Tensor Calculus ○

1
Introduction
Tensor is a generalized vector. So tensor analysis or calculus is the generaliza-
tion of the vector analysis or calculus. Tensor deals with the characteristics or
properties or relations of the transformations from one co-ordinate system to
another co-ordinate system of dimension N, where N is a positive integer.

Spaces of N-dimension
In three dimensional space a point is a set of three numbers called co-ordinates
determined by a particular co-ordinate system or frame of reference.
For examples (x, y, z), (ρ, ϕ, z), (r, θ, ϕ) are co-ordinates of a point in rect-
angular, cylindrical and spherical co-ordinate system respectively.
A point in N-dimensional space is by analogy a set of N numbers denoted
by (x1 , x2 , x3 , . . . , xN ), where 1, 2, 3, . . . , N are not as exponents or powers but
as superscripts (upper indeces) of x and N ≥ 2, a policy which we shall prove
useful.

LaTeX transcription of Dr. Khondokar Mezbahuddin Ahmed’s handwritten lecture notes

for MTH 404: Differential Geometry and Tensor Calculus by Atiq Rahman.
 ○
2
Riemannian space based on distance function. Euclidean space based on the point of
co-ordinates system.

Tensor
Tensors are based on coordinates of axes. If dimension 1 then tensor it represents a
vector and if dimension more than 1 then it represents tensor.

Definition of curve
A curve in VN (VN denotes set of points of N dimension) is defined as the totality of
points which satisfy the N equations given by xi = xi (t) where t is a parameter and
xi (t) denote functions of t.

Transformation of co-ordinates
Let (x1 , x2 , x3 , . . . , xN ) and (x̄1 , x̄2 , x̄3 , . . . , x̄N ) be a co-ordinates of a point in two
different frames of reference. Suppose there exists N independent relations between
the co-ordinates of two system having the following form:

x̄1 = x̄1 (x1 , x2 , . . . , xN ) 




x̄2 = x̄2 (x1 , x2 , . . . , xN ) 



T : .. → (1)
.





x̄N = x̄N (x1 , x2 , . . . , xN )



which can be indicated briefly by

T : x̄i = x̄i (x1 , x2 , . . . , xN ), i = 1, 2, . . . , N → (2)

The subscripts (superscripts or lower scripts) are used to identify the variables.
 ○
3
Here T is a real, single-valued reversible functional transformations. Because x1 , x2 , . . . , xN
are functions of t, where t is a parameter.
Since the transformation T in (1) is assumed to be reversible and one-to-one, we
can write briefly by

T −1 : xi = xi (x̄1 , x̄2 , . . . , x̄N ) → (3)

The relations (2) and (3) define a transformation of co-ordinates from one frame of
reference to another.

Summation Convention
In
PN writing an expression such as a1 x1 +a2 x2 +· · ·+aN xN , we can use the short notation
i i
i=1 ai x . An even shorter notation is simply to write it as ai x , where we adopt the
convention that whenever an index (superscript or lowerscript) is repeated in a given
term then we are to sum over that index from 1 to N unless otherwise specified. This
is called Summation Convention.

Dummy index or umbral index

Clearly, instead of using the index i, we could have used another letter say p and the
sum could be written as ap xp . Any index which is repeated in a given term, so that
the summation convention applies, is called a dummy index or umbral index.
 ○
4
Free index
An index occuring only once in a given term is called a free index and can stand for
any of the numbers 1, 2, . . . , N such as k in the two equations given below:

xk = xk (x̄1 , x̄2 , . . . , x̄N )

x̄k = x̄k (x1 , x2 , . . . , xN )

Contravariant tensor or vector of first order or rank one

If N quantities A1 , A2 , . . . , AN in a co-ordinate system (x1 , x2 , . . . , xN ) are related to
N other quantities Ā1 , Ā2 , . . . , ĀN in another co-ordinate system (x̄1 , x̄2 , . . . , x̄N ) by
the transformation equations
N
X ∂ x̄p q
Āp = A , p = 1, 2, . . . , N
q=1
∂xq

which by the convention adopted can simply be written as

∂ x̄p q
Āp = A
∂xq
they are called Components of Contravariant tensor or Contravariant vector of first
order or rank one.

Covariant tensor or vector of first order or rank one

If N quantities A1 , A2 , . . . , AN in a co-ordinate system (x1 , x2 , . . . , xN ) are related to
N other quantities Ā1 , Ā2 , . . . , ĀN in another co-ordinates system (x̄1 , x̄2 , . . . , x̄N ) by
the transformation equations
N
X ∂xq
Āp = Aq , p = 1, 2, . . . , N
q=1
∂ x̄p

So,
∂xq
Āp =
Aq
∂ x̄p
they are called components of Covariant tensor or Covariant vector of first order or
rank one.
 ○
5
Contravariant and covariant tensors of rank two
If N 2 quantities Ars in a co-ordinate system (x1 , x2 , . . . , xN ) are related to another N 2
quantities Āpq in another co-ordinate system (x̄1 , x̄2 , . . . , x̄N ) by the transformation
equations
N X N
X ∂ x̄p ∂ x̄q rs
Āpq = A , p, q = 1, 2, . . . , N
r=1 s=1
∂xr ∂xs
Which by the conventions adopted can simply be written as
∂ x̄p ∂ x̄q rs
Āpq = A
∂xr ∂xs
they are called Contravariant components of a tensor of second order or rank two.
Similarly, the N 2 quantities Ars are called Covariant Component of a tensor of
rank two if
N X N
X ∂xr ∂xs
Āpq = Ars , p, q = 1, 2, . . . , N.
r=1 s=1
∂ x̄p ∂ x̄q
So,
∂xr ∂xs
Āpq = Ars
∂ x̄p ∂ x̄q

Mixed tensor of rank two

The N 2 quantities Ars are called component of mixed tensor of rank two if
N X N
X ∂ x̄p ∂xs r
Āpq = As , p, q = 1, 2, . . . , N.
r=1 s=1
∂xr ∂ x̄q

So,
∂ x̄p ∂xs r
Āpq = As
∂xr ∂ x̄q
Mixed tensor of rank more than two ○
6
Tensor of rank more than two are defined as for example, Aijk
lm are the components of
a mixed tensor of rank 5, where contravariant of order 3 and covariant of order 2, if
they transform according to the relations

∂ x̄p ∂ x̄q ∂ x̄r ijk ∂xl ∂xm

Āpqr
st = A or,
∂xi ∂xj ∂xk lm ∂ x̄s ∂ x̄t
∂ x̄p ∂ x̄q ∂ x̄r ∂xl ∂xm ijk
Āpqr
st = A , p, q, r, s, t = 1, 2, . . . , N
∂xi ∂xj ∂xk ∂ x̄s ∂ x̄t lm
Similarly, Aijk
pqrs is a mixed tensor of rank 7, Contravariant of rank 3 and covariant of
rank 4, if they transform according to the law

∂ x̄l ∂ x̄m ∂ x̄n ∂xp ∂xq ∂xr ∂xs ijk

Ālmn
αβγδ = Apqrs
∂xi ∂xj ∂xk ∂ x̄α ∂ x̄β ∂ x̄γ ∂ x̄δ

Generalization of mixed tensor

A set of N r+s quantities Aqp11,q ,p2 ,...,pr
2 ,...,qs
co-ordinate system (x1 , x2 , . . . , xN ) is a mixed
tensor of rank (r + s) contravariant of rank r and covariant of rank s, if they are
related to the set of N r+s other quantities Ām 1 ,m2 ,...,mr
n1 ,n2 ,...,ns in another co-ordinate system
1 2 N
(x̄ , x̄ , . . . , x̄ ) by the transformation equation

∂ x̄m1 ∂ x̄m2 ∂ x̄mr ∂xq1 ∂xq2 ∂xqs p1 ,p2 ...pr

Ām 1 ,m2 ...mr
n1 ,n2 ...ns = . . . . . . Aq ,q ...q → (4)
∂xp1 ∂xp2 ∂xpr ∂ x̄n1 ∂ x̄n2 ∂ x̄ns 1 2 s

Type of tensor
The equation (4) type is called a tensor of the type (r,s). So, the tensor Apq is of the
type (2,0), the tensor Apq is of the type (0,2) and Apq is a tensor of the type (1,2).
Contraction of tensors ○
7
If one contravariant and one covariant index of tensor are set equal, the result indicates
that a summation over the equal indices is to be taken according to the summation
convention. This resulting sum is a tensor of rank two less than that of the original
tensor. This process is called Contraction of tensor. As for example, in the tensor of
rank 5, Apqr pqt pq
st set r = s to obtain Ast = Bt which is a tensor of rank 3. Further by
pq p
setting q = t, we obtain Bt = C which is a tensor of rank one.

The Kronecker delta

The Kronecker delta written as δji which is defined by
(
i 0, i ̸= j
δj =
1, i = j

As its notation indicates with two indices, it is a mixed tensor of one contravariant of
rank one and of one covariant of rank one and type is (1,1).

Basis and reciprocal basis vectors and metric

Any set of three linearly independent vectors e1 , e2 , e3 is said to be basis for a three-
dimensional space. The basis vectors e1 , e2 , e3 and e1 , e2 , e3 of two bases are called
reciprocal if ei · ej = δji i.e. if each vector of one basis is perpendicular to two vectors
of other.
The Riemannian metric and metric tensor ○
8
The metric written as gij which is defined by

gij = ei · ej = gji = ej · ei

and the reciprocal is defined by

g ij = g ji = ei · ej

In a rectangular system gij = 1 when i = j and gij = 0 when i ̸= j. The quantities gij
and g ij are the same as those appearing in the quadratic differential form which are
defined in the Riemannian metric. The quantities gij and g ij describe the geometric
characteristics of a space and it will be seen that both of them are components of
tensors of rank two. The quantities gij is known as fundamental metric tensor.

Associate tensor
The inner product of the fundamental tensor gij and Aj is a covariant vector gij Aj
which is said to be associated to Aj . We define

Ai = gij Aj ;
B i = g ij Bj

and we say that the vector B i is associate to the vector Bj .

 ○
9
The relation between a vector and its associate is reciprocal for the vector associate
to Aj is
g ij Aj = g ij gjk Ak = δik Ak = Ai
This process of association is often referred to lower the superscripts or raising the
subscripts respectively.

Symmetric and skew symmetric tensor

A tensor is called symmetric with respect to two contravariant or two covariant indices
if its components remain unaltered upon interchange of the indices. Thus if Apqr st =
Aqpr
st , the tensor is symmetric in p and q. So, if a tensor is symmetric with respect to
any two contravariant and any two covariant indices, it is called Symmetric.
A tensor is called skew-symmetric with respect to two contravariant or two co-
variant indices if its components change sign upon interchange of the indices. Thus
if Apqr
st = −Aqpr
st , the tensor is skew-symmetric in p and q. So, if a tensor is skew-
symmetric with respect to any two contravariant and any two covariant indices, it is
called skew-symmetric.
Example 1.01 ○
10
Show that every tensor of rank two can be expressed as the sum of two tensors,
one of which is symmetric and the other skew-symmetric in a pair of covariant or
contravariant indices.

Proof
Let us consider the contravariant tensor Aij of rank two. We can write
1 ij 1
Aij = (A + Aji ) + (Aij − Aji )
2 2
= B ij + C ij → (1) (1)

where
1 ij
B ij = (A + Aji ) → (2) (2)
2
1
and C ij = (Aij − Aji ) → (3) (3)
2
Interchanging i and j in (2), we have
1 ji 1
B ji = (A + Aij ) = (Aij + Aji ) = B ij
2 2
which shows that B ij is symmetric.
Again, interchanging i and j in (3), we have
1 ji 1
C ji = (A − Aij ) = − (Aij − Aji ) = −C ij
2 2
which shows that C ij is skew-symmetric. Thus in R.H.S. of (1) B ij is symmetric and
C ij is skew-symmetric. Hence every contravariant tensor of rank two can be expressed
as the sum of two tensors, one of which is symmetric and the other skew-symmetric
in a pair of contravariant indices.
Example 1.02 ○
11
Show that any covariant tensor of rank two can be expressed as the sum of symmetric
and skew-symmetric tensors.

Proof
Let us consider the covariant tensor Aij of rank two. We can write
1 1
Aij = (Aij + Aji ) + (Aij − Aji )
2 2
= Bij + Cij → (1)

where
1
Bij = (Aij + Aji ) → (2) (4)
2
1
and Cij = (Aij − Aji ) → (3) (5)
2
Interchanging i and j into (2), we have
1 1
Bji = (Aji + Aij ) = (Aij + Aji ) = Bij
2 2
which shows that Bij is symmetric.
Again, interchanging i and j into (3), we have
1 1
Cji = (Aji − Aij ) = − (Aij − Aji ) = Cij
2 2
which shows that Cij is skew-symmetric. Thus in R.H.S. of (1) Bij is symmetric and
Cij is skew-symmetric.
Hence every covariant tensor of rank two can be expressed as the sum of two
tensors, one of which is symmetric and the other skew-symmetric in a pair of covariant
indices.
Example 1.03 ○
12
If Apq pq
r and Br are tensors, prove that their sum and difference are tensors.

Proof
By hypothesis Apq pq
r and Br are tensors, so that we have

∂ x̄i ∂ x̄j ∂xr pq

Āij
k = Ar → (1) (6)
∂xp ∂xq ∂ x̄k
∂ x̄i ∂ x̄j ∂xr pq
and B̄kij = Br → (2) (7)
∂xp ∂xq ∂ x̄k
By adding (1) and (2), we have

∂ x̄i ∂ x̄j ∂xr pq

Āij ij
k + B̄k = (Ar + Brpq )
∂xp ∂xq ∂ x̄k
Thus Apq pq pq pq
r + Br is a tensor of the same rank and type as Ar and Br .
Again, by subtracting (1) and (2), we also have

∂ x̄i ∂ x̄j ∂xr pq

Āij ij
k − B̄k = (Ar − Brpq )
∂xp ∂xq ∂ x̄k
Thus Apq pq pq pq
r − Br is a tensor of the same rank and type as Ar and Br .

Example 1.04
pqs
If Apq s pq s
r and Bt are tensors, show that Crt = Ar Bt is also a tensor.

Proof
pqs
We have to prove that Crt is a tensor whose components of tensors Apq s
r and Bt .
pq s
Since Ar and Bt are tensors, we have

∂ x̄i ∂ x̄j ∂xr pq

Āij
k = Ar → (1) (8)
∂xp ∂xq ∂ x̄k
l ∂ x̄l ∂xt s
and B̄m = Bt → (2) (9)
∂xs ∂ x̄m
Multiplying (1) and (2) we have
 ○
13
∂ x̄i ∂ x̄j ∂xr ∂ x̄l ∂xt pq s
Āij l
k B̄m = Ar Bt
∂xp ∂xq ∂ x̄k ∂xs ∂ x̄m
pq s
which shows that Ar Bt is a tensor of rank 5 with contravariant indices p,q,s and
covariant indices r,t.
pqs
By using the notation Crt = Apq s
r Bt is also a tensor, and which is the outer
pq s
product of Ar and Bt .

Outer multiplications
Outer multiplication: the product of two tensor is a tensor whose rank is the sum of
the ranks of the given tensors. This product which involves ordinary multiplication of
components of the tensor is called the outer multiplication. As for example Apq s
r Bt =
pqs pq s
Crt is the outer product of Ar and Bt .

Remark 1.05
Every tensor can be written as a product of two tensors of lower rank. For two reason
division of tensors is not always possible.

Inner multiplications
Inner multiplication: By the process of outer multiplication of two tensors followed by
contraction we obtain a new tensor called an inner product of the given tensors. The
process is called inner multiplication. As for example, given the tensors Apq m
r , Bne , the
pq m
outer product Ar Bne .
 ○
14
Let r=m, q=n then we obtain the inner product Apn m
m Bne . If we let r=m, we obtain
pq m
the inner product Am Bnl .

Remark 1.06
The inner and outer multiplications of tensors are commutative and associative.

Scalars or invariants
Scalars or invariants: Suppose θ is a function of co-ordinates xk and let ϕ denote
functional value under a transformation to a new set of co-ordinates y k . Then θ is
called a scalar or invariant with respect to co-ordinate transformations if θ = ϕ.

Remark 1.07
A scalar or invariant is also a tensor of rank zero.

Tensor field
If to each point of a region in N-dimensional space there corresponds a definite tensor,
we say that a tensor field has been defined. Then we call that this is a vector field or
a scalar field according to as the tensor is of rank one or zero.
Riemannian space and metric space ○
15
In the three dimensional Euclidean space the distance between two contiguous points
(x, y, z) and (x + dx, y + dy, z + dz) is given by

ds2 = dx2 + dy 2 + dz 2 ,

where ds is called the element or linear element of the curve and the axes are rectan-
gular co-ordinates. Also the distance ds between two rectangular neighbouring points
with oblique curvilinear co-ordinates (u, v, w) and (u + du, v + dv, w + dw) is given by

ds2 = a du2 + b dv 2 + c dw2 + 2f dudv + 2g dvdw + 2h dwdu,

where a, b, c, f, g, h are functions of the coordinates u, v, w.

This idea of distance was extended by Riemann to space of N-dimension. The
quadratic differential form gij dxi dxj which expresses the distance between two adja-
cent points is called a Riemannian metric or metric or line element and gij is called
metric tensor or fundamental tensor.
An N-dimensional space characterized by a Riemannian metric is called a Rieman-
nian Space and it is denoted by Riemannian Vn .

Remark 1.07
Riemannian geometry is that part of geometry which is based upon Riemannian metric.
Product of two vectors A and B ○
16
A · B = gij Ai B j which is in contravariant form.
A · B = g ij Ai Bj which is in covariant form.

Remark 1.08
A · B = aα β
i Aα ai Bβ , Āi = aα
i Aα ; B̄i = aβi Bβ .

= aα β
i ai Aα Bβ

= g αβ Aα Bβ , g αβ = aα β
i ai

Length or magnitude of a vector

A2 = gαβ Aα Aβ = g αβ Aα Aβ = Aα Aα
p p √ √
∴ A = gαβ Aα Aβ = g αβ Aα Aβ = Aα Aα = Aα Aα

Angle between two vectors

A·B
A · B = AB cos θ ∴ cos θ =
AB

gαβ Aα B β
cos θ = p p
gαβ Aα Aβ gαβ B α B β
g αβ Aα Bβ
= p p
g αβ A
α Aβ g αβ Bα Bβ
α
A Bα
= √ p
Aα Aα B β Bβ
!
−1 Aα Bα
∴ θ = cos √ p
A Aα B β Bβ
α
Length of an arc PR = s ○
17
ds2 = gαβ dxα dxβ

Length of a curve in a Riemannian space Vn

b 1
dxα dxβ 2
Z 
s= gαβ dt, xi = xi (t)
a dt dt

Angle between two coordinate curves

Let θ be the angle. Then
β
ds1 · ds2 gαβ dxα
1 dx2
cos θ = =
ds1 ds2 ds1 ds2
q q
β β
Where ds1 = gαβ dxα
1 dx1 , ds2 = gαβ dxα2 dx2
So,
gαβ Aα B β
cos θ = p p
gαβ Aα Aβ gαβ B β B β
β
Where dxα α
1 = A and dx2 = B
β
Metric tensor and Reciprocal or Conjugate met-
ric tensor: ○
18
Relation between covariant and contravariant
We have that in an orthogonal system the distinction between the covariance and
contravariance vanishes so that we have Ai = Ai . Let x be our given system and x̄ be
an orthogonal system so that Āi = Āi . Now, Āi = Āi or,

∂xλ ∂ x̄i β
i
Aλ = A → (1)
∂ x̄ ∂xβ
∴ bλi Aλ = aiβ Aβ ,
λ i
Where bλi = ∂x , ai = ∂x
∂ x̄
. Multiplying both sides of (1) by aiα and sum over i. Then
P ∂i x̄iλ β Pβ i i β
aα aβ A or, δαλ Aλ = Aβ
P i i
we have aα bi Aλ = aα aβ or, Aα = Aβ gαβ , where
∂ x̄i ∂ x̄i
P i i
gαβ = aα aβ and aα = ∂xα , aβ = ∂xβ and x̄ is an orthogonal system and xα , xβ
i i i

are the co-ordinates of a given system. Now Aα = Aβ gαβ

A1 = Aβ g1β = A1 g11 + A2 g12 + · · · + AN g1N

A2 = Aβ g2β = A1 g21 + A2 g22 + · · · + AN g2N
..
.
AN = Aβ gN β = A1 gN 1 + A2 gN 2 + · · · + AN gN N
 ○
19
then
g11 g12 ... g1N
g21 g22 ... g2N
gαβ =
... ... ... ...
gN 1 gN 2 ... gN N
and this gαβ is called the fundamental metric or simply the metric.
Now,
 
g11 g12 . . . g1N
 g21 g22 . . . g2N 
[gαβ ] = 
...
 is called the metric tensor.
... ... ... 
gN 1 gN 2 . . . gN N

The reciprocal or conjugate metric tensor is

 11
g 12 g 1N

g ...
21
g g 22 ... 2N 
g 
[g αβ ] = 
... ... ... ... 
gN 1 gN 2 ... gN N
G(α,β)
where g αβ = g
, where g = |gαβ | ̸= 0 and G(α, β) are the cofactors of gαβ .

Theorem
Show that gij dxi dxj is an invariant.

Proof
Consider the transformation xi → x̄i . Since gij is a covariant tensor of rank two, so
we have
∂xi ∂xj
ḡpq = gij p
∂ x̄ ∂ x̄q
 ○
20
or,
∂xi ∂xj
ḡpq − gij =0
∂ x̄p ∂ x̄q
or,
∂xi ∂xj
 
ḡpq − gij dx̄p dx̄q = 0 [Multiplying both sides by dx̄p dx̄q ]
∂ x̄p ∂ x̄q
or,
∂xi ∂xj p q
ḡpq dx̄p dx̄q − gij dx̄ dx̄ = 0
∂ x̄p ∂ x̄q
or,  i  j 
∂x ∂x
ḡpq dx̄p dx̄q = gij dx̄ p
dx̄ q
∂ x̄p ∂ x̄q
∴ ḡpq dx̄p dx̄q = gij dxi dxj
i j
Hence gij dx dx is an invariant.

Theorem
The fundamental tensor gij is a covariant symmetrical tensor of rank 2.

Proof
The line element is given by

ds2 = gij dxi dxj → (1)

We consider a coordinate transformation from xi to x̄i (i=1,2,...,N) given by

xi = xi (x̄1 , x̄2 , . . . , x̄N )

So that the metric gij dxi dxj transforms to ḡij dx̄i dx̄j . But ds is invariant, so

ds2 = gij dxi dxj = ḡij dx̄i dx̄j → (2)

Now, we have to show that

(i) dxi and dxj are contravariant tensor.
(ii) gij is a covariant tensor of rank 2 and
(iii) gij is a symmetric tensor.
Proof of (i) ○
21
When xi are transformed to x̄i , then We have

∂ x̄i p
dx̄i = dx
∂xp
∂ x̄i p
∴ Āi = A , where dxp = Ap
∂xp
Which confirm that Ai , that is dxi is a Contravariant vector or tensor. Similarly, we
can show that dxj are Contravariant tensors.

Proof of (ii)
To prove gij is a covariant tensor of rank 2, we have

∂xi p ∂xj q
dxi = dx̄ and dxj = dx̄
∂ x̄p ∂ x̄q
Substituting these values in ds2 = gij dxi dxj = ḡij dx̄i dx̄j → (3) We get
 i  j 
∂x ∂x
gij dxi dxj = ḡij dx̄ p
dx̄ q
∂ x̄p ∂ x̄q
or,
∂xi ∂xj p q
gpq dx̄p dx̄q = ḡij dx̄ dx̄ , since gpq dx̄p dx̄q = gij dxi dxj
∂ x̄p ∂ x̄q
or,
∂xi ∂xj
 
gpq − ḡij p dx̄p dx̄q = 0
∂ x̄ ∂ x̄q
Since dx̄p and dx̄q are arbitrary vectors and dx̄p dx̄q ̸= 0

∂xi ∂xj
∴ gpq − ḡij =0
∂ x̄p ∂ x̄q
or,
∂xi ∂xj
gpq = ḡij
∂ x̄p ∂ x̄q
Which confirms that gij is a covariant tensor of rank two.
Proof of (iii) ○
22
Finally we have to show that gij is symmetric. We can write,
1 1
gij = (gij + gji ) + (gij − gji ) = Aij + Bij , → (4)
2 2
Where Aij = 12 (gij + gji ) is a symmetric tensor and Bij = 21 (gij − gji ) is a skew-
symmetric or antisymmetric tensor. Here we shall prove that Bij contributes nothing
to gij . Now gij dxi dxj = (Aij + Bij )dxi dxj → [by (4)]

∴ (gij − Aij )dxi dxj = Bij dxi dxj → (5)

Interchanging indices in Bij dxi dxj , we have

Bij dxi dxj = Bji dxj dxi

or,
Bij dxi dxj = −Bij dxi dxj [∵ Bij is antisymmetric]
or,
2Bij dxi dxj = 0
∴ Bij dxi dxj = 0 → (6)
Using (6) into (5), we get
(gij − Aij )dxi dxj = 0
or,
gij − Aij = 0, since dxi and dxj are arbitrary vectors.
∴ gij = Aij
But Aij is symmetric. Therefore gij is also symmetric. Thus gij is a covariant sym-
metric tensor of rank two and this tensor is called the fundamental covariant tensor
of rank two.
Problem 1 ○
23
Determine the metric tensor of first fundamental tensor in cylindrical coordinates.

Solution
Referred Cartesian coordinates the metric ds in three dimensional Euclidean space is
given by
ds2 = dx2 + dy 2 + dz 2 → (1)
Also we know that the relation the Cartesian coordinates (x,y,z) and the Cylinderical
coordinates (ρ, ϕ, z) are

x = ρ cos ϕ, y = ρ sin ϕ, z=z

∴ dx = −ρ sin ϕdϕ + cos ϕdρ

dy = ρ cos ϕdϕ + sin ϕdρ
dz = dz
 ○
24

∴ dx2 = ρ2 sin2 ϕdϕ2 + cos2 ϕdρ2 − 2ρ sin ϕ cos ϕdρdϕ → (2) (10)
2 2 2 2 2 2
dy = ρ cos ϕdϕ + sin ϕdρ + 2ρ sin ϕ cos ϕdρdϕ → (3) (11)
dz 2 = dz 2 → (4) (12)

Adding (2), (3) and (4), we get

ds2 = dx2 + dy 2 + dz 2 = ρ2 (sin2 ϕ + cos2 ϕ)dϕ2

+ (sin2 ϕ + cos2 ϕ)dρ2 + dz 2
∴ ds2 = dρ2 + ρ2 dϕ2 + dz 2 which is the metric in cylinderical coordinates.

Here g11 = 1, g22 = ρ2 , g33 = 1, g12 = g21 = 0, g31 = g13 = 0, g23 = g32 = 0. Hence
the metric tensor in cylinderical coordinate is
   
g11 g12 g13 1 0 0
[gij ] = g21 g22 g23  = 0 ρ2 0
g31 g32 g33 0 0 1

Problem 2
Determine the conjugate metric tensor or fundamental covariant tensor of rank two in
cylinderical coordinates.

Solution
The metric tensor in cylinderical coordinates is
   
g11 g12 g13 1 0 0
[gij ] = g21 g22 g23  = 0 ρ2 0
g31 g32 g33 0 0 1
 ○
25
1 0 0
∴ g = |gij | = 0 ρ2 0 = ρ2 ̸= 0
0 0 1
G(i,j)
We know that g ij = g

ρ2 0
G(1, 1) cofactor of g11 0 1 ρ2
g 11 = = = 2
= 2 =1
g g ρ ρ

G(2, 2) Cofactor of g22 1

g 22 = = = 2
g g ρ
G(3, 3) Cofactor of g33 ρ2
g 33 = = = 2 =1
g g ρ
G(1, 2) Cofactor of g12 0
g 12 = = = = 0 = g 21
g g g
G(1, 3) Cofactor of g13 0
g 13 = = = = 0 = g 31
g g g
Thus g ij = 0 if i ̸= j. Then g 23 = 0, g 32 = 0, g 12 = g 21 = 0, g 13 = g 31 = 0.
Hence the conjugate metric tensor in cylinderical coordinates is given by
 11
g 12 g 13
  
g 1 0 0
ij 21 22 23 1
[g ] = g g g  = 0 ρ2 0
g 31 g 32 g 33 0 0 1
Problem 3 ○
26
Find, g ij corresponding to the metric ds2 = 5(dx1 )2 + 3(dx2 )2 + 4(dx3 )2 − 6dx1 dx2 +
4dx2 dx3 .

Solution
The line element of ds is given by

ds2 = gij dxi dxj , i, j = 1, 2, . . . , N.

In our problem, i = j = 1, 2, 3.

∴ ds2 = g1j dx1 dxj + g2j dx2 dxj + g3j dx3 dxj
= g11 dx1 dx1 + g12 dx1 dx2 + g13 dx1 dx3
+ g21 dx2 dx1 + g22 dx2 dx2 + g23 dx2 dx3
+ g31 dx3 dx1 + g32 dx3 dx2 + g33 dx3 dx3 → (1) (13)

Given that,

ds2 = 5(dx1 )2 + 3(dx2 )2 + 4(dx3 )2 − 6dx1 dx2 + 4dx2 dx3 → (2)

Comparing (1) and (2), We have

g11 = 5, g22 = 3, g33 = 4, g12 = −3 = g21 ,

[∵ g12 , g21 are symmetric, 2g21 = −6 ∴ g21 = −3]
g32 = g23 = 2, [similarly 2g32 = 4, g32 = g23 = 2]
g13 = g31 = 0

Therefore,
g11 g12 g13 5 −3 0
g = |gij | = g21 g22 g23 = −3 3 2
g31 g32 g33 0 2 4
= 5(12 − 4) + 3(−12) = 40 − 36 = 4 ̸= 0
 ○
27
G(i,j) denotes the cofactors of gij . Therefore,

G(1, 1) (12 − 4) 8
g 11 = = = =2
g 4 4
G(1, 2) −(−12 − 0) 12
g 12 = = = =3
g 4 4
G(1, 3) −6 3
g 13 = = =−
g 4 2
G(2, 1) −(−12) 12
g 21 = = = =3
g 4 4
G(2, 2) 20
g 22 = = =5
g 4
G(2, 3) −(10) 5
g 23 = = =−
g 4 2
G(3, 1) −6 3
g 31 = = =−
g 4 2
G(3, 2) −(10) 5
g 32 = = =−
g 4 2
G(3, 3) 15 − 9 6 3
g 33 = = = =
g 4 4 2
Thus  11
g 12 g 13
  
g 2 3 −3/2
[g ij ] = g 21 g 22 g 23  =  3 5 −5/2
g 31 g 32 g 33 −3/2 −5/2 3/2
And also it will be shown that
[gij ][g ij ] = I
 ○
28
Now,   
5 −3 0 2 3 −3/2
ij
[gij ][g ] = −3 3 2  3 5 −5/2
0 2 4 −3/2 −5/2 3/2
 
(10 − 9) (15 − 15) (−15/2 + 15/2)
= (−6 + 9 − 3) (−9 + 15 − 5) (−9/2 − 15/2 + 6/2)
(6 − 6) (10 − 10) (−5 + 6)
 
1 0 0
= 0 1 0 = I, shown.
0 0 1

Problem 4
Find g and g ij corresponding to ds2 = 3(dx1 )2 + 2(dx2 )2 + 4(dx3 )2 − 6dx1 dx3 .

Solution
By the definition of the element ds in N-dimensional space, we have

ds2 = gij dxi dxj , i, j = 1, 2, . . . , N.

In our case i,j = 1,2,3; then

ds2 = g1j dx1 dxj + g2j dx2 dxj + g3j dx3 dxj
= g11 (dx1 )2 + g12 dx1 dx2 + g13 dx1 dx3
+ g21 dx2 dx1 + g22 (dx2 )2 + g23 dx2 dx3
+ g31 dx3 dx1 + g32 dx3 dx2 + g33 (dx3 )2 → (1) (14)

Given that ds2 = 3(dx1 )2 + 2(dx2 )2 + 4(dx3 )2 − 6dx1 dx3 → (2) Comparing (1) and
(2) We have

g11 = 3, g22 = 2, g33 = 4, g12 = g21 = 0, g13 = g31 = −3, g23 = g32 = 0.
 ○
29
g11 g12 g13 3 0 −3
∴ g = g21 g22 g23 = 0 2 0
g31 g32 g33 −3 0 4
= 3(8) + 0 + (−3)(6) = 24 − 18 = 6 ̸= 0 Ans
Let G(i,j) be the cofactors of gij

G(1, 1) 8 4
So, g 11 = = =
g 6 3
G(1, 2) −0
g 12 = = =0
g 6
G(1, 3) 6
g 13 = = =1
g 6
G(2, 1) −0
g 21 = = =0
g 6
G(2, 2) 12 − 9 3 1
g 22 = = = =
g 6 6 2
G(2, 3) −0
g 23 = = =0
g 6
G(3, 1) 6
g 31 = = =1
g 6
G(3, 2) −0
g 32 = = =0
g 6
G(3, 3) 6
g 33 = = =1
g 6
Thus  11
g 12 g 13
  
g 4/3 0 1
[g ij ] = g 21 g 22 g 23  =  0 1/2 0 Ans
g 31 g 32 g 33 1 0 1
Problem 5 ○
30
Find the metric of a Euclidean space reffered to spherical coordinates.

Solution
Reffered to Cartesian Coordinates, the metric in Euclidean space is given by

ds2 = dx2 + dy 2 + dz 2 → (1)

We know that the relation between the Cartesian coordinates (x,y,z) and the spherical
coordinates (r, θ, ϕ) is

x = r sin θ cos ϕ, y = r sin θ sin ϕ, z = r cos θ → (2)

∂x ∂y ∂z 
∴ = sin θ cos ϕ, = sin θ sin ϕ, = cos θ, 
∂r ∂r ∂r




∂x ∂y ∂z 
= r cos θ cos ϕ, = r cos θ sin ϕ, = −r sin θ, → (3)
∂θ ∂θ ∂θ 

∂x ∂y ∂z


= −r sin θ sin ϕ, = r sin θ cos ϕ, =0


∂ϕ ∂ϕ ∂ϕ
Let (x1 , x2 , x3 ) be the rectangular Cartesian coordinates and (x̄1 , x̄2 , x̄3 ) be the spher-
ical coordinates of a point in the Euclidean space.

x1 = x, x2 = y, x3 = z
)
Then We have → (4)
and x̄1 = r, x̄2 = θ, x̄3 = ϕ

The metric in Euclidean space for three dimensional space is given by

ds2 = gij dxi dxj , i, j = 1, 2, 3 → (5)

Comparing (5) with (1), we have

g11 = g22 = g33 = 1, g12 = g21 = 0, g13 = g31 = 0, g23 = g32 = 0

 ○
31
From the law of transformations, We have
∂xp ∂xq
ḡij = gpq , p, q = 1, 2, 3
∂ x̄i ∂ x̄j
p p
∂x ∂x
= gpp , since gpq = 0 if p ̸= q.
∂ x̄i ∂ x̄j
3
X ∂xp ∂xp
= gpp
p=1
∂ x̄i ∂ x̄j
3
X ∂xp ∂xp
= · 1, since gpp = 1, for p = 1, 2, 3.
p=1
∂ x̄i ∂ x̄j

∂x1 ∂x1 ∂x2 ∂x2 ∂x3 ∂x3

= + + → (6)
∂ x̄i ∂ x̄j ∂ x̄i ∂ x̄j ∂ x̄i ∂ x̄j

(1) Putting i=j=1 into (6), We get

2  2 2  3 2
∂x1

∂x ∂x
ḡ11 = + +
∂ x̄1 ∂ x̄1 ∂ x̄1
 2  2  2
∂x ∂y ∂z
= + + , using (4)
∂r ∂r ∂r
= (sin θ cos ϕ)2 + (sin θ sin ϕ)2 + cos2 θ
= sin2 θ(cos2 ϕ + sin2 ϕ) + cos2 θ
= sin2 θ + cos2 θ = 1
∴ ḡ11 = 1

(2) putting i=j=2 into (6), We get

2  2 2  3 2
∂x1

∂x ∂x
ḡ22 = + +
∂ x̄2 ∂ x̄2 ∂ x̄2
 2  2  2
∂x ∂y ∂z
= + +
∂θ ∂θ ∂θ
= (r cos θ cos ϕ)2 + (r cos θ sin ϕ)2 + (−r sin θ)2
= r2 cos2 θ(cos2 ϕ + sin2 ϕ) + r2 sin2 θ
= r2 (cos2 θ + sin2 θ) = r2
∴ ḡ22 = r2
 ○
32
(3) putting i=j=3 into (6), We get
2  2 2  3 2
∂x1

∂x ∂x
ḡ33 = + +
∂ x̄3 ∂ x̄3 ∂ x̄3
 2  2  2
∂x ∂y ∂z
= + +
∂ϕ ∂ϕ ∂ϕ
= (−r sin θ sin ϕ)2 + (r sin θ cos ϕ)2 + 02
= r2 sin2 θ(sin2 ϕ + cos2 ϕ)
∴ ḡ33 = r2 sin2 θ

(4) putting i=1, j=2 into (6), We get

∂x1 ∂x1 ∂x2 ∂x2 ∂x3 ∂x3

ḡ12 = 1 2
+ 1 2
+
∂ x̄ ∂ x̄ ∂ x̄ ∂ x̄ ∂ x̄1 ∂ x̄2
∂x ∂x ∂y ∂y ∂z ∂z
= + +
∂r ∂θ ∂r ∂θ ∂r ∂θ
= (sin θ cos ϕ)(r cos θ cos ϕ) + (sin θ sin ϕ)(r cos θ sin ϕ) + cos θ(−r sin θ)
= r sin θ cos θ(cos2 ϕ + sin2 ϕ) − r sin θ cos θ = 0
∴ ḡ12 = ḡ21 = 0

Similarly, we can show that g13 = g31 = 0, g23 = g32 = 0. Reffered to spherical
coordinates, the metric are the line element is given by

ds2 = ḡij dx̄i dx̄j , Since ḡij = 0 for i ̸= j

= ḡ11 (dx̄ ) + ḡ22 (dx̄2 )2 + ḡ33 (dx̄3 )2
1 2

= 1(dr)2 + r2 (dθ)2 + r2 sin2 θ(dϕ)2

∴ ds2 = dr2 + r2 dθ2 + r2 sin2 θdϕ2 which is the metric in spherical coordinates.

Question
Hence find the metric tensor of fundamental tensor in spherical coordinates.
 ○
33
Hence the metric tensor or fundamental metric tensor in spherical coordinates is
   
ḡ11 ḡ12 ḡ13 1 0 0
[ḡij ] = ḡ21 ḡ22 ḡ23  = 0 r2 0  Ans
ḡ31 ḡ32 ḡ33 0 0 r2 sin2 θ

Question
Determine the conjugate metric tensor (or the second fundamental covariant tensor)
in spherical coordinates.

Solution
The metric tensor in spherical coordinates is
   
g11 g12 g13 1 0 0
[gij ] = g21 g22 g23  = 0 r2 0 
g31 g32 g33 0 0 r2 sin2 θ

Let
g11 g12 g13 1 0 0
g = |gij | = g21 g22 g23 = 0 r2 0 = r4 sin2 θ ̸= 0
g31 g32 g33 0 0 r2 sin2 θ
G(i,j)
Now, g ij = g
, where G(i,j) is the cofactors of gij .
 ○
34
Now,  11
g 12 g 13

  g
ij G(i, j)
[g ] = = g 21 g 22 g 23  ,
g
g 31 g 32 g 33
where
G(1, 1) r4 sin2 θ
g 11 = = =1
g r4 sin2 θ
G(2, 2) r2 sin2 θ 1
g 22 = = 4 2 = 2
g r sin θ r
G(3, 3) r2 1
g 33 = = 4 2 = 2 2
g r sin θ r sin θ
G(1, 2) 0
g 12 = = = 0 = g 21
g g
Similarly, we can show that

g 13 = g 31 = 0, g 23 = g 32 = 0.

Hence the conjugate (or reciprocal) metric tensor in spherical coordinates is

 
1 0 0
ij
[g ] = 0 r12 0 
1
0 0 r 2 sin2 θ

and the conjugate metric in spherical coordinates is

1 2 1
ds2 = dr2 + dθ + 2 2 dϕ2
r2 r sin θ
Christoffel’s Symbols or brackets ○
35
The Christoffel’s symbols of the first kind are defined and denoted as follows:
 
1 ∂gik ∂gjk ∂gij
[ij, k] = Γij,k = + −
2 ∂xj ∂xi ∂xk

We shall write Γij,k = 21 (gik,j + gjk,i − gij,k ) The Christoffel’s symbols of the second
kind are defined and denoted as follows:

{kij } = Γkij = g kl Γij,l = g kl [ij, l]

Problem
Prove that
(a) [ij, k] = [ji, k]
(b) {kij } = {kji }
(c) [ij, h] = gkh {kij }
∂gik
(d) [ij, k] + [jk, i] = ∂xj

Proof (a)
By definition of Christoffel’s symbol of the first kind, we have
 
1 ∂gik ∂gjk ∂gij
[ij, k] = + − → (1)
2 ∂xj ∂xi ∂xk

Interchanging i and j into (1), we get

 
1 ∂gjk ∂gik ∂gji
[ji, k] = + −
2 ∂xi ∂xj ∂xk
 
1 ∂gik ∂gjk ∂gij
= + − [since gij is symmetric i.e. gij = gji ]
2 ∂xj ∂xi ∂xk

Hence, [ij, k] = [ji, k], proved.

Proof (b) ○
36
By definition of the Christoffel’s symbol of the second kind, we have

{kij } = g kh [ij, h] → (2)

Interchanging i and j into (2), we get

{kji } = g kh [ji, h]
= g kh [ij, h], since [i, j, h] = [ji, h]
= {kij }

Hence {kij } = {kji } → proved

(c)
By definition of Christoffel’s symbol of the second kind, we have

{kij } = g kh [ij, h] → (3)

Multiplying both sides of (3) by gkm , we get

gkm {kij } = gkm g kh [ij, h]

h
= δm [ij, h]
= [ij, m] → (4)

∴ gkm {kij } = [ij, m]

Replacing m by h into (4), we get

[ij, h] = gkh {kij }, proved

Problem ○
37
(d)
By definition of the Christoffel symbol of the first kind, we have
   
1 ∂gik ∂gjk ∂gij 1 ∂gki ∂gji ∂gjk
[ij, k] + [jk, i] = + − + + −
2 ∂xj ∂xi ∂xk 2 ∂xj ∂xk ∂xi
 
1 ∂gik ∂gjk ∂gij ∂gki ∂gji ∂gjk
= + − + + −
2 ∂xj ∂xi ∂xk ∂xj ∂xk ∂xi
1 ∂gik ∂gik
= ·2 j =
2 ∂x ∂xj
∂gik
∴ [ij, k] + [jk, i] = , proved
∂xj

Problem
√
Prove that {iij } = ∂
∂xj
log g

Proof
In matrix form gij may be represented by
 
g11 g12 ... g1N
 g21 g22 ... g2N 
[gij ] =  .
 
.. .. .. 
 .. . . . 
gN 1 gN 2 ... gN N

If g is the determinant of [gij ], then we have

g11 g12 ... g1N

g21 g22 ... g2N
g = |gij | = . .. .. ..
.. . . .
gN 1 gN 2 ... gN N

Differentiating both sides w.r. to xj , we get

∂g11 ∂g12 ∂g1N
∂xj ∂xj
... ∂xj
g11 g12 ... g1N
∂g g21 g22 ... g2N g21 g22 ... g2N
= .. .. .. .. + ··· + .. .. .. ..
∂xj . . . . . . . .
∂gN 1 ∂gN 2 ∂gN N
gN 1 gN 2 ... gN N ∂xj ∂xj
... ∂xj
Proof continued ○
38
clearly,
∂g11
Cofactor of = Cofactor of g11 in g = gg 11
∂xj
∂g12
Cofactor of = Cofactor of g12 in g = gg 12
∂xj
...
∂gN N
Cofactor of = Cofactor of gN N in g = gg N N
∂xj
Therefore,
∂g ∂g11 11 ∂g12 12 ∂gN N N N
= gg + gg + · · · + gg
∂xj ∂xj ∂xj ∂xj
∂gik
= gg ik , i, k = 1, 2, . . . , n
∂xj
∂gik
= gg ik ([ij, k] + [kj, i]) ∵ = [ij, k] + [kj, i]
∂xj
= gg ik [ij, k] + gg ik [kj, i]
= g{iij } + g{kkj }
= g{iij } + g{iij } [replacing k by i]
= 2g{iij }

∂ √ 1 ∂ √
∴ (log g) = √ ( g)
∂x g ∂x
1 ∂g √ 1 1 ∂g
or, {iij } = 2g ∂xj
= ∂
∂xj
(log g), proved = √ √
g 2 g ∂x
1 ∂g
=
2g ∂x
Problem ○
39
Problem
Prove that
∂g ik
= −g hk {ihj } − g hi {khj }
∂xj

Proof
We write
∂ ∂
(gpm g mi ) = (δpi ),
∂xj ∂xj
Where δpi is kronecker delta.

∂gpm mi ∂g mi
or, j
g + gpm =0
∂x ∂xj
∂gpm ∂g mi
∴ −g mi = g pm → (1)
∂xj ∂xj
Multiplying both sides of (1) by g pk , we get

∂gpm ∂g mi
−g pk g mi = g pk
gpm
∂xj ∂xj
mi
∂g pm ∂g
or, − g pk g mi = δmk
∂xj ∂xj
ki
∂gpm ∂g ∂gis
or, − g pk g mi = ∵ = [ik, j] + [jk, i] ∵ {ijk } = g il [jk, l]
∂xj ∂xj ∂xk
∂g ki
or, − g pk g mi ([pj, m] + [mj, p]) =
∂xj
∂g ki
or, − g pk {ipj } − g mi {kmj } =
∂xj
Replacing dummy indices m and p by h, we get

∂g ki
= −g hk {ihj } − g hi {khj }, proved
∂xj

Fundamental lemma of tensor calculus or lemma of Christof-

fel’s is given by
∂ 2 xρ
= bργ {r̄ik } − bα β
i bk {ραβ },
∂ x̄i ∂ x̄k
where {r̄ik } = ḡ γl [ik, l]
Example 1 ○
40
Example 1
Evaluate the Christoffel’s symbols of first kind in (a) rectangular (b) cylindrical (c)
spherical coordinates.

Solution
(a) In three dimensional space in rectangular coordinates let x1 = x, x2 = y, x3 = z.
Then the metric in three dimensional space is given by ds2 = gij dxi dxj , i, j, k =
1, 2, 3.

∴ ds2 =g11 (dx1 )2 + g12 dx1 dx2 + g13 dx1 dx3 + g21 dx2 dx1
+ g22 (dx2 )2 + g23 dx2 dx3 + g31 dx3 dx1 + g32 dx3 dx2 + g33 (dx3 )2 → (1)

Also the metric in three dimensional rectangular coordinates system is given by

ds2 = dx2 + dy 2 + dz 2 = (dx1 )2 + (dx2 )2 + (dx3 )2 → (2)

Comparing (1) and (2), we get

g11 = g22 = g33 = 1, g12 = g21 = 0, g13 = g31 = 0, g23 = g32 = 0.

Now the Christoffel’s symbols of the first kind are given by

 
1 ∂gjk ∂gik ∂gij
[ij, k] = + − , i, j, k = 1, 2, 3. → (3)
2 ∂xi ∂xj ∂xk

putting i = j = k = 1 into (3), we get

 
1 ∂g11 ∂g11 ∂g11 1 ∂g11 1 ∂
[11, 1] = + − = = (1) = 0
2 ∂x1 ∂x1 ∂x1 2 ∂x1 2 ∂x

∴ [11, 1] = 0
putting i = j = 2, k = 2 into (3), we get
 
1 ∂g22 ∂g22 ∂g22 1 ∂
[22, 2] = + − = (1) = 0
2 ∂x2 ∂x2 ∂x2 2 ∂y

∴ [22, 2] = 0, Similarly [33, 3] = 0

Solution continued ○
41
Again putting i=j=1, k=2 into (3), we get
 
1 ∂g12 ∂g12 ∂g11 1 ∂
[11, 2] = + − = {0 + 0 − (1)} = 0
2 ∂x1 ∂x1 ∂x2 2 ∂y

∴ [11, 2] = 0
Similarly, assigning i=j=k = 1, 2, 3 into (3), we can easily show that all other Christof-
fel’s Symbols of first kind
 in rectangular coordinates
 are zero.
∂gjk ∂gij ∂gij
Hence, [ij, k] = 12 ∂g ik
∂xj
+ ∂xi
− ∂xk
= 0, since ∂xk
= 0 as gij = constant.

(b)
We know that the metric in cylindrical coordinates is given by

ds2 = dρ2 + ρ2 dϕ2 + dz 2 → (1)

Also in three dimensional space for cylindrical coordinates we have x1 = ρ, x2 =

ϕ, x3 = z and the metric in three dimensional space is given by ds2 = gij dxi dxj , i, j =
1, 2, 3 → (2)

⇒ ds2 = g11 (dx1 )2 + g22 (dx2 )2 + g33 (dx3 )2 + 2g12 dx1 dx2 + 2g23 dx2 dx3 + 2g13 dx1 dx3 [since gij = gji ]
∴ ds2 = g11 dρ2 + g22 dϕ2 + g33 dz 2 + 2g12 dρdϕ + 2g23 dϕdz + 2g13 dρdz → (3)

Comparing (1) and (3), we obtain

g11 = 1, g22 = ρ2 , g33 = 1, g12 = g21 = 0, g23 = g32 = 0, g13 = g31 = 0.

Solution (b) ○
42
Now the Christoffel’s symbol first kind is given by
 
1 ∂gjk ∂gik ∂gij
[ij, k] = + − , i, j, k = 1, 2, 3. → (4)
2 ∂xi ∂xj ∂xk

If i=j=k=1, then (4) gives

 
1 ∂g11 ∂g11 ∂g11 1 ∂
[11, 1] = + − = (1) = 0
2 ∂x1 ∂x1 ∂x1 2 ∂ρ

Similarly from (4), we obtain

 
1 ∂g22 ∂g22 ∂g22 1 ∂ 2
[22, 2] = + − = (ρ ) = 0,
2 ∂x2 ∂x2 ∂x2 2 ∂ϕ
 
1 ∂g33 ∂g33 ∂g33 1 ∂
[33, 3] = + − = (1) = 0
2 ∂x3 ∂x3 ∂x3 2 ∂z
 
1 ∂g22 ∂g12 ∂g12 1 ∂ 2 1
[12, 2] = + − = (ρ ) = · 2ρ = ρ
2 ∂x1 ∂x2 ∂x2 2 ∂ρ 2
∴ [12, 2] = [21, 2] = ρ
   
1 ∂g21 ∂g21 ∂g22 1 ∂ ∂ ∂ 2 1
[22, 1] = + − = (0) + (0) − (ρ ) = − · 2ρ = −ρ
2 ∂x2 ∂x2 ∂x1 2 ∂ρ ∂ϕ ∂ρ 2
∴ [22, 1] = −ρ
 
1 ∂g12 ∂g12 ∂g11
[11, 2] = + − =0
2 ∂x1 ∂x1 ∂x2
 
1 ∂g23 ∂g13 ∂g12
[12, 3] = + − =0
2 ∂x1 ∂x2 ∂x3
Similarly we can show that all other Christoffel’s symbols of first kind in cylindrical
coordinates are zero.
Solution (c) ○
43
(c)
We know that the metric in spherical coordinates is given by

ds2 = dr2 + r2 dθ2 + r2 sin2 θdϕ2 → (1)

Also in three dimensional space for spherical coordinates we have x1 = r, x2 = θ, x3 =

ϕ. The metric for three dimensional space is given by ds2 = gij dxi dxj , i, j =
1, 2, 3 → (2)

⇒ ds2 = g11 (dx1 )2 + g22 (dx2 )2 + g33 (dx3 )2 + 2g12 dx1 dx2 + 2g13 dx1 dx3 + 2g23 dx2 dx3
⇒ ds2 = g11 (dr)2 + g22 (dθ)2 + g33 (dϕ)2 + 2g12 drdθ + 2g13 drdϕ + 2g23 dθdϕ → (3)

Comparing (1) and (3), we have

g11 = 1, g22 = r2 , g33 = r2 sin2 θ, g12 = g21 = 0, g13 = g31 = 0, g23 = g32 = 0

Now the Christoffel’s symbols of first kind are given by

 
1 ∂gjk ∂gik ∂gij
[ij, k] = + − , i, j, k = 1, 2, 3. → (4)
2 ∂xi ∂xj ∂xk

By putting the different values of i,j,k from 1 to 3 into (4), we obtain

 
1 ∂g22 ∂g12 ∂g12 1 ∂ 2 1
[12, 2] = + − = (r ) = 2r
2 ∂x1 ∂x2 ∂x2 2 ∂r 2

∴ [12, 2] = r = [21, 2]
   
1 ∂g21 ∂g21 ∂g22 1 ∂ 2
[22, 1] = + − = − (r ) = −r
2 ∂x2 ∂x2 ∂x1 2 ∂r
∴ [22, 1] = −r
Solution continued ○
44
 
1 ∂g33 ∂g13 ∂g13 1 ∂ 2 2
[13, 3] = + − = (r sin θ)
2 ∂x1 ∂x3 ∂x3 2 ∂r
1
= 2r sin2 θ = r sin2 θ
2
∴ [13, 3] = [31, 3] = r sin2 θ
 
1 ∂g31 ∂g31 ∂g33 1 ∂ 2 2
[33, 1] = + − =− (r sin θ)
2 ∂x3 ∂x3 ∂x1 2 ∂r
1
= − 2r sin2 θ = −r sin2 θ
2
 
1 ∂g33 ∂g23 ∂g23 1 ∂ 2 2
[23, 3] = + − = (r sin θ)
2 ∂x2 ∂x3 ∂x3 2 ∂θ
1 1
= r2 2 sin θ cos θ = r2 sin 2θ
2 2
1 2
∴ [23, 3] = [32, 3] = r sin 2θ
2
 
1 ∂g32 ∂g32 ∂g33 1 ∂ 2 2
[33, 2] = + − =− (r sin θ)
2 ∂x3 ∂x3 ∂x2 2 ∂θ
1 1
= − r2 2 sin θ cos θ = − r2 sin 2θ
2 2
1 2
∴ [33, 2] = − r sin 2θ
2
Similarly, we can show that for all other value of i,j,k from 1 to 3 the Christoffel’s
symbol of first kind in spherical coordinates are zero.
Example 2 ○
45
Example 2
Determine the Christoffel’s symbols of the second kind in (a) rectangular coordinates
(b) cylindrical coordinates (c) spherical coordinates.

Solution
We know that for orthogonal are rectangular coordinates, we have gij = 0 if i ̸= j. So,
we can use the results of first kind in this case.

(a) In rectangular coordinates

gii = 1 so that g ii = g1ii = 1 and {kij } = 0. As for example, by the definition of
Christoffel’s symbols of the second kind, we have

{kij } = g kl [ij, l]

= g k1 [ij, 1] + g k2 [ij, 2] + g k3 [ij, 3] → (1)

By putting k=i=j=1 into (1), we have

{111 } = g 11 [11, 1] + g 12 [11, 2] + g 13 [11, 3]

=1·0+0·0+0·0=0
∴ {111 } = 0
Similarly, by putting the different values of i,j,k from 1 to 3 into (1), we can show that
all other Christoffel’s symbols of second kind in rectangular system coordinates are
zero.
Solution (b) ○
46
(b) In cylindrical coordinates, We have
x1 = ρ, x2 = ϕ, x3 = z
2
and g11 = 1, g22 = ρ , g33 = 1, gij = 0 if i ̸= j.
Therefore,
1
g 11 = =1
g11
1 1
g 22 = = 2
g22 ρ
1
g 33 = =1
g33

and g ij = 0 if i ̸= j.
Now the Christoffel’s symbol of the second kind are given by

{kij } = g kl [ij, l]
= g k1 [ij, 1] + g k2 [ij, 2] + g k3 [ij, 3] → (1)

The only non-zero Christoffel’s symbols of the second kind can occur when i=2.
Let j=2, k=1 then (1) gives

{122 } = g 11 [22, 1] + 0 + 0 = 1 · (−ρ) = −ρ

Again let j=1, k=2 then (1) gives

1 1
{221 } = 0 + g 22 [21, 2] + 0 = ·ρ=
ρ2 ρ
1
∴ {221 } = {212 } =
ρ
We can show that other Christoffel’s symbols of the second kind will be zero.
Solution (c) ○
47
(c) In spherical coordinates we have
x1 = r, x2 = θ, x3 = ϕ
Also we know that

g11 = 1, g22 = r2 , g33 = r2 sin2 θ, gij = 0, i ̸= j

and
1 1
g 11 = 1, g 22 =, g 33 = 2 2 , g ij = 0, i ̸= j
r2 r sin θ
The Christoffel’s symbol of the second kind are given by

{kij } = g kl [ij, k]
= g k1 [ij, 1] + g k2 [ij, 2] + g k3 [ij, 3] → (1)

The Christoffel’s symbols can occur when i=2 or i=3 which are
 
1 ∂g22 1 ∂ 2
{122 } = g 11 − 1 =− (r ) = −r
2 ∂x 2 ∂r

∴ {122 } = −r
1 ∂g22 1 1 ∂ 2 1 1 1
{212 } = g 22 = 2 · (r ) = 2 · · 2r =
2 ∂x1 r 2 ∂r r 2 r
2 2 1
∴ {12 } = {21 } =
r
   
1 11 1 ∂g33 1 ∂ 2 2 1
{33 } = g − 1 = − (r sin θ) = − 2r sin2 θ = −r sin2 θ
2 ∂x 2 ∂r 2
∴ {133 } = −r sin2 θ
Solution continued ○
48
 
1 ∂g33 1 1 ∂ 2 2
{233 } = g 22 − =− 2 (r sin θ)
2 ∂x2 2 r ∂θ
1
= − 2 · r2 2 sin θ cos θ = − sin θ cos θ
2r
∴ {233 } = − sin θ cos θ
1 ∂g33 1 1 ∂ 2 2
{331 } = g 33 = (r sin θ)
2 ∂x1 2 r2 sin2 θ ∂r
1 1
= 2 2 sin2 θ · 2r =
2r sin θ r
3 3 1
∴ {31 } = {13 } =
r
3 33 1 ∂g33 1 1 ∂ 2 2
{23 } = g = (r sin θ)
2 ∂x2 2 r2 sin2 θ ∂θ
1 cos θ
= 2 2 · r2 2 sin θ cos θ = = cot θ
2r sin θ sin θ
∴ {323 } = {332 } = cot θ
and all others Christoffel’s symbols of the second are zero.
Tensor forms ○
49
Tensor form of gradient, divergence, Curl and Laplacian
1. Gradient: If ϕ is a scalar or invariant, then the gradient of ϕ is defined by ∇ϕ
∂ϕ
= grad ϕ = ϕ,p = , where ϕ,p is the
∂xp
Covariant derivative of ϕ with respect to xp .
2. Divergence: The divergence of Ap is the contraction of its covariant derivative
w.r. to xq that is the contraction of Ap,q ; then

1 ∂ √ k
div Ap = Ap,p = √ ( gA ).
g ∂xk

∂A ∂A
3. Curl: The curl of Ap is Ap,q − Aq,p = ∂xqp − ∂xpq , which is a tensor of rank two.
The curl is defined as Curl Ap = Ap,q − Aq,p .
4. Laplacian: The laplacian of ϕ is the divergence of a grad ϕ. as
   
∂ϕ 1 ∂ √ jk ∂ϕ
∇2 ϕ = div(ϕ,p ) = div = √ gg
∂xp g ∂xj ∂xk
√ √ √
If g < 0, g must be replaced by −g. If g > 0 or g < 0, g can be included
p
by writing |g|.
Geodesics ○
50
Geodesics
The distance s between two points t1 and t2 on a curve xr = xr (t) in a Riemannian
space is given by
Z t2 r
dxp dxq
s= gpq dt.
t1 dt dt
That curve in the space which makes s the distance a minimum is called a geodesic of
the space. By use of the calculus of variants the geodesics are found the differential
form equation
d2 xr dxp dxq
+ {rpq } = 0, where
ds2 ds ds
s is the arc length parameter.
As for examples the geodesic on a plane are straight lines where as the geodesics
on a sphere are arcs of great circles.
Covariant differentiation ○
51
Covariant differentiation
We know that the derivative of tensor is not in general a tensor. But we can form tensor
out of the ordinary derivative with the help of Christoffel lemma. So the rank of the
tensor obtained is one more than the rank of the original tensor by the covariant order
and it is called the covariant derivative of the original tensor. Hence the operation is
called Covariant differentiation.

Covariant differentiation of a covariant tensor of rank one

We know that
∂xα ∂xα
Āi = i
Aα = bαi Aα , bα
i =
∂ x̄ ∂ x̄i
∂ Āi ∂
∴ = (bα
i Aα )
∂ x̄k ∂ x̄k

∂xβ ∂xα
 
∂Aα ∂
= bα
i + Aα
∂x ∂ x̄k
β ∂ x̄k ∂ x̄i
∂Aα ∂ 2 xα
= bα β
i bk β
+ Aα → (1)
∂x ∂ x̄k ∂ x̄i
∂ 2 xρ
Now, A
∂ x̄k ∂ x̄i ρ
[Put α = ρ]
 
= bργ {γ̄ki } − bα β
k bi {ραβ } Aρ [using Christoffel lemma]

= bργ Aρ {γ̄ki } − bα β
k bi {ραβ }Aρ

= Āγ {γ̄ki } − bα β
k bi {ραβ }Aρ → (2)

Using (2) into (1), we get

Covariant derivatives ○
52
∂ Āi β ∂Aα
= bα
i bk + Āγ {γ̄ki } − bα β
k bi {ραβ }Aρ
∂ x̄k ∂xβ
 
∂ Āi α β ∂Aα
⇒ − Āγ {γ̄ ki } = bi b k − {ρ αβ }Aρ
∂ x̄k ∂xβ
∴ Āi,k = bα β
i bk Aα,β , → (3)
Where
∂ Āi
Āi,k = − Āγ {γ̄ki } → (3.1)
∂ x̄k
∂Aα
and Aα,β = − Aρ {ραβ } → (3.2)
∂xβ
Equation (3) is a mixed tensor of rank two and Aα,β is called the covariant derivative
of Aα of rank one with respect to xβ . Also Āi,k is called the covariant derivative of
Āi of rank one with respect to x̄k .

Covariant derivative of contravariant tensor of rank one.

Similar way as before, we obtain

Āi,k = aiα bβk Aα

,β , → (4)

Where
∂ Āi
Āi,k = + Āγ {ı̄γk } → (4.1)
∂ x̄k
∂Aα
and Aα
,β = + Aρ {αρβ } → (4.2)
∂xβ
Higher order tensors ○
53
Covariant derivative of a covariant tensor of rank two
Similar way We obtain
Āik,l = bα β λ
i bk bl Aαβ,λ → (5)
Where
∂ Āik
Āik,l = − Ākγ {γ̄il } − Āiγ {γ̄kl } → (5.1)
∂ x̄l
∂Aαβ
and Aαβ,λ = − Aρβ {ραλ } − Aαρ {ρβλ } → (5.2)
∂xλ

Covariant derivative of a contravariant tensor of rank two.

Similarly we get
Āik i k λ αβ
,l = aα aβ bl A,λ → (6)
Where
∂ Āik
Āik
,l = + Āγk {ı̄γl } + Āiγ {k̄γl } → (6.1)
∂ x̄l
∂Aαβ
and Aαβ
,λ = + Aρβ {αρλ } + Aαρ {βρλ } → (6.2)
∂xλ

Covariant derivative of one covariant and one contravariant

tensor (mixed tensor of rank two)
Similarly, we obtain
Āik,l = aiα bβk bλl Aα
β,λ → (7)
Where
∂ Āik
Āik,l = + Āγk {ı̄γl } − Āiγ {γ̄kl } → (7.1)
∂ x̄l
∂Aα β
and Aα
β,λ = + Aρβ {αρλ } − Aα ρ {ρβλ } → (7.2)
∂xλ
Examples ○
54
Ex. 1)
Let Ai and Bi be two covariant tensor. then show that
(i) (Ai + Bi ),k = Ai,k + Bi,k
(ii) (Ai − Bi ),k = Ai,k − Bi,k

Proof. (i)
Let Ai + Bi = Ci → (1)
By theorem of covariant derivative, we have
∂Aα
Aα,β = − Aρ {ραβ } → (2)
∂xβ
Differentiating Ci with respect to xk by using (2), we get

∂Ci
Ci,k = − Cl {lik }
∂xk
∂
= (Ai + Bi ) − (Al + Bl ){lik }
∂xk
∂Ai ∂Bi
= k
+ − Al {lik } − Bl {lik }
∂x
 ∂xk   
∂Ai l ∂Bi l
= − Al {ik } + − Bl {ik }
∂xk ∂xk
= Ai,k + Bi,k

Hence (Ai + Bi ),k = Ai,k + Bi,k

Ex. 2)
Let Ai and B i be two contravariant tensor. Then show that
(i) (Ai + B i ),k = Ai,k + B,k
i

(ii) (Ai − B i ),k = Ai,k − B,k

i
Examples continued ○
55
Proof. (i)
Let Ai − B i = C i → (1)
We have
∂Aα
Aα
,β = + A ρ {α
ρβ } → (2)
∂xβ
Differentiating C i w.r. to xk using (2), we get

i ∂C i
C,k = + C l {ikl }
∂xk
∂
= (Ai − B i ) + (Al − B l ){ikl }
∂xk
∂Ai ∂B i
   
l i l i
= + A {kl } − + B {kl }
∂xk ∂xk
= Ai,k − B,k
i

i
Hence (Ai − B i ),k = Ai,k − B,k

Ex. 3
Let Aij and Bji are two mixed tensors of rank two. Then show that
i
(i) (Aij + Bji ),k = Aij,k + Bj,k
i
(ii) (Aij − Bji ),k = Aij,k − Bj,k

Hints
By theorem we have
∂Aαβ
Aα
β,λ = − Aα ρ ρ α
ρ {βλ } + Aβ {ρλ }
∂xλ
and using this formula we can prove (i) and (ii).
Metric tensor properties ○
56
Ex. 1)
Prove that the covariant derivatives of the tensor gij , g ij and δji all vanish identically.
ij i
That is gij,k = 0, g,k = 0, δj,k =0
(i) By using the theorem of covariant derivative, we have

∂gij
gij,k = − glj {lik } − gil {ljk }
∂xk
∂gij
= − [ik, j] − [jk, i]
∂xk
∂gij
= − ([ik, j] + [jk, i])
∂xk
∂gij ∂gij
= −
∂xk ∂xk
∴ gij,k = 0, proved

ij ij
(ii) g,k = ∂g
∂xk
+ g il {jlk } + g jl {ilk } → (1)
Since, we know that
∂g ik
= −g kh {ihj } − g ih {khj }
∂xj
Interchanging j for k and writing h for l, we get

∂g ij
= −g il {jlk } − g jl {ilk }
∂xk
Now (1) gives
ij
g,k = −g jl {ilk } − g il {jlk } + g il {jlk } + g jl {ilk } = 0
ij
Hence g,k =0
Kronecker delta ○
57
(iii) By theorem of covariant derivative of mixed tensor, we have

∂Aij
Aij,k = − Ahj {ihk } + Aih {hjk }
∂xk

i ∂δji
∴ δj,k = − δjh {ihk } + δhi {hjk }
∂xk
∂δji
= − {ijk } + {ijk } → (1)
∂xk
(
i 1, i=j ∂δji
since δj = ∴ ∂xk
=0
0, i ̸= j
Now, from (1), we have
i
δj,k = −{ijk } + {ijk } = 0
i
Hence δj,k =0
Hence covariant derivative of gij , g ij , δji vanish identically.
Riemann Christoffel tensor ○
58
Riemann Christoffel’s tensor or curvature tensor
Let Ai be an arbitrary covariant vector. Then its covariant derivative w.r. to xj is
given by
∂Ai
Ai,j = − A α {α
ij } → (1)
∂xj
k
Again taking covariant derivative of Ai,j w.r. to x , we have
∂Ai,j
Ai,j,k = − Aα,j {α α
ik } − Ai,α {jk } → (2)
∂xk
Putting the values of Ai,j from (1) into (2), we get
     
∂ ∂Ai α ∂Aα β α ∂Ai β
Ai,jk = − A α { ij } − − A β { αj } { ik } − − A β { iα } {α
jk }
∂xk ∂xj ∂xj ∂xα

∂ 2 Ai ∂Aα α ∂ α ∂Aα ∂Ai

= − {ij } − Aα k {α
ij } − {ik } + {α β α
ik }{αj }Aβ − {jk } + {α β
jk }{iα }Aβ → (3)
∂xk ∂xj ∂xk ∂x ∂xj ∂xα
Interchanging j and k into (3), we get

∂ 2 Ai ∂Aα α ∂ α ∂Aα ∂Ai

Ai,kj = − {ik } − Aα j {α
ik } − {ij } + {α β α
ij }{αk }Aβ − {kj } + {α β
kj }{iα }Aβ → (4)
∂xj ∂xk ∂xj ∂x ∂xk ∂xα
Subtracting (4) from (3), we obtain
∂ α ∂
Ai,jk − Ai,kj = Aα {ik } − Aα k {α α β α β
ij } + {jk }{iα }Aβ − {ij }{αk }Aβ → (5)
∂xj ∂x
Curvature tensor ○
59
Interchanging β and α in the 2nd and 4th term into (5), we get
∂ α ∂ α
Ai,jk − Ai,kj = − {ij }Aα + {α β
ik }{αj }Aβ + {ik }Aα − {α β
ij }{βk }Aα
∂xk ∂xj
 
∂ α ∂ α β α β α
= Aα { ik } − { ij } + {ik }{βj } − {ij }{βk } → (6)
∂xj ∂xk
α
∴ Ai,jk − Ai,kj = Aα Rijk ,
α ∂ α ∂ α β α β α
Where Rijk = ∂x j {ik } − ∂xk {ij } + {ik }{βj } − {ij }{βk } → (7)
Since Ai,jk is a tensor of third order, so that left hand side of (6) is a tensor of third
order (being difference of two tensors of same rank or type). But Aα is an arbitrary
α
covariant vector and hence it follows from quotient law that Rijk is a mixed tensor
α
of fourth order. This tensor Rijk is called Riemann Christoffel tensor (or curvature
tensor).
α
The covariant tensor of Rhijk of the fourth order defined by Rhijk = ghα Rijk is
called the covariant curvature tensor and the symbols Rhijk is called the Riemann’s
symbols of the first kind.
Cyclic property ○
60
Ex. 1)
Prove that the cyclic property of Riemann Christoffel tensor. i.e. Prove that
α α α
Rijk + Rjki + Rkij =0

Proof:
From Riemann Christoffel tensor, We know that

α ∂ α ∂ α
Rijk = {ik } − {ij } + {α β α β
βj }{ik } − {βk }{ij } → (1)
∂xj ∂xk
α ∂ α ∂ α
Rjki = {ji } − {jk } + {α β α β
βk }{ji } − {βi }{jk } → (2)
∂xk ∂xi
α ∂ α ∂ α
Rkij = {kj } − {ki } + {α β α β
βi }{kj } − {βj }{ki } → (3)
∂xi ∂xj
Now, adding (1), (2) and (3), we get
α α α
Rijk + Rjki + Rkij = 0, proved

Theorem (Ricci tensor):

The curvature tensor can be calculated in two different ways: one of these leads to a
zero tensor and others to a symmetric tensor Rij = Rji Known as Ricci tensor.
Ricci tensor ○
61
Proof: First Way
From Riemann Christoffel tensor, We have

α ∂ α ∂ α
Rijk = {ik } − {ij } + {α β α β
βj }{ik } − {βk }{ij } → (1)
∂xj ∂xk
Replacing i by α into (1), we get

α ∂ α ∂ α
Rαjk = {αk } − {αj } + {α β α β
βj }{αk } − {βk }{αj }
∂xj ∂xk
∂2 √ ∂2 √
= j k
log( g) − log( g) + 0
∂x ∂x ∂xk ∂xj
√
Since the dummy indices α and β are interchangeable and {α
αk } =
∂
∂xk
log g
α
∴ Rαjk =0

Second way
Replacing k by α into (1), we get

α ∂ α ∂ α
Rijα = {iα } − {ij } + {α β α β
βj }{iα } − {βα }{ij }
∂xj ∂xα
∂2 √ ∂ α
∴ Rij = log( g) − {ij } + {βiα }{α β α
βj } − {jα }{βi } → (2)
∂xj ∂xi ∂xα
Interchanging indices i and j into (2), we get

∂2 √ ∂ α
Rji = (log g) − {ji } + {βjα }{α β α
βi } − {iα }{βj } → (3)
∂xi ∂xj ∂xα
∂2 √ ∂ α
= (log g) − {ji } + {α β α β
βi }{jα } − {βj }{iα } → (4)
∂xi ∂xj ∂xα
√
Since dummy indices α, β are interchangeable and {α ∂
αj } = ∂xj (log g). From (2) and
(4), it is evident that Rij is symmetric i.e. Rij = Rji which is called Ricci tensor.
Riemann symbols ○
62
Remark:
h h h
1. Rikj = −Rijk = −Rik
2. The scalar R is defined by g ij Rij = R

Ex. 2)
Prove that Riemann symbol of 1st kind

1 ∂ 2 ghk ∂ 2 ghi ∂ 2 gij ∂ 2 gik

 
h
Rhijk = geh Rijk = − + − +g lm ([ih, l][jk, m]−[jh, l][ik, m])
2 ∂xi ∂xj ∂xk ∂xj ∂xk ∂xh ∂xj ∂xh

Proof.
The components Rhijk of associate covariant tensor of fourth order which is defined
by
l
Rhijk = glh Rijk → (1)
and it is called Riemann symbol of first kind. Using Riemann symbol of 2nd kind into
(1), we have
 
∂ l ∂ l l m l m
Rhijk = glh {ik } − {ij } + {mj }{ ik } − {mk }{ij }
∂xj ∂xk

∂ l ∂
= glh {ik } − glh k {lij } + [mj, h]{m m
ik } − [mk, h]{ij } → (2)
∂xj ∂x
The relation between Christoffel symbols of 1st kind and 2nd kind is given by

geh {lik } = [ik, h]

or,
∂geh l ∂ ∂
{ik } + geh j {lik } = [ik, h] → (3)
∂xj ∂x ∂xj
∂ ∂ ∂geh ∂ ∂geh
∴ geh j {lik } = (geh {lik }) − {lik } = [ik, h] − {lik } → (3)
∂x ∂xj ∂xj ∂xj ∂xj
Riemann symbols proof ○
63
Using (3) into (2), We have
∂ ∂geh ∂ ∂geh
Rhijk = [ik, h] − {lik } − [ij, h] + {lij } k + [mj, h]{m m
ik } − [mk, h]{ij }
∂xj ∂xj ∂xk ∂x
∂ ∂
= [ik, h] − {lik }([lj, h] + [hj, l]) − [ij, h] + {lij }([lk, h] + [hk, l]) + [mj, h]{m m
ik } − [mk, h]{ij }
∂xj ∂xk
∂ ∂
= [ik, h] − [ij, h] + {lij }[lk, h] + {lij }[hk, l] − {lik }[lj, h] − {lik }[hj, l] + [mj, h]{m m
ik } − [mk, h]{ij }
∂xj ∂xk
Replacing m by l, we have
∂ ∂
= [ik, h] − [ij, h] + {lij }[lk, h] + {lij }[hk, l] − {lik }[lj, h] − {lik }[hj, l] + [lj, h]{lik } − [lk, h]{lij }
∂xj ∂xk
∂ ∂
= [ik, h] − [ij, h] + {lij }[hk, l] − {lik }[hj, l]
∂xj   ∂xk    
∂ 1 ∂gkh ∂gih ∂gik ∂ 1 ∂gjh ∂gih ∂gij
= + − − + − + {lij }[hk, l] − {lik }[hj, l]
∂xj 2 ∂xi ∂xk ∂xh ∂xk 2 ∂xi ∂xj ∂xh
1 ∂ 2 gkh ∂ 2 gih ∂ 2 gik ∂ 2 gjh ∂ 2 gih ∂ 2 gij
 
= + − − − + + {lij }[hk, l] − {lik }[hj, l]
2 ∂xj ∂xi ∂xj ∂xk ∂xj ∂xh ∂xk ∂xi ∂xk ∂xj ∂xk ∂xh
1 ∂ 2 gkh ∂ 2 gik ∂ 2 gjh ∂ 2 gij
 
= − − + + g lm [ij, m][hk, l] − g lm [ik, m][hj, l], proved
2 ∂xj ∂xi ∂xj ∂xh ∂xk ∂xi ∂xk ∂xh
Cyclic property of Riemann symbols ○
64
Ex. 3)
Prove that
Rhijk + Rhjki + Rhkij = 0

Proof.
From Riemann symbol of 1st kind we have
 2
∂ 2 gik ∂ 2 gij ∂ 2 gik

1 ∂ gjk
Rhijk = − + − +g lm ([ih, l][jk, m]−[jh, l][ik, m]) → (1)
2 ∂xk ∂xi ∂xk ∂xj ∂xj ∂xk ∂xj ∂xk
 2
∂ 2 gkh ∂ 2 gjk ∂ 2 gji

1 ∂ gki
Rhjki = − + − +g lm ([jh, l][ki, m]−[kh, l][ji, m]) → (2)
2 ∂xj ∂xh ∂xi ∂xj ∂xk ∂xh ∂xk ∂xh
 2
∂ 2 gih ∂ 2 gki ∂ 2 gkj

1 ∂ gij
Rhkij = − + − +g lm ([kh, l][ij, m]−[ih, l][kj, m]) → (3)
2 ∂xk ∂xh ∂xk ∂xj ∂xi ∂xh ∂xi ∂xh
By adding (1), (2) and (3), we obtain

Rhijk + Rhjki + Rhkij = 0, proved

Bianchi identity ○
65
Bianchi Identity
Statement:
α α α
Rijk,l + Rikl,j + Rilj,k =0
or equivalently
Rhijk,l + Rhikl,j + Rhilj,k = 0
which is known as Bianchi identity.

Proof.
Since the Christoffel’s symbols referred to the geodesic coordinate system vanish at
the origin (pole). So, [ij, k] = 0 and {kij } = 0. We know that

α ∂ α ∂ α
Rijk = {ik } − {ij } + {α β α β
βj }{ik } − {βk }{ij } → (1)
∂xj ∂xk
Under the coordinate conditions at the pole, taking the coordinate derivative of both
sides of (1) with respect to xl , we get

α ∂2 ∂2
Rijk,l = l j
{α
ik } − {α
ij } → (2)
∂x ∂x ∂xl ∂xk
Similarly, we have

α ∂2 α ∂2
Rikl,j = { il } − {α
ik } → (3)
∂xj ∂xk ∂xj ∂xl

α ∂2 ∂2
Rilj,k = k l
{α
ij } − {α
il } → (4)
∂x ∂x ∂xk ∂xj
Adding (2), (3) and (4); we obtain
α α α
Rijk,l + Rikl,j + Rilj,k =0 → (5)
Final examples ○
66
Multiplying both sides of (5) by ghα and summing over α, we get
α α α
ghα [Rijk,l + Rikl,j + Rilj,k ]=0

∴ Rhijk,l + Rhikl,j + Rhilj,k = 0

Ex. 1)
Show that
(i) Rhijk = −Rihjk
(ii) Rhijk = −Rhjki
(iii) Rhijk = Rkhij

Ex. 2)
Find g and g ij corresponding to the metric
1
ds2 = r2
dr2 + r2 (dθ2 + sin2 θdϕ2 )
1− a2

Ex. 3)
√
Show that divAp = √
1 ∂
g ∂xm
( gAm )

Proof.
We have
∂Ap
divAp = Ap,p = + Am {pmp }
∂xp
∂Ap ∂ √
= + Am m (log g)
∂xp ∂x
√
∂Ap m 1 ∂ g
= + A √
∂xp g ∂xm
1 ∂ √ m
= √ ( gA ), proved
g ∂xm
Laplacian ○
67
Ex 4)
Show that  
2 1 ∂ √ mp ∂ϕ
∇ ϕ= √ gg
g ∂xm ∂xp

Proof
We have

∇2 ϕ = div(grad ϕ)
= div(Gm ),

where Gm is the contravariant vector associated with the gradient ϕ, that is Gp . Then
Gm = g mp Gp .
From Ex. 3) We have

1 ∂ √ m
divAp = √ ( gA )
g ∂xm

1 ∂ √ m
∴ div(Gm ) = √ ( gG )
g ∂xm
1 ∂ √ mp
∴ ∇2 ϕ = √ ( gg Gp )
g ∂xm
 
1 ∂ √ mp ∂ϕ
= √ gg ,
g ∂xm ∂xp
∂ϕ
since Gp = grad ϕ = ∂xp
proved