Suppose that is a subgroup of . Given a representation of , we may construct a representation of by restriction. What about the other direction — given a representation of , is there a natural way to construct a representation of ? The answer is yes, and this is the induced representation.
More precisely, we start with a representation of and want to construct a representation of which contains as an -subrepresentation. Suppose we have such a thing. Then, for each ,
is a subspace of (not necessarily an -subrepresentation!). We should have that, if for some , then
so that only depends on the left coset . The representation is induced if there are “no more relations”. Formally:
Let be a finite group and let be a subgroup. Let be a representation of and let be a representation of . We say that is induced from if:
There is an -subrepresentation of that is isomorphic to ; and
If are a set of left coset representatives for in , then
The action of is determined by the direct sum decomposition: if and then we may write for some and some , and then
and is known since we have that is isomorphic to as an -representation.
One can show that induced representations always exist (we didn’t do this in class). Here is one construction: Let
with -action for all , . For any we can define its support
which will be a union of left cosets of . We can then take ; then and .
If is a representation of then any two representation of induced from are isomorphic.
We defer this proof until after the discussion of Frobenius reciprocity below. ∎
The significance of the lemma is that we can talk about ‘the’ induced representation, since it is unique up to isomorphism.
Let be the two-dimensional representation of such that and — irreducible if . Let be the one-dimensional representation of the subgroup with .
The cosets of in are . If then is a -subrepresentation isomorphic to , and . We clearly have , so this shows
Suppose that is the trivial representation of . Then the induced representation coincides with the permutation representation associated to the left action of on . Indeed, contains a vector fixed by (that is, a copy of the trivial representation of ), and if are the left cosets of in then
are a basis for . Moreover, if then for some and so
This shows that acts on these basis vectors ‘in the same way’ as it acts on the elements , which is what we have to prove.
Inducing multiplies the dimension by the index of the subgroup:
If are the cosets of in then
for some copy of in . The dimension formula follows. ∎