A representation of a group G is a group homomorphism G → Aut(V) = GL(V) into the automorphisms on a vector space. This is equivalent to saying that it's the structure of a kG-module on a k-vector space V.
A unitary representation is a representation that maps into the unitary operators on a Hilbert spaace G → U(H) We view ℂ G as an involutive structure, with (λ g)* = λ* g-1. Then a unitary representation is a ℂ G module with an inner product such that <g x, y> = <x, g* y> for all g ∈ kG In other words, V is a *-representation.
An irreducible representation is a representation such that there are no subspaces of V invariant under the action of G. From the module perspective, it's the same thing as a simple kG-module.
V is a direct sum of irreducible representations
V is a semisimple module
Or G acting on functions f: G → k via left multiplication
kG acting on itself
The trace of the representation matrix χ(g) = Tr(ρ(g))
A linear functional on kG from the trace of the endomorphism representation, vanishing on the commutator subspace [kG, kG]
Class functions (constant on conjugacy classes) are equivalent to the center Z(kG)
Irreducible modules correspond to ring maps Z(kG) → k
When we look at characters as linear functionals on kG, we can view them as corresponding to integrals; in particular, they correspond to integral kernels
This is particularly useful when we look at characters for infinite groups
If V is an irreducible representation, we define the specific character χ(g) = Tr(ρ(g)) as the irreducible character of this representation
Irreducible characters end up being equivalent to maps G → ℂ such that
From the module perspective, an irreducible character is a linear functional if and only if
U(1) : One-dimensional Irreducible Unitary
The 1-dim irreducible unitary characters are exactly maps into U(1)
We can define the inner product as a Frobenius pairing, where we take kG as the coalgebra with duplication coproduct, giving (kG)* an algebra structure. Then we can take the counit to be the average over the group if it's finite
for compact groups, you use the Haar measure and take L2(G) instead of the strict dual
locally compact groups also have a Haar measure, but the integral over the whole space is infinite You have a few options:
non locally compact groups are difficult:
since there's no Haar measure, by definition there's no translation-invariant measure
so you use a quasi-invariant measure: actions by the group elements preserve null sets
you then define a Radon-Nikodym derivative ρ(g,x) = (x)
and define a unitary action on L2(G, μ) with (U(g) ċ f) (x) = f(g-1 x)
and you integrate with respect to the quasi-invariant measure
An example of this is the Wiener measure!
Another solution is to take the colimit of compact subgroups G^• = G1 ≤ G2 ≤ …
Then you take the limit over the the haar measures (when this exists; which is only for certain characters) <φ, ψ> = lim ∫Gn φ(g) ψ*(g) d μn(g)
Over ℂ we have that the irreducible characters satisfy
<χi, χj> = ∑g χi(g) χj*(g) = δij
This allows us to decompose an arbitrary complex representation into irreducible pieces using projection
From the kG-module perspective, we have Wedderburn's Structure Theorem
If G is a finite group and
Let G be a finite group with |G| ≢ 0 (mod char k), and let V be a representation. Then every invariant subspace W ⊆ V has a complement subspace U with V = W ⊕ U
From the module perspective, this says that every finite group with |G| ≠ 0 (mod char k) has kG semisimple.
Every short exact sequence of kG-modules splits
Let φ1, φ2 be irreducible representations on V1 and V2, and let Ψ: φ1 → φ2 (commutative square) Then
From the perspective of kG-modules we have
If V1 and V2 are simple kG-modules then