1: \mag1200

2: \documentclass[matzei,numbook]{svjour}

3: \usepackage{amsmath,amsfonts,amssymb}

4: 

5: %------------------------------------------------------------------------------

6: 

7: \spnewtheorem*{Theorem}{Theorem}{\bf}{\it}

8: \spnewtheorem*{Proposition}{Proposition}{\bf}{\it}

9: \spnewtheorem*{Corollary}{Corollary}{\bf}{\it}

10: \spnewtheorem*{Lemma}{Lemma}{\bf}{\it}

11: 

12: \hoffset11truemm%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

13: %\voffset-4truemm%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

14: 

15: %------------------------------------------------------------------------------

16: 

17: \def\al{\alpha}

18: \def\AMN{{\rm A}_{NM}}

19: 

20: \def\Bla{{\rm B}_{\la\lat}}

21: \def\Bll{{\rm B}_{l\ts\lt}}

22: \def\Bmm{{\rm B}_{\ts m\mt}}

23: \def\Bnn{{\rm B}_{\ts n\nt}}

24: \def\Bt{\ts\widetilde{\ns B\ts}}

25: 

26: \def\bx{{\boxed{\phantom{\square}}\kern-.4pt}}

27: 

28: \def\CC{{\mathbb C}}

29: \def\com{\ts,\hskip-.5pt}

30: 

31: \def\de{\delta}

32: \def\De{\Delta}

33: 

34: \def\End{\operatorname{End}\ts}

35: 

36: \def\FLa{F_{\ns\La\Lat}}

37: \def\FOm{F_{\Om\Omt}}

38: 

39: \def\ga{\gamma}

40: \def\gap{\gamma^{\,\prime}}

41: \def\ge{\geqslant}

42: \def\glL{{\mathfrak{gl}_L}}

43: \def\glM{{\mathfrak{gl}_M}}

44: \def\glMN{{\mathfrak{gl}_{\ts N+M}}}

45: \def\glN{{\mathfrak{gl}_N}}

46: \def\GOm{G_{\Om\Omt}}

47: \def\GOmp{G_{\Om\Omt}^{\,\,\prime}}

48: 

49: \def\id{{\rm id}}

50: \def\Ill{{\rm I}_{\ts l\ts\lt}}

51: \def\Imm{{\rm I}_{\ts m\mt}}

52: \def\Inn{{\rm I}_{\ts n\nt}}

53: 

54: \def\io{\iota}

55: 

56: \def\kt{{\widetilde{k}}}

57: 

58: \def\la{\lambda}

59: \def\La{\Lambda}

60: \def\lap{\la^{\ts\prime}}

61: \def\Lap{\La^{\ts\prime}}

62: \def\Lapp{\La^{\ts\prime\prime}}

63: \def\lat{\widetilde{\la}}

64: \def\Lat{\tilde{\La}}

65: \def\Latp{\Lat^{\ts\prime}}

66: \def\Latpp{\Lat^{\ts\prime\prime}}

67: \def\lcd{\ts,\,\ldots,}

68: \def\le{\leqslant}

69: \def\lm{{\la/\ns\mu}}

70: \def\lmt{{\lat/\mut}}

71: \def\lt{{\tilde{l}}}

72: 

73: \def\mi{{\raise.5pt\hbox{-}}}

74: \def\mt{{\widetilde{m}}}

75: \def\mup{\mu^{\ts\prime}}

76: \def\mut{\widetilde{\mu}}

77: \def\mw{\kern-81pt}

78: 

79: \def\ns{\hskip-1pt}

80: \def\nt{{\widetilde{n}}}

81: \def\nup{\nu^{\ts\prime}}

82: \def\nut{\widetilde{\nu}}

83: 

84: \def\om{\omega}

85: \def\Om{\Omega}

86: \def\Omt{{\widetilde{\Om}}}

87: \def\op{\oplus}

88: \def\ot{\otimes}

89: 

90: \def\ph{\varphi}

91: \def\Pp{P^{\vee}}

92: \def\Pt{\ts\widetilde{\ns P\ts}}

93: 

94: \def\Qp{Q^{\,\vee}}

95: 

96: \def\Ra{R^{\hskip1pt\ast}}

97: \def\Rb{\,\overline{\hskip-2.5pt R\hskip-4.5pt\phantom{\bar{t}}}\ts}

98: \def\Rbp{\Rb^{\,\vee}}

99: \def\Rp{R^{\,\vee}}

100: \def\RR{{\mathbb R}}

101: \def\Rt{\ts\widetilde{\ns R\ts}}

102: 

103: \def\si{\sigma}

104: \def\sgn{\operatorname{sgn}\ts}

105: \def\Sll{S_{\ts\lt\ts l}}

106: 

107: \def\T{{\cal T}}

108: \def\th{\theta}

109: \def\Th{\Theta}

110: \def\ts{\hskip1pt}

111: 

112: \def\UL{\operatorname{U}(\glL)}

113: \def\UM{\operatorname{U}(\glM)}

114: \def\UMN{\operatorname{U}(\glMN)}

115: \def\UN{\operatorname{U}(\glN)}

116: \def\Up{\Upsilon}

117: \def\Upt{\widetilde{\Upsilon}}

118: 

119: \def\Vla{V_{\la\ts\lat}}

120: \def\VLa{V_{\ns\La\Lat\ts}}

121: \def\Vlm{V_{\ns\la\lat}^{\,\mu\ts\mut}}

122: \def\VOm{V_{\Om\Omt}}

123: 

124: \def\Wkk{W_{k\ts\kt}}

125: \def\Wkko{W_{k\ts\kt}^{\hskip1pt\circ}}

126: \def\Wll{W_{l\ts\lt}}

127: \def\Wllo{W_{l\ts\lt}^{\hskip.5pt\circ}}

128: \def\Wmm{W_{m\mt}}

129: \def\Wmmo{W_{m\mt}^{\ts\circ}}

130: \def\Wnn{W_{n\ts\nt}}

131: \def\Wnno{W_{n\ts\nt}^{\,\circ}}

132: \def\Wp{W^{\hskip.5pt\prime}}

133: 

134: \def\YL{\operatorname{Y}(\glL)}

135: \def\YMN{\operatorname{Y}(\glMN)}

136: \def\YN{\operatorname{Y}(\glN)}

137: 

138: \def\Zmm{Z_{m\mt}}

139: \def\Zp{Z^{\hskip.5pt\prime}}

140: \def\zt{{\widetilde{z}}}

141: \def\ZZ{{\mathbb Z}} 

142: 

143: %==============================================================================

144: 

145: \begin{document}

146: \title{Rational representations of Yangians\\

147: associated with skew Young diagrams\\}

148: \author{Maxim Nazarov}

149: \institute{Department of Mathematics, University of York,

150: York YO10 5DD, England\\\email{mln1@york.ac.uk\\ }}

151: \titlerunning{Rational representations of Yangians}

152: \authorrunning{Maxim Nazarov}

153: \maketitle

154: 

155: %------------------------------------------------------------------------------

156: 

157: \vskip-15pt

158: \begin{abstract}

159: {\bf\hskip-10pt.}\hfill  

160: Consider the general linear group $GL_M$ over the complex field. 

161: \newline

162: The irreducible rational representations of the group $GL_M$

163: can be labeled~by the pairs of partitions 

164: $\mu=(\mu_{\ts1},\mu_{\ts2}\ts,\,\ldots)$

165: and

166: $\mut=(\mut_{\ts1},\mut_{\ts2}\ts,\,\ldots)$

167: such that the~total number of non-zero parts of $\mu$ and $\mut$

168: does not exceed $M$. Let $V_{\mu\ts\mut}$ be the irreducible 

169: representation corresponding to such a pair.

170: 

171: Regard the direct product

172: $GL_N\times GL_M$ as a subgroup of $GL_{N+M}\ts$.

173: Take any irreducible rational representation

174: $V_{\la\lat}$ of $GL_{N+M}\,$. 

175: The vector space

176: $$

177: \Vlm={\rm Hom}_{\,GL_M}(\ts V_{\mu\ts\mut}\ts\com V_{\la\lat}\ts)

178: $$

179: comes with a natural action of $GL_N\ts$. Put 

180: $n=\la_1-\mu_1+\la_2-\mu_2+\ldots$

181: and $\nt=\lat_1-\mut_1+\lat_2-\mut_2+\ldots\ $.

182: For any pair of standard Young tableaux 

183: $\Om$ and $\Omt$

184: of skew shapes $\lm$ and $\lmt$ respectively,

185: we give a realization of $\Vlm$

186: as a subspace in the tensor product $\Wnn$ of $n$ copies of defining

187: representation $\CC^N$ of $GL_N\ts$, and of $\nt$ copies

188: of the contragredient representation $(\CC^N)^\ast$.

189: This subspace is determined as the image of a certain linear operator

190: $\FOm$ on $\Wnn\,$. We introduce this operator

191: by an explicit multiplicative formula.

192: 

193: When $M=0$ and $\Vlm=V_{\la\lat}$ is an irreducible representation of

194: $GL_N$, we recover the known realization of $V_{\la\lat}$

195: as a certain subspace in the space of all traceless tensors in $\Wnn\,$.

196: Then the operator $\FOm$ may be regarded as the rational analogue

197: of the classical Young symmetrizer, corresponding to the

198: standard tableau $\Om$ of shape $\la\ts$.

199: That symmetrizer is a certain linear operator on

200: the tensor product of $n$ copies of $\CC^N$,

201: with the image equivalent to the irreducible

202: polynomial representation of $GL_N$, 

203: corresponding to the partition $\la\,$. Even when

204: $M=0$, our formula for the operator $\FOm$ is new.

205: 

206: Our results are applications of the representation theory

207: of the Yangian $\YN$ of the Lie algebra $\glN\ts$.

208: In particular, $\FOm$ is an intertwining operator between

209: certain representations of the algebra $\YN$ on $\Wnn\,$.

210: We also introduce the notion of a rational representation

211: of the Yangian $\YN\ts$. As a representation of $\YN\ts$,

212: the image of $\FOm$ is rational and irreducible.

213: \end{abstract}

214: 

215: %======================================================================

216: 

217: \section{\hskip-3.5pt. Main results}

218: 

219: %------------------------------------------------------------------------------

220: 

221: \textbf{1.1.}

222: For any positive integer $N$, consider the general linear group $GL_N$

223: over the complex field $\CC\ts$. Consider also the corresponding

224: Lie algebra $\glN\ts$. Let us choose the Borel subalgebra in $\glN$ 

225: consisting of the upper triangular matrices, and fix the basis of the 

226: diagonal matrix units $E_{11}\lcd E_{NN}$ in the corresponding Cartan 

227: subalgebra of $\glN\ts$.

228: Now take any irreducible {\it rational} representation

229: of the group $GL_N\ts$. This representation is finite-dimensional,

230: and its matrix elements in any basis are rational functions of the

231: standard coordinates on $GL_N\ts$. Moreover, denominators of

232: these rational functions are powers of the determinant.

233: This representation can also be regarded

234: as a $\glN$-module.

235: The highest weight of this $\glN$-module is a non-increasing

236: sequence of $N$ integers. Let us write this sequence of integers as

237: $$

238: (\ts\nu_{\ts1}\lcd\nu_{p}

239: \com0\lcd0\com

240: -\nut_{q}\lcd-\nut_{\ts1}\ts)

241: $$

242: where $\nu_{\ts1}\ge\ldots\ge\nu_{p}>0$ and

243: $\nut_{\ts1}\ge\ldots\ge\nut_{q}>0\ts$.

244: It is convenient to put $\nu_{\ts i}=0$ for every index $i>p\ts$,

245: and $\nut_{\ts i}=0$ for every $i>q\ts$.

246: Thus our irreducible representation of $GL_N$ is

247: indexed by the pair of partitions

248: $$

249: \nu=(\ts\nu_{\ts1},\nu_{\ts2},\,\ldots\,)

250: \quad\text{and}\quad

251: \nut=(\ts\nut_{\ts1},\nut_{\ts2},\,\ldots\,)

252: $$

253: such that $\nup_{\ts1}+\widetilde\nu^{\,\prime}_{\ts1}\le N$.

254: As usual, $\nup=(\nup_{\ts1},\nup_{\ts2},\,\ldots\,)$

255: is the partition conjugate to $\nu\ts$.

256: In particular, here $\nup_{\ts1}=p$ is the number of non-zero parts of $\nu$.

257: 

258: Let us denote by $V_{\nu\ts\nut}$

259: irreducible representation of $GL_N$

260: corresponding to the partitions $\nu$ and $\nut$. Put 

261: $$

262: n=\nu_{\ts1}+\nu_{\ts2}+\ldots

263: \quad\text{and}\quad\,

264: \nt=\nut_{\ts1}+\nut_{\ts2}+\ldots\,

265: $$

266: so that $\nu$ and $\nut$ are partitions of the numbers $n$ and $\nt$

267: respectively.

268: The irreducible representation $V_{\nu\ts\nut}$ occurs in the tensor product

269: \begin{equation}\label{mix}

270: \Wnn=(\CC^N)^\ast\ot\ldots\ot(\CC^N)^\ast\ot\CC^N\ot\ldots\ot\CC^N

271: \end{equation}

272: of $n$ copies of the defining representation $\CC^N$ of the group $GL_N$,

273: and of $\nt$ copies of the contragredient representation $(\CC^N)^\ast\ts$.

274: The vector space at the right hand side of (\ref{mix}) is called a

275: space of {\it mixed tensors\/}. %of degree $N$.

276: If $\nut=(0\com0\ts,\ts\ldots\ts)\ts$ and $\nt=0\ts$, then

277: $V_{\nu\ts\nut}\ts$ is a {\it polynomial\/} representation of $GL_N\ts$;

278: it is then denoted by $V_\nu\ts$. 

279: If $\nu=(0\com0\ts,\ts\ldots\ts)\ts$ and $n=0\ts$, then

280: the representation $V_{\nu\ts\nut}\ts$ of $GL_N$ is contragredient to

281: $V_{\ts\nut}\,$; it is then denoted by $V_{\nut}^{\ts\ast}\ts$ accordingly. 

282: 

283: Consider the natural pairing  

284: $\langle\ ,\,\rangle$ between the vector spaces

285: $(\CC^N)^\ast$ and $\CC^N$. Take any two indices

286: $k\in\{1\lcd n\}$ and $l\in\{1\lcd\nt\ts\}\ts$.

287: By applying the pairing $\langle\ ,\,\rangle$ to a tensor $w\in\Wnn$

288: in the $l\ts$th factor $(\CC^N)^\ast$ and the $k\ts$th factor $\CC^N$,

289: we obtain a certain tensor $\widehat{w}\in W_{n-1,\nt-1}\ts$.

290: The tensor $w$ is called \textit{traceless}, 

291: if $\widehat{w}=0$ for all possible $k$ and $l\ts$.

292: Denote by $\Wnno$ the subspace in $\Wnn$ consisting of all

293: traceless tensors, this subspace is $GL_N\ts$-invariant.

294: For the above $n$ and $\nt$, the irreducible components of $\Wnn$

295: equivalent to $V_{\nu\ts\nut}\ts$ occur only in the subspace $\Wnno\ts$.

296: Moreover, any such component of $\Wnn$ has the~form 

297: \vglue-16pt

298: $$

299: (\ts V_{\nut}^{\ts\ast}\otimes V_\nu\ts)\,\cap\,\Wnno

300: $$

301: for certain embeddings of the irreducible representations $V_\nu$ and 

302: $V_{\nut}^{\ts\ast}$ to the tensor products of $n$ copies 

303: of the $\CC^N$ and of $\nt$ copies of $(\CC^N)^\ast$ respectively.

304: 

305: All these facts about irreducible rational

306: representations of the group $GL_N$

307: are well known, see for instance \cite{K}.

308: In the present article, we consider a particular decomposition 

309: of the subspace of traceless tensors $\Wnno\subset\Wnn$ into

310: a direct sum of irreducible components.

311: In the particular case $\nt=0\ts$, this is the decomposition 

312: of the tensor product of $n$ copies of $\CC^N$ by means of

313: Young symmetrizers as in \cite[Section IV.4]{W}\ts;

314: see also \cite[Subsection 2.1]{N2}. For arbitrary $\nt\ts$, we

315: give an explicit multiplicative formula for the projector in $\Wnn$

316: onto each irreducible component of $\Wnno\ts$.

317: These projectors may be regarded as generalizations of the Young

318: symmetrizers from polynomial to the rational representations of $GL_N\ts$.

319: These generalizations seem to be new. %cf.\ \cite{KW}.

320: 

321: Our multiplicative formulas for the generalized Young symmetrizers in

322: $\Wnn$ can be extended even further. Take any two pairs of partitions

323: $$

324: \la=(\ts\la_1,\la_2\ts,\,\ldots\,)

325: \quad\text{and}\quad

326: \lat=(\ts\lat_1,\lat_2\ts,\,\ldots\,)\,,

327: $$

328: $$

329: \mu=(\ts\mu_1,\mu_2\ts,\,\ldots\,)

330: \quad\text{and}\quad

331: \mut=(\ts\mut_1,\mut_2\ts,\,\ldots\,)\,.

332: $$

333: Then choose any integer $M$ such that 

334: \begin{equation}\label{ass}

335: \lap_{\ts1}+\widetilde\la^{\,\prime}_{\ts1}\le N+M

336: \quad\text{and}\quad

337: \mup_{\ts1}+\widetilde\mu^{\,\prime}_{\ts1}\le M\,.

338: \end{equation}

339: Consider the irreducible representations $V_{\la\lat}\ts$ 

340: and $V_{\mu\ts\mut}\ts$ of the groups $GL_{N+M}$ and $GL_M$

341: respectively. The decomposition $\ts\CC^{\ts N+M}=\CC^N\!\op\CC^M$ provides

342: an embedding of the direct product $GL_N\times GL_M$ to $GL_{N+M}$.

343: The vector space

344: \begin{equation}\label{1.0}

345: \Vlm\ts=\ts{\rm Hom}_{\,GL_M}(\,V_{\mu\ts\mut}\ts\com V_{\la\lat}\,)

346: \end{equation}

347: comes with a natural action of the group $GL_N$\ts.

348: This action of $GL_N$ may be reducible.

349: In the special case when $M=0$ and $\mu=\mut=(0\com0\ts,\ts\ldots\ts)\ts$,

350: we have $\Vlm=V_{\la\lat}\,$; we assume that $GL_M=\{1\}$ when $M=0$. 

351: 

352: The vector space $\Vlm$ is non-zero, if and only if

353: \begin{equation}\label{con}

354: \la_i\ge\mu_i\ ,\,

355: \lap_i-\mu^{\,\prime}_i\le N

356: \quad\text{and}\quad

357: \lat_i\ge\mut_i\ ,\,

358: \lat^{\,\prime}_i-\mut^{\ts\,\prime}_i\le N

359: \end{equation}

360: for every index $i\,$; see \cite[Section I.5]{M}\ts.

361: In this article, we give distinguished embeddings of $\Vlm$

362: into the space of mixed tensors $\Wnn$ where

363: $$

364: n=\la_1-\mu_1+\la_2-\mu_2+\ldots

365: \quad\text{and}\quad

366: \nt=\lat_1-\mut_1+\lat_2-\mut_2+\ldots\,.

367: $$

368: These embeddings will be compatible with the action of the group $GL_N\ts$.

369: Under any of these embeddings, the vector space $\Vlm$ will be realized

370: as the image of a certain operator on $\Wnn$ commuting with the action

371: of $GL_N\ts$. We will give

372: a multiplicative formula for each of these operators.

373: Setting $\mu=\mut=(0\com0\ts,\ts\ldots\ts)$ and $M=0$

374: in these multiplicative formulas, we will obtain projectors

375: onto $GL_N\ts$-irreducible components of $\Wnno\ts$.

376: However, for arbitrary $\mu$ and $\mut$ the images of

377: $\Vlm$ under our embeddings may be not contained in the subspace

378: of traceless tensors $\Wnno\subset\Wnn\ts$.

379: 

380: When referring to theorems, propositions, lemmas and corollaries,

381: we will indicate the subsections where they respectively appear. 

382: There will be no more than one of each of these in any subsection,

383: so our referring system should cause no confusion. 

384: For example, the proposition and the lemma 

385: in the next subsection will be referred to as Proposition 1.2 and Lemma 1.2.

386: 

387: %------------------------------------------------------------------------------

388: 

389: \smallskip\medskip\noindent\textbf{1.2.}

390: First consider the case $\lat=\mut=(0\com0\ts,\ts\ldots\ts)$

391: and $\nt=0\ts$. In this case the vector spaces (\ref{mix}) and

392: (\ref{1.0}) will be denoted simply

393: by $W_n$ and $V_\la^{\ts\mu}$.

394: Suppose that $\la_i\ge\mu_i$ for every index $i\ts$.

395: Consider the {\it skew Young diagram}

396: $$

397: \lm=\{\,(i\com j)\in\ZZ^2\ |\ i\ge1,\ \la_i\ge j>\mu_i\,\}\,.

398: $$

399: If $\mu=(0\com0\ts,\ts\ldots\ts)\ts$, this is the conventional Young diagram

400: of the partition $\la\ts$.

401: We will employ the standard

402: graphical representation \cite[Section I.1]{M}

403: of Young diagrams on the plane $\RR^2$ with two matrix style coordinates.

404: Here the first coordinate increases from top to bottom, while the second

405: coordinate increases from left to right. The element $(i\com j)\in\lm$

406: is represented by the unit box with the bottom right corner

407: at the point $(i\com j)\in\RR^2$. 

408: 

409: The set $\lm$ consists of $n$ elements.

410: A \textit{standard tableau} of shape $\lm$ is a bijection

411: $\Om:\lm\to\{1\lcd n\}$ such that $\Om(i\com j)<\Om(i+1\com j)$ and

412: $\Om(i\com j)<\Om(i\com j+1)$ for all possible $i$ and $j\ts$.

413: For every $k=1\lcd n$ put $c_k(\Om)=j-i$ if $k=\Om(i\com j)\ts$.

414: The difference $c_k(\Om)$ is the \textit{content} of the box

415: occupied by the number $k$ in the tableau $\Om\ts$.

416: Graphically, the tableau $\Om$ is represented by placing the

417: numbers $\Om(i\com j)$ into the corresponding boxes of $\lm$

418: on the plane $\RR^2$. 

419: By filling the boxes with the numbers

420: $1\lcd n$ by columns from the left to the right,

421: downwards in every column, we get

422: the \textit{column tableau} $\Om$ of shape $\lm\ts$.

423: This tableau is standard. Below on the left

424: we represent the column tableau $\Om$ for

425: $\la=(5,\ns3,\ns3,\ns3,\ns3,\ns0,\ns0,\ts\ldots)$ and

426: $\mu=(3,\ns3,\ns2,\ns0,\ns0,\ts\ldots)\ts$,

427: on the right we show the contents of the boxes of $\lm\ts$:

428: 

429: \bigskip

430: \vbox{

431: $$

432: \kern22.5pt\longrightarrow\,j\mw

433: $$

434: \vglue-25pt

435: $$

436: \kern80pt\vert

437: $$

438: \vglue-27pt

439: $$

440: \bigr\downarrow\mw\kern1pt

441: $$

442: \vglue-16pt

443: $$

444: i\mw

445: $$

446: \vglue-44pt

447: $$

448: \phantom{\bx}

449: \phantom{\bx}

450: \phantom{\bx}

451: {\bx}

452: {\bx}

453: \kern80pt

454: \phantom{\bx}

455: \phantom{\bx}

456: \phantom{\bx}

457: {\bx}

458: {\bx}

459: $$

460: \vglue-16.6pt

461: $$

462: \phantom{\bx}

463: \phantom{\bx}

464: \phantom{\bx}

465: \phantom{\bx}

466: \phantom{\bx}

467: \kern80pt

468: \phantom{\bx}

469: \phantom{\bx}

470: \phantom{\bx}

471: \phantom{\bx}

472: \phantom{\bx}

473: $$

474: \vglue-16.6pt

475: $$

476: \phantom{\bx}

477: \phantom{\bx}

478: {\bx}

479: \phantom{\bx}

480: \phantom{\bx}

481: \kern80pt

482: \phantom{\bx}

483: \phantom{\bx}

484: {\bx}

485: \phantom{\bx}

486: \phantom{\bx}

487: $$

488: \vglue-16.5pt

489: $$

490: {\bx}

491: {\bx}

492: {\bx}

493: \phantom{\bx}

494: \phantom{\bx}

495: \kern80pt

496: {\bx}

497: {\bx}

498: {\bx}

499: \phantom{\bx}

500: \phantom{\bx}

501: $$

502: \vglue-16.5pt

503: $$

504: {\bx}

505: {\bx}

506: {\bx}

507: \phantom{\bx}

508: \phantom{\bx}

509: \kern80pt

510: {\bx}

511: {\bx}

512: {\bx}

513: \phantom{\bx}

514: \phantom{\bx}

515: $$

516: \vglue-82.5pt

517: $$

518: \kern43pt8\kern9pt9\kern90pt\kern42pt3\kern9pt4

519: $$

520: \vglue-3.5pt

521: $$

522: 5\kern146pt0

523: $$

524: \vglue-16.5pt

525: $$

526: 1\kern9pt3\kern9pt6\kern116pt\mi3\kern6pt\mi2\kern6pt\mi1\kern26pt

527: $$

528: \vglue-16.5pt

529: $$

530: 2\kern9pt4\kern9pt7\kern116pt\mi4\kern6pt\mi3\kern6pt\mi2\kern26pt

531: $$

532: }

533: 

534: \bigskip\medskip\noindent

535: In this example $n=9$, and the sequence of contents

536: $c_1(\Om)\lcd c_9(\Om)$ is

537: $$

538: (\ts\mi3\com\mi4\com\mi2\com\mi3\com0\com\mi1\com\mi2\com3\com4\ts)\ts.

539: $$

540: 

541: Now introduce complex

542: variables $t_1(\Om)\lcd t_n(\Om)$ with the constraints

543: \begin{equation}\label{1.0002}

544: t_k(\Om)=t_l(\Om)

545: \text{\ \ if $k$ and $l$ occur in the same column of $\Om$\ts.}

546: \end{equation}

547: So the number of independent variables among $t_1(\Om)\lcd t_n(\Om)$ 

548: equals the number of non-empty columns of the diagram $\lm\ts$.

549: Order lexicographically

550: the set of all pairs $(k\com l)$ with $1\le k<l\le n$.

551: Take the ordered product over this set,

552: 

553: \vskip-20pt

554: \begin{equation}\label{1.3}

555: \prod_{1\le k<l\le n}^{\longrightarrow}\ 

556: \left(1-\frac{P_{kl}}{\ts c_k(\Om)-c_{\ts l}(\Om)+t_k(\Om)-t_l(\Om)}\ts\right)

557: \end{equation}

558: \noindent

559: where $P_{kl}$ denotes the operator on the vector space $W_n$

560: exchanging the $k\ts$th and $l\ts$th tensor factors. Let us consider 

561: (\ref{1.3}) as a rational function of the constrained variables 

562: $t_1(\Om)\lcd t_n(\Om)$. The next result goes back to \cite{C1}.

563: 

564: \begin{Lemma}

565: The product {\rm(\ref{1.3})} is regular at\/

566: $t_1(\Om)=\ldots=t_n(\Om)\ts$. 

567: \end{Lemma}

568: 

569: \noindent

570: Note that the function (\ref{1.3}) depends only on the differences

571: $t_k(\Om)-t_l(\Om)\ts$. The value of {\rm(\ref{1.3})} at

572: $t_1(\Om)=\ldots=t_n(\Om)\ts$ will be denoted by $F_\Om\,$. 

573: For the proof of Lemma 1.2 see \cite[Section 2]{NT2}.

574: This proof provides an explicit multiplicative formula for

575: the operator $F_\Om$ on the vector space $W_n\ts$.

576: 

577: Let us denote by $V_\Om$ the image of the operator $F_\Om$ on $W_n\ts$.

578: The subspace $V_\Om\subset W_n$ is stable under the action of

579: $GL_N\ts$, see the last paragraph of this subsection. 

580: %The following proposition also belongs to Cherednik \cite{C2}\ts; 

581: Proof of the following result was given in \cite[Subsection 4.6]{N2}.

582: 

583: \begin{Proposition}

584: The representations $V_\la^{\ts\mu}$ and $V_\Om$ of\/ $GL_N$ are equivalent. 

585: \end{Proposition}

586: 

587: \noindent

588: Note that the operator $F_\Om$ on the vector space $W_n$ does not depend

589: on $M$. It is well known that the dimension of the vector

590: space $V_\la^{\ts\mu}$ is the same for all integers $M$ such that

591: $\lap_1\le N+M$ and $\mup_1\le M$; see \cite[Section I.5]{M}.

592: 

593: The symmetric group $S_n$ acts on the tensor product $W_n$ of

594: $n$ copies of $\CC^N$ by permutations of the tensor factors. 

595: This action extends to the group ring $\CC S_n\ts$. 

596: The image of $\CC\ts S_n$ in the operator algebra 

597: $\End(W_n)$ coincides with

598: the commutant of the image of the group $GL_N\ts$;

599: see \cite[Section IV.4]{W}.

600: The operator $P_{kl}\in\End(W_n)$ in

601: (\ref{1.3}) corresponds to the transposition $(k\ts l)\in S_n\ts$.

602: The operator $F_\Om\in\End(W_n)$ corresponds to a certain element

603: $f_{\ts\Om}\in\CC S_n$ which does not depend on $N$. 

604: If $\mu=(0\com0\ts,\ts\ldots\ts)\ts$, then $f_{\ts\Om}$ is a diagonal

605: matrix element of the irreducible representation of the group $S_n$  

606: labeled by the partition $\la\,$; see Subsection 2.2 of the present article.

607: For an alternative definition of the element $f_{\ts\Om}\in\CC S_n$

608: in the case $\mu\neq(0\com0\ts,\ts\ldots\ts)\ts$, see Subsection 2.3;

609: in particular, see the equality (\ref{2.9}).

610: 

611: %------------------------------------------------------------------------------

612: 

613: \smallskip\medskip\noindent\textbf{1.3.}

614: Let us now formulate our results for the general partitions

615: $\lat$ and $\mut\ts$.

616: We will assume that $\la_i\ge\mu_i$ and $\lat_i\ge\mut_i$

617: for every index $k\ts$, otherwise we would have $\Vlm=\{0\}$.

618: Our embedding of the vector space $\Vlm$ into the space of mixed tensors

619: (\ref{mix}) will depend on the choice of two standard tableaux $\Om$ and $\Omt$

620: of the skew shapes $\lm$ and $\lmt$ respectively.

621: 

622: Choose any basis $v_1\lcd v_N$ in the vector space $\CC^N$.

623: Let $v_1^\ast\lcd v_N^{\ts\ast}$ be the dual basis in $(\CC^N)^\ast$ so that

624: $\langle\ts v_a^\ast\com v_b\ts\rangle=\de_{ab}$ for $a\com b=1\lcd N$.

625: The vector

626: 

627: \vskip-20pt

628: \begin{equation}\label{1.44444444}

629: w_0\,=\,\sum_{a=1}^N\,v_a^\ast\ns\ot v_a\,\in\,(\CC^N)^\ast\ns\ot\ts\CC^N

630: \end{equation}

631: 

632: \newpage\noindent

633: does not depend on the choice of the basis $v_1\lcd v_N$ and is invariant

634: under the action of $GL_N$ on $(\CC^N)^\ast\ns\ot\,\CC^N$.

635: Then consider the linear operator

636: \begin{equation}\label{1.45}

637: u\ot v\,\mapsto\,\langle\ts u\com v\ts\rangle\cdot w_0

638: \end{equation}

639: in the vector space

640: $(\CC^N)^\ast\ns\ot\,\CC^N$, it commutes with the action of $GL_N\ts$.

641: 

642: The tableaux $\Om$ and $\Omt$ define the sequences of contents

643: $c_1(\Om)\lcd c_n(\Om)\ns$ and $c_1(\Omt)\lcd c_{\ts\nt}(\Omt)\ts$.

644: For any indices $k\in\{1\lcd n\}$ and $l\in\{1\lcd\nt\ts\}\ts$,

645: denote by $Q_{\ts l,\nt+k}$ the linear operator on $\Wnn\ts$,

646: acting as (\ref{1.45}) in the $l\ts$th and $(\nt+k)\ts$th

647: tensor factors of (\ref{mix}), and acting

648: as the identity in the remaining $n+\nt-2$ tensor factors.

649: %This notation will agree with the notation of Section 4.

650: Consider the ordered product

651: \begin{equation}\label{1.333}

652: \GOm\ =\ 

653: \prod_{1\le k\le n}^{\longrightarrow}\ \biggl(\ 

654: \prod_{1\le l\le\nt}^{\longrightarrow}\ 

655: \biggl(1-\frac{Q_{\ts\nt-l+1\ts,\ts\nt+k}}

656: {\ts c_k(\Om)+{c_{\ts l}(\Omt)}^{\phantom{\prime}}\ns\!+N+M}

657: \,\biggr)\!\biggr)\,.

658: \end{equation}

659: Note that

660: $$

661: c_k(\Om)+{c_{\ts l}(\Omt)}\ge2-\lap_1-\lat_1^{\ts\prime}\ge2-N-M\ts,

662: $$

663: %by our choice of the integer $M$. 

664: so every denominator in (\ref{1.333}) is non-zero.

665: Also consider the ordered product

666: $$

667: \GOmp\ =\ 

668: \prod_{1\le k\le n}^{\longleftarrow}\ \biggl(\ 

669: \prod_{1\le l\le\nt}^{\longleftarrow}\ 

670: \biggl(1-\frac{Q_{\ts\nt-l+1\ts,\ts\nt+k}}

671: {\ts c_k(\Om)+{c_{\ts l}(\Omt)}^{\phantom{\prime}}\ns\!+N+M}

672: \,\biggr)\!\biggr)\,,

673: \hskip21pt

674: $$

675: where the factors corresponding to the indices $k=1\lcd n$ and $l=1\lcd\nt$

676: are arranged from right to left, as indicated by the reversed arrows.

677: 

678: The tableau $\Omt$ defines

679: the linear operator $F_\Omt$

680: in the tensor product $W_{\ts\nt}$ of $\nt$ copies of $\CC^N$,

681: see Subsection~1.2.

682: Identify the dual vector space $W_{\ts\nt}^{\ts\ast}$ with

683: the tensor product of $\nt$ copies of $(\CC^N)^\ast$, so that

684: \begin{equation}\label{den}

685: \Wnn=W_{\ts\nt}^{\ts\ast}\otimes W_n\,.

686: \end{equation}

687: Under the permutational action of the group $S_{\ts\nt}\ts$

688: in $W_{\ts\nt}\ts$, the operator $F_\Omt$ on $W_{\ts\nt}$ corresponds to 

689: a certain element $f_\Omt$ of the group ring $\CC S_{\ts\nt}\ts$. 

690: The group $S_{\ts\nt}$ also acts by permutations of the tensor factors

691: in $W_{\ts\nt}^{\ts\ast}\ts$. The operator on $W_{\ts\nt}^{\ts\ast}$

692: corresponding to the element $f_\Omt\in\CC S_{\ts\nt}\ts$ will also be

693: denoted by $F_\Omt\ts$. However, we will always distinguish the vector

694: spaces $W_{\ts\nt}$ and $W_{\ts\nt}^{\ts\ast}\ts$. The linear operators

695: on $W_{\ts\nt}$ and $W_{\ts\nt}^{\ts\ast}$ reversing the order of the

696: $\nt$ tensor factors $\CC^N$ and $(\CC^N)^\ast$ respectively, will be

697: denoted by the same symbol $P_{\ts\nt}\ts$.

698:  

699: The tableau $\Om$ defines the linear operator $F_\Om$

700: in the tensor product $W_n\ts$, as in Subsection 1.2.

701: Introduce the linear operator on $\Wnn$

702: \begin{equation}\label{maindef}

703: \FOm\ts=\,\GOm\ts\cdot\ts((\ts P_{\ts\nt}\ts F_\Omt\,P_{\ts\nt})\ot F_\Om)\,.

704: \end{equation}

705: 

706: \begin{Lemma}

707: We have the equality of operators on  the vector space $\Wnn$

708: $$

709: \FOm\ts=\ts((\ts P_{\ts\nt}\ts F_\Omt\,P_{\ts\nt})\ot F_\Om)\ts\cdot\ts\GOmp\,.

710: $$

711: \end{Lemma}

712: 

713: We will prove Lemma 1.3 in Subsection 4.2.

714: Denote by $\VOm$

715: the image of the operator $\FOm$ on $\Wnn\ts$.

716: Note that unlike the subspace $V_\Om\ts\subset W_n\ts$, 

717: the subspace $\VOm\subset\Wnn$ may depend on the choice of $M$.

718: The subspace $\VOm\subset\Wnn$ is stable under the action of 

719: $GL_N\ts$. Here is a generalization of Proposition 1.2 

720: from polynomial to the rational representations of $GL_N$.

721: 

722: \begin{Proposition}

723: The representations $\Vlm$ and $\VOm$ of\/ $GL_N$ are equivalent. 

724: \end{Proposition}

725: 

726: The proof of Proposition 1.3 is given in the end if Subsection 5.4. 

727: Denote by $V^{\ts\ast}_\Omt$ 

728: the image of the operator $F_\Omt$ in

729: $W_{\ts\nt}^{\ts\ast}\ts$.

730: Due to Lemma 1.3, the image $\VOm$ of the operator $\FOm$

731: is contained in the subspace

732: $$

733: (\ts P_{\ts\nt}\cdot V^{\ts\ast}_\Omt\ts)\ot V_\Om\,\subset\,

734: W_{\ts\nt}^{\ts\ast}\otimes W_n\,.

735: $$

736: If $M=0$, then $\mu=\mut=(0\com0\ts,\ts\ldots\ts)\ts$. Then

737: the vector spaces $V_\Om$ and $P_{\ts\nt}\cdot V^{\ts\ast}_\Omt$

738: are equivalent respectively to $V_\la$ and $V_{\ts\lat}^{\ts\ast}$ as

739: representations of the group $GL_N$. We have the equality

740: \begin{equation}\label{1.4444}

741: \VOm=

742: ((\ts P_{\ts\nt}\cdot V^{\ts\ast}_\Omt)\ot V_\Om)\ts\cap\ts\Wnno

743: \quad\text{if}\quad

744: M=0\,;

745: \end{equation}

746: see Subsection 3.4. If $M\neq0$,

747: the image $\VOm$ of the operator $\FOm$ may be not contained in

748: the subspace $\Wnno\subset\Wnn$ of traceless tensors.

749: Yet our proof of Proposition 1.3 is based on the equality (\ref{1.4444}).

750: 

751: If $M\neq0$, the operator $\FOm$

752: can also be defined via the equality (\ref{3.4444}),

753: cf.\ the equality (\ref{2.9}).

754: Both definitions of the operator $\FOm$ are motivated by the

755: representation theory of Yangians~\cite{MNO}, see the next subsection.

756: 

757: %------------------------------------------------------------------------------

758: 

759: \smallskip\medskip\noindent\textbf{1.4.}

760: The vector space $\Vlm$

761: is irreducible under the natural action of

762: the subalgebra of $GL_M$\ts-invariants in the enveloping algebra

763: $\UMN\ts$. Denote this subalgebra by $\AMN\ts$,

764: it coincides with the centralizer of the subalgebra $\UM\subset\UMN\ts$.

765: Our Theorem 1.5 below describes the action of the algebra $\AMN$ on $\Vlm$ 

766: explicitly, using the \textit{Yangian} $\YN$ of the Lie algebra $\glN\ts$.   

767: The Yangian $\YN$ is a deformation of the

768: enveloping algebra of the polynomial current Lie algebra $\glN[x]$

769: in the class of Hopf algebras. % \cite{D}.

770: 

771: The unital associative algebra $\YN$ has a 

772: family of generators $T_{ab}^{(i)}$ where $i=1,2,\ts\ldots\ts$ and

773: $a\com b=1\lcd N$. The defining relations for these generators

774: can be written by using the formal power series

775: \begin{equation}\label{1.31}

776: T_{ab}(x)=

777: \de_{ab}\cdot1+T_{ab}^{(1)}x^{-\ns1}+T_{ab}^{(2)}x^{-\ns2}+\,\ldots

778: \,\in\,\YN\,[[x^{-1}]]\,.

779: \end{equation}

780: Here $x$ is the formal parameter. Let $y$ be another formal parameter,  

781: then the defining relations in the associative algebra $\YN$

782: can be written as

783: \begin{equation}\label{1.32}

784: (x-y)\cdot[\ts T_{ab}(x)\ts,T_{cd}(y)\ts]\ts=\;

785: T_{cb}(x)\ts T_{ad}(y)-T_{cb}(y)\ts T_{ad}(x)\,,

786: \end{equation}

787: where $a\com b\com c\com d=1\lcd N\ts$.

788: The square brackets in (\ref{1.32}) denote the usual commutator.

789: If $N=1$, the algebra $\YN$ is commutative.

790: Using the series (\ref{1.31}),

791: the coproduct $\De:\YN\to\YN\ot\YN$ is defined by

792: \begin{equation}\label{1.33}

793: \De\bigl(T_{ab}(x)\bigr)\,=\,\sum_{c=1}^N\ T_{ac}(x)\ot T_{cb}(x)\,;

794: \end{equation}

795: the tensor product on the right hand side of the equality (\ref{1.33})

796: is taken over the subalgebra $\CC[[x^{-1}]]\subset\YN\,[[x^{-1}]]\ts$.

797: The counit homomorphism $\varepsilon:\YN\to\CC$ is determined by

798: the assignment $\,\varepsilon:\,T_{ab}(x)\ts\mapsto\ts\de_{ab}\cdot1$.

799: 

800: The antipode $\operatorname{S}$ on $\YN$ can be defined by using the element

801: \begin{equation}\label{1.71}

802: T(x)\,\ts=\sum_{a,b=1}^N\, E_{ab}\ot

803: T_{ab}(x)\in\End(\CC^N)\ot\YN\,[[x^{-1}]]\,,

804: \end{equation}

805: where the matrix units $E_{ab}$ are regarded as basis elements of the

806: algebra $\End(\CC^N)\ts$. The formal power series (\ref{1.71}) in

807: $x^{-1}$ is invertible, because its leading term is $1\ot1$.

808: The anti-automorphism $\operatorname{S}$ is defined by the assignment

809: \begin{equation}\label{antip}

810: \id\ot\operatorname{S}\ts:\ts T(x)\mapsto T(x)^{-1}\,.

811: \end{equation}

812: We will also use an automorphism $\om_N$ of the algebra $\YN$ defined by

813: the assignment

814: 

815: \vskip-20pt

816: \begin{equation}\label{1.51}

817: \id\ot\om_N\ts:\ts T(x)\mapsto T(-x)^{-1}\,.

818: \end{equation}

819: The automorphism $\om_N$ is clearly involutive.

820: For references and more details on the definition of the Yangian $\YN$

821: see \cite[Section 1]{MNO}\ts.

822: 

823: Take any formal power series $g(x)\in\CC[[x^{-1}]]$ with the leading

824: term $1$. The assignment

825: 

826: \vskip-20pt

827: \begin{equation}\label{1.61}

828: T_{ab}(x)\mapsto\,g(x)\cdot T_{ab}(x)

829: \end{equation}

830: defines an automorphism of the algebra $\YN\ts$,

831: see (\ref{1.31}) and (\ref{1.32}).

832: 

833: The Yangian $\YN$ contains the enveloping

834: algebra $\UN$ as a Hopf subalgebra. Regard 

835: the matrix units $E_{ab}\in\glN$ as generators of

836: the algebra $\UN\ts$. Due to the defining relations (\ref{1.32}),

837: an embedding of associative algebras $\UN\to\YN$ 

838: can be determined by the assignment

839: \begin{equation}\label{4.4}

840: E_{ab}\mapsto-\ts T_{ba}^{(1)}.

841: \end{equation}

842: This particular embedding will be used in Section 4, see Proposition 4.1.

843: This is a Hopf algebra embedding, because by (\ref{1.33})

844: we have the equality

845: \begin{equation}\label{Hopfemb}

846: \De\ts\bigl(\ts T_{ab}^{(1)}\bigr)\ts=\,

847: T_{ab}^{(1)}\ns\ot\ts1+1\ot\ts T_{ab}^{(1)}\,.

848: \end{equation}

849: The relations (\ref{1.32}) also show that the assignment

850: \begin{equation}\label{1.52}

851: \pi_N:\ts T_{ab}(x)\,\mapsto\,\de_{ab}\cdot1-E_{ba}\,x^{-1}

852: \end{equation}

853: defines a homomorphism $\pi_N:\YN\to\UN$

854: of associative algebras. By definition,

855: the homomorphism $\pi_N$ acts on the subalgebra

856: $\UN\subset\YN$ as the identity. However,

857: $\pi_N$ is not a Hopf algebra homomorphism.

858: 

859: The defining relations (\ref{1.32}) show that for any $z\in\CC\,$,

860: the assignment

861: \begin{equation}\label{tau}

862: \tau_z:\,T_{ab}(x)\ts\mapsto\,T_{ab}(x-z)

863: \quad\textrm{for all}\quad

864: a\com b=1\lcd N

865: \end{equation}

866: defines an automorphism $\tau_z$ of the algebra $\YN\ts$. Here the formal

867: power series $T_{ab}(x-z)$ in $(x-z)^{-1}$ should be re-expanded in $x^{-1}$.

868: Note that the automorphism $\tau_z$ acts on the subalgebra

869: $\UN\subset\YN$ as the identity.

870: 

871: By pulling the action of 

872: the enveloping algebra $\UN$ in the defining $\glN$-module $\CC^N\ts$

873: back through the homomorphism

874: $$

875: %\begin{equation}\label{eval}

876: \pi_N\circ\,\tau_z:\YN\to\UN\,,

877: %\end{equation}

878: $$

879: we obtain a module over the algebra $\YN\ts$, 

880: called an \textit{evaluation module\/}. 

881: To indicate the dependence on the

882: parameter $z\ts$, let us denote this $\YN\ts$-module by $V(z)\ts$.

883: As a $\glN$-module, this is still the defining module $\CC^N$;

884: here we use the embedding (\ref{4.4}). Finally, by

885: using the antipode $\operatorname{S}$ of $\YN\ts$, 

886: define the $\YN\ts$-module $V(z)^\ast$ dual to $V(z)\ts$. 

887: As a $\glN$-module, $V(z)^\ast$ is the contragredient module 

888: $(\CC^N)^\ast$. Here we again use the embedding (\ref{4.4}). 

889: 

890: The operator $\FOm$ on the space

891: $\Wnn$ admits the following interpretation in terms of the tensor products

892: of evaluation modules over the algebra $\YN\ts$. Here we will use

893: the comultiplication (\ref{1.33}) on $\YN\ts$. Let 

894: \begin{equation}\label{Pnn}

895: P_{\ts n+\nt}\ts:\ts

896: W_n\ot W_{\ts\nt}^{\ts\ast}\,\to\,

897: W_{\ts\nt}^{\ts\ast}\ot W_n

898: \end{equation}

899: be the linear

900: operator reversing the order of all the $\ts n+\nt\ts$ tensor factors. 

901: 

902: \begin{Proposition}

903: The operator\/ $\FOm\,P_{n+\nt}\ts$ is a\/ $\YN$-intertwiner

904: $$

905: V(c_n(\Om))\ot\ldots\ot V(c_1(\Om))\ot

906: V(-c_1(\Omt)\ns-\!M)^\ast\ot\ldots\ot V(-c_{\ts\nt}(\Omt)\ns-\!M)^\ast

907: $$

908: \vglue-10pt

909: $$

910: \Big\downarrow

911: $$

912: \vglue-14pt

913: $$

914: V(-c_{\ts\nt}(\Omt)\ns-\!M)^\ast\ot\ldots\ot V(-c_1(\Omt)\ns-\!M)^\ast\ot

915: V(c_1(\Om))\ot\ldots\ot V(c_n(\Om))\,.

916: $$

917: \end{Proposition}

918: 

919: \noindent

920: Thus the image $\VOm$ of the operator $\FOm$ on $\Wnn$

921: is a submodule of the target $\YN\ts$-module in Proposition 1.4.

922: If $\,\nt=0\ts$, this interpretation of $\FOm=F_\Om$ and $\VOm=V_\Om$

923: is due to Cherednik~\cite{C2}. In full generality,

924: the proof of Proposition 1.4 is given is Subsection 4.2

925: of the present article.

926: 

927: If $M=0$, then

928: $\mu=\mut=(0\com0\ts,\ts\ldots\ts)\ts$. 

929: The image $\VOm$ of the operator $\FOm$ is then equivalent to

930: $V_{\la\lat}$ as a representation of $GL_N$.

931: Proposition~1.4 then turns $V_{\la\lat}$ into $\YN\ts$-module. 

932: Note that the resulting $\YN\ts$-module can also be obtained from the

933: $\glN\ts$-module $V_{\la\lat}$ by pulling back through the

934: homomorphism $\pi_N\ts$, and then through a certain automorphism of $\YN$

935: of the form (\ref{1.61})\ts;

936: %of $\YN$ for a certain function $g(x)\in\CC(x)\ts$; 

937: this is a special case of Theorem~1.5 below.

938: 

939: %------------------------------------------------------------------------------

940: 

941: \smallskip\medskip\noindent\textbf{1.5.}

942: Olshanski \cite{O1} defined a homomorphism from the algebra $\YN$ 

943: to the subalgebra $\AMN$ of $GL_M$-invariants in $\UMN$,

944: for each non-negative integer $M$. Along with the centre

945: of the algebra $\UMN\ts$, the image of this homomorphism generates

946: the algebra $\AMN\ts$. We will use the following version of this homomorphism,

947: it will be denoted by $\pi_{NM}\ts$.

948: 

949: Let the indices $a\com b$ range over the set $\{1\lcd N+M\}\ts$.

950: Consider the basis of the matrix units $E_{ab}$

951: in the Lie algebra $\glMN\ts$. We assume that the subalgebras $\glN$ and

952: $\glM$ in $\glMN$ are spanned by elements $E_{ab}$ where

953: $$

954: %\begin{equation}\label{ind}

955: 1\le a\com b\le N

956: \quad\text{and}\quad

957: N+1\le a\com b\le N+M

958: %\end{equation}

959: $$

960: respectively. The subalgebra in the Yangian $\YMN$

961: generated by $T_{ab}^{(i)}$ where $1\le a\com b\le N\,$,

962: by definition coincides with the Yangian $\YN$.

963: Let us denote by $\ph_M$ this natural embedding $\YN\to\YMN$. 

964: Consider also the involutive automorphism $\om_{N+M}$

965: of the algebra $\YMN\ts$, see (\ref{1.51}).

966: The image of the homomorphism

967: $$

968: \pi_{N+M}\circ\ts\om_{N+M}\circ\ts\ph_M:\ts\YN\to\UMN

969: $$

970: belongs to the subalgebra $\AMN\subset\UMN\ts$.

971: Moreover, this image along with the centre

972: of the algebra $\UMN\ts$, generates the subalgebra $\AMN\ts$.

973: For the detailed proofs of these two assertions, see \cite[Section 2]{MO}.

974: In the present article, we use the homomorphism $\YN\to\UMN$

975: \begin{equation}\label{1.69}

976: \pi_{NM}=\,\pi_{N+M}\circ\ts\om_{N+M}\circ\ts\ph_M\circ\ts\om_N\ts.

977: \end{equation}

978: Note that when $M=0$, the homomorphism (\ref{1.69}) coincides with $\pi_N\ts$. 

979: Put

980: \begin{equation}\label{1.62}

981: g_\mu(x)\ =\ \prod_{i=1}^\infty\ 

982: \frac

983: {(\ts x-\mu_i+i\ts)(\ts x+i-1\ts)}

984: {(\ts x-\mu_i+i-1\ts)(\ts x+i\ts)}

985: \,\,.

986: \end{equation}

987: Here in the infinite product over $i\ts$, only finitely many

988: factors differ from $1$. Hence $g_\mu(x)$ is a rational function 

989: of $x$. We have $g_\mu(\infty)=1$, therefore $g_\mu(x)$

990: expands as a power series in $x^{-1}$ with leading term $1$. 

991: %%%See also (\ref{fact}).

992: 

993: Let us keep to the assumptions \ns(\ref{ass}) on the integer $M$.

994: The~vector~space $\Vlm$ comes

995: with a natural action of the algebra $\AMN\ts$. Let us regard $\Vlm$ as a 

996: $\YN\ts$-module, by using the composition of the homomorphism

997: $\pi_{NM}:\YN\to\AMN$ with the automorphism of $\YN$ defined by

998: (\ref{1.61}) where 

999: $$%\begin{equation}\label{1.62}

1000: g(x)\,=\,g_\mu(x)\,\ts

1001: g_{\ts\lat}(\ts-\ts x\ns-\!M)

1002: \ts/\ts

1003: g_{\ts\mut\ts}(\ts-\ts x\ns-\!M)\,.

1004: $$%\end{equation}

1005: By Proposition~1.4, the image $\VOm$ of the operator

1006: $\FOm$ can also be regarded as a $\YN\ts$-module.

1007: Our main result is the following theorem.

1008: 

1009: \begin{Theorem}

1010: The\/ $\YN$-modules\/ $\Vlm$ and\/ $\VOm$ are equivalent.

1011: \end{Theorem}

1012: 

1013: \noindent

1014: When $\nt=0\ts$, this theorem goes back to \cite[Theorem 2.6]{C2}.

1015: However, most of the results in \cite{C2} have been given without proofs.

1016: We prove Theorem~1.5 in Subsections 5.2 to 5.4 of the present article.

1017: The algebra~$\AMN$ acts on the vector space $\Vlm$ irreducibly

1018: \cite[Theorem 9.1.12]{D}.

1019: The central elements of $\UMN$ act on $\Vlm$

1020: as scalar operators. So Theorem 1.5 has a corollary.

1021: 

1022: \begin{Corollary}

1023: The $\YN\ts$-module $\VOm$ is irreducible.

1024: \end{Corollary}

1025: 

1026: The embedding $\UN\to\YN$ defined by (\ref{4.4}) 

1027: provides an action of $\UN$ on the $\YN\ts$-module $\Vlm\ts$.

1028: The vector space $\Vlm$ also comes with a natural action

1029: of $\UN\subset\AMN\ts$. This natural action of $\UN$ in

1030: $\Vlm$ coincides with its action as a subalgebra in $\YN\ts$,

1031: see Subsection~4.3.

1032: 

1033: The subspace $\VOm\subset\Wnn$

1034: is stable under the action of the group $GL_N\ts$.

1035: Hence the enveloping algebra $\UN$ acts naturally on the vector space

1036: $\VOm\ts$. This natural action of $\UN$ on $\VOm$ coincides with its action

1037: as a subalgebra in $\YN\ts$, see again Subsection 4.3.

1038: 

1039: The natural action of $\UN$ on 

1040: the vector space $\Vlm$ may be reducible.

1041: Using Theorem 1.5 and its Corollary 1.5,

1042: we can identify the vector space $\Vlm$

1043: with the subspace $\VOm\subset\Wnn$ uniquely, up to multiplication

1044: in $\VOm$ by a non-zero complex number.

1045: Theorem 1.5 can be regarded as sharpening of Proposition 1.3.

1046: Moreover, we will obtain Proposition 1.3 in the course of

1047: the proof of Theorem 1.5. In our proof, we will use Proposition 3.6.

1048: 

1049: %------------------------------------------------------------------------------

1050: 

1051: \smallskip\medskip\noindent\textbf{1.6.}

1052: In priniciple, the analysis of any rational representation 

1053: of the group $GL_N$ may be reduced to a polynomial representation,

1054: cf.\ \cite[Section IV.5]{W}. In particular,

1055: the irreducible rational

1056: representation $V_{\nu\ts\nut}$ of $GL_N$ as defined in Subsection~1.1,

1057: is equivalent to the tensor product of the one-dimensional

1058: representation where

1059: 

1060: \vskip-16pt

1061: $$

1062: X\ts\mapsto\,(\ts\det X)^{-\nut_1}

1063: $$

1064: for any $\ts X\in GL_N\ts$, and of the polynomial representation $V_\zeta$ 

1065: of $GL_N$ where

1066: $$

1067: \zeta_{\ts i}\ts=\,

1068: \left\{

1069: \begin{array}{ll}

1070: \nu_{i}-\nut_{\ts N-i+1}+\nut_{\ts1}

1071: &\ \ \textrm{if}\ \ \ i\le N,\\[2pt]

1072: 0

1073: &\ \ \textrm{if}\ \ \ i>N.

1074: \end{array}

1075: \right.

1076: $$

1077: However, realization of the representation $V_{\nu\ts\nut}$

1078: in the space of mixed tensors (\ref{mix}) is preferable,

1079: at least from the combinatorial point of view \cite{KW,S}\ts,

1080: and when $N\to\infty$ \cite{O2,VK}.

1081: This makes our explicit formula

1082: (\ref{maindef}) for the operator $\FOm$ on (\ref{mix})

1083: useful, already %in the special case 

1084: when $M=0$ and $\mu=\mut=(0\com0\ts,\ts\ldots\ts)\ts$.

1085: 

1086: Borrowing terminology from the representation theory of the group

1087: $GL_N$ let us call an irreducible

1088: module over the Hopf algebra $\YN$ \textit{polynomial\/}, 

1089: if it is equivalent to a submodule in the tensor product of

1090: evaluation modules $V(z_1)\ot\ldots\ot V(z_n)$ for some integer $n\ge0$ and

1091: some $z_1\lcd z_n\in\CC\ts$. 

1092: Let us call an irreducible $\YN\ts$-module \textit{rational\/}, 

1093: if it is equivalent to a submodule in a tensor product of 

1094: any number of evaluation

1095: and dual evaluation $\YN\ts$-modules. The tensor product here may be

1096: taken in any suitable order, since the Hopf algebra $\YN$ is not

1097: cocommutative for $N>1$. The irreducible $\YN\ts$-module $\VOm$ is then

1098: rational by definition. The $\YN\ts$-module $\Vlm$ as figured in

1099: Theorem 1.5 is then also rational due to Corollary 4.4.

1100: 

1101: Any irreducible rational

1102: $\YN\ts$-module $V$ can be made polynomial by pulling $V$ back through

1103: an automorphism of $\YN$ of the form (\ref{1.61}) for a suitable series 

1104: $g(x)\in\CC[[x^{-1}]]$ with leading term $1$, cf. \cite[Theorem 2.16]{CP}.

1105: Note that due to the definition (\ref{1.33}),

1106: this pullback of $V$ coincides with the tensor product of $V$

1107: with the one-dimensional $\YN\ts$-module such that

1108: \begin{equation}\label{onedimrep}

1109: T_{ab}(x)\,\mapsto\,\de_{ab}\cdot g(x)\,.

1110: \end{equation}

1111: 

1112: 

1113: Let us show how to make the $\YN\ts$-module $\VOm$ polynomial, 

1114: according to the general assertion above.

1115: The standard tableaux $\Om$ and $\Omt$ are of skew shapes

1116: $\lm$ and $\lmt$ respectively. Suppose $\VOm\neq\{0\}$,

1117: so that the conditions (\ref{con}) are satisfied.

1118: Put $r=\lat_1\ts$. Define 

1119: two partitions $\xi$ and $\eta$ by

1120: $$

1121: \xi_{\ts i}\ts=\,

1122: \left\{

1123: \begin{array}{ll}

1124: \la_{i}-\lat_{\ts N+M-i+1}+r

1125: &\ \ \textrm{if}\ \ \ i\le N+M,\\[2pt]

1126: 0

1127: &\ \ \textrm{if}\ \ \ i>N+M

1128: \end{array}

1129: \right.

1130: $$

1131: and

1132: $$

1133: \eta_{\ts i}\ts=\,

1134: \left\{

1135: \begin{array}{ll}

1136: \mu_{i}-\mut_{\ts M-i+1}+r

1137: &\ \ \textrm{if}\ \ \ i\le M,\\[2pt]

1138: 0

1139: &\ \ \textrm{if}\ \ \ i>M.

1140: \end{array}

1141: \right.

1142: $$

1143: Then we have

1144: $$

1145: \xi_{\ts i}\ge\eta_{\ts i}\ ,\,

1146: \xi^{\,\prime}_{\ts i}-\eta^{\,\prime}_{\ts i}\le N

1147: $$

1148: for every index $i$ due to (\ref{con}).

1149: In particular, the skew Young diagram $\xi/\eta$ is well defined.

1150: Choose any standard tableau $\Gamma$ of shape $\xi/\eta\ts$.

1151: Consider the corresponding $\YN\ts$-module $V_\Gamma\ts$,

1152: it is polynomial by definition. Denote by $V_\Gamma(-r)$ the pullback

1153: of $V_\Gamma$ relative to the automorphism $\tau_{-r}$ of $\YN\ts$.

1154: The $\YN\ts$-module $V_\Gamma(-r)$ is also polynomial by definition.

1155: The proof of the next proposition will be given in Subsection 4.5\ts;

1156: cf.\ \cite[Section A]{KW}.

1157: 

1158: \begin{Proposition}

1159: The\/ 

1160: $\YN$-module\/ $\VOm$ is equivalent to the

1161: pullback of the\/ $\YN\ts$-module\/

1162: $V_\Gamma(-r)$ relative to the automorphism\/

1163: {\rm(\ref{1.61})}, where

1164: $$

1165: g(x)\,=\,g_{\ts\lat}(\ts-\ts x\ns-\!M)

1166: \,\cdot\,

1167: \frac{\,x+M+r}{x+M}\ .

1168: $$

1169: \end{Proposition}

1170: 

1171: \noindent

1172: Together with Theorem 1.5, this proposition provides another realization

1173: of the rational $\YN\ts$-module $\Vlm$.

1174: However, the realization of $\Vlm$

1175: in the space (\ref{mix}) of mixed tensors provided directly by Theorem 1.5

1176: is preferable, as it was for the rational representation

1177: $V_{\nu\ts\nut}$ of the group $GL_N$. 

1178: 

1179: In the particular case when $\nt=0\ts$,

1180: the $\YN\ts$-module $\VOm$ is already polynomial.

1181: Proof of Theorem 1.5 in this case was given in \cite[\ns Section~4]{N2}. 

1182: The present article is a sequel to \cite{N2}, but can be read independently.

1183: 

1184: %==============================================================================

1185: 

1186: \section{\hskip-3.5pt. Fusion procedure}

1187: 

1188: %------------------------------------------------------------------------------

1189: 

1190: \textbf{2.1.}

1191: Let us begin this section with recalling a few classical facts

1192: about the irreducible representations of the symmetric group  $S_l$ over the

1193: field $\CC$.  These representations are labeled by partitions $\la$ of $l\ts$.

1194: We will identify partitions with their Young diagrams.

1195: Denote by $U_\la$ the irreducible representation of $S_l$

1196: corresponding to the partition $\la\ts$. We will also regard

1197: representations of the group $S_l$ as modules over the group ring $\CC S_l$.

1198: Fix the chain of subgroups $S_1\subset S_2\subset\ldots\subset S_l$

1199: with the standard embeddings.

1200: 

1201: There is a decomposition of the vector space $U_\la$ into a direct sum of

1202: one-dimensional subspaces, labeled by the {standard tableaux} of shape $\la$.

1203: The one-dimensional subspace $U_\La\subset U_\la$ corresponding to

1204: a standard tableau $\La$ is defined as follows. For any

1205: $m\in\{1\lcd l-1\}$ take the tableau obtained from $\La$ by

1206: removing the numbers $m+1\lcd l\ts$. Let the Young diagram $\mu$

1207: be the shape of the resulting tableau.

1208: Then the subspace $U_\La$ is contained in an irreducible

1209: $\CC S_m$-submodule of $U_\la$ corresponding to $\mu\ts$. Any basis of

1210: $U_\la$ formed by vectors $u_\La\in U_\La$ is called a 

1211: \textit{Young basis}. Fix an $S_l$-invariant inner product

1212: $(\,\,,\,)$ on $U_\la\ts$. All the subspaces

1213: $U_\La\subset U_\la$ are then pairwise orthogonal. We choose the vectors

1214: $u_\La\in U_\La$ so that $(\ts u_\La,u_\La\ts)=1$.

1215: 

1216: For any standard tableau $\La$ of shape $\la\ts$,

1217: we will work with the diagonal matrix element of the

1218: representation $U_\la$ corresponding to the vector $u_\La$\,,

1219: \begin{equation}\label{2.0}

1220: f_\La=

1221: \sum_{s\in S_l}\,

1222: (\,u_\La\com\ts s\ns\cdot\ns u_\La\,)

1223: \,s\,\in\,\CC S_l\,.

1224: \end{equation}

1225: Note the equality

1226: \begin{equation}\label{2.01}

1227: f_\La^{\ts2}\,=\ts f_\La\cdot l\ts!\ts/\dim U_\la\ts.

1228: \end{equation}

1229: 

1230: For any $k=1\lcd l-1$ let $s_k\in S_l$ be the transposition of $k$ and $k+1$.

1231: We will use the description \cite{Y2} of the action of the generators 

1232: $s_1\lcd s_{l-1}$ of the group $S_l$ on

1233: the vectors of the Young basis.

1234: Fix any standard tableau $\La\ts$. For every

1235: $k=1\lcd l$ let $c_k(\La)$ be the {content}

1236: of the box occupied by $k$ in $\La\ts$. That is, $c_k(\La)=j-i$ if

1237: $k=\La(i\com j)\ts$.

1238: Consider the tableau $s_k\La$ obtained from $\La$ by exchanging the numbers

1239: $k$ and $k+1$. The tableau $s_k\La$ may be non-standard, this

1240: happens exactly when $k$ and $k+1$ stand (next to each other) in the same 

1241: row or column of $\La\ts$. But $c_k(\La)\neq c_{k+1}(\La)$ always, put

1242: $h=(\ts c_{k+1}(\La)-c_k(\La))^{\ts-1}$. If $s_k\La$ is non-standard,

1243: then $h=1$ or $h=-1$.

1244: 

1245: So far the vector $u_\La$ has been determined

1246: up to a multiplier $z\in\CC$ with $|z|=1$. According to \cite{Y2}, all the 

1247: vectors of the Young basis can be further normalized so that

1248: for any standard tableau $\La$ and $k=1\lcd l-1$

1249: \begin{equation}\label{2.3}

1250: s_k\cdot u_\La=

1251: \left\{

1252: \begin{array}{ll}

1253: h\ts u_\La+\sqrt{1-h^2}\, u_{s_k\La}

1254: &\ \ \textrm{if}\ s_k\La\ \textrm{is standard,}\\[2pt]

1255: h\ts u_\La

1256: &\ \ \textrm{otherwise.}

1257: \end{array}

1258: \right.

1259: \end{equation} 

1260: This normalization fixes all the vectors of the Young basis

1261: up to a common multiplier $z\in\CC$ with $|z|=1$. If the tableau $s_k\La$

1262: is standard, then by (\ref{2.3}) 

1263: \begin{equation}\label{2.55555}

1264: \sqrt{1-h^2}\ts\,u_{s_k\La}=(s_k-h)\,u_\La\,.

1265: \end{equation}

1266: Then by the definition (\ref{2.0}) we have the identity

1267: \begin{equation}\label{2.4}

1268: (s_k-h)\,f_\La=f_{s_k\La}\,(s_k+h)\ts.

1269: \end{equation} 

1270: Note that for any two standard tableaux $\La$ and $\Lap$ there exists a

1271: sequence of transpositions $s_{k_1}\lcd s_{k_n}$ such that

1272: $\Lap=s_{k_n}\ldots s_{k_1}\La$ and the tableau $s_{k_m}\ldots s_{k_1}\La$

1273: is standard for each $m=1\lcd n-1$. Then the identities (\ref{2.4})

1274: provide a relation between the matrix elements $f_\La$ and $f_\Lap$

1275: for any $\La\neq\Lap$.

1276: 

1277: We end the present subsection with the following useful identity.

1278: Define the function $g_\la(x)\in\CC(x)$ via (\ref{1.61}).

1279: Then by \cite[Lemma 4.4]{N2} we have

1280: \begin{equation}\label{fact}

1281: g_\la(x)\ =\ 

1282: \prod_{k=1}^l\ 

1283: \biggl(

1284: 1-\frac1{(\ts x-c_k(\La))^2}

1285: \biggr)^{\!-1}\ts.

1286: \end{equation}

1287: 

1288: %------------------------------------------------------------------------------

1289: 

1290: \medskip\noindent\textbf{2.2.}

1291: The matrix element $f_\La\in\CC S_l$

1292: can also be obtained by the so called \textit{fusion procedure}.

1293: For every two distinct indices

1294: $i\com j\in\{1\lcd l\ts\}$ introduce the rational

1295: function of $x\com y\in\CC$

1296: \begin{equation}\label{2.45}

1297: f_{ij}(x\com y)=1-\frac{(\ts i\ts j\ts)}{x-y}\ ,

1298: \end{equation}

1299: valued in $\CC S_l$\,; 

1300: here $(\ts i\ts j\ts)\in S_l$ is the transposition of $i$ and $j\ts$.

1301: As direct calculation shows, these rational functions satisfy the relation

1302: \begin{equation}\label{2.5}

1303: f_{ij}(x\com y)\,f_{ik}(x\com z)\,f_{jk}(y\com z)=

1304: f_{jk}(y\com z)\,f_{ik}(x\com z)\,f_{ij}(x\com y)

1305: \end{equation}

1306: for any three pairwise distinct indices $i\com j\com k\in\{1\lcd l\ts\}\ts$. 

1307: Note that

1308: \begin{equation}\label{2.55}

1309: f_{\ts ij}(x\com y)\,f_{\ts ji}(y\com x)=1-\frac1{(x-y)^{\ts2}}\,.

1310: \end{equation}

1311: 

1312: Now take $l$ complex variables $x_1\lcd x_l$. Order lexicographically

1313: the set of all pairs $(i\com j)$ with $1\le i<j\le l$. The

1314: ordered product over this set,

1315: \begin{equation}\label{2.6}

1316: \prod_{1\le i<j\le l}^{\longrightarrow}\ 

1317: f_{ij}(\ts x_i\com x_j\ts)

1318: \end{equation}

1319: is a rational function of $x_1\lcd x_l$ with values in $\CC S_l\ts$.

1320: This rational function depends only on the differences $x_i-x_j\ts$.

1321: Further, take the complex variables $t_1(\La)\lcd t_l(\La)$ 

1322: constrained like in Subsection 1.2. That is,

1323: \begin{equation}\label{1.22}

1324: t_i(\La)=t_j(\La)

1325: \text{\ \ if $i$ and $j$ occur in the same column of $\La$\ts.}

1326: \end{equation}

1327: So the number of independent variables amongst $t_1(\La)\lcd t_l(\La)$ 

1328: is $\la_1\ts$. Set

1329: \begin{equation}\label{1.2}

1330: x_k=c_k(\La)+t_k(\La)

1331: \text{\ \ for each\ \ $k=1\lcd l$\ts.}

1332: \end{equation}

1333: 

1334: \begin{Proposition}

1335: Restriction of the rational function {\rm(\ref{2.6})} to {\rm(\ref{1.2})}

1336: is regular at $t_1(\La)=\ldots=t_l(\La)$. The value of this restriction at

1337: $t_1(\La)=\ldots=t_l(\La)$ coincides with the element $f_\La\in\CC S_l$.

1338: \end{Proposition}

1339: 

1340: \noindent

1341: This proposition goes back to \cite{C2}.

1342: In its present form, it has been proved in \cite[Section 2.2]{N1}.

1343: The proof actually provides an explicit multiplicative formula for the

1344: element $f_\La\in\CC S_l\ts$, different from the classical 

1345: formula \cite{Y1}.

1346: 

1347: %------------------------------------------------------------------------------

1348: 

1349: \medskip\smallskip\noindent\textbf{2.3.}

1350: We need a generalization of Proposition 2.2

1351: to standard tableaux of skew shapes \cite{C2}.

1352: Take any $m\in\{0\lcd l-1\}$\ts. 

1353: Let $\Up$ be standard tableau obtained from $\La$ by removing

1354: the boxes with numbers $m+1\lcd l$. Let $\mu$ be the shape of

1355: $\Up$. Define a standard tableau $\Om$ of skew shape

1356: $\lm$ by setting 

1357: \begin{equation}\label{Om}

1358: \Om(i\com j)=\La(i\com\ns j)-m

1359: \ \ \text{for each}\ \ 

1360: (i\com j)\in\lm\,.

1361: \end{equation}

1362: Every standard tableau $\Om$ of shape $\lm$

1363: can be obtained from a suitable $\La$ in this way.

1364: 

1365: Put $n=l-m$. Denote by $\io_m$ the embedding of the symmetric group $S_n$

1366: into $S_l$ as a subgroup preserving the subset $\{m+1\lcd l\}$\ts;

1367: we extend the mapping $\io_m$ to $\CC S_n$ by linearity.

1368: Denote by $S_{mn}$ the subgroup $S_m\times\io_m(S_n)$ in $S_l\ts$.

1369: Introduce the linear mapping

1370: \begin{equation}\label{2.85}

1371: \th_m:\CC S_l\to\CC S_{mn}:s\mapsto

1372: \left\{

1373: \begin{array}{ll}

1374: s&\ \textrm{if}\ s\in S_{mn}\ts;\\[2pt]

1375: 0&\ \textrm{otherwise.}

1376: \end{array}

1377: \right.

1378: \end{equation}

1379: By definition, the element $f_\La\in\CC S_l$ is divisible on the

1380: left and on the right by $f_{\ts\Up}\in\CC S_m$. Hence there exists an

1381: element $f_\Om\in\CC S_n$ such that

1382: \begin{equation}\label{2.9}

1383: \th_m(f_\La)=f_{\ts\Up}\cdot\io_m(f_\Om).

1384: \end{equation}

1385: The element $f_\Om\in\CC S_n$ does not depend on the choice of

1386: standard tableau $\Up$ of the shape $\mu$\ts, because the boxes with

1387: the numbers $1\lcd m$ have in $\La$ and $\Up$ the same contents\ts;

1388: see (\ref{2.4}). The generalization of Proposition 2.2 from $\La$

1389: to $\Om$ is based on the following simple observation.

1390: 

1391: \begin{Proposition}

1392: The image under the map\/ $\th_m$ of the product\/

1393:  {\rm(\ref{2.6})} equals

1394: \begin{equation}\label{2.7}

1395: \prod_{1\le i<j\le m}^{\longrightarrow}

1396: f_{ij}(\ts x_i\com x_j\ts)

1397: \hskip5pt\cdot\hskip-6pt

1398: \prod_{m<i<j\le l}^{\longrightarrow} 

1399: f_{ij}(\ts x_i\com x_j\ts)\,.

1400: \end{equation}

1401: \end{Proposition}

1402: 

1403: \noindent

1404: Now consider the ordered product on the right hand side~of~(\ref{2.7})\ts,

1405: \begin{equation}\label{2.10}

1406: \prod_{m<i<j\le l}^{\longrightarrow} 

1407: f_{ij}(\ts x_i\com x_j\ts)\,.

1408: \end{equation}

1409: 

1410: \begin{Corollary}

1411: Restriction of the rational function {\rm(\ref{2.10})} to {\rm(\ref{1.2})} 

1412: is regular at $t_{m+1}(\La)=\ldots=t_l(\La)$. 

1413: The value at $t_{m+1}(\La)=\ldots=t_l(\La)$ of this restriction

1414: coincides with $\io_m(f_\Om)\in\CC S_l$.

1415: \end{Corollary}

1416: 

1417: \noindent

1418: For the proof of Proposition 2.3 and Corollary 2.3,

1419: see \cite[Subsection 2.3]{N2}. 

1420: 

1421: Let us complete the current subsection with another simple observation.

1422: For any two standard tableaux $\La$ and $\Lap$ of the same shape $\la\ts$,

1423: consider the matrix element of the representation $U_\la$ corresponding

1424: to the pair of vectors $u_\La$ and $u_{\La'}$ of the Young basis,

1425: \begin{equation}\label{fll}

1426: f_{\ts\La}^{\ts\La'}=

1427: \sum_{s\in S_l}\,

1428: (\,u_{\La'}\com\ts s\cdot\ns u_\La\,)

1429: \,s\,\in\,\CC S_l\,.

1430: \end{equation}

1431: If $\La=\Lap\ts$, then we have the equality $f_{\ts\La}^{\ts\La'}=f_\La$

1432: by the definition (\ref{2.0}).

1433: Now take any $m\in\{0\lcd l-1\}$. Let $n=l-m$ as before.

1434: Consider the element

1435: $$

1436: \th_m(\ts f_{\ts\La}^{\ts\La'})\in S_{mn}\,,

1437: $$

1438: see (\ref{2.85}).

1439: Let $\Up$ and $\Up^{\ts\prime}$ be the standard tableaux obtained by removing

1440: the boxes with numbers $m+1\lcd l$ from the tableaux $\La$ and

1441: $\Lap$ respectively. 

1442: %The proof of the following lemma is contained in \cite[Subsection 2.4]{N2}. 

1443:  

1444: \begin{Lemma}

1445: We have\/ $\th_m(\ts f_{\ts\La}^{\ts\La'})=0$ unless\/ 

1446: $\Up$ and\/ $\Up^{\ts\prime}$ are of the same shape.

1447: \end{Lemma}

1448: 

1449: \begin{proof}

1450: By its definition, the element $ f_{\ts\La}^{\ts\La'}\in\CC S_l$ is divisible

1451: by $f_{\ts\Up}\in\CC S_m$ on the right, and by $f_{\ts\Up'}\in\CC S_m$ on

1452: the left. The image $\th_m(\ts  f_{\ts\La}^{\ts\La'})\in\CC S_{mn}$ 

1453: inherits these two divisibility properties.

1454: If the tableaux $\Up$ and\/ $\Up^{\ts\prime}$ are not of the same shape,

1455: the irreducible representations $U_\Up$ and $U_{\Up'}$ of $S_m$ are not 

1456: equivalent, and $f_{\ts\Up'}\ts s\,f_{\ts\Up}=0$ for any $s\in S_m$\qed

1457: \end{proof}

1458: 

1459: %------------------------------------------------------------------------------

1460: 

1461: \smallskip\noindent\textbf{2.4.}

1462: Here we collect a few results that

1463: we need for our proof of Theorem~1.5.

1464: Let the symmetric group $S_{l+1}$ act by permutations on $0\com1\lcd l\ts$.

1465: Then the subgroup $\io_1(S_l)\subset S_{l+1}$ still acts by permutations

1466: on $1\lcd l\ts$. 

1467: Consider the rational functions (\ref{2.45}) with

1468: pairwise distinct indices $i\com j\in\{0\com1\lcd l\ts\}$;

1469: these functions take values in the group ring $\CC S_{\ts l+1}\ts$.

1470: Take the matrix element $f_\La\in\CC S_l$ defined by (\ref{2.0}).

1471: Consider the image of $f_\La$ under the

1472: embedding $\io_{\ts1}:\ts\CC S_l\to\CC S_{\ts l+1}\ts$.

1473: For the proof of the next result see \text{\cite[Section 2]{N1}.}

1474: 

1475: \begin{Proposition}

1476: We have equality of rational functions in\/ $x\ts$,

1477: valued in\/ $\CC S_{\ts l+1}$

1478: $$

1479: \ f_{01}(x\com c_1(\La))\ts\ldots f_{0l}(x\com c_{\ts l}(\La))

1480: \cdot\io_{\ts1}(f_\La)

1481: \ts=\ts

1482: \biggl(1\ts-\sum_{k=1}^l\,\frac{(\ts0\ts k)}x\ts\biggr)

1483: \cdot\io_{\ts1}(f_\La)\,.

1484: $$

1485: \end{Proposition}

1486: 

1487: Take any $m\in\{0\lcd l-1\}$ and put $n=l-m\ts$, as in Subsection 2.3.

1488: Also fix any non-negative integer $\lt$. Then take 

1489: any $\mt\in\{0\lcd\lt-1\}$ and put $\nt=\lt-\mt\ts$.

1490: Let the symmetric group $S_{1\ns+\lt+l}$ act by permutations on

1491: \begin{equation}\label{inda}

1492: 0\com1\lcd\lt\com\lt+\ns1\lcd\lt+l\,.

1493: \end{equation}

1494: Define a linear map

1495: $\ga:\CC S_{l+1}\to\CC S_{l+1}$ as follows.

1496: Let the indices $i$ and $j$ range over the sequence (\ref{inda}).

1497: Then for $s\in S_{l+1}$ we set

1498: $\ga(s)=s\,$ if

1499: $$

1500: \lt<i\le\lt+m

1501: \ \Rightarrow\ 

1502: \lt<s(i)

1503: \ \quad\text{and}\ \quad

1504: \nt<j\le\lt

1505: \ \Rightarrow\ 

1506: 0<s^{-1}(j)\le\lt\,;

1507: $$

1508: otherwise $\ga(s)=0$.

1509: Furher, let $h_1\lcd h_{\lt+l}$ be arbitrary complex numbers.

1510: Consider the ordered product in the algebra $\CC S_{1+\lt+l}\,$,

1511: \begin{equation}\label{X}

1512: \prod_{1\le i\le\lt+l}^{\longrightarrow}\!

1513: ((\ts0\ts i\ts)+h_i)\,.

1514: \end{equation}

1515: 

1516: \begin{Lemma}

1517: The image of the product\/ {\rm(\ref{X})}

1518: under the linear map\/ $\ga$ equals

1519: $$

1520: \prod_{1\le i\le\nt}^{\longrightarrow}\,

1521: ((\ts0\ts i\ts)+h_i)

1522: \hskip10pt\cdot\hskip-6pt

1523: \prod_{\lt+m<i\le\lt+l}^{\longrightarrow}\hskip-7pt

1524: ((\ts0\ts i\ts)+h_i)\,.

1525: $$

1526: \end{Lemma}

1527: 

1528: \begin{proof}

1529: Let us expand the ordered product (\ref{X}) as a sum of the products

1530: of transpositions $s=(\ts0\ts i_1)\ldots(\ts0\ts i_a)$ with the coefficients

1531: from $\CC\ts$; the sum is taken over all subsequences

1532: $i_1\lcd i_a$ in the sequence $1\lcd\lt+l\ts$. Take any subequence such that

1533: $\lt<i_b\le\lt+m$ for some $b\le a$. Assume that the index $b$ is minimal

1534: with this property. Then $s(i_b)\le\lt$ so that $\ga(s)=0$.

1535: 

1536: Further, take any subsequence such that

1537: $\nt<i_c\le\lt$ for some $c\le a$. Assume that the index $c$ is maximal

1538: with this property. Then $s^{-1}(i_c)=0$ or 

1539: $s^{-1}(i_c)>l\ts$, so that $\ga(s)=0$

1540: \qed

1541: \end{proof}

1542: 

1543: \noindent

1544: We will also need a reformulation of this lemma.

1545: Let us define a linear map

1546: $\gap\ns:\ts\CC S_{l+1}\to\CC S_{l+1}$ by setting

1547: $\gap(s)=s\,$ if

1548: $$

1549: \nt<i\le\lt

1550: \ \Rightarrow\ 

1551: 0<s(i)\le\lt

1552: \ \quad\text{and}\ \quad

1553: \lt<j\le\lt+m

1554: \ \Rightarrow\ 

1555: s^{-1}(j)>\lt\,;

1556: $$

1557: otherwise $\gap(s)=0$.

1558: Consider the ordered product in the algebra $\CC S_{1+\lt+l}\,$,

1559: \begin{equation}\label{Y}

1560: \prod_{1\le i\le\lt+l}^{\longleftarrow}\!

1561: ((\ts0\ts i\ts)+h_i)

1562: \end{equation}

1563: where the factors corresponding to the indices $i=1\lcd\lt+l$

1564: are arranged from right to left, as indicated by the reversed arrow.

1565: 

1566: \begin{Corollary}

1567: The image of the product\/ {\rm(\ref{Y})}

1568: under the linear map\/ $\gap$ equals

1569: $$

1570: \prod_{\lt+m<i\le\lt+l}^{\longleftarrow}\hskip-7pt

1571: ((\ts0\ts i\ts)+h_i)

1572: \hskip8pt\cdot\,\ts

1573: \prod_{1\le i\le\nt}^{\longleftarrow}\,

1574: ((\ts0\ts i\ts)+h_i)\,.

1575: $$

1576: \end{Corollary}

1577: 

1578: \noindent

1579: This corollary is derived from Lemma 2.4 by using the

1580: anti-automorphism of the group ring $\CC S_{1+\lt+l}\,$, 

1581: such that $s\mapsto s^{-1}$ for every group element $s\ts$.

1582: 

1583: %==============================================================================

1584: 

1585: \section{\hskip-3.5pt. Traceless tensors}

1586: 

1587: %------------------------------------------------------------------------------

1588: 

1589: \textbf{3.1.}

1590: For any positive integer $L$ take the vector spaces $\CC^L$ and

1591: $(\CC^L)^\ast$. Then for any two non-negative integers $l$ and $\lt$

1592: consider the tensor product

1593: \begin{equation}\label{3.00}

1594: \Wll\ts=\ts(\CC^L)^\ast\ot\ldots\ot(\CC^L)^\ast\ot\CC^L\ot\ldots\ot\CC^L

1595: \end{equation}

1596: of $l$ copies of the defining representation $\CC^L$ of the group $GL_L\ts$,

1597: and of $\lt$ copies of the contragredient representation $(\CC^L)^\ast\ts$.

1598: Note that now our basic vector space is $\CC^L$ while

1599: in Section 1 it was $\CC^N\ns$, cf.\ (\ref{mix}).

1600: The space $\CC^N$ will reappear in Subsection 3.5.

1601: Denote by $\Bll$ the commutant of the image of $GL_L$

1602: in the algebra $\End(\Wll)\ts$.

1603: The operator algebra $\Bll$ may be regarded as a subalgebra of the 

1604: \textit{Brauer centralizer algebra\/} \cite{B}, 

1605: see for instance \cite{K}.

1606: 

1607: In this section, let us label the $\ts \lt+l\ts$ tensor factors 

1608: in (\ref{3.00}) by the indices

1609: \begin{equation}\label{ind}

1610: 1\lcd\lt\com\lt+\ns1\lcd\lt+l

1611: \end{equation}

1612: respectively from the left to the right.

1613: Let $Q$ be the linear operator on $(\CC^L)^\ast\ot\CC^L$

1614: defined similarly to the operator (\ref{1.45}) on $(\CC^N)^\ast\ot\CC^N$.

1615: For any indices $i\com j\in\{1\lcd\lt+l\ts\}$ such that 

1616: $i\le\lt<j\ts$, denote by $Q_{ij}$ the operator on $\Wll$ acting as $Q$

1617: on the $i\ts$th and $j\ts$th tensor factors,

1618: and acting as the identity on the remaining 

1619: $\lt\ns+l\ns-\!2\ts$ tensor factors.

1620: Each of the operators $Q_{ij}$ belongs to the commutant $\Bll\,$.

1621: Note the equality $Q_{ij}^{\ts2}=L\cdot Q_{ij}\,$.

1622: 

1623: The subgroup $\Sll=S_{\ts\lt}\times\io_{\ts\lt\ts}(S_l)$ of the

1624: symmetric group $S_{\ts\lt+l}$ acts on $\Wll$ by

1625: permutations of $\lt+l$ tensor factors. 

1626: The image of $\Sll$

1627: in the algebra $\End(\Wll)$ belongs to $\Bll\ts$.

1628: For two distinct indices $i\com j\in\{1\lcd\lt+l\ts\}$ the 

1629: transposition $(i\ts j)\in S_{\ts \lt+l}$ belongs to the subgroup $\Sll$

1630: %$\subset S_{\ts \lt+l}$

1631: if $\ts i\com j\le\lt\ts$ or if $\ts i\com j>\lt\ts$.

1632: Then denote by $P_{ij}$ the corresponding permutational operator on 

1633: $\Wll\,$. The centralizer algebra $\Bll$ is generated by operators $P_{ij}$

1634: and $Q_{ij}$ with all possible indices $i\com j\,$;

1635: see \cite[Lemma 1.2]{K}.   

1636: 

1637: The irreducible modules over the associative algebra $\Bll\ts$ 

1638: or equivalently, the irreducible representations of the general linear 

1639: group $GL_L$ that occur in the tensor product $\Wll\,$, can be indexed

1640: by the pairs of partitions $\la$~and~$\lat$ of respectively 

1641: $l-k$ and $\lt-k\ts$, where $k=0\lcd\min(\ts l\com\lt\ts)\ts$ and 

1642: $\lap_{\ts1}+\lat^{\ts\prime}_{\ts1}\le L\ts$;

1643: see for instance \cite[Corollary 4.7]{S}.

1644: In the present article, we consider only the irreducible

1645: $\Bll\,$-modules such that $k=0$.

1646: The corresponding irreducible representations of the group $GL_L$

1647: occur in the subspace of traceless tensors

1648: $%\begin{equation}\label{3.0}

1649: \Wllo\,\subset\,\Wll\ts\,.

1650: $%\end{equation}

1651: The images of the groups $GL_L$ and $\Sll$

1652: in the algebra $\End(\Wllo\ts)$

1653: span the commutants of each other

1654: \cite[Theorem 1.1]{K}. In other words,

1655: when acting on the vector space $\Wllo\,\ts$, 

1656: the groups $GL_L$ and $\Sll$ form a \textit{dual pair}.

1657: 

1658: By definition, all

1659: the operators $Q_{ij}$ on $\Wll$ vanish on the subspace $\Wllo\,$.

1660: Denote by $\Ill$ the two-sided ideal in $\Bll$ generated by all the operators

1661: $Q_{ij}\ts$. The quotient algebra $\Bll\ts/\,\Ill$ can be identified 

1662: with the image of the group ring

1663: $\CC\Sll$ in the algebra $\End(\Wllo\ts)\ts$.

1664: The image of the element $P_{ij}\in\Bll$ in the quotient algebra 

1665: is then identified with the operator on the subspace 

1666: $\Wllo\,\subset\,\Wll\ts$

1667: corresponding to $(i\ts j)\in\Sll\ts$. 

1668: 

1669: \vskip-2pt

1670: Take any pair of partitions $\la$ of $l$ and $\lat$ of $\lt\ts$,

1671: such that $\lap_{\ts1}+\lat^{\ts\prime}_{\ts1}\le L\ts$.

1672: Denote by $\Vla$ the corresponding irreducible rational representation 

1673: of the group $GL_L\ts$, this representation occurs in the subspace 

1674: $\Wllo\,\ts$.

1675: Further, take the irreducible representations $U_\la$ of $S_l$ and $U_{\lat}$ 

1676: of $S_{\ts\lt}\ts$. Consider the tensor product 

1677: $U_{\lat}\ot U_\la$ as a representation of the group $\Sll\ts$.

1678: This is the irreducible representation, corresponding to $\Vla$

1679: under the duality between $GL_L\ts$ and $\Sll\ts$ on $\Wllo\,\ts$.

1680: Regard $U_{\lat}\ot U_\la$ as a $\Bll\,$-module

1681: by using the homomorphism

1682: \begin{equation}\label{3.22}

1683: \Bll\,\longrightarrow\,\Bll\ts/\,\Ill\,.

1684: \end{equation}

1685: Then $U_{\lat}\ot U_\la$ is the irreducible $\Bll\,$-module

1686: indexed by the pair $\la\ts\com\lat\ts$.

1687: 

1688: %------------------------------------------------------------------------------

1689: 

1690: \smallskip\medskip\noindent\textbf{3.2.}

1691: Let us keep labeling the tensor factors 

1692: in (\ref{3.00}) by indices (\ref{ind}).

1693: For any two distinct indices $i\com j\in\{1\lcd\lt+l\}$ such that 

1694: $\ts i\com j\le\lt\ts$ or $\ts i\com j>\lt\ts$ put

1695: \begin{equation}\label{3.45}

1696: R_{\ts ij}(x\com y)=1-\frac{P_{ij}}{x-y}\,.

1697: \end{equation}

1698: As in Subsection 2.2, here $x$ and $y$ are complex variables. 

1699: The function (\ref{3.45}) of $x$ and $y$ 

1700: is called the \textit{Yang R-matrix\/}.

1701: It corresponds to the function (\ref{2.45}) which now

1702: takes values in $\CC S_{\ts\lt+l}\,$.

1703: By (\ref{2.5}),

1704: for any pairwise distinct indices $i\com j\com k\in\{1\lcd\lt+l\}$

1705: such that $i\com j\com k\le\lt$ or $i\com j\com k>\lt$ we have the relation

1706: \begin{equation}\label{3.5}

1707: R_{\ts ij}(x\com y)\,R_{\ts ik}(x\com z)\,R_{\ts jk}(y\com z)=

1708: R_{\ts jk}(y\com z)\,R_{\ts ik}(x\com z)\,R_{\ts ij}(x\com y)\,,

1709: \end{equation}

1710: it is called the \textit{Yang-Baxter relation\/}. Note that by (\ref{2.55}),

1711: \begin{equation}\label{3.55}

1712: R_{\ts ij}(x\com y)\,R_{\ts ji}(y\com x)=1-\frac1{(x-y)^{\ts2}}\,.

1713: \end{equation}

1714: 

1715: For any two indices $i\com j\in\{1\lcd\lt+l\ts\}$ such that 

1716: $i\le\lt<j\ts$, put

1717: \begin{equation}\label{3.555}

1718: \hskip40pt

1719: \Rt_{ij}(x\com y)=1-\frac{Q_{ij}}{x-y}

1720: \textrm{\hskip10pt and\hskip10pt}

1721: \Rb_{ij}(x\com y)=1-\frac{Q_{ij}}{x+y+L}\,\ts;

1722: \end{equation}

1723: then we have

1724: 

1725: \vskip-20pt

1726: \begin{equation}\label{3.6}

1727: \Rt_{ij}(-x\com y)\ \Rb_{ij}(x\com y)=1\,.

1728: \end{equation}

1729: 

1730: The linear operator $Q$ on $(\CC^L)^\ast\ot\CC^L$

1731: is obtained from the permutation operator on $\CC^L\ot\CC^L$

1732: by conjugation in the first tensor factor. Using this observation

1733: along with the relation (\ref{2.5}), we get the relation

1734: \begin{equation}\label{3.7}

1735: \ \qquad

1736: \Rt_{\ts ik}(x\com z)\,\Rt_{\ts ij}(x\com y)\,R_{\ts jk}(y\com z)=

1737: R_{\ts jk}(y\com z)\,\Rt_{\ts ij}(x\com y)\,\Rt_{\ts ik}(x\com z)

1738: \end{equation}

1739: for any indices $i\com j\com k\in\{1\lcd\lt+l\ts\}$ such that

1740: $i\le\lt<j\com k$ and $j\neq k\ts$.

1741: By changing $x$ to $-x$ in (\ref{3.7}) and then

1742: using (\ref{3.6}), we obtain the relation

1743: \begin{equation}\label{3.8}

1744: \ \,\qquad

1745: \Rb_{\ts ij}(x\com y)\,\Rb_{\ts ik}(x\com z)\,R_{\ts jk}(y\com z)=

1746: R_{\ts jk}(y\com z)\,\Rb_{\ts ik}(x\com z)\,\Rb_{\ts ij}(x\com y)\,.

1747: \end{equation}

1748: Similarly, when $i\com j\le\lt<k$ and $i\neq j$,

1749: we obtain the relation

1750: \begin{equation}\label{3.85}

1751: \ \,\qquad

1752: R_{\ts ji}(y\com x)\,\Rb_{\ts ik}(x\com z)\,\Rb_{\ts jk}(y\com z)=

1753: \Rb_{\ts jk}(y\com z)\,\Rb_{\ts ik}(x\com z)\ts R_{\ts ji}(y\com x)\ts.

1754: \end{equation}

1755: To obtain the relation (\ref{3.85}), we have also used the identity

1756: \begin{equation}\label{3.75}

1757: R_{ij}(-x\com\ns-y)=R_{ji}(y\com x)\,.

1758: \end{equation}

1759: The relations (\ref{3.5})\com\ts(\ref{3.55}) and

1760: (\ref{3.6}) to (\ref{3.85}) are equalities

1761: of functions that take their values in the operator algebra $\Bll\,$.

1762: 

1763: Now take $\lt+l$ complex variables

1764: $x_1\lcd x_{\ts\lt+l}\,$.

1765: Consider the rational function of these variables, 

1766: \begin{equation}\label{3.9}

1767: \prod_{1\le i<j\le\lt}^{\longrightarrow}\,

1768: R_{\ts ji}(\ts x_j\com x_i\ts)

1769: \ts\ \cdot\!\!\!

1770: \prod_{\substack{1\le i\le\lt\\\lt<j\le\lt+l}}^{\longrightarrow}\,\ts

1771: \Rb_{\ts ij}(\ts x_i\com x_j\ts)

1772: \ts\ \cdot\!\!\!\!\!\!

1773: \prod_{\lt<i<j\le\lt+l}^{\longrightarrow}\!\ns

1774: R_{\ts ij}(\ts x_i\com x_j\ts)

1775: \end{equation}

1776: taking values in the algebra $\Bll\,$. In each of the three

1777: ordered products in (\ref{3.9}), 

1778: the pairs $(i\com j)$ are ordered lexicographically. 

1779: By using the relations (\ref{3.85}), the entire product (\ref{3.9})

1780: can be rewritten as

1781: $$

1782: \prod_{1\le i\le\lt}^{\longleftarrow}\ \biggl(\ 

1783: \prod_{\lt<j\le\lt+l}^{\longrightarrow}\,

1784: \Rb_{\ts ij}(\ts x_i\com\ts x_j\ts)\biggr)

1785: $$

1786: \vglue-6pt

1787: \begin{equation}\label{3.999}

1788: \times\ 

1789: \prod_{1\le i<j\le\lt}^{\longrightarrow}\,

1790: R_{\ts ji}(\ts x_j\com x_i\ts)

1791: \ts\ \cdot\!\!\!\!\!\!

1792: \prod_{\lt<i<j\le\lt+l}^{\longrightarrow}\!\ns

1793: R_{\ts ij}(\ts x_i\com x_j\ts)\ .

1794: \end{equation}

1795: By using the relations (\ref{3.8}), the product (\ref{3.9})

1796: can also be rewritten as

1797: $$

1798: \prod_{1\le i<j\le\lt}^{\longrightarrow}\,

1799: R_{\ts ji}(\ts x_j\com x_i\ts)

1800: \ts\ \cdot\!\!\!\!\!\!

1801: \prod_{\lt<i<j\le\lt+l}^{\longrightarrow}\!\ns

1802: R_{\ts ij}(\ts x_i\com x_j\ts)

1803: $$

1804: \vglue-6pt

1805: \begin{equation}\label{3.9999}

1806: \times\ 

1807: \prod_{1\le i\le\lt}^{\longrightarrow}\ \biggl(\ 

1808: \prod_{\lt<j\le\lt+l}^{\longleftarrow}\,

1809: \Rb_{\ts ij}(\ts x_i\com x_j\ts)\biggr)\ .

1810: \end{equation} 

1811: In the product displayed in the first line of

1812: (\ref{3.999}), the factors corresponding to

1813: $i=1\lcd\lt$ are arranged from right to left, as indicated

1814: by the reversed arrow; a similar convention has been used in the

1815: display (\ref{3.9999}).

1816: 

1817: %------------------------------------------------------------------------------

1818: 

1819: \smallskip\medskip\noindent\textbf{3.3.}

1820: Denote by $W_l$ and $W_{\lt\ts}^{\ts\ast}$ the tensor products of 

1821: respectively

1822: $l$ and $\lt$ copies of the vector spaces $\CC^L$ and $(\CC^L)^\ast$.

1823: Choose two standard tableaux $\La$ and $\Lat\ts$, of shapes $\la$ and $\lat$

1824: respectively. Denote by $F_\La$ and $F_{\Lat}$ the linear operators on 

1825: respectively $W_l$ and $W_{\lt\ts}^{\ts\ast}$, 

1826: corresponding to the diagonal matrix elements

1827: $f_\La\in\CC S_l$ and $f_{\Lat}\in\CC S_{\ts\lt}\,\ts$; see Subsection 2.1. 

1828: Let $P_{\ts\lt}$ be the linear operator on $W_{\lt\ts}^{\ts\ast}\ts$, 

1829: reversing the order of $\lt$ tensor factors $(\CC^L)^\ast$.

1830: 

1831: Take $l$ complex variables $t_1(\La)\lcd t_l(\La)$ 

1832: constrained by (\ref{1.22}). Setting

1833: \begin{equation}\label{se}

1834: x_{\ts\lt+k}=c_{k}(\La)+t_{k}(\La)

1835: \text{\ \ for each\ \ $k=1\lcd l$}

1836: \end{equation}

1837: in the product

1838: $$

1839: \prod_{\lt<i<j\le\lt+l}^{\longrightarrow}\!\ns

1840: R_{\ts ij}(\ts x_i\com x_j\ts)

1841: $$

1842: we obtain a rational function of the constrained variables

1843: $t_1(\La)\lcd t_l(\La)\ts$. By Proposition 2.2, this function is regular at 

1844: \begin{equation}\label{or}

1845: t_1(\La)=\ldots=t_l(\La)=0\ts.

1846: \end{equation}

1847: The value of this function at (\ref{or})

1848: coincides with the operator $1\ot F_\La$ on

1849: \begin{equation}\label{del}

1850: \Wll\,=\,W_{\lt\ts}^{\ts\ast}\ot W_l\,.

1851: \end{equation}

1852: 

1853: Furthermore, take the $\lt$ constrained variables

1854: $t_1(\Lat)\lcd t_{\ts\lt\ts}(\Lat)\ts$. Setting

1855: \begin{equation}\label{set}

1856: x_{\ts\lt-k+1}=c_{k}(\Lat)+t_{k}(\Lat)

1857: \text{\ \ for each\ \ $k=1\lcd\lt$}

1858: \end{equation}

1859: in

1860: $$

1861: \prod_{1\le i<j\le\lt}^{\longrightarrow}\,

1862: R_{\ts ji}(\ts x_j\com x_i\ts)

1863: \ =\ 

1864: P_{\ts\lt}\hskip4pt\cdot\hskip-6pt

1865: \prod_{1\le i<j\le\lt}^{\longrightarrow}\,

1866: R_{\ts ij}(\ts x_{\ts\lt-i+1}\com x_{\ts\lt-j+1}\ts)\hskip4pt\cdot\hskip3pt

1867: P_{\ts\lt}

1868: $$

1869: we obtain a rational function of $t_1(\Lat)\lcd t_{\ts\lt\ts}(\Lat)$

1870: which is regular at

1871: \begin{equation}\label{ort}

1872: t_1(\Lat)=\ldots=t_{\ts\lt\ts}(\Lat)=0\ts.

1873: \end{equation}

1874: Here we apply Proposition 2.2 to the tableau $\Lat$ instead of $\La\ts$,

1875: and use the Yang-Baxter relation (\ref{3.5}) repeatedly.

1876: The value of this restriction at (\ref{ort}) equals the operator

1877: $(\ts P_{\ts\lt}\,F_{\Lat}\,P_{\ts\lt}\ts)\ot1$ 

1878: on the vector space $\Wll\,$.

1879: 

1880: Finally, note that for any $k\in\{1\lcd l\}$ and $i\in\{1\lcd\lt\ts\}$

1881: we have

1882: $$

1883: c_k(\La)+c_i(\Lat)\,\ge\,2-\lap_1-\lat^{\ts\prime}_1\,\ge\,2-L

1884: $$

1885: due to our assumptions on the partitions $\la$ and $\lat\ts$.

1886: Using observations made in this and the previous subsection,

1887: we arrive at the following proposition.

1888: 

1889: \begin{Proposition}

1890: Restriction of the rational function 

1891: {\rm(\ref{3.9})} to {\rm(\ref{se}),(\ref{set})} 

1892: is regular at {\rm(\ref{or}),(\ref{ort})}.

1893: The value of this restriction at {\rm(\ref{or}),(\ref{ort})} equals

1894: $$

1895: \prod_{1\le i\le\lt}^{\longrightarrow}\ \biggl(\ 

1896: \prod_{1\le k\le l}^{\longrightarrow}\ 

1897: \biggl(1-\frac{Q_{\ts\lt-i+1\ts,\ts\lt+k}}

1898: {\ts c_k(\La)+{c_i(\Lat)}^{\phantom{\prime}}\ns\!+L}

1899: \,\biggr)\ns\biggr)\ \cdot\ 

1900: (\ts P_{\ts\lt}\,F_{\Lat}\,P_{\ts\lt}\ts)\ot F_\La

1901: $$

1902: $$

1903: \ =\ (\ts P_{\ts\lt}\,F_{\Lat}\,P_{\ts\lt}\ts)\ot F_\La\ \,\cdot\,

1904: \prod_{1\le i\le\lt}^{\longleftarrow}\ \biggl(\ 

1905: \prod_{1\le k\le l}^{\longleftarrow}\ 

1906: \biggl(1-\frac{Q_{\ts\lt-i+1\ts,\ts\lt+k}}

1907: {\ts c_k(\La)+{c_i(\Lat)}^{\phantom{\prime}}\ns\!+L}

1908: \,\biggr)\ns\biggr)\,.

1909: $$

1910: \end{Proposition}

1911: 

1912: \noindent

1913: Let us define the linear operator $\FLa\ts$ on the vector space $\Wll$ 

1914: as either of the two (equal) products displayed in Proposition 3.3. 

1915: Using this

1916: definition of $\FLa$ along with the relations (\ref{2.4}) and (\ref{3.8}),

1917: we obtain the identity

1918: $$

1919: P_{\ts\lt+k\ts,\ts\lt+k+1}\ts 

1920: R_{\ts\lt+k+1\ts,\ts\lt+k}\ts(\ts c_{k+1}(\La)\com c_k(\La))

1921: \,\cdot\,\FLa

1922: $$

1923: \begin{equation}\label{kid}

1924: \hskip42pt

1925: \,=\ts\,F_{s_k\La,\Lat\ts}\,\cdot\,

1926: P_{\ts\lt+k\ts,\ts\lt+k+1}\ts 

1927: R_{\ts\lt+k\ts,\ts\lt+k+1}\ts(\ts c_{k}(\La)\com c_{k+1}(\La))

1928: \end{equation}

1929: for any $k\in\{1\lcd l-1\}$ such that the tableau $s_k\La$ is standard.

1930: Similarly, by using the relation (\ref{3.85}) instead of (\ref{3.8}), 

1931: we obtain the identity

1932: $$

1933: P_{\ts\lt-i\ts,\ts\lt-i+1}\ts 

1934: R_{\ts\lt-i\ts,\ts\lt-i+1}\ts(\ts c_{i+1}(\Lat)\com c_i(\Lat))

1935: \,\cdot\,\FLa

1936: $$

1937: \begin{equation}\label{iid}

1938: \hskip48pt

1939: \,=\ts\,F_{\La,s_i\Lat\ts}\,\cdot\,

1940: P_{\ts\lt-i\ts,\ts\lt-i+1}\ts 

1941: R_{\ts\lt-i+1\ts,\ts\lt-i}\ts(\ts c_{i}(\Lat)\com c_{i+1}(\Lat))

1942: \end{equation}

1943: for any index $i\in\{1\lcd\lt-1\}$ such that 

1944: the tableau $s_i\Lat$ is standard.

1945: The identities (\ref{kid}) and (\ref{iid}) provide relations

1946: between the operators $\FLa$ for different pairs $\La$ and $\Lat$,

1947: see the end of Subsection 2.1. 

1948: 

1949: Denote by $\beta$ the involutive antiautomorphism of the algebra $\Bll\,$,

1950: such that each of the generators $P_{ij}$ and $Q_{ij}$ of $\Bll$  

1951: is $\beta\ts$-invariant. The equality of two products displayed in

1952: Proposition 3.3 can be then reformulated~as

1953: 

1954: \begin{Corollary}

1955: The element\/ $\FLa\in\Bll\ts$ is\/ $\beta\ts$-invariant.

1956: \end{Corollary}

1957: 

1958: %------------------------------------------------------------------------------

1959: 

1960: \noindent\textbf{3.4.}

1961: Denote by $\VLa$ the image of the operator

1962: $\FLa$ on the vector space $\Wll\ts$.

1963: Denote respectively by $V_\La$ and $V_{\ns\Lat}^{\ts\ast}$ the images of

1964: the operators $F_\La$ and $F_{\Lat}$ on the vector spaces $W_l$ and 

1965: $W_{\lt\ts}^{\ts\ast}\ts$. Then

1966: \begin{equation}\label{inc}

1967: \VLa\ts\subset\ts

1968: (\ts P_{\ts\lt}\cdot V_{\Lat}^{\ts\ast})\ot V_\La

1969: \end{equation}

1970: due to the (second) definition of the operator $\FLa\,$. Take the subspace

1971: $\Wllo\ts$ of traceless tensors. The next proposition is pivotal

1972: for the present article.

1973: 

1974: \begin{Proposition}

1975: We have the equality of vector spaces

1976: $$

1977: \VLa\ts=\ts

1978: ((\ts P_{\ts\lt}\cdot V_{\ns\Lat}^{\ts\ast})\ot V_\La)

1979: \,\cap\,\Wllo\ts\,.

1980: $$

1981: \end{Proposition}

1982: 

1983: \begin{proof}

1984: By definition, each of the operators $Q_{ij}$ vanishes on the subspace

1985: $\Wllo\,$. So by the (second) definition of the operator $\FLa\ts$,

1986: its action on~the subspace $\Wllo\ts$ coincides with that

1987: of the operator $(\ts P_{\ts\lt}\,F_{\Lat}\,P_{\ts\lt}\ts)\ot F_\La\ts$.~Hence

1988: $$

1989: \VLa\,\supset

1990: ((\ts P_{\ts\lt}\,F_{\Lat}\,P_{\ts\lt}\ts)\ot F_\La\ts)\,\cdot\ts\Wllo

1991: \ts\,=\,

1992: ((\ts P_{\ts\lt}\cdot V_{\ns\Lat}^{\ts\ast})\ot V_\La)

1993: \,\cap\,\Wllo\,\ts.

1994: $$

1995: Due to (\ref{inc}), it now remains to show that $\VLa\subset\Wllo\,$.

1996: Equivalently, we have to show that $Q_{ij}\,\FLa=0$

1997: for any possible indices $i$ and $j\ts$.

1998: 

1999: By the (first) definition of $\FLa\,$, this operator

2000: is divisible on the left by

2001: $$

2002: 1-\frac{Q_{\ts\lt,\lt+1}}

2003: {\ts c_1(\La)+{c_1(\Lat)}^{\phantom{\prime}}\ns\!+L}

2004: \ =\ 1-\frac{\ts Q_{\ts\lt,\lt+1}\ts}{L}\,\ts.

2005: $$

2006: So the relation 

2007: $$

2008: Q_{\ts\lt,\lt+1}^{\ts2}\ns=\ts L\cdot Q_{\ts\lt,\lt+1}

2009: $$ implies

2010: the equality $Q_{ij}\,\FLa=0$ for $i=\lt\com j=\lt+1$ 

2011: and arbitrary $\La\com\Lat\ts$.

2012: 

2013: If for some $k\in\{1\lcd l-1\}\ts$

2014: the tableau $s_k\La$ is standard, then by (\ref{kid})

2015: $$

2016: P_{\ts\lt+k\ts,\ts\lt+k+1}\ts\FLa\,=\,

2017: \frac{\FLa}{c_{k+1}(\La)-c_k(\La)}\ \ts+

2018: $$

2019: $$

2020: F_{s_k\La,\Lat\ts}\cdot

2021: \biggl(

2022: P_{\ts\lt+k\ts,\ts\lt+k+1}-\frac1{\ts c_k(\La)-c_{k+1}(\La)\ts}

2023: \ts\biggr)\ts.

2024: $$

2025: If the tableau $s_k\La$ is not standard, then 

2026: $$

2027: P_{\ts\lt+k\ts,\ts\lt+k+1}\ts\FLa\,=\,\pm\,\FLa\ts\,;

2028: $$

2029: the latter relation follows from the (second) definition of $\FLa$ 

2030: by (\ref{2.3}).

2031: If for some $i\in\{1\lcd l-1\}\ts$

2032: the tableau $s_i\Lat$ is standard, then due to (\ref{iid})

2033: $$

2034: P_{\ts\lt-i\ts,\ts\lt-i+1}\ts\FLa\,=\,

2035: \frac{\FLa}{{c_{i+1}(\Lat)-c_i(\Lat)}^{\phantom{\prime}}\ns\!}\ \ts+

2036: $$

2037: $$

2038: F_{\La,s_i\Lat\ts}\cdot

2039: \biggl(

2040: P_{\ts\lt-i\ts,\ts\lt-i+1}-

2041: \frac1{\ts{c_i(\Lat)-c_{i+1}(\Lat)}^{\phantom{\prime}}\ns\!\ts}

2042: \ts\biggr)\,.

2043: $$

2044: If the tableau $s_i\Lat$ is not standard, then again due to (\ref{2.3})

2045: $$

2046: P_{\ts\lt-i\ts,\ts\lt-i+1}\ts\FLa\,=\,\pm\,\FLa\,.

2047: $$

2048: 

2049: For any permutation $s\in\Sll\ts$, let $P_s$ be

2050: the corresponding linear operator on the tensor product $\Wll\ts$.

2051: The four relations displayed above imply that

2052: for certain operators $R_{\La^\prime\ns\Lat^\prime}(s)$ on $\Wll\ts$, 

2053: we have the equality

2054: $$

2055: P_s\,\FLa=\ts

2056: \sum_{\La^\prime\ns,\Lat^\prime}\,\ts 

2057: F_{\La^\prime\ns\Lat^\prime}\,R_{\La^\prime\ns\Lat^\prime}(s)\,;

2058: $$

2059: here $\La^\prime\com\Lat^{\ts\prime}$ range

2060: over all standard tableaux of shapes $\la\ts\com\lat$ respectively.

2061: For any $i\in\{1\lcd\lt\ts\}$ and $j\in\{\lt+1\lcd\lt+l\ts\}$

2062: we can find a permutation $s\in\Sll$ so that $s(i)=\lt$ and

2063: $s(j)=\lt+1$. Then we get the equalities

2064: $$

2065: Q_{ij}\,\FLa\,=\,P_s^{\ts-1}\ts Q_{\ts\lt,\lt+1}\,P_s\ts\FLa

2066: \,=\,\sum_{\La^\prime\ns,\Lat^\prime}\,

2067: P_s^{\ts-1}\ts Q_{\ts\lt,\lt+1}\ts F_{\La^\prime\ns\Lat^\prime}\ts

2068: R_{\La^\prime\ns\Lat^\prime}(s)=0\quad\qed

2069: $$

2070: \end{proof}

2071: 

2072: \noindent

2073: By definition, the associative algebra $\Bll$ is semisimple.

2074: Let $\Bla$ be the simple ideal of $\Bll$

2075: corresponding to the irreducible $\Bll\,$-module $U_{\lat}\ot U_\la\ts$.

2076: Let $s_0$ be the element of

2077: the maximal length in the symmetric group $S_{\lt}\ts$.

2078: Proposition 3.4 provides the following characterization of $\FLa\ts$.

2079: 

2080: \begin{Corollary}

2081: The operator $\FLa\in\Bll$

2082: is the unique element of the simple ideal\/ $\Bla\,$,

2083: with the image under\/ {\rm(\ref{3.22})} corresponding to the element

2084: \begin{equation}\label{cor}

2085: (\ts s_0\ts f_{\Lat}\ts s_0)\ts\cdot\ts\io_{\lt}\ts(f_\La)\,\in\,\CC\Sll\,.

2086: \end{equation}

2087: \end{Corollary}

2088: 

2089: \begin{proof}

2090: By definition, we have 

2091: $$

2092: \FLa\ts\in\ts(\ts P_{\ts\lt}\,F_{\Lat}\,P_{\ts\lt}\ts)\ot F_\La+\,\Ill\,\ts.

2093: $$

2094: Hence the image of $\FLa$ under the 

2095: homomorphism {\rm(\ref{3.22})} corresponds to 

2096: the element (\ref{cor}).

2097: Due to Proposition 3.4, we also have $\FLa\in\Bla\ts$.

2098: But any element of the ideal $\Bla\subset\Bll$

2099: is uniquely determined by the image of this element under

2100: the homomorphism (\ref{3.22})\qed

2101: \end{proof}

2102: 

2103: \noindent

2104: Due to (\ref{2.01}), the above characterization of

2105: the operator $\FLa$ implies that

2106: $$%\begin{equation}\label{3.301}

2107: \FLa^{\,2}\,=\,\FLa\,\cdot\,

2108: l\ts!\ts/\dim U_\la\,\cdot\,

2109: \lt\ts!\ts/\dim U_{\lat}\,.

2110: $$%\end{equation}

2111: 

2112: %------------------------------------------------------------------------------

2113: 

2114: \smallskip\noindent\textbf{3.5.}

2115: We will now extend the results of Subsection 3.3 to standard tableaux

2116: of skew shapes. Take any $m\in\{0\lcd l-1\}$ and $\mt\in\{0\lcd\lt-1\}$\ts. 

2117: As in

2118: Subsection~2.3, denote by $\Up$ be standard tableau obtained from $\La$

2119: by removing the boxes with the numbers $m+1\lcd l$. Let $\mu$ be the shape of

2120: the tableau $\Up$. Define a standard tableau $\Om$ of the skew shape

2121: $\lm$ as in (\ref{Om}).

2122: Similarly, define the partition $\mut$ of $\mt\ts$, the standard tableau $\Upt$

2123: of shape $\mut\ts$, and the standard tableau $\Omt$ of shape $\lat/\mut\ts$.

2124: Put

2125: $$

2126: n=l-m

2127: \quad\text{and}\quad 

2128: \nt=\lt-\mt\,.

2129: $$

2130: 

2131: Take any $M\in\{0\lcd L-1\}$ and put $N=L-M$. 

2132: Fix a decomposition $\CC^L=\CC^N\oplus\CC^M$,

2133: it provides the decomposition $(\CC^L)^\ast=(\CC^N)^\ast\oplus(\CC^M)^\ast$.

2134: Consider the tensor product

2135: \begin{equation}\label{3.000}

2136: \Wmm\ts=\ts(\CC^M)^\ast\ot\ldots\ot(\CC^M)^\ast\ot\CC^M\ot\ldots\ot\CC^M

2137: \end{equation}

2138: of $m$ copies of the defining representation $\CC^M$ of the group $GL_M\ts$,

2139: and of $\mt$ copies of the representation $(\CC^M)^\ast\ts$.

2140: We will use the factorization

2141: $$%\begin{equation}\label{dem}

2142: \Wmm\,=\,W_{\mt\ts}^{\ts\ast}\ot W_m\,.

2143: $$%\end{equation}

2144: similar to (\ref{den}) and (\ref{del}).

2145: Consider also the subspace of traceless tensors

2146: $\Wmmo\subset\Wmm\,$.

2147: There is a unique $GL_M\ts$-equivariant projector

2148: $$

2149: I_{m\mt}:\,\Wmm\,\to\,\Wmmo\,;

2150: $$

2151: see Subsection 3.1.

2152: Assume that the irreducible rational representation $V_{\mu\ts\mut}$

2153: of $GL_M$ appears~in $\Wmmo\subset\Wmm\,$. 

2154: Equivalently, assume that $\mup_1+\mut_1^{\,\prime}\le M$.

2155: 

2156: In Section 1, we denoted respectively by $W_n$ and $W_{\ts\nt}^{\ts\ast}$ 

2157: the tensor products of $n$ and $\nt$ copies of $\CC^N$ and $(\CC^N)^\ast$,

2158: so that we had the factorization (\ref{den}). Let us now regard

2159: the tensor product of vector spaces $W_{\ts\nt}^{\ts\ast}\ot\Wmm\ot W_n$ 

2160: as a direct summand of $\Wll\ts$, 

2161: using the decompositions $\CC^L=\CC^N\oplus\CC^M$

2162: and $(\CC^L)^\ast=(\CC^N)^\ast\oplus(\CC^M)^\ast$ 

2163: of the $\lt+l$ tensor factors in (\ref{3.00}). Let

2164: $J_{m\mt}$ the projector of $\Wll$ to this direct summand.

2165: Further, consider the subspace

2166: \begin{equation}\label{3.41}

2167: \Zmm\,=\,

2168: W_{\ts\nt}^{\ts\ast}\ot\Wmmo\ot W_n

2169: \,\subset\,

2170: W_{\ts\nt}^{\ts\ast}\ot\Wmm\ot W_n

2171: \,\subset\,

2172: \Wll\,.

2173: \end{equation}

2174: Denote by $H_{m\mt}$ the composition of the operators $1\ot I_{m\mt}\ot1$

2175: and $J_{m\mt}\ts$, 

2176: this composition is a projector to the subspace $\Zmm\subset\Wll\,$.

2177: For any linear operator $A$ on $\Wll\ts$

2178: determine the operator on the same space,

2179: \begin{equation}\label{ave}

2180: A^\vee=\,H_{m\mt}\,A\,H_{m\mt}\,.

2181: \end{equation}

2182: 

2183: We may also regard $A^\vee$ as an operator on the subspace $\Zmm\ts$.

2184: For any $A\in\Bll$ let $\Th_{m\mt}(A)$ be the operator 

2185: on the vector space $\Wmmo\ot W_{\ts\nt}^{\ts\ast}\ot W_n$,

2186: obtained from $A^\vee$ by identifying this vector space with $\Zmm$

2187: %%%$W_{\ts\nt}^{\ts\ast}\ot\Wmmo\ot W_n$ 

2188: in~the natural way, that is by

2189: exchanging the tensor factors $\Wmmo$ and $W_{\ts\nt}^{\ts\ast}$.

2190: Let $\Bmm$ and $\Bnn$ be the commutants of the actions of the~groups

2191: $GL_M$ and $GL_N$ on the vector spaces $\Wmm$ and $\Wnn$ respectively.

2192: Consider also the two-sided ideal $\Imm\subset\Bmm$ 

2193: %%%and $\Inn\subset\Bnn\ts$, 

2194: defined as in Subsection 3.1.

2195: The projector $H_{m\mt}$ commutes with action of subgroup

2196: $GL_N\times GL_M\ts\subset\ts GL_L$ on $\Wll\,$. 

2197: Hence for $A\in\Bll\ts$ %%%we have

2198: $$

2199: \Th_{m\mt}(A)\in(\,\Bmm/\,\Imm)\ot\Bnn\,;

2200: $$

2201: Here the quotient algebra $\ts\Bmm/\,\Imm\ts$ is identified with 

2202: the image of $\CC S_{\mt m}$ in the operator algebra $\End(\Wmmo\ts)\ts$. 

2203: Thus we obtain a linear map

2204: $$%\begin{equation}\label{3.46}

2205: \Th_{m\mt}\ts:\,\Bll\,\rightarrow\,(\Bmm/\,\Imm)\ot\Bnn\,.

2206: $$%\end{equation}

2207: 

2208: As the mapping $A\mapsto A^\vee$ is linear, for any possible 

2209: $i\com j\in\{1\lcd\lt+l\}$

2210: $$

2211: \Rp_{ij}(x\com y)=1-\frac{\Pp_{ij}}{x-y}

2212: \qquad\textrm{and}\qquad

2213: \Rbp_{ij}(x\com y)=1-\frac{\Qp_{ij}}{x+y+L}\,.

2214: $$

2215: Note that the operator $\Pp_{ij}$ on the subspace $\Zmm$ coincides

2216: with restriction of the permutational operator $P_{ij}$ on $\Wll$

2217: to this subspace, whenever

2218: $$

2219: j<i\le\nt

2220: \quad\text{or}\quad

2221: \nt<j<i\le\lt

2222: \quad\text{or}\quad

2223: \lt<i<j\le\lt+m

2224: \quad\text{or}\quad

2225: \lt+m<i<j\,.

2226: $$

2227: Also note that if

2228: $$

2229: i\le\nt

2230: \quad\text{and}\quad

2231: \lt+m<j\,,

2232: $$ 

2233: then the operator $\Qp_{ij}$ on the subspace

2234: $\Zmm$ acts as (\ref{1.45}) in $i\ts$th tensor factor $(\CC^N)^\ast$

2235: and the $(j\ns-\lt-m\ts)\ts$th tensor factor $\CC^N\ns$, and acts

2236: as the identity in~all the remaining tensor factors.

2237: 

2238: Now consider the rational function (\ref{3.9}) with values

2239: in the algebra $\Bll\ts$, of the variables $x_1\lcd x_{\ts\lt+l}\,$.

2240: In the following proposition, these variables need not to be restricted

2241: by (\ref{se}) and (\ref{set}).

2242: 

2243: \begin{Proposition}

2244: The image of the product\/

2245: {\rm(\ref{3.9})} under the map\/ $A\mapsto A^\vee$~is

2246: $$%\begin{equation}\label{3.42}

2247: \prod_{1\le i<j\le\nt}^{\longrightarrow}

2248: \Rp_{\ts ji}(\ts x_j\com x_i\ts)

2249: \ts\ \cdot\hskip-12pt

2250: \prod_{\substack{1\le i\le\nt \\ \lt+m<j\le\lt+l}}^{\longrightarrow}\!\!

2251: \Rbp_{\ts ij}(\ts x_i\com x_j\ts)

2252: \ts\ \cdot\hskip-16pt

2253: \prod_{\lt+m<i<j\le\lt+l}^{\longrightarrow}\hskip-8pt

2254: \Rp_{\ts ij}(\ts x_i\com x_j\ts)

2255: $$%\end{equation}

2256: \begin{equation}\label{3.43}

2257: \times

2258: \prod_{\nt<i<j\le\lt}^{\longrightarrow}\,

2259: \Rp_{\ts ji}(\ts x_j\com x_i\ts)

2260: \ts\ \cdot\hskip-12pt

2261: \prod_{\lt<i<j\le\lt+m}^{\longrightarrow}\!\!\!

2262: \Rp_{\ts ij}(\ts x_i\com x_j\ts)\,.

2263: \end{equation}

2264: \end{Proposition}

2265: 

2266: \begin{proof}

2267: Denote by $W_{\ts\lt+l}$ the tensor product of $\lt+l$ copies of

2268: the vector space $\CC^L$. By conjugating

2269: any linear operator $B$ on the vector space $W_{\ts\lt+l}$ relative to each

2270: of the first $\lt$ tensor factors $\CC^L$, we obtain a certain linear

2271: operator on $\Wll\,$, which will be denoted by $\Bt$.

2272: 

2273: As observed in Subsection 3.2, the ordered product (\ref{3.9}) can also

2274: be written as (\ref{3.999}). The function (\ref{3.999}) 

2275: of the variables $x_1\lcd x_{\ts\lt+l}$ takes its values

2276: in the algebra of linear operators on $\Wll\,$.

2277: This function can be obtained by applying the conjugation

2278: $B\mapsto\Bt$ to the values of the function

2279: $$

2280: \prod_{1\le i<j\le\lt}^{\longleftarrow}\,

2281: R_{\ts ji}(\ts x_j\com x_i\ts)

2282: \ts\ \cdot\ns

2283: \prod_{1\le i\le\lt}^{\longleftarrow}\ \biggl(\ 

2284: \prod_{\lt<j\le\lt+l}^{\longleftarrow}\,

2285: R_{\ts ij}(\ts x_i\com-\ts x_j\ns-\!L\ts)\biggr)

2286: $$

2287: \begin{equation}\label{3.99}

2288: \times\!

2289: \prod_{\lt<i<j\le\lt+l}^{\longrightarrow}\!\ns

2290: R_{\ts ij}(\ts x_i\com x_j\ts)\ .

2291: \end{equation}

2292: In the entire product (\ref{3.99}), every factor is defined by

2293: (\ref{3.45}) where $P_{ij}$ is the linear operator on $W_{\ts\lt+l}$

2294: exchanging the $i\ts$th and $j\ts$th copies of $\CC^L$.

2295: Using the Yang-Baxter relation (\ref{3.5}) repeatedly, the

2296: product displayed in the second line of (\ref{3.99})

2297: can be rewritten as 

2298: $$

2299: \prod_{\lt<i<j\le\lt+l}^{\longleftarrow}\!\ns

2300: R_{\ts ji}(\ts-\ts x_j\ns-\!L\com-\ts x_i\ns-\!L\ts)\ .

2301: $$

2302: 

2303: For the current proof only, let us reorder the

2304: sequence of indices (\ref{ind}) as 

2305: $$

2306: \lt\com\lt-\ns1\lcd1\com\lt+l\lcd\lt+2\com\lt+1\,.

2307: $$

2308: We will employ the symbol $\prec$ when referring to this reordering.

2309: Then order lexicographically the set of pairs of indices $(i\com j)$ 

2310: with $i\prec j$ from the latter sequence; the lexicographical

2311: ordering is taken with respect to $\prec$ here. The entire product

2312: (\ref{3.99}) can then be rewritten as

2313: \begin{equation}\label{product}

2314: \prod_{\lt\ts\preccurlyeq\ts i\ts\prec\ts 

2315: j\ts\preccurlyeq\ts\lt+1}^{\longrightarrow}\hskip-6pt

2316: R_{\ts ij}(\ts y_i\com y_j\ts)

2317: \end{equation}

2318: where

2319: $$

2320: y_i\ts=\,

2321: \left\{

2322: \begin{array}{ll}

2323: x_i

2324: &\ \ \textrm{if}\ \ \ i=1\lcd\lt\,;\\[2pt]

2325: -x_i\!-\!L

2326: &\ \ \textrm{if}\ \ \ i=\lt+1\lcd\lt+l\,.

2327: \end{array}

2328: \right.

2329: $$

2330: 

2331: The ordered product (\ref{product}) can be expanded as a linear combination

2332: of the linear operators $P_s$ on $W_{\lt+l}$ corresponding

2333: to permutations $s\in S_{\lt+l}\,$, with the coefficients

2334: from the field $\CC(\ts y_1\lcd y_{\ts \lt+l}\ts)\ts$.

2335: By the definition (\ref{ave}), we have 

2336: $\Pt_s{\hskip-4.5pt}^\vee=0$ unless the permutation $s$ preserves each of

2337: the two subsets

2338: $$

2339: \{\nt+1\lcd\lt\}

2340: \ts\com\ts

2341: \{\lt+1\lcd\lt+m\}

2342: \subset

2343: \{1\lcd\lt+l\}\,.

2344: $$ 

2345: Using Proposition 2.3 twice, we now establish that the image of

2346: (\ref{product}) under the mapping $B\mapsto\Bt{}^\vee$

2347: coincides with the image of

2348: $$

2349: \prod_{\lt\preccurlyeq i\ts\prec j\prec\nt}^{\longrightarrow}

2350: R_{\ts ij}(\ts y_i\com y_j\ts)

2351: \ \ts\cdot\hskip-8pt

2352: \prod_{\nt\preccurlyeq i\prec j\prec\lt+m}^{\longrightarrow}\hskip-6pt

2353: R_{\ts ij}(\ts y_i\com y_j\ts)

2354: \ \ts\cdot\hskip-12pt

2355: \prod_{\lt+m\preccurlyeq i\prec j\preccurlyeq\lt+1}^{\longrightarrow}

2356: \hskip-8pt

2357: R_{\ts ij}(\ts y_i\com y_j\ts)\,.

2358: $$

2359: The latter image is easy to describe, because the three ordered products~over

2360: $$

2361: \lt\preccurlyeq i\ts\prec j\prec\nt\,,\ \  

2362: \nt\preccurlyeq i\prec j\prec\lt+m\ \ \text{and}\ \   

2363: \lt+m\preccurlyeq i\prec j\preccurlyeq\lt+1

2364: $$

2365: in the above display pairwise commute, and the image of

2366: the entire displayed product is multiplicative relative to

2367: the factorization into these three. Thus the image of

2368: (\ref{product}) under $B\mapsto\Bt{}^\vee$ equals

2369: $$

2370: \prod_{\nt\preccurlyeq i\preccurlyeq1}^{\longrightarrow}\ \biggl(\ 

2371: \prod_{\lt+l\preccurlyeq j\prec\lt+m}^{\longleftarrow}\,

2372: \Rbp_{\ts ij}(\ts x_i\com x_j\ts )\biggr)

2373: \ \cdot\hskip-2pt

2374: \prod_{\nt\preccurlyeq i\prec j\preccurlyeq1}^{\longleftarrow}

2375: \Rp_{\ts ji}(\ts x_i\com x_j\ts)

2376: \,\ \cdot\hskip-12pt

2377: \prod_{\lt+l\preccurlyeq i\prec j\prec\lt+m}^{\longrightarrow}\hskip-8pt

2378: \Rp_{\ts ji}(\ts x_j\com x_i\ts)

2379: $$

2380: $$

2381: \times

2382: \prod_{\lt\preccurlyeq i\ts\prec j\prec\nt}^{\longleftarrow}

2383: \Rp_{\ts ij}(\ts x_i\com x_j\ts)

2384: \ \ts\cdot\hskip-8pt

2385: \prod_{\lt+m\preccurlyeq i\prec j\preccurlyeq\lt+1}^{\longrightarrow}

2386: \hskip-8pt

2387: \Rp_{\ts ji}(\ts x_j\com x_i\ts)\ \,=

2388: $$

2389: $$

2390: \prod_{1\le i<j\le\nt}^{\longrightarrow}

2391: \Rp_{\ts ji}(\ts x_j\com x_i\ts)

2392: \ \ts\cdot\,\,

2393: \prod_{1\le i\le\nt}^{\longrightarrow}\ \biggl(\ 

2394: \prod_{\lt+m<j\le\lt+l}^{\longrightarrow}\hskip-3pt

2395: \Rbp_{\ts ij}(\ts x_i\com x_j\ts )\biggr)

2396: \ \ts\cdot\hskip-11pt

2397: \prod_{\lt+m<i<j\le\lt+l}^{\longrightarrow}\hskip-9pt

2398: \Rp_{\ts ij}(\ts x_i\com x_j\ts)

2399: $$

2400: $$

2401: \times

2402: \prod_{\nt<i<j\le\lt}^{\longrightarrow}

2403: \Rp_{\ts ji}(\ts x_j\com x_i\ts)

2404: \,\ \cdot\hskip-7pt

2405: \prod_{\lt<i<j\le\lt+m}^{\longrightarrow}\hskip-5pt

2406: \Rp_{\ts ij}(\ts x_i\com x_j\ts)\,\ts;

2407: $$

2408: here we used the analogues of the relations (\ref{3.5}) and (\ref{3.85}),

2409: for the operators $\Rp_{\ts ij}(\ts x\com y\ts)$ and

2410: $\Rbp_{\ts ij}(\ts x\com y\ts)$ instead of

2411: of $R_{\ts ij}(\ts x\com y\ts)$ and $\Rb_{\ts ij}(\ts x\com y\ts)$.

2412: At the right hand side of these equalities, we already employed

2413: the natural ordering of indices (\ref{ind}), and the usual

2414: lexicographical ordering of pairs of indices~\qed

2415: \end{proof}

2416: 

2417: \noindent

2418: Let us now make the substitutions (\ref{se}),(\ref{set}) and then the

2419: substitutions (\ref{or}),(\ref{ort}) in Proposition 3.5. Note that 

2420: $$

2421: c_k(\La)\ts=\,

2422: \left\{

2423: \begin{array}{ll}

2424: c_k(\ts\Up)

2425: &\ \ \textrm{if}\ \ \ k\le m\ts,\\[2pt]

2426: c_{k-m}(\Om)

2427: &\ \ \textrm{if}\ \ \ k>m

2428: \end{array}

2429: \right.

2430: $$

2431: and

2432: $$

2433: c_k(\Lat)\ts=\,

2434: \left\{

2435: \begin{array}{ll}

2436: c_k(\ts\Upt)

2437: &\ \ \textrm{if}\ \ \ k\le\mt\ts,\\[2pt]

2438: c_{k-\mt}(\Omt)

2439: &\ \ \textrm{if}\ \ \ k>\mt\,.

2440: \end{array}

2441: \right.

2442: $$

2443: After the substitutions, the rational function of

2444: $x_{\ts\nt+1}\lcd x_{\ts\lt+m}$

2445: displayed in the second line line of (\ref{3.43}) takes the operator value

2446: $$

2447: 1\ot((\,P_{\ts\mt}\,F_{\ts\Upt}\,P_{\ts\mt})\ot F_{\ts\Up})\,|\,\Wmmo\ts)

2448: \ot1\,,

2449: $$

2450: where $F_{\ts\Up}$ and $F_{\ts\Upt}$ are operators on respectively

2451: $W_m$ and $W_{\ts\mt}^{\ts\ast}$, while

2452: $P_{\ts\mt}$ is the operator on 

2453: $W_{\mt\ts}^{\ts\ast}$ reversing the order

2454: of tensor factors. Here we applied 

2455: Proposition 2.2 to the (non-skew) standard tableaux $\Up$

2456: and $\Upt$ instead of $\La\ts$.

2457: 

2458: Now consider the operator $\FOm$ on $\Wnn$ as defined in Subsection 1.3.

2459: %see also Subsection 1.2.

2460: The product displayed in the first line of (\ref{3.43}) can be rewritten as

2461: $$

2462: \prod_{1\le i\le\nt}^{\longleftarrow}\ \biggl(\ 

2463: \prod_{\lt+m<j\le\lt+l}^{\longrightarrow}\,

2464: \Rbp_{\ts ij}(\ts x_i\com\ts x_j\ts)\biggr)

2465: $$

2466: $$

2467: \times\ 

2468: \prod_{1\le i<j\le\nt}^{\longrightarrow}\,

2469: \Rp_{\ts ji}(\ts x_j\com x_i\ts)

2470: \ts\ \cdot\hskip-16pt

2471: \prod_{\lt+m<i<j\le\lt+l}^{\longrightarrow}\!\ns

2472: \Rp_{\ts ij}(\ts x_i\com x_j\ts)\ ;

2473: $$

2474: see Subsection 3.2. By using Proposition 3.3 and the definition of

2475: $\Th_{m\mt}\ts$, after

2476: the above substitutions in Proposition 3.5, we obtain the equality

2477: \begin{equation}\label{3.4444}

2478: \Th_{m\mt}(\FLa)=

2479: \bigl(

2480: (\ts P_{\ts\mt}\,F_{\ts\Upt}\,P_{\ts\mt})\ot F_{\ts\Up})\,|\,\Wmmo\ts

2481: \bigr)

2482: \ot\FOm\,.

2483: \end{equation}

2484: The linear operator $\FLa$ on $\Wll$ has been defined as

2485: either of the two (equal) products displayed in Proposition 3.3.

2486: When $m+\mt>0$, the equality (\ref{3.4444}) can be

2487: regarded as an alternative definition of the operator $\FOm$ on $\Wnn\ts$.

2488: 

2489: %------------------------------------------------------------------------------

2490: 

2491: \bigskip\noindent\textbf{3.6.}

2492: In this subsection, we need to indicate the basic vector 

2493: spaces $\CC^N\com\CC^L$ and $\CC^M$ corresponding to the tensor products

2494: (\ref{mix}),(\ref{3.00}) and (\ref{3.000}). For this reason,

2495: we will employ respectively

2496: the symbols $\Wnn\ts(N)\com\Wll\ts(L)$ and $\Wmm\ts(M)$ 

2497: to denote these tensor products. In the same style, we will write

2498: $W_n(N)\com W_{\nt}^{\ts\ast}(N)$ and $\Wnno\ts(N)\ts$.

2499: The linear operator (\ref{1.45}) on $(\CC^N)^\ast\ot\CC^N$

2500: will be denoted by $Q(N)\ts$. Following this style of notation

2501: throughout the article would make our formulas rather cumbersome,

2502: so we confine this style to the present subsection.  

2503: Recall that here we have $M=L-N$.

2504: 

2505: The image of the operator $\FOm$ on the vector space

2506: $\Wnn\ts(N)$ has been denoted by $\VOm\,$. 

2507: If $M=0$ then $\mu=\mut=(0\com0\ts,\ts\ldots\ts)\ts$ and

2508: $\Om=\La\com\Omt=\Lat\ts$.

2509: In this special case, we have the equality (\ref{1.4444}) by Proposition 3.3.

2510: The next proposition holds for arbitrary $M\ge0\ts$.

2511: 

2512: \begin{Proposition}

2513: If\/ $\Vlm\neq\{0\}$, then $\VOm\neq\{0\}$.

2514: \end{Proposition}

2515: 

2516: \begin{proof}

2517: Let us realize the irreducible

2518: representation $\Vla$ of the group $GL_L$

2519: from (\ref{1.0}) as the image $\VLa\subset\Wll\ts(L)$

2520: of the operator $\FLa\,$. Choose a basis $v_1\lcd v_L$ in $\CC^L$ such

2521: that the subspaces $\CC^N$ and $\CC^M$ of $\CC^L$ are spanned

2522: respectively by $v_1\lcd v_N$ and $v_{N+1}\lcd v_L\,$.

2523: Let $v_1^\ast\lcd v_L^\ast$ be the dual basis in $(\CC^L)^\ast$.

2524: For any non-negative integers $k,\kt$ and $t$ such that $k+t\le l$ and

2525: $\kt+t\le\lt\ts$, denote by $Z_{k\kt}^{\ts(t)}$ the subspace of 

2526: $\Wll\ts(L)$ spanned by the vectors

2527: $$

2528: \sum_{N<a_1,\ldots,\ts a_t\le L}

2529: u\ot 

2530: v_{a_t}^{\ts\ast}\ot\ldots\ot v_{a_1}^{\ts\ast}\ot

2531: w\ot

2532: v_{a_1}\ot\ldots\ot v_{a_t}\ot 

2533: v\,,

2534: $$

2535: for all $u\in W_{\ts\lt-\kt-t\ts}^{\ts\ast}(N)\ts$,

2536: all $v\in W_{\ts l-k-t\ts}(N)\ts$, 

2537: and all $w\in W_{\ts k\kt\ts}^{\ts\circ}(M)\ts$.

2538: 

2539: The vector space $\Wll\ts(L)$

2540: is the sum of its subspaces, obtained from all the subspaces

2541: of the form $Z_{k\kt}^{\ts(t)}$ 

2542: by the permutational operators $P_s\ts$

2543: for some $s\in\Sll\ts$. 

2544: This sum may be not direct, cf.\ \cite[Section V.6]{W}.

2545: However, every summand is preserved by 

2546: the action of the subgroup $GL_M\subset GL_L\ts$.

2547: 

2548: For instance, consider the subspace $Z_{k\kt}^{\ts(t)}$ itself. 

2549: Let $\bar{Z}_{k\kt}^{\ts(t)}$ be the subspace of $\Wll\ts(L)$ 

2550: spanned, for the same $u,v$ and $w$ as above, by the vectors

2551: $$

2552: \sum_{1\le a_1,\ldots,\ts a_t\le N}

2553: u\ot 

2554: v_{a_t}^{\ts\ast}\ot\ldots\ot v_{a_1}^{\ts\ast}\ot

2555: w\ot

2556: v_{a_1}\ot\ldots\ot v_{a_t}\ot 

2557: v\,.

2558: $$

2559: Then

2560: 

2561: \vskip-20pt

2562: $$

2563: \FLa\cdot Z_{k\kt}^{\ts(t)}\ts=\ts\FLa\cdot\bar{Z}_{k\kt}^{\ts(t)}\,\ts;

2564: $$

2565: 

2566: \smallskip\noindent

2567: this observation follows from from the equalities 

2568: $\FLa\,Q_{ij}=0$ for the pairs

2569: \begin{equation}\label{3.54}

2570: (i\com j)=

2571: (\ts\lt-\kt\com\lt+k+1)

2572: \lcd

2573: (\ts\lt-\kt-t+1\com\lt+k+t)\,;

2574: \end{equation}

2575: here $Q_{ij}$ are operators in $\Wll\ts(L)\ts$.

2576: These operator equalities are implied by

2577: Proposition 3.4 and Corollary~3.3. 

2578: Note that $\bar{Z}_{k\kt}^{\ts(t)}\subset Z_{k\kt}^{\ts(0)}\ts$. 

2579: Also note that

2580: $$

2581: {\rm Hom}_{\,GL_M}

2582: \bigl(\,V_{\ts\mu\mut}\ts\com\FLa\cdot Z_{k\kt}^{\ts(0)}\,\bigr)

2583: \neq\{0\}

2584: \ \ \Rightarrow\ \ k=m\hskip.5pt\com\kt=\mt\,.

2585: $$

2586: The subspace $Z_{m\mt}^{\ts(0)}\subset\Wll\ts(L)$

2587: coincides with the subspace $\Zmm\ts$, see (\ref{3.41}). 

2588: 

2589: By definition, the subspace $Z_{k\kt}^{\ts(t)}$ is contained in the 

2590: direct summand

2591: $$

2592: W_{\ts\lt-\kt-t\ts}^{\ts\ast}(N)\ot

2593: W_{\ts k+t\ts,\ts\kt+t\ts}(M)\ot

2594: W_{\ts l-k-t\ts}(N)\,\subset\,\Wll\ts(L)\ts.

2595: $$

2596: Let $J_{k+t,\kt+t}$ be the projector in $\Wll\ts(L)$ to this direct summand.

2597: Let $I_{k\kt}^{\ts(t)}$ be the operator on the tensor product

2598: $W_{k+t\ts,\ts\kt+t}(M)\ts$, acting as

2599: $Q(M)\ns/M$ on the $i\ts$th and $j\ts$th

2600: tensor factors for every pair of indices

2601: $$

2602: (i\com j)=

2603: (\ts t\com\kt+k+t+1)

2604: \lcd

2605: (1\com\kt+k+2t)\,,

2606: $$

2607: and acting as

2608: 

2609: \vskip-16pt

2610: $$

2611: I_{k\kt}:\,\Wkk\ts(M)\,\to\,\Wkko\ts(M)

2612: \nopagebreak

2613: $$

2614: on the remaining $\kt+k$ tensor factors of $W_{k+t\ts,\ts\kt+t}(M)\ts$;

2615: see Subsection 3.5.

2616: 

2617: Let $H_{k\kt}^{(t)}$ be

2618: the composition of operators $1\ot I_{k\kt}^{(t)}\ot1$

2619: and $J_{k+t,\kt+t}$ on $\Wll\ts(L)\ts$, 

2620: this composition is a $GL_M$-equivariant

2621: projector to the subspace~$Z_{k\kt}^{\ts(t)}$.

2622: In a similar way, by using $Q(N)\ns/N$ instead of $Q(M)\ns/M$, define

2623: a projector 

2624: $$

2625: \bar{H}_{k\kt}^{\ts(t)}:\,\Wll(L)\,\to\,\bar{Z}_{k\kt}^{\ts(t)}\,.

2626: $$

2627: Using the notation of Subsection 3.5, we have the equality

2628: $H_{k\kt}^{(0)}=H_{m\mt}\ts$.

2629: 

2630: If an irreducible representation of

2631: $GL_M$ equivalent to $V_{\ts\mu\mut}$ occurs in the projection

2632: $H_{k\kt}^{(t)}\cdot\VLa\ts$, 

2633: it also occurs in the projection 

2634: $\bar{H}_{k\kt}^{(t)}\cdot\VLa\ts$.

2635: This follows from the equalities 

2636: $Q_{ij}\ts F_\La=0$ for the pairs of indices $i$ and $j$ given by 

2637: (\ref{3.54}). Either occurence implies that $k=m$.

2638: But again, we have

2639: $$

2640: \bar{H}_{m\mt}^{(t)}\cdot\VLa\,\subset\,

2641: H_{m\mt}^{(0)}\cdot\VLa\,=\,

2642: H_{m\mt}\cdot\VLa\,.

2643: $$

2644: 

2645: Now suppose that $\Vlm\neq\{0\}$. 

2646: By the above argument, then

2647: there exist a pair of permutations $s^{\ts\prime}$ and $s^{\ts\prime\prime}$ 

2648: in $\Sll$ such that

2649: \begin{equation}\label{3.56}

2650: {\rm Hom}_{\,GL_M}(\,V_{\ts\mu\mut}\ts\com\ts 

2651: H_{m\mt}\,P_{s'}\,\FLa\,P_{s''}\cdot\Zmm\ts)\,\neq\,\{0\}\,.

2652: \end{equation}

2653: 

2654: Let $\Lap,\Lapp$ and $\Latp,\Latpp$ range over the sets of

2655: standard tableaux of shapes $\la$ and $\lat$ respectively. Let 

2656: 

2657: \vskip-20pt

2658: \begin{equation}\label{FLL}

2659: F_{\La''\ns\Lat''}^{\ts\La'\Lat'}\,\in\,\Bll

2660: \end{equation}

2661: be the unique element of the simple ideal

2662: $\Bla\subset\Bll\,$ with the image under the homomorphism 

2663: {\rm(\ref{3.22})} corresponding to the element 

2664: $$

2665: (\ts s_0\ts f_{\Lat''}^{\ts\Lat'}\ts s_0)

2666: \ts\cdot\ts

2667: \io_{\lt}\ts(\ts f_{\La''}^{\ts\La'})\,\in\,\CC\Sll\,;

2668: $$

2669: see (\ref{fll}). If $\Lap=\Lapp$ and $\Latp=\Latpp$, 

2670: then we have the equality

2671: $$

2672: F_{\La''\ns\Lat''}^{\ts\La'\Lat'}\,=\,F_{\La'\ns\Lat'}

2673: $$

2674: due to Corollary 3.4. 

2675: The product $P_{s'}\,\FLa\,P_{s''}$ in (\ref{3.56}) can be 

2676: written as a linear combination of the operators (\ref{FLL}), 

2677: with the coefficients from $\CC\ts$. By (\ref{3.56}), 

2678: there exists standard tableaux $\Lap,\Lapp$ and $\Latp,\Latpp$ such that

2679: \begin{equation}\label{3.57}

2680: {\rm Hom}_{\,GL_M}(\,V_{\ts\mu\mut}\ts\com\ts 

2681: H_{m\mt}\,F_{\La''\ns\Lat''}^{\ts\La'\Lat'}\cdot\Zmm\ts)\,\neq\,\{0\}\,.

2682: \end{equation}

2683: 

2684: By definition, the operator (\ref{FLL}) is divisible by

2685: $(\ts P_{\ts\lt}\,F_{\Lat'}\ts P_{\ts\lt}\ts)\ot F_{\La'}$ 

2686: on the left, and by

2687: $(\ts P_{\ts\lt}\,F_{\Lat''}\ts P_{\ts\lt}\ts)\ot F_{\La''}$

2688: on the right.

2689: Now consider the tableaux $\Up^{\ts\prime}$ and $\Up^{\ts\prime\prime}$,

2690: obtained by removing the boxes with the numbers $m+1\lcd l$ from the 

2691: tableaux $\Lap$ and $\Lapp$ respectively.

2692: Also consider the

2693: the tableaux $\Upt^{\ts\prime}$ and $\Upt^{\ts\prime\prime}$,

2694: obtained by removing the boxes with the numbers $\mt+1\lcd\lt$ from the 

2695: tableaux $\Latp$ and $\Latpp$. Then

2696: the restriction of the operator

2697: $$

2698: H_{m\mt}\,F_{\La''\ns\Lat''}^{\ts\La'\Lat'}

2699: $$

2700: to the subspace $\Zmm\subset\Wll(L)$ is divisible on the left by 

2701: $$

2702: 1\ot((\,P_{\ts\mt}\,F_{\ts\Upt'}\,P_{\ts\mt})\ot F_{\ts\Up'})

2703: \,|\,\Wmmo(M))

2704: \ot1\,,

2705: $$

2706: cf.\ the proof of Lemma 2.3. Similarly, it%this restriction

2707: is divisible on the right by 

2708: $$

2709: 1\ot((\,P_{\ts\mt}\,F_{\ts\Upt''}\,P_{\ts\mt})\ot F_{\ts\Up''})

2710: \,|\,\Wmmo(M))

2711: \ot1\,.

2712: $$

2713: Now the inequality (\ref{3.57}) implies that the shapes of the tableaux

2714: $\Up^{\ts\prime}$ and $\Up^{\ts\prime\prime}$ coincide.

2715: The same inequality implies that the shapes of

2716: $\Upt^{\ts\prime}$ and $\Upt^{\ts\prime\prime}$ coincide as well.

2717: Then 

2718: $$

2719: F_{\La''\ns\Lat''}^{\ts\La'\Lat'}\ts=\,F_{\ts\La'\ns\Lat'}\,F

2720: $$

2721: for the operator $F$ on $\Zmm$ corresponding to

2722: some invertible element $f$ of the group ring of

2723: $$

2724: S_{\ts\nt\mt}\times\io_{\lt}\ts(S_{mn})\,\subset\,S_{\lt+l}\,,

2725: $$

2726: see (\ref{2.3}). Then

2727: $$

2728: F_{\La''\ns\Lat''}^{\ts\La'\Lat'}\cdot\Zmm\,=\,F_{\ts\La'\ns\Lat'}\cdot\Zmm\,.

2729: $$

2730: By applying now the relation (\ref{3.4444}) to the tableaux $\Lap$ and

2731: $\Latp$ instead of $\La\ts$ and $\Lat\ts$, the inequality (\ref{3.57})

2732: implies that the tableaux $\Up^{\ts\prime}$ and $\Upt^{\ts\prime}$

2733: are of shapes $\mu$ and $\mut$ respectively.

2734: Moreover, the left hand side of (\ref{3.57}) then

2735: equals $V_{\Om'\ns\Omt'}$ for some standard tableaux $\Om^{\ts\prime}$

2736: and $\Omt^{\ts\prime}$ of skew shapes $\lm$ and $\lmt$ respectively.

2737: The inequality $V_{\Om'\ns\Omt'}\neq\{0\}$ implies that $\VOm\neq\{0\}$\qed

2738: \end{proof}

2739: 

2740: \noindent

2741: The vector space $\Vlm$ as defined by (\ref{1.0}),

2742: comes with a natural action of the subgroup $GL_N\subset GL_L\ts$.

2743: On the other hand,

2744: the subspace $\VOm\subset\Wnn(N)$ is preserved by the action of $GL_N\ts$, 

2745: because $\FOm\in\Bnn\ts$. Let us consider $\Vlm$ and $\VOm$ as

2746: representations of the group $GL_N$. In Subsection 5.4

2747: we will prove that

2748: these representations are equivalent, this is what Proposition~1.3 states.

2749: We will use Proposition~3.6 along with

2750: an explicit description of the action of the subalgebra

2751: $\AMN\subset\UMN\ts$ on the vector space $\VOm\,$.

2752: 

2753: %==============================================================================

2754: 

2755: \section{\hskip-3.5pt. Yangian representations}

2756: 

2757: %------------------------------------------------------------------------------

2758: 

2759: \noindent\textbf{4.1.}

2760: In this section we will work with the space of mixed tensors (\ref{mix}).

2761: Thus our basic vector space will be $\CC^N$ like it was in Section 1,

2762: and unlike it was in Section 3.  

2763: We will also use the tensor product of vector spaces

2764: \begin{equation}\label{aux}

2765: \CC^N\ns\ot\Wnn\ts=\ts

2766: \CC^N\ns\ot(\CC^N)^\ast\ot\ldots\ot(\CC^N)^\ast\ot\CC^N\ns\ot\ldots\ot\CC^N

2767: \end{equation}

2768: where the first tensor factor $\CC^N$ shall not be regarded

2769: as a representation of the group $GL_N$, it will play only an auxiliary role.

2770: The $1\ns+\ts\nt+n\ts$ tensor factors in (\ref{aux}) will be

2771: labeled by the indices

2772: $$

2773: 0\com1\lcd\nt\com\nt+\ns1\lcd\nt+n

2774: $$

2775: respectively from the left to right. 

2776: For two distinct indices $i$ and $j$ such that

2777: $$

2778: 0<i\com j\le\nt

2779: \quad\text{or}\quad

2780: \nt<i\com j

2781: \quad\text{or}\quad

2782: i=0

2783: \quad\text{and}\quad  

2784: \nt<j

2785: $$

2786: let $P_{ij}$ be the linear operator on (\ref{aux}), exchanging

2787: the $i\ts$th and $j\ts$th tensor factors. Then determine 

2788: the function $R_{ij}(x,y)$ by the same equality (\ref{3.45}) as before.

2789: But now $R_{ij}(x,y)$ takes values in the algebra of operators on

2790: (\ref{aux}).

2791: 

2792: Further, for any two indices

2793: $i$~and~$j$ such that $0<i\le\nt<j\ts$, let

2794: $Q_{ij}$ the operator on (\ref{aux}) acting as (\ref{1.45})

2795: on the $i\ts$th and $j\ts$th tensor factors,

2796: and acting as the identity on remaining 

2797: $\nt\ns+n\ns-\!1\ts$ factors. For such $i$ and $j$ put

2798: \begin{equation}\label{instead}

2799: \Rb_{ij}(x\com y)\,=\,1-\frac{Q_{ij}}{x+y+N}\,\ts;

2800: \end{equation}

2801: note the difference between the equality displayed above, 

2802: and the second equality in (\ref{3.555}). 

2803: Furthermore, for any index $i$ such that $0<i\le\nt\ts$, put

2804: $$

2805: \Ra_{\ts0j}(x,y)=

2806: \biggl(1+\frac{Q_{0i}}{x-y}\ts\biggr)

2807: \cdot

2808: \biggl(1-\frac1{(x-y)^{\ts2}}\biggr)^{\!-1}

2809: $$

2810: where $Q_{0i}$ is the linear 

2811: operator on (\ref{aux}) acting as (\ref{1.45})

2812: on the $i\ts$th and $0\ts$th tensor factors,

2813: and acting as the identity on remaining 

2814: $\nt\ns+n\ns-\!1\ts$ factors.

2815: 

2816: Note that for all possible indices $i\com j>0$ we have the relations

2817: \begin{equation}\label{4.001}

2818: R_{\ts0i}(x\com y)\,R_{\ts0j}(x\com z)\,R_{\ts ij}(y\com z)=

2819: R_{\ts ij}(y\com z)\,R_{\ts0j}(x\com z)\,R_{\ts0i}(x\com y)\,,

2820: \end{equation}

2821: \begin{equation}\label{4.002}

2822: \Ra_{\ts0i}(x\com y)\,\Ra_{\ts0j}(x\com z)\,R_{\ts ij}(y\com z)=

2823: R_{\ts ij}(y\com z)\,\Ra_{\ts0j}(x\com z)\,\Ra_{\ts0i}(x\com y)\,,

2824: \end{equation}

2825: \begin{equation}\label{4.003}

2826: \Ra_{\ts0i}(x\com-y)\,R_{\ts0j}(x\com z)\,\Rb_{\ts ij}(y\com z)\,=\,

2827: \Rb_{\ts ij}(y\com z)\,R_{\ts 0j}(x\com z)\,\Ra_{\ts 0i}(x\com-y)\,.

2828: \end{equation}

2829: Here (\ref{4.001}) is a Yang-Baxter relation, while

2830: (\ref{4.002}) and (\ref{4.003}) are derived from Yang-Baxter

2831: relations, by using (\ref{3.55}) and (\ref{3.6}) where $L$ is replaced by $N$. 

2832: 

2833: For any $y_1\lcd y_{\ts\nt}\in\CC$ and $z_1\lcd z_n\in\CC$ 

2834: consider the tensor product 

2835: \begin{equation}\label{mod}

2836: V(y_{\ts\nt})^\ast\ns\ot\ldots\ot V(y_1)^\ast\ns\ot 

2837: V(z_1)\ot\ldots\ot V(z_n)

2838: \end{equation}

2839: of evaluation and dual evaluation $\YN\ts$-modules, these have

2840: been defined in Subsection~1.4. The underlying space of

2841: the $\YN\ts$-module (\ref{mod}) is $\Wnn\,$. Denote by $\rho_{n\nt}$ 

2842: the homomorphism $\YN\to\End(\Wnn)$ corresponding to (\ref{mod}).

2843: Consider the element (\ref{1.71}). In what follows, we identify the algebras

2844: \begin{equation}\label{twoalg}

2845: \End(\CC^N)\ot\End(\Wnn)=

2846: \End(\ts\CC^N\ns\ot\Wnn)\,.

2847: \end{equation}

2848: 

2849: \begin{Proposition}

2850: Under the homomorphism

2851: $$%\begin{equation}\label{idrhozz}  

2852: \id\ot\rho_{n\nt}:\,

2853: \End(\CC^N)\ot\YN

2854: \,\to\,

2855: \End(\ts\CC^N\ns\ot\Wnn)\,,

2856: $$%\end{equation}

2857: $$%\begin{equation}\label{4.2}

2858: T(x)\,\mapsto\,

2859: \Ra_{\ts01}(x\com y_{\ts\nt})\ts\ldots\ts\Ra_{\ts0\nt}(x\com y_1)\,

2860: R_{\ts0,\nt+1}(x\com z_1)\ts\ldots\ts R_{\ts0,\nt+n}(x\com z_n)\,.

2861: $$%\end{equation}

2862: \end{Proposition}

2863: 

2864: \begin{proof}

2865: Firstly, consider the case when $n=1$ and $\nt=0$. Using the

2866: matrix units $E_{ab}\in\End(\CC^N)$, the element

2867: $P_{\ts01}\in\End(\ts\CC^N\ns\ot\CC^N)$ expands as 

2868: $$

2869: P_{\ts01}\,=

2870: \sum_{a,b=1}^N\, E_{ab}\ot E_{ba}\,. 

2871: $$

2872: Using definitions (\ref{1.52}) and (\ref{tau}), we prove that 

2873: under the homomorphism 

2874: $$

2875: \id\ot\rho_{10}:\,

2876: \End(\CC^N)\ot\YN

2877: \,\to\,

2878: \End(\ts\CC^N\ns\ot\CC^N)\,,

2879: $$

2880: we have $T(x)\mapsto R_{\ts01}(x\com z_1)$ as required.

2881: 

2882: Secondly, consider the case when $n=0$ and $\nt=1$.

2883: The operator $Q_{01}$ on $\CC^N\ns\ot(\CC^N)^\ast$ is obtained

2884: from the operator $P_{\ts01}$ on $\CC^N\ns\ot\CC^N$ by conjugation in the

2885: second tensor factor. Using our result from the previous case and

2886: the relation (\ref{3.55}), we prove that

2887: $T(x)\mapsto\Ra_{\ts01}(x\com y_1)$ 

2888: under the homomorphism 

2889: $$

2890: \id\ot\rho_{01}:\,

2891: \End(\CC^N)\ot\YN

2892: \,\to\,

2893: \End(\ts\CC^N\ns\ot(\CC^N)^\ast)\,.

2894: $$

2895: For arbitrary $n$ and $\nt$, Proposition~4.1 follows from its

2896: two particular cases considered here, and from the definition 

2897: (\ref{1.33}) of coproduct on $\YN$

2898: \qed

2899: \end{proof}

2900: 

2901: \noindent

2902: Consider also the product of evaluation and

2903: dual evaluation $\YN\ts$-modules

2904: \begin{equation}\label{modrev}

2905: V(z_n)\ot\ldots\ot V(z_1)\ns\ot

2906: V(y_1)^\ast\ns\ot\ldots\ot V(y_{\ts\nt})^\ast\,.

2907: \end{equation}

2908: Here the ordering of the tensor factors is opposite

2909: to that used in (\ref{mod}).

2910: The underlying vector space of the $\YN\ts$-module (\ref{modrev})

2911: can be identified with that of the $\YN\ts$-module (\ref{mod}) 

2912: by using the permutational operator $(\ref{Pnn})$. Let 

2913: $\sigma_{n\nt}:\YN\to\End(\Wnn)$ be the composition of

2914: the homomorphism corresponding to the $\YN\ts$-module (\ref{modrev}), with

2915: the conjugation by $(\ref{Pnn})$.

2916: Our proof of Proposition 4.1 then implies the following result.

2917: 

2918: \begin{Corollary}

2919: Under the homomorphism

2920: $$

2921: \id\ot\si_{n\nt}:\,

2922: \End(\CC^N)\ot\YN

2923: \,\to\,

2924: \End(\ts\CC^N\ns\ot\Wnn)\,,

2925: $$

2926: $$

2927: T(x)\,\mapsto\,

2928: R_{\ts0,\nt+n}(x\com z_n)\ts\ldots\ts R_{\ts0,\nt+1}(x\com z_1)\,

2929: \Ra_{\ts0\nt}(x\com y_1)\ts\ldots\ts\Ra_{\ts01}(x\com y_{\ts\nt})\,.

2930: $$

2931: \end{Corollary}

2932: 

2933: Now consider the standard action of the enveloping algebra

2934: $\UN$ on the space of mixed tensors $\Wnn\,$.

2935: Let us denote by $\varpi_{\ts n\nt}$ 

2936: the corresponding homomorphism $\UN\to\End(\Wnn)\,$.

2937: Using definitions (\ref{1.71})~and~(\ref{1.52}),

2938: we obtain the next result.

2939: 

2940: \begin{Lemma}

2941: Under the homomorphism

2942: $$%\begin{equation}\label{idvarpial}

2943: \id\ot(\ts\varpi_{\ts n\nt}\circ\ts\pi_N):\,\ts

2944: \End(\CC^N)\ot\YN

2945: \,\to\,

2946: \End(\ts\CC^N\ns\ot\Wnn)\,,

2947: $$%\end{equation}

2948: $$

2949: T(x)\,\mapsto\,1+

2950: \bigl(\ts 

2951: Q_{01}+\ldots+Q_{0\nt}-P_{\ts0,\nt+1}-\ldots-P_{\ts0,\nt+n}

2952: \ts\bigr)\,\ts

2953: x^{-1}\,.

2954: $$

2955: \end{Lemma}

2956: 

2957: %------------------------------------------------------------------------------

2958: 

2959: \smallskip\noindent\textbf{4.2.}

2960: Take any two standard tableaux $\Om$ and $\Omt\ts$, with respectively

2961: $n$ and $\nt$ boxes.

2962: Consider the operator $\FOm$ on the vector space $\Wnn\,$,

2963: as defined in Subsection 1.3. We denoted

2964: by $P_{\ts\nt+n}$ the operator on $\Wnn\ts$ reversing the order of

2965: all $\nt+n$ tensor factors, see (\ref{Pnn}).

2966: Let us prove Proposition~1.4.~Set

2967: \begin{equation}\label{setz}

2968: z_k=c_k(\Om)

2969: \text{\ \ for each\ \ }

2970: k=1\lcd n\ts;

2971: \end{equation}

2972: \begin{equation}\label{sety}

2973: y_k=-\ts c_k(\Omt)\ns-\!M

2974: \text{\ \ for each\ \ }

2975: k=1\lcd\nt\ts.

2976: \end{equation}

2977: In these settings, Proposition 4.1 describes 

2978: the action of the algebra~$\YN$ in the tensor product

2979: of evaluation and dual evaluation modules, displayed

2980: in \text{Proposition~1.4} at the bottom.

2981: The underlying space of the tensor product

2982: of evaluation and dual evaluation modules displayed

2983: in Proposition~1.4 at the top, can be identified with $\Wnn$

2984: by using the operator (\ref{Pnn}). 

2985: The action of $\YN$ in the resulting module is now described 

2986: by Corollary~4.1.

2987: 

2988: Now take $\nt+n$ complex variables $x_1\lcd x_{\ts\nt+n}\,$.

2989: Let $F(\ts x_1\lcd x_{\ts\nt+n})$ be the rational function of

2990: these variables, defined as the product

2991: $$

2992: \prod_{1\le i<j\le\nt}^{\longrightarrow}

2993: R_{\ts ji}(\ts x_j\com x_i\ts)

2994: \ \,\cdot\hskip-8pt

2995: \prod_{\substack{1\le i\le\nt\\\nt<j\le\nt+n}}^{\longrightarrow}

2996: \Rb_{\ts ij}(\ts x_i\com x_j\ts)

2997: \ts\ \cdot\hskip-13pt

2998: \prod_{\nt<i<j\le\nt+n}^{\longrightarrow}\hskip-6pt

2999: R_{\ts ij}(\ts x_i\com x_j\ts)\,.

3000: $$

3001: According to notation of Subsection 4.1, the values of this 

3002: function

3003: belong to the algebra (\ref{twoalg}). Using the relations (\ref{3.75}) and

3004: (\ref{4.001}) to (\ref{4.003}) repeatedly, 

3005: we obtain the equality of rational functions with values in 

3006: the algebra (\ref{twoalg}),

3007: $$

3008: \Ra_{\ts01}(\ts x\com-x_1)

3009: \ts\ldots\ts

3010: \Ra_{\ts0\nt}(\ts x\com-x_{\ts\nt})\,

3011: R_{\ts0,\nt+1}(\ts x\com x_{\ts\nt+1})

3012: \ts\ldots\ts

3013: R_{\ts0,\nt+n}(\ts x\com x_{\ts\nt+n})

3014: $$

3015: $$

3016: \times\ 

3017: F(\ts x_1\lcd x_{\ts\nt+n})

3018: \,=\,

3019: F(\ts x_1\lcd x_{\ts\nt+n})

3020: \ \times

3021: $$

3022: $$

3023: R_{\ts0,\nt+n}(\ts x\com x_{\ts\nt+n})

3024: \ts\ldots\ts

3025: R_{\ts0,\nt+1}(\ts x\com x_{\ts\nt+1})\,

3026: \Ra_{\ts0\nt}(\ts x\com-x_{\ts\nt})

3027: \ts\ldots\ts

3028: \Ra_{\ts01}(\ts x\com-x_1)\,.

3029: $$

3030: In this equality, let us substitute

3031: $$

3032: x_{\ts\nt+k}=c_{k}(\Om)+t_{k}(\Om)

3033: \text{\ \ for each\ \ $k=1\lcd n$}

3034: $$

3035: where the complex variables $t_1(\Om)\lcd t_n(\Om)$

3036: are constrained by (\ref{1.0002}). By using the variables

3037: $t_1(\Omt)\lcd t_{\ts\nt}(\Omt)$ constrained similarly, also substitute

3038: $$

3039: x_{\ts\nt-k+1}=c_{k}(\Omt)+M+t_{k}(\Omt)

3040: \text{\ \ for each\ \ $k=1\lcd\nt\ts$.}

3041: $$

3042: We then obtain an equality of rational functions of the

3043: constrained variables, and of the variable $x\ts$. 

3044: By setting in the resulting equality 

3045: $$

3046: t_1(\Om)=\ldots=t_n(\Om)=0

3047: \quad\text{and}\quad

3048: t_1(\Omt)=\ldots=t_{\ts\nt}(\Omt)=0\,,

3049: $$

3050: we obtain an equality of rational functions

3051: of the variable $x$ only. 

3052: In view of Proposition 4.1 

3053: and Corollary 4.1, 

3054: the last equality proves Proposition~1.4.

3055: Indeed, after all the above substitutions 

3056: $F(\ts x_1\lcd x_{\ts\nt+n})$ takes the value

3057: $1\ot\FOm\ts$ in the algebra (\ref{twoalg}), see 

3058: the definition (\ref{maindef}).

3059: Here we have used the equality between (\ref{3.9}) and (\ref{3.999}),

3060: established in Subsection 3.2. Note that

3061: the equality between (\ref{3.9}) and (\ref{3.9999}) now implies

3062: Lemma 1.3.

3063: 

3064: %------------------------------------------------------------------------------

3065: 

3066: \smallskip\medskip\noindent\textbf{4.3.}

3067: Consider the embedding $\UN\to\YN$ as defined by ({\ref{4.4}).

3068: By the definition (\ref{1.52}),

3069: the homomorphism $\pi_N:\YN\to\UN$ is identical on the subalgebra 

3070: $\UN\subset\YN\ts$.

3071: The automorphism $\om_N$ of $\YN$

3072: is also identical on this subalgebra, see the definition (\ref{1.51}).

3073: It follows that the restriction of $\pi_{NM}$ to the subalgebra

3074: $\UN\subset\YN$ coincides with the natural embedding $\UN\to\UMN\ts$.

3075: For the particular choice (\ref{1.62}) of the series $g(x)$,

3076: the automorphism (\ref{1.61}) is identical on 

3077: $\UN\subset\YN\ts$, because the coefficient at 

3078: $x^{-1}$ in the expansion of (\ref{1.62}) is $0\ts$.

3079: So the action of the subalgebra

3080: $\UN\subset\YN\ts$ in the $\YN\ts$-module $\Vlm$ coincides

3081: with its natural action, corresponding to the natural

3082: action of $GL_N$ in $\Vlm$.

3083: 

3084: Further, the action of the subalgebra

3085: $\UN\subset\YN\ts$ in any evaluation module $V(z)$ over $\YN\ts$,

3086: and in any dual evaluation module $V(z)^\ast$ over $\YN\ts$,

3087: coincides with the natural action of $\UN$

3088: in the underlying vector space $\CC^N$ of these modules.

3089: The assigment (\ref{4.4}) determines a Hopf algebra

3090: embedding, see (\ref{Hopfemb}).

3091: Hence the action of the subalgebra

3092: $\UN\subset\YN\ts$ in the tensor product of %evaluation and dual evaluation

3093: $\YN\ts$-modules displayed in Proposition 1.4 at the bottom,

3094: coincides with the natural action of $\UN$

3095: in the vector space $\Wnn\ts$. The $\YN\ts$-module $\VOm$ was

3096: defined as a submodule of that tensor product.

3097: Thus the action of $\UN\subset\YN\ts$ in this submodule

3098: coincides with the natural action of $\UN$ in the subspace

3099: $\VOm\subset\Wnn\ts$.

3100: 

3101: %------------------------------------------------------------------------------

3102: 

3103: \smallskip\medskip\noindent\textbf{4.4.}

3104: Take the $\YN\ts$-module (\ref{mod}) in the particular case

3105: when $\nt=0$~and 

3106: \begin{equation}\label{setext}

3107: z_k=z-k+1

3108: \text{\ \ for each\ \ }

3109: k=1\lcd n\ts.

3110: \end{equation}

3111: Here $z$ is arbitrary complex papameter. The underlying vector

3112: space of this module is the tensor product $W_n$ of $n$ copies of $\CC^N$.

3113: Let $A_n$ be the operator of antisymmetrization on $W_n\ts,$ 

3114: normalized so that $A_n^2=A_n\ts$. In the notation of

3115: Subsection 4.1 we then have the equality in the algebra (\ref{twoalg}),

3116: $$

3117: \prod_{1\le i<j\le n}^{\longrightarrow}

3118: R_{\ts ij}(\ts z_i\com z_j\ts)

3119: \ =\ n\ts!\,\cdot\,1\ot A_n\,;

3120: $$

3121: see the end of Subsection 1.2.

3122: The arguments of Subsection 4.2 now imply

3123: $$

3124: R_{\ts01}(\ts x\com z_1)

3125: \ts\ldots\ts

3126: R_{\ts0n}(\ts x\com z_n)

3127: \,\cdot\ts(\ts1\ot A_n) 

3128: $$

3129: \begin{equation}\label{ext}

3130: =\ 

3131: (\ts1\ot A_n)\,\cdot\ts

3132: R_{\ts0n}(\ts x\com z_n)

3133: \ts\ldots\ts

3134: R_{\ts01}(\ts x\com z_1)\,.

3135: \end{equation}

3136: Due to Propositition 4.1, the equality (\ref{ext}) indicates

3137: that the $n\ts$th exterior power of $\CC^N$ is a submodule

3138: of the $\YN\ts$-module (\ref{mod}), for $\nt=0$ and 

3139: the parameters $z_1\lcd z_n$ defined by (\ref{setext}).

3140: Denote by $V_n(z)$ this submodule.

3141: 

3142: Now consider the $\YN\ts$-module (\ref{mod}) in the case

3143: when $n=0$ and 

3144: \begin{equation}\label{setdext}

3145: y_{\ts\nt-k+1}=z-k+1

3146: \text{\ \ for each\ \ }

3147: k=1\lcd\nt\ts.

3148: \end{equation}

3149: The underlying vector

3150: space of this module is

3151: $W_{\ts\nt}^{\ts\ast}\ts$.

3152: Let $A_{\ts\nt}$ be the normalized operator of antisymmetrization

3153: on $W_{\ts\nt}^{\ts\ast}\ts$. Similarly to (\ref{ext}), we have

3154: $$

3155: \Ra_{\ts01}(\ts x\com y_{\ts\nt})

3156: \ts\ldots\ts

3157: \Ra_{\ts0\hskip.5pt\nt}(\ts x\com y_1)

3158: \,\cdot\ts(\ts1\ot A_{\ts\nt}) 

3159: $$

3160: \begin{equation}\label{dext}

3161: =\ 

3162: (\ts1\ot A_{\ts\nt})\,\cdot\ts

3163: \Ra_{\ts0\hskip.5pt\nt}(\ts x\com y_1)

3164: \ts\ldots\ts

3165: \Ra_{\ts01}(\ts x\com y_{\ts\nt})\,.

3166: \end{equation}

3167: Due to Corollary 4.1, the equality (\ref{dext}) shows

3168: that the $\nt\ts$th exterior power of $(\CC^N)^\ast$ is a submodule

3169: of the $\YN\ts$-module (\ref{mod}), for $n=0$ and for

3170: the parameters $y_1\lcd y_{\nt}$ defined by (\ref{setdext}).

3171: Denote by $V_{\ts\nt}(z)^{\ts\prime}$ this submodule.

3172: 

3173: \begin{Lemma}

3174: The\/ $\YN\ts$-module $V_n(z)^\ast$ dual to $V_n(z)\ts$, is equivalent to

3175: $V_n(z)^{\ts\prime}$.

3176: \end{Lemma}

3177: 

3178: \begin{proof}

3179: The action of $\YN$ on $V_n(z)$ can be described by the assignment

3180: $$

3181: T(x)\,\mapsto\,

3182: R_{\ts01}(\ts x\com z_1)

3183: \ts\ldots\ts

3184: R_{\ts0n}(\ts x\com z_n)

3185: \,\cdot\ts(\ts1\ot A_n)\,,

3186: $$ 

3187: cf.\ Proposition 4.1.

3188: The same action can also be described by the assignment

3189: $$

3190: T(x)^{-1}\,\mapsto\,

3191: R_{\ts0n}(\ts x\com z_n)^{-1}

3192: \ts\ldots\ts

3193: R_{\ts01}(\ts x\com z_1)^{-1}

3194: \,\cdot\ts

3195: (\ts1\ot A_n)

3196: $$

3197: $$

3198: =\ 

3199: (\ts1\ot A_n)

3200: \,\cdot\ts

3201: R_{\ts01}(\ts x\com z_1)^{-1}

3202: \ts\ldots\ts

3203: R_{\ts0n}(\ts x\com z_n)^{-1}\,,

3204: $$

3205: see (\ref{ext}).

3206: Using the definition (\ref{antip}) of the antipode ${\rm S}$

3207: along with (\ref{3.55}), the action of $\YN$ 

3208: on the module dual to $V_n(z)$ can then be described by

3209: $$

3210: T(x)\,\mapsto\,

3211: \Ra_{\ts01}(\ts x\com z_1)

3212: \ts\ldots\ts

3213: \Ra_{\ts0n}(\ts x\com z_n)

3214: \,\cdot\ts(\ts1\ot A_n)

3215: \quad\qed

3216: $$ 

3217: \end{proof}

3218: 

3219: \noindent

3220: Let us consider the one-dimensional

3221: $\YN\ts$-modules $V_N(z)$ and $V_N(z)^\ast\ts$.

3222: We will use the following observation, cf.\ \cite[Subsection 2.10]{MNO}.

3223: 

3224: \begin{Proposition}

3225: In the $\YN\ts$-modules\/ $V_N(z)$ and $V_N(z)^\ast$, respectively

3226: \begin{equation}\label{rat}

3227: T_{ab}(x)\,\mapsto\,\ts\de_{ab}\,\cdot\,\frac{x-z-1}{x-z}

3228: \,\quad\text{and}\quad\,\/

3229: T_{ab}(x)\,\mapsto\,\ts\de_{ab}\,\cdot\,\frac{x-z}{x-z-1}\ .

3230: \end{equation}

3231: \end{Proposition}

3232: 

3233: \begin{proof}

3234: It suffices to describe the action of $\YN$ only in the module $V_N(z)\ts$.

3235: The decription of the action of $\YN$ in $V_N(z)^\ast$ will then follow

3236: by the definition of the antipode $\operatorname{S}\ts$.

3237: Using Proposition 2.4, the left hand side of the equality

3238: (\ref{ext}) can be rewritten (\ts for any $n$) as

3239: \begin{equation}\label{lhs}

3240: \left(1-\frac{P_{\ts01}+\ldots+P_{\ts0n}}{x-z}\right)\,\cdot\ts(\ts1\ot A_n)\,\ts.

3241: \end{equation}

3242: If $n=N$, the sum $P_{\ts01}+\ldots+P_{\ts0n}$

3243: can be replaced by $1$ without affecting the value of (\ref{lhs})\ts;

3244: see for instance the proof of \cite[Theorem 2.3]{NT1}

3245: \qed

3246: \end{proof}

3247: 

3248: \noindent

3249: Hence the one-dimensional $\YN\ts$-module, described in the display 

3250: (\ref{rat}) first, is rational\ts; see Subsection 1.6. 

3251: By applying Lemma 4.4 when $n=N$, the second $\YN\ts$-module in

3252: (\ref{rat}) is also rational.

3253: By using the definition (\ref{1.33}) of the

3254: comultiplication on $\YN$, we now obtain the following result.

3255: 

3256: \begin{Corollary}

3257: Let\/ $g(x)\in\CC(x)$ be a rational function such that\/ $g(\infty)=1$, with

3258: all zeros and poles contained in $z+\ZZ$ for some common parameter $z\in\CC$.

3259: Then the one-dimensional\/ $\YN\ts$-module defined by {\rm(\ref{onedimrep})},

3260: is rational.

3261: \end{Corollary}

3262: 

3263: \noindent\textbf{4.5.}

3264: In this subsection, we shall derive Proposition 1.6 from Theorem 1.5.

3265: The (\ts independent) proof of Theorem 1.5 will be given in the next

3266: section. However, see the concluding remarks in Subsection 5.4.

3267: 

3268: Take the partition $\eta$ introduced in Subsection 1.6.

3269: Using the definition (\ref{1.62}), we obtain the equality  

3270: of rational functions

3271: \begin{equation}\label{gid}

3272: g_\eta(x+r)\,=\,

3273: \frac{g_\mu(x)}

3274: {g_{\ts\mut\ts}(\ts-\ts x\ns-\!M)}

3275: \,\cdot\,

3276: \frac{(\ts x+r)\ts(\ts x+M)}{x\ts(\ts x+M+r)}\ .

3277: \end{equation}

3278: Take the vector space $V_\xi^{\,\eta}$ defined as in

3279: Subsection 1.2. Regard this vector space as $\YN\ts$-module

3280: according to Subsection 1.5. The pullback of this module

3281: relative to the automorphism $\tau_{-r}$ is denoted by  

3282: $V_\xi^{\,\eta}(-r)\ts$.

3283: 

3284: By Theorem 1.5, the $\YN\ts$-module $\VOm$ 

3285: is equivalent to the $\YN\ts$-module $\Vlm$, 

3286: which was defined in Subsection 1.5. Using this definition,

3287: the latter $\YN\ts$-module is equivalent to the pullback of the module

3288: $V_\xi^{\,\eta}(-r)\ts$ relative to

3289: the automorphism (\ref{1.61}) of $\YN\ts$, where

3290: $$

3291: g(x)\,=\,

3292: \frac{g_\mu(x)\,\ts g_{\ts\lat}(\ts-\ts x\ns-\!M)}

3293: {g_{\ts\mut\ts}(\ts-\ts x\ns-\!M)\,g_\eta(x+r)}

3294: \cdot

3295: \frac{(\ts x+r)}x

3296: \,=\,g_{\ts\lat}(\ts-\ts x\ns-\!M)

3297: \cdot

3298: \frac{\,x+M+r}{x+M}\ .

3299: $$

3300: Here we used the definitions (\ref{1.52})\com(\ref{1.69})

3301: and the equality (\ref{gid}).

3302: Applying Theorem 1.5 to the $\YN\ts$-modules

3303: $V_\xi^{\,\eta}$ and $V_\Gamma$ instead of $\Vlm$ and $\VOm$ respectively,

3304: we now obtain Proposition 1.6.

3305: 

3306: %==============================================================================

3307: 

3308: \section{\hskip-3.5pt. Proof of Theorem 1.5}

3309: 

3310: %------------------------------------------------------------------------------

3311: 

3312: \noindent\textbf{5.1.}

3313: In this section, our basic vector space will be $\CC^L$ like

3314: it was in Section~3, and unlike it was in the previous section.

3315: Thus instead of (\ref{aux}),

3316: we will work with the tensor product of vector spaces

3317: \begin{equation}\label{aul}

3318: \CC^L\ns\ot\Wll\ts=\ts

3319: \CC^L\ns\ot(\CC^L)^\ast\ot\ldots\ot(\CC^L)^\ast\ot\CC^L\ns\ot\ldots\ot\CC^L

3320: \end{equation}

3321: where the first tensor factor $\CC^L$ will be auxiliary,

3322: and shall not be regarded as a representation of $GL_L\ts$.

3323: The $1\ns+\ts\lt+l\ts$ tensor factors in (\ref{aul}) will be labeled,

3324: respectively from the left to right, by the indices (\ref{inda}).

3325: For all possible indices $i\com j$ the operators $P_{ij}\com Q_{ij}$

3326: and the functions 

3327: $R_{ij}(x\com y)\ts$,

3328: ${}\,\Rb_{ij}(x\com y)\com\,\Ra_{0i}(x\com y)$

3329: will be defined as their counterparts in Subsection~4.1, 

3330: but $N$ will be replaced by $L\ts$.

3331: In particular, these functions will take values~in

3332: \begin{equation}\label{twoall}

3333: \End(\CC^L)\ot\End(\Wll)=

3334: \End(\ts\CC^L\ns\ot\Wll)\,.

3335: \end{equation}

3336: Instead of (\ref{instead}), we will use the second equality in

3337: the display (\ref{3.555}). We will set

3338: $L=N+M$, and fix a decomposition $\CC^{L}=\CC^N\ns\op\ts\CC^M$.

3339: 

3340: We begin the proof of Theorem 1.5 by choosing any standard tableau

3341: $\La$ of shape $\la\ts$, such that the tableau $\Om$ is obtained from $\La$

3342: by removing the boxes with numbers $1\lcd m\ts$. Here $\la$

3343: is a partition of $l\ts$, and $m=l-n\ts$. Similarly, choose

3344: any standard tableau $\Lat$ of shape $\lat\,$, such that the tableau $\Omt$

3345: is obtained from $\Lat$ by removing the boxes with numbers $1\lcd\mt\ts$.

3346: Again, here $\lat$ is a partition of $\lt\ts$, and $\mt=\lt-\nt\ts$.

3347: The standard tableaux of non-skew shapes $\mu$ and $\mut\ts$, 

3348: obtained by removing the boxes

3349: with the numbers $m+1\lcd l$ from $\La\ts$

3350: and the boxes with the numbers $\mt+1\lcd\lt$ from $\Lat\ts$,

3351: will be denoted respectively by $\Up$ and $\Upt\ts$.

3352: 

3353: In Section 3, we denoted by $F_\La$ the linear operator

3354: in the tensor product $W_l$ of $l$ copies of the vector space $\CC^L\ts$,

3355: corresponding to the matrix element $f_\La\in\CC S_l\ts$.

3356: Proposition 2.4 implies the equality of rational functions in~$x$

3357: $$

3358: R_{\ts0,\lt+1}(\ts x\com c_1(\La))

3359: \,\ldots\,

3360: R_{\ts0,\lt+l}(\ts x\com c_{\ts l}(\La))

3361: \ts\cdot\ts(\ts1\ot1\ot F_\La)

3362: $$

3363: \vglue-12pt

3364: \begin{equation}\label{4.61}

3365: =\,

3366: \biggl(1\ts-\!\sum_{\lt<j\le\lt+l}\frac{P_{\ts0j}}x\ts\biggr)

3367: \cdot\ts(\ts1\ot1\ot F_\La)

3368: \end{equation}

3369: with values in the algebra (\ref{twoall}).

3370: By applying Proposition 2.4 to the standard tableau $\Lat$ instead of

3371: $\La\ts$, and by changing $x$ to $-x\ts$, we obtain the equality

3372: $$

3373: \Ra_{\ts01}(\ts x\com-c_{\ts\lt\ts}(\Lat))

3374: \,\ldots\,

3375: \Ra_{\ts0\ts\lt\ts}(\ts x\com -c_1(\Lat))

3376: \ts\cdot\ts(\ts1\ot(\ts P_{\ts\lt}\,F_{\Lat}\,P_{\ts\lt}\ts)\ot1)

3377: $$

3378: \vglue-12pt

3379: \begin{equation}\label{4.62}

3380: =\ g_{\ts\lat}(-x)\ts\cdot

3381: \biggl(1\ts+\!\sum_{1\le i\le\lt}\frac{Q_{\ts0i}}x\ts\biggr)

3382: \cdot\ts(\ts1\ot(\ts P_{\ts\lt}\,F_{\Lat}\,P_{\ts\lt}\ts)\ot1)

3383: \end{equation}

3384: where $g_{\ts\lat}(-x)$ is defined via (\ref{1.62}). 

3385: To obtain (\ref{4.62}), we have also used (\ref{fact}).

3386: 

3387: Observe that for any indices $i$ and $j$ such that 

3388: $0<i\le\lt<j\ts$, we~have

3389: $$

3390: Q_{0i}\ts P_{0j}\ts\FLa\,=\,P_{0j}\,Q_{ij}\ts\FLa\,=\,0

3391: $$

3392: by Proposition 3.4. So the equalities

3393: (\ref{4.61})\com(\ref{4.62}) and Lemma 1.3 imply~that

3394: $$

3395: \Ra_{\ts01}(\ts x\com-c_{\ts\lt\ts}(\Lat))

3396: \ts\ldots\ts

3397: \Ra_{\ts0\ts\lt\ts}(\ts x\com -c_1(\Lat))\,\ts

3398: R_{\ts0,\lt+1}(\ts x\com c_1(\La))

3399: \,\ldots\,

3400: R_{\ts0,\lt+l}(\ts x\com c_{\ts l}(\La))

3401: $$

3402: \vglue-12pt

3403: \begin{equation}\label{4.6}

3404: \times\ \FLa\ =\ g_{\ts\lat}(-x)\ts\cdot

3405: \biggl(1\ts+\!\sum_{1\le i\le\lt}\frac{Q_{\ts0i}}x

3406: \ -\!\sum_{\lt<j\le\lt+l}\frac{P_{\ts0j}}x

3407: \ts\biggr)

3408: \cdot\ts\FLa\,\ts.

3409: \end{equation}

3410: 

3411: %------------------------------------------------------------------------------

3412: 

3413: \smallskip\noindent\textbf{5.2.}

3414: The equality (\ref{4.6}) is the starting point for our proof of Theorem 1.5. 

3415: Firstly, consider the case when $M=0$ and

3416: $\mu=\mut=(0\com0\ts,\ts\ldots\ts)\ts$. In this case $N=L$, $n=l$, $\nt=\lt$

3417: and the tableaux $\Om=\La$, $\Omt=\Lat$ have non-skew shapes.

3418: In this case, the left hand side of the equality (\ref{4.6})

3419: describes the action of the Yangian $\YL$ on the submodule $\VOm=\VLa$

3420: of the tensor product of the dual evaluation modules with

3421: the parameters $-\ts c_{\ts\lt}(\Lat)\lcd-\ts c_1(\Lat)$

3422: and the evaluation modules with

3423: the parameters $c_1(\La)\lcd c_{\ts l}(\La)\ts$.

3424: The tensor factors have to be taken in the order as specified here,

3425: see Proposition 4.1.

3426: 

3427: The right hand side of (\ref{4.6})

3428: describes the action of $\YL$ in the module $\Vlm=\Vla$ as

3429: defined in Subsection~1.5\ts; see Lemma 4.1.

3430: Indeed, here we have $g_\mu(x)=g_{\ts\mut}(x)=1$ and $\pi_{NM}=\pi_N$. 

3431: The image of the operator

3432: $\FOm=\FLa\,$, as a $\glL\ts$-submodule in $\Wll\,$,

3433: is equivalent to the $\glL\ts$-module $\Vla\ts$. 

3434: Thus the equality (\ref{4.6})

3435: implies Theorem 1.5 in the case when $M=0$.

3436: 

3437: Now suppose that $M\ge1$. 

3438: Consider again the tensor product of the dual evaluation modules over $\YL$

3439: with the parameters $-\ts c_{\ts\lt}(\Lat)\lcd-\ts c_1(\Lat)$ and 

3440: the evaluation modules with the parameters $c_1(\La)\lcd c_{\ts l}(\La)\ts$.

3441: Then consider the $\YL\ts$-module,

3442: obtained by pulling this tensor product

3443: back through the automorphism $\om_L$ of the algebra

3444: $\YL\ts$; see (\ref{1.51}). The action of

3445: $\YL$ in the latter module is described by the assignment 

3446: $$

3447: \sum_{a,b=1}^L\,E_{ab}\ot T_{ab}(x)\,\ts\mapsto\,

3448: R_{\ts0,\lt+l}(\ts-x\com c_{\ts l}(\La))^{\ts-1}

3449: \ldots\,

3450: R_{\ts0,\lt+1}(\ts-x\com c_1(\La))^{\ts-1}

3451: $$

3452: \vglue-12pt

3453: $$

3454: \times\ \,

3455: \Ra_{\ts0\ts\lt\ts}(\ts-x\com -c_1(\Lat))^{\ts-1}

3456: \ts\ldots\,

3457: \Ra_{\ts01}(\ts-x\com-c_{\ts\lt\ts}(\Lat))^{\ts-1}

3458: $$

3459: \vglue-6pt

3460: $$

3461: =\ \,

3462: g_\la(-x)\ g_{\ts\lat}(x)^{-1}\ts g_{\ts\lat}(x\ns-\ns L)^{-1}

3463: $$

3464: \vglue-8pt

3465: $$

3466: \times\ \,

3467: R_{\ts0,\lt+l}(\ts x\com-c_{\ts l}(\La))

3468: \,\ldots\,

3469: R_{\ts0,\lt+1}(\ts x\com-c_1(\La))

3470: $$

3471: \vglue-16pt

3472: \begin{equation}\label{wmod}

3473: \times\ \,

3474: \Ra_{\ts0\ts\lt\ts}(\ts x\com c_1(\Lat)\ns+\ns L)

3475: \,\ldots\,

3476: \Ra_{\ts01}(\ts x\com c_{\ts\lt\ts}(\Lat)\ns+\ns L)\,;

3477: \end{equation}

3478: 

3479: \medskip\noindent

3480: see (\ref{fact}) and (\ref{3.55}),(\ref{3.6}),(\ref{3.75}).

3481: Denote by $W$ the restriction of this $\YL\ts$-module

3482: to the subalgebra $\YN\subset\YL\ts$;

3483: we use the natural embedding $\ph_M:\YN\to\YL\ts$.

3484: That is, $\ph_M:\ts T_{ab}(x)\ts\mapsto\ts T_{ab}(x)$ 

3485: for $1\le a\com b\le N\ts$ by definition.

3486: Further, denote by $\Wp$ the $\YN\ts$-module

3487: obtained by pulling the $\YN\ts$-module $W$ back through

3488: the automorphism $\om_N$ of $\YN\ts$.

3489: Note that the vector space of the $\YN\ts$-modules

3490: $W$ and $\Wp$ is $\Wll\,$. 

3491: 

3492: The subspace $\Zmm\subset\Wll$ is defined by (\ref{3.41}).

3493: For every $i>\lt+m$ put

3494: \begin{equation}\label{Pw}

3495: R_{\ts0i}(x\com y)^{\,\wedge}\,=\,\ts

3496: 1-\frac{\,P_{\,0i}^{\ts\wedge}\,}{x-y}

3497: \end{equation}

3498: where $P_{\,0i}^{\ts\wedge}$ stands for the restriction

3499: of the permutational operator $P_{0i}$ on the vector space

3500: $\CC^L\ot\Wll$ to the subspace $\CC^N\ot\Zmm\subset\CC^L\ot\Wll\,$.

3501: If $0<i\le\nt\/$, let $Q_{\ts0i}^{\,\wedge}$ be the linear operator

3502: on $\CC^N\ot\Zmm\,$, acting as (\ref{1.45}) on the $i\ts$th tensor

3503: factor $(\CC^N)^\ast$ and the $0\ts$th

3504: tensor factor $\CC^N\ns$ of $\ts\CC^N\ot\Zmm$, and acting as the identity

3505: in other tensor factors of $\CC^N\ot\Zmm\,$. Then put

3506: $$

3507: \Ra_{\ts0i}(x,y)^{\,\wedge}\,=\,\ts

3508: \biggl(1+\frac{Q_{\ts0i}^{\,\wedge}}{x-y}\ts\biggr)

3509: \cdot

3510: \biggl(1-\frac1{(x-y)^{\ts2}}\biggr)^{\!-1}\ts.

3511: $$

3512: 

3513: One can define an action of $\YN$ in the vector space $\Zmm$~by

3514: $$

3515: \sum_{a,b=1}^N\,E_{ab}\ot T_{ab}(x)\,\ts\mapsto\,

3516: g_\la(-x)\ts\,g_{\ts\lat}(x)^{-1}\ts

3517: g_{\ts\lat}(\ts x-\ns L)^{-1}\ts

3518: g_{\ts\mut\ts}(\ts x-\ns L)

3519: $$

3520: \vglue-12pt

3521: $$

3522: \times\ \,

3523: R_{\ts0,\lt+l}(\ts x\com-c_n(\Om))^{\,\wedge}

3524: \,\ldots\,

3525: R_{\ts0,\lt+m+1}(\ts x\com-c_1(\Om))^{\,\wedge}

3526: \hskip-6pt

3527: $$

3528: \vglue-14pt

3529: \begin{equation}\label{zmod}

3530: \times\ \,

3531: \Ra_{\ts0\ts\nt\ts}(\ts x\com c_1(\Omt)\ns+\ns L)^{\,\wedge}

3532: \,\ldots\,

3533: \Ra_{\ts01}(\ts x\com c_{\ts\nt}(\Omt)\ns+\ns L)^{\,\wedge}\,\ts.

3534: \end{equation}

3535: 

3536: \smallskip\noindent

3537: This assertion will follow from the observation below; let us 

3538: denote by $Z$ the $\YN\ts$-module.

3539: Further, let us denote by $\Zp$ the $\YN\ts$-module

3540: obtained by pulling the $\YN\ts$-module $Z$ back through

3541: the automorphism $\om_N$ of $\YN\ts$. 

3542: The action of $\YN$ in $\Zp$ can be then described by the assignment

3543: $$

3544: \sum_{a,b=1}^N\,E_{ab}\ot T_{ab}(x)\,\ts\mapsto\,

3545: g_{\ts\lat}(-x)\ts\,

3546: g_{\ts\lat}(\ts-\ts x-\ns M)^{-1}\ts

3547: g_\mu(x)^{-1}\ts

3548: g_{\ts\mut\ts}(\ts-\ts x-\ns M)

3549: $$

3550: \vglue-12pt

3551: $$

3552: \times\ \,

3553: \Ra_{\ts01}(\ts x\com-\ts c_{\ts\nt}(\Omt)\ns-\ns M)^{\,\wedge}

3554: \,\ldots\,

3555: \Ra_{\ts0\ts\nt\ts}(\ts x\com-\ts c_1(\Omt)\ns-\ns M)^{\,\wedge}

3556: \hskip-24pt

3557: $$

3558: \vglue-8pt

3559: $$

3560: \times\ \,

3561: R_{\ts0,\lt+m+1}(\ts x\com c_1(\Om))^{\,\wedge}

3562: \,\ldots\,

3563: R_{\ts0,\lt+l}(\ts x\com c_n(\Om))^{\,\wedge}\,\,.

3564: $$

3565: 

3566: \smallskip\noindent

3567: Observe that the $\YN\ts$-module $\Zp$

3568: can also be obtained by pulling back the tensor product 

3569: of $\YN$-modules

3570: $$%\begin{equation}\label{Vcoc}

3571: V(\ts y_{\ts\nt})^\ast

3572: \ot\ts\ldots\ot 

3573: V(\ts y_1)^\ast

3574: \ot\Wmmo\ot

3575: V(\ts z_1)\ot\ts\ldots\ot V(z_n)

3576: $$%\end{equation}

3577: through the automorphism (\ref{1.61}) of the algebra $\YN\ts$, where 

3578: \begin{equation}\label{gx}

3579: g(x)\,=\,

3580: g_{\ts\lat}(-x)\,\ts

3581: g_{\ts\lat}(\ts-\ts x-\ns M)^{-1}\ts

3582: g_\mu(x)^{-1}\ts

3583: g_{\ts\mut\ts}(\ts-\ts x-\ns M)\,.

3584: \end{equation}

3585: Here we keep to the settings (\ref{setz}),(\ref{sety}).

3586: The Hopf algebra $\YN$ acts in the tensor factor $\Wmmo$ trivially,

3587: that is via the counit homomorphism $\,\varepsilon\ts$.

3588: 

3589: The underlying vector space of the $\YN\ts$-modules

3590: $Z$ and $\Zp$ is $\Zmm\,$, see (\ref{3.41}).

3591: In Subsection 3.5, we introduced the projector

3592: \begin{equation}\label{4.65}

3593: H_{m\mt}:\,\Wll\,\to\,\Zmm\,.

3594: \end{equation}

3595: %Note that the operator (\ref{4.65}) is $GL_N\times GL_M\ts$-equivariant.

3596: 

3597: \begin{Proposition}

3598: The projector\/ {\rm (\ref{4.65})} is\/

3599: a\/ $\YN\ts$-intertwiner\/ $W\ns\to Z$.

3600: \end{Proposition}

3601: 

3602: \begin{proof}

3603: This proposition follows

3604: by comparing the assignment (\ref{wmod})

3605: with the definition (\ref{zmod}) of the $\YN\ts$-module $Z\ts$.

3606: Indeed, let us expand the product 

3607: at the right hand side of the equality (\ref{wmod}) as the sum

3608: \begin{equation}\label{Sab}

3609: \sum_{a,b=1}^L\,E_{ab}\ot S_{ab}(x)

3610: \end{equation}

3611: for certain functions

3612: $S_{ab}(x)$ taking values in $\End(\Wll)\ts$.

3613: It suffices to show that the sum

3614: $$

3615: \sum_{a,b=1}^N\,E_{ab}\ot(\ts H_{m\mt}\,S_{ab}(x))

3616: $$

3617: is equal to the product at the right hand side of the assignment

3618: (\ref{zmod}), multiplied by $1\ot H_{m\mt}$ on the right.

3619: 

3620: To do this, let us apply to the values of the functions $S_{ab}(x)$

3621: in the sum (\ref{Sab})

3622: the operator conjugation with respect to each of the $\lt$ tensor

3623: factors $(\CC^L)^\ast$ of $\Wll\,$, cf.\ the beginning of

3624: our proof of Proposition 3.5. The resulting sum is the expansion

3625: of the product of the rational functions in $x\ts$,

3626: $$

3627: g_\la(-x)\ts\,g_{\ts\lat}(x)^{-1}\,\cdot\ts

3628: R_{\ts0,\lt+l}(\ts x\com-c_{\ts l}(\La))

3629: \,\ldots\,

3630: R_{\ts0,\lt+1}(\ts x\com-c_1(\La))

3631: $$

3632: \vglue-16pt

3633: \begin{equation}\label{A}

3634: \times\ \,

3635: R_{\ts0\ts\lt\ts}(\ts-x\com-c_1(\Lat)\ns-\ns L)

3636: \,\ldots\,

3637: R_{\ts01}(\ts-x\com-c_{\ts\lt\ts}(\Lat)\ns-\ns L)

3638: \end{equation}

3639: taking values in the algebra of operators on

3640: the tensor product %$W_{1+\lt+l}$ 

3641: of $1\ns+\lt+l$ copies

3642: of the vector space $\CC^L$. Denote by $W_{1+\lt+l}$ this tensor product.

3643: Each factor in (\ref{A}) is defined by

3644: (\ref{3.45}) where $P_{0j}$ is the linear operator on $W_{1+\lt+l}$

3645: exchanging the $0\ts$th and $j\ts$th copies of $\CC^L$.

3646: 

3647: The right hand side of (\ref{zmod}) is a rational function of $x$,

3648: taking values in the operator algebra $\End(\CC^N\ot\Zmm\ts)\ts$.

3649: By applying to these values the operator conjugation

3650: with respect to each of the $\nt$ tensor

3651: factors $(\CC^N)^\ast$ of $\CC^N\ot\Zmm\ts$, we obtain the function

3652: $$

3653: g_\la(-x)\ts\,g_{\ts\lat}(x)\,\cdot\ts

3654: R_{\ts0,\lt+l}(\ts x\com-c_n(\Om))^{\,\wedge}

3655: \,\ldots\,

3656: R_{\ts0,\lt+m+1}(\ts x\com-c_1(\Om))^{\,\wedge}

3657: $$

3658: \vglue-12pt

3659: \begin{equation}\label{B}

3660: \times\ \,

3661: R_{\ts0\ts\nt\ts}(\ts-x\com-c_1(\Omt)\ns-\ns L)^{\,\wedge}

3662: \,\ldots\,

3663: R_{\ts01}(\ts-x\com-c_{\ts\nt}(\Omt)\ns-\ns L)^{\,\wedge}

3664: \end{equation}

3665: taking values in the algebra of operators on

3666: $\CC^N\ot W_{\nt}\ot\Wmmo\ot W_n\,$.

3667: Each factor in (\ref{B}) is defined by (\ref{Pw})

3668: where $P_{\,0i}^{\ts\wedge}$ stands for the restriction

3669: of~the permutational operator $P_{0i}$ on the vector space 

3670: $W_{1+\lt+l}$ to the subspace

3671: $$

3672: \CC^N\ot W_{\nt}\ot\Wmmo\ot W_n\,\subset\,W_{1+\lt+l}\ts\,.

3673: $$

3674: Proposition 5.2 follows by comparing

3675: the products (\ref{A}) and (\ref{B}),

3676: and by using Corollary 2.4

3677: \qed

3678: \end{proof}

3679: 

3680: \begin{Corollary}

3681: The projector\/ {\rm (\ref{4.65})} is

3682: a\/ $\YN\ts$-intertwiner\/ $\Wp\!\to\ts\Zp$.

3683: \end{Corollary}

3684: 

3685: %------------------------------------------------------------------------------

3686: 

3687: \noindent\textbf{5.3.}

3688: Let us continue our proof of Theorem 1.5.

3689: Consider the image of the subspace $\Zmm$

3690: under the operator $\FLa$ in $\Wll\,$. Note that

3691: this image

3692: is contained in the subspace $\VLa\subset\Wll\,$.

3693: Also consider the $\YN\ts$-module $W$ defined in Subsection 5.2\ts;

3694: the vector space of this module is $\Wll\,$.

3695: 

3696: \begin{Proposition}

3697: The image of the subspace\/ $\Zmm\subset\Wll$ under the operator $\FLa$

3698: is a\/ $\YN$-submodule of\/ $W\ts$.

3699: \end{Proposition}

3700: 

3701: \begin{proof}

3702: The action of the coefficients of the series 

3703: $T_{ab}(x)$ with $1\le a\com b\le N$

3704: in the $\YN\ts$-module $W$ is described by the assignment (\ref{wmod}). 

3705: Here we use the natural embedding $\ph_M:\YN\to\YL\ts$. 

3706: Consider the product of the rational functions of $x$ at the

3707: right hand side of (\ref{wmod})\ts; these functions take values in the

3708: algebra (\ref{twoall}). We have an equality

3709: $$

3710: R_{\ts0,\lt+l}(\ts x\com-c_{\ts l}(\La))

3711: \,\ldots\,

3712: R_{\ts0,\lt+1}(\ts x\com-c_1(\La))

3713: $$

3714: \vglue-12pt

3715: $$

3716: \times\ \,

3717: \Ra_{\ts0\ts\lt\ts}(\ts x\com c_1(\Lat)\ns+\ns L)

3718: \,\ldots\,

3719: \Ra_{\ts01}(\ts x\com c_{\ts\lt\ts}(\Lat)\ns+\ns L)

3720: \,\cdot\,

3721: (\,1\ot\FLa\ts)

3722: \ =\ 

3723: (\,1\ot\FLa\ts)

3724: $$

3725: \vglue-8pt

3726: $$

3727: \times\ \,

3728: \Ra_{\ts01}(\ts x\com c_{\ts\lt\ts}(\Lat)\ns+\ns L)

3729: \,\ldots\,

3730: \Ra_{\ts0\ts\lt\ts}(\ts x\com c_1(\Lat)\ns+\ns L)

3731: $$

3732: \vglue-12pt

3733: \begin{equation}\label{C}

3734: \ts\ \times\ \,

3735: R_{\ts0,\lt+1}(\ts x\com-c_1(\La))

3736: \,\ldots\,

3737: R_{\ts0,\lt+l}(\ts x\com-c_{\ts l}(\La))\,;

3738: \end{equation}

3739: 

3740: \smallskip\noindent

3741: see our proof of Proposition 1.4, which was given in Subsection 4.2.

3742: To get the equality (\ref{C}), we also used the equality in (\ref{wmod}).

3743: Let us expand the product of $\ts\lt+l\ts$ factors in the last two lines of

3744: the display (\ref{C}), as the sum

3745: $$

3746: \sum_{a,b=1}^L\,E_{ab}\ot Y_{ab}(x)

3747: $$

3748: for certain functions $Y_{ab}(x)$ taking values in $\End(\Wll\ts)\ts$.

3749: Then consider the restrictions of the operator values of the

3750: functions $Y_{ab}(x)$ with $1\le a\com b\le N$

3751: to the subspace $\Zmm\subset\Wll\,$. Arguing like in the proof

3752: of Proposition 5.2, but using Lemma 2.4 itself instead of its

3753: Corollary 2.4, we prove that

3754: $$

3755: \sum_{a,b=1}^N\,E_{ab}\ot (\ts Y_{ab}(x)\,|\,\Zmm)

3756: \ = \ g_{\ts\mut\ts}(x\ns-\ns L)

3757: $$

3758: \vglue-8pt

3759: $$

3760: \times\ \,

3761: \Ra_{\ts01}(\ts x\com c_{\ts\nt\ts}(\Omt)\ns+\ns L)^{\,\wedge}

3762: \,\ldots\,

3763: \Ra_{\ts0\ts\nt\ts}(\ts x\com c_1(\Omt)\ns+\ns L)^{\,\wedge}

3764: $$

3765: \vglue-8pt

3766: $$

3767: \ts\ \times\ \,

3768: R_{\ts0,\lt+m+1}(\ts x\com-c_1(\Om))^{\,\wedge}

3769: \,\ldots\,

3770: R_{\ts0,\lt+l}(\ts x\com-c_n(\La))^{\,\wedge}\,\ts.

3771: $$

3772: 

3773: \smallskip\noindent

3774: In particular, the operator values of the functions $Y_{ab}(x)$

3775: with $1\le a,b\le N$ preserve the subspace $\Zmm\,$.

3776: Now Proposition~5.3 follows from (\ref{C})\qed

3777: \end{proof}

3778: 

3779: \noindent

3780: Since the $\YN\ts$-module $\Wp$ is obtained from $W$

3781: by pulling back through an automorphism of $\YN\ts$,

3782: Proposition 5.3 has the following corollary.

3783: 

3784: \begin{Corollary}

3785: The image of the subspace\/ $\Zmm\subset\Wll$ under the operator $\FLa$

3786: is a\/ $\YN$-submodule of $\Wp\ts$.

3787: \end{Corollary}

3788: 

3789: %------------------------------------------------------------------------------

3790: 

3791: \noindent\textbf{5.4.}

3792: In this subsection we complete the proof of Theorem 1.5.

3793: Let $V$ be the image of the subspace $\Zmm\subset\Wll$ 

3794: under the linear operator

3795: $$

3796: H_{m\mt}\ts\FLa:\,\Wll\,\to\,\Zmm\,.

3797: $$ 

3798: By identifying $\Zmm$ with the tensor product 

3799: $\Wmmo\ot W_{\ts\nt}^{\ts\ast}\ot W_n$ as we did in Subsection 3.5,

3800: that is by 

3801: exchanging the tensor factors $\Wmmo$ and $W_{\ts\nt}^{\ts\ast}$

3802: of $\Zmm\ts$, we identify $V$ with a certain subspace of

3803: $\Wmmo\ot W_{\ts\nt}^{\ts\ast}\ot W_n\,$. But due to the equality

3804: (\ref{3.4444}), the latter subspace coincides with the subspace 

3805: \begin{equation}\label{subspace}

3806: V_{\ts\Up\Upt}\ot\VOm\,\subset\,\Wmmo\ot W_{\ts\nt}^{\ts\ast}\ot W_n\,.

3807: \end{equation}

3808: 

3809: It follows from

3810: Corollaries 5.2 and 5.3, that the subspace $V\subset\Zmm$

3811: is a submodule in the $\YN\ts$-module $\Zp\ts$.

3812: Let us now regard $V$ as $\YN\ts$-module, by using

3813: the action of the algebra 

3814: $\YN$ in $V$ inherited from $\Zp\ts$. Then

3815: $V$ is a subquotient of the $\YN\ts$-module $\Wp$ by definition.

3816: 

3817: The observation on the $\YN\ts$-module $\Zp\ts$, made immediately

3818: before Proposition 5.2, yields the following

3819: description of the $\YN\ts$-module $V$. 

3820: Take the $\YN\ts$-module $\VOm$ as defined in Subsection 1.4. 

3821: Pull $\VOm$ back through the automorphism (\ref{1.61}) 

3822: of $\YN\ts$, where $g(x)$ is given by (\ref{gx}).

3823: Extend the resulting action of $\YN$ in %the vector space 

3824: $\VOm\ts$ to the subspace (\ref{subspace}) so that

3825: $\YN$ acts in $V_{\ts\Up\Upt}$ trivially.

3826: %%%that is via the counit homomorphism $\varepsilon\ts$.

3827: By exchanging the tensor factors $\Wmmo$ and $W_{\ts\nt}^{\ts\ast}$

3828: of $\Wmmo\ot W_{\ts\nt}^{\ts\ast}\ot W_n$ in (\ref{subspace}),

3829: we then obtain the $\YN\ts$-module $V$.

3830: 

3831: The subspace $V_{\ts\Up\Upt}\subset\Wmmo$ is equivalent to $V_{\ts\mu\mut}$

3832: as a representation of the group $GL_M\ts$.

3833: The subspace $\VLa\subset\Wll$ is equivalent to $V_{\la\lat}$ as a

3834: representation of the group $GL_L\ts$.

3835: Let us now regard $\VLa$ as $\YN\ts$-module

3836: by pulling back through the homomorphism

3837: $\pi_{NM}:\YN\to\UL\ts$, 

3838: and then through the automorphism (\ref{1.61}) of $\YN$ where

3839: $g(x)=g_{\ts\lat}(-x)\ts$.

3840: 

3841: \begin{Proposition}

3842: $\YN$-module $V$ is a subquotient of\/ $\YN\ts$-module\/ $\VLa\ts$.

3843: \end{Proposition}

3844: 

3845: \begin{proof}

3846: By the definition (\ref{1.69}), we have

3847: $$

3848: \pi_{NM}=\,\pi_L\circ\ts\om_L\circ\ts\ph_M\circ\ts\om_N\ts.

3849: $$

3850: Consider $\VLa$ as a $\YL\ts$-module, using the definition from

3851: Subsection 1.4. That is, the $\YL\ts$-module $\VLa$ is a submodule

3852: of the tensor product of the dual evaluation $\YN\ts-$modules with

3853: the parameters $-\ts c_{\ts\lt}(\Lat)\lcd-\ts c_1(\Lat)$

3854: and the evaluation $\YN\ts$-modules with

3855: the parameters $c_1(\La)\lcd c_{\ts l}(\La)\ts$.

3856: The equality (\ref{4.6}) implies, that

3857: the same $\YL\ts$-module can be obtained from the representation

3858: $\VLa$ of the group $GL_L\ts$, by pulling back through the homomorphism

3859: $\pi_{L}:\YL\to\UL\ts$, 

3860: and then through the automorphism (\ref{1.61}) of $\YN$ where

3861: $g(x)=g_{\ts\lat}(-x)\ts$.

3862: 

3863: Hence the $\YN\ts$-module $\VLa$ as defined immediately before stating

3864: Proposition 5.4, 

3865: can also be obtained by pulling the action of $\YL$ in $\VLa$

3866: back through the homomorphism

3867: $$

3868: \om_L\circ\ts\ph_M\circ\ts\om_N:\YN\to\YL\,.

3869: $$

3870: So $\VLa$ is a submodule in the $\YN\ts$-module $\Wp$.

3871: But by definition, $V$ is a quotient of a certain $\YN\ts$-submodule of

3872: $\Wp$. The latter submodule of $\Wp$ is contained in $\VLa$\qed

3873: \end{proof}

3874: 

3875: \noindent 

3876: Consider the restriction of the representation $\VLa$ of 

3877: the group $GL_L$

3878: to the subgroup $GL_M\subset GL_L\ts$. 

3879: Realize the vector space (\ref{1.0}) as

3880: \begin{equation}\label{4.10}

3881: {\rm Hom}_{\,GL_M}(\ts V_{\ts\Up\Upt}\ts\com\VLa\ts)\,.

3882: \end{equation}

3883: Since the image of the homomorphism $\pi_{NM}$ is contained in

3884: the subalgebra of $GL_M\ts$-invariants $\AMN\subset\UL\ts$, the action

3885: of the algebra $\YN$ in $\VLa$ from Proposition 5.4,

3886: induces an action of $\YN$ in (\ref{4.10}).

3887: This action of $\YN$ in the vector space (\ref{4.10})

3888: is irreducible, see \cite[Section 2]{MO}.

3889: 

3890: The operator (\ref{4.65}) is $GL_N\times GL_M\ts$-equivariant,

3891: and the vector space $V_{\ts\Up\Upt}\ot\VOm$ of the $\YN\ts$-module $V$

3892: comes with a natural action 

3893: of the groups $GL_N$ and $GL_M\ts$. The action of $GL_M$ in $V$

3894: commutes with the action of the algebra $\YN\ts$.

3895: By Proposition 5.4, the $\YN\ts$-module 

3896: \begin{equation}\label{4.11}

3897: {\rm Hom}_{\,GL_M}(\ts V_{\ts\Up\Upt}\ts\com V\ts)

3898: \end{equation}

3899: is a subquotient of %the $\YN\ts$-module 

3900: (\ref{4.10}). The $\YN\ts$-module (\ref{4.10})

3901: is irreducible, it must be equal to the $\YN\ts$-module

3902: (\ref{4.11}). Here we used Proposition~3.6.

3903: 

3904: The $\YN\ts$-module (\ref{4.11})  

3905: can also be obtained by pulling the $\YN\ts$-module $\VOm$

3906: as defined in Subsection 1.5, back through the automorphism

3907: (\ref{1.61}), where $g(x)$ is given by (\ref{gx}).

3908: The proof of Theorem 1.5 is complete. 

3909: 

3910: Note that (\ref{4.11}) is also a subquotient of (\ref{4.10})

3911: as a representation of the group $GL_N\ts$. Thus we obtain

3912: Proposition 1.3 together with Theorem 1.5.

3913: 

3914: Let us make a few concluding remarks. One can prove Proposition~1.6

3915: independently of Theorem 1.5, cf.\ \cite[Theorem A.4]{KW}. Then

3916: one can derive Theorem 1.5 from its particular case $\nt=0\ts$, 

3917: considered in \cite[Section 4]{N2}, by using Proposition 1.6.

3918: We chose the present proof of Theorem 1.5, because

3919: it generalizes the proof for $\nt=0\ts$.

3920: Moreover, the method of our proof of Theorem 1.5 extends from

3921: the group $GL_N$ to other classical Lie groups, 

3922: the orthogonal group $O_N$ and the symplectic group $Sp_{\ts N}\,$;

3923: see \cite[Section 5]{N2}.

3924: 

3925: %==============================================================================

3926: 

3927: \begin{acknowledgement}\hskip-9.5pt.

3928: I am grateful to Grigori Olshanski

3929: for numerous conversations. This work has been

3930: supported by the EC under the

3931: grant ERB-FMRX-CT97-0100.

3932: \end{acknowledgement}

3933: 

3934: %==============================================================================

3935: 

3936: \begin{thebibliography}{MNO}

3937: 

3938: \bibitem[B]{B}

3939: {R.\,Brauer},

3940: \textit{On algebras which are connected with the

3941: semisimple continuous groups},

3942: {Ann.\ Math.}

3943: \textbf{38}

3944: (1937),

3945: 857--872.

3946: 

3947: \bibitem[C1]{C1}

3948: {I.\,Cherednik},

3949: \textit{On special bases of irreducible finite-dimensional representations

3950: of the degenerate affine Hecke algebra},

3951: {Funct.\ Analysis Appl.}

3952: \textbf{20}

3953: (1986),

3954: 87--89.

3955: 

3956: \bibitem[C2]{C2}

3957: {I.\,Cherednik},

3958: \textit{A new interpretation of Gelfand-Zetlin bases},

3959: {Duke Math.\ J.}

3960: \textbf{54}

3961: (1987),

3962: 563--577.

3963: 

3964: \bibitem[CP]{CP}

3965: {V.\,Chari and A.\,Pressley},

3966: \textit{Fundamental representations of Yangians and

3967: singularities of $R$-matrices},

3968: {J.\ Reine Angew.\ Math.}

3969: \textbf{417}

3970: (1991),

3971: 87--128.

3972: 

3973: \bibitem[D]{D}

3974: {J.\,Dixmier},

3975: \textit{Alg\`ebres Enveloppantes},

3976: Gauthier-Villars, Paris, 1974.

3977: % {V.\,Drinfeld},

3978: % \textit{Hopf algebras and the quantum Yang-Baxter equation},

3979: % {Soviet Math.\ Dokl.}

3980: % \textbf{32}

3981: %(1985),

3982: % 254--258.

3983: 

3984: \bibitem[K]{K}

3985: {K.\,Koike},

3986: \textit{On the decomposition of tensor products of the representations

3987: of the classical groups: by means of universal characters}, 

3988: {Adv.\ Math.}

3989: \textbf{74}

3990: (1989),

3991: 57--86.

3992: 

3993: \bibitem[KW]{KW}

3994: {R.\,King and T.\,Welsh},

3995: \textit{Construction of $GL(n)$-modules using composite tableaux},

3996: {Linear and Multilinear Algebra}

3997: \textbf{34}

3998: (1993),

3999: 99--122.

4000: 

4001: \bibitem[M]{M}

4002: {I.\,Macdonald},

4003: \textit{Symmetric Functions and Hall Polynomials},

4004: Clarendon Press, Oxford, 1995.

4005: 

4006: \bibitem[MO]{MO}

4007: {A.\,Molev and G.\,Olshanski},

4008: \textit{Centralizer construction for twisted Yangians},

4009: {Selecta Math.}

4010: \textbf{6}

4011: (2000),

4012: 269--317.

4013: 

4014: \bibitem[MNO]{MNO}

4015: {\hskip2pt\hskip-4pt A.\,Molev, M.\,Nazarov and G.\,Olshanski},

4016: \textit{Yangians and classical Lie algebras},

4017: {Russian Math.\ Surveys}

4018: \textbf{51} 

4019: (1996),

4020: 205--282.

4021: 

4022: \bibitem[N1]{N1}

4023: {M.\,Nazarov},

4024: \textit{Yangians and Capelli identities},

4025: {Amer.\ Math.\ Soc.\ Translations}

4026: \textbf{181}

4027: (1998),

4028: 139--163.

4029: 

4030: \bibitem[N2]{N2}

4031: {M.\,Nazarov},

4032: \textit{Representations of twisted

4033: Yangians associated with skew Young diagrams},

4034: {\tt math.RT/0207115\ts}.

4035: 

4036: \bibitem[NT1]{NT1}

4037: {M.\,Nazarov and V.\,Tarasov},

4038: \textit{On irreducibility of tensor products of Yangian modules},

4039: {Internat.\ Math.\ Research Notices}

4040: (1998),

4041: 125--150.

4042: 

4043: \bibitem[NT2]{NT2}

4044: {M.\,Nazarov and V.\,Tarasov},

4045: \textit{On irreducibility of tensor products of Yangian modules

4046: associated with skew Young diagrams},

4047: {Duke Math.\ J.}

4048: \textbf{112}

4049: (2002),

4050: 342--378.

4051: 

4052: \bibitem[O1]{O1}

4053: {G.\,Olshanski},

4054: \textit{Extension of the algebra $U(g)$ for infinite-dimensional classical

4055: Lie algebras $g$, and the Yangians $Y(gl(m))$},

4056: {Soviet Math.\ Dokl.}

4057: \textbf{36}

4058: (1988),

4059: 569--573.

4060: 

4061: \bibitem[O2]{O2}

4062: {G.\,Olshanski},

4063: \textit{Representations of infinite-dimensional classical groups,

4064: limits of enveloping algebras, and Yangians},

4065: {Adv. Soviet Math.}

4066: \textbf{2} 

4067: (1991), 

4068: 1--66.

4069: 

4070: \bibitem[S]{S}

4071: {J.\,Stembridge},

4072: \textit{Rational tableaux and the tensor algebra of\/ $gl_n\ts$},

4073: {J.\ Comb.\ Theory}

4074: \textbf{A46}

4075: (1987),

4076: 79--120.

4077: 

4078: \bibitem[VK]{VK}

4079: {A.\,Vershik and S.\,Kerov},

4080: \textit{Characters and factor representations of the

4081: infinite unitary group},

4082: {Soviet Math.\ Dokl.}

4083: \textbf{26}

4084: (1982),

4085: 570--574.

4086: 

4087: \bibitem[W]{W}

4088: {H.\,Weyl},

4089: \textit{Classical Groups, their Invariants and Representations},

4090: Princeton University Press, Princeton, 1946.

4091: 

4092: \bibitem[Y1]{Y1}

4093: {A.\,Young},

4094: \textit{On quantitative substitutional analysis I\ts} and \textit{II\/},

4095: {Proc.\ London Math. Soc.}

4096: \textbf{33}

4097: (1901),

4098: 97--146  

4099: and

4100: \textbf{34}

4101: (1902),

4102: 361--397.

4103: 

4104: \bibitem[Y2]{Y2}

4105: {A.\ Young,}

4106: \textit{On quantitative substitutional analysis VI\ts},

4107: {Proc. London Math. Soc.}

4108: \textbf{34}

4109: (1932),

4110: 196--230.

4111: 

4112: \end{thebibliography}

4113: 

4114: %==============================================================================

4115: 

4116: \enddocument

4117: 

4118: %==============================================================================

4119: 

4120: $$

4121: \xi^{\,\prime}_{\ts i}\ts=\,

4122: \left\{

4123: \begin{array}{ll}

4124: M\ns+N\ns-\lat^{\ts\prime}_{\ts r-i+1}

4125: &\ \ \textrm{if}\ \ \ i\le r\ts,\\[2pt]

4126: \la^{\ts\prime}_{\ts i-r}

4127: &\ \ \textrm{if}\ \ \ i>r

4128: \end{array}

4129: \right.

4130: $$

4131: and

4132: $$

4133: \eta^{\,\prime}_{\ts i}\ts=\,

4134: \left\{

4135: \begin{array}{ll}

4136: M\ns-\mut^{\,\prime}_{\ts r-i+1}

4137: &\ \ \textrm{if}\ \ \ i\le r\ts,\\[2pt]

4138: \mu^{\,\prime}_{\ts i-r}

4139: &\ \ \textrm{if}\ \ \ i>r\ts.

4140: \end{array}

4141: \right.

4142: $$

4143: The terms in each of the sequences 

4144: $\xi^{\ts\prime}=(\ts\xi^{\ts\prime}_1,\xi^{\ts\prime}_2\ts,\,\ldots\,)$

4145: and

4146: $\eta^{\ts\prime}=(\ts\eta^{\ts\prime}_1,\eta^{\ts\prime}_2\ts,\,\ldots\,)$

4147: are not negative 

4148: and do not increase because of our assumptions (\ref{ass}) on the integer $M$.

4149: Since $\xi^{\ts\prime}$ and $\eta^{\ts\prime}$ are partitions,

4150: the partitions $\xi$ and $\eta$ can be determined as their conjugates.

4151: Furthermore, we have $\xi^{\,\prime}_{\ts i}\ge\eta^{\,\prime}_{\ts i}$ and

4152: $$

4153: \xi_{\ts i}\ge\eta_{\ts i}\ ,\,

4154: \xi^{\,\prime}_{\ts i}-\eta^{\,\prime}_{\ts i}\le N

4155: $$

4156: for every index $i\ge1$ due to (\ref{con}).

4157: 

4158: %------------------------------------------------------------------------------

4159: 

4160: \noindent

4161: By definition, we have $V_n^{\hskip.5pt\prime}(z)=\{0\}$ when $n>N$.

4162: Assume that $V_0(z)$ is the trivial $\YN\ts$-module $\CC\ts$, defined

4163: by using the counit $\varepsilon:\YN\to\CC\ts$.

4164: 

4165: \begin{Proposition}

4166: If\/ $n\le N$ then

4167: the\/ $\YN\ts$-module\/ $V_n^{\hskip.5pt\prime}(z)$

4168: is equivalent to the pullback of the\/ $\YN\ts$-module

4169: $V_{\ts N-n}(\ts z\ns-\ns n)$ 

4170: relative to the automorphism\/ {\rm(\ref{1.61})}, where

4171: $$

4172: g(x)\,=\,

4173: \frac{\,x-z}{x-z-1}\ .

4174: $$

4175: \end{Proposition}

4176: 

4177: %-----------------------------------------------------------------------------

4178: