(Chapter head:)Tensor Analysis

1 Points, vectors and tensors

Let E be an n-dimensional Euclidean space and let U be the space of n-
dimensional vectors associated with E. Points of E and vectors of U satisfy
the basic rules of vector alebra.
1.1 Inner product, norm and ortogonality
Let
u.v < u, v > (!)
denote the inner product of u and v. The norm of a vector u is de"ned as
|u| =

u.u (#)
and u is said to be a unit vector if
|u| = 1. ($)
A vector u is said to be orthoonal (or perpendicular) to a vector v if and
only if
u.v = 0. (%)
1.1.1 Ortogonal bases and cartesian coordinate frames
A set e
i
e
1
, e
2
, ...e
n
of n mutually orthoonal vectors satisfyin
e
i
.e
j
=
ij
(&)
where

ij
=
_
1 if i = j
0 if i ,= j
(')
is the (ronec)er delta* de"nes an orthonormal basis for U.
Any vector u U can be represented as
u = u
1
e
1
+u
2
e
2
...... +u
n
e
n
= u
i
e
i
(+)
where
u
i
=e
i
.u, i = 1...n (,)
are the cartesian components of u relative to the basis e
i
. Any vector of U is
uni-uely de"ned by its components relative to a iven basis. This allow us to
represent any vector u as a sinle column matri.* denoted u* of components:
u =
_

_
u
1
u
2
.
.
.
u
n
_

_
. (/)
!
An orthonormal basis e
i
* toether with an oriin point* x
o
E* de"ne a
cartesian coordinate frame. Thus* analoously to the representation of vectors*
any point x of E can be represented by an array:
x =
_

_
x
1
x
2
.
.
.
x
n
_

_
* (!0)
of cartesian coordinates of x. The cartesian coordinates x
i
of x are the cartesian
componets of the vector
u = x x
o
(!!)
with
u
i
= (x x
o
) .e
i
. (!#)
1.2 Linear operators on vectors. Second order tensors
1econd order tensors are linear transformations from U into U* i.e.* a second
order tensor T : U U maps each vector u U into a vector v U:
v = Tu. (!$)
The operation of sum and scalar multiplication of tensors are de"ned by:
(S +T) u = Su +Tu (!%)
(S) u = (Su)
where . 2n addition* the 3ero tensor* 0* and the identity tensor* I* are*
respectively* the tensors that satisfy
(0) u =

0 (!&)
(I) u = u
u U.
The product of two tensors S and T is the tensor ST de"ned by:
STu = S (Tu) . (!')
2n eneral*
ST ,= TS. (!+)
2f ST = TS* then S and T are said to commute.
#
1.2.1 The Transpose, Symmetric and Ske tensors
The transpose* T
T
* of a tensor T is the uni-ue tensor that satis"es:
Tu.v = u.T
T
v, u, v U. (!,)
2f T = T
T
then T is said to be symmetric. 2f T = T
T
then T is said to be
s)ew.
Any tensor T can be decomposed as the sum:
T = T
sym
+T
skew
(!/)
of its symmetric part
T
sym
=
1
2
_
T +T
T
_
(#0)
and its s)ew part
T
skew
=
1
2
_
T T
T
_
. (#!)
!asic properties The followin basic properties involvin the transpose and
the s)ew and symmetric parts of a tensor hold:
"i# (S +T)
T
= S
T
+T
T
;
"ii# (ST)
T
= T
T
S
T
;
"iii#
_
T
T
_
T
= T;
"iv# 2f T is symmetric* then
T
skew
= 0 and T
sym
= T4 (##)
"v# 2f T is s)ew* then
T
skew
= T and T
sym
= 0. (#$)
1.$ %ross product
2n the vector space V(= R
3
) of the translation of the pontual Euclidean space
E* we may de"ne the cross product of the vectors u and v
u v =
ijk
u
i
v
j
e
k
(#%)
where e
k
is the )-th cartesian base (cartesian base representation) and
ijk
is
the permutation symbol.
The cross product has the followin properties:
u v = v u
(u +v) w = (u w) + (v w)
$
u. (u v) = 0
(u v) . (u v) = (u.u) (v.v) (u.v)
2
*
for every u* v and w V and * R.
5rom the de"nition in (#%) we can see that
e
i
e
j
=
ijk
e
k
(#&)
and

ijk
= (e
i
e
j
) .e
k
. (#')
6ow* since

ijk

mnp
= [(e
i
e
j
) .e
k
] [(e
m
e
n
) .e
p
] (#+)
= det

e
i
.e
m
e
i
.e
n
e
i
.e
p
e
j
.e
m
e
j
.e
n
e
j
.e
p
e
k
.e
m
e
k
.e
n
e
k
.e
p

= det

im

in

ip

jm

jn

jp

km

kn

kp

*
placin sucessively: k = p4 k = p and j = n4 and k = p* j = n and i = m* we
derive

ijk

mnk
=
im

jn

in

jm
(#,)

ijk

mjk
= 2
im

ijk

ijk
= 2
ii
= 6
Let A denote the matri. represented in a cartesian coordinate system as
A =
_
_
u
1
v
1
w
1
u
2
v
2
w
2
u
3
v
3
w
3
_
_
.
Then the paralelepiped formed by the edes u* v and w is iven by
det [A] = (u v) . w =
ijk
u
i
v
j
w
k
(#/)
Two additional properties are iven by:
u (v w) = (u. w) v (u.v) w
(u v) w = ( w.u) v ( w.v) u
%
1.& Tensors and its components
7e denote a second order tensor a linear transformation T : V V that corre-
sponds to a iven vector v a vector u* i.e.*
u = Tv ($0)
1ince T is a linear transformation*
T (v
1
+v
2
) = Tv
1
+Tv
2
for v
1
* v
2
V and * R.
1.' Operations ith tensors
The set of tensors forms a linear space L(V, V).
1.'.1 Trace function
The function trace of a tensor T is de"ned as the sum of the diaonal ele-
ments of the matri. [T] that represents T with respect to a cartesian base
e
i
, i = 1, ...ni.e.
tr [T] = T
ii
($!)
The trace function is a linear transformation tr : L R* since
tr [ T + R] = tr [T] + tr [R] * T* R L and * R.
Properties:
tr [A] = 0* A Skew
tr [I] = 3
tr
_
A
T

= tr [A] , A L
tr [T
1
T
2
T
3
] = tr [T
3
T
1
T
2
] = tr [T
2
T
3
T
1
] (ciclic permutation)
1.'.2 Inner product
The inner product of two tensors T* R L may be de"ned as:
T.R = tr
_
TR
T

. ($#)
Let [T] and [R] be the matri. representation of the linear transformations T :
V V and R : V V with respect to a cartesian base e
i
, i = 1, ...n. Then*
the inner product of two tensors may be e.pressed as
T.R = T
ij
R
ij
At this point* we can notice that the trace function may also be de"ned as
tr [A] = A.I ($$)
where I is the identity transformation.
&
1.( Tensor Product
The tensor product of the vectors a and

b* represented by a

b* is the tensor
(linear transformation) de"ned by
_
a

b
_
v =
_

b.v
_
a* v V. ($%)
5rom the above de"nition* we may obtain the followin properties:

_
a +

b
_
= (a ) +
_

b
_
a
_

b +
_
=
_
a

b
_
+ (a )
for a*

b and V and * R.
The followin formulas are also valid:
tr
_
a

b
_
=a.

b
_
a

b
_
T
=
_

b a
_
_
a

b
__


!
_
=
_

b.
__
a

!
_
T
_
a

b
_
= (Ta)

b
_
a

b
_
T =a
_
T
T
b
_

i
(e
i
e
i
) = I, i.e., (e
i
e
i
) = I
($&)
for any a*

b* and

! V and T L.
1.(.1 Trace, inner product and )uclidean norm
5or any u, v U* the trace of the tensor (u v) is the linear map de"ned as
tr (u v) = u.v. ($')
5or a eneric tensor* T = T
ij
(e
i
e
j
)* it then follows that
tr (T) = T
ij
tr (e
i
e
j
) = T
ij

ij
= T
ii
* ($+)
that is* the trace of T is the sum of the diaonal terms of the matri. represen-
tation [T].
The inner product* S.T* between two second order tensors S and T is de"ned
as
S.T S : T = S
ij
T
ij
. ($,)
The Euclidean norm of a tensor T is de"ned as:
|T| =

T.T =
_
T
2
11
+T
2
12
T
2
nn
. ($/)
'
1.(.2 !asic properties
The followin basic properties involvin the internal product of tensors hold for
any tensors R, S, T and vectors ",

t, u and v:
"i# I.T = tr (T) ;
"ii# R. (ST) = S
T
R.T = RT
T
.S;
"iii# u.Sv = S. (u v) ;
"iv#
_
"

t
_
. (u v) =
_
u.

t
_
(".v) ;
"v# T
ij
= T. (e
i
e
j
) ;
"vi# (u v)
ij
= (u v) . (e
i
e
j
) = u
i
v
j
;
"vii# 2f S is symmetric* then S.T = S.T
T
= S.T
sym
;
"viii# 2f S is s)ew* then S.T = S.T
T
= S.T
skew
;
"i*# 2f S is symmetric and T is s)ew* then S.T = 0.
1.+ !asis in L
Let T L and e
i
, i = 1, ...n be a cartesian base of V. Then*
T = T
ij
(e
i
e
j
) (%0)
where
(e
i
e
j
) * i* j = 1, ...n (%!)
is a basis for L and T
ij
are the components of T with respect to this basis.
1.+.1 %artesian components and matri* representation
Any second order tensor T can be represented as:
T = T
11
(e
1
e
1
) +T
12
(e
1
e
2
) +... +T
nn
(e
n
e
n
) (%#)
= T
ij
(e
i
e
j
)
where
T
ij
=e
i
.Te
j
(%$)
are the cartesian components of T.
Any tensor is uni-uely de"ned by its cartesian components. Thus* by ar-
ranin the components T
ij
in a matri.* we may have the followin matri.
representation for T:
[T] =
_

_
T
11
T
12
. . . T
1n
T
21
T
22
. . . T
2n
.
.
.
.
.
.
.
.
.
.
.
.
T
n1
T
n2
T
nn
_

_
. (%%)
+
5or instance* the cartesian components of the identity tensor read:
I
ij
=
ij
* (%&)
so that its matri. representation is iven by:
[I] =
_

_
1 0 . . . 0
0 1 . . . 0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 1
_

_
. (%')
The cartesian components of the vector v = Tu are iven by:
v
i
= [T
lk
(e
l
e
k
) u
j
e
j
] .e
i
= T
ij
u
j
.
Thus* the array v of cartesian components of v is obtained from the matri.
vector product:
v =
_

_
v
1
v
2
.
.
.
v
n
_

_
=
_

_
T
11
T
12
. . . T
1n
T
21
T
22
. . . T
2n
.
.
.
.
.
.
.
.
.
.
.
.
T
n1
T
n2
T
nn
_

_
_

_
u
1
u
2
.
.
.
u
n
_

_
. (%+)
2t can be easily proved that the cartesian components T
T
ij
of the transpose T
T
of a tensor T are iven by:
T
T
ij
= T
ji
. (%,)
Thus* T
T
has the followin cartesian matri. representation:
_
T
T

=
_

_
T
11
T
21
. . . T
n1
T
12
T
22
. . . T
n2
.
.
.
.
.
.
.
.
.
.
.
.
T
1n
T
2n
T
nn
_

_
. (%/)
The s)ew part of the tensor product
_
a

b
_
is a tensor denoted by the
e.ternal product* a

b* i.e.* by de"nition
a

b =
_
a

b
_
skew
(&0)
=
1
2
_
_
a

b
_

_
a

b
_
T
_
=
1
2
__
a

b
_

b a
__
.
Also* a fourth order tensor D : L L is de"ned as:
# = D, #, L
where _
a

b

!
__
e

$
_
(.e)
_

!.

$
__
a

b
_
. (&!)
,
1.+.2 ,eterminant of a tensor
The determinant function is a scalar function with a tensor arument* de"ned
in a cartesian coordinate system by its components as

ijk

nqr
det [T] = det

T
ip
T
iq
T
ir
T
jp
T
jq
T
jr
T
kp
T
kq
T
kr

= det

T
pi
T
qi
T
ri
T
pj
T
qj
T
rj
T
pk
T
qk
T
rk

(&#)
i.e.

ijk
det [T] =
pqr
T
ip
T
jq
T
kr
(&$)
=
pqr
T
pi
T
qj
T
rk
= det

T
i1
T
i2
T
i3
T
j1
T
j2
T
j3
T
k1
T
k2
T
k3

= det

T
1i
T
2i
T
3i
T
1j
T
2j
T
3j
T
1k
T
2k
T
3k

and
det [T] =
1
6

ijk

pqr
T
ip
T
jq
T
kr
(&%)
=
1
6

ijk

pqr
T
pi
T
qj
T
rk
= det

T
11
T
12
T
13
T
21
T
22
T
23
T
31
T
32
T
33

= det

T
11
T
21
T
31
T
12
T
22
T
32
T
13
T
23
T
33

7ith these relations* we may derive

det[I] = 14
det
_
T
T

= det [T]4
det(T) =
3
det [T] , R* dim(V) = 34
det(u v) = 04
det(RT) = det (R) det (T).
/
1.- %ofactor of a tensor
Let %
T
= &$ [T] the cofactor of the tensor T* whose components %
Tij
are the
cofactors of the components T
ij
of tensor T. 8evelopin det [T] in terms of
cofactors* we derive
det [T] =
1
6

ijk

pqr
T
ip
T
jq
T
kr
=
1
3
T.%
T
=
1
3
T
ip
%
Tip
then* wrt a cartesian base* we have
%
Tip
=
1
2

ijk

pqr
T
jq
T
kr
(&&)
The tensor (%
T
)
T
= (&$ [T])
T
= &$
_
T
T

is denoted the ad9oint tensor* rep-

resented by a!j [T].
a!j [T] = (&$ [T])
T
= &$
_
T
T

(&')
This tensor has the followin properties:
T (a!j [T]) = (a!j [T]) T = I det [T] (&+)
2n fact*
%
Tip
T
mp
=
1
2

ijk

pqr
T
mp
T
jq
T
kr
=
1
2

mjk

ijk
det [T]
=
mi
det [T]
6otice that
det (T a!j [T]) = det (I det [T]) = (det [T])
3
= det [T] det [a!j [T]] .
Thus* if T is non-sinular* i.e.* det [T] ,= 0* then
det [a!j [T]] = (det [T])
2
. (&,)
:ultiplyin (&+) by (det [T])
1
we derive
(det [T])
1
T a!j [T] = I
what implies
T
1
=
1
det [T]
a!j [T] (&/)
!0
1.-.1 Inverse tensor and determinant
A tensor T is said to be invertible if its inverse* denoted T
1
* satisfyin
T
1
T = TT
1
= I
e.ists.
The determinant of a tensor T* denoted det (T)* is the determinat of the
matri. [T]. A tensor T is invertible if and only if
det (T) ,= 0.
A tensor T is said to be positive de"nite if
Tu.u 0* u ,= 0.
Any positive de"nite tensor is invertible.
!asic relations involving the determinant and inverse tensor ;elation
(i) below holds for any tensors S and T:
"i# det (ST) = det (S) det (T)4
"ii# det
_
T
1
_
= (det (T))
1
4
"iii# (ST)
1
= T
1
S
1
4
"iv#
_
T
1
_
T
=
_
T
T
_
1
.
1.-.2 .eometric interpretation of det [T]
7e have seen that the volume ' (() of a paralelepiped formed by the edes u*
v and w is iven by:
' (() = [(u v) . w[ = [
ijk
u
i
v
j
w
k
[ ('0)
A tensor T transform the paralelepiped into another paralelepiped iven by
)(()* i.e.*
)(() = [(Tu Tv) .T w[ ('!)
= [
pqr
T
pi
T
qj
T
rk
u
i
v
j
w
k
[
= [
ijk
u
i
v
j
w
k
det [T][
= [(u v) . w[ [det [T][
thus
)(() = [det [T][ ' (()
conse-uently
)(()
' (()
= [det [T][ =
[(Tu Tv) .T w[
[(u v) . w[
('#)
!!
1./ 0nalysis of Tensorial 1unctions
1./.1 ,erivative of scalar functions
The scalar functions $ : D R may have vector (D V) or tensor (D L) type
of aruments. Consider the function $ : D V R. 7e say that a scalar
function $ is diferentiable at x D (open set)* alon the direction u* when the
followin limit e.ists
*$ (x; u) = lim
h0
$ (x ++ u) $ (x)
+
=
!
! +
$ (x ++ u)

h=0
. ('$)
2f $ is di<erentiable* then
*$ (x; u) = $ (x) .u. ('%)
Consider the function $ : D L R. 7e say that a scalar function $ is
diferentiable at T D (open set)* alon the direction %* when the followin
limit e.ists
*$ (T; %) = lim
h0
$ (T ++ %) $ (T)
+
=
!
! +
$ (T ++ %)

h=0
('&)
2f $ is di<erentiable* then
*$ (T; %) =
T
$ (T) .% = tr
_

T
$ (T) %
T

('')
1./.2 )*ample
Consider the case
$ (T) = tr
_
T
k

('+)
Then* *$ (T; %) =
d
d h
$ (T ++ %)

h=0
. 6ow* $ (T ++ %) = tr
_
(T ++ %)
k
_
.
=ut* from the binomial formula we have
(T ++ %)
k
= T
k
++ k T
k1
% +
1
2
+
2
k (k 1) T
k2
%
2
+... ++
k
%
k
.
>ence* from the linearity of the trace function* we may write
tr
_
(T ++ %)
k
_
= tr
_
T
k

++ k tr
_
T
k1
%

+&
_
+
2
_
conse-uently
*$ (T; %) = k tr
_
T
k1
%

. (',)
Consider now the tensor T*
_
V = R
3
_
. The characteristic e-uation associated
with the tensor T is iven by
det [T ,I] = p (,) = 0
!#
i.e.
p (,) = ,
3
I
T
,
2
+II
T
, III
T
= 0 ('/)
The invariants of T* denoted by I
T
* II
T
and III
T
* are iven by
I
T
= tr [T] (+0)
II
T
=
1
2
_
I
2
T
tr
_
T
2
_
III
T
= det [T] =
1
6
_
(tr [T])
3
3tr
_
T
2

tr [T] + 2tr
_
T
3

_
>owever* the Cayley->amilton theorem states that any tensor T satis"es its
characteristics e-uation* i.e.*
T
3
I
T
T
2
+II
T
T III
T
I = 0 (+!)
5rom the above results* we may derive:
-I
T
-T
= I (+#)
-II
T
-T
= (I tr [T] T)
T
-III
T
-T
=
_
T
2
I
T
T +II
T
I
_
T
6otice that*
-I
T
-T
ij
=
-T
kk
-T
ij
=
ik

kj
=
ij
hence
-I
T
-T
= I.
:oreover* since tr [T] = I.T = I
T
-II
T
-T
ij
=
1
2
_
2 (tr [T])
-I
T
-T
ij

-
-T
ij
(T
rk
T
kr
)
_
= I
T

ij

1
2
_
-T
rk
-T
ij
T
kr
+
-T
kr
-T
ij
T
rk
_
= I
T

ij

1
2
(
ir

jk
T
kr
+
ik

jr
T
rk
)
= I
T

ij

1
2
(T
ji
+T
ji
)
= I
T

ij
T
ji
thus
-II
T
-T
= I
T
I T
T
5orm the Cayley->amilton theorem*
T
3
I
T
T
2
+TII
T
III
T
I = 0
!$
Ta)in the trace of the above function* one derives
tr
_
T
3

I
T
tr
_
T
2

+tr [T] II
T
3III
T
= 0
1ince I
T
= tr [T]
tr
_
T
3

I
T
tr
_
T
2

+I
T
II
T
3III
T
= 0
T
ir
T
rs
T
si
I
T
T
is
T
si
+I
T
II
T
3III
T
= 0
conse-uently
-
-T
ab
T
ir
T
rs
T
si
I
T
T
is
T
si
+I
T
II
T
3III
T
= 0
i.e.

ia

rb
T
rs
T
si
+T
ir

ra

sb
T
si
+T
ir
T
rs

sa

ib

ab
T
is
T
si
I
T

ia

sb
T
si
I
T
T
is

sa

ib
+
ab
II
T
+I
T
(I
T

ab
T
ba
) 3
-III
T
-T
ab
= 0
thus
T
bs
T
sa
+T
ia
T
bi
+T
br
T
ra

ab
T
is
T
si
I
T
T
ba
I
T
T
ba
+
ab
II
T
+I
T
(I
T

ab
T
ba
) 3
-III
T
-T
ab
= 0
1o* in a compact notation we have
3
_
T
2
_
T
tr
_
T
2

I 2I
T
T
T
+II
T
I +I
T
_
I
T
I T
T
_
= 3
-III
T
-T
.
=ut II
T
=
1
2
_
(tr [T])
2
tr
_
T
2

_
* then
3
-III
T
-T
= 3
_
T
2
_
T
+
_
I
2
T
tr
_
T
2
_
I +II
T
I 3I
T
T
T
= 3
_
T
2
_
T
+ 3II
T
I 3I
T
T
T
i.e.
-III
T
-T
=
_
T
2
_
T
+II
T
I I
T
T
T
what can be written as
-III
T
-T
=
_
T
2
I
T
T +II
T
I
_
T
.
6ow* T
3
I
T
T
2
+TII
T
III
T
I = 0* so
T
2
I
T
T +II
T
I III
T
T
1
= 0
!%
i.e.
T
2
I
T
T +II
T
I = III
T
T
1
which allow us to write
-III
T
-T
= III
T
T
T
. (+$)
Consider now the relation
!
!t
det [T (t)] .
Then* we have
!
!t
det [T (t)] =
!
!t
III
T
(t) (+%)
=
-III
T
-T
rs
!
!t
T
rs
(t)
=
-III
T
-T
.

T (t)
1ubstitutin (+$) into (+%) we derive
!
!t
det [T (t)] = det [T (t)] T
T
.

T
= det [T (t)]

T.T
T
= det [T (t)] tr
_

TT
1
_
conse-uently
!
!t
det [T (t)] = det [T] tr
_

TT
1
_
. (+&)
1./.$ ,erivatives of vetorial valued functions
The vector valued functions

$ : D V may have vector (D V) or tensor
(D L) type of aruments. Consider the vector valued function with a vector
arument:

$ : D V V. 7e say that

$ is diferentiable at x* alon the
direction u* when the followin limit e.ists
*

$ (x; u) = lim
h0

$ (x ++ u)

$ (x)
+
=
!
! +
_

$ (x ++ u)
_

h=0
(+')
2f

$ is di<erentiable* then
*

$ (x; u) = [$ (x)] u (++)

!&
,e2nition Consider the vector valued function with a vector arument:

$ :
D V V. Then* the diverence of

$ (x)* represented by !iv
_

$ (x)
_
* is de"ned
by
!iv
_

$ (x)
_
:= tr
_

$ (x)
_
(+,)
2n a cartesian coordinate system*
!iv
_

$ (x)
_
=
-$
i
-x
i
= $
i,i
(+/)
The rotational of

$ (x)* represented by r&t
_

$ (x)
_
* is de"ned by
r&t
_

$ (x)
_
v = 2
_

$ (x)
_
Skew
v = 2 w
f
v (,0)
where w
f
is the a.ial vector associated with the 1)ew part of $ (x)* with
_

$ (x)
_
Skew
=
1
2
_
$ (x) $ (x)
T
_
2n a cartesian coordinate system* we have
r&t [$ (x)]
i
=
ijk
-$
k
-x
j
(x) (,!)
6otice that* if a tensor A is 1)ew*
_
V = R
3
_
* then A = A
T
implies
A =
_
_
0 a
3
a
2
a
3
0 a
1
a
2
a
1
0
_
_
(,#)
where w
A
= (a
1
, a
2
, a
3
).
1./.& ,erivatives of tensorial valued functions
The tensor valued functions $ : D L may have vector (D V) or tensor
(D L) type of aruments. Consider the tensor valued function with a tensor
arument:

$ : D L L. 7e say that $ is diferentiable at T* alon the
direction %* when the followin limit e.ists
*$ (T; %) = lim
h0
$ (T ++ %) $ (T)
+
=
!
! +
$ (T ++ %)

h=0
(,$)
2f $ is di<erentiable* then
*$ (T; %) = [$ (T)] % (,%)
2n terms of a cartesian coordinate system*
*$ (T; %)
ij
= [$ (T)]
ijkl
%
kl
!'
,ivergent of a tensorial 2eld The diverence of a tensor "eld . (x) L*
at x* is the only vector de"ned by
!iv [. (x)] .v = !iv
_
[. (x)]
T
v
_
, v V (,&)
2n terms of a cartesian coordinate system*
!iv [. (x)]
i
=
-.
ij
-x
j
(x) (,')
3otational of a tensorial 2eld The rotational of a tensor "eld* at x* is a
tensor de"ned by
r&t [. (x)] v = r&t
_
[. (x)]
T
v
_
, v V (,+)
2n terms of a cartesian coordinate system*
r&t [. (x)]
ij
=
imk
-.
jk
-x
m
(x) (,,)
Laplacian of a scalar 2eld 2s the scalar valued function* / : x V R*
de"ned by
/ = !iv [/(x)] (,/)
2n terms of a cartesian coordinate system*
/ =
-
2
/
-x
2
i
(x) = /
,ii
(x) (/0)
Laplacian of a vector valued 2eld 2s the scalar valued function*

$ : x
V V* de"ned by

$ = !iv
_

$ (x)
_
(/!)
2n terms of a cartesian coordinate system*
$
i
=
-
2
$
i
-x
2
j
(x) = $
i,jj
(x) (/#)
Laplacian of a tensor valued 2eld 2s the tensor valued function* T : x
V L* de"ned by
[T (x)] v = ([T (x)] v) , v V (/$)
2n terms of a cartesian coordinate system*
[T (x)]
ij
=
-
2
T
ij
-x
2
k
(x) = T
ij,kk
(x) (/%)
!+
Properties 5orm the above relations* we may derive
!. !iv (u v) = 0ra! [u] v +u !iv [v]
#. !iv (/ u) = 0ra! [/] .u +/ !iv [u]
$. 0ra! [/ u] = u 0ra! [/] +/ 0ra! [u]
%. !iv
__
T
T

u
_
= u.!iv [T] +T .0ra! [u]
&. !iv (/ T) = / !iv [T] + [T] 0ra! [/]
'. r&t [/ u] = 0ra! [/] u +/ r&t [u]
+. r&t [u v] = 0ra! [u] v 0ra! [v] u +u !iv [v] v !iv [u]
,. !iv (u v) =v.r&t [u] u.r&t [v]
/. r&t [v] u = 0ra! [v] u 0ra! [v]
T
u
!0. r&t [0ra! [/]] = 0
!!. !iv [r&t [u]] = 0
!#. r&t [r&t [u]] = 0ra! [!iv [u]] u
1./.' ,erivative of tensorial and vetorial 2elds parametri4ed by a
scalar variable "t5time#

d
dt
__
a(t)

b (t)
__
=
_
d
dt
a (t)

b (t)
_
+
_
a(t)
d
dt

b (t)
_

d
dt
T
1
= T
1

T T
1
2n fact* since
T T
1
= I
we obtain* by di<erentiation that

T T
1
+T

T
1
= 0
thus

T
1
= T
1

T T
1
(/&)
!,
2 Integrals of Tensor 2elds
>ere* we are considerin tensorial "elds : D Y* where R are reular reions
contained in D V* i.e.* is a reion contained in D with a reular contour -R.
A reion is denoted reular if it is bounded* orientable and has a continuous
normal vector "eld* pointed outward from R.
The eneral e.pression for the diverence theorem is iven by
_
R

ijk...
n

!A =
_
R
-
ijk...
-x

!' (/')
where
ijk...
denote tensorial components of class %
1
in R.
5rom the eneral e.pression* we may derive the followin particular e.pres-
sions
2.6.( Scalar valued 2elds
Let /(x) : V R* then
_
R
/ n !A =
_
R
0ra! [/] !' (/+)
2n a cartesian coordinate system*
_
R
/ n

!A =
_
R
-/
-x

!' (/,)
2.6.+ 7ector valued 2elds
Let u(x) : V V* then
_
R
u.n !A =
_
R
!iv [u] !' (//)
2n a cartesian coordinate system*
_
R
u
i
n
i
!A =
_
R
-u
i
-x
i
!' (!00)
2.6.- Tensor valued 2elds
Let T (x) : V L* then
_
R
Tn !A =
_
R
!iv [T] !' (!0!)
2n a cartesian coordinate system*
_
R
T
ij
n
j
!A =
_
R
-T
ij
-x
j
!' (!0#)
!/
Let
ij
=
ijk
u
k
. Then
_
R

ijk
u
k
n
j
!A =
_
R

ijk
-u
k
-x
j
!' (!0$)
what may be written in a compact form as
_
R
n u !A =
_
R
r&t [u] !' . (!0%)
Also*
_
R
u
i
n
j
!A =
_
R
-u
i
-x
j
!' (!0&)
what may be written in a compact form as
_
R
u n !A =
_
R
0ra! [u] !' . (!0')
:oreover*
_
R
u Tn !A =
_
R
0ra! [u] T
T
+u !iv [T] !' . (!0+)
6ow* considerin u = 1/* where both 1 and / are scalar valued "elds* we
derive
_
R
1/.n !A =
_
R
!iv [1/] !'
=
_
R
(1./ +1!iv [/]) !'
=
_
R
(1./ +1/) !'
replacin 1 by / we derive
_
R
(1/ /1) .n !A =
_
R
(1/ /1) !' . (!0,)
2.1 Stokes Theorem
Consider now the reular surface S with a closed contour %. The eneral
_
C

ijk...
!x
i
=
_
S

pqr
-
rjk...
-x
q
n
p
!A (!0/)
where !x
i
are the components of the tanent vector to % at x.
#0
$ 8omeork 92
i) Consider that 2 R, v and u V and S L. 1how that:
!. (2v) = 2v +v 24
#. !iv (2v) = 2!iv (v) +v.24
$. !iv
_
S
T
v
_
= S.v +v.!iv (S)4
%. !iv (2S) = 2!iv (S) + [S] 24
&. r&t (2u) = 2 u +2r&t (u)4
'. !iv (u v) =v.r&t (u) u.r&t (v)4
ii) Let v and w V and S L. 1how that:
!.
_

v.Sn ! =
_

v.!iv (S) +S.v !

#.
_

v. ( w.n) ! =
_

v.!iv ( w) + [v] w !
$.
_

Sn v ! =
_

_
!iv (S) v + [S] [v]
T
_
!
iii) show that
!. 3
ijk
3
mjk
= 2
im
#. 3
ijk
3
ijk
= 6
$. r&t (u)
i
= 3
ijk
uk
xj
%. [r&t (T)]
ij
= 3
imk
T
jk
x
m
&.
_

$ (x)
_
i
=

2
fi
x
2
j
= $
i,jj
& Tensors Operations
&.1 Orthogonal tensors
A tensor 4 is said to be orthoonal if
4
T
= 4
1
. (!!0)
This de"nition implies that the determinat of any orthoonal tensor e-uals either
+1 or 1. An orthoonal tensor 4 with
det (4) = 1 (!!!)
#!
is called a proper orthoonal tensor (or rotation). The product 4
1
4
2
of any
two orthoonal tensors 4
1
and 4
2
is an orthoonal tensor. 5or all vectors u
and v* an orthoonal tensor 4 satis"es:
4u.4v = u.v (!!#)
;otations and chanes of basis
Let e
i
, i = 1...n and e

i
, i = 1...n be two orthoonal bases of U. 1uch
bases are related by:
e

j
= Re
j
* for j = 1...n* (!!$)
where R is a rotation (proper orthoonal tensor). Let T and u be* respectively*
a tensor and a vector with matri. representation [T] and u with respect to
the basis e
i
* i = 1...n. The matri. representations [T

] and u

of T and
u relative to the basis e

i
* i = 1...n are iven by the followin products of
matrices:
[T

] = [R]
T
[T] [R] and u

= [R]
T
u . (!!%)
E-uivalently* in component form* we have:
T

ij
= R
ki
T
kl
R
lj
and u

i
= R
ji
u
j
. (!!&)
The matri. [R] is iven by:
[R] =
_

_
e
1
.e

1
e
1
.e

2
. . . e
1
.e

n
e
2
.e

1
e
2
.e

2
. . . e
2
.e

n
.
.
.
.
.
.
.
.
.
.
.
.
e
n
.e

1
e
n
.e

2
e
n
.e

n
_

_
* (!!')
or* in component form*
R
ij
=e
i
.e

j
. (!!+)
&.1.1 )*ample: 0 rotation matri* in to dimensions
2n a two dimensional space* the rotation tensor has a simple cartesian repre-
sentation. Let the tensor R be a transformation that rotates all vectors of
the two-dimensional space by an (anti-cloc)wise positive) anle 5. The matri.
representation of R is iven as:
[R] =
_
cos (5) sin(5)
sin(5) cos (5)
_
. (!!,)
&.2 Spectral decomposition
?iven a tensor T* a non-3ero vector u is said to be an eienvector of T associated
with a iven eienvalue , if
Tu = , u. (!!/)
The space of all vectors u satisfyin the above relation is called the eien-space
(or characteristic space) of T correspondin to ,. The followin properties hold:
##
"i# The eienvalues of a positive de"nite tensor are strictly positive
Proof: Let ,
i
be an eienvalue of A* a positive de"nite tensor. Then*
v
i
* |v
i
| = 1* so that
Av
i
= ,
i
v
i
(!#0)
therefore
,
i
= ,
i
v
i
, v
i
= ,
i
|v
i
|
2
(!#!)
= Av
i
.v
i
> 0.
"ii# The characteristic spaces of a symmetric tensor are mutually orthoonal.
Proof: let 6 and , be distinct eienvalues of a symmetric tensor S* then
there e.ists v and w* with |v| = 1 and | w| = 1 so that
Sv = 6v
and
S w = , w
(!##)
then
6v. w = Sv. w =v.S w
and
, w.v = S w.v
(!#$)
subtractin both e-uations yield
(6 ,) v. w = 0. (!#%)
1ince (6 ,) ,= 0* we must have
v. w = 0. (!#&)
&.2.1 Spectral theorem
Let S be a symmetric tensor. Then S admits the representation
S =
n

i=1
7
i
(v
i
v
i
) * (!#')
where (v
i
, i = 1...n) is an orthonormal basis for U consistin e.clusively of
eienvectors of S and 7
i
are the correspondin eienvalues. The above rep-
resentation is called the spectral decomposition of S. ;elative to the basis
(v
i
, i = 1...n)* S has the followin diaonal representation
[S] =
_

_
7
1
0 . . . 0
0 7
2
. . . 0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 7
n
_

_
. (!#+)
#$
&.2.2 )igenpro;ections
Alternatively* with p n de"ned as the number of distinct eienvalues of S* we
may write
S =
p

i=1
7
i
E
i
, (!#,)
where the symmetric tensors E
i
are called the eienpro9ections of S. Each
eienpro9ection E
i
is the orthoonal pro9ection operator on the characteristic
space of S associated with 7
i
. The eienpro9ections have the property
I =
p

i=1
E
i
* (!#/)
and* if p = n (no multiple eienvalues)* then
E
i
= (v
i
v
i
) , for i = 1...n. (!$0)
Also* the eienpro9ections satisfy
E
i
=
_

p
j=1
j=i
1
ij
(S 7
j
I) if p > 1
I if p = 1
(!$!)
2n the particular case in which n = 3* we have:
"i# 2n the vase 7
1
,= 7
2
,= 7
3
* we have
E
i
= (v
i
v
i
) , for i = 1..3* (!$#)
where v
i
is the eienvector associated with 7
i
. Therefore
S =
3

i=1
7
i
E
i
. (!$$)
"ii# 2n the vase 7
1
,= 7
2
= 7
3
* we have
E
1
= (v
1
v
1
)
and
E
2
= I E
1
= I (v
1
v
1
)
(!$%)
where v
1
is the eienvector associated with 7
1
. Therefore
S = 7
1
(v
1
v
1
) +7
2
[I (v
1
v
1
)] (!$&)
"iii# 2n the vase 7
1
= 7
2
= 7
3
* we have
E
1
= I. (!$')
Therefore
S = 7
1
I. (!$+)
#%
&.2.$ %haracteristic e<uation and Principal invariants
Every eienvalue 7
i
satis"es the characteristic e-uation
p (7
i
) = det (S 7
i
I) = 0. (!$,)
2n the three-dimensional space* for any * det (S I) admits the followin
representation
det (S I) =
3
+
2
I
S
II
S
+III
S
* (!$/)
where I
S
, II
S
and III
S
are the principal invariants of S* de"ned by
I
S
= tr (S) = S
ii
4 (!%0)
II
S
=
1
2
_
tr
_
S
2
_
tr (S)
2
_
=
1
2
(S
ii
S
jj
S
ij
S
ji
) 4
III
S
= det (S) =
1
6
3
ijk
3
pqr
S
ip
S
jq
S
kr
.
2n this case* the characteristic e-uation reads
7
3
i
+7
2
i
I
S
7
i
II
S
+III
S
= 0 (!%!)
and the eienvalues 7
i
are the solution of this cubic e-uation.
2f S is symmetric* then its principal invariants can be e.pressed in terms of
its eienvalues as
I
S
= 7
1
+7
2
+7
3
4 (!%#)
II
S
= 7
1
7
2
+7
2
7
3
+7
1
7
3
4
III
S
= 7
1
7
2
7
3
.
&.2.& Polar decomposition
Let . be a positive de"nite tensor. Then there e.ists symmetric positive de"nite
tensors U and ' *and a rotation R such that
. = RU = ' R. (!%$)
The decomposition RU and ' R are uni-ue and are called* respectively* the
riht and left polar decompositions of .. The symmetric tensors U and ' are
iven by
U =

.
T
. and ' =

..
T
* (!%%)
where

denotes the tensor s-uare root. The s-uare root of a symmetric tensor
A is the uni-ue tensor 8 that satis"es
8
2
88 = A. (!%&)
Let
A =

i
,
a
i
(v
a
i
v
a
i
) (!%')
#&
with ,
a
i
and v
a
i
denotin* respectively* the eienvalues and the basis of eien-
vectors of A. The spectral decomposition of its s-uare root* 8* reads
8 =

i
_
,
a
i
(v
a
i
v
a
i
) . (!%+)
&.$ Special tensors
The deviator of a symmetric tensor T* denoted T
de!
* is de"ned as
T
de!
: = T
1
3
(I.T) I (!%,)
= T
1
3
T
!ol
I
with
T
!ol
:= tr (T) = I.T (!%/)
and it follows that
tr (T
de!
) = I.T
de!
= 0. (!&0)
The spherical part of T* denoted T
sph
* is de"ned as
T
sph
:= T T
de!
=
1
3
T
!ol
I =
1
3
[I I] T (!&!)
Assume that T is a ran)-one update of I. 2ts inverse can be computed
e.plicitly accordin to the 1herman-:orrison formula:
T = I +(u v) (!&#)
then
T
1
= I

1 +u, v
(u v) (!&$)
where u and v are arbitrary vectors and is an arbitrary scalar such that
,=
1
u, v
*
so that T is non-sinular.
Proof: Let
T
1
= I + (u v) . (!&%)
Then* in order to compute we impose that
TT
1
= T
1
T = I (!&&)
A straihtforward enerali3ation of the formula in (!&$) is the followin: 2f
T = U +(u v) (!&')
#'
then
T
1
= U
1

1+"
1
#u,#!
_
U
1
u U
T
v
_
i.e.
T
1
= U
1

1+"
1
#u,#!
U
1
(u v) U
1
(!&+)
where it is assumed that @ is a non-sinular tensor.
Proof: E.press T = U

T with

T = I +U
1
uv* such that T
1
=

T
1
U
1
*
and use (!&$).
' 8igher order tensors
1o far we have seen operations involvin scalars* that can be considered as 3ero
order tensors* vectors* which can be considered "rts order tensors* and second
order tensors* which are associated with linear operators (or transformations) on
vectors. Linear operators of hiher order* or hiher order tensors* are fre-uently
employed in continuum mechanics. 2n this section we introduce some basic
de"nitions and operations involvin hiher order tensors.
A third order tensor may be represented as
A = A
ijk
(e
i
e
j
e
k
) * (!&,)
with the de"nition _
a

b
_

! =
_
.

!
__
a

b
_
* (!&/)
'.1 1ourth order tensor
5ourth order tensors are particularly relevant in continuum mechanics. A en-
eral fourth order tensor D is represented as
D = D
ijks
(e
i
e
j
e
k
e
s
) . (!'0)
5ourth order tensors map second order tensors into second order tensors. They
also map vectors in third order tensors and third order tensors into vectors.
As a direct e.tension* we de"ne
_
a

b

!
_
e =
_
e.

!
__
a

b
_
* (!'!)
and the double contractions
_
a

b

!
__
e

$
_

_
a

b

!
_
:
_
e

$
_
(!'#)
= (.e)
_

!.

$
__
a

b
_
*
and
_
a

b

!
__
e

$ 0

+
_

_
a

b

!
_
:
_
e

$ 0

+
_
(!'$)
= (.e)
_

!.

$
__
a

b 0

+
_
*
with the above de"nitions* the followin reations are valid
#+
"i# D
ijkl
= D(e
k
e
l
) . (e
i
e
j
) (e
i
e
j
) : D: (e
k
e
l
)
"ii# Du = D
ijkl
u
l
(e
i
e
j
e
k
)
"iii# DS= D
ijks
S
ks
(e
i
e
j
) D: S
Consider the enerali3ed >oo)As law.
# = D. (!'%)
2n components* we have
# = #
ij
(e
i
e
j
) (!'&)
= D
ijks
(e
i
e
j
e
k
e
s
)
nl
(e
n
e
l
)
= D
ijks

nl
e
k
, e
n
e
s
, e
l
(e
i
e
j
)
= D
ijks

nl

kn

sl
(e
i
e
j
)
= D
ijks

ks
(e
i
e
j
)
"iv# D
T
S= D
ijks
S
ij
(e
k
e
s
) S: D4
6otice that
DU, S =

D
T
S, U
_
* U and S 9in(V, V) . (!'')
2n components
(D
ijks
U
ks
) S
ij
= (D
ijks
S
ij
) U
ks
(!'+)
"v# DT = D
ijmn
T
mnkl
(e
i
e
j
e
k
e
l
) D: T.
This represents the followin composition
(DT) U = D(TU) (!',)
where Dand T are linear transformations mappin 9in(V, V) 9in(V, V).
'.1.1 Symmetry
7e shall call symmetric any fourth order tensor that satis"es
DS, U = S, DU * U and S 9in(V, V) . (!'/)
2n dyadic we have
S: D: U = (D: S) : U
for any second order tensors S and U.
This de"nition is analoous to that of symmetric second order tensors. The
cartesian components of symmetric fourth order tensors satisfy
D
ijkl
= D
klij
. (!+0)
#,
2t should be noted that other symmetries are possible in fourth order tensors.
2f symmetry occurs in the last two indices* i.e.* if
D
ijkl
= D
ijlk
(!+!)
the tensor has the properties:
DS = DS
T
(!+#)
in dyadics
D: S = D: S
T
and S: D =(S: D)
T
for any S. 2f it is symmetric in the "rst two indices* i.e.*
D
ijkl
= D
jikl
(!+$)
then*
DS = (DS)
T
(!+%)
in dyadics
D: S = (D: S)
T
and S: D =S
T
: D.
'.1.2 %hange of basis transformation
Aain* let us consider the orthoonal basis e

i
* i = 1, ...n de"ned as
e

j
= Re
j
(!+&)
with R a rotation. The components D

ijkl
of a tensor D relative to the basis
de"ned by e

i
* i = 1, ...n are iven by
D

ijkl
= R
mi
R
nj
R
pk
R
ql
D
mnpq
(!+')
where D
mnpq
are the components of D relative to e
i
. 2n fact
D = D

ijkl
_
e

i
e

j
e

k
e

l
_
(!++)
= D
mnpq
(e
m
e
n
e
p
e
q
) .
6ow
e

j
= Re
j
(!+,)
= Re
j
, e
m
e
m
= R
mj
e
m
therefore
D = D

ijkl
_
e

i
e

j
e

k
e

l
_
(!+/)
= D

ijkl
R
mi
R
nj
R
pk
R
ql
(e
m
e
n
e
p
e
q
)
= D
mnpq
(e
m
e
n
e
p
e
q
)
#/
hence
D

ijkl
R
mi
R
nj
R
pk
R
ql
= D
mnpq
. (!,0)
Also
e
m
= R
T
e

m
(!,!)
=

R
T
e

m
, e

i
_
e

i
= R
T
im
e

i
= R
mi
e

i
therefore
D= D
mnpq
(e
m
e
n
e
p
e
q
) (!,#)
= D
mnpq
R
mi
R
nj
R
pk
R
ql
_
e

i
e

j
e

k
e

l
_
= D

ijkl
_
e

i
e

j
e

k
e

l
_
hence
D

ijkl
= R
mi
R
nj
R
pk
R
ql
D
mnpq
. (!,$)
'.1.$ Isotropic Tensors
A tensor is said to be isotropic if its components are invariant under any chane
of basis. The only second order isotropic tensors are the so-called spherical
tensors* i.e.* the tensors de"ned as
I (!,%)
with scalar .
Any isotropic fourth order tensor U can be constructed as a linear combina-
tion of three basic isotropic tensors* I* I
T
and (I I):
U = I +I
T
+: (I I) (!,&)
where * and : are scalars.
The tensor I is called the fourth order identity. 2ts components are:
I
ijkl
=
ik

jl
. (!,')
5or any second order tensor S* we have
IS = S* S 9in(V, V) (!,+)
in dyadics
I: S = S: I = S.
:oreover* for any fourth order tensor T
IT = TI = T* T (!,,)
$0
in dyadics
I: T = T: I = T.
The tensor I
T
is the transposition tensor. 2t maps any second order tensor
onto its transpose* i.e.*
I
T
S = S
T
* S 9in(V, V) (!,/)
in dyadics
I
T
: S = S: I
T
= S
T
for any S. The components of I
T
are
(I
T
)
ijkl
=
ij

kl
. (!/0)
5inally* the tensor (I I) has components
(I I)
ijkl
=
ij

kl
. (!/!)
7hen applied to any tensor T it ives
(I I) .T (I I) : T (!/#)
= tr (T) I.
Another important isotropic tensor that fre-uently appears in continuum me-
chanics is the tensor de"ned as
I
Sym
=
1
2
(I +I
T
) . (!/$)
This tensor maps any second order tensor into its symmetric part* i.e.*
I
Sym
T = T
sym
* T 9in(V, V) (!/%)
in dyadics
I
Sym
: T = T : I
Sym
= T
sym
* T 9in(V, V) .
This tensor is denoted as the symmetric pro9ection or symmetric identity. 2ts
components are iven by:
(I
Sym
)
ijkl
=
1
2
(
ik

jl
+
il

jk
) . (!/&)
=y analoy* we can de"ned I
Skew
as
I
Skew
=
1
2
(I I
T
) . (!/')
This tensor maps any second order tensor into its symmetric part* i.e.*
I
Skew
T = T
Skew
* T 9in(V, V) . (!/+)
$!
This tensor is denoted as the symmetric pro9ection or symmetric identity. 2ts
components are iven by:
(I
Skew
)
ijkl
=
1
2
(
ik

jl

il

jk
) . (!/,)
?eneric tensors of order m are de"ned as
=
i1i2im
(e
i1
e
i2
e
im
)
where* e.tendin the previous de"nitions of the tensor product* we have
(e
i1
e
i2
e
im
) u = (u.e
im
)
_
e
i1
e
i2
e
im1
_
for all u U. The de"nition of contraction operations are completely analoous
to those seen above for fourth order tensors.
'.2 )lementary algebra of &th order tensors
'.2.1 %omponent representation
The simplest form of a %th order tensor A is a -uad* which is de"ned as the
tensor product of two #nd order tensors T and U* i.e.
A = T U = [T
ij
(e
i
e
j
)] [U
kl
(e
k
e
l
)] = T
ij
U
kl
(e
i
e
j
e
k
e
l
) (!//)
The products (e
i
e
j
e
k
e
l
)* which are denoted the base -uadrads* form
the basis of the product space R
3
R
3
R
3
R
3
. The e.pression in (!//) is*
clearly* only a special case of the eneral representation of a %th order tensor
A = A
ijkl
(e
i
e
j
e
k
e
l
) (#00)
Any %th order tensor de"nes a linear mappin from R
3
R
3
to R
3
R
3
* since
AS = A
ijkl
(e
i
e
j
e
k
e
l
) S
rs
(e
r
e
s
) (#0!)
= A
ijkl
S
rs
(e
i
e
j
e
k
e
l
) (e
r
e
s
)
= A
ijkl
S
rs
e
k
, e
r
e
l
, e
s
(e
i
e
j
)
= A
ijkl
S
rs

kr

ls
(e
i
e
j
)
= A
ijkl
S
kl
(e
i
e
j
)
= U
ij
(e
i
e
j
)
= U
where we introduced the tensor @ with components U
ij
= A
ijkl
S
kl
.
@seful notations are the Boverline open productC and the Bunderline open
productC * which are de"ned via the component representations
T U := T
ik
U
jl
(e
i
e
j
e
k
e
l
)
and
T U := T
il
U
jk
(e
i
e
j
e
k
e
l
)
(#0#)
$#
@seful rules* that involve the open ( Tensor) product symbols* for #nd order
tensors U* ' and ; are:
[U ' ] ; = ', ; U
[U ' ] ; = U;'
T
[U ' ] ; = U;
T
'
T
(#0$)
in dyadics we also have
; : [U ' ] = U, ; '
; : [U ' ] = U
T
;'
; : [U ' ] =
_
U
T
;'
_
T
= '
T
;
T
U.
(#0%)
'.2.2 Symmetry and ske5symmetry
The ma9or transpose of a %th order tensor A is de"ned as
A
T
= A
klij
(e
i
e
j
e
k
e
l
) (#0&)
i.e. the transpose is associated with a Bma9or shiftC of indices. The ma9or-
symmetric part of A* denoted A
sym
* and the ma9or-s)ew-symmetric part of A*
denoted A
skew
* are de"ned as follows:
A
sym
=
1
2
_
A +A
T

and
A
skew
=
1
2
_
A A
T

(#0')
A possesses ma9or symmetry if A
sym
= A (and A
skew
= 0)* i.e. when A = A
T
.
2n component form*
A
ijkl
= A
klij
. (#0+)
A possesses ma9or s)ew-symmetry when A
skew
= A (and A
sym
= 0)* i.e. when
A = A
T
. 2n component form*
A
ijkl
= A
klij
* (#0,)
which (in particular) infers that
A
ijkl
= 0* for ij = k<. (#0/)
:oreover* A possesses !st and #nd minor symmetry if
A
ijkl
= A
jikl
and
A
ijkl
= A
ijlk
(#!0a)
respectively. Li)ewise* A possesses !st and #nd minor s)ew-symmetry if
A
ijkl
= A
jikl
and
A
ijkl
= A
ijlk
(#!!)
respectively.
$$