Basic Modular Representation Theory

Basic Modular Representation Theory

BASIC MODULAR REPRESENTATION THEORY GYUJIN OH Contents 1. Algebra 1 1.1. Related randos 2 2. cde triangle 2 3. Brauer characters 4 4. Green correspondence 5 5. Block and defect 7 6. Example: SL2(Fp) 9 6.1. Brauer tree 10 7. Ext 10 8. Auslander-Reiten theory 11 I scratched the surface of Schneider’s Modular representation theory of nite groups. 1. Algebra Here our base ring is unital and our modules are left modules. ∙ Any artinian and noetherian module has Jordan-Holder composition series. ∙ A semisimple module is a direct sum of simple modules. Every module has a unique maximal semisimple submodule which we call as socle. It is just the sum of all simple submodules. ∙ Radical is intersection of maximal submodules. So that the radical of base ring is the usual Jacobson radical. Semisimple = artinian + zero radical. ∙ If M is artinian and f ∈ EndR(M) is injective, then f is bijective (Fitting’s lemma). Same for noetherian/surjective. This uses increasing/decreasing chain of im(f n) and ker(f n). In particular this shows that, for any indecomposable finite length R, EndR(M) is local. ∙ Given any idempotent e ∈ R, eRe ⊂ R is a subring with unit element e. Decomposing e = e1 + ⋯ + er as sum of pairwise orthogonal idempotents is the same as decomposing Re = Re1 ⊕ ⋯ ⊕ Rer into sum of nonzero left ideals. Thus Re is indecomposable i e is primitive. ∙ If R is left noetherian, R has only nitely many central idempotents, primitive central idempotents are pairwise orthogonal, and they sum up to 1. Given a central primitive idempotent e, M is said to belong to e-block if eM = M. Then not only eM = M, but ex = x for all x ∈ M. 1 ∙ For a 2-sided ideal I ⊂ R such that either every element in I is (topologically) nilpotent. Then, any idempotent of R/I lifts to an idempotent of R. This operation can be done preserving primitivity and orthogonality. ∙ A denition of projective module is that P such that HomR(P, −) is an exact functor. A module hom f ∶ M → N is essential if it’s surjective but f (L) ≠ N for any proper L ⊂ M. A projective cover of M is an essential R-module hom f ∶ P → M where P is projective. There is at most one projective cover up to isomorphism. ∙ Conversely if R is complete and R/ Jac(R) is left artinian, then any fg module has projective cover, which mod Jac(R) is an iso. ∙ The Grothendieck group R(A) of category of nite length A-modules is the same as the free ℤ-module generated over simple A-modules (denoted as A). The Grothendieck group K0(A) of category of nitely generated projective A-modules is generated over ℤ by nitely generated indecomposable projective A-modules (denoted as A), if A is left noetherian. ∙ In particular, if A is semisimple, then any A-module is projective. ∙ If A is left artinian, then K0(A) is freely generated by A. Thus we can consider the com- posed homomorphism cA ∶ ℤ[A] → ℤ[A] called the Cartan homomorphism. ∙ If A is complete and A = A/ Jac(A) is left artinian, then Grothendieck groups for A and A are the same. So, A → A/ Jac(A) is a projective cover, and A = P1 ⊕ ⋯ ⊕ Pr , a decomposi- tion into a sum of nitely generated indecomposable A-modules, and every such module occurs as direct summand. ∙ So if A is left artinian, we have A = {P1, ⋯ ,Pt }, and thus A = {M1, ⋯ ,Mt } where Mi = Pi/ Jac(A)Pi. 1.1. Related randos. ∙ A Frobenius algebra is a nite-dimensional algebra over a eld with a nondegenerate symmetric bilinear form such that (xy, z) = (x, yz). In particular k[G] is a Frobenius algebra via (x, y) ↦ xy(1) (the coecient of 1 of xy). This I think is also referred as symmetric Frobenius algebra at some places. ∙ A module is projective and indecomposable i P/ rad(P) = soc(P). ∙ Given a ring hom f ∶ A → B, a B-module P is relatively projective wrt f if, for any commutative diagram P 0 ~ M / N where , are B-module hom’s while 0 is an A-module hom, there is a B-module hom ∶ P → M replacing 0 still making the diagram commute. ∙ For any A-module L0, B ⊗A L0 is relatively projective. More generally P is relatively pro- jective i it is a direct summand of B ⊗A L0 for some A-module L0. ∙ There is criterion for left R[G]-module to be relatively R[H]-projective (for H ≤ G, R any commutative ring). 2. cde triangle We do modular representation theory. Let R be a (0, p)-ring for k, where k is an algebraically closed eld of characteristic p > 0, R is a complete local commutative domain such that mR is 2 principal, residue eld is k and K = Frac(R) is of characteristic zero. Then there is the ramication e index e ≥ 1 such that Rp = mR. The trinity of the cde triangle is ∙ RK (G) ∶= R(K[G]) = K0(K[G]), ∙ Rk(G) ∶= R(k[G]), ∙ K0(k[G]). Here G is a nite group. ∙ The c. This is the Cartan homomorphism cG ∶ K0(k[G]) → Rk(G). This makes sense because k[G] is a nite-dimensional k-vector space so of course this is left artinian. ∙ The d. This map dG ∶ RK (G) → Rk(G) is something like “mod p reduction.” Of course as usual this is not a priori well-dened. Namely, one chooses a G-invariant lattice inside a G-representation over K (which is possible b/c of averaging) and then reduce mod p (or R), but the resulting representation might not be isomorphic. But considering Rk(G) is like we’re in a “semisimplied world” and in particular the resulting module in Rk(G) is always the same. This is almost the same as Brauer-Nesbitt theorem. This particular statement can be proved by passing to a common renement (after mutiplying by suciently high power of R). – Perhaps a more classical statement of Brauer-Nesbitt is that a semisimple represen- tation of G over k is completely determined by characteristic polynomials of g ∈ G’s. ¤ ¤ ¡ ∙ The e. By the Jacobson radical theory we did before, K0(R[G]) ≅ ℤ[R[G]], R[G] = k[G] ¡ and K0(k[G]) ≅ ℤ[k[G]]. Thus we can dene eG ∶ K0(k[G]) → RK (G) by ∼ K0(k[G]) ←ãããããããã K0(R[G]) → K0(K[G]). Thus the triangle is of form dG RK (G) / Rk(G) e 9 eG cG K0(k[G]) Quite obviously the triangle commutes. ∙ If p does not divide |G|, then cde are all isomorphisms. ∙ If G is a p-group, then Jac(k[G]) = Ik[G], the augmentation ideal. In particular, k[G] is a local ring, and the trivial module is the only simple k[G]-module (“p-group lemma” says that there is always a xed vector). Thus Rk(G) and K0(k[G]) are both ℤ, and the diagram becomes [V ]↦dimK V RK (G) / ℤ b ? 1↦K[G] ⋅|G| ℤ I am not doing Cliord theory. ∙ Induction gives a well-dened map G indH ∶ RF (H) → RF (G), [W ] ↦ [F[G] ⊗F[H] W ], 3 for any eld F and H ≤ G. This is not a homomorphism but rather the image is an ideal in RF (G). ∙ Induction commutes with d. ∙ Brauer induction theorem says that G Rk(G) = ∑ indH (Rk(H)), H∈He where He is the set of elementary subgroups of G. From this one sees that d is surjective. m m ∙ c is injective and has nite cokernel. Furthermore p Rk(G) ⊂ im(c), where p ‖|G|. Proof uses Brauer induction to reduce to case of elementary subgroups. ∙ e is injective and its image is a direct summand of RK (G). This uses two bilinear forms ([V ], [W ]) ↦ dimK HomK[G](V,W ) RK (G) × RK (G) ããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããã→ ℤ, ([P], [V ]) ↦ dimk Homk[G](P,V ) K0(k[G]) × Rk(G) ããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããããã→ ℤ. The second bilinear form is perfect, and the rst one is also perfect if K is a splitting eld for G. ∙ Furthermore this duality shows that, over suciently large eld, so that K0(k[G]) and Rk(G) are free over ℤ, the matrices representing d and e are transposes to each other. Therefore, the Cartan matrix corresponding to the natural bases is symmetric! 3. Brauer characters ∙ Usual charcter theory gives a homomorphism Tr ∶ Rk(G) → Cl(G, k), where Cl(G, k) means the space of class functions. By Artin-Wedderburn, Tr ∶ k ⊗ℤ Rk(G) → Cl(G, k) is injective. This is why in characteristic p the usual character theory is less useful; you lose information by k⊗ℤ. We use V as the character of V . ∙ Somehow quite surprisingly there is the notion of p-regular(∼semisimple)/p-unipotent elements (e.g. there is Jordan decomposition), and the general slogan is that mod p repre- sentation theory is encoded not in all conjugacy classes but rather in p-regular conjugacy classes. ∙ To be more precise, a p-regular conjugacy class is a conj. class whose elements have order prime to p. An element is p-unipotent if it is of order a power of p. Then, there is Jordan decomposition, namely for any g ∈ G, there are unique gr , gu ∈ G such that gr is p- regular, gu is p-unipotent, g = gr gu = gugr . The construction is quite simple, namely you just take appropriate powers of g.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    11 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us