1: \documentclass[11pt]{article}
2:
3: %\usepackage{times}
4: %\usepackage{geometry}
5: %\geometry{letterpaper}
6:
7: \usepackage{verbatim}
8: \usepackage{amsfonts}
9: \usepackage{amsthm}
10: \usepackage{amssymb}
11: \usepackage{amsmath}
12: %\usepackage[notref,notcite]{showkeys}
13: \usepackage[latin1]{inputenc}
14: %\usepackage[dvips]{graphics}
15: \usepackage{graphicx}
16: \usepackage{epstopdf}
17: %\usepackage[pdftex]{graphicx}
18: \DeclareGraphicsRule{.tif}{png}{.png}{`convert #1 `dirname #1`/`basename #1 .tif`.png}
19:
20: \addtolength{\textwidth}{2,5cm} \addtolength{\oddsidemargin}{-1cm}
21: \setlength{\topmargin}{-0,5cm} \addtolength{\textheight}{2cm}
22: \renewcommand{\baselinestretch}{1.1}
23:
24: \newcommand{\todo}[1]{\vspace{5 mm}\par \noindent
25: \marginpar{\textsc{ToDo}}
26: \framebox{\begin{minipage}[c]{0.95 \textwidth}
27: \tt #1 \end{minipage}}\vspace{5 mm}\par}
28:
29: %\renewcommand{\todo}[1]{}
30:
31: \newcommand{\idiot}[1]{\vspace{5 mm}\par \noindent
32: %\marginpar{\textsc{Note}}
33: \framebox{\begin{minipage}[c]{0.95 \textwidth}
34: \tt #1 \end{minipage}}\vspace{5 mm}\par}
35:
36: %\renewcommand{\idiot}[1]{}
37:
38:
39:
40:
41: \theoremstyle{plain}
42: \newcounter{cms}
43: \newtheorem{thm}{Theorem}[section]
44: \newtheorem{prop}[thm]{Proposition}
45: \newtheorem{lem}[thm]{Lemma}
46: \newtheorem{con}[thm]{Conjecture}
47: \newtheorem{cor}[thm]{Corollary}
48: \theoremstyle{definition}
49: \newtheorem{definition}[thm]{Definition}
50: \newtheorem{rem}[thm]{Remark}
51: \newtheorem{example}[thm]{Example}
52: \newtheorem{problem}[thm]{Problem}
53:
54: \newcommand{\SM}{{\mathcal S}}
55: \newcommand{\V}{{\mathcal V}}
56: \newcommand{\LE}{{\mathcal L}}
57: \newcommand{\la}{{\lambda}}
58: \newcommand{\g}{{\gamma}}
59: \newcommand{\al}{{\alpha}}
60: \newcommand{\be}{{\beta}}
61: \newcommand{\va}{{\varphi}}
62: \newcommand{\as}{{A}^{\s}}
63: \newcommand{\s}{{\sigma}}
64: \newcommand{\de}{{\delta}}
65: \newcommand{\sign}{\operatorname{sign}}
66: \newcommand{\imaj}{\operatorname{imaj}}
67: \newcommand{\Neg}{\operatorname{Neg}}
68: \newcommand{\NDes}{\operatorname{NDes}}
69: \newcommand{\IDes}{\operatorname{IDes}}
70: \newcommand{\Ddes}{\operatorname{Ddes}}
71: \newcommand{\maj}{\operatorname{maj}}
72: \newcommand{\fmaj}{\operatorname{fmaj}}
73: \newcommand{\Dmaj}{\operatorname{Dmaj}}
74: \newcommand{\inv}{\operatorname{inv}}
75: \newcommand{\me}{\operatorname{N}}
76: \newcommand{\DDes}{\operatorname{DDes}}
77: \newcommand{\nmaj}{\operatorname{nmaj}}
78: \newcommand{\dmaj}{\operatorname{dmaj}}
79: \newcommand{\des}{\operatorname{des}}
80: \newcommand{\ndes}{\operatorname{ndes}}
81: \newcommand{\ddes}{\operatorname{ddes}}
82: \newcommand{\Des}{\operatorname{Des}}
83: \newcommand{\Id}{\operatorname{Id}}
84: \newcommand{\Int}{\operatorname{Int}}
85: \newcommand{\Ad}{\operatorname{Ad}}
86: \newcommand{\For}{\operatorname{For}}
87: \newcommand{\NN}{\mathbb{N}}
88: \newcommand{\ZZ}{\mathbb{Z}}
89: \newcommand{\PP}{\mathbb{P}}
90: \newcommand{\zero}{\hat{0}}
91: \newcommand{\one}{\hat{1}}
92: \newcommand{\rk}{\operatorname{rk}}
93: \newcommand{\join}{\vee}
94: \newcommand{\AAA}{\textsf{A}}
95: \newcommand{\BBB}{\textsf{B}}
96: \newcommand{\un}{\underline{\nu}}
97: \newcommand{\ba}{\bar{\nu}}
98:
99:
100: %\newenvironment{proof}{\begin{trivlist}\item{\bf{Proof.}}}
101: % {\hfill\rule{2mm}{2mm}\end{trivlist}}
102:
103:
104:
105: \begin{document}
106:
107: \title{Equidistribution of negative statistics and quotients of Coxeter groups of type $B$ and $D$}
108: \author{Riccardo Biagioli \footnote{Dedicated to the memory of my friend and colleague Giulio Minervini.}\\
109: Universit\'e de Lyon, Universit\'e Lyon 1\\
110: Institut Camille Jordan - UMR 5208 du CNRS\\
111: 43, boulevard du 11 novembre 1918\\
112: F - 69622 Villeurbanne Cedex
113: }
114:
115: \date{}
116: \maketitle
117:
118: \begin{abstract}
119: We generalize some identities and $q$-identities previously known for the symmetric group to Coxeter groups of type $B$ and $D$. The extended results include theorems of Foata and Sch\"utzenberger, Gessel, and Roselle on various distributions of inversion number, major index, and descent number. In order to show our results we provide caracterizations of the systems of minimal coset representatives of Coxeter groups of type $B$ and $D$.
120: \end{abstract}
121:
122: \section{Introduction}
123:
124: A well known theorem of MacMahon \cite{MM} shows that the {\em length function} and the {\em major index} are equidistributed over the symmetric group $S_n$.
125: We recall that the length of a permutation $\sigma \in S_n$ is given by the {\em number of inversions}, denoted $\inv(\s):=|\{(i,j) \mid i<j , \; \s(i) > \s(j) \}|$, and the major index of $\s$ is the sum of all its {\em descents}. More precisely,
126: $$ \maj(\sigma):=\sum_{i \in \Des(\sigma)} i, $$
127: where $\Des(\s):=\{i \in [n-1] \mid \s(i)>\s(i+1)\}$. Foata gave a bijective proof of this equidistribution theorem in \cite{F}. He studied further his bijection and together with
128: %He defined a bijection $\varphi: S_n \rightarrow S_n$, with the following two properties
129: %\begin{equation*}\label{majinv}
130: %\maj(\s)=\inv(\varphi(\s)), \;\; {\rm and} \;\; \Des(\s^{-1})=\Des(\varphi(\s)^{-1}).
131: %\end{equation*}
132: Sch\"utzenberger derived the two following results \cite{FS}. The first one is a refinement of MacMahon's theorem, asserting the equidistribution of major index and number of inversions over descent classes.
133:
134: \begin{thm}[Foata-Sch\"utzenberger]\label{FS1}
135: Let $M=\{m_1,\ldots, m_t\}_< \subseteq \{1,\ldots, n-1\}$. Then
136: \[ \sum_{\{\s \in S_n \mid \Des(\s^{-1})=M\}} q^{\maj(\s)}= \sum_{\{\s \in S_n \mid \Des(\s^{-1})=M\}} q^{\inv(\s)}\]
137: \end{thm}
138: The second one concerns the symmetry of the distribution of the major index and the inversion number over the symmetric group.
139: \begin{thm}[Foata-Sch\"utzenberger] \label{FS2} The pairs of statistics $(\maj,\inv)$ and $(\inv,\maj)$ have the same distribution on $S_n$, namely
140: \[S_n(t,q):=\sum_{\s \in S_n} t^{\maj(\s)}q^{\inv(\s)}=\sum_{\s \in S_n} t^{\inv(\s)}q^{\maj(\s)}.\]
141: \end{thm}
142:
143:
144: Theorem \ref{FS1} has been extensively studied and generalized in many ways in the last three decades. Nevertheless, it still receives a lot of attention as shown by two recent papers of Hivert, Novelli, and Thibon \cite{HNT}, and of Adin, Brenti, and Roichman \cite{ABR2}, where a multivariate generalization and an extension to the hyperoctahedral group of it are provided. In the latter paper, the problem of finding an analogue of this Foata-Sch\"utzenberger theorem for the Coxeter groups of type $D$ is proposed \cite[Problem 5.6]{ABR2}.
145:
146: In this paper we answer this question. Actually, we show that the {\em negative major indices} ``$\nmaj$", introduced in \cite{ABR1} on Coxeter groups of type $B$, and ``$\dmaj$", defined in \cite{B} on Coxeter groups of type $D$, give generalizations of the first and second Foata-Sch\"utzenberger identities to $B_n$ and $D_n$. In our analysis we derive nice relations among {\em quotients}, or sets of minimal coset representatives, of $B_n$ and $D_n$ that are interesting in their own. Explicit maps between these quotients are shown, and used to compute some generating functions.
147:
148: Finally, we use our results, and the {\em negative descent numbers}, to give generalizations to $B_n$ and $D_n$ of two classical $q$-identities. The first one, due to Roselle \cite{Ros} (see also Rawlings \cite[(2.4)]{Raw}), is the generating function of the inversion number and major index over the symmetric group: for undefined notation see next section.
149: \begin{thm}[Roselle]\label{Roselle}
150: $$ \sum_{n\geq 0} S_n(t,q) \frac{u^n}{(t;t)_n (q;q)_n}=\frac{1}{(u;t,q)_{\infty,\infty}},$$
151: \end{thm}
152: \noindent where $S_0(t,q)=1$. The second one is the trivariate distribution of inversion number, major index, and {\em number of descents}, due to Gessel \cite[Theorem 8.4]{Ges}, (see also \cite{GG}).
153: \begin{thm}[Gessel]\label{Gessel}
154: $$\sum_{n\geq 0} \frac{u^n}{[n]_q !} \frac{\sum_{\s \in S_n} t^{\maj(\s)} q^{\inv(\s)} p^{\des(\s)}}{(t;q)_{n+1}}=\sum_{k \geq 0} p^{k} e[u]_q e[tu]_q\cdots e[t^k u]_q.$$
155: \end{thm}
156:
157: \section{Preliminaries and notation}
158:
159: In this section we give some definitions, notation and results that
160: will be used in the rest of this work. For $n \in \NN$ we let $[n]:= \{ 1,2, \ldots , n \} $ (where $[0]:= \emptyset $). Given $n, m \in \ZZ, \; n \leq m$, we let $[n,m]:=\{n,n+1, \ldots, m \}.$ We let $\PP:=\{1,2,3,\ldots\}.$
161: The cardinality of a set $A$ will be denoted by $|A|$ and we let ${[n] \choose 2}:=\{S \subseteq [n] \mid |S|=2 \}.$ Given a set $A$, we denote $A_<:=\{a_1,a_2,\ldots\}$ where $a_1<a_2<\ldots$.
162:
163: For our study we need notation for $q$-analogs of the factorial, binomial coefficient, and multinomial coefficient. These are defined by the following expressions
164: $$[n]_q:=1+q+q^2+\ldots + q^{n-1}; \;\;\;\;\; [n]_q!:=[n]_q[n-1]_q\cdots[2]_q[1]_q;$$
165: $$\begin{bmatrix} n \\ m \end{bmatrix}_q:=\frac{[n]_q!}{[m]_q![n-m]_q!}; \;\;\;\;\;\;
166: \begin{bmatrix} n \\ m_1, \; m_2, \ldots, m_t \end{bmatrix}_q:=\frac{[n]_q!}{[m_1]_q![m_2]_q! \cdots [m_t]_q!}.$$
167: As usual we let
168: \begin{eqnarray*}
169: (a;q)_0&:=&1 \\
170: (a;q)_n&:=&(1-a)(1-aq)\cdots (1-aq^{n-1})\\
171: (a;q)_\infty&:=&\prod_{n\geq 1}(1-aq^{n-1}).
172: \end{eqnarray*}
173: Moreover, for $r, s \in \NN$ we let
174: $$(a;t,q)_{r,s}:=\left\{\begin{array}{ll} \;\; 1 & {\rm if} \ r \ {\rm or} \ s \ {\rm are \ zero}\\
175: {\displaystyle\prod_{1\leq i\leq r}\prod_{1\leq j\leq s} (1-at^{i-1}q^{j-1})} & {\rm if} \ r,s \geq 1\end{array}\right.,$$
176: and
177: $$(a;t,q)_{\infty,\infty}:=\prod_{i\geq 1}\prod_{j\geq 1}(1-at^{i-1}q^{j-1}).$$
178: Finally,
179: $$e[u]_q:=\sum_{n\geq 0}\frac{u^n}{[n]_q !},$$
180: is the $q$-analogue of the exponential function. The following $q$-binomial theorem is well known (see e.g. \cite{An})
181: \begin{thm}\label{qbinomial}
182: $$(-xq;q)_n=\sum_{m=0}^n \begin{bmatrix} n \\ m \end{bmatrix}_q q^{{m+1 \choose 2}}x^m.$$
183: \end{thm}
184:
185:
186: \subsection{Coxeter groups of type $B$ and $D$}
187:
188: We denote by $B_{n}$ the group of all bijections $\be$ of the set
189: $[-n,n]\setminus \{0\}$ onto itself such that
190: \[\be(-i)=-\be(i)\]
191: for all $i \in [-n,n]\setminus \{0\}$, with composition as the
192: group operation. This group is usually known as the group of {\em
193: signed permutations} on $[n]$, or as the {\em hyperoctahedral
194: group} of rank $n$. If
195: $\be \in B_{n}$ then we write $\be=[\be(1),\dots,\be(n)]$ and we call this the
196: {\em window} notation of $\be$. As set of generators for $B_n$ we
197: take $S_B:=\{s_1^B,\ldots,s_{n-1}^B,s_0^B\}$ where for $i
198: \in[n-1]$
199: \[s_i^B:=[1,\ldots,i-1,i+1,i,i+2,\ldots,n] \;\; {\rm and} \;\; s_0^B:=[-1,2,\ldots,n].\]
200: It is well known that $(B_n,S_B)$ is a Coxeter system of type $B$ (see e.g., \cite[\S 8.1]{BB}).
201: \begin{figure}[htdb!]
202: \centering
203: \includegraphics[scale=.5]{115}
204: \caption{The Dynkin diagram of $B_n$}\label{graphB}
205: \end{figure}
206:
207: To give an explicit combinatorial description of the length function $\ell_B$ of $B_n$ with respect to $S_B$, we need the following statistics.
208: For $\be \in B_n$ we let
209: \begin{eqnarray*}
210: \me_1(\be) & := & |\{i \in [n] \mid \be(i)<0\}|, \ \ {\rm and}\\
211: \me_{2}(\be) &:=& \left|\left\{\{i,j\} \in {[n] \choose 2} \mid \be(i)+\be(j)<0 \right\}\right|.
212: \end{eqnarray*}
213: Note that, if $\be \in B_n$,
214: \begin{equation} \label{som}
215: \me_{1}(\be)+\me_{2}(\be)=-\sum_{\{i \in [n] \mid \be(i)<0\}} \be(i).
216: \end{equation}
217: \noindent For example if $\be=[-3,1,-6,2,-4,-5] \in B_{6}$ then
218: $\me_1(\be)=4$, and $\me_2(\be)=14$.
219: \smallskip
220:
221: The following characterizations of the length function, and of the right descent set of $\be \in B_n$ are well known \cite{BB}.
222:
223: \begin{prop} Let $\be \in B_n$. Then
224: \begin{eqnarray*}
225: \ell_B(\be)&=&\inv(\be)+\me_1(\be)+\me_2(\be), \ \ and\\
226: \Des_B(\be)&=&\{i \in [0,n-1] \mid \be(i) > \be(i+1)\},
227: \end{eqnarray*}
228: where $\be(0):=0$.
229: \end{prop}
230:
231: \bigskip
232:
233: We denote by $D_{n}$ the subgroup of $B_{n}$ consisting of all the
234: signed permutations having an even number of negative entries in
235: their window notation, more precisely
236: \[ D_{n} := \{\g \in B_{n} \mid \me_{1}(\g)\equiv 0 \; ({\rm mod} \; 2 ) \}. \]
237: It is usually called the {\em even-signed permutation group}. As a set of generators for $D_n$ we take
238: $S_D:=\{s_{0}^D,s_{1}^D,\dots,s_{n-1}^D\}$ where for $i \in [n-1]$
239: \[s_i^D:=s_i^B \;\; {\rm and} \;\; s_{0}^D:=[-2,-1,3,\ldots,n].\]
240: \begin{figure}[htdb!]
241: \centering
242: \includegraphics[scale=.5]{116}
243: \caption{The Dynkin diagram of $D_n$}
244: \label{graphD}
245: \end{figure}
246:
247: There is a well known direct combinatorial way to compute the length, and the right descent set of $\g \in D_{n}$, (see, e.g., \cite[\S 8.2]{BB}).
248: \begin{prop}Let $\g \in D_n$. Then
249: \begin{eqnarray*}\label{lD}
250: \ell_D(\g)&=&\inv(\g) + \me_{2}(\g),\; \; {\rm and} \\
251: \Des_D(\g)&=&\{i \in [0,n-1] \mid \g(i)>\g(i+1)\},
252: \end{eqnarray*}
253: where $\g(0):=-\g(2)$.
254: \end{prop}
255:
256: \subsection{Negative statistics}
257:
258: In \cite{ABR1}, Adin, Brenti and Roichman introduced the following statistics on $B_n$.
259: For $\be \in B_{n}$ let
260: \[\NDes(\be):=\Des(\be) \biguplus \{-\be(i) \mid \be(i)<0 \},\]
261: and define
262: \begin{eqnarray*}
263: \nmaj(\be):= \sum_{i \in \NDes(\be)}i , \ \ {\rm and} \ \
264: \ndes(\be):= |\NDes(\be)|.
265: \end{eqnarray*}
266: It follows from (\ref{som}) that
267: \begin{eqnarray}
268: \label{nmaj} \nmaj(\be)&=&\maj(\be)+\me_1(\be) +\me_2(\be), \ \ {\rm and} \\
269: \label{ndes} \ndes(\be)&=&\des(\be)+\me_1(\be).
270: \end{eqnarray}
271: For the element $\be=[-3,1,-6,2,-4,-5] \in B_{6}$, $\nmaj(\be)=29$, and $\ndes(\be)=7$.
272:
273: \bigskip
274:
275: In \cite{B}, a notion of descent multiset for $\g\in D_n$ is introduced
276: \[\DDes(\g):=\Des(\g) \biguplus \{ -\g(i)-1 \mid \g(i)<0 \} \setminus \{0\},\]
277: and the following statistics are defined
278: \begin{eqnarray*}
279: \dmaj(\g):= \sum_{i \in \DDes(\g)}i , \ \ {\rm and} \ \
280: \ddes(\g):= |\DDes(\g)|.
281: \end{eqnarray*}
282: It easily follows that
283: \begin{eqnarray}
284: \label{dmaj} \dmaj(\g)&=& \maj(\g) + \me_{2}(\g), \ \ {\rm and}\\
285: \label{ddes} \ddes(\g)&=&\des(\g)+ \me_1(\g)+\epsilon(\g),
286: \end{eqnarray}
287: where
288: \begin{equation}
289: \epsilon(\g) := \left \{\begin{array}{ll}
290: -1 & \mbox{if $1 \not \in \g([n])$} \\
291: 0 & \mbox{if $1 \in \g([n])$}.
292: \end{array} \right.
293: \end{equation}
294: For example if $\g=[-4,1,3,-5,-2,-6] \in D_{6}$
295: then $\dmaj(\g)=21$, and $\ddes(\g)=5$.
296: \bigskip
297:
298:
299: The statistics $\nmaj$ and $\dmaj$ are usually called {\em negative major indices}; $\ndes$ and $\ddes$ {\em negative descent numbers} for $B_n$ and $D_n$, respectively. The negative major indices are Mahonian statistics, namely they are equidistributed with the length over the group,
300: $$ \sum_{\be \in B_{n}} q^{\nmaj(\be)} = \sum_{\be \in B_{n}} q^{\ell_B(\g)}, \;\;\;\; {\rm and} \;\;\;\; \sum_{\g \in D_{n}} q^{\dmaj(\g)} = \sum_{\g \in D_{n}} q^{\ell_D(\g)}.$$
301: The pairs $(\ndes,\nmaj)$ and $(\ddes,\dmaj)$ give generalizations to $B_n$ and $D_n$ of a famous identity of Carlitz, see \cite[Theorem 3.2]{ABR1}, and \cite[Theorem 3.4]{B}.
302:
303: \subsection{Quotients of Coxeter groups}
304:
305: To show some of the next results we will need of the following decomposition that comes
306: from the general theory of Coxeter group. We refer the reader to \cite{BB} for any undefined notation.
307:
308: Let $(W,S)$ be a Coxeter
309: system, for $J \subseteq S$ we let $W_{J}$ be the {\em parabolic subgroup} of $W$
310: generated by $J$, and
311: $$
312: W^{J} := \{ w \in W \mid \ell(ws) > \ell(w) \; \; \mbox {for all} \; \; s \in J \},
313: $$
314: the set of minimal left coset representatives of $W_{J}$, or the {\em (right) quotient}. The quotient $W^{J}$ is a poset
315: according to the Bruhat order. The following is well known (see \cite[\S 2.4]{BB}).
316: \begin{prop}
317: \label{2.1} Let $(W,S)$ be a Coxeter system, and let $J \subseteq S$. Then:
318: \begin{itemize}
319: \item[i)] Every $w \in W$ has a unique factorization
320: $w=w^{J}w_{J}$ such that $w^{J} \in W^{J}$ and $w_{J} \in W_{J}$.
321: \item[ii)] For this factorization $\ell(w)=\ell(w^{J})+\ell(w_{J})$.
322: \end{itemize}
323: \end{prop}
324:
325: As a first application of this decomposition to the groups $B_n$ (and $D_n$),
326: let us consider the parabolic subgroup generated by $J:=S_B \setminus \{s_0^B\}$. In this case, by looking at the Dynkin diagram in Figure \ref{graphB}, we obtain that $B_J=S_n$. Moreover it is not hard to see that
327: \begin{eqnarray}
328: \label{T} B^J:=B_n^J=\{u \in B_{n} \mid u(1)<u(2)<\ldots<u(n) \}.
329: \end{eqnarray}
330: Hence from Proposition \ref{2.1} we get
331: \begin{equation}
332: \label{decomp} B_{n}= \biguplus_{\s \in S_{n}}\{ u \s \mid u \in B^J\},
333: \end{equation}
334: where $\biguplus$ denotes disjoint union. Note that in the case $D_n$, for $J:=S_D \setminus \{s_0^D\}$, a similar decomposition holds,
335: \begin{equation*}
336: D_{n}= \biguplus_{\s \in S_{n}}\{ u \s \mid u \in D^J\},
337: \end{equation*}
338: where once again $D_J=S_n$, and $D^J=\{u \in D_{n} \mid u(1)<u(2)<\ldots<u(n) \}.$
339:
340: \begin{rem}
341: The construction or right quotient can be mirrored, by considering {\em left descents}. Let $J\subseteq S$. A {\em left quotient} of $W$ is defined by
342: $$ ^J W:=\{w \in W \mid \ell(s w)>\ell(w)\ {\rm for \ all} \ s \in J \}.$$
343: Proposition \ref{2.1} holds for left quotients too, but the factorization in $i)$ becomes $w=w_J \cdot ^J\!w$, with $^J\! w \in$ $^J\! W$.
344: Left and right quotients are isomorphic posets, by means of the inversion map. In the next section, we will work with subsets of $B_n$ and $D_n$ that are left quotients. They are called descent classes for reasons that will be immediately clear.
345: \end{rem}
346:
347:
348:
349: \section{Combinatorial description of descent classes}
350:
351: Let us fix a subset of descents $M:=\{m_1, m_2, \ldots, m_t\}_< \subseteq [0,n-1]$. The set
352: \begin{equation}\label{Bdescentclass}
353: B(M):=\{\be \in B_n \mid \Des_B(\be^{-1}) \subseteq M\},
354: \end{equation}
355: is usually called a $B$-{\em descent class}. Note that this set is nothing but a left quotient of $B_n$. More precisely, it is the one corresponding to the subset $J=S \setminus \tilde{M}$, where $\tilde{M}:=\{s_i \mid i \in M\}$. The following result can be found in \cite[Lemma 4.1]{ABR2}.
356:
357: \begin{lem}\label{Bcombclass} Let $\be \in B_n$, and $M=\{m_1,\ldots,m_t\}_{<}\subseteq [0,n-1]$. Let $m_{t+1}:=n$. Then $\Des_B(\be^{-1})\subseteq M$ if and only if there exist (unique) integers $r_1,\ldots, r_t$ satisfying $m_i \leq r_i \leq m_{i+1}$ for all $i$, and such that $\be$ is a shuffle of the following increasing sequences:
358: \begin{equation}\label{shuffleB}
359: \begin{array}{l}
360: (1,2,\ldots,m_1), \\
361: \left(-r_1,-r_1+1,\ldots,-(m_1+1)\right), \\
362: (r_1+1,r_1+2,\ldots,m_2), \\
363: \hspace{2cm} \vdots \\
364: \big(-r_t,-r_t+1,\ldots,-(m_t+1)\big), \\
365: (r_t+1,r_t+2,\ldots,n).
366: \end{array}
367: \end{equation}
368: Some of these sequences may be empty, if $r_i=m_i$ or $r_i=m_{i+1}$ for some $i$, or if $m_i=0$.
369: \end{lem}
370:
371: The following one is an explicit description of $D$-descent classes
372: \begin{equation}\label{Ddescentclass}
373: D(M):=\{\g \in D_n \mid \Des_D(\g^{-1}) \subseteq M\}.
374: \end{equation}
375: \begin{lem}\label{Dcombclass} Let $\g \in D_n$, and $M=\{m_1,\ldots,m_t\}_{<}\subseteq [0,n-1]$. Let $m_{t+1}:=n$. Then $\Des_D(\g^{-1})\subseteq M$ if and only if there exist (unique) integers $r_1,\ldots, r_t$ satisfying $m_i \leq r_i \leq m_{i+1}$ for all $i$, and such that $\g$ is a shuffle of the following increasing sequences. There are three cases, and six possible ``blocks" of sequences.
376: \begin{itemize}
377: \item[1)] If $0 \in M$: $(m_1=0)$
378: \begin{equation}\label{shuffleD-O}
379: \begin{array}{l}
380: (-r_1,-r_1+1,\ldots,-2,-1), \\
381: (r_1+1,r_1+2,\ldots,m_2), \\
382: \hspace{2cm} \vdots \\
383: \big(-r_t,-r_t+1,\ldots,-(m_t+1)\big),\\
384: (r_t+1,r_t+2,\ldots,n), \\
385: \end{array}
386: \end{equation}
387: with \ {$\displaystyle \sum_{i=1}^t(r_i-m_i) \equiv 0 \; ({\rm mod} \ 2)$}.
388:
389: \item[2)] If $0, 1\not\in M$: (note $m_1\geq 2$)
390: \begin{equation}\label{shuffleD-01}
391: \begin{array}{lll}
392: (1,2,\ldots,m_1), & (-1,2,\ldots,m_1) &\\
393: \left(-r_1,-r_1+1,\ldots,-(m_1+1)\right), & (-r_1,-r_1+1,\ldots,-(m_1+1)) &\\
394: (r_1+1,r_1+2,\ldots,m_2), & (r_1+1,r_1+2,\ldots,m_2) & \\
395: \hspace{2cm} \vdots & \hspace{2cm} \vdots \\
396: \big(-r_t,-r_t+1,\ldots,-(m_t+1)\big),& \big(-r_t,-r_t+1,\ldots,-(m_t+1)\big) &\\
397: (r_t+1,r_t+2,\ldots,n) & (r_t+1,r_t+2,\ldots,n)&
398: \end{array}
399: \end{equation}
400: with
401: $$
402: \begin{array}{lll}
403: {\displaystyle \sum_{i=1}^t(r_i-m_i) \equiv 0 \; ({\rm mod} \ 2)}; & {\displaystyle \sum_{i=1}^t(r_i-m_i) \equiv 1 \; ({\rm mod} \ 2).} &
404: \end{array}$$
405: \item[3)] If $0 \not\in M$ and $1 \in M$: (note $m_2\geq 2$, and $r_1\geq 2$)
406: \begin{small}
407: \begin{equation}\label{shuffleD-1}
408: \begin{array}{llll}
409: (1) & & (-r_1,\ldots,-2,1) & \\
410: (2,3,\ldots,m_2), & (-1,2,3,\ldots,m_2) & (r_1+1,r_1+2,\ldots,m_2)&\\
411: \big(-r_2,-r_2+1,\ldots,-(m_2+1)\big), & \big(-r_2,-r_2+1,\ldots,-(m_2+1)\big), & \big(-r_2,-r_2+1,\ldots,-(m_2+1)\big) & \\
412: (r_2+1,r_2+2,\ldots,m_3), & (r_2+1,r_2+2,\ldots,m_3), & (r_2+1,r_2+2,\ldots,m_3) & \\
413: \hspace{2cm} \vdots & \hspace{2cm} \vdots & \hspace{2cm} \vdots \\
414: \big(-r_t,-r_t+1,\ldots,-(m_t+1)\big),& \big(-r_t,-r_t+1,\ldots,-(m_t+1)\big),& \big(-r_t,-r_t+1,\ldots,-(m_t+1)\big) &\\
415: (r_t+1,r_t+2,\ldots,n) & (r_t+1,r_t+2,\ldots,n) & (r_t+1,r_t+2,\ldots,n)&
416: \end{array}
417: \end{equation}
418: \end{small}
419: with
420: $$\begin{array}{llll}
421: {\displaystyle \sum_{i=2}^t(r_i-m_i) \equiv 0 \; ({\rm mod} \ 2)}; & {\displaystyle \sum_{i=2}^t(r_i-m_i) \equiv 1 \; ({\rm mod} \ 2)}; & {\displaystyle \sum_{i=1}^t(r_i-m_i) \equiv 0 \; ({\rm mod} \ 2).}& \end{array}$$
422: \end{itemize}
423: Some of these sequences may be empty, if $r_i=m_i$ or $r_i=m_{i+1}$ for some $i$, or if $m_i=0$.
424: \end{lem}
425:
426: \begin{proof}
427: The only difference with respect to the $B_n$ case is for the $0,1$ descents. They depend on the relative positions of $\pm 1$ and $\pm 2$ in the window notation of $\g$. The following are the $D$-descent classes of all elements of $B_2$. We have that
428: $$\begin{array}{llll}
429: D(\emptyset)&=\{[1,2], [-1,2]\}\\
430: D(\{0\})&=\{[2,-1], [-2,-1]\}\\
431: D(\{1\})&= \{[-2,1], [2,1]\}\\
432: D(\{0,1\})&= \{[1,-2], [-1,-2]\}.
433: \end{array}$$
434: From this, the parity conditions $\sum_{i=1}^t(r_i-m_i) \equiv 0 $ or $\equiv 1$ (mod $2$), and Lemma \ref{Bcombclass} the result follows.
435: \end{proof}
436:
437: \begin{rem}\label{rem33}
438: Let us fix a subset of descents $M:=\{m_1,m_2,\ldots, m_t\}$. Consider the decompositions of $B_n$ and $D_n$ given by Proposition \ref{2.1} by using left quotients.
439: % $$ B_n= B_M B(M) \;\;\; {\rm and} \;\;\; D_n=D_M D(M).$$
440: Recall that $|B_n|=2^n n!$ and that $|D_n|=2^{n-1} n!$. By looking at the Dynkin diagrams in Figure \ref{graphB} and Figure \ref{graphD}, it is easy to derive the following equalities.
441: \begin{itemize}
442: \item[{\small$\clubsuit$}] If $0 \in M$, then $|B(M)|=2\cdot |D(M)|$;
443: \item[{\small$\clubsuit$}] If $0,1 \not\in M$, then $|B(M)|=|D(M)|$;
444: \item[{\small$\clubsuit$}] If $0\not\in M$, and $1\in M$, then $|B(M)|=m_2\cdot |D(M)|$.
445: \end{itemize}
446: \end{rem}
447:
448: Now we make explicit these equalities by showing relations between $D$ and $B$ left quotients.
449: %In order to do so, we consider the following linear extension to $B_n$ of the Bruhat order on $D_n$. More precisely,
450: %for $\be, \g \in B_n$, we say that
451: %$$\be \prec \g \Longleftrightarrow \ell_D(\g) < \ell_D(h),$$
452: %and we call it the {\em $D$-length order.}
453:
454: \begin{prop}\label{L1}
455: Let $0 \in M$. Then
456: \begin{itemize}
457: \item[i)] $B(M)$ splits into the disjoint union
458: $$B(M)=D(M) \uplus \bar{D}(M),$$
459: where $\bar{D}(M):=\{\bar{\g}=(-\g(1),\g(2),\ldots,\g(n)) \mid \g\in D(M)\}= \{\g\cdot s_0^B \mid \g\in D(M)\}$.
460: \item[ii)] Moreover
461: $$\sum_{\be \in B(M)} q^{\ell_D(\be)} = 2 \cdot { \sum_{\g \in D(M)} q^{\ell_D(\g)}}.$$
462: \end{itemize}
463: \end{prop}
464:
465: \begin{proof}
466: Let $\g \in D(M)$. By Lemma \ref{Dcombclass} $\g$ is a shuffle of the sequences in (\ref{shuffleD-O}), and so it can also be obtained as a shuffle of the sequences in (\ref{shuffleB}). Hence $\g \in B(M)$. Now, let us change the sign to the first entry of $\g$, by getting $\bar{\g}$. We are changing the sign of $-r_i$, or of $r_i+1$ for $i \in [t]$, in one of the sequences in (\ref{shuffleD-O}). Note that this operation does not create a new $B$-descent for $\bar{\g}$. Hence $\bar{\g} \in B(M)\setminus D(M)$. More precisely, $\bar{\g}$ can be obtained by shuffling the same sequences that give $\g$ where the twos involving $r_i$ are replaced either by $(-r_i+1,\ldots,-(m_i+1))$ and $(r_i, r_i+1,\ldots,m_i)$, or by $(-r_i-1,\ldots,-(m_i+1))$ and $(r_i+2,\ldots,m_i)$, depending if it is the sign of $-r_i$, or of $r_i+1$, that changes. All those sequences belong to (\ref{shuffleB}). So $i)$ follows by Remark \ref{rem33}.
467:
468: Now, it is easy to see that for all $\g \in D(M)$, one has $\ell_D(\g)=\ell_D(\bar{\g})$. To see that, suppose $\g(1)>0$. Then
469: \begin{eqnarray*}
470: \inv(\bar{\g})=\inv(\g) - (\g(1)-1)\;\; {\rm and} \;\;\me_2(\bar{\g})=\me_2(\g)+(\g(1)-1),
471: \end{eqnarray*}
472: and so the length $\ell_D$ is stable. If $\g(1)<0$ a similar computations holds, hence $ii)$ follows.
473: \end{proof}
474: Note that the two subsets $D(M)$ and $\bar{D}(M)$ are not isomorphic as posets, when they are considered as sub-posets of $(B(M),<_B)$, where $<_B$ denote the $B$-Bruhat order . An example is given for $n=3$ and $M=\{0,2\}$.
475:
476: \begin{prop}\label{L2}
477: Let $0,1 \not \in M$. Then
478: \begin{itemize}
479: \item[i)] The map $\varphi : B(M) \longrightarrow D(M)$ defined by
480: $$ \beta \mapsto \left\{ \begin{array}{ll}
481: \be, & \text{if} \;\; \be \in D_n;\\
482: s^B_0 \cdot \be, & \text{otherwise},
483: \end{array} \right.$$
484: is a bijection.
485: \item[ii)] Moreover
486: $$\sum_{\be \in B(M)} q^{\ell_D(\be)} = \sum_{\g \in D(M)} q^{\ell_D(\g)}.$$
487: \end{itemize}
488: \end{prop}
489: \begin{proof}
490: Let $\be \in B(M)$, it is a shuffle of the sequences in (\ref{shuffleB}). If $\be \in D_n$, then it is also a shuffle of the sequences in the first block of (\ref{shuffleD-01}). Hence $\be \in D(M)$. Now suppose that $\be \not\in D_n$. Since $0 \not\in \Des_B(\be^{-1})$, then $1 \in \be[n]$. By multiplying on the left by $s_0^B$, we change the sign of $1$, and so the parity of $\beta$. Hence $s_0^B \cdot\beta \in D_n$. Actually, we obtain an element which is a shuffle of the sequences in second block of (\ref{shuffleD-01}). From Remark \ref{rem33} $i)$ follows.
491:
492: Since $\me_2(\g)=\me_2(\varphi(\g))$ and $\inv(\g)=\inv(\varphi(\g))$, one has $\ell_D(\g)=\ell_D(\varphi(\g))$, and so $ii)$ follows.
493: \end{proof}
494: The map $\varphi$ is not a poset isomorphism between $(B(M),<_B)$ and $(D(M),<_D)$, where $<_B$ and $<_D$ denote the corresponding Bruhat orders. When $n=3$, and $M=\{2\}$, $B(M)$ is a chain, while in $D(M)$ there are two elements not comparable.
495:
496: \begin{prop}\label{L3}
497: Let $0 \not\in M$, and $1\in M$. Then
498: \begin{itemize}
499: \item[i)] $B(M)$ splits as the disjoint union of the following $m_2$ subsets
500: $$B(M)=D_1(M)\uplus D_{12}(M) \uplus \ldots \uplus D_{12\ldots m_2}(M).$$
501: Each $D_{1\ldots i}(M)$ is in bijection with $D(M)$, and it is recursively defined as follows:
502:
503: 1) $D_1(M)$ is obtained by shuffling the sequences defining $D(M)$ where $-1$ (if present) is replaced with $1$.
504:
505: %2) $D_{12}(M)$ is obtained by shuffling the sequences defining $D_1(M)$ where: \\
506: %- $1$ and $\pm 2$ are switched if they are in the same sequence;\\
507: %- $\pm 2$ is replaced by $1$, and $1$ is replaced by $-2$, otherwise. This happens only when $1$ is at the beginning of a sequence.
508:
509: 2) For each $i\geq 2$, $D_{12\ldots i}(M)$ is obtained by shuffling the sequences defining $D_{12\ldots i-1}(M)$ where:
510: \begin{itemize}
511: \item[${\small \clubsuit}$] $1$ and $\pm i$ are swiched if they are in the same sequence;
512: \item[${\small \clubsuit}$] $ i$ is replaced by $1$, and $1$ is replaced by $-i$, otherwise. This case happens when $1$ is at the beginning of a sequence of type $(1,-(i-1),\ldots, -2)$, and $i$ is the initial value of the sequence $(i,i+1,\ldots, m_2)$.
513: \end{itemize}
514: \item[ii)] Moreover
515: $$ \sum_{\be \in B(M)} q^{\ell_D(\be)} = [m_2]_q \cdot \sum_{\g \in D(M)} q^{\ell_D(\g)}.$$
516: \end{itemize}
517: \end{prop}
518:
519: \noindent Before writing down the proof let us consider an example.
520:
521: \begin{example}
522: Consider $n=4$ and $M=\{1,3\}$. Then $D(M)$ is given by the shuffles of the following blocks of increasing sequences (written in column).
523: $$D(M)=\left\{
524: \begin{array}{rrrr}
525: (1) & (-1,2,3) & (-2,1) & (-3,-2,1) \\
526: (2,3); & (-4); & (3); & (4) \\
527: (4) & & (-4) & \\
528: \end{array}\right\}
529: $$
530: Then $B(M)$ splits as disjoint union of the following three subsets:
531: $$D_1(M)=\left\{
532: \begin{array}{rrrr}
533: (1) & (1,2,3) & (-2,1) & (-3,-2,1) \\
534: (2,3); & (-4); & (3); & (4) \\
535: (4) & & (-4) & \\
536: \end{array}\right\}
537: $$
538: $$D_{12}(M)=\left\{
539: \begin{array}{rrrr}
540: (-2) & (2,1,3) & (1,-2) & (-3,1,-2) \\
541: (1,3); & (-4); & (3); & (4) \\
542: (4) & & (-4) & \\
543: \end{array}\right\}
544: $$
545: $$D_{123}(M)=\left\{
546: \begin{array}{rrrr}
547: (-2) & (2,3,1) & (-3,-2) & (1,-3,-2) \\
548: (3,1); & (-4); & (1); & (4) \\
549: (4) & & (-4) & \\
550: \end{array}\right\}.
551: $$
552: \end{example}
553: \medskip
554:
555: \begin{proof}
556: The transformations defining $D_ {1\ldots i}(M)$ involve only the first two sequences of the three blocks of (\ref{shuffleD-1}). It is easy to see that $D_{1\ldots i}(M) \subseteq B(M)$ for all $i \in [m_2]$, and that $D_{1\ldots i}(M)$ and $D_{1\ldots j}(M)$ are disjoint if $i\neq j$. Hence the decomposition in $i)$ follows from Remark \ref{rem33}.
557:
558: Since changing $-1$ into $1$ in a signed permutation $\g$ affects neither $\inv(\g)$ nor $\me_2(\g)$, it follows that
559: $$\sum_{\g \in D(M)} q^{\ell_D(\g)} = \sum_{\g \in D_{1}(M)} q^{\ell_D(\g)}.$$
560: Now let us show that for all $i\geq 2$
561: $$\sum_{\g \in D_{1\ldots i}(M)} q^{\ell_D(\g)} = q \sum_{\g \in D_{1\ldots i-1}(M)} q^{\ell_D(\g)}.$$
562: Let $\g \in D_{1\ldots i-1}(M)$. Consider the block in (\ref{shuffleD-1}) whose a particular shuffle gives $\g$.
563:
564: If $1$ and $\pm i$ are in the same sequence, it can be either of the form $(\ldots, 1,i,\ldots,m_2)$, or of the form $(-r_1,\ldots, -i,1\ldots,-2)$. Now consider the shuffle giving $\g$, where $1$ has been switched with $\pm i$. We get a new element $\bar{\g}\in D_{1\ldots i}(M)$. It is clear that $\bar{\g}$ has one more inversion with respect to $\g$, and so the $D$-length go up by $1$. In fact, all other sequences in the block (whose shuffle gives $\g$) are made by elements that are either all bigger or all smaller of both $1$ and $\pm i$. Hence the difference between $\inv(\g)$ and $\inv(\bar{\g})$ depends only on the relative positions of $1$ and $\pm i$ within the same sequence.
565:
566: Suppose that $1$ and $i$ are not in the same sequence. This means that $1$ is at the beginning of the sequence $(1,-(i-1),\ldots,-2)$ and $i$ is at the beginning of the sequence $(i,i+1,\ldots,m_{2})$. So $\bar{\g} \in D_{1\ldots i}(M)$, the element corresponding to $\g$ after the switch, is obtained by shuffling a block that contains the following two sequences
567: $$(-i,-(i-1),\ldots,-2) \;\; \text{and}\;\; (1,i+1,\ldots,m_{2}).$$
568: Once again all other sequences of the block are made by elements that are either all smaller or bigger of both $1$ and $i$.
569: The difference between the values of $\inv(\bar{\g})$ and $\inv(\g)$ depends only on the relative positions of $1$ and $i$. Hence
570: $\bar{\g}$ loses $i-2$ inversions with respect to $\g$ (the ones given by the $1$ at the beginning of the sequence), and $\me_2(\bar{\g})= \me_2(\g) +(i-1)$ thanks to $-i$. So $\ell_D(\bar{\g})=\ell_D(\g)+1$.
571: \end{proof}
572:
573:
574: \section{Equidistribution over descent classes}
575:
576: In this section we show generalizations of Theorem \ref{FS1} to Coxeter groups of type $B$ and $D$. We need the following classical result; see \cite[Theorem 3.1]{GG}, and \cite[Example 2.2.5]{StaEC1} for a proof.
577:
578: \begin{thm}\label{stanley}
579: Let $n \in \PP$ and $M=\{m_1,m_2,\ldots, m_t\}_{<}\subseteq [n-1]$. Then
580: \begin{eqnarray*}\sum_{\{\s \in S_n \mid \Des(\s^{-1})\subseteq M\}}q^{\maj(\s)} \ = \sum_{\{\s \in S_n \mid \Des(\s^{-1})\subseteq M\}}q^{\inv(\s)} = \begin{bmatrix} n \\ m_1, \; m_2-m_1, \ldots, n-m_t \end{bmatrix}_q.
581: \end{eqnarray*}
582: \end{thm}
583:
584: \begin{thm}\label{mainB}
585: Let $n \in \PP$ and $M=\{m_1,m_2,\ldots,m_t\}_<\subseteq [0,n-1]$. Then
586: \begin{eqnarray}\label{th2}
587: \sum_{\{\be \in B_n \mid \Des_B(\be^{-1})\subseteq M\}} q^{\nmaj(\be)} &=& \sum_{\{\be \in B_n \mid \Des_B(\be^{-1})\subseteq M\}} q^{\ell_B(\be)} = \sum_{\{\be \in B_n \mid \Des_B(\be^{-1})\subseteq M\}}q^{\fmaj(\be)} \nonumber
588: \\
589: &=& \begin{bmatrix} n \\ m_1, \; m_2-m_1, \ldots, n-m_t \end{bmatrix}_q \cdot \displaystyle{\prod_{i=m_1+1}^{n} (1+q^i)}.
590: \end{eqnarray}
591: \end{thm}
592:
593: \begin{proof}
594: Let us denote by ${\rm Sh}(r_1,\ldots,r_t)$ the set of signed permutations obtained as shuffles of the sequences in (\ref{shuffleB}), with prescribed $r_1,\ldots,r_t$. From Theorem \ref{stanley} it follows that
595: \begin{equation}\label{crucialpoint}
596: \sum_{\beta \in {\rm Sh}(r_1,\ldots,r_t)}q^{\maj(\beta)}=\sum_{\beta \in {\rm Sh}(r_1,\ldots,r_t)}q^{\inv(\beta)}= \begin{bmatrix} n \\ m_1, \; r_1-m_1, \ldots, r_t-m_t, \; n-r_t \end{bmatrix}_q.
597: \end{equation}
598: In fact inversion number and major index of a shuffle depend only on the order of the elements in the shuffled sequences. From this, and the definitions of $\nmaj(\beta)=\maj(\beta)+\me_1(\beta)+\me_2(\beta)$ and of $\ell_B(\beta)=\inv(\beta)+\me_1(\beta)+\me_2(\beta)$, the first equality in (\ref{th2}) follows. The second equality and the sum have been computed in \cite{ABR2}. The symbol $\fmaj$ denote the {\em flag-major index} introduced by Adin and Roichman in \cite{AR}.
599: \end{proof}
600:
601: By the Principle of Inclusion-Exclusion we obtain
602: \begin{cor}\label{c1}
603: $$
604: \sum_{\{\beta \in B_n \mid \Des_B(\beta^{-1})=M\}}q^{\nmaj(\beta)}=\sum_{\{\beta \in B_n \mid \Des_B(\beta^{-1})=M\}}q^{\ell_B(\beta)}=\sum_{\{\beta \in B_n \mid \Des_B(\beta^{-1})=M\}}q^{\fmaj_B(\beta)}.$$
605: \end{cor}
606:
607:
608: The following lemma will be useful in the computation of our main result Theorem \ref{mainD}.
609:
610: \begin{lem}\label{calcoloserie}
611: Let $n \in \PP$ and $M=\{m_1,m_2,\ldots,m_t\}_<\subseteq [0,n-1]$. Then
612: \begin{eqnarray*}
613: \sum_{\{\be \in B_n \mid \Des_B(\be^{-1})\subseteq M\}} q^{\ell_D(\be)} = \begin{bmatrix} n \\ m_1, \; m_2-m_1, \ldots, n-m_t \end{bmatrix}_q \cdot \displaystyle{\prod_{i=m_1}^{n-1} (1+q^i)}.
614: \end{eqnarray*}
615: \end{lem}
616: \begin{proof}
617: Let $\be \in B(M)$. Recall that $\ell_D(\be)=\ell_B(\be)-\me_1(\be)$, and that $\ell_B(\beta)=\inv(\beta)+\sum_{\be(i)<0} |\be(i)|$. Note that $\be(i)<0$ if and only if there exists a $j$ such that $m_j+1
618: \leq |\be(i)| \leq r_j$. Therefore
619: \begin{eqnarray*}
620: \sum_{\be(i)<0} |\be(i)| & = & \sum_{i=1}^t (m_i+1) + \ldots + r_i\\
621: & = & \sum_{i=1}^t \left[(r_i - m_i)m_i + \frac{(r_i-m_i)(r_i-m_i+1)}{2}\right]\\
622: & = & \sum_{i=1}^t \frac{1}{2} (r_i-m_i)(r_i+m_i+1).
623: \end{eqnarray*}
624: Moreover $\me_1(\be) = \sum_{i=1}^t (r_i-m_i)$, and so
625: \begin{eqnarray*}
626: \ell_D(\be)& = & \inv(\be)+ \sum_{i=1}^t \frac{1}{2}(r_i-m_i)(r_i+m_i+1) - (r_i-m_i)\\
627: & = & \inv(\be) + \sum_{i=1}^t {r_i-m_i+1 \choose 2} + (r_i-m_i)(m_i-1)
628: \end{eqnarray*}
629: Hence by (\ref{crucialpoint})
630: \begin{eqnarray}
631: \sum_{\be \in B(M)} q^{\ell_{D}(\be)} & = & \sum_{r_1,\ldots,r_t} \ \sum_{\beta \in {\rm Sh}(r_1,\ldots, r_t)} q^{\inv(\be)} q^{\sum_{i=1}^t {r_i-m_i+1 \choose 2} + (r_i-m_i)} \nonumber \\
632: & = & \sum_{r_1,\ldots, r_t} \begin{bmatrix} n \\ m_1, \; r_1-m_1, \ldots, n-r_t \end{bmatrix}_q \cdot q^{\sum_{i=1}^t {r_i-m_i+1 \choose 2} + (r_i-m_i)(m_i-1) } \nonumber \\
633: & = & \begin{bmatrix} n \\ m_1, \; m_2-m_1, \ldots, n-m_t \end{bmatrix}_q \cdot \prod_{i=1}^t \sum_{r_i=m_i}^{m_{i+1}} \begin{bmatrix} m_{i+1}-m_i \\ r_i - m_i \end{bmatrix}_q \cdot q^{{r_i-m_i+1 \choose 2} + (r_i-m_i)(m_i-1)} \nonumber \\
634: & = & \begin{bmatrix} n \\ m_1, \; m_2-m_1, \ldots, n-m_t \end{bmatrix}_q \cdot \prod_{i=1}^t \prod_{j=m_i}^{m_{i+1}-1}(1+ q^j) \label{passo} \\
635: & = & \begin{bmatrix} n \\ m_1, \; m_2-m_1, \ldots, n-m_t \end{bmatrix}_q \cdot \prod_{j=m_1}^{n-1} (1+q^j) \nonumber
636: \end{eqnarray}
637: where the sum runs over $m_i\leq r_i \leq m_{i+1}$, and (\ref{passo}) is obtained by applying the $q$-binomial Theorem \ref{qbinomial} with $x=q^{(m_i-1)}$.
638: \end{proof}
639:
640: \begin{thm}\label{mainD}
641: Let $n \in \PP$ and $M=\{m_1,m_2,\ldots,m_t\}_<\subseteq [0,n-1]$. Then
642: \begin{eqnarray*}
643: \sum_{\g \in D(M)} q^{\dmaj(\g)}&=& \sum_{\g \in D(M)} q^{\ell_D(\g)}\\
644: &=& {\displaystyle \left\{\begin{array}{ll}
645: \begin{bmatrix} n \\ m_1, \; m_2-m_1, \ldots, n-m_t \end{bmatrix}_q \cdot \displaystyle{\prod_{i=1}^{n-1} (1+q^i)} \; & {\rm if} \; 0 \in M;\\
646: \begin{bmatrix} n \\ m_1, \; m_2-m_1, \ldots, n-m_t \end{bmatrix}_q \cdot \displaystyle{\prod_{i=m_1}^{n-1} (1+q^i)} \; & {\rm if} \; 0, 1 \not\in M;\\
647: \begin{bmatrix} n \\ m_1, \; m_2-m_1, \ldots, n-m_t \end{bmatrix}_q \cdot \displaystyle{\frac{\prod_{i=1}^{n-1} (1+q^i)}{[m_2]_q}} \; & {\rm if} \; 0 \not\in M, \; {\rm and} \; 1 \in M.
648: \end{array}
649: \right .}
650: \end{eqnarray*}
651: \end{thm}
652: \begin{proof} Once again the first equality follows from (\ref{crucialpoint}) and the definitions of $\dmaj$ and $\ell_D$.
653: The computation of the sum is now an easy application of Lemma \ref{calcoloserie}, together with Propositions \ref{L1}, \ref{L2}, and \ref{L3}.
654: \end{proof}
655: As corollary we obtain the desired generalization.
656: \begin{cor}\label{c2}
657: $$
658: \sum_{\{\g \in D_n \mid \Des_D(\g^{-1})=M\}}q^{\dmaj(\g)}=\sum_{\{\g \in D_n \mid \Des_D(\g^{-1})=M\}}q^{\ell_D(\g)}.
659: $$
660: \end{cor}
661:
662: \begin{rem}
663: If we replace $\Des_B$ with the usual descent set $\Des$, Corollary \ref{c1} is still valid. It easily follows from Theorem \ref{mainB} since $\Des(\be^{-1})\subseteq M$ if and only of $\Des_B(\be^{-1})\subseteq M \cup \{0\}$. Analogously, by replacing $\Des_D$ with $\Des$, Corollary \ref{c2} holds for the Coxeter group of type $D$.
664:
665: \noindent The two corollaries are not true if as descent set one choose $\NDes$ for $B_n$ and $\DDes$ for $D_n$.
666: \end{rem}
667:
668: \section{Symmetry of the joint distribution}
669:
670: In this section we find generalizations of Foata-Sch\"utzenberger Theorem \ref{FS2}, Roselle Theorem \ref{Roselle}, and Gessel Theorem \ref{Gessel}.
671: \smallskip
672:
673: The following is an easy computation.
674:
675:
676: \begin{lem}\label{lemmino} Let $n\in \PP$. Then
677: \[\sum_{u \in B^J}p^{\me_1(u)} q^{\me_1(u)+\me_2(u)}=\sum_{S\subseteq [n]}p^{|S|} q^{\sum_{i\in S}i}=\prod_{i=1}^n (1+pq^i)=(-pq;q)_n.\]
678: Moreover
679: \[\sum_{u \in D^J}p^{\me_1(u)+\epsilon(u)} q^{\me_2(u)}=\sum_{S\subseteq [n-1]} p^{|S|} q^{\sum_{i\in S}i}=\prod_{i=1}^{n-1} (1+pq^i)=(-pq;q)_{n-1}.\]
680: \end{lem}
681:
682: \begin{prop}\label{sopra}
683: The distribution of $({\rm nmaj},\ell_B)$ over $B_n$ is symmetric, namely
684: \[B_n(t,q):=\sum_{\be \in B_n} t^{\nmaj(\be)}q^{\ell_B(\be)}=\sum_{\be \in B_n} t^{\ell_B(\be)}q^{\nmaj(\be)}\]
685: \end{prop}
686: \begin{proof} Let consider the decomposition (\ref{decomp}) of $B_n$. Let $u \in B^J$ (or $D^J$) and $\s \in S_n$. Then the following equalities hold
687: \[\maj(u\s)=\maj(\s) \;\; {\rm and} \;\; \inv(u\s)=\inv(u).\]
688: Moreover
689: \[\me_1(u \s)=\me_1(u) \;\; {\rm and} \;\; \me_2(u\s)=\me_2(u).\]
690: Then from Theorem \ref{FS2} it follows
691: \begin{eqnarray*}
692: \sum_{\be \in B_{n}} t^{\ell_B(\be)}q^{\nmaj(\g)} & = & \sum_{u
693: \in B^J} \sum_{\s \in S_{n}} t^{\inv(u \sigma) + \me_1(u \s)+ \me_{2}(u\sigma
694: )} q^{\maj(u \sigma )+\me_1(u \s) + \me_{2}(u\sigma)} \\
695: & = & \sum_{u \in B^J}t^{\me_1(u)+\me_2(u)}q^{\me_1(u)+\me_2(u)}\sum_{\s \in S_n}t^{\inv(\s)} q^{maj(\s)}\\
696: & = & \sum_{u \in B^J}t^{\me_1(u)+\me_2(u)}q^{\me_1(u)+\me_2(u)}\sum_{\s \in S_n}t^{\maj(\s)} q^{\inv(\s)}\\
697: & = & \sum_{u
698: \in B^J} \sum_{\s \in S_{n}} t^{\maj(u \sigma) +\me_1(u\s)+\me_2(u\s)} q^{\inv(u \sigma)+ \me_1(u\s)+\me_2(u\s)} \\
699: & = & \sum_{\be \in B_{n}} t^{\nmaj(\be)}q^{\ell_B(\be)}.
700: \end{eqnarray*}
701: \end{proof}
702:
703: The analogous result holds for $D_n$. The proof is very similar to that of $B_n$ and is left to the reader.
704: \begin{prop}
705: The pair of statistics $(\dmaj,\ell_D)$ is symmetric, namely
706: \[D_n(t,q):=\sum_{\g \in D_n} t^{\dmaj(\g)}q^{\ell_D(\g)}=\sum_{\g \in D_n} t^{\ell_D(\g)}q^{\dmaj(\g)}.\]
707: \end{prop}
708:
709: Note that, the flag-major index and the {\em $D$-major index} \cite{BC} do not share with $\nmaj$ and $\dmaj$ this symmetric distribution property.
710: \smallskip
711:
712: The following identities are generalizations of Theorem \ref{Roselle} of Roselle to $B_n$ and $D_n$. They easily follow from the proof of Proposition \ref{sopra}, Lemma \ref{lemmino}, and from Theorem \ref{Roselle}.
713: \begin{prop}[Roselle Identities for $B_n$ and $D_n$]
714: \begin{eqnarray*}
715: \sum_{n\geq 0} B_n(t,q) \frac{u^n}{(t;t)_n (q;q)_n (-qt;qt)_{n}}&=&\frac{1}{(u;t,q)_{\infty,\infty}}, \ \ (B_0(t,q):=0);\\
716: 1+ \sum_{n\geq 1} D_n(t,q) \frac{u^n}{(t;t)_n (q;q)_n (-qt;qt)_{n-1}}&=&\frac{1}{(u;t,q)_{\infty,\infty}}.
717: \end{eqnarray*}
718: \end{prop}
719: \smallskip
720:
721: Similarly the following identities, which generalize Gessel formula, follow from the proof of Proposition \ref{sopra}, Lemma \ref{lemmino}, and Theorem \ref{Gessel}.
722: \begin{prop}[Gessel Identities for $B_n$ and $D_n$]
723: \begin{eqnarray*}
724: \sum_{n\geq 0} \frac{u^n}{[n]_q !} \frac{\sum_{\be \in B_n} t^{\nmaj(\s)} q^{\ell_B(\be)} p^{\ndes(\be)}}{(-tqp;tq)_{n} (t;q)_{n+1}}&=&\sum_{k \geq 0} p^{k} e[u]_q e[tu]_q\cdots e[t^k u]_q;\\
725: \frac{1}{1-t} +\sum_{n\geq 1} \frac{u^n}{[n]_q !} \frac{\sum_{\g \in D_n} t^{\dmaj(\g)} q^{\ell_D(\g)} p^{\ddes(\g)}}{(-tqp;tq)_{n-1}(t;q)_{n+1}}&=&\sum_{k \geq 0} p^{k} e[u]_q e[tu]_q\cdots e[t^k u]_q.
726: \end{eqnarray*}
727: \end{prop}
728:
729: \section{Concluding remarks}
730:
731: As we mentioned along the paper, there exists another family of statistics, the {\em flag-statistics}, defined on Coxeter groups of type $B$, $D$ (see \cite{AR} and \cite{BC}), and more generally on complex reflection groups \cite{BaB}. Several generating functions involving flag-statistics have already been computed. In particular, we refer to the series of papers of Foata and Han \cite{FH2,FH3,FH5}, for a complete overview on the argument.
732:
733: We remark that among the series computed, none involve a combination of flag-statistics and length. This is why we conclude the paper with the following interesting proposal.
734:
735: \begin{problem}
736: What kind of identities, generalizing the ones of Roselle and Gessel, might be obtained by using flag-statistics ?
737: \end{problem}
738:
739:
740:
741:
742:
743: \begin{thebibliography}{xx}
744:
745: \bibitem{AR}
746: R. M. Adin and Y. Roichman, {\em The Flag Major Index and Group Actions on Polynomial Rings}, Europ. J. Combinatorics, 22 (2001), 431-446.
747:
748: \bibitem{ABR1}
749: R. M. Adin, F. Brenti and Y. Roichman, {\em Descent Numbers and Major Indices for the Hyperoctahedral Group}, Adv. in Appl. Math., 27 (2001), 210-224.
750:
751: \bibitem{ABR2}
752: R. M. Adin, F. Brenti and Y. Roichman, {\em Equi-distribution over descent classes of the Hyperoctahedral Group}, J. Combin. Theory Ser. A, 113 (2006), 917-933.
753:
754: \bibitem{An}
755: G.E. Andrews, {\em The Theory of Partitions}, The Encyclopedia of Mathematics and Its Applications Series, Addison-Wesley Pub. Co., NY, 300 pp. (1976).
756:
757: \bibitem{BaB}
758: E. Bagno and R. Biagioli, {\em Colored-Descent Representations of Complex Reflection Groups $G(r,p,n)$}, Israel J. Math., to appear.
759:
760: \bibitem{B}
761: R. Biagioli, {\em Major and descent statistics for the even-signed permutation group},
762: Adv. in Appl. Math., 31 (2003), 163-179.
763:
764: \bibitem{BC} R. Biagioli and F. Caselli, \emph{Invariant algebras and major indices
765: for classical Weyl groups}, Proc. London Math. Soc., 88 (2004), 603-631.
766:
767: \bibitem{BB}
768: A. Bj\"{o}rner and F. Brenti, {\em Combinatorics of Coxeter Groups},
769: Grad. Texts in Math. 231, Springer-Verlag, Berlin, 2005.
770:
771: \bibitem{F}
772: D. Foata, {\em On the Netto Inversion Number of a Sequence}, Proc. Amer. Math. Soc., 19 (1968), 236-240.
773:
774: \bibitem{FH2}
775: D. Foata and G.N. Han, {\em Signed words and permutations, II: the Euler-Mahonian polynomials}, Electronic J. Combinatorics, 11(2), R22, (2005).
776:
777: \bibitem{FH3}
778: D. Foata and G.N. Han, {\em Signed words and permutations, III: the MacMahon Verfahren}, S\'em. Lothar. Combin., vol. 54a, (2006), 20 pages (The Viennot Festschrift)
779:
780: \bibitem{FH5}
781: D. Foata and G.N. Han, {\em Signed words and permutations, V: a sextuple distribution}, preprint.
782:
783: \bibitem{FS}
784: D. Foata and M.P. Sch\"utzenberger, {\em Major index and inversion number of permutations}, Math. Nachr., 83 (1978), 143-159.
785:
786: \bibitem{GG}
787: A. Garsia and I. Gessel, {\em Permutation statistics and partitions}, Adv. Math., 31 (1979), 288-305.
788:
789: \bibitem{Ges}
790: I. Gessel, {\em Generating functions and enumeration of sequences}, M.I.T. doctoral thesis, 1977.
791:
792: \bibitem{HNT}
793: F. Hivert, J.C. Novelli, and J.Y. Thibon, {\em Multivariate generalizations of the Foata-Sch\"utzenberger equidistribution}, Discr. Math. Theo. Comp. Sci. Proc. of Fourth Colloquium on Mathematics and Computer Science, Nancy, 2006.
794:
795: %\bibitem{Hum}
796: %J.E. Humphreys, {\em Reflection Groups and Coxeter Groups},
797: %Cambridge Stud. Adv. Math., no. 29, Cambridge Univ. Press,
798: %Cambridge, 1990.
799:
800: \bibitem{MM}
801: P.A. MacMahon, {\em Combinatory Analysis}, vol. 1, Cambridge Univ. Press, London, 1915.
802:
803: \bibitem{Raw}
804: D. Rawlings, {\em The combinatorics of certain products},
805: Proc. Amer. Math. Soc., 89 (1983), 560-562.
806:
807: \bibitem{Ros}
808: D.P. Roselle, {\em Coefficients associated with the expansion of certain products},
809: Proc. Amer. Math. Soc., 45 (1974), 144-150.
810:
811: \bibitem{StaEC1}
812: R.P. Stanley, {\em Enumerative Combinatorics}, vol. 1, Cambridge Stud. Adv. Math., no. 49, Cambridge Univ. Press, Cambridge, 1997.
813:
814: \end{thebibliography}
815:
816:
817:
818:
819:
820:
821:
822:
823: \end{document}