Tensor Analysis

1.1 Introduction

Physical laws: If they are to be valid, must be independently of any

particular coordinate system used to describe mathematically. A study
of the consequence of this requirements leads to tensor analysis which
is of great use in general relativity theory, differential geometry,
elasticity hydrodynamics, electromagnetic theory, and numerous other
fields of science and engineering.

1.2 Curvilinear coordinate:

e2

u2

e1

u3

u1
e3
Fig: Curvilinear system
The position vector of any point P is described by its corresponding
coordinate values. The coordinate system which is generated by three
surfaces say u1 = constant, u2 = constant, u3 = constant as shown in
figure above is called curvilinear coordinate system, where u 1, u2 and
u3 represents curve. The value of P(u 1, u2, u3) in such system is known
as curvilinear coordinate of point P. Let the transformation of
following type exists,
1
 x = x (u1, u2, u3)
y = y (u1, u2, u3)
z = z (u1, u2, u3)
also,
u1 = u1(x, y, z)
u2 = u2(x, y, z)
u3 = u3(x, y, z)
Now, arbitrary position vector in curvilinear system is,
r = r (u1, u2, u3)
dr - du1 + du2 + du3 …(1)
dr = n dui
The unit tangent vectors on the surfaces or curves u 1, u2, u3
respectively.
e1 = , e2 = , e3 =
Then, ds2 = dr . dr
= 3dui . 3 duj
= SS duiduj
= SS hiei. hjej dui duj …(A)
Where hI or hj is scale factor or metrical coefficient.
ds2 = S gij dui duj
Where,
gij = hi ei.hjej is the metric coefficient

For orthogonal curvilinear system

ei. ej = 1 i=j
0 ij
Then,
ds2 = g11(du1)2 + g22 (du2)2 + g33(du3)2
 ds2 = h (du1)2 + h (du2)2 + h (du3)2
Hence,
h1 = , h2 = , h3 =

2
Also, from equation (1)
dr = h1e1 du1 + h2e2 du2 + h3e3 du3…(2)
The matrix form gij can be written as,
gij =
In determinant form,
In orthogonal system
g = |gij| =
 h1 =, h2 = , h3 =
In spherical polar coordinate,
h1 = = 1, h2 = = r, h3 = = rsin
g=
In cylindrical system,
h1 = 1, h2 = , h3 = 1
g=
Volume element is
dV = ds1ds2 ds3
ds2 = h12du12 + h22 du22 + h32du32
ds2= ds12 + ds22 + ds32
 ds1 = h1du1, ds2 = h2du2, ds3 = h3du3
 dv = h1h2h3du1du2 du3
d = du1du2du3
The Jacobian is,
J = . ×
= h1e1. (h2e2 ×h3e3)
= h1 e1 . h2 h3 e1
= h1h2h3
 g = h12 h22 h32
 = h1h2h3
Also, from equation (A)
h1e1 = , h2 e2 = , h3e3 =

3
1.3 Gradient of scalar function in orthogonal
curvilinear coordinate system

Here, if  be the scalar function, then  is expressed in its

component form as,
 = e1 + e2 = + e3
where s1, s2, s3 are components of arc length s along u1 ¸u2, u3
cures.
Since
ds1 = h1du1, ds2= h2du2, sd3 = h3du3
 = + +
If  = u1 = constant surface
u1 =
Similarly,
u2 = , u3 =
Now,
. × = h1e1. (h2e2 × h2e3)
= h1. h2h3e1
= h1h2h3
And,
u1. (u2.u3)
= . ×
=
Hence, therefore,
[u1, u2 u3] = 1
Hence, they are reciprocal systems.

1.4 Coordinate Transformations

Let (x1, x2 …..xN) and (x',x2 …..xN) be coordinates of a point in two
frame of reference. Suppose there exists N independent relations
between the coordinates of the two systems having the form.
x' = x' (x', x2 …xN)
4
 x2 = x2 (x1, x2 ….xN)
| | |
| | |
xN = xN (x1, x2….xN)
Which we can indicate by,
xK = xK (x1, x2…xN), k = 1, 2, ….N  (1)
where it is supposed that the functions involved are single valued,
continuous and have continuous derivatives.
Then conversely, to each set of coordinates (x1,x2, ….xN), there
will correspond a unique set (x1, x2…..xN) given by,
xk = xk (x1,x2, …xN) …..(2)
K = 1, 2, …….N
The relations (2) or (3) define a transformation of coordinate from one
frame of reference to another.

1.49 Contravarient Tensor and Covariant Tensor

Suppose N quantities A1, A2 … An in a coordinate system (x1, x2, …
xN) are related to N other quantities A1, A2…. AN in another
coordinate system (x1,x2…..xN) by the transformation equations.
AP = N Aq
or, AP =  Aq
Which by the conventions adopted can simply be written as,
AP =
Then, they are called components of a contravarient vector of
contravarient tensor of the first rank or first order.
Covarient tensor
On other hand, suppose N quantities A 1, A2, …AN in a coordinate
system (x1, x2…xN) are related to N other quantities A1, A2… AN in
another system (x1,x2…..xN) by transformation equations,
AP = N Aq P = 1, 2, ….N
q
or, AP = A
Then they are called components of a covariant vector or covariant
tensor of first rank or first order.
# Note for Benefit

5
 Instead of speaking of a tensor whose components are A P or AP;
we shall often refer simply to the tensor AP or Ap. No confusion
should arise from this.

1.5 Mixed Tensors: TU 2054

Suppose N2 quantities Aqs in a coordinate system (x 1, x2, …..xN) are
related to N2 other quantities Apr in another coordinate system (x1,
x2…..AN) by the transformation equations,
Apr = N N Aqs
P,r = 1, 2, ….. N
or, A Pr = Aqs
They are called contravariant components of tensor of second rank or
rank two.
The N2 quantities Aqs are called covariant components of a tensor of
the second tank if,
Apr = Aqs
Similarly, the N2 quantities Asq are called components of a mixed
tensor of second rank if,
ArP = Aqs

1.6 Kronekar delta TU2056

The kronekar delta  = = 1 for i = j
= 0 for I  j
It is injvariant,
 = d = is a mixed tensor of rank 2.
Verification
Here, we have,
d==
= = 
So, Kronetar delta is mixed tensor of rank 2.

6
1.7 Metric tensor
The differential of arc length ds in rectangular coordinate (x, y, z). By
transformation of general curvilinear coordinates, this becomes.
ds2 = S gij dui duj
Also, we can write,
ds2 = gij dxi dxj
We know that ds2 is invariant i.e.,
ds2= gij dxidxj
= gpq dp dxq
Now,
gpq dxidxj = gij dxi dxj
gij = gpq
 gpq = gij
It is convarient tensor of rank 2.
gij can be symmetric;
Let,
ds2 = gjk dxj dxk …………..(1)
ds2 = gkj dxj dxk ……….(2)
Adding (1) and (2)
2ds2 = (gjk + gkj) dxj dxk
ds2 = (gjk + gkj) dxj dxk
= g dxj dxk
Where,
g = (gjk + gkj) = g is symmetric
Replacing ,  by J and K,

Hence, metric tensor is symmetric tensor.

1.8 Symmetric and skew-symmetric tensor

A tensor Aij is said to be symmetric if,
Aij = Aji
and skew-symmetric if,
Aij = - Aji

7
Q. A tensor is symmetric in one coordinate system is symmetric in all
coordinate system. TU 2055
i.e. prove Apq = Aqp if Apq = Aqp
Proof:
Let Apq be a tensor, then we have,
Apq = Aqp
Apq = Aij
= = Aji
 Apq = Aqp Hence the statement
Also,
Apq = Aij
=
 Apq = - Aqp
Q.  is mixed tensor of second rank. TU 2057
We have,

From, the definition of given coordinate (p, q) can be written in
terms of partial differentiation,
(p, q) =
Let d(k, ) be the transformed part of (p, q) in another coordinate
system
d(k, ) =
d(k, ) =
d(k, ) = (p, q)
 d= d
Hence,  is mixed tensor of second rank.
Q. Every tensor can be expressed as the sum of symmetric and
skewsymmetric tensor.
Let us consider the covariant tensor Apq,
Apq = (Apq + Aqp) + (Apq + Aqp)
 Apq = Bpq + Cpq …………..(1)
Here,
Bpq = (Apq + Aqp)

8
 = (Aqp + Apq)
 Bpq = Bqp
Hence, Bpq is symmetric tensor
Cpq = (Apq - Aqp)
= (Aqp - Apq)
 Cpq = - Cap
 Cpq is skew-symmetric tensor
Hence, therefore proved.

1.9 Contraction: T.U.2057

If one covariant and one contravarient index of a tensor are set equal,
then we get the tensor of rank less than two of original tensor. This
process sis called contraction.
If A is a tensor of rank 5
If q = s, then, B is resulting tensor, is tensor of rank 3.
As, A is a tensor, it must be transformed as,
A= A
Setting, q = s
A= A
= A

1.10 Outer multiplication TU2055,2060

The product of two tensor is a tensor whose rank is sum of given
individual tensor
For e.g. A . B = C is outer product of A and B

1.11 Inner multiplication

The outer multiplication followed by contraction, called inner
multiplication.
For e.g. given tensors, A and B
The outer product, = A A B
Let q = r = A B = C
=C

9
1.12 Quotient law: TU 2056, 2066, 2067
If the inner product of a physical quantity with an arbitrary tensor is a
tensor, then the quantity is also a tensor.
Proof: We consider a physical quantity A(i, j, k) so that,
A(i, j, k) B = C
Note for Benefit: Choose index in this way,
i j k 
p q r s
In a new coordinate system
A(p, q, r) B = C
A(p, q, r) B = C
or, A(p, q, r) B = A[i, j, k) B
A(p, q, r) = A(i, j, k)
Multiplying by on both sides,
A(p, q, r) = A(i, j, k)
A(p, q, r) = A(I, j, k)
d d A(p, q, r) = A(i, j,k)
Setting, p = m, q = n
A(m, n, r) = A(i, j,k)
This is the form of mixed tensor,
A = Ak
Since, transformation law of tensor satisfied. The quantity A(I, j, k) is
a mixed tensor of rank 3.

1.13 Conjugate or Reciprocal tensors: TU2069

Metric tensor is defined by,
ds2 = gij dxi dxj
Where ds2 is invariant.
gij
If g = |gij|  0, then conjugate or reciprocal defined as,
gij = =
=
Where G(i, j) is cofactor of gij in g

10
 gij G(i, j) = g
g21 G(2, 1) + g22 G(2, 2) + g23 G(2, 3) = g
Also, if,
gij G(, j) = 0 if i  
gij gj = 1 for i = 
= 0 for i = 
j
i.e. gij g = 
This is one of the important relation.
Q. Find g and gjk, if, TU 2054
2 2 2 2 3 2 1 2 2 3
ds = s(dx1) + 3(dx ) + 4(dx ) - 6 dx dx + 4 dx dx
Solution:
We have,
g = |gij| =
Then,
g11 = 5, g22 = 3, g33 = 4
g12 = g21 = - 3, g23 = g32 = 2
g31 = g1`3 = 0
g=
 g = 40 - 36 = 4
but gjk =
gij =
Cofactor of g11 = 8 Cofactor of g21 = - 12
Cofactor of g12 = 12 Cofactor of g22 = 20
Cofactor of g13 = - 6 Cofactor of g23 = - 10
Cofactor of g31 = - 6
Cofactor of g32 = - 10
Cofactor of g33 = 6
Now, gij =
=
 gij =
Conjugate metric tensor in spherical and cylindrical coordinates:-
(a) Cylindrical coordinates:
We know, x = Cos

11
 y =  sin
Z\Z
dx = -sind + cosd
dy = cosd + sin d
dz = z
Then, ds2 = dx2 + dy2 + dz2
= dP2 + 2d2 + dz2
Hence,
g11 = 1, g22 = 2, g33 = 1
 gij =
g = |gij| = = 2
g11 =
= =1
22
g = = =
g33 = = = 1
g12 = = = 0
Similarly, gjk = 0 if jk. In matrix form, the conjugate metric
tensor can be represented by,

(b) In spherical polar coordinate,

h1 = = 1, h2 = = r, h3 = = rsin
gij =
g = |gij| = r4 sinb2
g11 = =
= 1
22 33
g =, g =
and gjk = 0 if jk
Then, in matrix form

1.14 Associated Tensors

The tensors Aj and Aj are associated by the metric tensor,
Aj = gij Aj

12
 Similarly we can say that Bi is associated with Bj
Bi = gij Bj

1.15 Christoffel's symbols:

The following symbols,
(a) [ij, k] = (ij‚k) = ….(1
(b) =  = gk[ij, ] ………(2)
are called the christoffel symbols of the first and second kind,
respectively.

1.16 Transformation Laws of Christoffel's Symbols

(a) [ij‚k] = …………….(3)
k
(b) =g [ij‚l] ……………(4)
and,
[ij‚k] = [,] + g
……………..(5)
is the transformation law of first Christoffel symbol.
= + ……….(6)
is transformation law of second type Christoffes symbol.
Verification: TU 20760 (First type)
(d),
We know,
gij = g ……………(7)
Note for Benefit: Choose index in following order
i j k 
 
Differentiating (7) w.r. to xk
=
= g + g+

Let us change index in cyclic order,

j


k
i 13


 = g +  g
+
Again, change index in cyclic order I above equation,
= g+ g+

Now, substituting in equation (3), we get,

[ij‚k] = [,] + g
Hence, proved.
(b) To prove the transformation law of second kind, we, have
transformation of Christoffel symbol of first kind,
[ij‚k] = [,] + g
We know,
= xk[ij‚l]
= g
= g []
+ gg
 = +
Hence, proved.
Q. Prove, -
We have,
= +
Multiplying by , we get,
=  + 
= +
 = -
Hence, proved.
Short-cut idea:-
(a) Write the transformation rule of second kind as,
. [eqn (6)]
(b) Multiply both sides by , then write the final answer.

14
1.17 Prove:
(a) = [ik, j] + [jk, i]
R.H.S.
= [ik, j] + [jk, i]
=
=
(b) = - gj
Proof:
= =0
or, gpr + gqr = 0
gpr = - gqr
multiplying both sides by gqs
gqs gpr = - gqs gqr
gqs gpr [[(qk,r] + [rk,q]] = - 
or, gqs gpr [qk, r] + gqs gpr [rk, q] =
= - gqs - gpr
Replacing, i = p, j = s, q = , r = 
= - gj - gi Proved.
(c) = n
We have,
gpq = = ….(1)
gpq G(p,q ) = g
Differentiating with respect to gpq, assuming G(p, q) does not contain p, q
explicitly,
= G(p, q) [G(p, q) = ggpq]
=
= G(p, q) [[p, q] + [q, p]]
= ggpq [[p, q] + [q, p]]
= g+
= 2g For p = q
= = n
Replacing, p = i,  = j
= n Proved

15
1.18 TU 2067
Q. Find the Christoffel symbols of first kind, [22, 1], [12, 2] and
[21, 2] in,
(a) Cylindrical coordinate
We have,
[ij, k] =
In cylindrical coordinates (, , z) we have,
g= =
[22, 1] =
=
= -
[21, 2] =
=
= 
[12, 2] =
=
= 
Spherical polar coordinates
We have,
g= =
[22, 1] =
=
= -r
Similarly, [21, 2] = r

1.19
Q. Determine the Christeffel symbols of second kind in
(a) Cylindrical
(b) Spherical coordinates.
(a) Cylindrical coordinates:
x1 = , x2 = , x3 = z
also,
g11 = 1, g22 = 2, g33 = 1
The only non-zero christeffel symbols of the second kind can occur where P =
2, are,
= = 2 = - 

16
 = = = =
(b) In spherical coordinates,
x' = r, x2 = , x3 = 
and
g11 = 1, g22 = r2, g33 = r2sin2
The only non-zero chriostoffel symbols of the second kind can occur where
p = 2 or 3
= = (r2) = - r
= = = (r2) =
= - = (r2sin2) = - rsin2
= = (r2 sin2) = - sincos
= = = (r2sin2) =
= = = (r2 sin2) = cot

1.20 Geodesics: 2072 First term CDP, First Semester

In figure the shortest path s is called geodesics, is along
the curve xk = xk(t)
The path joining two points t 1 and t2 along the curve xk 
t2
= xk (t) in Riemanian space is given by, 
t1
Fig: Sphere
s= ds = dt
= dt ……….(1)
The integral should be minimum to find geodesies. The integral,
I = Ldt should be extremum
For the Lagrangian L = L(x, x, t) the equation is,
= 0 ………….(2)
Replacing L by s
= 0 ………(2a)
Here, s = ………..(3)
Now,
=
= gij
= gij 2  xj
and,
= xi + xi - gixi s
Again,

17
 = xi xj
Substituting in (2a)
xi + xixj - gi xi s
- xixj = 0
gi xi + xi xj - xi xj = gi xjs
gi xi + xi xj = gi xis
Let us choose s = 1, s = 0
gi xi + [ij, ] xi xj = 0
Multiplying both sides by gk
gk gixi + gk [ij, ] xi xj = 0
 xi + xi xj = 0
or, xk + xi xj = 0 [ds = dt]
 + =0
is required equation.

1.21 Covariant derivative:

The derivative of an invariant or scalar function produces a covariant vector
(tensor).
=
If we take derivative of a vector, we cannot obtain a tensor. The quantity is
the speed of particle through geodecis and it is also contraverient vector.
GThe product of Ai with ,
i.e. Ai is invariant quantity.
The rate,
=  (an invariant)
Ai + =  …….(1)
The equation of geodecies is,
+ = 0 ………(2
From (1) and (2)
- Ak + =
=
Since , both are contravarient tensors their coefficient represented by,
Aij = - Ak
is covarient tensor of rank 2.
The quantity Ai,j is the derivative of A i with respect to x j and known as
covariant derivative of Ai.
Also,

18
 We have covariant derivative of Aj w,r to xj
Ai,j = - Ak
Ai,j = - gkA
= A + gi - [ij, ] A
= A + gi
= [j, i] A + gi
Multiplying both sides by gik
gikAi,j = gik [, j, i] A + gik gi
A = A + 
A= + A
This is called covariant derivative of Ak w.r.to xj.

1.21a
Q. Write the covariant derivative with respect to x k of each of the
following tensors,
(a) gij
i.e., gij,k = - Asj - Asi
= [ik,j] + [jk, i] - [ik, j] - [jk, i]
= 0
(b) g = + gj + gi
=0
(c) 
= - + 
=  + 
= -+
= 0-+
= 0

1.21b TU 2059, 2069, 2071 First sem. mid-term

Q. Given Ap = Aq
and,
= -
Then prove
A = - Aq transforms like a tensor
Proof: We know,
Ap = Aq
19
 Differentiating w.r. to xr
= + Aq ……………(1
= +
or, = +
= - ……(2)
Substituting in (1)
= Aq - Aq
+
= d Aq - Aq
+
Replacing q by  , s by  in last term.
or, - At =
 Ap,r = A,
Hence, A = - Aq transforms like a tensor.

1.21c
(2) Similarly, we can prove
A = + A
We have contravarient tensor transforms like.
Ap = Aq
Differentiating with respect to xr
= + Aq
Let us copy and paste equation (2),
=
Interchanging x and x
= -
Then,
= Aq - Aq
+
Replacing, replacable indices, then we get
- A =
 A, = A,
Hence, + At transforms like a tensor.
[See schqum's outlines page no. 217]

20
1.22 dv = dx1, dx2 … dxN is invariant.
TU 2052, 62,63,64,65, 71

In matrix form metric tensor gij is written as,

gij =
Let, g = |gij| denotes the determinant of gij.
g = |gij| =
We know volume element in orthogonal system,
dv = h1h2h3 du1du2du3
In this case,
g12 = g23 = g31 = 0
g21 = g32 = g13 = 0
g = = g11g22g33
 = h1h2h3
Therefore, dv = dx1 dx2….. dxN
In primed coordinate,
dv = dx1 dx2 ….dxN
The metric tensor gij is transformed as,
gij = gpq
Taking determinant on both sides,
g = |gij| = |gpq|
2
g=Jg
=J
In the form, Jw = w, w is weight of tensor and J is Jacobian.
If w = 0, related tensor is called absolute tensor.
If w0, the tensor is called relative tensor.
Here, is relative tensor weight 1.
The volume element can be transformed as,
dv = dx1 dx2 …. dxN
= J dx1 dx2………xN
= dx1 dx2 …. dxN
= dx1 dx2…dxN
 dv = dv
Hence, the statement.

21
1.23 Tensor form of Gradient, Divergvence and
Curl
Gradient:
If  is a scalar or invariant, the gradient of  is given by
 = grad =
Divergence:
The divergence of vector Ai is the contraction of covariant derivative of Ai.
divAi = A = + AK
= + Ak
= +
Note for Benefit: Here we used the formulae,
= [See 1.17c]
div Ai =
 divAi = Ak
Curl
The curl of vector Ai is defined by,
Curl Ai = Ai,j - Aj,i
= -
Again,
Curl Ai = - ijk Ai,j
Where, ijk is permutation covariant tensor. It is defined by,
ijk = eijk
where eijk is the permutation symbol.
eijk = + 1 for ijk = 231, 123, 312
= - 1 for ijk =213, 321, 132
= 0 for any two indices being equivalent.
Laplacian: TU 2065, 2071 First sem mid term, 2071 end sem exam
We know, laplacian of any function is divergence of grad.
Let, Ak = gk
We have the divergence of any vector Ai is given by,
div Ai =
Then,
2 = div
 2 =
In case g<0, must be replaced by . Both cases g>0 and g<0 can be included
by writing in place of . So we can write,

22
 2 = gk
is required tensor form of Laplacian.

1.33 Physical Components:

The physical components of a vector A p or Ap denoted by Au,Av,Aw are the
projections of the vector on the tangents to the coordinate curves and are
given in the curve of orthogonal coordinates by,
Au = A' = , Av = A2 =
Aw = A3 =
Similarly, the physical components of a tensor Apq or Apq is,
Auu = g11A11 =
Auv = A12 =
Auw = A13 = etc.

1.34 Gradient Divergence, Curl, Laplacian in

spherical polar Coordiante [Tensor Form)
The physical components,
Ar = , A0 = A2 , A = A3
x1 = r, x2 = , x3 = 
g=
 = r2sin = r2sin
Ar = A1 = A1  A' = Ar
A = A2 = rA2  A2 =
A = A3 = rsinA3  A3 =
divAi = Ai
=
=
 divAi = (r2 Ar) + +
Laplacian: We know, TU 2072 end semester exam
2 =
=
+
where we used,
g11 = = 1, g22 = , g33 =
= + +
2

Curl:

23
 We know that for definition,
Curl Ai = Ai,j - Aj,i
= -
c1e1 + e2e2 + c3e3 = -
c1e1 = e1
where, e1 = =
c1e1 =
 c1e1 = er
Now,
e2 = e = =
c2e2 = e2
= e
Similarly,
e3 = = = e
 c3e3 = e3
=
 Curl Ai = e+
er + e

1.34 In Cylindrical Coordinate:

g = = 2
 =
A = A'  A1 = A
A = A2  A2 =
Az = A 3  A3 = Az
i
Div A =
 Div Ai =
Laplacian:
2 =
=
 2 = + +
Curl:
Curl Ai = Ai,j - Aj, i
= -
c1e1 + c2e2 + c3e3 = -
c1e1 = e1

24
 e1 = er
c1e1 =
 c1e1 = er ………..(i)
Similarly,
c2e2 = e2
where, e2 = e
c2e2 = e ………(ii)
Also,
c3e3 = e3
Where,
e3 = e = e
 c3e3 = e
Then, our final answer is,
Cul Ai = c1e1 + c2e2+ c3e3
= e + e +
e

1.35 Intrinsic or absolute derivative:

The intrinsic or absolute derivative of A i along the curve xj = xj(t) is denoted
by Ai/t and defined by as the inner product of covariant derivative of A i
with respect to xj and i.e.,
= Ai,j
=
= - Ak
 = - Ak
Eg: TU 2068
1. = ‚j
= =
2. = Aij,k
=
= - Aj - Ai
3. =A
=
= - A +A
Application: TU 2059, 2065
Prove, ai = +

25
 In Newtonian mechanics,
Fi = mai
=m
Here,
ai = = intrinsic derivative of velocity
i
a =
= v,
=
= + vk
 i
a = +
This satisfies the transformation laws of coordinate system in tensor analysis,
Hence therefore,
ai =
is contravariant tensorof rank 1.
ai = 0, means the intrinsic derivative of contravarient tensor v i is zero, the
tensor is said to move parallel along the curve xj = xj(t).
ai = is zero, covariant derivative along the geodecies is zero. Thus
geodecies is a tangent vector with zero covariant derivative along the curve.

1.36 Rieman-Christoffel tensor: TU2062,67,2071 Frist sem.

We can associate Rieman-christoffel tensor. R
Ap,qr - Ap,rq = R An
Here,
Ap,qr =(A,p,q), r = - Aj,q - Ap, j
= -
-
= Aj - -
+ Ak - + A
Interchanging q and r and substracting we find,
Ap,qr - Ap,rq = Ak - Aj - Ak
+ Aj
= A j - Aj - Aj
+ Aj
= R Aj
Note: Make clear understanding on covariant derivative, then it will be easy.
Where,
R= - -+

26
 Replace j by n then result follows:
 Ap,qr- Ap,rq = R An
Since, Ap,qr - Ap,rq is a tensor R An is q tensor, and since An is is an arbitrary
tensor, R is a tensor by trhe quotient law. This tensor is called the Riemann-
Chriatoffel tensor, and is sometimes written R.
Rpqrs = gns R is an associated tensor of R and thus is a tensor. It is called the
covariant curvature tensor and is of fundamental importance in Einsein's
general theory of relativity.
Short method:
We can associate Riemann Christoffel tensor R,
Ak,jk - Ai,klj = R A
Then,
Ai,jk = - A,j - Ai,
Interchanging j and k,
Ai,kj = - A, k - Aj,
Substracting, we get,
Aj,jk - Aj,kj = RA
The covariant curvature tensor is given by,
Rijk = gm R
Agaibn, Ricci tensor is given by,
R = Rik
is Ricci tensor.

1.37 Length of vector and angel between vector

T.U. 2051, 2052

The length of vector Ai or Aj is given by,

L2 = AiAj = gij Ai Aj = gij AjAj……(1)
We know,
ds2 = gij dxi dxj
= (gii)2 (dexi)2 [i = j]
=
The unit tangent vector T along the xi - curve,
T= = =  …………(2)
Here, for x1 - curv e, x1 = x3 = constant.
The angel between any two tangent vector T and T is,
cosij = g T T = g T T

27
 = g 
 cosij = ……..(3)
In three dimension the angle between two curves is,
cos12 = cos23 =
cos31 =
In orthogonal system, 12 = 23 = 31 = 90‫ۥ‬
i.e. g12 = g23 = g31 = 0
but, 11 = 22 = 33 = 0°
i.e. |gij| = g = g11 g22 g33
Since,
g=
in orthogonal system.

1.38 Alternatively
The inner product of Ai and Bj i.e. AjBj stands as a scalar product of two
vectors in Cartesian coordinate system. The angle between two vectors A j and
Bi is,
cos =
The angel ij between two vectors Ai and Bj
ij =
Then, cosij =
 cosij =
Then, follow from equation (3).

1.39 Stress tensor: TU 2053

Q. Prove stress tensor is contravarient tensor of rank 2.
x3

33
 31
32

23
 13

22
21
x2
 12
11

28
x1
 Let us consider a unit cube with three forces
F1 , F2 and F3 are applied.
It terms of components there three forces can be written as,
Fi = ij ej ………(1)
F1 = sij ej = 11e1 + 12 e2 + 13e3
F2 = 2j ej = 21e1 + 22 e2 + 33e3
F3 = 3j ej = 31e1 + 32 e2 + 33e3
Then,
ij =
is stress tensor of rank 2.
Where the diagonal elements are the normal stress and off diagonal element
are shear or tagential stress.
So, ij represents a tensor, called stress tensor of rank 2.
Also, we can derive that it is contravarient tensor of rank 2.
We know under the condition of equilibrium,
F = 0
F = ij n (ei . ej) ………..(2)
We can change the unit vector in accordance with coordinate transformation.
ej = M f, ej = M f
Where,
M=
So, equation (2) becomes,
F = ijn
= ij M M n ff
= sn ff
Where,
s = ij M M
 s = ij
Hence, therefore ij is stress tensor, contravcarient tensor of rank 2.
TU 2053
1.40Q A quantity A(j, k, , m) which is a function of coordinates x i
transforms to another coordinate system xi according to the rule,
A (p, q, r, s) = A(j, k, , m)
29
 (a) Is the quantity is tensor?
It follows coordinate transformation law of tensor, so it is tensor. i.e. the
given quantity is a tensor.
(b) If so write tensor in suitable notation?
A = A
(c) Find rank.
Here, rank of tensor is 4.
1.41Q What is metric tensor? Show that it can be choosen to be
symmetric? T.U. 2051

= See article 1.7

1.42Q Show that volume element, dv = dx' dx2 …dxN is invariant.
T.U. 2052, 2062, 2063, 2064, 2065, 2071, semester board exam)

= See 1.22
1.43Q Show that the angels 12, 23 and 31 between the coordinate curves
in a three dimensional system are given by
cos12 = , cos23 = , cos31 =
T.U. 2051, TU 2052
= See 1.37, 1.38
1.44Q Show that stress tensor is a tensor of second rank TU 2053
= See 1.39
1.45Q Write down contravarient, Covarrient and mixed tensors of rank
2. T.U. 2054
T.U. 2054
= Refer to the article 1.49 for definition of covariant, contravarient and
mixed tensor.
Covarient tensor of rank 2,
Aij = A
= m m A(matrix form)
Contravarient tensor of rank 2,
Aij = A
= MjMiA[matrix form]
Mixed tensor of rank 2,
A = A
d= A (kronekar delta)

30
1.46Q Show that the square of the element of arc length in general
curvilinear coordinates can be expressed as,
ds2 = SS gij dui duj
Where symbol carry usual meaning. T.U. 2057
Solution: See 1.2
1.47Q If Aj and AP represent the same contravarient vector in unprimed
and primed frame if reference respectively, then show that,
AP = Aj T.U. 2059
 Write detail about contravarient tensor see article 1.49
1.48Q Show that components of co-variant vector should transforms as
AP = A in going from one coordinate system to another.
= Read from 1.49
1.49Q Define conjugate tensor. Prove, g gv =  TU2069
= See 1.13
1.50Q Show that  = in all coordinate is a mixed tensor. What is the
associated tensor ij?
= Refer to the 1.6
1.51Q Let A be a tensor. Check whether A and A are tensor. It they are tensors,
determine their ranks. TU 2070
Solution:
Given, A is tensor, so
A= A
let j = n, we need to show A a tensor.
A= A
= A
=  A
= A
Hence, A is a tensor of rank 3.
Also, A
Let put j =n, k =m in equation (1)
A = A
= A
=  A
= A
Hence, A is tensor of rank 1y also denoted by Br, is followed by contraction.

31
1.52Q A covarient tensor has components xy 1, 2y-z2, xz in rectangular
coordinates. Find it's covariant components in spherical
coordinates. TU 2070
Solution: Let Aj denote the covariant components in rectangular coordinates,
x' = x, x2 = y x3 = z then,
A1 = xy = x'x2, A2 = 2y - z2 = 2x2- (x3)2
A3 = x1x3
Where care must be taken to distinguish between superscripts and exponents.
Let Ak denote the covariant components in spherical coordinates x1 = r,
x2 = , x3 =  then,
Ak = Aj …………..(1)
The transformation equations between coordinate systems are,
x1 = x1 s in x2 cosx3, x2 = x1 sin x2 sinx3
and x3 = x1 cosx2
Then equations (I) yields the require components,
Aj = A1 + A2 + A3
= (sin x2 cosx3) (x1 x2)+ (sinx2 sinx3) (2x2 - (x3)2) + (cosx2) x1x3
= (sin cos) (r2sin2 sin cos) + (sinsin) (2rsin) sin - r2 cos2)
+ cos (r2sin cos cos)
A2 = A1 + A2 + A3
= (r cos cos) (r2 sin2 sin cos) + (rcos sin)
(2r sinsin - r2 cos2) + (-rsin) (r2sin cos cos)
A3 = A1 + A2 + A3
= (- rsinsin) (r2sin2 sin cos) + (rsincos) (2rysinsin - r2cos2) +
(0) (r2sin2 cos cos)
[Se Schamu's series page no. 200]
1.53Q Find the covariant derivative of A B w.t. to xq. T.U. 2070

Here,
q = -AB- AB
+ AB+ AB+AB
=
B+A
=AB+AB
1.54 If a vector ui gets parallel transported along the curves then, u, =
0
32
 If the angle between two vectors ui nd vi is , show that getting
parallel transported does not change angle betgweentghem.
Solution: 2071 first semester first term
Since, ui and vi be the given two vectors, which undergoes parallel
displacement, we can write,
u =0
v =0
multiplying both equations by gii,
ui,j = 0
v i,j = 0
If the angle between two vectors ui and vi is , then
uivi = uvcos
(uivi) = - uv sin ……(1)
Also, we can write,
(uivi) = (uivi),j vi
= uivi,j + u vi
= ui.0 + 0.vi
= 0 ………………..(2)
Hence, (1) and (2) implies that,
uvsin = 0
=0
  = constant.
which implies angle is constant i.e. does not change.
TU 2071 First sem. Second term.
1.55 Show that   = 2

Use it to find 2, when,

ds2 = - at2 + a2(t) [dx2 + dy2 + dz2] or,
gij =
Solution: Given,
ds2 = - at2 + a2(t) [dx2 + dy2 dz2] and metric tensor is,
gij =
g = |gij| = a6(t)
 = a3(t)
Here, we replace by
 = a3(t)
You can do by taking also,

33
 2 =
=
=
 2 = 3 + -
TU2063
1.56 Show that ds2 = c2(dx4)2 - dxk dxk (k = 1,2, 3) is invariant under the
transformations,
x1 =  (x1 - vx4), x2 = x2 , x3 = x3
x4 =  where  = v /c,  =
Solution: We have to prove,
ds2 = c2 (dx4)2 - dxk dxk
ds = c2(dx4)2 - dx1 dx1 - dx2 dx2 - dx3dx3
2

= c2 2
- d{(x1 - vx4)}
d{(x12 - vx4)_} - dx2 dx2 - dx3 dx3
= c22 2 - 2 (dx1 - vdx4)2 - dx2 dx2 - dx3dx3
= c22 -  2
- dx2 dx2 - dx3 dx3
= c2 2 (dx4)2 - 2 2 cdx1 dx4 + r22 (dx1)2
- 2(dx1)2 + 22vdx1 dx4 - v22 (dx4)2 - dx2dx3
= r2 (c2 - v2) (dx4)2 - 2 2 vdx1 dx4 +  22(dx1)2
- 2(dx1)2 + 2 2v dx1 dx4 - dx2 dx2 - dx3dx3
= c2(dx4)2 - 2 (1 - 2) (dx1)2 - dx2dx+2 - dx3 dx3
= c2(dx4)2 - dxkdxk
= ds2
ds2 = ds2
Hence ds2 is invariant.
TU 2066
1.57 Show that,  = 0 i j is a tensor,
Whereas,
ij = 0 if i  j
= 1 if i = j
Soln:- is not a tensor.
 For First part see 1.6
For second part,
ij must be transformed as follows to become a tensor,

34
 drs = ij
For i = j, ij = 1
drs = .1
drs =
ij does not transforms like a tensor, so it is not a tensor.
Note for Benefit: Read by placing syllabus infront of you, and follow
syllabus and class note. Do not confuse with yearly systems old
questions, which are out of syllabus. And all problems discussed by
prof. are important. Read all problems discuss in class. You will secure
good marks.
1.60Q Write down the intrinsic derivative of following tensors, assumed
to be differential function of it.
(a)  (b) Aj (c) A
Solution: See article 1.35 examples.
1.61Q Derive transformation laws for the christoffel symbols of the first
kind? TU 2070
Solution: See 1.16
1.62Q Define covariant tensor of rank one and two interms of
transformation rules in going from one coordinate system to
another. Define covariant derivative of tensors of rank one and
two. Find an expression for R pr using A p,dr - Ap, rd = R A where
the symbols carry their usual meanings.
TU 2071 First semester
Solution: For First part see 1.4a, 1.45 and 1.21
For second part see 1.36
1.67Q Define conjuate of a tensor. Prove that for metric tensor g  gv = 
TU 2069
See 1.13
1.68Q Prove that for orthogonal coordinate system the off-diagonal
components of the metric tensor is zero. TU 2052
Solution: See 1.37
Some Important problems with solutions.
1.69Q Suppose A and B are tensors. Prove that their sum and difference
are tensors.
Solution: We have A and B are tensors. So it satisfy transformation rule.
A= A

35
 B= B
Adding (1) and (2)
A+B =
Substracting,
A-B =
Then and are tensor of the same rank and type as A and B
1.70Q Let AP = Aq prove that Aq = AP
Solution: We have,
AP = Aq
Multiplying by
or, AP = Aq
=  Aq
= Ar
Put, r = q
 Aq = AP
Hence, proved.
Q. Suppose Ap and Ap are tensors, show that,
A, = + As is a tensor.
Solution: This is same problem like 1.21c. Also you can proceed as follows.
Since, Aj = Ar ……..(1)
= + Ar ………….(2)
Also we have,
= - -
Interchanging x and x coordinates,
= -
Substituting in (2),
= + Ar
-
= +n Ar
-  Ar As - Aj
+ Ar
= As Aj
= + Ai =
As Aj
and is a mixed tensor of rank 2 as it is transformed like a tensor.
Tensor form of Maxwell equation

36
 (a) div B = 0
We define the tensor Bk. Then we can write above eqn,
B=0
(b) div D = 4
We define the tensor Dk , and '' is invariant.
D = 4
(c) ×E =
This can be written as,
- jkq Ek,q =
or, jkq Ek,q =
(d) ×H =
We can write in tensor form,
- jkq Hk,q =
or, jkq Hk,q =
Where, Hk and Ik are tensors and c is invariant. These equations form the basis
for electromagnetic gheory.
Q. Suppose the kinetic energy T of a particle of constant mass M
moving with velocity having magnitude U is given by t,
T = Mv2 = M gpq
Prove that,
- = Mak
Where ak denotes the covariant components of acceleration.
Solution: We have,
T = Mv2 = M gpq
Differentiating w.r.t to xk
= M
= Mgkq q
and,
=M
Now,
-
= M
= M
= M (gkq xq + [pq, k] xp xq)
= Mgkr
= Mgkr qr

37
 = Mqk
Hence,
- = Mak
Hence proved.
TU2072 First semesters Q.No. 5
Q. Explain why there are basically two kinds of tensors
(contravariant and covariant only). Mixed tensor can be explained
in terms of contravariant and covariant tensors.
Solution: Let us consider the rotational transformation of a vector,
A = A1e1 + A2e2 + A3e3
from the Cartesian system defined by ei (I = 1, 2, 3) into a rotated coordinate
system defined by e, with the same vector A then represented as,
A' = A e + A e + A e
the components of A and A' are related by,
A = S Aj ……….(1)
Where the coefficients are the projections of e in the ej directions. Because the
ei and the ej are linearly related, we can write,
A = S Aj …………(2)
We also know that the gradient of a scalar  has in the unrotated Cartesian
coordinates the components
()j = ej
meaning that in a rotated system, we have,
() = ) = = …………(3)
showing that the gradient has a transformation law that differs form that of eq n
(2) in that xi/xj has been replaced by xj/x'i
Quantities transforming according to equation (2) are called
contravariant vectors, while those transforming according to equation (3)
are termed covariant. Form above explanation it is clear that the we have
basically two kinds of tensors i.e. contra variant and covariant
Summarizing,
(A')i = S Aj, A a contravariant tensor
A'i = S Aj A a covariant tensor
For second part see 1.5.

38