1: % ----------------------------------------------------------------

2: % AMS-LaTeX Paper ************************************************

3: % **** -----------------------------------------------------------

4: \documentclass{amsart}

5: \usepackage{epsf}

6: %[active]{srcltx} % SRC Specials: DVI [Inverse] Search

7: % ----------------------------------------------------------------

8: \vfuzz2pt % Don't report over-full v-boxes if over-edge is small

9: \hfuzz2pt % Don't report over-full h-boxes if over-edge is small

10: % THEOREMS -------------------------------------------------------

11: 

12: \newtheorem{lemma}{Lemma}[section]

13: \newtheorem{propos}[lemma]{Proposition}

14: \newtheorem{example}[lemma]{Example}

15: \newtheorem{theorem}[lemma]{Theorem}

16: \newtheorem{conjec}[lemma]{Conjecture}

17: \newtheorem{corol}[lemma]{Corollary}

18: \theoremstyle{definition}

19: \newtheorem{defin}[lemma]{Definition}

20: \theoremstyle{remark}

21: \newtheorem{remark}[lemma]{Remark}

22: % \numberwithin{equation}{section}

23: 

24: 

25: 

26: 

27: % MATH -----------------------------------------------------------

28: 

29: \newcommand{\CC}{\hbox{{$\sl C$}}}

30: 

31: 

32: \newcommand{\cu}{\mathfrak{u}}  % lie algebras

33: \newcommand{\cg}{\mathfrak{g}}

34: \newcommand{\cv}{\mathfrak{v}}

35: 

36: \newcommand{\Q}{\mathbb{Q}}  % open font

37: \newcommand{\R}{\mathbb{R}}

38: \newcommand{\F}{\mathbb{F}}

39: \newcommand{\Z}{\mathbb{Z}}

40: \newcommand{\quat}{\mathbb{H}}

41: 

42: 

43: \newcommand{\del}{\partial}

44: \newcommand{\ev}{{\rm ev}}

45: \newcommand{\image}{{\rm image}}

46: 

47: \newcommand{\isom}{{\cong}}

48: \newcommand{\eps}{{\epsilon}}

49: \newcommand{\tens}{\mathop{\otimes}}

50: \newcommand{\la}{{\triangleright}}

51: \newcommand{\ra}{{\triangleleft}}

52: \newcommand{\Lin}{{\rm Lin}}

53: 

54: \newcommand{\Ad}{{\rm Ad}}

55: \newcommand{\id}{{\rm id}}

56: \newcommand{\<}{\langle}

57: \renewcommand{\>}{\rangle}

58: \newcommand{\End}{{\rm End}}

59: \newcommand{\from}{{\longleftarrow}}

60: \newcommand{\note[1]}{}

61: 

62: 

63: %\newcommand{\proof}{{\bf Proof\ }}

64: \newcommand{\eproof}{$\quad \diamond$\bigskip}

65: 

66: \newcommand{\und}[1]{{\underline {#1}}}

67: \newcommand{\eqn}[2]{\begin{equation}#2\label{#1}\end{equation}}

68: \newcommand{\cmath}[1]{\[\begin{array}{c} #1 \end{array}\]}

69: %\newcommand{\align}[1]{\begin{eqnarray*}#1\end{eqnarray*}}

70: % e.g.\align{LHS &&=RHS1\\ &&=RHS2\\ &&=RHS3....}

71: 

72: 

73: \begin{document}

74: 

75:  %\vskip 3cm

76:  %\hsize 15cm

77:  %\vsize 26cm

78:  %\textheight 26cm

79:  %\textwidth 16cm \topmargin -.4in

80:  %\headheight .5in

81:  %\headsep 0in

82:  %\oddsidemargin .5cm

83:  %\evensidemargin -.2in

84:  %\topskip -20pt

85: 

86: 

87: 

88: 

89: 

90: %\pagestyle{article}

91: 

92: 

93: 

94: \title{\rm\large BRAIDED LINE AND COUNTING FIXED POINTS OF $GL(d,\F_q)$}

95: \author{P.J. Cameron $+$ S. Majid}%

96: 

97: 

98: 

99: \address{School of Mathematical Sciences\\

100: Queen Mary, University of London\\ 327 Mile End Rd,  London E1

101: 4NS, UK}

102: %\email{}%

103: 

104: 

105: \thanks{S.M. is a Royal Society University Research Fellow}%

106: %\subjclass{}%

107: %\keywords{}%

108: 

109: %\date{December 2001}%

110: %\dedicatory{}%

111: %\commby{}%

112: 

113: \maketitle

114: % ----------------------------------------------------------------

115: 

116: \begin{abstract}We interpret a recent formula for counting orbits of

117: $GL(d,\F_q)$ in terms of counting fixed points as addition in the

118: affine braided line. The theory of such braided groups (or Hopf

119: algebras in braided categories) allows us to obtain the inverse

120: relationship, which turns out to be the same formula but with $q$

121: and $q^{-1}$ interchanged (a perfect duality between counting

122: orbits and counting fixed points). In particular, the probability

123: that an element of $GL(d,\F_q)$ has no fixed points is found to be

124: the truncated q-exponential

125: \[ {}_de_q^{-{1\over q-1}}=\sum_{i=0}^d {(-1)^i\over

126: (q^i-1)\cdots(q-1)}.\]

127: \end{abstract}

128: 

129: %\renewcommand{\thesection}{\Roman{section}}

130: %\renewcommand{\thesubsection}{\thesection.\Alph{subsection}}

131: 

132: 

133: \section{Introduction}

134: 

135: \bigskip

136: 

137: Let $\Sigma=\{1,\cdots,d\}$. It is known \cite{Bos} that one may

138: write the number $F_j$ of orbits of $G\subseteq S_d$ acting on the

139: set $\Sigma_j$ of distinct $j$-tuples in terms of the

140: probabilities $p_i$ that an element of $G$ has $i$ fixed points.

141: Writing generating functions

142: \[ F=\sum_{j=0}^d F_j {x^j\over j!},\quad p(x)=\sum_{i=0}^d p_i

143: x^i,\] the two are related by

144: \[ F(x)=p(x+1).\]

145: For example, when $G=S_d$, it is evident that $F_j=1$ for all $j$

146: and hence

147: \[ p_0={}_de^{-1}\equiv\sum_{i=0}^d {(-1)^i\over i!}.\]

148: 

149: 

150: Recently, one of us\cite{Cam:unp} has considered a similar problem

151: for $G\subseteq GL(d,\F_q)$ where $\F_q$ denotes the field of

152: prime-power order $q$. Let

153: \[ \Sigma_j=\{(a_1,\cdots,a_j),\quad a_i\in \F_q^d,\quad a_i\

154: {\rm linearly\ indept.}\},\quad L_j=\#\{{\rm Orbits\ of}\ G\ {\rm

155: in}\ \Sigma_j\},\] and let $P_i$ denote the probability that an

156: element of $G$ pointwise fixes exactly a subspace of $\F_q^d$ of

157: dimension $i$. Then similarly by the orbit-counting lemma one has

158:  \eqn{LP}{ L_j={1\over |G|}\sum_{g\in G}\#\{g-{\rm fixed\ points\

159: in}\ \Sigma_j\}=\sum_{i=j}^d

160: P_i(q^i-1)(q^i-q)\cdots(q^i-q^{j-1}).} The generating function

161: \eqn{Lx}{ L(x)=\sum_{j=0}^d{L_j x^j\over(q^j-1)\cdots (q-1)}} is

162: likewise written in terms of the $P_i$ and $q$-binomial

163: coefficients.  The latter are well-known (the Gauss formula) to

164: count the number of subspaces of a given dimension in an

165: $\F_q$-vector space. The $q$-factorial denominators in such

166: expressions are understood as $1$ for $j=0$.

167: \bigskip

168: 

169: On the other hand, for applications, we would really like the

170: inverse relationship $P_i$ in terms of $L_j$. In this note we

171: provide this inverse in a natural way. Thus:

172: 

173: \begin{propos} With the notations above the probabilities $P_i$

174: of an $i$-dimensional fixed subspace are given by

175: \[ P_j={q^{-{j(j-1)\over 2}}\over (q^j-1)\cdots (q-1)}\sum_{i=j}^d

176: {L_i (-1)^{i-j}\over (q^i-q^j)\cdots (q^{j+1}-q^j)}.\]

177: \end{propos}

178: 

179: For example, it is evident that if $G=GL(d,\F_q)$ then $L_j=1$ for

180: all $j$. Hence we find:

181: 

182: \begin{corol} For $G=GL(d,\F_q)$ the probability of no fixed

183: points is

184: \[ P_0=\sum_{i=0}^d {(-1)^i\over (q^i-1)\cdots (q-1)}\equiv

185: {}_de_q^{-{1\over (q-1)}}.\]

186: \end{corol}

187: 

188: Our standard notation for the (truncated) q-exponential here is

189: such that ${}_de^x_q\to {}_de^x$ as $q\to 1$, i.e. it is defined

190: with the factorial of $q$-integers $[i]_q={q^i-1\over q-1}$. In

191: our case $q\ge 2$ and these $(q-1)$ factors cancel but the

192: notation reminds us of a divergence if we were to try include the

193: finite group case as $q=1$.

194: 

195: There are of course many ways to invert a linear relationship. We

196: mention that one may also, for example, use the method of M\"obius

197: inversion with respect to a suitable function, see \cite[p.

198: 127]{Sta:enu}. On the other hand we will obtain the above

199: inversion in a way that is interesting because it is precisely

200: analogous to the computation for the $S_d$ case based on

201: generating functions. The only difference is that that one should

202: consider the generating function dummy variable $x$ for the

203: purposes of addition and subtraction as a braided one, i.e. the

204: polynomials $k[x]$ as a Hopf algebra, but in a braided category

205: where $x$ has braid statistics $q$ (the affine `braided line').

206: (The idea here is similar to that of a superspace but instead of a

207: `fermionic' factor $-1$ there is a factor $q$ when $x$ is

208: interchanged with another independent $x$.) The systematic passage

209: from permutation groups to linear groups over finite fields by

210: means of switching on a braiding would appear to be an interesting

211: direction suggested by the present work.

212: 

213: The required methods of Hopf algebras in braided categories or

214: `braided groups' have been introduced some years ago\cite{Ma:bg},

215: precisely as the the basic structure underlying quantum groups and

216: $q$-analysis, \cite{Ma:introm}\cite{Ma:book}. We recall what we

217: need in a self-contained preliminary Section~2. Working with $x$

218: as a braided variable we relate, in Section~3, the signed

219: generating function $\bar L(x)=L(-x)$ and $P(x)=\sum_{i=0}^dP_i

220: x^i$ by \eqn{barLP}{ \bar L(x)=P(y-_q x)|_{y=1},} where $-_q$ on

221: the right is subtraction under the additive braided group

222: structure of the braided line. We can then immediately invert the

223: relationship as \eqn{PbarL}{ P(x)=\bar L(y-_{q^{-1}} x)|_{y=1},}

224: where $-_{q^{-1}}$ is subtraction under the opposite additive

225: braided group structure (it is not commutative). This gives the

226: above inversion formula, essentially without computation. It also

227: exhibits a novel duality phenomenon whereby the coefficients of

228: $P(x)$ are given in terms of those of $\bar L(x)$ by exactly the

229: same formula as vice-versa, but with $q$ and $q^{-1}$

230: interchanged. This is discussed in Section~4.

231: 

232: 

233: \section{Preliminaries: the affine braided line}

234: 

235: We briefly recall the definition of braided groups and the braided

236: line. Once these are understood, the computation itself in the

237: next section will be immediate.

238: 

239: The formal definition of a braided category is in

240: \cite{JoyStr:bra}. It consists of a category $\CC$, a functor

241: $\tens:\CC\times\CC\to \CC$ and natural transformations

242: $\Psi:\tens\to \tens^{\rm op}$ and $\Phi:(\ \tens\ )\tens\to

243: \tens(\ \tens\ )$ obey respectively hexagon and pentagon

244: identities. In addition there is a trivial object $\und 1$ and

245: associated functorial isomorophisms of identity with respect to

246: $\tens$. A braided group\cite{Ma:bg}\cite{Ma:introm} is first of

247: all a unital algebra $B$ in such a category, defined in the

248: obvious way except that the identity must be viewed as a morphism

249: $\eta:\und 1\to B$. In addition, we require a coalgebra structure

250: $\Delta:B\to B\tens B$ (the coproduct) and $\eps:B\to \und 1$ (the

251: counit) obeying axioms which are the same as for an algebra but

252: with arrows reversed. These $\Delta,\eps$ are required to be

253: algebra homomorphisms, where $B\tens B$ has the braided tensor

254: product algebra structure. Its product $(B\tens B)\tens (B\tens

255: B)\to B\tens B$ consists of $\Psi$ to interchange the middle

256: factors followed by products $\cdot$ in $B$, as well as $\Phi$ as

257: needed to order brackets. Finally, there is an antipode or

258: `linearized inverse' morphism $S:B\to B$ obeying \eqn{antipode}{

259: \cdot\circ (\id\tens S)\circ

260: \Delta=\cdot\circ(S\tens\id)\circ\Delta=\eta\circ\eps.}

261: 

262: \begin{figure}

263: \[{\rm (a)}\quad \epsfbox{bialg.eps},\quad {\rm (b)}\quad

264: \epsfbox{antB.eps}\]

265: \caption{Axioms of a braided group: (a) associativity of the

266: product (similarly for $\Delta$) and the braided homomorphism

267: axiom, (b) the antipode axioms}

268: \end{figure}

269: 

270: The theory of braided groups can be developed similarly to that of

271: quantum groups or Hopf algebras. A useful (but not necessary) tool

272: is to use a diagrammatic notation as follows. One omits $\Phi$ and

273: brackets. One denotes $\tens$ by juxtaposition, the braiding by

274: $\Psi=\epsfbox{braid.eps}$ and its inverse by

275: $\Psi^{-1}=\epsfbox{braidinv.eps}$. Algebraic operations are then

276: drawn as `flow charts' flowing from the top of the page downwards.

277: General morphisms such as $S$ are drawn as nodes. All of this is

278: as for any braided category. Finally, we denote the product and

279: coproduct of $B$ by $\Delta=\epsfbox{deltafrag.eps}$ and

280: $\cdot=\epsfbox{prodfrag.eps}$. The principal axioms of a braided

281: group are shown in this notation in Figure~1. One of the key

282: lemmas that we will need is:

283: \begin{lemma}\cite{Ma:introm} $S$ is braided-antimultiplicative

284: and anticomultiplicative, \[ S\circ\cdot=\cdot\circ\Psi\circ

285: (S\tens S),\quad \Delta\circ S=(S\tens S)\circ\Psi\circ \Delta.\]

286: \end{lemma}

287: 

288: We will also need that if $B$ is a braided group in $\CC$ with $S$

289: invertible then

290: \eqn{deltaop}{\Delta^{op}=\Psi^{-1}\circ\Delta,\quad

291: S^{op}=S^{-1}}

292: with the original algebra and counit provide

293: another braided group structure $B^{\rm op}$ in the category

294: $\CC^{\rm op}$ defined as $\CC$ but with $\Psi,\Psi^{-1}$

295: interchanged (i.e. with reversed braid crossings).

296: 

297: In our elementary application we work in the category $\CC_q$ of

298: $\Z$-graded vector spaces $V,W$, etc. with its usual $\tens$ and

299:  (trivial) $\Phi$, but with

300: \eqn{Psiq}{\Psi_{V,W}(v\tens w)=w\tens v\ q^{{\rm deg}(v){\rm

301: deg(w)}}}extended linearly. Morphisms are degree preserving maps.

302: We work over any field $k$. The unit object is $\und 1=k$.

303: 

304: The algebra of polynomials $B=k[x]$ lives in this category (all

305: its structure maps are morphisms) with ${\rm deg}(x)=1$. We make

306: it into a braided group (the `affine braided line') with

307: \[ \Delta x=x\tens 1+1\tens x,\quad \eps x=0,\quad Sx=-x.\]

308: This looks like the usual additive group Hopf algebra structure

309: on an affine line but $\Delta$ extends differently as a

310: homomorphism to the braided tensor product. The braided tensor

311: algebra $B\tens B$ is generated by $y\equiv x\tens 1$ and

312: $x\equiv 1\tens x$ with the relations $xy=qyx$ (the so-called

313: `quantum plane'). Then

314: \[ \Delta f(x)=f(y+x),\quad \eps f(x)=f(0),\]

315: which exhibits $\Delta$ as the operation of addition with respect

316: to braided variables where independent $x,y$ $q$-commute (to

317: emphasise this one may write $f(y+_qx)$, which is the notation

318: indicated in (\ref{barLP}-(\ref{PbarL})). Working directly from

319: the braided anti/homomorphism properties of $S$ and $\Delta$ one

320: also has explicitly \eqn{DeltaSx}{ \Delta

321: x^i=\sum_{j=0}^i\left[{i\atop j}\right]_q x^j\tens x^{i-j},\quad

322: S(x^i)=q^{i(i-1)\over 2}(-x)^i,} where \eqn{qbinom}{\left[{i\atop

323: j}\right]_q ={(q^i-1)\cdots (q^{i-j+1}-1)\over(q^j-1)\cdots

324: (q-1)}={[i]_q!\over [j]_q![i-j]_q!}.} Clearly, the elementary

325: properties of braided groups mentioned above can easily be

326: verified directly from these definitions and boil down to a number

327: of useful identities for $q$-binomial coefficients, not all of

328: them obvious.

329: 

330: The braided line can be viewed as the `correct' setting for

331: q-analysis in one variable including Jackson's q-derivative and

332: q-integration\cite{Jac:int}, q-exponentials, as well as a new

333: theory of q-Fourier transforms\cite{Ma:fre}\cite{KemMa:alg}. Such

334: methods in particular solved the long-standing problem of

335: extending $q$-differential calculus to the multivariable case on

336: affine $q$-planes (these turn out to be additive braided groups).

337: 

338: 

339: \section{Inverting the map from $P[x]$ to $L(x)$}

340: 

341: As in \cite{Cam:unp} we write (\ref{LP}) as

342: \[ L(x)\equiv \sum_{j=0}^d {L_j\over (q^j-1)\cdots (q-1)}

343: x^j=\sum_{j=0}^d\sum_{i=j}^d

344: P_i{(q^i-1)\cdots (q^{i-j+1}-1)\over (q^j-1)\cdots

345: (q-1)}q^{j(j-1)\over 2}x^j\] or, changing the order of summation

346: and using the q-binomial notation,

347: \eqn{LPbinom}{L(x)=\sum_{i=0}^dP_i\sum_{j=0}^i\left[{i\atop

348: j}\right]_q q^{j(j-1)\over 2}x^j .}

349: 

350: Now consider the algebra homomorphism $\phi:k[x]\to k$ which

351: `evaluates at 1', \eqn{phi}{ \phi(x^m)=1,\quad \forall m.} Using

352: this and the structure of the braided line recalled above, we

353: clearly have \begin{eqnarray} \bar

354: L(x)&=&\sum_{i=0}^dP_i\sum_{j=0}^i\left[{i\atop j}\right]_q

355: q^{j(j-1)\over 2}(-x)^j\label{barLxP}\\ &=& (\phi\tens

356: S)\sum_{i=0}^dP_i\sum_{j=0}^i\left[{i\atop j}\right]_q

357: x^{i-j}\tens x^j=(\phi\tens S)\circ\Delta P(x)\nonumber

358: \end{eqnarray} which is the precise meaning of (\ref{barLP}).

359: 

360: \begin{lemma} The linear operator $T=(\phi\tens S)\circ\Delta$

361: which maps $P(x)$ to $\bar L(x)$

362: has inverse \[ T^{-1}=(\phi\circ S\tens\id)\circ\Delta\circ

363: S^{-1}.\]

364: \end{lemma}

365: \begin{figure}

366: \[\epsfbox{fixed.eps}\]

367: \caption{(a) The operator $T$ sending $P$ to $L$ and (b) its

368: inverse as proven in (c)}

369: \end{figure}

370: \proof The proof is shown in Figure~2 in the diagrammatic

371: notation. The composition $T^{-1}\circ T$ is depicted in the first

372: line in Figure~2(c). Here $T^{-1}$ is linear so we can move the

373: output of $\phi$ in $T$ to the end of the computation as shown

374: (remember that juxtaposition at the bottom of the line denotes

375: tensor product). We cancel $S^{-1}\circ S$, then use

376: coassociativity of $\Delta$. Finally we use that $\phi$ is an

377: algebra homomorphism and cancel an `antipode loop' using the axiom

378: of an antipode. The resulting counit combines with $\Delta$ to the

379: identity (the counity axiom). Note that $\phi$ is not a morphism

380: in $\CC_q$ but we do not use this (we apply $\phi$ only at the end

381: of each diagrammatic expression). Similarly for the lower line of

382: Figure~2(c). Since these techniques may be unfamiliar, we also

383: write the proof more conventionally. Thus,

384: \begin{eqnarray*} T^{-1}\circ T &=& (\phi\circ

385: S\tens\id)\circ\Delta\circ (\phi\tens\Delta)=(\phi\tens\phi\circ

386: S\tens\id)\circ(\id\tens\Delta)\circ\Delta\\

387: &=& (\phi\tens\phi\circ S\tens\id)\circ(\Delta\tens\id)\circ\Delta

388: =\left(\phi\circ\cdot\circ(\id\tens

389: S)\circ\Delta\tens\id\right)\circ\Delta\\

390: & =&\phi(1)(\eps\tens\id)\circ\Delta=\id\end{eqnarray*} and

391: similarly for $T\circ T^{-1}$. These proofs amount to a series of

392: q-identities among q-binomial coefficients by writing out $\Delta,

393: S$ explicitly on monomials $x^m$. \eproof

394: 

395: It remains only to compute $T^{-1}$ explicitly. We write $\bar

396: L=\sum_{i=0}^d \bar L_i x^i$ (without any factorial factors). Then

397: \[ P(x)=T^{-1}(\bar L(x))=(\phi\circ S\tens\id)

398: \sum_{i=0}^d \bar L_i (-1)^iq^{-i(i-1)\over 2}\sum_{j=0}^i

399: \left[{i\atop j}\right]_q x^{i-j}\tens x^j\]

400: \[=\sum_{i=0}^d \bar L_i (-1)^iq^{-i(i-1)\over 2}\sum_{j=0}^i

401: (-1)^{i-j}q^{(i-j)(i-j-1)\over 2}\left[{i\atop j}\right]_q x^j

402: \]

403: giving \eqn{PbarLx}{ P(x)=\sum_{i=0}^d \bar L_i \sum_{j=0}^i

404: \left[{i\atop j}\right]_q q^{-j(j-1)\over 2}q^{-j(i-j)}(-x)^j.}

405: 

406: We can then put in the form of $\bar L_i$ in terms of $L_i$ and

407: reorder the summation to extract the $P_i$. Thus

408: \[ P(x)=\sum_{j=0}^d x^jq^{-j(j-1)\over

409: 2}\sum_{i=j}^d L_i{(-1)^{i-j} q^{-j(i-j)}\over (q^j-1)\cdots

410: (q-1)(q^{i-j}-1)\cdots (q-1)}\]

411: \[=\sum_{j=0}^d x^j{q^{-j(j-1)\over

412: 2} \over (q^j-1)\cdots (q-1)}\sum_{i=j}^d L_i{(-1)^{i-j} \over

413: (q^i-q^j)\cdots (q^{j+1}-q^j)}\] or $P_i$ as stated in

414: Proposition~1.1. This completes our proof of that.

415: 

416: 

417: 

418: \section{Duality under $q\leftrightarrow q^{-1}$}

419: 

420: Using one of the basic results from braided group theory, namely

421: the second part of Lemma~2.1, we can also write \eqn{Tinvdua}{

422: T^{-1}=(\phi\tens S^{-1})\circ\Psi^{-1}\circ\Delta,} which is the

423: other expression shown in Figure~2(b) and the precise meaning of

424: (\ref{PbarL}). This computes more directly as (\ref{PbarLx}), with

425: $q^{-j(i-j)}$ coming from the braiding $\Psi^{-1}(x^j\tens

426: x^{i-j})$ and $q^{-j(j-1)\over 2}$ from $S^{-1}$.

427: 

428: \begin{corol} The operator $T^{-1}$ is exactly the same as the operator $T$

429: but with $q$ replaced by $q^{-1}$.

430: \end{corol}

431: \proof We see that $T^{-1}$ has just the same form as $T$ but with

432: $\Delta^{\rm op}$ and $S^{\rm op}$ in place of $\Delta, S$. This

433: is the opposite braided group structure on $k[x]$, namely the same

434: form on degree 1 but developed as a braided group in the category

435: $\CC_q^{\rm op}=\CC_{q^{-1}}$. One may also see this explicitly by

436: observing that

437: \[\left[{i\atop j}\right]_{q^{-1}}=q^{-j(i-j)}\left[{i\atop

438: j}\right]_q\] so that (\ref{PbarLx}) can be written as

439: \eqn{PbarLx2}{P(x)=\sum_{i=0}^d \bar L_i \sum_{j=0}^i

440: (-1)^j\left[{i\atop j}\right]_{q^{-1}}  q^{-j(j-1)\over 2} (-x)^j}

441: which has the same form as our original (\ref{barLxP}). \eproof

442: 

443: Finally, we mention another way to obtain the above duality using

444: braided group theory, this time in the form of `q-analysis'.

445: Namely, the displacement of a function by $1$ can be written as a

446: q-exponential of q-derivatives (the braided Taylor's

447: theorem\cite{Ma:fre}). In the present one-variable case it is

448: easily verified on monomials and takes the form

449: \[ (\phi\tens\id)\Delta =e_q^{\del_q}

450: =\sum_{i=0}^\infty {(\del_q)^i\over [i]_q!}\]

451: acting on polynomials (the sum is only finite).  Here

452: \[ \del_qf(x)={f(qx)-f(x)\over x(q-1)}\]

453: is Jackson's $q$-derivative\cite{Jac:int}. (Also in our case one

454: can cancel the $(q-1)$ factors here and in $[i]_q$! if one

455: prefers.) Moreover,

456: \[ e_q^Ae_{q^{-1}}^{-A}=1\]

457: for any nilpotent operator $A$ as one may see by

458: q-differentiating\cite[Lem. 3.2.2]{Ma:book}. Hence

459: \[ T=S\circ e_q^{\del_q}=e_q^{-\del_{q^{-1}}}\circ S.\]

460: The second equality here is easily verified on monomials. Then

461: \[ T^{-1}=e_{q^{-1}}^{-\del_q}\circ S^{-1}=S^{-1}\circ

462: e_{q^{-1}}^{\del_{q^{-1}}},\]

463: which has the same form with $q$ and $q^{-1}$ interchanged.

464: 

465: \baselineskip 14pt

466: %\bibliographystyle{unsrt}

467: %\bibliography{biblio}

468: %\end{document}

469: 

470: \begin{thebibliography}{10}

471: 

472: \bibitem{Bos}

473: N. Boston, W. Dabrowski, T. Foguel, P.J. Gies, J.~Leavitt,

474: D.T.~Ose and D.A. Jackson.

475: \newblock The proportion of fixed-point-free elements of a

476: transitive permutation group.

477: \newblock {\em Commun. Algebra} 21:3259--3275, 1993.

478: 

479: \bibitem{Cam:unp}

480: P.J.~Cameron.

481: \newblock Seminar, November 2000 (unpublished).

482: 

483: \bibitem{Sta:enu}

484: R.P.~Stanley.

485: \newblock {\em Enumerative Combinatorics, Vol 1}.

486: \newblock Cambridge University Press, 1997.

487: 

488: \bibitem{Ma:bg}

489: S.~Majid.

490: \newblock Braided groups.

491: \newblock {\em J. Pure and Applied Algebra}, 86:187--221, 1993.

492: 

493:  \bibitem{Ma:introm}

494: S.~Majid.

495: \newblock Algebras and {H}opf algebras in braided categories.

496: \newblock volume 158 of {\em Lec. Notes in Pure and

497: Appl. Math}, pages 55--105. Marcel Dekker, 1994.

498: 

499: \bibitem{Ma:book}

500: S.~Majid.

501: \newblock {\em Foundations of Quantum Group Theory}.

502: \newblock Cambridge Univeristy Press, 1995.

503: 

504: \bibitem{JoyStr:bra}

505: A.~Joyal and R.~Street.

506: \newblock Braided monoidal categories.

507: \newblock Mathematics Reports 86008, Macquarie University, 1986.

508: 

509: \bibitem{Jac:int}

510: F.H. Jackson.

511: \newblock {$q$}-{I}ntegration.

512: \newblock {\em Proc. Durham Phil. Soc.}, 7:182--189, 1927.

513: 

514: \bibitem{Ma:fre}

515: S.~Majid.

516: \newblock Free braided differential calculus, braided binomial theorem and the

517:   braided exponential map.

518: \newblock {\em J. Math. Phys.}, 34:4843--4856, 1993.

519: 

520: \bibitem{KemMa:alg}

521: A.~Kempf and S.~Majid.

522: \newblock Algebraic $q$-integration and {F}ourier theory on quantum and braided

523:   spaces.

524: \newblock {\em J. Math. Phys.}, 35:6802--6837, 1994.

525: \end{thebibliography}

526: 

527: 

528: 

529: \end{document}

530: