12—Tensors 377
Similarly
g^11 =g(~e^1 ,~e^1 ) = 1/ 2
g^12 =g(~e^1 ,~e^2 ) =− 1 /
√
2
g^21 =g(~e^2 ,~e^1 ) =− 1 /
√
2
g^22 =g(~e^2 ,~e^2 ) = 2
(
grc
)
=
(
1 / 2 − 1 /
√
2
− 1 /
√
2 2
)
The mixed components are
g^11 =g(~e^1 ,~e 1 ) = 1
g^12 =g(~e^1 ,~e 2 ) = 0
g^21 =g(~e^2 ,~e 1 ) = 0
g^22 =g(~e^2 ,~e 2 ) = 1
(
grc
)
=
(
δcr
)
=
(
1 0
0 1
)
(33)
I usedrandcfor the indices to remind you that these are the row and column variables. Multiply the first two
matrices together and you obtain the third one — the unit matrix. The matrix
(
gij
)
is therefore the inverse of
the matrix
(
gij
)
. This last result isnotgeneral, but is due to the special nature of the tensorg.
The question arises as before: how do you compute the components of a tensor in one basis when the
components are given in another basis. The question is answered as before: It is a problem in geometry to express
the new basis vectors in terms of the old one.
~e′i=aji~ej (34)
The covariant components in the new system are (say for a second rank tensor)
Tij′ =T(~e′i, ~ej′) =T(aki~ek, a`j~e`) =akia`jTk` (35)
Analogous equations will hold for the contravariant indices provided that you can find the reciprocal basis
vectors in the transformed system. That is, you need the vectors reciprocal to the~e′iin terms of the vectors
reciprocal to~ei. Assume a relation:
~e′i=bji~ej
You must find theb’s in terms of thea’s. The relationship must come from reciprocity relationships. Required:
~e′i.~e′j=δji=
(
bki~ek
)
.
(
~e`
)
=bkia`j~ek.~e`=bkia`jδk`=bkiakj (36)