1: \documentclass[11pt]{amsart}

2:  \usepackage{amsbsy,amssymb,amscd,amsfonts,latexsym,amstext,delarray,

3:  amsmath,graphicx}

4: \input{amssymb.sty}

5: \usepackage[dvips]{epsfig}

6: 

7: %%change height page

8: \setlength{\textwidth}{15.5cm}

9: %%change width page

10: \setlength{\oddsidemargin}{0.5cm} \setlength{\topmargin}{.5cm}

11: \setlength{\evensidemargin}{\oddsidemargin}

12: 

13: \input xypic

14: 

15: 

16: 

17: 

18: \newtheorem{thm}{Theorem}[section]

19: \newtheorem{prop}[thm]{Proposition}

20: \newtheorem{cor}[thm]{Corollary}

21: \newtheorem{lem}[thm]{Lemma}

22: \newtheorem{defn}[thm]{Definition}

23: \newtheorem{rem}[thm]{Remark}

24: \newtheorem{ex}[thm]{Example}

25: 

26: \numberwithin{equation}{section}

27: 

28: \def\bA{{\mathbb A}}

29: \def\bB{{\mathbb B}}

30: \def\bC{{\mathbb C}}

31: \def\bD{{\mathbb D}}

32: \def\bE{{\mathbb E}}

33: \def\bF{{\mathbb F}}

34: \def\bG{{\mathbb G}}

35: \def\bH{{\mathbb H}}

36: \def\bI{{\mathbb I}}

37: \def\bJ{{\mathbb J}}

38: \def\bK{{\mathbb K}}

39: \def\bL{{\mathbb L}}

40: \def\bM{{\mathbb M}}

41: \def\bN{{\mathbb N}}

42: \def\bO{{\mathbb O}}

43: \def\bP{{\mathbb P}}

44: \def\bQ{{\mathbb Q}}

45: \def\bR{{\mathbb R}}

46: \def\bS{{\mathbb S}}

47: \def\bT{{\mathbb T}}

48: \def\bU{{\mathbb U}}

49: \def\bV{{\mathbb V}}

50: \def\bW{{\mathbb W}}

51: \def\bX{{\mathbb X}}

52: \def\bY{{\mathbb Y}}

53: \def\bZ{{\mathbb Z}}

54: 

55: \def\A{{\mathbb A}}

56: \def\C{{\mathbb C}}

57: \def\R{{\mathbb R}}

58: \def\F{{\mathbb F}}

59: \renewcommand{\H}{{\mathbb H}}

60: \def\K{{\mathbb K}}

61: \def\N{{\mathbb N}}

62: \renewcommand{\P}{{\mathbb P}}

63: \def\Q{{\mathbb Q}}

64: \def\Z{{\mathbb Z}}

65: \def\Zb{{\mathbb Z}}

66: \def\s{\sigma}

67: 

68: \def\cal{\mathcal}

69: \def\gr{\mathrm{gr}}

70: \def\oplusinf{\mathop{\oplus}}

71: \def\otimesinf{\mathop{\otimes}}

72: \def\id{{\mbox{Id}}}

73: \def\ch{{\mbox{ch}}}

74: \def\im{{\mbox{Im}}}

75: \def\Aut{{\mbox{Aut}}}

76: \def\dim{{\mbox{dim}}}

77: \def\ker{{\mbox{Ker}}}

78: \def\Der{{\mbox{Der}}}

79: \def\Hom {{\mbox{Hom}}}

80: \def\hom {{\mbox{hom}}}

81: \def\End{{\mbox{End}}}

82: \def\Trace{{\mbox{Trace}}}

83: \def\Int{{\mbox{Int}}}

84: \def\alg{{\mbox{alg}}}

85: \def\Lie{{\mbox{Lie}}}

86: \def\Ad{{\mbox{Ad}}}

87: \def\Spin{{\mbox{Spin}}}

88: \def\wh{\widehat}

89: 

90: \def\cali{{\cal I}}

91: \def\calp{{\cal P}}

92: \def\cala{{\cal A}}

93: \def\calb{{\cal B}}

94: \def\calk{{\cal K}}

95: \def\call{{\cal L}}

96: \def\calm{{\cal M}}

97: \def\caln{{\cal N}}

98: \def\calo{{\cal O}}

99: \def\calh{{\cal H}}

100: \def\cals{{\cal S}}

101: \def\calc{{\cal C}}

102: \def\cale{{\cal E}}

103: \def\calu{{\cal U}}

104: \def\calf{{\cal F}}

105: \def\calr{{\cal R}}

106: \def\calV{{\cal V}}

107: \def\caly{{\cal Y}}

108: \def\fraca{{\mathfrak A}}

109:  \def\fraci{{\mathfrak I}}

110:  \def\fracg{{\mathfrak g}}

111: \def\frach{{\mathfrak h}}

112: \def\fracd{{\mathfrak d}}

113: \def\fracS{{\mathfrak {S}}}

114:   \def\fracd{{\mathfrak d}}

115:  \def\fracsu{{\mathfrak {su}}}

116:  \def\fracsl{{\mathfrak {sl}}}

117: \def\fracso{{\mathfrak {so}}}

118: \def\calg{{C_{\mathrm{alg}}}}

119: \def\bbbone{\mbox{\rm 1\hspace {-.6em} l}}

120: \def\ue{\underline{E}}

121: \def\uf{\underline{F}}

122: \def\uk{\underline{K}}

123: \def\ux{\underline{X}}

124: \def\gg{{\mathbf g}}

125: \def\ag{{\mathbf a}}

126: \def\ug{{\mathbf u}}

127: \def\mg{{\mathbf m}}

128: \def\vg{{\mathbf v}}

129: \def\ung{{\mathbf 1}}

130: \def\Yg{{\mathbf Y}}

131: \def\Tg{{\mathbf T}}

132: \def\Lg{{\mathbf L}}

133: \def\Rg{{\mathbf R}}

134: \def\vg{{\mathbf v}}

135: \def\hg{{\mathbf H}}

136: %\def\alg{{\mathbf{Alg}}}

137: \def\vect{{\mathbf{Vect}}}

138: \def\Homg {{\mathbf{Hom}}}

139: \def\homg {{\mathbf{hom}}}

140: \def\endg{{\mathbf{end}}}

141: 

142: 

143: \def\af{{\mathfrak a}}

144: \newcommand{\fg}{{\mathfrak{g}}}

145: \newcommand{\fh}{{\mathfrak{h}}}

146: \def\m{{\mathfrak m}}

147: \def\n{{\mathfrak n}}

148: \def\p{{\mathfrak p}}

149: \def\fU{{\mathfrak U}}

150: 

151: \def\cutint{{\int \!\!\!\!\!\! -}}

152: \def\qqq{{\,,\quad \forall}}

153: \def\cA{{\mathcal A}}

154: \def\cB{{\mathcal B}}

155: \def\cC{{\mathcal C}}

156: \def\cD{{\mathcal D}}

157: \def\cE{{\mathcal E}}

158: \def\cF{{\mathcal F}}

159: \def\cG{{\mathcal G}}

160: \def\cH{{\mathcal H}}

161: \def\cI{{\mathcal I}}

162: \def\cJ{{\mathcal J}}

163: \def\cK{{\mathcal K}}

164: \def\cL{{\mathcal L}}

165: \def\cM{{\mathcal M}}

166: \def\cN{{\mathcal N}}

167: \def\cO{{\mathcal O}}

168: \def\cP{{\mathcal P}}

169: \def\cQ{{\mathcal Q}}

170: \def\cR{{\mathcal R}}

171: \def\cS{{\mathcal S}}

172: \def\cT{{\mathcal T}}

173: \def\cU{{\mathcal U}}

174: \def\cV{{\mathcal V}}

175: \def\cX{{\mathcal X}}

176: \def\cY{{\mathcal Y}}

177: \def\cZ{{\mathcal Z}}

178: 

179: \newcommand{\ie}{{\it i.e.\/}\ }

180: \newcommand{\eg}{{\it e.g.\/}\ }

181: \newcommand{\cf}{{\it cf.\/}\ }

182: \newcommand{\opcit}{{\it op.cit.\/}\ }

183: \newcommand{\resp}{{\it resp.\/}\ }

184: \def\text{\hbox}

185: 

186: \def\Aut{{\rm Aut}}

187: \def\Coker{{\rm Coker}}

188: \def\dist{{\rm dist}}

189: \def\Dom{{\rm Dom}}

190: \def\End{{\rm End}}

191: \def\Ext{{\rm Ext}}

192: \def\Gal{{\rm Gal}}

193: \def\GL{{\rm GL}}

194: \def\Gr{{\rm Gr}}

195: \def\Hom{{\rm Hom}}

196: \def\id{{\rm id}}

197: \def\Ind{{\rm Ind}}

198: \def\Index{{\rm Ind}}

199: \def\Ker{{\rm Ker}}

200: \def\Lie{{\rm Lie}}

201: \def\PGL{{\rm PGL}}

202: \def\PSL{{\rm PSL}}

203: \def\rank{{\rm rank}}

204: \def\Res{{\rm Res}}

205: \def\sign{{\rm sign}}

206: \def\SL{{\rm SL}}

207: \def\Spec{{\rm Spec}}

208: \def\Sp{{\rm Spec}}

209: \def\spin{{\rm spin}}

210: \def\Spin{{\rm Spin}}

211: \def\Trace{{\rm Tr}}

212: \def\Tr{{\rm Tr}}

213: \def\tr{{\rm tr}}

214: 

215: \parindent 0in

216: 

217: \title[The noncommutative ``Pardis'']{A walk in the noncommutative garden}

218: \author[Connes]{Alain Connes}

219: \author[Marcolli]{Matilde Marcolli}

220: \address{A.~Connes: Coll\`ege de France \\

221: 3, rue d'Ulm \\ Paris, F-75005 France

222: \\ I.H.E.S. and Vanderbilt

223: University} \email{alain\@@connes.org}

224: \address{M.~Marcolli: Max--Planck Institut f\"ur Mathematik  \\

225: Vivatsgasse 7 \\

226: Bonn, D-53111 Germany} \email{marcolli\@@mpim-bonn.mpg.de}

227: 

228: \begin{document}

229: \maketitle

230: 

231: \tableofcontents

232: 

233: \newpage

234: 

235: \section{Introduction}\label{intro}

236: 

237: \begin{verse}

238: {\em If you cleave the hearth of one drop of water \\

239: a hundred pure oceans emerge from it.} \\

240: \smallskip

241: (Mahmud Shabistari, {\em Gulshan-i-raz})

242: \end{verse}

243: 

244: \bigskip

245: 

246: We have decided to contribute to the volume of the IPM lectures on

247: noncommutative geometry a text that collects a list of examples of

248: noncommutative spaces. As the quote of the Sufi poet here above suggests,

249: it is often better to approach a new subject by analyzing specific

250: examples rather than presenting the general theory. We hope that the

251: diversity of examples the readers will encounter in this text will

252: suffice to convince them of the fact that noncommutative geometry is

253: a very rich field in rapid evolution, full of interesting and yet

254: unexplored landscapes. Many of the examples collected here have not

255: yet been fully explored from the point of view of the general

256: guidelines we propose in Section \ref{road} and the main point of

257: this text is to provide a great number of open questions. The reader

258: should interpret this survey as a suggestion of possible interesting

259: problems to investigate, both in the settings described here, as

260: well as in other examples that are available but did not fit in this

261: list, and in the many more that still await to be discovered.

262: Besides the existing books on NCG such as \cite{Co90},

263: \cite{Manin91NC}, \cite{Co94}, \cite{khal},

264: \cite{landi} \cite{FGV} \cite{Mar}, two new

265: books are being written: one by the two authors of this paper \cite{CoMarBook},

266: and one by Connes and Moscovici \cite{CoMoBook}.

267: 

268: 

269: \section{Handling noncommutative spaces in the wild:

270: basic tools}\label{road}

271: 

272: We are going to see in many example how one obtains the algebra

273: of coordinates $\cA$ of a noncommutative space $X$.

274: Here we think of $\cA$ as being the algebra

275: of ``smooth functions'', which will usually be

276: a dense subalgebra of a $C^*$-algebra $\bar\cA$.

277: 

278: \smallskip

279: 

280: Here are some basic steps that one can perform in order to

281: acquire a good understanding of a given noncommutative space $X$

282: with algebra of coordinates $\cA$.

283: 

284: \begin{itemize}

285: 

286: \item[1)] Resolve the diagonal of $\cA$ and compute the cyclic cohomology.

287: 

288: \item[2)] Find a geometric model of $X$ up to homotopy.

289: 

290: \item[3)] Construct the spectral geometry $(\cA,\cH,D)$.

291: 

292: \item[4)] Compute the time evolution and analyze the thermodynamics.

293: 

294: \end{itemize}

295: 

296: \medskip

297: 

298: 1) The first step means finding a resolution of the $\cA$-bimodule

299: $\cA$ by projective $\cA$-bimodules making it possible to compute

300: the Hochschild homology of $\cA$ effectively. In general, such

301: resolutions will be of Kozsul type and the typical example is the

302: resolution of the diagonal for the algebra $C^\infty(X)$ of smooth

303: functions on a compact manifold as in the $C^\infty$ version

304: \cite{ConnesCH} of the Hochschild, Kostant, Rosenberg theorem (\cf

305: \cite{HKR}). It makes it possible to know what is the analogue of

306: differential forms and of de Rham currents on the space $X$ and to

307: take the next step of computing the cyclic homology and cyclic

308: cohomology of $\cA$, which are the natural replacements for the de

309: Rham theory. For foliation algebras

310: this was done long ago (\cf \cite{Co1}, \cite{BryNi}, \cite{Crai}).

311: It ties in with the natural double complex of transverse currents.

312: 

313: It is not always easy to perform this step of finding

314: a resolution and computing Hochschild and cyclic (co)homology.

315: For instance, in the case of algebras given by generators and 

316: relations this uses the whole theory of Kozsul duality,

317: which has been successfully extended to $N$-homogeneous algebras (\cf 

318: \cite{D-V}, \cite{D-V1}, \cite{D-V2}, \cite{D-V3}, \cite{D-V4},

319: \cite{BDV}). 

320: 

321: One specific example in which it would be very interesting to

322: resolve the diagonal is the modular Hecke algebras (Section

323: \ref{hecke}). In essence, finding a resolution of the diagonal in

324: the algebra of modular forms of arbitrary level, equivariant with

325: respect to the action of the group $\GL_2(\A_f)$ of finite adeles,

326: would yield formulas for the compatibility of Hecke operators with

327: the algebra structure. This is a basic and hard problem of the

328: theory of modular forms.

329: 

330: \medskip

331: 

332: Cyclic cohomology (and homology) is a well developed theory which

333: was first designed to handle the leaf spaces of foliations as well

334: as group rings of discrete groups (\cf \cite{khal}). The theory

335: admits a purely algebraic version which is at center stage in

336: ``algebraic" noncommutative geometry, but it is crucial in the

337: analytic set-up to construct cyclic cocycles with good compatibility

338: properties with the topology of the algebra. For instance, when the

339: domain of definition of the cocycle is a dense subalgebra stable

340: under holomorphic functional calculus, it automatically gives an

341: invariant of the $K$-theory of the underlying $C^*$-algebra (\cf

342: \cite{Co1}).

343: 

344: 

345: \smallskip

346: 

347: 2) The essence of the second step is that many noncommutative spaces

348: defined as a ``bad quotients'' (\cf Section \ref{quotients})

349: can be desingularized, provided one is

350: ready to work up to homotopy. Thus for instance if the space $X$ is

351: defined as the quotient

352: $$

353: X=\,Y/\,\sim

354: $$

355: of an ``ordinary" space $Y$ by an equivalence relation $\sim$ one

356: can often find a description of the same space $X$ as a quotient

357: $$

358: X=\,Z/\,\sim

359: $$

360: where the equivalence classes are now {\em contractible} spaces. The

361: homotopy type of $Z$ is then uniquely determined and serves as a

362: substitute for that of $X$ (see \cite{BauCo}).

363: 

364: For instance, if the equivalence relation on $Y$ comes from the free

365: action of a torsion free discrete group $\Gamma$, the space $Z$ is

366: simply a product over $\Gamma$ of the form

367: $$

368: Z=\,Y\times_\Gamma\,E\Gamma ,

369: $$

370: where $E\Gamma$ is a contractible space on which $\Gamma$ acts

371: freely and properly.

372: 

373: The main point of this second step is that it gives a starting point

374: for computing the $K$-theory of the space $X$ \ie of the

375: $C^*$-algebra $A=\bar\cA$ playing the role of the algebra of continuous

376: functions on $X$. Indeed, for each element of the $K$-homology

377: of the classifying space $Z$, there is a general construction

378: of an index problem for ``families

379: parameterized by $X$" that yields an assembly map (\cf \cite{BauCo})

380: \begin{equation}\label{mu-map}

381: \mu\;:\; K_*(Z)\to K_*(A)

382: \end{equation}

383: This Baum--Connes map is an isomorphism in a lot of cases (with

384: suitable care of torsion, \cf \cite{BCH}) including all

385: connected locally compact groups, all amenable groupoids

386: and all hyperbolic discrete groups.

387: It thus gives a computable guess for the $K$-theory of $X$.

388: 

389: The next

390: step is not only to really compute $K(A)$ but also to get a good

391: model for the ``vector bundles" on $X$ \ie the finite projective

392: modules over $A$. This step should then be combined with the above

393: first step to compute the Chern character using connections,

394: curvature, and eventually computing moduli spaces of Yang-Mills

395: connections as was done for instance for the NC-torus in

396: \cite{CoRi}.

397: 

398: \smallskip

399: 

400: 3) The third step makes it possible to pass from the soft part of

401: differential

402: geometry to the harder ``Riemannian" metric aspect. The sought for

403: spectral geometry $(\cA,\cH,D)$ has three essential features:

404: 

405: \begin{itemize}

406: 

407: \item The $K$-homology class of $(\cA,\cH,D)$.

408: 

409: \item The smooth structure.

410: 

411: \item The metric.

412: 

413: \end{itemize}

414: 

415: One should always look for a spectral triple whose $K$-homology

416: class is as non-trivial as possible. Ideally it should extend to a

417: class for the double algebra $\cA\otimes \cA^o$ and then be a generator

418: for Poincar\'e duality. In general this is too much to ask for, since

419: many interesting spaces do not fulfill Poincar\'e duality. The main

420: tool for determining the stable homotopy class of the spectral

421: triple is Kasparov's bivariant $KK$-theory. Thus it is quite

422: important to already have taken step 2 and to look for classes whose

423: pairing with $K$-theory is as non-trivial as can be. For the smooth

424: structure, there is often a natural guess for a subalgebra

425: $A^\infty\subset A$ of the $C^*$-algebra $A=\bar\cA$ that will play the role

426: of the algebra of smooth functions. It should in general contain the

427: original algebra $\cA$ but should have the further property of being

428: stable under the holomorphic functional calculus. This ensures that

429: the inclusion $A^\infty\subset A$ is an isomorphism in $K$-theory

430: and makes it possible to complete the classification of smooth vector bundles.

431: 

432: The role of the unbounded operator $D$ for the smooth structure is

433: that it defines the geodesic flow by the formula

434: $$

435: F_t(a)=\,e^{it|D|}\,a\,e^{-it|D|}\qqq a\in A^\infty

436: $$

437: and one expects that smoothness is governed by the smoothness of the

438: operator valued map $\R\ni t \mapsto F_t(a)$. The main result of the

439: general theory is the local index formula of \cite{CoMo}, which

440: provides the analogue of the Pontrjagin classes of smooth manifolds

441: in the noncommutative framework.

442: 

443: The problem of determining $D$ from the knowledge of the

444: $K$-homology class is very similar to the choice of a connection on

445: a bundle. There are general results that assert the existence of an

446: unbounded selfadjoint $D$ with bounded commutators with $\cA$ from

447: estimates on the commutators with the phase $F$. The strongest is obtained

448: (\cf \cite{Co94}) just assuming that the $[F,a]$ are in an ideal

449: called ${\rm Li}\,\cH$ and it ensures the existence of a theta-summable

450: spectral triple which is what one needs to get started.

451: 

452: It is not always possible to find a finitely-summable spectral

453: triple, first because of growth conditions on the algebra

454: \cite{Co-fr}, but also since the finitely-summable condition is very

455: analogous to  type II in the theory of factors. In very general

456: cases, like the noncommutative space coming from foliations, one can

457: however go from type III to type II by passing to the total space of

458: the space of transverse metrics and then use the theory of

459: hypoelliptic operators \cite{CoMoHopf}.

460: 

461: 

462: \medskip

463: 

464: 

465: 

466: Another way to attack the problem of determining $D$ is to consider

467: the larger algebra generated by $\cA$ and $D$, write a-priori

468: relations between $\cA$ and $D$ and then look for irreducible

469: representations that fall in the correct stable homotopy class.

470: Ideally one should minimize the spectral action functional

471: \cite{C-C2} in this homotopy class thus coming close to gravity. In

472: practise one should use anything available and the example of the

473: NC-space given by the quantum group $SU_q(2)$ shows that things can

474: be quite subtle \cite{SDLSV2}.

475: 

476: Once the spectral triple $(\cA,\cH,D)$ has been determined, the basic

477: steps are the following, one should compute

478: 

479: \begin{itemize}

480: 

481: \item The dimension spectrum $\Sigma\subset \C$.

482: 

483: \item The local index formula.

484: 

485: \item The inner fluctuations, scalar curvature, and spectral action.

486: 

487: \end{itemize}

488: 

489: 

490: \bigskip

491: 

492: 4) Often a noncommutative space comes with a measure class, which in

493: turn determines a time evolution $\sigma_t$, namely a 1-parameter

494: family of automorphisms of the $C^*$-algebra $A=\bar\cA$. In the type II

495: situation one can apply the discussion of step 3 above and, in the

496: finite dimensional case, use the operator $D$ to represent

497: functionals in the measure class in the form

498: $$

499: \varphi(a)=\, \int\!\!\!\!\!\!-\ \,a \,|D|^{-p} \qqq \,a\in \cA

500: $$

501: where $\int\!\!\!\!\!\!-\  $ is the noncommutative integral \ie the

502: Dixmier trace and $p$ is the ``dimension".

503: 

504: In the general case one should expect to be in the type III

505: situation in which  the time evolution $\sigma_t$ is highly

506: non-trivial. We shall see some examples, for instance in Section

507: \ref{Qlatt}. Given the data $(\bar\cA,\sigma_t)$ it is natural to regard

508: it as a quantum statistical mechanical system, with $\bar\cA$ as algebra

509: of observables and $\sigma_t$ as time evolution. One can then look

510: for equilibrium states for the system, for a given value $\beta$ of

511: the thermodynamic parameter (inverse temperature).

512: 

513: If the algebra $\bar\cA$ is concretely realized as an algebra of bounded

514: operators on a Hilbert space $\cH$, then one can consider the

515: Hamiltonian $H$, namely the (unbounded) operator on $\cH$ that is the

516: infinitesimal generator of the time evolution. If the operator

517: $\exp(-\beta H)$ is trace class, then one has equilibrium states for

518: the system $(\bar\cA,\sigma_t)$ written in the usual Gibbs form

519: $$

520: \varphi_\beta(a)=\frac{\Tr(a\exp(-\beta H))}{\Tr(\exp(-\beta H))},

521: $$

522: where $Z(\beta)=\Tr(\exp(-\beta H))$ is the partition function of the

523: system. The notion of equilibrium state continues to make sense

524: when $\exp(-\beta H)$ is not necessarily trace class, and is given by

525: the more subtle notion of KMS (Kubo--Martin--Schwinger) states.

526: 

527: These are states on $\bar\cA$, namely continuous functionals

528: $\varphi: \bar\cA\to \C$ with $\varphi(1)=1$ and $\varphi(a^*a)\geq

529: 0$, satisfying the KMS$_\beta$ condition that, for all $a,b\in

530: \bar\cA$ there exists a function $F_{a,b}(z)$ which is holomorphic

531: on the strip $0< \Re(z) <i\beta$ continuous and bounded on the

532: closed strip and such that, for all $t\in\R$,

533: \begin{equation}\label{KMScond}

534:   F_{a,b}(t)=\varphi(a\sigma_t(b)) \ \ \text { and } \ \

535: F_{a,b}(t+i\beta)=\varphi(\sigma_t(b)a).

536: \end{equation}

537: KMS states at zero temperature can be defined as weak limits as

538: $\beta\to \infty$ of KMS$_\beta$ states.

539: 

540: 

541: \begin{figure}

542: \begin{center}

543: \includegraphics[scale=0.9]{kmss.eps}

544: \end{center}

545: \caption{The KMS condition \label{FigKMS}}

546: \end{figure}

547: 

548: 

549: One can construct using KMS states very refined invariants of

550: noncommutative spaces. For a fixed $\beta$, the KMS$_\beta$ states

551: form a simplex, hence one can consider only the extremal KMS$_\beta$

552: states $\cE_\beta$, from which one recovers all the others by convex

553: combinations. An extremal KMS$_\beta$ state is always factorial and

554: the type of the factor is an invariant of the state. The simplest

555: situation is the type I. One can show under minimal hypothesis

556: (\cite{CCM}) that extremal KMS$_\beta$ states continue to survive

557: when one lowers the temperature \ie one increases $\beta$. Thus, in

558: essence, when cooling down the system this tends to become more and

559: more ``classical" and in the $0$-temperature limit $\cE_\beta$ gives

560: a good replacement of the notion of classical points for a

561: noncommutative space. We shall see in Section \ref{shimura} how, in

562: examples related to arithmetic, the ``classical points'' described

563: by the zero temperature KMS states of certain quantum statistical

564: mechanical systems recover classical arithmetic varieties. The

565: extremal KMS states at zero temperature, evaluated on suitable

566: arithmetic elements in the noncommutative algebra, can be shown in

567: significant cases to have an interesting Galois action, related to

568: interesting questions in number theory (\cf \cite{BC}, \cite{CMR},

569: \cite{CoMa}).

570: 

571: In joint work with Consani, we showed in \cite{CCM}

572: how to define an analog in characteristic

573: zero of the action of the Frobenius on the etale cohomology by a

574: process involving the above thermodynamics. One key feature is that

575: the analogue of the Frobenius is the ``dual" of the above time

576: evolution $\sigma_t$. The process involves cyclic homology and its

577: three basic steps are (\cite{CCM})

578: 

579: \begin{itemize}

580: 

581: 

582: \item Cooling.

583: 

584: \item Distillation.

585: 

586: \item Dual action of  $\R_+^*$ on the cyclic homology of the

587: distilled space.

588: 

589: \end{itemize}

590: 

591: 

592: \medskip

593: When applied to the simplest system

594: (the Bost--Connes system of \cite{BC})

595: this yields a cohomological interpretation of the spectral

596: realization of the zeros of the Riemann zeta function

597: (\cite{CoRH}, \cite{CCM}).

598: 

599: 

600: 

601: 

602: \bigskip

603: 

604: 

605: \section{Phase spaces of microscopic systems}\label{heisenberg}

606: 

607: What can be regarded historically as the first example of a

608: non--commutative space is the Heisenberg formulation of the

609: observational Ritz-Rydberg law of spectrocopy. In fact, quantum

610: mechanics showed that indeed the parameter space, or phase space

611: of the mechanical system given by a single atom fails to be a

612: manifold. It is important to convince oneself of this fact and to

613: understand that this conclusion is indeed dictated by the {\it

614: experimental findings of spectroscopy}.

615: 

616: At the beginning of the twentieth century a wealth of experimental

617: data was being collected on the spectra of various chemical

618: elements. These spectra obey experimentally discovered laws, the

619: most notable being the Ritz-Rydberg combination principle. The

620: principle can be stated as follows; spectral lines are indexed by

621: pairs of labels. The statement of the principle then is that

622: certain pairs of spectral lines, when expressed in terms of

623: frequencies, do add up to give another line in the spectrum.

624: Moreover, this happens  precisely when the labels are of the form

625: $i,j$ and $j,k$.

626: 

627: \begin{center}

628: \begin{figure}

629: \includegraphics[scale=0.9]{Ritz2.eps}

630: \caption{Spectral lines and the Ritz-Rydberg law.

631: \label{FigRitz}}

632: \end{figure}

633: \end{center}

634: 

635: In the seminal paper \cite{Heisenberg} of 1925, Heisenberg

636: considers the classical prediction for the radiation emitted by a

637: moving electron in a field, where the observable dipole moment can

638: be computed, with the motion of the electron given in Fourier

639: expansion. The classical model would predict (in his notation)

640: frequencies distributed according to the law

641: \begin{equation}\label{emissionCM}

642:  \nu(n,\alpha) = \alpha \nu(n) = \alpha \frac{1}{h} \frac{dW}{dn}.

643: \end{equation}

644: When comparing the frequencies obtained in this classical model

645: with the data, Heisenberg noticed that the classical law

646: \eqref{emissionCM} did not match the phenomenon observed.

647: 

648: The spectral rays provide a `picture' of an atom: if atoms were

649: classical systems, then the picture formed by the spectral lines

650: would be (in our modern mathematical language) a {\it group},

651: which is what \eqref{emissionCM} predicts. That is, the classical

652: model predicts that the observed frequencies should simply add,

653: obeying a group law, or, in Heisenberg's notation, that

654: \begin{equation}\label{emission2}

655:  \nu(n,\alpha)+ \nu(n,\beta)= \nu(n,\alpha+\beta).

656: \end{equation}

657: Correspondingly, observables would form the convolution algebra of

658: a group.

659: 

660: What the spectral lines were instead providing was the picture of

661: a {\it groupoid}. Heisenberg realized that the classical law of

662: (\ref{emissionCM}) \eqref{emission2} would have to be replaced by

663: the quantum--mechanical

664: \begin{equation}\label{emissionQM}

665:  \nu(n,n-\alpha) = \frac{1}{h} \left( W(n)- W(n-\alpha) \right).

666: \end{equation}

667: This replaces the group law with that of a groupoid, replacing the

668: classical \eqref{emission2} by the quantum--mechanical

669: \begin{equation}\label{emissionQ}

670:  \nu(n,n-\alpha)+ \nu(n-\alpha, n-\alpha-\beta)=

671: \nu(n,n-\alpha-\beta).

672: \end{equation}

673: Similarly, the classical Fourier modes $\fU_{\alpha}(n)

674: e^{i\omega(n)\alpha t}$ were replaced by $ \fU

675: (n,n-\alpha)e^{i\omega(n,n-\alpha)t}$.

676: 

677: The analysis of the emission spectrum given by Heisenberg was in

678: very good agreement with the Ritz--Rydberg law, or combination

679: principle, for spectral lines in emission or absorption spectra.

680: 

681: In the same paper, Heisenberg also extends his redefinition of the

682: multiplication law for the Fourier coefficients to coordinates and

683: momenta, by introducing transition amplitudes that satisfy similar

684: product rules. This is, in Born's words, ``his most audacious

685: step'': in fact, it is precisely this step that brings

686: non--commutative geometry on the scene.

687: 

688: It was Born who realized that what Heisenberg described in his

689: paper corresponded to replacing classical coordinates with

690: coordinates which no longer commute, but which obey the laws of

691: matrix multiplication. In his own words reported in

692: \cite{vdWaerden},

693: 

694: \begin{quote}

695: After having sent Heisenberg's paper to the Zeitschrift f\"ur

696: Physik for publication, I began to ponder about his symbolic

697: multiplication, and was soon so involved in it that I thought the

698: whole day and could hardly sleep at night. For I felt there was

699: something fundamental behind it ... And one morning ... I suddenly

700: saw light: Heisenberg's symbolic multiplication was nothing but

701: the matrix calculus.

702: \end{quote}

703: 

704: \bigskip

705: 

706: Thus, spectral lines are parameterized by two indices

707: $L_{\alpha\beta}$ satisfying a cocycle relation

708: $L_{\alpha\beta}+L_{\beta\gamma}= L_{\alpha\gamma}$, and a

709: coboundary relation expresses each spectral line as a difference

710: $L_{\alpha\beta}=\nu_\alpha -\nu_\beta$. In other words, the

711: Ritz--Rydberg law gives the groupoid law \eqref{emissionQ}, or

712: equivalently,

713: $$ (i,j)\bullet (j,k)= (i,k) $$

714: and the convolution algebra of the group is replaced by

715: observables satisfying the matrix product

716: $$ (AB)_{ik} = \sum_j A_{ij} B_{jk}, $$

717: for which in general commutativity is lost:

718: $$ AB\neq BA. $$

719: The Hamiltonian $H$ is a matrix with the frequencies on the

720: diagonal, and observables obey the evolution equation

721: $$ \frac{d}{dt} A = i [H,A]. $$

722: 

723: Out of Heisenberg's paper and Born's interpretation of the same in

724: terms of matrix calculus, emerged the statement of Heisenberg's

725: uncertainty principle in the form of a commutation relation of

726: matrices

727: $$ [P,Q] = \frac{h}{2\pi i} I. $$

728: The matrix calculus and the uncertainty principle were formulated

729: in the subsequent paper of Born and Jordan \cite{Born}, also

730: published in 1925. This viewpoint on quantum mechanics was later

731: somewhat obscured by the advent of the Schr\"odinger equation.

732: The Schr\"odinger approach shifted the emphasis back to the more

733: traditional technique

734: of solving partial differential equations, while the more modern

735: viewpoint of Heisenberg implied a much more serious change

736: of paradigm, affecting our most basic understanding of the notion of

737: space. Heisenberg's approach can be regarded as the historic origin of

738: noncommutative geometry.

739: 

740: \bigskip

741: 

742: 

743: \section{Noncommutative quotients}\label{quotients}

744: 

745: A large source of examples of noncommutative spaces is given by

746: quotients of equivalence relations. One starts by an ordinary

747: commutative space $X$ (\eg a smooth manifold or more generally a

748: locally compact Hausdorff topological space). This can be described

749: via its algebra of functions $C(X)$, and abelian $C^*$-algebra.

750: Suppose then that we are interested in taking a quotient $Y=X/\sim$

751: of $X$ with respect to an equivalence relation. In general, one

752: should not expect the quotient to be nice. Even when $X$ is a smooth

753: manifold, the quotient $Y$ need not even be a Hausdorff space. In

754: general, one would like to still be able to characterize $Y$ through

755: its ring of functions.  One usually defines $C(Y)$ to be functions

756: on $X$ that are invariant under the equivalence relation ,

757: \begin{equation}\label{commCY}

758: C(Y)= \{ f\in C(X): f(a)=f(b), \forall a\sim b \} .

759: \end{equation}

760: Clearly, for a ``bad'' equivalence relation one typically

761: gets this way only constant functions $C(Y)=\C$.

762: 

763: There is a better way to

764: associate to the quotient space $Y$ a ring of functions which is

765: nontrivial for any equivalence relation. This requires dropping the

766: commutativity requirement. One can then consider functions of two

767: variables $f_{ab}$ defined on the graph of the equivalence relation,

768: with a product which is no longer the commutative pointwise product,

769: but the noncommutative convolution product dictated by the groupoid of

770: the equivalence relation.

771: In general the elements in the algebra of functions

772: \begin{equation}\label{ncCY}

773: ``C(Y)\text{''}= \{ (f_{ab}) : a\sim b \}

774: \end{equation}

775: act as bounded operators on the Hilbert space $L^2$ of the

776: equivalence class. This also guarantees the convergence in the

777: operator norm of the convolution product

778: $$ \sum_{a\sim b\sim c} f_{ab} g_{bc}. $$

779: 

780: We give a few examples to illustrate the difference between the

781: traditional construction and the one of noncommutative geometry.

782: 

783: \begin{ex} {\em

784: Consider the space $Y=\{ x_0, x_1\}$ with the equivalence relation

785: $x_0\sim x_1$. With the first point of view the algebra of

786: functions on the quotient is $\C$, and in the second point of view

787: it is $\cB =M_2(\C)$, that is

788: \begin{equation}

789: \cB = \left\{ f = \left( \begin{array}{cc} f_{aa} & f_{ab} \\

790: f_{ba} & f_{bb} \end{array} \right) \right\} \, . \label{twopoint}

791: \end{equation}

792: These two algebras are not the same, though in this case they are

793: Morita equivalent. }\end{ex}

794: 

795: Notice that, when one computes the spectrum of the algebra

796: (\ref{twopoint}), it turns out that it is composed of only one

797: point, so the two points $a$ and $b$ have been identified.

798: This first trivial example represents the typical situation where

799: the quotient space is ``nice'': the two constructions give Morita

800: equivalent algebras. In this sense, Morita

801: equivalent algebras are regarded as ``the same'' (or better

802: isomorphic) spaces in non--commutative geometry.

803: 

804: \begin{ex} {\em

805: Consider the space $Y=[0,1] \times \{ 0,1\}$ with the equivalence

806: relation $(x,0)\sim (x,1)$ for $x\in (0,1)$. Then in the first

807: viewpoint the algebra of functions is again given just by the

808: constant functions $\C$, but in the second case we obtain

809: \begin{equation}  \{ f\in C([0,1])\otimes M_2(\C) : \, f(0) \text{ and

810: } f(1) \text{ diagonal } \}. \label{alg1} \end{equation} }

811: \label{ex2} \end{ex}

812: 

813: In this case, these algebras are not Morita equivalent. This can be

814: seen by computing their $K$--theory. This means that the approach of

815: non--commutative spaces produces something genuinely new, as soon as

816: the quotient space ceases to be ``nice'' in the classical sense.

817: 

818: In general, the first kind of construction of functions on the

819: quotient space is cohomological in nature: one seeks for functions

820: satisfying certain equations or constraints. Usually there are

821: very few solutions. The second approach, instead, typically

822: produces a very large class of functions.

823: 

824: \bigskip

825: 

826: \section{Spaces of leaves of foliations}\label{foliations}

827: 

828: There is a very rich collection of examples of noncommutative spaces

829: given by  the leaf spaces of foliations. The connection thus

830: obtained between noncommutative geometry and the geometric theory of

831: foliations is very far reaching for instance through the role of

832: Gelfand-Fuchs cohomology, of the Godbillon-Vey invariant and of the

833: passage from type III to type II using the transverse frame bundle.

834: It is this class of examples  that triggered the initial development

835: of cyclic cohomology (\cf \cite{khal} section 4), of the local index

836: formula in noncommutative geometry as well as the theory of

837: characteristic classes for Hopf algebra actions.

838: 

839: 

840: \medskip

841: 

842: The construction of the algebra associated to a foliation is a

843: special case of the construction of section \ref{quotients} but both

844: the presence of holonomy and the case when the graph of the

845: foliation is non-hausdorff require special care, so we shall recall

846: the basic steps below.

847: 

848: Let $V$ be a smooth manifold and $TV$ its tangent bundle, so that

849: for each $x \in V$, $T_x V$ is the tangent space of $V$ at $x$. A

850: smooth subbundle $F$ of $TV$ is called {\it integrable} iff one of

851: the following equivalent conditions is satisfied:

852: 

853: \smallskip

854: 

855: \begin{enumerate}

856: 

857: \item[a)] Every $x \in V$ is contained in a submanifold $W$ of $V$ such that

858: $$

859: T_y (W) = F_y \qquad \forall \, y \in W \, .

860: $$

861: 

862: \smallskip

863: 

864: \item[b)] Every $x \in V$ is in the domain $U \subset V$ of a

865: submersion $p : U \to {\mathbb R}^q$ ($q = {\rm codim} \, F$) with

866: $$

867: F_y = {\rm Ker} (p_*)_y \qquad \forall \, y \in U \, .

868: $$

869: 

870: \smallskip

871: \item[c)] $C^{\infty} (F) = \{ X \in C^{\infty} (TV) \, , \ X_x \in

872: F_x \quad \forall \, x \in V \}$ is a Lie algebra.

873: 

874: \smallskip

875: 

876: \item[d)] The ideal $J(F)$ of smooth exterior differential forms which

877: vanish on $F$ is stable by exterior differentiation.

878: \end{enumerate}

879: 

880: \medskip

881: 

882: Any $1$-dimensional subbundle $F$ of $TV$ is integrable, but for

883: $\dim F \geqq 2$ the condition is non trivial, for instance if $P

884: \overset{p}{\to} B$ is a principal $H$-bundle (with compact

885: structure group $H$) the bundle of horizontal vectors for a given

886: connection is integrable iff this connection is flat.

887: 

888: \smallskip

889: 

890: A foliation of $V$ is given by an integrable subbundle $F$ of $TV$.

891: The leaves of the foliation $(V,F)$ are the maximal connected

892: submanifolds $L$ of $V$ with $T_x (L) = F_x $, $\forall \, x \in L$,

893: and the partition of $V$ in leaves $$V = \cup

894: L_{\alpha}\,,\quad\alpha \in X$$ is characterized geometrically by

895: its ``local triviality'': every point $x \in V$ has a neighborhood

896: $U$ and a system of local coordinates

897: $(x^j)_{j = 1 , \ldots , \dim V}$ called

898: {\it foliation charts}, so

899: that the partition of $U$ in connected components of

900: leaves corresponds to the partition of $${\mathbb

901: R}^{\dim V} = {\mathbb R}^{\dim F} \times {\mathbb R}^{\text{codim}

902: \, F}$$ in the parallel affine subspaces ${\mathbb R}^{\dim F}

903: \times {\rm pt}$. These  are the leaves of the

904: restriction of $F$ and are called {\it plaques}.

905: 

906: The set $X=V/F$ of leaves of a foliation $(V,F)$

907: is in most cases a noncommutative space. In other words even though

908: as a set it has the cardinality of the continuum it is in general

909: not so at the effective level and it is in general impossible to

910: construct a countable set of measurable functions on $V$ that form a

911: complete set of invariants for the equivalence relation coming from

912: the partition of $V$ in leaves $V = \cup L_{\alpha}$. Even in the

913: simple cases in which the set $X=V/F$ of leaves is classical it

914: helps to introduce the associated algebraic tools in order to get a

915: feeling for their role in the general singular case.

916: 

917: \smallskip

918: 

919: To each foliation $(V,F)$ is canonically associated a $C^*$ algebra

920: $C^* (V,F)$ which encodes the topology of the space of leaves. The

921: construction is basically the same as the general one for quotient

922: spaces of Section \ref{quotients}, but there are interesting nuances

923: coming from the presence of holonomy in the foliation context. To

924: take this into account one first constructs a manifold $G$, $\dim

925: \,G = \dim \,V + \dim \,F$, called the graph (or holonomy groupoid)

926: of the foliation, which refines the equivalence relation coming from

927: the partition of $V$ in leaves $V = \cup L_{\alpha}$. This

928: construction is due to Thom, Pradines and Winkelnkemper,

929: see \cite{Winkel}.

930: 

931: An element $\gamma$ of $G$ is given by two points $x = s(\gamma)$,

932: $y = r(\gamma)$ of $V$ together with an equivalence class of smooth

933: paths: $\gamma (t)\in V$, $t \in [0,1]$; $\gamma (0) = x$, $\gamma

934: (1) = y$, tangent to the bundle $F$ ({\it i.e.} with $\dot\gamma (t)

935: \in F_{\gamma (t)}$, $\forall \, t \in {\mathbb R}$) up to the

936: following equivalence: $\gamma_1$ and $\gamma_2$ are equivalent iff

937: the {\it holonomy} of the path $\gamma_2 \circ \gamma_1^{-1}$ at the

938: point $x$ is the {\it identity}. The graph $G$ has an obvious

939: composition law. For $\gamma , \gamma' \in G$, the composition

940: $\gamma \circ \gamma'$ makes sense if $s(\gamma) = r(\gamma')$. If

941: the leaf $L$ which contains both $x$ and $y$ has no holonomy, then

942: the class in $G$ of the path $\gamma (t)$ only depends on the pair

943: $(y,x)$. The condition of trivial holonomy

944: is generic in the topological sense of dense $G_{\delta}$'s.

945: In general, if one fixes $x = s(\gamma)$, the map from $G_x

946: = \{ \gamma , s(\gamma) = x \}$ to the leaf $L$ through $x$, given

947: by $\gamma \in G_x \mapsto y = r(\gamma)$, is the holonomy covering

948: of $L$.

949: 

950: Both maps $r$ and $s$ from the manifold $G$ to $V$ are smooth

951: submersions and the map $(r,s)$ to $V \times V$ is an immersion

952: whose image in $V \times V$ is the (often singular) subset

953: \begin{equation}\label{subset}

954: \{ (y,x)\in V \times V: \, \text{ $y$ and $x$ are on the same leaf}\}.

955: \end{equation}

956: In first

957: approximation one can think of elements of $C^* (V,F)$ as continuous

958: matrices $k(x,y)$, where $(x,y)$ varies in the set \eqref{subset}.

959: We now describe this $C^*$ algebra in all details. We

960: assume, for notational convenience, that the manifold $G$ is

961: Hausdorff. Since this fails to be the case in very interesting

962: examples, we   also explain briefly how to remove this

963: hypothesis.

964: 

965: \smallskip

966: 

967: The basic elements of $C^* (V,F)$ are smooth half densities

968: $f \in C_c^{\infty} (G , \Omega^{1/2})$ with

969: compact support on $G$. The

970: bundle $\Omega_G^{1/2}$ of half densities on $G$ is obtained as

971: follows. One first defines a line bundle $\Omega_V^{1/2}$ on V. For

972: $x\in V$ one lets $\Omega_x^{1/2}$ be the one dimensional complex

973: vector space of maps from the exterior power $\wedge^k \, F_x$, $k =

974: \dim F$, to ${\mathbb C}$ such that

975: $$

976: \rho \, (\lambda \, v) = \vert \lambda \vert^{1/2} \, \rho \, (v)

977: \qquad \forall \, v \in \wedge^k \, F_x \, , \quad \forall \,

978: \lambda \in {\mathbb R} \, .

979: $$

980: Then, for $\gamma \in G$, one can identify $\Omega_{\gamma}^{1/2}$ with the one

981: dimensional complex vector space $\Omega_y^{1/2} \otimes

982: \Omega_x^{1/2}$, where $\gamma : x \to y$. In other words

983: $$

984: \Omega_G^{1/2}=\, r^*(\Omega_V^{1/2})\otimes s^*(\Omega_V^{1/2})\,.

985: $$

986:  Of course the bundle $\Omega_V^{1/2}$ is trivial on $V$, and we

987: could choose once and for all  a trivialisation $\nu$ turning

988: elements of $C_c^{\infty} (G , \Omega^{1/2})$ into functions.

989: Let us

990: however stress that the use of half densities makes all the

991: construction completely canonical.

992: 

993: For $f,g \in C_c^{\infty} (G , \Omega^{1/2})$, the convolution

994: product $f * g$ is defined by the equality

995: $$

996: (f * g) (\gamma) = \int_{\gamma_1 \circ \gamma_2 = \gamma}

997: f(\gamma_1) \, g(\gamma_2) \, .

998: $$

999: This makes sense because, for fixed $\gamma : x \to y$ and fixing $v_x

1000: \in \wedge^k \, F_x$ and $v_y \in \wedge^k \, F_y$, the product

1001: $f(\gamma_1) \, g(\gamma_1^{-1} \gamma)$ defines a $1$-density on

1002: $G^y = \{ \gamma_1 \in G , \, r (\gamma_1) = y \}$, which is smooth

1003: with compact support (it vanishes if $\gamma_1 \notin$ support $f$),

1004: and hence can be integrated over $G^y$ to give a scalar, namely $(f * g)

1005: (\gamma)$ evaluated on $v_x , v_y$.

1006: 

1007: The $*$ operation is defined by $f^* (\gamma) =

1008: \overline{f(\gamma^{-1})}$, {\it i.e.} if $\gamma : x \to y$ and

1009: $v_x \in \wedge^k \, F_x$, $v_y \in \wedge^k \, F_y$ then $f^*

1010: (\gamma)$ evaluated on $v_x , v_y$ is equal to

1011: $\overline{f(\gamma^{-1})}$ evaluated on $v_y , v_x$. We thus get a

1012: $*$ algebra $C_c^{\infty} (G , \Omega^{1/2})$. For each leaf $L$ of

1013: $(V,F)$ one has a natural representation of this $*$ algebra on the

1014: $L^2$ space of the holonomy covering $\tilde L$ of $L$. Fixing a

1015: base point $x \in L$, one identifies $\tilde L$ with $G_x = \{

1016: \gamma , s(\gamma) = x \}$ and defines

1017: $$

1018: (\pi_x (f) \, \xi) \, (\gamma) = \int_{\gamma_1 \circ \gamma_2 =

1019: \gamma} f(\gamma_1) \, \xi (\gamma_2) \qquad \forall \, \xi \in L^2

1020: (G_x),

1021: $$

1022: where $\xi$ is a square integrable half density on $G_x$. Given

1023: $\gamma : x \to y$ one has a natural isometry of $L^2 (G_x)$ on $L^2

1024: (G_y)$ which transforms the representation $\pi_x$ in $\pi_y$.

1025: 

1026: 

1027: By definition $C^* (V,F)$ is the $C^*$ algebra completion of

1028: $C_c^{\infty} (G , \Omega^{1/2})$ with the norm $$\Vert f \Vert =

1029: \sup_{x \in V} \, \Vert \pi_x (f) \Vert\,.$$

1030: 

1031: 

1032: Note that $C^* (V,F)$ is always norm separable and admits a natural

1033: smooth subalgebra, namely the algebra $C_c^\infty (V,F)=C_c^{\infty} (G ,

1034: \Omega^{1/2})$ of smooth compactly supported half densities.

1035: 

1036: If the leaf $L$ has trivial holonomy then the representation

1037: $\pi_x$, $x \in L$, is irreducible. In general, its commutant is

1038: generated by the action of the (discrete) holonomy group $G_x^x$ in

1039: $L^2 (G_x)$. If the foliation comes from a submersion $p : V \to B$,

1040: then its graph $G$ is $\{ (x,y) \in V \times V , \, p(x) = p(y)\}$

1041: which is a submanifold of $V \times V$, and $C^* (V,F)$ is identical

1042: with the algebra of the continuous field of Hilbert spaces $L^2

1043: (p^{-1} \{ x \})_{x \in B}$. Thus (unless $\dim \,F = 0$) it is

1044: isomorphic to the tensor product of $C_0 (B)$ with the elementary

1045: $C^*$ algebra of compact operators. If the foliation comes from an

1046: action of a Lie group $H$ in such a way that the graph is identical

1047: with $V \times H$ (this is not always true even for flows)  then

1048: $C^* (V,F)$ is identical with the reduced crossed product of $C_0

1049: (V)$ by $H$. Moreover the construction of $C^* (V,F)$ is local in

1050: the following sense.

1051: 

1052: 

1053: If $V' \subset V$ is an open set and $F'$ is the restriction of $F$

1054: to $V'$, then the graph $G'$ of $(V' , F')$ is an open set in the

1055: graph $G$ of $(V,F)$, and the inclusion $C_c^{\infty} (G' ,

1056: \Omega^{1/2}) \subset C_c^{\infty} (G , \Omega^{1/2})$ extends to an

1057: isometric $*$ homomorphism of $C^* (V' , F')$ in $C^* (V,F)$. The

1058: proof is straightforward and also applies in the case of non

1059: Hausdorff graph.

1060: 

1061: 

1062: Let us now briefly explain how the construction of the $C^*$ algebra

1063: $C^* (V,F)$ has to be done in the case when the graph of the

1064: foliation is not Hausdorff. This case is rather rare, since it never

1065: occurs if the foliation is real analytic. However, it does occur in

1066: cases which are topologically interesting for foliations, such as

1067: the Reeb foliation of the 3 sphere, which are constructed by

1068: patching together foliations of manifolds with boundaries $(V_i ,

1069: F_i)$ where the boundary $\partial V_i$ is a leaf of $F_i$. In fact

1070: most of the constructions done in geometry to produce smooth

1071: foliations of given codimension on a given manifold give a non

1072: Hausdorff graph. The $C^*$-algebra $C^* (V,F)$ turns out in this

1073: case to be obtained as a fibered product of the $C^* (V_i , F_i)$.

1074: 

1075: \begin{figure}

1076: \begin{center}

1077: \includegraphics[scale=0.6]{reeb.eps}

1078: \end{center}

1079: \caption{The Reeb foliation \label{FigReeb}}

1080: \end{figure}

1081: 

1082: 

1083: In the general non-Hausdorff case the graph $G$ of $(V,F)$, being

1084: non Hausdorff  may have only very few continuous fonctions with

1085: compact support. However, being a manifold, we can give a local chart

1086: $U \overset{\chi}{\to} {\mathbb R}^{\dim G}$. Take a smooth function

1087: $\varphi \in C_c^{\infty} ({\mathbb R}^{\dim G})$, ${\rm Supp} \,

1088: \varphi \subset \chi (U)$ and consider the function on $G$ equal to

1089: $\varphi \circ \chi$ on $U$ and to $0$ outside $U$. If $G$ were

1090: Hausdorff, this would generate all of $C_c^{\infty} (G)$ by taking

1091: linear combinations, and in general we take this linear span as the

1092: {\it definition} of $C_c^{\infty} (G)$. Note that we do not get {\it

1093: continuous} functions, since there may well be a sequence $\gamma_n

1094: \in U$ with two limits, one in ${\rm Supp} \, \varphi \circ \chi$

1095: one in the complement of $U$. The above definition of $C_c^{\infty}

1096: (G)$ obviously extends to get $C_c^{\infty} (G,\Omega^{1/2})$ the

1097: space of smooth $\frac{1}{2}$ densities on $G$. One then shows that

1098: the convolution $\varphi_1 * \varphi_2$ of $\varphi_1 , \varphi_2

1099: \in C_c^{\infty} (G , \Omega^{1/2})$ is in $C_c^{\infty} (G ,

1100: \Omega^{1/2})$.

1101: 

1102: 

1103: Then we proceed exactly as in the Hausdorff case, and construct the

1104: representation $\pi_x$ of the $*$ algebra $C_c^{\infty} (G ,

1105: \Omega^{1/2})$ in the Hilbert space $L^2 (G_x)$. We note that though

1106: $G$ is not Hausdorff, each $G_x$ is {\it Hausdorff},  being the

1107: holonomy covering of the leaf through $x$.

1108: 

1109: 

1110: For each $\varphi \in C_c^{\infty} (G , \Omega^{1/2})$ and $x \in

1111: V$, $\pi_x (\varphi)$ is an ordinary smoothing operator, bounded in

1112: $L^2 (G_x)$.

1113: 

1114: Exactly as in the Hausdorff case  $C^* (V,F)$ is defined as the

1115: $C^*$ completion of $C_c^{\infty} (G , \Omega^{1/2})$ with norm

1116: $\sup_{x \in V} \, \Vert \pi_x (\varphi) \Vert$.

1117: 

1118: \medskip

1119: 

1120: The obtained functor from foliations to $C^*$-algebras makes it

1121: possible first

1122: of all to translate from basic geometric properties to algebraic

1123: ones and the simplest examples of foliations already exhibit

1124: remarkable $C^*$-algebras. For instance the horocycle foliation of

1125: the unit sphere bundle of a Riemann surface of genus $>1$ gives a

1126: simple $C^*$-algebra without idempotent. The Kronecker foliation

1127: gives rise to the noncommutative torus, which we   describe

1128: in more detail in Section \ref{nctori}.

1129: 

1130: In the type II situation \ie in the presence of a holonomy invariant

1131: transverse measure $\Lambda$ the basic result of the theory is the

1132: longitudinal index theorem which computes the $L^2$-index of

1133: differential operators $D$ on the foliated manifold $(V,F)$ which

1134: are elliptic in the longitudinal direction (\ie $D$ restrict to

1135: the leaves $L$ as elliptic operators $D_L$). One starts with a pair of

1136: smooth vector bundles $E_1$, $E_2$ on $V$ together with a

1137: differential operator $D$ on $V$ from sections of $E_1$ to sections

1138: of $E_2$ such that:

1139: \begin{enumerate}

1140: \item[1)] $D$ restricts to leaves, \ie $(D \xi)_x$

1141: only depends on the restriction of $\xi$ to a neighborhood of

1142: $x$ in the leaf of $x$ ({\it i.e.} $D$ only uses partial

1143: differentiation in the leaf direction).

1144: \item[2)] $D$ is elliptic when restricted to any leaf.

1145: \end{enumerate}

1146: 

1147: \begin{thm}\label{longmeasindex} \cite{Co78}

1148: 

1149: {\rm a)} There exists a Borel transversal $B$ (resp. $B'$) such that

1150: the bundle $(\ell^2 (L \cap B))_{L \in V/F}$ is measurably

1151: isomorphic to the bundle $({\rm Ker} \, D_L)_{L \in V/F}$ (resp. to

1152: $({\rm Ker} \, D_L^*)_{L \in V/F})$.

1153: 

1154: \smallskip

1155: 

1156: \noindent {\rm b)} The scalar $\Lambda (B) < \infty$ is independent

1157: of the choice of $B$ and noted $\dim_{\Lambda} ({\rm Ker} \, (D))$.

1158: 

1159: \smallskip

1160: 

1161: \noindent  ${\rm c)}\qquad\dim_{\Lambda} ({\rm Ker} \, (D)) -

1162: \dim_{\Lambda} ({\rm Ker} \, (D^*)) = \,\varepsilon \, \langle {\rm

1163: ch} \, \sigma_D \, {\rm Td} \, (F_{\mathbb C}) , [C] \rangle$

1164: 

1165: 

1166:  ($\varepsilon = (-1) \, \frac{k(k+1)}{2}$, $k = \dim F$,

1167:  ${\rm Td} \, (F_{\mathbb C})=$  Todd genus,

1168: $\sigma_D=$ symbol of $D$) .

1169: 

1170: 

1171: \end{thm}

1172: 

1173: Here $[C]\in H_k(V,\C)$ is the homology class of the Ruelle-Sullivan

1174: current, a closed de-Rham current of dimension $k = \dim F$ which

1175: encodes the transverse measure $\Lambda$ by integration of a

1176: $k$-dimensional differential form $\omega$ on $V$ along the plaques

1177: of foliation charts.

1178: 

1179: \bigskip

1180: 

1181: 

1182: In particular the Betti numbers $\beta_j$ of a measured foliation

1183: were defined in \cite{Co78} and give the $L^2$-dimension of the

1184: space of $L^2$-harmonic forms along the leaves, more precisely one

1185: has the following result.

1186: 

1187: {\rm a)} For each $j = 0,1,2,\ldots ,\dim F$, there exists a Borel

1188: transversal $B_j$ such that the bundle $(H^j (L,{\mathbb C}))_{L \in

1189: V/F}$ of $j$-th square integrable harmonic forms on $L$ is

1190: measurably isomorphic to $(\ell^2 (L \cap B))_{L \in V/F}$.

1191: 

1192: \smallskip

1193: 

1194: \noindent {\rm b)} The scalar $\beta_j = \Lambda (B_j)$ is finite,

1195: independent of the choice of $B_j$, of the choice of the Euclidean

1196: structure on $F$.

1197: 

1198: \smallskip

1199: 

1200: \noindent {\rm c)} One has $\Sigma \, (-1)^j \, \beta_j = \chi

1201: (F,\Lambda)$.

1202: 

1203: Here the Euler characteristic is simply given by the pairing of the

1204: Ruelle-Sullivan current with the Euler class $e(F)$ of the oriented

1205: bundle $F$ on $V$.

1206: 

1207: Extending ideas of Cheeger and Gromov \cite{ChGr} in the case of discrete groups,

1208: D.~Gaboriau has shown in a remarkable recent work

1209: (\cf \cite{Gabo1}, \cite{Gabo2}) that the Betti

1210: numbers $\beta_j(F,\Lambda)$ of a foliation with contractible leaves

1211: are invariants of the measured equivalence relation $\mathcal

1212: R=\,\{(x,y)\,|\,y \in {\rm leaf}(x)\}$.

1213: 

1214: \medskip

1215: 

1216: In the general case one cannot expect to have a holonomy invariant

1217: transverse measure and in fact the simplest foliations are of type

1218: III from the measure theoretic point of view. Obtaining an analogue

1219: in general of Theorem \ref{longmeasindex} was the basic motivation

1220: for the construction of the assembly map (the second step of section

1221: \ref{road}). Let us now briefly state the longitudinal index

1222: theorem.

1223: 

1224: 

1225: Let $D$ be as above an elliptic differential operator along the

1226: leaves of the foliation $(V,F)$. Since $D$ is elliptic it has an

1227: inverse modulo $C^* (V,F)$ hence it gives an element ${\rm Ind}_a

1228: (D)$ of $K_0 (C^* (V,F))$ which is the {\em analytic index} of $D$.

1229: The {\em topological index} is obtained as follows. Let $i$ be an

1230: auxiliary imbedding of the manifold $V$ in ${\mathbb R}^{2n}$. Let

1231: $N$ be the total space of the normal bundle to the leaves: $N_x =

1232: (i_* (F_x))^{\perp} \subset {\mathbb R}^{2n}$. Let us foliate

1233: $\tilde V = V \times \R^{2n}$ by $\tilde F$, $\tilde

1234: F_{(x,t)} = F_x \times \{ 0 \}$, so that the leaves of $(\tilde V ,

1235: \tilde F)$ are just $\tilde L = L \times \{ t \}$, where $L$ is a

1236: leaf of $(V,F)$ and $t \in {\mathbb R}^{2n}$. The map $(x,\xi) \to

1237: (x,i(x) + \xi)$ turns an open neighborhood of the $0$-section in $N$

1238: into an open transversal $T$ of the foliation $(\tilde V , \tilde

1239: F)$. For a suitable open neighborhood $\Omega$ of $T$ in $\tilde V$,

1240: the $C^*$-algebra $C^* (\Omega , \tilde F)$ of the restriction of

1241: $\tilde F$ to $\Omega$ is (Morita) equivalent to $C_0 (T)$, hence

1242: the inclusion $C^* (\Omega , \tilde F) \subset C^* (\tilde V ,

1243: \tilde F)$ yields a $K$-theory map: $K^0 (N) \to K_0 (C^* (\tilde V

1244: , \tilde F))$. Since $C^* (\tilde V , \tilde F) = C^* (V,F) \otimes

1245: C_0 (\R^{2n})$, one has, by Bott periodicity, the equality

1246: $K_0 (C^* (\tilde V , \tilde F)) = K_0 (C^* (V,F))$.

1247: 

1248: Using the Thom isomorphism, $K^0 (F^*)$ is identified with $K^0 (N)$

1249: so that one gets by the above construction the topological index:

1250: $$

1251: {\rm Ind}_t : K^0 (F^*) \to K_0 (C^* (V,F)) \, .

1252: $$

1253: The {\em longitudinal index theorem} \cite{CoSk}  is the equality

1254: \begin{equation}\label{longind}

1255: {\rm Ind}_a (D) = {\rm Ind}_t ([\sigma_D]),

1256: \end{equation}

1257: where $\sigma_D$ is the

1258: longitudinal symbol of $D$ and $[\sigma_D]$ is its class in $K^0

1259: (F^*)$.

1260: 

1261: Since the group $K_0 (C^* (V,F))$ is still fairly hard to compute

1262: one needs computable invariants of its elements and this is where

1263: cyclic cohomology enters the scene. In fact its early development

1264: was already fully completed in 1981 for that precise goal  (\cf

1265: \cite{khal}). The role of the trace on $C^* (V,F)$ associated to the

1266: transverse measure $\Lambda$ is now played by cyclic cocycles on a

1267: dense subalgebra of $C^* (V,F)$. The hard analytic problem is to

1268: show that these cocycles have enough semi-continuity properties to

1269: define invariants of $K_0 (C^* (V,F))$. This was achieved for some

1270: of them in \cite{Co1} and makes it possible to formulate corollaries whose

1271: statements are independent of the general theory, such as

1272: 

1273: \smallskip

1274: \begin{thm}\label{transahat} \cite{Co1}

1275: Let $M$ be a compact oriented manifold and assume that the $\hat

1276: A$-genus

1277:  $\hat A (M)$ is non-zero (since $M$ is not

1278: assumed to be a spin manifold $\hat A (M)$ need not be an integer).

1279: Let then $F$ be an integrable Spin sub-bundle of $TM$. There exists

1280: no metric on $F$ for which the scalar curvature (of the leaves) is

1281: strictly positive $(\geq \varepsilon > 0)$ on $M$.

1282: \end{thm}

1283: 

1284: There is a very rich interplay between the theory of foliations and

1285: their characteristic classes and operator algebras even at the

1286: purely measure theoretic level \ie the classification of factors.

1287: 

1288: In a remarkable series of papers (see \cite{Hu} for references),

1289: J.~Heitsch and S.~Hurder have analyzed the interplay between the

1290: vanishing of the Godbillon-Vey invariant of a compact foliated

1291: manifold $(V,F)$ and the type of the von Neumann algebra of the

1292: foliation. Their work culminates in the following beautiful result

1293: of S.~Hurder (\cite{Hu}). If the von Neumann algebra is {\it

1294: semi-finite}, then the Godbillon-Vey invariant {\it vanishes}. We

1295: have shown, in fact, that cyclic cohomology yields a stronger

1296: result, proving that, if ${\rm GV} \ne 0$, then the central

1297: decomposition of $M$ contains necessarily factors $M$, whose virtual

1298: modular spectrum is of finite covolume in ${\mathbb R}_+^*$.

1299: 

1300: \begin{thm}\label{th3} \cite{Co1}

1301: Let $(V,F)$ be an oriented, transversally oriented, compact,

1302: foliated manifold, $({\rm codim} \, F = 1)$. Let $M$ be the

1303: associated von Neumann algebra, and ${\rm Mod}(M)$ be its flow of

1304: weights.

1305:  Then, if the Godbillon-Vey class of

1306: $(V,F)$ is different from $0$, there exists an invariant probability

1307: measure for the flow ${\rm Mod}(M)$.

1308: \end{thm}

1309: 

1310:  One actually constructs an invariant measure for the flow ${\rm

1311: Mod}(M)$, exploiting the following remarkable property of the

1312: natural cyclic 1-cocycle $\tau$ on the algebra $\cA$ of the

1313: transverse 1-jet bundle for the foliation. When viewed as a linear

1314: map $\delta$ from $\cA$ to its dual, $\delta$ is an unbounded

1315: derivation, which is {\em closable}, and whose domain extends to the

1316: center $Z$ of the von-Neumann algebra generated by $\cA$. Moreover,

1317: $\delta$ vanishes on this center, whose elements $h\in Z$ can then

1318: be used to obtain new cyclic cocycles $\tau_h$ on $\cA$. The pairing

1319: $$

1320: L(h)=\,\langle \tau_h,\, \mu(x)\rangle

1321: $$

1322: with the K-theory classes $\mu(x)$ obtained from the assembly map

1323: $\mu$, which we had constructed with P. Baum \cite{BauCo},

1324: then gives a measure on $Z$, whose invariance under the

1325: flow of weights follows from the discreteness of the K-group. To

1326: show that it is non-zero, one uses an index formula that evaluates

1327: the cyclic cocycles, associated as above to the Gelfand-Fuchs

1328: classes, on the range of  the assembly map $\mu$.

1329: 

1330: \medskip

1331: 

1332: The central question in the analysis of the noncommutative leaf

1333: space of a foliation is step 3) (of section \ref{road}), namely the

1334: metric aspect which entails in particular constructing a spectral

1335: triple describing the transverse geometry. The reason why the

1336: problem is really difficult is that it essentially amounts to doing

1337: ``metric" geometry on manifolds in a way which is ``background

1338: independent" to use the terminology of physicists \ie which is {\em

1339: invariant} under diffeomorphisms rather than covariant as in

1340: traditional Riemannian geometry. Indeed the transverse space of a

1341: foliation is a manifold endowed with the action of a large pseudo

1342: group of partial diffeomorphisms implementing the holonomy. Thus in

1343: particular no invariant metric exists in the general case and the

1344: situation is very similar to trying to develop gravity without

1345: making use of any particular ``background" metric that automatically

1346: destroys the invariance under the action of diffeomeorphisms (\cf

1347: \cite{CoMo-back}). Using the theory of hypoelliptic differential

1348: operators and the basic technique of reduction from type III to type

1349: II, a general construction of a spectral triple was done by

1350: Connes-Moscovici in \cite{CoMo}. The remaining problem of the

1351: computation of the local index formula in cyclic cohomology was

1352: solved in \cite{CoMoHopf} and led in particular to the discovery of

1353: new symmetries given by an action of a Hopf algebra which only

1354: depends upon the transverse dimension of the foliation.

1355: 

1356: This also led to the development of the noncommutative analogue of

1357: the Chern-Weil theory of characteristic classes \cite{CoMoHopf2} in

1358: the general context of Hopf algebra actions on noncommutative spaces

1359: and cyclic cohomology, a subject which  is undergoing rapid progress,

1360: in particular thanks to the recent works of M. Khalkhali and

1361: collaborators  \cite{kr1}, \cite{kr2}, \cite{kr3}, \cite{Hajac}.

1362: 

1363: \bigskip

1364: 

1365: \section{The noncommutative tori}\label{nctori}

1366: 

1367: This is perhaps considered as the prototype example of a

1368: noncommutative space, since it illustrates very clearly

1369: the properties and structures of noncommutative geometries.

1370: Noncommutative tori played a key role in

1371: the early developments of the theory in the 1980's (\cf

1372: \cite{C3}), giving rise to noncommutative analogues of

1373: vector bundles, connections, curvature, etc.

1374: 

1375: \smallskip

1376: 

1377: One can regard noncommutative tori as a special case of

1378: noncommutative spaces arising from foliations. In this case, one

1379: considers certain vector fields on the ordinary 2-dimensional

1380: real torus $T^2=\R^2/\Z^2$. In fact, one considers on $T^2$ the

1381: Kronecker foliation $dx =\theta dy$, where $\theta$ is a given real

1382: number. We are especially interested in the case where $\theta$ is

1383: irrational. That is, we consider the

1384: space of solutions of the differential

1385: equation,

1386: \begin{equation}

1387: dx = \theta dy, \ \ \forall x,y \in \R/\Z \label{Krfol}

1388: \end{equation}

1389: for $\theta \in ]0,1[$ is a fixed irrational number. In other words, we

1390: are considering the space of

1391: leaves of the Kronecker foliation on the torus (\cf Figure \ref{FigNCtori}).

1392: 

1393: \begin{figure}

1394: \begin{center}

1395: \includegraphics[scale=0.8]{NCtorus1.eps}

1396: \end{center}

1397: \caption{The Kronecker foliation and the noncommutative

1398: torus \label{FigNCtori}}

1399: \end{figure}

1400: 

1401: We can choose a transversal $T$ to the foliation, given by

1402: $$ T=\{ y=0 \}, \ \ \ \  T\cong S^1\cong\R/\Z. $$

1403: Two points of the transversal which differ by an integer multiple

1404: of $\theta$ give rise to the same leaf.

1405: We want to describe the further quotient

1406: \begin{equation}

1407: S^1 / \theta \Z \label{quottheta}

1408: \end{equation}

1409: by the equivalence relation which identifies any two points on the

1410: orbits of the irrational rotation

1411: \begin{equation}

1412: R_{\theta} x = x + \theta  \mod 1 \, .

1413: \label{Rtheta}

1414: \end{equation}

1415: 

1416: \smallskip

1417: 

1418: We can regard the circle $S^1$ and the quotient space

1419: \eqref{quottheta} at various levels of

1420: regularity (smooth, topological, measurable). This corresponds to

1421: different algebras of functions on the circle,

1422: \begin{equation}

1423: C^{\infty} (S^1) \subset C(S^1) \subset L^{\infty} (S^1) \, .

1424: \label{S1}

1425: \end{equation}

1426: 

1427: When passing to the quotient \eqref{quottheta}, if we just consider

1428: invariant functions we obtain a very poor algebra of functions, since,

1429: even at the measurable level, we would only have constant functions.

1430: If instead we consider the noncommutative algebra of functions

1431: obtained by the general recipe of ``noncommutative quotients''

1432: (functions on the graph of the equivalence relation with the

1433: convolution product), we obtain a very interesting and highly

1434: non--trivial algebra of functions describing the space of leaves

1435: of the foliation. This is given (in the topological category) by the

1436: ``irrational rotation algebra'', \ie the $C^*$-algebra

1437: \begin{equation} \label{algNCT}

1438: \cA_\theta: = \{ \left( a_{ij} \right)

1439: \, \, \, \, i,j\in T \, \, \text{ in the same leaf } \}.

1440: \end{equation}

1441: Namely, elements in the algebra $\cA_\theta$ associated to the

1442: transversal $T\simeq S^1$ are just matrices $(a_{ij})$ where the

1443: indices are arbitrary pairs of elements $i,j$ of $S^1$ belonging

1444: to the same leaf.

1445: 

1446: The algebraic rules are the same as for ordinary matrices. In the

1447: above situation, since the equivalence is given by a group action,

1448: the construction coincides with the crossed product. For

1449: instance, in the topological category,

1450: $\cA_\theta$ is identified with the crossed product $C^*$-algebra

1451: \begin{equation}\label{irr-rot-alg}

1452: \cA_\theta = C(S^1) \rtimes_{R_\theta} \Z.

1453: \end{equation}

1454: 

1455: \smallskip

1456: 

1457: The algebra (\ref{algNCT}) has two natural generators:

1458: \begin{equation} \label{U} U=\left\{ \begin{array}{ll}

1459: 1 & n=1 \\ 0 & \text{otherwise} \end{array}\right.  \end{equation}

1460: and

1461: \begin{equation} \label{VV} V=\left\{ \begin{array}{ll}

1462: e^{2\pi ia} & n=0 \\ 0 & \text{otherwise}

1463: \end{array} \right. \end{equation}

1464: In fact, an element $b=(a_{ij})$ of $\cA_\theta$ can be written as

1465: power series

1466: \begin{equation}

1467: b = \sum_{n \in \Z} b_n U^n ,

1468: \label{powerU}

1469: \end{equation}

1470: where each $b_n$ is an element of the algebra (\ref{S1}), with

1471: the multiplication rule given by

1472: \begin{equation}

1473: U h U^{-1} = h \circ R_{\theta}^{-1} \, . \label{RthetaU}

1474: \end{equation}

1475: The algebra \eqref{S1} is generated by the function $V$ on

1476: $S^1$,

1477: \begin{equation}

1478: V(\alpha) = \exp (2\pi i \alpha) \ \ \ \forall \alpha \in S^1

1479: \label{Valpha}

1480: \end{equation}

1481: and it follows that $\cA_\theta$ is generated by two unitaries

1482: $(U,V)$ with presentation given by the relation

1483: \begin{equation}

1484: VU = \lambda \, U V, \ \ \text{ with } \ \  \lambda = \exp (2\pi i \theta) \, .

1485: \label{[U,V]}

1486: \end{equation}

1487: 

1488: \smallskip

1489: 

1490: If we work in the smooth category, then a generic element

1491: $b$ in \eqref{S1} is given by a power series

1492: \begin{equation}

1493: b = \sum_{\Z^2} b_{nm} U^n V^m \ \in \cS (\Z^2)

1494: \label{smoothNCt}

1495: \end{equation}

1496: where $\cS (\Z^2)$ is the Schwartz space of sequences of rapid

1497: decay on $\Z^2$. We refer to the algebra of smooth functions \eqref{smoothNCt}

1498: as $\cC^\infty (\bT^2_\theta)$, where we think of

1499: $\bT^2_\theta$ as the (smooth) non--commutative torus.

1500: 

1501: \smallskip

1502: 

1503: Notice that in the definition (\ref{algNCT}) it is not necessary

1504: to restrict to the condition that $i,j$ lie on the transversal

1505: $T$. It is possible to also form an algebra

1506: \begin{equation} \cB_\theta=\{ \left( a_{ij}

1507: \right) \, \, \, \, i,j\in T^2 \, \, \text{ in the same leaf } \},

1508: \end{equation}

1509: where now the parameter of integration is no longer discrete. This

1510: ought to correspond to the same non--commutative space, and in

1511: fact the algebras are related by

1512: $$ \cB_\theta = \cA_\theta \otimes \cK, $$

1513: where $\cK$ is the algebra of all compact operators.

1514: 

1515: \medskip

1516: 

1517: The tangent space to the ordinary torus $T^2$ is spanned by the

1518: tangent directions $\frac{\partial}{\partial x}$ and

1519: $\frac{\partial}{\partial y}$. By choosing coordinates $U,V$, with

1520: $U=e^{2\pi i x}$ and $V=e^{2\pi i y}$, the tangent vectors are

1521: given by $\frac{\partial}{\partial x}=2\pi i

1522: U\frac{\partial}{\partial U}$ and $\frac{\partial}{\partial

1523: y}=2\pi i V\frac{\partial}{\partial V}$. These have analogs in

1524: terms of derivations of the algebra of the non--commutative torus.

1525: The two commuting vector fields which span the tangent space for

1526: an ordinary (commutative) 2-torus correspond algebraically to two

1527: commuting derivations of the algebra of smooth functions.

1528: 

1529: 

1530: These derivations continue to make sense when we replace the

1531: generators $U$ and $V$ of $C^{\infty} (\bT^2)$ by the generators of

1532: the algebra $\cC^{\infty} (\bT_{\theta}^2)$, which no longer

1533: commute, as shown in \eqref{[U,V]}. The derivations are still

1534: given by the same formulas as in the commutative case,

1535: \begin{equation}

1536: \delta_1 = 2\pi i U \, \frac{\partial}{\partial U} \ \ \ \

1537: \delta_2 = 2\pi i V \, \frac{\partial}{\partial V}

1538: \label{derivationsUV}

1539: \end{equation}

1540: so that $\delta_1 \left( \sum b_{nm} U^n V^m \right) = 2\pi i \sum

1541: n b_{nm} U^n V^m$, and similarly for $\delta_2$.

1542: 

1543: 

1544: The operators (\ref{derivationsUV}) are commuting derivations of

1545: the algebra $\cC^{\infty}(\bT_{\theta}^2)$.

1546: 

1547: 

1548: In fact, it is straightforward to verify that $\delta_1$ and

1549: $\delta_2$ satisfy

1550: \begin{equation}

1551: \delta_1 \delta_2 = \delta_2 \delta_1

1552: \label{der-commuting}

1553: \end{equation}

1554: and

1555: \begin{equation}

1556: \delta_j (bb') = \delta_j (b) b' + b \delta_j (b') \quad \forall b

1557: , b' \in \cA_\theta. \label{der-rule}

1558: \end{equation}

1559: 

1560: Just as in the classical case of a (commutative)

1561: manifold, what ensures that the derivations considered are enough

1562: to span the whole tangent space is the condition of ellipticity

1563: for the Laplacian

1564: $$ \Delta = \delta_1^2 + \delta_2^2. $$

1565: In Fourier modes the Laplacian is of the form $n^2 + m^2$, hence

1566: $\Delta^{-1}$ is a compact operator.

1567: 

1568: 

1569: \medskip

1570: 

1571: The geometry of the Kronecker foliation is closely related to the

1572: structure of the algebra. In fact, a choice of a {\it closed

1573: transversal} $T$ of the foliation corresponds

1574: canonically to a {\it finite projective module} over the algebra

1575: $\cA_\theta$.

1576: 

1577: In fact, the main result on finite projective module over the

1578: non--commutative tori $\bT_{\theta}^2$ is the following

1579: classification, which is obtained by combining the results of

1580: \cite{PV}, \cite{C3}, \cite{Rieffel}.

1581: 

1582: 

1583: \begin{thm}\label{NCTmodules}

1584: Finite projective modules over $\cA_{\theta}$ are classified up to

1585: isomorphism by a pair of integers $(p,q)$ such that $p+q \theta

1586: \geq 0$. For a choice of such pair, the corresponding module

1587: ${\mathcal H}_{p,q}^{\theta}$ is obtained from the transversal

1588: $T_{p,q}$ given by the closed geodesic of the torus $T^2$

1589: specified by $(p,q)$, via the following construction. Elements of

1590: the module associated to the transversal $T_{p,q}$ are rectangular

1591: matrices, $(\xi_{i,j})$ with $(i,j) \in T \times S^1$, and with

1592: $i$ and $j$ belonging to the same leaf. The right action of

1593: $(a_{i,j}) \in \cA_{\theta}$ is by matrix multiplication.

1594: \end{thm}

1595: 

1596: For instance, from the transversal $x=0$ one obtains the following

1597: right module over $\cA_\theta$. The underlying linear space is the

1598: usual Schwartz space

1599: \begin{equation}

1600: \cS (\R) = \{ \xi : \, \xi (s) \in \C, \quad \forall s \in \R

1601: \} \label{SRmodule}

1602: \end{equation}

1603: of complex valued smooth functions on $\R$, all of whose

1604: derivatives are of rapid decay.

1605: The right module structure is given by the action of

1606: the generators $U,V$

1607: \begin{equation}

1608: (\xi U) (s) = \xi (s+\theta) \ \ \ \ (\xi V) (s) = e^{2\pi is} \xi

1609: (s) \ \ \ \forall s \in \R \, .

1610: \label{actionUV}

1611: \end{equation}

1612: One of course checks that the relation \eqref{[U,V]} is satisfied,

1613: and that, as a right module over

1614: $\cA_\theta$, the space $\cS (\R)$ is {\it finitely generated}

1615: and {\it projective} (\ie it complements to a free module).

1616: 

1617: \smallskip

1618: 

1619: Finitely generated projective modules play an important role in

1620: noncommutative geometry, as they replace {\em vector bundles} in the

1621: commutative setting. In fact, in ordinary commutative geometry,

1622: one can equivalently describe vector bundles through their sections,

1623: which in turn form a finite projective module over the algebra of

1624: smooth functions. The

1625: notion of finite projective module continues to make sense in the

1626: non--commutative setting, and provides this way a good notion of

1627: ``non--commutative vector bundles''.

1628: 

1629: \smallskip

1630: 

1631: Suppose given a vector bundle $E$, described algebraically through its

1632: space of smooth sections $C^{\infty} (X,E)$. One can

1633: compute the dimension of $E$ by computing the trace of the identity

1634: endomorphism. In terms of the space of smooth sections, hence of finite

1635: projective modules, it is possible to recover the dimension of the

1636: vector bundle as a limit

1637: \begin{equation} \label{dimE}

1638: \dim_{\cA}(\cE)= \lim_{N\to \infty}\frac{1}{N}\left( \#

1639: \text{Generators of} \, \, \underbrace{\cE \oplus \cdots \oplus

1640: \cE}_{N \text{times}} \right).

1641: \end{equation}

1642: This method applies to the noncommutative setting. In the case

1643: of noncommutative tori, one finds that the Schwartz space $\cS(\R)$

1644: has dimension the real number

1645: \begin{equation}

1646: \dim_{\cB} (\cS) = \theta \, .

1647: \label{dimtheta}

1648: \end{equation}

1649: One similarly finds values $p+q\theta$ for the more general

1650: case of Theorem \ref{NCTmodules}.

1651: 

1652: \smallskip

1653: 

1654: The appearance of a real values dimension is related to the {\em

1655: density} of transversals in the leaves, that is, the limit of

1656: $$ \frac{\# B_R \cap S}{\text{size of}\, \, B_R}, $$

1657: for a ball $B_R$ of radius $R$ in the leaf. In this sense, the

1658: dimension $\theta$ of the Schwartz space measure the relative

1659: densities of the two transversals $S=\{ x=0 \}$ and $T=\{ y=0 \}$.

1660: 

1661: 

1662: In general, the appearance of non integral dimension is a basic

1663: feature of von

1664: Neumann algebras of type II. The dimension of a vector bundle

1665: is the only invariant that remains when one looks from the measure

1666: theoretic point of view, using the algebra of measurable functions

1667: $L^\infty(S^1)$ in (\ref{S1}). The von Neumann algebra which

1668: describes the quotient space from the measure theoretic point of

1669: view is the crossed product

1670: \begin{equation}

1671: R = L^{\infty} (S^1) \rtimes_{R_{\theta}} \Z.

1672: \label{measNCT}

1673: \end{equation}

1674: This is the well known hyperfinite factor of type II$_1$. In

1675: particular the classification of finite projective modules $\cE$

1676: over $R$ is given by a positive real number, the Murray and von

1677: Neumann dimension

1678: \begin{equation}

1679: \dim_R (\cE) \in \R_+ \, . \label{MvNR}

1680: \end{equation}

1681: 

1682: \medskip

1683: 

1684: The simplest way to describe the phenomenon of Morita equivalence

1685: for non--commutative tori is in terms of the Kronecker foliation,

1686: where it corresponds to reparameterizing the leaves space in terms

1687: of a different closed transversal. Thus, Morita equivalence of the

1688: algebras $\cA_{\theta}$ and $\cA_{\theta'}$ for $\theta$ and

1689: $\theta'$ in the same $\PGL(2,\Z)$ orbit becomes simply a

1690: statement that the leaf--space of the original foliation is

1691: independent of the transversal used to parameterize it. For

1692: instance, Morita equivalence between $\cA_{\theta}$ and

1693: $\cA_{-1/\theta}$ corresponds to changing the parameterization of

1694: the space of leaves from the transversal $T=\{ y=0 \}$ to the

1695: transversal $S=\{ x=0 \}$.

1696: 

1697: More generally, an explicit construction of bimodules

1698: $\cM_{\theta,\theta'}$ was obtained in \cite{C3}. These are

1699: given by the Schwartz space ${\mathcal S}(\R \times \Z/c)$, with

1700: the right action of $\cA_\theta$ given by

1701: $$ Uf\, (x,u)=f\left( x-\frac{c\theta+d}{c}, u-1\right) $$

1702: $$ Vf\, (x,u)=\exp(2\pi i(x-u d/c)) f(x,u) $$

1703: and the left action of $\cA_{\theta'}$

1704: $$ U'f\, (x,u)=f\left( x-\frac{1}{c}, u-a \right) $$

1705: $$ V'f\, (x,u)=\exp\left( 2\pi i \left(\frac{x}{c\theta+d}

1706: -\frac{u}{c}\right)\right) f(x,u). $$

1707: The bimodule $\cM_{\theta,\theta'}$ realizes the

1708: Morita equivalences between $\cA_\theta$ and

1709: $\cA_{\theta'}$ for

1710: $$ \theta'=\frac{a\theta + b}{c\theta + d}=g\theta $$

1711: with $g\in \PGL(2,\Z)$, \cf \cite{C3}, \cite{Ri2}.

1712: 

1713: \bigskip

1714: 

1715: 

1716: \section{Duals of discrete groups}\label{discgr}

1717: 

1718: 

1719: Noncommutative geometry provides naturally

1720: a generalization of Pontrjagin duality for

1721: discrete groups. While the Pontrjagin dual $\hat\Gamma$ of a

1722: finitely generated discrete abelian group is a compact abelian group,

1723: the dual of a more general finitely generated discrete group

1724: is a noncommutative space.

1725: 

1726: To see this, recall that the usual Pontrjagin duality assigns to

1727: a finitely generated discrete abelian group $\Gamma$ its group of

1728: characters $\hat\Gamma=\Hom(\Gamma,U(1))$. The duality

1729: is given by Fourier transform $e^{i\langle k, \gamma\rangle}$, for

1730: $\gamma\in \Gamma$ and $k\in \hat\Gamma$.

1731: 

1732: 

1733: In particular, Fourier transform gives an identification between

1734: the algebra of functions on

1735: $\hat\Gamma$ and the (reduced) $C^*$-algebra of

1736: the group $\Gamma$,

1737: \begin{equation}\label{CGammadual}

1738: C(\hat\Gamma) \cong C^*_r(\Gamma),

1739: \end{equation}

1740: where the reduced $C^*$-algebra $C^*_r(\Gamma)$ is the

1741: $C^*$-algebra generated by $\Gamma$ in the regular representation

1742: on $\ell^2(\Gamma)$.

1743: 

1744: 

1745: When $\Gamma$ is non-abelian Pontrjagin duality no

1746: longer applies in the classical sense. However, the left hand side of

1747: \eqref{CGammadual} still makes sense and it behaves ``like''

1748: the algebra of functions on the dual group. One can then

1749: say that, for a non-abelian group, the Pontrjagin dual

1750: $\hat\Gamma$ still exists as a noncommutative space whose algebra

1751: of coordinates is the $C^*$-algebra $C^*_r(\Gamma)$.

1752: 

1753: \medskip

1754: 

1755: As an example that illustrates this general philosophy we

1756: give a different version of Example \ref{ex2}.

1757: 

1758: \begin{ex}{\em

1759: The algebra (\ref{alg1}) of Example \ref{ex2}, is the group ring of

1760: the dihedral group $\Z \rtimes \Z/2 \cong \Z/2 * \Z/2$. }

1761: \label{dihedral}

1762: \end{ex}

1763: 

1764: In fact, first notice that to a representation of the group $\Z/2 *

1765: \Z/2$ (free product of two copies of the group with two elements) is

1766: the same thing as a pair of subspaces in the Hilbert space,

1767: $E,F\subset \cH$. The corresponding operators are $U=I-2P_E$,

1768: $V=I-2P_V$, with $P_E,P_V$ the projections. The operators $U,V$

1769: represent reflections, since $U=U^*$, $U^2=I$, $V=V^*$, $V^2=I$. The

1770: group $\Gamma =\Z/2 * \Z/2$ realized as words in the generators $U$

1771: and $V$ can equivalently be described as the semi-direct product

1772: $\Gamma =\Z \rtimes \Z/2$, by setting $X=UV$, with the action

1773: $UXU^{-1}=X^{-1}$. The regular representation of $\Gamma$ is

1774: analyzed using Mackey's theory for semi-direct products. 0ne

1775: considers first representations of the normal subgroup, and then

1776: orbits of the action of $\Z/2$. The irreducible representations of

1777: the normal subgroup $\Z$ are labeled by $S^1=\{z\in\C\,|z|=1\}$ and

1778: given by $X^n\mapsto z^n$. The action of $\Z/2$ is the involution given

1779: by conjugation $z \mapsto \bar z$. The quotient of $S^1$ by the

1780: $\Z/2$ action is identified with the interval $[-1,1]$ by the map

1781: $z\mapsto \Re( z)$. For points inside the interval the corresponding

1782: irreducible representation of $\Gamma$ is two dimensional. At each

1783: of the two endpoints $\pm 1$ one gets two inequivalent irreducible

1784: representations of $\Gamma$. Thus we recover the picture of Example

1785: \ref{ex2} and an isomorphism $C^*(\Gamma)\sim A$ where $A$ is the

1786: algebra \eqref{alg1}.

1787: 

1788: \medskip

1789:  The first two basic steps of the general theory are known

1790: for arbitrary discrete groups $\Gamma$, namely

1791: 

1792: 1) The

1793: resolution of the diagonal and computation of the cyclic cohomology

1794: are provided by the geometric model (due to Burghelea \cite{Burgh}) given by the

1795: free loop space of the classifying space $B\Gamma$.

1796: 

1797: 2) The assembly map (BC-map) of \cite{BauCo}

1798: from the K-homology of the classifying

1799: space $B\Gamma$ to the K-theory of the reduced $C^*$-algebra

1800: $C^*_r(\Gamma)$ is refined in \cite{BCH} to take care of torsion

1801: in the group

1802: $\Gamma$ and gives a pretty good approximation to the K-theory of

1803: $C^*_r(\Gamma)$ (\cf \cite{Skan1} and references therein).

1804: 

1805: In the presence of a natural

1806: smooth subalgebra of $C^*_r(\Gamma)$ containing the group ring and

1807: stable under holomorphic functional calculus, the combination of

1808: the two steps described above makes it possible  to prove an index

1809: theorem which is an higher dimensional form of Atiyah's $L^2$-index

1810: theorem for coverings. This gave the first proof of the Novikov

1811: conjecture for hyperbolic groups (\cite{CoMoind}). Since then the

1812: analysis of dense smooth subalgebras has played a key role, in

1813: particular in the ground breaking work of Vincent Lafforgue.

1814: See \cite{BauCo}, \cite{Julg}, \cite{Laff}, \cite{Skan1}, \cite{Skan2}.

1815: 

1816: 

1817: The next step, \ie the construction of a spectral geometry, is

1818: directly related to geometric group theory. In general one cannot

1819: expect to get a finite dimensional spectral triple since the growth

1820: properties of the group give (except for groups of polynomial

1821: growth) a basic obstruction (\cf \cite{Co-fr}). A general

1822: construction of a theta summable spectral triple was given in

1823: \cite{Co94} Section IV.9. Basically the transition from finitely

1824: summable spectral triples to the theta summable ones is the

1825: transition from finite dimensional geometry to the infinite

1826: dimensional case. In the theta summable case the Chern character is

1827: no longer a finite dimensional cyclic cocycle and one needs to

1828: extend cyclic cohomology using cocycles with infinite support in the

1829: $(b,B)$ bicomplex fulfilling a subtle growth condition. The general

1830: theory of {\em entire} cyclic cohomology was developed in

1831: \cite{Cotheta}. It is in general quite difficult to compute the

1832: Chern character in the theta summable case and one had to wait quite

1833: a long time until it was done for the basic example of discrete

1834: subgroups of semi-simple Lie groups. This has been achieved in a

1835: remarkable recent paper of M. Puschnigg \cite{Pusch} in the case of

1836: real rank one.

1837: 

1838: The fourth step \ie the thermodynamics might seem irrelevant in the

1839: type II context of discrete groups. However a small variant of the

1840: construction of the group ring, namely the Hecke algebra associated

1841: to an almost normal inclusion of discrete groups (in the sense

1842: considered in \cite{BC})

1843: suffices to meet the type III world. One of the open fields is to

1844: extend the above steps 1), 2) and 3) in the general context of

1845: almost normal inclusions of discrete groups, and to perform the

1846: thermodynamical analysis in the spirit of \cite{CCM} in that

1847: context.

1848: 

1849: 

1850: \bigskip

1851: 

1852: 

1853: \section{Brillouin zone and the quantum Hall effect}\label{qhe}

1854: 

1855: 

1856: 

1857: An important application to physics of the theory of

1858: non--commutative tori was the development of a rigorous

1859: mathematical model for the Integer Quantum Hall Effect (IQHE)

1860: obtained by Bellissard and collaborators \cite{Bellissard1},

1861: \cite{Bellissard2}, \cite{Co94}.

1862: 

1863: The classical Hall effect is a physical phenomenon first observed

1864: in the XIX century \cite{Hall}. A very thin metal sample is

1865: immersed in a constant uniform strong magnetic field orthogonal to

1866: the surface of the sample. By forcing a constant current to flow

1867: through the sample, the flow of charge carriers in the metal is

1868: subject to a Lorentz force perpendicular to the current and the

1869: magnetic field. The equation for the equilibrium of forces in the

1870: sample is of the form

1871: \begin{equation} \label{clHall} N e {\bf E} + {\bf j} \wedge {\bf B}

1872: =0, \end{equation} where ${\bf E}$ is the electric field, $e$ and

1873: $N$ the charge and number of the charge carriers in the metal,

1874: ${\bf B}$ the magnetic field, and ${\bf j}$ the current.

1875: 

1876: 

1877: \begin{center}

1878: \begin{figure}

1879: \includegraphics[scale=1.1]{qhe.eps}

1880: \caption{The quantum Hall effect experiment.

1881: \label{FigQHE}}

1882: \end{figure}

1883: \end{center}

1884: 

1885: 

1886: The equation \eqref{clHall} defines a linear relation: the ratio

1887: of the intensity of the Hall current to the intensity of the

1888: electric field defines the Hall conductance,

1889: \begin{equation} \label{condHall} \sigma_H = \frac{N e \delta}{B},

1890: \end{equation}

1891: with $B=|{\bf B}|$ the intensity of the magnetic field and

1892: $\delta$ the sample width. The dimensionless quantity

1893: \begin{equation} \label{filling} \nu_H = \frac{N\delta h}{B e} =

1894: \sigma_H R_H \end{equation} is called the filling factor, while

1895: the quantity $R_H = h/e^2$ is the Hall resistance. The filling

1896: factor measures the fraction of Landau level filled by conducting

1897: electrons in the sample. Thus, classically, the Hall conductance,

1898: measured in units of $e^2/h$, equals the filling factor.

1899: 

1900: In 1980, about a century after the classical Hall effect was

1901: observed, von Klitzing's experiment showed that, lowering the

1902: temperature below 1 K, the relation of Hall conductance to filling

1903: factor shows plateaux at integer values, \cite{vonKlitz}. The

1904: integer values of the Hall conductance are observed with a

1905: surprising experimental accuracy of the order of $10^{-8}$. This

1906: phenomenon of quantization of the Hall conductance is known as

1907: Integer Quantum Hall Effect (IQHE).

1908: 

1909: Laughlin first suggested that IQHE should be of a geometric origin

1910: \cite{Lau}. A detailed mathematical model of the IQHE, which

1911: accounts for all the important features of the experiment

1912: (quantization, localization, insensitivity to the presence of

1913: disorder, vanishing of direct conductance at plateaux levels)

1914: improving over the earlier Laughlin model, was developed by

1915: Bellissard and collaborators \cite{Bellissard1},

1916: \cite{Bellissard2}.

1917: 

1918: Bellissard's approach to the IQHE is based on non--commutative geometry.

1919: The quantization of the Hall conductance at integer values is

1920: indeed geometric in nature: it resembles another well known

1921: ``quantization'' phenomenon that happens in the more familiar

1922: setting of the geometry of compact 2--dimensional manifolds,

1923: namely the Gauss--Bonnet theorem, where the integral of the

1924: curvature is an integer multiple of $2\pi$, a property that is

1925: stable under deformations. In the same spirit, the values of the

1926: Hall conductance are related to the evaluation of a certain

1927: characteristic class, or, in other words, to an index theorem for

1928: a Fredholm operator.

1929: 

1930: More precisely, in the physical model one makes the simplifying

1931: assumption that the IQHE can be described by non--interacting

1932: particles. The Hamiltonian then describes the motion of a single

1933: electron subject to the magnetic field and an additional potential

1934: representing the lattice of ions in the conductor. In a perfect

1935: crystal and in the absence of a magnetic field, there is a group

1936: of translational symmetries. This corresponds to a group of

1937: unitary operators $U(a)$, $a\in G$, where $G$ is the locally

1938: compact group of symmetries. Turning on the magnetic field breaks

1939: this symmetry, in the sense that translates of the Hamiltonian

1940: $H_a = U(a) H U(a)^{-1}$ no longer commute with the Hamiltonian

1941: $H$. Since there is no preferred choice of one translate over the

1942: others, the algebra of observables must include all translates of

1943: the Hamiltonian, or better their resolvents, namely the bounded

1944: operators

1945: \begin{equation} \label{translR} R_a(z) = U(a) (zI - H)^{-1}

1946: U(a)^{-1}. \end{equation}

1947: 

1948: For a particle of (effective) mass $m$ and charge $e$ confined to

1949: the plane, subject to a magnetic field of vector potential ${\bf

1950: A}$ and to a bounded potential $V$, the Hamiltonian is of the form

1951: \begin{equation} \label{magnLapl} H= \frac{1}{2m} \sum_{j=1,2} (p_j -e

1952: A_j)^2 + V = H_0 + V, \end{equation} where the unperturbed part

1953: $H_0$ is invariant under the magnetic translations, namely the

1954: unitary representation of the translation group $\R^2$ given by

1955: $$  U(a) \psi(x) = \exp\left( \frac{-ieB}{2\hbar} \omega(x,a) \right)

1956: \psi(x-a), $$ with $\omega$ the standard symplectic form in the

1957: plane. The hull (strong closure) of the translates (\ref{translR})

1958: yields a topological space, whose homeomorphism type is

1959: independent of the point $z$ in the resolvent of $H$. This

1960: provides a non--commutative version of the Brillouin zone.

1961: 

1962: Recall that the Brillouin zones of a crystals are fundamental

1963: domains for the reciprocal lattice $\Gamma^\sharp$ obtained via

1964: the following inductive procedure. The {\em Bragg hyperplanes} of

1965: a crystal are the hyperplanes along which a pattern of diffraction

1966: of maximal intensity is observed when a beam of radiation (X-rays

1967: for instance) is shone at the crystal. The $N$-th Brillouin zone

1968: consists of all the points in (the dual) $\R^d$ such that the line

1969: from that point to the origin crosses exactly $(n-1)$ Bragg

1970: hyperplanes of the crystal.

1971: 

1972: \begin{figure}

1973: \begin{center}

1974: \includegraphics[scale=0.5]{2dBZonesBW.ps}

1975: \end{center}

1976: \caption{Brillouin zones for a 2-dimensional crystal

1977: \label{FigBrill}}

1978: \end{figure}

1979: 

1980: More precisely, in our case, if $e_1$ and $e_2$ are generators of the

1981: periodic lattice, we obtain a commutation relation

1982: $$ U(e_1) U(e_2) = e^{2\pi i \theta} U(e_2) U(e_1), $$

1983: where $\theta$ is the flux of the magnetic field through a

1984: fundamental domain for the lattice, in dimensionless units, hence

1985: the non-commutative Brillouin zone is described by a

1986: non-commutative torus.

1987: 

1988: This can also be seen easily in a discrete model, where the

1989: Hamiltonian is given by an operator

1990: \begin{equation}\label{discr-Ham}

1991: \begin{array}{rll}

1992: (H_a \, f)(m,n) = & e^{-ia_1 n} f(m+1,n) + & e^{ia_2 n} f(m-1,n)

1993: \\[3mm] + &

1994:  e^{-ia_2 m} f(m, n+1) + & e^{ia_2 m } f(m, n-1),

1995: \end{array}

1996: \end{equation}

1997: for $f\in L^2(\Z^2)$. This is a discrete version of the magnetic

1998: Laplacian. Notice then that \eqref{discr-Ham} can be written in

1999: the form

2000: $$ H_a = U+V + U^* + V^*, $$

2001: for

2002: $$ (U\, f)(m,n)= e^{-ia_2m} f(m,n+1) \ \ \ \ (V\, f)(m,n)=

2003: e^{-ia_1 n} f(m+1,n).

2004: $$

2005: These clearly satisfy the commutation relation \eqref{[U,V]} of

2006: $\bT^2_\theta$ with $\theta= a_2-a_1$.

2007: 

2008: 

2009: In the zero-temperature limit, the Hall conductance satisfies the

2010: Kubo formula

2011: \begin{equation}\label{kubo} \sigma_H = \frac{1}{2\pi i R_H}

2012: \tau(P_\mu [\delta_1 P_\mu, \delta_2 P_\mu]),

2013: \end{equation}

2014: where $P_\mu$ is a spectral projection of the Hamiltonian on

2015: energies smaller or equal to the Fermi level $E_\mu$, $\tau$ is

2016: the trace on $\cA_\theta$ given by

2017: \begin{equation} \label{trace-theta}

2018: \tau\left( \sum a_{n,m} U^n V^m \right) = a_{0,0}.

2019: \end{equation}

2020: and $\delta_1$, $\delta_2$ are as in

2021: \eqref{derivationsUV}. Here we assume that the Fermi level $\mu$

2022: is in a gap in the spectrum of the Hamiltonian. Then the spectral

2023: projections $P_\mu$ belong to the $C^*$-algebra of observables.

2024: 

2025: The Kubo formula \eqref{kubo} can be derived from purely physical

2026: considerations, such as transport theory and the quantum adiabatic

2027: limit.

2028: 

2029: The main result then is the fact that the integrality of the

2030: conductance observed in the Integer Quantum Hall Effect is explained

2031: topologically, that is, in terms of the integrality of the {\em

2032: cyclic cocycle} $\tau(a^0 (\delta_1 a^1 \delta_2 a^2 - \delta_2 a^1

2033: \delta_1 a^2))$ (\cf \cite{C3}).

2034: 

2035: \medskip

2036: 

2037: \begin{center}

2038: \begin{figure}

2039: \includegraphics[scale=0.6]{FQHE.eps}

2040: \caption{Observed fractions in the quantum Hall effect.

2041: \label{FigFQHE}}

2042: \end{figure}

2043: \end{center}

2044: 

2045: 

2046: The fractional QHE was discovered by Stormer and Tsui in 1982. The

2047: setup is as in the quantum Hall effect: in a high quality

2048: semi-conductor interface, which will be modelled by an infinite

2049: 2-dimensional surface, with low carrier concentration and

2050: extremely low temperatures $\sim 10 mK$, in the presence of a very

2051: strong magnetic field, the experiment shows that the graph of

2052: $\frac{h}{e^2}\sigma_H$ against the filling factor $\nu$ exhibits

2053: plateaux at certain fractional values.

2054: 

2055: The independent electron approximation that, in the case of

2056: the integer quantum Hall effect, reduces the problem to a single

2057: electron wavefunction is no longer viable in this case and one has to

2058: incorporate the Coulomb interaction between the electrons in a

2059: many-electron theory. Nonetheless, it is possible to use a crude

2060: approximation, whereby one alters the underlying geometry to

2061: account for an average effect of the multi-electron interactions.

2062: One can obtain this way a model of the fractional quantum Hall effect

2063: via noncommutative geometry (\cf \cite{MM1}, \cite{MM2}, \cite{MM3}),

2064: where one uses hyperbolic geometry to simulate the interactions.

2065: 

2066: The noncommutative geometry approach to the quantum Hall effect

2067: described above was extended to hyperbolic geometry in \cite{CHMM}.

2068: The analog of the operator \eqref{discr-Ham} is given by the Harper

2069: operator on the Cayley graph of a finitely generated discrete subgroup

2070: $\Gamma$ of $\PSL_2(\R)$. Given $\sigma: \Gamma \times

2071: \Gamma \to U(1)$ satisfying $\sigma(\gamma,1)=\sigma(1,\gamma)=1$

2072: and

2073: $$ \sigma(\gamma_1,\gamma_2)\sigma(\gamma_1\gamma_2,\gamma_3)=

2074: \sigma(\gamma_1,\gamma_2\gamma_3)\sigma(\gamma_2,\gamma_3), $$

2075: one considers the right $\sigma$-regular representation on

2076: $\ell^2(\Gamma)$ of the form

2077: \begin{equation}\label{rightreg}

2078: R^\sigma_\gamma

2079: \psi(\gamma') = \psi(\gamma' \gamma) \sigma(\gamma', \gamma)

2080: \end{equation}

2081: satisfying $$R^\sigma_\gamma

2082: R^\sigma_{\gamma'} = \sigma(\gamma,\gamma')

2083: R^\sigma_{\gamma\gamma'}.$$

2084: For $\{\gamma_i\}_{i=1}^r$ a symmetric set of generators of

2085: $\Gamma$, the Harper operator is of the form

2086: \begin{equation}\label{HarperGamma}

2087:  \cR_\sigma = \sum_{i=1}^r  R^\sigma_{\gamma_i},

2088: \end{equation}

2089: and the operator $r-\cR_\sigma$ is the discrete analog of the magnetic

2090: Laplacian (\cf \cite{Sun}).

2091: 

2092: The idea is that, by effect of the strong

2093: interaction with the other electrons, a single electron ``sees''

2094: the surrounding geometry as hyperbolic, with lattice sites that

2095: appear (as a multiple image effect) as the points in a lattice

2096: $\Gamma\subset \PSL_2(\R)$. Thus, one considers the general form

2097: of such a lattice

2098: \begin{equation}\label{FuchsianG}

2099: \Gamma = \Gamma(g; \nu_1, \ldots, \nu_n),

2100: \end{equation}

2101: with generators $a_i, b_i, c_j$, with $i=1,\ldots,g$ and

2102: $j=1,\ldots,n$ and a presentation of the form

2103: \begin{equation}\label{presentation}

2104: \Gamma(g; \nu_1, \ldots, \nu_n)=\langle a_i, b_i, c_j\,\,

2105: \left|\,\, \prod_{i=1}^g [a_i,b_i]c_1\cdots c_n =1, \,\,\,

2106: c_j^{\nu_j} =1 \right. \rangle.

2107: \end{equation}

2108: The quotient of the action of $\Gamma$ by isometrieson $\H$,

2109: \begin{equation}\label{hyporb}

2110: \Sigma(g; \nu_1, \ldots, \nu_n):= \Gamma \backslash \H,

2111: \end{equation}

2112: is a hyperbolic orbifold.

2113: 

2114: \begin{center}

2115: \begin{figure}

2116: \includegraphics[scale=0.6]{orbifold.eps}

2117: \caption{Hyperbolic orbifolds.

2118: \label{FigOrbi}}

2119: \end{figure}

2120: \end{center}

2121: 

2122: 

2123: Let $P_E$ denote denote the spectral projection associated to the

2124: Fermi level, \ie $P_E = \chi_{(-\infty, E]}(H)$. Then, in the zero

2125: temperature limit, the Hall conductance is given by

2126: \begin{equation}\label{sigmaE}

2127: \sigma_E = \tr_K (P_E, P_E, P_E),

2128: \end{equation}

2129: where $\tr_K$ denotes the conductance 2-cocycle. It is a

2130: cyclic 2-cocycle on the twisted group algebra $\C(\Gamma, \sigma)$

2131: of the form

2132: \begin{equation}\label{trK}

2133:  \tr_K(f_0, f_1, f_2) =

2134: \sum_{j=1}^g \tr(f_0  (\delta_j(f_1)\delta_{j+g}(f_2) -

2135:   \delta_{j+g}(f_1)\delta_j(f_2))),

2136: \end{equation}

2137: where the $\delta_j$ are derivations associated to the

2138: 1-cocycles $a_j$ associated to a symplectic basis

2139: $\{ a_j, b_j\}_{j=1,\ldots, g}$ of $H^1(\Gamma, \R)$ (\cf

2140: \cite{MM3}).

2141: 

2142: Within this model, one obtains the fractional values of the Hall

2143: conductance as integer multiples of orbifold Euler characteristics

2144: \begin{equation}\label{orbEulch}

2145: \chi_{orb}(\Sigma(g;\nu_1,\ldots,\nu_n))=2-2g +\nu -n  \in \Q.

2146: \end{equation}

2147: In fact, one shows (\cf \cite{MM2}, \cite{MM3}) that the conductance

2148: 2-cocycle is cohomologous to another cocycle, the area 2-cocycle, for

2149: which one can compute the values on $K$-theory (hence the value of

2150: \eqref{sigmaE}) by applying a twisted version of the Connes--Moscovici

2151: higher index theorem \cite{CoMoind}.

2152: 

2153: \smallskip

2154: 

2155: While in the case of the integer quantum Hall effect the

2156: noncommutative geometry model is completely satisfactory and explains

2157: all the physical properties of the system, in the fractional case

2158: the orbifold model can be considered

2159: as a first rough approximation to the quantum field theory that

2160: coverns the fractional quantum Hall effect. For instance, the geometry of

2161: 2-dimensional hyperbolic orbifolds is related to Chern--Simons theory

2162: through the moduli spaces of vortex equations. This remains an interesting

2163: open question.

2164: 

2165: \bigskip

2166: 

2167: 

2168: \section{Tilings}\label{tiles}

2169: 

2170: In general, by a {\em tiling} $\cT$ in $\R^d$ one means the following. One

2171: considers a finite collection $\{ \tau_1, \ldots, \tau_N \}$ of closed

2172: bounded subsets of $\R^d$ homeomorphic to the unit ball. These are

2173: called the {\em prototiles}. One usually assumes that the prototiles

2174: are polytopes in $\R^d$ with a single $d$-dimensional cell which is

2175: the interior of the prototile, but this assumption can be relaxed.

2176: A tiling $\cT$ of $\R^d$ is then a covering of $\R^d$ by sets with disjoint

2177: interior, each of which is a {\em tile}, that is, a translate of one

2178: of the prototiles.

2179: 

2180: \begin{center}

2181: \begin{figure}

2182: \includegraphics[scale=0.8]{PersianArt.eps}

2183: \caption{Prototiles for tilings.

2184: \label{FigTiles}}

2185: \end{figure}

2186: \end{center}

2187: 

2188: 

2189: Given a tiling $\cT$ of $\R^d$ one can form its orbit closure

2190: under translations. The metric on tilings is defined by saying that

2191: two tilings are close if they almost agree on a large ball centered

2192: at the origin in $\R^d$ (for more details and equivalent definitions

2193: see \eg \cite{AndPut}, \cite{BellC}).

2194: 

2195: Tilings can be periodic or aperiodic. There are many familiar examples

2196: of periodic tilings, while the best known examples of aperiodic

2197: tilings are the Penrose tilings \cite{Penrose}. Similar types of

2198: aperiodic tilings have been widely studied in the physics of

2199: quasicrystals (\cf \eg \cite{BaMo}, \cite{BellC}).

2200: 

2201: It was understood very early on in the development of noncommutative

2202: geometry (\cf \cite{Co-fr} and pp.5--7, pp.88--93, and pp.175--178 of

2203: \cite{Co94}) that Penrose tilings provide an interesting class of

2204: noncommutative spaces.

2205: 

2206: In fact, one can consider on the set $\Omega$ of tilings $\cT$ with

2207: given prototiles $\{ \tau_1, \ldots, \tau_N \}$ the equivalence relation

2208: given by the action of $\R^d$ by translations, \ie one identifies

2209: tilings that can be obtained from one another by translations. In the

2210: case of aperiodic tilings, this yields the type of quotient

2211: construction described in Section \ref{quotients}, which leads

2212: naturally to noncommutative spaces. An explicit description of this

2213: noncommutative space for the case of Penrose tilings can be found in

2214: \S II.3 of \cite{Co94}.

2215: 

2216: \begin{center}

2217: \begin{figure}

2218: \includegraphics[scale=0.6]{zellijs.eps}

2219: \caption{Quasiperiodic tilings and zellijs.

2220: \label{FigZell}}

2221: \end{figure}

2222: \end{center}

2223: 

2224: To simplify the picture slightly, we can consider the similar problem

2225: (dually) with arrangements of points in $\R^d$ instead of

2226: tilings. This is the formulation used in the theory of aperiodic

2227: solids and quasicrystals (\cf \cite{BellC}). Then, instead of tilings

2228: $\cT$, we consider discrete subsets of points $\cL\subset \R^d$. Such

2229: $\cL$ is a Delauney set if there are radii $r,R>0$ such that every

2230: open ball of radius $r$ meets $\cL$ in at most one point and every

2231: closed ball of radius $R$ meets $\cL$ in at least one point. One can

2232: describe $\cL$ by the counting measure

2233: $$ \mu_{\cL}(f) =\sum_{x\in\cL} f(x), $$

2234: and one can take the orbit closure $\Omega$ of the action of $\R^d$ by

2235: translations $$\mu_{\cL}\mapsto T_{-a}\mu_{\cL}=\mu_{\cL}\circ T_a, \

2236: \ \ \text{ for } a\in \R^d, $$

2237: in the space $\cM(\R^d)$ of Radon measures with the weak$^*$

2238: topology. The Hull of $\cL$ is the dynamical system $(\Omega,T)$,

2239: where $T$ denotes the action of $\R^d$ by translations.

2240: 

2241: This dynamical system determines a corresponding noncommutative space,

2242: describing the quotient of $\Omega$ by translations, namely the

2243: crossed product $C^*$-algebra

2244: \begin{equation}\label{crossOmega}

2245: \cA=C(\Omega)\rtimes_T \R^d.

2246: \end{equation}

2247: In fact, one can also consider the groupoid with set of units the

2248: transversal

2249: \begin{equation}\label{transvOmega}

2250: X=\{ \omega\in \Omega: 0\in {\rm Support}(\omega)\},

2251: \end{equation}

2252: arrows of the form $(\omega,a)\in \Omega\times \R^d$, with

2253: source and range maps $s(\omega,a)=T_{-a}\omega$, $r(\omega,a)=\omega$

2254: and $(\omega,a)\circ (T_{-a}\omega,b)=(\omega, a+b)$ (\cf

2255: \cite{BellC}). This defines a locally compact groupoid $\cG(\cL,X)$.

2256: The $C^*$-algebras $C^*(\cG(\cL,X))$ and $C(\Omega)\rtimes_T \R^d$ are

2257: Morita equivalent.

2258: 

2259: In the case where $\cL$ is a periodic arrangement of points with

2260: cocompact symmetry group $\Gamma\subset \R^d$, the space

2261: $\Omega$ is an ordinary commutative space, which is topologically

2262: a torus $\Omega=\R^d/\Gamma$. The $C^*$-algebra $\cA$ is in this case

2263: isomorphic to $C(\hat\Gamma)\otimes \cK$, where $\cK$ is the algebra

2264: of compact operators and $\hat\Gamma$ is the Pontrjagin dual of the

2265: abelian group $\Gamma\cong \Z^d$, isomorphic to $T^d$, obtained by taking

2266: the dual of $\R^d$ modulo the reciprocal lattice

2267: \begin{equation}\label{recLat}

2268:  \Gamma^\sharp = \{ k \in \R^d : \langle k, \gamma\rangle\in 2

2269: \pi \Z, \forall \gamma \in \Gamma \}.

2270: \end{equation}

2271: Thus, in physical language, $\hat\Gamma$ is identified with the

2272: Brillouin zone $B=\R^d/\gamma^\sharp$ of the periodic

2273: crystal $\cL$ (\cf Section \ref{qhe}). In this periodic case, the

2274: transversal $X=\cL/\Gamma$ is a finite set of points. The groupoid

2275: $C^*$-algebra $C^*(\cG(\cL,X))$ is in this case isomorphic to

2276: $C(\hat\Gamma)\otimes M_k(\C)$, where $k$ is the cardinality of the

2277: transversal $X$. Thus, the periodic case falls back into the realm of

2278: commutative spaces, while the aperiodic patterns give rise to truly

2279: noncommutative spaces, which are highly nontrivial and interesting.

2280: 

2281: \begin{center}

2282: \begin{figure}

2283: \includegraphics[scale=0.6]{muqarnas.eps}

2284: \caption{Quasiperiodic tilings and muqarnas.

2285: \label{FigMuq}}

2286: \end{figure}

2287: \end{center}

2288: 

2289: One of the richest sources of interesting tilings are the

2290: {\em zellijs} and {\em muqarnas} widely used in ancient architecture.

2291: Also collectively defined as ``arabesques'', not only these patterns

2292: exhibit highly nontrivial geometries, but they reflect

2293: the intricate interplay between philosophy, mathematics,

2294: and aesthetics (\cf \cite{ArdBak}, \cite{Bulatov}).

2295: Some of the best studies on

2296: zellijs and muqarnas concentrate on 2-dimensional {\em periodic}

2297: patterns. For instance we find in \cite{ArdBak}, p.43:

2298: \begin{quote}

2299: ``As Nature is based on rhythm, so the arabesque is rhythmic

2300: in concept. It reflects movement marked by the regular recurrence

2301: of features, elements, phenomena; hence it has periodicity.''

2302: \end{quote}

2303: It seems from this viewpoint that only the theory of periodic

2304: tilings (\ie commutative geometry) should be relevant in this

2305: context. However, more recent

2306: studies (\cf \cite{Bulatov}, \cite{Castera}, \cite{Castera2},

2307: \cite{Nagy}) suggest that the design of zellijs and muqarnas was not

2308: limited to 2-dimensional crystallographic groups, but, especially

2309: during the Timurid period, it

2310: involved also {\em aperiodic} patterns with fivefold symmetry, analogous to

2311: those observed in quasi-crystals. This is no accident and was

2312: certainly the result of a highly developed geometric theory:

2313: already in the historic textbook of Abu'l-Wafa' al-Buzjani (940-998)

2314: on geometric constructions \cite{Wafa} there is explicit mention of

2315: meetings and discussions where mathematicians were directly involved

2316: alongside artisans in the design of arabesque patterns.

2317: 

2318: The appearance of aperiodic

2319: tilings is documented in the anonymous Persian manuscript

2320: \cite{Anonym} ``On interlocking similar and congruent figures'', which

2321: dates back to the 11th-13th century. Some of these aperiodic aspects of

2322: zellijs and muqarnas were studied by Bulatov in the

2323: book \cite{Bulatov}, which also contains Vil'danova's Russian

2324: translation of the ancient Persian text.

2325: 

2326: 

2327: \bigskip

2328: 

2329: \section{Noncommutative spaces from dynamical systems}\label{dynsys}

2330: 

2331: 

2332: 

2333: We will look at some examples of noncommutative spaces associated to

2334: a discrete dynamical system $T$, for instance given by a self

2335: mapping of a Cantor set. Such noncommutative spaces

2336:  have been extensively studied in a

2337: series of papers (\cf \cite{GPS} and \cite{skau} for a survey) where

2338: C. Skau and his coworkers have obtained  remarkable results on the

2339: classification of minimal actions of $\Z$ on Cantor

2340:   sets using the  $K$-theory of the associated $C^*$-algebra.

2341: 

2342: It was found recently (\cf \cite{CM1}, \cite{CM2}, \S 4 of

2343: \cite{Mar} and \S 8 of \cite{MaPa})  that the mapping torus of such

2344: systems can be used to model the ``dual graph'' of the fibers at the

2345: archimedean primes of arithmetic surfaces, in Arakelov geometry, in

2346: the particular case in which the dynamical system $T$ is a subshift

2347: of finite type encoding the action of a Schottky group

2348: $\Gamma\subset \SL_2(\C)$ on its limit set $\Lambda_\Gamma\subset

2349: \P^1(\C)$. In fact, the results of \cite{CM1} were motivated by

2350: earlier results of Manin \cite{Man-hyp} that provided a geometric

2351: model for such dual graphs in terms of hyperbolic geometry and

2352: Schottky uniformizations.

2353: 

2354: 

2355: \smallskip

2356: 

2357: More generally, given an alphabet with letters $\{ \ell_1, \ldots, \ell_N \}$,

2358: the space $\cS^+_A$ of a subshift of finite type consists

2359: of all right-infinite {\em admissible} sequences

2360: \begin{equation}\label{admseqplus}

2361: w=a_0 a_1 a_2 \ldots a_n \ldots

2362: \end{equation}

2363: in the letters of the alphabet. Namely, $a_i\in \{ \ell_1, \ldots,

2364: \ell_N \}$ subject to an admissibility condition specified by an

2365: $N\times N$ matrix $A$ with entries in $\{0,1\}$. Two letters

2366: $\ell_i$ and $\ell_j$ in the alphabet can appear as consecutive digits

2367: $a_k$, $a_{k+1}$ in the word $w$ if and only if the entry $A_{ij}$ of

2368: the admissibility matrix $A$ is equal to $1$.

2369: One defines similarly the space $\cS_A$ as the set of doubly-infinite

2370: admissible sequences

2371: \begin{equation}\label{admseq}

2372: w=\ldots a_{-m}\ldots a_{-2} a_{-1} a_0 a_1 a_2 \ldots a_n \ldots

2373: \end{equation}

2374: The sets $\cS^+_A$ and $\cS_A$ have a natural choice of a topology.

2375: In fact, on $\cS_A$ we can put the topology generated by the

2376: sets $W^s(x,\ell) =\{ y\in {\mathcal S}_A | x_k = y_k, k\geq

2377: \ell \}$, and the $W^u(x,\ell)=\{ y\in {\mathcal S}_A | x_k = y_k, k\leq

2378: \ell \}$ for $x\in {\mathcal S}_A$ and $\ell \in \Z$. This induces a

2379: topology with analogous properties on ${\mathcal S}^+_A$ by realizing it as a

2380: subset of ${\mathcal S}_A$, for instance, by extending each sequence to

2381: the left as a constant sequence. One then considers on $\cS_A$ (or on

2382: $\cS_A^+$) the action of the two-sided (resp.~one-sided) shift

2383: $T$ defined by $(Tw)_k = a_{k+1}$, where the $a_k$ are the digits of

2384: the word $w$. Namely, the one-sided shift on $\cS_A^+$ is of the form

2385: \begin{equation}\label{shift+}

2386: T( a_0 a_1 a_2 \ldots a_\ell \ldots) = a_1 a_2 \ldots

2387: a_\ell \ldots

2388: \end{equation}

2389: while the two-sided shift on $\cS_A$ acts as

2390: \begin{equation}\label{shift}

2391: \begin{array}{rcccccccccl} T(& \ldots & a_{-m} & \ldots & a_{-1}& a_0

2392: & a_1 &

2393: \ldots & a_{\ell} & \ldots &) = \\ & \ldots & a_{-m+1} & \ldots & a_{0} &

2394: a_1 & a_2 & \ldots & a_{\ell+1} & \ldots & \end{array}

2395: \end{equation}

2396: Tipically spaces ${\mathcal S}^+_A$ and ${\mathcal S}_A$ are topologically

2397: Cantor sets. The one-sided shift $T$ of \eqref{shift+} is a

2398: continuous surjective map on ${\mathcal S}^+_A$, while the two-sided

2399: shift $T$ of \eqref{shift} is a homeomorphism of ${\mathcal S}_A$.

2400: 

2401: For example, let $\Gamma$ be a free group in $g$ generators $\{

2402: \gamma_1,\ldots, \gamma_g \}$. Consider the alphabet

2403: $\{ \gamma_1,\ldots, \gamma_g, \gamma_1^{-1}, \ldots, \gamma_g^{-1}

2404: \}$. Then one can consider the right-infinite, or doubly-infinite

2405: words in these letters, without cancellations, that is, subject to the

2406: admissibility rule that $a_{k+1}\neq a_k^{-1}$. This defines a

2407: subshift of finite type where the matrix $A$ is the symmetric

2408: $2g\times 2g$ matrix with $A_{ij}=0$ for $|i-j|=g$ and $A_{ij}=1$

2409: otherwise.  Suppose that $\Gamma$ is

2410: a Schottky group of genus $g$, \ie a finitely generated discrete

2411: subgroup $\Gamma\subset \SL_2(\C)$, isomorphic to a free group in $g$

2412: generators, where all nontrivial elements are hyperbolic.

2413: Then the points in $\cS_A^+$ parameterize points in the

2414: limit set $\Lambda_\Gamma\subset \P^1(\C)$ (the set of accumulation

2415: points of orbits of $\Gamma$).

2416: The points in $\cS_A$ parameterize geodesics in the

2417: three dimensional real hyperbolic space $\H^3$ with ends at points on

2418: the limit set $\Lambda_\Gamma$.

2419: 

2420: The pair $({\mathcal S}_A,T)$ is a typical example of an interesting

2421: class of dynamical systems, namely it is a {\em Smale

2422: space}. This means that locally ${\mathcal S}_A$ can be decomposed

2423: as the product of expanding and contracting directions for $T$.

2424: Namely, the following

2425: properties are satisfied.

2426: \begin{itemize}

2427: \item For every point $x\in {\mathcal S}_A$ there exist subsets $W^s(x)$

2428: and $W^u(x)$ of ${\mathcal S}_A$, such that $W^s(x) \times  W^u(x)$ is

2429: homeomorphic to a neighborhood of $x$.

2430: \item The map $T$ is contracting on $W^s(x)$ and expanding on

2431: $W^u(x)$, and $W^s(Tx)$ and $T(W^s(x))$ agree in some neighborhood of

2432: $x$, and so do $W^u(Tx)$ and $T(W^u(x))$.

2433: \end{itemize}

2434: 

2435: A construction of Ruelle shows that one can

2436: associate different $C^*$--algebras to Smale spaces (\cf \cite{Rue2},

2437: \cite{Putn}, \cite{PutSpi}). For Smale spaces like $({\mathcal

2438: S}_A,T)$ there are four basic possibilities: the crossed product algebra

2439: $C({\mathcal S}_A)\rtimes_T \Z$ and the ${\rm C}^*$--algebras ${\rm

2440: C}^*({\mathcal G}^s)\rtimes_T \Z$, ${\rm C}^*({\mathcal G}^u)\rtimes_T

2441: \Z$, ${\rm C}^*({\mathcal G}^a)\rtimes_T \Z$ obtained by considering

2442: the action of the shift $T$ on the groupoid ${\rm C}^*$--algebra

2443: associated to the groupoids ${\mathcal G}^s$, ${\mathcal G}^u$,

2444: ${\mathcal G}^a$ of the  stable, unstable, and asymptotic equivalence

2445: relations on $({\mathcal S}_A,T)$.

2446: 

2447: The first choice, $C({\mathcal S}_A)\rtimes_T \Z$, is

2448: closely related to the continuous dynamical system given by the

2449: mapping torus of $T$, while a choice like ${\rm C}^*({\mathcal

2450: G}^u)\rtimes_T \Z$ is related to the ``bad quotient'' of $\cS_A^+$ by

2451: the action of $T$. In the example of the Schottky group this

2452: corresponds to the action of $\Gamma$ on its limit set.

2453: 

2454: One can consider the suspension flow $\cS_T$ of a dynamical

2455: system $T$, that is, the mapping torus of the dynamical

2456: system $( {\mathcal S}_A, T)$, which is defined as

2457: \begin{equation} \label{suspensionT}

2458: {\mathcal S}_T := {\mathcal S}_A

2459: \times [0,1] / (x,0)\sim (Tx,1).

2460: \end{equation}

2461: 

2462: \begin{figure}

2463: \begin{center}

2464: \includegraphics[scale=0.65]{MapTorus.eps}

2465: \end{center}

2466: \caption{Mapping Torus \label{FigMapTorus}}

2467: \end{figure}

2468: 

2469: The first cohomology group of $\cS_T$ is

2470: the ``ordered cohomology'' of the dynamical system $T$, in the sense of

2471: \cite{BoHa} \cite{PaTu}.

2472: There is an identification of $H^1({\mathcal S}_T,\Z)$ with the

2473: $K_0$-group of the crossed product ${\rm C}^*$-algebra for the

2474: action of $T$ on ${\mathcal S}_A$,

2475: \begin{equation}

2476: H^1({\mathcal S}_T,\Z) \cong K_0({\rm C}({\mathcal

2477: S}_A) \rtimes_T \Z).  \label{H1K0}

2478: \end{equation}

2479: This can be seen from the Pimsner--Voiculescu exact sequence (\cf

2480: \cite{PV}) for the $K$-theory of a crossed product by $\Z$, which in

2481: this case reduces to

2482: \begin{equation}\label{PV} 0 \to K_1({\rm C}({\mathcal S})\rtimes_T

2483: \Z) \to {\rm C}({\mathcal S},\Z) \stackrel{I-T_*}{\to} {\rm

2484: C}({\mathcal S},\Z) \to K_0({\rm C}({\mathcal S})\rtimes_T \Z) \to

2485: 0, \end{equation}

2486: It can also be seen in terms of the Thom isomorphism

2487: of \cite{Co-Thom}, \cite{Co1}.

2488: 

2489: In fact, as we discussed in Section \ref{road}, one of the

2490: fundamental construction of noncommutative geometry (\cf

2491: \cite{Co1}) is that of {\em homotopy

2492: quotients}. These are commutative spaces, which provide, up to

2493: homotopy, geometric models for the corresponding noncommutative

2494: spaces. The noncommutative spaces themselves, as we are going to show

2495: in our case, appear as quotient spaces of foliations on the homotopy

2496: quotients with contractible leaves.

2497: 

2498: For the noncommutative space ${\mathcal S}_A/\Z$, with $\Z$ acting as

2499: powers of the invertible two-sided shift, the homotopy quotient

2500: is precisely the mapping torus \eqref{suspensionT},

2501: \begin{equation}\label{htpyquot2}

2502: {\mathcal S}_T = {\mathcal S}\times_\Z \R.

2503: \end{equation}

2504: The noncommutative space ${\mathcal S}/\Z$ can be

2505: identified with the quotient space of the natural foliation on

2506: \eqref{htpyquot2} whose generic leaf is contractible (a copy of

2507: $\R$).

2508: 

2509: Another noncommutative space associated to a subshift

2510: of finite type $T$ (which, up to Morita equivalence, correspond to

2511: another choice of the $C^*$-algebra of a Smale space, as mentioned

2512: above) is the Cuntz--Krieger algebra $\cO_A$,

2513: where $A$ is the admissibility matrix of the subshift finite type

2514: (\cf \cite{Cu} \cite{CuKrie}).

2515: 

2516: A partial isometry is a linear operator $S$ satisfying the relation

2517: $S= S S^* S$. The Cuntz--Krieger algebra ${\mathcal O}_A$ is defined

2518: as the universal $C^*$--algebra generated by partial isometries

2519: $S_1, \ldots, S_{N}$, satisfying the relations

2520: \begin{equation} \label{CK1rel} \sum_j S_j S_j^* =I \end{equation}

2521: \begin{equation} \label{CK2rel} S_i^* S_i =\sum_j A_{ij} \, S_j

2522: S_j^*. \end{equation}

2523: 

2524: 

2525: In the case of a Schottky group $\Gamma\subset \PSL_2(\C)$

2526: of genus $g$, the Cuntz--Krieger

2527: algebra ${\mathcal O}_A$ can be described in terms

2528: of the action of the free group $\Gamma$ on its limit set

2529: $\Lambda_\Gamma\subset \P^1(\C)$ (\cf \cite{Rob}, \cite{Spi}), so that we can regard

2530: ${\mathcal O}_A$ as a noncommutative space replacing the classical

2531: quotient $\Lambda_\Gamma / \Gamma$,

2532: \begin{equation} \label{treefree}

2533: {\mathcal O}_A \cong

2534: C(\Lambda_\Gamma) \rtimes \Gamma.

2535: \end{equation}

2536: The quotient space

2537: \begin{equation}\label{htpyquot1}

2538: \Lambda_\Gamma \times_\Gamma \H^3 = \Lambda_\Gamma \times_\Gamma

2539: \underline{E}\Gamma,

2540: \end{equation}

2541: is precisely the homotopy quotient of $\Lambda_\Gamma$ with respect to

2542: the action of $\Gamma$, with $\underline{E}\Gamma =\H^3$ and the

2543: classifying space $\underline{B}\Gamma = \H^3/\Gamma$. Here

2544: $\H^3/\Gamma$ is a hyperbolic 3-manifold of infinite volume, which

2545: is topologically a handlebody of genus $g$.

2546: In this case also we find

2547: that the noncommutative space $\Lambda_\Gamma /\Gamma$ is the quotient

2548: space of a foliation on the homotopy quotient \eqref{htpyquot1} with

2549: contractible leaves $\H^3$.

2550: 

2551: \bigskip

2552: 

2553: 

2554: \section{Noncommutative spaces from string theory}\label{string}

2555: 

2556: The main aspects of string and D-brane theory that

2557: involve noncommutative geometry are the bound states of

2558: configurations of parallel D-branes \cite{Wit1}, the matrix models

2559: for M-theory \cite{BFSS} and the strong coupling limit of

2560: string theory (\cf \eg \cite{Ard1}, \cite{Ard2}). It also plays an

2561: important role in the M-theory compactifications \cite{CDS}.

2562: We shall not discuss all these aspects in detail here. Since the

2563: focus of this review is on examples we   only mention a

2564: couple of examples of noncommutative spaces arising from

2565: string and D-brane theory.

2566: 

2567: \medskip

2568: 

2569: The noncommutative tori and the

2570: components of the Yang-Mills connections appear in the

2571: classification of the BPS states in M-theory \cite{CDS}.

2572: 

2573: Recall first that Yang--Mills theory on noncommutative tori

2574: can be formulated (\cf \cite{CoRi}) using suitable notions of

2575: connections and curvature for noncommutative spaces. In fact,

2576: the analogs of connection

2577: and curvature of vector bundles are straightforward to obtain

2578: (\cite{C3}): a connection is just given by the associated

2579: covariant differentiation $\nabla$ on the space of smooth

2580: sections. Thus here it is given by a pair of linear operators,

2581: \begin{equation}

2582: \nabla_j : \cS (\R) \rightarrow \cS (\R)

2583: \label{nabla}

2584: \end{equation}

2585: such that

2586: \begin{equation}

2587: \nabla_j (\xi b) = (\nabla_j \xi)b + \xi \delta_j (b) \quad

2588: \forall \xi \in \cS \ , b \in \cA_\theta \, .

2589: \label{connection}

2590: \end{equation}

2591: One checks that, as in the usual case, the trace of the curvature

2592: $$\Omega = \nabla_1 \nabla_2 - \nabla_2 \nabla_1, $$

2593: is independent of the choice of the connection.

2594: 

2595: We can make the following choice for the connection:

2596: \begin{equation}

2597: (\nabla_1 \xi) (s) = - \frac{2\pi i s}{\theta} \, \xi (s) \

2598: \,\,\,\,\,\, (\nabla_2 \xi)(s) = \xi' (s) \, . \label{connection2}

2599: \end{equation}

2600: Notice that, up to the correct powers of

2601: $2\pi i$, the total curvature of $\cS$ is an {\em integer}.

2602: In fact, the curvature $\Omega$ is constant, equal to $\frac{1}{\theta}$,

2603: so that the irrational number $\theta$ disappears in the total

2604: curvature, $\theta \times \frac{1}{\theta}$.

2605: This integrality phenomenon, where the pairing of dimension and

2606: curvature (both of which are non--integral) yields an integer:

2607: $$ \dim \times \Omega \sim \theta \times \frac{1}{\theta} =

2608: \text{integer}, $$ is the basis for the development of a theory of

2609: characteristic classes for non--commutative spaces. In the general

2610: case, this requires the development of more sophisticated tools,

2611: since analogs of the derivations $\delta_i$ used in the

2612: case of the noncommutative tori are not there in general.

2613: The general theory is obtained through cyclic

2614: homology, as developed in \cite{ConnesCH}.

2615: 

2616: \smallskip

2617: 

2618: Consider then the modules ${\mathcal H}^\theta_{p,q}$ described in

2619: Section \ref{nctori}. It is possible to define an $\cA_\theta$ valued

2620: inner product $\langle \cdot, \cdot \rangle_{\cA}$ on ${\mathcal

2621: H}^\theta_{p,q}$, as in \cite{Rieffel}, which is used to show that

2622: ${\mathcal H}^\theta_{p,q}$ is a projective module.

2623: Connections are required to be compatible with the metric,

2624: \begin{equation}\label{metr-conn}

2625:  \delta_j \langle\xi,\eta \rangle_{\cA} =\langle \nabla_j \xi,\eta

2626: \rangle_{\cA} +

2627: \langle \xi, \nabla_j \eta \rangle_{\cA}.

2628: \end{equation}

2629: It is proved in \cite{C3} that such connections always exist. The

2630: curvature $\Omega$ has values in $\cE=\End_{\cA}({\mathcal H})$.

2631: An $\cE$--valued inner product on ${\mathcal H}$ is given by

2632: $$ \langle \xi,\eta \rangle_{\cE} \zeta = \xi \langle \eta, \zeta

2633: \rangle_{\cA}, $$ and a canonical faithful trace $\tau_E$ is

2634: defined as

2635: $$ \tau_E(\langle \xi,\eta \rangle_{\cE}) = \tau(\langle \eta,\xi

2636: \rangle_{\cA}), $$ where $\tau$ is the trace on the algebra

2637: $\cA_\theta$, given by \eqref{trace-theta}.

2638: 

2639: The Yang--Mills action is defined (\cf \cite{CoRi}) as

2640: \begin{equation} \label{YM-action}

2641: \tau(\langle \Omega, \Omega\rangle_{\cE}).

2642: \end{equation}

2643: One seeks for minima of the Yang--Mills action among metric

2644: compatible connections \eqref{metr-conn}. The main result is

2645: that this recovers the classical moduli spaces

2646: of Yang--Mills connections on the ordinary torus (\cite{CoRi}):

2647: 

2648: \begin{thm} \label{NCT-YM}

2649: For a choice of a pair of integers $(p,q)$ with $p + q\theta \geq

2650: 0$, the moduli space of Yang--Mills connections on the

2651: $\cA_\theta$ module ${\mathcal H}^\theta_{pq}$ is a classical

2652: space given by the symmetric product

2653: $$ s^N(T^2)= (T^2)^N / \Sigma_N, $$

2654: where $\Sigma_N$ is the group of permutations in $N$--elements,

2655: for $N=\gcd (p,q)$.

2656: \end{thm}

2657: 

2658: \smallskip

2659: 

2660: In the matrix formulation of M-theory the basic equations to

2661: obtain periodicity of two of the basic coordinates $X_i$ turn out

2662: to be the

2663: \begin{equation}

2664: U_i X_j U_i^{-1} =X_j + a \delta_i^j , \ \ \  i= 1,2,  \label{Mtheory}

2665: \end{equation}

2666: where the $U_i$ are unitary gauge transformations.

2667: The multiplicative commutator $U_1 U_2 U_1^{-1}U_2^{-1}$ is then

2668: central and in the irreducible case its scalar value $ \lambda

2669: =\exp 2\pi i \theta $ brings in the algebra of coordinates on the

2670: noncommutative torus. The $X_j$ are then the components of the

2671: Yang-Mills connections. The same

2672: picture emerged from the other information one has about M-theory

2673: concerning its relation with 11 dimensional supergravity and that

2674: string theory dualities can then be interpreted using Morita

2675: equivalence, relating the values of $\theta$ on an orbit of $SL_2

2676: (\Z)$.

2677: 

2678: \smallskip

2679: 

2680: Nekrasov and Schwarz \cite{NS} showed that Yang-Mills gauge

2681: theory on noncommutative $\R^4$ gives a conceptual understanding of the

2682: nonzero B-field desingularization of the moduli space of

2683: instantons obtained by perturbing the ADHM equations.

2684: In \cite{Witten}, Seiberg and Witten exhibited the unexpected

2685: relation between the standard gauge theory and the noncommutative

2686: one, and clarified the limit in which the entire string dynamics

2687: is described by a gauge theory on a noncommutative space.

2688: Techniques from noncommutative differential and Riemannian geometry,

2689: in the sense discussed in Section \ref{road} were applied to

2690: string theory, for instance in \cite{Ard1}. The role of noncommutative

2691: geometry in the context of $T$-duality was considered in very

2692: interesting recent work of Mathai and collaborators, \cite{BMa1},

2693: \cite{BMa2}, \cite{RoMa}.

2694: 

2695: \medskip

2696: 

2697: Recently, in the context of the holographic description of

2698: type IIB string theory on the plane-wave background,

2699: Shahin M.M.~Sheikh-Jabbari obtained (\cf \cite{Shahin}) an interesting

2700: class of noncommutative spaces from the quantization of Nambu $d$-brackets.

2701: The classical Nambu brackets

2702: \begin{equation}\label{NambuC}

2703: \{ f_1,\ldots, f_k \}= \epsilon^{i_1\cdots i_k} \frac{\partial

2704: f_1}{\partial x^{i_1}}\cdots \frac{\partial

2705: f_k}{\partial x^{i_k}}

2706: \end{equation}

2707: of $k$ real valued functions of variables $(x^1,\ldots, x^k)$

2708: is quantized in the even case to the expression in $2k$

2709: operators

2710: \begin{equation}\label{NambuQev}

2711: \frac{1}{i^k} [ F_1,\ldots, F_{2k}]=\frac{1}{i^k (2k)!}

2712: \epsilon^{i_1\cdots i_{2k}} F_{i_1}\cdots F_{i_{2k}}.

2713: \end{equation}

2714: This generalizes the Poisson bracket quantization

2715: $\{f_1,f_2\} \mapsto \frac{-i}{\hbar} [F_1,F_2]$.

2716: The odd case is more subtle and it involves an additional operator

2717: related to the chirality operator $\gamma_5$. One sets

2718: \begin{equation}\label{NambuQodd}

2719: \frac{1}{i^k} [ F_1,\ldots, F_{2k-1}, \gamma]=\frac{1}{i^k (2k)!}

2720: \epsilon^{i_1\cdots i_{2k}} F_{i_1}\cdots F_{i_{2k-1}} \gamma,

2721: \end{equation}

2722: where $\gamma$ is the chirality operator in $2k$-dimensions.

2723: For example, for $k=2$ one gets

2724: $$ \begin{array}{rl}

2725: [F_1,F_1,F_3,\gamma]=& \frac{1}{24}([F_1,F_2][F_3,\gamma]

2726: +[F_3,\gamma][F_1,F_2] \\[2mm] & -([F_1,F_3][F_2,\gamma]+[F_2,\gamma][F_1,F_3])

2727: \\[2mm] & +[F_2,F_3][F_1,\gamma] +[F_1,\gamma][F_2,F_3]).

2728: \end{array} $$

2729: If one describes the ordinary $d$-dimensional sphere of radius $R$

2730: by the equation

2731: \begin{equation}\label{Sd1}

2732: \sum_{i=1}^{d+1} (x^i)^2=R^2,

2733: \end{equation}

2734: the coordinates satisfy

2735: \begin{equation}\label{Sd2}

2736: \{ x^{i_1},\ldots,x^{i_d}\}=R^{d-1} \epsilon^{i_1\cdots i_{d+1}}

2737: x^{i_{d+1}}.

2738: \end{equation}

2739: The equation \eqref{Sd1} and \eqref{Sd2} are then replaced by their

2740: quantized version, using the quantization of the Nambu bracket and

2741: the introduction of a quantization parameter $\ell$. This defines

2742: algebras generated by unitaries $X^i$ subject to the relations given

2743: by the quantization of \eqref{Sd1} and \eqref{Sd2}. Matrix

2744: representations of these algebra correspond to certain fuzzy spheres.

2745: It would be interesting to study the general structure of these

2746: noncommutative spaces from the point of view of the various steps

2747: introduced in Section \ref{road}, \cf also the discussion in Section

2748: \ref{deform}.

2749: 

2750: 

2751: \bigskip

2752: 

2753: 

2754: \section{Groupoids and the index theorem }

2755: \label{cotangent}

2756: 

2757: Since the construction of the $C^*$-algebra of foliations based on

2758: the holonomy groupoid (section \ref{foliations}), groupoids have

2759: played a major role in noncommutative geometry. In fact the original

2760: construction of matrix mechanics by Heisenberg (section

2761: \ref{heisenberg}) is exactly that of the convolution algebra of the

2762: groupoid of transitions imposed by experimental results. The

2763: convolution algebra of groupoids can be defined in the context of

2764: von-Neumann algebras and of $C^*$-algebras (\cf \cite{Co78} and

2765: \cite{Ren}). It is particularly simple and canonical in the context

2766: of smooth groupoids (\cf \cite{Co94} section II.5). One virtue of

2767: the general construction is that it provides a geometric mental

2768: picture of complicated analytical constructions. The prototype

2769: example is given by the tangent groupoid of a manifold (\cf

2770: \cite{Co94} section II.5). It is obtained by blowing up the diagonal

2771: in the square $V\times V$ of the manifold and as a set is given by

2772: $$

2773: G_V=\,V\times V\times ]0,1]\cup TV

2774: $$

2775: where $TV$ is the (total space of the) tangent bundle of $V$. A

2776: tangent vector $X\in T_x(V)$ appears as the limit of nearby triples

2777: $(x_1, x_2, \varepsilon)$ provided in any chart the ratios

2778: $(x_1-x_2)/\varepsilon$ converge to $X$. When $\varepsilon \to 0$

2779: the Heisenberg (matrix) law of composition :

2780: $$

2781: (x_1, x_2, \varepsilon)\circ (x_2, x_3, \varepsilon)=(x_1, x_3,

2782: \varepsilon)

2783: $$

2784: converges to the addition of tangent vectors, so that $G_V$ becomes

2785: a smooth groupoid. The functoriality of the construction $G\to

2786: C^*(G)$ of the convolution algebra for smooth groupoids $G$ is then

2787: enough to define  the Atiyah-Singer analytic index of

2788: pseudo-differential operators. It is simply given by the map in

2789: K-theory for the exact sequence of $C^*$-algebras associated to the

2790: geometric sequence

2791: $$

2792: V\times V\times ]0,1] \to G_V \supset TV

2793: $$

2794: where $TV$ is viewed as a closed subgroupoid of $G_V$. The

2795: corresponding exact sequence of $C^*$-algebras can be written as

2796: $$

2797: 0\to C_0(]0,1])\times \mathcal K\to C^*(G_V)\to C_0(T^*V)\to 0

2798: $$

2799: and is a geometric form of the extension of pseudo-differential

2800: operators. By construction the algebra $C_0(]0,1])$ is {\em

2801: contractible} and the same holds for the tensor product

2802: $C_0(]0,1])\times \mathcal K$ by the algebra $\mathcal K$ of compact

2803: operators. This shows that the restriction map $C^*(G_V)\to

2804: C_0(T^*V)$ is an isomorphism in K-theory :

2805: \begin{equation}\label{tstariso}

2806: K_0(C_0(T^*V))\sim K_0(C^*(G_V))

2807: \end{equation}

2808:  Since the K-theory of

2809: $\mathcal K$ is  $\Z$ for $K_0$, one gets the analytic index by the

2810: evaluation map

2811: $$

2812: C^*(G_V)\to \mathcal K\,,\quad K_0(C^*(G_V))\to K_0(\mathcal K)=\Z

2813: $$

2814: composed with the isomorphism \eqref{tstariso}. Using the Thom

2815: isomorphism yields a geometric proof (\cf \cite{Co94}) of the

2816: Atiyah-Singer index theorem, where all the analysis has been taken

2817: care of once and for all by the functor $G\to C^*(G)$.

2818: 

2819: This paradigm for a geometric set-up of the index theorem has been

2820: successfully extended to manifolds with singularities (\cf

2821: \cite{Month} \cite{Month1} and references there) and to manifolds

2822: with boundary \cite{ANS}.

2823: 

2824: \bigskip

2825: 

2826: 

2827: 

2828: 

2829: \section{Riemannian manifolds, conical singularities}\label{singconic}

2830: 

2831: A main property of the homotopy type of a compact

2832: oriented manifold is that it satisfies Poincar\'e duality not

2833: just in ordinary homology but also in $K$-homology.

2834: In fact, while

2835: Poincar\'e duality in ordinary homology is not sufficient to describe

2836: homotopy type of manifolds (\cf \cite{Mi-S}), Sullivan proved (\cf

2837: \cite{Sull}) that for simply connected PL

2838: manifolds of dimension at least $5$, ignoring 2-torsion,

2839: the same property in $KO$-homology suffices and the

2840: Chern character of the $KO$-homology

2841: fundamental class carries all the rational information on the

2842: Pontrjagin classes.

2843: 

2844: For an ordinary manifold the choice of the fundamental cycle in

2845: $K$-homology is a refinement of the choice of orientation of the

2846: manifold and, in its simplest form, it is a choice of

2847: Spin-structure. Of course the role of a spin structure is to allow

2848: for the construction of the corresponding Dirac operator which

2849: gives a corresponding Fredholm representation of the algebra of

2850: smooth functions. The choice of a square root involved in the

2851: Dirac operator $D$ corresponds to a choice of $K$-orientation.

2852: 

2853: $K$-homology admits a

2854: fairly simple definition in terms of Hilbert spaces and Fredholm

2855: representations of algebras. In fact,

2856: we have the following notion of Fredholm module (\cf \cite{Co94}):

2857: 

2858: \begin{defn}\label{Fred-mod-act}

2859: Let $\cA$ be an algebra, an {\em odd} Fredholm module over $\cA$

2860: is given by:

2861: \begin{enumerate}

2862: \item a representation of $\cA$ in a Hilbert space $\cH$.

2863: \item an operator $F = F^*$, $F^2 = 1$, on $\cH$ such that

2864: $$

2865: [F , a] \ \hbox{is a compact operator for any} \ a \in \cA \, .

2866: $$

2867: \end{enumerate}

2868: An {\it even} Fredholm module is given by an odd Fredholm module

2869: $(\cH , F)$ together with a $\Z / 2$ grading

2870: $\gamma$, $\gamma = \gamma^*$,

2871: $\gamma^2 = 1$ of the Hilbert space $\cH$ satisfying:

2872: \begin{enumerate}

2873: \item $\gamma a = a \gamma$, for all $a \in \cA$

2874: \item $\gamma F = -F \gamma$.

2875: \end{enumerate}

2876: \end{defn}

2877: 

2878: This definition is derived from

2879: Atiyah's definition \cite{AT} of abstract elliptic operators,

2880: and agrees with Kasparov's definition \cite{Kasp} for the cycles

2881: in $K$-homology, $KK(A,\C)$, when $A$ is a $C^*$-algebra.

2882: 

2883: The notion of Fredholm module can be illustrated by the following

2884: examples (\cf \cite{Co94}).

2885: 

2886: \begin{ex}\label{Fred-manifold} {\em If $X$ is a manifold, an

2887: elliptic operator on $X$ can be twisted with vector bundles, so as

2888: to give rise to an index map ${\rm Ind}: K_0(C(X)) \to \Z$. If $P$

2889: is an elliptic operator (the symbol is invertible) and a

2890: pseudodifferential operator of order zero, $P: L^2(X,E_+)\to

2891: L^2(X,E_-)$, then there exists a parametrix $Q$ for $P$. This is

2892: also an operator of order zero, and a quasi-inverse for $P$, in

2893: the sense that it is an inverse at the symbol level, namely $PQ-I$

2894: and $QP-I$ are compact operators. Consider then the operator

2895: $$ F=\left( \begin{array}{cc} 0 & Q \\ P & 0  \end{array} \right) $$

2896: on $\cH=L^2(X,E_+)\oplus L^2(X,E_-)$. The algebra $C(X)$ acts on

2897: $\cH$ and $[F,f]$ is a compact operator for all $f\in C(X)$. Since

2898: $F^2-I$ is compact, it is possible to add to $\cH$ a finite

2899: dimensional space to obtain $F^2=I$. Notice that the functions of

2900: $C(X)$ act differently on this modified space. In particular the

2901: function $f\equiv 1$ no longer acts as the identity: one recovers

2902: the index of $P$ this way.}

2903: \end{ex}

2904: 

2905: \begin{ex}\label{Fred-freeG} {\em Let $\Gamma=\Z * \Z$ a free

2906: group, and let $\cA =\C \Gamma$. Let $\cT$ be the tree of $\Gamma$

2907: with $\cT^0$ the set of vertices and $\cT^1$ the set of edges. Fix

2908: an origin $x_0$ in $\cT^0$. For any vertex $v\in \cT^0$ there

2909: exists a unique path connecting it to the origin $x_0$. This

2910: defines a bijection $\phi: \cT^0 \backslash \{ x_0 \} \to \cT^1$

2911: that assigns $v\mapsto \phi(v)$ with $\phi(v)$ this unique edge.

2912: Let $U_\phi$ be the unitary operator implementing $\phi$, and

2913: consider the operator

2914: $$ F = \left( \begin{array}{cc} 0 & U_\phi \\ U_\phi^* & 0

2915: \end{array} \right) $$

2916: acting on $\cH = \ell^2(\cT^0) \oplus \ell^2(\cT^1) \oplus \C$. By

2917: construction $\Gamma$ acts naturally on $\cT_j$ which gives a

2918: corresponding action of $\cA$ in $\cH$. The pair $(\cH,F)$ is a

2919: Fredholm module over $\cA$.}

2920: \end{ex}

2921: 

2922: 

2923: \begin{ex}\label{Hilb-transform} {\em On $S^1\simeq

2924: \bP^1(\R)$, consider the algebra of functions ${\rm

2925: C}(\bP^1(\R))$, acting on the Hilbert space $L^2(\R)$, as

2926: multiplication operators $(f\, \xi)(s) = f(s) \xi(s)$. Let $F$ be

2927: the Hilbert transform

2928: $$ (F\, \xi)(s)=\frac{1}{\pi i} \int \frac{f(t)}{s-t} \, dt. $$

2929: This multiplies by $+1$ the positive Fourier modes and by $-1$ the

2930: negative Fourier modes. A function $f\in {\rm C}(\bP^1(\R))$ has

2931: the property that $[F,f]$ is of finite rank if and only if $f$ is

2932: a rational function $f(s) =P(s)/Q(s)$. This is Kronecker's

2933: characterization of rational functions.}

2934: \end{ex}

2935: 

2936: \medskip

2937: 

2938: Besides the $K$-homology class, specified by a Fredholm module,

2939: one also wants to generalize to the noncommutative setting the

2940: infinitesimal line element $ds$ of a Riemannian manifold. In ordinary

2941: Riemannian geometry one deals rather with the $ds^2$ given by

2942: the usual local expression $g_{\mu \nu} \, dx^{\mu} \, dx^{\nu}$.

2943: However, in order to extend the notion of metric space to the

2944: noncommutative setting it is more natural to deal with $ds$, for

2945: which the ansatz is

2946: \begin{equation}

2947: ds = \times\!\!\!\!-\!\!\!-\!\!\!-\!\!\!\!\!\times \, ,

2948: \label{dsx-x}

2949: \end{equation}

2950: where the right hand side has the meaning usually attributed to it in

2951: physics, namely the Fermion propagator

2952: \begin{equation}

2953: \times\!\!\!\!-\!\!\!-\!\!\!-\!\!\!\!\!\times = D^{-1} ,

2954: \label{Fermprop}

2955: \end{equation}

2956: where $D$ is the Dirac operator. In other words, the presence of a

2957: spin (or spin$^c$) structure makes it possible to extract the square

2958: root of $ds^2$, using the Dirac operator as a

2959: differential square root of a Laplacian.

2960: 

2961: This prescription recovers the usual geodesic distance on a Riemannian

2962: manifold, by the following result (\cf

2963: \cite{Co-fr}).

2964: 

2965: \begin{lem}\label{geod-distance}

2966: On a Riemannian spin manifold the geodesic distance $d(x,y)$

2967: between two points is computed by the formula

2968: \begin{equation}

2969: d(x,y) = \sup  \{ \vert f(x) - f(y) \vert \, ; \ f \in \cA

2970: \, , \ \Vert [D,f] \Vert \leq 1 \} \label{geoddist}

2971: \end{equation}

2972: where $D$ is the Dirac operator, $D=ds^{-1}$, and $\cA$ is the

2973: algebra of smooth functions.

2974: \end{lem}

2975: 

2976: This essentially follows from the fact that the quantity $\Vert

2977: [D,f] \Vert$ can be identified with the Lipschitz norm of the

2978: function $f$,

2979: $$ \Vert [D,f] \Vert = {\rm ess} \sup_{x\in M} \| (\nabla f)_x \|

2980: = \sup_{x\neq y}\frac{ | f(x)-f(y) | }{d(x,y)}. $$

2981: 

2982: Notice that, if $ds$ has the dimension of a length $L$, then $D$

2983: has dimension $L^{-1}$ and the above expression for $d(x,y)$ also

2984: has the dimension of a length. On a Riemannian spin manifold $X$,

2985: the condition $\| [D,f] \|\leq 1$, for $D$ the Dirac operator, is

2986: equivalent to the condition that $f$ is a Lipschitz function with

2987: Lipschitz constant $c\leq 1$.

2988: 

2989: The advantage of the definition \eqref{dsx-x}, \eqref{Fermprop} of the

2990: line element is that it is of a {\em spectral} and operator theoretic

2991: nature, hence it extends to the noncommutative setting.

2992: 

2993: The structure that combines the $K$-homology fundamental cycle with

2994: the spectral definition of the line element $ds$ is the notion of {\em

2995: spectral triple} $(\cA,\cH,D)$ (\cf \cite{ConnesS3}, \cite{CoMo}).

2996: 

2997: \begin{defn}\label{def-triple}

2998: A (compact) noncommutative geometry is a triple

2999: \begin{equation}

3000: (\cA ,\cH ,D)

3001: \label{triple}

3002: \end{equation}

3003: where $\cA$ is a unital algebra

3004: represented concretely as an

3005: algebra of bounded operators on the Hilbert space $\cH$. The unbounded

3006: operator $D$ is the inverse of the line  element

3007: \begin{equation}

3008: ds=1/D.\label{dsD}

3009: \end{equation}

3010: Such a triple $(\cA ,\cH ,D)$ is requires to satisfy the properties:

3011: \begin{enumerate}

3012: \item $[D,a]$ is bounded for any $a \in A^\infty$, a dense subalgebra of

3013: the $C^*$-algebra $A=\bar\cA$.

3014: \item $D=D^*$ and $(D+\lambda)^{-1}$ is a compact operator, for

3015: all $\lambda \not\in \R$.

3016: \end{enumerate}

3017: We say that a spectral triple $(\cA ,\cH ,D)$ is even if the

3018: Hilbert space $\cH$ has a $\Z/2$-grading by an operator $\gamma$

3019: satisfying

3020: \begin{equation}\label{gammaZ2}

3021: \gamma =\gamma^*, \ \ \ \gamma^2 =1, \ \ \ \gamma\, D =- D \,

3022: \gamma, \ \ \ \gamma \, a = a\, \gamma \ \ \forall a \in A.

3023: \end{equation}

3024: \end{defn}

3025: 

3026: This definition is entirely spectral. The elements of the algebra (in

3027: general noncommutative) are operators and the line element is also an

3028: operator. The polar decomposition $D=|D|F$ recovers the Fredholm

3029: module $F$ defining the fundamental class in $K$-homology.

3030: The formula for the geodesic distance extends to this

3031: context as follows.

3032: 

3033: \begin{defn}\label{distanceNC}

3034: Let $\varphi_i: A \to \C$, for $i=1,2$, be states on $A$, \ie

3035: normalized positive linear functionals on $A$ with $\varphi_i (1) =

3036: 1$ and $\varphi_i (a^* a) \geq 0$ for all $a \in A$. Then the

3037: distance between them is given by the formula

3038: \begin{equation}

3039: d(\varphi_1 ,\varphi_2) = \sup  \{ \vert \varphi_1 (a) -

3040: \varphi_2 (a) \vert \ ;

3041: \ a\in A \ , \ \Vert [D,a]\Vert \leq 1 \} \, .

3042: \label{geodNC}

3043: \end{equation}

3044: \end{defn}

3045: 

3046: \smallskip

3047: 

3048: A spectral triple $(\cA,\cH,D)$ is of {\em metric dimension} $p$,

3049: or $p$-summable, if $|D|^{-1}$ is an infinitesimal of order $1/p$

3050: (\ie $|D|^{-p}$ is an infinitesimal of order one). Here $p<

3051: \infty$ is a positive real number. A spectral triple $(\cA,\cH,D)$ is

3052: $\theta$-summable if $\Tr(e^{-t D^2}) <\infty$ for all $t>0$.

3053: The latter case corresponds to an infinite dimensional geometry.

3054: 

3055: Spectral triples also provide a more refined notion of dimension

3056: besides the metric dimension (summability). It is given by the

3057: {\em dimension spectrum}, which is not a number but a subset of

3058: the complex plane.

3059: 

3060: More precisely, let  $(\cA,\cH,D)$ be

3061: a spectral triple satisfying

3062: the regularity hypothesis

3063: \begin{equation}

3064: a \ \hbox{and } \ [D,a] \ \in \ \cap_k {\rm Dom}( \delta^k ) , \ \forall \,

3065: a\in A^\infty ,

3066: \label{regS3}

3067: \end{equation}

3068: where $\delta$ is the derivation $\delta(T) = [\vert D \vert,T]$, for any

3069: operator $T$.

3070: Let $\cB$ denote the algebra generated by  $\delta^k (a)$ and $\delta^k

3071: ([D,a])$.

3072: The dimension spectrum of the triple $(\cA,\cH,D)$

3073: is the subset $\Sigma \subset \C$ consisting of all the singularities of

3074: the analytic functions $\zeta_b (z)$ obtained by continuation of

3075: \begin{equation}

3076: \zeta_b (z) = \Tr  (b \vert D \vert^{-z}), \ \ \ \Re(z)

3077: > p \ , \ \ \ b\in \cB \, . \label{zeta-b}

3078: \end{equation}

3079: 

3080: 

3081: \begin{ex} {\em

3082: Let $M$ be a smooth compact Riemannian spin manifold, and

3083: $(\cA,\cH,D)$ is the corresponding spectral triple given by the

3084: algebra of smooth functions, the space of spinors, and the Dirac

3085: operator. Then the metric dimension agrees with the

3086: usual dimension $n$ of $M$. The dimension spectrum of $M$ is the

3087: set $\{ 0,1,\ldots ,n\}$, where $n=\dim M$, and it is simple.

3088: (Multiplicities appear for singular manifolds.)}

3089: \end{ex}

3090: 

3091: It is interesting in the case of an ordinary Riemannian manifold $M$

3092: to see the meaning of the points in the dimension spectrum that are

3093: smaller than $n=\dim M$. These are dimensions in which the space

3094: ``manifests itself nontrivially'' with some interesting geometry.

3095: 

3096: For instance, at the point $n=\dim M$ of the dimension spectrum one

3097: can recover the

3098: volume form of the Riemannian metric by the equality (valid up to

3099: a normalization constant \cf \cite{Co94})

3100: \begin{equation}

3101: {\int \!\!\!\!\!\! -}f \,\vert ds\vert^n =  \, \int_{M_n} f \,

3102: \sqrt{g} \ d^n x \, ,                       \label{nvolume}

3103: \end{equation}

3104: where the integral $\cutint T$ is given (\cf \cite{Co94}) by the

3105: Dixmier trace (\cf \cite{Dix}) generalizing the Wodzicki residue of

3106: pseudodifferential operators (\cf \cite{Wo}).

3107: 

3108: One can also consider integration $\cutint ds^k$ in any other

3109: dimension in the

3110: dimension spectrum, with $ds=D^{-1}$ the line element.

3111: In the case of a Riemannian manifold one finds

3112: other important curvature expressions. For instance, if $M$ is a

3113: manifold of dimension $\dim M=4$, when one considers integration in

3114: dimension $2$ one finds the Einstein--Hilbert action. In fact,

3115: a direct computation

3116: yields the following result (\cf \cite{Kast} \cite{K-W}):

3117: 

3118: \begin{prop}

3119: Let $dv=\sqrt{g} \ d^4 x$ denote the volume form, $ds=D^{-1}$  the

3120: length element, \ie the inverse of the Dirac operator, and $r$ the

3121: scalar curvature. We obtain:

3122: \begin{equation}

3123: {\int \!\!\!\!\!\! -} \, ds^2 = \frac{-1}{48\pi^2} \, \int_{M_4} r

3124: \, \, dv   \, .                      \label{EHaction}

3125: \end{equation}

3126: \end{prop}

3127: 

3128: In general, one obtains the scalar curvature of an $n$-dimensional manifold

3129: from the integral $\cutint ds^{n-2}$.

3130: 

3131: \medskip

3132: 

3133: Many interesting examples of spectral triples just satisfy the

3134: conditions stated in Definition \ref{def-triple}. However, there are

3135: significant case where more refined properties of manifolds carry over

3136: to the noncommutative case, such as the presence of a real structure

3137: (which makes it possible to distinguish between $K$-homology and

3138: $KO$-homology) and the ``order one condition'' for the Dirac

3139: operator. These properties are described as follows (\cf \cite{Co3}

3140: and \cite{Coo2}).

3141: 

3142: \begin{defn}\label{realstr}

3143: A real structure on an $n$-dimensional spectral triple

3144: $(\cA,\cH,D)$ is an antilinear isometry $J: \cH \to \cH$, with the

3145: property that

3146: \begin{equation}\label{per8}

3147: J^2 = \varepsilon, \ \ \ \ JD = \varepsilon' DJ, \ \ \text{and} \ \

3148: J\gamma = \varepsilon'' \gamma

3149: J \, \text{(even case)}.

3150: \end{equation}

3151: The numbers $\varepsilon ,\varepsilon' ,\varepsilon'' \in \{ -1,1\}$

3152: are a function of $n \mod 8$ given by

3153: 

3154: \begin{center}

3155: \begin{tabular}

3156: {|c| r r r r r r r r|} \hline {\bf n }&0 &1 &2 &3 &4 &5 &6 &7 \\

3157: \hline \hline

3158: $\varepsilon$  &1 & 1&-1&-1&-1&-1& 1&1 \\

3159: $\varepsilon'$ &1 &-1&1 &1 &1 &-1& 1&1 \\

3160: $\varepsilon''$&1 &{}&-1&{}&1 &{}&-1&{} \\  \hline

3161: \end{tabular}

3162: \end{center}

3163: 

3164: 

3165: Moreover, the action of $\cA$ satisfies the commutation rule

3166: \begin{equation}\label{comm-rule}

3167: [a,b^0] = 0 \quad \forall \, a,b \in \cA,

3168: \end{equation}

3169: where

3170: \begin{equation}\label{b0}

3171: b^0 = J b^* J^{-1} \qquad \forall b \in \cA,

3172: \end{equation}

3173: and the operator $D$ satisfies

3174: \begin{equation}\label{order1}

3175: [[D,a],b^0] = 0 \qquad \forall \, a,b \in \cA \, .

3176: \end{equation}

3177: \end{defn}

3178: 

3179: The anti-linear isometry $J$ is given, in ordinary Riemannian

3180: geometry, by the charge conjugation operator acting on spinors. In

3181: the noncommutative case, this is replaced by the Tomita

3182: antilinear conjugation operator (\cf \cite{Tak}).

3183: 

3184: \smallskip

3185: 

3186: In \cite{Co3} and \cite{FGV} Theorem 11.2, necessary and sufficient

3187: conditions are given that a spectral triple $(\cA,\cH,D)$ (with real

3188: structure $J$) should fulfill in order to come from an ordinary

3189: compact Riemannian spin manifold:

3190: 

3191: \begin{enumerate}

3192: \item $ds=D^{-1}$ is an infinitesimal of order $1/n$.

3193: \item There is a real structure in the sense of Definition

3194: \ref{realstr}.

3195: \item The commutation relation \eqref{order1} holds (this is

3196: $[[D,a],b]=0$, for all $a,b\in \cA$, when $\cA$ is commutative).

3197: \item The regularity hypothesis of \eqref{regS3} holds:

3198: $a$ and $[D,a]$ are in $\cap_k {\rm Dom}( \delta^k)$ for all $a\in \cA_\infty$.

3199: \item There exists a Hochschild cycle $c\in Z_n(\cA_\infty,\cA_\infty)$ such

3200: that its representation $\pi(c)$ on $\cH$ induced by

3201: $$ \pi(a^0\otimes\cdots\otimes a^n)=a^0 [D,a^1]\cdots [D,a^n] $$

3202: satisfies $\pi(c)=\gamma$, for $\gamma$ as in \eqref{gammaZ2}, in

3203: the even case, and $\pi(c)=1$ in the odd case.

3204: \item The space $\cH^\infty= \cap_k {\rm Dom}( D^k)$ is a finite

3205: projective $\cA$-module, endowed with a $\cA$-valued inner product

3206: $\langle \xi, \eta \rangle_{\cA}$ defined by

3207: $$ \langle a \xi, \eta \rangle = \cutint a\, \langle \xi, \eta

3208: \rangle_{\cA}\, ds^n. $$

3209: \item The intersection form

3210: \begin{equation}

3211: K_* (\cA) \times K_* (\cA) \to \Z

3212: \label{intform}

3213: \end{equation}

3214: obtained from the Fredholm index of $D$ with

3215: coefficients in $K_* (\cA \otimes \cA^0)$

3216: is invertible.

3217: \end{enumerate}

3218: 

3219: When $\cA$ is commutative, the above conditions characterize a

3220: smooth Riemannian manifold $M$, with $\cA_\infty=\cC^\infty(M)$ (we

3221: refer to \cite{FGV} for the precise statement). However, the

3222: conditions can be stated without any commutativity assumption on

3223: $\cA$. They are satisfied, for instance, by the isospectral

3224: deformations of \cite{CLa}, which we discuss in Section

3225: \ref{isodef}. Another very significant noncommutative example is the

3226: standard model of elementary particles (\cf \cite{Co3}), which we

3227: discuss in Section \ref{stmod}.

3228: 

3229: 

3230: \bigskip

3231: 

3232: Another example of spectral triple associated to a classical

3233: space, which is not classically a smooth manifold, is the case of

3234: manifolds with singularities. In particular, one can consider the

3235: case of an isolated conical singularity. This case was studied by

3236: Lescure \cite{Lescure}.

3237: 

3238: \smallskip

3239: 

3240: 

3241: Let $X$ be a manifold with an isolated conical singularity. The

3242: cone point $c\in X$ has the property that there is a neighborhood

3243: $U$ of $c$ in $X$, such that $U\smallsetminus \{ c \}$ is of the

3244: form $(0, 1] \times N$, with $N$ a smooth compact manifold, and

3245: metric $g|_U = dr^2 + r^2 g_N$, where $g_N$ is the metric on $N$.

3246: 

3247: \smallskip

3248: 

3249: A natural class of differential operators on manifolds with

3250: isolated conical singularities is given by the elliptic operators

3251: of Fuchs type, acting on sections of a bundle $E$. These are

3252: operators whose restriction over $(0, 1] \times N$ takes the form

3253: $$ r^{-\nu} \sum_{k=0}^d a_k(r) (-r \partial_r)^k, $$

3254: for $\nu \in \R$ and $a_k \in C^\infty([0,1], {\rm Diff}^{d-k}(N,

3255: E|_N))$, which are elliptic with symbol $\sigma_M(D)=\sum_{k=0}^d

3256: a_k(0) z^k$ that is an elliptic family parameterized by $Im(z)$. In

3257: particular, operators of Dirac type are elliptic of Fuchs type. For

3258: such an operator $D$, which is of first order and symmetric, results

3259: of Chou \cite{chou}, Br\"uning, Seeley \cite{BruSee} and Lesch

3260: \cite{Lesch} show that its self-adjoint extension has discrete

3261: spectrum, with $(n+1)$-summable resolvent, for $\dim X= n$.

3262: 

3263: \smallskip

3264: 

3265: The algebra that is used in the construction of the spectral

3266: triple is $\cA= C^\infty_c(X) \oplus \C$, the algebra of functions

3267: that are smooth on $X\smallsetminus \{ c \}$ and constant near the

3268: singularity. The Hilbert space on which $D$ acts is chosen from a

3269: family of weighted Sobolev spaces. Roughly, one defines weighted

3270: Sobolev spaces that look like the standard Sobolev space on the

3271: smooth part and on the cone are defined locally by norms

3272: $$

3273: \| f \|_{s,\gamma}^2 = \int_{\R^*_+ \times \R^{m-1}} \left(1 + (\log

3274: t)^2 +\xi^2\right)^s \left| \widehat{(r^{-\gamma +1/2} f)}

3275: (t,\xi)\right|^2 \frac{dt}{t} d\xi,

3276: $$

3277: where $\hat f$ denotes Fourier transform on the group $\R_+^*

3278: \times \R^{m-1}$.

3279: 

3280: \smallskip

3281: 

3282: Then one obtains the following result (\cf \cite{Lescure})

3283: 

3284: \begin{thm}

3285: The data $(\cA, \cH, D)$ given above define a spectral triple. In

3286: particular, the zeta functions $\Tr(a |D|^{-z})$ admit analytic

3287: continuation to $\C \smallsetminus \Sigma$, where the dimension

3288: spectrum is of the form

3289: $$ \Sigma=\{ \dim X - k, k\in \N \}, $$

3290: with multiplicities $\leq 2$.

3291: \end{thm}

3292: 

3293: \smallskip

3294: 

3295: The analysis of the zeta functions uses the heat kernel

3296: $$ \Tr (| D |^{-z}) = \frac{1}{\Gamma(z/2)} \int_0^\infty t^{z/2

3297: -1} \Tr (e^{-t D^2}) dt, $$ for which one can rely on the results of

3298: \cite{chou} and \cite{Lesch}. The case of $\Tr(a |D|^{-z})$ of the

3299: form $\Tr(Q |D|^{-z})$ with $Q\in \Psi^\ell_c(E)$, is treated by

3300: splitting $Q|D|^{-z}$ as a sum of a contribution from the smooth

3301: part and one from the singularity.

3302: 

3303: \smallskip

3304: 

3305: The Chern character for this spectral triple gives a map

3306: $$ Ch: K_*(X) \to H_*(X,\C), $$

3307: where we have $K^*(X)\cong K_*(\cA)$ and $H_*(X,\C) \cong

3308: PHC^*(\cA)$, the periodic cyclic homology of the algebra $\cA$.

3309: 

3310: The cocycles $\varphi_n$ in the $(b,B)$-bicomplex for the algebra

3311: $\cA$ have also been computed explicitly and are of the form

3312: $$ \varphi_n (a_0,\ldots,a_n) =\nu_n \int_X a_0 da_1 \wedge \cdots

3313: \wedge da_n \wedge \hat A(X) \wedge Ch(E),

3314: $$

3315: for $n \geq 1$, while for $n=0$, $\lambda\in \C$,

3316: $\varphi_0(a+\lambda)=\int_X a \hat A(X)\wedge Ch(E) + \lambda

3317: \Ind(D_+)$.

3318: 

3319: \bigskip

3320: 

3321: \section{Cantor sets and fractals}\label{fractals}

3322: 

3323: An important class of $C^*$-algebras are those obtained as

3324: direct limits of a sequence of finite dimensional

3325: subalgebras and embeddings. These are called {\em approximately finite

3326: dimensional}, or simply AF-algebras.

3327: 

3328: An AF algebra $\cA$ is determined by a diagram of finite

3329: dimensional algebras and inclusions, its  Bratteli diagram

3330: \cite{Bratteli}, and from the diagram itself it is possible to

3331: read a lot of the structure of the algebra, for instance its ideal

3332: structure. Some simple examples of algebras that belong to this

3333: class are:

3334: 

3335: \begin{ex}\label{Cantorset} {\it

3336: An example of a commutative AF is the algebra of complex valued

3337: continuous functions on a Cantor set, where a Bratteli diagram is

3338: determined by a decreasing family of disjoint intervals covering

3339: the Cantor set. } \end{ex}

3340: 

3341: A non--commutative example of AF algebra is given by the algebra

3342: of the canonical anticommutation relations of quantum mechanics.

3343: 

3344: \begin{ex} {\it Consider a real Hilbert space $\cE$ and a

3345: linear map $\cE \to \cB(\cH)$, $f \mapsto T_f$, to bounded operators

3346: in a Hilbert space $\cH$, satisfying

3347: $$ T_f T_g +T_g T_f =0 $$

3348: $$ T_f^* T_g+ T_g T_f^* = \langle g,f \rangle I, $$

3349: and the algebra $\cA$ generated by all the operators $T_f$

3350: satisfying these relations. } \label{exCAR}

3351: \end{ex}

3352: 

3353: A survey with many examples of AF algebras and their properties is

3354: given for instance in \cite{Davidson}.

3355: 

3356: Let $\cA$ be a commutative AF ${\rm C}^*$-algebra. A

3357: commutative AF algebra $\cA$ is spanned by its projections,

3358: since finite dimensional commutative algebras are generated by

3359: orthogonal projections. This condition is equivalent to the

3360: spectrum $\Lambda=\Sp(\cA)$ of the algebra being a totally

3361: disconnected compact Hausdorff space, typically a Cantor set.

3362: Realizing such Cantor set as the intersection of a decreasing

3363: family of disjoint intervals covering $\Lambda$ also provides a

3364: Bratteli diagram for the AF algebra $\cA=C(\Lambda)$.

3365: 

3366: As described in \cite{Co94},

3367: in order to construct the Hilbert space $\cH$ for a Cantor set

3368: $\Lambda\subset \R$, let $J_k$ be the collection of bounded open

3369: intervals in $\R\setminus \Lambda$. We denote by $L=\{ \ell_k

3370: \}_{k\geq 1}$ the countable collection of the lengths of the

3371: intervals $J_k$. We can assume that the lengths are ordered

3372: \begin{equation}\label{ord-ell}

3373:  \ell_1 \geq \ell_2 \geq \ell_3 \geq \cdots \geq \ell_k \cdots

3374: >0.

3375: \end{equation}

3376: We also denote by $E=\{ x_{k,\pm} \}$ the set of the endpoints of

3377: the intervals $J_k$, with $x_{k,+}>x_{k,-}$. Consider the Hilbert

3378: space

3379: \begin{equation}\label{H-endpts}

3380: \cH := \ell^2 (E)

3381: \end{equation}

3382: 

3383: Since the endpoints of the $J_k$ are points of $\Lambda$, there is

3384: an action of ${\rm C}(\Lambda)$ on $\cH$ given by

3385: \begin{equation}\label{act-CS}

3386: f\cdot \xi (x) = f(x) \xi(x), \ \ \ \forall f\in {\rm

3387: C}(\Lambda),\, \, \forall \xi \in \cH,\, \, \forall x \in E.

3388: \end{equation}

3389: 

3390: A sign operator $F$ can be obtained (\cf \cite{Co94})

3391: by choosing the closed subspace

3392: $\hat\cH\subset \cH$ given by

3393: \begin{equation}\label{PH}

3394: \hat\cH=\{ \xi \in \cH : \xi(x_{k,-})=\xi(x_{k,+}), \, \, \forall

3395: k \}.

3396: \end{equation}

3397: Then $F$ has eigenspaces $\hat\cH$ with eigenvalue $+1$ and

3398: $\hat\cH^\perp$ with eigenvalue $-1$, so that, when restricted to

3399: the subspace $\cH_k$ of coordinates $\xi(x_{k,+})$ and

3400: $\xi(x_{k,-})$, the sign $F$ is given by

3401: $$ F|_{\cH_k} = \left(\begin{array}{cc} 0&1\\1&0

3402: \end{array}\right). $$

3403: 

3404: Finally, a Dirac operator $D=|D| F$ is obtained as

3405: \begin{equation}\label{fractalDirac}

3406:  D|_{\cH_k} \left(\begin{array}{c} \xi(x_{k,+})\\ \xi(x_{k,-})

3407: \end{array}\right) = \ell_k^{-1} \cdot \, \left(\begin{array}{c}

3408: \xi(x_{k,-})\\ \xi(x_{k,+}) \end{array}\right).

3409: \end{equation}

3410: 

3411: We then obtain the following result.

3412: 

3413: \begin{prop}

3414: Let $\Lambda\subset \R$ be a Cantor set. Let $\cA_\infty\subset

3415: C(\Lambda)$ be the dense subalgebra of locally constant functions on

3416: the Cantor set. Then the data $(\cA,\cH,D)$ form a spectral triple,

3417: with $\cH$ as in

3418: (\ref{H-endpts}), the action (\ref{act-CS}), and $D$ as in

3419: (\ref{fractalDirac}). The zeta function satisfies

3420: $$ \Tr( |D|^{-s} ) = 2 \zeta_L (s), $$

3421: where $\zeta_L(s)$ is the geometric zeta function of

3422: $L=\{ \ell_k \}_{k\geq 1}$, defined as

3423: \begin{equation}\label{zeta-fractal}

3424: \zeta_L(s):= \sum_k \ell_k^s.

3425: \end{equation}

3426: \label{propZCS}

3427: \end{prop}

3428: 

3429: \smallskip

3430: 

3431: These zeta functions are related to the theory of Dirichlet series

3432: and to other arithmetic zeta functions, and also to Ruelle's

3433: dynamical zeta functions (\cf \cite{LapFra}).

3434: 

3435: \smallskip

3436: 

3437: For example, for the classical middle-third Cantor set, we have

3438: set of lengths $\ell_k = 3^{-k}$ and multiplicities $m_k=2^{k-1}$,

3439: for $k\geq 1$, so that we obtain

3440: \begin{equation}\label{zeta-CS}

3441: \Tr(|D|^{-s})= 2\zeta_L(s)=\sum_{k\geq 1} 2^k 3^{-sk} =

3442: \frac{2\cdot 3^{-s}}{1-2\cdot 3^{-s}}.

3443: \end{equation}

3444: This shows that the dimension spectrum of the spectral triple of a

3445: Cantor set has points off the real line. In fact, the set of poles

3446: of (\ref{zeta-CS}) is

3447: \begin{equation}\label{poles-CS}

3448: \left\{ \frac{\log 2}{\log 3} + \frac{2\pi i n}{\log 3}

3449: \right\}_{n\in \Z}.

3450: \end{equation}

3451: 

3452: 

3453: In this case the dimension spectrum lies on a vertical line and it

3454: intersects the real axis in the point $D=\frac{\log 2}{\log 3}$

3455: which is the Hausdoff dimension of the ternary Cantor set. The

3456: same is true for other Cantor sets, as long as the self-similarity

3457: is given by a unique contraction (in the ternary case the original

3458: interval is replaced by two intervals of lengths scaled by 1/3).

3459: 

3460: If one considers slightly more complicated fractals in $\R$, where

3461: the self-similarity requires more than one scaling map, the

3462: dimension spectrum may be correspondingly more complicated. This

3463: can be seen in the case of the Fibonacci Cantor set, for instance (\cf

3464: \cite{LapFra}).

3465: 

3466: The Fibonacci Cantor set $\Lambda$ is obtained from the interval

3467: $I=[0,4]$ by successively removing $F_{n+1}$ open intervals

3468: $J_{n,j}$ of lengths $\ell_n =1/2^n$

3469: according to the rule of Figure \ref{FigFC}.

3470: We can associate to this Cantor set the

3471: commutative AF algebra $\cA=C(\Lambda)$.

3472: 

3473: \begin{center}

3474: \begin{figure}

3475: \includegraphics[scale=0.35]{FibCantor.eps}

3476: \caption{The Fibonacci Cantor set.

3477: \label{FigFC}}

3478: \end{figure}

3479: \end{center}

3480: 

3481: To obtain the Hilbert space we consider again the set $E$ of

3482: endpoints $x_{n,j,\pm}$ of the intervals $J_{n,j}$ and we take

3483: $\cH=\ell^2(E)$. We define the Dirac operator as in the previous

3484: case, and we again consider the dense involutive subalgebra

3485: $\cA_\infty$ of locally constant functions.

3486: 

3487: The data $(\cA,\cH,D)$ give a spectral triple. The Dirac

3488: operator is related to the geometric zeta function of the

3489: Fibonacci Cantor set by

3490: $$ Tr(|D|^{-s})= 2 \zeta_F(s) =\frac{2}{1-2^{-s}-4^{-s}}, $$

3491: where the geometric zeta function is $\zeta_F(s)=\sum_n F_{n+1}

3492: 2^{-ns}$, with $F_n$ the Fibonacci numbers.

3493: 

3494: A simple argument shows that the dimension

3495: spectrum is given by the set

3496: $$ \Sigma= \left\{

3497: \frac{\log\phi}{\log 2} + \frac{2\pi i n}{\log 2}  \right\}_{n\in

3498: \Z} \cup \left\{ -\frac{\log\phi}{\log 2} + \frac{2\pi i

3499: (n+1/2)}{\log 2} \right\}_{n\in \Z}, $$ where

3500: $\phi=\frac{1+\sqrt{5}}{2}$ is the golden ratio.

3501: 

3502: 

3503: \medskip

3504: 

3505: Recent results on the noncommutative geometry of fractals and Cantor

3506: sets and spectral triple constructions for AF algebras can be found

3507: in \cite{AntChris}, \cite{GuIso1}, \cite{GuIso2}. The construction 

3508: in \cite{AntChris} is in fact a spectral triple for the dual group

3509: of the Cantor set seen as the product of countably many copies of 

3510: the group $\Z/2$. The recent work

3511: \cite{IvChris} shows that it is easy to describe a compact metric

3512: space exactly (\ie recovering the metric) via a spectral triple,

3513: which is a sum of two-dimensional modules, but spectral triples

3514: carry much more information than just the one regarding the metric.

3515: 

3516: 

3517: \bigskip

3518: 

3519: \section{Spaces of dimension $z$ and DimReg}\label{DimReg}

3520: 

3521: In perturbative quantum field theory, one computes expectation values

3522: of observables via a formal series, where the terms are parameterized

3523: by Feynman graphs and reduce to ordinary finite dimensional integrals

3524: in momentum space of expressions assigned to the graphs by the

3525: Feynman rules. These expressions typically produce divergent

3526: integrals. For example, in the example of the scalar $\phi^3$ theory

3527: in dimension $D=4$ or $D=4+2\N$, one encounters a divergence already

3528: in the simplest one loop diagram, with corresponding integral (in

3529: Euclidean signature)

3530: \begin{equation}\label{Fint}

3531:   \hbox{\psfig{figure=feydiag.eps}\quad =}  \int \frac{1}{k^2+m^2}\,

3532:  \frac{1}{((p+k)^2+m^2)} \, d^D k.

3533: \end{equation}

3534: 

3535: One needs therefore a regularization procedure for these divergent

3536: integrals. The regularization most commonly adopted in quantum field

3537: theory computation is ``Dimensional Regularization (DimReg) and Minimal

3538: Subtraction (MS)''. The method was introduced in the '70s in

3539: \cite{BoGia} and \cite{tHV} and it has the advantage of preserving

3540: basic symmetries.

3541: 

3542: The regularization procedure of DimReg is essentially based on the use

3543: of the formula

3544:  \begin{equation}\label{gaussienne}

3545:  \int\,e^{-\lambda\,k^2}\,d^{d}

3546: k\,=\,{\pi^{d/2}}\,{\lambda^{-d/2}}\,,

3547:  \end{equation}

3548: to {\em define} the meaning of the integral in

3549: $d=(D-z)$-dimensions, for $z\in \C$ in a neighborhood of zero.

3550: For instance, in the case of \eqref{Fint}, the procedure of

3551: dimensional regularization yields the result

3552: $$ \pi^{(D-z)/2} \Gamma\left(\frac{4-D+z}{2}\right) \int_0^1 \left( (

3553: x-x^2)p^2 +m^2\right)^{\frac{D-z-4}{2}} \, dx .$$

3554: 

3555: \medskip

3556: 

3557: In the recent survey \cite{ManDim}, Yuri Manin refers to DimReg as

3558: ``dimensions in search of a space''\footnote{Nicely reminiscent of

3559: Pirandello's play ``Six characters in search of an author''.}.

3560: Indeed, in the usual approach in perturbative quantum field theory, the

3561: dimensional regularization procedure is just regarded as a formal rule

3562: of analytic continuation of formal (divergent) expressions in integral

3563: dimensions $D$ to complex values of the variable $D$.

3564: 

3565: However, using noncommutative geometry, it is possible to construct

3566: actual spaces (in the sense of noncommutative Riemannian geometry)

3567: $X_z$ whose dimension (in the sense of dimension spectrum) is a

3568: point $z\in \C$ (\cf \cite{CoMar-an}).

3569: 

3570: It is well known in the physics literature that there are problems

3571: related to using dimensionl regularization in chiral theory, because

3572: of how to give a consistent prescription on how to extend the

3573: $\gamma_5$ (the product of the matrices $\gamma^i$ when $D=4$) to

3574: noninteger dimension $D-z$. It turns out that a prescription

3575: known as Breitenlohner-Maison (\cite{BM}, \cite{Collins}) admits an

3576: interpretation in terms of the cup product of spectral triples, where

3577: one takes the product of the spectral triple associated to the ordinary

3578: geometry in the integer dimension $D$ by

3579: a spectral triple $X_z$ whose dimension spectrum is reduced

3580: to the complex number $z$ (\cf \cite{CoMar-an}).

3581: 

3582: We illustrate here the construction for the case where $z\in

3583: \R^*_+$. The more general case of $z\in \C$ is more delicate.

3584: 

3585: One needs to work in a slightly modified setting for spectral triples,

3586: which is given by the type II spectral triples (\cf \cite{BF},

3587: \cite{CPRS}, \cite{CPRS1}). In this setting the usual type I  trace

3588: of operators in $\cL(\cH)$ is replaced by the

3589: trace on a type II$_{\infty}$ von-Neumann algebra.

3590: 

3591: One considers a self-adjoint operator $Y$, affiliated to

3592: a type II$_{\infty}$ factor $N$, with spectral measure given by

3593: \begin{equation}\label{Sp-measure}

3594: \Tr_N (\chi_E(Y))=\,\frac{1}{2}\,

3595: \int_E\,dy

3596: \end{equation}

3597: for any interval $E\subset \R$, with characteristic function

3598: $\chi_E$.

3599: 

3600: If $Y=F\, |Y|$ is the polar decomposition of $Y$, one sets

3601: \begin{equation}\label{newdz}

3602: D_z=\, \rho(z) \,F\, |Y|^{1/z}

3603:  \end{equation}

3604: with the complex

3605: power $|Y|^{1/z}$ defined by the

3606: functional calculus.

3607: The normalization constant $ \rho(z)$

3608: is chosen to be

3609: \begin{equation}\label{rhoz}

3610:  \rho(z)=\,\pi^{-\frac{1}{2}}\,

3611: \left(\Gamma(\frac{z}{2}+1)\right)^{\frac{1}{z}}

3612: \end{equation}

3613: so that one obtains

3614: \begin{equation}\label{basic}

3615: \Tr \left(e^{-\lambda D^2}\right)=\,

3616: \,{\pi^{z/2}}\,{\lambda^{-z/2}} \ \ \ \forall \lambda \in \R^*_+\,.

3617: \end{equation}

3618: 

3619: This gives a geometric meaning to the basic formula \eqref{gaussienne}

3620: of DimReg. The algebra $\cA$ of the spectral triple $X_z$ can be made

3621: to contain any operator $a$ such that $[D_z,a]$ is bounded

3622: and both $a$

3623: and $[D_z,a]$ are smooth for the ``geodesic flow"

3624: \begin{equation}\label{geod}

3625: T\mapsto \,e^{it|D_z|}\,T\,e^{-it|D_z|}.

3626: \end{equation}

3627: The dimension spectrum of $X_z$ is reduced to the

3628: single point $z$, since

3629: \begin{equation}\label{trace'}

3630: {\rm Trace}'_N(

3631: (D^2_z)^{-s/2})=\,\rho^{-s}\, \int_1^\infty\,u^{-s/z}\,du=

3632: \,\rho^{-s}\,\frac{z}{s-z}

3633: \end{equation}

3634: has a single (simple) pole

3635: at $s=z$ and is absolutely convergent in the half space

3636: Re$(s/z)>1$. Here ${\rm Trace}'_N$ denotes the trace with an

3637: infrared cutoff (\ie integrating outside $|y|<1$).

3638: 

3639: \bigskip

3640: 

3641: \section{Local algebras in supersymmetric QFT}\label{susy}

3642: 

3643: 

3644: It is quite striking that the general framework of noncommutative

3645: geometry is suitable not only for handling  finite dimensional

3646: spaces (commutative or not, of non-integer dimension etc.) but is

3647: also compatible with infinite dimensional spaces. We already saw in

3648: Section \ref{discgr} that discrete groups of exponential growth

3649: naturally give rise to noncommutative spaces which are described by

3650: a $\theta$-summable spectral triple, but not by a finitely summable

3651: spectral triple. This is characteristic of an infinite dimensional

3652: space and in that case, as we saw for discrete groups, cyclic

3653: cohomology needs to be extended to {\em entire} cyclic cohomology. A

3654: very similar kind of noncommutative spaces arises from Quantum Field

3655: Theory in the supersymmetric context \cite{Co94} Section

3656: IV.9.$\beta$. We   briefly recall this below and then explain

3657: open questions also in the context of  supersymmetric theories.

3658: 

3659: The simplest example to understand the framework is that of the free

3660: Wess-Zumino model in two dimensions, a supersymmetric free field

3661: theory in a two dimensional space-time where space is compact

3662: (\cite{Co94}). Thus space is a circle $S^1$ and space-time is a

3663: cylinder $C=\,S^1\times \R$ endowed with the Lorentzian metric.  The

3664: fields are given by a complex scalar bosonic field $\phi$ of mass

3665: $m$ and a spinor field $\psi$ of the same mass. The Lagrangian of

3666: the theory is of the form $\cL=\,\cL_b +\cL_f$ where,

3667: $$

3668: \cL_b=\,\frac{1}{2}(|\partial_0 \phi|^2- |\partial_1 \phi|^2 - m^2

3669: |\phi|^2)

3670: $$

3671: and for the fermions,

3672: $$

3673: \cL_f=\,i\,\bar \psi\,\gamma^\mu\,\partial_\mu\,\psi-\,m\,\bar

3674: \psi\,\psi

3675: $$

3676: where the spinor field is given by a column matrix, with $\bar

3677: \psi=\,\gamma^0\,\psi^*$ and the $\gamma^\mu$ are two by two Pauli

3678: matrices, anticommuting, self-adjoint and of square $1$.

3679: 

3680: The Hilbert space of the quantum theory is the tensor product

3681: $\cH=\,\cH_b \otimes \cH_f$ of the bosonic one $\cH_b$ by the

3682: fermionic one $\cH_f$. The quantum field $\phi(x)$ and its conjugate

3683: momentum $\pi(x)$ are operator valued distributions in $\cH_b$ and

3684: the bosonic Hamiltonian is of the form

3685: $$

3686: H_b=\,\int_{S^1} \,:\,|\pi(x)|^2 + |\partial_1 \phi(x)|^2 + m^2

3687: |\phi(x)|^2  \,:\, dx

3688: $$

3689: where the Wick ordering takes care of an irrelevant additive

3690: constant. The fermionic Hilbert space $\cH_f$ is given by the Dirac

3691: sea representation which simply corresponds to a suitable spin

3692: representation of the infinite dimensional Clifford algebra

3693: containing the fermionic quantum fields $\psi_j(x)$. The fermionic

3694: Hamiltonian is then the positive operator in $\cH_f$ given by

3695: $$

3696: H_f=\,\int_{S^1}\,:\,\bar

3697: \psi\,\gamma^1\,i\,\partial\,\psi-\,m\,\bar \psi\,\psi \,:\,

3698: $$

3699: The full Hamiltonian of the non-interacting theory is acting in the

3700: Hilbert space $\cH=\,\cH_b \otimes \cH_f$ and is the positive

3701: operator

3702: $$

3703: H=\,H_b\otimes 1 +\,1\otimes H_f \,.

3704: $$

3705: This is were supersymmetry enters the scene in finding a

3706: self-adjoint square root of $H$ in the same way as the Dirac

3707: operator is a square root of the Laplacian in the case of finite

3708: dimensional manifolds. This square root, called the {\em

3709: supercharge} operator, is given by

3710: $$

3711: Q=\,\frac{1}{\sqrt{2}}\,\int_{S^1}(

3712: \psi_1(x)(\pi(x)-\partial\phi^*(x)-i m \phi(x))+

3713: \psi_2(x)(\pi^*(x)-\partial\phi(x)-i m \phi^*(x)) + {\rm h.c.})dx

3714: $$

3715: where the symbol $+{\rm h.c.}$ means that one adds the hermitian

3716: conjugate.

3717: 

3718: The basic relation with spectral triples is then given by the

3719: following result (\cite{Co94} Section IV).

3720: 

3721: \begin{thm} For any local region $\mathcal O\subset C$ let $\cA(\mathcal

3722: O)$ be the algebra of functions of quantum fields with support in

3723: $\mathcal O$ acting in the Hilbert space $\cH$. Then the triple

3724: $$

3725: (\cA(\mathcal O),\cH,Q)

3726: $$

3727: is an even $\theta$-summable spectral triple, with $\Z_2$-grading

3728: given by the operator $\gamma=\,(-1)^{N_f}$ counting the parity of

3729: the fermion number operator $N_f$.

3730: \end{thm}

3731: 

3732: 

3733: To be more specific the algebra $\cA(\mathcal O)$ is generated by

3734: the imaginary exponentials $e^{i(\phi(f)+\phi(f)^*)}$ and

3735: $e^{i(\pi(f)+\pi(f)^*)}$ for $f\in C_c^{\infty}(\mathcal O)$. As

3736: shown in \cite{Co94} Section IV.9.$\beta$, and exactly as in the

3737: case of discrete groups with exponential growth, one needs the

3738: entire cyclic cohomology rather than its finite dimensional version

3739: in order to obtain the Chern character of $\theta$-summable spectral

3740: triples. Indeed the index map is non polynomial in the above example

3741: of the Wess-Zumino model in two dimensions and the $K$-theory of the

3742: above local algebras is highly non-trivial. In fact it is in that

3743: framework that the JLO-cocycle was discovered by Jaffe-Lesniewski

3744: and Osterwalder \cite{JLO}.

3745: 

3746: 

3747: It is an open problem to extend the above result to interacting

3748: theories in higher dimension and give a full computation of the

3749: $K$-theory of the local algebras as well as of the Chern character

3750: in entire cyclic cohomology. The results of Jaffe and his

3751: collaborators on constructive quantum field theory yield many

3752: interacting non-trivial examples of supersymmetric two dimensional

3753: models. Moreover the recent breakthrough of Puschnigg in the case of

3754: lattices  of semi-simple Lie groups of rank one opens the way to the

3755: computation of the Chern character in entire cyclic cohomology.

3756: 

3757: \bigskip

3758: 

3759: \section{Spacetime and the standard model of elementary

3760: particles}\label{stmod}

3761: 

3762: The standard model of elementary particle physics provides a

3763: surprising example of a spectral triple in the noncommutative

3764: setting, which in addition to the conditions of Definition

3765: \ref{def-triple} also has a real structure satisfying all the

3766: additional conditions of Definition \ref{realstr}.

3767: 

3768: The noncommutative geometry of the standard model developed in

3769: \cite{Co3} (\cf also \cite{C-C}, \cite{C-C2}, \cite{CL},

3770: \cite{Kast2}) gives a concise conceptual way to describe, through

3771: a simple mathematical structure, the full complexity of the input

3772: from physics. As we   recall here, the model also allows for

3773: predictions.

3774: 

3775: The physics of the standard model can be described by a

3776: Lagrangian. We consider here the standard model minimally coupled

3777: to gravity, so that the Lagrangian we shall be concerned with is

3778: the sum

3779: \begin{equation}\label{Lagr-EHSM}

3780:   \cL = \cL_{EH} + \cL_{SM}

3781: \end{equation}

3782: of the Einstein--Hilbert Lagrangian $\cL_{EH}$ and the standard

3783: model Lagrangian $\cL_{SM}$.

3784: 

3785: The standard model Lagrangian $\cL_{SM}$ has a very complicated

3786: expression, which, if written in full, might take a full page (\cf

3787: \eg \cite{Velt2}). It comprises five types of terms,

3788: \begin{equation}\label{Lagr-SM}

3789: \cL_{SM} = \cL_G + \cL_{GH} + \cL_H + \cL_{Gf} + \cL_{Hf},

3790: \end{equation}

3791: where the various terms involve:

3792: \begin{itemize}

3793: \item spin $1$ bosons $G$: the eight gluons, $\gamma$, $W^\pm$, $Z$;

3794: \item spin $0$ bosons $H$ such as the Higgs fields;

3795: \item spin $1/2$ fermions $f$: quarks and leptons.

3796: \end{itemize}

3797: The term $\cL_G$ is the pure gauge boson part, $\cL_{GH}$ for the

3798: minimal coupling with the Higgs fields, and $\cL_H$ gives the

3799: quartic Higgs self interaction. In addition to the coupling

3800: constants for the gauge fields, the fermion kinetic term

3801: $\cL_{Gf}$ contains the hypercharges $Y_L$, $Y_R$. These numbers,

3802: which are constant over generations, are assigned

3803: phenomenologically, so as to obtain the correct values of the

3804: electromagnetic charges. The term $\cL_{Hf}$ contains the Yukawa

3805: coupling of the Higgs fields with fermions. A more detailed and

3806: explicit description of the various terms of \eqref{Lagr-SM} is

3807: given in \cite{Co94} \S VI.5.$\beta$. See also \cite{Velt2}.

3808: 

3809: \begin{figure}

3810: \begin{center}

3811: \includegraphics[scale=0.65]{standardmodel.eps}

3812: \end{center}

3813: \caption{Elementary particles \label{FigSM}}

3814: \end{figure}

3815: 

3816: 

3817: The symmetry group of the Einstein--Hilbert Lagrangian $\cL_{EH}$

3818: by itself would be, by the equivalence principle, the

3819: diffeomorphism group ${\rm Diff}(X)$ of the space-time manifold.

3820: In the standard model Lagrangian $\cL_{SM}$, on the other hand,

3821: the gauge theory has another huge symmetry group which is the

3822: group of local gauge transformations. According to our current

3823: understanding of elementary particle physics, this is given by

3824: \begin{equation}\label{gauge-gr}

3825: G_{SM}(X)= {\rm C}^\infty(X,U(1)\times SU(2)\times SU(3)).

3826: \end{equation}

3827: (At least in the case of a trivial principal bundle, \eg when the

3828: spacetime manifold $X$ is contractible.)

3829: 

3830: Thus, when one considers the Lagrangian $\cL$ of

3831: \eqref{Lagr-EHSM}, the full symmetry group $G$ will be a

3832: semidirect product

3833: \begin{equation}\label{symm-gr-SM}

3834: G(X) = G_{SM}(X) \rtimes {\rm Diff}(X).

3835: \end{equation}

3836: In fact, a diffeomorphism of the manifold relabels the gauge

3837: parameters.

3838: 

3839: To achieve a {\em geometrization} of the standard model, one would

3840: like to be able to exhibit a space $X$ for which

3841: \begin{equation}\label{diffG}

3842: G(X)={\rm Diff}(X).

3843: \end{equation}

3844: If such a space existed, then we would be able to say that the whole

3845: theory is pure gravity on $X$. However, it is impossible to find

3846: such a space $X$ among ordinary manifolds. In fact, a result of W.

3847: Thurston, D. Epstein and J. Mather (\cf \cite{mather}) shows that

3848: the connected component of the identity in the diffeomorphism group

3849: of a (connected) manifold is a simple group (see \cite{mather} for

3850: the precise statement). A simple group cannot have a nontrivial

3851: normal subgroup, so it cannot have the structure of semi-direct

3852: product like $G(X)$ in \eqref{symm-gr-SM}.

3853: 

3854: However, it is possible to obtain a space with the desired

3855: properties among {\em noncommutative spaces}. What plays the role of

3856: the  connected component of the identity in the diffeomorphism group

3857: ${\rm Diff}(X)$ in the noncommutative setting is the group

3858: $\Aut^+(\cA)$ of automorphisms of the (noncommutative) algebra that

3859: preserve the fundamental class in $K$-homology \ie that can be

3860: implemented by a unitary compatible with the grading and real

3861: structure.

3862: 

3863: When the algebra $\cA$ is not commutative, among its automorphisms

3864: there are, in particular, inner ones. They associate to an element

3865: $x$ of the algebra the element $uxu^{-1}$, for some $u\in \cA$. Of

3866: course $uxu^{-1}$ is not, in general, equal to $x$ because the

3867: algebra is not commutative. The inner automorphisms form a normal

3868: subgroup of the group of automorphisms. Thus, we see that the

3869: group $\Aut^+ (\cA)$ has in general the same type of structure as

3870: our desired group of symmetries $G(X)$, namely, it has a normal

3871: subgroup of inner automorphisms and it has a quotient. It is

3872: amusing how the physical and the mathematical vocabularies agree

3873: here: in physics one talks about internal symmetries and in

3874: mathematics one talks about inner automorphisms (one might as well

3875: call them internal automorphisms).

3876: 

3877: There is a very simple non commutative algebra $\cA$ whose group

3878: of inner automorphisms corresponds to the group of gauge

3879: transformations $G_{SM}(X)$, and such that the quotient $\Aut^+

3880: (\cA) / {\rm Inn} (\cA)$ corresponds exactly to diffeomorphisms

3881: (\cf \cite{Schu}). The noncommutative space is a product $X\times

3882: F$ of an ordinary spacetime manifold $X$ by a ``finite

3883: noncommutative space'' $F$. The noncommutative algebra $\cA_F$ is

3884: a direct sum of the algebras $\bC$, $\bH$ (here denoting the

3885: quaternions), and $M_3 (\bC)$ (the algebra of $3\times 3$ complex

3886: matrices).

3887: 

3888: The algebra $\cA_F$ corresponds to a {\it finite} space where the

3889: standard model fermions and the Yukawa parameters (masses of

3890: fermions and mixing matrix of Kobayashi Maskawa) determine the

3891: spectral geometry in the following manner. The Hilbert space

3892: $\cH_F$ is finite-dimensional and admits the set of elementary

3893: fermions as a basis. This comprises the generations of quarks

3894: (down--up, strange--charmed, bottom--top),

3895: \begin{equation}

3896: \begin{array}{ccccccccc}

3897: u_L & u_R & d_L & d_R& & \bar{u}_L & \bar{u}_R & \bar{d}_L & \bar{d}_R \\[2mm]

3898: c_L & c_R & s_L & s_R& & \bar{c}_L & \bar{c}_R & \bar{s}_L & \bar{s}_R \\[2mm]

3899: t_L & t_R & b_L & b_R& & \bar{t}_L & \bar{t}_R & \bar{b}_L &

3900: \bar{b}_R,

3901: \end{array}

3902: \label{quarks}

3903: \end{equation}

3904: with the additional color index $(y,r,b)$, and the generations of

3905: leptons (electron, muon, tau, and corresponding neutrinos)

3906: \begin{equation}

3907: \begin{array}{ccccccc}

3908: e_L& e_R& \nu^e_L& & \bar e_L& \bar e_R& \bar \nu^e_L

3909: \\[2mm]

3910: \mu_L& \mu_R& \nu^\mu_L& & \bar \mu_L& \bar \mu_R& \bar \nu^\mu_L

3911: \\[2mm]

3912: \tau_L& \tau_R& \nu^\tau_L& & \bar \tau_L& \bar \tau_R &

3913: \bar\nu^\tau_L

3914: \end{array}

3915: \label{leptons}

3916: \end{equation}

3917: (We discuss here only the minimal standard model with no right

3918: handed neutrinos.)

3919: 

3920: The $\Z/2$ grading $\gamma_F$ on the Hilbert space $\cH_F$ has

3921: sign $+1$ on left handed particles (\eg the $u_L$, $d_L$, etc.)

3922: and sign $-1$ on the right handed particles. The involution $J_F$

3923: giving the real structure is the charge conjugation, namely, if we

3924: write $\cH_F = \cE \oplus \bar \cE$, then $J_F$ acts on the

3925: fermion basis as $J_F(f,\bar h) = (h,\bar f)$. This satisfies

3926: $J_F^2=1$ and $J_F\gamma_F=\gamma_F J_F$, as should be for

3927: dimension $n=0$.

3928: 

3929: The algebra $\cA_F$ admits a natural representation in $\cH_F$

3930: (see \cite{Coo2}). An element $(z,q,m)\in \C \oplus \H \oplus

3931: M_3(\C)$ acts as

3932: $$ (z,q,m) \cdot \left(\begin{array}{c} u_R \\ d_R \end{array}\right)

3933: = \left(\begin{array}{c} z\, u_R \\ \bar z\, d_R

3934: \end{array}\right) \ \ \ \ (z,q,m) \cdot e_R = \bar z\,  e_R $$

3935: $$ (z,q,m) \cdot \left(\begin{array}{c} u_L \\ d_L \end{array}\right)

3936: = q \left(\begin{array}{c} u_L \\ d_L \end{array}\right) \ \ \ \

3937: (z,q,m) \cdot \left(\begin{array}{c} \nu^e_L \\ e_L

3938: \end{array}\right) = q \left(\begin{array}{c} \nu^e_L \\ e_L

3939: \end{array}\right), $$

3940: $$ (z,q,m) \cdot \left(\begin{array}{c} \bar e_L \\ \bar e_R

3941: \end{array}\right) = \left(\begin{array}{c} z\, \bar e_L \\ z\, \bar e_R

3942: \end{array}\right) $$

3943: $$ (z,q,m) \cdot \bar u_R  = m\,  \bar u_R \ \ \ \ (z,q,m) \cdot \bar

3944: d_R =m  \, \bar  d_R $$ and similarly for the other generations.

3945: Here $q\in \H$ acts as multiplication by the matrix

3946: $$ q= \left(\begin{array}{cc} \alpha & \beta \\

3947: -\bar \beta & \bar \alpha \end{array}\right),  $$ where $q=\alpha+

3948: \beta\, j$, with $\alpha,\beta\in \C$. The matrix $m\in M_3(\C)$

3949: acts on the color indices $(y,r,b)$.

3950: 

3951: The data $(\cA_F, \cH_F)$ can be completed to a spectral triple

3952: $(\cA_F,\cH_F,D_F)$ where the Dirac operator (in this finite

3953: dimensional case a matrix) is given by

3954: \begin{equation}\label{YukawaD}

3955: D_F = \left( \begin{array}{cc} Y & 0 \\ 0 & \bar Y

3956: \end{array}\right)

3957: \end{equation}

3958: on $\cH_F=\cE\oplus \bar \cE$, where $Y$ is the Yukawa coupling

3959: matrix, which combines the masses of the elementary fermions

3960: together with the Cabibbo--Kobayashi--Maskawa (CKM) quark mixing matrix.

3961: 

3962: The fermionic fields acquire mass through the spontaneous symmetry

3963: breaking produced by the Higgs fields. The Yukawa coupling matrix

3964: takes the form $Y=Y_q \otimes 1 \oplus Y_f$, where the matrix

3965: $Y_f$ is of the form

3966: $$ \left(\begin{array}{ccc} 0&0& M_e\\

3967: 0&0&0\\ M_e^* &0&0 \end{array}\right), $$ in the basis

3968: $(e_R,\nu_L,e_L)$ and successive generations, while $Y_q$ is of

3969: the form

3970: $$ \left(\begin{array}{cccc} 0&0& M_u & 0\\

3971: 0&0&0& M_d \\M_u^* &0&0&0 \\ 0& M_d^* & 0&0

3972: \end{array}\right), $$

3973: in the basis given by $(u_R,d_R,u_L,d_L)$ and successive

3974: generations. In the case of the lepton masses, up to rotating the

3975: fields to mass eigenstates, one obtains a mass term for each

3976: fermion, and the off diagonal terms in $M_e$ can be reabsorbed in

3977: the definition of the fields. In the quark case, the situation is

3978: more complicated and the Yukawa coupling matrix can be reduced to

3979: the mass eigenvalues and the CKM quark mixing. By rotating the

3980: fields, it is possible to eliminate the off diagonal terms in

3981: $M_u$. Then $M_d$ satisfies $V M_d V^*=M_u$, where $V$ is the CKM

3982: quark mixing, given by a $3\times 3$ unitary matrix

3983: $$ V= \left( \begin{array}{ccc} V_{ud} & V_{us} & V_{ub} \\

3984: V_{cd} & V_{cs} & V_{cb} \\

3985: V_{td} & V_{ts} & V_{tb} \end{array} \right) $$ acting on the

3986: charge $-e/3$ quarks (down, strange, bottom). The entries of this

3987: matrix can be expressed in terms of three angles

3988: $\theta_{12},\theta_{23},\theta_{13}$ and a phase, and can be

3989: determined experimentally from weak decays and deep inelastic

3990: neutrino scatterings.

3991: 

3992: The detailed structure of the Yukawa coupling matrix $Y$ (in

3993: particular the fact that color is not broken) allows one to check

3994: that the finite geometry $(\cA_F,\cH_F,D_F)$ satisfies all the

3995: axioms of Definition \ref{realstr} for a noncommutative spectral

3996: manifold. The key point is that elements $a \in \cA_F$ and

3997: $[D_f,a]$ commute with $J_F\, \cA_F\, J_F$. These operators

3998: preserve the subspace $\cE \subset \cH_F$. On this subspace, for

3999: $b=(z,q,m)$, the action of $J_F \, b^* \, J_F$ is by

4000: multiplication by $z$ or by the transpose $m^t$. It is then not

4001: hard to check explicitly the commutation with $a$ or $[D,a]$ (\cf

4002: \cite{Co94} \S VI.5.$\delta$). By exchanging the roles of $a$ and

4003: $b$, one sees analogously that $a$ commutes with $J_F b J_F$ and

4004: $[D,J_F b J_F]$ on $\bar \cE$, hence the desired commutation

4005: relations hold on all of $\cH_F$.

4006: 

4007: We can then consider the product $X\times F$, where $X$ is an

4008: ordinary $4$--dimensional Riemannian spin manifold and $F$ is the

4009: finite geometry described above. This product geometry is a

4010: spectral triple $(\cA,\cH,D)$ obtained as the cup product of a

4011: triple $(\cC^\infty(X),L^2(X,S), D_1)$, where $D_1$ is the Dirac

4012: operator on $X$ acting on square integrable spinors in $L^2(X,S)$,

4013: with the spectral triple $(\cA_F,\cH_F,D_F)$ described above.

4014: Namely, the resulting (smooth) algebra and Hilbert space are of

4015: the form

4016: \begin{equation}\label{algXF}

4017: \cA_\infty=\cC^\infty(X,\cA_F) \ \ \ \  \cH= L^2(X,S)\otimes

4018: \cH_F,

4019: \end{equation}

4020: and the Dirac operator is given by

4021: \begin{equation}\label{diracXF}

4022: D = D_1 \otimes 1 + \gamma \otimes D_F,

4023: \end{equation}

4024: where $\gamma$ is the usual $\Z/2$ grading on the spinor bundle

4025: $S$. The induced $\Z/2$ grading on $\cH$ is the tensor product

4026: $\gamma\otimes \gamma_F$, and the real structure is given by

4027: $J=C\otimes J_F$, where $C$ is the charge conjugation operator on

4028: spinors.

4029: 

4030: Notice that, so far, we have only used the information on the

4031: fermions of the standard model. We'll see now that the bosons,

4032: with the correct quantum numbers, are {\em deduced} as inner

4033: fluctuations of the metric of the spectral triple $(\cA,\cH,D)$.

4034: 

4035: It is a general fact that, for noncommutative geometries

4036: $(\cA,\cH,D)$, one can consider inner fluctuations of the metric

4037: of the form

4038: $$

4039: D\mapsto D+A+JAJ^{-1}

4040: $$

4041: where $A$ is of the form

4042: \begin{equation}\label{Aab}

4043: A= \sum a_i \, [D, a_i' ]  \ \ \ \  a_i,a_i' \in \cA.

4044: \end{equation}

4045: 

4046: In the case of the standard model, a direct computation of the

4047: inner fluctuations gives the standard model gauge bosons $\gamma ,

4048: W^{\pm} ,Z$, the eight gluons and the Higgs fields $\varphi$ with

4049: accurate quantum numbers (\cf \cite{Co3}). In fact, a field $A$ of

4050: the form \eqref{Aab} can be separated in a ``discrete part''

4051: $A^{(0,1)} = \sum a_i \, [\gamma\otimes D_F, a_i' ]$ and a

4052: continuous part $A^{(1,0)} = \sum a_i \, [D_1\otimes 1, a_i' ]$,

4053: with $a_i=(z_i,q_i,m_i)$ and $a_i'=(z_i',q_i',m_i')$,

4054: $q_i=\alpha_i + \beta_i j$ and $q_i'=\alpha_i' + \beta_i' j$. The

4055: discrete part gives a quaternion valued function

4056: $$ q(x) = \sum z_i \left( (\alpha_i'-z_i') + z_i \beta_i'\, j \right)

4057: = \varphi_1 + \varphi_2 \, j  $$ which provides the Higgs doublet.

4058: The continuous part gives three types of fields:

4059: \begin{itemize}

4060: \item A $U(1)$ gauge field $U =\sum z_i \, d z_i'$

4061: \item An $SU(2)$ gauge field $Q= \sum q_i \, d q_i'$

4062: \item A $U(3)$ gauge field $M=\sum m_i \, d m_i'$,

4063: which can be reduced to an $SU(3)$ gauge field $M'$ by subtracting

4064: the scalar part of the overall gauge field which eliminates

4065: inessential fluctuations that do not change the metric.

4066: \end{itemize}

4067: The resulting internal fluctuation of the metric $A+J A J^{-1}$ is

4068: then of the form (\cf \cite{Co3})

4069: $$ \left(\begin{array}{ccc} -2U & 0 & 0 \\

4070: 0 & Q_{11}-U & Q_{12} \\ 0& Q_{21} & Q_{22}-U

4071: \end{array} \right) $$ on the basis of leptons $(e_R,\nu_L,e_L)$

4072: and successive generations, and

4073: $$ \left(\begin{array}{cccc} \frac{4}{3} U + M' & 0 & 0 & 0 \\[2mm]

4074: 0 & \frac{-2}{3} U + M' & 0 & 0 \\[2mm]

4075: 0 & 0 & Q_{11} + \frac{1}{3} U + M' & Q_{12} \\[2mm]

4076: 0 & 0 & Q_{21} & Q_{22} + \frac{1}{3} U + M'

4077: \end{array}\right), $$

4078: on the basis of quarks given by $(u_R,d_R,u_L,d_L)$ and successive

4079: generations. A striking feature that these internal fluctuations

4080: exhibit is the fact that the expressions above recover all the

4081: exact values of the hypercharges $Y_L$, $Y_R$ that appear in the

4082: fermion kinetic term of the standard model Lagrangian.

4083: 

4084: Finally, one can also recover the bosonic part of the standard model

4085: Lagragian from a very general principle, the {\em spectral action

4086: principle} of Chamseddine--Connes (\cf \cite{C-C}, \cite{C-C2},

4087: \cite{C-C3}). The result is that the Hilbert--Einstein action

4088: functional for the Riemannian metric, the Yang--Mills action for the

4089: vector potentials, and the self interaction and the minimal coupling

4090: for the Higgs fields all appear with the correct signs in the

4091: asymptotic expansion for large $\Lambda$ of the number $N(\Lambda)$

4092: of eigenvalues of $D$ which are $\leq \Lambda$ (\cf \cite{C-C}),

4093: \begin{equation}

4094: N(\Lambda) = \# \ \hbox{eigenvalues of $D$ in} \

4095: [-\Lambda,\Lambda].

4096: \end{equation}

4097: 

4098: The spectral action principle, applied to a spectral triple

4099: $(\cA,\cH,D)$, can be stated as saying that the physical action

4100: depends only on $\Sp(D)\subset \R$. This spectral datum corresponds

4101: to the data $(\cH,D)$ of the spectral triple, independent of the

4102: action of $\cA$. Different $\cA$ that correspond to the same

4103: spectral data can be thought of as the noncommutative analog of

4104: isospectral Riemannian manifolds (\cf the discussion of isospectral

4105: deformations in Section \ref{isodef}). A natural expression for an

4106: action that depends only on $\Sp(D)$ and is {\em additive} for

4107: direct sums of spaces is of the form

4108: \begin{equation}\label{Sp-action}

4109: \Tr\, \chi (\frac{D}{\Lambda} )+ \langle \psi, D \, \psi \rangle,

4110: \end{equation}

4111: where $\chi$ is a positive even function and $\Lambda$ is a scale.

4112: 

4113: 

4114: In the case of the standard model, this formula \eqref{Sp-action} is

4115: applied to the full ``metric" including the internal fluctuations

4116: and gives the full standard model action minimally coupled with

4117: gravity. The Fermionic part of the action \eqref{Sp-action} gives

4118: (\cf \cite{C-C}, \cite{C-C2})

4119: \begin{equation}\label{fermionic}

4120: \langle \psi, D \, \psi \rangle= \int_X \left( \cL_{Gf} + \cL_{Hf}

4121: \right) \sqrt{|g|} d^4 x .

4122: \end{equation}

4123: 

4124: The bosonic part of  the action \eqref{Sp-action} evaluated via heat

4125: kernel invariants gives the standard model Lagrangian minimally

4126: coupled with gravity. Namely, one writes the function $\chi

4127: (\frac{D}{\Lambda})$  as the superposition of exponentials. One then

4128: computes the trace by a semiclassical approximation from local

4129: expressions involving the familiar heat equation expansion. This

4130: delivers all the correct terms in the action (\cf \cite{C-C2} for an

4131: explict calculation of all the terms involved).

4132: 

4133: Notice that here one treats the spacetime manifold $X$ in the

4134: Euclidean signature. The formalism of spectral triple can be

4135: extended in various ways to the Lorentzian signature (\cf \eg

4136: \cite{EHaw}). Perhaps the most convenient choice is to drop the

4137: self-adjointness condition for $D$ while still requiring $D^2$ to

4138: be self-adjoint.

4139: 

4140: \bigskip

4141: 

4142: 

4143: \section{Isospectral deformations}\label{isodef}

4144: 

4145: A very rich class of examples of noncommutative manifolds is

4146: obtained by considering isospectral deformations of a classical

4147: Riemannian manifold. These examples satisfy all the axioms of

4148: ordinary Riemannian geometry (\cf \cite{Co3}) except

4149: commutativity. They are obtained by the following result

4150: (Connes--Landi \cite{CLa}):

4151: 

4152: \begin{thm}\label{Theorem6}

4153: Let $M$ be a compact Riemannian spin manifold. Then if the

4154: isometry group of $M$ has rank $r \geq 2$, $M$ admits a

4155: non-trivial one parameter isospectral deformation to

4156: noncommutative geometries $M_{\theta}$.

4157: \end{thm}

4158: 

4159: \smallskip

4160: 

4161: The main idea of the construction is to deform the standard

4162: spectral triple describing the Riemannian geometry along a two

4163: torus embedded in the isometry group, to a family of spectral

4164: triples describing non-commutative geometries.

4165: 

4166: \smallskip

4167: 

4168: More precisely, under the assumption on the rank of the group of

4169: isometries of the compact spin manifold $X$, there exists a

4170: two-torus

4171: $$ T^2 \subset {\rm Isom}(X), $$

4172: where we identify $T^2 = \R^2 / (2\pi \Z)^2$. Let $U(s)$ be the

4173: unitary operators in this subgroup of isometries, for

4174: $s=(s_1,s_2)\in T^2$, acting on the Hilbert space $\cH = L^2(X,S)$

4175: of the spectral triple $$(C^\infty(X),L^2(X,S),D,J).$$

4176: Equivalently, we write $U(s)= \exp (i(s_1 P_1 + s_2 P_2))$, where

4177: $P_i$ are the corresponding Lie algebra generators, with ${\rm

4178: Spec}(P_i)\subset \Z$, satisfying $[D,P_i]=0$ and $P_i J = - J

4179: P_i$, so that $[U(s),D]=[U(s),J]=0$.

4180: 

4181: \smallskip

4182: 

4183: The action $\alpha_s(T)=U(s) T U(s)^{-1}$ has the following

4184: property. Any operator $T$ such that the map $s \mapsto

4185: \alpha_s(T)$ is smooth can be uniquely written as a norm

4186: convergent series

4187: \begin{equation}\label{T-bidegree}

4188: T = \sum_{n_1, n_2 \in \Z} \hat T_{n_1,n_2}

4189: \end{equation}

4190: where each term $\hat T_{n_1,n_2}$ is an operator of bi-degree

4191: $(n_1,n_2)$, that is,

4192: $$ \alpha_s(\hat T_{n_1,n_2})= \exp(i(s_1 n_1 + s_2 n_2)) \hat

4193: T_{n_1,n_2}, $$ for each $s=(s_1,s_2)\in T^2$, and the sequence of

4194: norms $\| \hat T_{n_1,n_2} \|$ is of rapid decay.

4195: 

4196: \smallskip

4197: 

4198: This property makes it possible to define left and right twists

4199: for such operators $T$, defined as

4200: \begin{equation}\label{T-left-twist}

4201: \ell (T):= \sum_{n_1,n_2} \hat T_{n_1,n_2} \exp\left( 2 \pi i

4202: \theta n_2 P_1 \right)

4203: \end{equation}

4204: and

4205: \begin{equation}\label{T-right-twist}

4206: r (T):=\sum_{n_1,n_2} \hat T_{n_1,n_2} \exp\left( 2 \pi i \theta

4207: n_1 P_2 \right).

4208: \end{equation}

4209: Both series still converge in norm, since the $P_i$ are

4210: self-adjoint operators.

4211: 

4212: \smallskip

4213: 

4214: It is then possible to introduce a (left) deformed product

4215: \begin{equation}\label{left-twist-prod}

4216: x * y = \exp ( 2 \pi i \theta n_1 ' n_2 ) xy,

4217: \end{equation}

4218: for $x$ a homogeneous operator of bi-degree $(n_1, n_2)$ and $y$ a

4219: homogeneous operator of bi-degree $(n_1', n_2')$. A (right)

4220: deformed product is similarly defined by setting $x *_r y = \exp (

4221: 2 \pi i \theta n_1  n_2' ) xy$. These deformed products satisfy

4222: $\ell(x)\ell(y)= x* y$ and $r(x)r(y)=x *_r y$.

4223: 

4224: \smallskip

4225: 

4226: The deformed spectral triples are then obtained by maintaining the

4227: same Hilbert space $\cH = L^2(X,S)$ and Dirac operator $D$, while

4228: modifying the algebra $C^\infty(X)$ to the non-commutative algebra

4229: $\cA_\theta :=\ell(C^\infty(X))$ and the involution $J$ that

4230: defines the real structure to $J_\theta := \exp(2\pi i \theta

4231: P_1P_2) J$.

4232: 

4233: 

4234: 

4235: \bigskip

4236: 

4237: \section{Algebraic deformations}\label{deform}

4238: 

4239: There is a very general context in which one constructs noncommutative

4240: spaces via deformations of commutative algebras. Unlike the

4241: isospectral deformations discussed in Section \ref{isodef}, here one

4242: proceeds mostly at a formal algebraic level, without involving the

4243: operator algebra structure and without invoking the presence of a

4244: Riemannian structure.

4245: 

4246: The idea of deformation quantization originates in the idea that

4247: classical mechanics has as setting a smooth manifold (phase space)

4248: with a symplectic structure, which defines a Poisson bracket

4249: $\{,\}$. The system is quantized by deforming the pointwise

4250: product in the algebra $\cA=\cC^\infty(M)$ (or in a suitable

4251: subalgebra) to a family $*_\hbar$ of products satisfying $f

4252: *_\hbar g \to fg$ as $\hbar\to 0$, which are associative but no

4253: longer necessarily commutative. These are also required to satisfy

4254: $$ \frac{f *_\hbar g - g*_\hbar f}{i\hbar} \to \{ f, g \}, $$

4255: as $\hbar\to 0$, namely, the ordinary product is deformed in the

4256: direction of the Poisson bracket. On the algebra $\cC^\infty(M)$ a

4257: Poisson bracket is specified by assigning a section $\Lambda$ of

4258: $\Lambda^2(TM)$ with the property that

4259: $$ \{ f, g \} = \langle \Lambda, df\wedge dg \rangle $$

4260: satisfies the Jacobi identity. Typically, this produces a {\it

4261: formal deformation}: a formal power series in $\hbar$. Namely, the

4262: deformed product can be written in terms of a sequence of

4263: bi--differential operators $B_k$ satisfying

4264: \begin{equation}

4265: f*g = fg + \hbar B_1(f,g) + \hbar^2 B_2(f,g) + \cdots

4266: \label{Poisson-deform}

4267: \end{equation}

4268: 

4269: \smallskip

4270: 

4271: Under this perspective, there is a good understanding of formal

4272: deformations. For instance, Kontsevich \cite{KontIHES97} proved

4273: that formal deformations always exist, by providing an explicit

4274: combinatorial formula that generates all the $\{ B_2, B_3, \ldots

4275: \}$ in the expansion from the $B_1$, hence in terms of the Poisson

4276: structure $\Lambda$. The formal solution (\ref{Poisson-deform})

4277: can then be written as

4278: $$ \sum_{n=0}^\infty \hbar^n \sum_{\Gamma\in G[n]} \omega_\Gamma

4279: B_{\Gamma,\Lambda}(f,g), $$ where $G[n]$ is a set of $(n(n+1))^n$

4280: labeled graphs with $n+2$ vertices and $n$ edges, $\omega_\Gamma$

4281: is a coeffcient obtained by integrating a differential form

4282: (depending on the graph $\Gamma$) on the configuration space of

4283: $n$ distinct points in the upper half plane, and

4284: $B_{\Gamma,\Lambda}$ is a bi--differential operator whose

4285: coefficients are derivatives of $\Lambda$ of orders specified by

4286: the combinatorial information of the graph $\Gamma$.

4287: 

4288: \smallskip

4289: 

4290: A setting of deformation quantization which is compatible with

4291: $C^*$--algebras was developed by Rieffel in \cite{Rieffel89}. We

4292: recall briefly Rieffel's setting. For simplicity, we restrict to

4293: the simpler case of a compact manifold.

4294: 

4295: \begin{defn}\label{Rieffel-def}

4296: A strict (Rieffel) deformation quantization of $\cA=\cC^\infty(M)$

4297: is obtained by assigning an associative product $*_\hbar$, an

4298: involution (depending on $\hbar$) and a $C^*$--norm $\| \cdot

4299: \|_{\hbar}$ on $\cA$, for $\hbar \in I$ (some interval containing

4300: zero), such that:

4301: 

4302: (i) For $\hbar=0$ these give the $C^*$--algebra $C(M)$,

4303: 

4304: (ii) For all $f,g\in \cA$, as $\hbar \to 0$,

4305: $$ \left\| \frac{f *_\hbar g - g*_\hbar f}{i\hbar} - \{ f, g \}

4306: \right\|_{\hbar} \to 0. $$

4307: 

4308: One denotes by $\cA_\hbar$ the $C^*$--algebra obtained by

4309: completing $\cA$ in the norm $\| \cdot \|_{\hbar}$.

4310: \end{defn}

4311: 

4312: The functions of $\hbar$ are all supposed to be analytic, so that

4313: formal power series expansions make sense.

4314: 

4315: \begin{rem}{\em The notion of a strict deformation quantization should be

4316: regarded as a notion of integrability for a formal solutions. }

4317: \end{rem}

4318: 

4319: \smallskip

4320: 

4321: Rieffel also provides a setting for compatible actions by a Lie

4322: group of symmetries, and proves that non--commutative tori (also

4323: of higher rank) are strict deformation quantizations of ordinary

4324: tori, that are compatible with the action of the ordinary torus as

4325: group of symmetry. Typically, for a given Poisson structure,

4326: strict deformation quantizations are not unique. This happens

4327: already in the case of tori.

4328: 

4329: \smallskip

4330: 

4331: In the same paper \cite{Rieffel89}, Rieffel uses a basic result of

4332: Wassermann \cite{Wassermann88} to produce an example where formal

4333: solutions are not integrable. The example is provided by the

4334: two--sphere $S^2$. There is on $S^2$ a symplectic structure, and a

4335: corresponding Poisson structure $\Lambda$ which is invariant under

4336: $SO(3)$. Rieffel proves the following striking result (Theorem 7.1

4337: of \cite{Rieffel89}):

4338: 

4339: \begin{thm} There are no $SO(3)$--invariant strict deformations of the

4340: ordinary product on $\cC^\infty(S^2)$ in the direction of the

4341: $SO(3)$--invariant Poisson structure.

4342: \end{thm}

4343: 

4344: In fact, the proof of this result shows more, namely that no

4345: $SO(3)$--invariant deformation of the ordinary product in $C(S^2)$

4346: can produce a non--commutative $C^*$--algebra. This rigidity

4347: result reflects a strong rigidity result for $SU(2)$ proved by

4348: Wassermann \cite{Wassermann88}, namely the only ergodic actions of

4349: $SU(2)$ are on von Neumann algebras of type I. The interest of

4350: this result lies in the fact that there are formal deformations of

4351: the Poisson structure that are $SO(3)$--invariant (see \eg

4352: \cite{GuVai}, \cite{BayFron}), but these only exist as a formal

4353: power series in the sense of (\ref{Poisson-deform}) and, by the

4354: results of Wassermann and Rieffel are not integrable.

4355: 

4356: \smallskip

4357: 

4358: Summarizing, we have the following type of phenomenon: on the one

4359: hand we have formal solutions, formal deformation quantizations

4360: about which a lot is known, but for which, in general, there may

4361: not be an integrability result. More precisely, when we try to

4362: pass from formal to actual solutions, there are cases where

4363: existence fails (the sphere), and others (tori) where uniqueness

4364: fails. The picture that emerges is remarkably similar to the case

4365: of formal and actual solutions of ordinary differential equations.

4366: 

4367: \smallskip

4368: 

4369: It is very instructive to build an analogy between the {\it problem

4370: of ambiguity} for formal solutions of ODE's and the present

4371: situation of formal non--commutative spaces and actual

4372: non--commutative spaces. The main conclusion to be drawn from this

4373: analogy is that there ought to be a {\it theory of ambiguity} which

4374: formulates precisely the relation between the formal

4375: non--commutative geometry and its integrated ($C^*$--algebraic)

4376: version.

4377: 

4378: \smallskip

4379: 

4380: To illustrate this concept, we take a closer look at the analogous

4381: story in the theory of ODE's. A good reference for a modern

4382: viewpoint is \cite{Ramis}. A formal solution of a differential

4383: equation is a power series expansion: for instance

4384: $\sum_{n=0}^\infty (-1)^n n! x^{n+1}$ is a formal solution of the

4385: Euler equation $x^2 y' +y=x$. Convergent series give rise to actual

4386: solutions, and more involved summation processes such as Borel

4387: summation can be used to transform  a given formal solution of an

4388: analytic ODE into an actual solution on a sufficiently narrow sector

4389: in $\C$ of sufficiently small radius, but such solution is in

4390: general {\it not unique}. It is known from several classical methods

4391: that some divergent series can  be ``summed'' modulo a function with

4392: exponential decrease of a certain order. This property (Gevrey

4393: summability) is also satisfied by formal solutions of analytic

4394: ODE's, and, stated in a more geometric fashion, it is essentially a

4395: cohomological condition. It also shows that, whereas on {\it small

4396: sectors} one has existence of actual solutions but not uniqueness,

4397: on {\it large sectors} one gains uniqueness, at the cost of possibly

4398: loosing existence. A complete answer to summability of formal

4399: solutions can then be given in terms of a more refined

4400: multi--summability (combining Gevrey series and functions of

4401: different order) and the Newton polygon of the equation.

4402: 

4403: \smallskip

4404: 

4405: The general flavor of this theory is surprisingly similar to the

4406: problem of formal solutions in non--commutative geometry. It is to

4407: be expected that an ambiguity theorem exists, which accounts for

4408: the cases of lack of uniqueness, or lack of existence, of actual

4409: solutions illustrated by the results of Rieffel.

4410: 

4411: 

4412: 

4413: 

4414: \medskip

4415: 

4416: Already in dealing with our first truly non--trivial example of

4417: noncommutative spaces, the noncommutative tori, we encountered

4418: subtleties related to the difference between the quotient and the

4419: deformation approach to the construction of non--commutative

4420: spaces.

4421: 

4422: \smallskip

4423: 

4424: In fact, the non--commutative tori we described in Section

4425: \ref{nctori} admit a description as algebras

4426: obtained as deformations of the ordinary product of functions, by

4427: setting

4428: \begin{equation}\label{NCTdeform}

4429:  (f*g)(x,y):= \left( e^{2\pi i\theta \frac{\partial}{\partial x}

4430: \frac{\partial}{\partial y'} } f(x,y) g(x',y') \right)_{x=x',y=y'} =

4431: \sum \frac{(i2\pi \theta)^n}{n!} D_1^n f D_2^n g.

4432: \end{equation}

4433: 

4434: \smallskip

4435: 

4436: Notice however that while $U\,\frac{\partial}{\partial U}$ and

4437: $V\,\frac{\partial}{\partial V}$ are derivations

4438:  for the algebra of the non--commutative torus, this is {\it not}

4439:  the case for

4440: $\frac{\partial}{\partial U}$ and $\frac{\partial}{\partial V}$. The

4441: same holds for the quantum plane (\cf \cite{[M]}) whose algebra of

4442: coordinates admits two generators $u,v$ with relation

4443: $$

4444: u\,v=\,q\,v\,u\,.

4445: $$

4446: These generators can be rotated ($u\mapsto \lambda

4447: \,u$, $v\mapsto \mu \,v$) without affecting the presentation but

4448: translations of the generators are not automorphisms of the algebra.

4449:  In other words, one can view the

4450: non--commutative torus as a deformation of an ordinary torus, which

4451: in turn is a quotient of the classical plane $\R^2$ by a lattice of

4452: translations, but the action of translations does not extend to the

4453: quantum plane. This is an instance of the fact that the general

4454: operations of quotient and deformation, in constructing

4455: non--commutative spaces, do not satisfy any simple compatibility

4456: rules and need to be manipulated with care.

4457: 

4458: Moreover, phenomena like the Morita equivalence between, for

4459: instance $\theta$ and $1/{\theta}$, are not  detectable in a purely

4460: deformation theoretic perturbative expansion like the one given by

4461: the Moyal product \eqref{NCTdeform}. They are non-perturbative and

4462: cannot be seen at the perturbative level of the star product.

4463: 

4464: \medskip

4465: 

4466: In this respect, a very interesting recent result is that of Gayral,

4467: Gracia-Bondia, Iochum, Sch\"ucker, and Varilly, \cite{GGISV}, where

4468: they consider a version of the structure of spectral triple for

4469: non-compact spaces. In that case, for instance, one no longer can

4470: expect the Dirac operator to have compact resolvent and one can only

4471: expect a local version to hold, \eg $a (D-i)^{-1}$ is compact for

4472: $a\in \cA$. Other properties of Definitions \ref{def-triple} and

4473: \ref{realstr} are easily adapted to a ``local version'' but become

4474: more difficult to check than in the compact case. They show that the

4475: Moyal product deformation of $\R^{2n}$ fits in the framework of

4476: spectral triples and provides an example of such non-unital spectral

4477: triples. Thus, it appears that the structure of noncommutative

4478: Riemannian geometry provided by spectral triples should adapt nicely

4479: to some classes of algebraic deformations.

4480: 

4481: \smallskip

4482: 

4483: It appears at first that spectral triples  may not be the right type

4484: of structure to deal with noncommutative spaces associated to

4485: algebraic deformations, because it corresponds to a form of

4486: Riemannian geometry, while many such spaces originate from K\"ahler

4487: geometry. However, the K\"ahler structure can often be also encoded

4488: in the setting of spectral triple, for example by considering also a

4489: second Dirac operator, as in \cite{BKLR} or through the presence of

4490: a Lefschetz operator as in \cite{CM1}.

4491: 

4492: \medskip

4493: 

4494: Noncommutative spaces obtained as deformations of commutative

4495: algebras fit in the context of a well developed algebraic theory of

4496: noncommutative spaces (\cf \eg \cite{Kontsevich93}

4497: \cite{Kontsevich00} \cite{Manin01RM} \cite{Manin91NC} \cite{[M]}

4498: \cite{Rosenberg98} \cite{Rosenberg95}, \cite{Soibelman01}). This

4499: theory touches on a variety of subjects like quantum groups and the

4500: deformation approach to non--commutative spaces and is interestingly

4501: connected to the theory of mirror symmetry. However, it is often not

4502: clear how to integrate this approach with the functional analytic

4503: theory of non--commutative geometry briefly summarized in section

4504: \ref{road}. Only recently, several results confirmed the existence

4505: of a rich interplay between the algebraic and functional analytic

4506: aspects of noncommutative geometry, especially through the work of

4507: Connes and Dubois-Violette (\cf \cite{CDV1}, \cite{CDV2}, \cite{CDV3}) 

4508: and of Polishchuk (\cf \cite{Poli}). Also, the work of Chakraborty and Pal

4509: \cite{ChakraPal} and Connes \cite{Co-Qgr} and more recently of van

4510: Suijlekom, Dabrowski, Landi, Sitarz, and Varilly \cite{SDLSV},

4511: \cite{SDLSV2} showed that quantum groups fit very nicely within the

4512: framework of noncommutative geometry described by spectral triples,

4513: contrary to what was previously belived. Ultimately, successfully

4514: importing tools from the theory of operator algebras into the realm

4515: of algebraic geometry might well land within the framework of what

4516: Manin refers to as a ``second quantization of algebraic geometry''.

4517: 

4518: 

4519: \bigskip

4520: 

4521: \section{Quantum groups}\label{qgroups}

4522: 

4523: 

4524: For a long time it was widely believed that quantum groups could not

4525: fit into the setting of noncommutative manifolds defined in terms of

4526: spectral geometry. On the contrary, recent work of Chakraborti and Pal

4527: showed in \cite{ChakraPal} that the quantum group

4528: $SU_q(2)$, for $0\leq q < 1$, admits a spectral triple with Dirac

4529: operator that is equivariant with respect to its own (co)action.

4530: 

4531: \smallskip

4532: 

4533: The algebra $\cA$ of functions on the quantum group $SU_q(2)$ is generated

4534: by two elements $\alpha$ and $\beta$ with the relations

4535: \begin{equation}\label{Qgr-alg}

4536: \begin{array}{c}

4537: \alpha^*\alpha +\beta^*\beta =1, \ \ \ \ \alpha\alpha^*+ q^2  \beta\beta^*

4538: =1, \\[2mm] \alpha\beta = q\beta\alpha, \ \ \  \alpha\beta^* =q

4539: \beta^*\alpha, \ \ \  \beta^*\beta =\beta\beta^*. \end{array}

4540: \end{equation}

4541: 

4542: \smallskip

4543: 

4544: By the representation theory of the quantum group $SU_q(2)$ (\cf

4545: \cite{KlimSchmu}) there exists a Hilbert space $\cH$ with orthonormal

4546: basis $e^{(n)}_{ij}$, $n\in \frac{1}{2} \N$, $i,j\in \{ -n,\ldots, n \}$, and

4547: a unitary representation

4548: \begin{equation}\label{Qgr-rep}

4549: \begin{array}{ll}

4550: \alpha\, e^{(n)}_{ij} = & a_+(n,i,j)\, e^{(n+1/2)}_{i-1/2, j-1/2}   +

4551: a_-(n,i,j)\, e^{(n-1/2)}_{i-1/2, j-1/2}  \\[4mm]

4552: \beta\, e^{(n)}_{ij} = & b_+(n,i,j)\, e^{(n+1/2)}_{i+1/2, j-1/2}   +

4553: b_-(n,i,j)\, e^{(n-1/2)}_{i+1/2, j-1/2},

4554: \end{array}

4555: \end{equation}

4556: with coefficients

4557: $$ \begin{array}{ll}

4558: a_+(n,i,j) = & q^{2n+i+j+1}\, Q(2n-2j+2,2n-2i+2,4n+2,4n+4) \\[2mm]

4559:  a_-(n,i,j) = & Q(2n+2j,2n+2i,4n,4n+2) \\[2mm]

4560:  b_+(n,i,j) = & -q^{n+j} \, Q(2n-2j+2,2n+2i+2,4n+2,4n+4) \\[2mm]

4561:  b_-(n,i,j) = & q^{n+i} \, Q(2n+2j,2n-2i,4n,4n+2),

4562: \end{array} $$

4563: where we use the notation

4564: $$ Q(n,m,k,r)=\frac{(1-q^{ n })^{1/2}(1-q^{ m })^{1/2}}{(1-q^{ k

4565: })^{1/2}(1-q^{ r })^{1/2}}. $$

4566: 

4567: \smallskip

4568: 

4569: Consider then, as in \cite{Co-Qgr}, the operator

4570: \begin{equation}\label{Qgr-Dirac}

4571: D\, e^{(n)}_{ij} = \left\{ \begin{array}{rr} -2n & n\neq i \\

4572: 2n & n=i. \end{array} \right.

4573: \end{equation}

4574: More generally, one can consider operators of the form

4575: $D\, e^{(n)}_{ij} = d(n,i)\, e^{(n)}_{ij}$, as in \cite{ChakraPal},

4576: with $d(n,i)$ satisfying the conditions

4577: $d(n+1/2,i+1/2)-d(n,i) =O(1)$ and $d(n+1/2,i-1/2)-d(n,i) = O(n+i+1)$.

4578: Then one has the following result (Chakraborti--Pal

4579: \cite{ChakraPal}):

4580: 

4581: \begin{thm}\label{Qgr-Sp3}

4582: The data $(\cA,\cH,D)$ ad above define an $SU_q(2)$ equivariant odd

4583: 3--summable spectral triple.

4584: \end{thm}

4585: 

4586: The equivariance condition means that there is an action on $\cH$ of

4587: the enveloping algebra $\cU=U_q(\SL(2))$, which commutes with the

4588: Dirac operator $D$. This is generated by operators

4589: $$ \begin{array}{rl} k\, e^{(n)}_{ij} =&  q^j \, e^{(n)}_{ij} \\[3mm]

4590: e\, e^{(n)}_{ij} =& q^{-n+1/2} (1-q^{2(n+j+1)})^{1/2}

4591: (1-q^{2(n-j)})^{1/2} (1-q^2)^{-1} \, e^{(n)}_{ij+1}, \end{array} $$

4592: satisfying the relations

4593: $$ ke=qek, \,\,\,\,  kf=q^{-1} fk, \,\,\,\,

4594: [e,f]=\frac{k^2-k^{-2}}{q-q^{-1}},

4595: $$

4596: with $f=e^*$, and with coproduct

4597: $$ \Delta(k)=k\otimes k, \,\,\,\, \Delta(e)=k^{-1}\otimes e + e\otimes

4598: k, \,\,\,\, \Delta(f) =k^{-1}\otimes f + f\otimes k. $$

4599: 

4600: \smallskip

4601: 

4602: It is interesting that, while the classical $SU(2)$ is of

4603: (topological and metric) dimension three, the topological dimension of

4604: the algebra $\cA$ of $SU_q(2)$ drops to one (\cf \cite{ChakraPal}),

4605: but the metric dimension of the spectral triple remains equal to

4606: three as in the classical case.

4607: 

4608: \smallskip

4609: 

4610: Chakraborti and Pal showed in \cite{ChakraPal} that the Chern

4611: character of the spectral triple is nontrivial. Moreover, Connes in

4612: \cite{Co-Qgr} gave an explicit formula for its local index cocycle,

4613: where a delicate calculation provides the cochain whose coboundary is

4614: the difference between the Chern character and the local version in

4615: terms of remainders in the rational approximation to the logarithmic

4616: derivative of the Dedekind eta function.

4617: 

4618: \smallskip

4619: 

4620: The local index formula is obtained by constructing a symbol map

4621: $$ \rho : \cB \to C^\infty(S^*_q), $$

4622: where the algebra $C^\infty(S^*_q)$ gives a noncommutative version of

4623: the cosphere bundle, with a restriction map

4624: $r:C^\infty(S^*_q) \to C^\infty(D_{q+}^2 \times D_{q-}^2)$ to the

4625: algebra of two noncommutative disks. Here

4626: $\cB$ is the algebra generated by the elements $\delta^k(a)$, $a\in

4627: \cA$, with $\delta(a) =[|D|,a]$. On the cosphere bundle there is a

4628: geodesic flow, induced by the group of automorphisms $a\mapsto e^{it

4629: |D|} \, a \, e^{-it |D|}$. Then $\rho(b)^0$ denotes the component of

4630: degree zero with respect to the grading induced by this flow.

4631: 

4632: \smallskip

4633: 

4634: The algebra $C^\infty(D^2_q)$ is an extension

4635: $$ 0 \to \cS \to C^\infty(D^2_q) \stackrel{\sigma}{\to} C^\infty(S^1)

4636: \to 0, $$

4637: where the ideal $\cS$ is the algebra of rapidly decaying

4638: matrices. There are linear

4639: functionals $\tau_0$ and $\tau_1$ on $C^\infty(D^2_q)$,

4640: $$ \tau_1(a) = \frac{1}{2\pi} \int_0^{2\pi} \sigma(a) d\theta, $$

4641: $$ \tau_0(a) = \lim_{N\to \infty} \sum_{k=0}^N \langle a\, \epsilon_k,

4642: \epsilon_k \rangle - \tau_1(a) N, $$

4643: where $\tau_0$ is defined in terms of the representation of

4644: $C^\infty(D^2_q)$ on the Hilbert space $\ell^2(\N)$ with o.n.~basis $\{

4645: \epsilon_k \}$.

4646: 

4647: \smallskip

4648: 

4649: Recall that (\cf \cite{Co94}) a cycle $(\Omega,d,\int)$ is a triple with

4650: where $(\Omega , d )$ is a  graded differential algebra,

4651: and $\int : \Omega^n \to \C$ is a closed graded trace on $\Omega$.

4652: A cycle over an algebra $\cA$ is given by a cycle $(\Omega , d , \int

4653: )$ together with a homomorphism $\rho :

4654: \cA \to \Omega^0$.

4655: 

4656: 

4657: In the case of the algebra $\cA$ of $SU_q(2)$, a cycle $(\Omega,d,\int)$

4658: is obtained in \cite{Co-Qgr} by considering $\Omega^1

4659: = \cA \oplus \Omega^{(2)}(S^1)$, with $\Omega^{(2)}(S^1)$ the space of

4660: weight two differential forms $f(\theta) d\theta^2$, with the

4661: $\cA$--bimodule structure

4662: $$ a\, (\xi,f)=(a\xi,\sigma(a) f) \ \ \ \  (\xi,f)\,a = (\xi a,

4663: -i\sigma(\xi) \sigma(a)' + f \sigma(a)), $$

4664: with differential

4665: $$ da = \partial a+ \frac{1}{2} \sigma(a)'' d\theta^2, $$

4666: with $\partial$ the derivation $\partial = \partial_\beta -

4667: \partial_\alpha$, and

4668: $$ \int \, (\xi,f) = \tau(\xi) + \frac{1}{2\pi i} \int f\, d\theta, $$

4669: where $\tau(a) = \tau_0(r_-(a^{(0)}))$, with $a^{(0)}$ the component

4670: of degree zero for $\partial$ and $r_-$ the restriction to

4671: $C^\infty(D^2_{q-})$. This definition of the cycle corrects for the

4672: fact that $\tau$ itself (as well as $\tau_0$) fails to be a trace.

4673: 

4674: \smallskip

4675: 

4676: The following result then holds (Connes \cite{Co-Qgr}):

4677: 

4678: \begin{thm}\label{loc-Qgr}

4679: \begin{enumerate}

4680: \item The spectral triple $(\cA,\cH,D)$ of Theorem \ref{Qgr-Sp3} has

4681: dimension spectrum $\Sigma=\{ 1,2,3 \}$.

4682: \item The residue formula for pseudodifferential operators $a\in

4683: \cB$ in terms of their symbol is given by

4684: $$ \begin{array}{ll} \cutint\,\, a \, |D|^{-3} = & (\tau_1\otimes \tau_1)

4685: (r\rho(a)^0) \\[2mm]

4686:  \cutint\,\, a \, |D|^{-2} = & (\tau_1\otimes\tau_0+\tau_0\otimes

4687: \tau_1)(r\rho(a)^0) \\[2mm]

4688: \cutint\,\, a \, |D|^{-1} = & (\tau_0\otimes\tau_0)(r\rho(a)^0)

4689: \end{array} $$

4690: \item The character $\chi(a_0,a_1)= \int a_0 \, da_1$ of the cycle

4691: $(\Omega,d,\int)$ is equal to the cocycle

4692: $$ \psi_1(a_0,a_1)= 2 \cutint a_0 \delta (a_1) P |D|^{-1} - \cutint a_0

4693: \delta^2 (a_1) P |D|^{-1}, $$

4694: with $P=(1+F)/2$. The local index formula is given by

4695: $$ \varphi_{odd} = \psi_1 + (b+B) \varphi_{even}, $$

4696: where $\varphi$ is the local index cocycle.

4697: \item The character $\Tr(a_0 [F,a_1])$ differs from the local form

4698: $\psi_1$ by the coboundary $b\psi_0$, with $\psi_0(a) = 2 \Tr (a P

4699: |D|^{-s})_{s=0}$. This cochain is determined by the values

4700: $\psi_0((\beta^*\beta)^n)$, which are of the form

4701: $$ \psi_0((\beta^*\beta)^n) = q^{-2n}(q^2 R_n(q^2) - G(q^2)), $$

4702: where $G$ is the logarithmic derivative of the Dedekind eta function

4703: \begin{equation}\label{DedEta}

4704:  \eta(q^2)=q^{1/12} \prod_{k=1}^\infty (1-q^{2k}),

4705: \end{equation}

4706: and the $R_n$ are rational functions with poles only at roots of

4707: unity.

4708: \end{enumerate}

4709: \end{thm}

4710: 

4711: More recently, another important breakthrough in the relation between

4712: quantum groups and the formalism of spectral triples was obtained by

4713: Walter van Suijlekom, Ludwik Dabrowski, Giovanni

4714: Landi, Andrzej Sitarz, Joseph C. Varilly, in \cite{SDLSV} and

4715: \cite{SDLSV2}.

4716: 

4717: They construct a $3^+$ summable spectral triple $(\cA,\cH,D)$,

4718: where $\cA$ is, as before, the algebra of coordinates of the

4719: quantum group $SU_q(2)$. The geometry in this case is an isospectral

4720: deformation of the classical case, in the sense that the Dirac

4721: operator is the same as the usual Dirac operator for the round metric

4722: on the ordinary 3-sphere $S^3$. Moreover, the spectral triple

4723: $(\cA,\cH,D)$ is especially nice, in as it is equivariant with

4724: respect to both left and right action of the Hopf algebra

4725: $\cU_q(su_q(2))$.

4726: 

4727: The classical Dirac operator for the round metric on $S^3$ has

4728: spectrum $\Sigma=\Sigma_+ \cup \Sigma_-$ with $\Sigma_+=\{ (2j+3/2):

4729: j=0,1/2,1,3/2,\ldots \}$ with multiplicities $(2j+1)(2j+2)$ and

4730: $\Sigma_-=\{ -(2j+1/2): j=1/2,1,3/2,\ldots\}$ with multiplicities

4731: $2j(2j+1)$. The Hilbert space is obtained by taking $V\otimes \C^2$,

4732: where $V$ is the left regular representation of $\cA$. It is very

4733: important here to take $V\otimes \C^2$ instead of $\C^2\otimes V$.

4734: Not only the latter violates the equivariance condition, but it was

4735: shown by Ghoswami that it produces unbounded commutators $[D,a]$,

4736: hence one does not obtain a spectral triple in that way.

4737: 

4738: The spectral triple contructed in \cite{SDLSV} and \cite{SDLSV2} has

4739: a real structure $J$  and the Dirac operator satisfies a weak form

4740: of the ``order one condition'' (\cf Section \ref{singconic} above).

4741: The local index formula of \cite{Co-Qgr} (\cf Theorem \ref{loc-Qgr}

4742: above) extends to the spectral triple of \cite{SDLSV}, as proved in

4743: \cite{SDLSV2} and the structures of the cotangent space and the

4744: geodesic flow are essentially the same.

4745: 

4746: 

4747: 

4748: 

4749: 

4750: \bigskip

4751: \bigskip

4752: 

4753: \section{Spherical manifolds}\label{ncspheres}

4754: 

4755: The noncommutative spheres $S^3_\varphi\subset{\mathbb R}^4_\varphi$

4756: are obtained as solutions of a very simple problem, namely the vanishing

4757: of the first component of the Chern character of a unitary $U\in

4758: M_2(\cA)$ where $\cA$ is the algebra of functions on the sphere and

4759: the Chern character is taken in the cyclic homology (b,B) bicomplex.

4760: The origin of this problem is to quantize the volume form of a three

4761: manifold (\cf \cite{CDV1}). The solutions are parameterized by three

4762: angles $\varphi_k$, $k\in \{1,2,3\}$ and the corresponding algebras

4763: are obtained by imposing the ``unit sphere relation"

4764: \begin{equation}

4765: \sum \,x_\mu^2=\,1

4766: \end{equation}

4767: to the four generators $x_0, x_1, x_2, x_3$ of the quadratic algebra

4768:  $C_{\mathrm{alg}}({\mathbb R}^4_\varphi)$

4769: with the six relations

4770: 

4771: \smallskip

4772: \begin{equation}

4773: \label{pres1} \sin (\varphi_k) \, [x_0 , x_k]_+ = i\; \cos

4774: (\varphi_{\ell} - \varphi_m) \, [x_{\ell} , x_m]

4775: \end{equation}

4776: \begin{equation}

4777: \label{pres2} \cos (\varphi_k) \, [x_0 , x_k] = i \;\sin

4778: (\varphi_{\ell} - \varphi_m) \, [x_{\ell} , x_m]_+ \, ,

4779: \end{equation}

4780: 

4781: \medskip

4782: 

4783: where  $ [a , b]_+=\,a\,b+\,b\,a$ is the anticommutator and by

4784: convention the indices $k,l,m\in \{1,2,3\}$ always appear in cyclic

4785: order.

4786: 

4787: The analysis of these algebras is a special case of the general

4788: theory of central quadratic forms for quadratic algebras developed

4789: in \cite{CDV2}, \cite{CDV3} and which we briefly recall below.

4790: 

4791: Let $\cala=A(V,R)=T(V)/(R)$  be a quadratic algebra where $V$ is the

4792: linear span of the generators and $(R)\subset T(V)$ the ideal

4793: generated by the relations. The geometric data $\{E\,,\,

4794: \sigma\,,\,\call\}$ is given by an algebraic variety $E$, a

4795: correspondence $\sigma$ on $E$ and a line bundle $\call$ over $E$.

4796: These data are defined so as to yield an homomorphism $h$ from

4797: $\cala$ to a crossed product algebra constructed from sections of

4798: powers of the line bundle $\call$ on the graphs of the iterations of

4799: the correspondence $\sigma$. This crossed product only involves the

4800: positive powers of the correspondence $\sigma$ and thus remains

4801: ``triangular" and far remote from the ``semi-simple" set-up of

4802: $C^*$-algebras.

4803: 

4804: This morphism  $h$ can be considerably refined using the notion of

4805: positive central quadratic form.

4806: 

4807: 

4808: \begin{defn} \label{cent}

4809: Let $Q \in S^2(V)$ be a symmetric bilinear form on $V^\ast$ and $C$

4810: a component of $E \times E$. We   say that $Q$ is

4811: \underline{central} on $C$ iff for all ($Z,\,Z'$) in $C$ and

4812: $\omega\in R$ one has,

4813: \begin{equation} \label{defcentral}

4814: \omega(Z,Z')\, Q(\sigma(Z'),\sigma^{-1}(Z))+Q(Z,Z')\,

4815: \omega(\sigma(Z'),\sigma^{-1}(Z)) =0

4816: \end{equation}

4817: \end{defn}

4818: 

4819: 

4820: This makes it possible to construct purely algebraically a crossed product

4821: algebra and an homomorphism from $\cala=A(V,R)$ to this crossed

4822: product \cite{CDV2}, \cite{CDV3}. 

4823: The relation with $C^*$-algebras arises from

4824: positive central quadratic forms which make sense on involutive

4825: quadratic algebras.

4826: 

4827: Let $\cala=A(V,R)$ be an {\sl involutive} quadratic algebra \ie an

4828: algebra over $\C$ which is a

4829:  $\ast$-algebra with  involution $x\mapsto x^\ast$ preserving the subspace

4830:   $V$ of the generators.

4831: The real structure of $V$ is given by the antilinear involution

4832: $v\mapsto j( v)$ restriction of $x\mapsto x^\ast$. As

4833: $(xy)^\ast=y^\ast x^\ast$ for $x,y\in \cala$, the space  $R$ of

4834: relations fulfills

4835: \begin{equation}

4836: (j \otimes j)( R)=t(R) \label{eq5.3}

4837: \end{equation}

4838:  in $V\otimes V$ where $t:V\otimes V\rightarrow V\otimes V$ is the

4839: transposition $v\otimes w \mapsto t(v\otimes w)=w\otimes v$. This

4840: implies that the characteristic variety is stable under the

4841: involution $j$ and one has

4842: $$

4843: \sigma( j(Z)) =\, j( \sigma^{-1}(Z))

4844: $$

4845: 

4846: 

4847: 

4848: \smallskip

4849: 

4850: Let then  $C$ be an invariant component of $E \times E$, we say that

4851: $C$ is $j$-{\em real} when it is globally invariant under the

4852: involution

4853: \begin{equation}\label{tildej}

4854: \tilde j(Z,\,Z'):=( j( Z'),\, j( Z))

4855: \end{equation}

4856: Let then $Q$ be a central qudratic form on $C$, we say that $Q$ is

4857: positive on $C$ iff

4858: 

4859: $$

4860: Q(Z,j(Z))> 0 \qqq Z\in K\,.

4861: $$

4862: 

4863: 

4864: \smallskip One can then endow the line bundle  $\call$ dual of the tautological bundle

4865: on $P(V^*)$ with the hermitian metric given by

4866: \begin{equation}\label{herm}

4867: \langle f\,L,\,g\, L'\rangle_Q(Z) =\,f(Z)\,\overline{g(Z)} \,\frac{

4868: L(Z)\,\overline{ L'(Z)}}{Q(Z,\,j(Z))} \qquad L, L' \in V,\quad Z \in

4869: K\, \,.

4870: \end{equation}

4871: ($\forall f,g \in C(K)$)

4872: 

4873: \medskip One then defines a generalized crossed product $C^*$-algebra

4874:  $C(K) \times_{\sigma,\,\call}

4875: \mathbb {Z} $ following M. Pimsner \cite{pims:1997}. Given a compact

4876: space $K$, an homeomorphism $\sigma$ of $K$ and a hermitian line

4877: bundle $\call$ on $K$ we define the  $C^\ast$-algebra $C(K)

4878: \times_{\sigma,\,\call} \mathbb {Z} $ as the twisted cross-product

4879: of $C(K)$ by the Hilbert $C^*$-bimodule associated to $\call$ and

4880: $\sigma$ (\cite{aba-eil-exel:1998}, \cite{pims:1997}).

4881: 

4882: We let for each $n \geq 0$, $\call^{\sigma^n}$ be the hermitian line

4883: bundle pullback of $\call$ by $\sigma^n$ and (cf.

4884: \cite{art-tat-vdb:1990}, \cite{smi-sta:1992})

4885: \begin{equation}

4886: \call_n := \call \otimes \call^{\sigma} \otimes \cdots \otimes

4887: \call^{\sigma^{n-1}} \label{gene2}

4888: \end{equation}

4889: We first define a $\ast$-algebra as the linear span of the monomials

4890: \begin{equation}

4891: \xi \, W^n\, , \quad W^{\ast n} \, \eta^\ast \,,\quad \xi\,,\eta \in

4892: C(K,\call_n) \label{gene}

4893: \end{equation}

4894: with product given as in (\cite{art-tat-vdb:1990},

4895: \cite{smi-sta:1992}) for $(\xi_1 \, W^{n_1})\,(\xi_2 \, W^{n_2})$ so

4896: that

4897: \begin{equation}

4898: (\xi_1 \, W^{n_1})\,(\xi_2 \, W^{n_2}):= (\xi_1 \otimes

4899: (\xi_2\circ{\sigma^{n_1}}) )\, W^{n_1+n_2} \label{gene3}

4900: \end{equation}

4901: We use the hermitian structure of $\call_n $ to give meaning to the

4902: products $\eta^\ast \,\xi$ and $\xi \;\eta^\ast$ for $\xi\,,\eta \in

4903: C(K,\call_n)$. The product then extends uniquely to an associative

4904: product of $\ast$-algebra fulfilling the following additional rules

4905: \begin{equation}

4906: (W^{\ast k} \, \eta^\ast)\,( \xi \, W^k):= \, (\eta^\ast\, \xi)\circ

4907: \sigma^{-k}\,,\qquad ( \xi \, W^k)\,(W^{\ast k} \, \eta^\ast)\,:= \,

4908: \xi \;\eta^\ast \label{gene1}

4909: \end{equation}

4910: 

4911: 

4912:  The $C^\ast$-norm of $C(K) \times_{\sigma,\,\call} \mathbb {Z} $

4913:  is defined as for ordinary cross-products and due to the amenability of the group $\mathbb {Z} $

4914: there is no distinction between the reduced and maximal norms. The

4915: latter is obtained as the supremum of the norms in involutive

4916: representations in Hilbert space. The natural positive conditional

4917: expectation on the subalgebra $C(K)$ shows that the $C^\ast$-norm

4918: restricts to the usual sup norm on $C(K)$.

4919: 

4920: 

4921: 

4922: \medskip

4923: 

4924: 

4925: \begin{thm}\label{C*}

4926:  Let $K \subset E$ be a compact

4927: $\sigma$-invariant subset and $Q$ be central and strictly positive

4928: on $\{(Z,\,\bar Z);\, Z\in K\}$. Let $\call$ be the restriction to

4929: $K$ of the dual of the tautological line bundle on $P(V^\ast)$

4930: endowed with the hermitian metric $\langle\;,\; \rangle_Q$.

4931: 

4932: 

4933:  (i) The equality $\sqrt{2}\,\theta(Y):= Y\, W + W^\ast\,\bar Y^\ast$

4934: yields a $\ast$-homomorphism $$\theta:\cala=A(V,R) \to C(K)

4935: \times_{\sigma,\,\call} \mathbb {Z} $$

4936: 

4937: 

4938:  (ii) For any $Y \in V$

4939: the $C^\ast$-norm of $\theta(Y)$ fulfills

4940: $$\sup_K \|Y\|\leq \sqrt{2}\| \,\theta(Y)\|

4941: \leq 2\sup_K \|Y\| $$

4942: 

4943: 

4944:  (iii) If $\sigma^4 \neq \bbbone$, then $\theta(Q)= 1$ where

4945: $Q$ is viewed as an element of $T(V)/(R)$.

4946: \end{thm}

4947: 

4948: \bigskip

4949: 

4950: 

4951: In the above case of the sphere $S^3_\varphi$ one lets $Q$ be the

4952: quadratic form

4953: \begin{equation}

4954:  Q(x,\,x'):=\sum x_\mu\,x'_\mu

4955: \label{quad}

4956: \end{equation}

4957: 

4958: In the generic case one has :

4959: 

4960: \begin{prop}  1) The characteristic variety is the union of 4 points with an elliptic curve $F_\varphi$.

4961: 

4962: 2) The quadratic form  $Q$ is central and positive on $F_\varphi

4963: \times F_\varphi$.

4964: \end{prop}

4965: 

4966: \smallskip

4967: In suitable coordinates the equations defining the elliptic curve

4968: $F_\varphi$ are

4969: \begin{equation}

4970: \frac{Z_0^2-Z_1^2}{

4971: s_1}=\frac{Z_0^2-Z_2^2}{s_2}=\frac{Z_0^2-Z_3^2}{s_3} \label{charZ}

4972: \end{equation}

4973: where $s_k := 1 + t_\ell\, t_m \, , \: t_k:= {\rm tan}\,\varphi_k$.

4974: 

4975: The positivity of $Q$ is automatic since in the coordinates  $x$ the

4976: involution $j_\varphi$

4977:  of the $\ast$-algebra $C_{\mathrm{alg}}({\mathbb R}^4_\varphi)$

4978: is just $j_\varphi(Z)= \bar Z$, so that $Q(X,\,j_\varphi(X))>0$ for

4979: $X \neq 0$.

4980: 

4981: 

4982: \begin{cor} \label{II}

4983: Let $K \subset F_\varphi $  be a compact $\sigma$-invariant subset.

4984: The homomorphism $\theta$ of Theorem \ref{C*} is a unital

4985:  $\ast$-homomorphism from $ C_{\mathrm{alg}}(S^3_\varphi)$

4986: to the cross-product $ C^{\infty}(K) \times_{\sigma,\,\call} \mathbb

4987: {Z} $.

4988: \end{cor}

4989: 

4990: 

4991: 

4992: 

4993:  It follows  that one obtains a

4994: non-trivial $C^\ast$-algebra $C^\ast(S^3_\varphi)$ as the completion

4995: of  $ C_{\mathrm{alg}}(S^3_\varphi)$ for the semi-norm,

4996: \begin{equation}

4997:  \| P \|:= \sup \| \,\pi(P) \|

4998: \label{norm}

4999: \end{equation}

5000: where $\pi$ varies through all unitary representations of $

5001: C_{\mathrm{alg}}(S^3_\varphi)$. It was clear from the start that

5002: \eqref{norm} defines a finite $C^\ast$-semi-norm on $

5003: C_{\mathrm{alg}}(S^3_\varphi)$ since the equation of the sphere

5004: $\sum x_\mu^2=1$ together with the self-adjointness

5005: $x_\mu=\,x_\mu^*$ show that in any unitary representation one has

5006: $$

5007: \|\, \pi(x_\mu) \|\leq 1\qqq \mu\,.

5008: $$

5009: What the above corollary gives is a lower bound for the

5010: $C^\ast$-norm such as that given by statement (ii) of Theorem

5011: \ref{C*} on the linear subspace $V$ of generators.

5012: 

5013: 

5014: \smallskip The correspondence

5015: $\sigma$ on $F_\varphi $,

5016:  is for generic  $\varphi$ a translation of module $\eta$ of the elliptic curve

5017:  $F_\varphi$ and one distinguishes two cases :  the {\em even} case when it preserves

5018:  the two real components of the curve $F_\varphi \cap

5019: P_3(\mathbb {R})$ and the odd case when it permutes them.

5020: 

5021: \begin{prop} \label{siginvar} Let $\varphi  $ be generic and even.

5022: 

5023: (i) The cross-product $ C(F_\varphi) \times_{\sigma,\,\call} \mathbb

5024: {Z} $ is isomorphic to the mapping torus of the automorphism $\beta

5025: $ of the noncommutative torus ${\mathbb T}_{\eta}^2 = C_\varphi

5026: \times_\sigma \mathbb {Z} $ acting on the generators by the  matrix

5027: $\left[

5028: \begin{array}{cc}

5029: 1& 4\\

5030: 0& 1

5031: \end{array}

5032: \right] $.

5033: 

5034: 

5035: (ii) The crossed product $F_\varphi \times_{\sigma,\,\call} \mathbb

5036: {Z} $ is a noncommutative $3$-manifold with an elliptic

5037:  action of the three dimensional

5038: Heisenberg Lie algebra $\frach_3$ and an invariant trace $\tau$.

5039: 

5040: 

5041: \end{prop}

5042: 

5043: \medskip

5044: 

5045: It follows that one is exactly in the framework developed in

5046: \cite{C3}. We refer to \cite{Rieffel89} and \cite{aba-exel:1997}

5047: where these noncommutative manifolds were analyzed in terms of

5048: crossed products by Hilbert $C^*$-bimodules.

5049: 

5050: 

5051:  Integration on the translation invariant volume form $dv$

5052: of $F_\varphi$ gives the $\frach_3$-invariant trace $\tau$,

5053: \begin{eqnarray}

5054:  \label{trace}

5055: \tau(f)& = & \int f dv\,,\quad \forall f \in C^{\infty}(F_\varphi)\nonumber\\

5056: \tau(\xi \, W^k)& = &\tau(W^{\ast k} \, \eta^\ast)\,=\,0\,,\quad

5057: \forall k\neq 0

5058:  \end{eqnarray}

5059: It follows in particular that the results of \cite{C3} apply to

5060: obtain the calculus. In particular the following gives the

5061: ``fundamental class" as a $3$-cyclic cocycle,

5062: \begin{equation}\label{3trace}

5063: \tau_3(a_0,\,a_1,\,a_2 ,\,a_3)=\,\sum

5064: \epsilon_{ijk}\,\tau(a_0\,\delta_i(a_1)\,\delta_j(a_2)

5065: \,\delta_k(a_3))

5066: \end{equation}

5067: where the $\delta_j$ are the generators of the action of $\frach_3$.

5068: 

5069: 

5070: \medskip The relation between the noncommutative spheres

5071: $S^3_\varphi$ and the noncommutative nilmanifolds $F_\varphi

5072: \times_{\sigma,\,\call} \mathbb {Z} $ is analyzed in 

5073: \cite{CDV2}, \cite{CDV3} thanks to the computation of the 

5074: Jacobian of the homomorphism $\theta$.

5075: 

5076: \bigskip

5077: 

5078: 

5079: 

5080: 

5081: \section{$\Q$-lattices}\label{Qlatt}

5082: 

5083: A class of examples of noncommutative spaces of relevance to

5084: number theory is given by the moduli spaces of $\Q$-lattices up to

5085: commensurability. These fall within the general framework of

5086: noncommutative spaces obtained as quotients of equivalence relations

5087: discussed in Section \ref{quotients}.

5088: 

5089: A $\Q$-lattice in $\R^n$ consists of a pair $( \Lambda , \phi) $

5090: of a lattice $\Lambda\subset \R^n$ (a cocompact free abelian

5091: subgroup of $\R^n$ of rank $n$) together with a system of labels

5092: of its torsion points given by a homomorphism of abelian groups

5093: \begin{equation}\label{phimap}

5094:  \phi :  \Q^n/\Z^n \longrightarrow \Q\Lambda / \Lambda.

5095: \end{equation}

5096: 

5097: Two $\Q$-lattices are commensurable,

5098: $$ (\Lambda_1, \phi_1) \sim (\Lambda_2, \phi_2), $$

5099: iff $\Q\Lambda_1=\Q\Lambda_2$ and

5100: $$ \phi_1 = \phi_2 \mod \Lambda_1 + \Lambda_2 $$

5101: 

5102: In general, the map $\phi$ of \eqref{phimap} is just a group

5103: homomorphism. A $\Q$-lattice is said to be {\em invertible}

5104: is $\phi$ is an isomorphism. Two invertible $\Q$-lattices are

5105: commensurable if and only if they are equal.

5106: 

5107: \begin{center}

5108: \begin{figure}

5109: \includegraphics[scale=0.95]{Qlat.eps}

5110: \caption{$\Q$-lattices: generic and invertible case.

5111: \label{FigQlat}}

5112: \end{figure}

5113: \end{center}

5114: 

5115: The space $\cL_n$ of commensurabilty classes of

5116: $\Q$-lattices in $\R^n$ has the typical property of noncommutative

5117: spaces: it has the cardinality of the continuum but one cannot

5118: construct a countable collection of measurable functions that

5119: separate points of $\cL_n$.

5120: Thus, one can use noncommutative geometry to describe the quotient

5121: space $\cL_n$ through a noncommutative $C^*$-algebra $C^*(\cL_n)$.

5122: 

5123: We consider especially the case of $n=1$ and $n=2$. One is also

5124: interested in the $C^*$-algebras describing $\Q$-lattices up to

5125: scaling, $\cA_1=C^*(\cL_1/\R_+^*)$ and $\cA_2=C^*(\cL_2/\C^*)$.

5126: 

5127: In the 1-dimensional case, a $\Q$-lattice can always be written in the

5128: form

5129: \begin{equation}\label{1dimQlat}

5130:  ( \Lambda , \phi) \, = (\lambda\, \Z,\lambda\,\rho)

5131: \end{equation}

5132: for some $\lambda>0$ and some

5133: \begin{equation}\label{rho}

5134: \rho \in \Hom(\Q/\Z,\Q/\Z)=\varprojlim \Z/n\Z = \hat\Z.

5135: \end{equation}

5136: By considering lattices up to scaling, we eliminate the factor

5137: $\lambda>0$ so that 1-dimensional $\Q$-lattices up to scale are

5138: completely specified by the choice of the element $\rho\in

5139: \hat\Z$. Thus, the algebra of coordinates of the space of

5140: 1-dimensional $\Q$-lattices up to scale is the commutative

5141: $C^*$-algebra

5142: \begin{equation}\label{ChatZ}

5143: C(\hat\Z)\simeq C^*(\Q/\Z),

5144: \end{equation}

5145: where we use Pontrjagin duality to get the identification in

5146: \eqref{ChatZ}.

5147: 

5148: The equivalence relation of commensurability is implemented by the

5149: action of the semigroup $\N^\times$ on $\Q$-lattices. The

5150: corresponding action on the algebra \eqref{ChatZ} is by

5151: \begin{equation}\label{Nact}

5152:  \alpha_n(f) (\rho)=\left\{ \begin{array}{lr} f(n^{-1} \rho) & \rho \in

5153:  n\hat\Z \\ 0 & \text{otherwise.} \end{array}\right.

5154: \end{equation}

5155: 

5156: Thus, the quotient of the space of 1-dimensional $\Q$-lattices up

5157: to scale by the commensurability relation and its algebra of

5158: coordinates is given by the semigroup crossed product

5159: \begin{equation}\label{semicross}

5160:  C^*(\Q/\Z)\rtimes \N^\times .

5161: \end{equation}

5162: This is the Bost--Connes $C^*$-algebra introduced in \cite{BC}.

5163: 

5164: It has a natural time evolution given by the covolume of a pair of

5165: commensurable $\Q$-lattices. It has symmetries (compatible with the

5166: time evolution) given by the group $\hat\Z^*=\GL_1(\A_f)/\Q^*$ and

5167: the KMS (Kubo--Martin--Schwinger) equilibrium states of the system

5168: have interesting arithmetic properties. Namely, the partition function

5169: of the system is the Riemann zeta function. There is a unique KMS

5170: state for sufficiently high temperature, while at low temperature the

5171: system undergoes a phase transition with spontaneous symmetry

5172: breaking. The pure phases (estremal KMS states) at low temperature

5173: are parameterized by elements in $\hat\Z^*$. They have an explicit

5174: expression in terms of polylogarithms at roots of unity. At zero

5175: temperature the extremal KMS states, evaluated on the elements of a

5176: rational subalgebra affect values that are algebraic numbers.

5177: The action on these values of the Galois group $\Gal(\bar\Q/\Q)$

5178: factors through its abelianization and

5179: is obtained (via the class field theory isomorphism $\hat\Z^* \cong

5180: \Gal(\Q^{ab}/\Q)$) as the action of symmetries on the algebra (\cf

5181: \cite{BC}, \cite{CoMa}, \cite{CoMajgp1} for details).

5182: 

5183: \medskip

5184: 

5185: In the 2-dimensional case, a $\Q$-lattice can be written in the form

5186: $$ (\Lambda,\phi)=(\lambda (\Z+\Z\tau),\lambda\rho), $$

5187: for some $\lambda\in \C^*$, some $\tau\in \H$, and some $\rho\in

5188: M_2(\hat\Z)=\Hom(\Q^2/\Z^2,\Q^2/\Z^2)$. Thus, the space of

5189: 2-dimensional $\Q$-lattices up to the scale factor $\lambda\in

5190: \C^*$ and up to isomorphisms, is given by

5191: \begin{equation}\label{2dQlat}

5192:  M_2(\hat\Z)\times \H  \mod \Gamma=\SL_2(\Z).

5193: \end{equation}

5194: The commensurability relation giving the

5195: space $\cL_2/\C^*$ is implemented by the partially

5196: defined action of $\GL_2^+(\Q)$.

5197: 

5198: One considers in this case the quotient of the space

5199: \begin{equation}\label{tildeU}

5200: \tilde \cU:= \{ (g,\rho,\alpha)\in \GL_2^+(\Q)\times

5201: M_2(\hat\Z)\times \GL_2^+(\R)\, :\,\, g\rho\in M_2(\hat\Z) \}

5202: \end{equation}

5203: by the action of $\Gamma \times \Gamma$ given by

5204: \begin{equation}\label{Gamma2action}

5205: (\gamma_1,\gamma_2) \, (g,\rho,\alpha)=(\gamma_1 g \gamma_2^{-1},

5206: \gamma_2 \rho, \gamma_2 \alpha).

5207: \end{equation}

5208: 

5209: The groupoid $\cR_2$ of the equivalence relation of

5210: commensurability on 2-dimensional $\Q$-lattices (not considered up

5211: to scaling for the moment) is a locally compact groupoid, which

5212: can be parameterized by the quotient of \eqref{tildeU} by

5213: $\Gamma\times \Gamma$ via the map $r: \tilde \cU \to \cR_2$,

5214: \begin{equation}\label{mapgr}

5215: r(g,\rho,\alpha)=\left( (\alpha^{-1}g^{-1} \Lambda_0, \alpha^{-1}

5216: \rho), (\alpha^{-1}\Lambda_0, \alpha^{-1} \rho)\right).

5217: \end{equation}

5218: 

5219: We then consider the quotient by scaling.

5220: The quotient $\GL_2^+(\R)/\C^*$ can be identified with the

5221: hyperbolic plane $\H$ in the usual way.

5222: If $(\Lambda_k,\phi_k)$ $k=1,2$ are a pair of commensurable

5223: 2-dimensional $\Q$-lattices, then for any $\lambda\in \C^*$, the

5224: $\Q$-lattices $(\lambda \Lambda_k, \lambda \phi_k)$ are also

5225: commensurable, with

5226: $$ r(g,\rho,\alpha\lambda^{-1})=\lambda r(g,\rho,\alpha). $$

5227: However, the action of $\C^*$ on $\Q$-lattices is not free due to

5228: the presence of lattices (such as $\Lambda_0$ above) with nontrivial

5229: automorphisms. Thus, the quotient $Z=\cR_2/\C^*$ is no longer a

5230: groupoid. Still, one can define a convolution algebra for $Z$ by

5231: restricting the convolution product of $\cR_2$ to homogeneous

5232: functions of weight zero, where a function $f$ has weight $k$ if

5233: it satisfies

5234: $$ f(g,\rho,\alpha\lambda)=\lambda^k f(g,\rho,\alpha), \ \ \

5235: \forall \lambda \in \C^*. $$

5236: The space $Z$ is the quotient of the space

5237: \begin{equation}\label{Uspace}

5238:  \cU:=\{ (g,\rho,z) \in \GL_2^+(\Q)\times M_2(\hat\Z)\times \H |

5239: g\rho\in M_2(\hat\Z) \}

5240: \end{equation}

5241: by the action of $\Gamma\times \Gamma$. Here the space

5242: $M_2(\hat\Z)\times \H$ has a partially defined action of

5243: $\GL_2^+(\Q)$ given by

5244: $$ g (\rho,z)=(g\rho, g(z)), $$

5245: where $g(z)$ denotes action as fractional linear transformation.

5246: 

5247: Thus, the algebra of coordinates $\cA_2$ for the noncommutative space

5248: of commensurability classes of 2-dimensional $\Q$-lattices up

5249: to scaling is given by the following convolution algebra.

5250: 

5251: Consider the space $C_c(Z)$ of continuous compactly supported

5252: functions on $Z$. These can be seen, equivalently, as functions on

5253: $\cU$ as in \eqref{Uspace} invariant under the

5254: $\Gamma\times\Gamma$ action $(g,\rho,

5255: z)\mapsto(\gamma_1g\gamma_2^{-1},\gamma_2 z)$.

5256: One endows $C_c(Z)$ with the convolution product

5257: \begin{equation}\label{Heckeprod2}

5258: (f_1*f_2)(g,\rho,z)= \displaystyle{\sum_{s\in \Gamma\backslash

5259: \GL_2^+(\Q): s\rho\in M_2(\hat\Z)}} f_1(gs^{-1},s\rho,s(z))

5260: f_2(s,\rho,z)

5261: \end{equation}

5262: and the involution

5263: $f^*(g,\rho,z)=\overline{f(g^{-1},g\rho,g(z))}$.

5264: 

5265: Again there is a time evolution on this algebra, which is given by the

5266: covolume,

5267: \begin{equation}\label{evolution2}

5268: \sigma_t(f) (g,\rho,z) = \det(g)^{it} \, f(g,\rho,z).

5269: \end{equation}

5270: The partition function for this $\GL_2$ system is given by

5271: \begin{equation}\label{partition2}

5272: Z(\beta)= \sum_{m\in \Gamma\backslash M_2^+(\Z)}\det(m)^{-\beta}

5273: =\sum_{k=1}^\infty \sigma(k)\,

5274: k^{-\beta}=\zeta(\beta)\zeta(\beta-1),

5275: \end{equation}

5276: where $\sigma(k)=\sum_{d|k} d$. The form of the partition

5277: function suggests the possibility that two distinct phase

5278: transitions might happen at $\beta=1$ and $\beta=2$.

5279: 

5280: The structure of KMS states for this system is analysed in

5281: \cite{CoMa}. The main result is the following.

5282: 

5283: \begin{thm}\label{GL2KMS}

5284: The KMS$_\beta$ states of the $\GL_2$-system have the following

5285: properties:

5286: \begin{enumerate}

5287: \item In the range $\beta\leq 1$ there are no KMS states.

5288: \item In the range $\beta>2$ the set of extremal KMS states is

5289: given by the classical Shimura variety

5290: \begin{equation}\label{Ekmsbeta2}

5291: \cE_\beta \cong \GL_2(\Q)\backslash \GL_2(\A) /\C^*.

5292: \end{equation}

5293: \end{enumerate}

5294: \end{thm}

5295: 

5296: The symmetries are more complicated than in the Bost--Connes case. In

5297: fact, in addition to symmetries given by automorphisms that commute

5298: with the time evolution, there are also symmetries by {\em

5299: endomorphisms} that play an important role. The resulting symmetry

5300: group is the quotient $\GL_2(\A_f)/\Q^*$. An important result of

5301: Shimura \cite{Shi} shows that this group is in fact the Galois group

5302: of the field $F$ of modular functions. The group $\GL_2(\A_f)$

5303: decomposes as a product

5304: \begin{equation}\label{SymmGL2}

5305: \GL_2(\A_f)=\GL_2^+(\Q) \GL_2(\hat\Z),

5306: \end{equation}

5307: where $\GL_2(\hat\Z)$ acts by automorphisms related to the

5308: deck transformations of the tower of the modular curves, while

5309: $\GL_2^+(\Q)$ acts by endomorphisms that move across levels in

5310: the modular tower.

5311: 

5312: The modular field $F$ is the field of modular functions over

5313: $\Q^{ab}$, namely the union of the fields $F_N$ of modular

5314: functions of level $N$ rational over the cyclotomic field

5315: $\Q(\zeta_n)$, that is, such that the $q$-expansion in powers of

5316: $q^{1/N}=\exp(2\pi i \tau/N)$ has all coefficients in $\Q(e^{2\pi

5317: i/N})$.

5318: 

5319: The action of the Galois group $\hat\Z^* \simeq \Gal(\Q^{ab}/\Q)$

5320: on the coefficients determines a homomorphism

5321: \begin{equation}\label{cyclhom}

5322: {\rm cycl}: \hat\Z^* \to \Aut(F).

5323: \end{equation}

5324: 

5325: If $\tau\in \H$ is a generic point, then the evaluation map

5326: $f\mapsto f(\tau)$ determines an embedding $F\hookrightarrow \C$. We

5327: denote by $F_\tau$ the image in $\C$. This yields an

5328: identification

5329: \begin{equation}\label{GalFtau}

5330:  \theta_\tau: \Gal(F_\tau/\Q)

5331: \stackrel{\simeq}{\to} \Q^* \backslash \GL_2(\A_f).

5332: \end{equation}

5333: 

5334: There is an arithmetic algebra $\cA_{2,\Q}$ (defined over $\Q$) of

5335: unbounded multipliers of the $C^*$-algebra $\cA_2$, obtained by

5336: considering continuous functions on $Z$ (\cf

5337: \eqref{Uspace}), with finite support in the variable $g\in

5338: \Gamma\backslash\GL_2^+(\Q)$ and with the following properties.

5339: Let $p_N: M_2(\hat\Z)\to M_2(\Z/N\Z)$ be the canonical

5340: projection. With the notation $f_{(g,\rho)}(z) = f(g,\rho,z)$,

5341: we say that $f_{(g,\rho)}\in C(\H)$ is of level $N$ if

5342: $$ f_{(g,\rho)} = f_{(g,p_N(\rho))} \ \ \ \ \forall (g,\rho). $$

5343: We require that elements of $\cA_{2,\Q}$ have the $f_{(g,\rho)}$ of

5344: finite level with $ f_{(g,m)} \in F$ for all $(g,m)$. We also

5345: require that the action \eqref{cyclhom} on the coefficients

5346: of the q-expansion of the $f_{(g,m)}$ satisfies

5347: $$ f_{(g,\alpha(u)m)} = {\rm cycl}(u) \, f_{(g,m)}, $$

5348: for all $g\in \GL_2^+(\Q)$ diagonal and all $u\in \hat\Z^*$, with

5349: $$ \alpha(u)=\begin{pmatrix} u& 0 \\ 0 & 1 \end{pmatrix}, $$

5350: to avoid some ``trivial'' elements that would spoil the Galois

5351: action on values of states (\cf \cite{CoMa}, \cite{CoMajgp1}).

5352: The action of symmetries extends to $\cA_{2,\Q}$.

5353: We have then the following result (\cite{CoMa}):

5354: 

5355: 

5356: \begin{thm}\label{GalGL2infty}

5357: Consider a state $\varphi=\varphi_{\infty,L}\in \cE_\infty$, for a

5358: generic invertible $\Q$-lattice $L=(\rho,\tau)$. Then the

5359: values of the state on elements of the arithmetic

5360: subalgebra generate the image in $\C$ of the modular field,

5361: \begin{equation}\label{values}

5362: \varphi(\cA_{2,\Q})\subset F_\tau,

5363: \end{equation}

5364: and the isomorphism

5365: \begin{equation}\label{thetaphi1}

5366: \theta_\varphi : \Gal(F_\tau/\Q)

5367: \stackrel{\simeq}{\longrightarrow} \Q^* \backslash \GL_2(\A_f),

5368: \end{equation}

5369: given by

5370: \begin{equation}\label{thetaphi2}

5371:  \theta_\varphi(\gamma)=\rho^{-1}\, \theta_\tau(\gamma) \, \rho,

5372: \end{equation}

5373: for $\theta_\tau$ as in \eqref{GalFtau}, intertwines the Galois

5374: action on the values of the state with the action of symmetries,

5375: \begin{equation}\label{intertwineGL2}

5376: \gamma\, \varphi(f) = \varphi( \theta_\varphi(\gamma) f), \ \ \ \

5377: \forall f\in \cA_{2,\Q}, \ \ \forall\gamma\in \Gal(F_\tau/\Q).

5378: \end{equation}

5379: \end{thm}

5380: 

5381: \medskip

5382: 

5383: A notion analogous to that of $\Q$-lattices can be given for other

5384: number fields $\K$. This notion was used in \cite{CMR} to construct a

5385: quantum statistical mechanical system for $\K$ an imaginary quadratic

5386: field. This system shares properties with both the Bost--Connes system

5387: of \cite{BC} and the $\GL_2$ system (2-dimensional $\Q$-lattices) of

5388: \cite{CoMa}.

5389: 

5390: We assume that $\K=\Q(\sqrt{-d})$, $d$ a positive integer. Let

5391: $\tau\in \H$ be such that

5392: $\K=\Q(\tau)$ and $\cO = \Z + \Z \tau$ is the ring of integers of $\K$.

5393: 

5394: A 1-dimensional $\K$-lattice

5395: $(\Lambda,\phi)$ is a finitely generated $\cO$-submodule $\Lambda\subset \C$,

5396: such that $\Lambda\otimes_\cO \K \cong \K$, together with a morphism of

5397: $\cO$-modules

5398: \begin{equation}\label{Kphi}

5399:  \phi : \K/\cO \to \K\Lambda/\Lambda.

5400: \end{equation}

5401: A 1-dimensional $\K$-lattice is {\em invertible} if $\phi$ is an

5402: isomorphism of $\cO$-modules.

5403: A 1-dimensional $\K$-lattice is, in particular, a 2-dimensional

5404: $\Q$-lattice.

5405: 

5406: We consider the

5407: notion of commensurability as in the case of $\Q$-lattices.

5408: Two 1-dimensional $\K$-lattices

5409: $(\Lambda_1,\phi_1)$ and

5410: $(\Lambda_2,\phi_2)$ are commensurable if $\K\Lambda_1=\K\Lambda_2$

5411: and $\phi_1=\phi_2$ modulo $\Lambda_1+\Lambda_2$.

5412: In particular, two 1-dimensional $\K$-lattices are commensurable iff

5413: the underlying $\Q$-lattices are commensurable.

5414: 

5415: The algebra of the corresponding noncommutative space is a restriction

5416: of the algebra of the $\GL_2$-system to the subgroupoid of the

5417: equivalence of commensurability restricted to $\K$-lattices.

5418: The time evolution is also a restriction from the $\GL_2$-system.

5419: 

5420: The resulting system has partition function the Dedekind zeta function

5421: $\zeta_\K(\beta)$ of the number field $\K$. Above the critical temperature

5422: $T=1$ there is a unique KMS state, while at lower temperatures the

5423: extremal KMS states are parameterized by elements of $\A_\K^*/\K^*$,

5424: where $\A_K=\A_{K,f}\times \C$ are the adeles of $\K$, with

5425: $\A_{K,f}=\A_f\otimes \K$. The KMS states at zero temperature,

5426: evaluated on the restriction to $\K$-lattices of the arithmetic

5427: algebra of the $\GL_2$-system, have an action of the Galois group

5428: $\Gal(\K^{ab}/\K)$ realized (via the class field theory isomorphism)

5429: through the action of symmetries (automorphisms and endomorphisms) of

5430: the system (\cf \cite{CMR}).

5431: 

5432: 

5433: \bigskip

5434: 

5435: \section{Modular Hecke algebras}\label{hecke}

5436: 

5437: Connes and Moscovici \cite{Co-Mosc1} defined modular Hecke algebras

5438: $\cA(\Gamma)$ of level $\Gamma$, a congruence subgroup of

5439: $\PSL_2(\Z)$. These extend both the ring of classical Hecke operators

5440: and the algebra of modular forms.

5441: 

5442: Modular Hecke algebras  encode two \textit{a priori} unrelated

5443: structures on modular forms, namely the algebra structure given by

5444: the pointwise product on one hand, and the action of the Hecke

5445: operators on the other. To any congruence subgroup $\Gamma$ of $\,

5446: \SL_2(\Z)$ corresponds a crossed product algebra

5447: $\cA(\Gamma)$, the \textit{modular Hecke algebra} of level $\Gamma$,

5448: which is a direct extension of both the ring of classical Hecke

5449: operators and of the algebra $\cM(\Gamma)$ of $\Gamma$-modular

5450: forms.

5451: 

5452: These algebras can be obtained by considering the action of

5453: $\GL^+_2(\Q)$ on the algebra of modular forms on the full (adelic)

5454: modular tower, which yields the ``holomorphic part" of the ``ring of

5455: functions" of the noncommutative space of commensurability classes of

5456: 2-dimensional $\Q$-lattices, introduced in Section \ref{Qlatt}.

5457: 

5458: With $\cM$ denoting the algebra of modular forms of arbitrary

5459: level, the elements of $\cA(\Gamma)$ are maps with finite support

5460: $$

5461:  F: \Gamma\backslash \GL^+ (2, \mathbb{Q}) \to \cM \, ,

5462: \qquad \Gamma  \alpha \mapsto F_{\alpha} \in \cM  \, ,

5463: $$

5464: satisfying the covariance condition

5465: \begin{equation}

5466: F_{\alpha \gamma} \, = \, F_{\alpha} \vert \gamma \qqq\alpha \in

5467: \GL^+ (2, \mathbb{Q}) \, , \gamma \in \Gamma \, \nonumber

5468: \end{equation}

5469: and their product is given by convolution.

5470: 

5471: More in detail,

5472: let $G=\PGL^+_2(\Q)$ and $\Gamma \subset \PSL_2(\Z)$ a finite

5473: index subgroup. The quotient map $\Gamma \backslash G \to \Gamma

5474: \backslash G / \Gamma$ is finite to one, and $\Gamma$ acts on $\C

5475: [\Gamma \backslash G]$. Let $\cH_k$ be the space of holomorphic

5476: functions $f: \H \to \C$ with polynomial growth, and with the

5477: action $|_k$, for $k\in 2\Z$, of $\PGL^+_2(\R)$ of the form

5478: $$ \left(f|_k \left(\begin{array}{cc} a& b\\ c& d \end{array}\right)\right)

5479: (z)= \frac{(ad-bc)^{k/2}}{(cz+d)^k} \, f\left( \frac{az+b}{cz+d}

5480: \right).

5481: $$

5482: This determines induced actions of $G$ and $\Gamma$ on $\cH_k$.

5483: The space of modular forms is obtained as $\cM_k(\Gamma)=

5484: \cH_k^\Gamma$, the invariants of this action.

5485: 

5486: One can then define

5487: \begin{equation}\label{AkGamma}

5488:  \cA_k (\Gamma) := \left( \C [ \Gamma \backslash G ] \otimes_\C

5489: \cH_k \right)^\Gamma,

5490: \end{equation}

5491: with respect to the right action of $\Gamma$,

5492: $$ \gamma : \sum_i (\Gamma g_i) \otimes f_i \mapsto \sum_i \Gamma g_i

5493: \gamma \otimes (f_i |_k \gamma).

5494: $$

5495: One considers the graded vector space $\cA_*(\Gamma) = \oplus_k

5496: \cA_k(\Gamma)$. The elements of $\cA_k(\Gamma)$ can be thought of

5497: as finitely supported $\Gamma$-equivariant maps

5498: $$ \phi: \Gamma\backslash G \to \cH_k  \ \ \ \sum_i (\Gamma

5499: g_i)\otimes f_i \mapsto f_i. $$

5500: We can embed

5501: $$ \cA_*(\Gamma)\subset \hat\cA_*(\Gamma):= \Hom_\Gamma

5502: (\C[\Gamma\backslash G], \cH_k), $$ where we think of

5503: $\cA_*=\cH_*[\Gamma\backslash G]$ as polynomials in

5504: $\Gamma\backslash G$ with $\cH_*$ coefficients, and of $\hat\cA_*=

5505: \cH_* [[ \Gamma\backslash G ]]$ as formal power series, that is,

5506: $\Gamma$-equivariant maps $\phi: \Gamma \backslash G \to \cH_k$.

5507: There is on $\cA_*(\Gamma)$ an associative multiplication (\cf

5508: \cite{Co-Mosc1}), which makes

5509: $\cA_*(\Gamma)$ into a noncommutative ring. This is given by a

5510: convolution product. For any $\phi \in \cA_k(\Gamma)$, we have

5511: $\phi_g = \phi_{\gamma g}$, with $\phi_g=0$ off a finite subset of

5512: $\Gamma \backslash G$, and $\phi_g | \gamma = \phi_{g\gamma}$, so

5513: these terms are left $\Gamma$-invariant and right

5514: $\Gamma$-equivariant. For $\phi \in \cA_k(\Gamma)$ and $\psi \in

5515: \cA_\ell (\Gamma)$ we then define the convolution product as

5516: \begin{equation}\label{convAprod}

5517: ( \phi * \psi )_g := \sum_{(g_1,g_2)\in G\times_\Gamma G, g_1 g_2

5518: =g} (\phi_{g_1} | g_2) \phi_{g_2},

5519: \end{equation}

5520: 

5521: The algebra $\cA_*(\Gamma)$ constructed this way has two

5522: remarkable subalgebras.

5523: \begin{itemize}

5524: \item $\cA_0(\Gamma)= \C [\Gamma \backslash G / \Gamma]$ is the

5525: algebra $\bT$ of Hecke operators.

5526: \item $\cM_k(\Gamma)\subset \cA_k(\Gamma)$ also gives a subalgebra

5527: $\cM_*(\Gamma) \subset \cA_*(\Gamma)$.

5528: \end{itemize}

5529: 

5530: In particular observe that all the coefficients $\phi_g$ are

5531: modular forms. In fact, they satisfy $\phi_g | \gamma =

5532: \phi_{g\gamma}$, hence, for $\gamma \in \Gamma$, this gives

5533: $\phi_g | \gamma = \phi_g$.

5534: 

5535: Notice however that the convolution product on $\cA_*$ does not

5536: agree with the Hecke action, namely the diagram

5537: \begin{eqnarray*}

5538: \diagram \cM_*(\Gamma) \otimes \bT \dto^{\iota}\rto^{H} &

5539: \cM_*(\Gamma) \dto^{\iota} \\

5540: \cA_*\otimes \cA_* \rto^{*}& \cA_*,

5541: \enddiagram

5542: \end{eqnarray*}

5543: with $\iota$ the inclusion of subalgebras and $H$ the Hecke

5544: action, is {\em not} commutative, nor is the symmetric one

5545: \begin{eqnarray*}

5546: \diagram \bT  \otimes  \cM_*(\Gamma) \dto^{\iota}\rto^{H} &

5547: \cM_*(\Gamma) \dto^{\iota} \\

5548: \cA_*\otimes \cA_* \rto^{*}& \cA_*.

5549: \enddiagram

5550: \end{eqnarray*}

5551: To get the correct Hecke action on modular forms from the algebra

5552: $\cA_*(\Gamma)$, one needs to introduce the augmentation map

5553: $$ \epsilon : \C [\Gamma \backslash G] \to \C $$

5554: extended to a map

5555: $$ \epsilon\otimes 1 : \C[\Gamma \backslash G] \otimes \cH_k \to

5556: \cH_k  \ \ \ \sum [g]\otimes \phi_g \mapsto \sum \phi_g. $$

5557: 

5558: One then obtains a commutative diagram

5559: \begin{eqnarray*}

5560: \diagram \bT  \otimes  \cM_*(\Gamma) \dto^{\iota}\rto^{H} &

5561: \cM_*(\Gamma)  \\

5562: \cA_*\otimes \cA_* \rto^{*}& \cA_* \uto^{\epsilon \otimes 1}.

5563: \enddiagram

5564: \end{eqnarray*}

5565: 

5566: In \cite{CoMoHopf} Connes and Moscovici introduced a Hopf algebra $\cH_1$

5567: associated to the transverse geometry of codimension one foliations.

5568: This is the universal enveloping algebra of a Lie algebra with basis

5569: $\{ \cX,\cY,\delta_n \, n\geq 1 \}$ satisfying, for $n,k,\ell \geq 1$,

5570: \begin{equation}\label{Liebra}

5571: [\cY,\cX]=\cX, \,\,\, [\cY,\delta_n] = n\,\delta_n, \,\,\,

5572: [\cX,\delta_n] =\delta_{n+1}, \,\,\, [\delta_k,\delta_\ell]=0,

5573: \end{equation}

5574: with coproduct an algebra homomorphism $\Delta: \cH_1 \to \cH_1

5575: \otimes \cH_1$ satisfying

5576: \begin{equation}\label{coprod-H1}

5577: \begin{array}{ll}

5578: \Delta \cY=& \cY \otimes 1 + 1 \otimes \cY, \\ \Delta \delta_1 =&

5579: \delta_1 \otimes 1 + 1 \otimes \delta_1, \\ \Delta \cX =& \cX\otimes 1

5580: + 1 \otimes \cX + \delta_1 \otimes \cY,

5581: \end{array}

5582: \end{equation}

5583: antipode the anti-isomorphism satisfying

5584: \begin{equation}\label{antipode-H1}

5585: S(\cY)=-\cY, \,\,\,\, S(\cX)=-\cX+\delta_1\cY, \,\,\,\,

5586: S(\delta_1)=-\delta_1,

5587: \end{equation}

5588: and co-unit $\epsilon(h)$ the constant term of $h\in \cH_1$.

5589: 

5590: The Hopf algebra $\cH_1$ acts as symmetries of the modular Hecke

5591: algebras. This is a manifestation of the general fact that,

5592: while symmetries of ordinary commutative spaces are encoded by group

5593: actions, symmetries of noncommutative spaces are given by Hopf

5594: algebras.

5595: 

5596: By comparing the actions of the Hopf algebra $\cH_1$, it is possible to

5597: derive an analogy (\cf \cite{Co-Mosc1}) between the modular Hecke

5598: algebras and the crossed product algebra of the action of a discrete

5599: subgroup of ${\rm Diff}(S^1)$ on polynomial functions on the frame

5600: bundle of $S^1$.

5601: 

5602: In fact, for $\Gamma$ a discrete subgroup of ${\rm Diff}(S^1)$, and

5603: $X$ a smooth compact 1-dimensional manifold, consider as in

5604: \cite{Co-Mosc1} the algebra

5605: \begin{equation}\label{Jet-alg}

5606: \cA_\Gamma = C^\infty_c (J^1_+(X)) \rtimes \Gamma,

5607: \end{equation}

5608: where $J^1_+(X)$ is the oriented 1-jet bundle. This has an action of

5609: the Hopf algebra $\cH_1$ by

5610: \begin{equation}\label{H1act1}

5611: \begin{array}{ll}

5612: \cY(f U_\phi^*)=& y_1 \frac{\partial f}{\partial y_1} U_\phi^* \\[3mm]

5613: \cX(f U_\phi^*)=& y_1 \frac{\partial f}{\partial y} U_\phi^* \\[2mm]

5614: \delta_n (f U_\phi^*)=& y_1^n \frac{d^n}{d y^n}\left( \log

5615: \frac{d\phi}{dy} \right) f U_\phi^*,

5616: \end{array}

5617: \end{equation}

5618: with coordinates $(y,y_1)$ on $ J^1_+(X)\simeq X\times \R^+$.

5619: The trace $\tau$ defined by the volume form

5620: \begin{equation}\label{trace-tau}

5621: \tau(f U_\phi^*)= \left\{ \begin{array}{ll} \int_{J^1_+(X)} f(y,y_1)

5622: \, \frac{dy\wedge dy_1}{y_1^2} & \phi=1 \\[3mm]

5623: 0 & \phi\neq 1 \end{array}\right.

5624: \end{equation}

5625: satisfies

5626: \begin{equation}\label{tau-nu}

5627: \tau(h(a)) = \nu(h) \tau(a) \, \, \, \, \forall h \in\cH_1,

5628: \end{equation}

5629: with $\nu\in \cH_1^*$ satisfying

5630: \begin{equation}\label{nu}

5631: \nu(\cY)=1, \,\,\,\,\,\, \nu(\cX)=0, \,\,\,\,\,\, \nu(\delta_n)=0.

5632: \end{equation}

5633: The twisted antipode $\tilde S=\nu * S$ satisfies $\tilde S^2=1$ and

5634: \begin{equation}\label{twistS}

5635: \tilde S(\cY)=-\cY+1, \,\,\,\,\,\, \tilde S(\cX)=-\cX +\delta_1 \cY,

5636: \,\,\,\,\,\, \tilde S(\delta_1)=-\delta_1.

5637: \end{equation}

5638: 

5639: 

5640: The Hopf cyclic cohomology of a Hopf algebra is another fundamental

5641: tool in noncommutative geometry, which was developed by Connes and

5642: Moscovici in \cite{CoMoHopf}. They applied it to the

5643: computation of the local index formula for tranversely

5644: hypoelliptic operators on foliations. An action of a Hopf algebra on

5645: an algebra induces a characteristic map from the Hopf cyclic

5646: cohomology of the Hopf algebra to the cyclic cohomology of the

5647: algebra, hence the index computation can be done in terms of Hopf

5648: cyclic cohomology. The periodic Hopf cyclic cohomology of the Hopf

5649: algebra of transverse geometry is related to the Gelfand-Fuchs

5650: cohomology of the Lie algebra of formal vector fields \cite{CoMoHopf2}.

5651: 

5652: In the case of the Hopf algebra $\cH_1$, there are three basic

5653: cyclic cocycles, which in the original context of transverse geometry

5654: correspond, respectively, to the Schwarzian derivative, the

5655: Godbillon-Vey class, and the transverse fundamental class.

5656: 

5657: In particular, the Hopf cyclic

5658: cocycle associated to  the Schwarzian derivative is of the form

5659: \begin{equation}\label{delta2'}

5660: \delta_2':= \delta_2 - \frac 12 \delta_1^2

5661: \end{equation}

5662: with

5663: \begin{equation}\label{delta2'act}

5664:  \delta_2' (f U_\phi^*)= y_1^2 \left\{ \phi(y); y\right\} \, f

5665: U_\phi^*

5666: \end{equation}

5667: \begin{equation}\label{schw-der}

5668: \left\{ F ; x \right\} : = \frac{d^2}{dx^2} \left( \log \frac{dF}{dx}

5669: \right) - \frac 12 \left( \frac{d}{dx} \left( \log \frac{dF}{dx}

5670: \right) \right)^2.

5671: \end{equation}

5672: 

5673: The action of the Hopf algebra $\cH_1$ on the modular Hecke algebra

5674: described in \cite{Co-Mosc1} involves the natural derivation on

5675: the algebra of modular forms initially introduced by Ramanujan, which

5676: corrects the ordinary

5677: differentiation by a logarithmic derivative of the Dedekind $\eta$

5678: function,

5679: \begin{equation}\label{Heck-rep}

5680: \cX:= \frac{1}{2\pi i} \frac{d}{dz} - \frac{1}{2\pi i}

5681: \frac{d}{dz}(\log \eta^4) \,\, \cY, \ \ \

5682: \cY (f) =\frac k2 \, f, \,\,\, \forall f \in \cM_k .

5683: \end{equation}

5684: The element $\cY$ is the grading operator

5685: that multiplies by $k/2$ forms of weight $k$, viewed as sections of

5686: the $(k/2)$th power of the line bundle of 1-forms. The element

5687: $\delta_1$ acts as multiplication by a form-valued cocycle on

5688: $\GL^+_2(\Q)$, which measures the

5689: lack of invariance of the section $\eta^4 dz$.

5690: More precisely, one has the following action of

5691: $\cH_1$ (Connes--Moscovici \cite{Co-Mosc1}):

5692: 

5693: \begin{thm}\label{actH1hecke}

5694: There is an action of the Hopf algebra $\cH_1$ on the modular Hecke

5695: algebra $\cA(\Gamma)$ of level $\Gamma$, induced by an action on

5696: $\cA_{G^+(\Q)}:= \cM \rtimes G^+(\Q)$, for

5697: $\cM=\varinjlim_{N\to\infty} \cM(\Gamma(N))$, of the form

5698: \begin{equation}\label{actH1GQ}

5699: \begin{array}{ll}

5700: \cY(fU^*_\gamma)= & \cY(f)\, U^*_\gamma \\[3mm]

5701: \cX(fU^*_\gamma)= & \cX(f)\, U^*_\gamma \\[2mm]

5702: \delta_n(fU^*_\gamma)= & \frac{d^n}{dZ^n} \left( \log \frac{d(Z |_0

5703: \gamma)}{dZ} \right) (dZ)^n \, fU^*_\gamma,

5704: \end{array}

5705: \end{equation}

5706: with $\cX(f)$ and $\cY(f)$ as in \eqref{Heck-rep}, and

5707: \begin{equation}\label{coordZ}

5708:  Z(z) = \int_{i\infty}^z \eta^4 dz.

5709: \end{equation}

5710: \end{thm}

5711: 

5712: The cocycle \eqref{delta2'} associated to the Schwarzian derivative is

5713: represented by an inner derivation of $\cA_{G^+(\Q)}$,

5714: \begin{equation}\label{delta2'heck}

5715: \delta_2'(a) = [a, \omega_4],

5716: \end{equation}

5717: where $\omega_4$ is the weight four modular form

5718: \begin{equation}\label{omega4}

5719: \omega_4 = -\frac{E_4}{72}, \,\,\,\, \text{ with } \,\,\, E_4(q)= 1 +

5720: 240 \sum_{n=1}^\infty n^3 \frac{q^n}{1-q^n}, \,\,\,\,\, q=e^{2\pi i z},

5721: \end{equation}

5722: which is expressed as a Schwarzian derivative

5723: \begin{equation}\label{Somega4}

5724:  \omega_4 = (2\pi i)^{-2} \, \{ Z; z\}.

5725: \end{equation}

5726: 

5727: This result is

5728: used in \cite{Co-Mosc1} to investigate perturbations of the Hopf algebra

5729: action. The freedom one has in modifying the action by

5730: a 1-cocycle corresponds exactly to the data introduced by Zagier

5731: in \cite{Zagier}, defining "canonical" Rankin-Cohen algebras,

5732: with the derivation $\partial$ and the element $\Phi$ in Zagier's

5733: notation corresponding, respectively, to the action of the generator

5734: $\cX$ on modular forms and to $\omega_4=2\Phi$.

5735: 

5736: The cocycle associated to the Godbillon-Vey class is described in

5737: terms of a 1-cocycle on $\GL^+_2(\Q)$ with values in

5738: Eisenstein series of weight two, which measures the lack of

5739: $\GL^+_2(\Q)$-invariance of the connection associated to the

5740: generator $\cX$. The authors derive from this an arithmetic presentation

5741: of the rational Euler class in $H^2(\SL_2(\Q),\Q)$ in

5742: terms of generalized Dedekind sums.

5743: 

5744: The cocycle associated to the transverse fundamental class, on the

5745: other hand, gives rise to a natural extension of the first

5746: Rankin-Cohen bracket \cite{Zagier} from modular

5747: forms to the modular Hecke algebras.

5748: 

5749: \medskip

5750: 

5751: Rankin--Cohen algebras can be treated in different perspectives:

5752: Zagier introduced them and studied them with a

5753: direct algebraic approach (\cf \cite{Zagier}). There appears to be an

5754: interesting and deep connection to vertex operator algebras, which

5755: manifests itself in a form of duality between these two types of

5756: algebras.

5757: 

5758: The Rankin--Cohen brackets are a family of brackets

5759: $[f,g\,]^{(k,\ell)}_n$ for $n\geq 0$, defined for $f,g \in R$, where

5760: $R$ is a graded ring with a derivation $D$. For

5761: $R=\oplus_{k\geq 0} R_k$, $D: R_k \to R_{k+2}$, $f\in R_k$ $g\in

5762: R_\ell$, the brackets $[,]_n : R_k\otimes R_\ell \to R_{k+\ell+2}$

5763: are given by

5764: \begin{equation}\label{RCbraD}

5765: [f,g\,]^{(k,\ell)}_n = \sum_{r+s=n} (-1)^r \left(\begin{array}{c}

5766: n+k-1 \\ s \end{array}\right) \left(\begin{array}{c} n+\ell -1 \\

5767: r \end{array}\right) D^r f \, D^s g.

5768: \end{equation}

5769: 

5770: These Rankin--Cohen brackets induced by $(R_*, D) \Rightarrow

5771: (R_*, [,]_*)$ give rise to a {\em standard Rankin--Cohen algebra}, in

5772: Zagier's terminology (\cf \cite{Zagier}).

5773: There is an isomorphism of categories between graded rings with a

5774: derivation and standard Rankin--Cohen algebras.

5775: 

5776: In the case of Lie algebras, one can first define a {\em standard

5777: Lie algebra} as the Lie algebra associated to an associative

5778: algebra $(A,*)\Rightarrow (A,[,])$ by setting $[X,Y]= X*Y-Y*X$ and

5779: then define an abstract Lie algebra as a structure $(A,[,])$ that

5780: satisfies {\em all} the algebraic identities satisfied by a

5781: standard Lie algebras, though it is not necessarily induced by an

5782: associative algebra. It is then a theorem that the antisymmetry of

5783: the bracket and the Jacobi identity are sufficient to determine

5784: all the other algebraic identities, hence one can take these as a

5785: definition of an abstract Lie algebra.

5786: 

5787: Just as in the case of Lie algebras, we can define a Rankin--Cohen

5788: algebra $(R_*, [,]_*)$ as a graded ring $R_*$ with a family of

5789: degree $2n$ brackets $[,]_n$ satisfying all the algebraic

5790: identities of the standard Rankin--Cohen algebra. However, in this

5791: case there is no simple set of axioms that implies all the

5792: algebraic identities.

5793: 

5794: The motivation for this structure lies in the fact that there is a

5795: very important example of a Rankin--Cohen algebra which is in fact

5796: non-standard. The example is provided by modular forms (\cf

5797: \cite{Zagier}).

5798: 

5799: If $f\in \cM_k$ is a modular form satisfying

5800: $$ f\left( \frac{az+b}{cz+d} \right) = (cz+d)^k f(z), $$

5801: then it derivative is no longer a modular form, due to the

5802: presence of the second term in

5803: $$ f' \left(\frac{az+b}{cz+d} \right) =(cz+d)^{k+2} f(z) + kc

5804: (cz+d)^{k+1} f(z).

5805: $$

5806: On the other hand, if we have $f\in \cM_k$ and $g\in \cM_\ell$,

5807: the bracket

5808: $$ [f,g](z) := \ell f'(z) g(z) - k f(z) g'(z) $$

5809: is a modular form in $\cM_{k+\ell +2}$. Similarly, we can define

5810: an $n$-th bracket $[,]_n: \cM_k \otimes \cM_\ell \to

5811: \cM_{k+\ell+2}$. Here's the first few brackets:

5812: $$ \begin{array}{ll} [f,g]_0 =& f g \\[2mm]

5813: [f,g]_1 =& k f g' - \ell f' g \\[2mm]

5814: [f,g]_2 = & \left(\begin{array}{c} k+1\\ 2\end{array}\right) f g''

5815: - (k+1) (\ell +1) f'g' + \left(\begin{array}{c} \ell +1\\

5816: 2\end{array}\right)f'' g. \end{array} $$

5817: 

5818: 

5819: Notice that for the graded ring of modular forms we have $\cM_*

5820: (\Gamma)\subset \cH$, where $\cH$ is the vector space $\cH = {\rm

5821: Hol}(\H)_{polyn}$ of holomorphic functions on the upper half

5822: plane $\H$ with polynomial growth. This is closed under

5823: differentiation and $(\cH, D)$ induces a standard Rankin--Cohen

5824: algebra $(\cH, [,]_*)$. The inclusion $(\cM_*,[,]_*) \subset (\cH,

5825: [,]_*)$ is not closed under differentiation but it is closed under

5826: the brackets.

5827: 

5828: A way of constructing non-standard Rankin--Cohen algebras is

5829: provided by Zagier's {\em canonical construction} (\cf

5830: \cite{Zagier}). One considers

5831: here the data $(R_*,D,\Phi)$, where $R_*$ is a graded ring with a

5832: derivation $D$ and with a choice of an element $\Phi \in R_4$, the

5833: {\em curvature}. One then defines the brackets by the formula

5834: \begin{equation}\label{RCbraDPhi}

5835: [f,g]^{(k,\ell)}_n = \sum_{r+s=n} (-1)^r \left(\begin{array}{c}

5836: n+k-1 \\ s \end{array}\right) \left(\begin{array}{c} n+\ell -1 \\

5837: r \end{array}\right)\, f_r \, g_s,

5838: \end{equation}

5839: where $f_0 = f$ and

5840: \begin{equation}\label{DPhi}

5841: f_{r+1} = D f_r + r(r+1) \Phi f_{r-1}.

5842: \end{equation}

5843: The structure $(R_*,[,]_*)$ obtained this way is a Rankin--Cohen

5844: algebra (see \cite{Zagier}).

5845: 

5846: 

5847: There is a gauge action on the curvature $\Phi$, namely, for any

5848: $\varphi \in R_2$ the transformation $D \mapsto D'$ and $\Phi

5849: \mapsto \Phi'$ with

5850: $$ D'(f) = D(f) + k \varphi f $$

5851: for $f \in \cM_k$ and

5852: \begin{equation}\label{gaugePhi}

5853:  \Phi' = \Phi + \varphi^2 - D(\varphi)

5854: \end{equation}

5855: give rise to {\em the same} Rankin--Cohen algebra. Thus, all the

5856: cases where the curvature $\Phi$ can be gauged away to zero

5857: correspond to the standard case.

5858: 

5859: \medskip

5860: 

5861: 

5862: The modular form $\omega_4$ of \eqref{Somega4} provides

5863: the curvature element $\omega_4 = 2\Phi$, and the gauge equivalence

5864: condition \eqref{gaugePhi} can be rephrased in terms of Hopf

5865: algebras as the freedom to change the $\cH_1$ action by a

5866: cocycle. In particular (\cf \cite{Co-Mosc1}), for the specified

5867: action, the resulting Rankin--Cohen structure is canonical but not

5868: standard, in Zagier's terminology.

5869: 

5870: The 1-form $dZ=\eta^4 dz$ is, up to scalars, the only holomorphic

5871: differential on the elliptic curve $E=X_{\Gamma(6)} \cong

5872: X_{\Gamma_0(36)}$ of equation $y^2 = x^3 +1$, so that

5873: $dZ=\frac{dx}{y}$ in Weierstrass coordinates.

5874: 

5875: \medskip

5876: 

5877: The Rankin--Cohen brackets on modular forms can be extended to

5878: brackets $RC_n$ on the modular Hecke algebra, defined in terms of the

5879: action of the Hopf algebra $\cH_1$ of transverse geometry. In fact,

5880: more generally, it is shown in \cite{Co-Mosc2} that it is possible to

5881: define such Rankin--Cohen brackets on any associative algebra $\cA$

5882: endowed with an action of the Hopf algebra $\cH_1$ for which there

5883: exists an element $\Omega\in \cA$ such that

5884: \begin{equation}\label{innerdelta2'}

5885: \delta_2'(a) = [ \Omega, a ], \,\,\,\, \forall a\in \cA,

5886: \end{equation}

5887: and with $\delta_2'$ as in \eqref{delta2'}, and

5888: \begin{equation}\label{deltan0}

5889: \delta_n(\Omega)=0, \,\,\, \forall n\geq 1.

5890: \end{equation}

5891: 

5892: \smallskip

5893: 

5894: Under these hypotheses, the following result holds (Connes--Moscovici

5895: \cite{Co-Mosc2}):

5896: 

5897: \begin{thm}\label{RCdef}

5898: Suppose given an associative algebra $\cA$ with an action of the Hopf

5899: algebra $\cH_1$ satisfying the conditions \eqref{innerdelta2'} and

5900: \eqref{deltan0}.

5901: \begin{enumerate}

5902: \item There exists Rankin--Cohen brackets $RC_n$ of the form

5903: \begin{equation}\label{RCnH1}

5904: RC_n(a,b)= \sum_{k=0}^n \frac{A_k}{k!} (2\cY+k)_{n-k}(a)\,

5905: \frac{B_{n-k}}{(n-k)!} (2\cY+n-k)_k (b),

5906: \end{equation}

5907: with $(\alpha)_r=\alpha(\alpha+1)\cdots(\alpha +r-1)$ and the

5908: coefficients $A_{-1}=0$, $A_0=1$, $B_0=1$, $B_1=\cX$,

5909: $$ A_{n+1}= S(\cX) A_n -n \Omega^0 \left(\cY - \frac{n-1}{2}\right)

5910: A_{n-1}, $$

5911: $$ B_{n+1}= \cX B_n - n \Omega \left(\cY-\frac{n-1}{2}\right) B_{n-1},

5912: $$

5913: and $\Omega^0$ the right multiplication by $\Omega$.

5914: \item When applied to the modular Hecke algebra $\cA(\Gamma)$, with

5915: $\Omega= \omega_4 = 2\Phi$, the

5916: above construction yields brackets \eqref{RCnH1} that are completely

5917: determined by their restriction to modular forms where they agree with

5918: the Rankin--Cohen brackets \eqref{RCbraDPhi}.

5919: \item The brackets \eqref{RCnH1} determine associative deformations

5920: \begin{equation}\label{assoc-def}

5921: a*b = \sum_n \hbar^n \, RC_n(a,b).

5922: \end{equation}

5923: \end{enumerate}

5924: \end{thm}

5925: 

5926: 

5927: \medskip

5928: 

5929: The first of the steps described in Section \ref{road}, namely

5930: resolving the diagonal in $\cA(\Gamma)$, is not yet

5931: done for the modular Hecke algebras and should shed light on the

5932: important number theoretic problem of the interrelation of the Hecke

5933: operators with the algebra structure given by the pointwise product.

5934: 

5935: The algebra $\cA(\Gamma)$ is deeply related to the algebra of the

5936: space of two dimensional $\Q$-lattices of Section \ref{Qlatt}.

5937: 

5938: 

5939: 

5940: \section{Noncommutative moduli spaces, Shimura

5941: varieties}\label{shimura}

5942: 

5943: 

5944: It appears from the study of some significant cases that an important

5945: source of interesting noncommutative spaces is provided by the

5946: ``boundary'' of classical (algebro-geometric) moduli spaces, when one

5947: takes into account the possible presence of degenerations of classical

5948: algebraic varieties that give rise to objects no longer defined within

5949: the context of algebraic varieties, but which still make sense as

5950: noncommutative spaces.

5951: 

5952: \smallskip

5953: 

5954: An example of algebro-geometric moduli spaces which is sufficiently

5955: simple to describe but which at the same time exhibits a very rich

5956: structure is that of the modular curves. The geometry of modular

5957: curves has already appeared behind our discussion of the 2-dimensional

5958: $\Q$-lattices and of the modular Hecke algebras, through an

5959: associated class of functions: the modular functions that appeared in

5960: our discussion of the arithmetic algebra for the quantum statistical

5961: mechanical system of 2-dimensional $\Q$-lattices and the

5962: modular forms in the modular Hecke algebras.

5963: 

5964: The modular curves, quotients of the hyperbolic plane $\H$ by the

5965: action of a subgroup $\Gamma$ of finite index of $\SL_2(\Z)$, are complex

5966: algebraic curves, which admit an arithmetic structure, as they are

5967: defined over cyclotomic number fields $\Q(\zeta_N)$. They are also

5968: naturally moduli spaces. The object they parameterize are elliptic

5969: curves (with some level structure). The modular curves have an

5970: algebro-geometric compactification obtained by adding finitely many

5971: cusp points, given by the points in $\P^1(\Q)/\Gamma$. These

5972: correspond to the algebro-geometric degeneration of the elliptic

5973: curve to $\C^*$. However, in addition to these degenerations, one can

5974: consider degenerations to noncommutative tori, obtained by a limit

5975: $q\to \exp(2\pi i \theta)$ in the modulus $q=\exp(2\pi i \tau)$ of the

5976: elliptic curve, where now $\theta$ is allowed to be also irrational.

5977: The resulting boundary $\P^1(\R)/\Gamma$ is a noncommutative space (in

5978: the sense of Section \ref{quotients}). It appeared in the string

5979: theory compactifications considered in \cite{CDS}. The arithmetic

5980: properties of the noncommutative spaces $\P^1(\R)/\Gamma$ were studied

5981: in \cite{ManMar}, \cite{Mar2} \cite{Mar3}.

5982: 

5983: The modular curves, for varying finite index $\Gamma\subset

5984: \SL_2(\Z)$, form a tower of branched coverings. The projective limit

5985: of this tower sits as a connected component in the more refined {\em

5986: adelic} version of the modular tower, given by the quotient

5987: \begin{equation}\label{Sh2}

5988: \GL_2(\Q)\backslash \GL_2(\A) /\C^*,

5989: \end{equation}

5990: where, as usual, $\A=\A_f\times \R$ denotes the adeles of $\Q$, with

5991: $\A_f=\hat\Z\otimes \Q$ the finite adeles.

5992: 

5993: The space \eqref{Sh2} is also a moduli space. In fact, it belongs to

5994: an important class of algebro-geometric moduli spaces of great

5995: arithmetic significance, the Shimura varieties $Sh(G,X)$, where the

5996: data $(G,X)$ are given by a reductive algebraic group $G$ and a

5997: hermitian symmetric domain $X$. The pro-variety \eqref{Sh2}

5998: is the Shimura variety $Sh(\GL_2,\H^\pm)$, where $\H^\pm=\GL_2(\R)

5999: /\C^*$ is the union of upper and lower half plane in $\P^1(\C)$.

6000: 

6001: \smallskip

6002: 

6003: We have mentioned above that the spaces $\P^1(\R)/\Gamma$ describe

6004: degenerations of elliptic curves to noncommutative tori. This type of

6005: degeneration corresponds to degenerating a lattice $\Lambda=\Z+\Z\tau$

6006: to a pseudolattice $L=\Z+\Z\theta$ (see \cite{Manin01RM} for a detailed

6007: discussion of this viewpoint and its implications in noncommutative

6008: geometry and in arithmetic). In terms of the space \eqref{Sh2}, it

6009: corresponds to degenerating the archimedean component, namely

6010: replacing $\GL_2(\R)$ by $M_2(\R)^\cdot=M_2(\R)\smallsetminus \{ 0

6011: \}$, nonzero $2\times 2$-matrices. However, when one is working with

6012: the adelic description as in \eqref{Sh2}, one can equally consider the

6013: possibility of degenerating a lattice at the non-archimedean

6014: components. This brings back directly the notion of $\Q$-lattices of

6015: Section \ref{Qlatt}.

6016: 

6017: In fact, it was shown in \cite{CMR2} that the notions of 2-dimensional

6018: $\Q$-lattices and commensurability can be reformulated in terms of

6019: Tate modules of elliptic curves and isogeny. In these terms, the space

6020: of $\Q$-lattices corresponds to non-archimedean degenerations of the

6021: Tate module, which correspond to the ``bad quotient''

6022: \begin{equation}\label{Sh2nc}

6023: \GL_2(\Q)\backslash M_2(\A_f)\times \GL_2(\R) /\C^*.

6024: \end{equation}

6025: The combination of these two types of degenerations yields a

6026: ``noncommutative compactification'' of the Shimura variety

6027: $Sh(\GL_2,\H^\pm)$ which is the algebra of the ``bad quotient''

6028: \begin{equation}\label{Sh2ncC}

6029: \GL_2(\Q)\backslash M_2(\A)^\cdot /\C^*,

6030: \end{equation}

6031: where $M_2(\A)^\cdot$ are the elements of $M_2(\A)$ with nonzero

6032: archimedean component. One recovers the Shimura variety

6033: $Sh(\GL_2,\H^\pm)$ as the set of classical points (extremal KMS states

6034: at zero temperature) of the quantum statistical mechanical system

6035: associated to the noncommutative space \eqref{Sh2nc} (\cf \cite{CoMa},

6036: \cite{CMR2}).

6037: 

6038: \medskip

6039: 

6040: More generally, Shimura varieties are moduli spaces for certain

6041: types of motives or as moduli spaces of Hodge structures (\cf \eg

6042: \cite{Mil2}). A Hodge structure $(W,h)$ is a pair of a finite

6043: dimensional $\Q$-vector space $W$ and a homomorphism $h:\bS\to

6044: \GL(W_\R)$, of the real algebraic group

6045: $\bS=\Res_{\C/\R}\bG_m$, with

6046: $W_\R=W\otimes \R$. This determines a decomposition

6047: $W_\R\otimes \C=\oplus_{p,q} W^{p,q}$ with

6048: $\overline{W^{p,q}}=W^{q,p}$ and $h(z)$ acting on $W^{p,q}$ by

6049: $z^{-p}\bar z^{-q}$. This gives a Hodge filtration and a weight

6050: filtration $W_\R=\oplus_k W_k$ where $W_k=\oplus_{p+q=m} W^{p,q}$.

6051: The Hodge structure $(W,h)$ has weight $m$ if $W_\R=W_m$. It is

6052: rational if the weight filtration is defined over $\Q$. A Hodge

6053: structure of weight $m$ is polarized if there is a morphism of

6054: Hodge structures $\psi:W\otimes W\to \Q(-m)$, such that

6055: $(2\pi i)^m\psi(\cdot, h(i)\cdot)$ is symmetric and positive

6056: definite. Here $\Q(m)$ is the rational Hodge structure of weight

6057: $-2m$, with $W=(2\pi i)^m \Q$ with the action $h(z)=(z\bar z)^m$.

6058: For a rational $(W,h)$, the subspace of $W\otimes \Q(m)$ fixed by

6059: the $h(z)$, for all $z\in \C^*$, is the space of ``Hodge cycles''.

6060: 

6061: One can then view Shimura varieties $Sh(G,X)$ as moduli spaces of

6062: Hodge structures in the following way. Let $(G,X)$ be a Shimura datum

6063: and $\rho:G \to \GL(V)$ a faithful representation. Since $G$ is

6064: reductive, there is a finite family of tensors $\tau_i$ such that

6065: \begin{equation}\label{tensors}

6066: G=\{ g\in \GL(V): g\tau_i=\tau_i \}.

6067: \end{equation}

6068: A point $x\in X$ is by construction a $G(\R)$ conjugacy class of

6069: morphisms $h_x: \bS\to G$, with suitable properties.

6070: 

6071: Consider data of the form $((W,h),\{ s_i \},\phi)$, where

6072: $(W,h)$ is a rational Hodge structure, $\{ s_i \}$ a finite

6073: family of Hodge cycles, and $\phi$ a $K$-level structure, for some

6074: $K\subset G(\A_f)$, namely a $K$-orbit of $\A_f$-modules {\em

6075: isomorphisms} $\phi: V(\A_f) \to W(\A_f)$, which maps $\tau_i$ to

6076: $s_i$. Isomorphisms of such data are isomorphisms $f:W\to W'$ of

6077: rational Hodge structures, sending $s_i\mapsto s_i'$,

6078: and such that $f\circ \phi=\phi' k$,  for some $k\in K$.

6079: 

6080: We assume that there exists an isomorphism of $\Q$-vector spaces

6081: $\beta:W\to V$ mapping $s_i \mapsto \tau_i$ and $h$ to $h_x$, for some

6082: $x\in X$.

6083: 

6084: One denotes by ${\rm Hodge}(G,X,K)$ the set data $((W,h),\{ s_i \},\phi)$.

6085: The Shimura variety $$Sh_K(G,X)=G(\Q)\backslash X\times

6086: G(\A_f)/K$$ is the moduli space of the isomorphism classes of data

6087: $((W,h),\{ s_i \},\phi)$, namely there is a map of ${\rm Hodge}(G,X,K)$

6088: to $Sh_K(G,X)$ (seen over $\C$), that descends to a bijection on

6089: isomorphism classes ${\rm Hodge}(G,X,K)/\sim$.

6090: 

6091: In such cases also one can consider degenerations of these data, both

6092: at the archimedean and at the nonarchimedean components. One then

6093: considers data $((W,h),\{ s_i \}, \phi,\tilde\beta)$, with

6094: a non-trivial {\em homomorphism} $\tilde\beta: W\to V$, which is a

6095: morphism of Hodge structures, and such that $\tilde\beta(\ell_{s_i})\subset

6096: \ell_{\tau_i}$. This yields noncommutative spaces inside which the

6097: classical Shimura variety sits as the set of classical points.

6098: 

6099: Quantum statistical machanical systems associated to Shimura varieties

6100: have been recently studied by Ha and Paugam in \cite{HaPa}. Given a

6101: faithful representation $\rho:G\to \GL(V)$ as above, there is an

6102: ``enveloping semigroup'' $M$, that is, a normal irreducible semigroup

6103: $M\subset \End(V)$ such that $M^\times =G$. Such semigroup can be used

6104: to encode the degenerations of the Hodge data described above.

6105: The data $(G,X,V,M)$ then determine a noncommutative space which

6106: describes the ``bad quotient'' $Sh^{nc}_K(G,X)=G(\Q)\backslash X\times

6107: M(\A_f)$ and is a moduli space for the

6108: possibly non-invertible data $((W,h),\{ s_i \},\phi)$.

6109: Its ``set of classical points'' is the Shimura variety

6110: $Sh(G,X)$.  The actual construction of the algebras involves

6111: some delicate steps, especially to handle the presence of stacky

6112: singularities (\cf \cite{HaPa}).

6113: 

6114: 

6115: \section{The adeles class space and the spectral realization}\label{adeles}

6116: 

6117: 

6118: 

6119: In this section we   describe a noncommutative space, the adele

6120: class space $X_\K$, associated to any global field $\K$, which leads

6121: to a spectral realization of the zeros of the Riemann zeta function

6122: for $\K=\Q$ and more generally of $L$-functions associated to Hecke

6123: characters. It also gives a geometric interpretation of the

6124: Riemann-Weil explicit formulas of number theory as a trace formula.

6125: This space is closely related for $\K=\Q$ with the space of

6126: commensurability classes of $\Q$-lattices described above. Rather

6127: than starting directly with its description we   first put the

6128: problem of finding the geometry of the set of prime numbers in the

6129: proper perspective.

6130: 

6131: \medskip

6132: {\bf The set of primes}

6133: \medskip

6134: 

6135: One of the main problems of arithmetic is to understand the

6136: distribution of the set of prime numbers

6137: $$

6138: \{2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59,

6139: 61, 67, 71, \ldots \}

6140: $$

6141: as a subset of the integers. To that effect one introduces the

6142: counting function

6143: $$

6144: \pi(n)=\,{\rm number}\;  {\rm of}\;  {\rm primes}\; p\leq n

6145: $$

6146: The problem is to understand the behavior of $\pi(x)$ when $x\to

6147: \infty$. One often hears that the problem comes from the lack of a

6148: ``simple" formula for $\pi(x)$. This is not really true and for

6149: instance in 1898 H. Laurent \cite{laurent} gave the following

6150: formula whose validity is an easy exercise in arithmetic,

6151: \begin{small}

6152: $$

6153: \pi(n)=\, 2+\,\sum_{k=5}^n\,\frac{e^{2\pi i \Gamma(k)/k}-1}{e^{-2\pi i

6154: /k}-1} \,,

6155: $$

6156: \end{small}

6157: where $ \Gamma(k)=(k-1)!$ is the Euler Gamma function.

6158: 

6159: \begin{figure}

6160: \begin{center}

6161: \includegraphics[scale=0.80]{pin3.eps}

6162: \end{center}

6163: \caption{Graphs of $\pi(x)$ and  Li$(x)$ \label{pili1}}

6164: \end{figure}

6165: 

6166: The problem with this formula is that it has no bearing on the

6167: asymptotic expansion of $\pi(x)$ when $x\to \infty$. Such an

6168: expansion was guessed by Gauss in the following form,

6169: $$

6170: \pi(x)=\, \int_0^x\,\frac{du}{\log(u)}\,+\;R(x)

6171: $$

6172: where the integral logarithm admits the asymptotic expansion,

6173: $$Li(x) =\int_0^x\,\frac{du}{\log(u)}

6174: \sim \sum \,(k-1)!\,\frac{x}{\log(x)^k}$$

6175: 

6176: \begin{figure}

6177: \begin{center}

6178: \includegraphics[scale=0.80]{pin4.eps}

6179: \end{center}

6180: \caption{Graph of $\pi(x)-$ Li$(x)$ \label{pili2}}

6181: \end{figure}

6182: 

6183: The key issue then is the size of the remainder $R(x)$.

6184: 

6185: \medskip

6186:  {\bf The Riemann hypothesis}

6187: \medskip

6188: 

6189: It asserts that this size is governed by

6190: \begin{equation}\label{rh}

6191: R(x)=\,O(\sqrt{x}\,\log(x)) .

6192: \end{equation}

6193: The Riemann Hypothesis is in fact a conjecture on the zeros of the

6194: zeta function

6195: \begin{equation}\label{zeta}

6196: \zeta(s)=\,\sum_1^\infty\,n^{-s} ,

6197: \end{equation}

6198: whose definition goes back to Euler,

6199: who showed the fundamental factorization

6200: \begin{equation}\label{eulerprod}

6201:  \zeta(s)=\,\prod_{\mathcal P}\,(1-\,p^{-s})^{-1}\,.

6202: \end{equation}

6203: 

6204: It extends to a meromorphic function in the whole complex plane $\C$

6205: and fulfills the functional equation

6206: \begin{equation}\label{functequ}

6207: \pi^{-s/2}\,\Gamma(s/2)\,\zeta(s)=\,\pi^{-(1-s)/2}\,\Gamma((1-s)/2)\,\zeta(1-s),

6208: \end{equation}

6209: so that the function

6210: \begin{equation}\label{zetaq}

6211: \zeta_\Q(s)=\pi^{-s/2}\,\Gamma(s/2)\,\zeta(s)

6212: \end{equation}

6213: admits the symmetry $s\mapsto 1-s$.

6214: 

6215: The Riemann conjecture asserts that all zeros of  $\zeta_\Q$ are on

6216: the critical line $\frac{1}{2} +\,i\,\R$. The reason why the

6217: location of the zeros of $\zeta_\Q$ controls the size of the

6218: remainder in \eqref{rh} is the explicit formulas that relate primes

6219: with these zeros. Riemann proved the following first instance of an

6220: ``explicit formula"

6221: \begin{equation}\label{explicit1}

6222: \pi'(x)=\,Li(x)-\,\sum_\rho\,Li(x^\rho)+\,\int_x^\infty\,\frac{1}{u^2-1}\;\frac{du}{u

6223: \log u}-\,\log 2

6224: \end{equation}

6225: where the sum is over the non-trivial (\ie complex) zeros of the

6226: zeta function, and where the function $\pi'(x)$ is given by

6227: $$

6228: \pi'(x)=\,\pi(x)\,+\,\,\frac{1}{2}\,\pi(x^{\frac{1}{2}})+

6229: \,\frac{1}{3}\,\pi(x^{\frac{1}{3}})\,+\cdots

6230: $$

6231: which gives by the Moebius inversion formula

6232: $$

6233: \pi(x)=\,\sum \,\mu(m)\,\frac{1}{m}\,\pi'(x^{\frac{1}{m}}) .

6234: $$

6235: 

6236: \bigskip

6237: \begin{figure}

6238: \begin{center}

6239: \includegraphics[scale=0.80]{rszeta.eps}

6240: \end{center}

6241: \caption{Zeros of zeta \label{zeros}}

6242: \end{figure}

6243: 

6244: 

6245: \medskip

6246: {\bf The generalized Riemann hypothesis}

6247: \medskip

6248: 

6249: The explicit formulas of Riemann were put in more modern form by A.

6250: Weil, as

6251: \begin{equation}\label{explicit}

6252: \,\wh h(0) + \wh h(1) -\sum_\rho\,\wh h(\rho)=\,\sum_{v}

6253: \int'_{\K^*_v} \, \frac{h(u^{-1})}{

6254:  |1-u|}\, d^* u \, ,

6255: \end{equation}

6256: where $\K$ is now an arbitrary global field, $v\in \Sigma_\K$ varies

6257: among the places of $\K$ and the integral is taking place over the

6258: locally compact field $\K_v$ obtained by completion of $\K$ at the

6259: place $v$. Also $\int'$ is  the pairing with the distribution on

6260: $\K_v$ which agrees with ${du \over \vert 1-u \vert}$ for $u \not= 1$

6261: and whose Fourier transform (relative to a selfdual choice of

6262: additive characters $\alpha_v$) vanishes at $1$. By definition a

6263: {\em global field} is a (countable) discrete cocompact subfield in a

6264: locally compact ring. This ring depends functorially on $\K$ and is

6265: called the ring $\A_\K$ of adeles of $\K$. The quotient

6266: 

6267: \begin{equation}\label{ideleclass}

6268: C_\K=\,{\rm GL}_1(\A_\K)/{\rm GL}_1(\K)

6269: \end{equation}

6270: is the locally compact group of idele classes of $\K$ which plays a

6271: central role in class field theory. In Weil's explicit formula the

6272: test function $h$ is in the Bruhat-Schwartz space $\cS(C_\K)$. The

6273: multiplicative groups ${\rm GL}_1(\K_v)=\K_v^*$ are embedded

6274: canonically as cocompact subgroups of $C_\K$. The sum on the left

6275: hand side takes place over the zeros of $L$-functions associated to

6276: Hecke characters. The function $\wh h$ is the Fourier transform of

6277: $h$. The generalized Riemann conjecture asserts that all the zeros

6278: of these $L$-functions are on the critical line $\frac{1}{2}

6279: +\,i\,\R$. This was proved by Weil when the global field $\K$ has

6280: non-zero characteristic, but remains open in the case when $\K$ is of

6281: characteristic zero, in which case it is a number field \ie a finite

6282: algebraic extension of the field $\Q$ of rational numbers.

6283: 

6284: 

6285: \medskip

6286:  {\bf

6287: Quantum Chaos $\to$ Riemann Flow ? }

6288: \medskip

6289: 

6290: For $E>0$ let $N(E)$ be the number of zeros of the Riemann zeta

6291: function $\zeta_\Q$ whose imaginary parts are in the open interval

6292: $]0,E[$. Riemann proved that the step function $N(E)$ can be written

6293: as the sum

6294: $$N(E)=\,  \langle N(E) \rangle  + N_{\rm osc} (E)$$

6295: of a smooth approximation $ \langle N(E) \rangle$  and a purely

6296: oscillatory function $N_{\rm osc} (E)$ and gave the following

6297: explicit form

6298: 

6299: \begin{equation}\label{riemcount}  \langle N(E) \rangle = \,\frac{E}{2\,\pi}\,(\log

6300: \frac{E}{2\,\pi} -1) + \frac{7}{8} +\,o(1)

6301: \end{equation}

6302: 

6303: for the smooth approximation.

6304: 

6305: There is a striking analogy between the behavior of the step

6306: function $N(E)$ and that of the function counting the number of

6307: eigenvalues of the Hamiltonian $H$ of the quantum system obtained

6308: after quantization of a chaotic dynamical system, which is at

6309: centerstage in the theory of {\em quantum chaos}. A  comparison of

6310: the asymptotic expansions of the oscillatory terms in both cases,

6311: namely

6312: 

6313: $$N_{\rm osc} (E) \sim \frac{1}{ \pi} \sum_{\gamma_p} \sum_{m=1}^{\infty}

6314: \frac{1}{ m} \, \frac{1}{ 2{\rm sh} \left( \frac{m\lambda_p }{

6315: 2}\right)} \, \sin (m \, E \, T_{\gamma}^{\#})

6316: $$

6317: 

6318: for the quantization of a chaotic dynamical system, and

6319: 

6320: 

6321: $$

6322: N_{\rm osc} (E) \sim \frac{-1 }{ \pi} \sum_p \sum_{m=1}^{\infty}

6323: \frac{1}{ m} \, \frac{1 }{ p^{m/2}} \, \sin \, (m \, E \, \log \, p)

6324: \,

6325: $$

6326: 

6327: for the Riemann zeta function, gives precious indications on the

6328: hypothetical {\em Riemann flow} that would make it possible to identify the

6329: zeros of zeta as the spectrum of an Hamiltonian. For instance the

6330: periodic orbits of the flow should be labeled by the prime numbers

6331: and the corresponding periods $T_p$ should be given by the $\log p$.

6332: However a closer look reveals an overall minus sign that forbids any

6333: direct comparison.

6334: 

6335: 

6336: \bigskip

6337: \begin{figure}

6338: \begin{center}

6339: \includegraphics[scale=0.90]{graph1.eps}

6340: \end{center}

6341: \caption{Counting zeros of zeta \label{zeros1}}

6342: \end{figure}

6343: 

6344: 

6345: \medskip

6346:  {\bf

6347: Spectral realization as an absorption spectrum}

6348: \medskip

6349: 

6350: 

6351: The above major sign obstruction was bypassed in \cite{CoRH} using

6352: the following basic distinction between observed spectra in physics.

6353: When the light coming from a hot chemical element is decomposed

6354: through a prism, it gives rise to bright emission lines on a dark

6355: background, and the corresponding frequencies are a signature of its

6356: chemical composition. When the light coming from a distant star is

6357: decomposed through a prism, it gives rise to dark lines, called

6358: absorption lines,  on a white background. The spectrum of the light

6359: emitted by the sun was the first observed example of an absorption

6360: spectrum. In this case the absorption lines were discovered by

6361: Fraunhofer. The chemicals in the outer atmosphere of the star absorb

6362: the corresponding frequencies in the white light coming from the

6363: core of the star.

6364: 

6365: \bigskip

6366: \begin{figure}

6367: \begin{center}

6368: \includegraphics[scale=0.8]{spectres7.eps}

6369: \end{center}

6370: \caption{The two kinds of Spectra \label{spectra}}

6371: \end{figure}

6372: 

6373: The simple idea then is that, because of the minus sign above, one

6374: should look for the spectral realization of the zeros of zeta not as

6375: a usual emission spectrum but as an absorption spectrum. Of course

6376: by itself this idea does not suffice to get anywhere since one needs

6377: the basic dynamical system anyway. The  adele class space, namely the

6378: quotient

6379: $$

6380: X_\K=\A_\K/ \K^*

6381: $$

6382: introduced in \cite{CoRH}, does the job as shown there. The action of

6383: the idele class group $C_\K$ on the adele class space is simply given

6384: by multiplication. In particular the idele class group $C_\K$ acts on

6385: the suitably defined Hilbert space $L^2(X_\K)$ and the  zeros of

6386: $L$-functions give the absorption spectrum, with non-critical zeros

6387: appearing as resonances.

6388: 

6389: \medskip

6390: 

6391: Exactly as adeles, the adele class space $X_\K$ involves all the

6392: places of $\K$. If in order to simplify, one restricts to a finite

6393: set of places one still finds a noncommutative space but one can

6394: analyze the action of the analogue of $C_\K$ and compute its trace

6395: after performing a suitable cutoff (necessary in all cases to see

6396: the missing lines of an absorption spectrum). One gets a trace

6397: formula

6398: $$

6399: {\rm Trace} \, (R_{\Lambda} \, U(h)) = 2h (1) \log' \Lambda +

6400: \sum_{v \in S} \int'_{\K^*_v} \frac{h(u^{-1})}{  \vert 1-u \vert} \,

6401: d^* u + o(1),

6402: $$

6403:  where the terms in the right hand side are exactly the same as  in Weil's explicit

6404:  formula \eqref{explicit}. This is very encouraging since at least it

6405:  gives  geometric meaning to the complicated terms of \eqref{explicit}

6406:  as the contributions of the periodic orbits to the computation of

6407:  the trace.

6408: 

6409: In particular it gives a perfect interpretation  of the smooth

6410: function $  \langle N(E) \rangle $ approximating the counting $N(E)$

6411: of the zeros of zeta, from counting the number of states of the

6412:  one dimensional quantum system with Hamiltonian

6413:  $$

6414: h (q, p) =  \, 2\pi q\,p

6415: $$

6416: which is just the generator of the scaling group. Indeed the

6417: function $\frac{E}{2\,\pi}\,(\log \frac{E}{2\,\pi} -1)$ of Riemann

6418: formula \eqref{riemcount} appears as the number of missing degrees

6419: of freedom in the number of quantum states for the above system, as

6420: one obtains from the simple computation of the area of the region

6421: $$

6422: B_+=\,\{(p,q)\in[0,\Lambda]^2\,;\,h(p,q)\leq E\}

6423: $$

6424: 

6425: $${\rm Area}(B_+)=\frac{E}{2\pi} \times 2 \log \Lambda\,

6426:  - \frac{E}{2\pi} \left( \log \frac{E}{2\pi} - 1 \right)

6427: $$

6428: 

6429: while the term $\frac{E}{2\pi} \times 2 \log \Lambda$ corresponds to

6430: the number of degrees of freedom of white light. A careful

6431: computation gives not only the correction term of $\frac{7}{8}$ in

6432: \eqref{riemcount} but all the remaining $o(1)$ terms.

6433: 

6434: 

6435: 

6436: \bigskip

6437: \begin{figure}

6438: \begin{center}

6439: \includegraphics[scale=0.8]{gr0.eps}

6440: \end{center}

6441: \caption{Counting quantum states}\label{countingstates}

6442: \end{figure}

6443: 

6444: 

6445: Finally it was shown in \cite{CoRH} that the generalized Riemann

6446: hypothesis is equivalent to the validity of a {\em global} trace

6447: formula, but this is only one of many equivalent reformulations of the

6448: Riemann hypothesis.\footnote{There is a running joke, inspired

6449: by the European myth of Faust, about a mathematician trying to

6450: bargain with the devil for a proof of the Riemann hypothesis ...}

6451: 

6452: \medskip

6453: 

6454: 

6455: 

6456: \section{Thermodynamics of endomotives and the Tehran program}\label{program}

6457: 

6458: In many ways the great virtue of a problem like RH comes from the

6459: developments that it generates. At first sight it does not appear as

6460: having any relation with geometry, and its geometric nature

6461: gradually emerged in the twentieth century mainly because of the

6462: solution of Weil in the case of global fields of positive

6463: characteristic.

6464: 

6465:  We    outline a program, current joint

6466: work of Katia Consani and the two authors, to adapt Weil's proof for

6467: the case of global fields of positive characteristic to the case of

6468: number fields.\footnote{This program was first announced in a

6469: lecture at IPM Tehran in September 2005, hence we refer to it as ``the Tehran

6470: program''.}

6471: 

6472: 

6473: 

6474: \medskip

6475:  {\bf

6476: Function Fields}

6477: \medskip

6478: 

6479: Given a global field $\K$ of positive characteristic, there exists a

6480: finite field $\F_q$ and a  smooth projective curve $C$ defined over

6481: $\F_q$ such that $\K$ is the field of $\F_q$-valued rational

6482: functions on $C$. The analogue (Artin, Hasse, Schmidt) of the zeta

6483: function is

6484: $$

6485: \zeta_\K(s)=\,\prod_{v\in\Sigma_K}\,(1-q^{-f(v)s})^{-1}

6486: $$

6487: where  $\Sigma_\K$ is the set of places of $\K$ and $f(v)$ is the

6488: degree (see below) of the place $v\in \Sigma_\K$.

6489: 

6490: The functional equation takes the form

6491: $$

6492: q^{(g-1)(1-s)}\,\zeta_\K(1-s)=\,q^{(g-1)s} \,\zeta_\K(s)

6493: $$

6494: where $g$ is the genus of  $C$.

6495: 

6496: The analogue of the Riemann conjecture for such global fields was

6497: proved by Weil (1942) who developed algebraic geometry in that

6498: context. Weil's proof rests on two steps.

6499: 

6500: \begin{itemize}

6501: 

6502: \item (A) Explicit Formula

6503: \medskip

6504: 

6505: \item (B) Positivity

6506: 

6507: 

6508: \end{itemize}

6509: 

6510: 

6511: Both are based on the geometry of the action of the Frobenius on the

6512: set $C(\bar \F_q)$ of points of $C$ over an algebraic closure $\bar

6513: \F_q$ of  $\F_q$. This set $C(\bar \F_q)$ maps canonically to the

6514: set $\Sigma_\K$ of places of $\K$ and the degree of a place $v\in

6515: \Sigma_\K$ is the number of points in the orbit of the

6516:  Frobenius acting on the fiber of the projection

6517: $$

6518: C(\bar \F_q) \to \Sigma_\K\,.

6519: $$

6520: The analogue

6521: $$\#\{C(\F_{q^j})\}=\,\sum\,(-1)^k\,

6522: {\rm Tr}({\rm Fr}^{* j}|H^k_{{\rm et}}(\bar C,\Q_\ell))$$

6523: of the Lefschetz fixed point formula makes it possible to compute

6524: the number $\#\{C(\F_{q^j})\}$ of points with coordinates in the

6525: finite extension $\F_{q^j}$ from the action of ${\rm Fr}^*$ in the

6526: etale cohomology group $H^1_{{\rm et}}(\bar C,\Q_\ell)$, which does

6527: not depend upon the choice of the $\ell$-adic coefficients

6528: $\Q_\ell$.

6529: 

6530: This shows that the zeta function is a rational fraction

6531: $$

6532: \zeta_\K(s)=\,\frac{P(q^{-s})}{(1-q^{-s})(1-q^{1-s})}

6533: $$

6534: 

6535: where the polynomial $P$ is the characteristic polynomial of  the

6536: action of ${\rm Fr}^*$ in $H^1$.

6537: 

6538: 

6539: 

6540: The analogue of the Riemann conjecture for global fields of

6541: characteristic  $p$ means that its eigenvalues  \ie the complex

6542: numbers $\lambda_j$ of the factorization

6543: $$

6544: P(T)=\,\prod \,(1-\lambda_j\, T)

6545: $$

6546: are of modulus $|\lambda_j|=q^{1/2}$.

6547: 

6548: The main ingredient in the proof of Weil is the notion of

6549: correspondence, given by divisors in $C\times C$. They can be viewed

6550: as multivalued maps,

6551: $$Z\;:\;C\to C,\ \ \ \  P\mapsto Z(P).$$

6552: 

6553: Two correspondences are equivalent if

6554: they differ by a principal divisor,

6555: $$

6556: U\sim V \Leftrightarrow U-V=\,(f)

6557: $$

6558: The composition of correspondences is

6559: $$

6560: Z=\,Z_1 \star Z_2 \,, \ \ \  Z_1 \star Z_2(P)=\,Z_1 ( Z_2(P))

6561: $$

6562: and the adjoint is given using the transposition

6563: $\sigma(x,y)=\,(y,x)$ by

6564: $$

6565: Z'=\, \sigma(Z)\,.

6566: $$

6567: The degree $d(Z)$ of a correspondence is defined, independently of a

6568: generic point $P\in C$ by,

6569: $$

6570: d(Z)=\,Z \,\bullet \,(P\times C)\,,

6571: $$

6572: where $\bullet$ is the intersection number. One has a similar

6573: definition of the codegree

6574: $$ \quad d^{\,'}(Z)=\,Z

6575: \,\bullet \,(C\times P) $$

6576: 

6577:  Weil defines the {\em Trace} of a

6578: correspondence as follows

6579: $$

6580: {\rm Trace}(Z)=\,d(Z)+\,d^{\,'}(Z)-\,Z\bullet \Delta

6581: $$

6582: where $\Delta$ is the identity correspondence. The main step in

6583: Weil's proof is

6584: 

6585: \smallskip

6586: 

6587: \begin{thm} (Weil)

6588: The following positivity holds : ${\rm Trace}(Z \star Z')>0$ unless

6589: $Z$ is a trivial class.

6590: \end{thm}

6591: 

6592: \smallskip

6593: 

6594:  Clearly

6595: if one wants to have any chance at imitating the steps of Weil's

6596: proof of RH for the case of number fields one needs to have an

6597: analogue of the points of $C(\bar \F_q)$ and the action of the

6598: Frobenius, of the etale cohomology and of the unramified extensions

6599: $\K\otimes_{\F_q}\,\F_{q^n}$ of $\K$.

6600: 

6601: \medskip

6602:  {\bf

6603:  Endomotives and Galois action}

6604: \medskip

6605: 

6606: The adele class space $X_\K$ of a global field admits a natural

6607: action of the idele class group $C_\K$ and as such is on the adelic

6608: side of the class field theory isomorphism. In order to obtain a

6609: description of this space which is closer to geometry one needs to

6610: pass to the Galois side of class field theory. In the case

6611:  $\K=\Q$, it is possible to present the adele class space in a fairly simple

6612:  manner not involving adeles thanks to its intimate relation with

6613:  the  space of $1$-dimensional $\Q$-lattices of section \ref{Qlatt}.

6614: The direct interpretation of the action of the Galois group of $\bar

6615: \Q/\Q$ on the values of fabulous states for the BC-system (section

6616: \ref{Qlatt}) then suggests that one should be able to construct

6617: directly the space $X_\Q$ with a canonical action of the Galois

6618: group of $\bar \Q/\Q$.

6619: 

6620: This was done in \cite{CCM} thanks to an extension of the notion of

6621: Artin motives, called {\em endomotives}. Following Grothendieck, one

6622: can reformulate Galois theory over a field $\K$ as the equivalence of

6623: the category of reduced commutative finite dimensional algebras over

6624: $\K$ with the category of continuous actions of the Galois group $G$

6625: of $\bar \K/\K$ on finite sets. By construction the algebra of the

6626: BC-system is a crossed product of a commutative algebra $A$ by a

6627: semi-group. When working over $\K=\Q$ which is essential in the

6628: definition of fabulous states, the algebra $A$ is simply the group

6629: ring $\Q[\Q/\Z]$ of the torsion group $\Q/\Z$. Thus

6630: $$

6631: A=\, \varinjlim \,A_n \,,\quad A_n=\,\Q[\Z/n\Z]

6632: $$

6633: and we are dealing with a projective limit of Artin motives. The key

6634: point then is to keep track of the corresponding action of the

6635: Galois group $G$ of $\bar \K/\K$, $\K=\Q$. The Galois-Grothendieck

6636: correspondence associates to a reduced commutative finite

6637: dimensional algebra $B$ over $\K$ the set of characters of $B$ with

6638: values in $\bar \K$ together with the natural action of $G$. This

6639: action is non-trivial for the algebras $A_n=\,\Q[\Z/n\Z]$ where it

6640: corresponds to the cyclotomic theory.

6641: 

6642: One can then recover the Bost--Connes system with its natural Galois symmetry

6643: in a conceptual manner which extends to the general context of

6644: semigroup actions on projective systems of Artin motives. These

6645: typically arise from self-maps of algebraic varieties.

6646: Given a pointed algebraic

6647: variety $(Y,y_0)$ over a field $\K$ and a countable unital abelian

6648: semigroup $S$ of finite endomorphisms of $(Y,y_0)$, unramified over

6649: $y_0\in Y$ one constructs a projective system of Artin motives $X_s$

6650: over $\K$ from these data as follows. For $s\in S$, one sets

6651: \begin{equation}\label{Xs}

6652: X_s=\{ y\in Y:\, s(y)=y_0 \}.

6653: \end{equation}

6654: For a pair $s,s'\in S$, with $s'=sr$, the map $\xi_{s',s}: X_{sr}\to

6655: X_s$ is given by

6656: \begin{equation}\label{Xsr}

6657: X_{sr} \ni y \mapsto r(y)\in X_s.

6658: \end{equation}

6659: This defines a projective system indexed by the semigroup $S$ itself

6660: with partial order given by divisibility. We let $X=\varprojlim_s

6661: X_s$.

6662: 

6663: Since $s(y_0)=y_0$, the base point $y_0$ defines a component $Z_s$

6664: of $X_s$ for all $s\in S$. Let $\xi_{s',s}^{-1}(Z_s)$ be the inverse

6665: image of $Z_s$ in $X_{s'}$. It is a union of components of $X_{s'}$.

6666: This defines a projection $e_s$ onto an open and closed subset

6667: $X^{e_s}$ of the projective limit $X$. One then shows (\cite{CCM})

6668: that the semigroup $S$ acts on the projective limit $X$ by partial

6669: isomorphisms $\rho_s: X \to X^{e_s}$ defined by the property that

6670: \begin{equation}\label{endoS}

6671: \xi_{su}(\rho_s(x))=\xi_u(x), \, \forall u\in S, \forall x\in X.

6672: \end{equation}

6673: 

6674: The BC-system is obtained from the pointed algebraic variety

6675: $(\bG_m(\Q),1)$ where  the affine group scheme $\bG_m$ is the

6676: multiplicative group. The semigroup $S$ is the semigroup of non-zero

6677: endomorphisms of $\bG_m$. These correspond to maps of the form

6678: $u\mapsto u^n$ for some non-zero $n\in \Z$, and one restricts to

6679: $n\in \N^*$.

6680: 

6681: In this class of examples one has an ``equidistribution'' property,

6682: by which the uniform normalized counting measures $\mu_s$ on $X_s$

6683: are compatible with the projective system and define a probability

6684: measure on the limit $X$. Namely, one has

6685: \begin{equation}\label{measlim}

6686: \xi_{s',s} \mu_s = \mu_{s'},\ \ \ \  \forall s,s'\in S.

6687: \end{equation}

6688: This follows from the fact that the number of preimages of a point

6689: under $s\in S$ is equal to $\deg s$. This provides exactly the data

6690: which makes it possible to perform the thermodynamical analysis of such

6691: endomotives. This gives a rather unexplored new territory since even

6692: the simplest examples beyond the BC-system  remain to be

6693: investigated. For instance let $Y$ be an elliptic curve defined over

6694: $\K$. Let $S$ be the semigroup of non-zero endomorphisms of $Y$.

6695: This gives rise to an example in the general class described above.

6696: When the elliptic curve has complex multiplication, this gives rise

6697: to a system which, in the case of a maximal order, agrees with the

6698: one constructed in \cite{CMR}. In the case without complex

6699: multiplication, this provides an example of a system where the

6700: Galois action does not factor through an abelian quotient.

6701: 

6702: 

6703: 

6704: \medskip

6705:  {\bf

6706:  Frobenius as dual of the time evolution}

6707: \medskip

6708: 

6709: The Frobenius is such a universal symmetry in characteristic $p$,

6710: owing to the linearity of the map $x\mapsto x^p$ that it is very hard to

6711: find an analogue of such a far reaching concept in characteristic

6712: zero. As we   now explain, the  classification of type III

6713: factors provides the basic ingredient which when combined with

6714: cyclic cohomology makes it possible to analyze  the thermodynamics of a

6715: noncommutative space and get an analogue of the action of the

6716: Frobenius on etale cohomology.

6717: 

6718: 

6719: 

6720: The key ingredient is that noncommutativity generates a time

6721: evolution at the ``measure theory" level.  While it had been long

6722: known by operator

6723:  algebraists that the theory of

6724:  von-Neumann algebras represents

6725:  a far reaching extension of measure theory, the

6726:   main surprise which occurred at the

6727: beginning of the seventies in (Connes \cite{Co_2}) following

6728: Tomita's theory is that such an algebra $M$ inherits

6729:  from its noncommutativity a god-given time evolution:

6730: \begin{equation}

6731: \delta\,: \quad \R \quad\longrightarrow \quad{\rm Out}\,M=\,{\rm

6732: Aut}\,M/{\rm Inn}\,M

6733: \end{equation}

6734: 

6735: where ${\rm Out}\,M ={\rm Aut}\,M/{\rm Inn}\,M$ is the quotient of

6736: the group of automorphisms of $M$ by the normal subgroup of inner

6737: automorphisms. This led in \cite{Co_2} to the reduction from type

6738: III to type II and their automorphisms and eventually to the

6739: classification of injective factors.

6740: 

6741: 

6742: \noindent They are classified by their module,

6743: \begin{equation}

6744: {\rm Mod} (M) \mathop{\subset}_{\sim} \ \R_+^* \, , \label{eq:(20)}

6745: \end{equation}

6746: which is a virtual closed subgroup of $\R_+^*$ in the sense of

6747: G.~Mackey, i.e. an ergodic action of $\R_+^*$, called the flow of

6748: weights \cite{CT}. This invariant was first defined and used in

6749: \cite{Co_2}, to show in particular the existence of hyperfinite

6750: factors which are not isomorphic to Araki-Woods factors.

6751: 

6752: The ``measure theory" level \ie the set-up of von-Neumann algebras

6753: does not suffice to obtain the relevant cohomology theory and one

6754: needs to be given a weakly dense subalgebra $\cA\subset M$ playing

6755: the role of smooth functions on the noncommutative space. This

6756: algebra will play a key role when cyclic cohomology is used at a

6757: later stage. At first one only uses its norm closure $A=\bar \cA$ in

6758: $M$ and assumes that it is globally invariant under the modular

6759: automorphism group $\sigma_t^\varphi$ of a faithful normal state

6760: $\varphi$ on $M$. One can then proceed with the thermodynamics of

6761: the $C^*$ dynamical system $(A,\sigma_t)$. By a very simple

6762: procedure assuming that KMS states at low temperature are of type I,

6763: one obtains a ``cooling morphism" $\pi$ which is a morphism of

6764: algebras from the crossed product $\hat \cA=\,\cA

6765: \rtimes_{\sigma}\R$ to a type I algebra of compact operator valued

6766: functions on a canonical $\R^*_+$-principal bundle

6767: $\tilde\Omega_\beta$ over the space $\Omega_\beta$ of type I

6768: extremal KMS$_\beta$ states fulfilling a suitable regularity

6769: condition (\cf \cite{CCM}). Any $\varepsilon\in \Omega_\beta$ gives

6770: an irreducible representation $\pi_\varepsilon$ of $\cA$ and the

6771: choice of its essentially unique extension to $\hat \cA$ determines

6772: the fiber of the $\R^*_+$-principal bundle $\tilde\Omega_\beta$. The

6773: cooling morphism is then given by,

6774: \begin{equation}

6775: \pi_{\varepsilon,H}(\int \,x(t)\,U_t\,dt)=\,\int

6776: \,\pi_\varepsilon(x(t))\,e^{itH}\,dt .

6777: \end{equation}

6778: This morphism is equivariant for the dual action $\theta_\lambda \in

6779: \Aut(\hat \cA)$ of $\R^*_+$,

6780: \begin{equation}\label{dualaction}

6781: \theta_\lambda(\int \,x(t)\,U_t\,dt)=\,\int

6782: \,\lambda^{it}\,x(t)\,U_t\,dt.

6783: \end{equation}

6784: 

6785: The key point is that the range of the morphism $\pi$ is contained

6786: in an algebra of functions on $\tilde\Omega_\beta$ with values in

6787: trace class operators. In other words modulo a Morita equivalence

6788: one lands in the commutative world provided one lowers the

6789: temperature.

6790: 

6791: 

6792: The interesting space is obtained by ``distillation" and is simply

6793: given by the cokernel of the cooling morphism $\pi$ but this does

6794: not make sense in the category of algebras and algebra homomorphisms

6795: since the latter is not even an additive category. This is where

6796: cyclic cohomology enters the scene : the category of cyclic modules

6797: is an abelian category with a natural functor from the category of

6798: algebras and algebra homomorphisms.

6799: 

6800: Cyclic modules are modules of the cyclic category  $\Lambda$ which

6801: is a small category, obtained by enriching with {\it cyclic

6802: morphisms} the familiar {\it simplicial category} $\Delta$ of

6803: totally ordered finite sets and increasing maps. Alternatively,

6804: $\Lambda$ can be defined by means of its ``cyclic covering'', the

6805: category $E \Lambda$. The latter has one object $(\Zb , n)$ for each

6806: $n \geq 0$ and the morphisms $f : (\Zb , n) \to (\Zb , m)$ are given

6807: by non decreasing maps $f : \Zb \to \Zb \ $, such that $ f(x+n) =

6808: f(x)+m \qqq  x \in \Zb$. One has $\Lambda = E \Lambda / \Zb$, with

6809: respect to the obvious action of $\Zb$ by translation. To any

6810: algebra $\cA$ one associates a module $\cA^{\natural}$ over the category

6811: $\Lambda$ by assigning to each $n$ the $(n+1)$ tensor power

6812: $\cA\otimes \cA\cdots \otimes \cA$. The cyclic morphisms correspond to the

6813: cyclic permutations of the tensors while the face and degeneracy

6814: maps correspond to the algebra product of consecutive tensors and

6815: the insertion of the unit. The corresponding functor $\cA\to

6816: \cA^\natural$ gives a linearization of the category of associative

6817: algebras and cyclic cohomology appears as a derived functor.

6818: 

6819: One can thus define the {\em distilled} module $D(\cA,\varphi)$ as

6820: the cokernel of the cooling morphism  and consider the action of

6821: $\R^*_+$ (obtained from the above equivariance) in the cyclic

6822: homology group $HC_0(D(\cA,\varphi))$. As shown in \cite{CCM} this

6823: in the simplest case of the BC-system gives a cohomological

6824: interpretation of the above spectral realization of the zeros of the

6825: Riemann zeta function (and of Hecke L-functions).

6826: 

6827: One striking feature is that the KMS strip (\cf Figure \ref{FigKMS})

6828: becomes canonically identified in the process with the critical

6829: strip of the zeta function (recall that $\beta >1$) by

6830: multiplication by $i=\sqrt{ -1}$.

6831: 

6832: This cohomological interpretation combines with the above theory of

6833: endomotives to give a natural action of the Galois group $G$ of

6834: $\bar\Q/\Q$ on the above cohomology. This action factorizes  to the

6835: abelianization $G^{ab}$ and the corresponding decomposition

6836: according to characters of $G^{ab}$ corresponds to the spectral

6837: realization of L-functions.

6838: 

6839: The role of the invariant $S(M)$ in the classification of factors or

6840: of the more refined flow of weights mentioned above, is very similar

6841: to the role of the module of local or global fields and the Brauer

6842: theory of central simple algebras. In fact there is a striking

6843: parallel (see \cite{CCM}) between the lattice of unramified

6844: extensions $\K\to \,\K\otimes_{\F_q}\,\F_{q^n}$ of a global field of

6845: characteristic $p$ and the lattice of extensions of a factor $M$ by

6846: the crossed product algebras  $M\to \,M\rtimes_{ \sigma_T}\Z$. Using

6847: the algebraic closure of $\F_q$ \ie the operation $\K\to

6848: \K\otimes_{\F_q}\,\bar \F_{q}$ corresponds to passing to the dual

6849: algebra $M\to \,M\rtimes_\sigma \R$ and the dual action corresponds

6850: to the Frobenius automorphism when as above the appropriate

6851: cohomological operations (distillation and $HC_0)$ are performed.

6852: 

6853: \bigskip

6854: \bigskip

6855: \begin{center}

6856: \begin{tabular}{|c|c|}

6857: \hline &  \\

6858:  \bf{Global field $\K$} &\ \bf{Factor $M$} \\

6859: &\\ \hline &  \\

6860: ${\rm Mod}\,\K\subset \R_+^*$  &\ ${\rm Mod}\,M\subset \R_+^*$  \\

6861: &\\

6862: \hline &  \\

6863:  $\K\to \,\K\otimes_{\F_q}\,\F_{q^n}

6864: $ &\ $M\to \,M\rtimes_{ \sigma_T}\Z$ \\

6865: &\\ \hline &  \\ $K\to

6866: \K\otimes_{\F_q}\,\bar \F_{q}$ &\ $M\to \,M\rtimes_\sigma \R$ \\

6867: &\\ \hline &  \\ \bf{Points}

6868: $C(\bar \F_{q})$ &\ $\Gamma\subset$ \bf{$X_\Q$} \\

6869: &\\ \hline

6870: \end{tabular}

6871: \end{center}

6872: 

6873: \bigskip

6874: \bigskip

6875: \bigskip

6876: 

6877: \medskip

6878: A notable difference with the original Hilbert space theoretic

6879: spectral realization of \cite{CoRH} is that while in the latter case

6880: only the critical zeros were appearing directly (the possible

6881: non-critical ones appearing as resonances), in the cyclic homology

6882: set-up it is more natural to use everywhere the ``rapid decay"

6883: framework (advocated in \cite{Meyer}) so that all zeros appear on

6884: the same footing. This eliminates the difficulty coming from the

6885: potential non-critical zeros, so that the trace formula is much

6886: easier to prove and reduces to the Riemann-Weil explicit formula.

6887: However, it was not obvious how to obtain a direct geometric proof

6888: of this formula from the $S$-local trace formula of \cite{CoRH}.

6889: This was done in \cite{Meyer}, showing that the noncommutative

6890: geometry framework makes it possible to give a geometric

6891: interpretation of the Riemann-Weil explicit formula. While the

6892: spectral side of the trace formula is given by the action on the

6893: cyclic homology of the distilled space, the geometric side is given

6894: as follows \cite{CCM}.

6895: 

6896: 

6897: \medskip

6898: 

6899: \begin{thm} \label{geom} Let $h

6900: \in S (C_\K)$. Then the following holds:

6901: \begin{equation}\label{trace3}

6902: \Trace \; \left(\vartheta(h)|_{ \cH^1} \right)=\,\wh h(0) + \wh h(1)

6903: -\,\Delta\bullet \Delta\;h(1) -\,\sum_{v} \int_{(\K^*_v,e_{\K_v})}'

6904: \, \frac{h(u^{-1})}{

6905:  |1-u|}\, d^* u \, .

6906: \end{equation}

6907: \end{thm}

6908: 

6909: \medskip

6910: 

6911: We refer to \cite{CCM} for the detailed notations which are

6912: essentially those of  \cite{CoRH}. The origin  of the terms in the

6913: geometric side of the trace formula  comes from the Lefschetz

6914: formula by Atiyah-Bott \cite{atbo} and its adaptation by

6915: Guillemin-Sternberg (\cf \cite{gui}) to the distribution theoretic

6916: trace for flows on manifolds, which is a variation on the theme of

6917: \cite{atbo}. For the action of $C_\K$ on the adele class space

6918: $X_\K$ the relevant periodic orbits on which the computation

6919: concentrates turn out to form also the classical points of the

6920: noncommutative space $X_\K \backslash C_\K$ distilled in the above

6921: sense from $X_\K$. This ``classical" subspace of $X_\K\backslash

6922: C_\K$ is given by

6923: \begin{equation}\label{defcurve}

6924: \Gamma_\K=\,\cup \; C_\K\,[v]\,,\quad v\in \Sigma_\K .

6925: \end{equation}

6926: where for each place $v\in\Sigma_\K$, one lets $[v]$ be the adele

6927: \begin{equation}\label{base}

6928: [v]_w=1\qqq w\neq v\,,\quad [v]_v=0\,.

6929: \end{equation}

6930: 

6931: 

6932: In the function field case, one has a {\em non-canonical}

6933: isomorphism of the following form.

6934: 

6935: 

6936: \begin{prop} \label{Zspace}

6937: Let $\K$ be the function  field of an algebraic curve $C$ over

6938: $\F_q$. Then the  action of the Frobenius on $Y=C(\bar \F_q)$ is

6939: isomorphic to the action of $q^\Z$ on the quotient

6940: $$

6941: \Gamma_\K/C_{\K,1}.

6942: $$

6943: \end{prop}

6944: 

6945: 

6946: \begin{center}

6947: \begin{figure}

6948: \includegraphics[scale=1]{primecurve4.eps}

6949: \caption{The classical points of the adeles class space

6950: \label{Figpoints}}

6951: \end{figure}

6952: \end{center}

6953: 

6954: 

6955: 

6956: In the case $\K=\Q$ the space $\Gamma_\Q/C_{\Q,1}$ appears as the

6957: union of periodic orbits of period $\log p$ under the action of

6958: $C_\Q/C_{\Q,1}\sim \R$ (\cf Figure \ref{Figpoints}). This gives a

6959: first approximation to the sought for space $Y=C(\bar \F_q)$ in

6960: characteristic zero. One important refinement is obtained from the

6961: subtle nuance between the adelic description of $X_\Q$ and the finer

6962: description in terms of  the endomotive obtained from the pointed

6963: algebraic variety $(\bG_m(\Q),1)$. The second description keeps

6964: track of the Galois symmetry and as in proposition \ref{Zspace} the

6965: isomorphism of the two descriptions is non canonical.

6966: 

6967: 

6968: 

6969: \smallskip

6970: 

6971: At this point we have, in characteristic zero,  several of the

6972: geometric notions which are the analogues of the ingredients of

6973: Weil's proof and it is natural to try and imitate the steps of his

6974: proof. The step (A), \ie the explicit formula is taken care of by

6975: Theorem \ref{geom}. That what remains is to prove a positivity is a

6976: well known result of A. Weil (\cf \cite{EB}) which states  that RH

6977: is equivalent to the positivity of the distribution entering in the

6978: explicit formulae. Thanks to the above $H^1$ obtained as the cyclic

6979: homology of the distilled module, Weil's reformulation can be stated

6980: as follows.

6981: 

6982: \begin{thm}  \label{pos} The following two conditions are

6983: equivalent.

6984: \begin{itemize}

6985: \item All $L$-functions with Gr\"ossencharakter on $\K$ satisfy the Riemann

6986: Hypothesis.

6987: \item $\Trace\,\vartheta(f\,\star\,

6988: f^\sharp)|_{ \cH^1} \, \geq \,0$, for all $f\in\cS (C_\K)$.

6989: \end{itemize}

6990: \end{thm}

6991: 

6992: 

6993: Here we used the notation

6994: \begin{equation}\label{conv}

6995: f=\,f_1 \star f_2 \,,\ \ \ \text{ with } \ (f_1 \star f_2)(g)=\,\int

6996: \,f_1(k)\, f_2(k^{-1}\,g) \,d^*g

6997: \end{equation}

6998: for the convolution of functions, using the multiplicative Haar

6999: measure $d^*g$, and for the adjoint

7000: \begin{equation}\label{adj1}

7001: f\to f^{\sharp}\,,\quad f^{\sharp}(g)=\,\vert\, g\vert^{-1}\,\bar

7002: f(g^{-1}) .

7003: \end{equation}

7004: 

7005: The role of the specific correspondences used in Weil's proof of RH

7006: in positive characteristic is played by the test functions $ f\in

7007: \cS (C_\K) $. More precisely the scaling map which replaces $f(x)$

7008: by $f(g^{-1} x)$ has a graph, namely the set of pairs $(x, g^{-1} x)

7009: \in X_\K \times X_\K$, which we view as a correspondence $Z_g$.

7010: Then, given a test function $f$ on the ideles classes, one assigns

7011: to $f$ the linear combination

7012: \begin{equation}\label{corresp}

7013: Z(f)=\,\int f(g) Z_g d^*g

7014: \end{equation}

7015: of the above graphs, viewed as a ``divisor" on $X_\K \times X_\K$.

7016: 

7017: The analogs of the degrees $d(Z)$ and codegrees $d^{\,'}(Z)=\,d(Z')$

7018: of correspondences in the context of Weil's proof are given, for the

7019: degree, by

7020: \begin{equation}\label{degree}

7021: d(Z(h))=\,\wh h(1)=\,\int\,h(u)\, \vert u \vert  \, d^* \, u ,

7022: \end{equation}

7023: so that the degree $d(Z_g)$ of the correspondence $Z_g$ is equal to

7024: $|g|$. Similarly, for the codegree one has

7025: \begin{equation}\label{codegree}

7026: d^{\,'}(Z(h))=\,d(Z(\bar h^\sharp))=\,\int\,h(u)\,  d^* \, u =\,\wh

7027: h(0) ,

7028: \end{equation}

7029: so that the codegree $d^{\,'}(Z_g)$ of the correspondence $Z_g$ is

7030: equal to $1$.

7031: 

7032: 

7033: One of the major difficulties is to find the replacement for the

7034: principal divisors which in Weil's proof play a key role as an ideal

7035: in the algebra of correspondences on which the trace vanishes. At

7036: least already one can see that there is an interesting subspace $V$

7037: of the linear space of correspondences described above on which the

7038: trace also vanishes. It is given by the subspace

7039: \begin{equation}\label{V}

7040: V\subset \cS(C_\K) \,,\quad V=\{\,g(x)=\, \sum\,\xi(k\,x)\,|\,\xi\in

7041: \cS(\A_\K)_0\} ,

7042: \end{equation}

7043: where the subspace $\cS(\A_\K)_0\subset \cS(\A_\K)$ is defined by

7044: the two boundary conditions

7045: $$

7046: \xi(0)=0 \,,\quad \int \,\xi(x)\,dx\,=\,0 .

7047: $$

7048: 

7049: \medskip

7050: 

7051: 

7052: \begin{lem} \label{vanish}

7053: For any $f\in V\subset \cS (C_\K)$, one has

7054: $$

7055: \vartheta(f)|_{ \cH^1}=\,0 .

7056: $$

7057: \end{lem}

7058: 

7059: \medskip

7060: 

7061: This shows that the Weil pairing of Theorem \ref{pos} admits a huge

7062: radical given by all functions which extend to adeles and gives

7063: another justification for working with the above cohomology $H^1$.

7064:   In particular one can

7065: modify arbitrarily the degree and codegree of the correspondence

7066: $Z(h)$ by adding to $h$ an element of the radical $V$ using a subtle

7067: failure of Fubini's theorem. We will show in a forthcoming paper

7068: \cite{CoCM} that several of the steps of Weil's proof can be

7069: transposed in the framework described above.

7070: 

7071: This constitutes a clear motivation to develop noncommutative

7072: geometry much further. One can write a very tentative form of a

7073: dictionary from the language of algebraic geometry (in the case of

7074: curves) and that of noncommutative geometry. The dictionary is

7075: summarized in the following table. It should be stressed that the

7076: main problem is to find the correct translation in the right column

7077: (non-commutative geometry) of the well established notion of

7078: principal divisor in the (algebraic geometry) left column. The table

7079: below is too rough in that respect since one does not expect to be

7080: able to work in the usual ``primary" theory which involves periodic

7081: cyclic homology and index theorems. Instead one expects that both

7082: the unstable cyclic homology and the finer invariants of spectral

7083: triples arising from transgression will play an important role. Thus

7084: the table below should be taken as a very rough first approximation,

7085: and a motivation for developing the missing finer notions in the

7086: right column.

7087: 

7088: \bigskip

7089: \bigskip\bigskip\bigskip\bigskip\bigskip

7090: 

7091: 

7092: 

7093: 

7094: \begin{small}

7095: \begin{center}

7096: \begin{tabular}{|c|c|}

7097:  \hline & \\

7098: Virtual correspondences  & bivariant $K$-theory class $\Gamma$ \\ & \\

7099: \hline & \\

7100: Modulo torsion & $KK(A,B\otimes {\rm II}_1)$ \\ & \\

7101: \hline & \\

7102: Effective correspondences & Epimorphism of $C^*$-modules \\

7103: & \\ \hline & \\

7104: Principal correspondences & Compact morphisms  \\

7105: & \\ \hline &  \\

7106: Composition & cup product in $KK$-theory \\

7107: & \\ \hline & \\

7108: Degree of  correspondence & Pointwise index $d(\Gamma)$ \\

7109: & \\ \hline & \\

7110: $\deg D(P)\geq g \Rightarrow \sim$ effective & $d(\Gamma)>0

7111: \Rightarrow \exists K, \Gamma+K$ onto \\

7112: & \\ \hline & \\

7113: Adjusting the degree  & Fubini step  \\

7114: by trivial correspondences & on the test functions \\

7115: & \\ \hline & \\

7116: Frobenius correspondence &  Correspondence $Z_g$ \\

7117: & \\ \hline & \\

7118: Lefschetz formula & bivariant Chern  of $Z(h)$ \\

7119: &

7120: (localization on  graph $Z(h)$) \\  & \\ \hline &  \\

7121: Weil trace unchanged & bivariant Chern unchanged

7122: \\  by principal divisors & by compact

7123: perturbations

7124: \\

7125:  & \\ \hline

7126: \end{tabular}

7127: \end{center}

7128: \end{small}

7129: 

7130: 

7131: \newpage

7132: 

7133: 

7134: 

7135: 

7136: \begin{thebibliography}{99}

7137: 

7138: \bibitem{ANS} J.~Aastrup, R.~Nest, E.~Schrohe, 

7139: {\it A Continuous Field of C*-algebras and the

7140: Tangent Groupoid for Manifolds with Boundary}, ArXiv

7141: math.FA/0507317.

7142: 

7143: 

7144: \bibitem{aba-exel:1997}

7145: B.~Abadie, R.~Exel, {\em Hilbert $C^*$-bimodules

7146: over commutative $C^*$-algebras and an isomorphism condition

7147: for quantum Heisenberg manifolds}, Rev. Math. Phys.  Vol.9 (1997)

7148: no. 4, 411--423.

7149: 

7150: \bibitem{aba-eil-exel:1998}

7151:  B.~Abadie, S.~Eilers, R.~Exel,

7152: {\em Morita equivalence for crossed products by Hilbert $C^*$-bimodules}.

7153: Trans. Amer. Math. Soc. Vol.350, (1998) no. 8, 3043--3054.

7154: 

7155: \bibitem{AndPut} J.E.~Anderson, I.F.~Putnam, {\em  Topological

7156: invariants for substitution tilings and their associated $C\sp

7157: *$-algebras}. Ergodic Theory Dynam. Systems 18 (1998), no. 3, 509--537.

7158: 

7159: \bibitem{Anonym} Anonymous Persian manuscript, {\em Fi tadakhul

7160: al-ashkal al-mutashabiha aw mutawafiqa}, Biblioth\`eque Nationale, Paris,

7161: Ancien fonds persan 169.

7162: 

7163: \bibitem{AntChris} C.~Antonescu, E.~Christensen, {\em Spectral triples

7164: for AF $C^*$-algebras and metrics on the Cantor set}, preprint

7165: arXiv math.OA/0309044.

7166: 

7167: \bibitem{Ard1} Farhad Ardalan, Amir H. Fatollahi, {\em Point-Like

7168: Structure in Strings and Non-Commutative Geometry},

7169: Phys.Lett. B408 (1997) 157-163.

7170: 

7171: \bibitem{Ard2}  F. Ardalan, H. Arfaei, N. Sadooghi, {\em

7172: On the Anomalies and Schwinger Terms in Noncommutative Gauge

7173: Theories}, hep-th/0507230.

7174: 

7175: \bibitem{ArdBak} Nader Ardalan and Laleh Bakhtiar, {\em The sense of

7176: unity: the Sufi tradition in Persian architecture}, The University of

7177: Chicago Press, 1973.

7178: 

7179: \bibitem{art-tat-vdb:1990}

7180: M.~Artin, J.~Tate, and M.~Van~den Bergh,

7181: {\em Some algebras associated to automorphisms of elliptic curves}. The

7182:   Grothendieck Festschrift. vol.I.

7183: Prog. Math., Vol.86:33--85, 1990.

7184: 

7185: \bibitem{atbo} M.F.~Atiyah and R.~Bott, {\it A Lefschetz fixed

7186: point formula for elliptic complexes: I}, Annals of Math, Vol.86

7187: (1967) 374--407.

7188: 

7189: \bibitem{AT} M.F.~Atiyah, {\it Global theory of elliptic

7190: operators}. Proc. Internat.  Conf. on {\it Functional Analysis and

7191: Related Topics} (Tokyo, 1969). University of Tokyo  press, Tokyo

7192: 1970, 21-30.

7193: 

7194: \bibitem{BaMo} Michael Baake and Robert V. Moody (Editors), {\em

7195: Directions in mathematical quasicrystals}, CRM Monograph Series, 13.

7196: American Mathematical Society, Providence, RI, 2000. viii+379 pp.

7197: 

7198: \bibitem{BFSS} T. Banks, W. Fischler, S.H. Shenker, L. Susskind,

7199: {\em M theory as a Matrix model: a conjecture},

7200: Phys.Rev. D55 (1997) 5112-5128. hep-th/9610043.

7201: 

7202: \bibitem{BauCo} P.~Baum and A.~Connes,

7203: {\it Geometric K-theory for Lie groups and

7204: foliations}, Preprint IHES (M/82/), 1982, L'enseignement

7205: Mathematique, t.46 (2000), 1--35.

7206: 

7207: \bibitem{BCH} P. Baum, A. Connes, and N. Higson, {\it Classifying space

7208: for proper -actions and K-theory of group $C^*$-algebras}, Contemp.

7209: Math. Vol.167 (1994), 241--291.

7210: 

7211: \bibitem{BayFron} F.~Bayen and C.~Fronsdal.

7212: {\em Quantization on the sphere}.

7213:  J. Math. Phys. Vol.22 (1981) N.7, 1345--1349.

7214: 

7215: \bibitem{Bellissard1}

7216: J.~Bellissard,

7217: {\it Noncommutative geometry and quantum Hall effect}.

7218: In: Proceedings of the International Congress of Mathematicians,

7219:   Vol.\ 1, 2 (Z\"urich, 1994), pages 1238--1246, Basel, 1995. Birkh\"auser.

7220: 

7221: \bibitem{Bellissard2}

7222: J.~Bellissard, A.~van Elst, and H.~Schulz-Baldes.

7223: {\it The noncommutative geometry of the quantum Hall effect}.

7224: J. Math. Phys., 35 (1994) N.10, 5373--5451.

7225: 

7226: \bibitem{BellC} J.~Bellissard, {\em The noncommutative geometry of

7227: aperiodic solids}, in ``Geometric and topological methods for

7228: quantum field theory (Villa de Leyva, 2001)'',  86--156, World

7229: Scientific, 2003.

7230: 

7231: \bibitem{BF}  M-T.~Benameur, T.~Fack,

7232: {\em  On von Neumann spectral triples},

7233: math.KT/0012233.

7234: 

7235: \bibitem{BDV} R.~Berger, M.~Dubois-Violette, M.~Wambst, {\it Homogeneous

7236: algebras}, math.QA/0203035.

7237: 

7238: \bibitem{BoGia} C.G.~Bollini, J.J.~Giambiagi {\em Lowest order

7239: divergent graphs in $\nu$-dimensional space} Phys.Lett. B40 (1972)

7240: 566--568.

7241: 

7242: \bibitem{EB} E.~Bombieri, {\em Problems of the Millenium:

7243:  The Riemann Hypothesis}, Clay mathematical Institute (2000).

7244: 

7245: \bibitem{Born} M.~Born and P.~Jordan.

7246: {\it Zur Quantenmechanik},

7247: Z. Phys., 34 (1925) 858--888.

7248: 

7249: \bibitem{BKLR} D.~Borthwick, S.~Klimek, A.~Lesniewski, M.~Rinaldi, {\em

7250: Supersymmetry and Fredholm modules over quantized spaces}.

7251: Comm. Math. Phys. 166 (1994), no. 2, 397--415.

7252: 

7253: \bibitem{BC} J.-B. Bost and A. Connes, {\it Hecke Algebras, Type

7254: III factors and phase transitions with spontaneous symmetry

7255: breaking in number theory}, Selecta Mathematica, New

7256: Series Vol.1, N.3 (1995), 411--457.

7257: 

7258: \bibitem{BMa1} P.~Bouwknegt, K.~Hannabuss, V.~Mathai, {\em $T$-duality

7259: for principal torus bundles}.  J. High Energy Phys.  2004,  no. 3,

7260: 018, 10 pp. (electronic).

7261: 

7262: \bibitem{BMa2}  P.~Bouwknegt, J.~Evslin, V.~Mathai, {\em $T$-duality:

7263: topology change from $H$-flux}.  Comm. Math. Phys.  249  (2004),

7264: no. 2, 383--415.

7265: 

7266: \bibitem{BoHa} M.~Boyle, D.~Handelman, {\em Orbit equivalence, flow

7267: equivalence, and ordered cohomology}, Israel J. Math. 95 (1996)

7268: 169--210.

7269: 

7270: \bibitem{Bratteli} O.~Bratteli.

7271: {\it Inductive limits of finite dimensional $C^*$-algebras}.

7272: Trans. Amer. Math. Soc., 171 (1972) 195--234.

7273: 

7274: \bibitem{BM} P.~Breitenlohner, D.~Maison, {\em Dimensional

7275: renormalization and the action principle}.  Comm. Math. Phys.

7276: Vol.52 (1977), N.1, 11--38.

7277: 

7278: \bibitem{BruSee} J.~Br\"uning, R.~Seeley, {\it Regular singular

7279: asymptotics}, Adv. in Math. 58 (1985) 133--148.

7280: 

7281: \bibitem{BryNi} J-L. Brylinski and V. Nistor, {\it Cyclic Cohomology

7282: of Etale Groupoids}, K-theory, Vol.8 (1994), 341--365.

7283: 

7284: \bibitem{Bulatov} M.S.~Bulatov, {\em Geometricheskaya garmonizatsiya v

7285: arkhitekture Srednei Azii IX-XV vv.}, Nauka, 2nd ed., Moskva 1988,

7286: 360 pp.

7287: 

7288: \bibitem{Burgh}

7289: D. Burghelea, {\it  The cyclic homology of the group rings}, Comment.

7290: Math. Helv. Vol.60 (1985) 354--365.

7291: 

7292: \bibitem{CHMM} A.~Carey, K.~Hannabuss, V.~Mathai, P.~McCann,

7293:  {\em Quantum Hall Effect on the hyperbolic plane},

7294: Commun. Math. Physics, Vol.190, no. 3 (1998) 629--673.

7295: 

7296: \bibitem{CPRS} Alan L.~Carey, J.~Phillips, A.~Rennie, F. A.~Sukochev

7297: {\em  The Local Index Formula in Semifinite von Neumann Algebras I:

7298: Spectral Flow},

7299: math.OA/0411021.

7300: 

7301: \bibitem{CPRS1} Alan L.~Carey, J.~Phillips, A.~Rennie, F. A.~Sukochev

7302: {\em  The Local Index Formula in Semifinite von Neumann Algebras II:

7303: The Even Case},

7304: math.OA/0411021.

7305: 

7306: \bibitem{Castera} J.M.~Castera, {\em Zellijs, muqarnas and

7307: quasicrystals}, ISAMA 99 Proceedings, pp.99--104, 1999.

7308: 

7309: \bibitem{Castera2}  J.M.~Castera, {\em Arabesques}, ACR editions,

7310: 1999, 486 pp.

7311: 

7312: \bibitem{ChakraPal} P.S.~Chakraborti, A.~Pal, {\em Equivariant

7313: spectral triple on the quantum $SU(2)$-group},  $K$-Theory  28

7314: (2003),  no. 2, 107--126.

7315: 

7316: \bibitem{Chakra} T.~Chakraborti, P.~Pietil\"anen,

7317: {\em The Quantum Hall Effects}, Second Edition, Springer 1995.

7318: 

7319: \bibitem{C-C}  A.~Chamseddine and A.~Connes, {\it Universal

7320: formulas for  noncommutative geometry actions}, Phys. Rev.

7321: Letters, Vol.77, N.24, (1996)  4868--4871.

7322: 

7323: \bibitem{C-C2} A.~Chamseddine and A.~Connes, {\it The spectral

7324: action principle}, Commun. Math. Phys. 186 (1997) 731--750.

7325: 

7326: \bibitem{C-C3} A.~Chamseddine and A.~Connes, {\it Scale 

7327: Invariance in the  spectral

7328: action}, Arxiv hep-th/0512169.

7329: 

7330: \bibitem{ChGr} J. Cheeger, M. Gromov, {\it $L^2$-cohomology and group

7331: cohomology}, Topology, Vol.25 (1986), 189--215.

7332: 

7333: \bibitem{chou} A.~Chou, {\it The Dirac Operator on Spaces

7334: with Conical Singularities

7335: and Positive Scalar Curvatures}, Trans. Am. Math. Soc. 289 (1985),

7336: no. 1, 1--40.

7337: 

7338: \bibitem{IvChris} E.~Christensen, C.~Ivan, {\it Sums of two

7339: dimensional spectral triples}, ArXiv math.OA/0601024

7340: 

7341: \bibitem{Collins} J.~Collins, {\em Renormalization}, Cambridge

7342:  Monographs in Math. Physics, Cambridge University Press, 1984.

7343: 

7344: \bibitem{Co_2} A.~Connes, {\it Une classification des facteurs de type III},

7345: Ann. Sci. Ecole Norm. Sup., {\bf 6}, n.~4 (1973), 133-252.

7346: 

7347: \bibitem{Co78} A.~Connes, {\it Sur la th\'eorie non commutative de

7348: l'int\'egration} Alg\`ebres d'op\'erateurs (S\'em., Les Plans-sur-Bex,

7349: 1978), p. 19-143, Lecture Notes in Math., 725, Springer, Berlin,

7350: 1979.

7351: 

7352: \bibitem{Co90} A.~Connes, {\it G\'eom\'etrie Noncommutative}.

7353: InterEditions, Paris, 1990.

7354: 

7355: \bibitem{Co94} A.~Connes, {\it Noncommutative geometry}.

7356: Academic Press, 1994.

7357: 

7358: \bibitem{Co-Thom} A.~Connes, {\em  An analogue of the Thom isomorphism

7359: for crossed products of a $C^*$-algebra by an action of $\R$},

7360: Adv. in Math. 39 (1981), no. 1, 31--55.

7361: 

7362: \bibitem{Co1} A.~Connes, {\it Cyclic cohomology and the

7363: transverse fundamental class of a foliation}. In: Geometric

7364: methods in operator algebras (Kyoto, 1983). Pitman Res. Notes in

7365: Math., 123, Longman, Harlow 1986, 52--144.

7366: 

7367: \bibitem{C3} A.~Connes, {\it  $C^*$ alg\`ebres et

7368: g\'eom\'etrie differentielle}.  C.R. Acad. Sci. Paris,

7369: Ser.~A-B , 290 (1980) 599--604.

7370: 

7371: \bibitem{ConnesCH} A.~Connes, {\em Noncommutative

7372: differential geometry}. Inst.  Hautes Etudes Sci. Publ.

7373: Math., 62, 1985, 257--360.

7374: 

7375: \bibitem{Co-fr}  A.~Connes, {\it Compact metric spaces, Fredholm

7376: modules, and hyperfiniteness}, Ergod. Th. Dynam. Sys. (1989) 9,

7377: 207--220.

7378: 

7379: \bibitem{Cotheta}  A.~Connes, {\it On the Chern character of $\theta$-summable

7380: Fredholm modules} Comm. Math. Phys. 139 (1991), n° 1, p. 171-181.

7381: 

7382: \bibitem{Co3} A.~Connes, {\it Gravity coupled with matter and

7383: foundation of  noncommutative geometry}. Commun. Math. Phys.,

7384: 182 (1996) 155--176.

7385: 

7386: \bibitem{Coo2} A.~Connes, {\it Noncommutative geometry and

7387: reality}. Journal of Math. Physics, 36, n.11 (1995) 6194--6231.

7388: 

7389: \bibitem{ConnesS3} A.~Connes, {\em Geometry from the spectral point of

7390: view}. Lett. Math. Phys. 34 (1995), no. 3, 203--238.

7391: 

7392: \bibitem{CoRH} A.~Connes,  {\it Trace formula in Noncommutative

7393: Geometry and the zeros of the Riemann zeta function}. Selecta

7394: Mathematica. New Ser. 5 (1999) 29--106.

7395: 

7396: \bibitem{Co-Qgr} A.~Connes, {\em Cyclic cohomology, quantum group

7397: symmetries and the local index formula for ${\rm SU}\sb q(2)$}.

7398: J. Inst. Math. Jussieu  3  (2004),  no. 1, 17--68.

7399: 

7400: \bibitem{CCM} A.~Connes, C.~Consani, M.~Marcolli,

7401: {\em Noncommutative geometry and motives:

7402:  the thermodynamics of endomotives}, math.QA/0512138.

7403: 

7404: \bibitem{CoCM} A.~Connes, C.~Consani, M.~Marcolli, {\em The Weil proof

7405: and the geometry of the adeles class space}, in preparation.

7406: 

7407: \bibitem{CDS} A.~Connes, M.~Douglas, A.~Schwarz,

7408: {\it Noncommutative geometry and  Matrix theory: compactification

7409: on tori}.

7410: J. High Energy Phys. (1998) no. 2, Paper 3, 35 pp. (electronic)

7411: 

7412: \bibitem{CDV1} A.~Connes, M.~Dubois-Violette, {\em Noncommutative

7413: finite-dimensional manifolds. I. Spherical manifolds and related

7414: examples}.  Comm. Math. Phys.  230  (2002),  no. 3, 539--579.

7415: 

7416: \bibitem{CDV2} A.~Connes, M.~Dubois-Violette, {\em Moduli space and

7417: structure of noncommutative 3-spheres}.  Lett. Math. Phys.  66

7418: (2003),  no. 1-2, 91--121.

7419: 

7420: \bibitem{CDV3} A.~Connes, M.~Dubois-Violette, {\em Non commutative

7421: finite-dimensional manifolds II. Moduli space and structure of

7422: noncommutative 3-spheres}.  ArXiv math.QA/0511337.

7423: 

7424: \bibitem{CLa} A.~Connes and G.~Landi, {\it Noncommutative manifolds,

7425: the instanton algebra and isospectral

7426: deformations}. Comm. Math. Phys. 221 (2001), no. 1, 141--159.

7427: 

7428: \bibitem{CL} A.~Connes, J.~Lott, {\em Particle models and noncommutative

7429: geometry}, in ``Recent advances in field theory''

7430: (Annecy-le-Vieux, 1990).  Nuclear Phys. B Proc. Suppl.  18B

7431: (1990), 29--47 (1991).

7432: 

7433: \bibitem{CoMa} A.~Connes, M.~Marcolli, {\em From Physics to Number

7434: Theory via Noncommutative Geometry.

7435: Part I: Quantum Statistical Mechanics of $Q$-lattices},

7436: to appear in the volume

7437: ``Frontiers in Number Theory Physics and Geometry''

7438: math.NT/0404128.

7439: 

7440: \bibitem{CoMajgp1} A.~Connes, M.~Marcolli, {\em $\Q$-lattices: quantum

7441: statistical mechanics and

7442: Galois theory}, Journal of Geometry and Physics, Vol. 56 (2005) N.1,

7443: 2--23.

7444: 

7445: \bibitem{CoMar-an} A.~Connes, M.~Marcolli, {\em Anomalies, Dimensional

7446: Regularization and Noncommutative Geometry}, in preparation.

7447: 

7448: \bibitem{CoMarBook} A.~Connes, M.~Marcolli, {\em Noncommutative geometry,

7449: from quantum fields to motives} (tentative title), book in preparation.

7450: To appear as a co-publication of

7451: the American Mathematical Society, Colloquium Publications Series, and

7452: Hindustan Book Agency, Texts and Readings in Mathematics Series.

7453: 

7454: \bibitem{CMR} A.~Connes, M.~Marcolli, N.~Ramachandran, {\em KMS states

7455: and complex multiplication}, to appear in Selecta Mathematica.

7456: 

7457: \bibitem{CMR2} A.~Connes, M.~Marcolli, N.~Ramachandran, {\em KMS states

7458: and complex multiplication. Part II}, to appear in the proceedings of

7459: the Abel Symposium 2005.

7460: 

7461: \bibitem{CoMoind} A.~Connes, H.~Moscovici,

7462:  {\em Cyclic cohomology, the Novikov conjecture and hyperbolic

7463:  groups}, Topology, Vol. 29 (1990) no. 3, 345--388.

7464: 

7465: \bibitem{CoMo}  A.~Connes and H.~Moscovici, {\it The local

7466: index formula in noncommutative geometry}. GAFA,  5 (1995)

7467: 174--243.

7468: 

7469: \bibitem{CoMoHopf} A.~Connes and H.~Moscovici, {\it Hopf Algebras,

7470: Cyclic Cohomology and the Transverse Index Theorem}.

7471: Commun. Math.  Phys. 198 (1998) 199--246.

7472: 

7473: \bibitem{CoMoHopf2} A.~Connes and H.~Moscovici, {\it  Differentiable

7474: cyclic cohomology and Hopf algebraic structures in transverse

7475: geometry}. Essays on geometry and related topics, Vol. 1, 2, 217--255,

7476: Monogr. Enseign. Math., 38, Enseignement Math., Geneva, 2001.

7477: 

7478: \bibitem{Co-Mosc1} A.~Connes and H.~Moscovici, {\em Modular Hecke

7479: algebras and their Hopf symmetry},  Mosc. Math. J.  4  (2004),  no. 1,

7480: 67--109, 310 (math.QA/0301089).

7481: 

7482: \bibitem{Co-Mosc2} A.~Connes and H.~Moscovici, {\em Rankin-Cohen

7483: Brackets and the Hopf Algebra of Transverse Geometry},  Mosc. Math. J.

7484: 4  (2004),  no. 1, 111--130, 311.

7485: (math.QA/0304316).

7486: 

7487: \bibitem{CoMo-back} A.~Connes and H.~Moscovici,

7488: {\em Background independent geometry and Hopf cyclic cohomology},

7489:  math.QA/0505475.

7490: 

7491: \bibitem{CoMoBook}  A.~Connes and H.~Moscovici, {\it Spectral Geometry

7492: of Noncommutative Spaces} (tentative title), book in preparation.

7493: 

7494: \bibitem{CoRi}  A.~Connes and M.~Rieffel, {\it Yang-Mills for

7495: noncommutative two tori}. Contemp. Math. Oper. Algebra Math. Phys.

7496: AMS (1987) 237--266.

7497: 

7498: \bibitem{CoSk}  A.~Connes and G.~Skandalis,

7499: {\it The longitudinal index theorem for

7500: foliations} Publ. Res. Inst. Math. Sci. 20 (1984), n° 6, p.

7501: 1139-1183.

7502: 

7503: \bibitem{CT}  A.~Connes and  M.~Takesaki,

7504: {\it The flow of weights on factors of type III}, Tohoku Math. J.

7505: {\bf 29} (1977), 473-575.

7506: 

7507: \bibitem{CM1} C.~Consani, M.~Marcolli,  {\em Noncommutative

7508: geometry, dynamics, and $\infty$-adic Arakelov geometry}.  Selecta

7509: Math. (N.S.)  10  (2004),  no. 2, 167--251.

7510: 

7511: \bibitem{CM2} C.~Consani, M.~Marcolli,  {\em New perspectives in

7512: Arakelov geometry}.  Number theory,  81--102, CRM Proc. Lecture Notes,

7513: 36, Amer. Math. Soc., Providence, RI, 2004.

7514: 

7515: \bibitem{Crai} M. Crainic,  {\it Cyclic Cohomology

7516: of Etale Groupoids, the general case}, K-theory, Vol.17

7517: (1999), 319--362.

7518: 

7519: \bibitem{Cu} J.~Cuntz, {\it A class of $C^*$--algebras and topological

7520: Markov chains II: reducible chains and the $Ext$--functor for

7521: $C^*$--algebras}, Invent. Math. 63 (1981) 25--40.

7522: 

7523: \bibitem{CuKrie} J.~Cuntz, W.~Krieger, {\it A class of $C^*$--algebras

7524: and topological Markov chains}, Invent. Math. 56 (1980) 251--268.

7525: 

7526: \bibitem{Davidson} K.R.~Davidson, {\it $C^*$-algebras by example}.

7527: American Mathematical Society, Providence, RI, 1996.

7528: 

7529: \bibitem{Dix} J.~Dixmier, {\it Existence de traces non

7530: normales}. C.R. Acad. Sci. Paris, Ser. A-B, 262 (1966) A1107--A1108.

7531: 

7532: \bibitem{D-V} M.~Dubois-Violette, {\it  Generalized differential 

7533: spaces with $d^N=0$ and

7534: the $q$-differential calculus},

7535:  Czech J. Phys. 46 (1997) 1227--1233.

7536: 

7537: \bibitem{D-V1} M.~Dubois-Violette, {\it

7538:  $d^N=0$ : Generalized homology},

7539: K-Theory, 14  (1998) 371--404.

7540: 

7541: \bibitem{D-V2}

7542: M.~Dubois-Violette, {\it Lectures on graded differential algebras

7543: and noncommutative geometry}, In Y.~Maeda and H.~Moriyoshi, editors,

7544: {\em Noncommutative

7545:   Differential Geometry and Its Applications to Physics}, pages 245--306.

7546:   Shonan, Japan, 1999, Kluwer Academic Publishers, 2001.

7547: 

7548: \bibitem{D-V3} M.~Dubois-Violette and M.~Henneaux, {\it

7549: Generalized cohomology for irreducible tensor fields of mixed Young

7550: symmetry type}

7551:  Lett. Math. Phys.  49 (1999) 245--252.

7552: 

7553: \bibitem{D-V4} M.~Dubois-Violette and M.~Henneaux, {\it

7554:  Tensor fields of mixed Young symmetry type and $N$-complexes},

7555:  Commun. Math. Phys.  226 (2002) 393--418.

7556: 

7557: \bibitem{FGV} H.~Figueroa, J.M.~Gracia-Bond\'ia, J.~Varilly, {\em

7558: Elements of Noncommutative Geometry}, Birkh\"auser, 2000.

7559: 

7560: \bibitem{Gabo1} D. Gaboriau, {\it Invariants $\ell^2$ de relations

7561: d'\'equivalence et de groupes},   Publ. Math. Inst. Hautes \'Etudes

7562: Sci.  No. 95 (2002), 93--150.

7563: 

7564: \bibitem{Gabo2} D. Gaboriau, {\it Approximate dimension for

7565: equivalence relations and groups}, in preparation.

7566: 

7567: \bibitem{GGISV} V.~Gayral, J.M.~Gracia-Bond\'ia, B.~Iochum,

7568: T.~Sch\"ucker,  J.C.~V\'arilly, {\em Moyal planes are spectral

7569: triples}.  Comm. Math. Phys.  246  (2004),  no. 3, 569--623.

7570: 

7571: \bibitem{GPS} T.~Giordano, I. F.~Putnam and C. Fr.~Skau, 

7572: {\it Topological orbit equivalence and

7573:  $C^*$-crossed products}. J. Reine Angew. Math. 469

7574: (1995), 51--111

7575: 

7576: \bibitem{GuIso1} D.~Guido, T.~Isola, {\em Dimensions and singular

7577: traces for spectral triples, with applications to fractals},

7578: Journ. Funct. Analysis, 203 (2003), 362-400. (math.OA/0202108)

7579: 

7580: \bibitem{GuIso2} D.~Guido, T.~Isola, {\em Dimensions and spectral

7581: triples for fractals in $R^N$}, math.OA/0404295.

7582: 

7583: \bibitem{gui} V.~Guillemin, {\it Lectures on spectral theory of

7584: elliptic operators}, Duke Math. J., Vol.44  (1977) N.3, 485--517.

7585: 

7586: \bibitem{GuVai} D.~Gurevich and L.~Vainerman, {\it Non-commutative

7587: analogues of $q$-special polynomials and a $q$-integral on a quantum sphere}.

7588: J. Phys. A, 31 (1998) 7, 1771--1780.

7589: 

7590: \bibitem{HaPa} E.~Ha, F.~Paugam, {\em Bost-Connes-Marcolli systems for

7591: Shimura varieties. I. Definitions and formal analytic properties},

7592: IMRP 2005, Vol.5 (2005) 237--286. (math.OA/0507101)

7593: 

7594: \bibitem{Hajac} P.M.~Hajac, M.~Khalkhali, B.~Rangipour,

7595: Y.~Sommerhaeuser, {\em Hopf-cyclic homology and cohomology

7596: with coefficients}. math.KT/0306288.

7597: 

7598: \bibitem{Hall}

7599: E.H.~Hall, {\it On a new action of the magnet on electric currents},

7600:  Amer. J. of Math. 2 (1879) 287.

7601: 

7602: \bibitem{EHaw} E.~Hawkins, {\em Hamiltonian gravity and noncommutative

7603: geometry}. Comm. Math. Phys.  187  (1997),  no. 2, 471--489.

7604: 

7605: \bibitem{Heisenberg} W.~Heisenberg, {\it \"Uber quantentheoretische

7606: Umdeutung kinematischer und

7607:   mechanischer Beziehungen},  Z. Phys. 33 (1925) 879--893.

7608: 

7609: \bibitem{HKR} G.~Hochschild, B.~Kostant, A.~Rosenberg, {\em Differential

7610: forms on regular affine algebras}.  Trans. Amer. Math. Soc.

7611: Vol.102  (1962) 383--408.

7612: 

7613: \bibitem{tHV} G.~'t Hooft, M.~Veltman, {\em Regularization and

7614: renormalization of gauge fields} Nuclear Physics B, Vol.44, N.1

7615: (1972), 189--213.

7616: 

7617: \bibitem{Hu} S.~Hurder, {\em Secondary classes and the von Neumann algebra

7618: of a foliation}, {\it Preprint MSRI Berkeley} (1983).

7619: 

7620: \bibitem{JLO} A.~Jaffe, A.~Lesniewski, K.~Osterwalder,

7621: {\em Quantum K-theory : the Chern character},

7622:  Commun. Math. Phys., 118 (1988)

7623: 1--14.

7624: 

7625: \bibitem{Julg} P.~Julg, {\em Travaux de N. Higson et

7626: G. Kasparov sur la conjecture de

7627: Baum-Connes}. Seminaire Bourbaki, (1997-98), Expose 841,

7628: Vol.252 (1998), 151--183.

7629: 

7630: \bibitem{K-W} W.~Kalau and M.~Walze, {\it Gravity,

7631: noncommutative geometry and the  Wod\-zi\-cki residue}.

7632:  J. of Geom. and Phys., 16 (1995) 327--344.

7633: 

7634: \bibitem{Kasp} G.G.~Kasparov, {\it The operator $K$-functor

7635: and extensions of $C^*$  algebras}. Izv. Akad. Nauk SSSR, Ser. Mat.,

7636: 44 (1980) 571--636; Math. USSR Izv., 16 (1981) 513--572.

7637: 

7638: \bibitem{Kast} D.~Kastler, {\it The Dirac operator and

7639: gravitation}. Commun. Math. Phys., 166 (1995) 633--643.

7640: 

7641: \bibitem{Kast2} D.~Kastler, {\em A detailed account of Alain Connes'

7642: version of the standard model in noncommutative geometry. I, II},

7643: Rev. Math. Phys.  5  (1993),  no. 3, 477--532. {\em III},

7644:  Rev. Math. Phys.  8  (1996),  no. 1, 103--165.

7645: 

7646: \bibitem{khal} M.~Khalkhali,{\em  Very Basic Noncommutative

7647: Geometry}, arXiv:math.KT/0408416. Lecture Notes Series IPM Tehran,

7648: Vol.5, 2005.

7649: 

7650: \bibitem{kr1} M.~Khalkhali and B.~Rangipour,

7651: {\em  A new cyclic module for Hopf

7652: algebras}, $K$-Theory  27  (2002),  no. 2, 111--131.

7653: 

7654: \bibitem{kr2} M.~Khalkhali and B.~Rangipour, {\em  Invariant cyclic homology},

7655:  $K$-Theory  28  (2003),  no. 2, 183--205.

7656: 

7657: \bibitem{kr3} M.~Khalkhali and B.~Rangipour,

7658: {\em  Cyclic cohomology of (extended) Hopf

7659: algebras},

7660: Noncommutative geometry and quantum groups,

7661: (Warsaw, 2001),  59--89, Banach Center Publ., 61,

7662: Polish Acad. Sci., Warsaw, 2003.

7663: 

7664: \bibitem{KlimSchmu} A.~Klimyk and K.~Schmuedgen, {\em Quantum groups

7665: and their representations}, Springer Verlag, 1998.

7666: 

7667: \bibitem{vonKlitz}

7668: K.~von Klitzing, G.~Dorda, M.~Pepper, {\em

7669: New method for high--accuracy determination of the fine--structure

7670:   constant based on quantized hall resistance}.

7671: Phys. Rev. Lett. 45 (1980) N.6, 494--497.

7672: 

7673: \bibitem{KontIHES97} M.~Kontsevich, {\it Deformation quantization of

7674: Poisson manifolds I}, IHES preprint M/97/72.

7675: 

7676: \bibitem{Kontsevich93} M.~Kontsevich, {\it Formal (non)commutative

7677: symplectic geometry}.

7678: In: The Gel'fand Mathematical Seminars, 1990--1992, pages

7679:   173--187. Birkh\"auser Boston, Boston, MA, 1993.

7680: 

7681: \bibitem{Kontsevich00} M.~Kontsevich and A.L.~Rosenberg, {\it

7682: Noncommutative smooth spaces}.

7683: In: The Gelfand Mathematical Seminars, 1996--1999, pages

7684:   85--108. Birkh\"auser Boston, Boston, MA, 2000.

7685: 

7686: \bibitem{Laff}  V.~Lafforgue,

7687: {\it K-th\'eorie bivariante pour les alg\`ebres de Banach et conjecture

7688: de Baum-Connes},  Inv. Math. Vol.149 (2002), 1--97.

7689: 

7690: \bibitem{landi} G.~Landi, {\em An Introduction to

7691: Noncommutative Spaces and their

7692: Geometry}. Lecture Notes in Physics: Monographs m51,

7693: Springer-Verlag, 1997. ISBN 3-540-63509-2.

7694: 

7695: \bibitem{LapFra} M.L.~Lapidus, M.~van Frankenhuysen, {\it Fractal

7696:   geometry and number theory. Complex dimensions of fractal strings and

7697:   zeros of zeta functions}. Birkh\"auser Boston, Inc., Boston, MA, 2000.

7698: 

7699: \bibitem{Lau} B.~Laughlin, {\it Quantized Hall conductivity in two

7700: dimensions}. Phys. Rev. B, 23 (1981) 5632.

7701: 

7702: \bibitem{laurent} H.~Laurent, {\it  Sur la th\'eorie des nombres

7703: premiers}.  C.R. Acad. Sci. Paris,  126 (1898) 809--810.

7704: 

7705: \bibitem{Lesch} M.~Lesch, {\it Operators of Fuchs type, conical

7706: singularities, and asymptotic methods}, B. G. Teubner

7707: Verlagsgesellschaft mbH, Stuttgart, 1997.

7708: 

7709: \bibitem{Lescure} J-M.~Lescure, {\em Triplets spectraux pour les

7710: vari\'et\'es \'a singularit\'e conique

7711: isol\'ee}. Bull. Soc. Math. France 129 (2001), no. 4, 593--623.

7712: 

7713: \bibitem{ManDim} Yu.I.~Manin, {\it The notion of dimension in geometry

7714: and algebra}, to appear in Bulletin of the AMS (math.AG/0502016).

7715: 

7716: \bibitem{Manin01RM} Yu.I.~Manin, {\it Real multiplication and

7717: noncommutative geometry},  The legacy of Niels Henrik Abel,  685--727,

7718: Springer, Berlin, 2004. (math.AG/0202109)

7719: 

7720: \bibitem{Manin91NC} Yu.I.~Manin, {\em Topics in noncommutative geometry}.

7721: Princeton University Press, Princeton, NJ, 1991.

7722: 

7723: \bibitem{Man-hyp} Yu.I.~Manin, {\em Three--dimensional hyperbolic geometry

7724: as $\infty$--adic Arakelov geometry}, Invent. Math. 104 (1991)

7725: 223--244.

7726: 

7727: \bibitem{[M]} Yu.I.~Manin, {\it Quantum groups and

7728: noncommutative geometry}. Centre Recherche Math. Univ.

7729: Montr\'eal 1988.

7730: 

7731: \bibitem{ManMar} Yu.I.~Manin, M.~Marcolli, {\em Continued fractions,

7732: modular symbols, and noncommutative geometry}.  Selecta Math. (N.S.)

7733: 8  (2002),  no. 3, 475--521.

7734: 

7735: \bibitem{MaPa} Yu.I.~Manin,  A.A.~Panchishkin, {\em Introduction to

7736: modern number theory. Fundamental problems, ideas and

7737: theories}. Second edition. Encyclopaedia of Mathematical Sciences,

7738: 49. Springer-Verlag, 2005. xvi+514 pp.

7739: 

7740: \bibitem{Mar} M.~Marcolli, {\em Arithmetic noncommutative geometry},

7741: University Lecture Series, 36. American Mathematical Society, 2005.

7742: xii+136 pp.

7743: 

7744: \bibitem{Mar2} M.~Marcolli, {\em Limiting modular symbols and the

7745: Lyapunov spectrum}.  J. Number Theory  98  (2003),  no. 2, 348--376.

7746: 

7747: \bibitem{Mar3} M.~Marcolli, {\em Modular curves, $C^*$-algebras, and

7748: chaotic cosmology}, math-ph/0312035.

7749: 

7750: \bibitem{MM1}  M.~Marcolli and V.~Mathai, {\em Twisted index theory on

7751: good orbifolds, II: fractional quantum numbers}, Communications in

7752: Mathematical Physics, Vol.217, no.1 (2001) 55--87.

7753: 

7754: \bibitem{MM2} M.~Marcolli and V.~Mathai, {\em Twisted index theory on good

7755: orbifolds, I: noncommutative Bloch theory}, Communications in

7756: Contemporary Mathematics, Vol.1 (1999)  553--587.

7757: 

7758: \bibitem{MM3} M.~Marcolli and V.~Mathai, {\em Towards the fractional

7759: quantum Hall effect: a noncommutative geometry perspective},

7760: cond-mat/0502356.

7761: 

7762: \bibitem{RoMa} V.~Mathai, J.~Rosenberg, {\em $T$-duality for

7763: torus bundles with $H$-fluxes via noncommutative topology}.

7764: Comm. Math. Phys.  253  (2005),  no. 3, 705--721

7765: 

7766: \bibitem{mather} J.~Mather, {\it Commutators of diffeomorphisms.

7767: II}, Comment. Math. Helv. 50 (1975), 33--40.

7768: 

7769: \bibitem{Meyer} R.~Meyer, {\em On a representation of the idele class

7770: group related to primes and zeros of $L$-functions}, preprint arXiv

7771: math.NT/0311468.

7772: 

7773: \bibitem{Mil2} J.S.~Milne, {\em Shimura varieties: the geometric

7774: side of the zeta function}, preprint available at

7775: www.jmilne.org

7776: 

7777: \bibitem{Mi-S} J.~Milnor and D.~Stasheff, {\it Characteristic

7778: classes}. Ann. of Math. Stud.,  Princeton

7779: University Press, Princeton, N.J. 1974.

7780: 

7781: \bibitem{Month} B.~Monthubert, {\it

7782: Pseudodifferential calculus on manifolds with corners and groupoids}

7783: Proc. Amer. Math. Soc. 127 (1999) no 10, 2871--2881. ArXiv

7784: math.KT/0507601.

7785: 

7786: \bibitem{Month1} B.~Monthubert and V.~Nistor, {\it

7787: A topological index theorem for manifolds with corners}

7788: 

7789: \bibitem{Nagy}  D.~Nagy, {\em Visual mathematics: a missing link in a

7790: split culture}. Vis. Math.  1  (1999),  no. 2, 20 pp. (electronic).

7791: 

7792: \bibitem{NS} N.~Nekrasov and A.~Schwarz, {\it Instantons in

7793: noncommutative $\R^4$ and  (2,0) superconformal six dimensional

7794: theory}.  Comm. Math. Phys. 198 (1998), no. 3, 689--703.

7795: 

7796: \bibitem{PaTu} W.~Parry, S.~Tuncel, {\it Classification problems in

7797: ergodic theory}, London Math. Soc. Lecture Notes Series 67, 1982.

7798: 

7799: \bibitem{Penrose} R.~Penrose, {\em Pentaplexity: a class of

7800: nonperiodic tilings of the plane}.  Math. Intelligencer  2  (1979/80),

7801: no. 1, 32--37.

7802: 

7803: \bibitem{pims:1997}

7804: M.~V.~Pimsner, {\em A class of C*-algebra

7805: generalizing both Cuntz-Krieger algebras and crossed products by

7806: $Z$}. Free probablility Theory, 189-212. Fields Inst. Comm. Vol.12,

7807: American Math. Society,  1997.

7808: 

7809: \bibitem{PV} M.~Pimsner and D.~Voiculescu, {\it Exact sequences

7810: for $K$ groups and  Ext group of certain crossed product

7811: $C^*$-algebras}. J. Operator Theory, 4 (1980) 93--118.

7812: 

7813: \bibitem{Poli} A.~Polishchuk, {\em Noncommutative two-tori with real

7814: multiplication as noncommutative projective varieties}.

7815: J. Geom. Phys.  50  (2004),  no. 1-4, 162--187.

7816: 

7817: \bibitem{Pusch} M.~Puschnigg, {\em Characters of Fredholm modules

7818: and a problem of Connes}, Preprint 2005.

7819: 

7820: \bibitem{Putn} I.~Putnam, {\it $C^*$--algebras from Smale spaces},

7821: Can. J. Math. 48 (1996) N.1 175--195.

7822: 

7823: \bibitem{PutSpi} I.~Putnam, J.~Spielberg, {\it The structure of ${\rm

7824: C}^*$--algebras associated with hyperbolic dynamical systems},

7825: J. Funct. Anal. 163 (1999) 279--299.

7826: 

7827: \bibitem{Ramis} J-P.~Ramis, {\it S\'eries divergentes et th\'eories

7828: asymptotiques}. Bull. Soc. Math. France, 121(Panoramas et Syntheses,

7829:   suppl.) 74, 1993.

7830: 

7831: \bibitem{Ren} J.~Renault, {\em A groupoid approach to $C^*$-algebras}.

7832: Lecture Notes in Math. 793, Springer, Berlin, 1980.

7833: 

7834: \bibitem{Ri2} M.A.~Rieffel, {\it $C^*$-algebras associated

7835: with irrational rotations}. Pacific J. Math., 93 (1981)

7836: 415--429.

7837: 

7838: \bibitem{Rieffel} M.A.~Rieffel, {\it The cancellation theorem

7839: for projective modules  over irrational rotation $C^*$-algebras}.

7840:  Proc. London Math. Soc., 47 (1983) 285--302.

7841: 

7842: \bibitem{Rieffel89} M.A.~Rieffel, {\it Deformation quantization of

7843: Heisenberg manifolds}. Comm. Math. Phys. 122 (1989) 4, 531--562.

7844: 

7845: \bibitem{Rob} G.~Robertson, {\it Boundary actions for affine buildings

7846: and higher rank Cuntz--Krieger algebras}, in J.~Cuntz, S.~Echterhoff

7847: (eds.) ``${\rm C}^*$--algebras'', Springer Verlag 2001, pp.~ 182--202.

7848: 

7849: \bibitem{Rosenberg98} A.L.~Rosenberg, {\em Noncommutative schemes}.

7850: Compositio Math. 112 (1998) N.1, 93--125.

7851: 

7852: \bibitem{Rosenberg95} A.L.~Rosenberg.

7853: {\em Noncommutative algebraic geometry and representations of

7854:   quantized algebras}.

7855: Kluwer Academic Publishers Group, Dordrecht, 1995.

7856: 

7857: \bibitem{Rue2} D.~Ruelle, {\it Non--commutative algebras for

7858: hyperbolic diffeomorphisms}, Invent. Math. 93 (1988) 1--13.

7859: 

7860: \bibitem{Schu} T.~Schucker, {\it Spin group and almost

7861: commutative geometry}. preprint hep-th/0007047.

7862: 

7863: \bibitem{Witten} N.~Seiberg and E.~Witten, {\it String theory

7864: and noncommutative geometry}. J. High Energy Phys. 1999, no. 9, Paper

7865: 32, 93 pp.

7866: 

7867: \bibitem{Shahin}  M.M.~Sheikh-Jabbari,

7868: {\em Tiny Graviton Matrix Theory: DLCQ of IIB Plane-Wave

7869: String Theory, A Conjecture}, JHEP 0409 (2004) 017.

7870: hep-th/0406214.

7871: 

7872: \bibitem{Shi} G.~ Shimura, {\em Arithmetic theory of automorphic

7873:     functions}, Iwanami Shoten and Princeton 1971.

7874: 

7875: 

7876: \bibitem{Skan1} G.~Skandalis, {\em Approche de la conjecture de Novikov

7877: par la

7878: cohomologie cyclique}. Seminaire Bourbaki, (1990-91), Expose 739,

7879: Ast\'erisque No.201-203 (1992), 299--320.

7880: 

7881: \bibitem{Skan2} G.~Skandalis, {\em Progr\`es r\'ecents

7882: sur la conjecture de Baum-Connes.

7883: Contribution de Vincent Lafforgue}, S\'eminaire Bourbaki, Vol.1999-2000,

7884: Expose 869,  Ast\'erisque  No. 276 (2002), 105--135.

7885: 

7886: \bibitem{skau} C.~ Skau, 

7887: {\it Orbit structure of topological dynamical systems and

7888: its invariants}, Operator algebras and quantum field theory (Rome,

7889: 1996), 533--544, Internat. Press, Cambridge, MA, 1997.

7890: 

7891: \bibitem{smi-sta:1992}

7892: S.P.~Smith and J.T. Stafford.

7893: {\em Regularity of the four dimensional Sklyanin algebra}.

7894: Compos. Math., Vol.83 (1992) 259--289.

7895: 

7896: \bibitem{Soibelman01} Y.~Soibelman, {\it Quantum tori, mirror symmetry

7897: and deformation theory}. Lett. Math. Phys. 56 (2001) 99--125.

7898: 

7899: \bibitem{Spi} J.~Spielberg, {\it Free--product groups, Cuntz--Krieger

7900: algebras, and covariant maps}. Internat. J. Math. 2 (1991), no. 4,

7901: 457--476.

7902: 

7903: \bibitem{SDLSV} W.~van Suijlekom, L.~Dabrowski, G.~Landi, A.~Sitarz,

7904: J.C.~Varilly, {\em Local index formula for

7905: $SU_q(2)$}, math.QA/0501287

7906: 

7907: \bibitem{SDLSV2} W.~van Suijlekom, L.~Dabrowski, G.~Landi, A.~Sitarz,

7908: J.C.~Varilly, {\em The Dirac operator on

7909: $SU_q(2)$}, math.QA/0411609

7910: 

7911: \bibitem{Sull} D.~Sullivan, {\it Geometric periodicity and

7912: the invariants of manifolds}. Lecture Notes in Math., 197,

7913: Springer, 1971.

7914: 

7915: \bibitem{Sun} T.~Sunada,

7916:  {\em A discrete analogue of periodic magnetic Schr\"odinger operators},

7917: Contemp. Math. Vol.173 (1994) 283--299.

7918: 

7919: \bibitem{Tak} M.~Takesaki, {\it Tomita's theory of modular

7920: Hilbert algebras and its applications}. Lecture Notes in Math., 28,

7921: Springer, 1970.

7922: 

7923: \bibitem{Velt2} M.~Veltman, {\em Diagrammatica}, Cambridge

7924: University Press, 1994.

7925: 

7926: \bibitem{vdWaerden} B.L.~van der Waerden,

7927: {\em Sources of Quantum Mechanics}. Dover, New York, 1967.

7928: 

7929: \bibitem{Wafa} Abu'l-Wafa' al-Buzjani, {\em Kitab fima yahtaju ilayhi

7930: al-sani` min a`mal al-handasa}. Milan manuscript (Arabic): Biblioteca

7931: Ambrosiana, Arab. 68.

7932: 

7933: \bibitem{Wassermann88} A.~Wassermann, {\em Ergodic actions of compact

7934: groups on operator

7935: algebras. III. Classification for ${\rm SU}(2)$}.

7936: Invent. Math. 93 (1988) 2, 309--354.

7937: 

7938: \bibitem{Winkel} H.E.~Winkelnkemper,

7939: {\it The  graph of a foliation}, Ann. Global Anal.

7940: Geom. 1 (1983), no 3, 51--75.

7941: 

7942: \bibitem{Wit1} E.~Witten, {\em Bound states of strings and p-Branes},

7943: Nuclear Physics B 460 (1996) 335--360.

7944: 

7945: \bibitem{Wo} M.~Wodzicki, {\it Noncommutative residue, Part

7946: I. Fundamentals}. In $K$-theory, arithmetic and geometry.

7947: Lecture Notes in Math., 1289, Springer, Berlin 1987.

7948: 

7949: \bibitem{Zagier} D.~Zagier, {\em Modular forms and differential

7950: operators}, Proc. Indian Acad. Sci. Math. Sci. (K.G.Ramanathan

7951: memorial issue) 104 (1994) N.1, 57--75.

7952: 

7953: \end{thebibliography}

7954: \end{document}

7955: \end

7956: