1: \documentclass[12pt]{article}

2: 

3: %\pagestyle{empty}  %%% for no page numbers

4: 

5: \textwidth 18 cm %% A4 paper 210 mm.

6: \textheight 23.2 cm %% A4 paper 297 mm. 

7: 

8: \oddsidemargin -1 cm 

9: %% this is 1 inch (2.54) cm less that the distance from the left

10: %% edge of the paper to the left margin of text of right hand pages 

11: 

12: \evensidemargin -1 cm  

13: %% same as above but for right hand pages

14: 

15: \topmargin 0.5 cm

16: %% this is 1 inch (2.54) cm (in fact more) less that the distance from

17: %% the top edge 

18: %% of the paper to the top of page head

19: 

20: \headsep 0em

21: \headheight 0em

22: 

23: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

24: % PACKAGES

25: 

26: \usepackage{amsmath}

27: \usepackage{amsthm}

28: \usepackage{amssymb}

29: \usepackage{graphics}

30: \usepackage{graphicx}

31: 

32: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

33: % Standard envionments

34: 

35: \newtheorem{theorem}{Theorem}

36: \newtheorem{lemma}[theorem]{Lemma}

37: \newtheorem{corrol}[theorem]{Corollary}

38: 

39: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

40: % Standard definitions

41: 

42: \def\defn#1{{\bf #1}}

43: \def\tabelm#1{{\begin{tabular}[t]{c} #1 \end{tabular}}}

44: \def\multi#1{{\begin{tabular}[t]{@{}l@{}}#1\end{tabular}}}

45: 

46: \def\Real{{\mathbb R}}

47: \def\Cmpx{{\mathbb C}}

48: \def\Intg{{\mathbb Z}}

49: \def\Natn{{\mathbb N}}

50: \def\cnj#1{{\overline{#1}}}

51: 

52: \def\sp{{\,,\quad}}

53: \def\sdotsc{,\!..,}

54: 

55: \def\Poisson(#1,#2){{\left\{#1,#2\right\}}}

56: \def\Set#1{{\left\{#1\right\}}}

57: 

58: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

59: % DEFINITIONS

60: \def\ket(#1){{\left|#1\right>}}

61: \def\bra(#1){{\left<#1\right|}}

62: 

63: \def\Aelement#1{{\boldsymbol{#1}}}

64: \def\Aa{\Aelement{a}}

65: \def\Aw{\Aelement{w}}

66: \def\Ax{\Aelement{x}}

67: \def\Ay{\Aelement{y}}

68: \def\Az{\Aelement{z}}

69: \def\Af{\Aelement{f}}

70: \def\Ag{\Aelement{g}}

71: \def\Av{\Aelement{v}}

72: \def\Au{\Aelement{u}}

73: \def\Ath{\Aelement{\theta}}

74: \def\Aeps{\Aelement{\varepsilon}}

75: \def\Aalpha{\Aelement{\alpha}}

76: \def\Ap{\Aelement{p}}

77: \def\Aq{\Aelement{q}}

78: 

79: \def\AF{\Aelement{F}}

80: \def\AC{\Aelement{C}}

81: 

82: \def\CC(#1){C_{#1}}

83: 

84: \def\Idn{{\boldsymbol{1}\hspace{-0.25em}\textup{\small I}}}

85: 

86: \def\Hil{{\cal H}}

87: \def\Lin{{\cal L}}

88: \def\Alg{{\cal A}}

89: \def\Balg{{\cal B}}

90: \def\Bpi{{\pi_\Balg}}

91: \def\man{{\cal M}}

92: \def\setU{{\cal U}}

93: \def\SetS{{\cal S}}

94: 

95: 

96: \title{A Noncommutative Geometric Analysis of a Sphere/Torus Topology

97: Change}

98: 

99: \author{J Gratus\thanks{email: j@gratus.net}}

100: 

101: \begin{document}

102: \maketitle

103: 

104: \begin{abstract}

105: A one parameter set of noncommutative complex algebras

106: is given. These may be considered deformation quantisation algebras.

107: The commutative limit of these algebras correspond to the

108: algebra of polynomial functions over a manifold or variety.  The

109: topology of the manifold or variety depends on the parameter, varying

110: from nothing, to a point, a sphere, a certain variety and finally a

111: torus. The irreducible adjoint preserving representations of the

112: noncommutative algebras are studied.  As well as typical

113: noncommutative sphere type representations and noncommutative torus

114: type representations, a new object is discovered and called a

115: Sphere-Torus.

116: \end{abstract}

117: 

118: \tableofcontents

119: 

120: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

121: 

122: \section{Introduction}

123: \label{ch_intr}

124: 

125: In noncommutative geometry we often wish to find analogues to

126: topological properties of a manifold such as compactness,

127: connectedness and genus, which we will consider here. For matrix

128: geometries~\cite{Madore_book} the question of genus is tricky since

129: both the sphere and torus have matrix analogues.

130: 

131: In deformation quantisation~\cite{Sternheimer1}, 

132: one can simply take the commutative limit

133: and ask what the genus of the underlying manifold is.  In the case of

134: the sphere and torus, we can also take the representation of the

135: noncommutative algebra and compare its properties with the

136: representations of the noncommutative sphere and torus.

137: 

138: In this article we present, in section \ref{ch_alg}, a one parameter

139: set of deformation algebras $\Alg(R)$ for $R\in\Real$. The commutative

140: limit of these algebras are $C^\omega_0(\man(R))$ the commutative

141: algebra of complex polynomials on the manifold (or variety)

142: $\man(R)$. This manifold has different topologies depending on the

143: value of $R$. Varying from nothing, to a point, a sphere, a variety

144: and finally a torus. This is described in section

145: \ref{ch_comm}. 

146: 

147: In section \ref{ch_rep} we look at the finite dimensional

148: representations of $\Alg(R)$. These can be classified as either

149: $S^2$-type representations or $T^2$-type representations by comparing

150: them to the representations for the noncommutative torus or the

151: noncommutative sphere. Depending on the value of $R$ one or other of

152: these representations exists. What we show in this article is that

153: there is a region of $R$ where both types of representation

154: exist. This region, which we shall name the \defn{sphere-torus}, is a

155: purely noncommutative region, it disappears in the commutative limit.

156: 

157: We summarise the various representation in the conclusion, section

158: \ref{ch_concl}. 

159: 

160: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

161: 

162: \subsection{Notation}

163: \begin{tabular}{lp{13cm}}

164: $\Real_+$ 

165: &

166: $\Set{t\in\Real\,|\,t>0}$

167: \\

168: $\man$

169: &

170: Manifold or variety.

171: \\

172: $C^\omega(\man)$, $C^\omega(\Real^r)$ 

173: &

174: Commutative algebra of complex

175: analytic functions over $\man$ or $\Real^r$.

176: \\

177: $C^\omega_0(\man)$, $C^\omega_0(\Real^r)$ 

178: &

179: Commutative algebra of complex

180: polynomials over $\man$ or $\Real^r$.

181: \\

182: $\Alg$, $\Balg$ 

183: &

184: Noncommutative algebras over $\Cmpx$.

185: \\

186: \multi{Unbold symbols:\\\quad $f,g,u,v,x_i,$\\\quad $F_s,w,x,y,z$} 

187: &

188: Elements of the commutative

189: algebras. i.e. Analytic or polynomial functions over $\man$ or

190: $\Real$.

191: \\

192: \multi{Bold symbols:\\\quad $\Af,\Ag,\Au,\Av,\Ax_i,$\\

193: \quad $\AF_s,\Aw,\Ax,\Ay,\Az$} 

194: &

195: Elements of the noncommutative algebras $\Alg$ and $\Balg$

196: \\

197: $\Aeps$ 

198: &

199: Element in the centre 

200: of the noncommutative algebras $\Alg$ and $\Balg$

201: \\

202: $\Hil$ 

203: &

204: Finite or infinite dimensional Hilbert space.

205: \\

206: $\Lin(\Hil,\Hil)$

207: &

208: Space of linear maps over $\Hil$. 

209: (Bounded or

210: unbounded operators.)

211: \\

212: $M_n(\Cmpx)$ 

213: &

214: Space of $n\times n$ complex matrices.

215: \\

216: $\Idn$

217: &

218: The identity in $\Lin(\Hil,\Hil)$ and the unit matrix in

219: $M_n(\Cmpx)$.

220: \\

221: $\ket(\theta)$

222: &

223: Bra-ket notation for vectors in $\Hil$.

224: \end{tabular}

225: 

226: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

227: 

228: \subsection{A one parameter set of immersions with a sphere, 

229: torus and variety}

230: \label{ch_comm}

231: 

232: \newlength{\pichigh}

233: \setlength{\pichigh}{3.2 cm}

234: \setlength{\unitlength}{\pichigh}

235: \begin{figure}

236: \begin{tabular}{|c|c|c|c|}

237: \hline

238: \begin{picture}(0.8,1.1)

239: \put(0,0){\includegraphics[height=\pichigh]{CrossSecA.eps}}

240: \put(0.4,0.8){\makebox(0,0)[rt]{$z\,$}}

241: \end{picture}

242: &

243: \begin{picture}(1.4,1.1)

244: \put(0,0){\includegraphics[height=\pichigh]{CrossSecB.eps}}

245: \put(0.7,0.9){\makebox(0,0)[rt]{$z\,$}}

246: \end{picture}

247: &

248: \begin{picture}(1.4,1.1)

249: \put(0,0){\includegraphics[height=\pichigh]{CrossSecC.eps}}

250: \put(0.7,0.9){\makebox(0,0)[rt]{$z\,$}}

251: \end{picture}

252: &

253: \begin{picture}(1.4,1.1)

254: \put(0,0){\includegraphics[height=\pichigh]{CrossSecD.eps}}

255: \put(0.7,0.9){\makebox(0,0)[rt]{$z\,$}}

256: \end{picture}

257: \\

258: $-1<R\le0$ 

259: &

260: $0<R<1$

261: &

262: $R=1$ 

263: &

264: $R>1$ 

265: \\

266: \hline

267: \end{tabular}

268: \caption{Shape of the slice of $\man$ setting $y=0$. The shape of

269: $\man$ is given by rotating slice about the $z$-axis.}

270: \label{fig_slices}

271: \end{figure}

272: 

273: Consider the immersions given by

274: %[

275: \begin{align}

276: \man(R)=\Set{(x,y,z)\in\Real^3\,\bigg|\,z^2+(x^2+y^2-R)^2=1}

277: \label{intr_man}

278: \end{align}

279: %]

280: It is obvious that for $R<-1$ there is no solution to (\ref{intr_man})

281: while for $R=-1$, $\man(R)$ consists of the single point at the origin

282: $(x,y,z)=(0,0,0)$. To picture $\man(R)$ for $R>-1$, we note that it is

283: axisymmetric about the $z$-axis. Therefore we can examine the shape of

284: $\man(R)$ by setting $y=0$ and rotating the subsequent 1 dimensional

285: variety about the $z$-axis. From figure \ref{fig_slices} we can see

286: that for $-1<R\le0$, $\man(R)$ is a convex manifold topologically

287: equivalent to the sphere.  For $0<R<1$, $\man(R)$ is a non convex

288: manifold topologically equivalent to the sphere.  For $R=1$, $\man(R)$

289: is not a manifold but instead a 2 dimensional variety, which is smooth

290: about all points except the origin $(x,y,z)=(0,0,0)$.  For $R>1$,

291: $\man(R)$ is a torus.

292: 

293: 

294: There is a Poisson structure on $\man$ is given by

295: %[

296: \begin{align}

297: \Poisson(x,y)=z \sp

298: \Poisson(z,x)=2(x^2+y^2-R)y=2wy \sp

299: \Poisson(y,z)=2(x^2+y^2-R)x=2wx 

300: \label{intr_poi}

301: \end{align}

302: %]

303: where $w=x^2+y^2-R$, which is consistent with (\ref{intr_man})

304: 

305: We can give a Darboux coordinate system $(p,q)$ such that

306: %[

307: \begin{align}

308: x = (R+\cos(2p))^{1/2}\cos(q)

309: \sp

310: y = -(R+\cos(2p))^{1/2}\sin(q)

311: \sp

312: z = \sin(2p)

313: \label{intr_darb}

314: \end{align}

315: %]

316: where $\Poisson(p,q)=1$. It is easy to see that (\ref{intr_man}) and

317: (\ref{intr_poi}) are satisfied. A necessary condition for these to be

318: valid is $R+\cos(2p)>0$. Thus

319: for the torus $R>1$ this is valid for all $p$. More

320: specifically we can patch coordinate systems with $0<q<2\pi$ and

321: $0<p<\pi$. For the variety $R=1$ we must exclude the point

322: $p=\tfrac12\pi$ which correspond to the point at the origin.  For

323: $-1<R<1$ we have

324: $-\tfrac12(\pi-\arccos(R))<p<\tfrac12(\pi-\arccos(R))$, and we must

325: exclude the two points $(x,y,z)=(0,0,\pm(1-R^2)^{1/2})$.

326: 

327: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

328: 

329: 

330: 

331: \subsection{Brief introduction to deformation quantisation}

332: 

333: We limit ourselves in this article to a the deformation quantisation

334: of algebraic manifolds and varieties with algebraic Poisson structures.

335: Let $\man$ smooth $m$ dimensional Poisson manifold or variety given by

336: %[

337: \begin{equation}

338: \begin{aligned}

339: \man=\Set{(x_1\sdotsc x_r)\in\Real^r

340: \bigg| 

341: F_s(x_1\sdotsc x_r)=0

342: ,\,

343: s=1\sdotsc r\!-\!m

344: }

345: \end{aligned}

346: \label{intr_gen_man}

347: \end{equation}

348: %]

349: where $F_s(x_1\sdotsc x_m)$ are polynomials. Let the Poisson structure

350: $\Poisson(\bullet,\bullet)$ be given by

351: $\Poisson(x_i,x_j)=C_{ij}(x_1\sdotsc x_m)$, where

352: $C_{ij}(x_1\sdotsc x_m)$ are also polynomials. Consistency implies

353: $\Poisson(x_i,F_s)=0$.  Let $C^\omega(\man)$ be the algebra of

354: complex analytic functions on $\man$ and let $C^\omega_0(\man)$ be the

355: subalgebra of polynomials in $(x_1\sdotsc x_m)$, this is dense in

356: $C^\omega(\man)$.

357: 

358: Let $\Balg$ be the free noncommutative

359: algebra generated by $\Set{\Ax_1\sdotsc \Ax_r,\Aeps}$, and define

360: the linear map $\Bpi:\Balg\mapsto C^\omega_0(\Real^r)$ where

361: $C^\omega_0(\Real^r)$ is the algebra of polynomials in

362: $(x_1\sdotsc x_n)$, via 

363: %[

364: \begin{align}

365: \Bpi(\Af+\Ag)=\Bpi(\Af)+\Bpi(\Ag)\sp

366: \Bpi(\Af\Ag)=\Bpi(\Af)\Bpi(\Ag)\sp

367: \Bpi(\lambda)=\lambda\sp

368: \Bpi(\Aeps)=0\sp

369: \Bpi(\Ax_i)=x_i

370: \end{align}

371: %]

372: Choose the elements $\AF_s\in\Balg$ for $s=1\sdotsc r\!-\!m$ and 

373: $\AC_{ij}\in\Balg$ for $i,j=1\sdotsc r$ such that 

374: $\Bpi(\AF_s)=F_s$ and $\Bpi(\AC_{ij})=C_{ij}$. 

375: We define the algebra $\Alg$ to be $\Balg$ quotiented by the noncommutative

376: polynomial relationships

377: %[

378: \begin{align}

379: [\Aeps,\Ax_i]=0\sp

380: \AF_s = 0 \sp

381: [\Ax_i,\Ax_j] = \Aeps\AC_{ij}

382: \label{intr_A_rels}

383: \end{align}

384: %]

385: for $i,j=1\sdotsc r$ and $s=1\sdotsc r-m$. 

386: The first equation in (\ref{intr_A_rels}) implies that $\Aeps$ is in

387: the centre of $\Alg$. 

388: We demand that $\Alg$ be an associative algebra. This imposes

389: restrictions on the possible choices of $\AF_s$ and 

390: $\AC_{ij}$, which we will not investigate here.

391: 

392: 

393: We can define the map

394: %[

395: \begin{align}

396: \pi:\Alg\mapsto C^\omega_0(\man)\,;\quad \pi(\Aeps)=0

397: \label{intr_pi}

398: \end{align}

399: %]

400: which is surjective. It is easy to see that this gives the Poisson 

401: structure via

402: %[

403: \begin{align}

404: \Poisson({\pi(\Af)},{\pi(\Ag)})=\pi\left(\frac{1}{i\Aeps}[\Af,\Ag]\right)

405: \end{align}

406: %]

407: Thus the following diagram commutes

408: %[

409: \begin{align}

410: \begin{array}{c}

411: \setlength{\unitlength}{1em}

412: \begin{picture}(17,6)

413: \put(0.7,5){$\Balg$}

414: \put(0,0){$C^\omega_0(\Real^r)$}

415: \put(1,4.5){\vector(0,-1){3.5}}

416: \put(-0.5,3){$\Bpi$}

417: %

418: \put(3.5,0.5){\vector(1,0){14}}

419: \put(2,5.5){\vector(1,0){15.5}}

420: \put(3,6){$/$\footnotesize 

421:            $\Set{[\Ax_i,\Aeps]=0,\, 

422:             \AF_s = 0,\, [\Ax_i,\Ax_j] = \Aeps\AC_{ij}}$}

423: \put(7,1){$/$\footnotesize $\Set{F_s = 0}$}

424: %

425: \put(18.5,5){$\Alg$}

426: \put(18,0){$C^\omega_0(\man)$}

427: \put(19,4.5){\vector(0,-1){3.5}}

428: \put(19.5,3){$\pi$}

429: \end{picture}

430: \end{array}

431: \end{align}

432: %]

433: We also demand that there is a conjugate structure $\dagger:\Alg\mapsto\Alg$

434: such that

435: %[

436: \begin{align}

437: (\Af\Ag)^\dagger=\Ag^\dagger\Af^\dagger

438: \sp

439: \lambda^\dagger=\cnj{\lambda} \text{ for } \lambda\in\Cmpx

440: \sp

441: \pi(\Af^\dagger)=\cnj{\pi(\Af)}

442: \sp

443: \Aeps^\dagger=\Aeps

444: \label{intr_conj}

445: \end{align}

446: %]

447: 

448: %%%%%%%%%%%%%%%%%%%%

449: 

450: We are interested in representations 

451: $\Psi:\Alg\mapsto\Lin(\Hil,\Hil)$ where $\Lin(\Hil,\Hil)$ is the space

452: of linear maps on the Hilbert space $\Hil$.

453: We demand that $\Psi$ is irreducible, and 

454: $\Psi(\Aeps)=\varepsilon\Idn$ where $\varepsilon\in\Real_+$.

455: If $\dim\Hil=n$ then $\Lin(\Hil,\Hil)\cong M_n(\Cmpx)$.  On the other hand if

456: $\dim\Hil=\infty$ then $\Lin(\Hil,\Hil)$ may contain unbounded

457: operators. We also demand that $\Psi$ preserve the conjugate structure

458: $\Psi(\Af^\dagger)=\Psi(\Af)^\dagger$ where the dagger on the right is

459: the Hermitian conjugate or adjoint.

460: 

461: We use the bra-ket notation so that if $\ket(\theta)\in\Hil$ then

462: $\Psi(\Af)\ket(\theta)$ the action of $\Af\in\Alg$ on $\ket(\theta)$

463: is written $\Af\ket(\theta)$.

464: Since $\Psi$ preserves the

465: conjugate we have

466: $\bra(\theta')\Af^\dagger\ket(\theta)=\cnj{\bra(\theta)\Af\ket(\theta')}$.

467: 

468: We sometimes want to recover the commutative structure from the matrix

469: algebras. This requires finding a sequence of representations

470: $\Psi_n:\Alg\mapsto M_n(\Cmpx)$, $\Psi_n(\Aeps)\to0$ as $n\to\infty$. 

471: However in general there is no canonical map $M_n(\Cmpx)\mapsto

472: M_{n+1}(\Cmpx)$. One exception being the noncommutative sphere

473: \cite{Gratus1}.

474: 

475: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

476: 

477: As stated above, in this article we give a shall give a one parameter

478: family of such algebras, whose representations can be compared to those for

479: the sphere and torus. Here we give a brief summary of these two

480: noncommutative geometries.

481: 

482: \vspace{1em}

483: 

484: \noindent

485: The \defn{Noncommutative Sphere} is generated by $\Ax,\Ay,\Az,\Aeps$ with 

486: %[

487: \begin{gather}

488: [\Ax,\Ay]=i\Aeps\Az\sp

489: [\Ay,\Az]=i\Aeps\Ax\sp

490: [\Az,\Ax]=i\Aeps\Ay\sp

491: \Ax^2+\Ay^2+\Az^2=1\sp

492: \Ax^\dagger=\Ax\sp

493: \Ay^\dagger=\Ay\sp

494: \Az^\dagger=\Az

495: \end{gather}

496: %]

497: The representations are the finite dimensional irreducible

498: representations of $SO(3)$. A basis for $\Hil$ is

499: $\ket(0)\sdotsc \ket(n-1)$ so that

500: %[

501: \begin{equation}

502: \begin{aligned}

503: &

504: \varepsilon=2(n^2-1)^{-1/2}

505: \sp &&

506: \Aa_+\ket(r)=\varepsilon(n-r-1)^{1/2}(r+1)^{1/2}\ket(r+1)

507: \sp\\ 

508: &

509: \Az\ket(r)=\varepsilon (r-\tfrac12(n-1))\ket(r)

510: \sp &&

511: \Aa_-\ket(r)=\varepsilon(n-r)^{1/2}r^{1/2}\ket(r-1)

512: \end{aligned}

513: \label{intr_S2_reps}

514: \end{equation}

515: %]

516: where $\Aa_+=\Ax+i\Ay$ and $\Aa_-=\Ax-i\Ay$ are called the ladder

517: elements. We note that the ladder operators, which are the

518: representations of the ladder elements, terminate $\Psi_n(\Aa_+)^n=0$.

519: We also note that $\Psi$ is unique for each $n$, and that

520: $\Psi_n(\Aeps)\to0$ as $n\to\infty$.

521: 

522: \vspace{1em}

523: 

524: In our language we the  \defn{Noncommutative Torus} is generated by

525: $\Set{\Ax_1,\Ax_2,\Ax_3,\Ax_4,\Aeps}$ with 

526: %[

527: \begin{equation}

528: \begin{aligned}

529: &\Ax_1^2+\Ax_2^2=1\sp \\

530: &\Ax_3^2+\Ax_4^2=1\sp \\

531: &[\Ax_1,\Ax_2]=0\sp \\

532: &[\Ax_3,\Ax_4]=0\sp 

533: \end{aligned}

534: \quad

535: \begin{aligned}

536: &[\Ax_1,\Ax_3]=-i\Aeps(\Ax_2\Ax_4+\Ax_4\Ax_2)\sp \\

537: &[\Ax_1,\Ax_4]=i\Aeps(\Ax_2\Ax_3+\Ax_3\Ax_2)\sp \\

538: &[\Ax_2,\Ax_3]=i\Aeps(\Ax_1\Ax_4+\Ax_4\Ax_1)\sp \\

539: &[\Ax_2,\Ax_4]=-i\Aeps(\Ax_1\Ax_3+\Ax_3\Ax_1)\sp 

540: \end{aligned}

541: \quad

542: \begin{aligned}

543: &\Ax_1^\dagger=\Ax_1\sp \\

544: &\Ax_2^\dagger=\Ax_2\sp \\

545: &\Ax_3^\dagger=\Ax_3\sp \\

546: &\Ax_4^\dagger=\Ax_4

547: \end{aligned}

548: \label{intr_T2_alg}

549: \end{equation}

550: %]

551: If we set $\Au=\Ax_1+i\Ax_2$, $\Av=\Ax_3+i\Ax_4$, and

552: $\Aq=(1+i\Aeps)/(1-i\Aeps)$ then we derive the usual noncommutative

553: torus.

554: %[

555: \begin{align}

556: \Au\Au^\dagger=\Au^\dagger\Au=1

557: \sp

558: \Av\Av^\dagger=\Av^\dagger\Av=1

559: \sp

560: \Au\Av=\Aq\Av\Au

561: \label{intr_T2_alg2}

562: \end{align}

563: %]

564: However as we have defined the algebra $\Alg$, the element $\Aq$ and

565: $\Aq^\dagger$ are not members of $\Alg$. To solve this problem we say

566: that $\Alg$ is generated by

567: $\Set{\Ax_1,\Ax_2,\Ax_3,\Ax_4,\Aeps,(1+\Aeps^2)^{-1}}$. 

568: 

569: There are both finite and infinite representations of the

570: noncommutative torus. The finite dimensional representations $\Psi_n$

571: have a basis $\ket(0)\sdotsc \ket(n-1)$

572: %[

573: \begin{equation}

574: \begin{aligned}

575: &\Psi_n(\Aq)=q\Idn\sp q=e^{2\pi i k/n} \sp 

576: \Au\ket(r)=e^{i(\beta + 2\pi r k i / n )}\ket(r) \sp

577: \\

578: &\Av\ket(r)=\ket(r+1) \sp \Av\ket(n-1)=\nu\ket(0)

579: \end{aligned}

580: \label{intr_T2_ref_fin}

581: \end{equation}

582: %]

583: where $n,k\in\Natn$ and where $\nu\in\Cmpx$, $|\nu|=1$ is a phase.

584: We impose that $n$ and $k$ are relatively prime, so that there are no

585: multiple eigenvalues of $\Psi(\Au)$. There also exist other more

586: complicated finite dimensional representations of the noncommutative

587: torus where $\Psi(\Au)$ has multiple eigenvalues.

588: 

589: We note that the ladder elements of this representation are given by

590: $\Av$ and $\Av^\dagger$ and that the ladder operators do not terminate

591: $\Psi_n(\Av)^m\ne0$ for all $n,m\in\Intg$.  To specify $\Psi_n$

592: completely requires giving $(n,k,\beta,\nu)$.  If $k=1$ then

593: $\Psi_n(\Aq)\to1$ as $n\to\infty$.

594: 

595: 

596: The infinite dimensional representations have a basis $\ket(r)$,

597: $r\in\Intg$, and are determined by the parameters

598: $\alpha,\beta\in\Real$ where $\alpha/2\pi$ is irrational

599: %[

600: \begin{equation}

601: \begin{aligned}

602: \Psi(\Aq)&=q\Idn\sp q=e^{i\alpha}

603: \sp

604: \Au\ket(r)=e^{i(r\alpha+\beta)}\ket(r)

605: \sp

606: \Av\ket(r)=\ket(r+1)

607: \end{aligned}

608: \label{intr_T2_ref_inf}

609: \end{equation}

610: %]

611: The eigenvalues of $\Psi(\Au)$ are dense on unit circle.

612: 

613:  

614: 

615: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

616: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

617: 

618: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

619: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

620: 

621: 

622: \section{$\Alg(R)$: The deformation algebra of polynomials on $\man(R)$}

623: \label{ch_alg}

624: 

625: In section \ref{ch_comm}, we define a one parameter set of immersions

626: $\man(R)\in\Real^3$. The commutative algebra of complex valued

627: polynomials in $(x,y,z)$ on $\man(R)$ is written $C^\omega_0(\man(R))$

628: and of course is dense in $C^\omega(\man(R))$. Here we give a one

629: parameter set of complex noncommutative algebras $\Alg(R)$ with

630: $R\in\Real$, which are the deformation quantisation of

631: $C^\omega_0(\man(R))$.  Each $\Alg(R)$ is generated by

632: $\Set{\Ax,\Ay,\Az,\Aeps,(1+\Aeps^2)^{-1}}$ with $\Aeps$ in the centre

633: of $\Alg(R)$. The reason for including $(1+\Aeps^2)^{-1}$ is similar

634: as for the noncommutative torus.  These are related via

635: %[

636: \begin{align}

637: [\Ax,\Ay] = i\Aeps\Az \sp

638: [\Ay,\Az] = i\Aeps(\Aw\Ax + \Ax\Aw)  \sp

639: [\Az,\Ax] = i\Aeps(\Aw\Ay + \Ay\Aw)

640: \label{alg_com}

641: \end{align}

642: %]

643: and

644: %[

645: \begin{align}

646: \Az^2+\Aw^2 = 1 

647: \label{alg_cass}

648: \end{align}

649: %]

650: where we define the element $\Aw\in\Alg(R)$ via

651: %[

652: \begin{align}

653: \Aw = \Ax^2 + \Ay^2 - R

654: \label{alg_def_w}

655: \end{align}

656: %]

657: The adjoint operation is given by

658: %[

659: \begin{align}

660: \Ax^\dagger = \Ax \sp

661: \Ay^\dagger = \Ay \sp

662: \Az^\dagger = \Az \sp

663: \Aw^\dagger = \Aw 

664: \label{alg_adj}

665: \end{align}

666: %]

667: It is easy to see that, assuming (\ref{alg_com}-\ref{alg_def_w})

668: define an associative algebra, then $\pi:\Alg(R)\mapsto

669: C^\omega_0(\man(R))$, where $\pi(\Ax)=x$, $\pi(\Ay)=y$,

670: $\pi(\Az)=z$. It also gives with the correct Poisson structure. We

671: shall show that these relationships define an associative algebra in

672: lemma \ref{thm_alg_ass} below. But first we need to define some new

673: elements of $\Alg(R)$ and derive some relationships which are valid if

674: $\Alg(R)$ is associative.

675: 

676: We define the ladder elements $\Aa_+,\Aa_-\in\Alg(R)$ via 

677: %[

678: \begin{align}

679: \Aa_+ = \Ax + i\Ay \sp 

680: \Aa_- = \Ax - i\Ay 

681: \label{alg_def_ap_am}

682: \end{align}

683: %]

684: In order to emphasise the circular nature of $\Aw$ and $\Az$ we shall

685: define the element $\Au\in\Alg(R)$ via

686: %[

687: \begin{align}

688: \Au = \Aw + i\Az \in\Alg(R)

689: \label{alg_def_u}

690: \end{align}

691: %]

692: and we show below that $\Au$ is unitary, i.e. $\Au^\dagger=\Au^{-1}$.

693: 

694: We define the ``pseudo element'' $\Aalpha$ via

695: %[

696: \begin{align}

697: \Aeps = \tan(\tfrac12\Aalpha)  

698: \end{align}

699: %]

700: and observe that $\Aalpha$ is not a member of $\Alg(R)$. However

701: since $\Aeps\in\Alg(R)$ and $(1+\Aeps^2)^{-1}\in\Alg(R)$ then

702: from the tan half angle formulae we have

703: $\sin(\Aalpha)=2\Aeps(1+\Aeps^2)^{-1}\in\Alg(R)$ and 

704: $\cos(\Aalpha)=(1-\Aeps^2)(1+\Aeps^2)^{-1}\in\Alg(R)$. Also 

705: $e^{\pm i n\Aalpha}=(\cos(\Aalpha)\pm i\sin(\Aalpha))^n$ so 

706: $e^{i n\Aalpha}\in\Alg(R)$ for all $n\in\Intg$.

707: 

708: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

709: Before showing that the algebra $\Alg(R)$ is associative we derive

710: some direct consequences of these definitions.

711: 

712: \begin{lemma}

713: From the above definitions, and the assumption that $\Alg(R)$ is

714: associative, we have the following relationships:

715: %[

716: \begin{align}

717: &

718: [\Az,\Aw] = 0

719: \label{alg_com_zw}

720: \\

721: &

722: \Au\Au^\dagger=\Au^\dagger\Au=1

723: \label{alg_u_udag}

724: \\

725: &

726: \Aa_+\Aa_- = \Aw + R +\Aeps \Az \sp

727: \Aa_-\Aa_+ = \Aw + R -\Aeps \Az

728: \label{alg_ap_am}

729: \\

730: &

731: [\Aw,\Aa_+] = -\Aeps(\Az\Aa_+ + \Aa_+\Az) \sp

732: [\Aw,\Aa_-] = +\Aeps(\Az\Aa_- + \Aa_-\Az)

733: \label{alg_com_wa}

734: \end{align}

735: %]

736: Also

737: %[

738: \begin{equation}

739: \begin{aligned}

740: &\Az\Aa_+ = \Aa_+(\cos(\Aalpha)\Az + \sin(\Aalpha)\Aw) 

741: \sp                                           

742: \Aw\Aa_+ = \Aa_+(-\sin(\Aalpha)\Az + \cos(\Aalpha)\Aw)

743: \\                                            

744: &\Az\Aa_- = \Aa_-(\cos(\Aalpha)\Az - \sin(\Aalpha)\Aw) 

745: \sp                                           

746: \Aw\Aa_- = \Aa_-(\sin(\Aalpha)\Az + \cos(\Aalpha)\Aw)

747: \end{aligned}

748: \label{alg_normal}

749: \end{equation}

750: %]

751: or alternatively

752: %[

753: \begin{align}

754: \Au\Aa_+=\Aa_+\Au e^{i\Aalpha} \sp

755: \Au\Aa_-=\Aa_-\Au e^{-i\Aalpha} \sp

756: \Au^\dagger\Aa_+=\Aa_+\Au^\dagger e^{-i\Aalpha} \sp

757: \Au^\dagger\Aa_+=\Aa_-\Au^\dagger e^{i\Aalpha}

758: \label{alg_normal_u}

759: \end{align}

760: %]

761: The general element $\Af\in\Alg(R)$ can be written uniquely in the form

762: %[

763: \begin{align}

764: \Af = \sum_{r=0}^N\sum_{s=-N}^N \Aa_+^r\Au^s\xi_{r,s}(\Aeps) +

765: \sum_{r=1}^N\sum_{s=-N}^N \Aa_-^r\Au^s\xi_{-r,s}(\Aeps) 

766: \label{alg_gen_el}

767: \end{align}

768: %]

769: for some $N\in\Natn$ where for all $r,s=-N\sdotsc N$ the function

770: $\xi_{r,s}(t)$ is a ration function in $t$ with denominator

771: $(1+t^2)^m$ for some $m\in\Natn$.

772: \end{lemma}

773: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

774: 

775: \begin{proof}

776: Equation (\ref{alg_ap_am}) follows automatically from

777: (\ref{alg_def_ap_am}),  (\ref{alg_def_w}) and the first equation 

778: in (\ref{alg_com}).

779: Equation (\ref{alg_com_wa}) follows from

780: %[

781: \begin{align*}

782: [\Aw,\Aa_+] 

783: = 

784: [\Aa_+\Aa_-  - R -\Aeps \Az,\Aa_+] 

785: =

786: \Aa_+[\Aa_-,\Aa_+] - \Aeps[\Az,\Aa_+]

787: =

788: -\Aeps(2\Aa_+\Az + [\Az,\Aa_+])

789: \end{align*}

790: %]

791: and likewise for $[\Aw,\Aa_-]$. For (\ref{alg_com_zw}) we have

792: %[

793: \begin{align*}

794: [\Az,\Aw] &= 

795: [\Az,\Aa_+\Aa_-  - R -\Aeps \Az] = 

796: [\Az,\Aa_+]\Aa_- + \Aa_+[\Az,\Aa_-] 

797: =

798: \Aeps ( \Aw\Aa_+\Aa_- + \Aa_+\Aw\Aa_-    

799:    - \Aa_+\Aw\Aa_- - \Aa_+\Aa_-\Aw) 

800: \\ &=

801: \Aeps ( \Aw\Aa_+\Aa_- - \Aa_+\Aa_-\Aw) =

802: -\Aeps [\Aa_+\Aa_- ,\Aw] =

803: -\Aeps[\Aw + R + \Aeps \Az,\Aw] =

804: -\Aeps^2[\Az,\Aw] 

805: \end{align*}

806: %]

807: hence $(1+\Aeps^2)[\Az,\Aw]=0$, so

808: $[\Az,\Aw]=(1+\Aeps^2)^{-1}(1+\Aeps^2)[\Az,\Aw]=0$.

809: The unitarity of $\Au$

810: (\ref{alg_u_udag}) follow from the definition of $\Au$ and the

811: commutativity of $\Az$ and $\Aw$.

812: 

813: From (\ref{alg_com}) and (\ref{alg_com_wa}) we can have 

814: %[

815: \begin{align*}

816: \Az\Aa_+ - \Aa_+\Az = + \Aeps \Aw\Aa_+ +\Aeps\Aa_+\Aw

817: \sp

818: -\Aeps\Az\Aa_+ - \Aeps\Aa_+\Az = \Aw\Aa_+ - \Aa_+\Aw

819: \end{align*}

820: %]

821: solving these as simultaneous equations and using the tan half angle

822: identity give the first of results of (\ref{alg_normal}). The other

823: identities in (\ref{alg_normal}) follow similarly. The identities in

824: (\ref{alg_normal_u}) are then the complex version of

825: (\ref{alg_normal}).

826: 

827: The generators all $\Alg(R)$ are all of the form (\ref{alg_gen_el}). Given

828: $\Af$ of this form, then $\Af\Au$, $\Af\Aeps$ and

829: $\Af(1-\Aeps^2)^{-1}$ are all of the form (\ref{alg_gen_el}). 

830: Also 

831: %[

832: \begin{align*}

833: \Af\Aa_+= 

834: \sum_{r=0}^N\sum_{s=-N}^N \Aa_+^{r+1}\Au^s e^{i s\Aalpha}\xi_{r,s}(\Aeps) +

835: \sum_{r=1}^N\sum_{s=-N}^N \Aa_-^{r-1}

836:   (\tfrac12(1+i\Aeps)\Au+\tfrac12(1-i\Aeps)\Au^{-1}+R)

837:    \Au^s e^{-i s\Aalpha}\xi_{-r,s}(\Aeps) 

838: \end{align*}

839: %]

840: So $\Af\Aa_+$ is of the form (\ref{alg_gen_el}), likewise for

841: $\Af\Aa_-$.

842: 

843: For uniqueness, from linearity, it is enough to show that if $\Af$ is

844: of the form (\ref{alg_gen_el}) and $\Af=0$ then $\xi_{r,s}=0$ for all

845: $r,s$. By multiplying $\Af$ with $(1+\Aeps^2)^M$ for sufficiently high

846: $M$ then we can assume $\xi_{r,s}$ are all polynomials. Let $m$ be the

847: largest degree of these polynomials. Now 

848: %[

849: \begin{align*}

850: 0 = \pi(\Af) = 

851: \sum_{r=0}^N\sum_{s=-N}^N \pi(\Aa_+)^r\pi(\Au)^s\xi_{r,s}(0) +

852: \sum_{r=1}^N\sum_{s=-N}^N \pi(\Aa_-)^r\pi(\Au)^s\xi_{-r,s}(0) 

853: \end{align*}

854: %]

855: so by looking at the coordinate system on $\man(R)$ this implies

856: $\xi_{r,s}(0)=0$ for all $r,s$. Thus

857: $\xi_{r,s}(\Aeps)=\Aeps\hat\xi_{r,s}(\Aeps)$ where

858: $\hat\xi_{r,s}(\Aeps)$ are polynomials of degree $\le m-1$. Continuing

859: this gives $\xi_{r,s}=0$

860: \end{proof}

861: 

862: \begin{lemma}

863: \label{thm_alg_ass}

864: The relationships (\ref{alg_def_w}) to (\ref{alg_normal_u}) define

865: an associative algebra.

866: \end{lemma}

867: \begin{proof}

868: Let $\SetS$ be set of all expressions of the form

869: (\ref{alg_gen_el}). We define a product $\cdot$ on $\SetS$ using the

870: above definitions.  

871: 

872: We wish to show that

873: $\Af_1\cdot(\Af_2\cdot\Af_3)=(\Af_1\cdot\Af_2)\cdot\Af_3$ for 

874: $\Af_1,\Af_2,\Af_3\in\SetS$. Where the inner bracket must be written

875: in the form (\ref{alg_gen_el}) first.

876: It is sufficient to show that 

877: $\Af_1\cdot(\Af_2\cdot\Af_3)=(\Af_1\cdot\Af_2)\cdot\Af_3$ 

878: where $\Af_1,\Af_2,\Af_3$ are

879: from the set $\Set{\Au,\Aa_+,\Aa_-,\Aeps,(1+\Aeps^2)^{-1}}$. This is

880: because any expression can be constructed from these five

881: elements. The element $\Au^\dagger=2\Aw-\Au$ with $\Aw$ given by

882: (\ref{alg_def_w}).

883: If $\Af_1,\Af_2$ or $\Af_3$ are either $\Aeps$ or $(1+\Aeps^2)^{-1}$

884: then the association relation holds since $\Aeps$ commutes with all

885: the generators.

886: 

887: The remaining 27 relationships must be proved in turn,

888: the interesting ones are

889: %[

890: \begin{align*}

891: (\Au\cdot\Aa_+)\cdot\Aa_- 

892: &= 

893: \Aa_+\Au e^{i\Aalpha} \cdot\Aa_- 

894: = 

895: (\tfrac12(1-i\Aeps)\Au+\tfrac12(1+i\Aeps)\Au^{-1}+R)\Au 

896: \\

897: &=

898: \Au(\tfrac12(1-i\Aeps)\Au+\tfrac12(1+i\Aeps)\Au^{-1}+R)

899: =

900: \Au\cdot(\Aa_+\cdot\Aa_-) 

901: \end{align*}

902: %]

903: and

904: %[

905: \begin{align*}

906: (\Aa_+\cdot\Aa_-)\cdot\Aa_+ 

907: &= 

908: (\tfrac12(1-i\Aeps)\Au+\tfrac12(1+i\Aeps)\Au^{-1}+R)\cdot\Aa_+ 

909: = 

910: \Aa_+(\tfrac12(1-i\Aeps)\Au e^{i\Aalpha}+

911:       \tfrac12(1+i\Aeps)\Au^{-1} e^{i\Aalpha}+R)

912: \\

913: &=

914: \Aa_+(\tfrac12(1+i\Aeps)\Au +

915:       \tfrac12(1-i\Aeps)\Au^{-1} +R)

916: =

917: \Aa_+\cdot(\Aa_-\cdot\Aa_+) 

918: \end{align*}

919: %]

920: 

921: \end{proof}

922: 

923: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

924: 

925: \section{Finite dimensional representations of $\Alg(R)$}

926: \label{ch_rep}

927: 

928: As mentioned in the introduction, we wish to find

929: irreducible representations over a finite Hilbert space $\Hil$ of 

930: $\Alg(R)$, i.e. $\Psi:\Alg(R)\mapsto \Lin(\Hil,\Hil)\cong M_n(\Cmpx)$. 

931: Such that $\Psi(\Aeps)=\varepsilon\Idn$ where

932: $\varepsilon\in\Real_+$ and such that $\Psi$

933: preserves the adjoint operator so that

934: $\Psi(\Af^\dagger)=\Psi(\Af)^\dagger$ where the dagger on the right

935: hand side is the Hermitian conjugate.

936: We note that since $\Psi(\Au)^\dagger\Psi(\Au)=\Idn$ we can

937: diagonalise $\Psi(\Au)$ and the eigenspaces of $\Psi(\Au)$ are

938: orthogonal. We shall further assume that $\Psi(\Au)$ has no multiple

939: eigenvalues, so the eigenspaces of $\Psi(\Au)$ are all one

940: dimensional.  This significantly simplifies the types of

941: representations.

942: 

943: \begin{theorem}

944: \label{th_reps}

945: Let $\Psi:\Alg(R)\mapsto\Lin(\Hil,\Hil)\cong M_n(\Cmpx)$ be an

946: irreducible adjoint preserving $n$ dimensional representation, such

947: that $\Psi(\Aeps)=\varepsilon\Idn$, $\varepsilon\in\Real_+$ and

948: $\Psi(\Au)$ has no multiple eigenvalues. Let $\alpha$ be given by

949: $\tan(\tfrac12\alpha)=\varepsilon$, $0<\alpha<\pi/2$. Then there

950: exists $\beta\in\Real$, such that we can label the orthonormal bases

951: for $\Hil$ which are the eigenspaces $\Psi(\Au)$

952: %[

953: \begin{align}

954: \ket(\beta)\,,\,

955: \ket(\beta+\alpha)\,,\,

956: \ket(\beta+2\alpha)\,,\,

957: \ldots\,,\,

958: \ket({\beta+(n-1)\alpha}) 

959: \label{rep_kets}

960: \end{align}

961: %]

962: where

963: %[

964: \begin{align}

965: \Au\ket(\beta+m\alpha) &= e^{i(\beta+m\alpha)}\ket(\beta+m\alpha)

966: \sp

967: m = 0\sdotsc n-1

968: \label{rep_u}

969: \end{align}

970: %]

971: There also exists a set of complex constants

972: $C_{\beta+m\alpha}\in\Cmpx$ for $m=0\sdotsc n$ satisfying

973: %[

974: \begin{align}

975: |C_{\beta+m\alpha}|^2 &= 

976: \sec(\tfrac12\alpha)\cos(\beta-\tfrac12\alpha+m\alpha) + R

977: \label{rep_C}

978: \end{align}

979: %]

980: and $\CC(\beta+m\alpha)\ne0$ for $m=1\sdotsc n-1$ so that

981: %[

982: \begin{equation}

983: \begin{aligned}

984: \Aa_-\ket(\beta+m\alpha) &= \CC(\beta+m\alpha)\ket({\beta+(m-1)\alpha})

985: \sp &&

986: m = 1\sdotsc n-1

987: \\

988: \Aa_+\ket(\beta+m\alpha) &= 

989: \cnj{\CC({\beta+(m+1)\alpha})}\ket({\beta+(m+1)\alpha})

990: \sp &&

991: m = 0\sdotsc n-2

992: \end{aligned}

993: \label{rep_a_pm}

994: \end{equation}

995: %]

996: 

997: The action of $\Aa_-$ and $\Aa_+$ on the first and last vectors of

998: (\ref{rep_kets}) respectively are given by either

999: %[

1000: \begin{align}

1001: \Aa_-\ket(\beta) = 0 \sp 

1002: \Aa_+\ket({\beta+(n-1)\alpha})=0

1003: \label{rep_S2_cond}

1004: \end{align}

1005: %]

1006: or

1007: %[

1008: \begin{align}

1009: \Aa_-\ket(\beta) =

1010: \CC(\beta)

1011: \ket({\beta+(n-1)\alpha})

1012: \sp

1013: \Aa_+\ket({\beta+(n-1)\alpha}) =

1014: \cnj{\CC(\beta)}

1015: \ket(\beta)

1016: \label{rep_T2_cond}

1017: \end{align}

1018: %]

1019: In the first case $\CC(\beta)=\CC({\beta+n\alpha})=0$ satisfies

1020: (\ref{rep_C}).

1021: In the second case $\CC(\beta)=\CC({\beta+n\alpha})\ne0$ satisfies

1022: (\ref{rep_C}) and

1023: %[

1024: \begin{align}

1025: n\alpha=2\pi k

1026: \label{rep_T2_cond_nk}

1027: \end{align}

1028: %]

1029: where $k\in\Natn$, $1\le k<n/2$ and $n$ and $k$ are relatively prime.

1030: \end{theorem}

1031: %

1032: 

1033: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1034: \begin{proof}

1035: 

1036: Let $\ket(\theta)$ be a normalised eigenvector of $\Psi(\Au)$ with

1037: eigenvalue $\lambda$ then

1038: $|\lambda|^2=\bra(\theta)\Au^\dagger\Au\ket(\theta)=1$ hence we can

1039: set $\lambda=e^{i\theta}$. From (\ref{alg_def_u}) we have

1040: $\Aw\ket(\theta)=\cos(\theta)\ket(\theta)$ and

1041: $\Az\ket(\theta)=\sin(\theta)\ket(\theta)$. 

1042: Let $N:\Real\mapsto\Real$ be given by

1043: $N(\theta)=\sec(\tfrac12\alpha)\cos(\theta-\tfrac12\alpha) + R$.

1044: From (\ref{alg_ap_am}) we have

1045: %[

1046: \begin{align*}

1047: \|\Aa_-\ket(\theta)\|^2 &=

1048: \bra(\theta)\Aa_+\Aa_-\ket(\theta)=

1049: \bra(\theta)(\Aw+\tan(\tfrac12\Aalpha)\Az+R)\ket(\theta)=

1050: \cos(\theta)+\tan(\tfrac12\alpha)\sin(\theta)+R = N(\theta)

1051: \\

1052: \|\Aa_+\ket(\theta)\|^2 &=

1053: \bra(\theta)\Aa_-\Aa_+\ket(\theta)=

1054: \bra(\theta)(\Aw-\tan(\tfrac12\Aalpha)\Az+R)\ket(\theta)=

1055: \cos(\theta)-\tan(\tfrac12\alpha)\sin(\theta)+R = N(\theta+\alpha)

1056: \end{align*}

1057: %]

1058: Thus if $\ket(\theta)$ is a eigenvector then $N(\theta)\ge0$ and

1059: $N(\theta+\alpha)\ge0$. Furthermore 

1060: $N(\theta)=0\ \Leftrightarrow\ \Aa_-\ket(\theta)=0$ and  

1061: $N(\theta+\alpha)=0\ \Leftrightarrow\ \Aa_+\ket(\theta)=0$.  

1062: 

1063: From (\ref{alg_normal_u}) we have

1064: %[

1065: \begin{align*}

1066: \Au\Aa_-\ket(\theta)=

1067: \Aa_-\Au e^{-i\Aalpha}\ket(\theta)=

1068: \Aa_- e^{-i\alpha}e^{i\theta}\ket(\theta)=

1069: e^{i(\theta-\alpha)}\Aa_-\ket(\theta)

1070: \end{align*}

1071: %]

1072: Hence if $N({\theta})\ne0$ then there must exist another normalised

1073: eigenvector $\ket(\theta')$ with eigenvalue $e^{i(\theta-\alpha)}$.

1074: If we let $\Aa_-\ket(\theta)=\CC(\theta)\ket(\theta')$ then

1075: $|\CC(\theta)|^2=N({\theta})$. Thus $\CC(\theta)$

1076: is determined from $N({\theta})$ up to a phase.

1077: 

1078: Similarly $\Au\Aa_+\ket(\theta')=

1079: e^{i(\theta'+\alpha)}\Aa_+\ket(\theta')=

1080: e^{i\theta}\Aa_+\ket(\theta')$. Since the eigenspaces of $\Psi(\Au)$

1081: are all one dimensional then $\Aa_+\ket(\theta')$ is parallel to

1082: $\ket(\theta)$. Thus setting $\Aa_+\ket(\theta')=D_\theta\ket(\theta)$ then 

1083: %[

1084: \begin{align*}

1085: D_\theta = 

1086: \bra(\theta')\Aa_+\ket(\theta) = 

1087: \cnj{\bra(\theta)\Aa_-\ket(\theta')} = 

1088: \cnj{\CC({\theta'})}

1089: \end{align*}

1090: %]

1091: hence $\Aa_+\ket(\theta')=\cnj{\CC({\theta'})}\ket(\theta)$.

1092: 

1093: 

1094: \vspace{1em}

1095: 

1096: {\bf Claim:} We now make the following claim.  Let $r\le n$ and there

1097: exists a set of independent normalised eigenvectors

1098: $\Set{\ket(\beta+m\alpha)\,|\,m=0\sdotsc r-1}$ such that

1099: $N(\beta+m\alpha)>0$ for $m=1\sdotsc r-1$.

1100: If either 

1101: %[

1102: \begin{align*}

1103: N(\beta)>0

1104: \textup{ and }

1105: \Aa_-\ket(\beta)=\CC({\beta})\ket({\beta+(r-1)\alpha}) 

1106: \textup{ for some choice of }\CC({\beta})

1107: \end{align*}

1108: %]

1109: or 

1110: %[

1111: \begin{align*}

1112: N(\beta)=0\,,

1113: \textup{ and }

1114: N(\beta+r\alpha)=0

1115: \end{align*}

1116: %]

1117: then $r=n$

1118: 

1119: \vspace{0.5em}

1120: 

1121: {\bf Proof of claim:} 

1122: Let $V_r=\textup{span}\Set{\ket(\beta+m\alpha)\,|\,m=0\sdotsc r-1}$.

1123: Since $N(\beta+m\alpha)>0$ for $m=1\sdotsc n-1$

1124: then we can define $\CC(\beta+m\alpha)\ne0$ for $m=1\sdotsc n-1$ 

1125: so that $\Aa_-\ket(\beta+m\alpha) =

1126: \CC(\beta+m\alpha)\ket({\beta+(m-1)\alpha})$ for 

1127: $m = 1\sdotsc r-1$ and 

1128: $\Aa_+\ket(\beta+m\alpha) = 

1129: \cnj{\CC({\beta+(m+1)\alpha})}\ket({\beta+(m+1)\alpha})$

1130: for $m = 0\sdotsc n-2$

1131: 

1132: For the first option 

1133: $\Aa_-\ket(\beta)=\CC({\beta})\ket({\beta+(r-1)\alpha})$  

1134: for so $\CC({\beta})\ne0$,

1135: so from the argument above

1136: $\Aa_+\ket({\beta+(r-1)\alpha})=\cnj{\CC({\beta})}\ket(\beta)$.

1137: 

1138: For the second option

1139: $\Aa_-\ket(\beta)=0$ and

1140: $\Aa_+\ket({\beta+(r-1)\alpha})=0$.

1141: 

1142: Hence, for both cases, the action of $\Au,\Aa_+,\Aa_-$ on $V_r$

1143: remains in $V_r$.  Thus if $r<n$ then it is obvious that $\Psi$ can be

1144: reduced to $V_r$.  This contradicts the irreducibility of $\Psi$ or the

1145: dimension $\Hil$ is $n$. Hence $r=n$.

1146: 

1147: \vspace{1em}

1148: {\bf Continuation of Proof:}

1149: 

1150: Since $N({\theta})\ge0$ for all eigenvectors $\ket(\theta)$ we have

1151: two possibilities. Either $N({\theta})>0$ for eigenvectors

1152: $\ket(\theta)$ or there exists a

1153: $\ket(\beta)$ such that $N(\beta)=0$. 

1154: 

1155: Taking the first case, that $N({\theta})|>0$ all eigenvectors

1156: $\ket(\theta)$. Choose $\beta$ so that $\ket(\beta)$ is any normalised

1157: eigenvector of $\Psi(\Au)$. We choose the phases of $\CC(\beta+m\alpha)$,

1158: and define $\ket(\beta+m\alpha)=\cnj{\CC({\beta+m\alpha})}^{-1}

1159: \Aa_+\ket({\beta+(m-1)\alpha})$ for $m=1\sdotsc n-1$.

1160: 

1161: From the claim above the $\Set{\ket(\beta+m\alpha)|m=0\sdotsc n-1}$

1162: are independent. Since $N(\beta)>0$ then $\Aa_-\ket(\beta)$ must be

1163: parallel to one of the this set. But from the claim the only

1164: possibility is

1165: $\Aa_-\ket(\beta)=\CC({\beta})\ket({\beta+(n-1)\alpha})$ for some

1166: choice of phase of $\CC({\beta})$.

1167: i.e.  (\ref{rep_T2_cond}).

1168: 

1169: Given condition (\ref{rep_T2_cond}) then

1170: %[

1171: \begin{align*}

1172: e^{i\beta}\cnj{\CC(\beta)}\ket(\beta) &=

1173: \Au \cnj{\CC(\beta)}\ket(\beta) = 

1174: \Au\Aa_+\ket({\beta+(n-1)\alpha})=

1175: e^{i\alpha}\Aa_+\Au\ket({\beta+(n-1)\alpha}) \\ &=

1176: e^{i\alpha}e^{i(\beta+(n-1)\alpha)}\Aa_+\ket({\beta+(n-1)\alpha}) = 

1177: e^{i(\beta+n\alpha)}\cnj{\CC(\beta)}\ket(\beta)

1178: \end{align*}

1179: %]

1180: Hence $e^{i n\alpha}=1$ so $n\alpha=2\pi k$ for some integer

1181: $k$. Clearly $k\le1$ and $k<n/2$ so that $0<\alpha<\pi$. Also $k$ and

1182: $n$ must be relatively prime so that $\Psi(\Au)$ has distinct

1183: eigenvalues 

1184: i.e. (\ref{rep_T2_cond_nk}).

1185: 

1186: Now consider the second possibility. That there exist a $\ket(\beta)$

1187: such that $N(\beta)=0$ and hence $\Aa_-\ket(\beta)=0$. 

1188: Again we choose the phases of $\CC({\beta+m\alpha})$ and define 

1189: $\ket(\beta+m\alpha)=\cnj{\CC({\beta+m\alpha})}^{-1}

1190: \Aa_+\ket({\beta+(m-1)\alpha})$

1191: for $m=1\sdotsc n-1$. These are all independent and none of the

1192: $\CC({\beta+m\alpha})=0$ by the claim above.

1193: 

1194: If $\Aa_+\ket({\beta+(n-1)\alpha})\ne0$ then by the claim above

1195: $\Aa_+\ket({\beta+(n-1)\alpha})=\cnj{\CC({\beta+n\alpha})}\ket(\beta)$.

1196: Hence

1197: %[

1198: \begin{align*}

1199: |\CC({\beta+n\alpha})|^2=

1200: \bra({\beta+(n-1)\alpha})\Aa_-\Aa_+\ket({\beta+(n-1)\alpha})=

1201: \cnj{\CC({\beta+n\alpha})}\,

1202: \bra({\beta+(n-1)\alpha})\Aa_-\ket(\beta)=

1203: 0

1204: \end{align*}

1205: %]

1206: hence $\CC({\beta+n\alpha})=0$ which contradicts

1207: $\Aa_+\ket({\beta+(n-1)\alpha})\ne0$. Hence (\ref{rep_S2_cond})

1208: \end{proof}

1209: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1210: 

1211: 

1212: By analogy with the representations of the noncommutative sphere these

1213: we shall call representations which satisfy (\ref{rep_S2_cond})

1214: \defn{$S^2$-type representations} and we say a 

1215: $S^2$-type representation is

1216: \defn{minimal} if

1217: %[

1218: \begin{align}

1219: n\alpha < 2\pi

1220: \label{rep_S2_min}

1221: \end{align}

1222: %]

1223: By analogy to the representations of the noncommutative torus we shall

1224: call representations which satisfy (\ref{rep_T2_cond})

1225: \defn{$T^2$-type representations} and we say a $T^2$-type representation

1226: is \defn{minimal} if

1227: %[

1228: \begin{align}

1229: n\alpha=2\pi

1230: \label{rep_T2_min}

1231: \end{align}

1232: %]

1233: We now wish to find what values of $(R,n,\alpha,\beta)$

1234: give rise to finite dimensional irreducible representations of

1235: $\Alg(R)$, with no multiple eigenvalues of $\Psi(\Au)$. 

1236: A summary is given in table \ref{fig_concl} below.

1237: 

1238: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1239: 

1240: \subsection{$S^2$-type representations}

1241: 

1242: Before we look at the different types of $S^2$-type representation and

1243: when they exist we shall give some basic facts about $S^2$-type

1244: representation.

1245: 

1246: \begin{lemma}

1247: \label{lm_S2_basic}

1248: If $\Psi$ is an $S^2$ representation with $(R,n,\alpha,\beta)$ given in

1249: theorem \ref{th_reps} then we can find a basis

1250: (\ref{rep_kets}) such that $\CC(\beta+m\alpha)\in\Real_+$.

1251: Also there exist an equivalent representation $\tilde\Psi$ with that same

1252: values of $(R,n,\alpha)$ and with $\beta$ replaced by $\tilde\beta$ 

1253: where $-2\pi<\tilde\beta-\tfrac12\alpha\le0$.

1254: 

1255: If $\Psi_1$ and $\Psi_2$ are $S^2$ representation with the same

1256: $(R,n,\alpha,\beta)$ then they are equivalent.

1257: \end{lemma}

1258: 

1259: \begin{proof}

1260: We can see that given $\alpha$ and $\beta$ the basis (\ref{rep_kets})

1261: is only defined up to a phase, likewise $\CC(\beta+m\alpha)$ is also

1262: only defined up to a phase. Thus given the set

1263: $\Set{\nu_m\in\Cmpx\,|\,\nu_m\cnj{\nu_m}=1\,,\,m=0\sdotsc n-1}$, we

1264: can always make the following replacement

1265: %[

1266: \begin{align}

1267: \ket({\beta+m\alpha})\to 

1268: \ket({\beta+m\alpha})' = 

1269: \nu_m\ket({\beta+m\alpha}) 

1270: \sp 

1271: \CC(\beta+m\alpha)\to 

1272: \CC(\beta+m\alpha)' = 

1273: \nu_{m-1}\cnj{\nu_m}\CC(\beta+m\alpha)

1274: \label{rep_phase}

1275: \end{align}

1276: %]

1277: without changing the equations (\ref{rep_u}) and

1278: (\ref{rep_a_pm}). Thus if we set

1279: $\nu_m=\nu_{m-1}\CC(\beta+m\alpha)/|\CC(\beta+m\alpha)|$ then all the

1280: $\CC(\beta+m\alpha)'$ are real and positive. 

1281: 

1282: Clearly replacing $\beta$ with $\tilde\beta\to\beta+2\pi k$ for 

1283: $k\in\Natn$ doesn't change the

1284: representation. Therefore we can place $\tilde\beta$ in any $2\pi$

1285: range. The one chosen makes the calculations below simpler.

1286: 

1287: If $\Psi_1$ and $\Psi_2$ are representation with the same

1288: $(R,n,\alpha,\beta)$ then, setting the $\CC(\beta+m\alpha)$ to be real

1289: and positive, the bases (\ref{rep_kets}) are the same (up

1290: to an overall choice phase), and the action (\ref{rep_u}),

1291: (\ref{rep_a_pm}) on these basis elements

1292: are the same, therefore the representations are equivalent.

1293: \end{proof}

1294: 

1295: For the rest of this subsection we assume we are given $R$ and $n$,

1296: and we wish to find $\alpha$ and $\beta$ so that

1297: $\Psi(R,n,\alpha,\beta)$ is $S^2$-type irreducible representation.  

1298: Given $\alpha$ and $\beta$ we write $\beta'=\beta-\tfrac12\alpha$.

1299: For a general $S^2$ representation (\ref{rep_C}) and 

1300: (\ref{rep_S2_cond}) imply

1301: %[

1302: \begin{align}

1303: \cos(\beta')=\cos(\beta'+n\alpha)=-R\cos(\tfrac12\alpha)

1304: \label{rep_lm_S2_cond}

1305: \end{align}

1306: %]

1307: The first equality is solved by setting $\beta'+n\alpha=2\pi k \pm

1308: \beta'$ for some $k\in\Natn$. This implies either 

1309: %[

1310: \begin{align}

1311: \alpha=2\pi k/n

1312: \label{rep_lm_S2_cond1}

1313: \end{align}

1314: %]

1315: or 

1316: %[

1317: \begin{align}

1318: \beta'=\pi k -\tfrac12 n \alpha

1319: \label{rep_lm_S2_cond2}

1320: \end{align}

1321: %]

1322: We can now place some simple constraints on $(R,n,\alpha,\beta)$ such

1323: that $\Psi(R,n,\alpha,\beta)$ is an irreducible $S^2$ representation.

1324: From (\ref{rep_lm_S2_cond}) we see that for $\Psi$ to be an $S^2$

1325: representation then  

1326: %[

1327: \begin{align}

1328: R\le\sec(\tfrac12\alpha)=(1+\varepsilon^2)^{1/2}

1329: \end{align}

1330: %]

1331: since $\varepsilon=\tan(\tfrac12\alpha)$. By looking at

1332: (\ref{rep_C}) we see that $\Psi(R,n,\alpha,\beta)$ is an irreducible 

1333: $S^2$ representation if and only if

1334: %[

1335: \begin{align}

1336: \cos(\beta'+m\alpha) + R\cos(\tfrac12\alpha) > 0

1337: \sp

1338: m=1\sdotsc n-1

1339: \label{rep_S2_ineq}

1340: \end{align}

1341: 

1342: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%]

1343: \subsubsection*{Minimal $S^2$ representation}

1344: 

1345: The first result is on the existence and uniqueness of minimal

1346: $S^2$-type representations.

1347: \begin{lemma}

1348: \label{lm_S2min}

1349: Given $R$\, and $n$ there exists minimal $S^2$-type representation if and only

1350: if \mbox{$-1<R<\sec(\pi/n)$}. This representation is unique and is given by

1351: %[

1352: \begin{align}

1353: \cos(\tfrac12n\alpha)+R\cos(\tfrac12\alpha)=0

1354: \sp

1355: 0<\alpha<2\pi/n

1356: \sp

1357: \beta'=-\tfrac12 n\alpha

1358: \label{rep_minS2}

1359: \end{align}

1360: %]

1361: \end{lemma}

1362: 

1363: \begin{proof}

1364: If $\Psi$ is minimal we must exclude (\ref{rep_lm_S2_cond1}).

1365: Since we choose $\beta'$ so that $-2\pi<\beta'\le0$ from

1366: (\ref{rep_lm_S2_cond2}) we have $-2<k-n\alpha/(2\pi)\le0$. Applying

1367: $0<n\alpha/(2\pi)<1$ we have

1368: %[

1369: \begin{align*}

1370: -2<n\alpha/(2\pi)<k\le 1+n\alpha/(2\pi) < 1

1371: \end{align*}

1372: %]

1373: therefore $k=0$ or $-1$. 

1374: 

1375: We shall exclude the case $k=-1$. If $k=-1$ so that

1376: $\beta'=-\pi-\tfrac12 n\alpha$, therefore $-2\pi<\beta'<-\pi$. Now

1377: $\beta'+\alpha=-\pi-\tfrac12(n-2)\alpha\le-\pi$ since $n\ge2$. Thus

1378: $-2\pi<\beta'<\beta'+\alpha\le -\pi$. Since $\cos$ is strictly

1379: decreasing in the range $-2\pi\ldots-\pi$ we have

1380: $-1\le\cos(\beta'+\alpha)<\cos(\beta')<1$. Thus

1381: $\cos(\beta'+\alpha)-\cos(\beta')=\cos(\beta'+\alpha)+R\cos(\tfrac12\alpha)<0$

1382: which contradicts (\ref{rep_S2_ineq}).

1383: 

1384: Thus if $\Psi$ is a minimal $S^2$ representation we have

1385: (\ref{rep_minS2}). 

1386: 

1387: Now consider the function

1388: $\hat{R}(\alpha)=-\cos(\tfrac12n\alpha)/\cos(\tfrac12\alpha)$ for the

1389: range $0<\alpha<2\pi/n$. 

1390: We observe that $\hat{R}(0)=-1$ and $\hat{R}(2\pi/n)=\sec(\pi/n)$ and

1391: %[

1392: \begin{align*}

1393: \hat{R}'(\alpha)&=

1394: (\cos(\tfrac12\alpha))^{-2}(

1395: \tfrac12n\alpha\cos(\tfrac12\alpha)\sin(\tfrac12n\alpha)-

1396: \tfrac12\alpha\cos(\tfrac12n\alpha)\sin(\tfrac12\alpha))

1397: \\

1398: &=

1399: (\cos(\tfrac12\alpha))^{-2}

1400: \tfrac12\alpha((n-1)\cos(\tfrac12\alpha)\sin(\tfrac12n\alpha)

1401: + \sin(\tfrac12(n-1)\alpha) ) > 0

1402: \end{align*}

1403: %]

1404: Thus

1405: $\hat{R}:\Set{\alpha|0<\alpha<2\pi/n}\mapsto\Set{R|-1<R<\sec(\pi/n)}$ 

1406: is an invertible function so we can uniquely solve

1407: (\ref{rep_minS2}). This also implies that $-1<R<\sec(\pi/n)$.

1408: 

1409: We now wish to show that an $\alpha,\beta'$ satisfying

1410: (\ref{rep_minS2}) is a representation. This simply requires showing

1411: (\ref{rep_S2_ineq}) is satisfied. Consider separately the cases

1412: $m=1\sdotsc\lfloor\tfrac12n\rfloor$ and

1413: $m=\lfloor\tfrac12n\rfloor+1\sdotsc n-1$. In the first case we have

1414: $-\tfrac12n<m-\tfrac12n\le0$ this implies $-\pi\le-\tfrac12n\alpha<

1415: (m-\tfrac12n)\alpha \le0$.  Since $\cos$ is strictly increasing in this

1416: range we have

1417: $-1\le\cos(-\tfrac12n\alpha)<\cos(m\alpha-\tfrac12n\alpha)\le1$ and

1418: hence (\ref{rep_S2_ineq}) is satisfied.  Likewise if

1419: $m=\lfloor\tfrac12n\rfloor+1\sdotsc n-1$ then $0\le

1420: m-\tfrac12n<\tfrac12n$ so $0\le

1421: m\alpha-\tfrac12n\alpha < \tfrac12n\alpha \le \pi$.  Since $\cos$ is

1422: strictly decreasing in this range we have

1423: $-1\le\cos(\tfrac12n\alpha)<\cos(m\alpha-\tfrac12n\alpha)\le1$ and

1424: hence (\ref{rep_S2_ineq}) is satisfied.

1425: \end{proof}

1426: 

1427: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1428: 

1429: \begin{figure}

1430: \setlength{\unitlength}{6cm}

1431: \begin{picture}(1,1.1)(0,-0.1)

1432: \put(0,0){

1433: \includegraphics[height=\unitlength]{min_S2_rng.eps}

1434: }

1435: \put(0.2,0.8){$\alpha$}

1436: \put(0.8,0.0){$R$}

1437: \put(0.7,0.6){\makebox(0,0)[lt]{$R=\sec(\alpha)$}}

1438: \put(0,0){\makebox(0,0)[lt]{Figure \ref{fig_min_S2}a}}

1439: \end{picture}

1440: %

1441: \qquad

1442: \setlength{\unitlength}{5cm}

1443: \begin{picture}(1,1.2)(0,-0.2)

1444: \put(0,0){

1445: \includegraphics[height=\unitlength]{min_S2A.eps}}

1446: \put(0,0){\makebox(0,0)[tl]

1447: {\tabelm{Figure \ref{fig_min_S2}b: $n=4,\,\alpha=.5,$

1448:          \\$R = -.557,\,\beta=-1$}}}

1449: \put(0.1,0.55){\tabelm{forbidden\\sector}}

1450: \end{picture}

1451: %

1452: \qquad

1453: \begin{picture}(1,1.2)(0,-0.2)

1454: \put(0,0){

1455: \includegraphics[height=\unitlength]{min_S2B.eps}}

1456: \put(0,0){\makebox(0,0)[tl]

1457: {\tabelm{Figure \ref{fig_min_S2}c: $n = 8\,, \alpha = .7,$

1458:                             \\$R = 1.003,\, \beta = -2.8$}}}

1459: \put(0.,0.55){\tabelm{forbidden\\sector}}

1460: \end{picture}

1461: \caption{The minimal $S^2$ representation. Figure \ref{fig_min_S2}a

1462: gives the relationship between $R$, $n$ and $\alpha$ for

1463: $n=2,3,\ldots$ Figures \ref{fig_min_S2}b and \ref{fig_min_S2}c are

1464: represent different representations.}

1465: \label{fig_min_S2}

1466: \end{figure}

1467: 

1468: The minimal $S^2$ representation may be said to be 

1469: the closest to the representation

1470: of the true noncommutative sphere, since they exist for $-1<R<1$ when

1471: the commutative limit is a topologically the sphere, 

1472: they are uniquely determined for each $R$

1473: by $n$, and $\varepsilon=\tan(\tfrac12\alpha)\to0$ as $n\to\infty$. 

1474: 

1475: As lemma \ref{lm_S2min} shows there exists a minimal

1476: $S^2$ representation of dimension $n$ for $-1<R<\sec(\pi/n)$. The

1477: relationship between $R$, $n$ and $\alpha$ is shown in Figure

1478: \ref{fig_min_S2}a,  $n=2,3,\ldots$. Given a permitted $R$ and $n$ we

1479: can draw a diagram to represent $\Psi$. This an open polygon inside a

1480: circle. The vertices are at the points

1481: $e^{i(\beta'+m\alpha)}$. Equations (\ref{rep_S2_ineq}) implies that

1482: there is a sector (called the ``forbidden sector'') and all the

1483: vertices must lie to the right of this sector. The open polygon begins

1484: and ends on the edge of this sector. Figures \ref{fig_min_S2}b and

1485: \ref{fig_min_S2}c are represent representations for different $n$ and

1486: $R$. Note for Figure \ref{fig_min_S2}c, $R>1$. 

1487: 

1488: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1489: \subsubsection*{Non minimal $S^2$ representation}

1490: 

1491: The situation for non minimal $S^2$ representation is more

1492: complicated. We have the following lemmas: 

1493: 

1494: \begin{lemma}

1495: \label{lm_S2_non_min_R}

1496: If $\Psi$ is a non minimal $S^2$-type representation then $R\ge1$.

1497: \end{lemma}

1498: \begin{proof}

1499: If $\ket(\theta)$ is a basis vector then from (\ref{rep_S2_ineq})

1500: $\cos(\theta)\ge-R\cos(\tfrac12\alpha)$. Therefore if we set

1501: $\theta'=(\theta)\!\!\!\!\mod 2\pi$ so that $0\le\theta'<2\pi$, then $\theta'$

1502: must lie either in the range $0\le\theta'\le\pi-\tfrac12\delta$ or

1503: $\pi+\tfrac12\delta\le\theta'<2\pi$ where

1504: $\cos(\tfrac12\delta)=R\cos(\tfrac12\alpha)$ and $0<\delta$. 

1505: The range

1506: $\Set{\theta'|\pi-\tfrac12\delta<\theta'<\pi+\tfrac12\delta}$ is

1507: called the forbidden sector.

1508: 

1509: Since $\Psi$ is non minimal there exists an $m$ such that

1510: $0\le(\beta'+m\alpha)\!\!\!\!\mod2\pi\le\pi-\tfrac12\delta$ and

1511: $\pi+\tfrac12\delta\le(\beta'+(m+1)\alpha)\!\!\!\!\mod 2\pi<2\pi$. Thus

1512: $0<\delta\le\alpha<\pi$. Since $\cos$ is decreasing we have

1513: $0<\cos(\tfrac12\alpha)\le\cos(\tfrac12\delta)<1$. Hence result.

1514: \end{proof}

1515: 

1516: 

1517: 

1518: \begin{figure}

1519: \begin{center}

1520: \setlength{\unitlength}{6cm}

1521: \begin{picture}(1,1.1)(0,0)

1522: \put(0,0){

1523: \includegraphics[height=\unitlength]{NS_rngA.eps}}

1524: \put(0.0,0.8){$\alpha$}

1525: \put(0.8,0.0){$R$}

1526: \put(0.3,0.7){\makebox(0,0)[lt]{$R=\sec(\alpha)$}}

1527: \put(0,0){\makebox(0,0)[lt]{Figure \ref{fig_non_min_S2}a}}

1528: \end{picture}

1529: \qquad

1530: \setlength{\unitlength}{6cm}

1531: \begin{picture}(1,1.1)(0,0)

1532: \put(0,0){

1533: \includegraphics[height=\unitlength]{NS_rngB.eps}}

1534: \put(0.0,0.8){$\alpha$}

1535: \put(0.95,0.0){$R$}

1536: \put(0.6,0.3){\makebox(0,0)[lt]{$R=\sec(\alpha)$}}

1537: \put(0,0){\makebox(0,0)[lt]{Figure \ref{fig_non_min_S2}b}}

1538: \end{picture}

1539: \end{center}

1540: \caption{Non minimal $S^2$ representation, the relationship between

1541: $R$ and $\alpha$ for $n=11$}

1542: \label{fig_non_min_S2}

1543: \end{figure}

1544: 

1545: \begin{figure}[t]

1546: %\setlength{\unitlength}{4.3cm}

1547: \setlength{\unitlength}{4.3cm}

1548: \begin{picture}(1,1.3)(0,-0.3)

1549: \put(0,0){

1550: \includegraphics[height=\unitlength]{NS_state_A.eps}}

1551: \put(0,0){\makebox(0,0)[tl]

1552: {\tabelm{

1553: Figure \ref{fig_S2_circ}a: (good) \\

1554: $N=11,\alpha=2.20$,\\$R = 1.97,\beta=-2.67$}}}

1555: \put(0.08,0.52){FS}

1556: \end{picture}

1557: %

1558: \begin{picture}(1,1.3)(0,-0.3)

1559: \put(0,0){

1560: \includegraphics[height=\unitlength]{NS_state_C.eps}}

1561: \put(0,0){\makebox(0,0)[tl]

1562: {\tabelm{

1563: Figure \ref{fig_S2_circ}b: (good) \\

1564: $N=11,\alpha=2\pi (3/11)$,\\$R = 1.50,\beta=-2.95$}}}

1565: \put(0.25,0.48){FS}

1566: \end{picture}

1567: %

1568: \begin{picture}(1,1.3)(0,-0.3)

1569: \put(0,0){

1570: \includegraphics[height=\unitlength]{NS_state_B.eps}}

1571: \put(0,0){\makebox(0,0)[tl]

1572: {\tabelm{

1573: Figure \ref{fig_S2_circ}c: (bad) \\

1574: $N=11,\alpha=2.40$,\\$R = 2.22,\beta=-3.77$}}}

1575: \put(0.05,0.48){FS}

1576: \end{picture}

1577: %

1578: \begin{picture}(1,1.3)(0,-0.3)

1579: \put(0,0){

1580: \includegraphics[height=\unitlength]{NS_state_D.eps}}

1581: \put(0,0){\makebox(0,0)[tl]

1582: {\tabelm{

1583: Figure \ref{fig_S2_circ}d: (bad) \\

1584: $N=11,\alpha=2\pi (3/11)$,\\$R = 1.10,\beta=-2.37$}}}

1585: \put(0.3,0.53){FS}

1586: \end{picture}

1587: \caption{Permissible and non permissible values of

1588: $n,R,\alpha,\beta$. (FS=forbidden sector)}

1589: \label{fig_S2_circ}

1590: \end{figure}

1591: 

1592: 

1593: \begin{lemma}

1594: If $R\le-1$ there exist no $S^2$-type representations.

1595: \end{lemma}

1596: \begin{proof}

1597: Follows from lemmas \ref{lm_S2min} and \ref{lm_S2_non_min}.

1598: \end{proof}

1599: 

1600: \begin{lemma}

1601: \label{lm_S2_non_min}

1602: If $R\ge1$ then there may exists many non minimal $S^2$-type

1603: representation. Given such a $\Psi(R,n,\alpha,\beta)$ then

1604: $1\le R\le\sec(\tfrac12\alpha)$ and

1605: $-\tfrac32\pi<\beta'<-\tfrac12\pi$ and also

1606: $(R,n,\alpha,\beta)$ obey either

1607: %[

1608: \begin{align}

1609: &

1610: \cos(\tfrac12n\alpha) = (-1)^{k+1} R\cos(\tfrac12\alpha)

1611: \sp

1612: 2\pi/n<\alpha<\pi/2

1613: \sp

1614: \beta'=\pi k - \tfrac12 n\alpha

1615: \sp

1616: k = [n\alpha/(2\pi)]-1

1617: \label{rep_S2_a}

1618: \end{align}

1619: %]

1620: or

1621: %[

1622: \begin{align}

1623: \alpha = 2k\pi/n

1624: \sp

1625: 1\le k \le (n-1)/2

1626: \sp

1627: \cos(\beta')=-R\cos(\pi k/n)

1628: \label{rep_S2_b}

1629: \end{align}

1630: %]

1631: \end{lemma}

1632: \begin{proof}

1633: From (\ref{rep_lm_S2_cond}) and $|\cos(\beta')|\le1$ we have 

1634: $R\le\sec(\tfrac12\alpha)$. Also since $R>0$ and

1635: $\cos(\tfrac12\alpha)>0$ then $\cos(\beta')<0$ then

1636: $-\tfrac32\pi<\beta'<-\tfrac12\pi$.

1637: 

1638: From (\ref{rep_lm_S2_cond}) we have either (\ref{rep_lm_S2_cond1}) or

1639: (\ref{rep_lm_S2_cond2}) is true. If (\ref{rep_lm_S2_cond2}) is true

1640: then from (\ref{rep_lm_S2_cond}) the first equation in

1641: (\ref{rep_S2_a}) is true. Also $\beta'/\pi=k-n\alpha/(2\pi)$ so

1642: $[\beta'/\pi]=-1=k-[n\alpha/(2\pi)]$.

1643: 

1644: If (\ref{rep_lm_S2_cond1}) is satisfied, then from

1645: (\ref{rep_lm_S2_cond}) equation (\ref{rep_S2_b}) is satisfied.  

1646: \end{proof}

1647: 

1648: 

1649: We note that the converse is to lemma \ref{lm_S2_non_min} is not

1650: true. That is given $(R,n,\alpha,\beta)$ which satisfies either

1651: (\ref{rep_S2_a}) or (\ref{rep_S2_b}) there need not be a corresponding

1652: representation $\Psi(R,n,\alpha,\beta)$. This is due to the

1653: requirement that (\ref{rep_S2_ineq}) is satisfied.  For example given

1654: $(R,n,\alpha,\beta)$ which is a solution to (\ref{rep_S2_a}) such that

1655: $n$ is even and $k$ is old, then for $m=n/2$ we have

1656: %[

1657: \begin{align*}

1658: \cos(\beta'+m\alpha) + R\cos(\tfrac12\alpha)=

1659: \cos(\pi k - \tfrac12 n \alpha + m \alpha) - \cos(\beta') =

1660: \cos(\pi k)-\cos(\beta') = -1-\cos(\beta') < 0

1661: \end{align*}

1662: %]

1663: hence (\ref{rep_S2_ineq}) is not satisfied.

1664: 

1665: In general that exact set of $(R,n,\alpha,\beta)$ which have non

1666: minimal $S^2$ representation is complicated. In figure

1667: \ref{fig_non_min_S2} we see the relationship between $R$ and $\alpha$

1668: for $n=11$. Figure \ref{fig_non_min_S2}b is simply a smaller

1669: region of $R$. Figure \ref{fig_S2_circ} gives four different representation

1670: with $n=11$, the first two are acceptable since (\ref{rep_S2_ineq}) is

1671: satisfied, whereas the second two are unacceptable since

1672: (\ref{rep_S2_ineq}) is not satisfied. Figures \ref{fig_S2_circ}a and

1673: \ref{fig_S2_circ}c correspond to (\ref{rep_S2_a}). Figures

1674: \ref{fig_S2_circ}b and \ref{fig_S2_circ}d correspond to

1675: (\ref{rep_S2_b})

1676: 

1677: 

1678: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1679: 

1680: \begin{figure}

1681: \setlength{\unitlength}{5cm}

1682: \begin{picture}(1.1,1.4)(-0.1,-0.4)

1683: \put(0,0){

1684: \includegraphics[height=\unitlength]{TT_min_rng.eps}}

1685: \put(0.0,0.8){$\alpha$}

1686: \put(0.9,-0.01){$R$}

1687: \put(1,1){\makebox(0,0)[rt]{$R=\sec(\alpha)$}}

1688: \put(0.1,-0.05){\makebox(0,0)[lt]

1689:        {\tabelm{Figure \ref{fig_T2}a: \\ 

1690:         Minimal $T^2$ representation,\\ 

1691:         Relationship between $R,n,\alpha$, \\

1692:         for $n=2,3,\ldots$}}}

1693: \end{picture}

1694: %

1695: \qquad

1696: \begin{picture}(1.1,1.3)(-0.1,-0.4)

1697: \put(0,0){

1698: \includegraphics[height=\unitlength]{TT_rng_A.eps}}

1699: \put(0.0,0.8){$\alpha$}

1700: \put(0.95,-0.01){$R$}

1701: \put(1,1){\makebox(0,0)[rt]{$R=\sec(\alpha)$}}

1702: \put(0.1,-0.05){\makebox(0,0)[lt]

1703:          {\tabelm{Figure \ref{fig_T2}b: \\

1704:          Relationship between $R,\alpha$ \\ for $n=11$.}}}

1705: \end{picture}

1706: %

1707: \qquad

1708: \begin{picture}(1,1.3)(0,-0.4)

1709: \put(0,0){

1710: \includegraphics[height=\unitlength]{TT_state_A.eps}}

1711: \put(0,0){\makebox(0,0)[tl]

1712: {\tabelm{

1713: Figure \ref{fig_T2}c: \\

1714: $N=11,\alpha=2\pi (3/11)$,\\$R = 1.10,\beta=-2.9$}}}

1715: \put(0.25,0.48){FS}

1716: \end{picture}

1717: \caption{Permissible region for $T^2$ representation and an example.}

1718: \label{fig_T2}

1719: \end{figure}

1720: 

1721: \subsection{$T^2$-type representations}

1722: 

1723: Again before we look at the different types of $T^2$-type

1724: representation and when they exist, we shall give some basic facts

1725: about $T^2$-type representation. Recall that for a $T^2$-type

1726: representation $\alpha=2\pi k/n$.

1727: 

1728: We note take we can still make the replacement given by

1729: (\ref{rep_phase}) for $m=0\sdotsc n-1$ with

1730: $\nu_{-1}=\nu_{n-1}$. However in this case the constant 

1731: $C_{\textup{prod}} = \prod_{m=0}^{n-1} \CC(\beta+m\alpha)$

1732: is unchanged under the replacement

1733: (\ref{rep_phase}). i.e. $C_{\textup{prod}}\to

1734: C_{\textup{prod}}'=C_{\textup{prod}}$. 

1735: Thus there is a resulting phase $\nu=C_{\textup{prod}}/|C_{\textup{prod}}|$.

1736: 

1737: 

1738: \begin{lemma}

1739: If $\Psi_1$ and $\Psi_2$ are $T^2$ representation with the same

1740: $(R,n,k,\beta,\nu)$ then they are equivalent.

1741: 

1742: If $\Psi$ is a $T^2$ representation we can find an

1743: equivalent representation $\tilde\Psi$ with the same 

1744: $(R,n,k,\nu)$ and with 

1745: $\beta$ replaced by $\tilde\beta$ such that 

1746: $\pi-2\pi/n<\tilde\beta-\tfrac12\alpha\le\pi$.

1747: \end{lemma}

1748: 

1749: \begin{proof}

1750: \label{lm_T2_basic}

1751: Since $(R,n,k,\nu)$ are the same for $\Psi_1$ and

1752: $\Psi_2$ then $\alpha=2\pi k/n$ is the same for $\Psi_1$ and

1753: $\Psi_2$.  By the same argument in lemma \ref{lm_S2_basic} we can

1754: choose basis elements (\ref{rep_kets}) such that

1755: $\CC(\beta+m\alpha)\in\Real_+$ for $m=1\sdotsc n-1$ and thus

1756: $\CC(\beta)=\nu|\CC(\beta)|$. Doing this for both $\Psi_1$ and

1757: $\Psi_2$ then the action (\ref{rep_u}), (\ref{rep_a_pm}) on these

1758: basis elements are the same, therefore the representations are

1759: equivalent.

1760: 

1761: In the proof of theorem \ref{th_reps} in the case of a $T^2$

1762: representation when all $\CC(\theta)\ne0$ we choose $\beta$ to be any

1763: value such that $e^{i\beta}$ was an eigenvalue of $\Psi(\Au)$. Since

1764: the $n$ roots are equally spaced around the circle we can choose any

1765: arc of length $2\pi/n$ to place $\beta$ in.  The one chosen makes the

1766: calculations below simpler.

1767: \end{proof}

1768: 

1769: For this subsection we assume we are given $(R,n,k,\nu)$, and

1770: we wish to find $\beta$ so that $\Psi(R,n,k,\beta,\nu)$

1771: is $T^2$-type irreducible representation. Again we define

1772: $\beta'=\beta-\tfrac12\alpha$.

1773: 

1774: \begin{lemma}

1775: Given $R$, $n$ and $k$ such that $R<\cos(\pi/n)\sec(\pi k/n)$ there

1776: are no $T^2$ representation $\Psi(R,n,k,\beta,\nu)$.

1777: 

1778: Given $R$, $n$ and $k$ such that 

1779: $\cos(\pi/n)\sec(\pi k/n)<R\le\sec(\pi k/n)$

1780: then there exist a one parameter set of $T^2$ representation

1781: $\Psi(R,n,k,\beta,\nu)$ with $\beta$ in the range

1782: %[

1783: \begin{align}

1784: \pi-2\pi/n+\tfrac12\delta<\beta'<\pi-\tfrac12\delta

1785: \sp

1786: \textup{ where }

1787: \cos(\tfrac12\delta)=R\cos(\pi k/n)

1788: \label{rep_T2_beta}

1789: \end{align}

1790: %]

1791: 

1792: Given $R$, $n$ and $k$ such that $R>\sec(\pi k/n)$ there exist a

1793: parameter set of irreducible $T^2$ representation

1794: $\Psi(R,n,k,\beta,\nu)$ with the full range of $\beta$,

1795: i.e. $\pi-2\pi/n<\beta'\le\pi$.

1796: \end{lemma}

1797: \begin{proof}

1798: By looking at (\ref{rep_C}) we see that $\Psi(R,n,k,\beta,\nu)$ is an

1799: irreducible $T^2$ representation if only if

1800: %[

1801: \begin{align}

1802: \cos(\beta'+2\pi mk/n) + R\cos(\pi k/n) > 0

1803: \sp

1804: m=0\sdotsc n-1

1805: \label{rep_T2_ineq}

1806: \end{align}

1807: %]

1808: hence if $R>\sec(\pi k/n)$, (\ref{rep_T2_ineq}) is satisfied for 

1809: all $\beta'$, hence $\Psi(R,n,k,\beta,\nu)$ is a representation for

1810: all $\beta$.

1811: 

1812: For $R\le\sec(\pi k/n)$ then

1813: similar to lemma \ref{lm_S2_non_min_R} if we set

1814: $\theta'=(\beta'+2\pi mk/n)\!\!\!\!\mod 2\pi$ then $\theta'$ must lie either

1815: in the range 

1816: $0\le\theta'<\pi-\tfrac12\delta$ or $\pi+\tfrac12\delta<\theta'<2\pi$

1817: where $\delta$ is given in

1818: (\ref{rep_T2_beta}).  Since the set $\Set{e^{i(\beta'+2\pi

1819: mk/n)}|m=0\sdotsc n=1}$ are equally spaced then $\delta<

1820: 2\pi/n$. Hence $0<\tfrac12\delta<\pi/n\le\tfrac12\alpha<\tfrac12\pi$

1821: and $\cos$ is strictly decreasing we have

1822: $0<\cos(\pi/n)<\cos(\tfrac12\delta)<1$. Since

1823: $\cos(\tfrac12\delta)=R\cos(\pi k/n)$ we have $R>\cos(\pi/n)\sec(\pi

1824: k/n)$.

1825: 

1826: Since $\beta'\le\pi$ then $\beta'\le\pi-\tfrac12\delta$. 

1827: Now there exists and $m$ such that

1828: $e^{i(\beta'+2\pi m k/n)}=e^{i(\beta'+2\pi/n)}$, hence

1829: $(\beta'+2\pi mk/n)\!\!\!\! \mod

1830: 2\pi=\beta'+2\pi/n$ hence $\beta'+2\pi/n$ must lie either in the range

1831: $0\le\beta'+2\pi/n<\pi-\tfrac12\delta$ or the range 

1832: $\pi+\tfrac12\delta<\beta'+2\pi/n<2\pi$ but since $\beta'+2\pi/n>\pi$

1833: we have $\beta'>\pi+\tfrac12\delta+2\pi/n$. Thus for $\Psi$ to be $T^2$

1834: representation we must have $\beta'$ given by (\ref{rep_T2_beta}).

1835: 

1836: If $\beta'$ is given by (\ref{rep_T2_beta}) then all the $\theta'$ lie

1837: in the permitted regions hence (\ref{rep_T2_ineq}) so it is a

1838: representation. 

1839: \end{proof}

1840: 

1841: We can see that $\Psi$ is a minimal $T^2$ representation if

1842: $k=1$. Thus we have the following corollary

1843: \begin{corrol}

1844: If  $R\le1$ there are no $T^2$ representation. 

1845: 

1846: Given $R$ and $n$ such that $1<R\le\sec(\pi/n)$ then there exist a one

1847: parameter set of irreducible $T^2$ minimal representation

1848: $\Psi(R,n,k=1,\beta,\nu)$ with $\beta$ given in (\ref{rep_T2_beta}).

1849: 

1850: Given $R$ and $n$ such that $R>\sec(\pi/n)$ there exist a one

1851: parameter set of irreducible $T^2$ representation

1852: $\Psi(R,n,k=1,\beta,\nu)$ with the full range of $\beta$,

1853: i.e. $\pi-2\pi/n<\beta'\le\pi$.

1854: \end{corrol}

1855: The range of possible $(R,\alpha)$ for minimal $T^2$ representation is

1856: simply $R>1$ and $\alpha=2\pi/n$. These are pictures in figure

1857: \ref{fig_T2}a. Figure \ref{fig_T2}b gives the range of possible

1858: $(R,\alpha)$ for $n=11$. Figure \ref{fig_T2}c gives an example of a

1859: $T^2$ representation.

1860: 

1861: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1862: 

1863: \subsection{Infinite dimensional $T^2$ representations}

1864: 

1865: Like the noncommutative torus $\Alg(R)$ also have infinite dimensional

1866: representations.

1867: We shall consider here only the infinite dimensional

1868: representations of the form (with $\Hil$ having the basis

1869: $\Set{\ket(\beta+m\alpha)\,|\,m\in\Intg}$)

1870: %[

1871: \begin{equation}

1872: \begin{aligned}

1873: &\Au\ket(\beta+m\alpha)=e^{i(\beta+m\alpha)}\ket(\beta+m\alpha) 

1874: \sp

1875: \Aa_-\ket(\beta+m\alpha) =

1876: \CC(\beta+m\alpha)\ket({\beta+(m-1)\alpha})

1877: \sp

1878: \\&

1879: \Aa_+\ket(\beta+m\alpha) =

1880: \CC({\beta+(m+1)\alpha})\ket({\beta+(m+1)\alpha})

1881: \end{aligned}

1882: \label{rep_inf}

1883: \end{equation}

1884: %]

1885: where 

1886: %[

1887: \begin{align}

1888: C_{\beta+m\alpha} &= 

1889: (\sec(\tfrac12\alpha)\cos(\beta'+m\alpha) + R)^{1/2}

1890: \sp

1891: \beta'=\beta-\tfrac12\alpha

1892: \end{align}

1893: %]

1894: and where $\alpha\ne2\pi n/k$ for any $n,k\in\Intg$.

1895: The eigenvalues of $\Psi(\Au)$ are dense on the circle, so there exist

1896: an infinite dimensional representations if only if

1897: $R\ge\sec(\alpha)$.

1898: 

1899: If $R=\sec(\alpha)$ there also exist semi-infinite

1900: dimensional $T^2$ representation. These are given by $\beta'=-\pi$. Thus

1901: $\Aa_-\ket(\beta)=0$ and $\Aa_+\ket(\beta-\alpha)=0$ so we can reduce the

1902: Hilbert space to the subspaces

1903: $\textup{span}\Set{\ket(-\pi+r\alpha),r\ge0}$ and 

1904: $\textup{span}\Set{\ket(-\pi+r\alpha),r\le-1}$. 

1905: 

1906: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1907: 

1908: \newcounter{fff}

1909: \def\footn{{\stepcounter{fff}${}^{\alph{fff}}$}}

1910: \def\footnn{{${}^{\alph{fff}}$}}

1911: 

1912: \newlength{\tlen}

1913: \newlength{\clen}

1914: \setlength{\tlen}{3.5cm}

1915: \setlength{\clen}{1.6cm}

1916: \setlength{\unitlength}{1em}

1917: 

1918: %\def\Qno{{n}}

1919: %\def\Qyes{{y}}

1920: \def\Qno{{\LARGE $\chi$}}

1921: 

1922: \def\Qyes{\makebox[3em][l]{\LARGE \raisebox{-0.2em}{

1923:            {\raisebox{-0.4em}{$\ddot\smile\!\!\!$}\circle{2}}}}}

1924: \def\Re{{R_\varepsilon}}

1925: 

1926: \begin{table}[t]

1927: \begin{minipage}{\textwidth}

1928: \begin{tabular}{|p{\tlen}|p{\clen}|p{\clen}|p{\clen}|p{\clen}|p{\clen}|p{\clen}|p{\clen}|}

1929: \hline

1930: &

1931: Null &

1932: Point &

1933: Sphere &

1934: Variety &

1935: \multi{Sphere-\\Torus} &

1936: \multi{Sphere-\\Torus\\bdd} &

1937: Torus 

1938: \\

1939: \hline

1940: \multi{$R$ range\\\ $\Re=(1\!+\!\varepsilon^2)^{1/2}$} &

1941: {\footnotesize{$R\le-1$}} &

1942: {\footnotesize{$R=-1$}} &

1943: {\footnotesize{$-1\!<\!R\!<\!1$}} & 

1944: {\footnotesize{$R=1$}} &

1945: \multi{\footnotesize$1<R<\Re$} &

1946: \multi{\footnotesize$R\!=\Re$} &

1947: \multi{\footnotesize$R\!>\Re$}

1948: \\

1949: \hline

1950: \multi{Topology of\\manifold $\man(R)$}

1951: &

1952: Null\footn

1953: &

1954: Point

1955: &

1956: $S^2$ top\footn

1957: &

1958: \multi{2dim\\variety\footn}

1959: &

1960: -\footn

1961: &

1962: -\footnn

1963: &

1964: $T^2$

1965: \\

1966: \hline

1967: \multi{Minimal $S^2$\\representations.}

1968: & \Qno

1969: & \Qno

1970: & \Qyes

1971: & \Qyes

1972: & \Qyes

1973: & \Qno

1974: & \Qno

1975: \\

1976: \hline

1977: \multi{Non minimal $S^2$\\representations.}

1978: & \Qno

1979: & \Qno

1980: & \Qno

1981: & \Qno

1982: & \Qyes

1983: & \Qyes

1984: & \Qno

1985: \\

1986: \hline

1987: \multi{Finite dim  $T^2$\\ representations}

1988: & \Qno

1989: & \Qno

1990: & \Qno

1991: & \Qno

1992: & \Qyes\footn

1993: & \Qyes\footn

1994: & \Qyes

1995: \\

1996: \hline

1997: \multi{Semi $\infty$-dim $T^2$ \\representations}

1998: & \Qno

1999: & \Qno

2000: & \Qno

2001: & \Qno

2002: & \Qno

2003: & \Qyes

2004: & \Qno

2005: \\

2006: \hline

2007: \multi{$\infty$-dim $T^2$ \\representations}

2008: & \Qno

2009: & \Qno

2010: & \Qno

2011: & \Qno

2012: & \Qno

2013: & \Qyes

2014: & \Qyes

2015: \\

2016: \hline

2017: \end{tabular}

2018: \footnotesize{ 

2019: \begin{tabular}{lll}

2020: Notes: & {a: No solution to equation} 

2021: & {b: $\man(R)$ is convex if $R<0$.} \\

2022: & {c: The point $(x,y,z)=(0,0,0)$ not smooth.} 

2023: & {d: Since $(1+\varepsilon^2)^{1/2}=1$ so no sphere-torus exists.} \\

2024: & {e: $\beta$ range is limited.} 

2025: & {f: $\beta$ range excludes one point.}

2026: \end{tabular}

2027: }

2028: \end{minipage}

2029: \caption{Summary of possible representations for different $R$ and

2030: $\varepsilon$.} 

2031: \label{fig_concl}

2032: \end{table}

2033: 

2034: \section{Conclusion and Discussion}

2035: \label{ch_concl}

2036: Table \ref{fig_concl} gives a list of all the possible

2037: representations. 

2038: We can see that representations reflect the topology, but not

2039: completely. 

2040: 

2041: There are loosely speaking four regions of $R$.

2042: If $R\le-1$ then there is either no manifold $\man(R)$ or

2043: it is just a point. Consequently there are no representations either.

2044: 

2045: The next region $-1<R<1$ the algebra $\Alg(R)$ is closest to the

2046: noncommutative sphere. The commutative limit is the sphere, and

2047: there exist only minimal $S^2$ representations

2048: which are the closest to the representations of the noncommutative

2049: sphere, in that they are parameterised by $n$. 

2050: 

2051: In the region $1<R<(1+\varepsilon^2)^{1/2}$ the algebra $\Alg(R)$ is a

2052: new object which we may call the ``sphere-torus''. This is a purely

2053: noncommutative object since setting $\varepsilon=0$ gives $1<R<1$ so

2054: there is no commutative analogue. In this region $\Alg(R)$ has

2055: the minimal $S^2$ representations, the non minimal $S^2$

2056: representations and the finite dimensional $T^2$ representations. 

2057: 

2058: Only when $R>(1+\varepsilon^2)^{1/2}$ can we say that

2059: $\Alg(R)$ is a deformation of the torus. There are no $S^2$

2060: representations and, as well as the finite dimensional $T^2$

2061: representations, the infinite $T^2$ representations exist.

2062: 

2063: If any definition of genus is applied to noncommutative geometry it

2064: would be interesting to see what values it would attain for the

2065: sphere-torus.

2066: 

2067: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2068: \section*{Acknowledgements}

2069: 

2070: The author would like to thank Arne Sletsj{\o}e Math Department 

2071: University of Oslo, John Madore Laboratoire de Physique Th{\'e}orique 

2072: Universit{\'e} de Paris Sud and Richard Kerner Laboratoire

2073: Physique Th{\'e}orique des Liquides, Universit{\'e} Pierre et Marie

2074: Curie, Paris VI for help in preparing this paper. 

2075: 

2076: The author would like to thank the the math department of the 

2077: University of Oslo for their facilities. 

2078: 

2079: 

2080: 

2081: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2082: \begin{thebibliography}{9}

2083: 

2084: \bibitem{Madore_book} J. Madore 1995, 

2085: {\it An Introduction to Noncommutative Differential Geometry and its

2086: Physical Applications},

2087:  Cambridge University Press.

2088: 

2089: \bibitem{Sternheimer1} D. Sternheimer,

2090: {\it Deformation Quatization: Twenty Years After}

2091: math.QA/9809056 (And references therein.)

2092: 

2093: \bibitem{Gratus1} J. Gratus, 

2094: {\it A Natural Basis of States for the Noncommutative Sphere 

2095: and its Moyal bracket} 

2096: J. Maths. Phys {\bf 38} (8), (Aug 1997) 4283 - 4300, 

2097: q-alg/9703038

2098: 

2099: \end{thebibliography}

2100: 

2101: \end{document}

2102: 

2103: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2104: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2105: