1.

Tutorial Sheet: Tensor Analysis

1) Show that if all the components of any tensor of any rank vanish in one
particular coordinate system, they vanish in all coordinate systems.

Note. This point takes on special importance in the four-dimensional

curved space of general relativity. If a quantity, expressed as a tensor,
exists in one coordinate system, it exists in all coordinate systems and is
not just a consequence of a choice of a coordinate system (as are centrifugal
and Coriolis forces in Newtonian mechanics).
2) The components of tensor A are equal to the corresponding components of
tensor B in one particular coordinate system, denoted by the superscript
0; that is,
A0ij = Bij0 .
Show that tensor A is equal to tensor B, Aij = Bij , in all coordinate
systems.
3) The last three components of a four-dimensional vector vanish in each of
two reference frames. If the second reference frame is not merely a rota-
tion of the first about the x0 axis, that is, if at least one of the coefficients
ai0 (i = 1, 2, 3) 6= 0, show that the zeroth component vanishes in all ref-
erence frames. Translated into relativistic mechanics this means that if
momentum is conserved in two Lorentz frames, then energy is conserved
in all Lorentz frames.
4) From an analysis of the behavior of a general second-rank tensor under
90◦ and 180◦ rotations about the coordinate axes, show that an isotropic
second-rank tensor in three- dimensional space must be a multiple of δij .
5) The four-dimensional fourth-rank Riemann–Christoffel curvature tensor of
general relativity, Riklm , satisfies the symmetry relations
Riklm = −Rikml = −Rkilm .

1
 With the indices running from 0 to 3, show that the number of independent
components is reduced from 256 to 36 and that the condition
Riklm = Rlmik
further reduces the number of independent components to 21. Finally, if
the components satisfy an identity Riklm + Rilmk + Rimkl = 0, show that
the number of independent components is reduced to 20.
Note. The final three-term identity furnishes new information only if all
four indices are different. Then it reduces the number of independent
components by one-third.
6) Tiklm is antisymmetric with respect to all pairs of indices. How many
independent components has it (in three-dimensional space)?

2
 2. Tutorial Sheet: Contraction and Direct
Product
1) If T···i is a tensor of rank n, show that ∂T···i /∂xj is a tensor of rank n + 1
(Cartesian coordinates).
Note. In non-Cartesian coordinate systems the coefficients aij are, in gen-
eral, functions of the coordinates, and the simple derivative of a tensor of
rank n is not a tensor except in the special case of n = 0. In this case the
derivative does yield a covariant vector (tensor of rank 1) by
∂ϕ0 X ∂ϕ ∂xj
=
∂x0i j
∂xj ∂x0i

∂Tijk··· /∂xj is a tensor of rank

P
2) If Tijk··· is a tensor of rank n, show that j
n − 1 (Cartesian coordinates).
3) The operator

2 1 ∂2
∇ − 2 2
c ∂t
may be written as
4
X ∂2
∂x 2,
i=1 i

using x4 = ict. This is the four-dimensional Laplacian, sometimes called

the d’Alem-bertian and denoted by 2 . Show that it is a scalar operator,
that is, is invariant under Lorentz transformations.

3
 3. Tutorial Sheet: Quotient Rule
1) The double summation Kij Ai Bj is invariant for any two vectors Ai and
Bj . Prove that Kij is a second-rank tensor.
Note. In the form ds2 (invariant) = gij dxi dxj , this result shows that the
matrix gij is a tensor.
2) The equation Kij Ajk = Bik holds for all orientations of the coordinate
system. If A and B are arbitrary second-rank tensors, show that K is a
second-rank tensor also.
3) The exponential in a plane wave is exp[i(~k · ~r − ωt)]. We recognize xµ =
(ct, x1 , x2 , x3 ) as a prototype vector in Minkowski space. If ~k · ~r − ωt is a
scalar under Lorentz transformations, show that k µ = (ω/c, k1 , k2 , k3 ) is a
vector in Minkowski space.
Note. Multiplication by ~ yields (E/c, p~) as a vector in Minkowski space.

4
 4. Tutorial Sheet: Pseudotensors and Dual
Tensors
1) An antisymmetric square array is given by
C 12 C 13
   
0 C3 −C2 0
−C3 0 C1  = −C 12 0 C 23 
C2 −C1 0 −C 13 −C 23 0
where (C1 , C2 , C3 ) form a pseudovector. Assuming that the relation
1
Ci = εijk C jk
2!
holds in all coordinate systems, prove that C jk is a tensor. (This is another
form of the quotient theorem.)
2) Show that the vector product is unique to three-dimensional space; that
is, only in three dimensions can we establish a one-to-one correspondence
between the components of an antisymmetric tensor (second-rank) and
the components of a vector.
3) Show that in R3
(a) δij = 3
(b) δij εijk = 0
(c) εipq εjpq = 2δij
(a) εijk εijk = 6
4) Show that in R3
εijk εpqk = δip δjq − δiq δjp
~ C
5) (a) Express the components of a cross-product vector C, ~ =A~ × B,
~ in
~ and B.
terms of εijk and the components of A ~
~·A
(b) Use the antisymmetry of εijk to show that A ~×B ~ = 0.
6) (a) Show that the inertia tensor (matrix) may be written
Iij = m(xi xj δij − xi xj )
for a particle of mass m at (x1 , x2 , x3 ).

5
 (b) Show that
Iij = −Mil Mlj = −mεilk xk εljm xm ,
where Mil = m1/2 εilk xk . This is the contraction of two second-rank
tensors and is identical with the matrix product.
~ and ∆ × ∆ϕ in tensor (index) notation in R3 so that it
7) Write ∆ · ∆ × A
becomes obvious that each expression vanishes.
8) Expressing cross products in terms of Levi-Civita symbols (εijk ), derive
the BAC − CAB rule, A ~ × (B~ × C)
~ = ω[B(
~ A~ · C)
~ − C(
~ A
~ · B)]
~
Hint. The relation of question 4 is helpful.
9) Verify that each of the following fourth-rank tensors is isotropic, that is,
that it has the same form independent of any rotation of the coordinate
systems.
(a) Aijkl = δij δkl ,
(b) Bijkl = δik δjl + δil δjk ,
(c) Cijkl = δik δjl − δil δjk .
10) Show that the two-index Levi-Civita symbol εij is a second-rank pseu-
dotensor (in two- dimensional space). Does this contradict the uniqueness
of δij (See Q 4 of the Tutorial on Tensor Analysis)?
11) Represent
  a 2 × 2 matrix, and using the 2 × 2 rotation matrix
εij by
cos ϕ sin ϕ
show that εij is invariant under orthogonal similarity
− sin ϕ cos ϕ
transformations.
12) Given Ak = 21 εijk B ij with B ij = −B ji , antisymmetric, show that
B mn = εmnk Ak

13) Show that the vector identity

~ × B)
(A ~ · (C
~ × D)
~ = (A
~ · C)(
~ B ~ · D)
~ − (A
~ · D)(
~ B ~ · C)
~

follows directly from the description of a cross product with εijk and the
identity of Q 4.

6
14) Generalize the cross product of two vectors to n-dimensional space for n =
4, 5, · · · Check the consistency of your construction and discuss concrete
examples.
Hint. First do it for the case of n = 2.

7
 5. Tutorial Sheet: General Tensors
1) Equations ε̂i = hi êi and ε̂r = êr = r̂, ε̂θ = rêθ = rθ̂, ε̂ϕ = r sin θêϕ =
r sin θϕ̂ use the scale factor hi . In Section Differential Vector Operator
(2.2), we had restricted ourselves to orthogonal coordinate systems, yet
Eq. ε̂i = hi êi holds for nonorthogonal systems. Justify the use of Eq.
ε̂i = hi êi for nonorthogonal systems.
2) (a) Show that ε̂i · ε̂j = δji .
(b) From the result of part (a) show that
F i = F~ · ε̂i , and Fi = F~ · ε̂i .
3) For the special case of three-dimensional space (ε1 , ε2 , ε3 defining a right-
handed coordinate system, not necessarily orthogonal), show that
ε̂j × ε̂k
ε̂i = , i, j, k = 1, 2, 3 and cyclic permutations.
ε̂j × ε̂k · ε̂i
Note. These contravariant basis vectors ε̂i define the reciprocal lattice
space.
4) Prove that the contravariant metric tensor is given by
g ij = ε̂i · ε̂j
5) If the covariant vectors ε̂i are orthogonal, show that
(a) gij is diagonal,
(b) g ii = 1/gii (no summation),
(c) |ε̂i | = 1/|ε̂i |.
6) Derive the covariant and contravariant metric tensors for circular cylindri-
cal coordinates.
7) Transform the right-hand side of,
∂ψ i
∇ψ = ε̂ ,
∂q i
into the êi basis, and verify that this expression agrees with the expression
of gradient for the orthogonal coordinates.

8
 8) Express the spherical polar unit vectors in Cartesian unit vectors.
9) (a) From the results of Q 8, calculate the partial derivatives of r̂, θ̂, and ϕ̂
with respect to r, θ , and ϕ.
(b) With ∇ given by
∂ 1 ∂ 1 ∂
r̂ + θ̂ + ϕ̂
∂r r ∂θ r sin θ ∂ϕ
(greatest space rate of change), use the results of part (a) to calculate
∇ · ∇ψ. This is an alternate derivation of the Laplacian.
Note. The derivatives of the left-hand ∇ operate on the unit vectors
of the right-hand ∇ before the unit vectors are dotted together.
10) Evaluate ∂εi /∂q j for spherical polar coordinates, and from these results
calculate Γkij for spherical polar coordinate. for spherical polar coordinates.
Note. Q 9 offers a way of calculating the needed partial derivatives. Re-
member,
ε̂1 = r̂, but ε̂2 = rθ̂ and ε̂3 = r sin θϕ̂

11) Show that the covariant derivative of a covariant vector is given by

∂Vi
Vi;j ≡ j
− Vk Γkij
∂q
Hint. Differentiate
ε̂i · ε̂j = δji

12) Verify that Vi;j = gik V;jk by showing that

 k 
∂Vi ∂V
j
− Vs Γsij = gik j
+ V m Γkmj
∂q ∂q
13) From the circular cylindrical metric tensor gij , calculate the Γkij for circular
cylindrical coordinates.
Note. There are only three nonvanishing Γ.
14) Using the Γkij from Q 13, write out the covariant derivatives V;ji of a vector
V~ in circular cylindrical coordinates.

9
15) A triclinic crystal is described using an oblique coordinate system. The
three covariant base vectors are
ε̂1 = 1.5x̂,
ε̂2 = 0.4x̂ + 1.6ŷ,
ε̂3 = 0.2x̂ + 0.3ŷ + 1.0ẑ.
(a) Calculate the elements of the covariant metric tensor gij .
(b) Calculate the Christoffel three-index symbols, Γkij . (This is a “by in-
spection” calculation.)
(c) From the cross-product form of Q 3 calculate the contravariant base
vector ε̂3 .
(d) Using the explicit forms ε̂3 and ε̂i , verify that ε̂3 · ε̂i = δi3 .
Note. If it were needed, the contravariant metric tensor could be deter-
mined by finding the inverse of gij or by finding the ε̂i and using g ij = ε̂i · ε̂j
.
16) Verify that
 
1 ∂gik ∂gjk ∂gij
[ij, k] = + − k
2 ∂q j ∂q i ∂q
∂ ε̂i
Hint. Substitute Eq. [ij, k] = ε̂k · ∂q j into the right-hand side and show
that an identity results.
ij
17) Show that for the metric tensor gij;k = 0, g;k = 0.
18) Show that parallel displacement δdq i = d2 q i along a geodesic. Construct
a geodesic by parallel displacement of δdq i .
19) Construct the covariant derivative of a vector V i by parallel transport
starting from the limiting procedure
V i (q j + dq j ) − V i (q j )
lim
dqj →0 dqJ

10
 6. Tutorial Sheet: Tensor Derivative
Operators
1) Verify Eq.,
∂g im ∂gim
= gg ,
∂q k ∂q k
for the specific case of spherical polar coordinates.
2) Starting with the divergence in tensor notation, Eq.,
1 ∂
∇ · V~ = V;ii = 1/2 k (g 1/2 V k )
g ∂q
develop the divergence of a vector in spherical polar coordinates, Eq.
 
1 ∂ ∂ ∂V ϕ
∇ · V~ = 2 2
sin θ (r Vr ) + r (sin θVθ ) + r
r sin θ ∂r ∂θ ∂ϕ

3) The covariant vector Ai is the gradient of a scalar. Show that the difference
of covariant derivatives Ai;j − Aj;i vanishes.

11