1: \documentclass[12pt]{article}

2: \usepackage{amsmath}

3: \usepackage{amsthm}

4: \usepackage{amsfonts}

5: \usepackage{amssymb}

6: \usepackage[all]{xypic}

7: 

8: \makeatletter

9: 

10: \newdimen\p@renwd

11: \newdimen\tmp

12: \setbox0=\hbox{B} \p@renwd=\wd0 % width of the big left (

13: \def\bm#1{\begingroup \m@th

14:   \setbox\z@\vbox{\def\cr{\crcr\noalign{\kern2\p@\global\let\cr\endline}}%

15:     \ialign{$##$\hfil\kern2\p@\kern\p@renwd&\thinspace\hfil$##$\hfil

16:       &&\quad\hfil$##$\hfil\crcr

17:       \omit\strut\hfil\crcr\noalign{\kern-\baselineskip}%

18:       #1\crcr\omit\strut\cr}}%

19:   \setbox\tw@\vbox{\unvcopy\z@\global\setbox\@ne\lastbox}%

20:   \setbox\tw@\hbox{\unhbox\@ne\unskip\global\setbox\@ne\lastbox}%

21:   \setbox\tw@\hbox{$\kern\wd\@ne\kern-\p@renwd\left(\kern-\wd\@ne

22:     \global\setbox\@ne\vbox{\box\@ne\kern2\p@}%

23:     \vcenter{\tmp=-\ht\@ne\unvbox\z@\kern-\baselineskip\kern\tmp}\,\right)$}%

24:   \null\;\vbox{\kern\ht\@ne\box\tw@}\endgroup}

25: 

26: 

27: 

28: \makeatother

29: 

30: \newcounter{appendix}

31: \setcounter{appendix}{0}

32: \def\appendix#1{\stepcounter{appendix}\goodbreak\vspace{12pt plus 3pt}

33: {\begin{center} 

34: \Huge Appendix \Alph{appendix}\par\vspace{8pt plus 2pt}

35: \bfseries\Large{#1}\end{center}}

36: \addcontentsline{toc}{section}{Appendix \Alph{appendix}. #1}\setcounter{subsection}{0}%

37: \vspace{8pt plus 2pt}

38: \nobreak}

39: 

40: \let\phi\varphi

41: \let\epsilon\varepsilon

42: \def\C{\mathbb C}

43: \def\CP{\mathbb {CP}}

44: \def\Z{\mathbb Z}

45: \def\N{\mathbb N}

46: \def\g{\mathfrak g}

47: \def\e{e}

48: \def\F{\mathcal F}

49: \def\M{\mathcal M}

50: \def\E{\mathcal E}

51: \def\slg{\mathfrak{sl}}

52: 

53: \def\mmod{\ \mathrm{mod}\ }

54: \let\wt\widetilde

55: \let\wh\widehat

56: \def\Im{\mathop{\mathrm{Im}}}

57: \def\sn{\mathop{\mathrm{sn}}}

58: \def\dn{\mathop{\mathrm{dn}}}

59: \def\cn{\mathop{\mathrm{cn}}}

60: \def\reg{\mathop{\mathrm{reg}}}

61: \def\Aut{\mathop{\mathrm{Aut}}}

62: \def\Mat{\mathop{\mathrm{Mat}}\nolimits}

63: \def\Dr{\mathop{\mathrm{Dr}}\nolimits}

64: \def\End{\mathop{\mathrm{End}}}

65: \def\NOD{\mathop{\mathrm{gcd}}}

66: 

67: 

68: \multlinegap=0pt

69: 

70: \newtheorem{prop}{Proposition}

71: 

72: \renewcommand{\thesubsection}{\arabic{subsection}}

73: 

74: \renewcommand{\thesection}{\protect\S\arabic{section}}

75: 

76: \theoremstyle{remark}

77: 

78: \newtheorem*{remark}{Remark}

79: \newtheorem*{remarks}{Remarks}

80: 

81: \theoremstyle{definition}

82: 

83: \newtheorem*{definition}{Definition}

84: 

85: \title{Elliptic algebras}

86: \author{Alexander Odesskii}

87: 

88: \date{}

89: 

90: \begin{document}

91: 

92: \sloppy

93: 

94: \maketitle

95: 

96: 

97: \begin{abstract}

98: The survey is devoted to associative $\Z_{\ge0}$-graded algebras presented by $n$

99: generators and $\frac{n(n-1)}2$ quadratic relations and satisfying the so-called 

100: Poincare-Birkhoff-Witt condition (PBW-algebras). We consider examples of such 

101: algebras depending on two continuous parameters (namely, on an elliptic curve 

102: and a point on this curve) which are flat deformations of the polynomial ring in 

103: $n$ variables. Diverse properties of these algebras are described, together with

104: their relations to integrable systems, deformation quantization, moduli spaces 

105: and other directions of modern investigations.

106: \end{abstract}

107: 

108: \newpage

109: 

110: \tableofcontents

111: 

112: \section*{Introduction}

113: \addcontentsline{toc}{section}{Introduction}

114: 

115: In the paper [45] devoted to study the $XYZ$-model and the representations of 

116: the corresponding algebra of monodromy matrices, Sklyanin introduced the family 

117: of associative algebras with four generators and six quadratic relations which 

118: are nowadays called Sklyanin algebras (see also Appendix D.1). The algebras of 

119: this family are naturally indexed by two continuous parameters, namely, by an 

120: elliptic curve and a point on this curve, and each of them is a flat deformation 

121: of the polynomial ring in four variables in the class of $\Z_{\ge0}$-graded 

122: associative algebras. On the other hand, a family of algebras with three generators 

123: (and three quadratic relations) with the same properties arose in [2], [34] (see also

124: [52]). In what follows it turned out (see [10], [17]-[22], [32]-[38]) that such 

125: algebras exist for arbitrarily many generators. The algebras in question are 

126: associative algebras of the following form. Let $V$ be a linear space of dimension 

127: $n$ over the field $\C$. Let $L\subset  V\otimes  V$ be a subspace of dimension 

128: $\frac{n(n-1)}2$. Let us construct an algebra $A$ with the space of generators 

129: $V$ and the space of defining relations $L$, that is, $A=T^*V/(L)$, where $T^*V$ 

130: is the tensor algebra of the space $V$ and $(L)$ is the two-sided ideal generated 

131: by $L$. It is clear that the algebra $A$ is $\Z_{\ge0}$-graded because the ideal 

132: $(L)$ is homogeneous. We have $A=\C\oplus  A_1\oplus A_2\oplus\dots$, where 

133: $A_1=V$,  $A_2=V\otimes

134: V/L$, $A_3=V\otimes V\otimes V/V\otimes L+L\otimes V$, etc.

135: 

136: \begin{definition}

137: We say that $A$ is a PBW-algebra (or satisfies the Poincare-Birkhoff-Witt condition) 

138: if $\dim A_\alpha=\frac{n(n+1)\dots(n+\alpha-1)}{\alpha!}$.

139: \end{definition}

140: 

141: Thus, a PBW-algebra is an algebra with $n$ generators and $\frac{n(n-1)}2$ 

142: quadratic relations for which the dimensions of the graded components are equal to 

143: those of the polynomial ring in $n$ variables.

144: 

145: Algebras of this kind arise in diverse areas of mathematics: in the theory of 

146: integrable systems [45], [46], [28], [9], moduli spaces [20], deformation quantization 

147: [12], [26], non-commutative geometry [2], [3], [11], [27], [47]-[49], [51], 

148: cohomology of algebras [8], [29], [41]-[44], [50], and quantum groups and 

149: $R$-matrices [45], [46], [25], [16], [14], [23], [31]. See Appendix D.

150: 

151: Since there are no classification results in the theory of PBW-algebras (for $n>3$), 

152: we deal with specific examples only. The known examples can conditionally be 

153: divided into two classes, namely, rational and elliptic algebras. Let us present 

154: examples of rational algebras.

155: 

156: 1. {\it  Skew polynomials.} This is the algebra with the generators $\{x_i;i=1,\dots,n\}$ 

157: and the relations $x_ix_j=q_{i,j}x_jx_i$, where $i<j$ and $q_{i,j}\ne0$.

158: 

159: One can readily see that the monomials 

160: $\{x_1^{\alpha_1}\dots x_n^{\alpha_n};\alpha_1,\dots,\alpha_n\in\Z_{\ge0}\}$ 

161: form a basis of the algebra of skew polynomials, which implies the PBW condition. 

162: Since $q_{i,j}$ are arbitrary non-zero numbers, we have obtained an 

163: $\frac{n(n-1)}2$-parameter family of algebras.

164: 

165: 2. {\it Projectivization of Lie algebras.} Let $\g$ be a Lie algebra of dimension $n-1$ 

166: with a basis $\{x_1,\dots,x_{n-1}\}$. We construct an algebra with $n$ generators 

167: $\{c,x_1,\dots,x_{n-1}\}$ and the relations $cx_i=x_ic$ and 

168: $x_ix_j-x_jx_i=c[x_i,x_j]$.

169: 

170: The condition PBW follows from the Poincare-Birkhoff-Witt theorem for the algebra 

171: $\g$. 

172: 

173: 3. {\it Drinfeld algebra.} A new realization of the quantum current algebra 

174: $U_q(\wh{\slg}_2)$ was suggested in [13] (see also [25]). Namely, the generators 

175: $x_k^\pm,h_k$ ($k\in\Z$) similar to the ordinary basis of the Lie algebra 

176: $\wh{\slg}_2$ were introduced. It is assumed that the elements $x_k^+$ satisfy 

177: the quadratic relations 

178: \begin{equation}

179: x_{k+1}^+x_l^+-q^2x_l^+x_{k+1}^+=q^2x_k^+x_{l+1}^+-x_{l+1}^+x_k^+.

180: \end{equation}

181: The elements $x_k^-$ satisfy similar relations. The algebra 

182: $\Dr_n(q)\subset U_q(\wh{\slg}_2)$ generated by $x_1^+,\dots,x_n^+$,

183: $n\in\N$, $q\in\C^*$, is a PBW-algebra.

184: 

185: In the elliptic case the algebra depends on two continuous parameters, namely, an 

186: elliptic curve $\E$ and a point $\eta\in\E$. Just these algebras are the subject 

187: of our survey. Their structure constants are elliptic functions of $\eta$ with 

188: modular parameter $\tau$. Our main example is given by the algebras 

189: $Q_{n,k}(\E,\eta)$, where $n\ge3$ is the number of generators, $k$ is a positive 

190: integer coprime to $n$, and $1\le  k<n$. We define the algebra $Q_{n,k}(\E,\eta)$ 

191: by the generators $\{x_i;i\in\Z/n\Z\}$ and the relations 

192: \begin{equation}

193: \sum_{r\in\Z/n\Z}\frac{\theta_{j-i+r(k-1)}(0)}

194: {\theta_{kr}(\eta)\theta_{j-i-r}(-\eta)}x_{j-r}x_{i+r}=0.

195: \end{equation}

196: The structure of these algebras depends on the expansion of the number $n/k$ in the 

197: continued fraction, and therefore we first study the simplest case $k=1$ and then 

198: pass to the general case. The fact that the algebra $Q_{n,k}(\E,\eta)$ belongs to 

199: the class of PBW-algebras is proved only for generic parameters $\E$ and $\eta$ 

200: (see \S2.6 and \S3). However, we conjecture that this holds for any $\E$ and $\eta$. 

201: A possible way to prove this conjecture is to produce an analog of the functional 

202: realization (see \S2.1) for arbitrary $k$ by using the constructions in \S5.

203: 

204: As we consider, the algebras $Q_{n,k}(\E,\eta)$ are a typical example of elliptic 

205: algebras; however, they are far from exhausting the list of all elliptic algebras. 

206: The simplest example of an elliptic algebra that does not belong to this class 

207: (and even is not a deformation of the polynomial ring) can be constructed as follows. 

208: Let the group $(\Z/2\Z)^2$ with the generators $g_1,g_2$ act by automorphisms on 

209: the algebra $Q_4(\E,\eta)$ as follows:

210: $g_1(x_i)=x_{i+2}$,

211: $g_2(x_i)=(-1)^ix_i$.

212: The same group acts on the algebra of ($2\times2$) matrices,

213: $g_1(\gamma)=\begin{pmatrix}-1&0\\0&1\end{pmatrix}\gamma\begin{pmatrix}-1&0\\0&1\end{pmatrix}^{-1}$, 

214: $g_2(\gamma)=\begin{pmatrix}0&1\\1&0\end{pmatrix}\gamma\begin{pmatrix}0&1\\1&0\end{pmatrix}^{-1}$.

215: This gives an action on the tensor product of associative algebras

216: $Q_4(\E,\eta)\otimes\Mat_2$. Let $\wt Q_4(\E,\eta)\subset

217: Q_4(\E,\eta)\otimes\Mat_2$ consist of elements invariant with respect to the group

218: action. One can readily see that the dimension of the graded components of 

219: $\wt Q_4(\E,\eta)$ coincide with those of $Q_4(\E,\eta)$, and therefore 

220: $\wt Q_4(\E,\eta)$ is a PBW-algebra. For another example of PBW-algebra (with $3$ 

221: generators), see the end of \S1.

222: 

223: Let us now describe one of the main constructions of PBW-algebras. Let 

224: $\lambda(x,y)$ be a meromorphic function of two variables. We construct an 

225: associative graded algebra $\F_\lambda$ as follows. Let the underlying linear space 

226: of $\F_\lambda$ coincide with $\F_\lambda=\C\oplus   F_1\oplus  F_2\oplus\dots$, 

227: where $F_1=\{f(u)\}$ is the space of meromorphic functions of one variable and 

228: $F_\alpha=\{f(u_1,\dots,u_\alpha)\}$ is the space of symmetric meromorphic functions 

229: of $\alpha$ variables. The space $F_\alpha$ is a natural extension of the symmetric 

230: power  $S^\alpha F_1$. The multiplication in the algebra $\F_\lambda$ is defined as 

231: follows: for $f\in F_\alpha$, and $g\in F_\beta$ the product $f*g\in

232: F_{\alpha+\beta}$ is 

233: 

234: \medskip

235: $f*g(u_1,\dots,u_{\alpha+\beta})=$

236: \begin{equation}

237: =\frac1{\alpha!\beta!}\sum_{\sigma\in

238: S_{\alpha+\beta}}f(u_{\sigma_1},\dots,u_{\sigma_\alpha})

239: g(u_{\sigma_{\alpha+1}},\dots,u_{\sigma_{\alpha+\beta}})

240: \prod_{\begin{subarray}{c}1\le                i\le\alpha\\\alpha+1\le

241: j\le\alpha+\beta\end{subarray}}\lambda(u_{\sigma_i},u_{\sigma_j}).

242: \end{equation}

243: In particular, if $f,g\in F_1$, then 

244: \begin{equation}

245: f*g(u_1,u_2)=f(u_1)g(u_2)\lambda(u_1,u_2)+

246: f(u_2)g(u_1)\lambda(u_2,u_1).

247: \end{equation}

248: One can readily see that the multiplication $*$ is associative for any $\lambda(x,y)$.

249: 

250: We now assume that $\lambda(x,y)=\frac{x-qy}{x-y}$, where $q\in\C^*$. Let 

251: $F_1^{(n)}=\{1,u,\dots,u^{n-1}\}\subset    F_1$  be the space of polynomials 

252: of degree less than $n$. Let $F_\alpha^{(n)}=S^\alpha

253: F_1^{(n)}\subset  F_\alpha$ be the space of symmetric polynomials in $\alpha$ 

254: variables of degree less than $n$ with respect to any variable. One can readily see 

255: that $F_\alpha^{(n)}*F_\beta^{(n)}\subseteq

256: F_{\alpha+\beta}^{(n)}$. Therefore, the algebra $\F_\lambda^{(n)}=\oplus_\alpha

257: F_\alpha^{(n)}$ is a subalgebra of $\F_\lambda$. Moreover, for $q=1$ the algebra 

258: $\F_\lambda^{(n)}$ is the polynomial ring $S^*F_1^{(n)}$ because $\lambda(x,y)=1$ 

259: in this case. Therefore, the algebra $\F_\lambda^{(n)}$ is a PBW-algebra for generic 

260: $q$. This algebra is isomorphic to the Drinfeld algebra $\Dr_n(q)$, and an 

261: isomorphism is given by the rule $u^k\mapsto x_{k+1}^+$. The algebra 

262: $Q_n(\E,\eta)$ can be obtained in a similar way with the only modification that the 

263: polynomials are replaced by theta functions (see \S2.1). A similar construction [38], 

264: [22] enables one to construct quantum moduli spaces $\M(\E,B)$ (see Appendix D.3) 

265: for any Borel subgroup $B$. The construction of algebras $Q_{n,k}(\E,\eta)$ for 

266: $k>1$ (and, more generally, quantum moduli spaces $\M(\E,P)$ for a parabolic 

267: subgroup $P$) is more complicated and involves exchange algebras (see \S5 and [21]) 

268: or elliptic $R$-matrices (see \S4).

269: 

270: Let us now describe the contents of the survey. In \S1 we describe the simplest 

271: elliptic PBW-algebras, namely, algebras $Q_3(\E,\eta)$ with three generators. These 

272: algebras were studied in many papers, see, for instance, [2], [3]. The section is 

273: of illustrative nature; we intend to explain some methods of studying elliptic 

274: algebras by the simplest example. The main attention in the survey is paid to the 

275: algebras $Q_n(\E,\eta)$, which are discussed in \S2. We give an explicit construction 

276: of these algebras, present natural families of their representations (which are 

277: studied in [19] in more detail), and describe the symplectic leaves of the 

278: corresponding Poisson algebra (we recall that $Q_n(\E,0)$ is the polynomial ring 

279: in $n$ variables). 

280: 

281: The structure of the algebras $Q_{n,k}(\E,\eta)$, $k>1$, is more complicated, and 

282: the detailed description of their properties is beyond the framework of the survey 

283: (see [35], [20]). The main properties of these algebras are described in \S3. In 

284: \S4 we explain the relationship between these algebras and Belavin's elliptic 

285: $R$-matrices. In \S5 we establish a relation of the algebras $Q_{n,k}(\E,\eta)$ to 

286: the so-called exchange algebras (see (24), (25), and also [36], [24], [33]). In 

287: Appendices A, B, C we present the notation we need and the properties of theta 

288: functions of one and several variables. Appendix D contains a brief survey of 

289: relations of elliptic algebras with other areas of mathematics. We tried to make 

290: this part independent of the main text.

291: 

292: In conclusion we say a few words concerning the facts that remain outside the survey 

293: but are immediately connected with its topic. In [37] the algebras 

294: $Q_{n,k}(\E,\eta)$ are studied provided that $\eta\in\E$ is a point of finite order. 

295: In this case the properties of the algebras $Q_{n,k}(\E,\eta)$ are similar to those 

296: of quantum groups if $q$ is a root of unity; in particular, these algebras are 

297: finite-dimensional over the centre. In [32] we study rational degenerations of the 

298: algebras $Q_{n,k}(\E,\eta)$ occurring if the elliptic curve $\E$ degenerates into 

299: the union of several copies of $\CP^1$ or into $\CP^1$ with a double point.

300: 

301: The algebras $Q_{n,k}(\E,\eta)$ are obtained when quantizing the components of the 

302: moduli spaces $\M(P,\E)$ (see Appendix D.3) that are isomorphic to $\CP^{n-1}$. The 

303: quantization of other components leads to elliptic algebras of more general form. 

304: These algebras were constructed in [38], [22] if $P$ is a Borel subgroup of an 

305: arbitrary group $G$. The case in which $P\subset   GL_m$ is an arbitrary parabolic 

306: subgroup of $GL_m$ is studied in [21].

307: 

308: The symplectic leaves of a Poisson manifold corresponding to the family of algebras 

309: $Q_{n,k}(\E,\eta)$ in a neighbourhood of $\eta=0$ and for a fixed elliptic curve 

310: $\E$ were studied in [20].

311: 

312: The corresponding Poisson algebras belong to the class of algebras with regular 

313: structure of symplectic leaves; these algebras were studied in [39].

314: 

315: 

316: \section{Algebras with three generators}

317: 

318: In this section we consider the simplest examples of elliptic PBW-algebras, namely, 

319: the algebras with three generators. Let us first study the quadratic Poisson 

320: structures on $\C^3$. Let $x_0,x_1,x_2$ be the coordinates on $\C^3$ and let there 

321: be a Poisson structure that is quadratic in these coordinates. We construct the 

322: polynomial $C=x_0\{x_1,x_2\}+x_1\{x_2,x_0\}+x_2\{x_0,x_1\}$. This is a homogeneous 

323: polynomial of degree three because the Poisson structure is quadratic. It is clear 

324: that the form of this polynomial is preserved under linear changes of coordinates 

325: (up to proportionality). Let us restrict ourselves to the non-degenerate case in 

326: which the equation  $C=0$ defines a non-singular projective manifold. It is clear 

327: that this is an elliptic curve. Moreover, by a linear change of variables one can 

328: reduce the polynomial $C$ to the form $C=x_0^3+x_1^3+x_2^3+3kx_0x_1x_2$, where 

329: $k\in\C$. In this case, as one can readily see by using the definition of $C$ and 

330: the Jacoby identity, the Poisson structure must be of the form (up to proportionality):

331: \begin{equation}

332: \{x_0x_1\}=x_2^2+kx_0x_1,\quad\{x_1x_2\}=x_0^2+kx_1x_2,\quad

333: \{x_2x_0\}=x_1^2+kx_2x_0.

334: \end{equation}

335: Moreover, $\{x_i,C\}=0$, and every central element is a polynomial in $C$. We recall 

336: that each Poisson manifold can be partitioned into the so-called symplectic leaves, 

337: which are Poisson submanifolds, and the restrictions of the Poisson structure to 

338: these submanifolds are non-degenerate. In our case, the symplectic leaves are as 

339: follows:

340: 

341: 1) the origin $x_0=x_1=x_2=0$;

342: 

343: 2) the homogeneous manifold $C=0$ without the origin;

344: 

345: 3) the manifolds $C=\lambda$, where $\lambda\in\C$, $\lambda\ne0$.

346: 

347: It is clear that our Poisson structure admits the automorphisms 

348: $x_i\mapsto\epsilon^ix_i$ and $x_i\mapsto  x_{i+1}$, where $\epsilon^3=1$,

349: $i\in\Z/3\Z$. It is natural to assume that the quantization of the Poisson structure 

350: (see Appendix D.2) is the family of associative algebras with the generators 

351: $x_0,x_1,x_2$ and three quadratic relations admitting the same automorphisms. 

352: However, each generic three-dimensional space of quadratic relations which is 

353: invariant with respect to these automorphisms is of the form

354: \begin{equation}

355: \begin{aligned}

356: x_0x_1-qx_1x_0&=px_2^2,\\

357: x_1x_2-qx_2x_1&=px_0^2,\\

358: x_2x_0-qx_0x_2&=px_1^2,

359: \end{aligned}

360: \end{equation}

361: where $p,q\in\C$ are complex numbers. We denote by $A_{p,q}$ the algebra with the 

362: generators $x_0,x_1,x_2$ and the defining relations (6). It is clear that the 

363: algebra $A_{p,q}$ is $\Z_{\ge0}$-graded, that is, $A_{p,q}=\C\oplus    F_1\oplus

364: F_2\oplus\dots$, where $F_\alpha F_\beta\subseteq F_{\alpha+\beta}$. Here $F_\alpha$ 

365: stands for the linear space spanned by the (non-commutative) monomials in 

366: $x_0,x_1,x_2$ of degree $\alpha$. It is natural to expect that the dimension of 

367: $F_\alpha$ is equal to that of the space of polynomials in three variables of degree 

368: $\alpha$, that is, $\dim F_\alpha=\frac{(\alpha+1)(\alpha+2)}2$.

369: 

370: Moreover, the Poisson algebra (5) has a central function 

371: $C=x_0^3+x_1^3+x_2^3+3kx_0x_1x_2$, and the centre is generated by the element $C$. 

372: Therefore, it is natural to expect that for generic $p$ and $q$ the algebra 

373: $A_{p,q}$ has a central element of the form $C_{p,q}=\phi x_0^3+\psi x_1^3+\mu

374: x_2^3+\lambda x_0x_1x_2$, where $\phi,\psi,\mu$, and $\lambda$ are functions of $p$ 

375: and $q$ (one can verify the existence of an element $C_{p,q}$ by the immediate 

376: calculation), and the centre is generated by $C_{p,q}$.

377: 

378: The standard technique of proving such assertions (for instance, the Poincare-Birkhoff

379: -Witt theorem for Lie algebras) makes use of the filtration on an algebra and the 

380: study of the graded adjoint algebra. In our case the algebra is already graded, and 

381: one cannot proceed by the ordinary induction on the terms of lesser filtration; 

382: therefore we use another technique. Namely, we shall study a certain class of modules 

383: over the algebra $A_{p,q}$ and try to obtain results on the algebra $A_{p,q}$ by 

384: using an information on the modules. The following class of modules is useful for 

385: our purposes.

386: 

387: \begin{definition}

388: A module over a $\Z_{\ge0}$-graded algebra $A$ is said to be linear if it is 

389: $\Z_{\ge0}$-graded as an $A$-module, generated by the space of degree $0$, and the 

390: dimensions of all components are equal to $1$.

391: \end{definition}

392: 

393: Let us study the linear modules over the algebra $A_{p,q}$. By definition, a linear 

394: module $M$ admits a basis $\{v_\alpha,\alpha\ge0\}$ with the following action of 

395: the generators:

396: $$

397: x_0v_\alpha=x_\alpha     v_{\alpha+1},\quad      x_1v_\alpha=y_\alpha

398: v_{\alpha+1},\quad x_2v_\alpha=z_\alpha v_{\alpha+1},

399: $$

400: where $x_\alpha,y_\alpha,z_\alpha$ are sequences, and $x_\alpha,y_\alpha,z_\alpha$ 

401: do not vanish simultaneously for any $\alpha$ (we want $M$ be generated by $v_0$). 

402: A change of the basis of the form $v_\alpha\to\lambda_\alpha       v_\alpha$ 

403: multiplies the triple $(x_\alpha,y_\alpha,z_\alpha)\in\C^3$ by 

404: $\frac{\lambda_{\alpha+1}}{\lambda_\alpha}$, that is, the module $M$ is defined by 

405: the sequence of points $(x_\alpha:y_\alpha:z_\alpha)\in\CP^2$ uniquely up to 

406: isomorphism of graded modules. It is clear that a sequence of points 

407: $(x_\alpha:y_\alpha:z_\alpha)\in\CP^2$ defines a module over the algebra $A_{p,q}$ 

408: if and only if the relations (6) hold for the operators on $M$ corresponding to this 

409: sequence. This is equivalent to the following relations:

410: \begin{equation}

411: \begin{aligned}

412: x_{\alpha+1}y_\alpha-qy_{\alpha+1}x_\alpha&=pz_{\alpha+1}z_\alpha,\\

413: y_{\alpha+1}z_\alpha-qz_{\alpha+1}y_\alpha&=px_{\alpha+1}x_\alpha,\\

414: z_{\alpha+1}x_\alpha-qx_{\alpha+1}z_\alpha&=py_{\alpha+1}y_\alpha.

415: \end{aligned}

416: \end{equation}

417: 

418: The relations (7) form a system of linear equations for $x_\alpha,y_\alpha,z_\alpha$ 

419: which has a non-zero solution (by the assumption on the module $M$), and therefore 

420: the determinant $\begin{vmatrix}-qy_{\alpha+1}&x_{\alpha+1}&-pz_{\alpha+1}\\

421: -px_{\alpha+1}&-qz_{\alpha+1}&y_{\alpha+1}\\

422: z_{\alpha+1}&-py_{\alpha+1}&-qx_{\alpha+1}\end{vmatrix}$ must vanish. Similarly, the 

423: relations (7) form a system of linear equations on 

424: $x_{\alpha+1},y_{\alpha+1},z_{\alpha+1}$ that has a non-zero solution, and therefore 

425: $\begin{vmatrix}y_\alpha&-qx_\alpha&-pz_\alpha\\

426: -px_\alpha&z_\alpha&-qy_\alpha\\

427: -qz_\alpha&-py_\alpha&x_\alpha

428: \end{vmatrix}=0$. One can readily see that these determinants give the same cubic 

429: polynomial in three variables. We see that for any $\alpha\ge0$ the point with 

430: the coordinates $(x_\alpha:y_\alpha:z_\alpha)$ belongs to the cubic 

431: \begin{equation}

432: x_\alpha^3+y_\alpha^3+z_\alpha^3+\frac{p^3+q^3-1}{pq}

433: x_\alpha y_\alpha z_\alpha=0.

434: \end{equation}

435: Moreover, if a point $(x_\alpha:y_\alpha:z_\alpha)$ belongs to this cubic, then, 

436: solving the system of linear equations (7) with respect to 

437: $x_{\alpha+1},y_{\alpha+1},z_{\alpha+1}$, we obtain a new point  

438: $(x_{\alpha+1}:y_{\alpha+1}:z_{\alpha+1})$ on the same cubic (because the 

439: determinant of the system (7) must be equal to $0$). Thus, the system (7) defines 

440: an automorphism of the projective manifold (8). Let us choose some 

441: $k=\frac{p^3+q^3-1}{pq}$. Then, varying $q$, we obtain a one-parameter family 

442: of automorphisms of the projective curve in $\CP^2$ given by the equation 

443: $x^3+y^3+z^3+kxyz=0$. As is known, for generic $k$ this equation defines an 

444: elliptic curve. Let this curve be $\E=\C/\Gamma$, where $\Gamma$ is an integral 

445: lattice generated by $1$ and $\tau$, where $\Im\tau>0$. The parameter $k$ is a 

446: function of $\tau$. If $k$ is chosen, then, passing to the limit as $q\to1$, we 

447: see that $p\to0$, and the automorphism defined by (7) tends to the identity 

448: automorphism. Therefore, our family of automorphisms of the elliptic curve $\E$ 

449: given by the equation (8) is a deformation of the identity automorphism. Thus, 

450: every automorphism of this family is a translation, of the form $u\to u+\eta$, 

451: where $u,\eta\in\E=\C/\Gamma$. Let $u_\alpha\in\E=\C/\Gamma$ be a point with the 

452: coordinates $(x_\alpha:y_\alpha:z_\alpha)$. We see that $u_{\alpha+1}=u_\alpha+\eta$, 

453: where $\eta$ depends only on the algebra, that is, on $p$ and $q$. Hence, 

454: $u_\alpha=u+\alpha\eta$, where $u\in\E$ is the parameter of the module $M$. We 

455: have obtained the following result.

456: 

457: \begin{prop}

458: The linear modules over the algebra $A_{p,q}$ are indexed by a point of the 

459: elliptic curve $\E\subset\CP^2$ given by the equation $x^3+y^3+z^3+k_{p,q}xyz=0$, 

460: where $k_{p,q}=\frac{p^3+q^3-1}{pq}$. The module $M_u$ corresponding to a point 

461: $u\in\E$ is given by the formulas

462: $$

463: x_0v_\alpha=x_\alpha     v_{\alpha+1},\quad      x_1v_\alpha=y_\alpha

464: v_{\alpha+1},\quad x_2v_\alpha=z_\alpha v_{\alpha+1},

465: $$

466: where $(x_\alpha:y_\alpha:z_\alpha)$ are the coordinates of the point 

467: $u+\alpha\eta\in\E$. Here the shift $\eta$ is determined by $p$ and $q$.

468: \end{prop}

469: 

470: We note that, when studying linear modules, for an algebra $A_{p,q}$ we have 

471: constructed both an elliptic curve $\E\subset\CP^2$ and a point $\eta\in\E$. In 

472: what follows we shall see that, conversely, the algebra $A_{p,q}$ can be  

473: reconstructed from $\E$ and $\eta$. Thus, two continuous parameters, $\E$ 

474: (that is $\tau$) and $\eta$, give a natural parametrization of the algebras 

475: $A_{p,q}$. Therefore, we change the notation and denote the algebra $A_{p,q}$ by 

476: $Q_3(\E,\eta)$.

477: 

478: Let us now apply a uniformization of the elliptic curve $\E\subset\CP^2$ given 

479: by the equation (8) by theta functions of order three (see Appendix A). A point 

480: $u\in\E=\C/\Gamma$ has the coordinates $(\theta_0(u):\theta_1(u):\theta_2(u))\in\CP^2$. 

481: In this notation, the module $M_u$ is given by the formulas

482: $$

483: x_0v_\alpha=\theta_0(u+\alpha\eta)v_{\alpha+1},\quad

484: x_1v_\alpha=\theta_1(u+\alpha\eta)v_{\alpha+1},\quad

485: x_2v_\alpha=\theta_2(u+\alpha\eta)v_{\alpha+1}.

486: $$

487: Let $\e$ be the linear operator in the space with basis $\{v_\alpha,\alpha\ge0\}$ 

488: given by the formula $\e v_\alpha=v_{\alpha+1}$. Let $u$ be the diagonal operator 

489: in the same space such that $\e u=(u-\eta)\e$. We have 

490: $uv_\alpha=(u_0+\alpha\eta)v_\alpha$ for some $u_0\in\C$. It is clear that the  

491: generators of the algebra $Q_3(\E,\eta)$ in the representation $M_u$ become 

492: $$

493: x_0=\theta_0(u)\e,\quad

494: x_1=\theta_1(u)\e,\quad

495: x_2=\theta_2(u)\e.

496: $$

497: 

498: This gives the following reformulation of the description of linear modules.

499: 

500: \begin{prop}

501: Let us consider the $\Z_{\ge0}$-graded algebra $B(\eta)=\C\oplus

502: B_1\oplus B_2\oplus\dots$, where  $B_\alpha=\{f(u)\e^\alpha\}$, $f$ ranges over 

503: all holomorphic functions, and the multiplication is given by the formula \emph:

504: $f(u)\e^\alpha\cdot

505: g(u)\e^\beta=f(u)g(u-\alpha\eta)\e^{\alpha+\beta}$. Then there is an algebra 

506: homomorphism $\phi\colon Q_3(\E,\eta)\to B(\eta)$ such that 

507: $x_0\to\theta_0(u)\e$,           $x_1\to\theta_1(u)\e$,

508: $x_2\to\theta_2(u)\e$.

509: \end{prop}

510: 

511: Proposition 2 provides a lower bound for the dimension $\dim F_\alpha$ of the 

512: graded components of the algebra $Q_3(\E,\eta)$. Really, the homomorphism $\phi$ 

513: preserves the grading, that is, $\phi(F_\alpha)\subset B_\alpha$. We have

514: $$

515: \phi(x_{i_1}\dots

516: x_{i_\alpha})=\theta_{i_1}(u)\e\dots\theta_{i_\alpha}(u)\e=

517: \theta_{i_1}(u)\theta_{i_2}(u-\eta)\dots

518: \theta_{i_2}(u-(\alpha-1)\eta)\e^\alpha.

519: $$

520: Thus, $\phi(F_\alpha)$ is the linear space (of holomorphic functions) spanned by 

521: the functions $\{\theta_{i_1}(u),\dots,\theta_{i_2}(u-(\alpha-1)\eta)\};

522: i_1,\dots,i_\alpha=0,1,2$. It is clear that all these functions are theta 

523: functions of order $3\alpha$ and belong to the space 

524: $\Theta_{3\alpha,\frac{\alpha(\alpha-1)}23\eta}(\Gamma)$. One can readily prove 

525: that the image $\phi(F_\alpha)$ coincides with the entire space 

526: $\Theta_{3\alpha,\frac{\alpha(\alpha-1)}23\eta}(\Gamma)$, and hence 

527: $\dim\phi(F_\alpha)=3\alpha$. We have obtained the bound $\dim  F_\alpha\ge3\alpha$. 

528: On the other hand, we know that $\dim F_\alpha\le\frac{(\alpha+1)(\alpha+2)}2$ 

529: because the relations in $Q_3(\E,\eta)$ are deformations of the relations in the 

530: polynomial ring in three variables. We expect that the equality $\dim

531: F_\alpha=\frac{(\alpha+1)(\alpha+2)}2$ holds for generic $\tau$ and $\eta$. Let 

532: us compare these numbers:

533: $$

534: \begin{array}{c|c|c|c|c}

535: \alpha&1&2&3&4\\

536: \hline

537: \dim F_\alpha(\text{conjecture})&3&6&10&15\\

538: \hline

539: \dim\phi(F_\alpha)&3&6&9&12

540: \end{array}

541: $$

542: We see that the first discrepancy holds for $\alpha=3$; possibly $\phi$ has a 

543: one-dimensional kernel on the space $F_3$. It can be shown that, really, there is 

544: a cubic element $C\in Q_3(\E,\eta)$ such that $C\ne0$ and $\phi(C)=0$. The 

545: element $C$ turns out to be central, that is, $x_\alpha    C=Cx_\alpha$   for

546: $\alpha=0,1,2$. Passing to the limit as $\eta\to0$ (for a fixed $\tau$), we see 

547: that $C\to    x_0^3+x_1^3+x_2^3+kx_0x_1x_2$ because the $\theta_i(u)$s uniformize 

548: the elliptic curve, that is, $\theta_0^3+\theta_1^3+\theta_2^3+k\theta_0\theta_1\theta_2=0$. 

549: Further, if $C$ is central and is not a zero divisor (the latter obviously holds 

550: for generic $\tau$ and $\eta$), then every element $\ker\phi$ must be divisible 

551: by $C$ according to the dimensional considerations. Really, the graded linear 

552: space $\oplus_{\alpha\ge0}F_\alpha$ turns out to be not smaller than 

553: $\left(\oplus_{\alpha\ge0}\Theta_{3\alpha,\frac{\alpha(\alpha-1)}23\eta}\right)\otimes\C[C]$, where $\deg C=3$. 

554: One can readily see that the component of degree $\alpha$ of this tensor product 

555: of graded linear spaces is of dimension $\frac{(\alpha+1)(\alpha+2)}2$. However, 

556: we know that $\dim F_\alpha\le\frac{(\alpha+1)(\alpha+2)}2$, which implies 

557: $\dim  F_\alpha=\frac{(\alpha+1)(\alpha+2)}2$. We have obtained the following 

558: result.

559: 

560: \begin{prop}

561: For generic $\tau$  and  $\eta$ the algebra $Q_3(\E,\eta)$ has a cubic central 

562: element $C$. The quotient algebra $Q_3(\E,\eta)/(C)$ is isomorphic to 

563: $\oplus_{\alpha\ge0}\Theta_{3\alpha,\frac{\alpha(\alpha-1)}23\eta}(\Gamma)$, where the product 

564: of elements $f\in\Theta_{3\alpha,\frac{\alpha(\alpha-1)}23\eta}(\Gamma)$ and 

565: $g\in\Theta_{3\beta,\frac{\beta(\beta-1)}23\eta}(\Gamma)$ is given by the formula 

566: $f*g(u)=f(u)g(u-3\alpha\eta)$.

567: \end{prop}

568: 

569: It follows from our description of $Q_3(\E,\eta)/(C)$ that this algebra is 

570: centre-free for generic $\eta$. Therefore, the centre of the algebra $Q_3(\E,\eta)$ 

571: is generated by the element $C$.

572: 

573: Let us now find the relations in the algebra $Q_3(\E,\eta)$, that is, let us 

574: express $p$ and $q$ in term of $\tau$ and $\eta$. We have 

575: $x_ix_{i+1}-qx_{i+1}x_i-px_{i+2}^2=0$ (these are the relations in (6)). Applying 

576: the homomorphism $\phi$, we obtain

577: $$

578: \theta_i(u)\theta_{i+1}(u-\eta)-q\theta_{i+1}(u)\theta_i(u-\eta)-

579: p\theta_{i+2}(u)\theta_{i+2}(u-\eta)=0.

580: $$

581: Hence (see (28) in Appendix A), $q=-\frac{\theta_1(\eta)}{\theta_2(\eta)}$,

582: $p=-\frac{\theta_0(\eta)}{\theta_2(\eta)}$.

583: 

584: The similar investigation of the Sklyanin algebra with four generators (see 

585: Appendix D.1) gives the following result.

586: 

587: \begin{prop}

588: For a generic Sklyanin algebra $S$ with four generators and the relations (39) 

589: one can find an elliptic curve $\E=\C/\Gamma$ defined by two quadrics in $\CP^3$ 

590: and a point $\eta\in\E$ such that \emph:

591: 

592: \emph{1)} there is a graded algebra homomorphism $\phi\colon

593: S\to B(\eta)$;

594: 

595: \emph{2)} the image of this homomorphism in $B_\alpha$ is 

596: $\Theta_{4\alpha,\frac{\alpha(\alpha-1)}24\eta+\frac\alpha2}(\Gamma)$;

597: 

598: \emph{3)} the kernel of this homomorphism is generated by two quadratic elements 

599: $C_1$ and $C_2$.

600: 

601: Thus, $S/(C_1,C_2)=\oplus_{\alpha\ge0}\Theta_{4\alpha,\frac{\alpha(\alpha-1)}24\eta+\frac\alpha2}(\Gamma)$.

602: \end{prop}

603: 

604: The Sklyanin algebra $S$ can be reconstructed from $\E$ and $\eta$. Let us 

605: denote this algebra by $Q_4(\E,\eta)$.

606: 

607: The following natural question arises: Does there exist a similar algebra 

608: $Q_n(\E,\eta)$ for any $n$?

609: 

610: To answer this question, the information concerning linear modules is insufficient 

611: because these modules are too small to reconstruct the algebra $Q_n(\E,\eta)$ 

612: for any $n$. Really, the algebra  $Q_n(\E,\eta)$ must have the functional 

613: dimension $n$, whereas the linear modules are of dimension one. Therefore, these 

614: modules can be used only when reconstructing a quotient algebra of $Q_n(\E,\eta)$. 

615: To overcome these difficulties, it is natural to study more general modules. 

616: Namely, let us study modules over the algebra $Q_3(\E,\eta)$ with a basis 

617: $\{v_{i,j};i,j\in\Z_{\ge0}\}$ and such that the generators of the algebra 

618: $Q_3(\E,\eta)$ take any element $v_{ij}$ to a linear combination of 

619: $v_{i+1,j}$ and $v_{i,j+1}$. Calculations show that every such module is of 

620: the form

621: $$

622: x_iv_{\alpha,\beta}=\frac{\theta_i(u_1+(\alpha-2\beta)\eta)}

623: {\theta(u_1-u_2+3(\alpha-\beta)\eta)}v_{\alpha+1,\beta}+

624: \frac{\theta_i(u_2+(\beta-2\alpha)\eta)}

625: {\theta(u_2-u_1+3(\beta-\alpha)\eta)}v_{\alpha,\beta+1},

626: $$

627: where $i\in\Z/3\Z$, $\alpha,\beta\in\Z_{\ge0}$, and $u_1,u_2\in\C$. Thus, the 

628: modules of this kind are indexed by a pair of points $u_1,u_2\in\E$. If we now 

629: assume that the algebra $Q_n(\E,\eta)$ has analogous modules (see (15)), then 

630: the above information uniquely defines the algebra $Q_n(\E,\eta)$.

631: 

632: \begin{remarks}

633: 1. One can pose the following more general problem. Let $M\subset\CP^{n-1}$ be 

634: a projective manifold and let $T$ be an automorphism of $M$. For a point $u\in  M$ 

635: we denote by $z_i(u)$ (where $i=0,\dots,n-1$) the homogeneous coordinates of $u$. 

636: Does there exist a PBW-algebra with $n$ generators $\{x_i,i=0,\dots,n-1\}$ that 

637: has a linear module $L_u$ (for any point $u\in   M$) given by the formula 

638: $x_iv_\alpha=z_i(T^\alpha

639: u)v_{\alpha+1}$? Here $T^\alpha u$ stands for $T(T(\dots  T(u)\dots)$. The algebras 

640: $Q_{n,k}(\E,\eta)$ are a solution of this problem for some $M$ and $T$, namely, 

641: if $M=\E^p$ is a power of a curve $\E$ and $T$ a translation (see \S5, Proposition 12). 

642: Here $p$ stands for the length of the expansion of $n/k=n_1-\frac1{n_2-\ldots-\frac1{n_p}}$ 

643: in the continued fraction.

644: 

645: 2. Let \footnote{This example was communicated to the author by Oleg Ogievetsky 

646: [1], [15], [40].} $A_3$ be the algebra with the generators $x,y,z$ and the 

647: relations $\epsilon 

648: zx+\epsilon^5y^2+xz=0$, $\epsilon^2z^2+yx+\epsilon^4xy=0$, and 

649: $zy+\epsilon^7yz+\epsilon^8x^2=0$, where $\epsilon^9=1$. This PBW-algebra 

650: corresponds to the case in which $M\subset\CP^2$ is an elliptic curve given by 

651: the equation $x^3+y^3+z^3=0$ and $T$ is an automorphism corresponding to the 

652: complex multiplication on $M$. The algebra $A_3$ is not a quantization of any 

653: Poisson structure on $\C^3$.

654: \end{remarks}

655: 

656: \section{Algebra $Q_n(\E,\eta)$}

657: 

658: \subsection{Construction}

659: 

660: For any $n\in\N$, any elliptic curve $\E=\C/\Gamma$, and any point $\eta\in\E$ 

661: we construct a graded associative algebra 

662: $Q_n(\E,\eta)=\C\oplus F_1\oplus  F_2\oplus\dots$, where $F_1=\Theta_{n,c}(\Gamma)$ 

663: and $F_\alpha=S^\alpha\Theta_{n,c+(\alpha-1)n}(\Gamma)$. By construction, 

664: $\dim  F_\alpha=\frac{n(n+1)\dots(n+\alpha-1)}{\alpha!}$. It is clear that the 

665: space $F_\alpha$ can be realized as the space of holomorphic symmetric functions 

666: of $\alpha$ variables $\{f(z_1,\dots,z_\alpha)\}$ such that 

667: \begin{equation}

668: \begin{aligned}

669: f(z_1+1,z_2,\dots,z_\alpha)&=f(z_1,\dots,z_\alpha),\\

670: f(z_1+\tau,z_2,\dots,z_\alpha)&=(-1)^ne^{-2\pi i(nz_1-c-(\alpha-1)n)}f(z_1,\dots,z_\alpha).

671: \end{aligned}

672: \end{equation}

673: 

674: For $f\in  F_\alpha$ and $g\in  F_\beta$ we define the symmetric function $f*g$ 

675: of $\alpha+\beta$ variables by the formula 

676: \begin{multline*}

677: f*g(z_1,\dots,z_{\alpha+\beta})=\frac1{\alpha!\beta!}

678: \sum_{\sigma\in S_{\alpha+\beta}}

679: f(z_{\sigma_1},\dots,z_{\sigma_\alpha})

680: g(z_{\sigma_{\alpha+1}}-2\alpha\eta,\dots,z_{\sigma_{\alpha+\beta}}-

681: 2\alpha\eta)\times\\

682: \times\prod_{\begin{subarray}{c}1\le                i\le\alpha\\\alpha+1\le

683: j\le\alpha+\beta\end{subarray}}

684: \frac{\theta(z_{\sigma_i}-z_{\sigma_j}-n\eta)}

685: {\theta(z_{\sigma_i}-z_{\sigma_j})}.

686: \end{multline*}

687: In particular, for $f,g\in F_1$ we have

688: $$

689: f*g(z_1,z_2)=f(z_1)g(z_2-2\eta)\frac{\theta(z_1-z_2-n\eta)}

690: {\theta(z_1-z_2)}+f(z_2)g(z_1-2\eta)\frac{\theta(z_2-z_1-n\eta)}

691: {\theta(z_2-z_1)}.

692: $$

693: Here $\theta(z)$ is a theta function of order one (see Appendix A).

694: 

695: \begin{prop}

696: If $f\in  F_\alpha$ and  $g\in F_\beta$, then $f*g\in F_{\alpha+\beta}$. The 

697: operation $*$ defines an associative multiplication on the space 

698: $\oplus_{\alpha\ge0}F_\alpha$

699: \end{prop}

700: 

701: \begin{proof}

702: Let us show that $f*g\in F_{\alpha+\beta}$. It immediately follows from the 

703: assumptions (9) concerning $f$ and $g$ and also from the properties of $\theta(z)$ 

704: (see Appendix A) that every summand in the formula for $f*g$ satisfies condition 

705: (9) for $F_{\alpha+\beta}$. Hence, $f*g$ is a meromorphic symmetric function 

706: satisfying condition (9). This function can have a pole of order not exceeding 

707: one on the diagonals $z_i-z_j=0$ and also for $z_i-z_j\in\Gamma$ because $\theta(z)$ 

708: has zeros for $z\in\Gamma$. However, the order of a pole of a symmetric function 

709: on the diagonal must be even. This implies that the function $f*g$ is holomorphic 

710: for $z_i=z_j$, and it follows from (9) that $f*g$ is holomorphic for $z_i-z_j\in\Gamma$ 

711: as well.

712: 

713: One can immediately see that the multiplication $*$ is associative.

714: \end{proof}

715: 

716: \subsection{Main properties of the algebra $Q_n(\E,\eta)$}

717: 

718: By construction, the dimensions of the graded components of the algebra   

719: $Q_n(\E,\eta)$ coincide with those for the polynomial ring in $n$ variables. For 

720: $\eta=0$ the formula for $f*g$ becomes

721: $$

722: f*g(z_1,\dots,z_{\alpha+1})=\frac1{\alpha!\beta!}\sum_{\sigma\in

723: S_{\alpha+\beta}}f(z_{\sigma_1},\dots,z_{\sigma_\alpha})

724: g(z_{\sigma_{\alpha+1}},\dots,z_{\sigma_{\alpha+\beta}}).

725: $$

726: This is the formula for the ordinary product in the algebra $S^*\Theta_{n,c}(\Gamma)$, 

727: that is, in the polynomial ring in $n$ variables. Therefore, for a fixed elliptic 

728: curve $\E$ (that is, for a fixed modular parameter $\tau$) the family of algebras 

729: $Q_n(\E,\eta)$ is a deformation of the polynomial ring. In particular (see 

730: Appendix D.2), there is a Poisson algebra, which we denote by $q_n(\E)$. One can 

731: readily obtain the formula for the Poisson bracket on the polynomial ring from 

732: the formula for $f*g$ by expanding the difference $f*g-g*f$ in the Taylor series 

733: with respect to $\eta$. It follows from the semicontinuity arguments that the 

734: algebra $Q_n(\E,\eta)$ with generic $\eta$ is determined by $n$ generators and 

735: $\frac{n(n-1)}2$ quadratic relations. One can prove (see \S2.6) that this is the 

736: case if $\eta$ is not a point of finite order on $\E$, that is, $N\eta\not\in\Gamma$ 

737: for any $N\in\N$.

738: 

739: The space $\Theta_{n,c}(\Gamma)$ of the generators of the algebra $Q_n(\E,\eta)$ 

740: is endowed with an action of a finite group $\wt{\Gamma_n}$ which is a central 

741: extension of the group $\Gamma/n\Gamma$ of points of order $n$ on the curve $\E$ 

742: (see Appendix A). It immediately follows from the formula for the product $*$ 

743: that the corresponding transformations of the space $F_\alpha=S^\alpha\Theta_{n,c}(\Gamma)$ 

744: are automorphisms of the algebra $Q_n(\E,\eta)$.

745: 

746: \subsection{Bosonization of the algebra $Q_n(\E,\eta)$}

747: 

748: The main approach to obtain representations of the algebra $Q_n(\E,\eta)$ is to 

749: construct homomorphisms from this algebra to other algebras with simple structure 

750: (close to Weil algebras) which have a natural set of representations. These 

751: homomorphisms are referred to as bosonizations, by analogy with the known 

752: constructions of quantum field theory.

753: 

754: Let $B_{p,n}(\eta)$ be a $\Z^p$-graded algebra whose space of degree 

755: $(\alpha_1,\dots,\alpha_p)$ is of the form  

756: $\{f(u_1,\dots,u_p)\e_1^{\alpha_1}\dots\e_p^{\alpha_p}\}$, where $f$ ranges over 

757: the meromorphic functions of $p$ variables and $\e_1,\dots,\e_p$ are elements of 

758: the algebra $B_{p,n}(\eta)$. Let $B_{p,n}(\eta)$ be generated by the space of 

759: meromorphic functions $f(u_1,\dots,u_p)$ and by the elements $\e_1,\dots,\e_p$ 

760: with the defining relations 

761: \begin{equation}

762: \begin{gathered}

763: \e_\alpha

764: f(u_1,\dots,u_p)=

765: f(u_1-2\eta,\dots,u_\alpha+(n-2)\eta,\dots,u_p-2\eta)\e_\alpha,\\

766: \e_\alpha\e_\beta=\e_\beta\e_\alpha,\quad

767: f(u_1,\dots,u_p)g(u_1,\dots,u_p)=g(u_1,\dots,u_p)f(u_1,\dots,u_p)

768: \end{gathered}

769: \end{equation}

770: 

771: We note that the subalgebra of $B_{p,n}(\eta)$ consisting of the elements of 

772: degree $(0,\dots,0)$ is the commutative algebra of all meromorphic functions of 

773: $p$ variables with the ordinary multiplication.

774: 

775: \begin{prop}

776: Let $\eta\in\E$ be a point of infinite order. For any $p\in\N$ there is a 

777: homomorphism $\phi_p\colon  Q_n(\E,\eta)\to B_{p,n}(\eta)$ that acts on the 

778: generators of the algebra $Q_n(\E,\eta)$ by the formula \emph:

779: \begin{equation}

780: \phi_p(f)=\sum_{1\le\alpha\le

781: p}\frac{f(u_\alpha)}{\theta(u_\alpha-u_1)\dots\theta(u_\alpha-u_p)}

782: \e_\alpha.

783: \end{equation}

784: Here $f\in\Theta_{n,c}(\Gamma)$ is a generator of $Q_n(\E,\eta)$ and the product 

785: in the denominator is of the form 

786: $\prod_{i\ne\alpha}\theta(u_\alpha-u_i)$.

787: \end{prop}

788: 

789: \begin{proof}

790: We write $\xi_\alpha=\frac1{\theta(u_\alpha-u_1)\dots\theta(u_\alpha-u_p)}

791: \e_\alpha$. It is clear that the elements $\xi_1,\dots,\xi_p$ together with the 

792: space of meromorphic functions $\{f(u_1,\dots,u_p)\}$ generate the algebra 

793: $B_{p,n}(\eta)$. The relations (10) become 

794: \begin{gather*}

795: \xi_\alpha

796: f(u_1,\dots,u_p)=

797: f(u_1-2\eta,\dots,u_\alpha+(n-2)\eta,\dots,u_p-2\eta)\xi_\alpha\\

798: \xi_\alpha\xi_\beta=-\frac{e^{2\pi

799: i(u_\beta-u_\alpha)}\theta(u_\alpha-u_\beta+n\eta)}

800: {\theta(u_\beta-u_\alpha+n\eta)}\xi_\beta\xi_\alpha

801: \end{gather*}

802: The formula (11) can be represented as 

803: \begin{equation}

804: \phi_p(f)=\sum_{1\le\alpha\le p}f(u_\alpha)\xi_\alpha.

805: \end{equation}

806: Using (12) and the formula for the multiplication in the algebra $Q_n(\E,\eta)$ 

807: and assuming that $\phi_p$ is a homomorphism, one can readily evaluate the 

808: extension of the map $\phi_p$ to the entire algebra. For instance, in the grading 

809: $2$ we have

810: \begin{align*}

811: \phi_p(f*g)&=\sum_{1\le\alpha\le p}f(u_\alpha)\xi_\alpha\cdot

812: \sum_{1\le\beta\le p}g(u_\beta)\xi_\beta=

813: \sum_{1\le\alpha,\beta\le                     p}f(u_\alpha)\xi_\alpha

814: f(u_\beta)\xi_\beta=\\

815: &=\sum_{\begin{subarray}{c}1\le\alpha,\beta\le

816: p\\\alpha\ne\beta\end{subarray}}

817: f(u_\alpha)g(u_\beta-2\eta)\xi_\alpha\xi_\beta+

818: \sum_{1\le\alpha\le p}f(u_\alpha)g(u_\alpha+(n-2)\eta)\xi_\alpha^2.

819: \end{align*}

820: The first sum is 

821: \begin{align*}

822: \sum_{1\le\alpha<\beta\le

823: p}&(f(u_\alpha)g(u_\beta-2\eta)\xi_\alpha\xi_\beta+

824: f(u_\beta)g(u_\alpha-2\eta)\xi_\beta\xi_\alpha)=

825: \end{align*}

826: \begin{align*}

827: =\sum_{1\le\alpha<\beta\le

828: p}\left(f(u_\alpha)g(u_\beta-2\eta)\xi_\alpha\xi_\beta-

829: f(u_\beta)g(u_\alpha-2\eta)\frac{e^{2\pi

830: i(u_\alpha-u_\beta)}\theta(u_\beta-u_\alpha-n\eta)}

831: {\theta(u_\alpha-u_\beta+n\eta)}\xi_\alpha\xi_\beta\right)=

832: \end{align*}

833: \begin{align*}

834: =\sum_{1\le\alpha<\beta\le

835: p}\frac{\theta(u_\alpha-u_\beta)}{\theta(u_\alpha-u_\beta-n\eta)}\times

836: \end{align*}

837: \begin{align*}

838: \times\left(f(u_\alpha)g(u_\beta-2\eta)

839: \frac{\theta(u_\alpha-u_\beta-n\eta)}{\theta(u_\alpha-u_\beta)}+

840: f(u_\beta)g(u_\alpha-2\eta)\frac{\theta(u_\beta-u_\alpha-n\eta)}

841: {\theta(u_\beta-u_\alpha)}\right)\xi_\alpha\xi_\beta=

842: \end{align*}

843: \begin{align*}

844: =\sum_{1\le\alpha<\beta\le

845: p}\frac{\theta(u_\alpha-u_\beta)}{\theta(u_\alpha-u_\beta-n\eta)}

846: f*g(u_\alpha,u_\beta)\xi_\alpha\xi_\beta\quad\text{ }

847: \end{align*}

848: Moreover, $f(u_\alpha)g(u_\alpha+(n-2)\eta)=

849: \frac{\theta(-n\eta)}{\theta(-2n\eta)}f*g(u_\alpha,u_\alpha+n\eta)$. We finally 

850: obtain

851: \medskip

852: \begin{equation}

853: \phi_p(f*g)=

854: \end{equation}

855: $$=\sum_{1\le\alpha<\beta\le

856: p}\frac{\theta(u_\alpha-u_\beta)}{\theta(u_\alpha-u_\beta-n\eta)}

857: f*g(u_\alpha,u_\beta)\xi_\alpha\xi_\beta+\frac{\theta(-n\eta)}

858: {\theta(-2n\eta)}\sum_{1\le\alpha\le

859: p}f*g(u_\alpha,u_\alpha+n\eta)\xi_\alpha^2$$

860: We see that the map $\phi_p$ can be extended to the quadratic part of the algebra 

861: $Q_n(\E,\eta)$ because the right-hand side of (13) depends on $f*g$ only but not 

862: on $f$ and $g$ separately. Thus implies the assertion for generic $\eta$ because 

863: in this case the algebra  $Q_n(\E,\eta)$ is defined by quadratic relations. To 

864: prove a more exact assertion (for the case in which $\eta$ is a point of infinite 

865: order), one must continue the above calculation. We obtain the following formula: 

866: if $f\in F_\alpha$, then 

867: \begin{equation}

868: \begin{aligned}

869: \phi_p(f)&=\sum_{\begin{subarray}{c}i_1,\dots,i_p\ge0,\\ i_1+\ldots+i_p=\alpha\end{subarray}}

870: B_{i_1,\dots,i_p}f(u_1,u_1+n\eta,\dots,u_1+(i_1-1)n\eta,\\

871: &\qquad\qquad\qquad\qquad u_2,u_2+n\eta,\dots,u_2+(i_2-1)n\eta,\dots)

872: \xi_1^{i_1}\dots\xi_p^{i_p},\\

873: &\text{where}\quad

874: B_{i_1,\dots,i_p}=\prod_{\begin{subarray}{c}1\le\lambda\le\lambda'\le

875: p\\0\le\mu<i_\lambda\\0\le\mu'<i_{\lambda'}\\\mu<\mu'\    \text{for}\

876: \lambda=\lambda'\end{subarray}}

877: \frac{\theta(u_\lambda+n\mu\eta-u_{\lambda'}-n\mu'\eta)}

878: {\theta(u_\lambda+n\mu\eta-u_{\lambda'}-n\mu'\eta-n\eta)}.

879: \end{aligned}

880: \end{equation}

881: This product can be represented as $\prod_{1\le

882: i<j<p}\frac{\theta(v_i-v_j)}{\theta(v_i-v_j-n\eta)}$, where 

883: $(v_1,\dots,v_p)=(u_1,u_1+n\eta,\dots)$ are the arguments of the function $f$ in 

884: the formula (14) for $\xi_1^{i_1}\dots\xi_p^{i_p}$.

885: 

886: The formula (14) makes sense if $\eta$ is a point of infinite order, and in 

887: this case the direct calculation shows that $\phi_p$ is a homomorphism.

888: \end{proof}

889: 

890: \subsection{Representations of the algebras $Q_n(\E,\eta)$}

891: 

892: The formula (14) shows that the image of the homomorphism $\phi_p$ is contained 

893: in the subalgebra $B_{p,n}^{\reg}(\eta)\subset  B_{p,n}(\eta)$ consisting of the 

894: elements $\sum_{\alpha_1,\dots,\alpha_p}f_{\alpha_1,\dots,\alpha_p}

895: \e^{\alpha_1}\dots\e_p^{\alpha_p}$, where the functions $f_{\alpha_1,\dots,\alpha_p}$ 

896: are holomorphic outside the divisors of the form 

897: $u_i-u_j-\lambda n\eta\in\Gamma$, $\lambda\in\Z$. 

898: 

899: Let $v_1,\dots,v_p\in\C$ be such that $v_i-v_j-\lambda

900: n\eta\not\in\Gamma$ for $\lambda\in\Z$. We construct a representation  

901: $M_{v_1,\dots,v_p}$ of the algebra $B_{p,n}^{\reg}(\eta)$ as follows. Let the 

902: representation $M_{v_1,\dots,v_p}$ have a basis 

903: $\{w_{\alpha_1,\dots,\alpha_p};\alpha_1,\dots\alpha_p\in\Z_{\ge0}\}$ in which the 

904: elements $\e_1,\dots,\e_p$ act by the rule $\e_iw_{\alpha_1,\dots,\alpha_p}=

905: w_{\alpha_1,\dots,\alpha_i+1,\dots,\alpha_p}$. Thus, 

906: $w_{\alpha_1,\dots,\alpha_p}=\e_1^{\alpha_1}\dots\e_p^{\alpha_p}w$, where 

907: $w=w_{0,\dots,0}$. The action of the commutative subalgebra of $B_{p,n}^{\reg}(\eta)$ 

908: consisting of the elements of degree $0$ is diagonal in this basis. We set 

909: $fw=f(v_1-(n-2)\eta,\dots,v_p-(n-2)\eta)w$, and hence $fw_{\alpha_1,\dots,\alpha_p}=

910: f(v_1+(2\alpha_1+\ldots+2\alpha_p-n\alpha_1-(n-2))\eta,\dots,

911: v_p+(2\alpha_1+\ldots+2\alpha_p-n\alpha_p-(n-2))\eta)

912: w_{\alpha_1,\dots,\alpha_p}$. It is clear that these formulas really define a 

913: representation of the algebra $B_{p,n}^{\reg}(\eta)$ in the space $M_{v_1,\dots,v_p}$, 

914: and, thanks to the homomorphism $\phi_p$, we have a representation of the algebra 

915: $Q_n(\E,\eta)$ as well. One can readily see that the space $M_{v_1,\dots,v_p}$ 

916: admits a basis $\{v_{\alpha_1,\dots,\alpha_p};\alpha_1,\dots,\alpha_p\in\Z_{\ge0}\}$, 

917: in which the action of the generators of the algebra $Q_n(\E,\eta)$ can be 

918: represented in the following form: if $f\in\Theta_n(\Gamma)$, then 

919: \begin{equation}

920: fv_{\alpha_1,\dots,\alpha_p}=\sum_{1\le                          i\le

921: p}\frac{f(v_i+(2\alpha_1+\ldots+2\alpha_p-n\alpha_i)\eta)}

922: {\theta(v_i-v_1-n(\alpha_i-\alpha_1)\eta)\dots

923: \theta(v_i-v_p-n(\alpha_i-\alpha_p)\eta)}

924: v_{\alpha_1,\dots,\alpha_i+1,\dots,\alpha_p}.

925: \end{equation}

926: The vectors $v_{\alpha_1,\dots,\alpha_p}$ are proportional to the vectors 

927: $w_{\alpha_1,\dots,\alpha_p}$. In particular, for $p=1$ we obtain modules $M_v$ 

928: with a basis $\{v_\alpha;\alpha\in\Z_{\ge0}\}$ and the action 

929: $fv_\alpha=f(v-(n-2)\alpha\eta)v_{\alpha+1}$. Thus, the algebra $Q_n(\E,\eta)$ 

930: has a family of linear modules parametrized by the elliptic curve $\E\subset\CP^{n-1}$, 

931: where the embedding is carried out by theta functions of order $n$.

932: 

933: 

934: \subsection{Symplectic leaves}

935: 

936: We recall that $Q_n(\E,0)$ is the polynomial ring $S^*\Theta_{n,c}(\Gamma)$. For 

937: a fixed elliptic curve $\E=\C/\Gamma$ we obtain the family of algebras $Q_n(\E,\eta)$, 

938: which is a flat deformation of the polynomial ring. We denote the corresponding 

939: Poisson algebra by $q_n(\E)$. We obtain a family of Poisson algebras, depending 

940: on $\E$, that is, on the modular parameter $\tau$. Let us study the symplectic 

941: leaves of this algebra. To this end, we note that, when passing to the limit as 

942: $\eta\to0$, the homomorphism $\phi_p$ of associative algebras gives a homomorphism 

943: of Poisson algebras. Namely, let us denote by $b_{p,n}$ the Poisson algebra formed 

944: by the elements 

945: $\sum_{\alpha_1,\dots,\alpha_p\ge0}

946: f_{\alpha_1,\dots,\alpha_p}(u_1,\dots,u_p)

947: \e_1^{\alpha_1}\dots\e_p^{\alpha_p}$, where $f_{\alpha_1,\dots,\alpha_p}$ are 

948: meromorphic functions and the Poisson bracket is 

949: $$

950: \{u_\alpha,u_\beta\}=\{\e_\alpha,\e_\beta\}=0;\quad

951: \{\e_\alpha,u_\beta\}=-2\e_\alpha;\quad

952: \{\e_\alpha,u_\alpha\}=(n-2)\e_\alpha,

953: $$

954: where $\alpha\ne\beta$.

955: 

956: The following assertion results from Proposition 6 in the limit as $\eta\to0$.

957: 

958: \begin{prop}

959: There is a Poisson algebra homomorphism $\psi_p\colon

960: q_n(\E)\to b_{p,n}$ given by the following formula: if $f\in\Theta_n(\Gamma)$, 

961: then

962: $\psi_p(f)=\sum_{1\le\alpha\le

963: p}\frac{f(u_\alpha)}{\theta(u_\alpha-u_1)\dots\theta(u_\alpha-u_p)}

964: \e_\alpha$.

965: \end{prop}

966: 

967: Let $\{\theta_i(u);i\in\Z/n\Z\}$ be a basis of the space 

968: $\Theta_{n,c}(\Gamma)$ and let $\{x_i;i\in\Z/n\Z\}$ be the corresponding basis 

969: in the space of elements of degree one in the algebra $Q_n(\E,\eta)$ (this space 

970: is isomorphic to $\Theta_{n,c}(\Gamma)$). For an elliptic curve $\E\subset\CP^{n-1}$ 

971: embedded by means of theta functions of order $n$ (this is the set of points with 

972: the coordinates $(\theta_0(z):\ldots:\theta_{n-1}(z))$) we denote by $C_p\E$ the 

973: variety of $p$-chords, that is, the union of projective spaces of dimension $p-1$ 

974: passing through $p$ points of $\E$. Let $K(C_p\E)$ be the corresponding homogeneous 

975: manifold in $\C^n$. It is clear that $K(C_p\E)$ consists of the points with the 

976: coordinates 

977: $x_i=\sum_{1\le\alpha\le

978: p}\frac{\theta_i(u_\alpha)}{\theta(u_\alpha-u_1)\dots

979: \theta(u_\alpha-u_p)}\e_\alpha$, where $u_\alpha,\e_\alpha\in\C$.

980: 

981: Let $2p<n$. Then one can show that $\dim K(C_p\E)=2p$ and $K(C_{p-1}\E)$ is the 

982: manifold of singularities of $K(C_p\E)$. It follows from Proposition 7 and from 

983: the fact that the Poisson bracket is non-degenerate on $b_{p,n}$ for $2p<n$ and 

984: $\e_\alpha\ne0$ that the non-singular part of the manifold $K(C_p\E)$ is a $2p$-

985: dimensional symplectic leaf of the Poisson algebra $q_n(\E)$.

986: 

987: Let $n$ be odd. One can show that the equation defining the manifold $K(C_{\frac{n-1}2}\E)$ 

988: is of the form $C=0$, where $C$ is a homogeneous polynomial of degree $n$ in the 

989: variables $x_i$. This polynomial is a central function of the algebra $q_n(\E)$. 

990: 

991: Let $n$ be even. The manifold $K(C_{\frac{n-2}2}\E)$ is defined by equations 

992: $C_1=0$ and $C_2=0$, where $\deg C_1=\deg C_2=n/2$. The polynomials $C_1$ and $C_2$ 

993: are central in the algebra $q_n(\E)$.

994: 

995: \subsection{Free modules, generations, and relations}

996: 

997: Let $\eta$ be a point of infinite order.

998: 

999: \begin{prop}

1000: Let numbers $v_1,\dots,v_n\in\C$ be in general position. Then the module  

1001: $M_{v_1,\dots,v_n}$ is generated by $v_{0,\dots,0}$ and is free over $Q_n(\E,\eta)$.

1002: \end{prop}

1003: 

1004: \begin{proof}

1005: By construction, the dimensions of graded components of $M_{v_1,\dots,v_n}$ 

1006: coincide with those of the algebra $Q_n(\E,\eta)$. Let us show that the module is 

1007: generated by the vector $v=v_{0,\dots,0}$. Let 

1008:  

1009: $$f_i=\prod_{\alpha\ne

1010: i}\theta(z-v_\alpha)\cdot(\theta(z+v_1+\ldots+v_n-v_i-c).$$

1011: It is clear that $f_i\in\Theta_{n,c}(\Gamma)$ for $1\le i\le n$. Therefore, the 

1012: $f_i$s are elements of degree $1$ of the algebra $Q_n(\E,\eta)$. It follows from 

1013: the formula (15) that $f_iv$ is non-zero and proportional to 

1014: $v_{0,\dots,1,\dots,0}=\e_iv$. Similarly, one can readily construct elements 

1015: $f_{i;\alpha_1,\dots,\alpha_n}\in\Theta_{n,c}(\Gamma)$ such that 

1016: $f_{i;\alpha_1,\dots,\alpha_n}v_{\alpha_1,\dots,\alpha_n}$ is non-zero and 

1017: proportional to $v_{\alpha_1,\dots,\alpha_i+1,\dots,\alpha_n}$. Namely,

1018: $f_{i;\alpha_1,\dots,\alpha_n}=\prod_{\beta\ne

1019: i}\theta(z-v_\beta-(2\alpha_1+\ldots+2\alpha_n-n\alpha_\beta)\eta)

1020: \cdot\theta(z+v_1+\ldots+v_n-v_i+(n-2)(\alpha_1+\ldots+\alpha_n)\eta

1021: -c)$.

1022: Thus, all elements $v_{\alpha_1,\dots,\alpha_n}$ are obtained from $v$ by the action 

1023: of elements of degree one in $Q_n(\E,\eta)$.

1024: \end{proof}

1025: 

1026: \begin{prop}

1027: The algebra $Q_n(\E,\eta)$ is presented by $n$ generators and $\frac{n(n-1)}2$ 

1028: quadratic relations.

1029: \end{prop}

1030: 

1031: \begin{proof}

1032: It follows from the proof of Proposition 8 that the algebra $Q_n(\E,\eta)$ is generated 

1033: by the elements of degree one. It is clear from the construction of the elements 

1034: $f_{i;\alpha_1,\dots,\alpha_n}$ that these elements admit quadratic relations of 

1035: the form

1036: \begin{equation}

1037: f_{j;\alpha_1,\dots,\alpha_i+1,\dots,\alpha_n}

1038: f_{i;\alpha_1,\dots,\alpha_n}=

1039: c_{i,j;\alpha_1,\dots,\alpha_n}

1040: f_{i;\alpha_1,\dots,\alpha_j+1,\dots,\alpha_n}

1041: f_{j;\alpha_1,\dots,\alpha_n},

1042: \end{equation}

1043: where $c_{i,j;\alpha_1,\dots,\alpha_n}\in\C^*$. To prove this relation, one must 

1044: apply it to the vector $v_{\alpha_1,\dots,\alpha_n}$. Let us show that these 

1045: quadratic relations imply the other ones. Let a relation be of the form 

1046: $\sum_\alpha a_t^{(\alpha)}a_{t-1}^{(\alpha)}\dots a_1^{(\alpha)}=0$. We expand 

1047: the element $a_1^{(\alpha)}$ in the basis $\{f_i\}$. The relation becomes 

1048: $\sum_{\beta,i}b_t^{(\beta)}\dots b_2^{(\beta)}f_i=0$. Let us now expand $b_2^{(\beta)}$ 

1049: in the basis $\{f_{i;0,\dots,1,\dots,0}\}$, where $1$ stands at the $i$th place. 

1050: Continuing this procedure, we eventually represent the relation in the form $\sum

1051: c_{i_1,\dots,i_t}f_{i_1;\alpha_1,\dots,\alpha_n}

1052: f_{i_2;\alpha_1,\dots,\alpha_{i_2}-1,\dots,\alpha_n}\dots f_{i_t}=0$. It is clear 

1053: that this relation follows from the relations (16).

1054: \end{proof}

1055: 

1056: \begin{prop}

1057: The relations in the algebra $Q_n(\E,\eta)$ are of the form 

1058: \begin{equation}

1059: \sum_{r\in\Z/n\Z}\frac1{\theta_{j-i-r}(-\eta)\theta_r(\eta)}x_{j-r}

1060: x_{i+r}=0,\quad\text{ÓÔÕ}\ i\ne j;\ i,j\in\Z/n\Z.

1061: \end{equation}

1062: \end{prop}

1063: 

1064: \begin{proof}

1065: Let us apply the formula for the multiplication in the algebra $Q_n(\E,\eta)$ 

1066: (see \S2.1). Since $x_i=\theta_i(z)$, the relations (17) becomes 

1067: \begin{multline*}

1068: \sum_{r\in\Z/n\Z}\frac1{\theta_{j-i-r}(-\eta)\theta_r(\eta)}

1069: \biggl(\theta_{j-r}(z_1)\theta_{i+r}(z_2-2\eta)

1070: \frac{\theta(z_1-z_2-n\eta)}{\theta(z_1-z_2)}+\\

1071: +\theta_{j-r}(z_2)\theta_{i+r}(z_1-2\eta)

1072: \frac{\theta(z_2-z_1-n\eta)}{\theta(z_2-z_1)}\biggr)=0.

1073: \end{multline*}

1074: This relation immediately follows from the relation (30) (see Appendix A).

1075: \end{proof}

1076: 

1077: \section{Main properties of the algebra $Q_{n,k}(\E,\eta)$}

1078: 

1079: We again assume that $\E=\C/\Gamma$ is an elliptic curve and $\eta\in\E$. Let 

1080: $n$ and $k$ be coprime positive integers such that $1\le k<n$. Let us present 

1081: the algebra $Q_{n,k}(\E,\eta)$ by the generators $\{x_i;i\in\Z/n\Z\}$ and the 

1082: relations 

1083: \begin{equation}

1084: \sum_{r\in\Z/n\Z}\frac{\theta_{j-i+r(k-1)}(0)}

1085: {\theta_{kr}(\eta)\theta_{j-i-r}(-\eta)}x_{j-r}x_{i+r}=0.

1086: \end{equation}

1087: As is known (see \S4), this is a PBW-algebra for generic $\E$ and $\eta$. We 

1088: conjecture that this holds for any $\E$ and $\eta$. For generic $\E$ and $\eta$ 

1089: the centre of the algebra $Q_{n,k}(\E,\eta)$ is the polynomial ring in $c=\NOD(n,k+1)$ 

1090: elements of degree $n/c$ (see [20]). Hypothetically, this is the case for any 

1091: $\E$ and $\eta$, where $\eta$ is a point of infinite order. If $\eta\in\E$ is a 

1092: point of finite order, then the algebra $Q_{n,k}(\E,\eta)$ is finite-dimensional 

1093: over its centre (see [37]). The following properties can readily be verified:

1094: 

1095: \smallskip

1096: 

1097: 1) $Q_{n,k}(\E,0)=\C[x_1,\dots,x_n]$ is commutative;

1098: 

1099: 2) $Q_{n,n-1}(\E,\eta)=\C[x_1,\dots,x_n]$ is commutative for any $\eta$;

1100: 

1101: 3) $Q_{n,k}(\E,\eta)\simeq    Q_{n,k'}(\E,\eta)$, where $kk'\equiv1\pmod n$;

1102: 

1103: 4) the maps $x_i\mapsto x_{i+1}$ and $x_i\mapsto\epsilon^ix_i$ (where $\epsilon$ 

1104: is a primitive root of unity of degree $n$) define automorphisms of the algebra 

1105: $Q_{n,k}(\E,\eta)$.

1106: 

1107: \smallskip

1108: 

1109: It follows from the results of \S5 (see Proposition 11) that the space of generators 

1110: of the algebra $Q_{n,k}(\E,\eta)$ is naturally isomorphic to the space of theta 

1111: functions $\Theta_{n/k}(\Gamma)$ (see Appendix B). Moreover, this space of generators 

1112: is dual to the space $\Theta_{n/n-k}(\Gamma)$  (see Proposition 14). For a 

1113: description of this duality between the spaces of theta functions, see Appendix C.

1114: 

1115: The algebra $Q_{n,k}(\E,\eta)$ is not a Hopf algebra and admits no comultiplications. 

1116: However, there are homomorphisms of the algebra $Q_{n,k}(\E,\eta)$ to tensor 

1117: products of other algebras of this kind (see [35; \S3]). To describe these 

1118: homomorphisms, we need the notation of Appendix B. Moreover, we denote by 

1119: $L_m(\E,\eta)=\C\oplus\Theta_{m,0}(\Gamma)\oplus

1120: \Theta_{2m,m\eta}(\Gamma)\oplus\dots$ the $\Z_{\ge0}$-graded algebra with the 

1121: multiplication $*$ given by the formula $f*g(z)=f(z+\beta\eta)g(z)$, where $\beta$ 

1122: is a power of $g$. As we know from \S2, $L_n(\E,(n-2)\eta)$ is a quotient algebra 

1123: of $Q_n(\E,\eta)$.

1124: 

1125: Let $A$ be an associative algebra and let $G\subset\Aut  A$. We denote by 

1126: $A^G\subset  A$ the subalgebra consisting of the elements invariant with respect to 

1127: $G$.

1128: 

1129: There are the following algebra homomorphisms:

1130: 

1131: \smallskip

1132: 

1133: a)

1134: $Q_{n,k}(\E,\eta)\to

1135: \left(L_{kn}\left(\E,\frac{n-k-1}k\eta\right)\otimes

1136: Q_{k,\l}\left(\E,\frac   nk\eta\right)\right)^{\wt{\Gamma_k}}$,

1137: where $\l=d(n_3,\dots,n_p)$ and the generators are taken to elements of bidegree 

1138: $(1,1)$.

1139: 

1140: b)

1141: $Q_{n,k}(\E,\eta)\to

1142: \left(L_{nk'}\left(\E,\frac{n-k'-1}{k'}\eta\right)\otimes

1143: Q_{k',\l'}\left(\E,\frac

1144: n{k'}\eta\right)\right)^{\wt{\Gamma_{k'}}}$,

1145: where $\l'=d(n_1,\dots,n_{p-2})$ and the generators are taken to elements of bidegree 

1146: $(1,1)$.

1147: 

1148: c)

1149: $Q_{n,k}(\E,\eta)\to

1150: \left(Q_{a,\alpha}\left(\E,\frac n\alpha\eta\right)\otimes

1151: L_{abn}\left(\E,\frac{n-a-b}{ab}\eta\right)\otimes

1152: Q_{b,\beta}\left(\E,\frac

1153: nb\right)\right)^{\wt{\Gamma_{ab}}}$,

1154: where $a=d(n_1,\dots,n_{i-1})$, $b=d(n_{i+1},\dots,n_p)$, $\alpha=d(n_1,\dots,n_{i-2})$, and  

1155: $\beta=d(n_{i+2},\dots,n_p)$ for some $i$; the generators are taken to elements 

1156: of multidegree $(1,1,1)$.

1157: 

1158: \smallskip

1159: 

1160: Let us describe the map c) geometrically (the description of the maps a) and b) 

1161: is the same). Let $f(z_1,\dots,z_p)\in\Theta_{n/k}(\Gamma)$. For some $i$  ($1<i<p$) 

1162: we choose a $z_i$, then $f$ (regarded as a function of $z_1,\dots,z_{i-1}$) 

1163: belongs to a space isomorphic to $\Theta_{a/\alpha}(\Gamma)$. Similarly, when 

1164: regarded as a function of $z_{i+1},\dots,z_p$, $f$ belongs to a space isomorphic to 

1165: $\Theta_{b/\beta}(\Gamma)$. Thus, for a fixed $z_i$ we have 

1166: $f\in\Theta_{a/\alpha}(\Gamma)\otimes\Theta_{b/\beta}(\Gamma)$. A family of linear 

1167: maps $\Theta_{n/k}(\Gamma)\to\Theta_{a/\alpha}(\Gamma)\otimes

1168: \Theta_{b/\beta}(\Gamma)$ arises. With regard to the dependence on $z_i$, we 

1169: obtain a linear map $\Theta_{n/k}(\Gamma)\to\Theta_{a/\alpha}(\Gamma)\otimes

1170: \Theta_{nab,0}(\Gamma)\otimes\Theta_{b/\beta}(\Gamma)$. The homomorphism c) corresponds 

1171: to this map (the space of generators of the algebra $Q_{n,k}(\E,\eta)$ is 

1172: $\Theta_{n/k}(\Gamma)$, the space of generators of the algebra 

1173: $L_{nab}\left(\E,\frac{n-a-b}{ab}\eta\right)$ is $\Theta_{nab,0}(\Gamma)$, etc.).

1174: 

1175: \section{Belavin elliptic $R$-matrix and the algebra $Q_{n,k}(\E,\eta)$}

1176: 

1177: Let $V$ be a vector space of dimension $n$. For each $u\in\C$ we denote by $V(u)$ 

1178: a vector space canonically isomorphic to $V$. Let $R$ be a meromorphic function 

1179: of two variables with values in $\End(V\otimes V)$. It is convenient to regard 

1180: $R(u,v)$ as a linear map 

1181: $$

1182: R(u,v)\colon V(u)\otimes V(v)\to V(v)\otimes V(u).

1183: $$

1184: 

1185: We recall that by the Yang-Baxter equation one means the condition that the 

1186: following diagram is commutative:

1187: 

1188: {\footnotesize

1189: 

1190: $$

1191: \xymatrix{&V(v)\otimes V(u)\otimes V(w)\ar[r]^{1\otimes R(u,w)}&

1192: V(v)\otimes V(w)\otimes V(u)\ar[dr]^{R(v,w)\otimes1}\\

1193: V(u)\otimes V(v)\otimes V(w)\ar[ur]^{R(u,v)\otimes1}\ar[dr]^{1\otimes

1194: R(v,w)}&&&

1195: V(w)\otimes V(v)\otimes V(u)\\

1196: &V(u)\otimes V(w)\otimes V(v)\ar[r]^{R(u,w)\otimes1}&

1197: V(w)\otimes V(u)\otimes V(v)\ar[ur]_{1\otimes R(u,v)}}

1198: $$}

1199: 

1200: A solution of the Yang-Baxter equation is called $R$-matrix.

1201: 

1202: Let $\{x_i;i=1,\dots,n\}$ be a basis in the space $V$ and let $\{x_i(u)\}$ be the 

1203: corresponding basis in the space $V(u)$. Let $R_{ij}^{\alpha\beta}(u,v)$ be an 

1204: entry of an $R$-matrix $R(u,v)$, that is, $R(u,v)\colon     x_i(u)\otimes     x_j(v)\to

1205: R_{ij}^{\alpha\beta}(u,v)x_\beta(v)\otimes x_\alpha(u)$.

1206: 

1207: Let an $R$-matrix $R(u,v)$ satisfy the relation $R(u,v)R(v,u)=1$. By the 

1208: Zamolodchikov algebra $Z_R$ one means the algebra with the generators 

1209: $\{x_i(u);i=1,\dots,n;u\in\C\}$ and the defining relations 

1210: \begin{equation}

1211: x_i(u)x_j(v)=R_{ij}^{\alpha\beta}(u,v)x_\beta(v)x_\alpha(u).

1212: \end{equation}

1213: It is clear that the elements $\{x_{i_1}(u_1)\dots

1214: x_{i_m}(u_m);1\le i_1,\dots,i_m\le n\}$ of the Zamolodchikov algebra are linearly 

1215: independent for generic $u_1,\dots,u_m$. Thus, Zamolodchikov algebras are 

1216: infinite-dimensional analogues of PBW-algebras. We recall that by the classical 

1217: $r$-matrix one means a Poisson structure of the form 

1218: $\{x_i(u),x_j(v)\}=r_{ij}^{\alpha\beta}(u,v)x_\alpha(u)x_\beta(v)$. It is clear 

1219: that the Zamolodchikov algebra is a quantization of this Poisson structure if the 

1220: $R$-matrix depends on an additional parameter $\hbar$ and the relations (19) are 

1221: of the form 

1222: $$

1223: x_i(u)x_j(v)=x_j(v)x_i(u)+\hbar

1224: r_{ij}^{\alpha\beta}(u,v)x_\beta(v)x_\alpha(u)+o(\hbar).

1225: $$

1226: 

1227: The Yang-Baxter equation has elliptic solutions, which are referred to as Belavin 

1228: $R$-matrices. Let $n,k\in\N$ be coprime and $1\le k<n$. For any $n$ and $k$ 

1229: there is a two-parameter family of $R$-matrices $R_{n,k}(\E,\eta)$ depending on 

1230: an elliptic curve $\E=\C/\Gamma$ and a point $\eta\in\E$. Namely,

1231: $$

1232: R_{n,k}(\E,\eta)(u-v)                       (x_i(u)\otimes

1233: x_j(v))=\frac1{p(u-v)}\sum_{r\in\Z/n\Z}

1234: \frac{\theta_{j-i+r(k-1)}(v-u+\eta)}

1235: {\theta_{kr}(\eta)\theta_{j-i-r}(v-u)}x_{j-r}(v)\otimes x_{i+r}(u),

1236: $$

1237: where $p(u-v)=

1238: \frac{\theta_1(0)\dots\theta_{n-1}(0)

1239: \theta_0(v-u+\eta)\dots\theta_{n-1}(v-u+\eta)}

1240: {\theta_0(\eta)\dots\theta_{n-1}(\eta)

1241: \theta_0(v-u)\dots\theta_{n-1}(v-u)}$;     $i,j\in\Z/n\Z$.

1242: One can readily see that $\det

1243: R_{n,k}(\E,\eta)(u-v)=

1244: \left(\frac{\theta_0(v-u-\eta)\dots\theta_{n-1}(v-u-\eta)}

1245: {\theta_0(v-u+\eta)\dots\theta_{n-1}(v-u+\eta)}\right)

1246: ^{\frac{n(n-1)}2}$. 

1247: Thus, the operator $R_{n,k}(\E,\eta)(-\eta)$ has a kernel. Let $L_{n,k}(\E,\eta)\subset

1248: V\otimes  V$ and $L_{n,k}(\E,\eta)=\ker R_{n,k}(\E,\eta)(-\eta)$. According to [10] 

1249: we have $\dim(L_{n,k}(\E,\eta))=\frac{n(n-1)}2$, and $L_{n,k}(\E,0)=\Lambda^2V$ 

1250: for $\eta=0$. Let $Q_{n,k}(\E,\eta)=T^*V/(L_{n,k}(\E,\eta))$ be the algebra with 

1251: the generators $\{x_i;i\in\Z/n\Z\}$ and the defining relations $L_{n,k}(\E,\eta)$. 

1252: The dimensions of the graded components of the algebra $Q_{n,k}(\E,\eta)$ coincide 

1253: with those of the polynomial ring $S^*V$ (see [10]). It follows from the formula 

1254: for $R_{n,k}(\E,\eta)(u-v)$ that the defining relations in the algebra $Q_{n,k}(\E,\eta)$ 

1255: are 

1256: \begin{equation}

1257: \sum_{r\in\Z/n\Z}\frac{\theta_{j-i+r(k-1)}(0)}

1258: {\theta_{kr}(\eta)\theta_{j-i-r}(-\eta)}x_{j-r}x_{i+r}=0.

1259: \end{equation}

1260: In particular, we see that $Q_{n,1}(\E,\eta)=Q_n(\E,\eta)$ if $\eta\in\E$ is a 

1261: point of infinite order.

1262: 

1263: It follows from the relations (20) that $Q_{n,k}(\E,\eta)\simeq

1264: Q_{n,k'}(\E,\eta)$, where $kk'\equiv1\pmod  n$. Moreover, $Q_{n,n-1}(\E,\eta)$ is 

1265: commutative for any $\E,\eta$. It is also clear that $Q_{n,k}(\E,0)$ is 

1266: commutative.

1267: 

1268: \section{Algebras $Q_{n,k}(\E,\eta)$ and the exchange algebras}

1269: 

1270: \subsection{Homomorphisms of algebras $Q_{n,k}(\E,\eta)$ into dynamical exchange 

1271: algebras}

1272: 

1273: The algebras $Q_{n,k}(\E,\eta)$ for arbitrary $k$ have representations similar to 

1274: the homomorphisms $\phi_p$ related to the case $k=1$ (see \ \S2.3,  (11)). 

1275: However, the structure of the algebra similar to $B_{p,n}(\eta)$ for $k=1$ turns 

1276: out to be more complicated for $k>1$. Let $n/k=n_1-\frac1{n_2-\ldots-\frac1{n_q}}$ 

1277: be the expansion of the number $n/k$ in the continued fraction in which $n_\alpha\ge2$ 

1278: for $1\le\alpha\le q$. It is clear that such an expansion exists and is unique. 

1279: We recall that $1\le    k<n$, where $n$ and $k$ are coprime. Let 

1280: $d(m_1,\dots,m_t)=\det M$, where $M=(m_{ij})$ be a ($t\times t$) matrix with the 

1281: entries $m_{ii}=m_i$,  $m_{i,i+1}=m_{i+1,i}=-1$, and $m_{ij}=0$ for $|i-j|>1$. 

1282: For $t=0$ we set $d(\varnothing)=1$. It is clear that 

1283: $n=d(n_1,\dots,n_q)$ and $k=d(n_2,\dots,n_q)$.

1284: 

1285: \begin{prop}

1286: There is an algebra homomorphism of $Q_{n,k}(\E,\eta)$ into the algebra $C_{n,k}(\eta)$ 

1287: generated by the commutative subalgebra $\{f(y_1,\dots,y_q)\}$ of holomorphic 

1288: functions of $q$ variables and by an element $\e$ with the defining relations of 

1289: the form $\e f(y_1,\dots,y_q)=f(y_1+\eta_1,\dots,y_q+\eta_q)\e$, where  

1290: $\eta_\alpha=(d(n_1,\dots,n_q)-d(n_1,\dots,n_{\alpha-1})-

1291: d(n_{\alpha+1},\dots,n_q))\eta$.  Moreover, 

1292: $x_i\to w_i(y_1,\dots,y_q)\e$, where

1293: $w_i\in\Theta_{n/k}(\Gamma)$ \emph(see Appendix B).

1294: \end{prop}

1295: 

1296: This is a special case of Proposition 13 below.

1297: 

1298: The algebra $C_{n,k}(\eta)$ has a family of modules $L_{v_1,\dots,v_q}$ with a 

1299: basis $\{v_\alpha;\alpha\in\Z_{\ge0}\}$ and with the action given by $\e

1300: v_\alpha=v_{\alpha+1}$ and

1301: $f(y_1,\dots,y_q)v_\alpha=

1302: f(v_1-\alpha\eta_1,\dots,v_q-\alpha\eta_q)v_\alpha$. This implies the following 

1303: assertion.

1304: 

1305: \begin{prop}

1306: The algebra $Q_{n,k}(\E,\eta)$ hes a family of modules $L_{u_1,\dots,u_q}$ with 

1307: a basis $\{v_\alpha;\alpha\in  Z_{\ge0}\}$ and with the action \emph:

1308: $x_iv_\alpha=w_i(u_1-\alpha\eta_1,\dots,u_q-\alpha\eta_q)

1309: v_{\alpha+1}$, where $w_i\in\Theta_{n/k}(\Gamma)$      and

1310: $\eta_j=(d(n_1,\dots,n_q)-d(n_1,\dots,n_{j-1})-

1311: d(n_{j+1},\dots,n_q))\eta$.

1312: \end{prop}

1313: 

1314: In particular, we see that the algebra $Q_{n,k}(\E,\eta)$ has a family of linear 

1315: modules depending on the point of $\E^q$, and the space of generators of the 

1316: algebra $Q_{n,k}(\E,\eta)$ is isomorphic to $\Theta_{n/k}(\Gamma)$.

1317: 

1318: Let $C_{m_1,\dots,m_q;n,k}(\E,\eta)$ be the algebra generated by the commutative 

1319: subalgebra $\{\phi(y_{1,1},\dots,y_{m_1,1};\dots;y_{1,q},\dots,y_{n_q,q})\}$, 

1320: where $\phi$ are the meromorphic functions with the ordinary multiplication, and 

1321: by elements $\{\e_{\alpha_1,\dots,\alpha_q};1\le\alpha_t\le m_t\}$. The defining 

1322: relations for the algebra $C_{m_1,\dots,m_q;n,k}(\E,\eta)$ look as follows:

1323: \begin{align*}

1324: \e_{\alpha_1,\dots,\alpha_q}y_{\beta,\nu}&=

1325: (y_{\beta,\nu}-(d(n_1,\dots,n_{\nu-1})+d(n_{\nu+1},\dots,n_q))\eta)

1326: \e_{\alpha_1,\dots,\alpha_q},\quad\ \alpha_\nu\ne\beta,\\

1327: \e_{\alpha_1,\dots,\alpha_q}y_{\alpha_\nu,\nu}&=

1328: (y_{\alpha_\nu,\nu}+(n-d(n_1,\dots,n_{\nu-1})-d(n_{\nu+1},\dots,n_q))

1329: \eta)\e_{\alpha_1,\dots,\alpha_q}.

1330: \end{align*}

1331: These relations mean that the $y_{\beta,\nu}$s are dynamical variables. This 

1332: immediately implies the relations between the elements $\e_{\alpha_1,\dots,\alpha_q}$ 

1333: and the meromorphic functions of the variables $y_{\beta,\nu}$. The remaining 

1334: relations are quadratic in $\e_{\alpha_1,\dots,\alpha_q}$ with the coefficients 

1335: depending on the dynamical variables $y_{\beta,\nu}$. The relations in "general 

1336: position" are as follows:

1337: \begin{equation}

1338: \e_{\alpha_1,\dots,\alpha_q}\e_{\beta_1,\dots,\beta_q}=

1339: \Lambda\e_{\beta_1,\dots,\beta_q}\e_{\alpha_1,\dots,\alpha_q}+

1340: \sum_{1\le t\le q-1}\Lambda_{t,t+1}

1341: \e_{\beta_1,\dots,\beta_t,\alpha_{t+1},\dots,\alpha_q}

1342: \e_{\alpha_1,\dots,\alpha_t,\beta_{t+1},\dots,\beta_q},

1343: \end{equation}

1344: where $\alpha_1\ne\beta_1$, \dots, $\alpha_q\ne\beta_q$ and

1345: $$\begin{aligned}

1346: \Lambda&=\frac{e^{-2\pi in\eta}\theta(y_{\beta_1,1}-y_{\alpha_1,1})

1347: \theta(y_{\beta_q,q}-y_{\alpha_q,q}+n\eta)}

1348: {\theta(y_{\beta_1,1}-y_{\alpha_1,1}-n\eta)

1349: \theta(y_{\beta_q,q}-y_{\alpha_q,q})},\\

1350: \Lambda_{t,t+1}&=\frac{e^{-2\pi in\eta}\theta(n\eta)

1351: \theta(y_{\beta_1,1}-y_{\alpha_1,1})}

1352: {\theta(y_{\beta_1,1}-y_{\alpha_1,1}-n\eta)}\cdot

1353: \frac{\theta(y_{\beta_t,t}+y_{\beta_{t+1},t+1}-

1354: y_{\alpha_t,t}-y_{\alpha_{t+1},t+1})}

1355: {\theta(y_{\beta_t,t}-y_{\alpha_t,t})

1356: \theta(y_{\beta_{t+1},t+1}-y_{\alpha_{t+1},t+1})}.

1357: \end{aligned}$$

1358: The relations of non-general position occur if some $\alpha_\nu$s are equal to  

1359: $\beta_\nu$s. These relations exist for any subset of the form 

1360: $\{\psi+1,\dots,\psi+\phi\}$, where $0\le\psi$, $\psi+\phi\le q$, and 

1361: $\alpha_\psi=\beta_\psi$               (if               $0<\psi$),

1362: $\alpha_{\psi+\phi+1}=\beta_{\psi+\phi+1}$  (if $\psi+\phi<q$), and

1363: $\alpha_{\psi+1}\ne\beta_{\psi+1}$,                            \dots,

1364: $\alpha_{\psi+\phi}\ne\beta_{\psi+\phi}$. These relations are of the form

1365: \begin{multline}

1366: \e_{\mu_1,\dots,\mu_\psi,\alpha_1,\dots,\alpha_\phi,

1367: \gamma_1,\dots,\gamma_p}

1368: \e_{\mu_1',\dots,\mu_{\psi-1}',\mu_\psi,\beta_1,\dots,\beta_\phi,

1369: \gamma_1,\gamma_2',\dots,\gamma_p'}=\\

1370: =\Lambda\e_{\mu_1,\dots,\mu_\psi,\beta_1,\dots,\beta_\phi,

1371: \gamma_1,\dots,\gamma_p}

1372: \e_{\mu_1',\dots,\mu_\psi,\alpha_1,\dots,\alpha_\phi,

1373: \gamma_1,\dots,\gamma_p'}+\\

1374: +\sum_{1\le t<\phi}\Lambda_{t,t+1}\e_{\mu_1,\dots,\mu_\psi,

1375: \beta_1,\dots,\beta_t,\alpha_{t+1},\dots,\alpha_\phi,

1376: \gamma_1,\dots,\gamma_p}

1377: \e_{\mu_1',\dots,\mu_\psi,\alpha_1,\dots,\alpha_t,

1378: \beta_{t+1},\dots,\beta_\phi,\gamma_1,\dots,\gamma_p'}.

1379: \end{multline}

1380: Here $\alpha_1\ne\beta_1$,   \dots,    $\alpha_\phi\ne\beta_\phi$, and

1381: $\phi+\psi+p=q$. The coefficients $\Lambda,\Lambda_{t,t+1}$ are defined by (21). 

1382: If $\psi=p=0$, then this relation coincides with (21), that is, becomes a relation 

1383: in general position.

1384: 

1385: We note that, if $\eta=0$, then the algebra $C_{m_1,\dots,m_q;n,k}(\E,0)$ is 

1386: commutative and does not depend on the elliptic curve $\E$. For $q=1$ and $q=2$ 

1387: the algebra $C_{m_1,\dots,m_q;n,k}(\E,0)$ is the polynomial ring in the variables 

1388: $\{\e_{\alpha_1,\dots,\alpha_q}\}$ over the field of meromorphic functions of the 

1389: variables $\{y_{\alpha,\beta}\}$. For $q>2$ the algebra $C_{m_1,\dots,m_q;n,k}(\E,0)$ 

1390: is no longer a polynomial ring because additional relations occur. Namely, the 

1391: relations (22) for $\eta=0$ become 

1392: \begin{multline}

1393: \e_{\mu_1,\dots,\mu_\psi,\alpha_1,\dots,\alpha_\phi,

1394: \gamma_1,\dots,\gamma_p}

1395: \e_{\mu_1',\dots,\mu_{\psi-1}',\mu_\psi,\beta_1,\dots,\beta_\phi,

1396: \gamma_1,\gamma_2',\dots,\gamma_p'}=\\

1397: =\e_{\mu_1,\dots,\mu_\psi,\beta_1,\dots,\beta_\phi,

1398: \gamma_1,\dots,\gamma_p}

1399: \e_{\mu_1',\dots,\mu_{\psi-1}',\mu_\psi,\alpha_1,\dots,\alpha_\phi,

1400: \gamma_1,\gamma_2',\dots,\gamma_p'}.

1401: \end{multline}

1402: The algebra $C_{m_1,\dots,m_q;n,k}(\E,\eta)$ is a flat deformation of the function 

1403: algebra on the manifold defined by the relations (23). Moreover, $y_{\alpha,\beta}$ 

1404: are dynamical variables. It is clear that the manifold defined by the equations 

1405: (23) is rational, and the general solution of these equations is 

1406: $\e_{\alpha_1,\dots,\alpha_q}=\e_{\alpha_1,\alpha_2}^{(1)}

1407: \e_{\alpha_2,\alpha_3}^{(2)}\dots\e_{\alpha_{q-1},\alpha_q}^{(q-1)}$,

1408: where $\{\e_{\alpha,\beta}^{(t)}\}$ are independent variables. 

1409: 

1410: \begin{prop}

1411: There is an algebra homomorphism 

1412: $\phi\colon  Q_{n,k}(\E,\eta)\to

1413: C_{m_1,\dots,m_q;n,k}(\E,\eta)$ that acts on the generators of the algebra 

1414: $Q_{n,k}(\E,\eta)$ as follows: \emph:

1415: \begin{equation}

1416: \phi(x_i)=\sum_{\begin{subarray}{c}1\le\alpha_1\le

1417: m_1\\\dots\\1\le\alpha_q\le m_q\end{subarray}}

1418: w_i(y_{\alpha_1,1},\dots,y_{\alpha_q,q})

1419: \e_{\alpha_1,\dots,\alpha_q}.

1420: \end{equation}

1421: Here $w_i\in\Theta_{n/k}(\Gamma)$.

1422: \end{prop}

1423: 

1424: \begin{proof}

1425: The algebra $Q_{n,k}(\E,\eta)$ is defined by the relations (18). Let us show that 

1426: the images $\phi(x_i)$ satisfy the same relations. We have

1427: \begin{multline*}

1428: \sum_{r\in\Z/n\Z}\frac{\theta_{j-i+r(k-1)}(0)}

1429: {\theta_{kr}(\eta)\theta_{j-i-r}(-\eta)}\phi(x_{j-r})\phi(x_{i+r})=\\

1430: =\sum_{\begin{subarray}{c}r\in\Z/n\Z\\\\1\le\alpha_1,\beta_1\le

1431: m_1\\\dots\\1\le\alpha_q,\beta_q\le m_q\end{subarray}}

1432: \frac{\theta_{j-i+r(k-1)}(0)}{\theta_{kr}(\eta)\theta_{j-i-r}(-\eta)}

1433: w_{j-r}(y_{\alpha_1,1},\dots,y_{\alpha_q,q})

1434: \e_{\alpha_1,\dots,\alpha_q}

1435: w_{i+r}(y_{\beta_1,1},\dots,y_{\beta_q,q})

1436: \e_{\beta_1,\dots,\beta_q}.

1437: \end{multline*}

1438: Using the relations (21) and (22), we obtain an expression of the form 

1439: $$

1440: \sum_{\begin{subarray}{l}\alpha_1\le\beta_1;\\\alpha_2\le\beta_2\

1441: \text{for}\

1442: \alpha_1=\beta_1;\\

1443: \dots\end{subarray}}

1444: \psi_{\alpha_1,\dots,\alpha_q,\beta_1,\dots,\beta_q}

1445: \e_{\alpha_1,\dots,\alpha_q}\e_{\beta_1,\dots,\beta_q}.

1446: $$

1447: We must prove that the coefficients are equal to $0$. Let us restrict ourselves 

1448: to the case $\alpha_1\ne\beta_1$, \dots, $\alpha_q\ne\beta_q$. In this case we 

1449: have by direct calculation

1450: \begin{multline*}

1451: \psi_{\alpha_1,\dots,\beta_q}=\sum_{r\in\Z/n\Z}

1452: \frac{\theta_{j-i+r(k-1)}(0)}{\theta_{kr}(\eta)\theta_{j-i-r}(-\eta)}

1453: \biggl(w_{j-r}(y_{\alpha_1,1},\dots,y_{\alpha_q,q})

1454: w_{i+r}(y_{\beta_1,1}+t_1,\dots,y_{\beta_q,q}+t_q)+\\

1455: +\frac{e^{-2\pi in\eta}\theta(y_{\alpha_1,1}-y_{\beta_1,1})

1456: \theta(y_{\alpha_q,q}-y_{\beta_q,q}+n\eta)}

1457: {\theta(y_{\alpha_1,1}-y_{\beta_1,1}-n\eta)

1458: \theta(y_{\alpha_q,q}-y_{\beta_q,q})}

1459: w_{j-r}(y_{\beta_1,1},\dots,y_{\beta_q,q})\times\\

1460: \times w_{i+r}(y_{\alpha_1,1}+t_1,\dots,y_{\alpha_q,q}+t_q)+\\

1461: +\sum_{1\le t\le q-1}\frac{e^{-2\pi in\eta}\theta(n\eta)

1462: \theta(y_{\alpha_1,1}-y_{\beta_1,1})}

1463: {\theta_{\alpha_1,1}-y_{\beta_1,1}-n\eta)}\cdot

1464: \frac{\theta(y_{\alpha_t,t}+y_{\beta_{t+1},t+1}-y_{\beta_t,t}-

1465: y_{\alpha_{t+1},t+1})}{\theta(y_{\alpha_t,t}-y_{\beta_t,t})

1466: \theta(y_{\beta_{t+1},t+1}-y_{\alpha_{t+1},t+1})}\times\\

1467: \times w_{j-r}(y_{\beta_1,1},\dots,y_{\beta_t,t},

1468: y_{\alpha_{t+1},t+1},\dots,,y_{\alpha_q,q})\times\\

1469: \times w_{i+r}(y_{\alpha_1,1}+t_1,\dots,y_{\alpha_t}+t_t,

1470: y_{\beta_{t+1}}+t_{t+1},\dots,y_{\beta_q,q}+t_q)\biggr).

1471: \end{multline*}

1472: Here

1473: $t_\alpha=-(d(n_1,\dots,n_{\alpha-1})+d(n_{\alpha+1},\dots,n_q))\eta$.

1474: 

1475: The equality $\psi_{\alpha_1,\dots,\beta_q}=0$ immediately follows from the 

1476: identity (35) proved in Appendix B.

1477: \end{proof}

1478: 

1479: \subsection{Homomorphism of the exchange algebra into the algebra

1480:  $Q_{n,k}(\E,\eta)$}

1481: 

1482: Let $q'\in\N$ and $\mu_1,\dots,\mu_{q'},\mu\in\C$. We define an associative algebra 

1483: $Y_{q'}(\E,\mu;\mu_1,\dots,\mu_{q'})$ as follows. The algebra 

1484: $Y_{q'}(\E,\mu;\mu_1,\dots,\mu_{q'})$ is presented by the generators 

1485: $\{\e(u_1,\dots,u_{q'});u_1,\dots,u_{q'}\in\C\}$ and the defining relations 

1486: \begin{multline*}

1487: \frac{\theta(v_1-u_1+\mu)}{\theta(v_1-u_1)} \e(u_1,\dots,u_{q'})

1488: \e(v_1+\mu_1,\dots,v_{q'}+\mu_{q'})=\\

1489: =\sum_{1\le t<q'}\frac{\theta(\mu)\theta(v_t-u_t+u_{t+1}-v_{t+1})}

1490: {\theta(v_t-u_t)\theta(u_{t+1}-v_{t+1})}\times\\

1491: \times\e(v_1,\dots,v_t,u_{t+1},\dots,u_{q'})

1492: \e(u_1+\mu_1,\dots,u_t+\mu_t,v_{t+1}+\mu_{t+1},\dots,v_{q'}+\mu_{q'})

1493: +\\

1494: +\frac{\theta(v_{q'}-u_{q'}+\mu)}{\theta(v_{q'}-u_{q'})}

1495: \e(v_1,\dots,v_{q'})\e(u_1+\mu_1,\dots,u_{q'}+\mu_{q'}).

1496: \end{multline*}

1497: 

1498: For $\mu=\mu_1=\ldots=\mu_{q'}=0$ the algebra $Y_{q'}(\E,0;0,\dots,0)$ is the 

1499: polynomial ring in infinitely many variables 

1500: $\{\e(u_1,\dots,u_{q'});u_1,\dots,u_{q'}\in\C\}$. One can show that the algebra 

1501: $Y_{q'}(\E,\mu;\mu_1,\dots,\mu_{q'})$ is a flat deformation of this polynomial 

1502: ring. 

1503: 

1504: Let $\frac     n{n-k}=n_1'-\frac1{n_2'-\ldots-\frac1{n_{q'}'}}$ be an expansion 

1505: in the continued fraction in which $n_\alpha'\ge2$ for $1\le\alpha\le q'$. It is 

1506: clear that such an expansion exists and unique. For the relationship between the 

1507: expansions in continued fractions of the numbers $\frac  nk$ and $\frac n{n-k}$, 

1508: see Appendix C. 

1509: 

1510: \begin{prop}

1511: There is an algebra homomorphism 

1512: $$\psi\colon

1513: Y_{q'}(\E,\mu;\mu_1,\dots,\mu_{q'})\to Q_{n,k}(\E,\eta),$$

1514:  where $\mu=n\eta$ and

1515: $\mu_\alpha=(d(n_1',\dots,n_{\alpha-1}')-d(n_{\alpha+1}',

1516: \dots,n_{q'}')\eta$. This homomorphism is of the form \emph:

1517: \begin{equation}

1518: \psi\colon\e(u_1,\dots,u_{q'})\to\sum_{\alpha\in\Z/n\Z}

1519: w_\alpha(u_1,\dots,u_{q'})x_{1-\alpha},

1520: \end{equation}

1521: where $\{x_i;i\in\Z/n\Z\}$ are the generators of the algebra $Q_{n,k}(\E,\eta)$, and 

1522: $\{w_\alpha;\alpha\in\Z/n\Z\}$ is a basis in the space of theta functions 

1523: $\Theta_{n/n-k}(\Gamma)$ (see Appendix B).

1524: \end{prop}

1525: 

1526: \begin{proof}

1527: Let us apply the map $\psi$ to the difference between the left- and right-hand 

1528: sides of the relations in the algebra $Y_{q'}(\E,\mu;\mu_1,\dots,\mu_{q'})$. We 

1529: must verify the resulting relation in the algebra $Q_{n,k}(\E,\eta)$. We have 

1530: \begin{multline*}

1531: \frac{\theta(v_1-u_1+\mu)}{\theta(v_1-u_1)}

1532: \psi(\e(u_1,\dots,u_{q'}))\psi(\e(v_1+\mu_1,\dots,v_{q'}+\mu_{q'}))-

1533: \\

1534: -\sum_{1\le t<q'}\frac{\theta(\mu)\theta(v_t-u_t+u_{t+1}-v_{t+1})}

1535: {\theta(v_t-u_t)\theta(u_{t+1}-v_{t+1})}\times\\

1536: \times\psi(\e(v_1,\dots,v_t,u_{t+1},\dots,u_{q'}))

1537: \psi(\e(u_1+\mu_1,\dots,u_t+\mu_t,v_{t+1}+\mu_{t+1},\dots,

1538: v_{q'}+\mu_{q'}))-\\

1539: -\frac{\theta(v_{q'}-u_{q'}+\mu)}{\theta(v_{q'}-u_{q'})}

1540: \psi(\e(v_1,\dots,v_{q'}))\psi(\e(u_1+\mu_1,\dots,u_{q'}+\mu_{q'}))=

1541: \\

1542: =\sum_{\alpha,\beta\in\Z/n\Z}x_{1-\alpha}x_{1-\beta}\times\\

1543: \times\biggl(

1544: \frac{\theta(v_1-u_1+\mu)}{\theta(v_1-u_1)}

1545: w_\alpha(u_1,\dots,u_{q'})

1546: w_\beta(v_1+\mu_1,\dots,

1547: v_{q'}+\mu_{q'})-\\

1548: -\sum_{1\le t<q'}\frac{\theta(\mu)\theta(v_t-u_t+u_{t+1}-v_{t+1})}

1549: {\theta(v_t-u_t)\theta(u_{t+1}-v_{t+1})}\times\\

1550: \times w_\alpha(v_1,\dots,v_t,u_{t+1},\dots,u_{q'})

1551: w_\beta(u_1+\mu_1,\dots,u_t+\mu_t,v_{t+1}+\mu_{t+1},\dots,

1552: v_{q'}+\mu_{q'})-\\

1553: -\frac{\theta(v_{q'}-u_{q'}+\mu)}{\theta(v_{q'}-u_{q'})}

1554: w_\alpha(v_1,\dots,v_{q'})w_\beta(u_1+\mu_1,\dots,u_{q'}+\mu_{q'})

1555: \biggr).

1556: \end{multline*}

1557: By using the identity (35) in Appendix B together with the relations (18) in the 

1558: algebra $Q_{n,k}(\E,\eta)$, one can readily see that this expression is equal to $0$.

1559: \end{proof}

1560: 

1561: \newpage

1562: 

1563: \appendix{Theta functions of one variable}

1564: 

1565: Let $\Gamma\subset\C$ be an integral lattice generated by $1$ and

1566: $\tau\in\C$, where $\Im\tau>0$. Let $n\in\N$ and $c\in\C$. We denote

1567: by  $\Theta_{n,c}(\Gamma)$ the space of the entire functions

1568: of one variable satisfying the following relations:

1569: $$

1570: f(z+1)=f(z),\quad f(z+\tau)=(-1)^ne^{-2\pi i(nz-c)}f(z).

1571: $$

1572: As is known [30], $\dim\Theta_{n,c}(\Gamma)=n$, every

1573: function $f\in\Theta_{n,c}(\Gamma)$ has exactly $n$ zeros modulo $\Gamma$   

1574: (counted according to their multiplicities), and the sum of these zeros 

1575:  modulo $\Gamma$ is equal

1576: to $c$. Let  $\theta(z)=\sum_{\alpha\in\Z}(-1)^\alpha

1577: e^{2\pi i\left(\alpha

1578: z+\frac{\alpha(\alpha-1)}2\tau\right)}$. It is clear that

1579: $\theta(z)\in\Theta_{1,0}(\Gamma)$. It follows from what was

1580:  said above that  $\theta(0)=0$, and this is the only zero 

1581: modulo~$\Gamma$. One can readily see that

1582: $\theta(-z)=-e^{-2\pi iz}\theta(z)$. Moreover, as is known, the function

1583: $\theta(z)$ can be expanded as the infinite product as follows:

1584: $$

1585: \theta(z)=\prod_{\alpha\ge1}(1-e^{2\pi

1586: i\alpha\tau})\cdot(1-e^{2\pi

1587: iz})\cdot\prod_{\alpha\ge1}(1-e^{2\pi

1588: i(z+\alpha\tau)})(1-e^{2\pi i(\alpha\tau-z)}).

1589: $$

1590:  

1591: Let us introduce the following linear operators $T_{\frac1n}$ and $T_{\frac1n\tau}$ 

1592: acting on the space of functions of one variable:

1593: $$

1594: T_{\frac1n}f(z)=f\left(z+\frac1n\right),\quad

1595: T_{\frac1n\tau}f(z)=e^{2\pi

1596: i\left(z+\frac1{2n}-\frac{n-1}{2n}\tau\right)}

1597: f\left(z+\frac1n\tau\right).

1598: $$

1599: One can readily see that the space $\Theta_{n,\frac{n-1}2}(\Gamma)$ is invariant 

1600: with respect to the operators $T_{\frac1n}$ and $T_{\frac1n\tau}$. Moreover, 

1601: $T_{\frac1n}T_{\frac1n\tau}=e^{\frac{2\pi i}n}T_{\frac1n\tau}T_{\frac1n}$. The 

1602: restriction of these operators to the space $\Theta_{n,\frac{n-1}2}(\Gamma)$ satisfy 

1603: the relations $T_{\frac1n}^n=T_{\frac1n\tau}^n=1$. Let $\wt{\Gamma_n}$ be the 

1604: group with the generators $a,b,\epsilon$ and the defining relations 

1605: $ab=\epsilon  ba$, $a\epsilon=\epsilon  a$,  $b\epsilon=\epsilon  b$, and

1606: $a^n=b^n=\epsilon^n=e$. The group $\wt{\Gamma_n}$ is a central extension of the 

1607: group $\Gamma_n=\Gamma/n\Gamma\simeq(\Z/n\Z)^2$, namely, the element $\epsilon$ 

1608: generates a normal subgroup $C_n=\Z/n\Z$, and $\wt{\Gamma_n}/C_n=\Gamma_n$. The 

1609: formulas $a\mapsto   T_{\frac1n}$,

1610: $b\mapsto   T_{\frac1n\tau}$, and  $\epsilon\mapsto{}$(multiplication by

1611: $e^{\frac{2\pi   i}n}$) define an irreducible representation of the group 

1612: $\wt{\Gamma_n}$ in the space $\Theta_{n,\frac{n-1}2}(\Gamma)$. Let us choose a 

1613: basis $\{\theta_\alpha;\alpha\in\Z/n\Z\}$ in the space $\Theta_{n,\frac{n-1}2}(\Gamma)$ 

1614: in which our operators act as follows: $T_{\frac1n}\theta_\alpha=e^{2\pi i\frac\alpha

1615: n}\theta_\alpha$, and $T_{\frac1n\tau}\theta_\alpha=\theta_{\alpha+1}$. It is 

1616: clear that this choice can be carried out uniquely up to multiplication by a 

1617: common constant. The functions $\theta_\alpha(z)$ are of the form 

1618: $$

1619: \theta_\alpha(z)=\theta\left(z+\frac\alpha n\tau\right)

1620: \theta\left(z+\frac1n+\frac\alpha n\tau\right)\dots

1621: \theta\left(z+\frac{n-1}n+\frac\alpha n\tau\right)

1622: e^{2\pi                                                 i\left(\alpha

1623: z+\frac{\alpha(\alpha-n)}{2n}\tau+\frac\alpha{2n}\right)}.

1624: $$

1625: One can readily see that $\theta_\alpha(z)\in\Theta_{n,\frac{n-1}2}(\Gamma)$,

1626: $\theta_{\alpha+n}(z)=\theta_\alpha(z)$, and

1627: \begin{equation}

1628: \begin{aligned}

1629: \theta_\alpha\left(z+\frac1n\right)&=e^{2\pi            i\frac\alpha

1630: n}\theta_\alpha(z),\\

1631: \theta_\alpha\left(z+\frac1n\tau\right)&=e^{-2\pi

1632: i\left(z+\frac1{2n}-\frac{n-1}{2n}\tau\right)}\theta_{\alpha+1}(z)

1633: \end{aligned}

1634: \end{equation}

1635: It is clear that the functions $\left\{\theta_\alpha\left(z-\frac1nc-\frac{n-1}{2n}\right);

1636: \alpha\in\Z/n\Z\right\}$ form a basis in the space $\Theta_{n,c}(\Gamma)$.

1637: 

1638: We need some identities:

1639: \begin{equation}

1640: \theta(nz)=\frac{n\theta_0(z)\dots\theta_{n-1}(z)e^{-2\pi

1641: i\frac{n(n-1)}2z}}{\theta_1(0)\dots\theta_{n-1}(0)

1642: \theta\left(\frac1n\right)\dots\theta\left(\frac{n-1}n\right)}.

1643: \end{equation}

1644: 

1645: \begin{proof}

1646: One can readily see by using relations (26) that the functions on both sides of 

1647: the equation belong to the space $\Theta_{n^2,\frac{n(n-1)}2\tau}(\Gamma)$. 

1648: Moreover, it is clear that the zeros of both functions coincide, namely, these 

1649: are $n^2$ points $\left\{\frac\alpha

1650: n+\frac\beta  n\tau;\alpha,\beta\in\Z\right\}$ modulo $\Gamma$. Hence, the 

1651: functions on the left- and right-hand sides of the equation differ by a constant 

1652: multiple, which can be evaluated by dividing (27) by $\theta(z)$ and passing to the 

1653: limit as $z\to0$.

1654: \end{proof}

1655: 

1656: Let $\theta_0,\theta_1,\theta_2\in\Theta_{3,0}(\Gamma)$. For $z,\eta\in\C$ and 

1657: $\alpha\in\Z/3\Z$ we have

1658: \begin{equation}

1659: \theta_0(\eta)\theta_\alpha(z+\eta)\theta_\alpha(z)+

1660: \theta_1(\eta)\theta_{\alpha+2}(z+\eta)

1661: \theta_{\alpha+1}(z)+

1662: \theta_2(\eta)\theta_{\alpha+1}(z+\eta)

1663: \theta_{\alpha+2}(z)=0.

1664: \end{equation}

1665: 

1666: \begin{proof}

1667: It is clear that $\theta_\alpha(z+\eta)\theta_\beta(z)\in\Theta_{6,-3\eta}(\Gamma)$ 

1668: as a function of the variable $z$. There must be three linear relations among 

1669: these nine functions in a six-dimensional space. With regard to the action of the 

1670: group $\wt{\Gamma_3}$, we see that the relations must be of the form

1671: $a(\eta)\theta_\alpha(z+\eta)\theta_\alpha(z)+

1672: b(\eta)\theta_{\alpha+1}(z+\eta)\theta_{\alpha+2}(z)+

1673: c(\eta)\theta_{\alpha+2}(z+\eta)\theta_{\alpha+1}(z)=0$. Really, every  

1674: three-dimensional space of relations invariant with respect to the translations 

1675: $z\to  z+\frac13$ and $z\to z+\frac13\tau$ (see (26)) is of this form, where 

1676: $a,b,c$ do not depend on $\alpha$. By setting  $\alpha=1$ and $z=0$, we obtain 

1677: $\frac{c(\eta)}{a(\eta)}=\frac{\theta_1(\eta)}{\theta_0(\eta)}$. By setting 

1678: $\alpha=2$ and $z=0$, we obtain $\frac{b(\eta)}{a(\eta)}=\frac{\theta_2(\eta)}{\theta_0(\eta)}$.

1679: \end{proof}  

1680: 

1681: Let $\theta_\alpha\in\Theta_{n,c}(\Gamma)$. Then

1682: \begin{multline}

1683: \frac{\theta(y-z+nv-nu)}{\theta(y-z)\theta(nv-nu)}

1684: \theta_\alpha(y+u)\theta_\beta(z+v+\eta)+

1685: \frac{\theta(z-y+n\eta)}{\theta(z-y)\theta(n\eta)}

1686: \theta_\alpha(z+u)\theta_\beta(y+v+\eta)=\\

1687: =\frac1n\theta\left(\frac1n\right)\dots\theta\left(\frac{n-1}n\right)

1688: \sum_{r\in\Z/n\Z}\frac{\theta_{\beta-\alpha}(v-u+\eta)}

1689: {\theta_r(\eta)\theta_{\beta-\alpha-r}(v-u)}\theta_{\beta-r}(y+v)

1690: \theta_{\alpha+r}(z+u+\eta).

1691: \end{multline}

1692: 

1693: \begin{proof}

1694: This is a special case of the relation (31) (for $p=1$) proved in Appendix B.

1695: \end{proof}

1696: 

1697: By setting $u=v+\eta$ in the relation (29) and making the change of variables 

1698: $y+v\to  y$, $z+v\to z$, we obtain

1699: \begin{multline}

1700: \frac{\theta(z-y+n\eta)}{\theta(z-y)\theta(n\eta)}

1701: (\theta_\alpha(z+\eta)\theta_\beta(y+\eta)-\theta_\alpha(y+\eta)

1702: \theta_\beta(z+\eta))=\\

1703: =\frac1n\theta\left(\frac1n\right)\dots\theta\left(\frac{n-1}n\right)

1704: \theta_{\beta-\alpha}(0)\sum_{r\in\Z/n\Z}

1705: \frac1{\theta_r(\eta)\theta_{\beta-\alpha-r}(-\eta)}

1706: \theta_{\beta-r}(y)\theta_{\alpha+r}(z+2\eta).

1707: \end{multline}

1708: 

1709: \appendix{Some theta functions of several variables associated with a power of an 

1710: elliptic curve}

1711: 

1712: Let $n$ and $k$ be coprime positive integers such

1713: that $1\le

1714: k<n$. We expand the ratio $\frac nk$ in a continued fraction of the

1715: form: $\frac

1716: nk=n_1-\frac1{n_2-\frac1{n_3-\ldots-\frac1{n_p}}}$,

1717: where $n_\alpha\ge2$ for any $\alpha$. It is clear that such an 

1718: expansion exists and is unique. We denote by $d(m_1, \dots , m_q)$ the determinant of the ($q

1719: \times q$) matrix~$(m_{\alpha\beta})$, where

1720: $m_{\alpha\alpha}=m_\alpha$,

1721: $m_{\alpha,\alpha+1}=m_{\alpha+1,\alpha}=-1$, and

1722: $m_{\alpha,\beta}=0$ for $|\alpha-\beta|>1$.  For 

1723: $q = 0$ we set $d(\varnothing)=1$. It follows from

1724: the elementary theory of continued fractions that $n = d(n_1, \dots ,

1725: n_p)$ and $k = d(n_2, \dots , n_p)$.

1726: 

1727:        Let $\Gamma\subset\C$ be an integral lattice   

1728: generated by 1 and~$\tau$ again, where $\Im\tau>0$.

1729: 

1730:       We denote by $\Theta_{n/k}(\Gamma)$ the space of entire functions (of $p$ 

1731: variables) satisfying the following relations:

1732: \begin{align*}

1733: f(z_1,\dots,z_\alpha+1,\dots,z_p)&=f(z_1,\dots,z_p),\\

1734: f(z_1,\dots,z_\alpha+\tau,\dots,z_p)&=(-1)^{n_\alpha}e^{-2\pi

1735: i(n_\alpha

1736: z_\alpha-z_{\alpha-1}-z_{\alpha+1}-(\delta_{1,\alpha}-1)\tau)}

1737: f(z_1,\dots,z_p).

1738: \end{align*}

1739: Here $1\le\alpha\le p$ and $z_0=z_{p+1}=0$, and $\delta_{1,\alpha}$

1740: stands for the Kronecker delta. Thus, the functions $f\in\Theta_{n/k}(\Gamma)$

1741: are periodic with respect to each of the variables with period 1 and quasiperiodic with period~$\tau$. 

1742: By the periodicity, each  function in the space $\Theta_{n/k}(\Gamma)$ can be expanded 

1743: in a Fourier series of the form

1744: $f(z_1,\dots,z_p)=\sum_{\alpha_1,\dots,\alpha_p\in\Z}

1745: a_{\alpha_1\dots\alpha_p}e^{2\pi i(\alpha_1z_1+\ldots+\alpha_pz_p)}$. 

1746: By the quasiperiodicity, the coefficients satisfy the system of linear 

1747: equations

1748: $$

1749: a_{\alpha_1,\dots,\alpha_{\nu-1}-1,\alpha_\nu+n_\nu,

1750: \alpha_{\nu+1}-1,\dots,\alpha_p}=

1751: (-1)^{n_\alpha}e^{2\pi i(\alpha_\nu+\delta_{1,\alpha}-1)\tau}

1752: a_{\alpha_1\dots\alpha_p}.

1753: $$

1754: 

1755:        This system clearly has $n = d(n_1, \dots ,

1756: n_p)$ linearly independent solutions

1757: each defining (for $\Im\tau>0$;

1758: $n_1,\dots,n_p\ge2$) a function in the space $\Theta_{n/k}(\Gamma)$ .

1759: 

1760:       For $k = 1$ we have the space

1761: of functions of one variable $\Theta_n(\Gamma)=\Theta_{n,0}(\Gamma)$ (see Appendix A) with

1762: a basis

1763: $\left\{w_\alpha(z)=\theta_\alpha\left(z+\frac{n-1}2\right),\

1764: \alpha\in\Z/n\Z\right\}$. A similar basis can be constructed in the space

1765: $\Theta_{n/k}(\Gamma)$ for an arbitrary~$k$. Let us define the operators

1766: $T_{\frac1n}$ and~$T_{\frac1n\tau}$ in the space of functions of $p$

1767: variables as follows:

1768: \begin{align*}

1769: T_{\frac1n}f(z_1,\dots,z_p)&=f(z_1+r_1,\dots,z_p+r_p),\\

1770: T_{\frac1n\tau}f(z_1,\dots,z_p)&=e^{2\pi i(z_1+\phi)}

1771: f(z_1+r_1\tau,\dots,z_p+r_p\tau).

1772: \end{align*}

1773: Here $r_\alpha=\frac{d(n_{\alpha+1},\dots,n_p)}{d(n_1,\dots,n_p)}$ and   

1774: $\phi\in\C$ is a constant.

1775: 

1776:     It is clear that $T_{\frac1n}T_{\frac1n\tau}=e^{2\pi i\frac

1777: kn}T_{\frac1n\tau}T_{\frac1n}$. As in the case of theta functions of one  

1778: variable, the space

1779:  $\Theta_{n/k}(\Gamma)$ is invariant with respect to the operators $T_{\frac1n}$

1780: and $T_{\frac1n\tau}$, and the restriction of these operators

1781: to~$\Theta_{n/k}(\Gamma)$ satisfies the relations $T_{\frac1n}^n=1$ and 

1782: $T_{\frac1n\tau}^n=\mu$, where $\mu\in\C$. Let us choose a $\phi$  in such a way that

1783: $\mu=1$; clearly, this can be done uniquely up to multiplication of

1784: $T_{\frac1n\tau}$ by a root of unity of degree $n$.

1785: 

1786: \begin{prop} 

1787: There is a basis $\bigl\{w_\alpha(z_1,\dots,z_p);\,\alpha\in\Z/n\Z\bigr\}$ in  

1788: $\Theta_{n/k}(\Gamma)$ such that

1789: $$

1790: T_{\frac1n}w_\alpha=e^{2\pi i\frac kn\alpha}w_\alpha,\quad

1791: T_{\frac1n\tau}w_\alpha=w_{\alpha+1}.

1792: $$

1793: This basis is defined uniquely up to multiplication by a common constant.

1794: \end{prop}

1795: 

1796: \begin{proof}

1797:  Let $f\in\Theta_{n/k}(\Gamma)$ be an eigenvector of the operator $T_{\frac1n}$ with 

1798:  an eigenvalue~$\lambda$. Since $T_{\frac1n}^n=1$ on the space $\Theta_{n/k}(\Gamma)$,

1799: we have  $\lambda^n=1$. Moreover, $T_{\frac1n}T_{\frac1n\tau}f=e^{2\pi

1800: i\frac

1801: kn}T_{\frac1n\tau}T_{\frac1n}f=e^{2\pi i\frac kn}\lambda

1802: T_{\frac1n\tau}f$, and hence $T_{\frac1n\tau}f$ is also

1803: an eigenvector with the eigenvalue $e^{2\pi i\frac

1804: kn}\lambda$. Since $n$ and $k$ are coprime, $e^{2\pi i\frac

1805: kn}$ is a

1806: primitive root of unity of degree $n$. Thus, the vectors $\bigl\{T_{\frac1n\tau}^\alpha

1807: f;\,\alpha=0,1,\dots,n-1\bigr\}$

1808:  are eigenvectors for the operator $T_{\frac1n}$ with

1809: different eigenvalues, and every of root of unity of degree $n$ is an

1810: eigenvalue for some $T_{\frac1n\tau}^\alpha f$. Let $w_0$

1811: be such that $T_{\frac1n}w_0=w_0$. We set

1812: $w_\alpha=T_{\frac1n\tau}^\alpha w_0$. It is clear that

1813: $T_{\frac1n}w_\alpha=e^{2\pi i\frac

1814: kn\alpha}w_\alpha$ and

1815: $T_{\frac1n\tau}w_\alpha=w_{\alpha+1}$. Moreover,

1816: $w_{\alpha+n}=w_\alpha$ because $T_{\frac1n\tau}^n=1$

1817: on the space $\Theta_{n/k}(\Gamma)$.

1818: \end{proof}

1819: 

1820:     We note that, as in the case of theta functions of one variable, the group 

1821:     $\wt{\Gamma_n}$ irreducibly acts on

1822: the space $\Theta_{n/k}(\Gamma)$ by the rule $a\mapsto T_{\frac1n}$, $b\mapsto

1823: T_{\frac1n\tau}$, and $\epsilon\mapsto\text{(multiplication by $e^{2\pi

1824: i\frac kn}$})$.

1825: 

1826: \begin{remark}

1827:  Let $L$ be the group of linear automorphisms on the space of functions 

1828:  of $p$

1829: variables of the form

1830: $$

1831: gf(z_1,\dots,z_p)=e^{2\pi

1832: i(\phi_1z_1+\ldots+\phi_pz_p+\lambda)}f(z_1+\psi_1,\dots,z_p+\psi_p)

1833: $$

1834: for $g\in L$. It is clear that $L$ is a $(2p + 1)$-dimensional Lie

1835: group. Let $L'\subset L$ be the subgroup of

1836: transformations preserving the space~$\Theta_{n/k}(\Gamma)$, that is,

1837: $L'=\bigl\{g\in

1838: L;\,g(\Theta_{n/k}(\Gamma))=\Theta_{n/k}(\Gamma)\bigr\}$.

1839: Let $L''\subset L'$  consist of the elements preserving each point of

1840: $\Theta_{n/k}(\Gamma)$, that is, $L''=\bigl\{g\in

1841: L';\,gf=f\text{ for any }f\in\Theta_{n/k}(\Gamma)\bigr\}$. One can see that 

1842: the quotient group $L'/L''=\wt G_n$  

1843:  is generated by the elements

1844: $T_{\frac1n}$ and $T_{\frac1n\tau}$ and by the multiplications by

1845: constants.

1846: \end{remark}

1847: We shall use the notation $w_\alpha^{n/k}(z_1,\dots,z_p)$ if it is not clear 

1848: from the context what are the theta functions in use.

1849: 

1850: We need the following identity relating theta functions in the spaces 

1851: $\Theta_{1,0}(\Gamma)$, $\Theta_{n,\frac{n-1}2}(\Gamma)$, and $\Theta_{n/k}(\Gamma)$: 

1852: \begin{align}

1853: &\frac{\theta(y_1-z_1+nv-nu)}{\theta(nv-nu)\theta(y_1-z_1)}\nonumber\\

1854: &{}\times w_\alpha(y_1+m_1u,\dots,y_p+m_pu)

1855: w_\beta(z_1+m_1v+l_1,\dots,z_p+m_pv+l_p)\nonumber\\

1856: &+\sum_{1\le t\le p-1}

1857: \frac{\theta(z_t-y_t+y_{t+1}-z_{t+1})}

1858: {\theta(z_t-y_t)\theta(y_{t+1}-z_{t+1})}\nonumber\\

1859: &{}\times w_\alpha(z_1+m_1u,\dots,z_t+m_tu,y_{t+1}+m_{t+1}u,\dots,y_p+m_pu)

1860: \nonumber\\ &{}\times

1861: w_\beta(y_1+m_1v+l_1,\dots,y_t+m_tv+l_t,z_{t+1}+m_{t+1}v+l_{t+1},

1862: \dots,z_p+m_pv+l_p)\nonumber\\

1863: &{}+\frac{\theta(z_p-y_p+n\eta)}{\theta(z_p-y_p)\theta(n\eta)}

1864: w_\alpha(z_1+m_1u_1,\dots,z_p+m_pu)

1865: \nonumber\\ &{}\times w_\beta(y_1+m_1v+l_1,\dots,y_p+m_pv+l_p)\nonumber\\

1866: &{}=\frac1n\theta\left(\frac1n\right)\dots\theta\left(\frac{n-1}n\right)

1867: \nonumber\\ &{}\times

1868: \sum_{r\in\Z/n\Z}\frac{\theta_{\beta-\alpha+ r(k-1)}(v-u+\eta)}

1869: {\theta_{ r k}(\eta)\theta_{\beta-\alpha- r}(v-u)}

1870: w_{\beta- r}(y_1+m_1v,\dots,y_p+m_pv)

1871: \nonumber\\ &{}\times

1872: w_{\alpha+ r}(z_1+m_1u+l_1,\dots,z_p+m_pu+l_p).

1873: \end{align}

1874: Here $m_\alpha=d(n_{\alpha+1},\dots,n_p)$  and 

1875: $l_\alpha=d(n_1,\dots,n_{\alpha-1})\eta$.

1876: 

1877: \begin{proof}

1878:  We denote by   

1879: $\phi_{\alpha,\beta}(\eta,u,v,y_1,\dots,y_p,z_1,\dots,z_p)$

1880: the difference between the right- and left-hand 

1881: sides of the formula (31). The calculation shows that this function 

1882: satisfies the following relations:

1883: \begin{equation}

1884: \begin{aligned}

1885: \phi_{\alpha,\beta}(\eta,\dots,y_\alpha+1,\dots,z_p)&=

1886: \phi_{\alpha,\beta}(\eta,\dots,z_p),\\

1887: \phi_{\alpha,\beta}(\eta,\dots,y_\alpha+\tau,\dots,z_p)&=

1888: -e^{-2\pi i(n_\alpha y_\alpha-y_{\alpha-1}-y_{\alpha+1}+\delta_{\alpha,1}v)}

1889: \phi_{\alpha,\beta}(\eta,\dots,z_p),\\

1890: \phi_{\alpha,\beta}(\eta,\dots,z_\alpha+1,\dots,z_p)&=

1891: \phi_{\alpha,\beta}(\eta,\dots,z_p),\\

1892: \phi_{\alpha,\beta}(\eta,\dots,z_\alpha+\tau,\dots,z_p)&=

1893: -e^{-2\pi i(n_\alpha z_\alpha-z_{\alpha-1}-z_{\alpha+1}+

1894: \delta_{\alpha,1}u+\delta_{\alpha,p}\eta)}

1895: \phi_{\alpha,\beta}(\eta,\dots,z_p).

1896: \end{aligned}

1897: \end{equation}

1898: Here $y_0=y_{p+1}=z_0=z_{p+1}=0$, and   $\delta_{\alpha,\beta}$  stands for the Kronecker

1899:  delta. Moreover, an evaluation shows

1900: that there are no poles on the divisors $nv-nu\in\Gamma$,

1901: $n\eta\in\Gamma$,

1902: $y_1-z_1\in\Gamma$, \dots, $y_p-z_p\in\Gamma$,

1903: and hence the function   $\phi_{\alpha,\beta}$   is holomorphic everywhere 

1904: on~$\C^{2p+3}$. However, it is clear that the

1905: functions $\bigl\{w_\lambda(y_1+m_1v,\dots,y_p+m_pv)

1906: w_\nu(z_1+m_1u+l_1,\dots,z_p+m_pu+l_p);\,

1907: \lambda,\nu\in\Z/n\Z\bigr\}$ form

1908: a basis in the space of holomorphic functions (of the variables $y_1,\dots,y_p,z_1,\dots,z_p$) 

1909:  satisfying the conditions (32).

1910: Therefore, the function   $\phi_{\alpha,\beta}$  is of the form

1911: \begin{multline}

1912: \varphi_{\alpha,\beta}(\eta,u,v,y_1,\dots,z_p)=\\

1913: {}=\sum_{\lambda,\nu\in\Z/n\Z}

1914: \psi_{\lambda,\nu}(\eta,u,v)w_\lambda(y_1+m_1v,\dots,y_p+m_pv)

1915: \\ {}\times

1916: w_\nu(z_1+m_1u+l_1,\dots,z_p+m_pu+l_p).

1917: \end{multline}

1918: 

1919:         Here the functions   $\psi_{\lambda,\nu}(\eta,u,v)$ are holomorphic and 

1920: satisfy the relations

1921: \begin{equation}

1922: \begin{gathered}

1923: \psi_{\lambda,\nu}(\eta+1,u,v)=\psi_{\lambda,\nu}(\eta,u+1,v)=

1924: \psi_{\lambda,\nu}(\eta,u,v+1)=\psi_{\lambda,\nu}(\eta,u,v),\\

1925: \psi_{\lambda,\nu}(\eta+\tau,u,v)=

1926: e^{-2\pi in(v-u)}\psi_{\lambda,\nu}(\eta,u,v),\\

1927: \psi_{\lambda,\nu}(\eta,u+\tau,v)=e^{2\pi in\eta}\psi_{\lambda,\nu}(\eta,u,v),\\

1928: \psi_{\lambda,\nu}(\eta,u,v+\tau)=e^{-2\pi in\eta}\psi_{\lambda,\nu}(\eta,u,v).

1929: \end{gathered}

1930: \end{equation}

1931: 

1932:         These relations are verified by the immediate calculation, namely, 

1933: one must compare the multipliers at the translations by 1

1934: and~$\tau$   in the formulas (32) and (33).

1935: 

1936:         However, every holomorphic function of the variables $\eta$, $u$ and $v$

1937:  that satisfies relations (34) is vanishes. 

1938:  Really, since this function is periodic, it admits the expansion in the Fourier 

1939: series

1940: $$

1941: \psi_{\lambda,\nu}(\eta,u,v)=\sum_{\alpha,\beta,\gamma\in\Z}

1942: a_{\lambda,\nu,\alpha,\beta,\gamma}

1943: e^{2\pi i(\alpha\eta+\beta u+\gamma v)}.

1944: $$

1945: Further, it follows from the quasiperiodicity that the

1946: coefficients $a_{\lambda,\nu,\alpha,\beta,\gamma}$  are equal to 0.

1947: \end{proof}

1948: 

1949: By setting $u=v+\eta$ in the identity (31) and making the change of variables 

1950: $y_1\to

1951: y_1-m_1v$,  $z_1\to  z_1-m_1v$,  \dots,  $y_p\to  y_p-m_pv$,  $z_p\to

1952: z_p-m_pv$, we obtain

1953: \begin{multline}

1954: \frac{\theta(y_1-z_1-n\eta)}{\theta(-n\eta)\theta(y_1-z_1)}

1955: w_\alpha(y_1+m_1\eta,\dots,y_p+m_p\eta)

1956: \theta_\beta(z_1+\l_1,\dots,z_p+\l_p)+\\

1957: +\sum_{1\le t<p}\frac{\theta(z_t-y_t+y_{t+1}-z_{t+1})}

1958: {\theta(z_t-y_t)\theta(y_{t+1}-z_{t+1})}

1959: w_\alpha(z_1+m_1\eta,\dots,z_t+m_t\eta,

1960: y_{t+1}+m_{t+1}\eta,\dots,y_p+m_p\eta)\times\\

1961: \times

1962: w_\beta(y_1+\l_1,\dots,y_t+\l_t,z_{t+1}+\l_{t+1},\dots,z_p+\l_p)+\\

1963: +\frac{\theta(z_p-y_p+n\eta)}{\theta(z_p-y_p)\theta(n\eta)}

1964: w_\alpha(z_1+m_1\eta,\dots,z_p+m_p\eta)

1965: w_\beta(y_1+\l_1,\dots,y_p+\l_p)=\\

1966: =\frac1n\theta\left(\frac1n\right)\dots\theta\left(\frac{n-1}n\right)

1967: \times\\\times\sum_{r\in\Z/n\Z}\frac{\theta_{\beta-\alpha+r(k-1)}(0)}

1968: {\theta_{rk}(\eta)\theta_{\beta-\alpha-r}(-\eta)}

1969: w_{\beta-r}(y_1,\dots,y_p)

1970: w_{\alpha+r}(z_1+m_1\eta+\l_1,\dots,z_p+m_p\eta+\l_p).

1971: \end{multline}

1972: 

1973: 

1974: 

1975: \appendix{Duality between the spaces  $\Theta_{n/k}(\Gamma)$ and

1976: $\Theta_{n/n-k}(\Gamma)$} 

1977: 

1978: Let us construct a canonical element 

1979: $\Delta_{n,k}\in\Theta_{n/k}(\Gamma)\otimes\Theta_{n/n-k}(\Gamma)$

1980: carring out the duality between these spaces (see (36)).

1981: 

1982: \begin{prop}

1983:  Let

1984: $$

1985: \frac nk=n_1-\frac1{n_2-\ldots-\frac1{n_p}},\quad \frac

1986: n{n-k}=n_1'-\frac1{n_2'-\ldots-\frac1{n_{p'}'}}

1987: $$ 

1988: be the expansions in continued fractions, where $n_\alpha\ge2$ and

1989: $n_\beta'\ge2$  for $1\le\alpha\le p$ and $1\le\beta\le p'$, respectively.

1990: Here $p$ and $p'$  stand for the lengths of the continued fractions. Then 

1991: $p'=n_1+\ldots+n_p-2p+1$

1992: and $n_1'+\ldots+n_{p'}'=2(n_1+\ldots+n_p)-3p+1$. Moreover,

1993: $n_1'+\ldots+n_\alpha'=2\alpha+\beta$   for

1994: $n_1+\ldots+n_\beta-2\beta+1\le\alpha\le

1995: n_1+\ldots+n_{\beta+1}-2\beta-2$. In other words, the Young diagrams for the 

1996: partitions

1997: $(n_1-1,n_1+n_2-3,\dots,n_1+\ldots+n_\alpha-2\alpha+1,\dots)$ and 

1998: $(n_1'-1,n_1'+n_2'-3,\dots,n_1'+\dots+n_\beta'-2\beta+1,\dots)$

1999: are dual to each other.

2000: \end{prop}

2001: 

2002: \begin{remark} 

2003: For $k = 1$, $p = 1$, and 

2004: $n_1 = n$ we have $p'  = n - 1$ and $n_1'=\ldots=n_{n-1}'=2$. 

2005: For $p > 1$, if $n_2, \dots , n_{p-1}\ge   3$, then the sequence

2006: $(n_1',\dots,n_p')$ becomes

2007: $(2^{(n_1-2)},3,2^{(n_2-3)},3,\dots,3,2^{(n_{p-1}-3)},3,2^{(n_p-2)})$.

2008: Here $2^{(t)}$, $t\ge0$, stands for a sequence of $t$ twos. This

2009: formula remains valid without the assumption that $n_2,\dots,n_p\ge3$ if we agree that

2010:  the

2011: sequence $(m_1,2^{(-1)},m_2)$ is of length $1$ and is equal to $(m_1+m_2-2)$. 

2012: This rule must be

2013: applied in succession to all occurrences $n_\alpha=2$ for $2\le\alpha\le p-1$.

2014: \end{remark}

2015: 

2016: \begin{proof}

2017:  The proof can be carried out by induction on $\min(p,p')$. For $p = 1$, one must 

2018:  prove that

2019: $\frac n{n-1}=2-\frac1{2-\ldots-\frac12}$  is of length 

2020: $n - 1$. Let $p,p'>1$ and let, say, $n_1>2$. We

2021: have     $\frac

2022: k{d(n_3,\dots,n_p)}=n_2-\frac1{n_3-\ldots-\frac1{n_p}}$. By assumption,

2023: $$

2024: \frac k{k-d(n_3,\dots,n_p)}=

2025: n_{n_1-1}'-1-\frac1{n_{n_1}'-\frac1{n_{n_1+1}'-\ldots-\frac1{n_{p'}'}}}.

2026: $$

2027: and the sequence $(n_1',\dots,n_{n_1-2}')$ is~$(2^{(n_1-2)})$.

2028: Further, one must show that

2029: $n_1'-\frac1{n_2'-\ldots-\frac1{n_{p'}'}}=\frac n{n-k}$. Here it is 

2030: used that 

2031: $d(n_1,\dots,n_p)=n$,

2032: $d(n_2,\dots,n_p)=k$,

2033: $

2034: \frac{d(n_1,\dots,n_p)}{d(n_2,\dots,n_p)}=

2035: n_1-\frac1{n_2-\ldots-\frac1{n_p}},

2036: $ 

2037: and $d(n_1,\dots,n_p)=n_1d(n_2,\dots,n_p)-d(n_3,\dots,n_p)$.

2038: \end{proof}

2039: 

2040: \begin{prop} 

2041: Let a function  $\Delta_{n.k}(z_1,\dots,z_p;z_1',\dots,z_{p'}')$

2042:  of $p+p'$  variables $z_1,\dots,z_p,z_1',\dots,z_{p'}'$

2043:   be defined by the formula

2044: \begin{multline*}

2045: \Delta_{n,k}(z_1,\dots,z_p,z_1',\dots,z_{p'}')=\\

2046: {}=e^{2\pi

2047: iz_1'}\theta(z_1-z_1')\theta(z_p+z_{p'}')\cdot\prod_{1\le\alpha\le

2048: p'-1}\theta(z_\alpha'-z_{\alpha+1}'+z_{n_1'+\ldots+n_\alpha'-2\alpha+1})

2049: \\

2050: {}\times\prod_{1\le\beta\le

2051: p-1}\theta(z_\beta-z_{\beta+1}+z_{n_1+\ldots+n_\beta-2\beta+1}').

2052: \end{multline*}

2053: This function satisfies the following relations:

2054: $$

2055: \Delta_{n,k}(z_1,\dots,z_\alpha+1,\dots,z_{p'}')=

2056: \Delta_{n,k}(z_1,\dots,z_\beta'+1,\dots,z_{p'}')=

2057: \Delta_{n,k}(z_1,\dots,z_{p'}'),\\

2058: $$

2059: \begin{multline*}

2060: \Delta_{n,k}(z_1,\dots,z_\alpha+\tau,\dots,z_{p'}')\\

2061: =(-1)^{n_\alpha}e^{-2\pi i

2062: (n_\alpha z_\alpha-z_{\alpha-1}-z_{\alpha+1}-(\delta_{\alpha,1}-1)\tau)}

2063: \Delta_{n,k}(z_1,\dots,z_{p'}'),

2064: \end{multline*}

2065: \begin{multline*}

2066: \Delta_{n,k}(z_1,\dots,z_\beta'+\tau,\dots,z_{p'}')\\

2067: =(-1)^{n_\beta'}e^{-2\pi i

2068: (n_\beta'z_\beta'-z_{\beta-1}'-z_{\beta+1}'-(\delta_{\beta,1}-1)\tau)}

2069: \Delta_{n,k}(z_1,\dots,z_{p'}').

2070: \end{multline*}

2071: Here $z_0=z_{p+1}=z_0'=z_{p'+1}'=0$ and $\delta_{\alpha,1}$  stands for the

2072: Kronecker delta.

2073: \end{prop}

2074: 

2075:       The proof immediately follows from our description of the duality 

2076:  between the sequences

2077: $(n_1,\dots,n_p)$ and $(n_1',\dots,n_{p'}')$.

2078: 

2079: \begin{prop}                                                           

2080: \begin{equation}

2081: \Delta_{n,k}(z_1,\dots,z_p;z_1',\dots,z_{p'}')=c_{n,k}

2082: \sum_{\alpha\in\Z/n\Z}w_\alpha^{n/k}(z_1,\dots,z_p)

2083: w_{1-\alpha}^{n/n-k}(z_1',\dots,z_{p'}').

2084: \end{equation}

2085: Here $c_{n,k}\in\C$ is a constant.

2086: \end{prop}

2087: 

2088: \begin{proof} 

2089: It follows from the previous proposition that the function $\Delta_{n,k}$  

2090:  belongs to the space~$\Theta_{n/k}(\Gamma)$ when regarded as a function of the 

2091: variables $z_1,\dots,z_p$. Similarly, $\Delta_{n,k}$  belongs to~$\Theta_{n/n-k}(\Gamma)$

2092: as a function of $z_1',\dots,z_{p'}'$. Therefore,  

2093: $$

2094: \Delta_{n,k}(z_1,\dots,z_p;z_1',\dots,z_{p'}')=

2095: \sum_{\alpha,\beta\in\Z/n\Z}\lambda_{\alpha,\beta}

2096: w_\alpha^{n/k}(z_1,\dots,z_p)

2097: w_\beta^{n/n-k}(z_1',\dots,z_{p'}').

2098: $$

2099: However, one can readily see that

2100: $$

2101: \Delta_{n,k}(z_1+ r_1,\dots,z_p+ r_p;z_1'+ r_1',\dots,z_{p'}'+ r_{p'}')=

2102: e^{\frac{2\pi i}n}\Delta_{n,k}(z_1,\dots,z_{p'}'),

2103: $$

2104: where $ r_\alpha=\frac{d(n_1,\dots,n_{\alpha-1})}n$ and $

2105: r_\beta'=\frac{d(n_1',\dots,n_{\beta-1}')}n$. Hence,

2106: $\lambda_{\alpha,\beta}=0$ for  $\alpha+\beta\not\equiv1\mmod n$

2107: (because $w_\alpha(z_1+ r_1,\dots,z_p+ r_p)=e^{2\pi i\frac\alpha

2108: n}w_\alpha(z_1,\dots,z_p)$ and $w_\beta(z_1'+ r_1',\dots,z_{p'}'+

2109: r_{p'}')=e^{2\pi i\frac\beta

2110: n}w_\beta(z_1',\dots,z_{p'}')$). Thus,   

2111: $\lambda_{\alpha,\beta}=\lambda_\alpha\delta_{\alpha+\beta,1}$. Similarly,

2112: \begin{multline*}

2113: \Delta_{n,k}(z_1+ r_1\tau,\dots,z_p+ r_p\tau;z_1'+ r_1'\tau,

2114: \dots,z_{p'}'+ r_{p'}'\tau)\\

2115: {}=e^{2\pi i\left(\frac1n\tau-z_p-z_{p'}'\right)}

2116: \Delta_{n,k}(z_1,\dots,z_{p'}').

2117: \end{multline*}

2118: Hence, $\lambda_\alpha=\lambda_{\alpha+1}$, that is,

2119: $\lambda_\alpha$    does not depend on~$\alpha$.

2120: \end{proof}

2121: 

2122: \appendix{}

2123: 

2124: \subsection{Integrable system, quantum groups, and $R$-matrices}

2125: 

2126: One of the main methods in the investigation of exactly solvable models [6] in 

2127: quantum and statistical physics is the inverse problem method (see [45]). This 

2128: method leads to the study of representations of algebras of monodromy matrices, 

2129: that is, to the study of meromorphic matrix functions $L(u)$ satisfying the relations

2130: \begin{equation}

2131: R(u-v)L^1(u)L^2(v)=L^2(v)L^1(u)R(u-v).

2132: \end{equation}

2133: Here $R(u)$ is a chosen solution of the Yang-Baxter equation in the class of 

2134: meromorphic matrix-valued functions,

2135: \begin{equation}

2136: R^{12}(u-v)R^{13}(u)R^{23}(v)=R^{23}(v)R^{13}(u)R^{12}(u-v).

2137: \end{equation}

2138: We note that $R(u)$ takes the values in ($n^2\times n^2$) matrices with a fixed 

2139: decomposition $\Mat_{n^2}=\Mat_n\otimes\Mat_n$. We use the standard notation, 

2140: namely, $L^1=L\otimes1$, $L^2=1\otimes L$, $R^{12}=R\otimes1$, etc. (see [45]). 

2141: 

2142: In [45] Sklyanin studies the solutions of the equation (37) for the simplest 

2143: elliptic solution of the equation (38), that is, for the so-called Baxter $R$-matrix, 

2144: which is of the form

2145:  $R(u)=

2146: 1+\sum_{\alpha=1}^3W_\alpha(u)\sigma_\alpha\otimes\sigma_\alpha$, where

2147: $\sigma_1=\begin{pmatrix}0&1\\1&0\end{pmatrix}$,

2148: $\sigma_2=\begin{pmatrix}0&-i\\i&0\end{pmatrix}$, and

2149: $\sigma_3=\begin{pmatrix}1&0\\0&-1\end{pmatrix}$ are the Pauli matrices, and the 

2150: coefficients $W_\alpha(u)$ can be expressed in terms of the Jacobi elliptic 

2151: functions as follows:

2152: \begin{align*}

2153: W_1(u)&=\frac{\sn(i\eta,k)}{\sn(u+i\eta,k)},\quad

2154: W_2(u)=\frac{\dn}{\sn}(u+i\eta,k)\frac{\sn}{\dn}(i\eta,k),\\

2155: W_3(u)&=\frac{\cn}{\sn}(u+i\eta,k)\frac{\sn}{\cn}(i\eta,k).

2156: \end{align*}

2157: The functions $W_\alpha(u)$ uniformize the elliptic curve 

2158: $\frac{W_\alpha^2-W_\beta^2}{W_\gamma^2-1}=J_{\alpha,\beta}$, where the 

2159: $J_{\alpha,\beta}$s do not depend on $u$ and satisfy the relation 

2160: $J_{12}+J_{23}+J_{31}+J_{12}J_{23}J_{31}=0$. Here $\alpha,\beta$, and $\gamma$ 

2161: are pairwise distinct and $J_{\beta,\alpha}=-J_{\alpha,\beta}$. 

2162: 

2163: We note that this elliptic curve is the complete intersection of two quadrics, 

2164: for instance, $w_1^2-w_2^2=J_{12}(w_3^2-1)$ and $w_2^2-w_3^2=J_{23}(w_1^2-1)$. 

2165: 

2166: Sklyanin discovered that the equation (37) for the Baxter $R$-matrix has a solution 

2167: of the form $L(u)=S_0+\sum_{\alpha=1}^3W_\alpha(u)S_\alpha$, where $S_0$ 

2168: and $S_\alpha$ are matrices that do not depend on $u$ and satisfy the following 

2169: relations:

2170: \begin{equation}

2171: \begin{aligned}\relax

2172: [S_\alpha,S_0]_-&=-iJ_{\beta\gamma}[S_\beta,S_\gamma]_+,\\

2173: [S_\alpha,S_\beta]_-&=i[S_0,S_\gamma]_+,\\

2174: \end{aligned}

2175: \end{equation}

2176: where $[a,b]_\pm=ab\pm ba$. 

2177: 

2178: Sklyanin further studies the algebra with the generators $S_0,S_\alpha$ and the 

2179: relations (39); he denotes this algebra by $\F_{\eta,k}$. The main assumption 

2180: concerning this algebra is that it satisfies the PBW condition. Moreover, Sklyanin 

2181: finds the quadratic central elements of the algebra $\F_{\eta,k}$ and the 

2182: finite-dimensional representations of the algebra $\F_{\eta,k}$ by difference 

2183: operators in some function space (see [46]).

2184: 

2185: In our notation, the Sklyanin algebra $\F_{\eta,k}$ is the algebra $Q_4(\E,\eta)$, 

2186: where $\E$ is an elliptic curve given by the functions $W_\alpha(u)$, that is, a 

2187: complete intersection of two quadrics in $\C^3$.

2188: 

2189: The Yang-Baxter equation has other elliptic solutions generalizing the Baxter 

2190: solution (see [7]). The result of [7] can be described as follows: for any pair 

2191: of positive integers $n$ and $k$ such that $1\le  k<n$ and $n$ and $k$ are coprime 

2192: there is a family of solutions $R_{n,k}(\E,\eta)(u)$ of the equation (38). Here 

2193: $\E$ is an elliptic curve and $\eta\in\E$, as above. The Baxter solution is obtained 

2194: for $n=2$ and $k=1$.

2195: 

2196: According to [10], the Sklyanin result can be generalized to an arbitrary solution 

2197: $R_{n,k}(\E,\eta)(u)$. In our notation, the results of [10] look as follows: there 

2198: is a homomorphism of the algebra of monodromy matrices for the $R$-matrix 

2199: $R_{n,k}(\E,\eta)$ into the algebra $Q_{n^2,nk-1}(\E,\eta)$. Correspondingly, the 

2200: algebra $Q_{n^2,nk-1}(\E,\eta)$ is a deformation of the projectivization of the 

2201: Lie algebra $\slg_n$. Moreover, there is a homomorphism of the algebra of monodromy 

2202: matrices into the algebra $Q_{dn^2,dnk-1}(\E,\eta)$ for any $d\in\N$. It can be 

2203: conjectured that every finite-dimensional representation of the algebra of monodromy 

2204: matrices can be obtained from a representation of the algebra $Q_{dn^2,dnk-1}(\E,\eta)$. 

2205: 

2206: Another relationship between the elliptic solutions of the Yang-Baxter equation 

2207: and the elliptic algebras follows from the results of [10]. The multiplication in 

2208: the algebra $Q_{n,k}(\E,\eta)$ is defined by the so-called Young projections 

2209: $S^\alpha V\otimes S^\beta V\to S^{\alpha+\beta}V$ corresponding to 

2210: $R_{n,k}(\E,\eta)(u)$ (see [10]). Moreover, $Q_{n,k}(\E,\eta)=\sum_\alpha S^\alpha V$.

2211: 

2212: We also note that the study of algebras $Q_{n,k}(\E,\eta)$ and their representations 

2213: led to deeper understanding of the structure of $R$-matrices $R_{n,k}(\E,\eta)(u)$ 

2214: and of the corresponding algebraic objects (the Zamolodchikov algebra and the 

2215: algebra of monodromy matrices). For this topic, see [33].

2216: 

2217: \subsection{Deformation quantization}

2218: 

2219: Let $M$ be a manifold ($C^\infty$, analytic, algebraic, etc.) and let $\F(M)$ 

2220: be a function algebra on $M$. In the "physical" language, $M$ is the state space 

2221: of the system and $\F(M)$ is the algebra of observables. In [5] the following 

2222: approach to the quantization was suggested: the underlying vector space of the 

2223: quantum algebra of observables coincide with that of $\F(M)$, but the multiplication 

2224: is deformed and is no longer commutative (though still associative). Moreover, 

2225: the multiplication depends on the deformation parameter (Planck constant). For 

2226: $\hbar=0$ we have the ordinary commutative multiplication. Since the Planck 

2227: constant is small, we do not notice that the observables in classical mechanics 

2228: are non-commutative. Expanding the multiplication in the series in powers of $\hbar$ 

2229: we obtain $f*g=fg+\{f,g\}\hbar+o(\hbar)$. The operation 

2230: $\{{\cdot},{\cdot}\}\colon\F(M)\otimes\F(M)\to\F(M)$ is bilinear, and, applying 

2231: the gauge transformations, one can make it anticommutative, $\{f,g\}=-\{g,f\}$. 

2232: Moreover, since the multiplication $*$ is associative, we see that 

2233: $\{f,gh\}=\{fg\}h+\{f,h\}g$ (the Leibniz rule) and 

2234: $\{f,\{g,h\}\}+\{h,\{f,g\}\}+\{g,\{h,f\}\}=0$ (the Jacobi identity). The Leibniz 

2235: rule means that $\{f,g\}=\langle w,df\wedge dg\rangle$, where $w$ is a bivector 

2236: field on $M$, and the Jacobi identity means that $[w,w]=0$. Thus, every quantization 

2237: defines a Poisson structure $w$ (or $\{{\cdot},{\cdot}\}$) on the manifold $M$. 

2238: The inverse problem arises: construct a quantization $*$ from a manifold $M$ 

2239: with Poisson structure (a Poisson manifold). This problem was solved in [26] at 

2240: the formal level. Namely, bidifferential operators $B_n$  ($n\ge2$) were constructed 

2241: on a Poisson manifold $M$ in such a way that the formal series 

2242: \begin{equation}

2243: f*g=fg+\{f,g\}\hbar+\sum_{n\ge2}B_n(f,g)\hbar^n

2244: \end{equation}

2245: satisfies the condition $(f*g)*h=f*(g*h)$. Moreover, the operators $B_n$ are 

2246: constructed from $\{{\cdot},{\cdot}\}$ by an explicit formula. Thus, the problem 

2247: of formal quantization was solved; however, the problem on the convergence of the 

2248: series (40) and on its identification remains open. This problem seems to be very 

2249: complicated. For instance, let $M=\C^n$,  $\F(M)=S^*(\C^n)=\C[x_1,\dots,x_n]$, 

2250: and let the Poisson bracket be quadratic ($\{x_i,x_j\}=\sum_{\alpha,\beta}c_{ij}^{\alpha\beta}x_\alpha

2251: x_\beta$, where $c_{ij}^{\alpha\beta}\in\C$ are symmetric with respect to $\alpha,\beta$ 

2252: and antisymmetric with respect to $i,j$). In this case, the multiplication $*$ 

2253: must be homogeneous, that is, $S^\alpha(\C^n)*S^\beta(\C^n)\subset

2254: S^{\alpha+\beta}(\C^n)$. Therefore, according to (40), the structure constants 

2255: of the multiplication $*$ in the basis $\{x_1^{\alpha_1}\dots

2256: x_n^{\alpha_n};\alpha_1,\dots,\alpha_n\in\Z_{\ge0}\}$ turn out to be formal series 

2257: in $\hbar$ which baffle the explicit evaluation even in the simplest case $n=2$,

2258: $\{x_1,x_2\}=\alpha x_1x_2$. In this case it is natural to assume that the quantum 

2259: algebra must be defined by the relation $x_1x_2=e^{-\alpha\hbar}x_2x_1$. On the 

2260: other hand, the algebras $Q_{n,k}(\E,\eta)$ introduced above are examples of the 

2261: quantization of $M=\C^n$, where $\eta$ plays the role of Planck constant because 

2262: $Q_{n,k}(\E,0)=\C[x_1,\dots,x_n]$. Moreover, the structure constants of the 

2263: algebra $Q_{n,k}(\E,\eta)$ turn out to be elliptic functions of $\eta$. There are 

2264: also rational and trigonometric limits of the algebras $Q_{n,k}(\E,\eta)$ in which 

2265: the structure constants are rational (trigonometric) functions of $\eta$ (see [32]).

2266: 

2267: \subsection{Moduli spaces}

2268: 

2269: Let $G$ be a semisimple Lie group and let $P\subset G$ be a parabolic subgroup. 

2270: Let $\M(\E,P)$ be the moduli space of the holomorphic $P$-bundles over an elliptic 

2271: curve $\E$ [4]. According to [20], every connected component of the space $\M(\E,P)$ 

2272: admits a natural Poisson structure. The main property of this structure is as 

2273: follows: the preimages of the natural map $\M(\E,P)\to\M(\E,G)$ corresponding to 

2274: forgetting the $P$-structure are symplectic leaves of the structure. The quantization 

2275: problem for the Poisson manifold $\M(\E,P)$ arises. The solution of this problem 

2276: could establish a relationship between the natural algebro-geometric problem of 

2277: studying $P$-bundles (and the corresponding $G$-bundles) and the problem to study 

2278: representations of the quantum function algebra on $\M(\E,P)$ because the representations 

2279: correspond to symplectic leaves.

2280: 

2281: In [38], [22] these quantum algebras were constructed provided that $P=B$ is a 

2282: Borel subgroup of an arbitrary group $G$. In [21] the quantum algebras were 

2283: constructed in the case of $G=GL_m$ and an arbitrary parabolic subgroup $P$. Here 

2284: the algebra $Q_n(\E,\eta)$ corresponds to the case $G=GL_2$, and the algebra 

2285: $Q_{n,k}(\E,\eta)$ to the case $G=GL_{k+1}$, where $P$ consists of upper block 

2286: triangular matrices of the form

2287: $\vcenter{\raisebox{5pt}{\bm{\mathstrut\cr

2288: k&&*&&*\cr\mathstrut\cr1&&0&&*\cr&&k&&1}}

2289: \vspace{20pt}.}$

2290: 

2291: \subsection{Non-commutative algebraic geometry}

2292: 

2293: One of the main ideas of algebraic geometry is to study the geometry of a manifold 

2294: by using the algebraic properties of a ring of functions on this manifold. The 

2295: non-commutative algebraic geometry extends these methods and the geometric intuition 

2296: to an appropriate class of non-commutative rings. In [51], [48] the non-commutative 

2297: algebraic geometry is developed in small dimensions. From this point of view, 

2298: the algebras $Q_{n,k}(\E,\eta)$ give examples of non-commutative vector spaces. 

2299: Similar examples of non-commutative Grassmannians ans other varieties are also 

2300: known [11], [19], [22], [49].

2301: 

2302: \subsection{Cohomology of algebras}

2303: 

2304: Cohomology properties of quadratic algebras are studied in [29], [41]-[44], [50]. 

2305: For generic $\eta$, the algebras $Q_{n,k}(\E,\eta)$ are examples of Koszul algebras. 

2306: One can readily prove this fact for $k=1$ by using the construction of a free 

2307: module in \S2.6. The constructions of dual algebras $Q_{n,k}^!(\E,\eta)$ are 

2308: given in [36].

2309: 

2310: \begin{thebibliography}{XX} 

2311: \addcontentsline{toc}{section}{References}

2312: 

2313: 

2314: \item %\key 1

2315:  M.~Artin and W.~Schelter,

2316:   ``Graded algebras of global dimension~$3$'',

2317:  Adv. Math.

2318:  66

2319:  (1987),

2320:  171--216.

2321: 

2322: 

2323: \item %\key 2

2324:  M.~Artin, J.~Tate, and M.~Van den Bergh,

2325:   ``Some algebras associated to automorphisms of elliptic curves'',

2326:  The Grothendieck Festschrift.~\rm{Vol.~I}, 

2327:  (Progr. Math., Vol.~86) Birkh\"auser Boston, Boston, MA, 

2328:  1990, pp.

2329:  33--85.

2330: 

2331: 

2332: \item %\key 3

2333:  M.~Artin, J.~Tate, and M.~Van den Bergh,

2334:   ``Modules over regular algebras of dimension~$3$'',

2335:  Invent. Math.

2336:  106

2337:  (1991),

2338: %issue 2

2339:  335--388.

2340: 

2341: 

2342: \item %\key 4

2343:  M.~F.~Atiyah,

2344:   ``Vector bundles over an elliptic curve'',

2345:  Proc. London Math. Soc.~\rm{(3)}

2346:  7

2347:  (1957),

2348:  414--452.

2349: 

2350: 

2351: \item %\key 5

2352:  F.~Bayen, M.~Flato, C.~Fr\o nsdal, A.~Lichnerowicz, and D.~Sternheimer,

2353:   ``Deformation theory and quantization. \rm{I.}

2354: Deformations of symplectic structures'',

2355:  Ann. Physics

2356:  111

2357:  (1978),

2358: %issue 1

2359:  61--110.

2360: 

2361: 

2362: \item %\key 6

2363:  R.~J.~Baxter,

2364:  Exactly solved models in statistical mechanics,

2365:  London-New York,~

2366:  Academic Press,

2367:  1982;

2368:  Russian transl.

2369:  Moscow,~

2370:  Mir,

2371:  1985.

2372: 

2373: 

2374: \item %\key 7

2375:  A.~A.~Belavin,

2376:   ``Discrete groups and the integrability of quantum systems'',

2377:  Funktsional. Anal. i Prilozhen.

2378:  14 

2379:  (1980), no. 4,

2380:  18--26;

2381:  English transl. in {\it Funct. Anal. Appl.}

2382:  14

2383:  (1981),

2384:  18--26.

2385: 

2386: 

2387: \item %\key 8

2388:  G.~Bergman,

2389:   ``The diamond lemma for ring theory'',

2390:  Adv. Math. 29 

2391:  (1978), 

2392: %issue 2

2393:  178--218.

2394: 

2395: 

2396: \item %\key 9

2397:  H.~W.~Braden, A.~Gorsky, A.~Odesskii, and V.~Rubtsov,

2398:   ``Double-elliptic dynamical systems from generalized

2399: Mukai--Sklyanin algebras'',

2400:  Nuclear Phys.~B 633

2401:  (2002),

2402: %issue 3

2403:  414--442.

2404: 

2405: 

2406: \item %\key 10

2407:  I.~V.~Cherednik,

2408:   ``On $R$-matrix quantizations of current group'',

2409:  Group-Theoretic Methods in Physics, Vol.~2

2410: (Proc. of Yurmala Conference, May, 1985),

2411:  Moscow,~

2412:  Nauka,

2413:  1980,

2414:  pp. 218--232

2415: 

2416: 

2417: \item %\key 11

2418:  A.~Connes and M.~Dubois-Violette,

2419:  Noncommutative finite-dimensional manifolds.

2420: \rm{I.}~Spherical manifolds and related examples, 

2421:  math.QA/0107070.

2422: 

2423: 

2424: \item %\key 12

2425:  V.~G.~Drinfel'd, 

2426:   ``Quadratic commutation relations in the quasi-classical case'',

2427:  in Collection of Sci. Works:

2428: Mathematical Physics and Functional Analysis,

2429:  Kiev,~

2430:  Naukova Dumka,

2431:  1986,

2432:  pp. 25--34;

2433:  English transl. in {\it Sel. Math. Soviet},

2434: vol. 11, 1992,

2435:  pp. 317--326.

2436: 

2437: 

2438: \item %\key 13

2439:  V.~G.~Drinfel'd,

2440:   ``A new realization of Yangians and quantized affine algebras'',

2441:  Dokl. Akad. Nauk SSSR 269

2442:  (1987), 

2443: no. 1,

2444:  13--17;

2445:  English transl. in {\it Soviet Math. Dokl.} 36

2446:  (1988), 

2447: %issue 2

2448:  212--216.

2449: 

2450: 

2451: \item %\key 14

2452:  B.~Enriquez and V.~N.~Rubtsov,

2453:   ``Quantum groups in higher genus

2454: and Drinfeld's new realizations method (${\frak{sl}}_2$ case)'',

2455:  Ann. Sci. \'Ecole Norm. Sup.~\rm{(4)}

2456:  30

2457:  (1997),

2458: %issue 6

2459:  821--846.

2460: 

2461: 

2462: \item %\key 15

2463:  H.~Ewen and O.~Ogievetsky,

2464:  Classification of quantum matrix groups in $3$ dimensions,

2465:  Preprint MPI-Ph/94-1993.

2466: 

2467: 

2468: \item %\key 16

2469:  B.~Feigin and E.~Frenkel,

2470:   ``Quantum $\ W$-algebras and elliptic algebras'',

2471:  Comm. Math. Phys.

2472:  178

2473:  (1996),

2474: %issue 3

2475:  653--678.

2476: 

2477: 

2478: \item %\key 17

2479:  B.~Feigin, M.~Jimbo, T.~Miwa, A.~Odesskii, and Ya.~Pugai,

2480:   ``Algebra of screening operators for the deformed $W_n$ algebra'',

2481:  Comm. Math. Phys.

2482:  191

2483:  (1998),

2484: %issue 3

2485:  501--541.

2486: 

2487: 

2488: \item %\key 18

2489:  B.~Feigin and A.~Odesskii,

2490:   ``A family of elliptic algebras'',

2491:  Internat. Math. Res. Notices 11

2492:  (1997), 

2493:  531--539.

2494: 

2495: 

2496: \item %\key 19

2497:  B.~L.~Feigin and A.~V.~Odesskii,

2498:   ``Coordinate ring of the quantum Grassmanian

2499: and intertwiners for the representations of Sklyanin algebras'',

2500:  Topics in Quantum Groups and Finite-Type Invariants (B.~Feigin et al, eds.), 

2501:  (Amer. Math. Soc. Transl. Ser.~2, Vol.~185) Amer. Math. Soc., Providence, RI, 

2502:  1998, pp. 

2503:  55--64.

2504: 

2505: 

2506: \item %\key 20

2507:  B.~L.~Feigin and A.~V.~Odesskii,

2508:   ``Vector bundles on an elliptic curve and Sklyanin algebras'',

2509:  Topics in Quantum Groups and Finite-Type Invariants

2510:  (B.~Feigin et al, eds.), 

2511:  (Amer. Math. Soc. Transl. Ser.~2, Vol.~185) Amer. Math. Soc., 

2512:  Providence,~RI~,

2513:  1998, pp.

2514:  65--84, q-alg/9509021.

2515: 

2516: 

2517: \item %\key 21

2518:  B.~L.~Feigin and A.~V.~Odesskii,

2519:   ``Functional realization of some elliptic Hamiltonian

2520: structure and bosonization of the corresponding quantum algebras'',

2521:  Integrable Structures of Exactly Solvable Two-Dimensional

2522: Models of Quantum Field Theory (S.~Pakuliak et al, eds.),

2523:   (NATO Sci. Ser.~II Math. Phys. Chem., Vol.~35) Kluwer, Dordrecht~,

2524:  2001, pp.

2525:  109--122.

2526: 

2527: 

2528: \item %\key 22

2529:  B.~L.~Feigin and A.~V.~Odesskii,

2530:   ``Quantized moduli spaces of the bundles on

2531: the elliptic curve and their applications'',

2532:  Integrable Structures of Exactly Solvable Two-Dimensional

2533: Models of Quantum Field Theory (S.~Pakuliak et al, eds.),

2534:  (NATO Sci. Ser.~II Math. Phys. Chem., Vol.~35) Kluwer, Dordrecht~, 

2535:  2001, pp.

2536:  123--137

2537: 

2538: 

2539: \item %\key 23

2540:  G.~Felder,

2541:   ``Elliptic quantum groups'',

2542:  Proc. of the XIth International Congress on

2543: Mathematical Physics (Paris, July 18--23, 1994),

2544:  International Press,

2545:  Cambridge,~MA, 1995, pp. 211--218.

2546:  

2547: 

2548: 

2549: \item %\key 24

2550:  G.~Felder and V.~Pasquier,

2551:   ``A simple construction of elliptic $R$-matrices'',

2552:  Lett. Math. Phys.

2553:  32

2554:  (1994),

2555:  167--171.

2556: 

2557: 

2558: \item %\key 25

2559:  M.~Jimbo,

2560:   ``Topics from Representations of $U_q(\frak G)$.

2561: An introductory guide to physicists'',

2562:  Quantum Group and Quantum Integrable Systems,

2563:  River Edge,~World Scientific,

2564:   1992, pp.

2565:  1--61.

2566: 

2567: 

2568: \item %\key 26

2569:  M.~Kontsevich,

2570:  Deformation quantization of Poisson manifolds.~\rm{I},

2571:  math/9709180.

2572: 

2573: 

2574: \item %\key 27

2575:  M.~Kontsevich and A.~Rosenberg,

2576:   ``Noncommutative smooth spaces'',

2577:  The Gelfand Mathematical Seminars, 1996--1999

2578:  (I.~M.~Gelfand et al, eds.), 

2579:  (Gelfand Math. Sem.) Birkh\"auser Boston, Boston,~MA,~

2580:  2000, pp.

2581:  85--108.

2582: 

2583: 

2584: \item %\key 28

2585:  I.~Krichever and A.~Zabrodin, 

2586:   ``Spin generalization of the Ruijsenaars--Schneider model,

2587: the non-Abelian $2D$ Toda chain, and representations of the Sklyanin algebra'',

2588:  Uspekhi Mat. Nauk

2589:  50

2590:  (1995),

2591: no. 6,

2592:  3--56;

2593:  English transl. in {\it Russian Math. Surveys}

2594:  50 (1995), 

2595:  1101--1150.

2596:  

2597: 

2598: 

2599: \item %\key 29

2600:  Yu.~I.~Manin,

2601:   ``Some remarks on Koszul algebras and quantum groups'',

2602:  Ann. Inst. Fourier \rm{(Grenoble)}

2603:  37 (1987),

2604: no. 4,

2605:  191--205.

2606:  

2607: 

2608: 

2609: \item %\key 30

2610:  D.~Mumford,

2611:  Lectures on theta functions, 

2612:  Novokuznetsk Phys. Math. Inst., Novokuznetsk, 

2613:  1998.

2614: % D.~Mumford

2615: % Tata lectures on theta.

2616: % I: Introduction and motivation: Theta functions in one variable.

2617: % Basic results on theta functions in several variables.

2618: % (Progr. Math., Vol. 28) Bikhauser, Boston-Basel-Stuttgart, 1983

2619: %

2620: % Tata lectures on theta.

2621: % II: Jacobian theta functions and differential equations.

2622: % (Progr. Math., Vol. 43) Bikhauser, Boston-Basel-Stuttgart, 1984

2623: 

2624: 

2625: \item %\key 31

2626:  A.~V.~Odesskii,

2627:   ``An analog of the Sklyanin algebra'',

2628:  Funktsional. Anal. i Prilozhen.

2629:  20 (1986),

2630: no. 2,

2631:  78--79;

2632:  English transl. in {\it Funct. Anal. Appl.}

2633:  20

2634: %issue 2

2635:  (1986),

2636:  152--154.

2637: 

2638: 

2639: \item %\key 32

2640:  A.~V.~Odesskii,

2641:   ``Rational degeneration of elliptic quadratic algebras'',

2642:  Infinite Analysis, Part A,~B (Kyoto, 1991),

2643:  (Adv. Ser. Math. Phys., Vol.~16) World Scientific

2644:  River Edge,~NY,~

2645:  1992, pp.

2646:  773--779.

2647: 

2648: 

2649: \item %\key 33

2650:  A.~V.~Odesskii,

2651:   ``Belavin elliptic $R$-matrices and exchange algebras'',

2652:  Funktsional. Anal. i Prilozhen. 36

2653:  (2002), 

2654: no. 1,

2655:  59--74;

2656:  English transl. in {\it Funct. Anal. Appl.}

2657:  36

2658:  (2002),

2659: %issue 1

2660:  49--61.

2661: 

2662: 

2663: \item %\key 34

2664:  A.~V.~Odesskii and B.~L.~Feigin,

2665:  Sklyanin's algebras associated to elliptic curves, 

2666:  Inst. Theor. Phys., Akad. Nauk Ukrain. SSR, Kiev, 

2667:  1988. 

2668:  (Russian)

2669: 

2670: 

2671: \item %\key 35

2672:  A.~V.~Odesskii and B.~L.~Feigin, 

2673:   ``Sklyanin's elliptic algebras'', 

2674:  Funktsional. Anal. i Prilozhen. 23

2675:  (1989),

2676: no. 3,

2677:  45--54;

2678:  English transl. in {\it Funct. Anal. Appl.}

2679:  23

2680:  (1990),

2681: %issue 3

2682:  207--214.

2683: 

2684: 

2685: \item %\key 36

2686:  A.~V.~Odesskii and B.~L.~Feigin,

2687:   ``Constructions of elliptic Sklyanin algebras

2688: and of quantum $R$-matrices'',

2689:  Funktsional. Anal. i Prilozhen.

2690:  27 

2691:  (1993), no.1, 

2692:  37--45;

2693:  English transl. in {\it Funct. Anal. Appl.}

2694:  27

2695:  (1993),

2696: %issue 1

2697:  31--38.

2698: 

2699: 

2700: \item %\key 37

2701:  A.~V.~Odesskii and B.~L.~Feigin,

2702:   ``Sklyanin's elliptic algebras.

2703: The case of a point of finite order'',

2704:  Funktsional. Anal. i Prilozhen. 29

2705:  (1995), 

2706: no. 2,

2707:  9--21;

2708:  English transl. in {\it Funct. Anal. Appl.}

2709:  29

2710:  (1995),

2711: %issue 2

2712:  81--90.

2713: 

2714: 

2715: \item %\key 38

2716:  A.~V.~Odesskii and B.~L.~Feigin,

2717:   ``Elliptic deformations of current algebras

2718: and their representations by difference operators'',

2719:  Funktsional. Anal. i Prilozhen. 31

2720:  (1997), 

2721: no. 3,

2722:  57--70;

2723:  English transl. in {\it Funct. Anal. Appl.}

2724:  31

2725:  (1998),

2726: %issue 3

2727:  193--203.

2728: 

2729: 

2730: \item %\key 39

2731:  A.~V.~Odesskii and V.~Rubtsov,

2732:   ``Polynomial Poisson algebras with regular structure

2733: of symplectic leaves'',

2734:  Teoret. Mat. Fiz. 133

2735:  (2002), 

2736: no. 1,

2737:  3--23;

2738:  English transl. in {\it Theoret. and Math. Phys.} 133

2739:  (2002),

2740: %issue 1

2741:  1321--1337.

2742: 

2743: 

2744: \item %\key 40

2745:  O.~Ogievetsky,

2746:   ``Uses of quantum spaces'',

2747:  Quantum Symmetries in Theoretical Physics and Mathematics

2748: (Bariloche, 2000)

2749:  (R.~Coquereaux et al, eds.), 

2750:  (Contemp. Math., Vol.~294) Amer. Math. Soc., Providence,~RI, 2002, pp.

2751:  161--232.

2752:  

2753: 

2754: 

2755: \item %\key 41

2756:  A.~Polishchuk and L.~Positselski,

2757:  Quadratic algebras,

2758:  Preprint,

2759:  1996.

2760: 

2761: 

2762: \item %\key 42

2763:  S.~B.~Priddy,

2764:   ``Koszul resolutions'',

2765:  Trans. Amer. Math. Soc.

2766:  152 (1970),

2767: no. 1,

2768:  39--60.

2769:  

2770: 

2771: 

2772: \item %\key 43

2773:  J.-E.~Roos,

2774:   ``On the characterisation of Koszul algebras.

2775: Four counter-examples'',

2776:  C.~R.~Acad. Sci. Paris S\'er.~\rm{I} Math.

2777:  321 (1995),

2778: no. 1,

2779:  15--20.

2780: 

2781: 

2782: \item %\key 44

2783:  B.~Shelton and C.~Tingey,

2784:   ``On Koszul algebras and a new construction

2785: of Artin--Schelter regular algebras'',

2786:  J.~Algebra

2787:  241

2788:  (2001),

2789: %issue 2

2790:  789--798.

2791: 

2792: 

2793: \item %\key 45

2794:  E.~K.~Sklyanin,

2795:   ``Some algebraic structures connected with

2796: the Yang--Baxter equation'',

2797:  Funktsional. Anal. i Prilozhen. 16

2798:  (1982), 

2799: no. 4,

2800:  27--34;

2801:  English transl. in {\it Funct. Anal. Appl.}

2802:  16 (1983),

2803: %issue 4

2804:  263--270.

2805:  

2806: 

2807: 

2808: \item %\key 46

2809:  E.~K.~Sklyanin,

2810:   ``Some algebraic structures connected with

2811: the Yang--Baxter equation. Representations of a quantum algebra'',

2812:  Funktsional. Anal. i Prilozhen. 17

2813:  (1983), 

2814: no. 4,

2815:  34--48;

2816:  English transl. in {\it Funct. Anal. Appl.}

2817:  17

2818:  (1983),

2819: %issue 4

2820:  273--284.

2821: 

2822: 

2823: \item %\key 47

2824:  S.~P.~Smith and J.~Tate,

2825:   ``The centre of the $3$-dimensional and $4$-dimensional

2826: Sklyanin algebras'',

2827:  $K$-Theory 8

2828:  (1994), 

2829: no. 1,

2830:  19--63.

2831: 

2832: 

2833: \item %\key 48

2834:  J.~T.~Stafford and M.~Van den Bergh,

2835:   ``Noncommutative curves and noncommutative surfaces'',

2836:  Bull. Amer. Math. Soc. \rm{(N.S.)}

2837:  38

2838:  (2001),

2839: %issue 2

2840:  171--216.

2841: 

2842: 

2843: \item %\key 49

2844:  J.~T.~Stafford and J.~J.~Zhang,

2845:   ``Examples in non-commutative projective geometry'',

2846:  Math. Proc. Cambridge Philos. Soc.

2847:  116

2848:  (1994),

2849: %issue 3

2850:  415--433.

2851: 

2852: 

2853: \item %\key 50

2854:  J.~Tate,

2855:   ``Homology of Noetherian rings and local rings'',

2856:  Illinois J.~Math.

2857:  1

2858:  (1957),

2859:  14--27.

2860: 

2861: 

2862: \item %\key 51

2863:  M.~Van den Bergh,

2864:  Blowing up of non-commutative smooth surfaces, 

2865:  (Mem. Amer. Math. Soc., Vol.~734) Amer. Math. Soc., Providence,~RI, 

2866:  2001.

2867: 

2868: 

2869: \item %\key 52

2870:  A.~M.~Vershik,

2871:   ``Algebras with quadratic relations'',

2872:  Spectral Theory of Operators and Infinite-Dimensional Analysis, 

2873:  Inst. Math., Akad. Nauk Ukrain. SSR, Kiev, 

2874:  1984.

2875:  (Russian)

2876: 

2877: 

2878: \end{thebibliography}

2879: 

2880: \bigskip

2881: \begin{flushleft}

2882: L. D. Landau Institute of Theoretical Physics,\\

2883: Russian Academy of Sciences\\

2884: \medskip

2885: e-mail: odesskii@mccme.ru

2886: 

2887: \end{flushleft}

2888: 

2889: \typeout{LaTeX Warning: Label(s) may have changed. Rerun}

2890: \end{document}

2891: