41

PROBLEMS FOR PART B

2.19 A transformation T operates on any vector a to give Ta ¼ a/jaj, where jaj is the magnitude of a. Show
that T is not a linear transformation.
2.20 (a) A tensor T transforms every vector a into a vector Ta ¼ m  a, where m is a specified vector. Show
that T is a linear transformation. (b) If m ¼ e1 þ e2, find the matrix of the tensor T.
2.21 A tensor T transforms the base vectors e1 and e2 such that Te1 ¼ e1 þ e2 , Te2 ¼ e1  e2 .
If a ¼ 2e1 þ 3e2 and b ¼ 3e1 þ 2e2 , use the linear property of T to find (a) Ta, (b) Tb, and (c) T(aþb).
2.22 Obtain the matrix for the tensor T, that transforms the base vectors as follows: Te1 ¼ 2e1 þ e3 ;
Te2 ¼ e2 þ 3e3 ; Te3 ¼ e1 þ 3e2 .
2.23 Find the matrix of the tensor T that transforms any vector a into a vector b ¼ m(a  n) where
pffiffiffi pffiffiffi
2 2
m¼ ðe1 þ e2 Þ and n ¼ ðe1 þ e3 Þ.
2 2
42 CHAPTER 2 Tensors

2.24 (a) A tensor T transforms every vector into its mirror image with respect to the plane whose normal is
e2. Find the matrix of T. (b) Do part (a) if the plane has a normal in the e3 direction.
2.25 (a) Let R correspond to a right-hand rotation of angle y about the x1-axis. Find the matrix of R. (b) Do
part (a) if the rotation is about the x2-axis. The coordinates are right-handed.
2.26 Consider a plane of reflection that passes through the origin. Let n be a unit normal vector to the plane
and let r be the position vector for a point in space. (a) Show that the reflected vector for r is given by
Tr ¼ r  2ðr  nÞn, pwhere
ffiffiffi T is the transformation that corresponds to the reflection. (b) Let
n ¼ ðe1 þ e2 þ e3 Þ= 3; find the matrix of T. (c) Use this linear transformation to find the mirror image
of the vector a ¼ e1 þ 2e2 þ 3e3 .
2.27 Knowing that the reflected vector for r is given by Tr ¼ r  2ðr  nÞn (see the previous problem),
where T is the transformation that corresponds to the reflection and n is the normal to the mirror, show
ffiffiffi T ¼ I  2nn and find the matrix of T if the nor-
that in dyadic notation the reflection tensor is givenpby
mal of the mirror is given by n ¼ ðe1 þ e2 þ e3 Þ= 3.
2.28 A rotation tensor R is defined by the relation Re1 ¼ e2 ; Re2 ¼ e3 ; Re3 ¼ e1 . (a) Find the matrix of R
and verify that RTR ¼ I and det R ¼ 1 and (b) find a unit vector in the direction of the axis of rotation
that could have been used to effect this particular rotation.
2.29 A rigid body undergoes a right-hand rotation of angle y about an axis that is in the direction of
the unit vector m. Let the origin of the coordinates be on the axis of rotation and r be the position
vector for a typical point in the body. (a) Show that the rotated vector of r is given
by: Rr ¼ ð1  cos ypÞðm
ffiffiffi  rÞm þ cos y r þ sin y ðm  rÞ, where R is the rotation tensor. (b) Let
m ¼ ðe1 þ e2 þ e3 Þ= 3, find the matrix for R.
2.30 For the rotation about an arbitrary axis m by an angle y, (a) show that the rotation tensor is given
by R ¼ ð1  cos y ÞðmmÞ þ cos y I þ sin yE, where mm denotes that dyadic product of m and m,
and E is the antisymmetric tensor whose dual vector (or axial vector) is m, (b) find RA, the antisym-
metric part of R and (c) show that the dual vector for RA is given by (sin y)m. Hint:
Rr ¼ ð1  cos y Þðm  rÞm þ cos y r þ sin y ðm  rÞ (see previous problem).
2.31 (a) Given a mirror whose normal is in the direction of e2, find the matrix of the tensor S, which first trans-
forms every vector into its mirror image and then transforms them by a 45 right-hand rotation about the
e1-axis. (b) Find the matrix of the tensor T, which first transforms every vector by a 45 right-hand rotation
about the e1-axis and then transforms them by a reflection with respect to a mirror (with normal e2).
(c) Consider the vector a ¼ e1 þ 2e2 þ 3e3 ; find the transformed vector by using the transformation S.
(d) For the same vector a ¼ e1 þ 2e2 þ 3e3 , find the transformed vector by using the transformation T.
2.32 Let R correspond to a right-hand rotation of angle y about the x3-axis; (a) find the matrix of R2.
(b) Show that R2 corresponds to a rotation of angle 2y about the same axis. (c) Find the matrix of
Rn for any integer n.
2.33 Rigid body rotations that are small can be described by an orthogonal transformation R ¼ I þ eR ,
where e ! 0 as the rotation angle approaches zero. Consider two successive small rotations, R1 and
R2; show that the final result does not depend on the order of rotations.
2.34 Let T and S be any two tensors. Show that (a) TT is a tensor, (b) TT þ ST ¼ ðT þ SÞT , and
(c) ðTSÞT ¼ ST TT .
2.35 For arbitrary tensors T and S, without relying on the component form, prove that (a) ðT1 ÞT ¼ ðTT Þ1
and (b) ðTSÞ1 ¼ S1 T1 .
 Problems for Part B 43

2.36 Let fei g and fei0 g be two rectangular Cartesian base vectors. (a) Show that if ei0 ¼ Qmi em , then
ei ¼ Qim em0 . (b) Verify Qmi Qmj ¼ dij ¼ Qim Qjm .
2.37 The basis fei0 g is obtained by a 30 counterclockwise rotation of the fei g basis about the e3 axis.
(a) Find the transformation matrix ½Q relating
pffiffiffi the two sets of basis. (b) By using the vector transforma-
tion law, find the components of a ¼ 3e1 þ e2 in the primed basis, i.e., find ai0 and (c) do part (b)
geometrically.
2.38 Do the previous problem with the fei0 g basis obtained by a 30 clockwise rotation of the fei g basis
about the e3 axis.
2.39 The matrix of a tensor T with respect to the basis fei g is
2 3
1 5 5
½T ¼ 4 5 0 0 5:
5 0 1
0 0 0
Find T11 ; T12 and T31 with respect to a right-handed basis fei0 g where e10 is in the direction of e2 þ 2e3
0
and e2 is in the direction of e1.
2.40 (a) For the tensor of the previous problem, find ½Tij0 , i.e., ½Te 0 where fei0 g is obtained by a 90 right-
i
hand rotation about the e3 axis and (b) obtain Tii0 and the determinant jTij0 j and compare them with Tii
and jTij j.
2.41 The dot product of two vectors a ¼ ai ei and b ¼ bi ei is equal to ai bi . Show that the dot product is a sca-
lar invariant with respect to orthogonal transformations of coordinates.
2.42 If Tij are the components of a tensor, (a) show that Tij Tij is a scalar invariant with respect to
orthogonal transformations of coordinates, (b) evaluate Tij Tij with respect to the basis fei g for
2 3 2 3
1 0 0 0 0 1
½T ¼ 4 1 2 5 5, (c) find ½T 0 if ei0 ¼ Qei , where ½Q ¼ 4 1 0 0 5, and (d) verify for the above that
1 2 3 ei 0 1 0 ei
0 0
Tij Tij ¼ Tij Tij .
2.43 Let ½T and [T]0 be two matrices of the same tensor T. Show that det½T ¼ det½T 0 .
2.44 (a) If the components of a third-order tensor are Rijk , show that Riik are components of a vector. (b) If
the components of a fourth-order tensor are Rijkl , show that Riikl are components of a second-order ten-
sor. (c) What are components of Riik... if Rijk... are components of a tensor of nth order?
2.45 The components of an arbitrary vector a and an arbitrary second tensor T are related by a triply sub-
scripted quantity Rijk in the manner ai ¼ Rijk Tjk for any rectangular Cartesian basis {ei}. Prove that
Rijk are the components of a third-order tensor.
2.46 For any vector a and any tensor T, show that (a) a  TA a ¼ 0 and (b) a  Ta ¼ a  TS a, where TA and TS
are antisymmetric and symmetric part of T, respectively.
2.47 Any tensor can be decomposed into a symmetric part and an antisymmetric part, that is, T ¼ TS þ TA .
Prove that the decomposition is unique. (Hint: Assume that it is not true and show contradiction.)
2 3
1 2 3
2.48 Given that a tensor T has the matrix ½T ¼ 4 4 5 6 5, (a) find the symmetric part and the antisym-
7 8 9
metric part of T and (b) find the dual vector (or axial vector) of the antisymmetric part of T.
44 CHAPTER 2 Tensors

2.49 Prove that the only possible real eigenvalues of an orthogonal tensor Q are l ¼ 1. Explain the direc-
tion of the eigenvectors corresponding to them for a proper orthogonal (rotation) tensor and for an
improper orthogonal (reflection) tensor.
" #
1 1 2 2
2.50 Given the improper orthogonal tensor ½Q ¼ 2 1 2 . (a) Verify that det ½Q ¼ 1.
3 2 2 1
(b) Verify that the eigenvalues are l ¼ 1 and 1. (c) Find the normal to the plane of reflection (i.e.,
eigenvectors corresponding to l ¼ 1) and (d) find the eigenvectors corresponding to l ¼ 1 (vectors
parallel to the plane of reflection).
2.51 Given that tensors R and S have the same eigenvector n and corresponding eigenvalues r1 and s1,
respectively, find an eigenvalue and the corresponding eigenvector for T ¼ RS.
2.52 Show that if n is a real eigenvector of an antisymmetric tensor T, then the corresponding eigenvalue
vanishes.
2.53 (a) Show that a is an eigenvector for the dyadic product ab of vectors a and b with eigenvalue a  b,
(b) find the first principal scalar invariant of the dyadic product ab and (c) show that the second and
the third principal scalar invariant of the dyadic product ab vanish, and that zero is a double eigenvalue
of ab.
2.54 For any rotation tensor, a set of basis fei0 g may be chosen with e30 along the axis of rotation so that
Re10 ¼ cos ye10 þ siny e20 ; Re20 ¼ sin y e10 þ cos y e20 ; Re30 ¼ e30 , where y is the angle of right-hand rota-
tion. (a) Find the antisymmetric part of R with respect to the basis fei0 g, i.e., find ½RA e 0 . (b) Show that
i
the dual vector of RA is given by tA ¼ sin y e30 and (c) show that the first scalar invariant of R is given
by 1 þ 2 cos y . That is, for any given rotation tensor R, its axis of rotation and the angle of rotation can
be obtained from the dual vector of RA and the first scalar invariant of R.
2.55 The rotation of a rigid body is described by Re1 ¼ e2 ; Re2 ¼ e3 ; Re3 ¼ e1 . Find the axis of rotation
and the angle of rotation. Use the result of the previous problem.
" #
1 0 0
2.56 Given the tensor ½Q ¼ 0 1 0 . (a) Show that the given tensor is a rotation tensor. (b) Verify
0 0 1
that the eigenvalues are l ¼ 1 and 1. (c) Find the direction for the axis of rotation (i.e., eigenvectors
corresponding to l ¼ 1). (d) Find the eigenvectors corresponding to l ¼ 1 and (e) obtain the angle of rotation
using the formula I1 ¼ 1 þ 2 cos y (see Prob. 2.54), where I1 is the first scalar invariant of the rotation
tensor.
2.57 Let F be an arbitrary tensor. (a) Show that FTF and FFT are both symmetric tensors. (b) If
F ¼ QU ¼ VQ, where Q is orthogonal, show that U2 ¼ FT F and V2 ¼ FFT . (c) If l and n are eigen-
value and the corresponding eigenvector for U, find the eigenvalue and eigenvector for V.
Tii Tjj Tij Tji
2.58 Verify that the second principal scalar invariant of a tensor T can be written: I2 ¼  .
2 2
2.59 A tensor T has a matrix [T] given below. (a) Write the characteristic equation and find the principal
values and their corresponding principal directions. (b) Find the principal scalar invariants. (c) If
n1 ; n2 ; n3 are the principal directions, write ½Tni . (d) Could the following matrix [S] represent the same
" # " #
5 4 0 7 2 0
tensor T with respect to some basis? ½T ¼ 4 1 0 , ½S ¼ 2 1 0 :
0 0 3 0 0 1
 2.26 Tensor-Valued Functions of a Scalar 45

2 3
3 0 0
2.60 Do the previous problem for the following matrix: ½T ¼ 4 0 0 4 5.
0 4 0
2.61 A tensor T has a matrix given below. Find the principal values and three mutually perpendicular prin-
cipal directions.
2 3
1 1 0
½T ¼ 4 1 1 0 5:
0 0 2

PART C: TENSOR CALCULUS

2.26 TENSOR-VALUED FUNCTIONS OF A SCALAR
Let T ¼ T(t) be a tensor-valued function of a scalar t (such as time). The derivative of T with respect to t is
defined to be a second-order tensor given by:
dT Tðt þ DtÞ  TðtÞ
¼ lim : (2.26.1)
dt Dt!0 Dt
The following identities can be easily established:
d dT dS
ðT þ SÞ ¼ þ ; (2.26.2)
dt dt dt
d da dT
ðaðtÞTÞ ¼ T þ a ; (2.26.3)
dt dt dt
d dT dS
ðTSÞ ¼ SþT ; (2.26.4)
dt dt dt
d dT da
ðTaÞ ¼ aþT ; (2.26.5)
dt dt dt
T
d T dT
ðT Þ ¼ : (2.26.6)
dt dt

We shall prove here only Eq. (2.26.5). The other identities can be proven in a similar way. Using the
definition given in Eq. (2.26.1), we have

d Tðt þ DtÞaðt þ DtÞ  TðtÞaðtÞ

ðTaÞ ¼ limDt!0
dt Dt
Tðt þ DtÞaðt þ DtÞ  TðtÞaðtÞ  TðtÞaðt þ DtÞ þ TðtÞaðt þ DtÞ
¼ limDt!0
Dt
Tðt þ DtÞaðt þ DtÞ  TðtÞaðt þ DtÞ þ TðtÞaðt þ DtÞ  TðtÞaðtÞ
¼ limDt!0
Dt
Tðt þ DtÞ  TðtÞ aðt þ DtÞ  aðtÞ
¼ limDt!0 aðt þ DtÞ þ lim TðtÞ :
Dt Dt!0 Dt
dðTaÞ d T da
Thus, ¼ aþT .
dt dt dt