MODULAR COVARIANTS of CYCLIC GROUPS of ORDER P 11

MODULAR COVARIANTS of CYCLIC GROUPS of ORDER P 11

MODULAR COVARIANTS OF CYCLIC GROUPS OF ORDER p JONATHAN ELMER Abstract. Let G be a cyclic group of order p and let V; W be |G-modules. G ∗ G We study the modules of covariants |[V ] (W ) = (S(V ) ⊗ W ) . For V indecomposable with dimension 2, and W an arbitrary indecomposable mod- G G ule, we show |[V ] (W ) is a free |[V ] -module (recovering a result of Broer G and Chuai [1]) and we give an explicit set of covariants generating |[V ] (W ) G freely over |[V ] . For V indecomposable with dimension 3 and W an arbi- G trary indecomposable module, we show that |[V ] (W ) is a Cohen-Macaulay G |[V ] -module (again recovering a result of Broer and Chuai) and we give an G explicit set of covariants which generate |[V ] (W ) freely over a homogeneous G set of parameters for |[V ] . 1. Introduction Let G be a group, | a field, and V and W |G-modules. Then G acts on the set of functions V ! W according to the formula g · φ(v) = gφ(g−1v) for all g 2 G and v 2 V . Classically, a covariant is a G-equivariant polynomial map V ! W . An invariant is the name given to a covariant V ! | where | denotes the trivial indecomposable |G-module. If the field | is infinite, then the set of polynomial maps V ! W can be identified with S(V ∗) ⊗ W , where the action on the tensor product is diagonal and the action on S(V ∗) is the natural extension of the action on V ∗ by algebra automorphisms. Then the natural pairing S(V ∗) × S(V ∗) ! S(V ∗) is compati- ble with the action of G, and makes the invariants S(V ∗)G a |-algebra, and the covariants (S(V ∗) ⊗ W )G a S(V ∗)G-module. If G is finite and the characteristic of | does not divide jGj, then Schur's lemma implies that every covariant restricts to an isomorphism of some direct summand of S(V ∗) onto W . Thus, covariants can be viewed as \copies" of W inside S(V ∗). Otherwise, the situation is more complicated. The algebra of polynomial maps V ! | is usually written as |[V ]. In this article we will write |[V ]G for the algebra of G-invariants on V , and |[V ]G(W ) for the module of V − W covariants. We are interested in the structure of |[V ]G(W ) as a |[V ]G-module. Throughout, G denotes a finite group. This question has been considered by a number of authors over the years. For example, Chevalley and Sheppard-Todd [2], [10] showed that if the characteristic of | does not divide jGj and G acts as a reflection group on V , then |[V ]G is a polynomial algebra and |[V ]G(W ) is free. More generally, Eagon and Hochster [7] showed that if the characteristic of | does not divide jGj then |[V ]G(W ) is a Cohen-Macaulay module (and |[V ]G a Cohen-Macaulay ring in particular). In the modular case, Hartmann [5] and Hartmann-Shepler [6] gave necessary and sufficient Date: June 28, 2018. 2010 Mathematics Subject Classification. 13A50. Key words and phrases. modular invariant theory, covariants, free module, cohen-macaulay, hilbert series. 1 2 JONATHAN ELMER conditions for a set of covariants to generate |[V ]G(W ) as a free |[V ]G-module, provided that |[V ]G is polynomial and W =∼ V ∗. Broer and Chuai [1] remove the restrictions on both W and |[V ]G. 2. Preliminaries The purpose of this article is to investigate modular covariants for cyclic groups of order p. We are particularly interested in what happens when |[V ]G(W ) is not a free |[V ]G-module. Accordingly, from this point onwards we let G be a cyclic group of order p and | a field of characteristic p. We fix a generator σ of G. Recall that, up to isomorphism, there are exactly p indecomposable |G-modules V1;V2;:::;Vp, where the dimension of Vi is i. The isomorphism class of Vi is usually represented by a module of column vectors on which σ acts as left-multiplication by a single ∼ ∼ Jordan block of size i. Suppose V = Vm and W = Wn. We will choose bases v1; v2; : : : ; vm and w1; w2; : : : ; wn of V and W respectively for which the action of G is given by σw1 = w1 σw2 = w2 − w1 σw3 = w2 − w2 + w1 . σwn = wn − wn−1 + wn−2 − ::: ± w1 with analgous formula for the action on V (thus, the action of σ−1 is given by left-multiplication by a upper-triangular Jordan block). Now let x1; : : : ; xm be the dual basis to v1; v2; : : : ; vm; the action of G on these variables is then given by σx1 = x1 + x2; σx2 = x2 + x3; σx3 = x3 + x4: . σxm = xm: Note that |[V ] = |[x1; x2; : : : ; xm], and a general element of |[V ](W ) = |[V ]⊗W is given by φ = f1w1 + f2w2 + ::: + fnwn where each fi 2 |[V ]. We define the support of φ by Supp(φ) = fi : fi 6= 0g: The operator ∆ = σ − 1 2 |G will play a major role in our exposition. ∆ is a σ-twisted derivation on |[V ]; that is, it satisfies the formula (1) ∆(fg) = f∆(g) + ∆(f)σ(g) for all f; g 2 |[V ]. Further, using induction and the fact that σ and ∆ commute, one can show ∆ satisfies a Leibniz-type rule k X n (2) ∆k(fg) = ∆i(f)σn−i(∆n−i(g)): i i=0 MODULAR COVARIANTS OF CYCLIC GROUPS OF ORDER p 3 For any f 2 |[V ] we define the weight of f: wt(f) = minfi > 0 : ∆i(f) = 0g: Notice that ∆wt(f)−1 2 ker(∆) = |[V ]G for all f 2 |[V ]. Note also the following crucial observation: for all i = 1; : : : ; n − 1 we have (3) ∆(wi+1) + σ(wi) = 0 and ∆(w1) = 0. From this we obtain a simple characterisation of covariants: Proposition 1. Let G φ = f1w1 + f2w2 + ::: + fnwn 2 |[V ] (W ): i−1 Then there exists f 2 |[V ] with weight ≤ n such that fi = ∆ (f) for all i = 1; : : : ; n. Proof. We have n ! X 0 =∆ fiwi i=1 n X = (fi∆(wi) + ∆(fi)σ(wi)) i=1 n−1 X = (∆(fi) − fi+1)σ(wi)) + ∆(fn)σ(wn) i=1 where we used (3) in the final step. Now note that σ(wi)) = wi + (terms in wi+1; wi+2; : : : ; wn) for all i = 1; : : : ; n. Thus, equating coefficients of wi, for i = 1; : : : ; n gives ∆(f1) = f2; ∆(f2) = f3;:::; ∆(fn−1) = fn; ∆(fn) = 0: i−1 n Putting f = f1 gives fi = ∆ (f) for all i = 1; : : : ; n and 0 = ∆ (f) as required. Note that the support of any covariant is therefore of the form f1; 2; : : : ; ig for some i ≤ n. Where no confusion will arise, we write Supp(φ) = i if φ is a covariant and Supp(φ) = f1; 2; : : : ; ig. Introduce notation n Kn := ker(∆ ) and n In := im(∆ ): These are |[V ]G-modules lying inside |[V ]. Now we can define a map G Θ: Kn ! |[V ] (W ) n X i−1 (4) Θ(f) = ∆ (f)wi: i=1 Clearly Θ is injective, and Proposition 1 implies it is also surjective. Therefore Θ is a degree-preserving bijective map. 4 JONATHAN ELMER 3. Hilbert series Let | be a field and let S = ⊕i≥0Si be a positively graded |-vector space. The dimension of each graded component of S is encoded in its Hilbert Series X i H(S; t) = dim(Si)t : i≥0 G The observations following Proposition 1 imply that H(|[V ] (W ); t) = H(Kn; t). In this section we will outline a method for computing H(Kn; t). Each homogeneous component |[V ]i of |[V ] is a |G-module. As such, each one decomposes as a direct sum of modules isomorphic to Vk for some values of k. Write µk(|[V ]i) for the multiplicity of Vk in |[V ]i and define the series X i Hk(|[V ]) = µk(|[V ]i)t : i≥0 The series Hk(|[Vm]) were studied by Hughes and Kemper in [8]. They can G G also be used to compute the Hilbert series of |[Vm] ; since dim(Vk ) = 1 for all k = 1; : : : ; p we have p G X (5) H(|[Vm] ; t) = Hk(|[Vm]; t): k=1 Now observe that n n ≥ k dim(ker(∆nj )) = Vk k otherwise. Therefore n−1 p X X H(Kn; t) = kHk(|[V ]; t) + nHk(|[V ]; t): k=1 k=n We can also write this as as a series not depending on p: n−1 G X (6) H(Kn; t) = nH(|[V ] ; t) − ( (n − k)Hk(|[V ]; t)): k=1 using equation (5). G G We will need the Hilbert Series of In = |[V ] \ In in a later section. For all k = 1; : : : ; p we have 1 n > k dim(∆n(V ))G = k 0 otherwise. Therefore p G X H(In ; t) = Hk(|[V ]; t); k=n+1 which we can write independently of p as n G G X (7) H(In ; t) = H(|[V ] ; t) − ( Hk(|[V ]; t)): k=1 Note that G (8) H(Kn−1; t) + H(In−1; t) = H(Kn; t): We will make use of this in the final section. MODULAR COVARIANTS OF CYCLIC GROUPS OF ORDER p 5 4. Decomposition Theorems In this section we will compute the series Hk(|[V2]; t) and Hk(|[V3]; t) for all k = 1; : : : ; p − 1. Hughes and Kemper [8, Theorem 3.4] give the formula m−1 X γ − γ−1 1 − γp(m−1)tp Y (9) H ( [V ]; t) = γ−k (1 − γm−1−2jt)−1; k | m 2p 1 − tp γ2M2p j=0 where M2p represents the set of 2pth roots of unity in C.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    12 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