Scribd
Upload
374 views4 pages

Module-1: Tensor Algebra: Lecture-7: The Skewsymmetric Tensor

The document defines a skew-symmetric tensor as a second-order tensor W where W T = -W. It is shown that for any skew-symmetric tensor W, there exists a unique vector w corresponding to W such that W u = w x u for any vector u. Conversely, any vector w has a unique associated skew-symmetric tensor W satisfying this property. Skew-symmetric tensors can be represented by their axial vector w and have complex eigenvalues of i|w| and -i|w|, with the axial vector corresponding to the zero eigenvalue. In higher dimensions, skew-symmetric tensors cannot always be associated with a vector.

Uploaded by

Ankush Pratap
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
374 views4 pages

Module-1: Tensor Algebra: Lecture-7: The Skewsymmetric Tensor

The document defines a skew-symmetric tensor as a second-order tensor W where W T = -W. It is shown that for any skew-symmetric tensor W, there exists a unique vector w corresponding to W such that W u = w x u for any vector u. Conversely, any vector w has a unique associated skew-symmetric tensor W satisfying this property. Skew-symmetric tensors can be represented by their axial vector w and have complex eigenvalues of i|w| and -i|w|, with the axial vector corresponding to the zero eigenvalue. In higher dimensions, skew-symmetric tensors cannot always be associated with a vector.

Uploaded by

Ankush Pratap
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

NPTEL – Mechanical Engineering – Continuum Mechanics

Module-1: Tensor Algebra


Lecture-7: The Skewsymmetric Tensor

Definition of skewsymmetric tensor:


A second-order tensor W is said to be skewsymmetric if W T = −W . Let u and v be
arbitrary vectors in vector space V then

(u, W v) = (W T u, v) = −(W u, v) = −(v, W u) (1)

Choosing u = v, we have

(u, W u) = −(u, W u)
=⇒ (u, W u) = 0, ∀u ∈ V (2)

From Eq. (2), we conclude that the vector W u is either orthogonal to the vector u or
W u = 0. Choosing u = ei and v = ej in Eq. (1), we get

Wij = −Wji .

Clearly, Wij = 0 if i = j. Let {e1 , e2 , e3 } be canonical basis. Then the skewsymmetric


tensor can be represented by
 
0 W12 W13
W =  −W12 0 W23  (3)


−W13 −W23 0

Thus, any skewsymmetric tensor defined over three dimensional space has three indepen-
dent components.

Problem 1. Given any skewsymmetric tensor W ∈ V 2 , there exist a unique w ∈ V


corresponding to W such that

W u = w × u, ∀u ∈ V (4)

Conversely, every vector w ∈ V has association with a unique skewsymmetric second-order


tensor W such that Eq. (4) holds. The vector w is known as axial vector corresponding
to W .

Proof. Consider the determinant of skewsymmetric tensor W ,

det(W ) = det(W T ) = det(−W ) = (−1)3 det(W ) = −det(W )


=⇒ det(W ) = 0 (5)

Joint initiative of IITs and IISc – Funded by MHRD, Government of India 1


NPTEL – Mechanical Engineering – Continuum Mechanics

Since the determinant of tensor W is zero there must exist at least one zero eigenvalue.
Let p be a unit eigenvector corresponding to zero eigenvalue i.e., W p = 0. Consider
q and r are two mutually orthogonal unit vectors such that they are orthogonal to the
vector p. Thus, the set {p, q, r} forms an orthonormal basis to the vector space V. Since
W is a second-order tensor we can have the following representation.

W = W11 p ⊗ p + W12 p ⊗ q + W13 p ⊗ r + W21 q ⊗ p + W22 q ⊗ q + W23 q ⊗ r +


W31 r ⊗ p + W32 r ⊗ q + W33 r ⊗ r (6)

Since the set {p, q, r} is an orthonormal basis, the components of W can be written as

W11 = p · W p, W12 = p · W q, W13 = p · W r, W21 = q · W p, W22 = q · W q,


W23 = q · W r, W31 = r · W p, W32 = r · W q, W33 = r · W r.

Using Eq. (2), we get (p, W p) = (q, W q) = (r, W r) = 0. The relation W p = 0 give
rise to (q, W p) = (r, W p) = 0. In addition, using skewsymmetric nature of tensor W ,
we have

(p, W q) = (W T p, q) = −(W p, q) = −(q, W p) = 0


(p, W r) = (W T p, r) = −(W p, r) = −(r, W p) = 0.

In summary, the components of tensor in orthonormal basis {p, q, r}: W11 = (p, W p) =
0, W12 = −W21 = (p, W q) = 0, W13 = −W31 = (p, W r) = 0, W22 = (q, W p) = 0,
W33 = (r, W r) = 0, W23 = (q, W r) = −(r, W q) = −W32 . If W is non-zero then W23
and W32 are only non-zero components. Furthermore, W23 = −W32 . Substituting these
components in Eq. (6), we obtain

W = W23 (q ⊗ r − r ⊗ q)

Consider the following product,

W u = W23 (q ⊗ r − r ⊗ q)u
= W23 ((r · u)q − (q · u)r)
= W23 ((r × q) × u)

Let the basis {p, q, r} be a right hand coordinate frame, i.e., p = q × r, q = r × p,


r = p × q. Let us choose W23 = −γ then we have

W u = γ((q × r) × u) = (γp) × u. (7)

Let w = γp then
W u = w × u, ∀u ∈ V.
Thus, there is an axial vector w corresponding to every skewsymmetric tensor W .

Conversely, if the vector w is given then we need to show that there exists a corre-
sponding skewsymmetric tensor W . For every vector u ∈ V there is a vector v such
that
w × u = v.

Joint initiative of IITs and IISc – Funded by MHRD, Government of India 2


NPTEL – Mechanical Engineering – Continuum Mechanics

There is an output vector for every input vector and hence, the operation with given
vector is a transformation from vector space to vector space. The linearity of the preceding
operation follows from the distributive property of cross product.
w × (αu + βv) = α(w × u) + β(w × v), ∀u, v ∈ V and ∀α, β ∈ <
Therefore, the cross product with a given vector w is a linear transformation. Conse-
quently, there should exist a second-order tensor corresponding to the operation. Let W
be a second-order tensor such that
W u = w × u, ∀u ∈ V.
Writing the preceding equation in indicial notation, we get
(W u)i = (w × u)i
Wij uj = ikj wk uj
Wij uj = −ijk wk uj
Arbitrariness of vector u implies Wij = −ijk wk . It is easy to see that Wij = −Wji . There-
fore, the second-order tensor W is a skewsymmetric tensor. Thus, there is a skewsym-
metric tensor W associated with a given vector w.

Component wise relations:


In summary, we have the following relations between the skewsymmetric tensor W and
its axial vector w.
Wij = −ijk wk (8)
1
wi = − ijk Wjk (9)
2
W = |w|(r ⊗ q − q ⊗ r) (10)
where q and r are two orthogonal unit vectors such that
w
=q×r
|w|
Eigenvalues of skewsymmetric tensor:
Let W and w be skewsymmetric tensor and its axial vector, respectively. Let I1 , I2 and
I3 be principal invariants of W . Then
I1 = Wii = ei · W ei = 0
1
I2 = imn ipq Wmp Wnq
2
1
= (δmp δnq − δmq δnp )Wmp Wnq (Using epsilon-delta identity)
2
1
= (Wmm Wnn − Wmn Wnm )
2
1
= Wmn Wmn (Since Wmm = Wnn = 0 and Wnm = −Wmn )
2
1
= (mni wi )(mnj wj ) (Using the relation Wpq = pqr wr )
2
1
= (2δij )wi wj (Using pqi pqj = 2δij )
2
= wi wi = w · w = |w|2
I3 = det(W ) = 0

Joint initiative of IITs and IISc – Funded by MHRD, Government of India 3


NPTEL – Mechanical Engineering – Continuum Mechanics

Using principal invariants, we get the following characteristic equation for the tensor W .

λ3 + |w|2 λ = 0 (11)

where λ is the eigenvalue. This cubic equation has one zero root and two purely imaginary
√
roots. If i is imaginary number −1 then the eigenvalues can be expressed as λ1 = 0,
λ2 = i|w| and λ3 = −i|w|. It can be verified that eigenvalues satisfy the characteristic
equation. Geometrically, there are no real eigenvectors for the skewsymmetric tensor
except the axial vector. Furthermore, the axial vector also maps to the zero vector as it
is corresponding to zero eigenvalue. In conclusion, the skewsymmetric tensor W assumes
a diagonal form if the field is extended to the complex numbers.
Skewsymmetric tensor in higher dimensional vector space:
Let us consider a skewsymmetric tensor in n-dimensional vector space. The following
property is true for any dimension,

(u, W u) = 0, ∀u ∈ V

From the geometrical interpretation, all eigenvalues are either complex numbers or zeros.
In other words either the directions undergo orthogonal rotation or the directions have
zero scale factors.
It is easy to see that any skewsymmetric tensor defined over n-dimensional vector space
has n(n − 1)/2 independent components. Consequently, the skewsymmetric tensor can
not be associated with a vector in the vector space other than three dimensional space.
Thus, three dimensional space has special property that every skewsymmetric tensor can
be associated with a unique vector and vice versa.

Reference

1. C. S. Jog, Continuum Mechanics: Foundations and Applications of Mechanics, Vol-


ume I, Third edition, 2015, Cambridge University Press.

Joint initiative of IITs and IISc – Funded by MHRD, Government of India 4

You might also like