Printer Friendly

0-Hecke algebra action on the Stanley-Reisner ring of the Boolean algebra.

1 Introduction

Let F be any field. The symmetric group [G.sub.n] naturally acts on the polynomial ring F[X] := F[[x.sub.1], ..., [x.sub.n]] by permuting the variables [x.sub.1], ..., [x.sub.n]. The invariant algebra [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], which consists of all the polynomials fixed by this 6n-action, is a polynomial algebra generated by the elementary symmetric functions [e.sub.1], ..., [e.sub.n]. The coinvariant algebra [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], with [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], is a vector space of dimension n! over F, and when F has characteristic larger than n the coinvariant algebra carries the regular representation of [G.sub.n]. A well known basis for [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] consists of the descent monomials. Garsia [7] obtained this basis by transferring a natural basis from the Stanley-Reisner ring F[[B.sub.n]] of the Boolean algebra [B.sub.n] to the polynomial ring F[X]. Here the Boolean algebra [B.sub.n] is the set of all subsets of [n] := {1,2, ..., n} partially ordered by inclusion, and the Stanley-Reisner ring F[[B.sub.n]] is the quotient of the polynomial algebra F [[y.sub.A]: A [subset or equal to] [n]] by the ideal ([y.sub.A][y.sub.B]: A and B are incomparable in [B.sub.n]).

The 0-Hecke algebra [H.sub.n](0) (of type A) is a deformation of the group algebra of [G.sub.n]. It acts on F[X] by the Demazure operators, also known as the isobaric divided difference operators, having the same invariant algebra as the [G.sub.n]-action on F[X]. In our earlier work [13], we showed that the coinvariant algebra [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is also isomorphic to the regular representation of [H.sub.n](0), for any field F, by constructing another basis for [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] which consists of certain polynomials whose leading terms are the descent monomials. This and the previously mentioned connection between the Stanley-Reisner ring F[[B.sub.n]] and the polynomial ring F[X] motivate us to define an [H.sub.n](0)-action on F[[B.sub.n]].

It turns out that our [H.sub.n](0)-action on F[[B.sub.n]] has similar properties to the [H.sub.n](0)-action on F[X]. It preserves an [N.sup.n+1]-multigrading of F[[B.sub.n]] and has invariant algebra equal to a polynomial algebra F[[THETA]], where [THETA] is the set of rank polynomials [[theta].sub.i] (the usual analogue of [e.sub.i] in F[[B.sub.n]]). We show that the [H.sub.n](0)-action is [THETA]-linear and thus descends to the coinvariant algebra F[[B.sub.n]]/([THETA]). Using the descent monomials in F[[B.sub.n]] it is not hard to see that F[[B.sub.n]]/([THETA]) carries the regular representation of [H.sub.n](0).

It is well known that every finite dimensional (complex) [G.sub.n]-representation is a direct sum of simple (i.e. irreducible) [G.sub.n]-modules, and the simple [G.sub.n]-modules are indexed by partitions [lambda] of n, which correspond to the Schur functions [s.sub.[lambda]] via the Frobenius characteristic map. Hotta-Springer [11] and Garsia-Procesi [9] discovered that the cohomology ring of the Springer fiber indexed by a partition p of n is isomorphic to certain quotient ring of F[X], which admits a graded [G.sub.n]-module structure corresponding to the modified Hall-Littlewood symmetric function [[??].sub.[mu]] (x; t) via the Frobenius characteristic map. The coinvariant algebra of [G.sub.n] is nothing but [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

In our previous work [13] we established a partial analogue of the above result by showing that the [H.sub.n](0)-action on F[X] descends to [R.sub.[mu]] if and only if [mu] = ([1.sup.k], n - k) is a hook, and if so then [R.sub.[mu]] has graded quasisymmetric characteristic equal to HM (x; t) and graded noncommutative characteristic [[??].sub.[mu]] (x; t). Here [[??].sub.[alpha]] (x; t) is the noncommutative modified Hall-Littlewood symmetric function introduced by Bergeron and Zabrocki [3] for any composition [alpha] of n. Using an analogue of the nabla operator Bergeron and Zabrocki [3] also introduced a (q, t)-analogue [[??].sub.[alpha]](x; q,t) for any composition a. Now we provide in Theorem 1.1 below a complete representation theoretic interpretation for [[??].sub.[alpha]](x; t) and [[??].sub.[alpha]](x; q, t) by the [H.sub.n](0)-action on F[[B.sub.n]].

To state our result, we first recall the two characteristic maps for representations of [H.sub.n](0) introduced by Krob and Thibon [14], which we call the quasisymmetric characteristic and the noncommutative characteristic. The simple [H.sub.n](0)-modules are indexed by compositions [alpha] of n and correspond to the fundamental quasisymmetric functions [F.sub.[alpha]] via the quasisymmetric characteristic; the projective indecomposable [H.sub.n](0)-modules are also indexed by compositions [alpha] of n and correspond to the noncommutative ribbon Schur functions sa via the noncommutative characteristic. See [section] 2 for details.

Theorem 1.1 Let [alpha] be a composition of n. Then there exists a homogeneous [H.sub.n](0)-invariant ideal [I.sub.[alpha]] of the multigraded algebra F[[B.sub.n]] such that the quotient algebra F[[B.sub.n]]/[I.sub.[alpha]] becomes a projective [H.sub.n](0)-module with multigraded noncommutative characteristic equal to

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

One has [[??].sub.[alpha]](x; t, [t.sup.2], ..., [t.sup.n-1]) = [[??].sub.[alpha]](x; t), and obtains [[??].sub.[alpha]](x; q, t) from [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] by taking [t.sub.i] = [t.sup.i] for all i [member of] D([alpha]), and [t.sub.i] = [q.sup.n-i] for all i [member of] [n - 1]\D([alpha]).

Here D([alpha]) is the set of partial sums of [alpha], the notation [beta][??][alpha] means [alpha] and [beta] are compositions of n with D([beta]) [subset or equal to] D([alpha]), and [[t.bar].sup.S] denotes the product [[PI].sub.i[member of]S][t.sub.i] over all elements i in a multiset S, including the repeated ones. Taking [alpha] = ([1.sup.n]) shows that F[[B.sub.n]]/([THETA]) carries the regular representation of [H.sub.n](0).

Specializations of [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] include not only [[??].sub.[alpha]](x; q, t), but also a more general family of noncommutative symmetric functions depending on parameters associated with paths in binary trees introduced recently by Lascoux, Novelli, and Thibon [15].

Next we study the quasisymmetric characteristic of F[[B.sub.n]]. We combine the usual [N.sub.n+1]-multigrading of F[[B.sub.n]] (recorded by [t.bar] := [t.sub.0], ..., [t.sub.n]) with the length filtration of [H.sub.n](0) (recorded by q) and obtain an N x [N.sup.n+1]-multigraded quasisymmetric characteristic for F[[B.sub.n]].

Theorem 1.2 The N x Nn+1-multigraded quasisymmetric characteristic of F[[B.sub.n]] is

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Here we identify [F.sub.I] with [F.sub.[alpha]] if D([alpha]) = I [subset or equal to] [n - 1]. The set Com(n, k) consists of all weak compositions of n with length k, i.e. all the sequences [alpha] = ([[alpha].sub.1], ..., [[alpha].sub.k]) of k nonnegative integers with [absolute value of [alpha]] := [[summation].sup.k.sub.i=1] [[alpha].sub.i] = n. The descent multiset of the weak composition [alpha] is the multiset

D([alpha]) := {[[alpha].sub.1], [[alpha].sub.1] + [[alpha].sub.2], ..., [[alpha].sub.1] + ... + [[alpha].sub.k-1]}.

We also define [G.sup.[alpha]] := {w [member of] [G.sub.n]: D(w) [subset or equal to] D([alpha])}. The set [[k + 1].sup.n] consists of all words of length n on the alphabet [k + 1]. Given p = ([p.sub.1], ..., [p.sub.n]) [member of] [[k + 1].sup.n], we write [p'.sub.i] := #{j: [p.sub.j] [less than or equal to] i}, inv(p) := #{(i,j): 1 [less than or equal to] i [less than or equal to] j [less than or equal to] n: [p.sub.i] > [p.sub.j]}, and D(p) := {i: [p.sub.i] > [p.sub.i+1]}.

Let [ps.sub.q;l]([F.sub.[alpha]]) := [F.sub.[alpha]](1, q, [q.sup.2], ..., [q.sup.l-1], 0, 0, ...). Applying the linear transformation [[summation].sub.l[greater than or equal to]0][u.sup.l.sub.1][ps.sub.q1;l+1] and the specialization [t.sub.i] = [q.sup.i.sub.2] [u.sub.2] for all i = 0,1, ..., n to Theorem 1.2, we recover a result of Garsia and Gessel [8, Theorem 2.2] on the generating function of the joint distribution of five permutation statistics:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

Here [(u; q).sub.n]: [[PI].sub.0[less than or equal to]i[less than or equal to]n](1 - [q.sup.i]u) the set B(l, k) consists of pairs of weak compositions [lambda] = ([[lambda].sub.1], ..., [[lambda].sub.n]) and [mu] = ([[mu].sub.1], ..., [[mu].sub.n]) satisfying the conditions l [greater than or equal to] [[lambda].sub.1] [greater than or equal to] ... [greater than or equal to] [[lambda].sub.n], max{[[mu].sub.i]: 1 [less than or equal to] i [less than or equal to] n} [less than or equal to] k, and [[lambda].sub.i] = [[lambda].sub.i+1] [right arrow] [u.sub.i] [greater than or equal to] [[mu].sub.i+1] (such pairs ([lambda], [mu]) are sometimes called bipartite partitions), and inv([mu]) is the number of inversion pairs in [mu]. Some further specializations of Theorem 1.2 imply identities of Carlitz-MacMahon [6, 17] and Adin-Brenti-Roichman [1].

The structure of this paper is as follows. Section 2 reviews the representation theory of the 0-Hecke algebra. Section 3 studies the Stanley-Reisner ring of the Boolean algebra. Section 4 defines a 0-Hecke algebra action on the Stanley-Reisner ring of the Boolean algebra. The noncommutative and quasisymmetric characteristics are discussed in Section 5 and Section 6. Finally we give some remarks and questions for future research in Section 7, including a generalization to an action of the Hecke algebra of any finite Coxeter group on the Stanley-Reisner ring of the Coxeter complex.

2 Representation theory of the 0-Hecke algebra

We review the representation theory of the 0-Hecke algebra in this section. The (type A) Hecke algebra [H.sub.n](q) is the associative F(q)-algebra generated by [T.sub.1], ..., [T.sub.n-1] with relations

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (2)

It has an F(q)-basis {[T.sub.w]: w [member of] [G.sub.n]} where [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is a reduced expression.

Specializing q = 1 gives the group algebra of [G.sub.n], with [s.sub.i] = [T.sub.i[absolute value of q=1] and w = [T.sub.w[absolute value of q=1]]. Let w [member of] [S.sub.n]. The length of w equals inv(w) := #{(i, j): 1 [less than or equal to] i < j [less than or equal to] n, w(i) > w(j)}, and the descent set of w is D(w) = {i: 1 [less than or equal to] i [less than or equal to] n - 1, w(i) > w(i + 1)}. We write des(w) := [absolute value of D(w)] and maj(w) := [[sumamtion].sub.i[member of]D(w)]i.

Let [alpha] be a (weak) composition of n, and let [[alpha].sup.c] be the composition of n with D([[alpha].sup.c]) = [n - 1]\D([alpha]). The parabolic subgroup [G.sub.[alpha]] is the subgroup of [G.sub.n] generated by {[s.sub.i]: i [member of] D([[alpha].sup.c])}. The set of all minimal [G.sub.[alpha]]-coset representatives is [G.sup.[alpha]] := {w [member of] [G.sub.n]: D(w) [subset or equal to] D([alpha])}. The descent class of a consists of the permutations in [G.sub.n] with descent set equal to D([alpha]), and turns out to be an interval under the (left) weak order of [G.sub.n], denoted by [[w.sub.0]([alpha]), [w.sub.1]([alpha])]. One sees that [w.sub.0]([alpha]) is the longest element of the parabolic subgroup [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is the longest element in [G.sup.[alpha]] (c.f. Bjorner and Wachs [5, Theorem 6.2]).

Another interesting specialization of [H.sub.n](q) is the 0-Hecke algebra [H.sub.n](0), with generators [[bar.[pi]].sub.i] = [T.sub.i|q=0] for i = 1, ..., n - 1, and an F-basis {[[bar.[pi]].sub.w] = [T.sub.w|q=0]: w [member of] [G.sub.n]}. Let [[pi].sub.i] := [[bar.[pi]].sub.i] + 1. Then [[pi].sub.1], ..., [[pi].sub.n-1] form another generating set for [H.sub.n](0), with the same relations as (2) except [[pi].sup.2.sub.i] = [[pi].sub.i], 1 [less than or equal to] i [less than or equal to] n - 1. The element [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is well defined for any w [member of] [G.sub.n] with a reduced expression [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is another F-bases for [H.sub.n](0). One can check that [[pi].sub.w] equals the sum of [[bar.[pi]].sub.u] over all u less than or equal to w in the Bruhat order of [G.sub.n]. In particular, [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is the sum of [[bar.[pi]].sub.u] for all [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Norton [18] decomposed the 0-Hecke algebra [H.sub.n](0) into a direct sum of projective indecomposable submodules [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] for all [alpha] |= n (i.e. compositions of n). Each [P.sub.[alpha]] has an F-basis [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. Its radical rad Pa is the unique maximal [H.sub.n](0)-submodule spanned by [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. Although [P.sub.[alpha]] itself is not necessarily simple, its top [C.sub.[alpha]] := [P.sub.[alpha]]/rad [P.sub.[alpha]] is a one-dimensional simple [H.sub.n](0)-module with the action of [H.sub.n](0) given by

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

It follows from general representation theory of algebras (see e.g. [2, [section] I.5]) that {[P.sub.[alpha]]: [alpha] |= n} and {[C.sub.[alpha]]: [alpha] |= n} are the complete lists of pairwise non-isomorphic projective indecomposable and simple [H.sub.n](0)-modules, respectively.

Krob and Thibon [14] introduced a correspondence between [H.sub.n](0)-representations and the dual Hopf algebras QSym and NSym, which we review next. The Hopf algebra QSym has a free Z-basis of fundamental quasisymmetric functions [F.sub.[alpha]], and the dual Hopf algebra NSym has a dual basis of non- commutative ribbon Schur functions [s.sub.[alpha]], for all compositions [alpha].

Let M = [M.sub.0] [contains or equal to] [M.sub.1] [contains or equal to] ... [contains or equal to] [M.sub.k] [contains or equal to] [M.sub.k+1] = 0 be a composition series of [H.sub.n](0)-modules with simple factors [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. Then the quasisymmetric characteristic of M is

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

The noncommutative characteristic of a projective [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

It is not hard to extend these characteristic maps to [H.sub.n](0)-modules with gradings and filtrations.

3 Stanley-Reisner ring of the Boolean algebra

In this section we study the Stanley-Reisner ring of the Boolean algebra. The Boolean algebra [B.sub.n] is the ranked poset of all subsets of [n] := {1,2, ..., n} ordered by inclusion, with minimum element [phi] and maximum element [n]. The rank of a subset of [n] is defined as its cardinality. The Stanley-Reisner ring F[[B.sub.n]] of the Boolean algebra [B.sub.n] is the quotient of the polynomial algebra F [[y.sub.A]: A [subset or equal to] [n]] by the ideal ([y.sub.A][y.sub.B]: A and B are incomparable in [B.sub.n]). It has an F-basis {[y.sub.M]} indexed by the multichains M in [B.sub.n], and is multigraded by the rank multisets r(M) of the multichains M.

The symmetric group [G.sub.n] acts on the Boolean algebra [B.sub.n] by permuting the integers 1, ..., n. This induces an [G.sub.n]-action on the Stanley-Reisner ring F[[B.sub.n]], preserving its multigrading. The invariant algebra [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] consists of all elements in F[[B.sub.n]] invariant under this 6n- action. One can show that [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], where [THETA] := {[[theta].sub.0], ..., [[theta].sub.n]}. Garsia [7] showed that F[[B.sub.n]] is a free F[[THETA]]-module on the basis of descent monomials

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (3)

There is an analogy between the Stanley-Reisner ring F[[B.sub.n]] and the polynomial ring F[X] via the transfer map [tau]: F[[B.sub.n]] [right arrow] F[X] defined by

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

for all multichains M = ([A.sub.1] [subset or equal to] ... [subset or equal to] [A.sub.k]) in [B.sub.n]. It is not a ring homomorphism (e.g. [y.sub.{1}][y.sub.{2}] = 0 but [x.sub.1][x.sub.2] [not equal to] 0). Nevertheless, it induces an isomorphism [tau]: F[[B.sub.n]]/([[theta].sub.0]) [??] F[X] of [G.sub.n]-modules. Moreover, it sends the rank polynomials [[theta].sub.1], ..., [[theta].sub.n] to the elementary symmetric polynomials [e.sub.1], ..., [e.sub.n], and sends the descent monomials [Y.sub.w] in F[[B.sup.*.sub.n]] defined by (3) to the corresponding descent monomials in F[X] for all w [member of] [S.sub.n].

Example 3.1 The Boolean algebra [B.sub.3] consists of all subsets of {1, 2, 3}. Its Stanley-Reisner ring F[[B.sub.3]] is a free F[[THETA]]-module with a basis of descent monomials [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], where [THETA] consists of the rank polynomials [[theta].sub.0] = [y.sub.[phi]], [[theta].sub.1] := [y.sub.1] + [y.sub.2] + [y.sub.13], [[theta].sub.2] := [y.sub.12] + [y.sub.13] + [y.sub.23], [[theta].sub.3] := [y.sub.123]. The transfer map [tau] sends [[theta].sub.1], [[theta].sub.2], [[theta].sub.3] to [e.sub.1], [e.sub.2], [e.sub.3], and sends the six descent monomials in F[[B.sub.3]] to the six descent monomials 1, [x.sub.2], [x.sub.3], [x.sub.1][x.sub.3], [x.sub.2] [x.sub.3], [x.sub.2][x.sub.2] in F[[x.sub.1], [x.sub.2], [x.sub.3]].

The homogeneous components of F[[B.sub.n]] are indexed by multisets with elements in {0, ..., n}, or equivalently by weak compositions [alpha] of n. The [alpha]-homogeneous component F[[[B.sub.n]].sub.[alpha]] has an F- basis {[y.sub.M]: r(M) = D([alpha])}. Denote by Com(n, k) the set of all weak compositions of n with length k. If M = ([A.sub.1] [subset or equal to] ... [subset or equal to] [A.sub.k]) is a multichain of length k in [B.sub.n] then we set [A.sub.0] := [phi] and [A.sub.k+1] := [n] by convention. Define [alpha](M) := ([[alpha].sub.1], ..., [[alpha].sub.k+1]), where [[alpha].sub.i] = [absolute value of [A.sub.i]] - [absolute value of [A.sub.i-1]] for all i [member of] [k + 1]. Then [alpha](M) [member of] Com(n, k + 1) and D([alpha](M)) = r(M), i.e. [alpha](M) indexes the homogeneous component containing [y.sub.M]. Define [sigma](M) to be the minimal element in [G.sub.n] which sends the standard multichain [[[alpha].sub.1]] [subset or equal to] [[[alpha].sub.1] + [[alpha].sub.2]] [subset or equal to] ... [subset or equal to] [[[alpha].sub.1] + ... + [[alpha].sub.k]] with rank multiset D([alpha](M)) to M. Then [sigma](M) [member of] [G.sup.[alpha](M)].

The map M [right arrow] ([alpha](M), [sigma](M)) is a bijection between multichains of length k in [B.sub.n] and the pairs ([alpha], [sigma]) of [alpha] [member of] Com(n, k + 1) and [sigma] [member of] [G.sup.[alpha]]. A short way to write down this encoding of M is to insert bars at the descent positions of [sigma](M). For example, the length-4 multichain {2} [subset or equal to] {2} [subset or equal to] {1,2,4} [subset or equal to] [4] in B4 is encoded by 2[parallel] 14[absolute value of 3].

There is another way to encode the multichain M. Let [p.sub.i](M) := min{j[member of]G [k + 1]: i [member of] [A.sub.j]}, 1 [less than or equal to] i [less than or equal to] n. So [p.sub.i](M) is the first position where i appears in M. One checks that

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (4)

The map M [right arrow] p(M) := ([p.sub.1](M), ..., [p.sub.n](M)) is an bijection between the set of multichains with length k in [B.sub.n] and the set [[k + 1].sup.n] of all words of length n on the alphabet [k +1], for any fixed integer k [greater than or equal to] 0.

Let p(M) = p = ([p.sub.1], ..., [p.sub.n]) [member of] [[k + 1].sup.n]. Then inv(p(M)) = inv([sigma](M)). Let p' := ([p'.sub.1], ..., [p'.sub.k]) where [p'.sub.i] := [absolute value of {j: [p.sub.j](M) [less than or equal to] i}] = [absolute value of [A.sub.i]]. Then the rank multiset of M consists of [p'.sub.1], ..., [p'.sub.k]. Define

D(p) := {i [member of] [n - 1]: [p.sub.i] > [p.sub.i+1]}. For example, the multichain 3[absolute value of 14][absolute value of 2]5 corresponds to p = (2,4,1,2,5) [member of] [[5].sup.5], and one has p' = (1, 3, 3, 4), D(2, 5, 1, 2, 4) = {2}.

These two encodings (with slightly different notation) were already used by Garsia and Gessel [8] in their work on generating functions of multivariate distributions of permutation statistics.

4 0-Hecke algebra action

We saw an analogy between F[[B.sub.n]] and F[X] in the last section. The usual [H.sub.n](0)-action on the polynomial ring F[X] is via the Demazure operators

[[bar.[pi]].sub.i](f) := [x.sub.i+1]f - [x.sub.i+1][s.sub.i]f/[x.sub.i] - [x.sub.i+1], [for all f [member of] F[X], 1 [less than or equal to] i [less than or equal to] n - 1. (5)

The above definition is equivalent to

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (6)

Here m is any monomial in F[X] containing neither [x.sub.i] nor [x.sub.i+1]. Denote by [[bar.[pi]]'.sub.i] the operator obtained from (6) by taking only the leading term (underlined) in the lexicographic order of the result. Then [[bar.[pi]]'.sub.1], ..., [[bar.[pi]]'.sub.n-1] realize another [H.sub.n](0)-action on F[X]. We call it the transferred [H.sub.n](0)-action because it can be obtained by applying the transfer map [tau] to our [H.sub.n](0)-action on F[[B.sub.n]], which we now define.

Let M = ([A.sub.1] [subset or equal to] ... [subset or equal to] [A.sub.k]) be a multichain in [B.sub.n]. Recall that [p.sub.i](M) := min{j [member of] [k + 1]: i [member of] [A.sub.j]}, 1 [less than or equal to] i [less than or equal to] n. We define

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (7)

for i = 1, ..., n - 1. Applying the transfer map [tau] one recovers [[bar.[pi]]'.sub.i]. For instance, when n = 4 one has

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

One can check that [[bar.[pi]].sub.1], ..., [[bar.[pi]].sub.n-1] realize an [H.sub.n](0)-action on F[[B.sub.n]] preserving the multigrading of F[[B.sub.n]]. If one set [t.sub.i] = [t.sup.i] for i = 1, ..., n, then there is an isomorphism F[[B.sub.n]]/([empty set]) [??] F[X] of graded [H.sub.n](0)-modules (which can be given explicitly, but not via the transfer map t).

It is not hard to show that [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is the invariant algebra of the [H.sub.n](0)-action on F[[B.sub.n]], defined as

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Proposition 4.1 The [H.sub.n](0)-action on F[[B.sub.n]] is [THETA]-linear.

Therefore the coinvariant algebra F[[B.sub.n]]/([THETA]) is a multigraded [H.sub.n](0)-module, which is isomorphic to the regular representation of [H.sub.n](0) by Theorem 1.1. This cannot be obtained simply by applying the transfer map [tau], since [tau] is not a map of [H.sub.n](0)-modules.

5 Noncommutative characteristic

In this section we use the [H.sub.n](0)-action on the Stanley-Reisner ring F[[B.sub.n]] of the Boolean algebra [B.sub.n] to provide a noncommutative analogue of the following remarkable result.

Theorem 5.1 (Hotta-Springer [11], Garsia-Procesi [9]) For any partition [mu] = (0 < [[mu].sub.1] [less than or equal to] ... [less than or equal to] [[mu].sub.k]) of n, there exists an [G.sub.n]-invariant ideal JM of C[X] such that C[X]/[J.sub.[mu]] is isomorphic to the cohomology ring of the Springer fiber indexed by p and has graded Frobenius characteristic equal to the modified Hall-Littlewood symmetric function

[[??].sub.[mu]](X; t) = [summation over [lambda]][t.sup.n(mu])][K.sub.[lambda][mu]]([t.sup.-1])[s.sub.[lambda]] inside Sym[t]

where n([mu]) := [[mu].sub.k-1] + 2[[mu].sub.k-2] + ... + (k - 1)[[mu].sub.1] and [K.sub.[lambda][mu]](t) is the Kostka-Foulkespolynomial.

Example 5.2 Tanisaki [19] gives a construction for the ideal [J.sub.[mu]]. If p = ([1.sup.k], n - k) is a hook then [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is generated by [e.sub.1], ..., [e.sub.k] and all monomials [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] with 1 [less than or equal to] [i.sub.1] < ... < [i.sub.k+1] [less than or equal to] n.

Now consider a composition [alpha] = ([[alpha].sub.1], ..., [[alpha].sub.l]). The major index of [alpha] is maj([alpha]) := [[summation].sub.i[member of]D([alpha])] i. Viewing a partition [??] = (0 < [[mu].sub.1] [less than or equal to] [[mu].sub.2] [less than or equal to] ...) as a composition one has maj([mu]) = n([mu]). Recall that [??] := ([[alpha].sub.l], ..., [[alpha].sub.l]) and ac is the composition of n with D([[alpha].sup.c]) = [n - 1]\D([alpha]). We define [alpha]' := [[??].sub.c] = [([??]).sup.c]. One can identify a with a ribbon diagram, i.e. a connected skew Young diagram without 2 by 2 boxes, which has row lengths [[alpha].sub.1], ..., [[alpha].sub.l], ordered from bottom to top. Note that a ribbon diagram is a Young diagram if and only if it is a hook. One can check that [alpha]' is the transpose of a; see the example below.

[alpha] = (2, 3,1, 1) [[alpha].sup.c] = (1, 2,1, 3) [alpha]' = (3,1, 2, 1)

Bergeron and Zabrocki [3] introduced a noncommutative modified Hall-Littlewood symmetric function

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (8)

and a (q, t)-analogue

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (9)

for every composition [alpha], where [s.sub.[beta]] is the noncommutative ribbon Schur function indexed by [beta], and c([alpha], [beta]) := [[summation].sub.i[member of]D([alpha])[intersection]D([beta])] i. In our earlier work [13] we provided a partial representation theoretic interpretation for [[??].sub.a](x; t) when [alpha] = ([1.sup.k], n - k) is a hook, using the [H.sub.n](0)-action on the polynomial ring F[X] by the Demazure operators.

Theorem 5.3 ([13]) The ideal [J.sub.[mu]]of F[X] is [H.sub.n](0)-invariant if and only if [mu] = ([1.sup.n-k], k) is a hook, and if that holds then F[X]/[J.sub.[mu]] becomes a graded projective [H.sub.n](0)-module with

[ch.sub.t](F[X]/[J.sub.[mu]]) = [[??].sub.[mu]](x; t),

[Ch.sub.t](F[X]/[J.sub.[mu]]) = [[??].sub.[mu]](x; t).

Now we switch to the Stanley-Reisner ring F[[B.sub.n]] and define [I.sub.[alpha]] to be its ideal generated by

[[THETA].sub.[alpha]] := {[[theta].sub.i]: i [member of] D([alpha]) [union] {n}} and {[y.sub.A]: A [subset or equal to] [n], [absolute value of A] [not member of] D([alpha]) [union] {n}}

for any composition [alpha] of n. The following result is a restatement of Theorem 1.1.

Theorem 5.4 Let [alpha] be a composition of n. Then F[[B.sub.n]]/[I.sub.[alpha]] is a projective [H.sub.n](0)-module with multi-graded noncommutative characteristic equal to

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

One has [[??].sub.[alpha]](x; t, [t.sup.2], ..., [t.sup.n-1]) = [[??].sub.[alpha]](x; t), and one obtains [[??].sub.[alpha]](x; q, t) from [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] by taking [t.sub.i] = [t.sup.i] for all i [member of] D([delta]), and [t.sub.i] = [q.sup.n-i] for all i [member of] [n - 1]\D([alpha]).

Proof: There is an F-basis for F[[B.sub.[alpha]]]/([[THETA].sub.[alpha]]) given by the descent monomials [Y.sub.w] defined in (3) for all w [member of] [G.sup.[alpha]]. The result follows from the [H.sub.n](0)-action on this basis and (4).

The proof of this theorem is actually simpler than the proof of our partial interpretation for Ha (x; t) in [13]. This is because [[bar.[pi]].sub.i] sends a descent monomial in F[[B.sub.n]] to either 0 or [+ or -] 1 times a descent monomial, but sends a descent monomial in F[X] to a polynomial in general (whose leading term is still a descent monomial). We view the Stanley-Reisner ring F[[B.sub.n]] (or F[[B.sub.n]]/([empty set])) as a q = 0 analogue of the polynomial ring F[X]. For an odd (i.e. q = -1) analogue, see Lauda and Russell [16].

Remark 5.5 If [alpha] = ([1.sup.k], n - k) is a hook, one can check that the ideal [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] has generators [[theta].sub.1], ..., [[theta].sub.k] and all [y.sub.A] with A [subset or equal to] [n] and [absolute value of A] [not member of] [k]. By Example 5.2, the images of these generators under the transfer map t are the Tanisaki generators for the ideal [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], but [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

For any composition [alpha] |= n, one can view [[??].sub.[alpha]](x; [t.sub.1], ..., [t.sub.n-1]) as a modified version of

[H.sub.[alpha]] = [H.sub.[alpha]](x; [t.sub.1], ..., [t.sub.n-1]) := [summation over B[??][alpha]] [[t.bar].sup.D([alpha])\D([beta])][s.sub.[beta]].

Below are some properties satisfied by [H.sub.[alpha]], generalizing the properties of [H.sub.[alpha]](x; t) given in [3].

Proposition 5.6 Let [alpha] and [beta] be two compositions.

(I) [H.sub.[alpha]](0, ..., 0) = [s.sub.[alpha]], [H.sub.[alpha]](1, ..., 1) = [h.sub.[alpha]].

(ii) [[union].sub.n[greater than or equal to]0] {[H.sub.[alpha]]: [alpha] |= n} is a basis for NSym[[t.sub.1], [t.sub.2], ...].

(iii) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] for any pair of compositions a and ft.

(iv) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

(v) If n = [absolute value of [alpha]] and t|n := ([t.sub.1], ..., [t.sub.n-1], 1, [t.sub.1], ..., [t.sub.n-1], 1, ...) then

[H.sub.[alpha]](x; [t.sub.1], ..., [t.sub.n-1])[H.sub.[beta]](x; t|n) = [H.sub.[alpha][beta]](t|n).

6 Quasisymmetric characteristic

Now we study the quasisymmetric characteristic of F[[B.sub.n]]. The following lemma follows easily from (4).

Lemma 6.1 Let [alpha] be a weak composition of n. Then the [alpha]-homogeneous component F[[[B.sub.n]].sub.a] of the Stanley-Reisner ring F[[B.sub.n]] is an [H.sub.n](O)-submodule of F[[B.sub.n]] with homogeneous multigrading [[t.bar].sup.D([alpha])] and isomorphic to the cyclic module [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Since F[[[B.sub.n]].sub.[alpha]] is a cyclic multigraded [H.sub.n](0)-module, we get an N x [N.sup.n+1]-multigraded quasisymmetric characteristic

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (10)

where q keeps track of the length filtration and [t.bar] keeps track of the multigrading of F[[[B.sub.n]].sub.[alpha]]. This defines an N x [N.sup.n+1]-multigraded quasisymmetric characteristic for the Stanley-Reisner ring F[[B.sub.n]], which is explicitly given in Theorem 1.2 and restated below.

Theorem 6.2 The N x [N.sup.n+1]-multigraded quasisymmetric characteristic of F[[B.sub.n]] is

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Proof: Use the two encodings of the multichains in [B.sub.n] as well as the free F[[THETA]]-basis {[Y.sub.w]: w [member of] [G.sub.n]} of descent monomials for F[[B.sub.n]] discussed in Section 3.

Next we explain here how this theorem specializes to (1), a result of Garsia and Gessel [8, Theorem 2.2] on the multivariate generating function of the permutation statistics inv(w), maj(w), des(w), maj([w.sup.-1]), and des([w.sup.-1]) for all w [member of] [G.sub.n]. First recall that

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Let [ps.sub.q;l]([F.sub.[alpha]]) := [F.sub.[alpha]](1, q, [q.sup.2], ..., [q.sup.l-1], 0, 0, ...), and [(u; q).sub.n] := (1 - u)(1 - qu)(1 - [q.sup.2]u) ... (1 - [q.sup.n]u). It is not hard to check (see Gessel and Reutenauer [10, Lemma 5.2]) that

[summation over l[greater than or equal to]0] [u.sup.l][ps.sub.q;l+1]([F.sub.[alpha]]) = [q.sup.maj([alpha])][u.sup.des([alpha])]/[(u; q).sub.n]

Then applying the linear transformation 0 [[summation].sub.l[greater than or equal to]0][u.sup.l.sub.1][ps.sub.q1;l+1] and the specialization [t.sub.i] = [q.sup.i.sub.2][u.sup.2] for all i = 0,1, ..., n to Theorem 6.2 we recover (1).

A further specialization of Theorem 6.2 gives a well known result which is often attributed to Carlitz [6] but actually dates back to MacMahon [17, Volume 2, Chapter 4].

Corollary 6.3 (Carlitz-MacMahon) Let [[k + 1].sub.q] := 1 + q + [q.sup.2] + ... + [q.sup.k]. Then

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Theorem 6.2 also implies the following result, which was obtained by Adin, Brenti, and Roichman [1] from the Hilbert series of the coinvariant algebra [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Corollary 6.4 (Adin-Brenti-Roichman) Let Par(n) be the set of weak partitions [lambda] = ([[lambda].sub.1], ..., [[lambda].sub.n]) with [[lambda].sub.1] [greater than or equal to] ... [greater than or equal to] [[lambda].sub.n] [greater than or equal to] 0, and let m([lambda]) = ([m.sub.0]([lambda]), [m.sub.1]([lambda]), ...), where [m.sub.j]([lambda]) := #{1 [less than or equal to] i [less than or equal to] n: [[lambda].sub.i] = j}.

Then

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

7 Remarks and questions for future research

7.1 Hecke algebra action

It is well known that the symmetric group Sn is the Coxeter group of type [A.sub.n-1]. The Stanley-Reisner ring of [B.sub.n] is essentially the Stanley-Reisner ring of the Coxeter complex of [G.sub.n]. The Hecke algebra [H.sup.W](q) can be defined for any finite Coxeter group W. We can generalize our action [H.sub.n](0)-action on F[[B.sub.n]] to an [H.sub.W](q)-action on the Stanley-Reisner ring F(q)[[DELTA](W)] of the Coxeter complex [DELTA](W) of any finite Coxeter group W. We show similar results for this [H.sub.W](q)-action.

7.2 Gluing the group algebra and the 0-Hecke algebra

The group algebra FW of a finite Coxeter group W naturally admits both actions of W and [H.sub.W](0). Hivert and Thiery [12] defined the Hecke group algebra of W by gluing these two actions. In type A, one can also glue the usual actions of [G.sub.n] and [H.sub.n](0) on the polynomial ring F[X], but the resulting algebra is different from the Hecke group algebra of [G.sub.n].

Now one has a W-action and an [H.sub.W](0)-action on the Stanley-Reisner ring F[[DELTA](W)]. What can we say about the algebra generated by the operators [s.sub.i] and [[bar.[pi]].sub.i] on F[[DELTA](W)]? Is it the same as the Hecke group algebra of W? If not, what properties (dimension, bases, presentation, simple and projective indecomposable modules, etc.) does it have?

7.3 Tits Building

Let [DELTA](G) be the Tits building of the general linear group G = GL(n, [F.sub.q]) and its usual BN-pair over a finite field [F.sub.q]; see e.g. Bjorner [4]. The Stanley-Reisner ring F[[DELTA](G)] is a q-analogue of F[[B.sub.n]]. The nonzero monomials in F[[DELTA](G)] are indexed by multiflags of subspaces of [F.sup.n.sub.q], and there are [q.sup.inv(w)] many multiflags corresponding to a given multichain M in [B.sub.n], where w = [sigma](M). Can one obtain the multivariate quasisymmetric function identities in Theorem 1.2 by defining a nice [H.sub.n](0)-action on F[[DELTA](G)]?

Acknowledgements

The author is grateful to Victor Reiner for providing valuable suggestions. He also thanks Ben Braun and Jean-Yves Thibon for helpful conversations and email correspondence.

References

[1] R. Adin, F. Brenti, and Y. Roichman, Descent representations and multivariate statistics, Trans. Amer. Math. Soc. 357 (2005) 3051-3082.

[2] I. Assem, D. Simson, and A. Skowronski, Elements of the representation theory of associative algebras, vol. 1: Techniques of representation theory, London Mathematical Society Student Texts, vol. 65, Cambridge University Press, Cambridge, 2006.

[3] N. Bergeron and M. Zabrocki, q and q, t-analogs of non-commutative symmetric functions, Discrete Math. 298 (2005) 79-103.

[4] A. Bjorner, Some combinatorial and algebraic properties of Coxeter complexes and Tits buildings, Adv. Math. 52 (1984) 173-212.

[5] A. Bjorner and M. Wachs, Generalized quotients in Coxeter groups, Trans. Amer. Math. Soc. 308 (1988) 1-37.

[6] L. Carlitz, A combinatorial property of q-Eulerian numbers, Amer. Monthly, 82 (1975) 51-54.

[7] A. Garsia, Combinatorial methods in the theory of Cohen-Macaulay rings, Adv. Math. 38 (1980) 229-266.

[8] A. Garsia and I. Gessel, Permutation statistics and partitions, Adv. Math. 31 (1979) 288-305.

[9] A. Garsia and C. Procesi, On certain graded Sn-modules and the q-Kostka polynomials, Adv. Math. 94 (1992) 82-138.

[10] I. Gessel and C. Reutenauer, Counting permutations with given cycle structure and descent set, J. Combin. Theory Ser. A 64 (1993) 189-215.

[11] R. Hotta and T. A. Springer, A specialization theorem for certain Weyl group representations and an application to the Green polynomials of unitary groups, Invent. Math. 41 (1977), 113-127.

[12] F. Hivert and N. Thiery, The Hecke group algebra of a Coxeter group and its representation theory, J. Algebra, 321(2009) 2230-2258.

[13] J. Huang, 0-Hecke algebra actions on coinvariants and flags, to appear in J. Algebraic Combin.

[14] D. Krob and J.-Y. Thibon, Noncommutative symmetric functions IV: Quantum linear groups and Hecke algebras at q = 0, J. Algebraic Combin. 6 (1997) 339-376.

[15] A. Lascoux, J.-C. Novelli, and J.-Y. Thibon, Noncommutative symmetric functions with matrix parameters, J. Algebr. Comb. 37 (2013) 621-642.

[16] A. Lauda and H. Russell, Oddification of the cohomology of type A Springer varieties, arXiv:1203.0797.

[17] P.A. MacMahon, Combinatorial Analysis, Vols. 1 and 2, Cambridge University Press 1915-1916; reprinted by Chelsea 1960.

[18] P.N. Norton, 0-Hecke algebras, J. Austral. Math. Soc. A 27 (1979) 337-357.

[19] T. Tanisaki, Defining ideals of the closures of conjugacy classes and representations of the Weyl groups, Tohoku Math. J. 34 (1982), 575-585.

Jia Huang *

School of Mathematics, University of Minnesota, Minneapolis, MN 55455, USA

* Email: jiahuangustc@gmail.com. Supported by NSF grant DMS-1001933.
COPYRIGHT 2014 DMTCS
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Huang, Jia
Publication:DMTCS Proceedings
Article Type:Report
Date:Jan 1, 2014
Words:6901
Previous Article:l-restricted Q-systems and quantum affine algebras.
Next Article:An equivariant rim hook rule for quantum cohomology of Grassmannians.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters