34 3 Symmetry of Second Rank Tensors, Cross Product
5.1 Isotropic and Symmetric Traceless Parts
The symmetric part of a second rank tensorAcan be decomposed further into an
isotropic partproportional to the product of the isotropic tensorδand thetrace
trA=Aλλand a symmetric traceless partAdefined by
Aμν =
1
2
(
Aμν+Aνμ
)
−
1
3
Aλλδμν. (3.3)
Thus the tensorAis decomposed into its isotropic, antisymmetric and symmetric
traceless parts according to
Aμν=
1
3
Aλλδμν+Aasyμν+Aμν. (3.4)
This decomposition is invariant under a rotation of the coordinate system.
The symbol...used to indicate the symmetric traceless part of a tensor, was
introduced by Ludwig Waldmann around 1960. Compared with the double arrow
←→, which also occurs in printing, the...has the advantage that it can be drawn
in one stroke. For second rank tensors,...first appeared in print in [20], and in
[21], it was applied for irreducible tensors of any rank. Alternative notations used in
the literature for symmetric traceless tensors are mentioned in Sect.3.1.7.
3.1.3 Trace of a Tensor
The isotropic part involves thetrace of the tensor
tr(A)=Aλλ=A 11 +A 22 +A 33. (3.5)
It is a scalar (tensor of rank=0), i.e., it is invariant under a rotation of the
coordinate system. The proof is: the tensor propertyA′μν=UμκUνλAκλimplies
A′μμ =UμκUμλAκλ, and due to the orthogonality (2.31) of the transformation
matrix, one hasA′μμ=δκλAκλ=Aκκ.
The termisotropicis used since the unit tensorδμνhas no directional properties,
it is not affected by a rotation of the coordinate system. Here the other orthogonality
(2.36) is used for the proof:δ′μν=UμκUνλδκλ=UμλUνλ=δμν.
Notice: the antisymmetric part of the tensor does not contribute to the trace:
tr(A)=Aλλ=A
sym
λλ =tr(A
sym). (3.6)
The trace of a second rank tensor is also given by the total contraction of this tensor
with the unit tensor:
tr(A)=δμνAμν=δμνAνμ=Aνν. (3.7)