Lectures in Representation Theory

Arun Ram
Department of Mathematics and Statistics
University of Melbourne
Parkville, VIC 3010 Australia
aram@unimelb.edu.au

Last update: 27 August 2013

Lecture 17

We saw last time that the Bratteli diagram describes the branching rules for restriction from irreducible modules over B to irreducible modules over A. Note that by Schur’s Lemma and Frobenius Reciprocity, gλμ=dimℂ  ( homA (Vλ,Wμ↓AB) ) =dimℂ  ( homB (Vλ↑AB,Wμ) ) , whence that Vλ↑AB≃ ∑μ∈Bˆ gλμWμ, so that the Young lattice also describes branching rules for induction of irreducible A-modules.

For A semisimple, we could produce the irreducible modules Vλ by noting that Vλ=Aeiiλ, where the eijλ were the preimages of the Eij in the λ-th block within A≃⨁λ∈AˆMdλ(ℂ).

Conversely, given an idempotent p in A, we can write p=∑pi, where the pi are minimal idempotents, and Ap≃⨁Api, with each Api irreducible. For each minimal idempotent p∈A, there exists a unique λ∈A with Ap≃Vλ; let us indicate this association by denoting our idempotents as pλ. Now if we set Ap=V=⨁λ∈Aˆ(Vλ)⊕mλ, one has that p=∑λ∈Aˆ ∑i=1mλ pλi and further Ap = A∑λ∈Aˆ ∑i=1mλ pλi ≃ ⨁λ∈Aˆ ∑i=1mλ Apλi

Note that Ap↑AB≃Bp, as B-modules. (To prove this, one needs to produce a bijection B⊗AAp⇆Bp. The appropriate maps are the C-linear extensions of b⊗ap↦bap and bp↦bp⊗p.)

Now we know that Vλ↑AB≃ ApλA↑AB≃ BpλA, for pλA a minimal idempotent in A (and naturally an idempotent of some sort in B). The branching rule (the induction version) now tells us that Vλ↑AB≃ ∑μ∈Bˆ gλμWμ. Thus if we express the idempotent pλA as a sum of minimal idempotents of B, we will obtain pλA=∑μ∈Bˆ∑i=1gλμqμiB, where the qμiB are minimal idempotents in B with BqμiB≃Wμ.

Let V be a representation of A, and χV the associated character, with V≃∑λ∈AˆmλVλ and χV=∑λ∈AˆmλχAλ We would like to produce a formula for χV↑AB. As B-modules, we know that V↑AB ≃ ∑λ∈Aˆ mλVλ ↑AB ≃ ∑λ∈Aˆ mλ∑μ∈Bˆ gλμWμ, and by taking traces one then obtains the formula χV↑AB= ∑λ∈Aˆ ∑μ∈Bˆ mλgλμ χBμ. This formula, while interesting, is not necessarily so nice for computational uses. With this in mind, let us define z=∑λ∈Aˆ mλdλA zλA∈A, where (as usual) the zλA are the minimal central idempotents of A, the dλA are the dimensions of the irreducible A-modules, and the mλ are the multiplicities of these irreducibles. Let {a}=𝒜 be a basis for A, and {a*}=𝒜* be its dual basis with respect to the trace of the regular representation of A. Now recall that zλA=∑a∈𝒜 dλAχAλ (a)a*, whence that z = ∑λ∈Aˆ mλdλA zλA = ∑λ∈Aˆ mλdλA ∑a∈𝒜 dλAχAλ (a)a* = ∑λ∈Aˆ ( ∑a∈𝒜 mλχAλ (a) ) a* = ∑λ∈Aˆ χV(a)a* We find then that the values of χV are encoded in this distinguished element z. Can we produce an element of B within which the values of χV↑AB are similarly encoded?

Let {b}=ℬ be a basis for B, let t→=(tλ) be any nondegenerate trace for B, and let {b*}=ℬ* be the associated dual basis. Define for each x∈B [x]=∑b∈ℬ bxb*.

Proposition 2.60 Given the above definitions, [z]= ∑b∈ℬ χV↑AB (b)b* .

Proof.

[z] = ∑λ∈Aˆ mλdλA [zλA] = ∑λ∈Aˆ mλdλA ∑i=1dλ [pλiA] = ∑λ∈Aˆ mλdλA ∑i=1dλ ∑μ∈Bˆ ∑j=1gλμ [qμjB] (The second equality is expressing zλA in terms of minimal idempotents of A, the third is expressing each of those minimal A-idempotents in terms of the minimal idempotents of B.) Note that if we take ek,kμ∈B and symmetrize it, we have [ek,kμ] = ∑1≤i,j≤dνBν∈Bˆ eijν ekkμ (eijν)* = ∑1≤i,j≤dνBν∈Bˆ eijνekkμ ejiνtν = ∑1≤i≤dνBj=k ν=μ eijν ekkμ ejiν tν = ∑i=1dν eiiνtν =1tνzνB Thus we have that [z] = ∑λ∈Aˆ mλdλA ∑i=1dλ ∑μ∈Bˆ ∑j=1gλμ [qμjB] = ∑λ∈Aˆ mλdλA ∑i=1dλ ∑μ∈Bˆ ∑j=1gλμ 1tμzμB = ∑λ∈Aˆ ∑μ∈Bˆ mλdλA dλAgλμ 1tμzμB = ∑λ∈Aˆ ∑μ∈Bˆ mλdλA gλμzμB = ∑λ∈Aˆ ∑μ∈Bˆ mλdλA gλμ ∑b∈ℬtμ χBμ(b)b* ⏟zμB = ∑b∈ℬ ∑λ∈Aˆ ∑μ∈Bˆ mλgλμ χBμ(b) ⏟χV↑AB(b)   b* [z] = ∑b∈ℬ χV↑AB (b)b*

□

Notes and References

This is a copy of lectures in Representation Theory given by Arun Ram, compiled by Tom Halverson, Rob Leduc and Mark McKinzie.

page history