Scalar and Vector Products 59

= ijk pqk ai bj ap bq

= (δip δjq − δiq δjp )ai bj ap bq

= (δip ai bp )(δjq bj bq ) − (δiq ai bq )(δjp bj ap )

= (ap ap )(bq bq ) − (aq aq )(bp bp )

= OA2 OB 2 − (OA.OBcosθ)2

= (OA.OBsinθ)2 .

Thus, the length of ck is OA.OBsinθ.

Again, we have
 
a1 a2 a3
 
ck ak = ijk ai bj ck = 
 a 2 b 2 b 3
 = 0,

a3 c 3 c 3

and similarly

ck bk = ijk ai bj bk = 0.

Thus, the vector ck is perpendicular to the both the vectors ai , bj .

−→
Finally, we may see that the directed line segments OA,
−−→ −→
OB, OC representing the three vectors ai , bj , ck from a right
haded or left handed set according as the set of axes is right-
handed or left handed. This follows from the fact that the
determinant
60 Tensor in Analytic Solid Geometry
 
a1 a2 a3
 

 b1 b2 b3 

a2 b 3 − a3 b 2 a3 b 1 − a1 b 3 a1 b 2 − a2 b 1

formed by the compoents of the three vectors in the order given is

positive.

4.4.1 Scalar Triple Product

The scalar ijk ai bj ck is the scalar product of ck with the vector product
ijk ai bj of ai , bj . The properties of the scalar triple product can all be
easily shown. Consequences of the properties of the alternate tensor
ijk . In fact, they may not even the paid any special attention, if we
work in suffix notation.

4.4.2 Vector Triple Products

The vector ijk ai bj is the vector product of the vectors ai , bj and

kpq ijk ai bj cp is the vector product of the vector of ai , bj with
cp . Here, i, j, k, p are dummy suffixes and q is the suffix. We
now have

kpq ijk ai bj cp = pqk ijk ai bj cp

= (δpi δqj − δpj δqi )ai bj cp

= δpi δqj ai bj cp − δpj δqi ai bj cp

 4.5 Tensor Formulation of Analytical Solid Geometry 61

= (δpi ai )(δqj bj )cp − (δpj bj )(δqi ai )cp

= ap b q c p − b p ap c q

= ap c q b q − b p c p aq .

Here ap cp and bp cp are the scalar products of the vectors ai , bj and

bj , ck respectively.

4.5 Tensor Formulation of Analytical Solid Geometry

We have seen that point and displacement are both cartesian tensors
of order one. With the help of this fact, we shall briefly consider the
tensorial formulation of linear analytic solid geometry.

4.5.1 Distance Between Two Points P(xi ) and Q(yi )

We have

−→ −→
OP = XI , OQ = YI .
−→ −→ −→
P Q = OQ − OP = yi − xi ,

P Q2 = (yi − xi )(yi − xi ).

4.5.2 Angle Between Two Lines with Direction Cosines

Let li , mi are two points P, Q on the line through O parallel to the given
lines and unit distance from O. Thus
62 Tensor in Analytic Solid Geometry
−→ −→
OP = lI , OQ = mi .

cosθ = li mi .

Where θ is the angle between two lines.

4.5.3 The Equation of Plane

Let li be the direction cosines of the normal to the given plane and
p the length of the perpendicular from the origin to the plane. Take
any point P (xi ) on the plane. Let K be the foot of the perpendicular
from the origin to the plane. The projection of OP on OK is
OQ=p. Also

−→
OP = xi

li xi = p,

is the required equation. If the equation of a plane is

axi + d = 0.

the direction cosine of the normal to the plane are proportional to ai

and the length of the perpendicular from the origin to the plane is

|d|
√ .
a1 a2
 4.6 Exercises 63

4.5.4 Condition for Two Line Coplanar

Let two lines are xi = ai + sli , xi = bi + lmi . The line intersects,

if and only if there exist scalar values s, t such that

ai + sli = bi + lmi

ai − bi = tmi − sli ,

To eliminate s, t, we multiply with pq li mk and also contract

identifying i, p; j, q; k, r and obtain

ijk (ai − bi )lj mk = tijk mi li mk − sijk li li mk .

Thus, the condition is

ijk (ai − bi )li mk = 0. (4.3)

As xi is any point on the second line, this condition can be written

as

ijk (xi − ai )li mk = 0. (4.4)

Thus, every point on the second line satisfied Eq.(4.4). Thus,

Eq.(4.3) is the condition of coplanarity and assuming the condition
to be satisfied, Eq.(4.4) is that of the plane through the two lines.
64 Tensor in Analytic Solid Geometry

4.6 Exercises

(1) Show that rotations through 90o about the lines OX1 , OX2 are
given by
   
1 0 0 0 0 1
   
|aij | = 
 0 0 −1 ,
 |bij | = 
 0 1 0 

0 1 0 −1 0 0

Show also that the resultants of these two rotations taken

in the two orders are rotations through lines with direction
cosines.

1 1 1 1 1 1
√ , √ , −√ ; √ , √ , −√ ,
3 3 3 3 3 3

through angle 180o

(2) Verify that the tensor aij , which corresponds to a rotation through
an angle θ about line li is orthogonal.
(3) Find the condition that the line li xi + p = 0 may touch the
quadric S.

References

[1] Harold Jeeffreys (1931), Cartesian Tensor, PP(16-23), Combridge

University Press (New York).
[2] David C. Kay, Theory and Problem of Tensor Calculus, PP(1-3),
McGraw Hill, Washinton, D.C.
 References 65

[3] Shanti Narayan (1961), Cartesian Tensor, PP(54-80), S. Chand,

New Delhi.
[4] Barry Spain (1960), Tensor Calculus, PP(17-23), Dover
Publication, Newyork.
[5] Zefer Ahson(2000). Tensor Analysis with Application, Anamaya
Publisher, New Delhi.
 5
General Tensor

In this chapter, we will discuss the basic concepts of general

tensor in three-dimensional Euclidean space. The notation of general
tensor, related to arbitrary coordinate systems in curvilinear system
of coordinates. In notation suffixes both as superscripts as well as
subscripts and the coordinate of a point will be given by superscripts,
according the coordinate of a point will be denoted by

x1 , x2 , x3 .

Here superscripts must on no account be confused. In this context,

we shall also modified the summation convention and say that if in
any symbol the same suffix appears both as a superscript as well
as a subscript, then the symbol in question denotes a sum of three-
numbers obtained by giving values 1,2,3 to the repeated suffix, for
example

aii = a11 + a22 + a33 ,

67
68 General Tensor

aiji = a1j1 + a2j2 + a3j3 .

5.1 Curvilinear Coordinates

Consider arbitrary region R of space and three continuously

differential functions

x1 = f 1 (y 1 , y 2 , y 3 ), x2 = f 2 (y 1 , y 2 , y 3 ), x3 = f 3 (y 1 , y 2 , y 3 )

in general

xi = f i (y 1 , y 2 , y 3 ) (5.1)

defined in R: y 1 , y 2 , y 3 is the rectangular Cartesian coordinates of

any point p. Thus, to each point y 1 , y 2 , y 3 of R. The equation (5.1)
associated a set of numbers x1 , x2 , x3 . We say that x1 , x2 , x3 are the
rectilinear coordinate of the point P.

5.2 Coordinate Transformation Equation

Consider now three equations

xj = φj (x1 , x2 ,¯x3 ); j = 1, 2, 3. (5.2)

where φj are three continuously differential functions. These equations

associate to each point Q by x1 , x2 , x3 another set of three numbers
x¯1 , x¯2 , x¯3
 5.3 Contravariant and Covariant Tensor 69

Also we suppose that this association is one-one. No two different

unbarred sets correspond to the same barred set. It is known from
analysis that this fact will be guaranteed if the determinant

∂x.j
6= 0
∂xi

∂ x¯1 ∂ x¯1 ∂ x¯1

∂x1 ∂x2 ∂x3
∂ x¯2 ∂ x¯2 ∂ x¯2
6= 0
∂x1 ∂x1 ∂x1
∂ x¯3 ∂ x¯3 ∂ x¯3
∂x1 ∂x1 ∂x1

At any point P. In the following determinant is not zero. Thus, x¯j

can also be thought of as the curvilinear coordinate of the point P. The
equation constitutes transformation equation between the two systems
of coordinate xi and x¯j .

5.3 Contravariant and Covariant Tensor

First, we shall consider two typical cases of transformation law, which

will play an important role in the following discussion, consider two-
systems of coordinate xi and x¯j and any point Q,
(1) we have

¯j
∂x ¯j
∂x ¯j
∂x
dxj = dx i
+ dx 2
+ dx3
∂xi ∂x2 ∂x3
70 General Tensor
¯j
∂x
= dxi . (5.3)
∂xi
¯ j associated with the same
Here, we have two sets dxi and dx
point in the two system of coordinates xi and x¯j such that
the equation (5.3) constitutes is the law of transformation
from one set to the another. The coefficient of transformation
depends only on the point Q and the transformation for the
coordinate.
(2) Now any function φ of the point Q so that to each point P
correspond a number φ can be thought of indifferently as a
function of xi as well as of x¯j such that two sets xi and x¯j giving
the coordinate of the same point correspond to the same value of
φ, we have

∂φ ∂φ ∂x1 ∂φ ∂x2 ∂φ ∂x3

= + +
∂ x¯j ∂x1 ∂ x¯j ∂x2 ∂ x¯j ∂x3 ∂ x¯j

∂φ ∂xi
= . (5.4)
∂xi ∂ x¯j
∂φ ∂φ
Here we have two sets ∂xi
and ∂ x¯j
associated with the same point
in the two coordinates xi and x¯j such that the equation(5.49)
constitute is the law of transformation from one set to the other,
the coefficients of transformation depend only on the point and
the equation of transformation of coordinates. In the two cases,
it will be seen that the law of transformation are in general
different in either case member of one set are expressed as
 5.5 Covariant Vector or Covariant Tensor of Order-One 71

linear combination of the member of other. The matrix of the

coefficients of transformation is constant for different pairs of
∂φ ∂φ
sets dxi , , , provided only that we do not change the
∂xi ∂xj
point Q.

5.4 Contravariant Vector or Contravariant Tensor of

Order-One

Contravariant vector, if at each point Q, is represented by a set of three

numbers such that if Ai , Āp of coordinate of two sets representing the
entity relative to two systems of coordinates xi , xp , then

∂xp i
Āp = A
∂ x̄i

Contravariant vector or Contravariant tensor of second order

∂x¯p ∂xq ij
Apq = A
∂ x̄i ∂ x¯j

5.5 Covariant Vector or Covariant Tensor of Order-One

Covariant vector, if at each point Q it is represented by a set of three

numbers such that if Ai , Āp of coordinate of two sets representing the
entity to two system of coordinates xi , xp , then

∂xi
Āp = Ai .
∂ x¯p

It should be always noted that the upper suffix denotes

contravariant and lower suffix denotes covariant.
72 General Tensor

Covariant vector or covariant tensor second order

∂x¯i ∂xj
Apq = Aij.
∂ x¯p ∂ x¯q

5.6 Mixed Second-Order Tensor

A mixed second order tensor, if at each point Q, is represented by a set

Aij , such that if Aij , Āpq two sets representing the entity Relative to the
two system of coordinate xi , x¯j , then

∂ x¯p ∂xi i
Āpq = A.
∂xi ∂ x¯q j

The distinction between, contravariance and covariance does not

exist in relation to the cartesian tensor.

5.7 General Tensor of Any Order

We can now easily define a tensor of any order with any number of
contravariant and any number of covariant suffixes. For example, if we
have an entity representable by sets Ai , A¯pqr Relative to coordinate xi ,
jk

x¯j , such that

∂ x¯p ∂xi ∂xk i

Āpqr = A ,
∂xi ∂ x¯q ∂ x¯r jk

then we say that Aijk is a third-order tensor with one contravariant and
two covariant suffixes.
 5.8 Metric Tensor 73

5.8 Metric Tensor

Consider a rectangular cartesian system y i and any curvilinear system

xi , consider two points P and Q near each other with coordinate in the
two system.
y i , y i + dy i

xi , xi + dxi .

• Let ds denote the length PQ. We call ds the element of length or

the line element, we have

(ds)2 = dy 1 dy 1 + dy 2 dy 2 + dy 3 dy 3 = dy i dy.i (5.5)

Also
∂y i m
dy i = dx
∂xm

∂y i m ∂y i n
dy i dy i = dx dx
∂xm ∂xn

= gmn dxm dxn ,

where
∂y i ∂y i
gmn = .
∂xm ∂xn

Clearly,

gmn = gnm
74 General Tensor

Thus, we have

(ds)2 = gmn dxm dxn ,

where (ds)2 is invariant and dxm is an arbitrary vector, it follows by

applying quotient law twice that gmn is a covariant second-order tensor.
This is known as metric tensor for the space. It is easily seen that

∂y i ∂y 2 ∂y 3 ∂y i ∂y i ∂y i
∂xi ∂xi ∂xi ∂xi ∂x2 ∂x3
∂y i ∂y 2 ∂y 3 ∂y 2 ∂y 2 ∂y 2
gmn =
∂x2 ∂x2 ∂x2 ∂xi ∂x2 ∂x3
∂y i ∂y 2 ∂y 3 ∂y 3 ∂y 3 ∂3
∂x3 ∂x3 ∂x3 ∂xi ∂x2 ∂x3
2
∂y i
= 6= 0. (5.6)
∂xi

It is important to remember that the tensor gmn is essentially

a characteristic of the Euclidean space which will have different
components for different coordinate system to which the space may
be referred.

5.9 Associate Contravariant Metric Tensor

We denote by g mn the cofactor of gmn in the matrix [gmn ] , divided by

the determinant of the matrix. We have

g mn gnp = δpm .
 5.10 Associate Metric Tensor 75

As gnp is not an arbitrary covariant tensor, we cannot apply

Quotient law to show that g mn is a contravariant second-order tensor.
Let up be any arbitary contravariant vector, then

gnp up = vn ,

is also an arbitrary covariant vector. We have

g mn vn = g mn gnp up = δpm up = um .

Here vn is any arbitary convariant vector and um is also a vector,

hence by quotiant law g mn is a contravariant second-order tensor. The
tensor g mn may be called contravariant metric tensor.

5.10 Associate Metric Tensor

Raising and lowering of suffixes, we shall discuss the important

process of raising and lowering the suffixes of a tensor in order to
obtain. New tensor associated with those given. Let now ur be any
contravariant vector, then gmr ur is a covariant vector. We say that this
vector has been obtained on lowering the index and denote it again by
the symbol um . Thus, we write

um = gmr ur .

Similarly, the contravariant vector

v m = gmr ur
76 General Tensor

is said to be obtained by raising the index. The process of lowering

and raising of suffixes can be applied in a tensor of every type. For
example, consider the tensor

uijk pq

and suppose that we wish to lower j. This is given on inner

multiplication with gmj . Thus, the requried tensor is

gmj uijk pq.

Two tensors are said to be Associated. If either is obtained from

the other by any combination of the process of raising and lowering
the suffixes. It is important to notice that the sets of components of
two associated vector Relative to any cartesian rectangular system
of coordinates are the same. This if un be the components of a
contravariant vectors relative to a cartesian rectangular system, the
components of the associated covariant vector relatively to the same
system are

δmn un = um

Thus, the sets of components are the same.

5.11 Christoffel Symbols of the First and Second -Kind

Consider any Curvilinear system of coordinate and any curve so that

the coordinate xi of any point on the curve is given as a function of t.
 5.11 Christoffel Symbols of the First and Second -Kind 77

If ui and uj be the components of a vector at any point of the curve

Relative to the two systems of axes, we have

i ∂xi m
u = u . (5.7)
∂xm

Here, ui , um , xi , xm are all functions of t. Suppose that the set of

vectors associated with different points of the curve are equal and
parallel so that ui are the constant.
Differentiating Eq.(5.7) with respect to t, we get

du ∂ 2 xi dxn m ∂xi dum

0= = u + .
dt ∂xm ∂xn dt ∂xm dt
∂x i
On inner multiplication with g rp ∂xp , we get

dur ∂ 2 xi ∂xi dxn m

+ g rp m n p u = 0. (5.8)
dt ∂x ∂x ∂x dt

Here, the summation for i=1,2,3 is implied even through the

repeated suffix of i at the same level. We have

∂xi ∂xi dum dum r du

r
g rp = g r
gmp = δm .
∂xp ∂xm dt dt dt

Eq.(5.8) gives the condition to be satisfied by the set of

equal and parallel vectors uj along a given curve. We saw
now in Eq.(5.1) to an important information by introducing
what are known as christoffel symbols of the first and second
kind. In this connection, we consider the following expression
in Eq.(5.1)
78 General Tensor

∂ 2 xi ∂xi
.
∂xm ∂xn ∂xp

Now, we have

∂xi ∂xi
gmn =
∂xm ∂xn

∂gmn ∂ 2 xi ∂xi ∂ 2 xi ∂xi

= + . (5.9)
∂xp ∂xm ∂xp ∂xn ∂xn ∂xp ∂xm

Interchanging m, n, p cyclically in Eq.(5.9), we get

∂g ∂ 2 xi ∂xi ∂ 2 xi ∂xi
= + (5.10)
∂xm ∂xn ∂xm ∂xp ∂xp ∂xm ∂xn

∂gpm ∂ 2 xi ∂xi ∂ 2 xi ∂xi

= + . (5.11)
∂xn ∂xp ∂xn ∂xm ∂xm ∂xn ∂xp

Equations (5.9),(5.10) and (5.11), we get

∂gnp ∂gpm ∂gmn ∂ 2 xi ∂xi

+ − = 2 . (5.12)
∂xm ∂xn ∂x ∂xm ∂xn ∂xp

and we write

∂gnp ∂gpm ∂gmn

[mn, p] = + − . (5.13)
∂xm ∂xn ∂x

Thus, with the help of Eq.(5.12) and Eq.(5.13), we re-write

Eq.(5.1) as

dur dxn
+ g rp [mn, p] um = 0.
dt dt
 5.12 Covariant Derivative of a Covariant Vector 79

Again writing
" #
rp r
g [mn, p] =
mn

we obtained
" #
r r n
du m dx
+ u =0 (5.14)
dt mn dt

as the required form of condition, if ur is a set of equal" and

#
r
parallel vectors along the curve. The symbols [mn, p] and
mn
are known as christoffel symbols of the first and second kind
respectively. Both symbols are easily seen to be symmetrical
in m, n.

5.12 Covariant Derivative of a Covariant Vector

Let X be any covariant vector field, we take a parallel vector field ur

so that ur satisfied the equation
" #
dur r
+ um = 0. (5.15)
dxm mn

∂(xr ur )
Then xr ur is a scalar, is a covariant vector, now
∂xi

∂(xr ur ) ∂ur r ∂xr

= x r + u . (5.16)
∂xi ∂xi ∂xi
80 General Tensor

From Eq.(5.15)
" #
∂ur r
=− um . (5.17)
∂xi mn

From Eq. (5.16) and Eq.(5.17), we have

" #
∂(xr ur ) r ∂xr
i
= xr um + ur i
∂x mn ∂x
" #
∂xm r
= − xr um .
∂xi ms

∂(xr ur )
Here, um is an arbitrary contravariant vector and is a
∂xi
covariant vector. Thus
" #
∂xm r
− xr
∂xi ms

is a covariant second-order tensor called the covariant derivative of xr ,

we write
" #
∂xm r
xmrs = − xr .
∂xi ms

5.13 Covariant Derivative of a Contravariant Vector

If xr any contravariant vector, then it can be shown that

" #
∂xr r
+ xm
∂xi ms
 5.14 Exercises 81

is a mixed second-order tensor. It is called the covariant derivative of

xr and we write
" #
r r
∂x
xrms = + xm .
∂xs ms

This can be proved by taking a contravariant vector field xr along

a curve C and parallel field of covariant vector ur along C and
considering the invariance of xr ur .
Thus, we have the two following important results:
" #
∂xm r
Xmrs = − Xrs (5.18)
∂xs ms
" #
r ∂xr r
Xms = + X m. (5.19)
∂xs ms

5.14 Exercises

(1) Find the components of the Christoffell symbols for the

metric

ds2 = r2 dθ2 + r2 sin2 θdφ2

(2) Compute all the Christoffell symbols for a space whose line
element is given by

ds2 = −eλ dr2 − r2 dθ2 − r2 sin2 θdφ2 + eµ dt2

82 General Tensor

(3) The length ds of a line element in a two-dimensional surface θ, φ

is given by

ds2 = R2 dθ2 + R2 sin2 θdφ2

where R is constant. Find all the components of the metric

tensor gµv and the christoffel symbols of first kind for this
surface.
(4) Prove that

gij,k = [ik, l] + [jk, i].

(5) Compute the christoffel symbols of the first and the second kind
in a plane in terms of polar coordinate r, θ.
(6) If Aijk is a skew-symmetric tensor, show that
( ) ( ) ( )
i l l
Aijk = Aijk = Aijk .
ij jk ik

References

[1] Harold Jeeffreys (1931), Cartesian Tensor, PP(43-57), Combridge

University Press (New York).
[2] David C. Kay, Theory and Problem of Tensor Calculus, PP(1-3),
McGraw Hill, Washinton, D.C.
[3] Shanti Narayan (1961), Cartesian Tensor, PP(129-154), S. Chand,
New Delhi.
 References 83

[4] Barry Spain (1960), Tensor Calculus, PP(27-37), Dover

Publication, Newyork.
[5] Zefer Ahson(2000). Tensor Analysis with Application, Anamaya
Publisher, New Delhi.
[6] U.C. De (2010) Tensor Calculus. PP(62-135), Narosa Publishing
House, New Delhi.
 6
Tensor in Relativity

6.1 Special Theory of Relativity

In classical mechanics, the position of a point in a space at which

any point P can be determine by its three space coordinate x1 , x2 , x3
referred to some rectangular cartesian system. Also an observer can
measured the time t at which the event takes place by mean of clock,
space and time cordinate comprised of the four numbers x1 , x2 ,
x3 and t.
Einstein examined the concept of simultaneity and comes to the
conclusion that simultaneous events at different points has no meaning
without qualification. Einstein arrived at the special theory of relativity
which he based on the two principles.
(1) It is impossible to detect the unaccelerated translating motion of
a system through space.
(2) The velocity c of a ray of light is a constant, which does not
depend on the relative velocity of its source and the observer.
Let us assume two systems S ans S , , which coincide at the

85
86 Tensor in Relativity

time t=0, such that S , moves with constant velocity v along

x axis of the system. Then the Lorentz transformation, which
can be deduced from the two principles of special relativity,
connects the space coordinates and the time of both system by the
equation

 vx, 
x1, = β(x1 − vt), x,2 = x2 , x,3 = x3 , t. = 1 − 2
c
v 2 12
where β = (1 − c2
) .
We can verify that

− (dx.1 )2 − (dx.2 )2 − (dx.3 )2 + c2 (dt. )2

= (dx1 )2 − (dx2 )2 − (dx3 )2 + c2 (dt. )2 . (6.1)

The invariance of this equation with respect to lorentz

transformation suggests that the Minkowski space defined by the
metric

dσ 2 = −(dx1 )2 − (dx2 )2 − (dx3 )2 + c2 (dx4 )2 (6.2)

where we have written x4 = t is approximate for the geometrical

discussion of special relativity. We denote the line element of this
four dimensional space by dσ is the physical distance between two
near points. The velocity u of a particle, which is at the point
dxi
xi has the components ui = dt
referred to the system S. It
follows that
 6.1 Special Theory of Relativity 87

dxi 1 v2 1
= (1 − 2 )− 2 . (6.3)
dt c c

The four-dimensional Minkowski momentum vector is defined by

α
mo c dx
dσ
, where mo is a constant. The special theory identifies the
4
fourth components mo c dx
dσ
with the mass m of the moving particle.
We have

v2 − 1
m = m0 (1 − ) 2 (6.4)
c2

the constant mo is the mass when u=0 and it is called the

rest mass of the particle. The mass m, which clearly increase
the velocity is called the relativistic mass of the particle. The
components

dxi dxi dt dxi

mo c = mo c =m
dσ dt dσ dt

are the generalization of the newtonion momentum vector. We

defined the fourth dimensional Minikowaski momentum
vector F α by

d 2 xα
F α = mo c2
dσ 2

d dxα
= c2 (mo )
dσ dσ

v2 − 1 d dxα
= (1 − ) 2 (mo ). (6.5)
c2 dt dt
88 Tensor in Relativity

The Newtonian force vector is

d dxi v2 1
xi = (mo ), F i = (1 − 2 )− 2 xi .
dt dt c

The motion of a particle which moves under the action of some

force system can be represented in Minkowski space by a curve called
the world line of the particle. If no force acts on the particle, we have
d2 xα
= 0, thus the world line of a free particle is a geodesic of the
dσ 2
minkowski space.
The velocity of a light ray is the constant c. In order to
discuss the mechanics of a continuous medium, we introduce
the symmetrical four dimensional energy momentum Tensor T αβ
defined by

T ij = T ji = ρui uj − E ij ; T i4 = T 4i = ρui ; T 44 = ρ

where ρ is the density and E ij is the cartesian stress tensor. If we

change to spherical polar coordinate r, θ and φ the metric of minkowski
space becomes

dσ 2 = −dr2 − r2 dθ2 − r2 sin2 θdφ2 + c2 dt2 (6.6)

6.2 Four-Vectors in Relativity

Four-vectors leading to two different types contravariant and

covariant four vector. Let us consider a four-vector A. It has
 6.2 Four-Vectors in Relativity 89

four component, one time Al and three components Ax , Ay , Az .

In the four-dimensional real Minikowski space, because of the
introduction of the metric tensor gµv , complications arise. There is
two types of four-vectors, a contravariant four vector denoted by a
suprescript

Aµ ; Av , A1 , A2 , A3 = At , Ax , Ay , Az .

and a covariant four vector denoted by a subscript

Aµ ; Ao , A1 , A2 , A3 = At , −Ax , −Ay , −Az .

The metric tensor is a 4 × 4 diagonal matrix

 
1 0 0 0
 
 0 1 0 0 
gµv = . (6.7)
 
 0 0 −1 0 
 
0 0 0 −1

It can be written in conscise form as


0 µ 6= v





gµv = g µv = 1 µ=v=0 . (6.8)



 −1 µ = v = 1, 2, 3


Using the metric tensor, one can convert a contravariant four vector
into a covariant four vector and vice versa.
90 Tensor in Relativity
X
Aµ = gµv Av , µ = 0, 1, 2, 3. (6.9)
v
X
Aµ = g µv Av , µ = 0, 1, 2, 3. (6.10)
v

If we use Einstein’s convention that repeated index involves the

summation symbol that we used above. The scalar product of any two
four-vector A and B can be written as

A.B = Aµ Bµ = A0 B0 + A1 B1 + A2 B2 + A3 B3

= At Bt − Ax Bx − Ay By − Az Bz

At Bt − A.B. (6.11)

The differential operators in the four-dimensional space time xµ

transforms as the components of a covariant four-vector.

a axµ a
= (6.12)
axµ ax0 µ axµ

 
a a
aµ = = ∇ . (6.13)
axµ axµ

The corresponding contravariant four vector is

 
µ µv a
a = g av = ∇ . (6.14)
ax0

It is clear that gµv contains all the information about the geometry
of the space. In the case of Minkowski’s space time. If we confine the
 6.3 Maxwell’s Equations 91

special theory of relativity the metric tensor gµv plays a passive role but
it play an active role in general relativity since the space time geometry
is not fixed in advance and can be curved depending on the distribution
of matter. So in special relativity one can avoid the distinction between
the contravariant and covariant vectrors and define the scalar product
of four vectors by

A.B = Ao B0 − A.B = Ao Bo − Ax Bx − Ay By − Az Bz (6.15)

6.3 Maxwell’s Equations

The classical theory of electrodynamics according to Lorentz is

specified by the electric potential φ, which is scalar and the magnetic
potential Ai , which is vector. Electric field strength vector Ei and
the magnetic vector Hi are derived from there-potential by the
equation

1δA
Ei = −gradφ − (6.16)
cδt

Hi = curlAi . (6.17)

using electrostatic units, maxwell’s equations are

divEi = 4πρ (6.18)

divHi = 0 (6.19)

1δHi
curlEi + =0 (6.20)
cδt
92 Tensor in Relativity
1δEi 4πJi
curlHi − = (6.21)
cδt c

where Ji is the current density vector and ρ is the charge density.

In Minkowski space, let us form the four-dimension potential
vector φα and the four-dimension current density vector J α defined
respectively by

φα = (−A1 , −A2 , −A3 , cφ)

Jα = (−J1 , −J2 , −J3 , ρ)

with respect to a particular coordinate system. Next we introduce the

skew symmetric tensor

ηαβ = φαβ − φβα

δφα δφβ
= −
δxβ δxα

and we immediately calculate that its non-vanishing components in the

given coordinate system are

η23 = −η32 = H1 ; η31 = −η13 = H2 ; η12 = −η21 = H3 ;

η14 = −η41 = cE1 ; η24 = −η42 = cE2 ; η34 = −η43 = cE3 ;

The non-vanishing contravariant components η αβ may now be

obtained and are
 6.4 General Theory of Relativity 93

η 23 = −η 32 = H1 ; η 31 = −η 13 = H2 ; η 12 = −η 21 = H3 ;

−E1 24 E2 34 E3
η 14 = −η 41 = ; η = −η 42 = ; η = −η 43 = ;
c c c

We now write Maxwell’s equation in terms of η and J and the

results are readily verified to be respectively

∂η 41 ∂η 42 ∂η 43 4π 4
1
+ 2
+ 3
= J
∂x ∂x ∂x c

∂η23 ∂η31 ∂η12

+ + =0
∂x1 ∂x2 ∂x3

∂η1j ∂ηj4 ∂η4j

+ + =0
∂x4 ∂xi ∂xj

∂η i1 ∂η i2 ∂η i3 ∂η i4 4π i
1
+ 2
+ 3
+ 4
= J.
∂x ∂x ∂x ∂x c

The first and last of these equations combine together into

the form

4π α
ηβαβ = J (6.22)
c

ηζβ,γ + ηβγ,α + ηγα,β = 0. (6.23)

we have accordingly written, maxwell’s equation in tensor form in

minkowski space. Thus, the are invariant under the Lorentz group of
transformations.
94 Tensor in Relativity

6.4 General Theory of Relativity

General theory of relativity, which was developed by Einstein in order

to discuss gravitation, he postulated the principle of covariance, which
shows that the law of physics must be independent of the space
time coordinates. This swept away the privileged role of the Lorentz
transformation. As a result, Minkowski space was replaced by the
Riemannian V4 with the general metric

dσ 2 = gαβ dxα dxβ . (6.24)

Einstein also introduced the principle of equivalence, which in

essence states that the fundamental tensor gαβ can be chosen to account
for the presence of a gravitational field. That is gαβ depends on the
distribution of matter and energy in physical space. Matter and energy
can be specified by the energy momentum tensor T αβ , which in the
special theory satisfied the equation Tααβ = F β . Einstein’s tensor
defined by

1
Gαβ = g αγ Rβγ − Rδβα (6.25)
2

Satisfies the equation Gαβα = 0. This equation of motion requires

α
Tβα = 0, but very remarkably Gαβα = 0. is an identity in Riemannian
geometry. This led Einstein to propose the relation

kTβα + Gαβ = 0. (6.26)

 6.5 Spherically Symmetrical Metric 95

In effect, these equations form the link between the physical energy
momentum tensor Tβα and the geometerical tensor Gαβ of the V4 of
general relativity. In the special theory , the world line of free particles
and of light rays are respectively the geodesics and the null geodesics
of the Minkowski space. The principle of equivalance demands that all
particles be regarded as free particle, when gravitation is the only force
under consideration. Then it follows from the principle of covariance
that the world line of a particle under the action of gravitational forces
is geodesic of the V4 with the metric.

6.5 Spherically Symmetrical Metric

General relativity discusses several important problems in which the

coordinate systems r, θ, φ and t are such that the metric takes the form

dσ 2 = −eλ dr2 − r2 dθ2 − r2 sin2 θd2 φ + c2 ev d2 t (6.27)

where λ and v are functions of r. A metric of this type is said to be

spherically symmetrical. It is a generalization of the special relativity
metric is expressed in spherical polars. The coefficients of dr2 and dt2
have been selected as exponential in order to ensure that the signature
of dσ 2 is -2. Let us write x1 = r, x2 = θ, x3 = φ and x4 = ct. The non
zero components of the fundamental tensor are

g11 = −eλ , g22 = −r2 , g33 = −r2 sin2 θ

and g44 = ev .
96 Tensor in Relativity

The determinant g becomes

g = −r4 sin2 θeλtv

and hence the non zero components of the conjugate symmetric

fundamental tensor are

1 33 1
g 11 = −e−λ , g 22 = − 2
, g =− 2 2
r r sin θ

and g 44 = e−v .
We obtained the components of the Einstein tensor for the
spherically metric
 
1 1 1 0
G11 = − 2 + e−λ + v , (6.28)
r r2 r
 
−λ 1 1 0 0 1 00 1 0 1 02
G22 = G33 =e − 2r λ − λ v + v + v + v , (6.29)
4 2 2r 4
 
1 1 1 0
G44 = − 2 + e−λ − λ , (6.30)
r r2 r

Gαβ = 0 f or α 6= β (6.31)

6.6 Planetary Motion

Let us investigate the motion of a planet in the gravitational field

of the sun. The sun will be selected as a gravitational particle and
the planet as a free particle whose mass is so small that it does
 6.7 Exercises 97

not affect the metric differential equation. for the planet of our solar
system is
d2 u m 3mu2
+ u = + . (6.32)
dφ2 c2 h2 c2

m 3mu2
The term c2 h2
is much larger than c2
, but when, we neglected
this latter term, we obtained Newton’s equation for the motion of a
planet. Thus the first approximation to the solution of Eq.(6.32) is the
Newtonian solution is

m
u= (1 + ecos(φ − ),
c2 h2

where e is the eccentricity of the elliptic orbit and  is the longitude

of perihelion. A second approximation to the solution can then be
obtained in the form

m
u= (1 + ecos(φ −  − 4),
c2 h2

where
3m2 φ
∆ =
c4 h2

This means that the major axes of the elliptic orbit is slowly
rotating about its focus (sun).

6.7 Exercises

(1) Show that (a) E.B and (b) E 2 − B 2 are invariant under Lorentz
transformation.
98 Tensor in Relativity

(2) Prove that the curvature invariant of Einstein’s universe is R =

6/R”2 .

References

[1] J.A. Schouten, Der Ricci-Kalkul, (1954).

[2] Barry Spain, Tensor Calculus, (1952).
[3] H. Weyl, Space Time Matter, (1922).
[4] I.S. Sakolinkoff, Mathmatical Theory of Elasticity, (1946).
[5] A. Einstein, Theory of Relativity, (1958).
[6] W. Pauli, Theory of Relativity, (1958).
[7] I.S. Sakolinkof, Tensor Analysis, (1967).
[8] A.W. Joshi, Matrices and Tensor in Physics, (1975).
 7
Geodesics and Its Coordinate

7.1 Families of Curves

Family of curves given by y = f (x, t). For every t, we obtain a

particular curve, if t is fixed and x varies using the symbol d to
represent the differential. If we keep x is fixed and that u varies. Then
the symbol δ respresent the differentials and this differential will be a
variation. The commutative law of partial derivatives

φxt = φtx , (7.1)

now

dδφ = d(φt δt) = (dδt)δt = (φtx dx)δt = φtx dxδt

δdφ = δ(φx dx) = (δφx )dx = (φxt δt)dx = φxt δtdx.

Hence,
dδφ = δdφ.

99
100 Geodesics and Its Coordinate

7.2 Euler’s Form

Let A(x1 , y1 ) and B(x2 , y2 ) are two points. This is a particular curve
for a family, if the curve

y = y(x, t) f or t = 0.

If all the curves have the same extremities A and B, then we have
δy = 0.If δt is the small values of neighbouring curves, then by
Leibnitz’s rule

Zx2
(δφ)dx = 0. (7.2)
x1

now

0 0
δφ = φx δx + φy δy + φy0 δy = φy δy + φy0 δy .

since δx = 0 and δy = 0 at A and B

Zx2
0
(φy δy + φy0 δy )δx = 0.
x1

Euler’s condition

d
= φy − φ 0.
dx y
 7.3 Geodesics 101

7.3 Geodesics

n-dimensional Riemannian space Vn , considers a curve given by xi =

xi (t). The length of the arc is
Z
s= φdt (7.3)

where

q
φ= gij x0 i y 0 j .

This curve is called a geodesic for each i and we can find the Euler’s
equation
 
aφ d aφ
− = 0. (7.4)
ax i dt ax0 i

Now
aφ 1 0 0

i
= agjk
x jx k
ax 2s axi

aφ 1 0
0i = gik x k .
ax s

Therefore,
 
d aφ s 0 1 agik j 0 0 k 1 0
=− gik x k + x x + gik x k .
dt ax0 i s 2 s ax j s
102 Geodesics and Its Coordinate

Euler’s equation becomes

1 agjk 0 j 0 k s 0 1 agik 0 j 0 k 1 00

i
x x + 2 gik x k − j
x x − gik x k = 0. (7.5)
2s ax s s ax s

If the length of the arc s as parameter t then

0 00
s =1 s = 0.

The above equation becomes

1 agjk 0 j 0 k agik 0 j 0 k 00

i
x x − j
x x − gik x k = 0. (7.6)
2 ax ax

or

0 0
gik x k + [jk, i] x j xk = 0, (7.7)

where [jk, i] is the christoffel symbol of first kind.

Now transversing with gil , we get
( )
l 0 0 00
x k x j + x l = 0, (7.8)
jk
( )
l
where is the Christoffel symbol of second kind. Now we
jk
write, equation of geodesic
( )
2 l
dx l dxi dxk
+ (7.9)
ds2 jk ds ds
 7.4 Geodesic Form of the Line Elements 103

Again the arc lenght if p of given as

Zt
dp(t)
s(t) = | |dt = c(t − to ). (7.10)
dt
to

It is proportional to the parameter of the geodesic if c=1, the p is

normalized. Setting p(t) = (x1 (t), x2 (t)....., xn (t). By application of
the definition, we get

!
dxi dxj d2 xk
 
D dp X a
= Γkij + =0 (7.11)
dt dt ij
dt dt dt2 axk

which is satisfied, if

d2 xk X k dxi dxj
+ Γij =0 (7.12)
dt2 ij
dt dt

for all k. This ordinary, nonlinear second order differential equation is

called the Geodesic equation.

7.4 Geodesic Form of the Line Elements

If φ(p) be a scalar function so that ∇φ = 0, φ(x) = 0,

0
and if u is the tangent vector along such a geodesic, then
we have

0 0
u = (dx , 0, 0, 0.......0).
104 Geodesics and Its Coordinate
0
The length of vector u is given by

u2 = gij ui uj

0
(dx )2 = gii dxi dxi , g ii = 1.

0 0
If v is the tangent vector to the hypersurface, x = 0, then
we have
0
v = (0, dx2 , dx3 , ........dxn ).

0 0
Also the vectors u and v are orthogonal vectors, then
0 0
gij u v = 0

0 0 0
gij u v = 0 if (u = 0 f or i = 1, 2, 3.......n)

0 0
gij v = 0 if (u 6= 0)

gij = 0 if (v j 6= 0 f or j = 2, 3, ..n)

gij = 0 f or j = 2, 3, ..n).

Now, the differential equation of geodesic is

( )
d 2 xi i dxj dxk
+ = 0. (7.13)
ds2 jk ds ds

If t is the unit tangent vector to a geodesic at any point, then

0 0
t =1 and t = 0 f or i 6= 1
 7.5 Geodesic Coordinate 105
0
i dxi dxi dxi dxi
t = = =1 and = 0 f or i 6= 1,
ds dxi ds ds
0
!
d 2 xi d dxi
2
= = 0 f or i = 1, 2, .......n.
ds ds dxi

Using Eq.(7.13), we get

" #
i
tj tk = 0
11

" #
i h i
=0 g ij 11 j = 0.
11

The line element namely

ds2 = gij dxi dxj

= gi1 dxi dxi + gjk dxj dxk

= (dxi )2 + gjk dxj dxk . (7.14)

Then ds2 is called the geodesic form of line element.

7.5 Geodesic Coordinate

agij
The components of metric tensors gij are constant, and then axk
=
0 for every i, j, k. In Riemannian space, there exist a coordinate
system with respect to which the christoffel symbol vanishes at given
point, and system is called a geodesic coordinate system and the
106 Geodesics and Its Coordinate

point is called the pole of the given system. suppose a system S

whose curvilinear coordinates are u1 , u2 , and a point P (u1 , u2 ) on
S. If v α , α = 1, 2 are the coordinates of S, then the transformation
is given by

uα = uα (v 1 , v 2 ), α = 1, 2. (7.15)

The second derivative formula yields the relation

( ) ( )
a2 uα α auβ auγ γ auα
+ = (7.16)
av λ v µ β γ av λ auµ λ µ av γ

( ) ( )
α γ
where and are the Christoffel symbols in u
β γ λ µ
( )
γ
coordinates systems and v coordinate system. In Eq.(6.15),
λ µ
vanishes at P, then for the particular point
( )
a2 uα α auβ auγ
+ = 0. (7.17)
av λ v µ β γ av λ auµ

The solution of Eq.(6.17) is transformation of Eq.(6.15) to a

coordinate system v α . The second degree polynomial
( )
α
uα = uαp + v α − vλvµ. (7.18)
γ µ
 7.6 Exercise 107

On differentation of (6.18), we get

( )
auα α
µ
= δµα − vλ
av λµ

( )
a2 uα α
=−
av λ v µ β γ
p

( )
auα a2 uα
   
α
= δµα and =− . (7.19)
av µ av λ v µ p γ µ
p

Using the formula

( ) ( )
α α
uαλµ p δγα


= − (7.20)
γµ β γ
p

We see that Eq.(6.19) satisfied Eq.(6.17) at point P and Eq.(6.18)

at P is the new coordinate given by v α = 0 are geodesic
coordinate.

7.6 Exercise
0
(1) Find the geodesics of a sphere of radius u determined by the
equation x1 = u1 sinu2 cosu3 , x2 = u1 sinu2 sinu3 and x3 =
u1 cosu2.
(2) Find the geodesics of a cylindrical coordinate of x1 = u1 cosu2 ,
x2 = u1 sinu2 and x3 = u3 .
108 Geodesics and Its Coordinate

(3) Find the geodesic equation for the metric

0 0
ds2 = (dx )2 + ((x2 )2 − (x )2 )(dx2 )2 .

References

[1] Harold Jeeffreys (1931), Cartesian Tensor, PP(1-66), Combridge

University Press (New York).
[2] David C. Kay, Theory and Problem of Tensor Calculus, PP(1-3),
McGraw Hill, Washinton, D.C.
[3] Shanti Narayan (1961), Cartesian Tensor, PP(37-51), S. Chand,
New Delhi.
[4] Barry Spain (1960), Tensor Calculus, PP(1-55), Dover Publication,
Newyork.
[5] Zefer Ahson (2000). Tensor Analysis with Application, Anamaya
Publisher, New Delhi.
 Index

A D
Algebraic Operations on Divergence of Vector Point
Tensors, 27 Function, 34
Alternate Tensor, 36 Dummy suffixes, 27
Acceleration, 53
E
Associate Contravariant Metric
Tensor, 87 Euler’s Form, 100
Associate Metric Tensor, 88 Einstein’s convention, 102
F
C
First-Order Tensors, 12
Cartesian Tensor, 7
Force, 40
Contraction Theorem, 9
Four-Vectors in Relativity, 102
Curvilinear Coordinates, 68, 81
Families of Curves, 112
Coordinate Transformation
Equation, 68, 81 G
Contravariant and Covariant Gradient of Tensor Field, 32
Tensor, 69 General Tensor, 67
Christoffel Symbols of the First General Theory of Relativity, 94
and Second Kind, 76 Geodesic Coordinate, 105
Covariant Derivative of a
I
Covariant Vector , 79
Isotropic Tenso, 51
Covariant Derivative of a
Inner Product, 42
Contravariant Vector, 80
Inner multiplication, 90
Curl of Vector Point
Function, 47 K
Covariant Vector, 84 Kronecker Tensor, 24

109
110 Index

Kinematics of Single Particle, 39

Summation Convention, 3
Kinetic Energy and Potential Symmetric and Skew-Symmetric
Energy, 41 Tensor, 21, 34
Strain Tensor at Any Point, 46
L
Stress Tensor at Any Point, 49
Length and Angle between Two
Simple Stress, 50
Vector, 57
Shearing Stress, 50
Line Element, 94, 99, 118
Spherically Symmetrical
M Metric, 95
Maxwell’s Equations, 91 Sum and Difference of
Metric Tensor, 85, 88 Tensors, 28
Mixed Second-Order Tensor, 72 T
Minkowski Space, 106, 107 Tensor Fields, 32
N Tensorial Formulation of Gauss’s
Null Geodesics, 107 Theorem, 35
Tensorial Formulation of Stoke’s
P Theorem, 35
Product of Tensor, 16, 28 Tensor in Physics, 39
Product of Two Matrices, 41 Tensor Formulation of Analytical
Product of Inertia, 59 Solid Geometry, 61
Tensor in Relativity, 85
Q
Quotient Law of Tensors, V
17, 61, 63 Vector Product, 31, 56
Vector as Directed Line
R
Segments, 55
Riemannian geometry, 107
Riemannian space, 118 W
Work Function and Potential
S
Energy, 41, 54
Symbols Multi-Suffix, 2
Scalar Triple Product, 73 Z
Scalar Product, 103 Zero Order, 126
 About the Authors

This book presents basic introduction of tensors with an emphasis on

the understanding of the fundamentals. It develops an appreciation of
the tensor application in both undergraduate and post graduate students
of Physics and Mathematics.

Bipin Singh Koranga is a graduate from Kumaun University,

Nainital. He has been with the Theoretical Physics Group, IIT Bombay
since 2006 and received the Ph.D. degree in Physics (Neutrino Masses
and Mixings) from the Indian Institute of Technology Bombay in
2007. He has been teaching basic courses in Physics and Mathematical
Physics at the graduate level for the last 12 years. His research interests
include the origin of universe, Physics beyond the standard model,
theoretical nuclear Physics, quantum mechanical neutrino oscillation
and few topics related to astrology. He has published over 42 scientific
papers in various International Journals. His present research interest
includes the neutrino mass models and related phenomenology.

Sanjay Kumar Padaliya is presently Head, Department of

Mathematics, S.G.R.R. (P.G) College, Dehradun. He received his
Ph.D. degree in Mathematics (Fixed Point Theory) from Kumaun
University, Nainital. He has been teaching basic courses in

111
112 About the Authors

Mathematics at graduate and postgraduate level for the last 20

years. His present research interest includes the Fixed Point Theory
and Fuzzy Analysis. He has published over 25 scientific papers in
various International Journals of repute and also presented his works
at National and International conferences. Dr. Padaliya supervised
05 research scholars for Ph.D. He is also a life member of
Indian Mathematical Society, Ramanujan Mathematical Society and
International Academy of Physical Sciences.