1: \documentclass[12pt]{article}

2: %\usepackage[section]{placeins}

3: \usepackage{graphicx}

4: \usepackage{amsfonts}

5: \usepackage{amsmath}

6: \usepackage{mathrsfs}

7: %\def\endequation{\eqno \hbox{\@eqnnum}$$\@ignoretrue}

8: \topmargin -.65cm

9: \setlength{\oddsidemargin}{2mm}

10: \textheight=227mm

11: \textwidth=155mm

12: \def\beq{\begin{equation}}

13: \def\eeq{\end{equation}}

14: \def\bea{\begin{eqnarray}}

15: \def\eea{\end{eqnarray}}

16: \def\nn{\nonumber}

17: \def\ba{\begin{array}}

18: \def\ea{\end{array}}

19: \def\d{\partial}

20: \def\v{\vert}

21: \def\l{\langle}

22: \def\r{\rangle}

23: %

24: \def\one{1\hskip -1mm{\rm l}}

25: 

26: \renewcommand{\thefootnote}{\fnsymbol{footnote}}

27: \setcounter{footnote}{0}

28: 

29: \begin{document}

30: 

31: \begin{center}

32: {\large \bf \sf

33: Exact partition function of $SU(m|n)$ supersymmetric \\

34: Haldane-Shastry spin chain }

35: 

36: \vspace{1.3cm}

37: 

38: {\sf B. Basu-Mallick\footnote{ 

39: e-mail address: bireswar.basumallick@saha.ac.in}

40: and Nilanjan Bondyopadhaya\footnote{e-mail address:

41: nilanjan.bondyopadhaya@saha.ac.in } }

42: 

43: \bigskip

44: 

45: {\em Theory Group, \\

46: Saha Institute of Nuclear Physics, \\

47: 1/AF Bidhan Nagar, Kolkata 700 064, India } \\

48: \bigskip

49: 

50: \end{center}

51: 

52: 

53: \vspace {2 cm}

54: \baselineskip=18pt

55: \noindent {\bf Abstract }

56: 

57: By taking the freezing limit of a spin Calogero-Sutherland model

58: containing `anyon like' representation of the permutation algebra, 

59: we derive the exact partition function of $SU(m|n)$ supersymmetric

60: Haldane-Shastry (HS) spin chain. 

61: This partition function allows us to study global properties 

62: of the spectrum like level density distribution and nearest neighbour

63: spacing distribution. It is found that, for supersymmetric HS spin chains 

64: with large number of lattice sites,  continuous part of the 

65: energy level density obeys Gaussian distribution with a 

66: high degree of accuracy. The mean value and standard deviation

67: of such Gaussian distribution can be calculated exactly.

68: We also conjecture that the partition function of supersymmetric HS spin 

69: chain satisfies a duality relation under the exchange of 

70: bosonic and fermionic spin degrees of freedom. 

71: 

72: \baselineskip=16pt

73: \vspace {.6 cm}

74: \noindent PACS No. : 02.30.Ik, 75.10.Jm, 05.30.-d, 03.65.Fd 

75: 

76: \vspace {.1 cm}

77: \noindent Keywords : 

78: Haldane-Shastry spin chain, partition function,

79: level density distribution, supersymmetry 

80: 

81: 

82: \newpage 

83: \baselineskip=18pt

84: \noindent \section {Introduction }

85: \renewcommand{\theequation}{1.{\arabic{equation}}}

86: \setcounter{equation}{0}

87: 

88: \medskip

89: 

90: 

91: Haldane-Shastry (HS) spin chain is 

92: a well known quantum integrable model, where equally spaced 

93: spins on a circle interact with each other through  

94: pairwise exchange interactions inversely proportional

95: to the square of their chord distances. 

96: Study of such HS spin-$\frac{1}{2}$ chain with long-range interaction

97: was originally motivated from the fact that the exact

98: ground state wavefunction of this model coincides with the 

99: $U \rightarrow \infty$ limit of Gutzwiller's

100: variational wave function for the Hubbard model,

101: and also with the one-dimensional version 

102: of the resonating valence bond state proposed by Anderson [1,2].

103: Remarkably,  HS spin chain can be explicitly solved

104: in much greater detail than integrable 

105:  spin chains with short-range interactions, has a Yangian 

106: quantum group symmetry and

107: interestingly shares many of the characteristics of

108: an ideal gas, but with fractional statistics [3-5].  

109: The Hamiltonian of $SU(m)$ HS model with $N$ number of lattice sites 

110: is given by

111: \beq

112:  H=\frac{1}{2}\sum_{1\leq j <k \leq N}

113:  \frac{(1+P_{jk})}{\sin^2(\xi_j - \xi_k )}  \, ,

114: \label {a1}

115: \eeq

116: where $\xi_j=j\pi / N$ and $P_{jk}$ is the exchange operator

117: interchanging the `spins' (taking $m$ possible values) on $j$-th and

118: $k$-th lattice sites.

119: 

120: By using the motif representations associated with

121: $Y(gl_m)$ Yangian symmetry of HS spin chain (\ref{a1}), 

122: one can find out its complete spectrum including the degeneracy factor 

123: for each energy level [6-8]. 

124: However, in practice, the computation of such degeneracy factors

125: becomes very cumbersome for $m>2$ and large values of $N$. 

126: Therefore, it is difficult to express 

127:  the partition function of HS spin chain in a simple form 

128: (for arbitrary values of $N$ and $m$) with the help of motif representations. 

129: Due to this reason, it is worthwhile to explore other approaches 

130: for calculating the partition function of HS spin chain. 

131: In fact, a rather simple expression for the exact partition function 

132: of $SU(m)$ HS spin chain has been obtained recently [9]

133: by applying the so called freezing trick [10-12].

134: This  freezing trick utilizes the connection between 

135: $SU(m)$ HS spin Hamiltonian 

136: and $SU(m)$ spin Calogero-Sutherland (CS) model which has dynamical 

137:  as well as spin degrees of freedom. More precisely, 

138: one takes the strong coupling limit of spin CS Hamiltonian, 

139: so that the particles freeze at their classical equilibrium positions 

140: of the scalar part of the potential and

141: spins get decoupled from the dynamical degree of freedom.

142: As a result, one can derive the partition function of HS spin chain

143: by `modding out' the partition function of spinless CS model from that of

144: the spin CS model.  By using this partition function

145: of $SU(m)$ HS spin chain, it is possible to study the energy level density

146: distribution and the nearest neighbour spacing distribution 

147: for fairly large values of $N$ [9]. Interestingly, 

148: it has been found that,  the continuous part of such 

149: energy level density follows Gaussian distribution to

150: a high degree of approximation. 

151: 

152: In this context it may be noted that,  there

153: exists a  $SU(m|n)$ supersymmetric extension of HS spin chain [3], where 

154: each site is occupied by either one of the $m$

155: type of bosonic states or one of the 

156: $n$ type of fermionic states.  Such supersymmetric spin chains

157: play an important role in describing some

158: correlated systems of condensed matter physics, 

159: where holes moving in the dynamical background of spin behave as 

160: bosons and spin-$\frac{1}{2}$ electrons behave as fermions [13]. It is worth

161: noting that the supersymmetric $SU(m|n)$ HS spin chain exhibits 

162: $Y(gl_{(m|n)})$ super-Yangian symmetry [3],  

163: which is also the quantum group symmetry of supersymmetric

164: $SU(m|n)$ Polychronakos spin chain [14,15]. Consequently,

165: by using the motif representations and skew-Young 

166: diagrammes associated with supersymmetric 

167: Polychronakos spin chain [15], one can in principle calculate

168: the degeneracy factors for all energy eigenvalues of supersymmetric

169: HS spin chain. However, similar to the nonsupersymmetric case, 

170: this method for finding the full spectrum and 

171: related partition function becomes very complicated for large values 

172: of $N$. 

173: 

174: The aim of the present article is to 

175: find out the exact partition function for supersymmetric $SU(m|n)$

176: HS model by applying the 

177: freezing trick and also to study global properties like level density 

178: distribution of the corresponding spectrum. For this purpose, 

179: it is convenient to map the supersymmetric HS model to a usual spin chain  

180: containing an `anyon like' representation of the permutation algebra

181: as spin dependent interactions [16,17]. 

182: In Sec.2 we describe this mapping and also show how the freezing trick 

183: can be applied for the case of $SU(m|n)$ HS spin chain by embedding it in a  

184: spin CS model containing the same 

185: anyon like representation of the permutation algebra. In Sec.3,  we 

186: find out the complete spectrum of such spin CS model

187: including the degeneracy factors for all energy levels. In Sec.4,

188: we calculate the partition function of this spin CS model 

189: at the strong coupling limit and divide it by that of the spinless

190: CS model to finally obtain 

191: the partition function of $SU(m|n)$ HS spin chain. 

192: In this section, we also discuss about the motif representation for 

193: $SU(m|n)$ HS spin chain and find that, due to the lifting 

194: of a selection rule, some extra energy levels appear in the 

195: spectrum in comparison with the case of $SU(m)$ spin chain. 

196:  Subsequently, we conjecture that the partition function of 

197: $SU(m|n)$ HS model satisfies a duality relation under the 

198: exchange of bosonic and fermionic spin degrees of freedom. 

199: In Sec.5, we study the level density distribution

200: and the nearest neighbour spacing distribution for the spectrum of 

201: $SU(m|n)$ HS spin chain by using its exact partition function. 

202: It is found that, for sufficiently large values of $N$,  

203:  continuous part of the energy level density

204: satisfies the Gaussian distribution with a high degree of accuracy.

205: We also derive exact expressions for the mean value and 

206: standard deviation which characterize such 

207: Gaussian distribution. Sec.6 is the concluding section. 

208: 

209: 

210: 

211: 

212: 

213: \noindent \section {Application of the freezing trick }

214: \renewcommand{\theequation}{2.{\arabic{equation}}}

215: \setcounter{equation}{0}

216: 

217: For the purpose of defining the  $SU(m|n)$ supersymmetric HS spin chain,  

218: let us consider a set of operators like

219: $C_{j \alpha}^\dagger$($C_{j \alpha}$) which creates (annihilates)

220: a particle of species $\alpha$ on the $j$-th lattice site.

221: These creation (annihilation) operators are assumed to be bosonic when 

222: $\alpha \in [1,2,....,m]$ and fermionic when $\alpha \in [m+1,m+2,....,m+n]$.

223: Thus,  the parity of $C_{j \alpha}^\dagger$($C_{j \alpha}$) is defined as 

224: \bea

225:  &&p(C_{j \alpha})=p(C_{j \alpha}^\dagger)=0 ~

226: \mathnormal{for}~ \alpha \in [1,2,....,m] \, , \nn \\

227:  &&p(C_{j \alpha})=p(C_{j \alpha}^\dagger)=1 ~

228:  \mathnormal{for}~ \alpha \in [m+1,m+2,....,m+n] \, . \nn

229: \eea

230: These operators satisfy commutation (anti-commutation) relations like 

231: \beq

232: [C_{j \alpha},C_{k \beta}]_{\pm}=0 \, ,~ 

233: [C_{j \alpha}^\dagger,C_{k \beta}^\dagger]_{\pm}=0 \, , ~

234: [C_{j \alpha},C_{k \beta}^\dagger]_{\pm}=\delta_{jk}\delta_{\alpha \beta} \, ,

235: \label{b1}

236: \eeq

237: where $[A,B]_{\pm} \equiv AB- (-1)^{p(A)p(B)}BA$.

238: Next, we focus our attention to a subspace of 

239: the related Fock space, for

240: which the total number of particles per site is always one:

241: 

242: \beq

243:  \sum_{\alpha=1}^{m+n} C_{j\alpha}^{\dagger} C_{j\alpha}=1,

244: \label{b2}

245: \eeq

246: for all $j$. On the above mentioned subspace, 

247: one can define supersymmetric exchange operators as 

248: \beq

249: \hat{P}_{jk}^{(m|n)} \equiv

250: \sum_{\alpha,\beta=1}^{m+n} C_{j \alpha}^\dagger

251: C_{k \beta}^\dagger C_{j \beta}C_{k \alpha} \, ,

252: \label{b3}

253: \eeq

254: where $1 \leq j <k \leq N$.  These 

255: $\hat{P}_{jk}^{(m|n)}$'s yield 

256: a realization of the permutation algebra given by

257: \beq

258: \mathcal{P}_{jk}^2=1 \, , 

259: ~\mathcal{P}_{jk}\mathcal{P}_{kl}=\mathcal{P}_{jl}\mathcal{P}_{jk}

260: =\mathcal{P}_{kl}\mathcal{P}_{jl} \, ,

261:  ~[\mathcal{P}_{jk},\mathcal{P}_{lm}]=0 \, ,

262: \label{b4}

263: \eeq

264: where $j,~k,~l,~m$ are all distinct indices.

265:  Replacing $P_{jk}$ by $\hat{P}_{jk}^{(m|n)}$ in eqn.(\ref{a1}),

266:  one obtains the Hamiltonian of $SU(m|n)$ supersymmetric 

267:   HS model as [3]

268: \beq

269: \mathcal{H}^{(m|n)}= \frac{1}{2} \sum_{1\leq j<k\leq N} 

270: \frac{ \left(1+\hat{P}_{jk}^{(m|n)}\right)}{\sin^2(\xi_j-\xi_k)}.

271: \label{b5}

272: \eeq

273: 

274: Now we want to describe how this $SU(m|n)$ supersymmetric 

275: HS model (\ref{b5}), containing creation-annihilation operators,   

276: can be transformed to a spin chain. 

277: To this end, we consider a particular type of 

278: anyon like representation 

279: of permutation algebra (\ref{b4}), which acts on 

280: a spin state like $\v \alpha_1 \alpha_2 \dots \alpha_N \r $

281:  (with $\alpha_j \in [1,2,\dots,m+n]  $) as [16,17]

282: \beq

283: \tilde{P}_{jk}^{(m|n)} \v \alpha_1  \dots \alpha_j \dots 

284: \alpha_k \dots \alpha_N \r

285: = e^{i \Phi (\alpha_j,\alpha_{j+1},\dots,\alpha_k)}

286: \v \alpha_1  \dots \alpha_k \dots \alpha_j \dots \alpha_N \r,

287: \label{b6}

288: \eeq

289: where 

290: $e^{i \Phi(\alpha_j,\alpha_{j+1},\dots,\alpha_k)}= 1 $ 

291: if $\alpha_j,\alpha_k \in [1,2,\dots,m]$,

292: $e^{i \Phi(\alpha_j,\alpha_{j+1},\dots,\alpha_k)}= -1 $ 

293: if $\alpha_j,\alpha_k \in [m+1,m+2,\dots,m+n]$, and 

294: $e^{i \Phi(\alpha_j,\alpha_{j+1},\dots,\alpha_k)}= 

295: (-1)^{{\pi \sum_{p=j+1}^{k-1} \sum_{\tau=m+1}^{m+n} 

296: \delta_{\alpha_p, \tau }}} $ 

297: if $\alpha_j\in [1,2,\dots,m]$  and 

298: $ \alpha_k \in [m+1,m+2, \dots,m+n]$ or  vice versa.

299: For the purpose of interpreting the phase factor 

300: $e^{i \Phi(\alpha_j,\alpha_{j+1},\dots,\alpha_k)}$ in a physical way, 

301: it is convenient to call $ \alpha_i $  a `bosonic' spin when $ 

302: \alpha_i \in [1,2,\dots,m] $ and 

303:  a `fermionic' spin when $\alpha_i \in [m+1,m+2,\dots,m+n] $.

304: From eqn.(\ref {b6}) it follows that, the exchange of two bosonic (fermionic)

305: spins produces a phase factor of $ 1 \, (-1)$ irrespective of the nature of 

306: spins situated in between the $j$-th and $k$-th lattice sites. However, if

307: we exchange one bosonic spin with one fermionic spin,  then the phase factor

308: becomes $(-1)^\rho$ where $\rho$ is the total 

309: number of fermionic spins situated in between the

310: $j$-th and $k$-th lattice sites. 

311: 

312: Next we observe that,  due to the constraint (\ref{b2}), 

313: the Hilbert space associated with $SU(m|n)$ HS Hamiltonian (\ref{b5}) 

314: can be spanned through the following 

315: orthonormal basis

316: vectors: $ C_{1 \alpha_1}^\dagger C_{2 \alpha_2}^\dagger 

317: \dots C_{N \alpha_N}^\dagger \v 0 \r $, where $\v 0\r$ is the

318:  vacuum state and $\alpha_j \in [1,2,\dots,m+n]$. Consequently, it is 

319: possible to define a one-to-one mapping  between 

320: these basis vectors  and those of the 

321:  above mentioned spin chain as

322: \beq

323: \v\alpha_1 \alpha_2 \dots \alpha_N \r \leftrightarrow C_{1 \alpha_1}^\dagger

324: C_{2 \alpha_2}^\dagger \dots C_{N \alpha_N}^\dagger \v 0 \r.

325: \label{b7}

326: \eeq

327: With the help of commutation (anti-commutation) relations (\ref{b1}),  

328:  one can easily verify that 

329: \bea

330: &&\hat{P}_{jk}^{(m|n)} \, C_{1 \alpha_1}^\dagger 

331: \dots C_{j \alpha_j}^\dagger \dots

332: C_{k \alpha_k}^\dagger \dots C_{N \alpha_N}^\dagger \v 0 \r  \nn \\

333: &&~~~~~~~~~~~~~~~~~~~~~~~~~~

334: \, = \, e^{i\Phi(\alpha_j,\dots,\alpha_k)} \, C_{1 \alpha_1}^\dagger 

335:  \dots C_{j \alpha_k}^\dagger \dots C_{k \alpha_j}^\dagger 

336: \dots C_{N \alpha_N}^\dagger \v 0\r, ~~~~

337: \label{b8}

338: \eea

339: where $e^{i\Phi(\alpha_j,\dots,\alpha_k)}$ is the same phase factor 

340: which appeared in eqn.(\ref{b6}). 

341: Comparison of eqn.(\ref{b8}) with eqn.(\ref{b6}) through the mapping 

342: (\ref{b7}) reveals that, the anyon like representation 

343:  $\tilde{P}_{jk}^{(m|n)}$  is equivalent to the supersymmetric 

344: exchange operator $\hat{P}_{jk}^{(m|n)}$.

345: Hence, if we define a spin chain Hamiltonian through

346:  $\tilde{P}_{jk}^{(m|n)}$ as

347: \beq

348: H^{(m|n)}= \frac{1}{2} \sum_{1\leq j<k\leq N} 

349: \frac{ \left(1+\tilde{P}_{jk}^{(m|n)}\right)}{\sin^2(\xi_j-\xi_k)},

350: \label{b9}

351: \eeq

352: that would be completely equivalent to the

353: supersymmetric $SU(m|n)$ HS model (\ref{b5}) [16]. 

354: Clearly,  for the special case $n=0$, $\tilde{P}_{jk}^{(m|n)}$ reproduces 

355: the original spin exchange operator $P_{jk}$ and 

356: ${H}^{(m|n)}$ (\ref{b9}) reduces to the 

357: Hamiltonian of $SU(m)$ HS spin chain (\ref{a1}). 

358: Since it is convenient to apply the freezing trick to 

359: the spin chain Hamiltonian (\ref {b9}), for the rest of this 

360: article we shall deal with  this form of 

361: supersymmetric $SU(m|n)$ HS model 

362: instead of its original form (\ref{b5}).

363: 

364: By using the anyon like representation

365:  $\tilde{P}_{jk}^{(m|n)}$, one can construct a spin CS model like 

366: \beq

367: H^* =-\sum_{j=1}^N \frac{\partial^2}{\partial x_j^2} + 

368: 2a\sum_{1\leq j < k \leq N} \frac{ \big( \, a+\tilde{P}_{jk}^{(m|n)} \, \big) }

369: {\sin^2(x_j-x_k)} \, ,

370: \label{b10}

371: \eeq

372: which contains spin as well as dynamical degrees of freedom and 

373: the positive parameter $a$ as coupling constant.  

374: With the help of mapping (\ref {b7}) it can be shown that, 

375: this spin CS model is equivalent to a supersymmetric 

376: spin CS model [18] with $Y(gl_{(m|n)})$ super-Yangian symmetry.  

377: The spin CS Hamiltonian $H^*$ (\ref {b10}) might be formally written as 

378: \beq

379: H^*=H_0+4a \mathsf{H}^{(m|n)},

380: \label{b11}

381: \eeq

382: where $H_0$ is the Hamiltonian of spinless CS model given by [19] 

383: \beq

384: H_0=-\sum_{j=1}^N \frac{\partial^2}{\partial x_j^2} + 

385: 2a(a-1)\sum_{1\leq j < k \leq N} \frac{1}{\sin^2(x_j-x_k)} \, ,

386: \label{b12}

387: \eeq

388: and 

389: $\mathsf{H}^{(m|n)}$ is obtained from $H^{(m|n)}$ (\ref{b9})

390:  by the replacement

391: $ \xi_j \rightarrow x_j$.

392: Now the decoupling of the dynamical degrees of freedom of $H^*$ (\ref {b10})

393: from its spin degrees of freedom can be achieved by using 

394: the freezing trick [10-12]. This trick is based

395: on the fact that in the limit $a \rightarrow \infty $,

396: particles freeze at the

397: equilibrium positions of $H_0$, which are simply the lattice points ($\xi_j$)

398: of the spin chain in eqn.(\ref{b9}). 

399: Consequently,  by using eqn.(\ref{b11})

400: at freezing limit, we find that the energy levels of $H^*$

401: are approximately given by 

402: \beq

403: E^*_{jk}\simeq E_{0,j}+4aE_k,

404: \label{b13}

405: \eeq

406: where $E_{0,j}$ and $E_k$ are any two levels of $H_0$  

407: and $H^{(m|n)}$ respectively. 

408: Hence, we obtain a relation like 

409: \beq

410: Z^{(m|n)}(T)=\lim_{a\rightarrow \infty} \frac{Z^*(4aT)}{Z_0(4aT)},

411: \label{b14}

412: \eeq

413: where $Z^{(m|n)}$, $ Z^*$ and $Z_0$ denote the partition functions 

414: corresponding to the Hamiltonians

415: $H^{(m|n)}$,  $H^*$ and $H_0$ respectively. Thus the

416:  freezing trick allows us to compute the partition function of

417:  $SU(m|n)$ supersymmetric HS spin chain, by modding out   

418:  the contribution of spinless CS model from the partition function of

419:  spin CS model (\ref{b10}).  Due to the 

420: Gallelian invariance of $H^*$ and $H_0$ it follows that, if 

421: $\psi$ is an eigenstate of any one of these Hamiltonians with momentum $p$, 

422: then $\psi'=e^{2i\tau \sum_{j=1}^N x_j} \psi$ will also be an 

423: eigenstate of the same Hamiltonian with momentum $(p+2 \tau N)$.

424: As a result, we can always adjust the 

425: parameter $\tau$ such that $\psi'$ will be an eigenfunction of

426:  $H^*$ or $H_0$ with zero momentum.

427: In this article, we shall always 

428: consider eigenstates of these Hamiltonians with zero momenta and evaluate the 

429: partition functions $Z^*$ as well as $Z_0$ at the center of mass 

430: frame. Since both $Z^*$ and $Z_0$ get 

431: modified by the same multiplicative factor due to a Gallelian transformation, 

432: $Z^{(m|n)}$ does not depend on the choice of the reference frame. 

433: 

434: 

435: \noindent \section {Spectrum of spin CS model }

436: \renewcommand{\theequation}{3.{\arabic{equation}}}

437: \setcounter{equation}{0}

438: In this section our aim is to find out the complete spectrum of 

439: spin CS model (\ref{b10})

440: containing anyon like representation of the permutation algebra. 

441: Even though the spectrum of such spin CS model has been studied

442: earlier [17],   multiplicities of degenerate eigenfunctions 

443: corresponding to all energy levels 

444: have not been found. Since these numbers are required 

445: for calculating the partition function of this model, 

446: here we want to derive a general expression  

447:  for the degeneracy factors of all energy levels. 

448: It is well known that the eigenfunctions of  spin 

449: CS Hamiltonian (\ref {b10}) can be written in a factorised form like 

450: \beq

451: \psi (x_1,\dots,x_N;\alpha_1,\dots,\alpha_N)

452: =\Gamma^a \phi(x_1,\dots,x_N ; \alpha_1,\dots,\alpha_2),

453: \label{c1}

454: \eeq

455: where $\Gamma=\prod_{i<j} \sin(x_i-x_j)$.

456:  By operating $H^*$ (\ref {b10}) on the above form of $\psi$, 

457:  we find that

458: \beq

459: H^*\psi=\Gamma^a \tilde{H}^* \phi,

460: \label{c2}

461: \eeq

462: where

463: \bea

464: \tilde{H}^*=4\biggl[ \sum_j \biggl( z_j \frac{\partial}

465: {\partial z_j}\biggr) ^2 + 

466: a \sum_{k<j} \frac{z_k+z_j}{z_k-z_j} 

467: \biggl(z_k\frac{\partial}{\partial z_k}-z_j\frac{\partial}

468: {\partial z_j}\biggr) ~~~~~~~~~~\nn \\

469: ~~~~~~~~-2a \sum_{k<j} (1+\tilde{P}^{(m|n)}_{jk}) \frac{z_jz_k}

470: {(z_j-z_k)^2} + \frac{a^2}{12} N(N^2-1)\biggr],

471: \label{c3}

472: \eea

473: with $z_j=e^{2ix_j}$. Equation (\ref {c2}) implies 

474: that,  if $\phi$ is an eigenvector of $\tilde{H}^*$ with 

475:  eigenvalue $E$, then 

476: $\Gamma^a \phi$  would be an eigenvector of $H^*$ with the 

477: same eigenvalue. Thus the diagonalisation problem of 

478:  $H^*$ is reduced to the diagonalisation problem of $\tilde{H}^*$.

479: 

480: For solving $\tilde{H}^*$, it is convenient to introduce 

481: another operator $\mathcal{H}$ 

482: which acts only on the coordinate degree of freedom and

483:  may be given by [7]

484: \bea

485: \mathcal{H}=4\biggl[ \sum_j \biggl( z_j \frac{\partial}

486: {\partial z_j}\biggr) ^2 + a \sum_{k<j} 

487: \frac{z_k+z_j}{z_k-z_j} \biggl(z_k\frac{\partial}

488: {\partial z_k}-z_j\frac{\partial}{\partial z_j}\biggr) ~~~~~~~~~~ \nn \\

489: ~~~~~~~~-2a \sum_{k<j} (1- K_{jk}) \frac{z_jz_k}{(z_j-z_k)^2} 

490: + \frac{a^2}{12} N(N^2-1)\biggr],

491: \label{c4}

492: \eea

493: where the $K_{jk}$ is the coordinate exchange operator 

494: which exchanges the coordinates of $j$-th and $k$-th particle:

495: \beq

496: K_{jk}f(x_1,\dots,x_j,\dots,x_k,\dots,x_N)=

497: f(x_1,\dots,x_k,\dots,x_j,\dots,x_N).\nn

498: \eeq

499: It may be observed that,  $\tilde{H}^*$ (\ref{c3}) 

500: can be reproduced from the expression of 

501: $\mathcal{H}$  (\ref{c4}) through the substitution

502: $K_{jk}\rightarrow -\tilde{P}_{jk}^{(m|n)}$. This 

503: connection between $\tilde{H}^*$ and $\mathcal{H}$ will play a 

504: crucial role in our calculation for finding the spectrum of $\tilde{H}^*$. 

505: Let us first consider state vectors given by monomials like

506: \beq

507: \xi_{\bf{p}}=z_1^{p_1}z_2^{p_2}\dots z_N^{p_N},

508: \label{c5}

509: \eeq

510: where ${\bf p} \equiv \{p_1,p_2,\dots,p_N\} \in \mathbb{R}^N$

511:  satisfies the constraints: 

512: (i) $(p_i-p_j)$ are integers for all $i,j$ and 

513: (ii) $\sum_{i=1}^N p_i=0$. The last condition implies that 

514: these monomials represent state vectors with zero total momentum.  

515: In particular, one can consider 

516: $\xi_{\bf\hat{p}}$

517: corresponding to a nonincreasing vector

518: ${\bf\hat{p}} \equiv (\hat{p}_1,\hat{p}_2,\dots,\hat{p}_N)$,  

519: whose elements satisfy the conditions: 

520: (i) $l_i\equiv \hat{p}_{i}-\hat{p}_{i+1}$ is a nonnegative integer

521: for $i \in [1,\dots,N-1]$ and 

522: (ii) $\sum_{i=1}^N {\hat p}_i=0$. 

523: It is evident that, 

524: $N-1$ number of nonnegative integers ($l_i$'s) are sufficient to specify  

525: a nonincreasing vector ${\bf\hat{p}}$. 

526: Given two distinct nonincreasing vectors 

527: ${\bf\hat{p}}$ and 

528: ${\bf\hat{p}'}$, we shall write ${\bf\hat{p}} \prec {\bf\hat{p}'}$ 

529: if $\hat{p}_1-\hat{p}_1'=\dots=\hat{p}_{i-1}-\hat{p}_{i-1}'=0$ 

530: and $\hat{p}_i<\hat{p}_i'$.

531: A partial ordering can be defined on monomials like 

532: $\xi_{\bf{p}}$ (\ref {c5}) in the following way. 

533: By permuting the elements of any  ${\bf p}$, 

534: one can always construct a  unique nonincreasing vector 

535: ${\bf\hat{p}}$.  

536: The basis element $\xi_{\bf p}$ would precede  $\xi_{\bf p'}$ 

537: if $ {\bf \hat{p}} \prec {\bf \hat{p}'} $, where 

538: $ {\bf \hat{p}}$ and $ {\bf \hat{p}'}$ are 

539:  nonincreasing vectors obtained

540:  from ${\bf p}$ and ${\bf p'}$ respectively by permuting 

541: their components. The above defined ordering is effectively a 

542: partial ordering, since it does not induce an

543: ordering between $\xi_{\bf p}$ and $\xi_{\bf p'}$ 

544: when ${\bf \hat{p}}= {\bf\hat{p}'}$. It can be shown 

545: that the action of $\mathcal{H}$ (\ref {c4}) on the state  

546: vector $\xi_{\bf p}$ yields [7,9]

547: %\begin{displaymath}

548: \beq

549: \mathcal{H } \xi_{\bf p} \, = \,  E_{\bf \hat{p}}  \xi_{\bf p} + 

550: \sum_{\substack{{\bf p'}\\({\bf \hat{p}'}<

551: {\bf \hat{p}})}}c_{{\bf p} {\bf p'}} \xi_{\bf p'} \, ,

552: \label{c6}

553: \eeq

554: %\end{displaymath}

555: where

556: \beq

557:  E_{\bf \hat{p}}=\sum_{i=1}^N \left \{2 \hat{p_i}+a(N+1-2i) \right \}^2.

558: \label{c7}

559: \eeq

560: Thus it is clear that, if one constructs a Hilbert

561: space through basis vectors of the form (\ref{c5}) and 

562: partially order them in the

563: above mentioned way, then $\mathcal{H}$ will act as a 

564: triangular matrix on this space.

565: 

566: Next we want to construct another partially ordered Hilbert space,

567: on which $\tilde{H}^*$ (\ref{c3}) can be represented as a triangular matrix.

568: To this end, we define a set of permutation operators as 

569: $\Pi_{jk}^{(m|n)}=\tilde{P}_{jk}^{(m|n)}K_{jk}$.

570: Since both  $\tilde{P}_{jk}^{(m|n)}$ and $K_{jk}$ 

571: satisfy an algebra of the form (\ref{b4}), while acting 

572: on the spin and coordinate spaces respectively, the newly defined 

573: operator $\Pi_{jk}^{(m|n)}$ also yields 

574: a representation of the same permutation algebra 

575: on the direct product of coordinate and spin spaces.

576: Hence, by using this representation of permutation 

577: algebra, we can construct a `generalized'

578: antisymmetric projection operator $\Lambda^{(m|n)} $ 

579: satisfying the relation 

580: \beq

581: \Pi_{jk}^{(m|n)}\Lambda^{(m|n)}=

582: \Lambda^{(m|n)} \Pi_{jk}^{(m|n)} =-\Lambda^{(m|n)} \, ,

583: \label{c8}

584: \eeq

585: or, equivalently, $\tilde{P}_{jk}^{(m|n)} 

586: \Lambda^{(m|n)}= -K_{jk}\Lambda^{(m|n)}$ [16,17]. 

587: Even though  

588:  $\Lambda^{(m|n)} $ can be expressed as a function of 

589: $\Pi_{jk}^{(m|n)}$,  explicit form of this projection operator

590: is not necessary for our present purpose. However it may be noted that, 

591: since  both $K_{jk}$ and $\tilde{P}_{jk}^{(m|n)}$ 

592: commute with $\mathcal{H}$ (\ref {c4}), 

593: $\Lambda^{(m|n)}$ also satisfies the relation 

594: \beq

595: [\mathcal{H},\Lambda^{(m|n)}]=0 \, .

596: \label{c9}

597: \eeq

598: With the help of projection operator $\Lambda^{(m|n)}$, 

599: we define a state vector on the direct product of coordinate and 

600: spin spaces as 

601: \bea

602: \phi^{\boldsymbol{\alpha}}_{\bf p} \equiv 

603: \phi^{\alpha_1 \dots \alpha_N}_{p_1 \dots p_N}= \Lambda^{(m|n)} 

604: \{ \xi_{\bf p} \v \alpha_1 \dots \alpha_N \r \} .

605: \label{c10}

606: \eea

607: Using  eqns.(\ref{c8}) and (\ref{b6}) it can be shown that

608: \bea

609: \phi^{\alpha_1\dots\alpha_j\dots\alpha_k \dots 

610: \alpha_N}_{ p_1 \dots \,  p_j\dots \,  p_k \dots \, p_N} 

611: \hskip -.56 cm  

612: &&=-\Lambda^{(m|n)} 

613:  K_{jk} {\tilde P}_{jk}^{(m|n)}

614:  \{ z_1^{p_1} \dots z_j^{p_j} \dots z_k^{p_k} \dots z_N^{p_N} 

615: \v \alpha_1 \dots\alpha_j \dots \alpha_k \dots \alpha_N \r \} \nn \\

616: &&= -e^{i \Phi(\alpha_j,\dots,\alpha_k)}\Lambda^{(m|n)}  

617:  \{ z_1^{p_1} \dots z_j^{p_k} \dots z_k^{p_j} \dots z_N^{p_N} 

618: \v \alpha_1 \dots \alpha_k \dots \alpha_j \dots \alpha_N  \r \} \nn \\

619: &&= -e^{i \Phi(\alpha_j,\dots,\alpha_k)} \phi^{\alpha_1 \dots 

620: \alpha_k\dots \alpha_j \dots \alpha_N}_{p_1 \dots \, p_k \dots

621: \, p_j \dots \, p_N} \, .

622: \label{c11}

623: \eea

624: By repeatedly using the above equation we find that 

625: \beq

626: \phi_{\bf p}^{\boldsymbol{\alpha}}= \epsilon ({\boldsymbol{\alpha}}, {\bf p})

627: \, \phi_{\bf \hat{p}}^{\boldsymbol{\alpha'}},

628: \label{c12}

629: \eeq

630: where $ \epsilon ({\boldsymbol{\alpha}}, {\bf p}) = \pm 1$, ${\bf \hat p}$

631: is the nonincreasing vector corresponding to ${\bf p}$ and 

632: ${\boldsymbol{\alpha'}} $ is a spin vector which

633: is obtained by permuting the components of ${\boldsymbol{\alpha}}$. 

634: Hence, all state vectors of the form (\ref {c10}) can be obtained 

635: by choosing ${\bf p}$ from the set of nonincreasing vectors only. 

636: 

637: Corresponding to any given nonincreasing vector ${\bf \hat{p}}$, 

638: one can define a vector space as

639: \beq

640: \mathbb{V}_{\bf \hat{p}} \equiv 

641: \l \, \phi_{\bf \hat{p}}^{\boldsymbol{\alpha}}~~|

642: \, \alpha_1 , \dots ,\alpha_N \in [1,2, \dots m+n] \, \r \,  . 

643: \label{c13}

644: \eeq

645: It is important to note that,  different values of 

646: ${\boldsymbol{\alpha}}$ may lead to the same 

647: $\phi_{\bf \hat{p}}^{\boldsymbol{\alpha}}$ which is a 

648: basis element of $\mathbb{V}_{\bf \hat{p}} $.

649:  To see this thing in a simple way, let us take a nonincreasing

650:  sequence ${ \bf \hat{p}}$ satisfying the condition  

651: $ \hat{p}_i = \hat{p}_j = \hat{p}$ (say). 

652: For this special case, eqn.(\ref{c11}) reduces to 

653: \beq

654: \phi^{\alpha_1 \dots \alpha_i \dots \alpha_j \dots \alpha_N}_{ \,

655: \hat{p}_1 \,  \dots \, \hat{p} \, \dots \, \hat{p} \, \dots \,  \hat{p}_N }

656:  = -\, e^{i \Phi (\alpha_i,\dots ,\alpha_j)} \,

657: \phi^{\alpha_1 \dots \alpha_j \dots \alpha_i \dots \alpha_N}_{ \, 

658: \hat{p}_1 \,  \dots \, \hat{p} \, \dots \, \hat{p} \, \dots \,  \hat{p}_N }\, .

659: \label{c14}

660: \eeq

661: Clearly 

662: $\phi^{\alpha_1 \dots \alpha_i \dots \alpha_j \dots \alpha_N}_{ \, 

663: \hat{p}_1 \,  \dots \, \hat{p} \, \dots \, \hat{p} \, \dots \,  \hat{p}_N }$

664: and $\phi^{\alpha_1 \dots \alpha_j \dots \alpha_i \dots \alpha_N}_{ \,  

665: \hat{p}_1 \,  \dots \, \hat{p} \, \dots \, \hat{p} \, \dots \,  \hat{p}_N }$

666: represent the same state vector (up to a phase factor), 

667:  although they correspond to different values of ${\boldsymbol{\alpha}}$.

668: For a given $\phi_{\bf \hat{p}}^{\boldsymbol{\alpha}}$, 

669: we say that two spin components of 

670: ${\boldsymbol{\alpha}}$ belong to the same `sector' if 

671: the corresponding two components of ${ \bf \hat{p}}$ are equal to 

672: each other. For example, the spin components 

673: $\alpha_i$ and $ \alpha_j$ appearing in the state

674: $\phi^{\alpha_1 \dots \alpha_i \dots \alpha_j \dots \alpha_N}_{ \, 

675: \hat{p}_1 \,  \dots \, \hat{p} \, \dots \, \hat{p} \, \dots \,  \hat{p}_N }$ 

676: belong to the same sector according to this convention. 

677: Since $e^{i \Phi}=1$, for $\alpha_i,\alpha_j 

678: \in [1,2,\dots,m]$, it is clear from eqn.(\ref{c14})

679: that bosonic spins within the same sector 

680: obey `fermionic statistics' after antisymmetrisation. In particular, 

681: two bosonic spins of same flavour can not coexist within a single sector. 

682: Similarly, since $e^{i \Phi}=-1$ for $\alpha_i,

683: \alpha_j \in [m+1,m+2,\dots,m+n]$,

684: one can find from eqn.(\ref{c14}) that fermionic spins within the 

685: same sector obey `bosonic statistics' after antisymmetrisation.

686: Therefore, any number of fermionic spins having the same flavour can

687: be accommodated within a single sector. 

688: 

689: Now we want to find out the dimensionality of the space 

690: $\mathbb{V}_{\bf \hat{p}}$. For this purpose, it is useful to write

691: ${\bf \hat{p}}$ in the form 

692: \beq

693: \bf{\hat{p}} \equiv \big ( \overbrace{\rho_1,\dots,\rho_1}^{k_1},

694: \dots,\overbrace{\rho_i,\dots,\rho_i}^{k_i},

695: \dots,\overbrace{\rho_r,\dots,\rho_r}^{k_r} \big ),

696: \label{c15}

697: \eeq 

698: where $\rho_1>\dots>\rho_i>\dots>\rho_r$, $\sum_{i=1}^r k_i=N$, 

699: and $r$ is an integer which can take any value from $1$ to 

700: $N$. It is obvious that  ${\bf k} \equiv \{ k_1,\dots,k_r\}$, which  

701: belongs to the set $\mathcal{P}_N$ of ordered partitions of $N$, 

702:  may be treated as a function of 

703: $\bf{\hat{p}}$. For a given $\phi_{\bf \hat{p}}^{\boldsymbol{\alpha}}$,

704:  clearly the components of $\boldsymbol{\alpha}$ are 

705:  separated into $r$ different sectors where the 

706: $i$-th sector contains $k_i$ number of spins. 

707: It is evident that the dimensionality of the space 

708: $\mathbb{V}_{\bf \hat{p}}$ may be obtained by counting 

709: the number of independent ways one can distribute total 

710: $N$ number of spins within $r$ sectors. To this end, let us first 

711: try to find out the number of independent ways of filling up the $i$-th sector 

712: through $j_1$ number of bosonic spins and $j_2$ number of fermionic  

713: spins, where $j_1+j_2=k_i$. Using eqn.(\ref{c14}) we have already seen that, 

714: bosonic and fermionic spins within the same sector obey fermionic and 

715: bosonic statistics respectively. Therefore, we can pick up 

716: $j_1$ number of bosonic spins

717: from $m$ different flavours in $^m C_{j_1}$ different 

718: ways and $j_2$ number of fermionic spins

719: from $n$ different flavours in $^{j_2+n-1}C_{j_2}$ ways,  where $^p 

720: C_l=\frac{p\, !}{l\hskip .01 cm ! \, (p-l)!} $  for $l \leq p$ and 

721:  $^{p}C_l=0$ for $l>p $.  

722: Thus the number of independent ways of 

723: filling up the $i$-th sector through $j_1$ number of bosonic spins and 

724: $j_2$ number of fermionic spins is given by 

725: $$

726: ^mC_{j_1} \, ^{n+j_2-1}C_{j_2} \, .

727: $$

728: Summing up these numbers for all possible values of 

729: $j_1$ and $j_2$, we obtain the total number of independent ways 

730: of filling up the $i$-th sector through  $k_i$ number of spins 

731: as

732: \bea

733: d^{(m|n)}(k_i)  

734: &=& \sum_{j_1= ~0}^{k_i} 

735:  \big(\,  ^mC_{j_1} \, ^{n+k_i-j_1-1}C_{k_i-j_1} \, \big)  \nn \\

736: &=& \sum_{j_1= ~0}^{{\rm min} \, (m,k_i)} 

737: \frac{m!~(n+k_i -j_1 -1)!}{j_1! ~(m-j_1)!~(n-1)!~(k_i-j_1)!}\, .

738: \label{c16}

739: \eea

740: Since two spins belonging to different sectors

741: do not follow any exchange relation like (\ref {c14}),

742: the number of independent ways we can distribute total 

743: $N$ number of spins within $r$ different sectors is

744: given by the product of all $d^{(m|n)}(k_i)$. 

745: Therefore, by using  (\ref{c16}), we finally obtain 

746: the dimension of $\mathbb{V}_{\bf \hat{p}}$ as 

747: \beq

748: d^{(m|n)}({\bf k})=\prod_{i=1}^r 

749: d^{(m|n)}(k_i)=\prod_{i=1}^r 

750:  \left( \sum_{j=0}^{{\rm min}\, (m,k_i)} \,

751:    {}^mC_j \, {}^{n+k_i-j-1}C_{k_i-j}  \right)  \, .

752: \label{c17}

753: \eeq

754: Even though this expression is derived by assuming that bosonic and 

755: fermionic spin degrees of freedom (i.e., $m$ and $n$ respectively) take 

756: nonzero values, it is also possible 

757: to obtain the dimension of $\mathbb{V}_{\bf \hat{p}}$ for the

758: $SU(n)$ fermionic case by putting $m=0$ in eqn.(\ref{c17}):

759: \beq

760: d^{(0|n)}({\bf k})= \prod_{i=1}^r 

761: ~^{n+k_i-1}C_{k_i} \, .

762: \label{c18}

763: \eeq

764: Furthermore, by putting $n=0$ in 

765: eqn.(\ref{c17}), subsequently using the relation $^{p}C_l=0$ for $l>p\geq 0$,  

766: and also assuming that $^{-1}C_0=1$, 

767: one can reproduce the

768: dimension of $\mathbb{V}_{\bf \hat{p}}$ for the $SU(m)$ 

769: bosonic case [9] as

770: \beq

771: d^{(m|0)}({\bf k})= \prod_{i=1}^r 

772: ~^{m} C_{k_i} .

773: \label{c19}

774: \eeq

775: It is interesting to observe that, while 

776: $d^{(m|0)}({\bf k})$ (\ref {c19}) can take a nonzero value 

777: only if $k_i \leq m $ for all $i$, both 

778: $d^{(m|n)}({\bf k})$ (\ref{c17}) and 

779: $d^{(0|n)}({\bf k})$ (\ref{c18}) take nonzero values 

780: for any ${\bf k} \in \mathcal{P}_N$. Consequently, 

781: $\mathbb{V}_{\bf \hat{p}}$ will represent a nontrivial vector space

782: for the $SU(m)$ bosonic case only if at most $m$ components of 

783: ${\bf \hat{p}}$ take the same value. On the other hand, 

784: $\mathbb{V}_{\bf \hat{p}}$ will represent a nontrivial vector space

785: for all possible values of ${\bf \hat{p}}$ when at least one 

786: fermionic spin degrees of freedom is present.  

787: 

788: The Hilbert space associated with 

789: $\tilde{H}^*$ (\ref{c3}) may now be defined by taking the direct sum of

790:  $\mathbb{V}_{\bf \hat{p}}$ (\ref {c13})  for all allowed values of 

791: ${\bf \hat{p}}\, $: 

792: \beq 

793: \mathbb{V} 

794: = \mathop{\oplus}_{\bf\hat{p}} \mathbb{V}_{\bf \hat{p}} \, .

795: \label{c20}

796: \eeq

797: We define a partial ordering in this Hilbert space by saying that 

798: the basis element 

799: $\phi_{\bf \hat{p}}^{\boldsymbol{\alpha}}$

800: precedes  

801: $\phi_{\bf \hat{p'}}^{\boldsymbol{\alpha'}}$ if 

802: $ {\bf \hat{p}} \prec {\bf \hat{p}'} $. 

803: By consecutively applying the relations (\ref{c10}), (\ref{c8}),  

804: (\ref{c9}), (\ref{c6}) and (\ref{c12}), it is easy to check that

805: \bea

806: \tilde{H}^* (\phi^{\boldsymbol{\alpha}}_{\bf \hat{p}})

807: &=&\Lambda^{(m|n)} \, \mathcal{H} \, \xi_{\bf \hat{p}} 

808: \v \alpha_1 \dots \alpha_N \r \nn \\

809: &=&\Lambda^{(m|n)} \biggl( E_{\bf \hat{p}}  \xi_{\bf \hat{p}} + 

810: \sum_{\substack{{\bf p'}\\({\bf \hat{p}'}

811: <{\bf \hat{p}})}}c_{{\bf \hat{p}}

812:  {\bf p'}} \,  \xi_{\bf p'}\biggl) \v \alpha_1 

813: \dots \alpha_N \r \nn \\

814: &=&E_{\bf \hat{p}}  

815: \phi_{\bf \hat{p}}^{\boldsymbol{\alpha}} \,  + \, 

816: \sum_{\substack{{\bf p'}\\({\bf \hat{p}'}

817: <{\bf \hat{p}})}}

818: \epsilon ({\boldsymbol{\alpha}}, {\bf p'}) \,

819: c_{{\bf \hat{p}} {\bf p'}} \,

820: \phi_{\bf \hat{p}'}^{\boldsymbol{\alpha'}} \, .

821: \label{c21}

822: \eea

823: Hence  $\tilde{H}^*$ is represented as a triangular matrix on  

824: $\mathbb{V}$. Diagonal elements of this triangular matrix yield

825: the eigenvalues of $\tilde{H}^*$ as  

826: \beq

827: E^*({\bf \hat{p}},\boldsymbol{\alpha}) \equiv 

828: E_{\bf \hat{p}}= \sum_{i=1}^N \big\{  2 \hat{p}_i +a(N+1-2i)  \big\}^2.

829: \label{c22}

830: \eeq

831: Consequently, the eigenvalues of spin CS Hamiltonian 

832: $H^*$ (\ref {b10}) are also given by 

833: $E^*({\bf \hat{p}},\boldsymbol{\alpha})$ in the above equation.

834: 

835: Since $E^*({\bf \hat{p}},\boldsymbol{\alpha})$ in eqn.(\ref{c22})

836: does not really depend on the spin vector $\boldsymbol{\alpha}$, 

837: the number of degenerate energy eigenstates associated with the

838:  quantum number ${\bf \hat{p}}$ would coincide with the dimension of the 

839: space $\mathbb{V}_{\bf \hat{p}} $. Thus the degeneracy factor 

840: of the energy eigenvalue 

841:  $E^*({\bf \hat{p}},\boldsymbol{\alpha})$ corresponding to the 

842:  quantum number ${\bf \hat{p}}$ is given by 

843:  $d^{(m|n)}({\bf k})$ appearing in eqn.(\ref{c17}). 

844: We have already seen that, in contrast to the pure bosonic case, 

845:  $d^{(m|n)}({\bf k})$ takes nonzero values

846:  for all possible ${\bf k} \in \mathcal{P}_N$

847: when at least one fermionic spin degrees of freedom is present. 

848: Consequently, the presence of fermionic spin degrees of freedom in 

849:  $H^*$ (\ref {b10}) would lead to a spectrum with 

850: many additional energy levels in comparison with

851: the spectrum of bosonic spin CS model. 

852: 

853: Finally let us briefly comment about the known spectrum

854: of spinless CS Hamiltonian $H_0$ (\ref {b12}) [19].

855: Using the fact that the eigenfunctions of $H_0$ 

856: can be written in a factorised form like

857: $ \psi_0=\Gamma^a \phi_0 (x_1,\dots,x_N)$,

858: it is possible to transform $H_0$ 

859: into $\tilde{H}_0$ as

860: $$

861: H_0 \psi_0=\Gamma^a \tilde{H}_0 \phi_0 \, ,

862: $$

863: where $\tilde{H}_0$ can be obtained from

864: $\mathcal{H}$ (\ref{c4}) through

865: the substitution $K_{ij} \rightarrow 1$.

866: For constructing the Hilbert space associated with 

867: $\tilde{H}_0$, one may consider elements like 

868: $ \phi_{\bf p} \equiv \Lambda_0 (\xi_{\bf p})$, 

869: where $ \Lambda_0$ is the symmetriser

870: in the coordinate space: $K_{jk}\Lambda_0 =  \Lambda_0$.

871: Since $\phi_{\bf p}=\phi_{\bf \hat{p}}$, where ${\bf \hat{p}}$ 

872: is the nonincreasing vector corresponding to 

873: ${\bf p}$, the Hilbert space of $\tilde{H}_0$ is defined 

874: through independent basis vectors $\phi_{\bf \hat{p}}$ for 

875: all values of ${\bf \hat{p}}$. 

876: An ordering can be defined among these state vectors by saying that 

877: $\phi_{\bf \hat{p}}$ precedes  

878: $\phi_{\bf \hat{p}'}$ if $ {\bf \hat{p}} \prec {\bf \hat{p}'} $. 

879: Using eqn.(\ref{c6}) it can be shown that, 

880: $\tilde{H}_0$ acts as a triangular matrix on these completely ordered 

881: basis vectors and the eigenvalues of $H_0$ are also 

882: given by $E_{\bf \hat{p}}$ in eqn.(\ref{c22}). However, 

883: due to the absence of spin degrees of freedom, only one energy eigenstate 

884: is obtained corresponding to each quantum number ${\bf {\hat p}}$

885: in this case.

886: 

887: \noindent \section {Partition function of $SU(m|n)$ HS spin chain }

888: \renewcommand{\theequation}{4.{\arabic{equation}}}

889: \setcounter{equation}{0}

890: 

891: 

892: By using the freezing trick we have seen that, 

893: the partition function of supersymmetric $SU(m|n)$ HS spin chain

894: can be obtained by dividing the partition function of spin CS model (\ref {b10}) 

895: at the strong coupling limit through that of the spinless CS model 

896: (\ref {b12}).  To execute this programme, 

897: let us first briefly recapitulate the calculation for the partition 

898: function of spinless CS model (\ref{b12}) at $a \rightarrow \infty$ limit [9].

899: It should be noted that,  the eigenvalues in eqn.(\ref{c22}) 

900:  can be expanded in powers of $a$ as

901: \beq

902: E^*({\bf \hat{p}},\boldsymbol{\alpha}) \equiv 

903: E({\bf \hat{p}})=a^2E_0+4a \sum_{i=1}^N (N+1-2i)\hat{p}_i +O(1),

904: \label{d1}

905: \eeq

906: where $ E_0=\frac{1}{3} N(N^2-1) $ .

907: Since $E_0$ does not depend on ${\bf\hat{p}}$ or 

908: $\boldsymbol{\alpha}$, the effect of this $E_0$  will be manifested as the same

909: overall multiplicative factor in the partition functions of 

910: spin CS model and its spinless counterpart. 

911: Hence, by dropping the first term

912: in eqn.(\ref{d1}), and neglecting the $O(1)$ term in the limit

913: $a \rightarrow \infty $, one can write down 

914: the partition function of spinless CS model (\ref{b12}) as

915: \beq

916: Z_0(4aT) \simeq \sum_{{\bf \hat{p}} } q^{\sum_i \hat{p}_i(N+1-2i)} \, ,

917: \label{d2}

918: \eeq

919: where $q=e^{-1/(k_BT)}$. Using $N-1$ number of nonnegative integers

920: ($l_i$'s) which uniquely determine ${{\bf\hat{p} }}$, 

921: one can further simplify this partition function as [9]

922: \beq

923: Z_0(4aT)  \simeq \sum_{l_1, \dots , l_{N-1} 

924: \geq 0} \prod_{j=1}^{N-1} q^{j(N-j)l_j}

925:  = \prod_{j=1}^{N-1}\frac{1}{ 1- q^{j(N-j)}}.

926: \label{d3}

927: \eeq

928: 

929: Next, we want to calculate the partition function 

930: of spin CS Hamiltonian (\ref{b10}) at $a \rightarrow \infty $ limit.

931: Dropping again the first term as well as $O(1)$ term 

932: from the right hand side of expansion (\ref{d1}), 

933: and expressing the nonincreasing vector 

934: ${\bf \hat{p}}$ through eqn.(\ref{c15}), 

935: $E^*({\bf \hat{p}},\boldsymbol{\alpha})$ can be written as 

936: \beq

937: E^*({\bf \hat{p}},\boldsymbol{\alpha}) \simeq 4a \sum_{i=1}^r

938:  \rho_i \sum _{j=K_{i-1}+1}^{K_i} (N+1-2j),

939: \label{d4}

940: \eeq

941: where $K_i=\sum_{j=1}^i k_j$ denote the partial 

942: sums corresponding to the partition 

943: ${\bf k} \in \mathcal{P}_N$ and $K_0=0$.

944: Using a set of 

945: variables like $\nu_j \equiv \rho_j-\rho_{j+1}$ for

946: $j \in [ 1,\dots,r-1 ] $ (since $\rho_j > \rho_{j+1}$,

947: all $\nu_j$'s are positive integers), one can express the

948: energy eigenvalue in eqn.(\ref{d4}) as 

949: \beq

950: E^*({\bf \hat{p}},\boldsymbol{\alpha})

951:  \simeq 4a\sum_{j=1}^{r-1} \nu_j N_j,

952: \label{d5}

953: \eeq

954: where $N_j=K_j(N-K_j)$. It may be noted that, 

955: due to the condition $\sum_{i=1}^N \hat{p}_i =0$,  $r-1$ number of

956: $\nu_j$'s uniquely determine the nonincreasing vector

957: ${\bf \hat{p}}$  in eqn.(\ref{c15}).  Consequently, the single sum 

958: $\sum_{\bf\hat{p}}$  can be replaced by the double sum 

959: $\sum_{\mathbf{k} \in ~\mathcal{P}_N} 

960: \sum_{\nu_1,\dots,\nu_{r-1}>0}$ in the expression of the partition 

961: function.  By using the eigenvalue relation (\ref{d5}) and the  

962: corresponding degeneracy factor $d^{(m|n)}({\bf k})$ (\ref {c17}), 

963: we obtain the partition function 

964: of spin CS Hamiltonian (\ref{b10}) at $a \rightarrow \infty $ limit as

965: \bea

966: Z^*(4aT)& \simeq & \sum _{\bf\hat{p}}

967: d^{(m|n)}({\bf k})

968:  \prod_{j=1}^{r-1} q^{N_j \nu_j} \nn \\

969: &=& \sum _{\mathbf{k} \in 

970: ~\mathcal{P}_N} d^{(m|n)}({\bf k})

971: \sum_{\nu_1,\dots,\nu_{r-1}>0}

972:  \prod_{j=1}^{r-1} q^{N_j \nu_j} \nn \\

973: &=&\sum _{\mathbf{k} \in 

974: ~\mathcal{P}_N} d^{(m|n)}({\bf k}) 

975: \prod_{j=1}^{r-1} \frac{q^{N_j}}{1-q^{N_{j}}} ~ .

976: \label{d6}

977: \eea

978: Using eqns.(\ref{b14}), 

979:  (\ref{d3}) and (\ref{d6}), we finally obtain the partition function of 

980: $SU(m|n)$ HS spin chain as

981: \beq

982: Z^{(m|n)}(T)=\prod_{l=1}^{N-1} \left(1- q^{l(N-l)} \right)

983: \sum _{\mathbf{k} \in 

984: ~\mathcal{P}_N} d^{(m|n)}({\bf k}) 

985: \prod_{j=1}^{r-1} \frac{q^{N_j}}{(1-q^{N_{j}})}.

986: \label{d7}

987: \eeq

988: Since the partial sums $ K_1, K_2,  \dots , K_{r-1}$ 

989: associated with  ${\bf k}$

990: are natural numbers obeying $1\leq K_1< \dots<K_{r-1}\leq N-1$, 

991: one can define their complements ($K_i' \,$'s) as elements of the set: 

992: $ \{1,\dots,N-1\}-\{K_1,\dots ,K_{r-1}\}$,  

993: which satisfy the ordering $ K'_1<\dots<K'_{N-r}$.

994: Hence one can rearrange the product

995:  $\prod_{l=1}^{N-1} (1-q^{l(N-l)})$ into two terms as [9]

996: \beq

997: \prod_{l=1}^{N-1} (1-q^{l(N-l)})= 

998: \prod_{j=1}^{r-1} (1-q^{N_j})\prod_{i=1}^{N-r}(1-q^{N_i'}),

999: \label{d8}

1000: \eeq

1001: where $N_i'=K_i'(N-K_i')$. By substituting this relation to eqn.(\ref{d7}),

1002: we get a simplified expression for the

1003: partition function of $SU(m|n)$ HS model as

1004: \beq

1005: Z^{(m|n)}(T)=\sum _{\mathbf{k}

1006: \in ~\mathcal{P}_N} d^{(m|n)} ({\bf k})~q^{\sum

1007: \limits^{r-1}_{j=1} N_j}\prod_{i=1}^{N-r}(1-q^{N'_{i}}) \, .

1008: \label{d9}

1009: \eeq

1010: We have already seen that,  both 

1011: $d^{(m|n)}({\bf k})$ (\ref {c17}) and 

1012: $d^{(0|n)}({\bf k})$ (\ref {c18}) take nonzero values 

1013: for any $\bf{k} \in \mathcal{P}_N$.

1014: Consequently, in contrast to the restricted choice of 

1015: $\bf{k}$ for the case of bosonic spin chain [9], 

1016: all possible $\bf{k} \in \mathcal{P}_N$

1017: will contribute to the partition function (\ref{d9}) in the 

1018: case of supersymmetric as well as fermionic HS spin chain.

1019: 

1020: It is well known that the spectrum of bosonic $SU(m)$ HS spin chain 

1021: (\ref {a1}) containing $N$ number of lattice sites can be obtained from motifs 

1022: like $\delta \equiv (0, \delta_1, \dots ,\delta_{N-1}, 0)$, 

1023: where each $\delta_j$ is either $0$ or $1$ [6-8]. The form of these 

1024: motifs and corresponding eigenvalues can be reproduced by using

1025: the partition function of bosonic $SU(m)$ HS spin chain [9]. 

1026: Now we want to explore how the motifs associated with $SU(m|n)$

1027: supersymmetric HS spin chain emerge naturally from the expression 

1028: of partition function (\ref{d9}). To this end, 

1029: we define a motif corresponding to the partition ${\bf k}$ 

1030: by using the following rule: $\delta_j=0$ if

1031: $j$ coincides with one of the partial sums $K_i$ and $\delta_j=1$ otherwise. 

1032: Furthermore, it is assumed that 

1033: the lowest power of $q$ in eqn.(\ref{d9}) for the partition 

1034: ${\bf k}$ gives the energy eigenvalue $E(\delta)$ of the above motif $\delta$.

1035: In this way we obtain the energy levels of $SU(m|n)$ 

1036: supersymmetric HS spin chain as 

1037: \beq

1038: E(\delta)=

1039: \sum_{i=1}^{r-1}N_i =

1040: \frac{N(N^2-1)}{6} + \sum_{j=1}^{N-1} \delta_j ~j(j-N)\, ,

1041: \label{d10}

1042: \eeq

1043: which apparently coincides with that of the bosonic 

1044: HS spin chain. However it should be observed that, 

1045: for the case of $SU(m)$ spin chain, only those 

1046: ${\bf k}$ would contribute in the 

1047: partition function for which $K_{j}-K_{j-1} = k_j \leq m$ [9].

1048: This leads to a selection rule which prohibits the occurrence of 

1049: $m$ or more consecutive 1's within the corresponding motifs. 

1050: On the other hand, since all $\bf{k} \in \mathcal{P}_N$

1051: contribute to the partition function (\ref{d9})

1052: of supersymmetric HS spin chain, 

1053: it is possible to place any number of 

1054: consecutive 1's or 0's within a motif $\delta$. Consequently, 

1055: the selection rule occurring in the bosonic case is lifted for the case 

1056: of supersymmetric HS spin chain and 

1057: many extra energy levels appear in the corresponding spectrum. 

1058: This absence of selection rule 

1059: in the spectrum of supersymmetric HS spin chain was previously 

1060: observed by Haldane on the basis of numerical calculations [3]. 

1061: By using the expression of $E(\delta)$ in eqn.(\ref{d10}), we can easily 

1062: evaluate the maximum and minimum energy eigenvalues of this system. 

1063: From the expression of 

1064: $E(\delta)$ it is evident that,  the motif 

1065: $\delta \equiv (0,0,\dots,0,0)$ would correspond to the maximum 

1066: energy $E_{max}= \frac{N(N^2-1)}{6}$.

1067:  Similarly for the motif $\delta \equiv (0,1,\dots,1,0)$, we 

1068: obtain the minimum energy of the system as

1069: $E_{min}= \frac{N(N^2-1)}{6} + \sum_{j=1}^{N-1} j(j-N)= 0$.

1070: It is interesting to note that these maximum and minimum 

1071: energy eigenvalues of $SU(m|n)$ supersymmetric HS spin chain 

1072: do not depend on the values of $m$ and $n$.  

1073: Moreover, the lifting of the selection rule 

1074: is responsible for the zero minimum energy of supersymmetric 

1075: HS spin chain.

1076: 

1077: Using Mathematica we find that,  for a wide range of 

1078: values of $m$, $n$ and $N$,  the partition function (\ref{d9}) of $SU(m|n)$ 

1079: HS model satisfies a duality relation of the form

1080: \beq

1081: Z^{(m|n)}(q)=q^{\frac{N(N^2-1)}{6}}Z^{(n|m)}(q^{-1}).

1082: \label{d11}

1083: \eeq

1084: This result motivates us to conjecture that the above duality relation, 

1085: involving the interchange of bosonic and fermionic spin degrees of freedom,

1086: is valid for all possible values of $m$, $n$ and $N$. It may be noted that, 

1087:  for the particular case $n=0$, eqn.(\ref{d11})

1088:  relates the partition function of  $SU(m)$

1089:  bosonic HS spin chain to that of $SU(m)$ fermionic spin chain. 

1090: By applying the relation 

1091: ${\tilde P}_{jk}^{(m|0)}= -{\tilde P}_{jk}^{(0|m)}$ and 

1092: the summation formula 

1093: $\sum_{1\leq j <k \leq N} \frac{1}{\sin^2( \xi_j- \xi_k)}=

1094: \frac{N(N^2-1)}{6} $ [9,20], we find that 

1095:  the Hamiltonians of bosonic and fermionic spin chains are connected as 

1096: $$

1097: H^{(m|0)}= \frac{N(N^2-1)}{6} -  H^{(0|m)}  \, .

1098: $$ 

1099: Using the above relation along with the definition of partition function 

1100: given by $Z^{(m|n)}(q)=tr[q^{H^{(m|n)}}]$,

1101: one can easily prove eqn.(\ref{d11}) for the particular case $n=0$. It would 

1102: be interesting  to explore whether eqn.(\ref{d11}) can be also proved

1103: for the general case  by establishing some relation between 

1104: ${\tilde P}_{jk}^{(m|n)}$ and ${\tilde P}_{jk}^{(n|m)}$.

1105: Comparing the coefficients of 

1106: same power of $q$ from both sides of eqn.(\ref{d11}), we find that 

1107: the energy levels of $SU(n|m)$ spin chain can be obtained from those of 

1108: $SU(m|n)$  spin chain through the transformation 

1109: $E_i \rightarrow \frac{N(N^2-1)}{6}-E_i $ and also get the relation 

1110: \beq

1111: \mathcal{D}^{(m|n)}\left(E_i\right)

1112: =\mathcal{D}^{(n|m)} \left( \frac{N(N^2-1)}{6}-E_i \right),

1113: \label{d12}

1114: \eeq

1115: where $\mathcal{D}^{(m|n)}(E_i)$ denotes 

1116: the degeneracy factor corresponding to energy $E_i$ of $SU(m|n)$

1117: HS spin chain.  Thus it is evident that, 

1118: the spectrum of $SU(n|m)$ spin chain can be

1119: obtained from that of $SU(m|n)$ spin chain through 

1120: an inversion and overall shift of all energy levels.

1121: Such relation between the spectra of 

1122: supersymmetric HS spin chains was empirically found 

1123: by Haldane with the help of numerical analysis [3].

1124: 

1125: 

1126: \noindent \section {Spectral properties of $SU(m|n)$ HS spin chain }

1127: \renewcommand{\theequation}{5.{\arabic{equation}}}

1128: \setcounter{equation}{0}

1129: In this section we shall explore some spectral properties 

1130: of supersymmetric HS model by 

1131: using its exact partition function $Z^{(m|n)}(T)$ (\ref{d9}). 

1132: It has been already mentioned that, 

1133: calculation for the degeneracy factors associated with the energy 

1134: eigenvalues of this spin chain becomes very cumbersome by using the 

1135: motif representations for large values of $N$. However, with the help of

1136: a symbolic software package like Mathematica, it is possible to 

1137: express the partition function (\ref{d9}) as a polynomial of $q$ and

1138: explicitly find out the degeneracy factors of all energy levels for relatively

1139: large values of $N$. In this way,  we can study 

1140: properties like level density distribution and  

1141: nearest-neighbour spacing (NNS) distribution for the spectrum 

1142: of supersymmetric HS spin chain.

1143: 

1144: For the case of $SU(m)$ bosonic spin chain, it has been found earlier

1145: that the continuous part of the energy level density obeys

1146: Gaussian distribution to a very high degree of accuracy for $N>>1$ [9]. 

1147: At present, our aim is to study the level density distribution in the spectrum

1148: of $SU(m|n)$ supersymmetric HS spin chain and investigate whether it 

1149: exhibits a similar behaviour. To begin with, let us consider

1150: the simplest case of $SU(1|1)$ supersymmetric HS spin chain.  

1151: In this case,  the degeneracy factor in eqn.(\ref{c17})

1152: reduces to a simple form given by 

1153: $ d^{(1|1)}({\bf k})= 2^r$. 

1154: By substituting this degeneracy factor to eqn.(\ref{d9}), taking some  

1155: specific value for the number of lattice sites like $N=15$ and

1156: using Mathematica, we express the partition 

1157: function of $SU(1|1)$ spin chain as a polynomial of $q$. The 

1158: coefficient of $q^{E_i}$ in such polynomial evidently gives the

1159: degeneracy factor  $\mathcal{D}^{(1|1)}(E_i)$ corresponding to the 

1160: energy eigenvalue $E_i$, which we plot in Fig.1. This figure

1161: clearly indicates that the energy level distribution 

1162: obey Gaussian approximation but with some local fluctuations. 

1163: Similar behaviour of energy level distribution has been found by

1164: studying $SU(m|n)$ HS spin chain with other values of $m,~n$ and sufficiently

1165: large values of $N$. 

1166: 

1167: From the above discussion it is apparent that, if we decompose 

1168: the energy level density associated with $SU(m|n)$ HS spin chain 

1169: as a sum of continuous part and fluctuating part, the continuous part will 

1170: obey Gaussian distribution for large values of $N$. 

1171: This behaviour of the continuous part

1172: can be measured in a quantitative way

1173:  by studying the cumulative level density [9], 

1174:  which eliminates the fluctuating part of the level density distribution. 

1175: For the case of $SU(m|n)$ HS spin chain, 

1176: cumulative level density of the spectrum is defined as

1177: \beq

1178:  F(E)=\frac{1}{(m+n)^N} \sum_{E_i\leq E} \mathcal{D}^{(m|n)} (E_i) \, .

1179: \label {e1}

1180: \eeq

1181: Obviously, this $F(E)$ can also be obtained by expressing 

1182: the exact partition function (\ref{d9}) as a polynomial of $q$. 

1183: We want to check whether this $F(E)$ agrees well  

1184: with the error function given by

1185: \beq

1186: G(E)=\frac{1}{2}\left[1+erf\left(\frac{E-\mu}{\surd{2} \sigma 

1187: }\right)\right],

1188: \label{e2}

1189: \eeq

1190: where $\mu $ and $\sigma $

1191: are respectively the mean value and the standard deviation associated with the 

1192: energy level density distribution. 

1193: These parameters are related to the Hamiltonian $H^{(m|n)}$

1194: (\ref{b9}) as 

1195: \bea

1196: ~~~~~~~~~~~~~~~~~~~~\mu=\frac{tr\left[ 

1197: H^{(m|n)}\right]}{{(m+n)}^N}\,, ~~~~~\sigma^2=  \frac{tr\left[(H^{(m|n)})^2 

1198: \right]}{{(m+n)}^N} \, - \, \mu^2. 

1199: ~~~~~~~~~~~~~~~~~~~~~(5.3a,b) \nn

1200: \eea

1201: 

1202: For the purpose of comparing $F(E)$ with $G(E)$,  it is 

1203: necessary to express the parameters $\mu$ and $\sigma$ as some 

1204: functions of $m$, $n$ and $N$. 

1205: To this end, we need the following trace formulas:

1206: \bea

1207: \begin{aligned}

1208: ~~~& tr\left[ 

1209: (\tilde{P}_{ij}^{(m|n)})^2\right]=tr\left[ \one \right]=

1210: s^N,~~~~tr\left[ \tilde{P}_{ij}^{(m|n)} 

1211: \right]=s^{N-2}t \, ,

1212: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~(5.4 a,b ) \nn \\   

1213: ~~~&tr \left[ \tilde{P}_{ij}^{(m|n)}\tilde{P}_{il}^{(m|n)} \right] 

1214: =tr\left[ \tilde{P}_{ij}^{(m|n)}

1215: \tilde{P}_{jl}^{(m|n)}\right] =tr\left[ 

1216: \tilde{P}_{ij}^{(m|n)}\tilde{P}_{kj}^{(m|n)}\right]=s^{N-2},

1217:  ~~~~~~~~~~~~~~~~(5.4 c )  \nn \\

1218: ~~~&tr\left[\tilde{P}_{ij}^{(m|n)}\tilde{P}_{kl}^{(m|n)}\right]=s^{N-4}t^2 ,

1219: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~(5.4d) \nn

1220: \end{aligned}

1221: \eea

1222: where $s=m+n$, $t=m-n$ and $i,j,k,l$ are all different indices. 

1223:  Derivation of these trace formulas is given 

1224: in Appendix A of this article. For obtaining the functional form of $\mu$ 

1225: and $\sigma$, it is also required to evaluate summations like 

1226: \bea

1227: \begin{aligned}

1228:  ~~~&R_0 \equiv \sum_{i<j} \frac{1}{\sin^2(\xi_i-\xi_j)} \, ,

1229: ~~~R_1 \equiv \sum_{i<j} 

1230: \frac{1}{\sin^4(\xi_i-\xi_j)}  ,~~~~~~~~~~~~~~~

1231: ~~~~~~~~~~~~~~~~~~(5.5a,b) \nn \\

1232:  ~~~~~&R_2 \equiv \sum_{i<j} 

1233: \sum_{\substack{k<l \\ (k,l \neq i,j)}}

1234: \frac{1}{\sin^2(\xi_i-\xi_j) \sin^2 

1235: (\xi_k-\xi_l)} \, ,

1236: ~~~~~~~~~~~~~~~~~~~~~~~~~

1237: ~~~~~~~~~~~~~~~~~~~(5.5c) \nn \\

1238: ~~~~& R_3 \equiv  2 \sum_{i<j} \sum_{j<l}  

1239: \frac{1}{\sin^2(\xi_i-\xi_j)\sin^2(\xi_j-\xi_l)} 

1240:  \, + \, \sum_{i<j} \sum_{i<l}  

1241: \frac{1}{\sin^2(\xi_i-\xi_j)\sin^2(\xi_i-\xi_l)}  \nn \\

1242: &~~~~+ \, \sum_{i<j}  \sum_{k<j}  

1243: \frac{1}{\sin^2(\xi_i-\xi_j)\sin^2(\xi_k-\xi_j)} \,   .

1244: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~(5.5d) \nn 

1245: \end{aligned}

1246: \eea

1247: %where the notation 

1248: %$\sideset{}{^{(i,j)}_{k<l}} \sum$ in (5.5c)

1249: %indicates that both 

1250: %  $i$ and $j$ are excluded from the ranges of 

1251: %summation variables $k$ and $l$. 

1252: It is easy to see that

1253: the above defined $R_0$, $R_1$, $R_2$ and $R_3$ satisfy the relation

1254: \addtocounter{equation}{3}

1255: \beq

1256: R_0^2=R_1+R_2+R_3.

1257: \label{e6}

1258: \eeq

1259: Using some summation formulas given in Ref.20, it can be shown that 

1260: %\beq

1261: %R_0^2=R_1+R_2+R_3,

1262: %\eeq

1263: \bea

1264: \begin{aligned}

1265: ~~~~~~~~~~~~&R_0=\frac{N(N^2-1)}{6}\, ,

1266: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1267: ~~~~~~~~~~~~~~~~~~~~~~~~(5.7a) \nn \\

1268: ~~~~~~~~~~~~&R_1=\frac{N(N^2-1)(N^2+11)}{90}\, ,

1269: ~~~~~~~~~~~~~~~~~~~~\,~~~~~~~~~~

1270: ~~~~~~~~~~~~~~~~~~~~~~~~~(5.7b) \nn \\

1271: ~~~~~~~~~~~~&R_2=\frac{N(N^2-1)^2(N-4)}{36}+

1272: \frac{N(N^2-1)(N^2+11)}{90} \, ,~~~~~~\,~~~~~~~~~~~~

1273: ~~~~~~~(5.7c) \nn \\

1274: ~~~~~~~~~~~~&R_3=\frac{4N(N^2-1)(N^2-4)}{45}\,.

1275: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1276: ~~~~~~~~~~~(5.7d) \nn 

1277: \end{aligned}

1278: \eea

1279: \addtocounter{equation}{1}

1280: Derivation of these relations is discussed in Appendix B of this article.

1281: 

1282: Now, by using eqns.(5.3a), (5.4a,b) and (5.7a), we can express

1283: $\mu$ as a function of $m$, $n$ and $N$ given by

1284: \beq

1285: \mu \, = \,  \frac{s^2+t}{2s^2}\, R_0 \, = \, \frac{s^2+t}{12s^2}\,  N(N^2-1).

1286: \label{e8}

1287: \eeq

1288: Next, by using the trace formulas (5.4a,b,c,d), we obtain

1289: \beq

1290: tr\left[(H^{(m|n)})^2\right]\, =\, 

1291: \frac{s^{N-2}\left(s^2+2t\right)}{4} \, R_0^2  + \frac{s^N}{4} \, R_1 

1292: + \frac{s^{N-4}t^2}{4} \, R_2 + \frac{s^{N-2}}{4} \, R_3 \, .

1293: \label{e9}

1294: \eeq

1295: Substituting the expressions for $\mu$ in eqn.(\ref{e8}) 

1296: and $tr\left[(H^{(m|n)})^2\right]$  in  eqn.(\ref{e9}) 

1297: to eqn.(5.3b), and subsequently using (\ref{e6}), it can be shown that

1298: \beq

1299: \sigma^2 \, = \, \frac{s^4-t^2}{4s^4} \, R_1  + 

1300: \frac{s^2-t^2}{4s^4} \, R_3.

1301: \label{e10}

1302: \eeq

1303: Finally, by substituting the values of $R_1$ 

1304: (5.7b) and $R_3$ (5.7d) to eqn.(\ref{e10}), we can express 

1305: $\sigma$ as a function of $m$, $n$ and $N$ given by

1306: \beq

1307: \sigma \, = \, \left \{ \frac{s^4-t^2}{360s^4} \, N(N^2-1)(N^2+11)

1308:   +   \frac{s^2-t^2}{45s^4} \,  N(N^2-1)(N^2-4) \right \}^\frac{1}{2} \, .

1309:  \label{e11}

1310: \eeq

1311: Thus we are able to find out the functional forms of the 

1312: parameters  $\mu$ and $\sigma$ for the case of $SU(m|n)$ HS spin chain. 

1313: It may be observed that, in the special 

1314: case $n=0$ (for which one gets $s=t=m$), 

1315: eqns.(\ref{e8}) and (\ref{e11}) reproduce the forms of

1316:  $\mu$ and $\sigma$ corresponding to the $SU(m)$ bosonic

1317: HS spin chain [9]. 

1318: 

1319: Now we can compare the cumulative level density $F(E)$ (\ref {e1}) 

1320: with the error function $G(E)$ (\ref{e2}), where the values of 

1321: $\mu$ and $\sigma$ are obtained from 

1322: eqns.(\ref{e8}) and  (\ref{e11}) respectively 

1323: for any given $m$, $n$ and $N$.  

1324: In Fig.2, we plot such $F(E)$

1325: and $G(E)$  for the particular 

1326: case of $SU(1|1)$ spin

1327:  chain with $N=15$ lattice sites. From this figure 

1328: it is evident that $F(E)$ follows

1329: $G(E)$ to a high degree of approximation. One can

1330: also quantify the agreement between $F(E)$ and

1331:  $G(E)$ by calculating the 

1332: corresponding mean square error (MSE), which for the above mentioned 

1333: case is given by $8.46 \times 10^{-6}$. 

1334: It may be noted that, the agreement between $F(E)$ and $G(E)$ improves 

1335: rapidly with increasing values of $N$. For example, 

1336: in the case of $SU(1|1)$ model,

1337: MSE between $F(E)$ and $G(E)$ decreases from $5.17 \times 10^{-5}$ 

1338: to $8.46 \times 10^{-6}$ when the value of $N$ is increased from $10$ to $15$.

1339: Next, we consider the particular cases of $SU(1|2)$ as well as 

1340: $SU(2|1)$ supersymmetric spin chain with $N=15$ lattice sites and 

1341: also the $SU(3)$ bosonic spin chain with same number of lattice sites 

1342: for the sake of comparison. 

1343: In Fig.3, we plot 

1344: $F(E)$ and $G(E)$ for $SU(1|2)$, $SU(2|1)$ and 

1345: $SU(3)$ HS spin chains and find that the

1346: corresponding MSEs are given by  $2.38 \times 10^{-5}$,

1347: $1.2 \times 10^{-5}$ and $8.61 \times 10^{-6}$ respectively.

1348: Again $F(E)$ shows very good agreement with $G(E)$ for all of these cases. 

1349: Such agreement also improves rapidly with increasing values of $N$.  

1350: For example, in the case of $SU(1|2)$ spin chain,

1351: MSE between $F(E)$ and $G(E)$ decreases from

1352: $8.23 \times 10^{-5}$  to $2.38 \times 10^{-5}$ 

1353: when the value of $N$ is increased from $10$ to $15$. Analysing 

1354: many other particular cases with different values of $m, ~n$

1355: and sufficiently large values of $N$, we find that 

1356: $F(E)$ follows $G(E)$ with a high degree of approximation 

1357: for all of these cases. 

1358: 

1359: From the above discussion it is evident that 

1360: the local fluctuations in energy level distribution,

1361:  as shown in Fig.1 for the particular

1362: case of $SU(1|1)$ spin chain, get cancelled very

1363:  rapidly whenever we take the cumulative sum of such 

1364:  distribution.  Furthermore, for sufficiently large values of $N$, 

1365:   continuous part of the level density distribution 

1366:  in the spectrum of supersymmetric HS spin chain

1367: satisfies the Gaussian approximation at the same high level of accuracy 

1368: as in the pure bosonic case. 

1369: It may be noted that, the level density of embedded Gaussian orthogonal 

1370: ensemble (GOE) also follows Gaussian distribution at the limit 

1371: $N \rightarrow \infty$, provided the number of one-particle states tends 

1372: to infinity faster than $N$ [21].  However, in our case of $SU(m|n)$ HS spin 

1373: chain, the number of one-particle states (i.e., $m+n$ ) remains fixed for 

1374: all values of $N$.

1375: 

1376: Next we want to study the NNS distribution in the 

1377: spectrum of $SU(m|n)$ HS spin chain. 

1378: To eliminate the effect of level density variation in the

1379: calculation of NNS distribution for the full energy range, 

1380: it is necessary to apply an unfolding mapping to the `raw' spectrum [22].

1381: This unfolding mapping may be defined by using the continuous part of the

1382: cumulative level density distribution. We have already seen that,

1383: for the case of $SU(m|n)$ HS spin chain, the continuous part of cumulative

1384: level density is given by $G(E)$ (\ref{e2}) with a high degree of approximation. 

1385: So we transform each energy $E_i$, $i=1,\cdots ,l \, $, into an

1386:  unfolded energy $\xi_i \equiv  G(E_i)$. The function $p(u)$ 

1387: is defined as the density of the normalized 

1388: spacings $u_i=(\xi_{i+1}-\xi_i)/\Delta$, where 

1389: $\Delta=(\xi_l-\xi_1)/(l-1)$ is the mean spacing 

1390: of the unfolded energy. To get rid of local fluctuations occurring in $p(u)$,

1391: again we study the cumulative NNS

1392: distribution given by $P(u)= \int_0^u p(x)dx$, instead of $p(u)$.

1393:  In this context it may be noted that,

1394:  NNS distributions corresponding to the cases of

1395: classical GOE as well as embedded GOE obey the Wigner's law [23]:

1396: \beq

1397: p(u)= \frac{\pi}{2} u \exp (-\pi u^2/4) \, . \nn

1398: \eeq

1399: On the other hand, from the conjecture of Berry and Tabor one may 

1400: expect that the NNS distribution for an integrable model will

1401: obey Poisson's law given by $p(u)= \exp(-u)$ [24]. However, it has been

1402: found that the NNS distribution for $SU(m)$ bosonic 

1403: HS spin chain does not follow either

1404:  Wigner's law or Poisson's law within a wide range

1405:  of $N$ [9]. Instead, the cumulative NNS distribution

1406: for this bosonic spin chain can be well approximated 

1407: by a function like

1408: \beq

1409: \tilde{P}(u)=v^{\alpha} [ 1- \gamma (1-v)^{\beta} ],

1410: \label{e12}

1411: \eeq

1412: where $v=u/u_{max}$ with $u_{max}$ being the largest normalized spacing, 

1413:  $\alpha$ and $\beta$ are

1414: two free parameters taking values within the range

1415: $0<\alpha,~\beta<1$, and the value of $\gamma$ is fixed by requiring that 

1416: the average normalized spacing be equal to 1.

1417: In our study we also find that,  NNS distribution 

1418: in the spectrum of $SU(m|n)$ HS spin chain does not follow either

1419:  Wigner's law or Poisson's law within a wide range of $N$. 

1420: In particular, it is observed that the slope of cumulative NNS distribution

1421: diverges for both $u \rightarrow 0$ and

1422:  $u \rightarrow u_{max}$, which can not be explained 

1423: from Wigner's or Poisson's distribution. Furthermore, we find 

1424: that the cumulative NNS distribution for $SU(m|n)$ HS spin chain 

1425: can be fitted well by $\tilde{P}(u)$ in eqn.(\ref{e12}) within a 

1426: range of $N$.

1427: For example, in the particular case of $SU(1|1)$ spin chain,  it is

1428: checked that $P(u)$ agrees well with $\tilde{P}(u)$ 

1429: within the range $N\leq 20$. In Fig.4, 

1430: we plot such $P(u)$ and $\tilde{P}(u)$ for 

1431: $N=17$ lattice sites ($u_{max}=2.626$ in this case) and found a 

1432: good agreement with  MSE $=0.0234$ when 

1433: the values of free parameters are taken as  $\alpha=0.39$ and $\beta=0.29$.

1434: However, it is possible that the appearance of such non-Poissonian 

1435: NNS distribution in the spectrum of $SU(m|n)$ HS spin chain is an

1436: artifact of finite-size effect, which requires further

1437: investigation.

1438: 

1439: Finally we want to make a comment about the behaviour of 

1440: parameters $\mu$ in eqn.(\ref{e8}) and $\sigma$ in eqn.(\ref{e11})

1441: under the exchange of bosonic and fermionic spin degrees of freedom.

1442: Since  $s \rightarrow s$ and $t \rightarrow -t$ under this exchange,  

1443: we find that $\sigma$ remains invariant and

1444: $\mu$ changes to $\bar{\mu}$ given by

1445:  $\bar {\mu}=\frac{N(N^2-1)}{6} - {\mu} \,$. 

1446: It is interesting to note that this relation between $\mu $ and 

1447: $\bar {\mu}$ can also be obtained by applying eqn.(\ref{d12}): 

1448: \bea

1449: {\mu} &=&\frac{1}{s^N} \sum_{E_i}\mathcal{D}^{(m|n)} (E_i) E_i  \nn \\

1450: &=&\frac{1}{s^N} \sum_{E_i}\mathcal{D}^{(n|m)} 

1451: \left(\frac{N(N^2-1)}{6}-E_i \right) 

1452: E_i 

1453: =\frac{N(N^2-1)}{6}-\bar{\mu} \, . \nn

1454: %\label{e13}

1455: \eea

1456: This agreement clearly gives a support to our conjecture

1457: (\ref{d11}).  By using this conjecture we have found in Sec.4 that, 

1458:  the spectrum of $SU(n|m)$ spin chain can be

1459:  obtained from that of $SU(m|n)$ spin chain through 

1460: an inversion and overall shift of all energy levels.

1461: Since none of these operations change the

1462: standard deviation of level density distribution, $\sigma$

1463: should take the same value for $SU(m|n)$ and $SU(n|m)$ 

1464: HS spin chain. Hence, the observation that 

1465: $\sigma$ in eqn.(\ref{e11}) remains invariant 

1466: under the exchange of bosonic and fermionic spin degrees of freedom, 

1467: is also consistent with our conjecture (\ref{d11}).

1468: 

1469: 

1470: \noindent \section {Conclusion }

1471: 

1472: Here we derive an exact expression for the partition function

1473: of $SU(m|n)$ supersymmetric HS spin chain by using the freezing trick

1474: and also study some properties of the related spectrum.

1475: For applying the freezing trick, we consider a spin CS model containing 

1476: an anyon like representation of the permutation algebra as spin dependent 

1477: interaction. We find out the complete spectrum of such spin CS model

1478: including the degeneracy factors of all energy eigenvalues.

1479: At the strong coupling limit, this spin CS model 

1480: reduces to the sum of spinless CS model with only dynamical degrees

1481: of freedom and $SU(m|n)$ supersymmetric HS spin chain. 

1482: Consequently, by factoring out the contribution due to dynamical

1483: degrees of freedom from partition function of this spin CS model, 

1484: we obtain the partition function of $SU(m|n)$ supersymmetric HS spin chain. 

1485: By using this partition function, we study the motif representation for 

1486: $SU(m|n)$ HS spin chain and find that, due to the lifting 

1487: of a selection rule, some additional energy levels appear in the 

1488: spectrum in comparison with the case of $SU(m)$ bosonic spin chain. 

1489: 

1490: By using Mathematica we observe that,  the partition function of 

1491: $SU(m|n)$ HS model satisfies the duality relation (\ref{d11})

1492: for many values of $m,$ $n$ and $N$. This observation 

1493: motivates us to conjecture that this duality relation, involving 

1494: the interchange of bosonic and fermionic spin degrees of freedom,

1495: is valid for all possible values of $m,$ $n$ and $N$.  It would 

1496: be interesting if this duality relation can be proved 

1497: analytically by using the motif representations and skew-Young 

1498: diagrammes associated with the $Y(gl_{(m|n)})$ quantum group.  

1499: Furthermore, it is known that, 

1500: the partition functions of $SU(m)$ and $SU(m|n)$ 

1501: Polychronakos spin chains are intimately connected with 

1502: Rogers-Szeg\"{o} (RS) polynomial [8,15],

1503: which appears in the theory of partitions [25].

1504: Since, HS spin chain share the same quantum 

1505: group symmetry with Polychronakos spin chain, 

1506: it might be promising to investigate mathematical

1507: structures connected with the partition functions of $SU(m)$ as well as

1508: $SU(m|n)$ HS spin chain and explore whether some 

1509: new RS type polynomials can be generated in this way. 

1510: 

1511: By using the partition function of $SU(m|n)$ HS spin chain, 

1512: we study global properties of its spectrum like level

1513: density distribution and  NNS distribution.

1514: It is found that, similar to the case of $SU(m)$ bosonic HS spin chain,

1515:  continuous part of the energy level density

1516: satisfies the Gaussian distribution with a high degree of accuracy

1517: for sufficiently large values of $N$.   

1518: We also derive exact expressions for the mean value and the

1519: standard deviation which characterize such 

1520: Gaussian distribution.  It would be interesting to provide 

1521: an explanation for this behaviour of energy level density distribution 

1522: in the framework of random matrix theory

1523: and explore whether the underlying quantum group symmetry 

1524: of HS spin chain plays some role in this matter. 

1525: 

1526: \vskip 1 cm

1527: \noindent{\bf Acknowledgements}

1528: \smallskip

1529: 

1530: We would like to thank Palash B. Pal for some helpful discussions. 

1531: 

1532: \newpage

1533: 

1534: 

1535: 

1536: \newpage

1537: \renewcommand{\theequation}{A-\arabic{equation}}

1538: \setcounter{equation}{0}

1539: \section*{\begin{large} Appendix A. 

1540: \end{large}\begin{normalsize} Evaluation of trace formulas \end{normalsize}} 

1541: 

1542: Here we shall derive the trace formulas (5.4a,b,c,d)

1543:  by assuming that $\v \alpha_1 \dots \alpha_N \r$

1544: (with $\alpha_j \in [1,\dots,s]$) are orthonormal 

1545: set of vectors.  Since the trace of identity operator is given by the dimension 

1546: of the Hilbert space, eqn.(5.4a) is really a trivial relation.

1547: Using eqn.(\ref{b6}), it can be shown that

1548: \bea

1549: \l\alpha_1 \dots \alpha_i \dots \alpha_j \dots \alpha_N 

1550: \v  \tilde{P}_{ij}^{(m|n)}

1551: \v \alpha_1 \dots \alpha_i \dots \alpha_j \dots \alpha_N \r

1552: =(-1)^{\epsilon (\alpha_i) } \delta_{\alpha_i\alpha_j} , \nn

1553: \eea

1554: where $\epsilon(\alpha_i)=0~(1)$ when $\alpha_i$ is a

1555:  bosonic (fermionic) spin.

1556: With the help of above equation, we derive 

1557: the trace relation (5.4b) as 

1558: \bea

1559: tr \left[\tilde{P}_{ij}^{(m|n)}\right]

1560: \hskip -.20 cm &=& \hskip -.20 cm 

1561: \sum_{\alpha_1\dots\alpha_N=1}^{s }

1562:  \l \alpha_1 \dots  \alpha_i \dots \alpha_j \dots 

1563: \alpha_N \v 

1564: \tilde{P}_{ij}^{(m|n)}\v \alpha_1 \dots  \alpha_i 

1565: \dots \alpha_j \dots \alpha_N  \r  \nn \\

1566: &=& ~ \sideset{}{^{(\alpha_i,\alpha_j)}}\sum_{\alpha_1

1567: \dots\alpha_N=1}^{s }

1568: \sum_{\alpha_i,\alpha_j=1}^s  \, \l \alpha_1 \dots  

1569: \alpha_i \dots \alpha_j \dots \alpha_N \v 

1570: \tilde{P}_{ij}^{(m|n)}\v \alpha_1 \dots  \alpha_i 

1571: \dots \alpha_j \dots \alpha_N  \r  \nn \\

1572: &=& ~ \sideset{}{^{(\alpha_i,\alpha_j)}}\sum_{\alpha_1

1573: \dots\alpha_N=1}^{s}

1574: \sum_{\alpha_i=1}^s (-1)^{\epsilon 

1575: (\alpha_i)}=s^{N-2}t , \nn 

1576: \eea

1577: where the notation

1578: $\sideset{}{^{(\alpha_i,\alpha_j)}}\sum\limits_{\alpha_1 \dots\alpha_N=1}^{s}$

1579: represents summation  over all spin components 

1580: $\alpha_1,\dots,\alpha_N$ except $\alpha_i$ and

1581: $\alpha_j$. 

1582: 

1583: Next, by using eqn.(\ref{b6}), it is found that

1584: \bea

1585: ~~~~~~~~~~\l\alpha_1 \dots  \alpha_i \dots \alpha_j 

1586: \dots \alpha_l \dots\alpha_N \v \tilde{P}_{ij}^{(m|n)} 

1587: \tilde{P}_{il}^{(m|n)} 

1588: \v \alpha_1 \dots  \alpha_i \dots \alpha_j 

1589: \dots \alpha_l \dots \alpha_N \r 

1590: =  \delta_{\alpha_i\alpha_j} \delta_{\alpha_i\alpha_l}.\nn

1591: \eea

1592: Applying the above equation, we obtain 

1593:  a trace relation in eqn.(5.4c) as 

1594: \begin{equation}

1595: \begin{split}

1596: & tr\left[ \tilde{P}_{ij}^{(m|n)}\tilde{P}_{il}^{(m|n)} 

1597: \right] \\

1598: &~~~~ =\sum_{\alpha_1\dots\alpha_N=1}^{s } 

1599: \l \alpha_1 \dots \alpha_i \dots \alpha_j 

1600: \dots \alpha_l \dots \alpha_N \v 

1601: \tilde{P}_{ij}^{(m|n)}\tilde{P}_{il}^{(m|n)} 

1602: \v \alpha_1 \dots \alpha_i \dots \alpha_j 

1603: \dots \alpha_l \dots  \alpha_N \r \nn \\

1604: &~~~~=~~\sideset{}{^{(\alpha_i,\alpha_j,\alpha_l)}}

1605: \sum_{\alpha_1\dots\alpha_N=1}^{s } 

1606: \sum_{\alpha_i=1}^s 1

1607: =s^{N-2}.

1608: \end{split}

1609: \end{equation}

1610:  Other trace relations in (5.4c) can be proved in a similar way.

1611: 

1612: By using eqn.(2.6), it is also found that

1613: \bea

1614: ~~~~~~&\l \alpha_1 \dots \alpha_i \dots \alpha_j \dots 

1615: \alpha_k \dots \alpha_l \dots \alpha_N \v \, 

1616: \tilde{P}_{ij}^{(m|n)}\tilde{P}_{kl}^{(m|n)} \, \v

1617:  \alpha_1 \dots \alpha_i \dots \alpha_j \dots 

1618: \alpha_k \dots \alpha_l \dots \alpha_N \r ~~~~~~~~~~~~~~~\nn \\

1619: &~~~~~~~~~~~~~~~~~~~~~~~~=(-1)^{\epsilon(\alpha_i)+\epsilon(\alpha_k)} 

1620: \, \delta_{\alpha_i \alpha_j} \delta_{\alpha_k 

1621: \alpha_l}\, . \nn

1622: \eea

1623: With the help of this equation, we obtain the trace relation (5.4d) as

1624: \begin{equation}

1625: \begin{split}

1626: &tr\left[ \tilde{P}_{ij}^{(m|n)}\tilde{P}_{kl}^{(m|n)} 

1627: \right] \\

1628: &=\sum_{\alpha_1\dots\alpha_N=1}^{s }

1629: \l\alpha_1 \dots \alpha_i \dots \alpha_j 

1630: \dots \alpha_k \dots \alpha_l \dots \alpha_N\v

1631: \tilde{P}_{ij}^{(m|n)}\tilde{P}_{kl}^{(m|n)}

1632: \v \alpha_1 \dots \alpha_i \dots \alpha_j \dots 

1633: \alpha_k \dots \alpha_l \dots \alpha_N \r   \\

1634: &=~~\sideset{}{^{(\alpha_i,\alpha_j,\alpha_k,

1635: \alpha_l)}}\sum_{\alpha_1\dots\alpha_N=1}^s 

1636: \sum_{\alpha_i ,\alpha_k=1}^s

1637: (-1)^{\epsilon (\alpha_i)+ \epsilon(\alpha_k)}= s^{N-4}t^2.  \nn

1638: \end{split}

1639: \end{equation}

1640: 

1641: \renewcommand{\theequation}{B-\arabic{equation}}

1642: \setcounter{equation}{0}

1643: \section*{\begin{large} Appendix B. \end{large} \begin{normalsize}                   

1644:               Evaluation of summation formulas\end{normalsize}} 

1645: Here we briefly describe the way of calculating known

1646: summation formulas (5.7a) and 

1647: (5.7b) [9],  and subsequently present our derivation 

1648: for new ones like (5.7c) and (5.7d). 

1649: From the work of Calogero et. al [20], it is 

1650:  known that

1651: \beq

1652: \sum_{j=1}^{N-1} \frac{1}{\sin^2 ( \frac{j \pi}{N})}= 

1653: \frac{N^2-1}{3}\, ,

1654: \label{B1}

1655: \eeq

1656: and

1657: \beq

1658: \sum_{j=1}^{N-1} \frac{1}{\sin^4 (\frac{j \pi }{N})}= 

1659: \frac{(N^2-1)(N^2+11)}{45}\, .

1660: \label{B2}

1661: \eeq

1662: Using the translational invariance on a 

1663: circular lattice and summation relation (\ref {B1}),

1664: one can obtain eqn.(5.7a) as  

1665: \beq

1666: R_0=\frac{1}{2}\sum_{i\neq j} \frac{1}{\sin^2(\xi_i-\xi_j)}

1667: =\frac{N}{2}\sum_{j=1}^{N-1} \frac{1}{\sin^2 \left(\frac{j\pi}{N}\right)}

1668: =\frac{N(N^2-1)}{6}\,. \nn

1669: %\label{B3}

1670: \eeq

1671: Similarly, by using (\ref{B2}), one obtains eqn.(5.7b) as 

1672: \beq

1673: R_1=\frac{1}{2}\sum_{i\neq j} \frac{1}{\sin^4(\xi_i-\xi_j)}

1674: =\frac{N}{2}\sum_{j=1}^{N-1} \frac{1}{\sin^4 \left(\frac{j\pi}{N}\right)}

1675: =\frac{N(N^2-1)(N^2+11)}{90}. \nn

1676: %\label{B4} 

1677: \eeq

1678: 

1679: For the purpose of calculating  $R_2$ in eqn.(5.5c), 

1680: we note that $R_0$ in eqn.(5.5a) can be expressed as

1681: \beq

1682: R_0= \sum_{\substack{k<l \\ (k,l \neq i,j)}}

1683: \frac{1}{\sin^2(\xi_k-\xi_l)}+

1684: \sum_{\substack{r=1 \\ (r \neq i)}}^N \frac{1}{\sin^2(\xi_i-\xi_r)} 

1685: + \sum_{\substack{r=1 \\ (r \neq j)}}^N  \frac{1}{\sin^2(\xi_j-\xi_r)}

1686: - \frac{1}{\sin^2(\xi_i-\xi_j)}. \nn

1687: \eeq

1688: Substituting the value of $R_0$ given in (5.7a) to the above relation and

1689: also using (\ref {B1}), we find that 

1690: \beq

1691: \sum_{\substack{k<l \\ (k,l \neq i,j)}} \frac{1}{\sin^2(\xi_k-\xi_l)}

1692: =\frac{(N^2-1)(N-4)}{6} 

1693: + \frac{1}{\sin^2(\xi_i-\xi_j)}\, .  \nn

1694: \eeq

1695: By substituting this expression 

1696: to $R_2$ in  eqn.(5.5c), and subsequently using eqns.(5.7a) as well as 

1697: (5.7b), we derive the value of $R_2$ given in eqn.(5.7c) as 

1698: \bea

1699: R_2 &=& \sum_{i<j} \frac{1}{\sin^2(\xi_i-\xi_j)} 

1700: \biggl[\frac{(N^2-1)(N-4)}{6} + 

1701: \frac{1}{\sin^2(\xi_i-\xi_j)} \biggr] \nn \\

1702: &=& \frac{N(N^2-1)^2(N-4)}{36} + \frac{N(N^2-1)(N^2+11)}{90} \,. \nn

1703: \eea

1704: Finally, by substituting the values of  $R_0$ (5.7a), 

1705: $R_1$ (5.7b) and $R_2$ (5.7c) to the relation (\ref{e6}), we 

1706: easily obtain the value of $R_3$  appearing in eqn.(5.7d).

1707: 

1708: \newpage

1709: \vskip 2 cm 

1710: \leftline {\large \bf References}

1711: \medskip

1712: \begin{enumerate}

1713: 

1714: \item F. D. M. Haldane, Phys. Rev. Lett.  60 (1988) 635.

1715: \item B. S. Shastry, Phys. Rev. Lett.  60 (1988) 639.

1716: \item F. D. M Haldane, in Proc. 16th Taniguchi Symp., Kashikojima, Japan 

1717: (1993), eds.  A. Okiji and N. Kawakami (Springer, 1994).

1718: \item Z.N.C. Ha, {\it Quantum many-body systems in one dimension} 

1719: (Series on Advances in Statistical Mechanics, Vol.12), 

1720: (World Scientific,1996). 

1721: \item A.P. Polychronakos,  {\it Generalized statistics in one dimension}, 

1722: Les Houches 1998 lectures, hep-th/9902157.

1723: \item F. D. M. Haldane, Z.N.C. Ha, J.C. Talstra, 

1724:  D. Benard and V. Pasquier, Phys. Rev. Lett. 69 (1992) 2021. 

1725: \item D. Benard, M. Gaudin, F. D. M. Haldane, and V. Pasquier, 

1726: J. Phys.  A26 (1993) 5219.

1727: \item  K.Hikami, Nucl. Phys. B441[FS] (1995) 530.

1728: \item F. Finkel, A. Gonz\'{a}lez-L\'{o}pez, Phys. Rev. 

1729:  B72 (2005) 174411.

1730: \item A.P. Polychronakos, Phys. Rev. Lett. 70 (1993) 2329; Nucl. Phys. 

1731: B419 (1994) 553.

1732: \item Bill Sutherland and B. S. Shastry, Phys. Rev. Lett. 71 (1993) 5.

1733: \item A.P. Polychronakos, 

1734: {\it Physics and mathematics of Calogero particles,} hep-th/0607033.

1735: \item P. Schlottmann, Int. Jour. Mod. Phys.  B11 (1997) 355. 

1736: \item B. Basu-Mallick, H. Ujino and M. Wadati, Jour. Phys. Soc. Jpn. 68

1737: (1999) 3219. 

1738: \item K. Hikami, and B. Basu-Mallick, Nucl. Phys.  B566

1739: [PM] (2000) 511.

1740: \item B. Basu-Mallick, Nucl. Phys.  B540 [FS] (1999) 679.

1741: \item B. Basu-Mallick, Nucl. Phys.  B482 [FS] (1996) 713.

1742: \item C. Ahn and W. M. Koo, Phys. Lett. B365 (1996) 105. 

1743: \item B. Sutherland, Phys. Rev. A5 (1972) 1372. 

1744: \item F. Calogero and A.M. Perelomov, Commun. Math. Phys. 59 (1978) 109.

1745: \item K.K. Mon and J.B. French, Ann. Phys. 95 (1975) 90. 

1746: \item F. Haake, {\it Quantum signatures of Chaos}

1747:  (Springer-verlag, 2001).

1748: \item V.K.B. Kota, Phys. Rep. 347 (2001) 223. 

1749: \item M.V. Berry and M. Tabor, Proc. R. Soc. Lond. A356 (1977) 375. 

1750: \item G.E. Andrews, {\it The theory of partitions}

1751:       (Addison-Wesley, Reading, M.A., 1976).

1752: 

1753: \end{enumerate}

1754: 

1755: \newpage

1756: 

1757: \begin {figure} [h]

1758: \centering

1759: %\includegraphics[scale=0.50,angle=270]{fig1.ps}

1760: \includegraphics[scale=0.87]{fig1.ps}

1761: %\includegraphics{fig1.ps}

1762: \caption{Energy levels $E_i$ and degeneracies $\mathcal{D}^{(1|1)}(E_i)$ 

1763: of the $SU(1|1)$ HS spin chain for $N=15$.}

1764: \end{figure}

1765: \newpage

1766: \begin {figure} [h]

1767: %\centering 

1768: \hskip -.25 cm 

1769: \includegraphics[scale=0.65,angle=270]{fig2.ps}

1770: \caption{Continuous curve represents the error function G(E) 

1771: and crosses represent the 

1772: cumulative distribution function F(E) (at its discontinuity points)

1773: for $SU(1|1)$ spin chain with $N=15$.} 

1774: \end{figure}

1775: \newpage

1776: \begin {figure} [h]

1777: \centering

1778: \includegraphics[scale=0.60,angle=270]{fig3.ps}

1779: \caption{Left continuous curve represents G(E) for $SU(1|2)$ 

1780: spin chain with $N=15$ and the corresponding F(E) 

1781: is given by crosses. Middle continuous curve 

1782: represents G(E) for $SU(2|1)$ spin chain

1783: with $N=15$ and the corresponding F(E) is given by dots.

1784: Right continuous curve represents G(E) for $SU(3)$ spin chain

1785: with $N=15$ and the corresponding F(E) is given by triangles.}

1786: \end{figure}

1787: \newpage

1788: \begin {figure} [h]

1789: \centering

1790: \includegraphics[scale=0.60,angle=270]{fig4.ps}

1791: \caption{Dotted curve represents the cumulative NNS distribution

1792: $P(u)$ for $SU(1|1)$ spin chain with $N=17$ and the continuous

1793: curve represents the approximate distribution function $\tilde{P}(u)$.}

1794: \end{figure}

1795: 

1796: \end{document}

1797: