2.5 Definitions of Vectors and Tensors in Physics 25
with the quantityC′linked withCby
C′μν=UμλUνκCλκ. (2.50)
The relation (2.50) proves:Cis a second rank tensor.
Examples for linear relations like (2.48) are those between the angular momentum
and the angular velocity of a solid body, where the moment of inertia tensor occurs,
and between the electric polarization and the electric field in a “linear medium”.
Here, the susceptibility tensor plays the role ofC.
Similarly, the linear relationbμν=Cμνλκaλκbetween two second rank tensors
aandbimplies that, in this case,Cis a tensor of rank 4. The elasticity and the
viscosity tensors linking the stress tensor or the pressure tensor with the gradient of
the displacement and of the velocity field, respectively, are of this type.
The generalization of (2.48) is a linear relation between a tensorbof rankwith
tensoraof rankkof the form
bμ 1 μ 2 ...μ=Cμ 1 μ 2 ...μν 1 ν 2 ...νkaν 1 ν 2 ...νk. (2.51)
HereCis a tensor of rank+k.
In physics, examples for linear relations linking tensors of rank 1 with tensors of
rank 1, 2, 3 and of tensors of 2 with tensors of rank 1, 2, 3 of tensors, and so on, were
already discussed over hundred years ago in the bookLehrbuch der Kristallphysik
where Woldemar Voigt introduced the notiontensor.
Therelation(2.51)isalinear mappingofaonb. Nevertheless, the physical
content may describe non-linear effects, when the tensorastands for a product of
tensors. Examples occur in non-linear optics. For strong electric fields, the induced
electric polarization contains not only the standard term linear in the field, but also
contributions bilinear and of third order in the electric field. The material coefficient
characterizing these effects, called higher order susceptibilities, are tensors of rank
3 and 4.
2.6 Parity
2.6.1 Parity Operation
In addition to their rank, tensors are classified by theirparity. The parity is either equal
to1or−1 when the physical quantity considered is an eigenfunction of theparity
operatorP.Theparity operationis an active transformation where the position
vectorris replaced by−r, cf. Fig.2.6. This active ‘mirroring’ should not be confused
with the mirroring of the coordinate system as described by the transformation matrix
Uμν=−δμν.