3.11. Lecture 11
We now turn to various ways of constructing new representations out of old ones. The methods will split into two general classes:
-
(i)
Linear algebraic
-
(ii)
Group theoretic
We start with linear algebra:
3.11.1. The dual representation
Given a vector space , recall that (the dual space of ) is the vector space consisting of all linear functionals from to , i.e. . We define a representation of by letting (for every and ) be the functional defined via the formula
Note that this may be seen as a special case of some representations we have seen before:
-
(i)
is a subrepresentation of the permutation representation on the space of -valued functions on that arises from the linear -action of on .
-
(ii)
Recall the representation (cf. Lemma 2.9.1). The dual representation is the special case .
Being a special case of , Lemma 2.9.2 (b) gives . Problem 3.11.4 below gives another proof of this fact.
3.11.2. Quotient representations
Let be a representation of a group and a subrepresentation of . Recall that the quotient space is the vector space modulo the equivalence relation . We normally write elements of in the form . We define a representation of on via the formula . Note that it is critical that is a subrepresentation for this to make sense (i.e. ).
The following proposition will let us avoid using quotient representations too often:
Proposition 3.11.1.
If , then
As a consequence of the above, we have
Proposition 3.11.2.
- Let be a finite dimensional representation of a finite group and a subrepresentation with corresponding quotient representation . Then
The proofs of both these results are left as exercises below.
We won’t need to use quotient representations too often, but they are required to make what we do next rigorous.
3.11.3. Tensor Products
We now define the tensor product of two representations; this is a way of ‘multiplying’ two or more representations. Throughout this section, let and be two representations of a group .
Firstly, we consider the permutation representation of on the free vector space ; we write this as , i.e. the elements of are finite formal sums
for , where all but finitely many are non-zero. Note that since is an infinite set, is an infinite-dimensional vector space, with basis given by all possible pairs . The group acts via the formula
This may seem like a sensibly defined representation however note that in , the vectors and are different. In fact we don not have any relations that allow us to split up brackets or pull out scalar factors. We therefore define the vector space
We claim that is a subrepresentation of , and define the tensor product representation as the quotient representation
Taking the quotient space by allows us to identify vectors that differ by elements of . If we consider our example notice that they differ by which is one of the spanning elements of , hence
as elements of . The same holds for the other linear relations in both components of . In other words, modding out the subspace is a way of forcing desired linear relations on the space .
If and , then we write for the element in .
Important: Not every element of may be written in the form (see Problem 3.11.8 below). However, the vectors do span , and the following proposition gives a convenient way of finding a basis from this spanning set:
Proposition 3.11.3.
Let and be finite-dimensional vector spaces, with being a basis of and being a basis of . Then
is a basis of . In particular, .
Example 3.11.4.
The above basically says .
Remark 3.11.5.
An element in of the form (, ) is called a pure tensor.
The previous proposition can be used to find the characters of tensor products of representations:
Proposition 3.11.6.
Proof.
Let be an eigenbasis of for ( ) and be an eigenbasis of for ). Then from the Proposition 3.11.3, is a basis of , and is in fact an eigenbasis for , since
This gives . ∎
3.11.4. Exercises
Problem 64. Let be a finite-dimensional representation of a finite group with dual representation .
-
(a)
Let be a basis of and the corresponding dual basis of , i.e.
Show that and .
-
(b)
Show that is irreducible if and only if is.
Problem 65. Let be a representation of a group and a subrepresentation of . Verify that defining
gives a -representation .
Problem 67. Consider the representation of on , where , given by
with .
-
(a)
Write down the matrices of and with respect to the basis
of (where and are the standard basis of ).
-
(b)
Write the character of as a sum of irreducible characters.
-
(c)
For each of the irreducible characters occurring in the expression of used in the previous part, find a subrepresentation of with that character.
-
(d)
Find an isomorphism from to .
-
(e)
Find an isomorphism from to .
-
(f)
Let with factors. Decompose into irreducible representations.
Problem 68. Let , , and be three finite-dimensional representations of a finite group . Show that
-
(a)
-
(b)
Remark 3.11.8.
Note that (a) lets us write unambiguously (and similarly for larger products of representations).
Problem 69. Let form a basis of . Show that cannot be written in the form .
Problem 70. For two finite-dimensional representations and of a finite group , construct an isomorphism to show that
Problem 71. Given a finite group , let denote the permutation representation of on associated to the conjugation action of on itself, i.e.
-
(a)
Show that , where is the conjugacy class belongs to. Hint: use Problem 2.7.2.
-
(b)
Then show that