1: \documentclass[11pt]{article}

2: %\documentclass{casabs}

3: \usepackage{amssymb}\usepackage{amsmath}

4: %\usepackage{latexsym}

5: \usepackage[all]{xy}

6: \usepackage[dvips]{graphicx}

7: \usepackage{pstricks}

8: \usepackage{pst-plot}

9: 

10: %\usepackage{showlabels}

11: 

12: \textwidth = 6.1 in

13: \textheight = 9in

14: \oddsidemargin = 0.0 in

15: \evensidemargin = 0.0 in

16: \topmargin = 0.0 in

17: \headheight = 0.0 in

18: \headsep = 0.0 in

19: 

20: 

21: \numberwithin{equation}{section}

22: 

23: \def\be{\begin{equation}}

24: \def\ee{\end{equation}}

25: \def\ba{\begin{align}}

26: \def\ea{\end{align}}

27: 

28: \def\p{\partial}

29: 

30: \def\yboxit#1#2{\vbox{\hrule height #1 \hbox{\vrule width #1

31: \vbox{#2}\vrule width #1 }\hrule height #1 }}

32: \def\fillbox#1{\hbox to #1{\vbox to #1{\vfil}\hfil}}

33: \def\ybox{{\lower 1.3pt \yboxit{0.4pt}{\fillbox{8pt}}\hskip-0.2pt}}

34: %

35: % degree of W

36: \def\d{\kappa}

37: \def\hSigma{{\hat\Sigma}}

38: %

39: \def\CY#1{{\rm CY$_{#1}$}}

40: \def\Coh{{\bf Coh\ }}

41: \def\Cat{{\bf Cat}}

42: \def\l{\left}

43: \def\r{\right}

44: \def\mapr{\mathop{\longrightarrow}\limits}

45: \def\grade{\varphi}

46: \def\comments#1{}

47: \def\cc{{\rm c.c.}}

48: \def\cof{{\rm cof~}}

49: \def\Nbar{{\bar N}}

50: \def\NRR{{N_{RR}}}

51: \def\NNS{{N_{NS}}}

52: \def\tM{\tilde M}

53: \def\ib{{\bar i}}

54: \def\jb{{\bar j}}

55: \def\kb{{\bar k}}

56: \def\lb{{\bar l}}

57: \def\mb{{\bar m}}

58: \def\nb{{\bar n}}

59: \def\ab{{\bar a}}

60: \def\bb{{\bar b}}

61: \def\cb{{\bar c}}

62: \def\Cb{{\bar C}}

63: \def\db{{\bar d}}

64: \def\Dbar{{\bar D}}

65: \def\zb{{\bar z}}

66: \def\Zb{{\bar Z}}

67: \def\phib{{\bar \phi}}

68: \def\taub{{\bar \tau}}

69: \def\QC{\Bbb{C}}

70: \def\QH{\Bbb{H}}

71: \def\QM{\Bbb{M}}

72: \def\QR{\Bbb{R}}

73: \def\QX{\Bbb{X}}

74: \def\QZ{\Bbb{Z}}

75: \def\p{\partial}

76: \def\delbar{{\bar\partial}}

77: \def\tilp{\tilde\partial}

78: \def\eps{\epsilon}

79: \def\half{\frac{1}{ 2}}

80: \def\Tr{{{\rm Tr~ }}}

81: \def\tr{{\rm tr\ }}

82: \def\Re{{\rm Re\hskip0.1em}}

83: \def\Im{{\rm Im\hskip0.1em}}

84: \def\even{{\rm even}}

85: %\def\odd{{\rm odd}}

86: \def\lcm{{\rm lcm}}

87: \def\diag{{\rm diag}}

88: \def\bra#1{{\langle}#1|}

89: \def\ket#1{|#1\rangle}

90: \def\bbra#1{{\langle\langle}#1|}

91: \def\kket#1{|#1\rangle\rangle}

92: \def\vev#1{\langle{#1}\rangle}

93: \def\bigvev#1{\bigg\langle{#1}\bigg\rangle}

94: \def\Dslash{\rlap{\hskip0.2em/}D}

95: \def\intersect{\cdot}

96: \def\cA{{\cal A}}

97: \def\cC{{\cal C}}

98: \def\cD{{\cal D}}

99: \def\cE{{\cal E}}

100: \def\CF{{\cal F}}

101: \def\cF{{\cal F}}

102: \def\cG{{\cal G}}

103: \def\cH{{\cal H}}

104: \def\cI{{\cal I}}

105: \def\cK{{\cal K}}

106: \def\cL{{\cal L}}

107: \def\cM{{\cal M}}

108: \def\cN{{\cal N}}

109: \def\cO{{\cal O}}

110: \def\cP{{\cal P}}

111: \def\cQ{{\cal Q}}

112: \def\cS{{\cal S}}

113: \def\cT{{\cal T}}

114: \def\cU{{\cal U}}

115: \def\cV{{\cal V}}

116: \def\cW{{\cal W}}

117: \def\cX{{\cal X}}

118: \def\Det{{\rm Det}}

119: \def\CP{{\cal P}}

120: \def\CM{{\cal M}}

121: \def\CN{{\cal N}}

122: \def\CR{{\cal R}}

123: \def\CD{{\cal D}}

124: \def\CF{{\cal F}}

125: \def\CJ{{\cal J}}

126: \def\CP{{\cal P }}

127: \def\CL{{\cal L}}

128: \def\CV{{\cal V}}

129: \def\CO{{\cal O}}

130: \def\CZ{{\cal Z}}

131: \def\CE{{\cal E}}

132: \def\CG{{\cal G}}

133: \def\CH{{\cal H}}

134: \def\CC{{\cal C}}

135: \def\CB{{\cal B}}

136: \def\CS{{\cal S}}

137: \def\CA{{\cal A}}

138: \def\CC{{\cal C}}

139: \def\CK{{\cal K}}

140: \def\CP{{\cal P }}

141: \def\CQ{{\cal Q }}

142: \def\CE{{\cal E }}

143: \def\CV{{\cal V }}

144: \def\CZ{{\cal Z }}

145: \def\CS{{\cal S }}

146: \def\CY{{\cal Y }}

147: \def\CW{{\cal W}}

148: \def\CI{{\cal I}}

149: \def\ad#1#2{\frac{\delta}{\delta\sigma^{#1}(#2)}}

150: \def\ppt{\frac{\partial}{\partial t}}

151: \def\comment#1{[#1]}

152: \def\nl{\hfill\break}

153: \def\ap{\alpha'}

154: \def\floor#1{{#1}}

155: \def\sgn{{\rm sgn\ }}

156: \def\P{\BP}

157: \def\I{I}

158: \def\IA{IA}

159: \def\II{\relax{I\kern-.10em I}}

160: \def\IIa{{\II}a}

161: \def\IIb{{\II}b}

162: \def\TeV{{\rm TeV}}

163: \def\AdS{{\rm AdS}}

164: %

165: \def\imp{$\Rightarrow$}

166: \def\IZ{\relax\ifmmode\mathchoice

167: {\hbox{\cmss Z\kern-.4em Z}}{\hbox{\cmss Z\kern-.4em Z}}

168: {\lower.9pt\hbox{\cmsss Z\kern-.4em Z}}

169: {\lower1.2pt\hbox{\cmsss Z\kern-.4em Z}}\else{\cmss Z\kern-.4em

170: Z}\fi}

171: \def\IB{\relax{\rm I\kern-.18em B}}

172: \def\IC{{\relax\hbox{$\inbar\kern-.3em{\rm C}$}}}

173: \def\ID{\relax{\rm I\kern-.18em D}}

174: \def\IE{\relax{\rm I\kern-.18em E}}

175: \def\IF{\relax{\rm I\kern-.18em F}}

176: \def\IG{\relax\hbox{$\inbar\kern-.3em{\rm G}$}}

177: \def\IGa{\relax\hbox{${\rm I}\kern-.18em\Gamma$}}

178: \def\IH{\relax{\rm I\kern-.18em H}}

179: \def\II{\relax{\rm I\kern-.18em I}}

180: \def\IK{\relax{\rm I\kern-.18em K}}

181: \def\IN{\relax{\rm I\kern-.18em N}}

182: \def\IP{\relax{\rm I\kern-.18em P}}

183: %\def\IX{\relax{\rm X\kern-.01em X}}

184: %this doesn't work

185: \def\IX{{\bf X}}

186: %

187: \def\inbar{\,\vrule height1.5ex width.4pt depth0pt}

188: \def\mod{{\rm\; mod\;}}

189: \def\ndt{\noindent}

190: \def\p{\partial}

191: \def\pab{\pb_{\bar A} }

192: \def\pb{{\bar \p}}

193: \def\pgp{\pb g g^{-1}}

194: \font\cmss=cmss10 \font\cmsss=cmss10 at 7pt

195: \def\IR{\relax{\rm I\kern-.18em R}}

196: \def\pbar{\bar{\p}}

197: \def\qmvw{\CM_{\vec \zeta}(\vec v, \vec w) }

198: \def\sdtimes{\mathbin{\hbox{\hskip2pt\vrule

199: height 4.1pt depth -.3pt width .25pt\hskip-2pt$\times$}}}

200: \def\im{{\rm im\ }}

201: \def\ker{{\rm ker\ }}

202: \def\cok{{\rm cok\ }}

203: \def\End{{\rm End\ }}

204: \def\vol{{\rm vol}}

205: \def\id {{\bf 1}}

206: \def\ch{{\rm ch}}

207: \def\chern{{\rm c}}

208: \def\rank{{\rm rank}}

209: \def\aroof{{\rm \hat A}}

210: %

211: \def\CP{\IP}

212: \def\CM{{\cal M}}

213: \def\CN{{\cal N}}

214: \def\CZ{{\cal Z}}

215: %

216: \def\BR{\IR}

217: \def\BZ{\IZ} % for now

218: \def\BP{\IP}

219: \def\BR{\IR}

220: \def\BC{\IC}

221: \def\BM{\QM}

222: \def\BH{\QH}

223: \def\BX{\QX}

224: %

225: \def\ls{l_s}

226: \def\ms{m_s}

227: \def\gs{g_s}

228: \def\lp10{l_P^{10}}

229: \def\lp11{l_P^{11}}

230: \def\R11{R_{11}}

231: 

232: \def\ch{{\rm ch}}

233: 

234: \def\codim{{\mathop{\rm codim}}}

235: \def\cok{{\rm cok}}

236: \def\coker{{\mathop {\rm coker}}}

237: 

238: \def\diff{{\rm diff}}

239: \def\Diff{{\rm Diff}}

240: \def\Det{{\rm Det}}

241: \def\Tr{{\rm Tr}}

242: 

243: \def\gpg{g^{-1} \p g}

244: \def\ib{{\bar i}}

245: \def\zb {\bar{z}}

246: \def\xb {\bar{x}}

247: \def\mb {\bar{m}}

248: \def\Jb {\bar{J}}

249: \def\Xb {\bar{X}}

250: \def\Hb {\bar{H}}

251: \def\Gb {\bar{G}}

252: \def\jb {\bar{j}}

253: \def\psib{\bar{\psi}}

254: \def\wb {\bar{w}}

255: \def\qb {\bar{q}}

256: \def\G{\Gamma}

257: 

258: %% MORE MACROS

259: 

260: \font\manual=manfnt

261: \def\dbend{\lower3.5pt\hbox{\manual\char127}}

262: \def\danger#1{\smallskip\noindent\rlap\dbend%

263: \indent{\bf #1}\par\vskip-1.5pt\nobreak}

264: \def\c{\cdot}

265: \def\IZ{\relax\ifmmode\mathchoice

266: {\hbox{\cmss Z\kern-.4em Z}}{\hbox{\cmss Z\kern-.4em Z}}

267: {\lower.9pt\hbox{\cmsss Z\kern-.4em Z}} {\lower1.2pt\hbox{\cmsss

268: Z\kern-.4em Z}}\else{\cmss Z\kern-.4em Z}\fi}

269: \def\half {\frac{1}{ 2}}

270: \def\sdtimes{\mathbin{\hbox{\hskip2pt\vrule height 4.1pt depth -.3pt

271: width .25pt \hskip-2pt$\times$}}  }

272: 

273: \def\p{\partial}

274: \def\pb{\bar{\partial}}

275: \def\bar{\overline}

276: \def\CS{{\cal S}}

277: \def\CN{{\cal N}}

278: \def\CH{{\cal H}}

279: \def\rt2{\sqrt{2}}

280: \def\irt2{\frac{1}{\sqrt{2}}}

281: \def\rtt{\sqrt{3}}

282: \def\hrtt{\frac{\sqrt{3}}{ 2}}

283: \def\t{\tilde}

284: \def\ndt{\noindent}

285: \def\s{\sigma}

286: \def\b{\beta}

287: \def\a{\alpha}

288: \def\w{\omega}

289: \def\psid{\psi^{\dagger}}

290: 

291: \def\lieg{{\bf \underline{g}}}

292: \def\liet{{\bf \underline{t}}}

293: \def\mod{{\rm mod}}

294: \def\Det{{\rm Det}}

295: 

296: 

297: %% MORE MACROS

298: \font\cmss=cmss10

299: \font\cmsss=cmss10 at 7pt

300: \def\IL{\relax{\rm I\kern-.18em L}}

301: \def\IH{\relax{\rm I\kern-.18em H}}

302: \def\IR{\relax{\rm I\kern-.18em R}}

303: \def\inbar{\vrule height1.5ex width.4pt depth0pt}

304: \def\IC{\relax\hbox{$\inbar\kern-.3em{\rm C}$}}

305: \def\rlx{\relax\leavevmode}

306: %%

307: \def\ZZ{\rlx\leavevmode\ifmmode\mathchoice{\hbox{\cmss Z\kern-.4em Z}}

308:  {\hbox{\cmss Z\kern-.4em Z}}{\lower.9pt\hbox{\cmsss Z\kern-.36em Z}}

309:  {\lower1.2pt\hbox{\cmsss Z\kern-.36em Z}}\else{\cmss Z\kern-.4em

310:  Z}\fi}

311: %%% misc.

312: \def\IZ{\relax\ifmmode\mathchoice

313: {\hbox{\cmss Z\kern-.4em Z}}{\hbox{\cmss Z\kern-.4em Z}}

314: {\lower.9pt\hbox{\cmsss Z\kern-.4em Z}}

315: {\lower1.2pt\hbox{\cmsss Z\kern-.4em Z}}\else{\cmss Z\kern-.4em

316: Z}\fi}

317: 

318: %% MORE MACROS

319: \def\codim{{\mathop{\rm codim}}}

320: \def\cok{{\rm cok}}

321: \def\coker{{\mathop {\rm coker}}}

322: 

323: \def\ch{{\rm ch}}

324: \def\diff{{\rm diff}}

325: \def\Diff{{\rm Diff}}

326: \def\Det{{\rm Det}}

327: \def\Tr{{\rm Tr}}

328: \def\gpg{g^{-1} \p g}

329: \def\ib{{\bar i}}

330: \def\zb {\bar{z}}

331: \def\nb {\bar{n}}

332: \def\cb {\bar{c}}

333: \def\wb {\bar{w}}

334: \def\qb {\bar{q}}

335: \def\G{\Gamma}

336: 

337: 

338: \font\manual=manfnt

339: \def\dbend{\lower3.5pt\hbox{\manual\char127}}

340: \def\danger#1{\smallskip\noindent\rlap\dbend%

341: \indent{\bf #1}\par\vskip-1.5pt\nobreak}

342: \def\c{\cdot}

343: \def\IZ{\relax\ifmmode\mathchoice

344: {\hbox{\cmss Z\kern-.4em Z}}{\hbox{\cmss Z\kern-.4em Z}}

345: {\lower.9pt\hbox{\cmsss Z\kern-.4em Z}} {\lower1.2pt\hbox{\cmsss

346: Z\kern-.4em Z}}\else{\cmss Z\kern-.4em Z}\fi}

347: \def\half {\frac{1}{ 2}}

348: \def\sdtimes{\mathbin{\hbox{\hskip2pt\vrule height 4.1pt depth -.3pt

349: width .25pt \hskip-2pt$\times$}}  }

350: 

351: \def\pb{\bar{\partial}}

352: \def\bar{\overline}

353: 

354: \def\rt2{\sqrt{2}}

355: \def\irt2{\frac{1}{\sqrt{2}}}

356: \def\rtt{\sqrt{3}}

357: \def\hrtt{\frac{\sqrt{3}}{ 2}}

358: \def\t{\tilde}

359: \def\T{\widetilde}

360: \def\ndt{\noindent}

361: \def\s{\sigma}

362: \def\bB {{\bf B}}

363: \def\sNSNS{\text{\tiny{NSNS}}}

364: \def\sRR{\text{\tiny{RR}}}

365: \def\sNS{\text{\tiny{NS}}}

366: \def\sR{\text{\tiny{R}}}

367: \def\sId{\text{\tiny{Id}}}

368: \def\scont{\text{\tiny{cont}}}

369: \def\sdisc{\text{\tiny{disc}}}

370: 

371: \input{epsf}

372: 

373: \usepackage{epsfig}

374:   

375: \title{\Large{\bf Non-compact Gepner Models, Landau-Ginzburg Orbifolds and

376:     Mirror Symmetry}} \author{Sujay K. Ashok$^{a,b}$, Raphael Benichou$^{c}$

377:   and Jan Troost$^{c}$ } \date{}

378: \begin{document}

379: \maketitle

380: 

381: \begin{center}

382:   $^{a}$Institute of Mathematical Sciences\\

383:   C.I.T Campus, Taramani\\

384:   Chennai, India 600113\\

385:   $^{b}$Perimeter Institute for Theoretical Physics\\

386:   Waterloo, Ontario, ON N$2$L$2$Y$5$, Canada \\

387:   $^{c}$Laboratoire de Physique Th\'eorique\footnote{Unit\'e Mixte du CRNS et

388:     de l'Ecole Normale Sup\'erieure associ\'ee \`a l'universit\'e Pierre et

389:     Marie Curie 6, UMR

390:     8549. Preprint LPTENS-07/51.}, Ecole Normale Sup\'erieure \\

391:   $24$ Rue Lhomond Paris $75005$, France

392: \end{center}

393: 

394: \begin{abstract}

395:   We study non-compact Gepner models that preserve sixteen 

396: or eight

397:   supercharges in type II string theories. In particular, we develop an

398:   orbifolded Landau-Ginzburg description of these models analogous to the

399:   Landau-Ginzburg formulation of compact Gepner models. The Landau-Ginzburg

400:   description provides an easy and direct access to the geometry of the

401:   singularity associated to the non-compact Gepner models.  Using these tools,

402:   we are able to give an intuitive account of the chiral rings of the models,

403:   and of the massless moduli in particular. By studying orbifolds of the

404:   singular linear dilaton models, we describe mirror pairs of non-compact

405:   Gepner models by suitably adapting the Greene-Plesser

406:   construction of 

407: mirror pairs for the compact case. For particular models, we

408:   take a large level, low curvature limit in which we can

409:   analyze corrections to a flat space orbifold approximation of the

410:   non-compact Gepner models. This gives rise to a counting of moduli which

411:   differs from the toric counting in a subtle way.

412: \end{abstract}

413: 

414: \section{Introduction}

415: Mirror symmetry for Calabi-Yau $3$-folds is a subject of great interest 

416: to physicists as well as mathematicians \cite{Yau:1998yt}\cite{Greene:1997ty}.

417: Mirror pairs were first exhibited \cite{Greene:1990ud}

418: by studying orbifolds of the quintic at the Gepner point \cite{Gepner:1987qi}

419: in the Calabi-Yau moduli space. This was explored in more detail in

420: \cite{Candelas:1990rm} which contained the first mathematical predictions from

421: mirror symmetry. Many generalizations were found using the Landau-Ginzburg formulation that were applicable in non-geometric regimes \cite{Berglund:1991pp}.

422: 

423: In this paper, we wish to extend the study of mirror symmetry to non-compact

424: Gepner models. Various approaches towards this problem have been followed.

425: Toric Calabi-Yau manifolds have been very well studied in the literature

426: following the construction of mirror pairs for hypersurfaces in toric

427: varieties \cite{Batyrev} and for non-compact Calabi-Yau manifolds (see

428: \cite{Chiang:1999tz} and references therein). In the physics literature, this

429: has been reformulated using a gauged linear sigma model \cite{Witten:1993yc,

430:   Hori:2000kt, Hori:2002cd}. Progress has also been made for more general

431: $c=9$ theories via conformal field theory techniques \cite{Eguchi:2004yi}. In

432: particular, the work on T-duality for the $N=2$ cigar and the Liouville

433: conformal field theories is relevant in this context \cite{Giveon:1999px,

434:   Hori:2001ax, Tong:2003ik, Israel:2004jt}.

435: 

436: One of the gaps we aim to fill in this paper is to clarify the link between

437: non-compact Gepner models and their Landau-Ginzburg description. For compact

438: Calabi-Yau manifolds that are hypersurfaces in weighted projective space, the

439: Landau-Ginzburg description is closely tied to the geometry

440: \cite{Greene:1988ut}. These are two of the many phases in a gauged linear

441: sigma-model description \cite{Witten:1993yc}. The Landau-Ginzburg model is also extremely useful for providing a simple setting to do computations. For instance, it has served well in the

442: past for establishing the link between compact Gepner models and compact

443: Calabi-Yau manifolds \cite{Greene:1988ut} as well as to provide the first list

444: of Calabi-Yau hypersurfaces in $W\IC\IP^4$ \cite{Candelas:1989hd}.

445: 

446: Our approach in this paper is to stress the ingredients of the analysis

447: of Gepner models that survive the transition from going from the compact to

448: the non-compact case. We alo point out the characteristics that differ

449: in the two cases.

450: Moreover we study in detail the orbifolded Landau-Ginzburg description of these models

451: \cite{Vafa:1989xc, Intriligator:1990ua}. This allows us to 

452: compare our models with geometric backgrounds which can be used as

453: internal spaces for type II strings on $\IR^{3,1}$.

454: 

455: In section \ref{noncompactgepner} we will review the asymptotic

456: partition function of non-compact Gepner models \cite{Eguchi:2000tc,Hanany:2002ev,Murthy:2003es,Eguchi:2004yi,Israel:2004ir}. The asymptotic partition

457: function, which is proportional to the divergent volume of space-time,

458: otherwise behaves much as

459: the partition function in the compact case. In particular we show here that

460: the $\beta$-method of Gepner \cite{Gepner:1987qi} for constructing modular

461: invariants can be adapted to the non-compact case. We then start the

462: discussion of the localized spectrum

463: \cite{Hanany:2002ev,Eguchi:2004yi,Israel:2004ir}, which is

464:  characteristic of non-compact models.

465: 

466:  In section \ref{LG} we will link the non-compact Gepner models to

467:  Landau-Ginzburg models. We briefly remark on the difference between the

468:  compact and non-compact Landau-Ginzburg model \cite{Ooguri:1995wj}. For

469:  instance, the former gives rise to a unital chiral ring, while the latter

470:  gives rise to a ring without unit element. We continue the analysis in

471:  section \ref{orbifoldedLG} with a discussion of the orbifolds of the

472:  Landau-Ginzburg models that are necessary to implement the GSO projection,

473:  and other orbifold groups. We will see that the formalism for counting chiral

474:  ring elements largely carries over from the compact case

475:  \cite{Vafa:1989xc,Intriligator:1990ua}. However, it will become intuitive

476:  that the Landau-Ginzburg potentials with negative powers exclude some of the

477:  (anti-)chiral ring elements as the potential renders them non-normalizable.

478:  Continuing the formal counting exercise will turn out to be useful

479:  nevertheless. It gives rise to a natural picture of mirror symmetry in

480:  conformal field theories, that is strongly reminiscent of its compact

481:  counterpart \cite{Greene:1990ud, Eguchi:2004yi}. We will discuss this point

482:  in section \ref{CFTmirror}.

483: 

484:  In section \ref{examples} we provide new examples of mirror conformal field

485:  theories. The techniques developed to analyze orbifolded Landau-Ginzburg

486:  models come in handy when treating these more complicated models. The mirror

487:  pairs of conformal field theories are best understood as linear dilaton

488:  backgrounds with $N=2$ superconformal symmetry. These %in turn 

489: can be deformed or resolved, to give rise to perturbatively well-defined mirror string backgrounds.  Finally, in section \ref{geometrymirror}, we discuss examples in which we can relate the mirror conformal field theories to geometries. It will turn out that we can approximate certain conformal field theories at large level with orbifold singularities that admit a toric description. At infinite level, we find agreement between the conformal field theory and the geometric results. At finite level, we find that the conformal field theory description takes into account various modifications to the background that can lift some moduli. In section \ref{conclusions} we discuss further applications of our results.

490: 

491: \section{Non-compact Gepner Models}

492: \label{noncompactgepner}

493: We start out with a rather brief but technical review of non-compact Gepner models (see

494: e.g. \cite{Eguchi:2000tc,Murthy:2003es,Eguchi:2004yi,Israel:2004ir})

495: in order to clarify the fact that the asymptotic partition function of

496: non-compact Gepner models can be constructed using the same

497: tools as in the compact case. 

498: We can then lay bare properties of the models along the same lines as in the

499: compact Gepner models.  In the second part of this section we give a preview

500: of the ingredients that go into analyzing the localized part of the partition

501: function.

502: 

503: \subsection{The Asymptotic Non-compact Gepner Models}

504:  \label{asymptotic}

505: 

506: 

507: For simplicity we will work with type II string theory on $\IR^{3,1}$ times an

508: internal (non-compact) conformal field theory of central charge

509: nine\footnote{The constructions can be extended to lower-dimensional flat

510:   spaces and to heterotic string theories, with little effort and lots of

511:   indices.}.  The internal conformal field theory is built from a product of

512: $N=2$ superconformal field theories. These can be split into three classes

513: depending on whether their central charge is smaller, larger 

514: than or equal to

515: three.

516: 

517: \begin{itemize}

518: \item[\_]The minimal $N=2$ superconformal field theories  (see

519:   e.g. \cite{Gepner:1986hr})

520: have central

521: charge smaller than three. They can be

522: viewed as coset conformal field theories of the form $SU(2)_{k-2} \times

523: U(1)_2 / U(1)_k$ of central charge $c_{MM}=3 -\frac{6}{k}$. The level $k$ is

524: the supersymmetric level of the total $SU(2)$ current algebra present in the

525: parent $N=1$ Wess-Zumino-Witten model. It is a positive integer

526: larger than or equal to two.

527: 

528:  Since the minimal $N=2$ superconformal models

529: have been reviewed frequently 

530: (\cite{Greene:1996cy,Maldacena:2001ky}), and since they are

531: standard in the construction of Gepner models \cite{Gepner:1987qi}, we only

532: very briefly recall some of their properties. The primaries of the model can

533: be labeled by three quantum numbers: the spin $j$ under the $SU(2)_{k-2}$

534: current algebra, the $\IZ_{2k}$ valued chiral momentum $n$ under the $U(1)_k$

535: current algebra and the $\IZ_4$ valued chiral momentum $s$ labeling a

536: representation of the $U(1)_2$ current algebra.  They satisfy the selection

537: rule: $2j\equiv n+s\ [2]$. We moreover have an equivalence between the

538: following representations: $(j,n,s) \equiv (\frac{k-2}{2}-j,n-k,s+2)$.  The

539: left-moving $U(1)_R$ charge $Q_{MM}$ for a primary is equal to $Q_{MM}=

540: \frac{n}{k} + \frac{s}{2}$.

541: 

542: \item[\_]The second class of theories has central charge

543: larger than three. An example in this class is an $N=2$ linear dilaton theory

544: with a slope such that the central charge is equal to $c=3+\frac{6}{l}$ with

545: positive and real values for the parameter $l$.

546: The superconformal algebra with central charge larger than three has

547: continuous unitary representations which are conveniently labeled

548: \cite{Dixon:1989cg} by a Casimir $j=1/2 + ip $ where $p \in \IR^+$, by an

549: integer momentum $2m \in \IZ$ and a $\IZ_4$ fermionic quantum number $s$. The

550: left-moving R-charge $Q_{nc}$ of a primary is $Q_{nc} = \frac{2m}{l}

551: +\frac{s}{2}$. (See e.g. \cite{Israel:2004jt} for a detailed discussion.)

552: 

553: \item[\_]The $N=2$ superconformal

554: algebra with $c=3$ is exceptional, and can be represented for instance by free

555: scalars (which can realize compact or non-compact target space directions).

556: \end{itemize}

557: 

558: 

559: In order to mimic the Gepner construction, we will assume that in the case

560: $c>3$, the chiral

561: algebra has some more structure (see e.g. \cite{Eguchi:2000tc}). For simplicity we will

562: work under the assumption that the parameter $l$ is a positive integer\footnote{It is sufficient to suppose that the central

563: charge is parameterized by a positive fractional level $l$ \cite{Eguchi:2000tc}.}.

564: We can then add to

565: the chiral algebra the generator of spectral flow on the $N=2$ superconformal

566: algebra by $2 l$ units.  The characters of the extended $N=2$ superconformal

567: algebra in the continuous representations are given by:

568: \begin{eqnarray}

569: Ch_{cont} (j,2m,s) &=& q^{\frac{p^2}{l}}   \frac{1}{\eta^3(\tau)}

570: \Theta_{s,2}(\tau) 

571: \Theta_{2m,l} (\tau).

572: \end{eqnarray}

573: It is crucial to us that the modular transformation properties of the

574: characters hinge upon the presence of the  

575: $\theta$-functions at levels $2$ and $l$.

576: 

577: 

578: \subsubsection{Levels And Charge Lattice}

579: 

580: We observe that the characters of the $N=2$ minimal models at

581: level $k$ transform as $\theta$-functions at levels $2$ and $-k$,

582: while the extended characters of the $N=2$ models with central charge

583: $c=3+\frac{6}{l}$ transform as $\theta$-functions at level $2$ and

584: $+l$. We thus note a first important sign difference 

585: in the transformation rules of the characters. Modular

586: invariants in the continuous sector of the theory can  be based on

587: modular invariants of $\theta$-functions. For one $U(1)_k$ current algebra

588: at level $k$

589: these are well-known to correspond to the divisors of $k$ via

590: orbifolding of the diagonal modular invariant. For a product of

591: $\theta$-functions, the analysis is more complicated,

592:  but a large class of modular invariants can be constructed by

593: taking products of modular invariants of the factors, and then

594: orbifolding.

595: 

596: In order to write down the modular invariant partition functions,

597: it is useful to introduce a charge lattice for the various $U(1)$

598: current algebras in

599: the theory. We introduce a vector of levels 

600: $(2,2, \dots, 2; k_1, \dots , k_p ; l_1, \dots, l_q)$ 

601: where $p$ is the number of

602: minimal model factors, $q$ is the number of non-compact factors, while

603: the number of fermionic levels equal to $2$ is $1+q+p$. Indeed, we work

604:  in light-cone gauge on $\IR^{3,1}$ such that there is one complex fermion

605: associated to the flat space directions (and there is

606: one complex fermion per factor

607: model). 

608: The charge lattice is periodic. In each direction, the periodicity is twice

609: the level.

610: A point in the lattice is defined by a charge vector $r= (s_0,s_1, \dots,

611: s_{p+q}; n_1,\dots,n_p; 2m_1, \dots, 2m_q)$ where we used the

612: notation

613: $s_0$ for the charge of the flat

614: space fermions under the $U(1)_2$ current algebra, and

615: similarly for the other fermions, while we copy 

616: the traditional notation for the

617: chiral momentum quantum numbers of compact and non-compact factors

618: that we introduced above (including their normalization). 

619: The scalar product on the charge lattice is defined as follows:

620: \begin{eqnarray}

621: r^{(1)} \cdot r^{(2)} &=& -\frac{s_0^{(1)} s_0^{(2)}}{4} - \frac{s_1^{(1)} s_1^{(2)}}{4} - \dots

622: - \frac{s_{p+q}^{(1)} s_{p+q}^{(2)}}{4} 

623: \nonumber \\ 

624: & & + \frac{n_1^{(1)} n_1^{(2)}}{2k_1} + \dots  + \frac{n_p^{(1)} n_p^{(2)}}{2k_p}

625: \nonumber \\

626: & & - \frac{(2m_1^{(1)})(2m_1^{(2)})}{2l_1} -\dots - \frac{(2m_q^{(1)})(2m_q^{(2)})}{2l_q}\,.                               

627: \end{eqnarray}

628: The contribution of the chiral momenta corresponding to the non-compact factors comes with an opposite sign from those of the compact factors. The all-important signature of the

629: quadratic form is therefore $(-,\dots,-; +, \dots, + ; - , \dots, -)$.

630: 

631: \subsubsection{A Canonical Vector}

632: 

633: We introduce the following vector $\beta_0$ in the charge lattice\footnote{We

634:   take the entries for the fermions to be minus one, in order to accord with

635:   the convention that the left-moving $U(1)_R$ charge 

636: is given for instance for a minimal

637:   model factor by $Q_{MM}=\frac{n}{k}+\frac{s}{2}$.}:

638: \begin{eqnarray}

639: \beta_0 &=& (-1,-1, \dots, -1; 1,1,\dots,1; -1,-1, \dots,-1).

640: \end{eqnarray}

641: We have that the left-moving R-charge $Q$ for a primary state with charge

642: vector $r$ is equal to $Q=2 \beta_0 \cdot r$. We moreover have that

643: \begin{eqnarray}

644: \beta_0 \cdot \beta_0 &=& -\frac{1+p+q}{4} + \sum_{i=1}^p \frac{1}{2k_i} - \sum_{j=1}^q \frac{1}{2l_i}

645: \nonumber \\

646: &=& -1

647: \end{eqnarray}

648: where we used that the total central charge of the light-cone gauge conformal field theory is equal to 

649: \be

650: 12 = 3(1+p+q) -  \sum_{i=1}^p \frac{6}{k_i} + \sum_{j=1}^q \frac{6}{l_i} \,.

651: \ee

652: In summary, the vector $\beta_0$ is useful to measure the R-charge, and

653: squares to minus one.

654: 

655: For the right-movers we will always take identical conventions to the

656: left-movers. In particular, the $N=2$ superconformal algebras have the

657: same structure constants. We will also define the right-moving

658: R-charge of the minimal model factors to be $\tilde{Q}_{MM} =

659: \frac{\tilde{n}}{k} + \frac{\tilde{s}}{2}$ and for the non-compact factors

660: $\tilde{Q}_{nc}=\frac{2 \tilde{m}}{k} + \frac{\tilde{s}}{2}$, while the charge vector

661: for the right-movers is $\tilde{r} = ( \tilde{s}_0, \dots, ; \dots ;

662: \dots, 2 \tilde{m}_{p+q})$. So, for the right-movers as well we have

663: that the total $U(1)_R$ charge is given by $\tilde{Q}= 2 \beta_0 \cdot

664: \tilde{r}$ with the same vector $\beta_0$. 

665: 

666: 

667: \subsubsection{Products of $\theta$-functions}

668: We define the following notation for the product of characters of the

669:  flat space fermions, the minimal model and the non-minimal $N=2$

670:  superconformal field theories. Since the characters transform like

671:  $\theta$-functions, we introduce the symbol:

672: \begin{eqnarray}

673: \Theta_{r} (\tau) &=& \Theta_{s_0,2}(\tau) \prod_{i=1}^p \chi_{j_i,n_i,s_i} (\tau)

674:  \prod_{i=1}^q Ch_{cont}(j_{p+i},2m_i,s_{p+i}) (\tau)

675: \end{eqnarray}

676: where $r$ is the total charge vector, and the first factor corresponds

677: to the flat space fermions, while the following $p$ factors correspond to

678: minimal model characters, 

679: and the final $q$ factors to the non-compact continuous

680: extended characters. In the $\Theta$ symbol we have left implicit the

681: labels corresponding to the levels, as well as those corresponding to

682: the Casimirs of the minimal and non-minimal models. The important point is that

683: the $\Theta$-functions transform as a product of ordinary $\theta$-functions

684: under modular transformations.

685: 

686: We introduce now the first modular invariant partition function which is the

687: diagonal modular invariant:

688: \begin{eqnarray}

689: Z_{diag} &=& \sum_{r} \Theta_{r} (\tau) \Theta_r (\bar{\tau})\,, 

690: \end{eqnarray}

691: where the diagonal sum over $r$ is over all inequivalent charges in the charge

692: lattice.  Implicitly, we take a diagonal A-type modular invariant for the

693: Casimir invariants $j$ for all factors\footnote{It would be interesting to

694:   study the more general modular invariants of type D and/or type E for the

695:   compact factors.}. We have suppressed the divergent non-compact volume factor

696: in the formula for the asymptotic partition function.

697: 

698: \subsubsection{Locality Orbifold}

699: 

700: Now that we have set-up our theory in a way which is very analogous to

701: \cite{Gepner:1987qi}, we can follow that reference closely. Along the way, we

702: reformulate a few minor points in a more modern orbifold language.

703: 

704: Locality in string theory 

705: only allows for fermions having either

706: Neveu-Schwarz (NS) or Ramond (R) boundary conditions for all factors simultaneously for the

707: left- or the right-movers. We implement that locality constraint by

708: orbifolding the diagonal partition function by a diagonal $(\IZ_2)^{p+q}$ group.

709: The i'th $\IZ_2$ acts on a state with charge vectors $r$ and $\tilde{r}$ as

710: $(-1)^{(s_0+s_i+\tilde{s}_0 + \tilde{s}_i)/2}$ for $i=1, \dots, p+q$. The action

711: can be summarized by introducing vectors $\beta_i$ which have a $2$ as

712: the first and the $i+1$'th entry for $i=1, \dots, p+q$. 

713: Then the action can equivalently be written as $(-1)^{\beta_i \cdot (r+\tilde{r})}$.

714: On an untwisted state with charges $s_0$ and $s_i$, the action is the

715: multiplication by a phase $(-1)^{s_0+s_i}$. 

716: 

717: The states invariant under $(\IZ_2)^{p+q}$ will be purely NS or purely R for

718: the left-movers. Since the partition function was diagonal, the same condition

719: will hold for the right-movers.  The orbifold also introduces twisted states.

720: These have left- and right-moving charge vectors that differ by multiples of

721: the vectors $\beta_i$ \cite{Gepner:1987qi}. Indeed, since the fermion number

722: $s_i$ is defined modulo four, this introduces a twisted state sector for each

723: $\IZ_2$ orbifold factor, and moreover, since $\beta_i^2=2$, we have that the

724: twisted states we introduce in this fashion are also orbifold invariant

725: \cite{Gepner:1987qi}.  Each of the $\IZ_2$ orbifolds introduces twisted

726: states, which renders the sum over left- and right-moving fermion numbers $s_i$

727: and $\tilde{s_i}$ independent, except for the fact that they need to be of the

728: same parity as the flat space fermion numbers $s_0$ and $\tilde{s}_0$

729: respectively. The above prescription is equivalent to the standard orbifold

730: procedure and produces a new modular invariant partition function.  Note that

731: the flat space fermion quantum numbers $s_0$ and $\tilde{s}_0$ are still of

732: equal parity.  We therefore only have NSNS and RR states at this point.  The

733: sum over the left- and right-moving worldsheet fermion numbers $s_0$ and

734: $\tilde{s}_0$ will become decoupled after performing a final $\IZ_2$ GSO

735: projection, thus introducing fermions.

736: 

737: 

738: \subsubsection{Integer R-charge Orbifold}

739: \label{integerR}

740: 

741: The standard GSO projection in string theory is based on the fact that

742: the partition function only has integer R-charges. In order to ensure

743: this condition in a Gepner model, we perform yet another orbifold. The

744: orbifold action on a state with charge $r$ and $\tilde{r}$ will be $exp

745: (2 \pi i \beta_0 \cdot (r + \tilde{r}))$. We first note that $\beta_0

746: \cdot \beta_i$ is an integer, such that the action of the orbifold on

747: $\beta_i$ twisted sectors is identical to the action on  $\beta_i$ untwisted

748: sectors. The order of the $2 \beta_0$ orbifold is therefore the order

749: of the operator $e^{ 2 \pi i Q } $ in the untwisted theory (where

750: $Q$ is the total left-moving R-charge in light-cone gauge)\footnote{The theory

751: we start out with obeys the charge relation $r=\tilde{r}$. 

752: Since $\beta_0 \cdot \beta_i \in \IZ$, we

753: still have the relation $e^{ 2 \pi i (Q+\tilde{Q})/2 } = e^{ 2 \pi i Q } $ after the

754: $(\IZ^{p+q}_2)$ orbifold.}. The order of

755: that orbifold is the smallest common divisor $d$ of all the levels in the

756: theory (including the fermionic level $2$ when $p+q$ is even and 

757: not when $p+q$ is odd).

758: 

759: It is clear in the untwisted sector that the orbifold forces the

760:  left-moving (and right-moving) $U(1)_R$ charge to be integer. There

761:  are also $d-1$ twisted sectors which have charge vectors which differ

762:  by multiples of $ 2 \beta_0$. Since $\beta_0^2=-1$, we have that

763:  these twisted sectors also have integer left- and right-moving

764:  R-charges. Thus, the orbifold has provided us with a new modular

765:  invariant partition function with integer R-charges for both left-

766:  and right-movers.  If we introduce the lattice $\Lambda$ generated

767: by the vectors $\beta_i$ and $2 \beta_0$, then we can write the

768:  partition function of the theory

769:  as\footnote{This partition function for non-compact models

770: is the analogue of $Z_0$ in

771:  \cite{Greene:1990ud} for compact Gepner models.}:

772: \begin{eqnarray}

773: Z_0 &=& \sum_{r-\tilde{r} \in \Lambda} \Theta_r (\tau) \Theta_{\tilde{r}} (\bar{\tau})\,,

774: \end{eqnarray}

775: where we restrict the sum to invariant states, namely states obeying the

776: conditions 

777: $ r \cdot \beta_i \in \IZ$ (purely NS or purely R) and 

778: $ r \cdot 2 \beta_0 \in \IZ$ (integer R-charges). Again we have left implicit the diagonal

779: A-type invariant for the minimal models as well as the diagonal integral over

780: radial momenta for the compact factors that diverges like the volume of space-time.

781: 

782: \subsubsection{The Standard GSO Projection}

783: 

784: The partition function now has integer R-charges for all states, and can be

785: GSO projected in the same manner as the flat space partition function. We

786: project onto odd R-charges, i.e. we satisfy the condition $2 \beta_0 \cdot r

787: \in 2 \IZ +1$ as well as $2 \beta_0 \cdot \tilde{r} \in 2 \IZ+1$.

788: The charge difference between left- and right-movers for the twisted states is

789: proportional to an odd multiple of $\beta_0$. The twisted sectors are the NS-R

790: and R-NS sectors of the theory, which correspond to space-time fermions, and

791: contribute negatively to the space-time partition function thus implementing

792: space-time statistics.  For type II theories, we have obtained an

793: asymptotic supersymmetric partition function in the Gepner formalism. Strictly

794: speaking we have a type IIB partition function, since we have made no

795: distinction between left- and right-movers.  It can be transformed easily into

796: a type IIA partition function by flipping the chirality of the final

797: GSO projection for the right-movers in the R-sector.

798: 

799: \subsubsection{Discrete Symmetries}

800: \label{symGroup}

801:  In this subsection, we can follow  \cite{Greene:1990ud} closely

802: since we have set up our model as in the compact case.

803: It is interesting to single out a particular symmetry group of the

804: model. 

805: 

806: The symmetry of the compact models contains a $\IZ_{k_i} \times \IZ_{k_i}$

807: group. Only the diagonal $\IZ_{k_i}$ subgroup has a non-trivial action on a

808: model with diagonal spectrum.

809: This subgroup acts as 

810: \be

811: \Phi_{r,\tilde{r}} \rightarrow exp(2 \pi i

812: (n_i+\tilde{n}_i)/ 2 k_i) \Phi_{r,\tilde{r}}\,.

813: \ee 

814: We can introduce a vector

815: \be 

816: \gamma_i=(0,...,0;0,...,0,2,0,...,0; 0,...,0)

817: \ee 

818: where $2$ is in the $i$'th entry after the first semi-column. The vector codes the action of the symmetry

819: group as follows: 

820: \be

821: \Phi_{r,\tilde{r}} \rightarrow exp(\pi i \gamma_i \cdot (r +

822: \tilde{r}))\Phi_{r,\tilde{r}}\,.

823: \ee 

824: We have a similar action in the non-compact theories. There is 

825: a symmetry group $\IZ_{l_i} \times\IZ_{l_i}$, of which only the diagonal subgroup acts nontrivially on the

826: states. In the full Gepner model, we can think of the product of the diagonal

827: subgroups 

828: \be

829: D=\prod_{i=1}^p \IZ_{k_i} \times \prod_{j=1}^q \IZ_{l_j}

830: \ee 

831: as mapped into the charge lattice via the maps $\gamma_{i=1, \dots, p+q}$.

832: 

833: Now we define the operator $g_0$ that acts by multiplication by $\exp(2\pi i

834: \beta_0 \cdot(r+\tilde{r}))$. It generates a group $\IZ_d$, that we used

835: previously to perform the integer R-charge orbifold.

836: This orbifold group $\IZ_d$ contains a subgroup

837: generated by $g_0^2$, which is a subgroup of

838: $D$. Indeed, the

839: element $g_0^2$ corresponds to a $\beta$-vector that has zero fermionic

840: entries.  When the order $d$ of the group $\IZ_d$ is even, the subgroup

841: generated by $g_0^2$ has order $n=d/2$ and it has order $n=d$ when $d$ is odd

842: (see also \cite{Greene:1990ud}).  The part of the diagonal symmetry group $D$ that

843: still acts after the projection onto integer R-charges is 

844: \be

845: G=(\prod_{i=1}^p

846: \IZ_{k_i} \times \prod_{j=1}^q \IZ_{l_j})/\IZ_n\,.

847: \ee

848: 

849: Following \cite{Greene:1990ud} we then define the maximal subgroup $H$ of $G$

850: which preserves supersymmetry in space-time. This is the subgroup

851: corresponding to all vectors $\beta_m$ in the charge lattice that satisfy the

852: equation

853: $2

854: \beta_m \cdot \beta_0 \in 2 \IZ$. 

855: This condition ensures that the left- and right- moving R-charges only differ

856: by even integers, as required by supersymmetry.

857: If we write 

858: \be

859: \beta_m = \sum_i c_m^i \gamma_i\,,

860: \ee 

861: then this condition boils down to the condition that \cite{Greene:1990ud}

862: \be

863: \sum \frac{c_m^i}{k_i} + \sum \frac{c_m^j}{l_j} \in \IZ \,.

864: \ee 

865: The subgroup $H$ generated by the vectors $\beta_m$ that satisfy this condition is the

866: maximal orbifold group consistent with space-time supersymmetry.

867: 

868: Any subgroup $F$ of $H$ can be used to generate new supersymmetric orbifold

869: models. Precisely as in the compact case, modding out the original

870: model by the maximal subgroup $H$ generates the mirror model

871: (in the conformal field theory sense). Moreover, orbifolding by a subgroup $F$

872: will generate a model that is mirror to the orbifold of the original model by

873: $H/F$. This was argued in \cite{Greene:1990ud} for the compact case, and we

874: will show in section \ref{CFTmirror} that it is also true for the non-compact

875: models. Note that the mirror symmetry in the

876: conformal field theory that we have set up above holds for {\em

877:   undeformed} models. We will see that deformations of a model are mapped to

878: resolutions of the mirror. We will discuss this important point in greater detail

879: later on.

880: 

881: \subsection{The Deep-throat Region Of Non-compact Gepner Models}

882: \label{local}

883: Until now we have discussed only the continuous part of the spectrum of

884: the non-compact Gepner models. 

885: The contribution of these states to the partition function is proportional to

886: the volume of the target space.

887: 

888: A good starting point for the rest of our discussion will be to think

889: of the initial model as based on a product of $N=2$ superconformal

890: minimal models and $N=2$ linear dilaton theories.

891: 

892: It is important to observe that although the exact torus partition

893: function exists for these conformal field theories, 

894: they only describe the

895: asymptotic spectrum of the corresponding string 

896: theory. Indeed, there is a region in space-time that is strongly

897: coupled (due to the linearly growing dilaton in one or several

898: directions). In that region, the one-loop spectrum is not a meaningful

899: quantity.

900: 

901: Nevertheless, we can get a handle on possible deformations of that

902: singular string theory by working under the following hypotheses. We

903: look for local deformations in the strongly coupled region deep

904: in the throat(s).

905: Secondly, we focus on 

906: marginal deformations of the worldsheet theory that

907: preserve supersymmetry in space-time. The $N=2$ superconformal algebra on the

908: worldsheet will be preserved, and the deformations will be based on chiral (or

909: anti-chiral) primaries. Morever, we suppose that the deformations cannot have

910: $U(1)_R$ quantum numbers that differ from those already appearing in the

911: asymptotic partition function, namely, the charges are quantized as in the

912: asymptotic partition function.  We believe all of these are mild assumptions,

913: given space-time supersymmetry. (See e.g. \cite{Gukov:1999ya}

914: \cite{Shapere:1999xr} for similar reasonings, mostly from a space-time

915: perspective.)

916: 

917: Using the fact that there is a map between $N=2$

918: superconformal algebra representations and reprentations of $SL(2,\IR)$

919: \cite{Dixon:1989cg}, we can 

920: reformulate the above conditions as saying that the marginal operators in the

921: full theory should

922: be based on chiral primaries of dimension smaller or equal to one half in

923: the linear dilaton factors. The operator in the linear dilaton factor

924:  should have a conformal dimension equal to $h=|m|/k$ where the $U(1)_R$

925: charge is given by $Q=2m/k$, and where $2m$ is an allowed $U(1)_R$ quantum

926: number given the (fixed) asymptotic partition function.

927: We get an upper cut-off: $2|m| \le k$ from the requirement of relevance on the

928: worldsheet. 

929: Moreover, we want to study operators that are

930: normalizable at weak coupling.

931: Strict normalisability requires $2 |m|>1$. In particular

932: the value $2 m=0$ is excluded in the non-compact factors, since this would

933: correspond to a non-normalizable operator for a conformal field theory with a

934: non-compact target space.

935: In summary, the quantum number $2|m|$ has to lie in the range

936: $1 < 2|m| \le k$. 

937: It should be noted that the operator with $2|m|=1$ is on the border

938: of being normalizable in the sense that it lies at the endpoint of a line of

939: delta-function normalizable operators \cite{Dixon:1989cg}. It will play a special role in what

940: follows, and we will call this type of operator almost-normalizable. We will flesh out the above analysis considerably in the following sections.

941: 

942: \begin{figure} \centering

943: \includegraphics{defres}

944: \caption{

945: The asymptotic region in a non-compact Gepner  model is identical to that of a linear dilaton conformal field theory. The strong coupling singularity can be cured either by turning on a Liouville potential, or by capping the cylinder with a cigar type deformation. 

946: }

947: \label{defres}

948: \end{figure}

949: 

950: \section{Landau-Ginzburg Models}

951: \label{LG}

952: One of our goals in this section is to obtain, in a simple manner, the chiral ring of localized operators in 

953: the non-compact Gepner models we described in section \ref{noncompactgepner}.

954: For the compact case, this is most carried out by associating a Landau-Ginzburg model to

955: each of the factor conformal field theories. The underlying idea is that at low energies,

956: the Landau-Ginzburg model flows to the conformal field theory that corresponds to the minimal model. 

957: 

958: Furthermore, the GSO projection that we discussed for the non-compact Gepner

959: models maps to an orbifold of the Landau-Ginzburg theory. Therefore the

960: techniques derived in \cite{Vafa:1989xc, Intriligator:1990ua} to obtain the

961: spectrum in the Landau-Ginzburg models and orbifolds thereof will be crucial

962: for our analysis.

963: 

964: In this section and the next, we give a similar Landau-Ginzburg

965: description of our non-compact Gepner models. We will find that much of the

966: technology used in the compact case can be used for the non-compact case

967: as well. However there are subtle differences in reading off the spectrum,

968: because some of the operators are not normalizable.

969: 

970: \subsection{Landau-Ginzburg Potentials For Minimal Models}

971: 

972: For $N=2$ superconformal minimal models, the flow between Landau-Ginzburg and

973: superconformal minimal models is well-studied

974: \cite{Martinec:1988zu, Vafa:1988uu} and we merely give a

975:  brief

976: reminder. A Landau-Ginzburg model with $N=(2,2)$ supersymmetry and a chiral

977: superfield $\Phi$ with superpotential \be\label{simpleLG} W_{MM} = \Phi^k \ee

978: flows in the infrared to an $N=2$ superconformal minimal model. The chiral

979: ring of both models match one-to-one. The chiral unital ring of the Landau-Ginzburg

980: model is

981:  $\, \, \IC[\Phi]/\partial_\Phi W$ which is linearly generated by the $k-1$

982: elements $\Phi^0, \Phi^1, \dots, \Phi^{k-2}$, which have (both left- and

983: right-moving) R-charge equal to $0,\frac{1}{k}, \dots , \frac{k-2}{k}$. These

984: match one-to-one to the chiral-chiral primaries of the diagonal minimal model.

985: Further evidence for this identification of the Landau-Ginzburg model fixed

986: point is provided by the matching of the elliptic genus of these models

987: \cite{Witten:1993jg}.

988: 

989: The T-dual or mirror Landau-Ginzburg model based on an twisted chiral

990: superfield \cite{Rocek:1991ps} flows to the anti-diagonal minimal model. This

991: can be written as a

992: $\IZ_k$ orbifold of the model in 

993: equation \eqref{simpleLG}. We will discuss such orbifolding methods to compute mirrors in later sections.

994: 

995: We can summarize the chiral ring of the Landau-Ginzburg model by specifying a

996: Poincare polynomial 

997: which is the trace over the chiral-chiral ring weighted

998: by the $U(1)$ R-charges \cite{Lerche:1989uy}:

999: \begin{eqnarray}

1000: Tr_{(c,c)} t^{Q} \tilde{t}^{\tilde{Q}} &=& 1+ (t \tilde{t})^{\frac{1}{k}} + (t \tilde{t})^{\frac{2}{k}}

1001: + \dots +  (t \tilde{t})^{\frac{k-2}{k}}

1002: \nonumber \\

1003: & =& \frac{ 1- (t \tilde{t})^{\frac{k-1}{k}} }{ 1 - (t \tilde{t})^{\frac{1}{k}}}.

1004: \end{eqnarray}

1005: In the Ramond sector, this gives rise to a polynomial that keeps track of the

1006: R-charges

1007: of the Ramond-Ramond ground states:

1008: \begin{eqnarray}

1009: Tr_{RR} t^{Q} \tilde{t}^{\tilde{Q}} &=&(t \tilde{t})^{-\frac{1}{2}+\frac{1}{k}} + (t

1010: \tilde{t})^{-\frac{1}{2} +\frac{2}{k}}+ \dots + (t \tilde{t})^{+\frac{1}{2}-\frac{1}{k}}.

1011: \end{eqnarray}

1012: Note that the charges fill out the range from $-c/6$ to $+c/6$, and lie

1013: inside

1014: the interval $]-\half,+\half[$. See figure \ref{Rgroundstatescomp}. 

1015: \begin{figure}

1016: \centering

1017: \includegraphics{Rgroundstatescomp}

1018: \caption{The R-charges of the ground states in the Ramond sector

1019: for a $N=2$ superconformal theory with central charge $c<3$.\label{Rgroundstatescomp}}

1020: \end{figure}

1021: 

1022: \subsection{Non-compact Landau-Ginzburg Models}

1023: 

1024: Now we want to discuss the link between the non-compact $N=2$ superconformal field

1025:  theories and the IR fixed point of Landau-Ginzburg models with a superpotential of the form:

1026: \begin{eqnarray}

1027: W_{nc} &=& \Phi^{-l}

1028: \end{eqnarray}

1029: where $l$ is a positive integer. This was introduced\footnote{See also e.g.

1030:   \cite{Ghoshal:1993qt, Hanany:1994fi} for an older and related use of

1031:   these models in the context of two-dimensional gravity.} in

1032: \cite{Ooguri:1995wj} and we would like to understand how far we can argue for

1033: a formal analogy with the compact case, and if possible borrow the techniques

1034: that have been extensively used in that 

1035: context. See also

1036:   \cite{Hori:2000kt, Hori:2002cd}

1037: for a more detailed analysis of the renormalization group flow with fixed asymptotics. 

1038: The central charge of this theory with fixed asympotics 

1039: is $c= 3+ \frac{6}{l}$. %%

1040: 

1041: It is natural to assume that the field $\Phi$ cannot take the value zero, and

1042: that moreover the point at infinity should remain a regular point. We can then

1043: associate to the model an operator ring $\, \IC [\Phi^{-1}]$ which should

1044: again be divided by the ideal generated by the derivative of the

1045: superpotential. This gives rise to a ring spanned by the $l+1$ elements

1046: $\Phi^{0}, \Phi^{-1}, \dots, \Phi^{-l}$.  Because the target-space is

1047: non-compact, $\Phi^{0}$ is not normalizable. We exclude the operator from the

1048: ring.  The ring of elements spanned by $\Phi^{-1}, \Phi^{-2}, \dots,

1049: \Phi^{-l}$ is a ring without unit element. This fact contrasts with the

1050: compact case, and it is associated to the non-existence of an $SL(2,R)$

1051: invariant ground state in the conformal field theory. The operators

1052: $\Phi^{-1}, \Phi^{-2}, \dots, \Phi^{-l}$ have R-charge

1053: $\frac{1}{l},\frac{2}{l},\dots,\frac{l}{l}$.  We can match the operators

1054: $\Phi^{-2}, \dots, \Phi^{-l}$ onto the chiral ring of the diagonal linear

1055: dilaton theory with $N=2$ superconformal symmetry (under the assumptions of

1056: relevance and normalizability, as discussed in section \ref{local}).  We

1057: moreover expect the operator $\Phi^{-1}$ to become the almost-normalizable

1058: chiral-chiral primary in the infra-red fixed point theory.

1059: 

1060: Again, we can summarize the chiral-chiral spectrum in a Poincar\'e polynomial

1061:  for the $(c,c)$ ring which for the case of

1062: (almost-)normalizable elements is:

1063: \begin{eqnarray}

1064: Tr_{(c,c)} t^{Q} \tilde{t}^{\tilde{Q}} &=& (t \tilde{t})^{\frac{1}{l}} + (t \tilde{t})^{\frac{2}{l}}

1065: + \dots +  (t \tilde{t})^{\frac{l}{l}}

1066: \nonumber \\

1067: & =&  (t \tilde{t})^{\frac{1}{l}}  \frac{ 1- (t \tilde{t}) }{ 1 - (t \tilde{t})^{\frac{1}{l}}}.

1068: \end{eqnarray}

1069: For the strictly normalizable elements, we should use:

1070: \begin{eqnarray}

1071: Tr_{(c,c)} t^{Q} \tilde{t}^{\tilde{Q}} &=& (t \tilde{t})^{\frac{2}{l}} + (t \tilde{t})^{\frac{3}{l}}

1072: + \dots +  (t \tilde{t})^{\frac{l}{l}}

1073: \nonumber \\

1074: &=&  (t \tilde{t})^{\frac{2}{l}}  \frac{ 1- (t \tilde{t})^{\frac{l-1}{l}} }{ 1 - (t \tilde{t})^{\frac{1}{l}}}.

1075: \end{eqnarray}

1076: Moreover, the normalizable Ramond ground states now have charges that go from $-1/2+1/l$ to

1077: $+1/2-1/l$ which again lie inside the interval $]-\half, +\half[$. The almost-normalizable ground state is an outlier at charge $-1/2$. It has a spectrally flowed partner at opposite charge $+1/2$. The Ramond ground states do not reach the charge $-c/6$ and $+c/6$. See figure

1078: \ref{Rgroundstatesnoncomp}.

1079: \begin{figure}

1080: \centering

1081: \includegraphics{Rgroundstatesnoncomp}

1082: \caption{The R-charges of the ground states in the Ramond sector

1083: for a $N=2$ superconformal theory with central charge $c>3$.

1084: There are almost-normalizable ground states at charges $\pm \frac{1}{2}$.}

1085: \label{Rgroundstatesnoncomp}

1086: \end{figure}

1087: The R-charges of the strictly normalizable 

1088: Ramond-Ramond ground states can be coded in the

1089: polynomial:

1090: \begin{eqnarray}

1091: Tr_{RR} t^{Q} \tilde{t}^{\tilde{Q}} &=&(t \tilde{t})^{-\frac{1}{2}+\frac{1}{l}} + (t

1092: \tilde{t})^{-\frac{1}{2} +\frac{2}{l}}+ \dots + (t \tilde{t})^{+\frac{1}{2}-\frac{1}{l}}.

1093: \end{eqnarray}

1094: 

1095: To summarize, we can associate a Landau-Ginzburg model to each factor of a

1096: non-compact Gepner model. The

1097: total superpotential will be the sum of the individual superpotentials. The

1098: Landau-Ginzburg models we discuss will be of the Fermat type. However, we note

1099: at this stage that there is a difference between the product

1100: of minimal and linear dilaton theories and the corresponding product of

1101: Landau-Ginzburg theories. Indeed, while the diagonal A-type minimal models are

1102: identified as infra-red fixed points of Landau-Ginzburg models, the

1103: non-compact

1104: Landau-Ginzburg model gives rise to a deformation of the

1105: $N=2$ linear dilaton theory.

1106: To the linear dilaton asymptotics, we add a deforming 

1107: potential.

1108: 

1109: So far we have studied simple Landau-Ginzburg theories (and their direct

1110: product theories). However, to provide Landau-Ginzburg analogues of the Gepner

1111: conformal field theories, we need to discuss orbifolded Landau-Ginzburg

1112: models.

1113: 

1114: \section{Landau-Ginzburg Orbifolds}

1115: \label{orbifoldedLG}

1116: 

1117: Our discussion of orbifolded Landau-Ginzburg models is mainly based on

1118: \cite{Vafa:1989xc, Intriligator:1990ua}. The orbifolding can arise due to the

1119: GSO projection in string theory, or due to a further geometric orbifolding of

1120: the resulting theory. In this section, we start out by discussing the orbifold

1121: action as being independent of possible actions on flat space factors,

1122: following \cite{Intriligator:1990ua}. We will comment on the relation to the

1123: full GSO projection (which also acts on the flat space factors) 

1124: in the appendix. Since

1125: our discussion will be closely related to the long discussion in the

1126: literature of the compact case, we will briefly review that discussion and we

1127: will only treat in more detail the crucial differences that exist in the

1128: non-compact case.

1129: 

1130: For each compact Landau-Ginzburg model with superpotential $\Phi_i^{k_i}$,

1131: there is a canonical diagonal action \be \Phi_i \rightarrow e^{ \frac{2 \pi i

1132:   }{k_i} } \Phi_i \,, \ee which generates a $\IZ_{k_i}$ group. The exponent is

1133: determined by the $U(1)_R$ charge of the field $\Phi_i$. For every non-compact

1134: factor with superpotential $\Phi_j^{-l_j}$, the canonical action is \be \Phi_j

1135: \rightarrow e^{- \frac{2 \pi i}{l_j}} \Phi_j\,.  \ee We act with the opposite

1136: phase, since the $U(1)_R$ charge of the field $\Phi_j$ is negative (becuase of

1137: the negative power in the superpotential). We introduce the charges $q_i$ for

1138: the fields $\Phi_i$ for both compact and non-compact factors, 

1139: which are equal

1140: to \be \left(\frac{1}{k_1}, \frac{1}{k_2}, \dots ; -\frac{1}{l_1},

1141:   -\frac{1}{l_2}, \dots\right)\,.  \ee Then the above actions on the fields

1142: can be written as \be \Phi_i \rightarrow e^{2 \pi i q_i} \Phi_i\,.  \ee The full

1143: symmetry group is then \be D=\prod_i \IZ_{k_i} \times \prod_j \IZ_{l_j}\,.

1144: \ee Note that it is identical to the group we identified previously in the

1145: context of non-compact Gepner models in section \ref{symGroup}.

1146: 

1147: We study orbifolds of the theory by a subgroup of the above diagonal group $D$.  For instance, we can choose to orbifold by a group generated by a single element, which has different weights in

1148: each of the factors, or by a product of such groups. For an element $h$ of the orbifold group, the action on each superfield can be written as 

1149: \be

1150: \Phi_i \rightarrow e^{2 \pi i \Theta_i^h} \Phi_i\,.

1151: \ee 

1152: The phases $\Theta_i^h$ parameterize the group elements $h$.

1153: 

1154: The integer R-charge orbifold discussed in section \ref{integerR} is special,

1155: since it is part of the GSO projection. In this case, we orbifold by the

1156: group generated by $g_{0}$. The action of this operator on the superfields is coded as

1157: $\Theta_i^{g_{0}}=q_i$.

1158: 

1159: Our main objective is to compute the number and charges of the

1160: chiral/chiral $(c,c)$ and anti-chiral/chiral $(a,c)$ states, as well as their

1161: $(a,a)$ and $(c,a)$ partners, in the orbifolded Landau-Ginzburg theory.

1162: In particular, those with

1163: $U(1)_R$ charges $\pm 1$ will lead to massless fields in

1164: spacetime. In the case where the geometric approximation is valid, marginal

1165: $(c,c)$ states correspond to complex structure deformations, and marginal

1166: $(a,c)$ states to K\"ahler moduli.

1167: 

1168: A crucial observation is that for the Calabi-Yau examples we consider in this paper, $(c,c)$ states are in one-to-one correspondence with Ramond-Ramond ground states \cite{Lerche:1989uy}. More precisely, a Ramond-Ramond ground state becomes a $(c,c)$ state under the action of half-unit left-right symmetric spectral flow. Similarly, $(a,c)$ states are derived from Ramond-Ramond ground

1169: states by half-unit left-right antisymmetric spectral flow. It will prove a good strategy to work in the R-R sector rather than in the NS-NS sector. The procedure we will use is the following 

1170: \cite{Intriligator:1990ua}:

1171: \begin{itemize}

1172: \item[\_]First identify the unprojected Ramond-Ramond ground states in each twisted sector and compute their R-charge.

1173: \item[\_]Then flow these states to the NS-NS sector, and keep those that are

1174:   invariant under the orbifold action.

1175: \end{itemize}

1176: Let us study the two steps of this procedure in greater detail.

1177: 

1178: \subsection{R-charges of Ramond-Ramond ground states}

1179: We first extend a result of reference \cite{Vafa:1989xc} for the twisted

1180: sector Ramond-Ramond ground states. Consider the sector of the theory twisted

1181: by $h$ (the following is also valid in the untwisted sector, with $h=1$).

1182: We define a particular Ramond-Ramond ground state $|0 \rangle^h_R$, which

1183: is the Ramond-Ramond ground state with the lowest left-moving R-charge for the

1184: compact factors, and the highest left-moving R-charge for the non-compact

1185: factors. The left-moving R-charges with respect to the individual factors of

1186: that

1187: Ramond-Ramond ground state is:

1188:  \begin{align}\label{twistedRchargeleft}

1189: Q &= +\sum_{\Theta_i^h \notin Z} \left( \Theta_i^h - {[}  \Theta_i^h {]} -

1190: \frac{1}{2} \right) \cr

1191: & \qquad+ \sum_{\Theta_i^h \in Z} (q_i - \frac{1}{2})\,,

1192: \end{align} 

1193: and the right-moving R-charge is:

1194:  \begin{align}\label{twistedRchargeright}

1195: \tilde{Q} &= -\sum_{\Theta_i^h \notin Z} \left(  \Theta_i^h - {[}  \Theta_i^h {]} -

1196: \frac{1}{2}\right)\cr

1197: & \qquad+ \sum_{\Theta_i^h \in Z} (q_i - \frac{1}{2})\,.

1198: \end{align} 

1199: Here, the expression $[\Theta]$ is defined as  the

1200: greatest integer smaller than $\Theta$ (for $\Theta$ not an integer). Remember

1201: that $\Theta_i^h$ is the phase that defines the action of $h$ on $\Phi_i$, and

1202: $q_i$ is the R-charge of $\Phi_i$.

1203: 

1204: The arguments leading to the first line of these formulae are elaborate

1205: \cite{Vafa:1989xc}. We refer to that reference for details. Note that it is

1206: natural to look for the Ramond-Ramond ground state in the $h$-twisted sector

1207: by twisting the left-movers half-way in one direction, and twisting the

1208: right-movers half-way in the other direction, due to the symmetry between

1209: left- and right-movers in the original theory. The subtraction of the integer

1210: part of the twist arises because of the fact that we can otherwise find a

1211: state with smaller R-charge, between $-1/2$ and $+1/2$ that will have lower

1212: conformal dimension.  See also figures \ref{Rgroundstatescomp} and

1213: \ref{Rgroundstatesnoncomp}. Furthermore, as argued in \cite{Vafa:1989xc}, the

1214: R-charges behave very much like fermion number, which is indeed lifted from

1215: $-1/2$ in the compact sector, by the contribution of $\Theta_i^h$. The crucial

1216: difference with the compact sector is only that for the non-compact sector we

1217: need to keep in mind that $\Theta_i^h$ is negative, and that therefore it is

1218: more natural to think of the state with fermion number or $U(1)_R$ charge

1219: moving {\em down} to a charge just below $+1/2$. That is precisely what is

1220: automatically coded in the above formula, since for $\Theta$ negative but

1221: bigger than $-1$, we have that $\Theta-[\Theta]-1/2=+1/2+\Theta$.

1222: 

1223: The second line in each of these formulae \eqref{twistedRchargeleft} and \eqref{twistedRchargeright} arises from spectral flowing the untwisted factor NS-NS ground state to the R-R sector. Note that the

1224: superfield living in an untwisted factor can be given a nonzero vev. This is

1225: not possible in the twisted factor, since a constant does not satisfy the

1226: twisted boundary conditions. In this 

1227: way, other Ramond-Ramond ground states can be generated in the same

1228: sector. For example, if the $i$-th factor is untwisted, the state

1229: $\Phi_i^{p}|0 \rangle_{R}^{h}$ 

1230: ($p$ integer) is another valid Ramond-Ramond ground state. It has left- and

1231: right-moving R-charge $(Q+p|q_i|,\tilde{Q}+p|q_i|)$, where we took into

1232: account the contribution of $\Phi_i^{p}$. In a compact factor, $p$ is

1233: restricted to the bounds $0 \le p \le k_i-2$. In a non-compact factor,

1234: strictly normalizable states have $2 \le -p \le l_i$, and almost-normalizable

1235: states have $p=-1$. These bounds follow from our discussion in section \ref{LG}.

1236: 

1237: The upshot is that the formulas \ref{twistedRchargeleft} and \ref{twistedRchargeright},

1238: derived in \cite{Vafa:1989xc} for compact models, are also valid for non-compact models.

1239: 

1240: Once this  hurdle is ovecome, one can extend the reasoning of \cite{Intriligator:1990ua} to

1241: cover the non-compact case as well. We will not repeat the whole analysis

1242: here. Let us observe that all the information about the unprojected

1243: Ramond-Ramond ground states in the $h$-twisted sector is conveniently encoded

1244: in a Poincare polynomial equal to:

1245: \begin{align}

1246: \Tr_{R, \text{twisted, unprojected}} t^{Q} \tilde{t}^{\tilde{Q}} & = 

1247: \left(\frac{t}{\tilde{t}} \right)^{ \sum_{\Theta_i^h \notin Z} ( \Theta_i^h  -  [  \Theta_i^h  ] - 1/2)}

1248: \left(t \tilde{t} \right)^{\sum_{\Theta_i^h \in Z} (q_i-1/2) }  \nonumber \\

1249: & \times \prod_{\begin{array}{c} \Theta_i^h \in Z \\ \scriptstyle{compact}\\ \scriptstyle{factors}

1250:     \end{array}}

1251: \frac{1-(t\tilde{t})^{\frac{k_i-1}{k_i}}}{1-(t\tilde{t})^{\frac{1}{k_i}}}

1252: \prod_{\begin{array}{c} \Theta_i^h \in Z \\ \scriptstyle{noncompact}\\ \scriptstyle{factors}

1253:     \end{array}}

1254: (t\tilde{t})^{\frac{2}{l_i}}\frac{1-(t\tilde{t})^{\frac{l_i-1}{l_i}}}{1-(t\tilde{t})^{\frac{1}{l_i}}}

1255: .

1256: \end{align}

1257: 

1258: In the untwisted sector, the above analysis simplifies. The particular

1259: Ramond-Ramond ground state we discussed earlier had charges $(-c/6,-c/6)$. It

1260: is the state one gets by half-unit spectral flow of the NS-NS vacuum. All the

1261: superfields are untwisted, and can be given 

1262: constant values. The relevant Poincar\'e

1263: polynomial is simply the product of the Poincar\'e polynomials of each

1264: factors.

1265: 

1266: \subsection{Spectral flow to the NS-NS sector}

1267: In this way we can enumerate all the Ramond-Ramond ground states,

1268: and their R-charges. Then we can generate all the NS-NS (anti-)chiral states,

1269: by spectral flow.

1270: 

1271: We obtain $(c,c)$ states by symmetric half-unit spectral flow. Their R-charges are easily computed from those of the Ramond-Ramond ground state: we add $(\frac{c}{6},\frac{c}{6})$ to the R-charges of the ground state. This operation does not change the twist: $h$-twisted Ramond-Ramond ground states flow to $h$-twisted $(c,c)$ states.

1272: 

1273: In a similar way, we get $(a,c)$ states by anti-symmetric half-unit spectral

1274: flow. This time, the Ramond-Ramond ground state R-charges are shifted by 

1275: $(-\frac{c}{6},\frac{c}{6})$. Note that anti-symmetric spectral flow changes the

1276: twist \cite{Vafa:1989xc} such that $h$-twisted Ramond-Ramond ground states flow to 

1277: $h g_{0}$-twisted $(a,c)$ states.

1278: 

1279: Among all these states, we keep only those that are invariant under the orbifold

1280: action. In the case of the integer R-charge orbifold, the selection is easy:

1281:  we keep the states that have integer R-charges. But in other cases, 

1282: the procedure is harder. We refer to \cite{Intriligator:1990ua} for a generic discussion of

1283:  this projection.

1284: 

1285: \subsection{Normalizability of non-compact twisted states}

1286: 

1287: For the twisted Ramond-Ramond ground states, we encounter a subtle point and

1288: this is the crucial difference between the compact and the non-compact case.

1289: When we twist a coordinate, and restrict to constant modes, the field is set

1290: to zero. This will not give rise to a normalizable state when the

1291: Landau-Ginzburg potential has a negative power, since the potential blows up

1292: at zero.  Although we could therefore immediately discard the states that have

1293: a non-compact twisted factor as not being contained in the normalizable

1294: spectrum, we advocate keeping an open mind -- we will continue the

1295: investigation into these non-normalizable states, in view of the possibility

1296: that we can tune the coefficient of the negative power potential to zero,

1297: which will render these states normalizable.

1298: 

1299: Strictly speaking, the models where we do take into account such twisted

1300: sector deformations should be thought of as existing in the well-defined

1301: linear dilaton conformal field theories (which give rise to locally strongly

1302: coupled string theory backgrounds). We believe we provide ample justification

1303: for this procedure later on. In particular in section \ref{CFTmirror} we will

1304: find that these states are needed in the construction of mirror pairs of

1305: conformal field theories, by generalizing the Greene-Plesser prescription of

1306: finding mirrors by orbifolding. More precisely, states untwisted in the

1307: non-compact directions will be mapped in the mirror model to states twisted in

1308: the non-compact directions.

1309: 

1310: To summarize, the full set of chiral and anti-chiral ring elements of the

1311: Landau-Ginzburg orbifold, including those arising in both the untwisted and

1312: twisted sectors, is interpreted as the spectrum of the linear dilaton theory.

1313: The Landau-Ginzburg model is a complex structure deformation of this linear

1314: dilaton background. This deformation removes from the spectrum the RR states

1315: twisted in the non-compact directions and the de-singularized string

1316: background only contains purely untwisted states. As we will see later on, in

1317: the mirror model, the deformation is mapped to a resolution of the theory. The

1318: spectrum of the resolved theory will, instead, only contain the states twisted

1319: in the non-compact directions.

1320: 

1321: \section{Concrete Models}

1322: \label{examples}

1323: In this section we will apply what we have learned in the previous sections to

1324: some concrete models. We  supplement the analysis of the

1325: asymptotic partition function of section \ref{asymptotic}

1326: with an analysis of the localized states (which we briefly touched upon 

1327: in section \ref{local}). We start out with some

1328: observations on the nature of non-compact Gepner models, and how they differ

1329: from their compact counterparts. 

1330: We will then compute the allowed deformations of $(c,c)$ and $(a,c)$ type in

1331: particular examples, in the language of orbifolded Landau-Ginzburg models. In the conformal field theory language, the conditions to be satisfied for chiral primaries have been written out in a series of papers \cite{Eguchi:2003ik, Eguchi:2004yi, Eguchi:2004ik}. We compare our results with the conformal field theory formalism in appendix \ref{CFTanalysis}.

1332: 

1333: 

1334: 

1335: \subsection{Gepner Points At Large Level}

1336: 

1337: We start out with some comments that allow us to single out some particularly

1338: interesting models.

1339: Compact Gepner models have factors with a central charge which is always

1340: smaller than three. A large level limit for all factors

1341:  necessarily increases the central charge of the Gepner model, and

1342: therefore is not consistent with the criticality condition for string

1343: theory. Thus, compact Gepner models are necessarily at large curvature (and small

1344: volume).

1345: 

1346: Non-compact Gepner models are of a different type. In particular we can have

1347: non-compact Gepner models that contain factor conformal field theories at

1348: central charge smaller {\it and} larger than three. We can therefore cancel

1349: off the difference in a large level limit, if we wish.  Assuming that we

1350: demand the existence of a limit in which all levels are large, we immediately

1351: conclude that such a non-compact Gepner model necessarily has three

1352: factors (since the total central charge is $c=9=3\times 3$).

1353: 

1354: There are two such classes of models (if we do not admit models with central

1355: charge

1356: precisely equal to three). One is where we have two minimal

1357: model factors and one non-compact factor, and the other has one

1358: minimal model factor and two non-compact factors. There are no other

1359: possibilities at central charge $c=9$. (When we consider a non-compact

1360: Gepner model at central charge $c=3.D$, with $D$ an integer different from

1361: $D=3$,

1362: there are other possibilities which can also easily be classified.)

1363: 

1364: The class of models with two minimal models can be parameterized by the 

1365: integer levels $k_1$ and $k_2$ of the minimal models, up to

1366: an initial choice of ADE modular invariant and a possible orbifold by a

1367: symmetry group.

1368: The level $l$ of the non-compact factor is then fixed to be $l

1369: = \frac{k_1 k_2}{k_1 + k_2}$ which can be integer or fractional.

1370: 

1371: 

1372: The other class of models is parameterized by two levels $l_1$ and

1373: $l_2$ for the non-compact models, and the combination $k= \frac{l_1

1374: l_2}{l_1+l_2}$ then needs to be integer, in order for the compact

1375: model at level $k$ to exist. The levels $l_i$ can a priori be

1376: fractional or real. As we have indicated before, we

1377: concentrate on the case where all levels are integer.

1378: 

1379: 

1380: \subsection*{Remarks}

1381: \begin{itemize}

1382: \item We note that a general class of models can be found by demanding

1383:   $\frac{k_1 k_2}{k_1 + k_2}$ to be integer (with $k_{1}$ and $k_2$ positive

1384: integers). Suppose we isolate the 

1385: greatest common divisor $d$  of $k_1=d \tilde{k}_1$ and $k_2= d \tilde{k}_2$

1386: (with $\tilde{k}_1$ and $\tilde{k}_2$ mutually prime). Then it can be shown

1387: that the level $ \frac{k_1 k_2}{k_1 + k_2}$ is integer if and only if $d$ is a

1388:   divisor of $\tilde{k}_1+ \tilde{k}_2$. That easily generates a large class of

1389:   models of which we will only study a few.

1390: \item We note that although the local curvature would seem to become small in

1391:   the large level limit, this reasoning does not take into account the GSO

1392:   orbifold that still needs to be performed. In specific examples it can be

1393:   checked that the GSO orbifold recreates small radii in the geometry (see

1394:   e.g. \cite{Israel:2004ir}\cite{Israel:2004jt} for a detailed discussion).

1395:   We can therefore generically expect the large level limit to correspond to

1396:   an {\em orbifolded}  weakly curved background.

1397: \end{itemize}

1398: 

1399: In the following we will mainly concentrate on the set of models that have

1400: the special property of allowing for a large level limit (although our

1401: formalism does apply more widely).

1402: We study in detail the set of models with three factors and levels $(2k,2k;k)$

1403: or $(k;2k,2k)$ for the minimal model and non-compact factors respectively, and

1404: orbifolds thereof. 

1405: As a warm-up exercise however, we treat the instructive

1406: example at complex dimension $D=2$ with the compact model at level $k$ and the

1407: non-compact model at the same 

1408: level $k$.

1409: 

1410: \subsection{The $(k; k)$ Model}

1411: 

1412: Let us apply the formalism for orbifolded Landau-Ginzburg models of section

1413: \ref{orbifoldedLG} to the case of two factors, one compact and one non-compact

1414: at equal levels $k$. We refer to the model as the $(k;k)$ model. 

1415: 

1416: The Landau-Ginzburg model has a potential 

1417: \be

1418: W_{LG}=\Phi_1^k + \Phi_2^{-k}

1419: \ee

1420: for two chiral superfields $\Phi_{1}$ and $\Phi_2$. The orbifold group that is

1421: necessary to implement GSO is generated by 

1422: \be

1423: g_{0}: (\Phi_1,\Phi_2) \rightarrow (e^{2 \pi i /k} \Phi_1,e^{- 2 \pi i /k} \Phi_2)\,. 

1424: \ee 

1425: We will be mostly interested in obtaining the numbers and types of

1426: deformations of the conformal field theory. In order to do this, one computes the left and right

1427: R-charges of the Ramond-Ramond ground states in all the twisted sectors using

1428: equations \eqref{twistedRchargeleft} and \eqref{twistedRchargeright}. Then, after

1429: spectral flow, one can compute the R-charges of the operators in the $(c,c)$

1430: and $(a,c)$ rings. For this $c=6$ theory, half-unit spectral flow amounts to adding $\pm

1431: 1$ unit of R-charge to the RR states.

1432: 

1433: We recall that for the $(a,c)$ states, asymmetric spectral flow from the RR

1434: sector adds a twist by $g_{0}$.

1435: We tabulate the R-charges of the relevant states below, and indicate with a

1436: star the sectors in which 

1437: non-zero constant modes can be given to the fields. We label by $\alpha$ the sector

1438: twisted by $g_{0}^{\alpha}$.

1439: \be

1440: \begin{array}{c|c|c|c}

1441: \alpha & RR & (c,c) & (a,c) \cr

1442: \hline

1443: 0 & (-1,-1)* & (0,0)* & (-1,+1) \cr

1444: 1 & (0,0) & (+1,+1) & (-2,0)* \cr

1445: 2 \le \alpha \le k-1 & (0,0) & (+1,+1)& (-1,+1) \cr

1446: \end{array}

1447: \ee

1448: We summarize the results:

1449: \begin{itemize}

1450: \item In the untwisted sector, the fields can have non-zero

1451:   constant modes. The Ramond-Ramond ground state flows in the $(c,c)$ ring to the

1452:   identity operator, with charges $(0,0)$. We moreover find $k-1$ marginal

1453:   $(c,c)$ states in this sector. These

1454:   arise from the monomials $\Phi_1^n \Phi_2^{-k+n}$ for $n=0,1, \dots, k-2$.

1455:   They can be checked to be invariant under the orbifold projection.

1456: \item Asymmetric spectral flow from the RR untwisted sector gives

1457:   $g_{0}$-twisted $(a,c)$ states. Therefore we get $k-1$ marginal $(a,c)$

1458:   states in this sector. The fact that these states are marginal is a

1459:   consequence of the special value of the central charge, $c=6$.

1460: \item For any value of the twist $\alpha=1,\dots,k-1$, we have no untwisted

1461:   fields

1462: in the RR sector. 

1463: Namely the fields $\Phi_1$ and $\Phi_2$ always have twisted boundary

1464: conditions in any twisted sector. Using the formula (\ref{twistedRchargeleft}), we find that the R-charges

1465:   of the $\alpha$-twisted sector Ramond ground states is $(\alpha/k -0 -1/2) +

1466:   (-\alpha/k+1-1/2) = 0$ on the left and $0$ on the right. We already see a

1467:   phenomenon typical to our non-compact Gepner models. The twist contribution

1468:   to the Ramond sector charges can cancel between compact and non-compact

1469:   factors. We are in a special case, in which the twisted R-charges of all

1470:   ground states  are zero.

1471: \item After flowing symmetrically to the NS-NS sector, we find a single $(c,c)$ state with charges

1472: $(+1,+1)$ in each twisted sector. They give $k-1$ marginal $(c,c)$ states in total in the twisted

1473: sectors.

1474: \item  Asymmetric spectral flow of the twisted RR ground states leads to $k-1$

1475:    $(a,c)$ deformations, one in each sector (except for $\alpha=1$).

1476: \end{itemize}

1477: 

1478: Therefore we have a total of $k-1$ marginal $(c,c)$ and $k-1$ marginal $(a,c)$

1479:  states from the spectral flow of the untwisted RR sector ground states. In

1480:  the Landau-Ginzburg model with potential $W_{LG}=\Phi_1^k+\Phi_2^{-k}$ these

1481:  are the only admissible localized modes. The potential term $\Phi_2^{-k}$

1482:  makes sure that the untwisted polynomials are allowed in the sense that they

1483:  are normalizable at weak coupling, and have a mild behaviour at the $\Phi_2

1484:  \approx 0$ end compared to the potential. These give rise to a $4(k-1)$

1485:  real-dimensional moduli space of backgrounds in string theory with sixteen

1486:  supercharges. 

1487: In contrast, the twisted operators are not normalizable in the

1488:  Liouville deformed model.

1489: 

1490: 

1491: \subsection{The $(2k,2k;k)$ Model.}\label{twoMM}

1492: 

1493: For this slightly more complicated example, we list the full set of

1494: (unprojected)

1495: $(c,c)$

1496: and $(a,c)$ states and their charges, and then pick out those that are marginal (and invariant

1497: with respect to the orbifold projection). Again, we  perform this

1498: exercise in the formalism of section \ref{orbifoldedLG} for orbifolded

1499: Landau-Ginzburg models.

1500: We consider a Landau-Ginzburg model with fields $\Phi_{1,2,3}$ and

1501: superpotential 

1502: \be

1503: W_{LG} = \Phi_1^{2k} + \Phi_2^{2k} + \Phi_3^{-k}\,.

1504: \ee 

1505: We perform the integer R-charge orbifold, generated by the $g_{0}$:

1506: \be

1507: g_{0}: (\Phi_1,\Phi_2,\Phi_3) \rightarrow (e^{2 \pi i /2k} \Phi_1, e^{2 \pi i

1508:   /2k} \Phi_2, e^{- 2 \pi i/k} \Phi_3) \,. 

1509: \ee

1510: In order to consider all the marginal operators, we will write down the

1511: relevant Poincare polynomials in each sector. First of all, in the untwisted

1512: sector we have the (unprojected, strictly

1513: normalizable) Poincare polynomial: 

1514: \begin{eqnarray}

1515: \left( \frac{1-(t \tilde{t})^{(2k-1)/2k}}{1-(t

1516:     \tilde{t})^{\frac{1}{2k}}}\right)^2  (t \tilde{t})^{\frac{2}{k}}

1517: \frac{1-(t \tilde{t})^{\frac{k-1}{k}}}{1-(t \tilde{t})^{\frac{1}{k}}}

1518: \end{eqnarray}

1519: which contains $(c,c)$ states only.

1520: When we label the twisted sectors by $\alpha=1,2, \dots, 2k-1$, we find that

1521: for $\alpha$ smaller than $k$ there are further twisted $(c,c)$ states in the

1522: NSNS 

1523: sector determined by the polynomials:

1524: \be

1525: \left( \frac{t}{\tilde{t}} \right)^{-1/2} (t \tilde{t})^{3/2}

1526: \ee

1527: and when $\alpha$ is larger than $k$ by the polynomial

1528: \be

1529: \left( \frac{t}{\tilde{t}} \right)^{+1/2} (t \tilde{t})^{3/2},

1530: \ee

1531: while at $\alpha=k$ we find the polynomial

1532: \be

1533: (t\tilde{t})^{-1/k-1/2}  (t \tilde{t})^{+3/2} (t \tilde{t})^{\frac{2}{k}}

1534:  \frac{(1-(t \tilde{t}))^{\frac{k-1}{k}}}{(1-(t \tilde{t}))^{\frac{1}{k}}}.

1535: \ee

1536: where the last factor is due to the fact that $\Phi_3$ is untwisted in this sector.

1537: The twisted $(a,c)$ states are determined by the polynomials:

1538: \be

1539: \left( \frac{t}{\tilde{t}} \right)^{-2} (t \tilde{t})^0 .1

1540: \ee

1541: when $\alpha$ is smaller than $k$  and

1542: \be

1543: \left( \frac{t}{\tilde{t}} \right)^{-1} (t \tilde{t})^0 .1

1544: \ee

1545: when $\alpha$ is larger than $k$.

1546: When $\alpha=k$, we get 

1547: \be

1548: (t\tilde{t})^{-1/k-1/2}  \left(\frac{t}{ \tilde{t}}\right)^{-3/2} (t \tilde{t})^{\frac{2}{k}}

1549:  \frac{(1-(t \tilde{t}))^{\frac{k-1}{k}}}{(1-(t \tilde{t}))^{\frac{1}{k}}}.

1550: \ee

1551: One can straightforwardly determine amongst these the states that are

1552: invariant under the orbifold action: they have integer R-charges.

1553: 

1554: \subsection*{Marginal Deformations}

1555: 

1556: We now want to look for exactly marginal deformations in these rings. These

1557: need to have left- and right R-charge equal to $\pm 1$. As for the $(k;k)$

1558: example, let us tabulate the R-charges of the ground states and their images

1559: under spectral flow. Once again, $\alpha$ labels the twist. A star indicates

1560: that some fields are untwisted and can be given nonzero vev's.

1561: \be

1562: \begin{array}{c|c|c|c}

1563:  \alpha& RR & (c,c) & (a,c) \cr

1564:  \hline

1565:  0 & (-3/2,-3/2)* & (0,0)* & (-1,1) \cr

1566:  1 & (-1/2,1/2) & (1,2) & (-3,0)* \cr

1567:  2 \le \alpha <k & (-1/2,1/2) & (1,2) & (-2,2) \cr

1568:  k & (-1/k-1/2,-1/k-1/2)* &(-1/k+1,-1/k+1)* & (-2,2)\cr 

1569:  k+1 & (1/2,-1/2) &(2,1) & (-1/k-2,-1/k+1)*\cr

1570:  \alpha>k+1 & (1/2,-1/2) & (2,1) & (-1,1) 

1571:  \end{array}

1572: \ee

1573: The table agrees with the Poincar\'e polynomials listed previously. We find

1574: the following marginal deformations:

1575: 

1576: \begin{itemize}

1577: \item The untwisted $(c,c)$ marginal states are straightforwardly

1578:   enumerated. They are given by invariant combinations of the $\Phi_i$ acting

1579:   on the vacuum: $\Phi_1^a \Phi_2^b \Phi_3^{-c} |0 \rangle_{NS}$,

1580:   with the bounds $0 \le a,b \le 2k-2$ and $2 \le c \le k$. The marginality

1581:   condition is:  $a+b+2c=2k$.\\

1582: Let's count these states. For a given $c$, each $a$ in the range $0 \le a \le

1583:   2k-2c$ gives exactly one solution. So the total number of states is:

1584: \be \sharp (c,c) = \sum_{c=2}^{k}(2k-2c+1) = (k-1)^2 \ee

1585: 

1586: \item A further search for marginal $(c,c)$ states gives a negative result.

1587:   When $\alpha$ is smaller than $k$, we have $(c,c)$ states that survive

1588:   projection, but they are not marginal since they have charges $(1,2)$. When

1589:   $\alpha=k$, the charges left and right can also never be both equal to one. When

1590:   $\alpha$ is larger than $k$, the charges are $(2,1)$ which also never leads to

1591:   marginality.

1592:   

1593: \item Let us look for marginal $(a,c)$ states in the twisted sector. We get

1594:   charges $(-2,2)$ when $\alpha-1$ is smaller than $k$, and therefore no marginal

1595:   states in these sectors. When $\alpha-1$ is larger

1596:   than $k$, we get charges $(-1,+1)$ which are marginal. So we get $(k-1)$

1597:   $(a,c)$ states from the twisted sectors, labeled by $\{k+2,\ldots 2k\}$.

1598:   There are no other marginal states in this theory.

1599: 

1600: \end{itemize}

1601: In summary, we find $(k-1)^2$ untwisted marginal $(c,c)$ states, and we find

1602: $k-1$ marginal $(a,c)$ states. Once again, only those states that arise from

1603: the spectral flow of RR ground states with untwisted non-compact factors are

1604: retained in the theory deformed with the Liouville potential. So all

1605: the $(c,c)$ states are in the spectrum of the deformed theory, but none of the

1606: $(a,c)$

1607:  states are.

1608: 

1609: \subsubsection{Orbifolds}\label{twoMMorbifold}

1610: 

1611: We will now consider orbifolds of the above model.  As we will discuss in

1612: detail in the next section, this exercise is useful since it generates an

1613: infinite number of mirror theories. The logic will be analogous to the

1614: Greene-Plesser analysis for the Gepner point in the quintic Calabi-Yau.  Under

1615: the hypothesis that the level $k$ is not prime, we can write the level as a

1616: product $k=k_1.k_2$ for two positive integers $k_1$ and $k_2$. Then we can

1617: perform a $\IZ_{k_1}$ orbifold of the three-factor model that we discussed

1618: above.

1619: 

1620: The Landau-Ginzburg model has superpotential 

1621: $$W_{LG}= \Phi_1^{2k} + \Phi_2^{2k} + \Phi_3^{-k}.$$ 

1622: The full symmetry group of $W_{LG}$ is $D=\IZ_{2k}

1623: \times \IZ_{2k} \times \IZ_{k}$, where each factor acts by phase

1624: multiplication on one of the superfields. The integer R-charge operator $g_{0}$ generates

1625: a $\IZ_{2k}$ subgroup. It acts as: \be g_{0}: (\Phi_1,\Phi_2,\Phi_3)

1626: \rightarrow (e^{\frac{2 i \pi}{2k}}\Phi_1,e^{\frac{2 i

1627:     \pi}{2k}}\Phi_2,e^{-\frac{2 i \pi}{k}}\Phi_3) \,.  \ee Now consider the

1628: $g_{1}$ operator with the following action: \be g_{1}:

1629: (\Phi_1,\Phi_2,\Phi_3) \rightarrow (e^{+\frac{2 i \pi

1630:     k_2}{2k}}\Phi_1,e^{-\frac{2 i \pi k_2}{2k}}\Phi_2,\Phi_3) \,.  \ee Earlier

1631: we orbifolded the Landau-Ginzburg model by the group generated by $g_{0}$. Consider the group of order

1632: $2k k_1$ generated by $g_{0}$ and $g_{1}$. We now orbifold the

1633: Landau-Ginzburg theory by that group. This is compatible with

1634: supersymmetry.

1635:  Let us find marginal deformations in this orbifold model,

1636: following the Landau-Ginzburg method. First

1637: we tabulate the R-charges of the Ramond-Ramond ground states. We label the sector

1638: twisted by $g_{0}^{\alpha} g_{1}^{\beta}$ with $\alpha$ and $\beta$, within the

1639: bounds $0 \le \alpha \le 2k-1$ and $0 \le \beta \le k_1-1$. A star means that

1640: some fields are untwisted.

1641: \be

1642: \begin{array}{c|c|c}

1643: \alpha & \beta & RR  \cr

1644: \hline

1645: \alpha=0 & \beta=0 & (-3/2,-3/2)* \cr

1646: 0 < \alpha <k &  \beta k_2 < \alpha &  (-1/2,1/2) \cr

1647: 0 < \alpha <k &  \alpha < \beta k_2  & (1/2,-1/2) \cr

1648: k<\alpha & \beta k_2 <2k-\alpha & (1/2,-1/2) \cr

1649: k<\alpha &  2k-\alpha<\beta k_2 & (-1/2,1/2)  \cr

1650: \alpha=0 & \beta \neq 0 & (-1/2-1/k,-1/2-1/k)* \cr

1651: \alpha=k & \beta & (-1/2-1/k,-1/2-1/k)*  \cr

1652: \alpha=\beta k_2 & \beta \neq 0 & (-1/2+1/2k,-1/2+1/2k)*  \cr

1653: \alpha=2k-\beta k_2 & \beta \neq 0 & (-1/2+1/2k,-1/2+1/2k)*  \cr

1654: \end{array}

1655: \ee

1656: Let us count the marginal operators:

1657: \begin{itemize}

1658: \item In the untwisted sector, we find $(c,c)$ chiral primaries.

1659:   Remember that in the model orbifolded by $g_{0}$, the $(k-1)^2$ $(c,c)$ states are labeled by three

1660:   integers $a,b,c$, such that $0 \le a,b \le 2k-2$, $2 \le c \le k$ and

1661:   $a+b+2c=2k$. The $g_{1}$ projection keeps only those that have $a \equiv b

1662:   \ [2k_1]$.\\

1663: Let us work at given $c$. We want to count the number of solutions to the

1664:   equation $a+b=2k-2c$, with  $a \equiv b\ [2k_1]$. We write $b=a+2d k_1 $,

1665:   with $d$ integer. Then we express $a$ and $b$ in terms of $c$ and $d$ only:

1666:   $a=k-c-d k_1$, and $b=k-c+d k_1$. The bounds on

1667:   $a$ and $b$ imply $-k+c \le d k_1 \le k-c$. Thus we have one solution for each

1668:   integer $d$ between $\frac{c-k}{k_1}$ and $\frac{k-c}{k_1}$.\\

1669: So the total number of $(c,c)$ states is:

1670: \be \sharp (c,c) = \sum_{c=2}^{k} \left( \left\lfloor \frac{k-c}{k_1}  \right\rfloor - \left\lceil

1671:   \frac{c-k}{k_1} \right\rceil + 1 \right) \ee

1672: where $\lfloor x \rfloor$ is the largest integer smaller or equal to $x$, and

1673: $\lceil x \rceil$ is the smallest integer bigger or equal to $x$.\\

1674:  This sum can be evaluated explicitly :

1675: \be \sharp (c,c) = 

1676: 2 \left( (k_2-1)(k_1-1) + \sum_{\check{c}=2}^{k_2} (k_2-\check{c})k_1  \right) + k-1 =

1677: k_1 k_2^2 - 2k_2 +1 \ee

1678: \item In the sector twisted by $g_{0}^{\alpha\mp1}\ g_{1}^\beta$, we find

1679:   $(c,a)$ and $(a,c)$ states\footnote{ The $\a\mp1$ occurs due to the shift in

1680:     the labeling of the twisted sectors when flowing asymmetrically from the

1681:     RR to the $(c,a)$ or $(a,c)$ sectors.}:

1682: \begin{itemize}

1683: \item If $0 \le \beta k_2  < \alpha <k$, or $\alpha > 2k- \beta k_2$, 

1684:  we find one marginal $(c,a)$ state. Let's count the number of such

1685:   twisted sectors. It can be written as:

1686: \be \sharp(c,a) = \sum_{\alpha=1}^{k-1} \left( \left\lfloor \frac{\alpha}{k_2}

1687:  \right\rfloor +1 \right) 

1688: + \sum_{\alpha=k+1}^{2k-1} \left( k_1-1- \left\lfloor \frac{2k-\alpha}{k_2}

1689:  \right\rfloor\right) \ee

1690: This sum can be computed explicitly as:

1691: \be \sharp(c,a) = \left( (k_2-1)\sum_{\hat{\alpha}=1}^{k_1} \hat{\alpha} \right) +

1692: \left( (k_1-1)(k-1) - (k_2-1)\sum_{\check{\alpha}=0}^{k_1-1} \check{\alpha}

1693: \right) = k_2 k_1^2 -2k_1+1 \ee

1694: \item Symmetrically, if $0 < \alpha < \beta k_2 $, or $k < \alpha < 2k-\beta k_2$, 

1695:  we find one $(a,c)$ state. The same counting shows that there are also $k_2 k_1^2 -2k_1+1$ such sectors.

1696: \item Eventually, if $\alpha=0$, $\alpha=k$ or $\beta k_2 = \pm \alpha \

1697:   [2k]$, we find no marginal deformation. 

1698:   \end{itemize}

1699: \end{itemize}

1700: 

1701: In summary, we have in this orbifold model: 

1702: \begin{itemize}

1703: \item $k_1 k_2^2 -2k_2+1$ marginal $(c,c)$ states.

1704: \item $k_2 k_1^2 -2k_1+1$ marinal $(a,c)$ states.

1705: \end{itemize}

1706: Once again, only the $(c,c)$ states are present in the spectrum of the

1707: deformed theory.

1708: 

1709: 

1710: \subsection{The $(k;2k,2k)$ Model.} \label{twocigars}

1711: This example will be very similar to the previous one. For this reason our discussion will be brief and we will focus on marginal deformations. The Landau-Ginzburg model has superpotential 

1712: \be

1713: W_{LG}= \Phi_1^k + \Phi_2^{-2k} + \Phi_3^{-2k}\,.

1714: \ee 

1715: The integer R-charge operator $g_{0}$ acts on the superfields as:

1716: \be

1717: g_{0}: (\Phi_1,\Phi_2,\Phi_3) = (e^{\frac{2 i \pi}{k}}\Phi_1,e^{-\frac{2 i \pi}{2k}}\Phi_2,e^{-\frac{2 i \pi}{2k}}\Phi_3) \,.

1718: \ee

1719: We tabulate the R-charges of the Ramond ground states and their spectral

1720: flows.

1721: A star means that some fields are untwisted, and $\alpha$ labels the sector

1722: twisted by $g_{0}^{\alpha}$:

1723: \be

1724: \begin{array}{c|c|c|c}

1725: \alpha& RR & (c,c) & (a,c) \cr

1726: \hline

1727: 0 & (-3/2,-3/2)* & (0,0)* & (-2,2) \cr

1728: 1 & (1/2,-1/2) & (2,1) & (-3,0)* \cr

1729: 2 \le \alpha<k & (1/2,-1/2) & (2,1) & (-1,1) \cr

1730: k& (1/k-1/2,1/k-1/2)* &(1/k+1,1/k+1)* & (-1,1)\cr 

1731: k+1& (-1/2,1/2) &(1,2) & (1/k-2,1/k+1)*\cr

1732: \alpha>k+1 & (-1/2,1/2) & (1,2) & (-2,2) 

1733: \end{array}

1734: \ee

1735: 

1736: \begin{itemize}

1737: \item In the untwisted sector, we find 

1738: $(k-1)^2$ $(c,c)$ states.

1739: \item In the sector twisted by $g_{0}^\alpha$ ($0<\alpha<2k-1$), we find one

1740:   $(a,c)$ state if $2\le \alpha \le k$, and one $(c,a)$ state if $k \le

1741:   \alpha\le 2k-2$.

1742: \end{itemize}

1743: To summarize, this model has $(k-1)^2$ marginal $(c,c)$ deformations and $k-1$

1744: margial $(a,c)$ deformations.

1745: 

1746: 

1747: \subsubsection{Orbifolds}\label{twocigarsorbifold}

1748: 

1749: We repeat the analysis of section \ref{twoMMorbifold}, and study the orbifolds

1750: of this model. The Landau-Ginzburg model has superpotential $W_{LG}= \Phi_1^k + \Phi_2^{-2k} + \Phi_3^{-2k}$. The full group of symmetries of $W_{LG}$ is $\IZ_k \times \IZ_{2k} \times \IZ_{2k}$, where each factor acts by phase multiplication on one superfield.

1751: 

1752: The integer R-charge operator $g_{0}$ generates a $\IZ_{2k}$ subgroup and acts as follows:

1753: \be 

1754: g_{0}: (\Phi_1,\Phi_2,\Phi_3) \rightarrow (e^{\frac{2 i \pi}{k}}\Phi_1,e^{-\frac{2 i \pi}{2k}}\Phi_2,e^{-\frac{2 i \pi}{2k}}\Phi_3) 

1755: \ee

1756: Now we consider the $g_{1}$ operator with the following action:

1757: \be 

1758: g_{1}: (\Phi_1,\Phi_2,\Phi_3) \rightarrow (\Phi_1,e^{-\frac{2 i \pi k_2}{2k}}\Phi_2,e^{+\frac{2 i \pi k_2}{2k}}\Phi_3)

1759: \ee

1760: As in the $(2k,2k;k)$ model, we assume that $k=k_1. k_2$, orbifold the theory

1761: by the subgroup generated by $g_{0}$ and $g_{1}$ and look for marginal

1762: deformations. The 

1763: counting is very similar 

1764: to the $(2k,2k;k)$ case. We find

1765: \begin{itemize}

1766: \item $k_1 k_2^2 -2k_2+1$ marginal $(c,c)$ states.

1767: \item $k_2 k_1^2 -2k_1+1$ marginal $(a,c)$ states.

1768: \end{itemize}

1769: 

1770: \subsection{The $(3,3,3;2)$ Model} 

1771: 

1772: All our previous examples share an interesting feature. The marginal $(c,c)$

1773: operators are obtained by half-unit spectral flow of untwisted RR-ground

1774: states. On the other 

1775: hand, the marginal $(a,c)$ operators are obtained by

1776: half-unit asymetric spectral flow of twisted RR-ground states. A direct

1777: consequence is that in these examples, the deformed theory only has

1778: $(c,c)$ moduli in its spectrum. 

1779: This statement looks general, since untwisted ground states have equal left

1780: and right R-charges, and 

1781: tend to flow to chiral operators. Similarly,

1782: twisted ground states have different left and right R-charges, and would be 

1783: expected to flow to anti-chiral operators. However, we will show in the present example

1784: that there are exceptions to this rule.

1785: 

1786: We consider the Gepner model with three compact factors at level 3, and one

1787: non-compact factor at level 2. The superpotential of the corresponding

1788: Landau-Ginzburg model is \be W=\Phi_1^3+\Phi_2^3+\Phi_3^3+\Phi_4^{-2}+\Phi_5^2

1789: \,, \ee and the model is orbifolded by the group generated by $g_{0}$: \be g_0

1790: : (\Phi_1,\Phi_2,\Phi_3,\Phi_4,\Phi_5)\rightarrow (e^{\frac{2i\pi}{3}}\Phi_1,

1791: e^{\frac{2i\pi}{3}}\Phi_2, e^{-\frac{2i\pi}{3}}\Phi_3, e^{-i\pi}

1792: \Phi_4,e^{i\pi}\Phi_5) \,.  \ee The addition of $\Phi_5$ is the simplest means

1793: to ensure that the Calabi-Yau condition is maintained while setting all the

1794: phase factors in \cite{Intriligator:1990ua} to zero. Using the by now familiar

1795: techniques, we tabulate the R-charges of the ground states in various sectors:

1796: \be

1797: \begin{array}{c|c|c|c}

1798: \alpha & RR & (c,c) & (a,c) \\

1799: \hline 

1800: 0 & (-3/2,-3/2)* \diamond & (0,0)* \diamond &(-1,1) \\

1801: 1 & (-1/2,1/2) & (1,2) & (-3,0)* \diamond \\

1802: 2 & (-1/2,-3/2)* & (1,0)* & (-2,2) \\

1803: 3 & (-1/2,-1/2)\diamond & (1,1)\diamond & (-2,0)* \\

1804: 4 & (-3/2,-1/2)* & (0,1)* & (-2,1) \diamond\\

1805: 5 & (1/2,-1/2) & (2,1) & (-3,1)* \\

1806: \end{array}

1807: \ee

1808: The star and the diamond respectively mean that the non-compact and the

1809: compact fields are untwisted. More precisely, $\Phi_{1,2,3}$

1810: can have zero modes in the $\alpha=0,3$ twisted sectors for the $(c,c)$ ring

1811: while they can have zero modes in the $\alpha=1,4$ twisted sectors in the

1812: $(a,c)$ ring. Similarly, $\Phi_{4,5}$ can have zero modes in the $\alpha=0,2,4$

1813: sectors in the $(c,c)$ ring, while it can have zero modes in the $\alpha=1,3,5$ sectors in the $(a,c)$ ring.

1814: 

1815: >From this, we see that there are 2 $(c,c)$ moduli:

1816: \begin{align}

1817: \Phi_4^{-2}\ket{0}_{c,c}^{\alpha=0} \quad \text{and} \quad \ket{0}_{c,c}^{\alpha=3}

1818: \cr

1819: \end{align}

1820: and $2$ $(a,c)$ moduli:

1821: \be

1822: \ket{0}_{a,c}^{\alpha=0} \quad \text{and}\quad \Phi_4^{-2}\ket{0}_{a,c}^{\alpha=3} \,.

1823: \ee

1824: Notice that the $(c,c)$ modulus in the third twisted sector occurs in a sector

1825: in which the fields that corresponds to the non-compact direction, $\Phi_4$,

1826: is twisted. Moreover, the $(a,c)$ modulus in the third twisted sector appears

1827: while the non-compact direction is untwisted. As a consequence, the theory deformed by

1828: the Liouville potential has only one $(c,c)$ modulus in its spectrum, {\em

1829:   and} it 

1830: also has one $(a,c)$ modulus. 

1831: (On the other hand, the resolved theory will have

1832: one $(a,c)$ modulus, together with one $(c,c)$ modulus.) It can be

1833: checked by direct calculation that this is consistent with the equations

1834: analyzed in \cite{Eguchi:2004yi}. 

1835: The Gepner model analysis of this model is discussed briefly in appendix \ref{3332}.

1836: 

1837: 

1838: \section{Mirror Symmetry For Non-compact Gepner Models}\label{CFTmirror}

1839: 

1840: In this section we address the question of identifying mirror pairs. In the

1841: case of compact Gepner models, when we specify the diagonal model as our

1842: starting point, we obtain the mirror model by modding out by the maximal

1843: discrete subgroup $H$ of the  diagonal group $G$ that is

1844: consistent with space-time supersymmetry \cite{Greene:1990ud}. Subgroups $F$ of $H$ give rise to

1845: models that are mirror to models modded out by $H/F$.

1846: 

1847: In the following we will argue that non-compact Gepner models behave very

1848: similarly, in their undeformed guise. In particular, we shall show that modding

1849: out by the maximal subgroup consistent with supersymmetry, we exchange $(c,c)$

1850: and $(c,a)$ deformations of the undeformed theory.

1851: 

1852: A corollary of 

1853: this statement is that a theory deformed by a given operator, will map after mirror symmetry to the mirror 

1854: theory, deformed by the mirror operator. Thus, mirror symmetry applies to the deformed, regular theories as well. 

1855: 

1856: The main difference with the compact models is therefore that the mirror map

1857: includes the specification of the action of the mirror map on the deforming

1858: operator(s). Naturally, the specification is that one changes the right-moving

1859: R-charge of the deforming operator to find the mirror deformation. The mirror map extends to subgroups of $H$ as in the compact case.

1860: 

1861: 

1862: We believe it is best to illustrate the above general framework in a few

1863: examples. In the following section, we will then revisit these examples and

1864: see to what extent we can interpret mirror symmetry of the conformal field

1865: theories in a geometric framework.

1866: 

1867: \subsection{Sixteen Supercharges}

1868: 

1869: We recall that compact Gepner models at central charge $c=6$ lie in the moduli

1870: space of $K3$ compactifications of string theory. It is well-known

1871: \cite{Aspinwall:1994rg} that the mirror transform acts as an automorphism of

1872: the K3 moduli space and there are special points in the moduli space where

1873: there are fixed points. For our non-compact $(k;k)$ model at central charge

1874: $c=6$, we see that in its singular guise, the particular $8(k-1)$ deformations

1875: that we identified are indeed self-mirror. This can be seen in figure

1876: \ref{ALE}. Changing the sign of the right-moving R-charge exchanges the two

1877: individual sets of $4(k-1)$ deformation parameters that we distinguished

1878: previously. We note that this property of self-mirroring holds only for the

1879: singular model.\footnote{We do not advocate that the reader take the

1880:   undeformed (well-defined) conformal field theory as a

1881:  good description

1882:   of the strongly coupled string background. We use it as a formal tool in

1883:   arguing for the precise points of analogy and difference with compact Gepner

1884:   models.}

1885: 

1886: \begin{figure}

1887: \centering

1888: \includegraphics[scale=.92]{kkdef}

1889: \caption{Chiral-chiral and chiral-anti-chiral

1890: primaries in the $(k;k)$ model. Each dot corresponds to one chiral primary,

1891: identified by its non-compact left- and right-moving

1892: quantum numbers $2m$ and $2\tilde{m}$.\label{ALE}}

1893: \end{figure}

1894: 

1895: As a consequence, we can discuss mirror symmetry for the weakly coupled deformed

1896: model. When we deform the $(k;k)$ model with a Liouville potential (consistent

1897: with sixteen supercharges), we are left with $4 (k-1)$ deformation parameters

1898: and a regular theory. It is mirror dual to the singular theory deformed by the

1899: winding potential (which geometrically gives rise to the cigar theory). The latter

1900: theory also has $4(k-1)$ deformation parameters, which can be mapped

1901: individually to their mirror images (using their worldsheet R-charges for each

1902: factor of the model).

1903: 

1904: \subsection{Eight Supercharges}

1905: 

1906: For compact Gepner models at central charge $c=9$, mirror symmetry maps one model onto another, exchanging $(c,c)$ and $(c,a)$ states. We will illustrate that this is the case for non-compact Gepner models as well. First we will treat a case in which we simply mod out by the maximal group, and thus

1907: obtain the mirror theory. Then we will show that one can mod out by a subgroup of the maximal subgroup and obtain mirror pairs. We will thus provide  large classes of mirror non-compact Gepner models.

1908: 

1909: \subsubsection{The $(k;2k,2k)$ Model}\label{zkoneorb}

1910: 

1911: We note that the $(k;2k,2k)$ model, which is modded out only by the GSO

1912: $\IZ_{2k}$ group (in the Landau-Ginzburg formulation) has $(k-1)^2$ marginal

1913: $(c,c)$ deformations and $k-1$ marginal $(a,c)$ deformations, as discussed in

1914: section \ref{twocigars}. Orbifolds of this model were treated in section

1915: \ref{twocigarsorbifold}.  The maximal group that we can divide out by

1916: consistent with supersymmetry is the orbifold group when $k_1=k$ and $k_2=1$.

1917: Then, substituting in the formulae for the massless moduli computed in that

1918: section, we get $k-1$ marginal $(c,c)$ deformations and $(k-1)^2$ marginal

1919: $(a,c)$ deformations. The orbifold indeed gives rise to the mirror theory.

1920: This is exactly analogous to the Greene-Plesser discussion of mirror conformal

1921: field theories associated to compact Calabi-Yau threefolds at a Gepner point.

1922: 

1923: We want to compare these models to those in the literature. The $(k;2k,2k)$

1924: model modded out by the maximal group and deformed by the sum of Liouville

1925: potentials corresponds to the non-compact Calabi-Yau studied in

1926: \cite{Lerche:2000uy} to geometrically engineer pure $SU(k)$ gauge theory. The

1927: orbifold group restricts all the possible $(c,c)$ deformations to the $k-1$

1928: moduli studied in that paper which span the Coulomb branch of the $SU(k)$

1929: gauge theory. We will discuss the relation between our non-compact Gepner

1930: model, the associated Landau-Ginzburg model and this geometry in more detail

1931: in the next section.

1932: 

1933: Moreover, we find that this model is mirror to a model that has no orbifold

1934: except the GSO projection, and deformed by the winding condensates in the two

1935: non-compact directions. That gives rise to the two-cigar model of

1936: \cite{Eguchi:2004ik}, as argued in that reference. From our perspective, we

1937: see that the double cigar deformations disallow all $(k-1)^2$ marginal $(c,c)$

1938: deformations in the unorbifolded model, and leaves only $k-1$ K\"ahler

1939: deformations.  That matches precisely the analysis in \cite{Eguchi:2004ik}.

1940: Note that this gives a confirmation of our methodology: the counting in

1941: \cite{Eguchi:2004ik} is based on the spectrum identified by considering a

1942: regularized partition function \cite{Maldacena:2000kv, Israel:2004ir,

1943:   Hanany:2002ev, Eguchi:2004yi}. Thus we find that the regularized partition

1944: function agrees with our intuitive arguments which find their basis in the

1945: Landau-Ginzburg model with negative power potentials (see sections \ref{LG}

1946: and \ref{orbifoldedLG}).

1947: 

1948: As an example of the power of our simple description, we note that the model

1949: with Liouville deformations and no orbifold action is not directly related to

1950: the known models described in \cite{Lerche:2000uy, Eguchi:2004ik}.

1951: Nevertheless, we readily identify its mirror to be the orbifolded theory with

1952: cigar deformations in both the non-compact factors.

1953: 

1954: \subsubsection{The Orbifolded $(2k,2k;k)$ Models}

1955: 

1956: It should be clear now that the examples in sections \ref{twoMMorbifold} and

1957: \ref{twocigarsorbifold} give rise to two infinite classes of mirror

1958: non-compact Gepner models. We will illustrate this for one of the two classes

1959: since the other class behaves in almost every respect analogously.\footnote{We

1960:   will signal the exception in the next section.}

1961: 

1962: Recall that the $(2k,2k;k)$ model allows for a maximal $\IZ_k$

1963: orbifold

1964: consistent with supersymmetry. When the level $k$ is a product of two positive

1965: integers

1966: $k=k_1k_2$,

1967: we can orbifold by a non-maximal subgroup $F=\IZ_{k_1}$ which gives rise to a

1968: model which is mirror to a $\IZ_{k_2}$ orbifold of the same model. Indeed, the

1969: counting of $(k_1k_2^2-2k_2+1)$ marginal $(c,c)$ operators and

1970: $(k_2^2k_1-2k_1+1)$ marginal $(c,a)$ states in the first model (the

1971: $\IZ_{k_1}$ orbifold) is precisely the mirror of the counting of the model

1972: with $\IZ_{k_2}$ orbifold group, as can be seen by a exchanging the role of

1973: $k_1$ and $k_2$ (see figure \ref{mirrorsymmetryfigure}). Thus, we have very

1974: simply but explicitly demonstrated the existence of an infinite number of mirror pairs of singular non-compact Gepner models.

1975: 

1976: We note that the model $(k;2k,2k)$ can be treated analogously. The proof

1977: of the generic fact that models modded out by 

1978: a subgroup $F$ of $H$

1979: are mirror to models modded

1980: out by $H/F$ runs along very much the same lines as in the compact case \cite{Greene:1990ud}, for

1981: the undeformed theory. For the deformed theory we need to remember that the

1982: mirror map will mirror the deformation as well.

1983: \begin{figure}

1984: \centering

1985: \includegraphics[scale=.92]{mirrorsymmetryfigure}

1986: \caption{An explicit example: the $(2k,2k;k)$ models and its orbifolds by

1987:   $\IZ_{k_1}$, with $k=30$.}

1988: \label{mirrorsymmetryfigure}

1989: \end{figure}

1990: 

1991: \section{Mirror Non-compact Geometries and their Relation to $\,\IC^n/\Gamma$ Orbifolds}

1992: \label{geometrymirror}

1993: In this section we want to discuss some geometric realizations of the mirror

1994: map that we identified in the non-compact Gepner models above. We will see

1995: that we can identify various models with non-compact Calabi-Yau geometries,

1996: and approximate their mirror duals with abelian orbifolds of $\, \IC^3$ at

1997: large levels. At finite level, we find that the results of toric geometry 

1998: acquire important modifications that lift certain

1999: moduli in our models.

2000: 

2001: \subsection{The $(k;k)$ Model}

2002: 

2003: In this model, the space-time background has sixteen supercharges. In

2004: analogy with the compact case, we identify the Calabi-Yau manifold that

2005: corresponds to the Landau-Ginzburg model by writing down the superpotential

2006: (augmented with the appropriate number of quadratic coordinates) in the non-compact weighted projective space $W\IC\IP^{4}$:

2007: \be

2008: w_1^k + w_2^2 + w_3^2 + c w_4^{-k} = 0.

2009: \ee

2010: The constant $c$ (multiplied by the coefficient of the first monomial)

2011: measures the strength of the Liouville deformation.

2012: By scaling the $\, \IC^\ast$ valued coordinate $w_4$ to one, we recuperate the

2013: equation:

2014: \begin{eqnarray}

2015: z_1^k + z_2^2 + z_3^2 + c &=& 0,

2016: \end{eqnarray}

2017: which describes an ALE space which is deformed by $c$ from its singular

2018: orbifold limit. We thus associate that geometry to the Landau-Ginzburg model, but we should

2019: keep in mind that this association is local, i.e. near the singularity. 

2020: Asymptotically the spaces differ. We will see an example of the consequences of

2021: this difference later on.

2022: The matching of the $4(k-1)$ marginal deformations to the geometric moduli is

2023: well-known.

2024: 

2025: \subsubsection{The Mirror Theory}

2026: 

2027: The conformal field theory mirror to the above theory was argued to be the

2028: $(k;k)$ model with cigar deformation. In the appendix we recall, following

2029: \cite{Ooguri:1995wj, Kutasov:1995te}, that after a single T-duality, this

2030: model is mapped to a configuration of NS5-branes spread on a circle in a

2031: transverse plane. Moreover, using the explicit geometric description, it can

2032: be argued in great detail (see the appendix) that the NS5-branes spread on the

2033: circle map under that T-duality to an orbifold singularity of the type $\, \IC^2/\IZ_k$ at the tip

2034: of the cigar and the center of the minimal model disc. In particular, we see

2035: that the $\IZ_k$ orbifold that arises from the GSO projection on the side of

2036: the cigar mirror conformal field theory acts {\em geometrically} on the cigar

2037: and minimal model coordinates close to the tip of the cigar and the center of

2038: the disc.

2039: 

2040: One should contrast this geometric action to the lack of such an action in the

2041: mirror geometry. Moreover, in this example, it becomes 

2042: manifest that when the cigar/winding deformation is turned on in the singular

2043: theory, the deformation caps off the linear dilaton cylinder. As a consequence

2044: it gives rise to a (geometric) fixed point which allows for the localization

2045: of twisted sector states. That agrees with our prescription for keeping the

2046: $4(k-1)$ twisted sector states when turning on the cigar deformation.

2047: 

2048: Note that there is a one-to-one match of the marginal supersymmetric

2049: deformations of the cigar times minimal model conformal field theory to the

2050: marginal supersymmetric deformations of the $\, \IC^2/\IZ_k$ orbifold (with

2051: $B$-field on the vanishing cycles) \cite{Ooguri:1995wj}.

2052: 

2053: In the strictly infinite level limit, the match is expected, since then the

2054: cigar and disc flatten out completely. The two models become identical in that

2055: limit. However, at finite level $k$, the match is due to 

2056: the rigidity of the sixteen supercharge construction, as will 

2057: become clear when we discuss models 

2058: with less supersymmetry.

2059: 

2060: \subsection{The $(k;2k,2k)$ Model}

2061: 

2062: Models with eight supercharges will show various new and interesting features.

2063: We will discuss two models in detail with increasing level of complexity. The

2064: model with levels $(k;2k,2k)$ and double Liouville deformation has a

2065: singularity that is well-described by a hypersurface in the

2066: (non-compact)weighted projective space of the form \cite{Eguchi:2004ik}:

2067: \begin{eqnarray}

2068: w_1^k + w_2^2 + w_3^2 +  w_4^{-2k} + w_5^{-2k}  &=& 0,

2069: \end{eqnarray}

2070: where the two coordinates $w_{4,5}$ are $\, \IC^\ast$ valued.  The deforming

2071: monomials are in one-to-one correspondence with the $(k-1)^2$ marginal $(c,c)$

2072: deformations that we identified previously. After performing the $\IZ_{k}$

2073: orbifold, we find that only the complex structure

2074: deformations of the form $w_1^n (w_4 w_5)^{-k+n}$ for $n=0,1,\dots,k-2$ are

2075: invariant under the orbifold action. As mentioned earlier, this is precisely

2076: the geometry discussed in \cite{Lerche:2000uy, Hori:2002cd, Eguchi:2004ik}.

2077: 

2078: \subsubsection{The Mirror Theory }

2079: 

2080: Let us now turn to the mirror conformal field theory and see whether we can count the number of K\"ahler deformations of the mirror theory using geometric means. The first observation we make is that, just as in the case with sixteen supercharges, there is an infinite level limit in which the model flattens out, up to an overall orbifold action. We refer the reader to appendix \ref{NS5} for the basic arguments in favour of such a description. In that (strict) limit, the model, locally, near the tips of the cigars and the center of

2081: the minimal model disc becomes equivalent to the orbifold $\, \IC^3/\IZ_{2k}$. Again, we use that on this side of the mirror symmetry, the GSO projection acts geometrically and infer that the

2082: action of the $\IZ_{2k}$ on the three factors of $\, \IC^3$ is 

2083: weighted

2084: as $\frac{1}{2k} (2k-1,2k-1,2)$. This orbifold is toric and the number of

2085: K\"ahler deformations of the geometry can be counted using toric geometry

2086: techniques\footnote{See, for instance, \cite{Aspinwall:1994ev, lustetal} and references

2087:   therein for a review of these methods.} as follows.

2088: 

2089: Consider the supersymmetric orbifold $\,\IC^3/\Gamma$, where the orbifold group of order $N$ is generated by 

2090: \be \theta: (z_1,z_2, z_3) \longrightarrow (\omega^{a_1} z_1,\omega^{a_2} z_2, \omega^{a_3} z_3) \quad \text{such that}\quad \sum_{i}a_i = 0 \quad \text{mod} \quad N \,.  

2091: \ee 

2092: The counting of K\"ahler deformations proceeds as follows.

2093: 

2094: \begin{itemize}

2095: \item Consider all powers of $\theta$ and list their exponents in multiples of

2096:   $2 \pi i$ such that they fall in the range $1/N \times (0,N-1)$. Label them

2097:   $(g_1, g_2, g_3)$.

2098:   

2099: \item The K\"ahler moduli are in one to one correspondence with those powers

2100:   of $\theta$ that satisfy the following conditions: \be\label{constraint}

2101:   \sum_{i=1}^{n} g_i = 1 \quad \text{with} \quad 0\le g_i < 1 \,.  \ee The

2102:   power of $\theta$ tells you in which twisted sector the modulus appears.

2103: \end{itemize}

2104: 

2105: Before we proceed further let us also briefly recall how one draws the toric

2106: diagram corresponding to the non-compact orbifold. This will turn out to be

2107: useful to compare and contrast the spectrum of these non-compact toric

2108: orbifolds with the Landau-Ginzburg computation of the spectrum of moduli for

2109: the non-compact Gepner models.

2110: 

2111: Given an action of $\theta=(\theta_1, \theta_2, \theta_3)$ on $\,\IC^3$ as

2112: above, we first find basis vectors $\{D_1, D_2, D_3\}$ that satisfy \be

2113: \sum_{i=1}^{3}\theta_i \, (D_i)_a = 0 \quad \text{mod}\quad N \,.  \ee These

2114: three basis vectors generate the toric fan. One solution is  $(D_i)_3 = 1\ 

2115: \forall\ i$. One can find two other linearly independent solutions to this

2116: equation. Thus, neglecting the third coordinate of the vectors and plotting only the other two solutions to the above equation, one gets points on a plane. That this must be so follows from the Calabi-Yau condition. These points define the Newton polyhedron corresponding to the non-compact Calabi-Yau. The basis vectors $D_i$ form a cone over the polyhedron which allows one to draw the toric diagram for the Calabi-Yau threefold in the plane. Now, for each power of $\theta^{n}=(g^{(n)}_1,g^{(n)}_2,g^{(n)}_3)$ that satisfies the above two constraints (i.e. for every K\"ahler deformation),we add another vector (interior or boundary point in the toric diagram) $E_n$ given by 

2117: \be 

2118: E_n = \sum_{i=1}^{3}\, g^{(n)}_i D_i \,.  

2119: \ee

2120: 

2121: Let us apply the above algorithm to compute the K\"ahler moduli and draw the

2122: toric diagram, for our case where $\Gamma=\IZ_{2k}$ (i.e. $N=2k$) and \be

2123: (\theta_1, \theta_2, \theta_3) = \left(\frac{2k-1}{2k}, \frac{2k-1}{2k},

2124:   \frac{2}{2k}\right)\,.  \ee One can check that there are $k$ K\"ahler

2125: deformations which arise from the twisted sectors $k, k+1, \dots, 2k-1$. This

2126: is one more than the number of $(c,c)$ deformations we obtained from the

2127: Landau-Ginzburg computation in section \ref{twocigarsorbifold}. Let us draw

2128: the toric diagram corresponding to this orbifold. We choose a basis of vectors

2129: \be D_1 = (0,1,1) \quad D_2 =(0,-1,1) \quad D_3=(k,0,1) \,.  \ee 

2130: We add $k$ points, corresponding to the $k$ K\"ahler deformations, which are elements

2131: in the $(c,a)$ ring, in the twisted sectors, whose coordinates are given by

2132: \be E_{k}=(0,0,1)\,, \quad E_{k+1} = (1,0,1)\,, \quad E_{k+2} = (2,0,1)\,, \quad \ldots

2133: \quad E_{2k-1} = (k-1,0,1) \,.  \ee

2134: \begin{figure}

2135: \centering

2136: \includegraphics[scale=.75]{toric.eps}

2137: \caption{The toric diagram for $\,\IC^3/\IZ_{2k}$ for $k=4$. Note hat $E_{k=4}$ corresponds to a boundary point while all other added points lie in the interior of the Newton polyhedron. We have also shown a possible triangulation of the polygon, corresponding to a particular resolution of the singularity.\label{toric}}

2138: \end{figure}

2139: These points are plotted in figure \ref{toric}. 

2140: Since $E_k=\half(D_1+D_2)$ in our $\IZ_{2k}$ orbifold it follows that this

2141: point will always be on the boundary of the toric diagram. 

2142: 

2143: When we compare the

2144: two calculations of the moduli, we see that the difference arises in 

2145: the $2k-1$st twisted sector of the GSO orbifold, since, in the conformal field

2146: theory, we did not find a marginal K\"ahler deformation in this twisted sector.

2147: Let us study in somewhat more detail how this difference comes about.

2148: 

2149: 

2150: It is well known that the points in the toric diagram can also be associated

2151: to exceptional divisors of the resolution of the orbifold. This is the twist-field-divisor map discussed in \cite{Aspinwall:1994ev}. The topology and

2152: intersection numbers of these divisors can be obtained in a straightforward

2153: manner by using the dual toric diagram. The boundary point corresponds to

2154: non-compact divisors and have the topology $\IP^1 \times\, \IC$. Therefore,

2155:  the existence of the resolution corresponding to 

2156: $E_k$ in the conformal field theory might seem

2157: surprising, given that the (strictly) normalizable deformations in the conformal field theory can be

2158: associated to compact cycles in the geometry. This can be understood as

2159: follows: the $k$th power of the orbifold action is trivial on one of the three

2160: complex directions in $\, \, \IC^3$, and creates a singularity that stretches along

2161: a complex line in $\, \, \IC^3$. The deformation is therefore akin to a K\"ahler

2162: deformation of a $\, \, \IC^2/\Gamma$ orbifold, embedded in $\, \IC^3$. However, it is

2163: important to note that the $\,\IC$ direction that is left invariant in the

2164: $k$th twisted sector, is the direction that is compactified 

2165: in the conformal

2166: field theory by the addition of the Landau-Ginzburg

2167: potential.

2168:  So in the

2169: conformal field theory, all the exceptional divisors become compact. A second

2170: effect of the compactification is that only $k-1$ of 

2171: the deformations are

2172: linearly independent. The Landau-Ginzburg counting of chiral primaries gives an easy method

2173: to understand which of the exceptional divisors are chosen as a basis in the

2174: conformal field theory.

2175: 

2176: \begin{figure}

2177: \centering

2178: \includegraphics[scale=0.75]{dual.eps}

2179: \caption{The dual toric diagram for $\,\IC^3/\IZ_{2k}$ with $k=4$. We use the same alphabet to denote the point in the toric diagram and the divisor corresponding to it in the dual toric diagram. Note that $E_{k}$ (which is $E_4$ in our case) is non-compact while all the other exceptional divisors are compact. \label{dual}}

2180: \end{figure}

2181: 

2182: Note that the $(c,a)$ deformation which is the identity operator in the

2183: minimal model, sets the volume of the compact factor and simultaneously

2184: the volume of the two-cycle at the $\, \IC^2/\IZ_{2}$ singularity.  The operator has

2185: charges $r=(0,0,0;0;k,k)$ and $\tilde{r} = (0,0,0;0;-k,-k)$ as can be seen

2186: from the Landau-Ginzburg model description, or appendix \ref{k2k2k}.

2187: 

2188: \subsubsection{The $(2k,2k;k)$ Model}

2189: 

2190: The toric abelian orbifold that is related to the $(2k,2k;k)$ model is the

2191: same as the one we discussed before. The differences with the previous model

2192: lie in the fact that we now compactify two coordinates, which reduces the

2193: number of K\"ahler moduli by two. By analyzing the spectrum of the conformal field theory as

2194: before, we find that in the toric diagram corresponding to the flat space

2195: orbifold, the twist fields associated to the divisors $E_{k}$ and $E_{k+1}$ are

2196: excluded in the conformal field theory.

2197: 

2198: However, we gain a K\"ahler modulus in the $0$th twisted sector which sets the

2199: overall volume of the two compact factors. It is the identity operator in the

2200: minimal model factors, and the winding operator in the cigar factor.  It has

2201: charges $r=(0,0,0;0,0;k)$ and $\tilde{r}=(0,0,0;0,0;-k)$ as can be seen from

2202: the Landau-Ginzburg description or from appendix \ref{2k2kk}.  This is

2203: consistent with the picture developed in \cite{Giveon:1999px}: the effective

2204: string coupling at the tip of the cigar sets the volume of the resolved

2205: cycles. 

2206: 

2207: Note that in the previous example, we had a slight refinement of the picture

2208: developed in \cite{Giveon:1999px}. Namely, in the presence of the two

2209: non-compact directions, it is the modulus associated to the $\IZ_2$ orbifold

2210: singularity at the tips of the cigars that sets the volume of the internal compact space.

2211: 

2212: \subsubsection{The $\IZ_{k_1}$ Orbifold Of The $(k;2k,2k)$ Model}

2213: In order to understand some more general features of the spectra of the

2214: conformal field theory and how they fit into the spectrum of the flat space

2215: orbifold approximation, let us study the orbifold models studied in sections

2216: \ref{twoMMorbifold} and \ref{twocigarsorbifold}.

2217: 

2218: The flat space approximation to the conformal field theory is given by a $\,\IC^3/\Gamma$ orbifold, generated by the elements

2219: \begin{align}

2220: g_0 &= \frac{1}{2k}(2k-1,2k-1,2) \quad\text{and}\cr

2221: g_1 &= \frac{1}{2k}(k_2,2k-k_2,0) \,.

2222: \end{align}

2223: Using the algorithm discussed earlier, one can easily draw the toric diagram

2224: associated to this singularity as in figure \ref{orbtoric}.

2225: \begin{figure}

2226: \centering

2227: \includegraphics[scale=0.75]{orbdots1}

2228: \caption{The toric diagram for $\,\IC^3/(\IZ_{2k}\times \IZ_{k_1})$ with $k=6$ and $k_1=2$. \label{orbtoric}}

2229: \end{figure}

2230: \begin{figure}

2231: \centering

2232: \includegraphics[scale=0.75]{orbdots3}

2233: \caption{Spectrum of the $\IZ_{k_1}$ orbifold of the $(k;2k,2k)$ conformal field theory for $k=6$ and $k_1=2$ are denoted by the filled dots. The unfilled dots show those points in the flat space approximation which are not included in the conformal field theory. \label{orbthree}}

2234: \end{figure}

2235: \begin{figure}

2236: \centering

2237: \includegraphics[scale=0.75]{orbdots2}

2238: \caption{Spectrum of the $\IZ_{k_1}$ orbifold of the $(2k,2k;k)$ conformal field theory for $k=6$ and $k_1=2$ are denoted by the filled dots. The unfilled dots show those points in the flat space approximation which are not included in the conformal field theory. \label{orbtwo}}

2239: \end{figure}

2240: The spectrum of the exact conformal field theory has already been discussed in the main part of

2241: the paper. If we now plot the spectrum of the exact conformal field theory that corresponds to the $(k;2k,2k)$ model along the same lines, we get figure \ref{orbthree}.  We have shown which elements of the flat space approximation get lifted in going to the exact conformal field theory description.

2242: 

2243: \subsubsection{The $\IZ_{k_1}$ Orbifold Of The $(2k,2k;k)$ Model}

2244: For this model the flat space approximation is identical to the one discussed

2245: above\footnote{For the $(2k,2k;k)$ example, the canonical group element $g_0$

2246:   is given by the inverse of the one we have written here but the group

2247:   generated and the orbifold of $\,\IC^3$ associated to it is unchanged.  }

2248: and it is drawn in figure \ref{orbtoric}. The conformal field theory analysis

2249: is, however, different and the K\"ahler deformations which are kept are shown

2250: in the figure \ref{orbtwo}.

2251: 

2252: 

2253: We mentioned before that in the toric diagram for the flat space orbifolds,

2254: the boundary points correspond to non-compact divisors. Nevertheless as we saw

2255: in the simpler orbifold example, in the exact conformal field theory

2256: description, some of these directions are compactified. These directions are

2257: those along which a potential of the form $\Phi^n$, $n > 0$ is turned on and

2258: which flow into a minimal model in the IR. However, the twist fields that correspond to divisors (via the twist-field-divisor map) which extend along directions that remain non-compact should be excluded in the conformal field theory as these do not lead to normalizable deformations. That this is

2259: so can be checked in our examples.

2260: 

2261: For instance, in the $(2k,2k;k)$ model, in figure \ref{orbtwo}, the excluded

2262: points on the boundary of the toric diagram correspond to K\"ahler

2263: deformations which are in the twisted sectors $g_1^{\b}$ and

2264: $g_0^{k}g_1^{\b}$, with $\b = 1, \ldots k_1-1$. Their respective $U(1)$

2265: charges are given by \be \frac{1}{2k}(\b k_2,2k-\b k_2,0)\quad\text{and}\quad

2266: \frac{1}{2k}(k+\b k_2,k-\b k_2,0)\quad \text{for}\quad n\in \{1, \ldots, k_1-1

2267: \}\,.  \ee As one can see, the twist fields are uncharged under the $U(1)$

2268: that acts on the non-compact direction and the divisors that correspond to

2269: these fields are non-compact. These are subsequently excluded from the

2270: conformal field theory spectrum.

2271: 

2272: However, further work is required to understand in full generality which states of the flat space orbifold are retained in a given conformal field theory of the type studied in this article.

2273: 

2274: \subsection{Relation to NS5-brane set-ups}

2275: The relation of the above orbifold approximations to NS5-brane set-ups has been discussed in the literature. These toric abelian orbifolds of $\, \, \IC^3$ can  be

2276: mapped one-to-one to NS5-brane configurations that wrap holomorphic curves. The

2277: holomorphic curves can in turn be described by dimers that are

2278: systematically reconstructed from the abelian orbifold group. In the

2279: particular case above, one obtains hexagon tilings of the plane with 

2280: labelings determined by the weights of the GSO projection and further

2281: orbifolding. They can be determined by straightforwardly generalizing  

2282: the examples in \cite{Hanany:2005ve, Franco:2005rj}.

2283: 

2284: In the sixteen supercharge case, one can show explicitly that the non-compact

2285: Gepner model captures the near-horizon doubly scaled limit of the backreacted

2286: NS5-brane geometry (as we have recalled in the appendix).

2287:  In the case of eight supercharges, our non-compact Gepner

2288: models capture a near-horizon doubly scaled limit of the backreaction of

2289: NS5-branes wrapped on the holomorphic curves coded by the dimer corresponding

2290: to our non-compact Gepner model.

2291: 

2292: \section{Conclusions and future directions}\label{conclusions}

2293: 

2294: We have shown that the Gepner formalism for constructing modular invariant

2295: partition functions carries over to the asymptotic partition function of

2296: non-compact Gepner models. The analogy was then further developed in a

2297: discussion of the symmetry groups of the Gepner models, a classification of

2298: subgroups consistent with supersymmetry, and the discussion of how orbifolding

2299: by the maximal group can give rise to mirror models.

2300: 

2301: Secondly, we discussed the deep throat region of the non-compact Gepner models

2302: and how to obtain the chiral primary states that are normalizable at weak

2303: coupling from the conformal field theory. We then discussed Landau-Ginzburg

2304: descriptions of both compact and non-compact Gepner models and extended the

2305: existing techniques to analyze Landau-Ginzburg orbifolds such that they

2306: applied to the non-compact models under discussion. This led to an intuitive

2307: understanding of which modes become normalizable, and which modes are lifted

2308: by a momentum or winding potential. The counting of deformations becomes very

2309: tractable in the Landau-Ginzburg formalism. For completeness, we matched it

2310: onto a more intricate conformal field theory counting.

2311: 

2312: We used these results to argue that mirror symmetry can be implemented in non-compact Gepner models. When taking into account all possible deformations of the linear dilaton theory, it becomes analogous to the compact case. However, one always needs to keep in mind that the choice of deformation needs to be mirrored when discussing the deformed theories.  As expected, we saw that in conformal field theory, mirror symmetry is implemented by a change in sign of the right-moving R-charge. The systematic treatment of symmetry groups allowed us to generate infinite classes of mirror pairs.

2313: 

2314: Indeed, in non-compact Gepner models, one can cancel off positive and negative contributions to the central charges of the individual factors, thus allowing for infinite classes of models that also have a small curvature limit. In such small curvature limits, we argued that one recuperates flat space with an

2315: overall orbifold action. We identified such a limit, along with the orbifold, and showed that the conformal field theory matches with a flat space orbifold in the infinite level limit.  At finite level, we identified subtle differences in the spectrum of the conformal field theory and the toric abelian orbifold singularity. It would be interesting to find a general rule that tells us, a priori, which modes of the toric orbifold are retained in the conformal field theory. 

2316: 

2317: There are a large number of future directions that one can pursue.  For

2318: instance, one can generalize the Landau-Ginzburg models to models with

2319: fractional levels in the non-compact directions. One can also apply these

2320: conformal field theory techniques to describe the spectrum of chiral primaries

2321: and mirror theories for the heterotic string on non-compact Landau-Ginzburg

2322: orbifolds.

2323: 

2324: Another direction that we had in mind while embarking upon this investigation

2325: is the following. We have an orbifold approximation to particular non-compact

2326: Gepner models.  Setting fractional or regular (i.e. physical) branes at such

2327: toric orbifold singularities is one way to engineer interesting quiver gauge

2328: theories. The toric data allows us to compute the superpotential on the brane

2329: at the orbifold singularity using techniques that are very well developed.

2330: Extending these results to determine the worldvolume superpotentials for

2331: branes in non-compact Gepner models would be extremely interesting as such

2332: results are not yet available using the exact boundary state description of

2333: D-branes in these models.

2334: 

2335: Furthermore, Seiberg duality is well-understood in the context of the toric

2336: quiver gauge theories \cite{Feng:2000mi, Beasley:2001zp, Cachazo:2001sg}. We

2337: would like to study whether one can understand Seiberg duality for D-branes in

2338: these almost toric spaces \cite{Eguchi:2003ik, Ashok:2005py,

2339:   Fotopoulos:2005cn} microscopically, as in \cite{Murthy:2006xt,

2340:   Ashok:2007sf}. The fact that we have a microscopic description of the

2341: near-horizon doubly scaled limit of these backgrounds as well as a tunable

2342: level, should give us further computational control.

2343: 

2344: \section*{Acknowledgments}

2345: We are grateful to Agostino Butti, Eleonora Dell'Aquila, Bogdan Florea,

2346: Davide Forcella, Amihay Hanany and Ruben Minasian for helpful conversations.

2347:  Our work was supported in part

2348: by the EU under the contract MRTN-CT-2004-005104. 

2349: 

2350: \appendix

2351: 

2352: \section{Non-compact Gepner model analysis}\label{CFTanalysis}

2353: 

2354: In this appendix we relate the counting of marginal deformations that we

2355: performed

2356: in a Landau-Ginzburg language to a more elaborate enumeration of states

2357: in a standard conformal field theory formalism.

2358:  See \cite{Eguchi:2004ik}

2359: for a detailed discussion and further examples.

2360: 

2361: A state in the $(k_1,...,k_p;l_1,...,l_q)$ model (restricted to the internal

2362: conformal field theory only) is associated to the left and right charges

2363: $$r=(s_1,...,s_{p+q};n_1,...,n_p;2m_1,...,2m_q)$$

2364: $$\tilde{r}=(\tilde{s}_1,...,\tilde{s}_{p+q};\tilde{n}_1,...,\tilde{n}_p;2\tilde{m}_1,...,2\tilde{m}_q)$$

2365: as well as the compact spins $j_1,...,j_{p}$ and the non-compact spins

2366: $j_{p+1},...,j_{p+q}$.  The spins are the same on the left and on the right,

2367: since we consider diagonal partition functions. The $2 \beta_0$-orbifold

2368: imposes integral R-charges, and allows the difference of left and right

2369: charges $r-\tilde{r}$ to be an even multiple of the Gepner vector $\beta_0$.

2370: The orbifolds $\beta_i$ that align the periodicities of the fermions allow

2371: additional even differences between the left fermion numbers $s_i$ and the

2372: right fermion numbers $\tilde{s}_i$.

2373: 

2374: Chiral primary operators are chiral primaries in each conformal field theory factor. From unitarity of the non-compact Gepner models (see appendix \ref{unitarity}) it  follows that in each factor separately we satisfy the equation $\pm Q_i = 2 h_i$ for chiral (respectively anti-chiral) primaries and similarly for the right-movers. Marginality of the deformations implies that we need to satisfy the equations $$ Q = 2 \beta_0\cdot r = \pm 1 \quad \tilde{Q} = 2 \beta_0\cdot \tilde{r} = \pm 1. $$

2375: 

2376: Solving this set of equations is straightforward but tedious because of the

2377: equivalences that exist for the minimal model quantum numbers:

2378: $$n_i \equiv n_i +2k_i,\quad s_i \equiv s_i+4,\quad (j,n,s) \equiv (\frac{k-2}{2}-j,n

2379: -k,s+2)$$

2380: The same kind of equivalences hold in the

2381: non-compact factor:

2382: $$2m_i \equiv 2m_i +2l_i,\quad s_i \equiv s_i+4,\quad (j,2m,s) \equiv (\frac{k+2}{2}-j,2m

2383: -k,s+2)$$

2384:  That is one technical reason why the Landau-Ginzurg

2385: method is more efficient to count (anti)chiral operators.

2386: However it is  possible to perform the counting in each individual example that we

2387: treated in the bulk of the paper. We do this analysis example by example.

2388: 

2389: \subsection{The $(k;k)$ Model}

2390: In the $(k;k)$ model, the Gepner vector is $\beta_0=(-1,-1;1;-1)$. Looking for

2391: chiral primary operators, we find 

2392: \begin{itemize}

2393: \item $k-1$ $(c,c)$ states in the untwisted sector with charges:

2394: $$ r=(0,0;n;k-n)=\tilde{r}, \quad j_1=n, \quad j_2=k-n, \quad 0 \le n \le k-2 $$

2395: \item $k-1$ $(a,c)$ states in the first twisted sector:

2396: $$ r=(0,0;-n;-k+n), \quad \tilde{r}=r-2\beta_0,\quad j_1=n, \quad j_2=k-n,  \quad 0 \le n \le k-2 $$

2397: \item 1 $(c,c)$ state in each $\alpha$-twisted sector ($1 \le \alpha \le k-1$):

2398: $$ r=(0,0;\alpha-1;k-\alpha+1), \quad \tilde{r}=r-2\alpha \beta_0,\quad

2399: j_1=\alpha-1, \quad j_2=k-\alpha+1 $$

2400: \item 1 $(a,c)$ state in each $(\alpha+1)$-twisted sector ($1 \le \alpha \le k-1$):

2401: $$ r=(0,0;-\alpha+1;-k+\alpha-1), \quad \tilde{r}=r-2(\alpha+1) \beta_0, \quad

2402: j_1=\alpha-1, \quad j_2=k-\alpha+1 $$

2403: \end{itemize}

2404: For each $(c,c)$ state we also find a $(a,a)$ state

2405: and

2406: each $(c,a)$ state is similarly paired with an $(a,c)$ state.

2407: It is straightforward to match all these states with the ones we found more

2408: fluently in the bulk of the paper with the Landau-Ginzburg methods. 

2409: 

2410: For each state, described by its quantum numbers $r$, $\tilde{r}$, $j_1$ and

2411: $j_2$, we can write the corresponding closed string vertex

2412: operator:

2413: \be V_{r,\tilde{r}}^{j_1,j_2} = V^{j_1}_{n,s_1;\tilde{n},\tilde{s}_1}

2414: V^{j_2}_{2m,s_2;2\tilde{m},\tilde{s}_2} \,. \ee

2415: Here $V^{j_1}_{n,s_1;\tilde{n},\tilde{s}_1}$ is a vertex operator of

2416: quantum numbers $(j_1,n,s_1)$ in the minimal model, 

2417: while $V^{j_2}_{2m,s_2;2\tilde{m},\tilde{s}_2}$ is a vertex operator  

2418: of quantum numbers $(j_2,n,s_2)$ in the noncompact factor. 

2419: If we denote the asymptotic coordinates along the cylinder  as $\rho$ and

2420: $\theta$

2421:  and the bosonized complex fermion by $H$, we have the asympotic expression for the

2422:  vertex

2423: operators (see e.g. \cite{Hosomichi:2004ph}):

2424: \be V^{j}_{2m,s;2\tilde{m},\tilde{s}} = \exp\left[ \sqrt{\frac{2}{k}}

2425:   \left( (j-1)\rho +im \theta_L +i\tilde{m} \theta_R \right)

2426:   +isH_L+i\tilde{s}H_R  \right] \,.\ee

2427: 

2428: \subsection{The $(2k,2k;k)$ Model}

2429: \label{2k2kk}

2430: Let us briefly mention the subtlety that the $\beta_0$ vector of Gepner is defined in

2431: principle in all factors of the models. The reason we can consider

2432: its action separately in the internal factors only without encountering

2433: further difficulties lies in the fact that we first of all only consider even

2434: multiples of $\beta_0$, and that moreover an even multiple of $\beta_0$ is

2435: equivalent to the same vector with only non-zero internal entries, up to the

2436: vectors $\beta_i$. That is true for all three-factor models we consider (and

2437: it is therefore also true for the two-factor model when we incorporate two

2438: further flat directions). 

2439: 

2440: Having dispensed of that subtlety in comparing the two methods, we

2441:  can again find the marginal deformations directly in the non-compact Gepner model:

2442: \begin{itemize}

2443: \item The $(k-1)^2$ untwisted $(c,c)$ states have quantum numbers:

2444: $$ r=(0,0,0;a,b;c)=\tilde{r}, \quad j_1=a, \quad j_2=b, \quad j_3=c$$

2445: $$ 0 \le a,b \le 2k-2, \quad 2 \le c \le k, \quad a+b+2c=2k $$

2446: \item The $k-1$ twisted $(a,c)$ states have quantum numbers:

2447: $$r=(0,0,0;-2k+\alpha,-2k+\alpha;k-\alpha),\quad \tilde{r}=r-2\alpha \beta_0$$

2448: $$j_1=2k-\alpha, \quad j_2=2k-\alpha, \quad j_3=\alpha-k, \quad k+2 \le \alpha \le 2k \,.$$

2449: where $\alpha$ labels the twisted sector in which these states live.

2450: \end{itemize}

2451: \noindent

2452: The corresponding closed string vertex operators are given by

2453: \be V_{r,\tilde{r}}^{j_1,j_2,j_3} = V^{j_1}_{n_1,s_1;\tilde{n}_1,\tilde{s}_1}

2454: V^{j_2}_{n_2,s_2;\tilde{n}_2,\tilde{s}_2}

2455: V^{j_3}_{2m,s_3;2\tilde{m},\tilde{s}_3}\,.\ee

2456: 

2457: \subsection{The $(k; 2k, 2k)$ Model}\label{k2k2k}

2458: 

2459: The marginal deformations are written as follows:

2460: \begin{itemize}

2461: \item The $(k-1)^2$ untwisted $(c,c)$ states have quantum numbers:

2462: $$ r=(0,0,0;a;b,c)=\tilde{r}, \quad j_1=a, \quad j_2=b, \quad j_3=c $$

2463: $$ 0 \le a \le k-2, \quad 2 \le b,c \le 2k, \quad 2a+b+c=2k $$

2464: \item The $k-1$ twisted $(a,c)$ states have quantum numbers:

2465: $$ r=(0,0,0;-k+\alpha;-\alpha,-\alpha), \quad \tilde{r}=r-2\alpha \beta_0$$

2466: $$j_1=k-\alpha, \quad j_2=\alpha, \quad j_3=\alpha, \quad 2 \le \alpha \le k $$

2467: where $\alpha$ labels the twisted sectors as before.

2468: \end{itemize}

2469: \noindent

2470: The corresponding closed string vertex operators are given by:

2471: \be V_{r,\tilde{r}}^{j_1,j_2,j_3} = V^{j_1}_{n,s_1;\tilde{n},\tilde{s}_1}

2472: V^{j_2}_{2m_1,s_2;2\tilde{m}_1,\tilde{s}_2}

2473: V^{j_3}_{2m_2,s_3;2\tilde{m}_2,\tilde{s}_3}\,.\ee

2474: 

2475: For the orbifold models as well, one can perform the tedious exercise, thus

2476: affirming that the Landau-Ginzburg formalism is indeed more efficient.

2477: 

2478: \subsection{The $(3,3,3;2)$ Model}\label{3332}

2479: 

2480: There is an extra subtlety that we need to address for this model. From the

2481: conformal field theory point of view, it is easiest to 

2482: ignore the quadratic Landau-Ginzburg model with a trivial infra-red fixed

2483: point,

2484: and to work with an even number of internal conformal field theory factors. It then follows that the vector 

2485: \begin{equation}

2486: 2\beta_0=(-2,-2,-2,-2,-2;2,2,2;-2)

2487: \end{equation}

2488: in the full light-cone conformal field theory is not equivalent modulo the

2489: vectors $\beta_i$ to the vector $2 \gamma_0 =(0,0,0,0,0;2,2,2;-2)$ which

2490: represents the $g_0$ action on the Landau-Ginzburg internal conformal field

2491: theory. Yet, it can be show that the Landau-Ginzburg model does correctly

2492: count the conformal field theory chiral-chiral and chiral-anti-chiral states.

2493: Since it is only the $(c,a)$ state in the $3$-twisted sector that is crucial

2494: to us in this example, let us show how to identify that state in the conformal

2495: field theory. It corresponds to the state with charges $r=(0,0,0,0,0;0,0;2)$

2496: on the left, as indicated by the Landau-Ginzburg model. The right-moving

2497: charges are computed by observing that we are in the $3$-twisted sector. We

2498: obtain the charge (up to equivalences in the charge lattice)

2499: $\tilde{r}=(0,0,0,0,-2;0,0;0)$ (and we remain diagonal in the non-compact

2500: quantum number $j=1$). Indeed, the state with these charges is in the spectrum

2501: of the theory.

2502: 

2503: Note that we had to adjust the precise identification of the state in the conformal field theory. This is typical of the model with an even number of factors. The final vertex operator is chiral and bosonic on the left, and anti-chiral and purely made of fermions on the right.

2504: 

2505: \section{T-duality to NS5-branes revisited}\label{NS5}

2506: In this appendix, we recall the relation of the $(SU(2)/U(1) \times

2507: SL(2,R)/U(1))/\IZ_k$ coset model to the near-horizon geometry of a particular

2508: constellation of NS5-branes \cite{Ooguri:1995wj, Sfetsos:1998xd, Giveon:1999px}. It is useful to revisit this exercise because we will be able to explicitly identify the region of

2509: space-time in which the NS5-branes reside with the presence of a

2510: patch isomorphic to $\,\IC^2/\IZ_k$ in the T-dual. This fact is used as an

2511: argument in 

2512: section \ref{geometrymirror}.

2513: 

2514: We recall that the supergravity background generated by

2515: parallel NS5-branes stretching in

2516:  the $x^{\mu=0,1,2,3,4,5}$-directions in the string frame is:

2517: \begin{eqnarray}

2518: ds^2 &=& \eta_{\mu\nu} dx^{\mu} dx^{\nu} + H(x^i) dx^i dx_i \nonumber \\

2519: e^{2 \Phi}  &=& g_s^2 \, H(x^i) \nonumber \\

2520: H_{ijk} &=& - {\epsilon^l}_{ijk} \partial_l H(x^i)

2521: \end{eqnarray}

2522: where the harmonic function $H$ is determined by the positions of

2523: the NS5-branes $x^{i=6,7,8,9}_a$:

2524: \begin{eqnarray}

2525: H(x^i) &=& 1 + \sum_{a=1}^k \frac{\alpha'}{|x^i-x^i_a|^2}

2526: \end{eqnarray}

2527: We concentrate on $k$ NS5-branes spread evenly

2528: on a topologically trivial circle of coordinate 

2529: radius $\rho_0$ in the $(x^6,x^7)$

2530: plane (see figure \ref{planeALF}). We recall from 

2531: \cite{Sfetsos:1998xd, Israel:2005fn} that after the coordinate

2532: change $r \ge 0$ and $\theta \in {[} 0 , \pi/2 {]} $

2533: \begin{eqnarray}

2534: (x^6,x^7) &=& \rho_0 \cosh r \sin \theta \, (\cos \psi,\sin \psi)  \nonumber \\

2535: (x^8,x^9) &=& \rho_0 \sinh r \cos \theta \, (\cos \phi,\sin \phi)\,

2536: \label{cartcoords}

2537: \end{eqnarray}

2538: and taking a near-horizon doubly scaled limit in which $\rho_0 \rightarrow 0$

2539: and $g_s \rightarrow 0$ and in which

2540:  $\rho_0 g_s / \sqrt{\alpha'}$ (and $\alpha'$) is kept fixed \cite{Giveon:1999px},

2541: and after neglecting the localization of the NS5-branes on the circle,

2542: we obtain the NS5-brane background:

2543: \begin{eqnarray}

2544: ds^2 &=& dx^{\mu} dx_{\mu} + \alpha' k \,

2545: \left[ dr^2+d \theta^2 + \frac{\tanh^2 r \ d \phi^2 + \tan^2 \theta \ d \psi^2

2546: }{1+\tan^2 \theta \tanh^2 r} \right] \, ,

2547: \nonumber \\

2548: e^{ 2 \Phi} &=& \frac{g_\textsc{eff}^2 }{\cosh^2 r - \sin^2 \theta} \, ,

2549:  \nonumber \\

2550: B  &=& \frac{\alpha' k }{1+\tan^2 \theta \tanh^2 r} \ d \phi \wedge d \psi \,

2551: \label{tdualNS5geom}

2552: \end{eqnarray}

2553: where the effective string coupling constant is 

2554: \begin{equation} 

2555: g_\textsc{eff} = \frac{\sqrt{k \alpha'}g_s}{\rho_0}. 

2556: \end{equation}

2557: We refer to \cite{Israel:2005fn} for the detailed calculation.

2558: 

2559: A first observation to make is that the NS5-branes are located

2560: at $r=0$ and $\theta=\pi/2$. Moreover, the fact that the coordinate

2561: radius of the NS5-brane ring has gone to zero, has been compensated by the

2562: fact that the harmonic function and radial metric

2563: component has blown up 

2564: and in such a way that the interior of the NS5-brane ring still has a

2565: proper size of order $\sqrt{k \alpha'}$. It is difficult to press together

2566: NS5-branes. 

2567: That's an important aspect of the solution, since in the interior, near $r=0$

2568: and $\theta=0$, we simply find a portion of flat space.

2569: 

2570: After T-duality in the angular direction $\psi$ around which

2571: the NS5-branes have been sprinkled, we found the T-dual geometry:

2572: \begin{eqnarray}

2573: ds^2 &=& \alpha' k \left[ d r^2 + \tanh^2 r \, \left( \frac{d\chi}{k}\right)^2

2574: + d\theta^2 + \mathrm{cotan}^2 \theta \, \left( \frac{d\chi}{k}-d\phi \right)^2 \right], \nonumber \\

2575: e^{2 \Psi}&=& \frac{g_\textsc{eff}}{k} \ \frac{1}{\cosh^2 r \ \sin^2 \theta}

2576: \end{eqnarray}

2577: where $\chi$ is an angle parameterizing the T-dual circle.

2578: The background is recognized as a vector $\IZ_k$ orbifold of the 

2579: product of coset conformal field theory geometries

2580: $SU(2)_k /U(1) \, \times SL(2,R)_k /U(1)$.

2581: 

2582: We want to add a couple of remarks to the analysis in \cite{Israel:2005fn}

2583:  to which

2584: we refer the reader for further comments. In

2585: particular, we first note the  singularity in both the metric

2586: and the dilaton at $\theta=0$. The origin of this singularity is the fact that

2587: we have performed an angular T-duality with a fixed point. Around the fixed

2588: point, as we have pointed out, the original background behaves like flat

2589: space. Thus, the T-dual behavior is recognized as the same type of

2590: singularity that one obtains in T-dualizing flat space in cylindrical

2591: coordinates. Since the original theory was regular near the origin, we expect the T-dual

2592: theory to behave well at this point as well. 

2593: 

2594: On the other hand, we observe that the region at $\theta= \pi/2$ and $r=0$

2595: where the NS5-branes resides has locally become identical to a $\,\IC^2/\IZ_k$

2596: orbifold. Thus, it is the orbifold singularity of order $k$ that codes the

2597: presence of the NS5-branes in the T-dual. 

2598: 

2599: We can make the above analysis  more precise by performing the

2600: following mental exercise. Cut out from the $6-7$ plane a little disc at the

2601: origin. Topologically, the space transverse to the NS5-branes has become

2602: $\IR^3 \times S^1$, with NS5-branes spread on the $S^1$ (see figure \ref{planeALF}).

2603: \begin{figure}

2604: \centering

2605: \includegraphics[width=0.8\textwidth]{planeALF}

2606: \caption{NS5-branes spread on a topologically trivial circle (drawn on the left). 

2607: When we cut out a little disc at the center, the configuration becomes topologically equivalent to NS5-branes spread on a topologically non-trivial circle (drawn on the right).}

2608: \label{planeALF}

2609: \end{figure}

2610: When we neglect the localization of the NS5-branes on the circle, the configuration is T-dual to

2611: an ALF space \cite{Ooguri:1995wj}\cite{Kutasov:1995te}, which in the particular case where the

2612: NS5-branes coincide in $\IR^3$ develops a $\,\IC^2/\IZ_k$ orbifold singularity. 

2613: That reasoning is another version of the one above, which uses local

2614:  fiberwise T-duality. Localizing the NS5-branes in

2615:  the original geometry on the circle is known to be equivalent to taking into

2616:  account worldsheet instanton corrections for the ALF space  \cite{Gregory:1997te}\cite{Tong:2002rq}\cite{Okuyama:2005gx}, and this

2617:  equivalence is conjectured to be valid for the NS5-branes on a topologically

2618:  trivial circle as well \cite{Israel:2005fn}.

2619: 

2620: We take away several facts from this analysis. Firstly, the $\IZ_k$

2621: orbifold projection acting on the  $SU(2)/U(1) \times

2622: SL(2,R)/U(1)$ coset has fixed points and gives rise locally

2623: to a physical $\IC^2/\IZ_k$ singularity that is T-dual to the presence of 

2624: $k$ NS5-branes. 

2625: And secondly, 

2626: in the geometry of the coset conformal field theory, the $\IZ_k$ GSO

2627: projection acts geometrically on the cigar coordinates,

2628: and this in contrast to the fact that the GSO projection in Gepner

2629: models does not orbifold the coordinates

2630:  of weighted projective space. We recalled

2631: the above detailed example because we will use 

2632: these facts as arguments in section \ref{geometrymirror} and they are particularly

2633: manifest in the above example.

2634: 

2635: \section{The consequences of unitarity}\label{unitarity}

2636: 

2637: We list in this appendix a number of properties that hold in unitary modules

2638: of the $N=2$ superconformal algebra, irrespective of the value of the central

2639: charge (and in particular also when the central charge is equal or greater

2640: than three). Our conventions for the $N=2$ superconformal algebra coincide

2641: with those of \cite{Polchinski:1998rr, Lerche:1989uy}.

2642: \subsection*{Properties}

2643: The following properties hold -- the proofs in the references that

2644: we give depends only on the unitarity of the 

2645: representation spaces as we checked on a case-by-case basis:

2646: \begin{enumerate}

2647: \item All states in the NS-sector with conformal dimension $h$ and R-charge

2648: $Q$ satisfy the inequality $h \ge |Q|/2$ \cite{Lerche:1989uy}.

2649: \item

2650: An NS-sector field of conformal dimension $h$ and R-charge $Q$ is a

2651: chiral primary 

2652: if and only if $h=Q/2$ and anti-chiral if and only if $h=-Q/2$

2653: \cite{Lerche:1989uy}. 

2654: \item Chiral primary fields satisfy the inequality $h \le c/6$ \cite{Lerche:1989uy}.

2655: \item Ramond sector ground states have R-charges $Q$ in the range

2656: $ -c/6 \le Q \le +c/6$ \cite{Lerche:1989uy}.

2657: \item An NS-sector state $|\phi \rangle$ 

2658: satisfies the equations 

2659: $G^+_{-l-1/2} | \phi \rangle = 0 = G^-_{l+1/2} |\phi \rangle$

2660: if and only if its conformal dimension

2661:  $h$ and R-charge $Q$ satisfy the relation $h=(l+1/2) Q-c(l^2+l)/6$ \cite{Vafa:1989xc}.

2662: \item A chiral ring exists \cite{Lerche:1989uy}.

2663: \end{enumerate}

2664: \subsection*{Remarks}

2665: Note that we performed a non-trivial exercise. For instance, for conformal

2666: field theories with central charge $c<3$, it is argued in \cite{Lerche:1989uy}

2667:  that there

2668: always exist a chiral primary field in the conformal field theory

2669: with conformal dimension $h=c/6$. That is a consequence of the existence of 

2670: the unit operator in the theory (combined with spectral flow).

2671:  In other words, there is a normalizable

2672: $SL(2,R)$ invariant ground state in these conformal field theories. That is

2673: not  a consequence of unitarity only, and it does {\em not}

2674:  hold generically for unitary $N=2$ superconformal field theories. In fact, the situation is

2675: subtle. For instance there are examples of bulk $N=2$ 

2676: superconformal field theories with central charge $c>3$ for which the unit

2677: operator does not exist when the conformal field theory is

2678: defined on a Riemann surface without boundary,

2679: while it does exist for instance on the boundary of a

2680: disc with particular boundary

2681: conditions (see e.g. \cite{Eguchi:2003ik}\cite{Israel:2004jt}). 

2682: It is due to these kind of subtleties that it is useful to have a

2683: (partial) list of properties that we can indiscriminantly use for unitary 

2684: $N=2$ theories with any central charge.

2685: 

2686:  \begin{thebibliography}{99} 

2687:  

2688: %\cite{Yau:1998yt}\cite{Greene:1997ty}

2689: \bibitem{Yau:1998yt}

2690:   S.~T.~Yau,

2691:   ``Mirror symmetry I,''

2692: %\href{http://www.slac.stanford.edu/spires/find/hep/www?irn=4308182}{SPIRES entry}

2693: {\it  Providence, USA: AMS (1998) 444 p}

2694: %\cite{Greene:1997ty}

2695: \bibitem{Greene:1997ty}

2696:   B.~Greene and S.~T.~Yau,

2697:   ``Mirror symmetry II,''

2698: %\href{http://www.slac.stanford.edu/spires/find/hep/www?irn=4019032}{SPIRES entry}

2699: {\it  Providence, USA: AMS (1997) 844 p. (AMS/IP studies in advanced mathematics. 1)}

2700: 

2701:  %\cite{Greene:1990ud}

2702: \bibitem{Greene:1990ud}

2703:   B.~R.~Greene and M.~R.~Plesser,

2704:   %``DUALITY IN CALABI-YAU MODULI SPACE,''

2705:   Nucl.\ Phys.\  B {\bf 338}, 15 (1990).

2706:   %%CITATION = NUPHA,B338,15;%%

2707: 

2708: %\cite{Gepner:1987qi}

2709: \bibitem{Gepner:1987qi}

2710:   D.~Gepner,

2711:   %``Space-Time Supersymmetry in Compactified String Theory and Superconformal

2712:   %Models,''

2713:   Nucl.\ Phys.\  B {\bf 296} (1988) 757.

2714:   %%CITATION = NUPHA,B296,757;%%

2715: 

2716: %\cite{Candelas:1990rm}

2717: \bibitem{Candelas:1990rm}

2718:   P.~Candelas, X.~C.~De La Ossa, P.~S.~Green and L.~Parkes,

2719:   %``A pair of Calabi-Yau manifolds as an exactly soluble superconformal

2720:   %theory,''

2721:   Nucl.\ Phys.\  B {\bf 359} (1991) 21.

2722:   %%CITATION = NUPHA,B359,21;%%

2723: 

2724: %\cite{Berglund:1991pp}

2725: \bibitem{Berglund:1991pp}

2726:   P.~Berglund and T.~Hubsch,

2727:   %``A generalized construction of mirror manifolds,''

2728:   Nucl.\ Phys.\  B {\bf 393}, 377 (1993)

2729:   [arXiv:hep-th/9201014].

2730:   %%CITATION = NUPHA,B393,377;%%

2731: 

2732: \bibitem{Batyrev}

2733: V.~Batyrev,

2734: %"Dual Polyhedra and Mirror Symmetry for Calabi-Yau Hypersurfaces in Toric Varieties,"

2735: arXiv:alg-geom/9310003.

2736: 

2737: %\cite{Chiang:1999tz}

2738: \bibitem{Chiang:1999tz}

2739:   T.~M.~Chiang, A.~Klemm, S.~T.~Yau and E.~Zaslow,

2740:   %``Local mirror symmetry: Calculations and interpretations,''

2741:   Adv.\ Theor.\ Math.\ Phys.\  {\bf 3}, 495 (1999)

2742:   [arXiv:hep-th/9903053].

2743:   %%CITATION = 00203,3,495;%%

2744: 

2745: %\cite{Witten:1993yc}

2746: \bibitem{Witten:1993yc}

2747:   E.~Witten,

2748:   %``Phases of N = 2 theories in two dimensions,''

2749:   Nucl.\ Phys.\  B {\bf 403}, 159 (1993)

2750:   [arXiv:hep-th/9301042].

2751:   %%CITATION = NUPHA,B403,159;%%

2752: 

2753: %\cite{Hori:2000kt}\cite{Hori:2002cd}\cite{Hori:2001ax}

2754: \bibitem{Hori:2000kt}

2755:   K.~Hori and C.~Vafa,

2756:   %``Mirror symmetry,''

2757:   arXiv:hep-th/0002222.

2758:   %%CITATION = HEP-TH/0002222;%%

2759:   %\cite{Hori:2002cd}\cite{Hori:2001ax}

2760: \bibitem{Hori:2002cd}

2761:   K.~Hori and A.~Kapustin,

2762:   %``Worldsheet descriptions of wrapped NS five-branes,''

2763:   JHEP {\bf 0211}, 038 (2002)

2764:   [arXiv:hep-th/0203147].

2765:   %%CITATION = JHEPA,0211,038;%%

2766: %\cite{Eguchi:2004yi}

2767: \bibitem{Eguchi:2004yi}

2768:   T.~Eguchi and Y.~Sugawara,

2769:   %``SL(2,R)/U(1) supercoset and elliptic genera of non-compact Calabi-Yau

2770:   %manifolds,''

2771:   JHEP {\bf 0405}, 014 (2004)

2772:   [arXiv:hep-th/0403193].

2773:   %%CITATION = JHEPA,0405,014;%%

2774: 

2775: %\cite{Giveon:1999px}\cite{Eguchi:2004yi}

2776: \bibitem{Giveon:1999px}

2777:   A.~Giveon and D.~Kutasov,

2778:   %``Little string theory in a double scaling limit,''

2779:   JHEP {\bf 9910} (1999) 034

2780:   [arXiv:hep-th/9909110].

2781:   %%CITATION = JHEPA,9910,034;%%

2782: 

2783: %\cite{Hori:2001ax}

2784: \bibitem{Hori:2001ax}

2785:   K.~Hori and A.~Kapustin,

2786:   %``Duality of the fermionic 2d black hole and N = 2 Liouville theory as

2787:   %mirror symmetry,''

2788:   JHEP {\bf 0108}, 045 (2001)

2789:   [arXiv:hep-th/0104202].

2790:   %%CITATION = JHEPA,0108,045;%%

2791: 

2792: %\cite{Tong:2003ik}

2793: \bibitem{Tong:2003ik}

2794:   D.~Tong,

2795:   %``Mirror mirror on the wall: On two-dimensional black holes and Liouville

2796:   %theory,''

2797:   JHEP {\bf 0304}, 031 (2003)

2798:   [arXiv:hep-th/0303151].

2799:   %%CITATION = JHEPA,0304,031;%%

2800: 

2801: %\cite{Israel:2004jt}

2802: \bibitem{Israel:2004jt}

2803:   D.~Israel, A.~Pakman and J.~Troost,

2804:   %``D-branes in N = 2 Liouville theory and its mirror,''

2805:   Nucl.\ Phys.\  B {\bf 710}, 529 (2005)

2806:   [arXiv:hep-th/0405259].

2807:   %%CITATION = NUPHA,B710,529;%%

2808: 

2809: %\cite{Greene:1988ut}\cite{Vafa:1989xc}\cite{Intriligator:1990ua}

2810: \bibitem{Greene:1988ut}

2811:   B.~R.~Greene, C.~Vafa and N.~P.~Warner,

2812:   %``Calabi-Yau Manifolds and Renormalization Group Flows,''  

2813:   Nucl.\ Phys.\  B {\bf 324} (1989) 371.

2814:   %%CITATION = NUPHA,B324,371;%%

2815: %\cite{Vafa:1989xc}\cite{Intriligator:1990ua}

2816: 

2817: %\cite{Candelas:1989hd}

2818: \bibitem{Candelas:1989hd}

2819:   P.~Candelas, M.~Lynker and R.~Schimmrigk,

2820:   %``Calabi-Yau Manifolds in Weighted P(4),''

2821:   Nucl.\ Phys.\  B {\bf 341}, 383 (1990).

2822:   %%CITATION = NUPHA,B341,383;%%

2823: 

2824: \bibitem{Vafa:1989xc}

2825:   C.~Vafa,

2826:   %``String Vacua and Orbifoldized L-G Models,''

2827:   Mod.\ Phys.\ Lett.\  A {\bf 4}, 1169 (1989).

2828:   %%CITATION = MPLAE,A4,1169;%%

2829: %\cite{Intriligator:1990ua}

2830: \bibitem{Intriligator:1990ua}

2831:   K.~A.~Intriligator and C.~Vafa,

2832:   %``LANDAU-GINZBURG ORBIFOLDS,''

2833:   Nucl.\ Phys.\  B {\bf 339} (1990) 95.

2834:   %%CITATION = NUPHA,B339,95;%%

2835: 

2836: %\cite{Eguchi:2000tc}\cite{Murthy:2003es}\cite{Hanany:2002ev}\cite{Eguchi:2004yi}\cite{Israel:2004ir}

2837: \bibitem{Eguchi:2000tc}

2838:   T.~Eguchi and Y.~Sugawara,

2839:   %``Modular invariance in superstring on Calabi-Yau n-fold with A-D-E

2840:   %singularity,''

2841:   Nucl.\ Phys.\  B {\bf 577}, 3 (2000)

2842:   [arXiv:hep-th/0002100].

2843:   %%CITATION = NUPHA,B577,3;%%

2844: 

2845: %\cite{Hanany:2002ev}

2846: \bibitem{Hanany:2002ev}

2847:   A.~Hanany, N.~Prezas and J.~Troost,

2848:   %``The partition function of the two-dimensional black hole conformal  field

2849:   %theory,''

2850:   JHEP {\bf 0204}, 014 (2002)

2851:   [arXiv:hep-th/0202129].

2852:   %%CITATION = JHEPA,0204,014;%%

2853: %\cite{Murthy:2003es}

2854: \bibitem{Murthy:2003es}

2855:   S.~Murthy,

2856:   %``Notes on non-critical superstrings in various dimensions,''

2857:   JHEP {\bf 0311}, 056 (2003)

2858:   [arXiv:hep-th/0305197].

2859:   %%CITATION = JHEPA,0311,056;%%

2860: 

2861: %\cite{Israel:2004ir}

2862: \bibitem{Israel:2004ir}

2863:   D.~Israel, C.~Kounnas, A.~Pakman and J.~Troost,

2864:   %``The partition function of the supersymmetric two-dimensional black hole

2865:   %and little string theory,''

2866:   JHEP {\bf 0406}, 033 (2004)

2867:   [arXiv:hep-th/0403237].

2868:   %%CI

2869: 

2870: %\cite{Ooguri:1995wj}

2871: \bibitem{Ooguri:1995wj}

2872:   H.~Ooguri and C.~Vafa,

2873:   %``Two-Dimensional Black Hole and Singularities of CY Manifolds,''

2874:   Nucl.\ Phys.\  B {\bf 463}, 55 (1996)

2875:   [arXiv:hep-th/9511164].

2876:   %%CITATION = NUPHA,B463,55;%%

2877: 

2878: %\cite{Maldacena:2000kv}

2879: \bibitem{Maldacena:2000kv}

2880:   J.~M.~Maldacena, H.~Ooguri and J.~Son,

2881:   %``Strings in AdS(3) and the SL(2,R) WZW model. II: Euclidean black hole,''

2882:   J.\ Math.\ Phys.\  {\bf 42}, 2961 (2001)

2883:   [arXiv:hep-th/0005183].

2884:   %%CITATION = JMAPA,42,2961;%%

2885: 

2886: %\cite{Gepner:1986hr}

2887: \bibitem{Gepner:1986hr}

2888:   D.~Gepner and Z.~a.~Qiu,

2889:   %``Modular Invariant Partition Functions for Parafermionic Field Theories,''

2890:   Nucl.\ Phys.\  B {\bf 285} (1987) 423.

2891:   %%CITATION = NUPHA,B285,423;%%

2892: 

2893: %\cite{Greene:1996cy}

2894: \bibitem{Greene:1996cy}

2895:   B.~R.~Greene,

2896:   %``String theory on Calabi-Yau manifolds,''

2897:   arXiv:hep-th/9702155.

2898:   %%CITATION = HEP-TH/9702155;%%

2899: 

2900: %\cite{Maldacena:2001ky}

2901: \bibitem{Maldacena:2001ky}

2902:   J.~M.~Maldacena, G.~W.~Moore and N.~Seiberg,

2903:   %``Geometrical interpretation of D-branes in gauged WZW models,''

2904:   JHEP {\bf 0107} (2001) 046

2905:   [arXiv:hep-th/0105038].

2906:   %%CITATION = JHEPA,0107,046;%%

2907: 

2908: %\cite{Dixon:1989cg}

2909: \bibitem{Dixon:1989cg}

2910:   L.~J.~Dixon, M.~E.~Peskin and J.~D.~Lykken,

2911:   %``N=2 Superconformal Symmetry and SO(2,1) Current Algebra,''

2912:   Nucl.\ Phys.\  B {\bf 325}, 329 (1989).

2913:   %%CITATION = NUPHA,B325,329;%%

2914: 

2915: %\cite{Gukov:1999ya}

2916: \bibitem{Gukov:1999ya}

2917:   S.~Gukov, C.~Vafa and E.~Witten,

2918:   %``CFT's from Calabi-Yau four-folds,''

2919:   Nucl.\ Phys.\  B {\bf 584} (2000) 69

2920:   [Erratum-ibid.\  B {\bf 608} (2001) 477]

2921:   [arXiv:hep-th/9906070].

2922:   %%CITATION = NUPHA,B584,69;%%

2923: %\cite{Shapere:1999xr}

2924: \bibitem{Shapere:1999xr}

2925:   A.~D.~Shapere and C.~Vafa,

2926:   %``BPS structure of Argyres-Douglas superconformal theories,''

2927:   arXiv:hep-th/9910182.

2928:   %%CITATION = HEP-TH/9910182;%%

2929: 

2930: %\cite{Martinec:1988zu}\cite{Vafa:1988uu}\cite{Witten:1993jg}

2931: \bibitem{Martinec:1988zu}

2932:   E.~J.~Martinec,

2933:   %``Algebraic Geometry and Effective Lagrangians,''

2934:   Phys.\ Lett.\  B {\bf 217} (1989) 431.

2935:   %%CITATION = PHLTA,B217,431;%%

2936: %\cite{Vafa:1988uu}

2937: \bibitem{Vafa:1988uu}

2938:   C.~Vafa and N.~P.~Warner,

2939:   %``Catastrophes and the Classification of Conformal Theories,''

2940:   Phys.\ Lett.\  B {\bf 218} (1989) 51.

2941:   %%CITATION = PHLTA,B218,51;%%

2942: 

2943: %\cite{Witten:1993jg}

2944: \bibitem{Witten:1993jg}

2945:   E.~Witten,

2946:   %``On the Landau-Ginzburg description of N=2 minimal models,''

2947:   Int.\ J.\ Mod.\ Phys.\  A {\bf 9}, 4783 (1994)

2948:   [arXiv:hep-th/9304026].

2949:   %%CITATION = IMPAE,A9,4783;%%

2950: 

2951: %\cite{Rocek:1991ps}

2952: \bibitem{Rocek:1991ps}

2953:   M.~Rocek and E.~P.~Verlinde,

2954:   %``Duality, quotients, and currents,''

2955:   Nucl.\ Phys.\  B {\bf 373}, 630 (1992)

2956:   [arXiv:hep-th/9110053].

2957:   %%CITATION = NUPHA,B373,630;%%

2958: 

2959: %\cite{Lerche:1989uy}

2960: \bibitem{Lerche:1989uy}

2961:   W.~Lerche, C.~Vafa and N.~P.~Warner,

2962:   %``Chiral Rings in N=2 Superconformal Theories,''

2963:   Nucl.\ Phys.\  B {\bf 324}, 427 (1989).

2964:   %%CITATION = NUPHA,B324,427;%%

2965: 

2966: %\cite{Ghoshal:1993qt}\cite{Hanany:1994fi}

2967: \bibitem{Ghoshal:1993qt}

2968:   D.~Ghoshal and S.~Mukhi,

2969:   %``Topological Landau-Ginzburg model of two-dimensional string theory,''

2970:   Nucl.\ Phys.\  B {\bf 425} (1994) 173

2971:   [arXiv:hep-th/9312189].

2972:   %%CITATION = NUPHA,B425,173;%%

2973: %\cite{Hanany:1994fi}

2974: \bibitem{Hanany:1994fi}

2975:   A.~Hanany, Y.~Oz and M.~Ronen Plesser,

2976:   %``Topological Landau-Ginzburg Formulation And Integrable Structure Of 2-D

2977:   %String Theory,''

2978:   Nucl.\ Phys.\  B {\bf 425} (1994) 150

2979:   [arXiv:hep-th/9401030].

2980:   %%CITATION = NUPHA,B425,150;%%

2981: 

2982: %\cite{Ghoshal:1995wm}

2983: \bibitem{Ghoshal:1995wm}

2984:   D.~Ghoshal and C.~Vafa,

2985:   %``C = 1 String As The Topological Theory Of The Conifold,''

2986:   Nucl.\ Phys.\  B {\bf 453}, 121 (1995)

2987:   [arXiv:hep-th/9506122].

2988:   %%CITATION = NUPHA,B453,121;%%

2989: 

2990:    %\cite{Aspinwall:1994rg}

2991: \bibitem{Aspinwall:1994rg}

2992:   P.~S.~Aspinwall and D.~R.~Morrison,

2993:   %``String theory on K3 surfaces,''

2994:   arXiv:hep-th/9404151.

2995:   %%CITATION = HEP-TH/9404151;%%

2996: 

2997: %\cite{Lerche:2000uy}

2998: \bibitem{Lerche:2000uy}

2999:   W.~Lerche,

3000:   %``On a boundary CFT description of nonperturbative N = 2 Yang-Mills

3001:   %theory,''

3002:   arXiv:hep-th/0006100.

3003:   %%CITATION = HEP-TH/0006100;%%

3004: 

3005: %\cite{Eguchi:2004ik}

3006: \bibitem{Eguchi:2004ik}

3007:   T.~Eguchi and Y.~Sugawara,

3008:   %``Conifold type singularities, N = 2 Liouville and SL(2,R)/U(1) theories,''

3009:   JHEP {\bf 0501} (2005) 027

3010:   [arXiv:hep-th/0411041].

3011:   %%CITATION = JHEPA,0501,027;%%

3012: 

3013: %\cite{Kutasov:1995te}

3014: \bibitem{Kutasov:1995te}

3015:   D.~Kutasov,

3016:   %``Orbifolds and Solitons,''

3017:   Phys.\ Lett.\  B {\bf 383}, 48 (1996)

3018:   [arXiv:hep-th/9512145].

3019:   %%CITATION = PHLTA,B383,48;%%

3020: 

3021: %\cite{Aspinwall:1994ev}

3022: \bibitem{Aspinwall:1994ev}

3023:   P.~S.~Aspinwall,

3024:   %``Resolution of orbifold singularities in string theory,''

3025:   arXiv:hep-th/9403123.

3026:   %%CITATION = HEP-TH/9403123;%%

3027: 

3028: %\cite{Lust:2006zh}

3029: \bibitem{lustetal}

3030:   D.~Lust, S.~Reffert, E.~Scheidegger and S.~Stieberger,

3031:   %``Resolved toroidal orbifolds and their orientifolds,''

3032:   arXiv:hep-th/0609014.

3033:   %%CITATION = HEP-TH/0609014;%%

3034: 

3035: %\cite{Hanany:2005ve}\cite{Franco:2005rj}

3036: \bibitem{Hanany:2005ve}

3037:   A.~Hanany and K.~D.~Kennaway,

3038:   %``Dimer models and toric diagrams,''

3039:   arXiv:hep-th/0503149.

3040:   %%CITATION = HEP-TH/0503149;%%

3041: %\cite{Franco:2005rj}

3042: \bibitem{Franco:2005rj}

3043:   S.~Franco, A.~Hanany, K.~D.~Kennaway, D.~Vegh and B.~Wecht,

3044:   %``Brane dimers and quiver gauge theories,''

3045:   JHEP {\bf 0601} (2006) 096

3046:   [arXiv:hep-th/0504110].

3047:   %%CITATION = JHEPA,0601,096;%%

3048: 

3049: %\cite{Feng:2000mi}\cite{Beasley:2001zp}\cite{Cachazo:2001sg}

3050: \bibitem{Feng:2000mi}

3051:   B.~Feng, A.~Hanany and Y.~H.~He,

3052:   %``D-brane gauge theories from toric singularities and toric duality,''

3053:   Nucl.\ Phys.\  B {\bf 595} (2001) 165

3054:   [arXiv:hep-th/0003085].

3055:   %%CITATION = NUPHA,B595,165;%%

3056: %\cite{Beasley:2001zp}\cite{Cachazo:2001sg}

3057: \bibitem{Beasley:2001zp}

3058:   C.~E.~Beasley and M.~R.~Plesser,

3059:   %``Toric duality is Seiberg duality,''

3060:   JHEP {\bf 0112} (2001) 001

3061:   [arXiv:hep-th/0109053].

3062:   %%CITATION = JHEPA,0112,001;%%

3063: %\cite{Cachazo:2001sg}

3064: \bibitem{Cachazo:2001sg}

3065:   F.~Cachazo, B.~Fiol, K.~A.~Intriligator, S.~Katz and C.~Vafa,

3066:   %``A geometric unification of dualities,''

3067:   Nucl.\ Phys.\  B {\bf 628} (2002) 3

3068:   [arXiv:hep-th/0110028].

3069:   %%CITATION = NUPHA,B628,3;%%

3070: 

3071: %\cite{Eguchi:2003ik}

3072: \bibitem{Eguchi:2003ik}

3073:   T.~Eguchi and Y.~Sugawara,

3074:   %``Modular bootstrap for boundary N = 2 Liouville theory,''

3075:   JHEP {\bf 0401}, 025 (2004)

3076:   [arXiv:hep-th/0311141].

3077:   %%CITATION = JHEPA,0401,025;%%

3078: 

3079: %\cite{Ashok:2005py}

3080: \bibitem{Ashok:2005py}

3081:   S.~K.~Ashok, S.~Murthy and J.~Troost,

3082:   %``D-branes in non-critical superstrings and minimal super Yang-Mills in

3083:   %various dimensions,''

3084:   Nucl.\ Phys.\  B {\bf 749}, 172 (2006)

3085:   [arXiv:hep-th/0504079].

3086:   %%CITATION = NUPHA,B749,172;%%

3087: 

3088: %\cite{Fotopoulos:2005cn}

3089: \bibitem{Fotopoulos:2005cn}

3090:   A.~Fotopoulos, V.~Niarchos and N.~Prezas,

3091:   %``D-branes and SQCD in non-critical superstring theory,''

3092:   JHEP {\bf 0510}, 081 (2005)

3093:   [arXiv:hep-th/0504010].

3094:   %%CITATION = JHEPA,0510,081;%%

3095: 

3096: %\cite{Murthy:2006xt}

3097: \bibitem{Murthy:2006xt}

3098:   S.~Murthy and J.~Troost,

3099:   %``D-branes in non-critical superstrings and duality in N = 1 gauge theories

3100:   %with flavor,''

3101:   JHEP {\bf 0610}, 019 (2006)

3102:   [arXiv:hep-th/0606203].

3103:   %%CITATION = JHEPA,0610,019;%%

3104: 

3105: %\cite{Ashok:2007sf}

3106: \bibitem{Ashok:2007sf}

3107:   S.~K.~Ashok, S.~Murthy and J.~Troost,

3108:   %``D-branes in unoriented non-critical strings and duality in SO(N) and Sp(N)

3109:   %gauge theories,''

3110:   JHEP {\bf 0706}, 047 (2007)

3111:   [arXiv:hep-th/0703148].

3112:   %%CITATION = JHEPA,0706,047;%%

3113: 

3114: 

3115: %\cite{Hosomichi:2004ph}

3116: \bibitem{Hosomichi:2004ph}

3117:   K.~Hosomichi,

3118:   %``N = 2 Liouville theory with boundary,''

3119:   JHEP {\bf 0612} (2006) 061

3120:   [arXiv:hep-th/0408172].

3121:   %%CITATION = JHEPA,0612,061;%%

3122: 

3123: %\cite{Sfetsos:1998xd}

3124: \bibitem{Sfetsos:1998xd}

3125:   K.~Sfetsos,

3126:   %``Branes for Higgs phases and exact conformal field theories,''

3127:   JHEP {\bf 9901} (1999) 015

3128:   [arXiv:hep-th/9811167].

3129:   %%CITATION = JHEPA,9901,015;%%

3130: 

3131: %\cite{Israel:2005fn}

3132: \bibitem{Israel:2005fn}

3133:   D.~Israel, A.~Pakman and J.~Troost,

3134:   %``D-branes in little string theory,''

3135:   Nucl.\ Phys.\  B {\bf 722}, 3 (2005)

3136:   [arXiv:hep-th/0502073].

3137:   %%CITATION = NUPHA,B722,3;%%

3138: 

3139: %\cite{Gregory:1997te}\cite{Tong:2002rq}\cite{Okuyama:2005gx}

3140: \bibitem{Gregory:1997te}

3141:   R.~Gregory, J.~A.~Harvey and G.~W.~Moore,

3142:   %``Unwinding strings and T-duality of Kaluza-Klein and H-monopoles,''

3143:   Adv.\ Theor.\ Math.\ Phys.\  {\bf 1}, 283 (1997)

3144:   [arXiv:hep-th/9708086].

3145:   %%CITATION = 00203,1,283;%%

3146: %\cite{Tong:2002rq}\cite{Okuyama:2005gx}

3147: \bibitem{Tong:2002rq}

3148:   D.~Tong,

3149:   %``NS5-branes, T-duality and worldsheet instantons,''

3150:   JHEP {\bf 0207}, 013 (2002)

3151:   [arXiv:hep-th/0204186].

3152:   %%CITATION = JHEPA,0207,013;%%

3153: %\cite{Okuyama:2005gx}

3154: \bibitem{Okuyama:2005gx}

3155:   K.~Okuyama,

3156:   %``Linear sigma models of H and KK monopoles,''

3157:   JHEP {\bf 0508} (2005) 089

3158:   [arXiv:hep-th/0508097].

3159:   %%CITATION = JHEPA,0508,089;%%

3160: 

3161: %\cite{Polchinski:1998rr}

3162: \bibitem{Polchinski:1998rr}

3163:   J.~Polchinski,

3164:   ``String theory. Vol. 2: Superstring theory and beyond,''

3165: %\href{http://www.slac.stanford.edu/spires/find/hep/www?irn=4634802}{SPIRES entry}

3166: {\it  Cambridge, UK: Univ. Pr. (1998) 531 p}

3167: 

3168:  \end{thebibliography}

3169: \end{document}

3170: 

3171: