Tensors

References:

P. Chadwick, Contiuum Mechanics, Dover Publications, 1976.

M. E. Gurtin, An Introduction to Continuum Mechanics,
Academic Press, 1981.

Acknowledgements:

H. Garmestani, Georgia Tech
A. Rollett, CMU
Mappings
Sets: Collection of Objects.
Elements: Constituents of a Set.

Mapping (or Transformation):
f maps D in C
Domain of f: Set D
Codomain of f: Set C
Range of f:
Range is a subset of codomain: range
f maps D onto C: range
One-to-one mapping:
A e a
A e e
{ } c b a A , , =
C D f :
} ) ( { D x x f f e =
C f _
C f =
2 1 2 1
) ( ) ( x x x f x f = =
Tensors
Tensors : mathematical representation of
physical entities or phenomena (e.g. Stress,
Strain, Elastic Stiffness)
Tensors of different rank are used to
represent different types of physical
quantities
Tensors of a given rank have certain
common properties and follow certain rules
(e.g. coordinate transformation laws)
Tensors are not simply sets or matrices
Tensors do not require definition of a
coordinate system
Scalars: Tensors of Rank Zero

Scalars may be specified completely by
a single number (e.g. density,
temperature, energy)

Scalars are not constants, but may
actually be functions of position and/or
time.
Points: Rank One Tensors
a
b
c
d
e
f
g
Euclidean Point Space E

Set of all points in three
dimensional space

Labeling of the points does not
require a coordinate frame!

Operations on points are not
meaningful
Vectors: Rank One Tensors
Defined as difference for a
given ordered pair (a,b): v =
b a
Euclidean Vector Space V
Still no need of a reference
frame!
a
b
c
d
e
f
g
Operations on Vectors:
Dot Product
v u = o
u
v
u cos v u v u =
u
u u u =
Magnitude or Norm
V e = v u v u u v ,
9 e e + = + | o | o | o , , , ) ( V w v u w v w u w v u
V e > u u u 0 0 u u u = = iff 0
No need for a reference frame yet!
t u s s 0
Operations on Vectors:
Dot Product
1 = n Unit Vector:
Orthogonal Vectors:
0 = v u
Several physical quantities possess the
characteristics described here for Vectors;
therefore they are all classified as Rank One
Tensors.
e.g. Force, Displacement, Velocity
These quantities require not only an assignment
of magnitude, but also a specification of direction.
Coordinate Frames
Right-handed, orthogonal, Cartesian Coordinate system.

Set of orthonormal basis vectors
}

{ }

{
3 2 1 i
e e e e =
Origin o
ij j i
j i if
j i if
o =
)
`

=
=
=
0
1

e e
Kronecker delta o
ij

Coordinate Frames

v

=
i
i i
v e v
Components of a vector v:
v e =
i i
v
( )
i
i
i i
v u e v u w

+ = + =
Adding Vectors (parallelogram law):
u
Indicial Notation:
i i i
v u w + =
( )
2 / 1
i i
u u = = u u u
Einsteins Summation
Convention
i i
v u = v u
w
Coordinate Transformation
Matrices: 2-D
Or, inverting
The graph above represents a transformation of coordinates when the system is
rotated at an angle u CCW

)
`

'
'
(

=
)
`

' + ' =
' ' =
y
x
y
x
or
y x y
y x x
u u
u u
u u
u u
cos sin
sin cos

sin cos
sin cos
)
`

=
)
`

'
'
+ = '
+ = '
y
x
y
x
or
y x y
y x x
u u
u u
u u
u u
cos sin
sin cos

cos sin
sin cos
Coordinate Transformation
Matrices: 3-D
} { } , , {
3 2 1 i
e e e e =
First Frame:
} { } , , {
3 2 1 i
e e e e
'
=
' ' '
Second Frame:
i i i i
v v e e v
' '
= =
j ij i j i i j i i
v v v v
'
=
'
=
'
=
'

' '
o e e e e
( )
i ji i j i j
v Q v v =
'
=
'
e e
{ } | |{ }
j i ij
Q v Q v e e
'
= =
'
Coordinate Transformation
Matrices: 3-D
ij jk ik kl jl ik l jl k ik j i
Q Q Q Q Q Q o o = = = =
'

'
e e e e
Six equations
Three independent parameters in [Q]
I =
T
QQ
T
Q Q =
1
1 det = Q
Coordinate Transformations:
Euler Angles
A sequence of three rotations (called Euler
angles) are needed to completely specify a
arbitrary transformation.
e
1
e
2
e
3
e

1
e

2
e

3
Rotation 1: Angle |
1
about e
3
so
that the new orientation of e
1
is
perpendicular to e
3

.
Rotation 2: Angle u about the new
e
1
so that the new orientation of e
3
is
parallel to e
3

.
Rotation 3: Angle |
2
about e
3

to
bring the two frames in coincidence.
Bunge Euler Angles
g ={
1
,|,
2
}
(0 s
1
s 2t, 0 s| st, 0s
2
s2t)
(1) ) 2 ( ) 3 ( e
lj
e
kl
e
ik
e
ij
R R R R =
(
(
(

+
+
=
| | |
| | |
| | |
cos sin cos sin sin
cos sin cos cos cos sin sin cos cos sin sin cos
sin sin sin cos cos cos sin sin cos sin cos cos
1 1
2 2 1 2 1 2 1 2 1
2 2 1 2 1 2 1 2 1
e
ij
R
Invariants
k k l k kl l k il ik l il k ik i i
u u u u u u Q Q u Q u Q u u = = = =
' '
o
All tensorial quantities possess certain
measures that are independent of the choice of
the coordinate reference frame. For example,
rank one tensors (or vectors) have one
invariant norm or magnitude of the vector.
Vector (cross) Product
v u w =
u
v
u sin v u w =
u
V e = v u u v v u ,
t u s s 0
w
V e + = + w v u w v w u w v u , , ) ( ) ( ) ( | o | o
V e = v u v u u , 0 ) (
V e = v u v u v v u u v u v u , ) ( ) )( ( ) ( ) (
2
w is perpendicular to u and v
Vector (cross) Product
, , ,
2 1 3 1 3 2 3 2 1
e e e e e e e e e = = =
k ijk j i
e e e e =
Permutation symbol
ijk
e

=
= =e =e e
=
= =e =e e
=
) ., . (
) , 3 , 2 , 1 ( ) , , ( 0
1 ., .
: ) 3 , 2 , 1 ( ) , , ( 1
) 1 ., . (
: ) , 3 , 2 , 1 ( ) , , ( 1
213 321 132
312 231 123
same the are indices two any if e i
of n permutatio a not is k j i if
e i
of n permutatio odd an is k j i if
e i
of n permutatio even an is k j i if
Vector (cross) Product
( ) ( ) ( )
j i j i j j i i
v u v u e e e e v u = =
k ijk j i
v u e v u e =
Dyadic (Outer) Product
Rank Two Tensor:
j j i
c b a = = ) ( ) ( c b a c b a
( )
ij nj m mn i j n m mn i j i ij
W W W W = = = = o e e e e e e We e
| | | |
(
(
(

=
(
(
(

=
3 3 2 3 1 3
3 2 2 2 1 2
3 1 2 1 1 1
3 2 1
3
2
1
b a b a b a
b a b a b a
b a b a b a
b b b
a
a
a
W
) ( b a W =
j i ij
b a W =
j i ij j i j i j j i i
W b a b a e e e e e e W = = =
Rank Two Tensors: Coordinate
Transformations
s r ij sj ri s sj r ri ij
j ij i s r rs j i ij
T Q Q Q Q T
Q T T
e e e e
e e e e e e T
'

'
=
'

'
=
=
' '

' '
= =
) ( ) (
] ][ ][ [ ] [
T
ij sj ri rs
Q T Q T T Q Q T =
'
=
'

or order tens n
or order tens third
or order tens second
vector
scalar
th
... ...

n
n
mnr kr jn im ijk
mnr kr jn im ijk
mn jn im ij
m im i
T Q Q Q T
T Q Q Q T
T Q Q T
a Q a
=
'
=
'
=
'
=
'
=
'
o o
Tensor (Inner) Products
(
(
(

(
(
(

=
(
(
(

3
2
1
33 32 31
23 22 21
13 12 11
3
2
1
a
a
a
T T T
T T T
T T T
b
b
b
b Ta =
( )( )
( )
i j ij i jk k ij i k j k ij
k k j i ij
a T a T a T
a T
e e e e e
e e e Ta
= = =
=
o
i j ij
b a T =
Tensor (Inner) Products
( ) ( )
k jk ij
a S T Sa T a TS =
ST TS =
(
(
(

(
(
(

(
(
(

3
2
1
33 32 31
23 22 21
13 12 11
33 32 31
23 22 21
13 12 11
a
a
a
S S S
S S S
S S S
T T T
T T T
T T T
Rank Two Tensors
Unit (spherical) Tensor or Identity Tensor:
i i j i ij
e e e e I = = o
Isotropic Tensor =
i i j i ij
C C e e e e = o
SSS S SS S I SS
0 0A 0v A IA v Iv
= = =
= = = =
3 2 1
0
Transpose of a Tensor:
i ij j i j ij
T
v S u v u S = = = v S u v Su
( ) ( ) ( ) S S S T ST T S T S = = + = +
T
T T T
T
T T
T
Rank Two Tensors
Symmetric Tensor:
T
ij
= T
ji
T = T
T


Skew-symmetric or Antisymmetric:
W
ij
= -W
ji
W = -W
T

Decomposition of a general tensor:
W S L + =
( ) ( )
T T
L L W L L S = + =
2
1
2
1
A linear transformation, from rank one tensor space V
into itself defines a rank two tensor
Tensors as Transformations
(Mappings)
c Tb =
Ta a T
Tb Ta b) T(a
o o =
+ = +
) (
R V e e + = + | o | o | o , ) ( b a, Tb Ta b a T
Mapping:
Linear Mapping:
Tensors
Yes
tensor? a is then , 3 If T a Ta =
Ta a T a a Ta
a a a T
Tb Ta b) T(a b a Tb Ta
b a b a b) T(a
o o o o o
o o o
= = =
= =
+ = + + = +
+ = + = +
) ( 3 ) 3 (
3 ) ( 3 ) (
3 3
3 3 ) ( 3
Tensors
No
tensor? a is then , 2 3 If
1
T e a Ta + =
( ) ( )
Tb Ta b) T(a
e b e a Tb Ta
e b a e b a b) T(a
+ = +
+ + + = +
+ + = + + = +
1 1
1 1
2 3 2 3
2 3 3 2 ) ( 3
Tensors as Mappings
Examples
| |
T
Q R ] [
1 0 0
0 cos sin
0 sin cos
=
(
(
(

= u u
u u
Transform vectors by a right-hand
rotation about the e
3
axis by an angle u.
u = 90
3 3
2 1 2
2 1 1
cos sin
sin cos
a b
a a b
a a b
=
+ =
=
u u
u u
3 3
1 2
2 1
a b
a b
a b
=
=
=
| |
(
(
(

=
1 0 0
0 0 1
0 1 0
R
e
1
e
2
a
b
u
V e = a b Ra
e
1
e
2
a b
Q: Let R correspond to a 90 right-hand
rigid body rotation about the x
3
axis.
Find the rotation tensor R.
Tensor (Inner) Products
Q: Let S correspond to a 90 right-
hand rigid body rotation about the x
1

axis. Find the rotation tensor S.
Q: Let W correspond to a 90 right-
hand rigid body rotation about the
x
3
axis, then a 90 right-hand rigid
body rotation about the x
1
axis. Find
the rotation tensor W
| |
| |
SR W
S
R
=
(
(
(

=
(
(
(

=
0 1 0
1 0 0
0 0 1
1 0 0
0 0 1
0 1 0
Tensor (Inner) Products
T S = o
ij ij
S S = = S S S
Magnitude or Norm:
S T T S =
W T W S W T S + = + | o | o ) (
0 > S S
0 S S S = = iff 0
( )
j i j i
d c b a = = ) )( ( ) ( d b c a d c b a
( ) ( )
( )( )
ij ij jl ik kl ij l j k i kl ij
l k kl j i ij
T S T S e e e e T S
e e T e e S
= = =
=
o o
T S
Trace of a Tensor
b a b a = ) ( tr
) (
) (
) ( ) ( ) (
33 22 11
T
ii ij ij j i ij
j i ij j i ij
T tr
T T T
T T T
tr T T tr tr
=
+ + =
= = =
= =
o e e
e e e e T
) ( T S T S
T
tr =
) (T T I tr =
T T
T T
A W A W W W
A S A S S S
= =
= =
Tensors
S RT T R) (S ST R
T T
= =
( ) v u S Sv u =
3 2 1
det
k j i ijk
S S S =e S
( )
( )
( ) ( )
1
1
det det
det det
det det det

=
=
=
S S
S S
T S ST
T
( ) 1 det = R
Inverse of a Tensor
( )
1 1
1

= S T ST
Inverse of a tensor T exists if it is non-singular
I SS =
1
( ) ( )
1
1

=
T
T
S S
Positive Definite Tensors
0 v v Sv v = e > , 0 V
Positive
Definite
0 v v Sv v = e > , 0 V
Positive
Semi-definite
Dual Vector of an Skew Tensor
| |
(
(
(

=
0
0
0
1 2
1 3
2 3
w w
w w
w w
W
V e = v v w Wv
{ }

=
3
2
1
w
w
w
w
Axial vector of W
Notations
Lin = the set of all rank two tensors
Lin
+
= subset of Lin with positive determinants
Sym = subset of Lin that are symmetric
Skw = subset of Lin that are skew
Psym = subset of Sym that are positive definite
Orth = set of all orthogonal tensors
Orth
+
= set of all rotations
Eigenvalues and Eigenvectors
a Ta =
Second rank tensors provide a linear transformation of
the vector space into itself. However, there would be
special vectors that satisfy the following condition:
{} are eigen values and {a} are eigen vectors

{} and {a} are independent of choice of coordinate frame

Multiple solutions are possible
Eigenvalues and Eigenvectors
0 a I T = ) (
a Ta =
0 ) det( = I T
Non-trivial Solution
exists when
0 ) ( ) ( ) (
3 2
2
1
3
= + T T T i i i
Characteristic
equation
Eigenvalues and Eigenvectors
0 ) ( ) ( ) (
3 2
2
1
3
= + T T T i i i
ii
T tr i = = T T) (
1
( ) | |
ji ij ii
T T T tr tr i = =
2
2 2
2
2
1
)] ( ) [(
2
1
) ( T T T
3 2 1 3
det ) (
k j i ijk
T T T i e = = T T
( ) ) ( ), ( ), (
3 2 1
T T T i i i
List of principal
invariants
Cayley-Hamilton Theorem
0 I T T T T T T = + ) ( ) ( ) (
3 2
2
1
3
i i i
Roots of Characteristic Equation
0 ) ( ) ( ) (
3 2
2
1
3
= + T T T i i i
The characteristic equation can produce
up to three real roots.
Multiple vectors are possible for each .
Rotation Tensors have only one real root!
Roots of Characteristic Equation:
Symmetric Tensors
i i
i
i j i ij
S e e e e S = =

=
3
1

A symmetric tensor S possesses three eigen values
(
1
,
2
,
3
) and at least one orthonormal set of eigen
vectors (e
1
,e
2
,e
3
) associated with (
1
,
2
,
3
) respectively.
b Sb a Sa S S e
T
= = =
0 = = = = a b a b a Sb Sa b b a e
( ) ( ) ij j j j i j i ij
S o = = = e e Se e
Symmetric Tensors
i i
i
i
e e S =

=
3
1

) (
) (
) (
3 2 1 3
1 3 3 2 2 1 2
3 2 1 1



=
+ + =
+ + =
S
S
S
i
i
i
| |
( )
(
(
(

=
3
2
1
0 0
0 0
0 0

i
S
e
Eigen Values and Eigen Vectors
0 ,
3
1
> =

=
i i i
i
i
e e C
A positive definite symmetric tensor C possesses
three positive eigen values (
1
,
2
,
3
).
Psym
i i
i
i
e = =

=
U e e C U ,
3
1

i i
i
i
e e C =

=

3
1
1 1

Skew Tensors
V e = v v w Wv
0 ) (
1
= = W W tr i
2
2 2
2
)] ( ) [(
2
1
) ( w W W W = = tr tr i
0 det ) (
3
= = W W i
W has only one real eigenvalue, which is
equal to zero
Polar Decomposition
1 det ,
, ,
= = =
e = =
= = e
+
R I R R RR
V U, FF V F F U
VR RU F F
T T
T T
Psym
Lin


=

= =
= =
= = =
= =
3
1
1 1
3
1
3
1
3
1
3
1
, ,
,
i
i i i
i
i i i
i
i i
i i
i
i i i
i
i
l r F r l F r l R
l l V r r U