1: \documentclass[12pt,epsf]{article}

2: \usepackage{amssymb,amscd}

3: \usepackage{graphics}

4: \usepackage{epsfig}

5: 

6: \def\baselinestretch{1.2}

7: \parskip 6 pt

8: 

9: \textheight 22.5 cm

10: %%\topmargin -.5 cm

11: \topmargin -.8 cm

12: %\topmargin -1.5cm

13: \textwidth 16cm

14: \oddsidemargin 0 in

15: \evensidemargin 0 in

16: 

17: 

18: 

19: \DeclareFontFamily{U}{rsf}{}

20: \DeclareFontShape{U}{rsf}{m}{n}{

21:   <5> <6> rsfs5 <7> <8> <9> rsfs7 <10-> rsfs10}{}

22: \DeclareMathAlphabet\Scr{U}{rsf}{m}{n}

23: 

24: 

25: 

26: %

27: %

28: % Putting citations together

29: %

30: \catcode`\@=11

31: %------------

32: %NAME: cite.sty

33: %

34: %              Replacement for \@citex

35: %

36: %     Allow, but strongly discourage, line breaks within a long

37: %     series of citations.  Compress lists of successive numbers

38: %     to one number range, e.g., 5,6,7,8,9 --> 5--9.  Compatible

39: %     with versions of \@cite that use exponents.

40: %              -- Donald Arseneau  1989

41: %

42: \def\@citex[#1]#2{%

43: \if@filesw \immediate \write \@auxout {\string \citation {#2}}\fi

44: \@tempcntb\m@ne \let\@h@ld\relax \def\@citea{}%

45: \@cite{%

46:   \@for \@citeb:=#2\do {%

47:     \@ifundefined {b@\@citeb}%

48:       {\@h@ld\@citea\@tempcntb\m@ne{\bf ?}%

49:       \@warning {Citation `\@citeb ' on page \thepage \space undefined}}%

50: %   else

51:       {\@tempcnta\@tempcntb \advance\@tempcnta\@ne%

52:       \@tempcntb\number\csname b@\@citeb \endcsname \relax%

53:       \ifnum\@tempcnta=\@tempcntb %Number follows previous--hold on to it

54:         \ifx\@h@ld\relax%

55: %         % first pair of successive

56:           \edef \@h@ld{\@citea\csname b@\@citeb\endcsname}%

57:         \else%

58: %         % compressible list of successive

59:           \edef\@h@ld{\ifmmode{-}\else--\fi\csname b@\@citeb\endcsname}%

60:         \fi%

61:       \else%   %  non-successor--dump what's held and do this one

62:         \@h@ld\@citea\csname b@\@citeb \endcsname%

63:         \let\@h@ld\relax%

64:       \fi}%

65:     \def\@citea{,\penalty\@highpenalty\,}%

66:   }\@h@ld

67: }{#1}}

68: 

69: %    \@cite macro using []'s:

70: %

71: \def\@citeb#1#2{{[#1]\if@tempswa , #2\fi}}

72: %

73: %    \@cite macro using exponents:

74: %

75: \def\@citeu#1#2{{$^{#1}$\if@tempswa , #2\fi }}

76: %

77: %    \@cite macro with nothing:

78: %

79: \def\@citep#1#2{{#1\if@tempswa , #2\fi}}

80: 

81: %

82: %   Style of citations:

83: %

84: \def\bcites{         % cite with []'s

85:         \catcode`\@=11

86:         \let\@cite=\@citeb

87:         \catcode`\@=12

88: }

89: 

90: \def\upcites{         % cite with exponents

91:         \catcode`\@=11

92:         \let\@cite=\@citeu

93:         \catcode`\@=12

94: }

95: 

96: \def\plaincites{      % cite without brackets

97:         \catcode`\@=11

98:         \let\@cite=\@citep

99:         \catcode`\@=12

100: }

101: 

102: %

103: %   Draft stuff

104: %

105: 

106: \newcount\hour

107: \newcount\minute

108: \newtoks\amorpm

109: \hour=\time\divide\hour by 60

110: \minute=\time{\multiply\hour by 60 \global\advance\minute by-\hour}

111: \edef\standardtime{{\ifnum\hour<12 \global\amorpm={am}%

112:         \else\global\amorpm={pm}\advance\hour by-12 \fi

113:         \ifnum\hour=0 \hour=12 \fi

114:         \number\hour:\ifnum\minute<10 0\fi\number\minute\the\amorpm}}

115: \edef\militarytime{\number\hour:\ifnum\minute<10 0\fi\number\minute}

116: 

117: \def\draftlabel#1{{\@bsphack\if@filesw {\let\thepage\relax

118:    \xdef\@gtempa{\write\@auxout{\string

119:       \newlabel{#1}{{\@currentlabel}{\thepage}}}}}\@gtempa

120:    \if@nobreak \ifvmode\nobreak\fi\fi\fi\@esphack}

121:         \gdef\@eqnlabel{#1}}

122: \def\@eqnlabel{}

123: \def\@vacuum{}

124: \def\marginnote#1{}

125: \def\draftmarginnote#1{\marginpar{\raggedright\scriptsize\tt#1}}

126: \overfullrule=0pt

127: 

128: \def\draft{

129:         \pagestyle{plain}

130:         \overfullrule=2pt

131:         \oddsidemargin -.5truein

132:         \def\@oddhead{\sl \phantom{\today\quad\militarytime} \hfil

133:         \smash{\Large\sl DRAFT} \hfil \today\quad\militarytime}

134:         \let\@evenhead\@oddhead

135:         \let\label=\draftlabel

136:         \let\marginnote=\draftmarginnote

137:         \def\ps@empty{\let\@mkboth\@gobbletwo

138:         \def\@oddfoot{\hfil \smash{\Large\sl DRAFT} \hfil}

139:         \let\@evenfoot\@oddhead}

140:         \def\@eqnnum{(\theequation)\rlap{\kern\marginparsep\tt\@eqnlabel}%

141:         \global\let\@eqnlabel\@vacuum}  }

142: 

143: 

144: 

145: %

146: %    Personal choice here

147: %

148: \def\section{\@startsection {section}{1}{\z@}{3.ex plus 1ex minus

149:  .2ex}{2.ex plus .2ex}{\Large\bf}}

150: \def\subsection{\@startsection{subsection}{2}{\z@}{2.75ex plus 1ex minus

151:  .2ex}{1.5ex plus .2ex}{\large\bf}}        

152: \def\subsect#1{\par\penalty1000{\noindent \bf #1}\par\penalty500}

153: \def\appendix{{\newpage\section*{Appendix}}\let\appendix\section%

154:         {\setcounter{section}{0}

155:         \gdef\thesection{\Alph{section}}}\section}

156: \def\thefootnote{\arabic{footnote}}

157: \def\abstract{\if@twocolumn

158: \section*{Abstract}

159: \else %\small

160: \begin{center}

161: {\bf Abstract\vspace{-.5em}\vspace{0pt}}

162: \end{center}

163: \quotation

164: \fi}

165: 

166: \catcode`\@=12

167: 

168: 

169: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

170: \newcommand{\beq}{\begin{equation}}

171: \newcommand{\eeq}{\end{equation}}

172: \newcommand{\beqa}{\begin{eqnarray}}

173: \newcommand{\eeqa}{\end{eqnarray}}

174: \newcommand{\dd}{{\rm d}}

175: \newcommand{\N}{{\bf N}}

176: \newcommand{\NN}{{\mathbb N}}

177: \newcommand{\Z}{{\bf Z}}

178: \newcommand{\ZZ}{{\mathbb Z}}

179: \newcommand{\Q}{{\bf Q}}

180: \newcommand{\QQ}{{\mathbb Q}}

181: \newcommand{\R}{{\bf R}}

182: \newcommand{\RR}{{\mathbb R}}

183: \newcommand{\C}{{\bf C}}

184: \newcommand{\CC}{{\mathbb C}}

185: \newcommand{\PP}{{\mathbb P}}

186: \newcommand{\e}{\,{\rm e}}

187: \newcommand{\CP}{{\CC\PP}}

188: \newcommand{\bra}{\langle}

189: \newcommand{\ket}{\rangle}

190: \newcommand{\no}{\nonumber}

191: \newcommand{\hsp}{\hskip5mm}

192: \newcommand{\vsp}{\vskip10mm}

193: \newcommand{\ds}{\displaystyle}

194: \newcommand{\ts}{\textstyle}

195: 

196: %  XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

197: %

198: %                Equation Definitions

199: %

200: \newcommand{\be}{\begin{equation}}

201: \newcommand{\ee}{\end{equation}}

202: \newcommand{\bea}{\begin{eqnarray}}

203: \newcommand{\eea}{\end{eqnarray}}

204: 

205: 

206: %

207: %                 Math Relations

208: %

209: \def\da{\downarrow}

210: \def\ua{\uparrow}

211: \def\upda{\updownarrow}

212: \def\to{\rightarrow}

213: \def\implies{\Rightarrow}

214: \def\To{\longrightarrow}

215: \def\longlongrightarrow{\relbar\joinrel\relbar\joinrel\rightarrow}

216: \def\ridiculousrightarrow{\relbar\joinrel\relbar\joinrel\relbar%

217: \joinrel\rightarrow}

218: 

219: \def\underarrow#1{\mathrel{\mathop{\longrightarrow}\limits_{#1}}}

220: \def\onnearrow#1{\mathrel{\mathop{\nearrow}\limits^{#1}}}

221: \def\undernearrow#1{\mathrel{\mathop{\nearrow}\limits_{#1}}}

222: \def\onarrow#1{\mathrel{\mathop{\longrightarrow}\limits^{#1}}}

223: \def\onArrow#1{\mathrel{\mathop{\longlongrightarrow}\limits^{#1}}}

224: \def\OnArrow#1{\mathrel{\mathop{\ridiculousrightarrow}\limits^{#1}}}

225: \def\lae{\mathrel{\mathop{\smash{\lower .5 ex \hbox{$\stackrel<\sim$}}}}}

226: \def\lae{\mathrel{\mathop{\smash{\lower .5 ex \hbox{$\stackrel>\sim$}}}}}

227: \def\eqq{\stackrel?=}

228: 

229: %                        Math Stuff

230: 

231: \def\ket#1{\left| #1 \right\rangle}

232: \def\bra#1{\left\langle #1 \right|}

233: \def\vev#1{\left\langle #1 \right\rangle}

234: \def\VEV#1{\left\langle #1 \right\rangle}

235: \def\f{\frac}

236: \def\pa{\partial}

237: \def\pb{\bar\pa}

238: \def\na{\nabla}

239: \def\Tr{{\rm Tr}}

240: \def\l:{\mathopen{:}\,}

241: \def\r:{\,\mathclose{:}}

242: \def\sech{\mathop{\rm sech}\nolimits}

243: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

244: 

245: %               DEFAULT SECTION SETUP

246: 

247: \catcode`\@=11

248: \def\theequation{\arabic{equation}}

249: %

250: %    for sections, subsections take from the following

251: %

252: %\def\theequation{\thesection.\arabic{equation}}

253: %\@addtoreset{equation}{section}

254: %\@addtoreset{footnote}{section}

255: %\@addtoreset{footnote}{subsection}

256: \catcode`\@=12

257: 

258: %                DEFAULT SETUP

259: 

260: 

261: \bcites

262: 

263: 

264: %                SETUP

265: 

266: %\draft

267: 

268: 

269: %             SECTION SETUP

270: 

271: \catcode`\@=11

272: \def\theequation{\thesection.\arabic{equation}}

273: \@addtoreset{equation}{section}

274: \@addtoreset{footnote}{section}

275: \@addtoreset{footnote}{subsection}

276: \catcode`\@=12

277: 

278: 

279: %               INFORMATION

280: 

281: \typeout{}

282: \typeout{}

283: \typeout{ }

284: \typeout{}

285: \typeout{}

286: \typeout{Ilka Brunner and Kentaro Hori}

287: \typeout{}

288: \typeout{}

289: \typeout{THIS IS A LATEX FILE: LATEX TWICE, AS USUAL. }

290: \typeout{}

291: \typeout{}

292: 

293: %

294: 

295: 

296: 

297: 

298: %

299: 

300: 

301: %

302: %           PAPER STARTS HERE

303: %

304: %load the bulking definitions etc.

305: %\input {shell}

306: 

307: 

308: \newcommand{\eps}{\epsilon}

309: \newcommand{\opsi}{\overline{\psi}}

310: \newcommand{\oQ}{\overline{Q}}

311: \newcommand{\btheta}{\overline{\theta}}

312: \newcommand{\cQ}{{\cal Q}}

313: \newcommand{\bepsilon}{\overline{\epsilon}}

314: \newcommand{\bPhi}{\overline{\Phi}}

315: \newcommand{\bP}{\overline{P}}

316: \newcommand{\bp}{\overline{p}}

317: \newcommand{\bphi}{\overline{\phi}}

318: \newcommand{\bpsi}{\overline{\psi}}

319: \newcommand{\bF}{\overline{F}}

320: \newcommand{\bJ}{\overline{J}}

321: \newcommand{\bh}{\overline{h}}

322: \newcommand{\bm}{\overline{m}}

323: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

324: \newcommand{\bi}{\overline{\imath}}

325: \newcommand{\bj}{\overline{\jmath}}

326: \newcommand{\bz}{\overline{z}}

327: \newcommand{\bchi}{\overline{\chi}}

328: %\newcommand{\blambda}{\overline{\lambda}}

329: \newcommand{\bsigma}{\overline{\sigma}}

330: \newcommand{\bSigma}{\overline{\Sigma}}

331: \newcommand{\bareta}{\overline{\eta}}

332: \newcommand{\ba}{\overline{a}}

333: \newcommand{\bb}{\overline{b}}

334: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

335: \newcommand{\lrd}{\overleftarrow{\partial}\!\!\!\!\!

336: \overrightarrow{\vbox to 8.35pt{}}}

337: \newcommand{\lrD}{\overleftarrow{D}\!\!\!\!\!

338: \overrightarrow{\vbox to 8.2pt{}}}

339: \newcommand{\bD}{\overline{D}}

340: \newcommand{\bcQ}{\overline{\cal Q}}

341: \newcommand{\cD}{{\cal D}}

342: \newcommand{\rp}{{{\rm Re}\,p}}

343: \newcommand{\vp}{\varphi_P}

344: \newcommand{\cJ}{{\cal J}}

345: \newcommand{\rre}{{\rm Re}}

346: \newcommand{\iim}{{\rm Im}}

347: \newcommand{\ra}{\rightarrow}

348: \newcommand{\wh}{\widehat}

349: \newcommand{\bpartial}{{\bar\partial}}

350: \newcommand{\bA}{\overline{A}}

351: \newcommand{\tY}{{\tilde Y}}

352: \newcommand{\bY}{{\overline Y}}

353: \newcommand{\tL}{\tilde{L}}

354: %\newcommand{\tG}{\tilde{G}}

355: \newcommand{\tJ}{\tilde{J}}

356: \newcommand{\tih}{\tilde{h}}

357: \newcommand{\tm}{\tilde{m}}

358: \newcommand{\tpsi}{\widetilde{\psi}}

359: \newcommand{\btpsi}{\overline{\widetilde{\psi}}}

360: \newcommand{\tr}{{\rm tr}}

361: \newcommand{\nn}{\nonumber}

362: \newcommand{\bartial}{\overline{\partial}}

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

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

365: \newcommand{\ft}{\mathfrak{t}}

366: \newcommand{\fu}{\mathfrak{u}}

367: \newcommand{\fm}{\mathfrak{m}}

368: \newcommand{\fgl}{\mathfrak{gl}}

369: \newcommand{\fsl}{\mathfrak{sl}}

370: \newcommand{\fsu}{\mathfrak{su}}

371: \newcommand{\fa}{\mathfrak{a}}

372: \newcommand{\fb}{\mathfrak{b}}

373: \newcommand{\whfg}{\widehat{\fg}}

374: \newcommand{\whfh}{\widehat{\fh}}

375: \newcommand{\wtJ}{\widetilde{J}}

376: \newcommand{\wtL}{L}

377: \newcommand{\bLambda}{\bar{\Lambda}}

378: \newcommand{\blambda}{\bar{\lambda}}

379: \newcommand{\V}{\widehat{V}}

380: \newcommand{\wtV}{\widehat{V}}

381: \newcommand{\B}{\Scr{H}}

382: \newcommand{\wtB}{\Scr{H}}

383: \newcommand{\bG}{\overline{G}}

384: \newcommand{\tG}{\widetilde{G}}

385: \newcommand{\btG}{\overline{\tG}}

386: \newcommand{\bq}{\overline{q}}

387: \newcommand{\bQ}{\overline{Q}}

388: \newcommand{\half}{{1\over 2}}

389: \newcommand{\wtq}{\widetilde{q}}

390: \newcommand{\s}{\sigma}

391: \newcommand{\bk}{\overline{k}}

392: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

393: %Ilka macros

394: \def\ew{\hspace*{-1mm}}

395: \def\ppe{\hspace*{-2.5mm}}

396: \newcommand{\Fus}[6]{F_{{\scriptstyle #1}{\scriptstyle #2}}

397:   \hspace*{.3mm}\displaystyle{[} \ew \begin{array}{ll} {\scriptstyle #3 }

398:   \ppe & {\scriptstyle #4} \ppe \\[-2mm] {\scriptstyle #5}\ppe &

399:   {\scriptstyle #6}\ew \end{array}\displaystyle{]}}

400: \newcommand{\Br}[6]{B_{{\scriptstyle #1}{\scriptstyle #2}}

401:   \hspace*{.3mm}\displaystyle{[} \ew \begin{array}{ll} {\scriptstyle #3 }

402:   \ppe & {\scriptstyle #4} \ppe \\[-2mm] {\scriptstyle #5}\ppe &

403:   {\scriptstyle #6}\ew \end{array}\displaystyle{]}}

404: \newcommand{\Brf}[4]{B

405:   \hspace*{.3mm}\displaystyle{[} \ew \begin{array}{ll} {\scriptstyle #1 }

406:   \ppe & {\scriptstyle #2} \ppe \\[-2mm] {\scriptstyle #3}\ppe &

407:   {\scriptstyle #4}\ew \end{array}\displaystyle{]}}

408: \newcommand{\kket}[1]{\vert \Scr{B}, #1\rangle\!\rangle}

409: \newcommand{\bbra}[1]{\langle\!\langle \Scr{B},#1 \vert}

410: \newcommand{\cket}[1]{\vert \Scr{C},#1\rangle\!\rangle}

411: \newcommand{\cbra}[1]{\langle\!\langle \Scr{C},#1 \vert}

412: 

413: \begin{document}

414: 

415: 

416: \begin{titlepage}

417: 

418: 

419: \begin{center}

420: 

421: \hfill

422: CERN-TH/2002-196\\

423: \hfill

424: hep-th/0208141\\

425: \hfill

426: 

427: 

428: \vskip 1.5 cm

429: {\large \bf Notes on Orientifolds of Rational Conformal Field Theories}

430: \vskip 2 cm 

431: {Ilka Brunner${}^*$ and Kentaro Hori${}^{\dag}$}\\

432: \vskip 0.6cm

433: {\it ${}^*$Theory Division, CERN,

434: CH-1211 Geneva 23, Switzerland}\\[0.3cm]

435: {\it ${}^{\dag}$Institute for Advanced Study,

436: Princeton, New Jersey 08540, U.S.A.\\

437: and\\

438: University of Toronto,

439: Toronto, Ontario M5S 1A7, Canada}

440: 

441: \end{center}

442: 

443: 

444: 

445: \vskip 2 cm

446: \begin{abstract}

447: We review and develop the construction of crosscap states

448: associated with parity symmetries

449: in rational conformal field theories.

450: A general method to construct crosscap states

451: in abelian orbifold models is presented.

452: It is then applied to rational $U(1)$ and parafermion

453: systems, where in addition we study  the geometrical interpretation of

454: the corresponding parities.

455: \end{abstract}

456: 

457: 

458: \vskip 3.5 cm

459: CERN-TH/2002-196

460: 

461: August 2002

462: 

463: 

464: 

465: \end{titlepage}

466: 

467: 

468: 

469: 

470: 

471: \section{Introduction}\label{sec:intro}

472: 

473: Four-dimensional

474: string compactifications

475: with ${\mathcal N}=1$ supersymmetry

476: allowing non-abelian gauge symmetries and chiral matter contents

477: are phenomenologically appealing.

478: %For a long time, $E_8\times E_8$ hetertic strings compactified on

479: %Calabi-Yau three-folds had been the leading candidates

480: %to achieve this. The situation has dramatically changed

481: %since some knowledge of

482: %the non-perturbative aspects of string theory were obtained.

483: Recently, D-branes on Calabi--Yau threefolds

484: were studied extensively, partly because one may obtain ${\mathcal N}=1$

485: supersymmetry in four dimensions if the D-branes extend in the

486: dimensions transverse to the Calabi-Yau.

487: However, consistency conditions require either

488: the Calabi--Yau to be non-compact or the tadpoles

489: to be cancelled by some other objects.

490: In the former case,

491: we are left with four-dimensional theories where

492: gravity is essentially decoupled.

493: Although such systems are interesting in their own right,

494: our main concern is the theory in four dimensions

495: with a {\it finite} Newton's constant.

496: Thus, we need to consider compact internal spaces.

497: The only known candidates to cancel the tadpoles while maintaining

498: ${\mathcal N}=1$ supersymmetry are orientifold planes.

499: 

500: Orientifolding means to gauge a parity symmetry of the worldsheet theory.

501: The basic example is the gauging of the worldsheet

502: orientation reversal $\Omega$ of the Type IIB superstring, 

503: resulting in the Type I superstring.

504: One is free, however, to consider more general parity actions

505: where $\Omega$ is combined with some action on 

506: space-time with a necessary consistency condition that this combination is

507: a symmetry of the underlying theory

508: \cite{Stalk,BSI,BSII,HOI,HOII} (see \cite{ASreview} for

509: a recent review).

510: The fixed-point sets of the space-time action, called orientifold planes,

511: can carry tension and RR charges

512: opposite to those of D-branes

513: and can be used for tadpole cancellation.

514: A crosscap state is associated to each parity symmetry, just as a

515: boundary state is associated with a boundary condition or D-brane

516: \cite{CLNY,PC}. These crosscap states encode the physical data, such as  

517: tension and RR charges of the orientifold planes.

518: 

519: One approach to study orientifolds of Calabi--Yau manifolds is to consider

520: special points in the moduli space

521: where the worldsheet theory is exactly solvable.

522: One class of such models are toroidal orbifolds, which have been

523: extensively studied \cite{GP,PS-GP,torus}.

524: Another important class of systems are Gepner models whose basic building

525: blocks are rational ${\mathcal N}=2$ superconformal field theories

526: (see \cite{BlWi,ABPSS} for earlier work on orientifolds of Gepner models).

527: However, general methods to

528: study parity symmetries and orientifolds of such models

529: are not developed to the same extent as in the case of D-branes.

530: 

531: The purpose of the present paper is to collect and

532: review the known techniques to study orientifolds

533: of rational conformal field theories (RCFTs) and further develop

534: them.

535: We present a coherent method to construct

536: parity symmetries and the corresponding crosscap states in RCFTs 

537: and their orbifolds.

538: The method is then applied to two simple examples,

539: the rational $U(1)$ and the parafermions $SU(2)/U(1)$.

540: This serves as a warm-up

541: to the ${\mathcal N}=2$ models, which will be reported in a forthcoming paper

542: \cite{BH2}.

543: Along the way, we also find the geometrical interpretation

544: of the parity symmetries of these examples. 

545: 

546: 

547: In Section \ref{general} we describe boundary

548: and crosscap states and the corresponding parities in RCFTs. 

549: We review the construction of a universal crosscap state by

550: Pradisi--Sagnotti--Stanev (PSS) \cite{PSS},

551: which applies in any RCFT with the charge conjugation modular invariant.

552: The corresponding parities can be combined with the discrete symmetries

553: of the system, giving rise to the class of crosscap states considered 

554: in \cite{HSS}. These crosscap states, as well as the rational boundary

555: states constructed by Cardy, preserve the diagonal subalgebra ${\cal A}$

556: of the full symmetry algebra ${\cal A} \otimes {\cal A}$. 

557: 

558: We then proceed to study parity symmetries of orbifold models

559: and provide a general new method to construct the crosscap states.

560: The emerging picture is much cleaner than the construction

561: of boundary states, which suffers from the fixed-point resolution problem.

562: Our method can be used to explain the result of an earlier paper

563: \cite{FHSSW}, which also studied the same subject

564: (see \cite{HuSc,HSSI} for earlier work concerning various

565: special cases). 

566: 

567: In the last subsection, we consider D-branes and parities preserving

568: the subalgebra ${\cal A}$ embedded into the symmetry algebra ${\cal A}

569: \otimes {\cal A}$ through an automorphism $\omega$ of ${\cal A}$:

570: $W \mapsto A\otimes 1 + 1\otimes \omega(W)$, in particular

571: the mirror automorphism that acts as charge conjugation.

572: Extending the terminology

573: of the ${\mathcal N}=2$ supersymmetry algebra \cite{OOY,BH2},

574: we call them B-branes/B-parities while the ones preserving the ordinary

575: diagonal subalgebra shall be called A-branes/A-parities.

576: Sometimes,

577: an orbifold model is the mirror of the original model. In such cases,

578: B-branes and B-parities can be obtained by applying the mirror map

579: to A-branes/A-parities of the mirror, which are constructed using

580: orbifold techniques.

581: 

582: 

583: Sections \ref{sec:arbR}, \ref{sec:ratR}, \ref{sec:gWZW} and 

584: \ref{sec:paraf}

585: are devoted to examples. 

586: In Section~\ref{sec:arbR} we revisit

587: the free boson compactified

588: on a circle of arbitrary radius, including an extension of

589: the standard construction \cite{CLNY,PC} to non-involutive parities.

590: We prove that the orientifold corresponding to the parity

591: where the target space action consists of a half-period shift

592: of the circle is T-dual to the orientifold associated to the reflection

593: $X\to -X$ of the circle coordinate with one 

594: $SO$- and one $Sp$-type orientifold plane (see \cite{ASreview} for a 

595: recent related discussion).

596: Section~\ref{sec:ratR} is concerned with the special case 

597: that the radius of the circle

598: is $R=\sqrt{k}$, $k$ a positive integer.

599: In this case, the theory becomes rational and one can apply the

600: method of Section \ref{general}. It is instructive to see how the

601: RCFT data encodes geometrical and physical information of the

602: orientifolds in this simple case.

603: 

604: Our second example is the parafermion system, which is

605: discussed in Sections~\ref{sec:gWZW} and \ref{sec:paraf}. This model

606: has a lagrangian description in terms of a $SU(2)_k$ mod

607: $U(1)_k$  gauged WZW model that is 

608: particularly well adapted to a study of the geometrical interpretation of 

609: parity actions. Geometrically, the parafermion theory can be understood

610: as a sigma-model with a disk target space parametrized by a

611: complex coordinate $z$ with $|z|\leq 1$. D-branes in this model

612: have been studied in \cite{MMS}.

613: A-type parities act as antiholomorphic involutions of the target

614: geometry, the basic example being $z\to \bar{z}$.  It is

615: possible to combine this with

616: an element of the $\Z_k$ symmetry of the theory,

617: which acts as a phase multiplication on the target space coordinate. 

618: Accordingly, the corresponding

619: orientifold planes are  located along diameters of the disk. B-type parities

620: act holomorphically on the target space, the fundamental B-type parity

621: being $z\to z$. This involution leaves the whole disk fixed

622: and therefore corresponds to an orientifold 2-plane.

623: Combining this with phase rotations leads generically

624: to non-involutive parities,

625: which we also consider. In the case where the level $k$ of the

626: parafermion theory is even, there is a second involutive parity,

627: $z\to -z$, which leaves only the center of the disk fixed and hence

628: describes an orientifold 0-plane. Finally we discuss

629: the same model purely in terms of rational conformal field theory

630: and give a detailed map of RCFT results to geometrical properties.

631: The questions discussed in Section~\ref{sec:paraf}

632: have been partially

633: addressed in \cite{hikida}, but we disagree with some of the 

634: results in that paper. In particular, we find that the geometric 

635: interpretation of the PSS crosscap states is different.

636: 

637: 

638: 

639: 

640: 

641: 

642: 

643: 

644: \section{Crosscaps in RCFT}\label{general}

645: 

646: \newcommand{\tilW}{\widetilde{W}}

647: \newcommand{\kke}{\rangle\!\rangle}

648: \newcommand{\bbr}{\langle\!\langle}

649: \newcommand{\ICk}{\kke^{C}}

650: \newcommand{\IBk}{\kke}

651: \newcommand{\ICb}{{}^{C}\bbr}

652: \newcommand{\IBb}{\bbr}

653: \newcommand{\shalf}{\mbox{$1\over 2$}}

654: \newcommand{\whchi}{\widehat{\chi}}

655: \newcommand{\bg}{\overline{g}}

656: 

657: 

658: We begin by describing the construction of boundary and crosscap states

659: of rational conformal field theories. A review and extension of 

660: previous work in

661: \cite{Cardy,Ishibashi,PSS,HSS,FHSSW,PSSII,bantay,bantay2,FSbranes} 

662: is followed by

663: developing new techniques to construct crosscap states in orbifolds.

664: 

665: We consider a quantum field theory in $1+1$ dimensions.

666: Let ${\mathcal H}_g$ be the space of states

667: of the system formulated on a circle

668: with the $g$-twisted periodic

669: boundary condition, where $g$ is an internal symmetry.

670: Let ${\mathcal H}_{\alpha_1,\alpha_2}$ be

671: the space of states on a segment

672: with the boundary conditions $\alpha_1$ and $\alpha_2$

673: at the left and the right ends.

674: We denote by $|\Scr{B}_{\alpha}\rangle$ and

675: $|\Scr{C}_{P}\rangle$

676: the boundary and crosscap states corresponding to

677: a boundary condition $\alpha$ and a parity symmetry $P=\tau\Omega$

678: \footnote{$\tau$ is an internal transform and $\Omega$ is the space

679: coordinate inversion. If the system has fermions,

680: $\Omega$ is assumed to include the exchange of left and right components.}.

681: The cylinder, Klein bottle (KB), and M\"obius strip (MS)

682: amplitudes are expressed in two ways

683: \beqa

684: &&

685: \Tr\!\!\mathop{}_{{\mathcal H}_{\alpha_1,\alpha_2}}g\e^{-\beta H_o(L)}

686: =\mathop{}_{g}\!\langle \Scr{B}_{1}|

687: \e^{-LH_c(\beta)}|\Scr{B}_{2}\rangle\!\!\mathop{}_{g},

688: \label{BB}\\

689: &&

690: \Tr\!\!\mathop{}_{{\mathcal H}_{P_1P_2^{-1}}}\!\!

691: P_2\e^{-\beta H_c(L)}

692: =\langle \Scr{C}_1|\e^{-{L\over 2}H_c(2\beta)}|\Scr{C}_2\rangle,

693: \label{CC}\\

694: &&

695: \Tr\!\!\mathop{}_{{\mathcal H}_{\alpha, P(\alpha)}}\!\!

696: P\e^{-\beta H_o(L)}

697: =\mathop{}_{P^{2}}\!\langle \Scr{B}_{\alpha} |\e^{-{L\over 2}H_c(2\beta)}

698: |\Scr{C}_P\rangle.

699: \label{BC}

700: \eeqa

701: Here $H_c(\ell)$ and $H_o(\ell)$ are the Hamiltonians of the system

702: on a circle of circumference $\ell$ and

703: segment of length $\ell$ respectively.

704: (The Hilbert spaces and boundary/crosscap states also depend

705:  on the lengths which are omitted for notational simplicity.)

706: The subscripts of the boundary states

707: show the periodicity of the boundary circle.

708: For instance, $|\Scr{B}_{2}\rangle_g$ consists of elements

709: in ${\mathcal H}_g$.

710: Note that $|\Scr{C}_P\rangle$ has a periodicity determined by $P^2$.

711: (Eq~(\ref{CC}) makes sense only if $P_1^2=P_2^2$.)

712: In (\ref{BC}), $P(\alpha)$ stands for the $P$-image of

713: the boundary condition $\alpha$.

714: The left and the right hand sides of (\ref{BB})--(\ref{BC})

715: may be referred to as {\it loop channel}

716: and

717: {\it tree channel} expressions respectively.

718: 

719: 

720: \begin{figure}[tb]

721: \centerline{\includegraphics{LT.eps}}

722: \caption{Cylinder, Klein Bottle and M\"obius Strip}

723: \label{LT}

724: \end{figure}

725: 

726: In what follows, we will consider

727: {\it conformally invariant} quantum field theories

728: and study boundary conditions and parity symmetries

729: that preserve the conformal invariance.

730: In such a theory, one can rescale the lengths

731: $(L,\beta)\to (\lambda L,\lambda \beta)$

732: without changing the amplitudes.

733: It is customary to choose the circumference of the closed string

734: to be $2\pi$ and the length of the open string to be $\pi$.

735: Suppose we choose $(L,\beta)=(\pi, -2\pi i\tau),

736: (2\pi,-2\pi i\tau), (\pi,-2\pi i\tau)$ in the loop channel expressions of

737: (\ref{BB}), (\ref{CC}), (\ref{BC}) respectively, where $\tau$

738: is a complex number on the positive imaginary axis.

739: Then in the tree-channel expressions, we take

740: $(L,\beta)=(-\pi i/\tau,2\pi),

741: (-\pi i/\tau,\pi), (-\pi i/2\tau,\pi)$.

742: In string theory, Eqs.~(\ref{BB})-(\ref{BC}) are referred to

743: as {\it loop/tree channel duality} and have played an important role

744: (see e.g. \cite{GP}).

745: 

746: 

747: 

748: \subsubsection*{\it The System We Consider}

749: 

750: 

751: We consider an RCFT based on a chiral algebra ${\mathcal A}=\{W_n^{(r)}\}$

752: with a set of representations $\{\Scr{H}_i\}$.

753: We primarily consider

754: the model ${\mathcal C}$

755: with the Hilbert space of states

756: \beq

757: {\mathcal H}^{\mathcal C}=\bigoplus_i\Scr{H}_i\otimes\Scr{H}_{\bi}.

758: \label{CCM}

759: \eeq

760: where $\bi$ is the BPZ conjugate of $i$. 

761: This model has ${\mathcal A}\otimes{\mathcal A}$ symmetry algebra

762: generated by $W^{(r)}=W^{(r)}\otimes 1$ and

763: $\tilW^{(r)}=1\otimes W^{(r)}$.

764: For each representation $i$, we fix an antiunitary operator

765: $U:\Scr{H}_i\to \Scr{H}_{\bi}$

766: such that

767: \beq

768: UW_n^{(r)}U^{-1}=(-1)^{s_r}{W_{-n}^{(r)}}^{\dag} \quad\forall r,

769: \eeq

770: where $s_r$ is a spin of the current $W^{(r)}$.

771: 

772: 

773: 

774: \subsection{Symmetry-Preserving D-branes/Orientifolds}

775: 

776: 

777: We first study D-branes and orientifolds that preserve

778: a diagonal subalgebra of the ${\mathcal A}\otimes{\mathcal A}$

779: symmetry.

780: On the Minkowski worldsheet with time and space coordinates

781: $(t,x)$, ``symmetry-preserving'' means the following:

782: for D-branes the associated boundary conditions (say, at $x=0$)

783: are such that $W^{(r)}(t,0)=\tilW^{(r)}(t,0)$,

784: while orientifolds should be associated with parity symmetries

785: that map $W^{(r)}(t,x)$ to $\tilW^{(r)}(t,-x)$.

786: 

787: \subsubsection{Constraints on Boundary/Crosscap Coefficients}

788: 

789: A Wick rotation followed by a 90${}^{\circ}$ rotation show

790: that the boundary and crosscap states obey

791: \beqa

792: (W^{(r)}_n-(-1)^{s_r}\tilW^{(r)}_{-n})|\Scr{B}\rangle=0,

793: \label{Bcond}\\

794: (W^{(r)}_n-(-1)^{s_r+n}\tilW^{(r)}_{-n})|\Scr{C}\rangle=0.

795: \label{Ccond}

796: \eeqa

797: The basic set of solutions to these equations

798: was found by Ishibashi \cite{Ishibashi}.

799: Let us denote by $\{|i,N\rangle\}_N$ an orthonormal basis

800: of the representation $\Scr{H}_i$.

801: Equations

802: (\ref{Bcond}) and (\ref{Ccond}) are solved respectively by

803: \beqa

804: &&\kket{i}:=\sum_N |i,N\rangle\otimes U|i,N\rangle,\\

805: &&\cket{i}:=\e^{\pi i(L_0-h_i)}\kket{i}.

806: \eeqa

807: It follows from the definition that

808: \beqa

809: &&\bbra{i}\e^{2\pi i \tau H_c}\kket{j}=\delta_{i,j}\chi_i(2\tau),

810: \label{bbchi}\\

811: &&\cbra{i}\e^{2\pi i \tau H_c}\cket{j}=\delta_{i,j}\chi_i(2\tau),

812: \label{ccchi}\\

813: &&\bbra{i}\e^{2\pi i \tau H_c}\cket{j}=\delta_{i,j}

814: \whchi_i(2\tau),

815: \label{bcchi}

816: \eeqa

817: where $H_c=L_0+\widetilde{L}_0-c/12$

818: and $\whchi_i(\tau)=\e^{-\pi i (h_i-c/24)}\chi_i(\tau+{1\over 2})$.

819: The actual boundary and crosscap states are linear

820: combinations of these Ishibashi-states:

821: \beqa

822: &&|\Scr{B}_{\alpha}\rangle=\sum_in_{\alpha i}\kket{i},\\

823: &&|\Scr{C}_{\mu}\rangle=\sum_i \gamma_{\mu i}\cket{i}.

824: \eeqa

825: Here $\alpha$ and $\mu$ are the labels for the boundary conditions

826: and parity symmetries.

827: We first assume that the parity symmetries are involutive

828:  $P_{\mu}^2=1$.

829: (We will later treat those that are not involutive.)

830: A set of constraints on the coefficients $n_{\alpha b}$ and $\gamma_{\mu b}$

831: are found by using the loop/tree channel duality (\ref{BB})--(\ref{BC}):

832: \beqa

833: &&

834: \Tr\!\!\mathop{}_{{\mathcal H}_{\alpha,\beta}}\!\e^{2\pi i\tau H_o}

835: =\langle \Scr{B}_{\alpha}|\e^{-{\pi i\over\tau} H_c}|\Scr{B}_{\beta}\rangle,

836: \label{TrBB}

837: \\

838: &&

839: \Tr\!\!\mathop{}_{{\mathcal H}_{g_{\mu\nu}}}\!\!

840: P_{\nu}\e^{2\pi i\tau H_c}

841: =\langle \Scr{C}_{\mu}|\e^{-{\pi i\over 2\tau}H_c}|\Scr{C}_{\nu}\rangle,

842: \label{TrCC}

843: \\

844: &&

845: \Tr\!\!\mathop{}_{{\mathcal H}_{\alpha,\mu(\alpha)}}\!\!

846: P_{\mu}\e^{2\pi i\tau H_o}

847: =\langle \Scr{B}_{\alpha}|\e^{-{\pi i\over 4\tau} H_c}|\Scr{C}_{\mu}\rangle,

848: \label{TrBC}

849: \eeqa

850: where

851: $g_{\mu\nu}$ is the internal symmetry $P_{\mu}P_{\nu}^{-1}$

852: that commutes with the chiral algebra ${\mathcal A}\otimes {\mathcal A}$,

853: and $\mu(\alpha)$ is the $P_{\mu}$-image of the boundary condition $\alpha$.

854: 

855: 

856: 

857: 

858: 

859: Since the diagonal symmetry ${\mathcal A}$ is preserved by

860: the boundary conditions $\alpha$ and $\beta$,

861: open string states fall into a sum of irreducible representations

862: \beq

863: {\mathcal H}_{\alpha,\beta}=\bigoplus_{i}

864: n_{\alpha\beta}^i\Scr{H}_i

865: \eeq

866: on which $H_o$ acts as $L_0-{c\over 24}$, where

867: $n_{\alpha\beta}^i$ are non-negative integers.

868: Using (\ref{bbchi}), (\ref{TrBB}) is expressed as

869: \beq

870: \sum_in_{\alpha\beta}^i\chi_i(\tau)=\sum_in_{\alpha i}^*n_{\beta i}

871: \chi_i(-1/\tau).

872: \label{Retrbb}

873: \eeq

874: 

875: 

876: For a symmetry $g$ that commutes with the chiral algebra

877: ${\mathcal A}\otimes {\mathcal A}$,

878: the space ${\mathcal H}_{g}$ of $g$-twisted closed string states

879: can be decomposed into the representations of ${\mathcal A}\otimes

880: {\mathcal A}$,

881: \beq

882: {\mathcal H}_g=\bigoplus_{ij}

883: h^{ij}_g \Scr{H}_i\otimes\Scr{H}_j,

884: \eeq

885: where $h^{ij}_g$ are non-negative integers.

886: Note that $P_{\nu}$ transforms the $g$-twisted boundary condition

887: into the $\tau_{\nu}g^{-1}\tau_{\nu}^{-1}$-twisted boundary condition.

888: Since $P_{\nu}$ is a symmetry-preserving parity,

889: $P_{\nu}W_n^{(r)}=\tilW_n^{(r)}P_{\nu}$,

890: it acts on the closed string states essentially by the exchange

891: of the left and right factors.

892: To be more precise, it maps the subspace

893: $h^{ij}_g\Scr{H}_i\otimes\Scr{H}_j$

894: of ${\mathcal H}_g$ to a subspace of

895: ${\mathcal H}_{\tau_{\nu}g^{-1}\tau_{\nu}^{-1}}$ as

896: \beq

897: P_{\nu}:\xi\otimes u\otimes v\in \C^{h^{ij}_g}\otimes

898: \Scr{H}_i\otimes \Scr{H}_{j}\mapsto

899: K_{\nu}^{ij}(g)\xi\otimes  v\otimes u\in

900: \C^{h^{ji}_{\tau_{\nu}g^{-1}\tau_{\nu}^{-1}}}

901: \otimes\Scr{H}_{j}\otimes \Scr{H}_{i},

902: \eeq

903: where $K_{\nu}^{ij}(g)$ is a matrix acting on the multiplicity space

904: $\C^{h^{ij}_g}\cong \C^{h^{ji}_{\tau_{\nu}g^{-1}\tau_{\nu}^{-1}}}$.

905: (Note that it has to be the case that

906: $h^{ij}_{g}=h^{ji}_{\tau_{\nu}g^{-1}\tau_{\nu}^{-1}}$.)

907: $P_{\nu}^2=1$ requires

908: $K_{\nu}^{ji}(\tau_{\nu}g^{-1}\tau_{\nu}^{-1})K_{\nu}^{ij}(g)=1$.

909: In particular, $K_{\nu}^{ii}(g_{\mu\nu})$ is a matrix that squares to 1

910: and therefore its eigenvalues must be $\pm 1$.

911: Thus, using (\ref{ccchi}), (\ref{TrCC})

912: is expressed as

913: \beq

914: \sum_{i}k_{\mu\nu}^{i}

915: \chi_i(2\tau)=\sum_i\gamma_{\mu i}^{*}\gamma_{\nu i}

916: \chi_i(-1/2\tau),

917: \label{Retrcc}

918: \eeq

919: where $k_{\mu\nu}^i=\tr K_{\nu}^{ii}(g_{\mu\nu})$.

920: Since $K_{\nu}^{ii}(g_{\mu\nu})$ squares to $1$,

921: the number $k_{\mu\nu}^i$ must be an integer such that

922: $|k_{\mu\nu}^i|\leq h^{ii}_{g_{\mu\nu}}$ and

923: $k_{\mu\nu}^i\equiv h^{ii}_{g_{\mu\nu}}$ mod 2.

924: 

925: 

926: Let us next consider the action of $P_{\mu}$

927: on open string states. Since it exchanges the left and right

928: boundaries of the string,

929: the symmetry-preserving condition becomes

930: $P_{\mu} W_n^{(r)}=(-1)^n W_n^{(r)} P_{\mu}$.

931: It therefore has to transform the subspace

932: $\Scr{H}_i^{\oplus n_{\alpha\beta}^i}$ of ${\mathcal H}_{\alpha,\beta}$

933: to a subspace of ${\mathcal H}_{\mu(\beta),\mu(\alpha)}$

934: as

935: $$

936: P_{\mu}:\eta\otimes u\in \C^{n_{\alpha\beta}^i}\otimes \Scr{H}_i\mapsto

937: M_{\alpha\beta,\mu}^i\eta\otimes

938: \e^{\pi i (L_0-h_a)}u\in \C^{n_{\mu(\beta)\mu(\alpha)}^i}\otimes

939: \Scr{H}_i,

940: $$

941: where $M_{\alpha\beta,\mu}^i$

942: is a matrix acting on $\C^{n_{\alpha\beta}^i}$.

943: (Note that $n_{\alpha\beta}^i$ must be equal to

944: $n_{\mu(\beta)\mu(\alpha)}^i$ for any $i$.)

945: $P_{\mu}^2=1$ requires

946: $M_{\mu(\beta)\mu(\alpha)}^iM_{\alpha\beta,\mu}^i=1$.

947: In particular, the eigenvalues of the matrix $M_{\alpha\mu(\alpha)}^i$

948: have to be $\pm 1$.

949: Using this, we can rewrite (\ref{TrBC}) as

950: \beq

951: \sum_i m_{\alpha,\mu}^i\whchi_i(\tau)

952: =\sum_i n_{\alpha i}^*\gamma_{\mu i}\whchi_i(-1/4\tau),

953: \label{Retrbc}

954: \eeq

955: where $m_{\alpha\mu}^i=\tr M_{\alpha\mu(\alpha),\mu}^i$.

956: For $P_{\mu}^2=1$, the number

957: $m_{\alpha\mu}^i$ must be an integer such that

958: $\pm m_{\alpha\mu}^i\leq n_{\alpha\mu(\alpha)}^i$ and

959: $m_{\alpha\mu}^i\equiv n_{\alpha\mu(\alpha)}^i$ mod 2.

960: 

961: 

962: We have found the constraints (\ref{Retrbb}), (\ref{Retrcc})

963: and (\ref{Retrbc}). At this stage, we use the modular transformation property

964: of characters

965: \beqa

966: &&\chi_j(-1/\tau)=\sum_i\chi_i(\tau)S_{ij},

967: \nn\\

968: &&

969: \chi_j(\tau+1)=\sum_{i}\chi_i(\tau)T_{ij},

970: \nn\\

971: &&\widehat{\chi}_j(-1/4\tau)

972: =\sum_i\widehat{\chi}_i(\tau)P_{ij},

973: \nn

974: \eeqa

975: where

976: \beq

977: P=\sqrt{T}ST^2S\sqrt{T},

978: \eeq

979: in which $\sqrt{T}_{ij}=\delta_{i,j}\e^{\pi i(h_i-{c\over 24})}$.

980: Here we are in the standard convention, $STS=T^{-1}ST^{-1}$.

981: The three constraints can then be rewritten as

982: \beqa

983: &&n_{\alpha\beta}^i=\sum_jn_{\alpha j}^*n_{\beta j}S_{ij},

984: \label{Ca1}

985: \\

986: &&k_{\mu\nu}^i

987: =\sum_j\gamma_{\mu j}^{*}\gamma_{\nu j}S_{ij},

988: \label{Ca2}

989: \\

990: &&m_{\alpha\mu}^i

991: =\sum_jn_{\alpha j}^{*}\gamma_{\mu j}P_{ij},

992: \label{Ca3}

993: \eeqa

994: where $n_{\alpha\beta}^i$,

995: $k_{\mu\nu}^i$ and

996: $m_{\alpha\mu}^i$ are integers such that

997: \beq

998: \begin{array}{l}

999: \bullet\,\,\,\,n_{\alpha\beta}^i\geq 0,\\

1000: \bullet\,\,\,\,-h^{ii}_{g_{\mu\nu}}\leq

1001: k_{\mu\nu}^i\leq h^{ii}_{g_{\mu\nu}},\quad

1002: k_{\mu\nu}^i\equiv h^{ii}_{g_{\mu\nu}}\,\,(\mbox{mod 2}),\\

1003: \bullet\,\,\,\,-n_{\alpha\mu(\alpha)}^i\leq

1004: m_{\alpha\mu}^i\leq n_{\alpha\mu(\alpha)}^i,\quad

1005: m_{\alpha\mu}^i\equiv n_{\alpha\mu(\alpha)}^i\,\,(\mbox{mod 2}),\\

1006: \end{array}

1007: \label{Ca4}

1008: \eeq

1009: where $h_g^{ij}$ is the multiplicity of $\Scr{H}_i\otimes\Scr{H}_j$

1010: in the space ${\mathcal H}_g$ of the $g$-twisted closed string states.

1011: 

1012: 

1013: \subsubsection{Cardy--PSS Solution}

1014: 

1015: A simple solution to the above constraints that applies to all RCFTs

1016: has been found by Cardy \cite{Cardy} for boundary

1017: states and by Pradisi--Sagnotti--Stanev \cite{PSS} for a crosscap state.

1018: 

1019: Cardy's boundary

1020: conditions carry the same labels as the representations $\{i\}$.

1021: The coefficients $n_{ij}$ and the multiplicities $n_{ij}^k$

1022: are given in terms of the modular $S$-matrix and the fusion coefficients:

1023: \beqa

1024: &&n_{ij}={S_{ij}\over \sqrt{S_{0j}}},\\

1025: &&n_{ij}^k=N_{\bi j}^{\bk}.

1026: \eeqa

1027: The constraints (\ref{Ca1}) translate into the Verlinde formula

1028: \cite{verlinde}.

1029: 

1030: PSS have found a crosscap state that corresponds to a parity symmetry

1031: transforming the Cardy boundary states as

1032: \beq

1033: P_0:i\mapsto \bi.

1034: \label{P0mapC}

1035: \eeq

1036: We label this state by ``0'' for a reason that will become clear shortly.

1037: The coefficients $\gamma_{0i}$ and the numbers $k_{00}^i$, $m_{i0}^{j}$

1038: are given by

1039: \beqa

1040: &&\gamma_{0i}={P_{0i}\over\sqrt{S_{0i}}},\\

1041: &&k_{00}^i=Y_{i0}^0,\label{Ce2}\\

1042: &&m_{i0}^j=Y_{\bi 0}^{\bj}\label{Ce3},

1043: \eeqa

1044: where $Y_{ij}^k$ is defined by

1045: \beq

1046: Y_{ij}^k=\sum_l{S_{il}P_{jl}P_{kl}^*\over S_{0l}}.

1047: \eeq

1048: It is straightforward to show that the constraints (\ref{Ca2}) and (\ref{Ca3})

1049: are satisfied for the PSS crosscap state. 

1050: To see that they obey (\ref{Ca4}), we first note that

1051: $g_{00}=1$ and therefore $h^{ii}_{g_{00}}=h^{ii}=\delta_{i,\bi}$

1052: because ${\mathcal H}_1={\mathcal H}=\oplus_i\Scr{H}_i\otimes \Scr{H}_{\bi}$.

1053: The conditions (\ref{Ca4}) therefore require $k_{00}^i=\pm \delta_{i,\bi}$.

1054: Also, since $n_{iP(i)}^j=n_{i\bi}^j=N_{\bi\bi}^{\bj}$,

1055: the number $m_{i0}^j$ must be an integer such that

1056: $|m_{i0}^j|\leq N_{\bi\bi}^{\bj}$ and $m_{i0}^j\equiv N_{\bi\bi}^{\bj}$

1057: (mod 2).

1058: These constraints are obeyed by the above solution because of

1059: 

1060: \noindent

1061: {\bf Bantay's Relation} \cite{bantay}

1062: \beq

1063: \begin{array}{l}

1064: |Y_{i0}^j|\leq N_{ii}^j,\\

1065: Y_{i0}^j\equiv N_{ii}^j\quad \mbox{mod 2}.

1066: \end{array}

1067: \label{bantays}

1068: \eeq

1069: 

1070: \noindent

1071: Indeed the condition on $m_{i0}^j=Y_{\bi 0}^{\bj}$ is nothing

1072: but Bantay's relation.

1073: The condition on $k_{00}^i=Y_{i0}^0$ also follows from this since

1074: $N_{ii}^0=\delta_{i,\bi}$.

1075: 

1076: The number $Y_{i0}^0$ (which is $\pm 1$ if $i=\bi$ and 0 otherwise)

1077: is a CFT analog of the Frobenius--Schur indicator for the theory

1078: of group representations

1079: \footnote{The Frobenius-Schur indicator of an irreducible representation $R$

1080: of a finite group $G$

1081: is defined to be $+1$ in the case when $R$ is real, $0$ when $R$ is complex

1082: and $-1$ when $R$ is pseudo-real.}.

1083: 

1084: 

1085: \subsubsection{Dressing by Discrete Symmetries}\label{sec:dressing}

1086: 

1087: If an RCFT has a discrete symmetry of a certain type,

1088: one can find additional parity symmetries together with

1089: the corresponding crosscap states.

1090: 

1091: Let $\Scr{G}$ be the finite abelian group generated by the simple currents $\{g\}$.

1092: We recall that a simple current $g$ is a representation such that

1093: the fusion product of

1094: $g$ with any representation $i$ contains only one representation, which we

1095: denote by $g(i)$.

1096: Let us introduce the number

1097: \beq

1098: Q_g(i)=h_i+h_g-h_{g(i)} \quad\mbox{mod 1}.

1099: \label{defQ}

1100: \eeq

1101: The map $g\mapsto \e^{2\pi iQ_g(i)}$ defines for each $i$ a homomorphism

1102: $\Scr{G}\to U(1)$

1103: (some of the properties of $Q_g(i)$ are summarized in Appendix \ref{app:Q}).

1104: We can thus define a representation of $\Scr{G}$

1105: on the Hilbert space 

1106: ${\mathcal H}$ in such a way that $g$ acts by the

1107: phase multiplication

1108: $\e^{2\pi i Q_g(i)}\times$ on the subspace $\Scr{H}_i\otimes\Scr{H}_{\bi}$.

1109: The Cardy state

1110: $|\Scr{B}_i\rangle$ is mapped by $g$ as

1111: \beq

1112: |\Scr{B}_i\rangle=\sum_{j}{S_{ij}\over\sqrt{S_{0j}}}\kket{j}

1113: \stackrel{g}{\mapsto}

1114: \sum_j{S_{ij}\over\sqrt{S_{0j}}}\e^{2\pi i Q_g(j)}\kket{j}

1115: =\sum_j{S_{g(i)j}\over \sqrt{S_{0j}}}\kket{j}=|\Scr{B}_{g(i)}\rangle,

1116: \label{mapofC}

1117: \eeq

1118: where we have used $S_{g(i)j}=\e^{2\pi i Q_g(j)}S_{ij}$.

1119: %Thus, the symmetry $g$ maps the Cardy brane $i$ to the Cardy brane $g(i)$.

1120: 

1121: 

1122: 

1123: We now find new parity symmetries $P_g$ that act on

1124: ${\mathcal H}$ as $g\circ P_0$.

1125: It follows from (\ref{P0mapC}) and (\ref{mapofC}),

1126: that these parities should map the Cardy branes as

1127: \beq

1128: P_g:i\mapsto g(\bi).

1129: \label{Pgi}

1130: \eeq

1131: The crosscap coefficients and the numbers

1132: $k_{g_1g_2}^i$, $m_{ig}^j$ are given by

1133: \beqa

1134: &&\gamma_{gi}={P_{gi}\over\sqrt{S_{0i}}},\\

1135: &&k_{g_1g_2}^i=Y_{i g_2}^{g_1}

1136: \label{zg1g2}\\

1137: &&m_{ig}^j=Y_{\bi g}^{\bj}.

1138: \eeqa

1139: It is easy to show that this solves the constraints (\ref{Ca2}) and (\ref{Ca3})

1140: \cite{PSS,HSS}.

1141: The Y-tensors can be rewritten as

1142: \beq

1143: Y_{i g_2}^{g_1}=\e^{\pi i (\hat{Q}_{g_2}(g_2^{-1}g_1)

1144: -2Q_{g_2}(i))}Y_{i0}^{g_2^{-1}g_1},\quad

1145: Y_{\bi g}^{\bj}

1146: =\e^{\pi i (\hat{Q}_g(g^{-1}(\bj))-2Q_g(\bi))}Y_{\bi 0}^{g^{-1}(\bj)},

1147: \eeq

1148: where $\hat{Q}_g(i):=h_i+h_g-h_{g(i)}$ (not just modulo integers).

1149: Since

1150: $\e^{\pi i (\hat{Q}_g(b)-2Q_g(a))}=\pm 1$ if

1151: $Y_{a 0}^{b}\ne 0$, the integrality of $k_{g_1g_2}^i$

1152: and $m_{ig}^j$ is also satisfied.

1153: 

1154: In order to show that the last constraint (\ref{Ca4}) is satisfied, we need to 

1155: find the multiplicities

1156: $h_{g_{12}}^{ii}$ and $n_{iP_g(i)}^j$, where

1157: $g_{12}=P_1P_2^{-1}=g_1g_2^{-1}$ and $P_g(i)=g(\bi)$ by (\ref{Pgi}).

1158: The space of $g$-twisted closed string states is given by

1159: ${\mathcal H}_g=\oplus_i\Scr{H}_i\otimes \Scr{H}_{g(\bi)}$ 

1160: (see Appendix \ref{subap:gtw}),

1161: and therefore $h_g^{ii}=\delta_{i,g(\bi)}$.

1162: Thus, (\ref{Ca4}) requires $k_{g_1g_2}^i=\pm \delta_{i,g_1g_2^{-1}(\bi)}$.

1163: We note that $N_{ii}^{g_2^{-1}g_1}=N^0_{i g_1^{-1}g_2(i)}

1164: =\delta_{\bi,g_1^{-1}g_2(i)}=\delta_{i,g_1g_2^{-1}(\bi)}$, and hence

1165: $Y_{i 0}^{g_2^{-1}g_1}=\pm\delta_{i,g_1g_2^{-1}(\bi)}$

1166: using Bantay's relation

1167: (\ref{bantays}).

1168: This indeed shows that

1169: $k_{g_1g_2}^i=

1170: Y_{i g_2}^{g_1}=\pm Y_{i 0}^{g_2^{-1}g_1}

1171: =\pm \delta_{i,g_1g_2^{-1}(\bi)}$.

1172: On the other hand, the open string multiplicity is

1173: $n_{ig(\bi)}^j=N_{\bi g(\bi)}^{\bj}=N_{\bi\bi}^{g^{-1}(\bj)}$.

1174: Then the claimed solution

1175: $m_{ig}^i=\pm Y_{\bi 0}^{g^{-1}(\bj)}$

1176: obeys the last condition of (\ref{Ca4}),

1177: again due to Bantay's relation (\ref{bantays}).

1178: 

1179: 

1180: 

1181: \subsubsection{Summary}

1182: 

1183: We have found D-branes $\Scr{B}_i$ and parity symmetries $P_g$

1184: ($g\in\Scr{G}$),

1185: which preserve the diagonal symmetry ${\mathcal A}\subset

1186: {\mathcal A}\otimes{\mathcal A}$.

1187: The corresponding boundary and crosscap states are given by

1188: \beqa

1189: &&|\Scr{B}_i\rangle

1190: =\sum_j{S_{ij}\over\sqrt{S_{0j}}}\kket{j},\\

1191: &&|\Scr{C}_{P_g}\rangle

1192: =\sum_i{P_{gj}\over\sqrt{S_{0j}}}\cket{j}.

1193: \eeqa

1194: The cylinder, Klein bottle and M\"obius strip amplitudes are given by

1195: \beqa

1196: &&\Tr\!\mathop{}_{{\mathcal H}_{i,i'}}\!\e^{2\pi i\tau H}

1197: =\langle \Scr{B}_i|\e^{-{\pi i\over \tau}H}|\Scr{B}_{i'}\rangle

1198: =\sum_j N_{\bi i'}^{\bj}\chi_j(\tau),

1199: \label{trr1}\\

1200: &&\Tr\!\mathop{}_{{\mathcal H}_{gh^{-1}}}\!P_h\e^{2\pi i\tau H}

1201: =\langle \Scr{C}_g|\e^{-{\pi i\over 2\tau}H}|\Scr{C}_h\rangle

1202: =\sum_j Y_{jh}^g\chi_j(2\tau),

1203: \label{trr2}\\

1204: &&\Tr\!\mathop{}_{{\mathcal H}_{i,g(\bi)}}\!P_g\e^{2\pi i\tau H}

1205: =\langle \Scr{B}_i|\e^{-{\pi i\over 4\tau}H}|\Scr{C}_g\rangle

1206: =\sum_j Y_{\bi g}^{\bj}\widehat{\chi}_j(\tau),

1207: \label{trr3}

1208: \eeqa

1209: where a shorthand notation $|\Scr{C}_g\rangle$ for

1210: $|\Scr{C}_{P_g}\rangle$ has been used.

1211: We can simplify  expression (\ref{trr1}) by using

1212: $N_{\bi i'}^{\bj}=N_{i'j}^i$.

1213: Also, taking the complex conjugate of (\ref{trr3})

1214: and using $(Y_{\bi g}^j)^*=Y_{ij}^g$, we have

1215: \beq

1216: \Tr\!\mathop{}_{{\mathcal H}_{g(\bi),i}}\!P_g\e^{2\pi i\tau H}

1217: =\langle \Scr{C}_g|\e^{-{\pi i\over 4\tau}H}|\Scr{B}_i\rangle

1218: =\sum_j Y_{ij}^g\widehat{\chi}_j(\tau).

1219: \eeq

1220: This can also be obtained from (\ref{trr3}) by replacing

1221: $i\to g(\bi)$ and using $Y_{\overline{g(\bi)} g}^{\bj}=Y_{ij}^g$,

1222: which can be derived by using $P_{\bj k}^*=P_{jk}$ and

1223: $\e^{-2\pi i Q_g(k)}P_{gk}=P_{gk}^*$.

1224: 

1225: 

1226: 

1227: 

1228: \subsection{Crosscaps in Orbifolds}\label{subsec:orb}

1229: 

1230: 

1231: 

1232: We have constructed parity symmetries together with the crosscap states

1233: for the charge-conjugate modular invariant ${\mathcal C}$.

1234: We now turn  to the orbifold model ${\mathcal C}/G$

1235: where $G$ is a group of simple currents, $G\subset \Scr{G}$.

1236: To define a consistent orbifold theory, 

1237: $G$ must have a symmetric bilinear form

1238: $q(g_1,g_2)$ with values in $\R/\Z$ such that

1239: \beq

1240: Q_{g_1}(g_2)=2q(g_1,g_2)\qquad\mbox{mod 1},

1241: \eeq

1242: and $q(g,g)=-h_g$.

1243: The orbifold model we consider has modular invariant partition function

1244: \beq

1245: Z^{{\mathcal C}/G}

1246: ={1\over |G|}\sum_{i,g_1,g_2}

1247: \e^{2\pi i(Q_{g_2}(i)-q(g_2,g_1))}

1248: \chi_i(\tau)\overline{\chi_{g_1^{-1}(i)}(\tau)}.

1249: \label{orbPF}

1250: \eeq

1251: We will find $|\Scr{G}|$ symmetry preserving

1252: parities/crosscaps in 1-to-1 correspondence to

1253: the number $|\Scr{G}|$ of simple currents, just as in the original model

1254: ${\mathcal C}$.

1255: 

1256: 

1257: 

1258: Let us first present the basic idea behind the construction of

1259: a crosscap state in the orbifold model ${\mathcal C}/G$

1260: for the parity symmetry that is induced by a parity

1261: $P$ of the original model ${\mathcal C}$.

1262: The twisted partition function with respect to the induced parity,

1263: denoted again by $P$,

1264: is expressed as

1265: \beq

1266: \Tr\!\!\mathop{}_{{\mathcal H}^{{\mathcal C}/G}}Pq^H

1267: =\sum_{g_1\in G}\Tr\!\!\mathop{}_{{\mathcal H}_{g_1}}\left(

1268: \Bigl({1\over |G|}\sum_{g_2\in G}g_2\Bigr)\,Pq^H\right),

1269: \eeq

1270: where the first sum is over the $G$-twisted spaces and

1271: ${1\over |G|}\sum_{g_2\in G}g_2$ is the projection onto

1272: the subspace of $G$-invariant states.

1273: Let us rearrange the sum as $(1/|G|)\sum_{g_1,g_2}\Tr_{{\mathcal H}_{g_1}}

1274: g_2Pq^H$, and make a replacement $g_1\to g_1g_2^{-1}$.

1275: At this point we recall from (\ref{CC}) that

1276: $$

1277: \Tr\!\!\mathop{}_{{\mathcal H}_{g_1g_2^{-1}}}(

1278: g_2Pq^H)

1279: ={}^{\mathcal C}\!\langle \Scr{C}_{g_1P}|

1280: q_t^H|\Scr{C}_{g_2P}\rangle^{\!{\mathcal C}},

1281: $$

1282: where the superscript shows that the state

1283: $|-\rangle^{\mathcal C}$ is in the theory ${\mathcal C}$.

1284: Then, we find

1285: \beq

1286: \Tr\!\!\mathop{}_{{\mathcal H}^{{\mathcal C}/G}}Pq^H

1287: ={1\over |G|}\sum_{g_1,g_2}{}^{\mathcal C}\!\langle \Scr{C}_{g_1P}|

1288: q_t^H|\Scr{C}_{g_2P}\rangle^{\!{\mathcal C}}.

1289: \eeq

1290: This implies that the crosscap state

1291: for the induced parity $P$

1292: is given by

1293: \beq

1294: |\Scr{C}_P\rangle^{\!{\mathcal C}/G}

1295: ={1\over\sqrt{|G|}}\sum_{g\in G}|\Scr{C}_{gP}\rangle^{\!{\mathcal C}}.

1296: \label{cand}

1297: \eeq

1298: 

1299: 

1300: \subsubsection{PSS Parities Induced on Orbifold Models}

1301: 

1302: We would like to apply this construction by identifying

1303: $P$ as one of the parity symmetries

1304: of ${\mathcal C}$ obtained in the previous section,

1305: say the original PSS parity $P_0$.

1306: We know that $gP_0$ is equal to $P_g$ at least in the action on

1307: the untwisted states ${\mathcal H}^{\mathcal C}$ and we know the

1308: crosscap states for all $P_g$.

1309: Thus, (\ref{cand}) appears to be an ideal formula

1310: for constructing a crosscap

1311: state in the orbifold theory.

1312: However, one has to be careful when identifying $P_0$ as $P$

1313: and $P_g$ as $gP_0$.

1314: The subtleties are:\\

1315: (i) $P$ may differ from $P_0$ in the action

1316: on the twisted Hilbert space

1317: ${\mathcal H}_g$ by a $g$ dependent phase.\\

1318: (ii) $P_g$ and $g\circ P_0$ may differ in the

1319: action on the twisted Hilbert space

1320: ${\mathcal H}_{g'}$ by a $g$ and $g'$ dependent phase.\\

1321: Let us first examine (ii).

1322: Using $k_{g_1g_2}^i$ in (\ref{zg1g2})

1323: and the $g_2$-action on ${\mathcal H}_{g_1}$ in (\ref{TTT}),

1324: it is straightforward to find

1325: \beq

1326: P_{g_2}=\e^{\pi i(\hat{Q}_{g_2}(g_1)-2q(g_2,g_1))}g_2P_0\qquad

1327: \mbox{on ${\mathcal H}_{g_1}$}.

1328: \eeq

1329: To accommodate the possibility (i) we suppose that $P$ and $P_0$ are related

1330: as

1331: \beq

1332: P=\e^{\pi i \theta(g)}P_0\qquad

1333: \mbox{on ${\mathcal H}_{g}$}.

1334: \eeq

1335: Then the above procedure is modified as follows

1336: \beqa

1337: \Tr\!\!\mathop{}_{{\mathcal H}^{{\mathcal C}/G}}Pq^H

1338: &=&{1\over |G|}

1339: \sum_{g_1,g_2}\Tr\!\!\mathop{}_{{\mathcal H}_{g_1}} g_2Pq^H

1340: \nn\\

1341: &=&{1\over |G|}\sum_{g_1,g_2}

1342: \e^{\pi i \theta(g_1)}

1343: \e^{-\pi i (\hat{Q}_{g_2}(g_1)-2q(g_2,g_1))}

1344: \Tr\!\!\mathop{}_{{\mathcal H}_{g_1}} P_{g_2}q^H

1345: \nn\\

1346: &=&{1\over |G|}\sum_{g_1,g_2}

1347: \e^{\pi i \theta(g_1)}

1348: \e^{-\pi i (\hat{Q}_{g_2}(g_1)-2q(g_2,g_1))}

1349: {}^{\mathcal C}\!\langle\Scr{C}_{P_{g_1g_2}}|q_t^H

1350: |\Scr{C}_{P_{g_2}}\rangle^{\!{\mathcal C}}.

1351: \eeqa

1352: We would like the phase on the RHS to be of the form

1353: $\e^{-i\omega_{g_1g_2}+i\omega_{g_2}}$,

1354: so that the partition function can be expressed

1355: as $\langle -|q_t^H|-\rangle$, where

1356: $$|-\rangle={1\over

1357: \sqrt{|G|}}

1358: \sum_g\e^{i\omega_g}|\Scr{C}_{P_{g}}\rangle^{\mathcal C}.$$

1359: Thus, we need to have

1360: $$

1361: \e^{\pi i(\theta(g_1)-\hat{Q}_{g_2}(g_1)+2q(g_2,g_1))}

1362: =\e^{-i\omega_{g_1g_2}+i\omega_{g_2}}.

1363: $$

1364: Setting $g_2=1$ we find

1365: $\e^{\pi i \theta(g)}=\e^{-i\omega_g+i\omega_1}$.

1366: Inserting this relation,

1367: we find the constraint on $\theta(g)$:

1368: \beq

1369: \theta(g_1g_2)=\theta(g_1)+\theta(g_2)-\hat{Q}_{g_2}(g_1)+2q(g_2,g_1)\qquad

1370: \mbox{mod 2}.

1371: \label{thetaEq}

1372: \eeq

1373: For each solution $\theta(g)$ to this constraint,

1374: we find the crosscap state

1375: \beq

1376: |\Scr{C}_{P^{\theta}}\rangle^{\!{\mathcal C}/G}

1377: ={\e^{i\omega_1}\over\sqrt{|G|}}

1378: \sum_{g\in G}

1379: \e^{-\pi i\theta(g)}|\Scr{C}_{P_g}\rangle^{\!{\mathcal C}}.

1380: \label{orbcross1}

1381: \eeq

1382: Let us count the number of solutions to (\ref{thetaEq}).

1383: If we find one solution, $\theta_*(g)$, the other solutions take the form

1384: $\theta_*(g)+{\mit\Delta}\theta(g)$,

1385: where ${\mit\Delta}\theta(g)$ obey the homogeneous equation

1386: ${\mit\Delta}\theta(g_1g_2)

1387: ={\mit\Delta}\theta(g_1)+{\mit\Delta}\theta(g_2)$ mod 2.

1388: Note that $g\to \e^{i\pi {\mit\Delta}\theta(g)}$

1389: defines a representation of the group $G$ into $U(1)$.

1390: Since there are $|G|$ such representations, we find that 

1391: Eq.~(\ref{thetaEq}) has $|G|$ solutions.

1392: 

1393: 

1394: We could have chosen another $P$ in this construction.

1395: In the above, $P$ was equal to $P_0$ when acting

1396: on the untwisted Hilbert space.

1397: Replacing $P_0$ here by $P_{g_1}$ does nothing new if $g_1\in G$,

1398: since the average over $G$ will be taken.

1399: However, replacing $P_0$ by $P_{g'}$ with $g'\in \Scr{G}\setminus G$

1400: will make a difference.

1401: Repeating the above procedure, we find

1402: parity symmetries of the orbifold theory

1403: induced from such a $P$.

1404: There are $|G|$ of them:

1405: one for each solution $\theta(g)$ of (\ref{thetaEq})

1406: which acts on the states as

1407: \beq

1408: P^{\theta}_{g'}=\e^{\pi i (\theta(g)-\hat{Q}_{g'}(g))}P_{g'}

1409: \qquad

1410: \mbox{on ${\mathcal H}_{g}$},

1411: \eeq

1412: and has the crosscap state

1413: \beq

1414: |\Scr{C}_{P^{\theta}_{g'}}\rangle^{\!{\mathcal C}/G}

1415: ={\e^{i\omega_{g'}}\over\sqrt{|G|}}

1416: \sum_{g\in G}

1417: \e^{-\pi i(\theta(g)-\hat{Q}_{g'}(g))}

1418: |\Scr{C}_{P_{gg'}}\rangle^{\!{\mathcal C}}.

1419: \label{orbcross2}

1420: \eeq

1421: Again, replacing $g'$ by $g'g_1$ with $g_1\in G$ makes no difference.

1422: 

1423: To summarize,

1424: for each $P$ we find $|G|$ parities

1425: from the choice of solutions to (\ref{thetaEq}),

1426: and there are $|\Scr{G}/G|$ choices for $P$ itself.

1427: Thus, we have found as many parity symmetries as

1428: $$

1429: |G|\times |\Scr{G}/G|=|\Scr{G}|.

1430: $$ 

1431:  

1432: \subsubsection*{\it The square of $P_{g'}^{\theta}$}

1433: 

1434: 

1435: The parity symmetries obtained this way are not necessarily

1436: involutive. Since $P_{g'}$ is involutive,

1437: the square of $P^{\theta}_{g'}$ is given by

1438: \beq

1439: (P^{\theta}_{g'})^2=\e^{2\pi i (\theta(g)-Q_{g'}(g))}\times 

1440: \qquad

1441: \mbox{on ${\mathcal H}_{g}$}.

1442: \label{SqP}

1443: \eeq

1444: We note that

1445:  $\theta(g)$ obeys

1446: $\theta(g_1g_2)=\theta(g_1)+\theta(g_2)$ modulo 1,

1447: since $\hat{Q}_{g_1}(g_2)-2q(g_1,g_2)$ is an integer.

1448: Note also that

1449: $Q_{g'}(g_1g_2)=Q_{g'}(g_1)+Q_{g'}(g_2)$ modulo 1.

1450: Thus,

1451: we find that $g\mapsto

1452: \e^{2\pi i (\theta(g)-Q_{g'}(g))}$ is a homomorphism of

1453: $G$ to $U(1)$, namely a character of $G$.

1454: Therefore, $(P^{\theta}_{g'})^2$ is a quantum symmetry

1455: of the orbifold model. In particular, the crosscap state $|\Scr{C}_{P_{g'}^{\theta}}\rangle$

1456: must be a state on the circle with the boundary condition

1457: twisted by this quantum symmetry.

1458: This means, as shown in Appendix~\ref{subap:quantums},

1459: that the state must transform under the action of $g$ as

1460: \beq

1461: |\Scr{C}_{P_{g'}^{\theta}}\rangle

1462: \,\stackrel{g}{\longmapsto}\,

1463: \e^{2\pi i (\theta(g)-Q_{g'}(g))}

1464: |\Scr{C}_{P_{g'}^{\theta}}\rangle,

1465: \eeq

1466: which can also be confirmed by a direct computation.

1467: 

1468: \subsubsection{Boundary States in Orbifolds}\label{subsubsec:Borb}

1469: 

1470: An idea to obtain D-branes in the orbifold model

1471: is to pick a D-brane $i$ in the original system and

1472: to take the ``average'' over the image branes $g(i)$, $g\in G$.

1473: The corresponding boundary states are given by

1474: \beq

1475: |\Scr{B}_{[i]}\rangle^{{\mathcal C}/G}

1476: ={1\over \sqrt{|G|}}

1477: \sum_{g\in G}|\Scr{B}_{g(i)}\rangle^{\mathcal C}.

1478: \label{bi0}

1479: \eeq

1480: The normalization factor

1481: $1/\sqrt{|G|}$ is required for the open string partition function

1482: to count the $i$-$g(i)$ string just once.

1483: (A more careful treatment is required if $g(i)=i$

1484: for some $g\ne {\rm id}$, see below.)

1485: Since $g|\Scr{B}_i\rangle=|\Scr{B}_{g(i)}\rangle$ (\ref{mapofC}),

1486: the state $|\Scr{B}_{[i]}\rangle^{{\mathcal C}/G}$ is $G$-invariant

1487: and belongs to the Hilbert space ${\mathcal H}^{{\mathcal C}/G}$.

1488: Obviously the brane $\Scr{B}_{[i]}$ is the same as

1489: $\Scr{B}_{[g(i)]}$.

1490: 

1491: 

1492: 

1493: Since the parities

1494: $P_{g'}^{\theta}$ are not involutive, but

1495: square to quantum symmetries (\ref{SqP}),

1496: one is motivated to consider the boundary states on the circle with twisted

1497: boundary condition.

1498: Let $g_{\rho}$ be the quantum symmetry associated with the character

1499: $g\mapsto \e^{2\pi i \rho(g)}$.

1500: We claim that the $g_{\rho}$-twisted boundary state for the brane

1501: $\Scr{B}_{[i]}$ takes the form

1502: \beq

1503: |\Scr{B}_{[i]}\rangle_{g_{\rho}}^{{\mathcal C}/G}

1504: ={\e^{i\lambda}\over \sqrt{|G|}}

1505: \sum_{g\in G}\e^{-2\pi i\rho(g)}|\Scr{B}_{g(i)}\rangle^{\mathcal C}.

1506: \label{bir}

1507: \eeq

1508: Indeed, $g\in G$ transforms it as

1509: \beq

1510: |\Scr{B}_{[i]}\rangle_{g_{\rho}}^{{\mathcal C}/G}

1511: \stackrel{g}{\longmapsto} 

1512: {\e^{i\lambda}\over \sqrt{|G|}}

1513: \sum_{g'\in G}\e^{-2\pi i\rho(g')}|\Scr{B}_{gg'(i)}\rangle^{\mathcal C}

1514: =\e^{2\pi i \rho(g)}|\Scr{B}_{[i]}\rangle_{g_{\rho}}^{{\mathcal C}/G}.

1515: \eeq

1516: As shown in Appendix~\ref{subap:quantums},

1517: this means that $|\Scr{B}_{[i]}\rangle_{g_{\rho}}^{{\mathcal C}/G}$

1518: is a state on the $g_{\rho}$-twisted circle.

1519: 

1520: 

1521: As mentioned above,

1522: when $g(i)\ne i$ if $g\ne 1$ for some $i$, the argument has to be further refined.

1523: This is known as ``the fixed-point problem''.

1524: Resolved boundary states have been constructed in 

1525: \cite{FHSSW,FSbreakI,FSbreakII,BFS}.

1526: In this paper, we do not try to reproduce a general solution, but will revisit

1527: the resolutions

1528: in the concrete models we consider later.

1529: 

1530: 

1531: 

1532: 

1533: \subsubsection{Constraints on Discrete Torsion}

1534: 

1535: In general, there can be more than one models of orbifold

1536: ${\mathcal C}/G$. We have chosen a particular one

1537: with the partition function (\ref{orbPF}), but one could change

1538: the model by turning on a ``discrete torsion'' \cite{Vafa:DT}.

1539: This means adding an extra phase factor $\e^{2\pi i e(g_2,g_1)}$

1540: for each summand of (\ref{orbPF}), where $e(g_2,g_1)$ is an

1541: antisymmetric bilinear form of $G$ with values in $\R/\Z$

1542: such that $e(g,g)=0$.

1543: Let us see how this modifies the above story.

1544: 

1545: The discrete torsion shifts the bilinear form $q$

1546: as 

1547: $$q(g_2,g_1)\to q(g_2,g_1)-e(g_2,g_1).$$

1548: The argument above goes through

1549: without modification until (\ref{thetaEq}),

1550: at which point one has to be careful.

1551: We note that $\hat{Q}_{g_2}(g_1)=h_{g_2}+h_{g_1}-h_{g_1g_2}$

1552: is symmetric under the exchange $g_1\leftrightarrow g_2$.

1553: Thus, (\ref{thetaEq}) is possible only if $2(q(g_2,g_1)-e(g_2,g_1))$

1554: is symmetric (mod 2). Since $q(g_2,g_1)$ is already symmetric, we find that

1555: $2e(g_2,g_1)$ has to be symmetric modulo 2, or

1556: $e(g_2,g_1)$ has to be symmetric modulo 1.

1557: Since $e(g_2,g_1)$ is antisymmetric at the same time,

1558: it may appear that no discrete torsion is allowed.

1559: However, since $-{1\over 2}

1560: \equiv {1\over 2}$ mod $\Z$,

1561: a symmetric form with

1562: $0$ or ${1\over 2}$ entries is at the same time antisymmetric modulo 1.

1563: Thus, special types of discrete torsion are indeed allowed.

1564: We shall call them {\it $\Z_2$ discrete torsions}.

1565: 

1566: \noindent

1567: {\bf Remark 1.}

1568: In general, choices are involved in finding a symmetric bilinear form

1569: $q(g_1,g_2)$ such that $2q(g_1,g_2)=Q_{g_1}(g_2)$ mod 1

1570: and $q(g,g)=-h_g$.

1571: A different choice corresponds exactly to the

1572: modification by a $\Z_2$ discrete torsion.

1573: \\

1574: {\bf Remark 2.}

1575: The restriction on the discrete torsion in orientifold models 

1576: is not new, if one recalls that the discrete torsion is a kind of

1577: $B$-field: Type I string theory projects out the NS-NS B-field modes

1578: \cite{GSW}.

1579: Furthermore, it is also known that special kinds of B-fields

1580: (with period $\pi$) are allowed \cite{SS}. 

1581: (See also \cite{AAADS})

1582: \\

1583: {\bf Remark 3.}

1584: It may appear natural to relate the $\Z_2$ discrete torsions to

1585: the group cohomology classes $\alpha\in H^2(G,\Z_2)$ in the standard way:

1586: $\e^{2\pi i e(g,h)}=\alpha(g,h)\alpha(h,g)^{-1}$.

1587: However, unlike the ordinary case where both

1588: $\e^{2\pi i e(g,h)}$ and $\alpha(g,h)$ take values in $U(1)$,

1589: it is not always true that the map

1590: $\alpha\mapsto \e^{2\pi i e}$ is one-to-one

1591: \footnote{

1592: Let $A$ be an abelian group.

1593: $Ext(G,A)=\{\alpha\in H^2(G,A)|{\rm symmetric}\}$, the kernel of the map

1594: $\alpha(g,h)\to\epsilon(g,h)=\alpha(g,h)\alpha(h,g)^{-1}$, is the set of

1595: {\it abelian} extensions of $G$ by $A$.

1596: It is trivial for $A=U(1)$ but not always for other $A$.

1597: For example $\Z_4$ and $\Z_2\times \Z_2$ are inequivalent

1598: $\Z_2$ extensions of $\Z_2$. 

1599: (For a product $G=G_1\times G_2\times \cdots\times G_s$, 

1600: $Ext(G,A)\cong\prod_{i=1}^s Ext(G_i,A)$ (Cor 3.17 of \cite{GCbook}).

1601: Also $Ext(\Z_n,A)=H^2(\Z_n,A)=A/A^n$ (Theorem 3.1 of \cite{GCbook}).

1602: Thus, if $G$ has a $\Z_n$ factor with $n$ even,

1603: $Ext(G,\Z_2)$ cannot be trivial.)}.

1604: Thus, just from the above consideration, one cannot conclude

1605: that $H^2(G,\Z_2)$ characterizes the $\Z_2$ discrete torsion.

1606: However, there is a claim \cite{KlRa} that this is indeed the

1607: case in certain models.

1608: 

1609: 

1610: 

1611: 

1612: 

1613: 

1614: 

1615: \subsection{New Crosscaps from Mirror Symmetry}

1616: 

1617: 

1618: \subsubsection{Twisting the Symmetry by

1619: Automorphisms}

1620: 

1621: Let $\omega$ be an automorphism of the chiral algebra ${\mathcal A}$

1622: that acts trivially on the Virasoro subalgebra $\{L_n\}$.

1623: The space $\Scr{H}_i$ acted on by ${\mathcal A}$ through $\omega$,

1624: $W:v\to \omega(W)v$, can be viewed as another representation

1625: $\Scr{H}_{\omega(i)}$ of ${\mathcal A}$.

1626: In other words, there is a unitary isomorphism

1627: \beq

1628: V_{\omega}:\Scr{H}_i\to \Scr{H}_{\omega(i)},

1629: \eeq

1630: such that $\omega(W)=V_{\omega}^{-1}WV_{\omega}$.

1631: % on $\Scr{H}_i$.

1632: 

1633: 

1634: 

1635: 

1636: The algebra ${\mathcal A}$ can be embedded into the symmetry algebra

1637: ${\mathcal A}\otimes {\mathcal A}$ as

1638: $W\mapsto W\otimes 1+1\otimes \omega(W)$.

1639: We can then consider D-branes and orientifolds that preserve

1640: such ``$\omega$-diagonal'' subalgebras \cite{FSbranes,RSGep}.

1641: They are associated with  boundary conditions such that

1642: $W^{(r)}(t,0) 

1643: =\omega \widetilde{W}^{(r)}(t,0)$

1644: and parity symmetries that map

1645: $W^{(r)}(t,x)$ to $\omega\widetilde{W}^{(r)}(t,-x)$.

1646: The conditions on the corresponding boundary and crosscap states

1647: are twisted accordingly: $\widetilde{W}_{-n}^{(r)}$

1648: in (\ref{Bcond}) and (\ref{Ccond})

1649: are replaced by $\omega(\widetilde{W}_{-n}^{(r)})$.

1650: The linear basis of solutions is given by the ``$\omega$-type

1651: Ishibashi states''

1652: \beqa

1653: &&

1654: \kket{i}_{\omega}=(V_{\omega}\otimes {\rm id})\kket{\omega^{-1}(i)},\\

1655: &&

1656: \cket{i}_{\omega}

1657: =(V_{\omega}\otimes {\rm id})\cket{\omega^{-1}(i)}=\e^{\pi i

1658: (L_0-h_i)}\kket{i}_{\omega},

1659: \eeqa

1660: which are sums of elements in

1661: $\Scr{H}_i\otimes\Scr{H}_{\overline{\omega^{-1}(i)}}$.

1662: These states have the same mutual inner-products

1663: as the ordinary Ishibashi states (\ref{bbchi})-(\ref{bcchi}).

1664: Inner products of states with different $\omega$'s

1665: (say $\omega=1$ and $\omega\ne 1$)

1666: are given in terms of so-called ``twining characters''.

1667: For instance,

1668: \beqa

1669: \bbra{j} q^{H} \kket{i}_\omega

1670: &=& \sum_{N,M} \langle j,N|V_{\omega}q^{L_0-\frac{c}{24}}

1671: \ket{\omega^{-1}(i),M}

1672: \langle j,N|q^{L_0-{c\over 24}}

1673: \ket{\omega^{-1}(i),M}^{\dag} \\ \no

1674: &=& \delta_{i,j}\delta_{i,\omega(i)} \ 

1675: \tr_{{\cal H}_i} V_\omega q^{2L_0-\frac{c}{12}} = 

1676: \delta_{i,j}\delta_{i,\omega(i)} \ \chi_j^{(0)} (2\tau).

1677: \eeqa

1678: 

1679: Boundary and crosscap states are linear combinations of

1680: the $\omega$-type Ishibashi-states.

1681: For boundary states,

1682: there is a long list of works that aim at determining the appropriate

1683: linear combinations.

1684: For crosscaps, the same amount of investigation has not been done.

1685: Here, we do not attempt to determine the appropriate combinations

1686: in full generality.

1687: However, we will find that this can be done

1688: in the case where an orbifold is ``mirror'' to the original (in the sense

1689: described below).

1690: The knowledge on the crosscaps for orbifolds turns out useful here.

1691: 

1692: 

1693: 

1694: \subsubsection{Mirror Symmetry}

1695: 

1696: 

1697: 

1698: \newcommand{\mapr}[1]{%

1699:   \smash{\mathop{%

1700:   \hbox to 1cm{\rightarrowfill}}\limits^{#1}}}

1701: \newcommand{\mapdl}[1]{\Big\downarrow

1702:   \llap{$\vcenter{\hbox{$\scriptstyle#1\,$}}$}}

1703: \newcommand{\mapdr}[1]{\Big\downarrow

1704:   \rlap{$\vcenter{\hbox{$\scriptstyle#1\,$}}$}}

1705: 

1706: An automorphism $\omega$ that conjugates the representations

1707: $\omega(i)=\bi$ (any $i$) is called a mirror automorphism.

1708: Two CFTs of symmetry algebra ${\mathcal A}\otimes {\mathcal A}$

1709: are said to be mirror to each other when they are equivalent as 2d quantum

1710: field theories and

1711: the action of $W\otimes W'\in {\mathcal A}\otimes {\mathcal A}$

1712: in one theory

1713: is mapped to the action of $\omega(W)\otimes W'\in

1714: {\mathcal A}\otimes  {\mathcal A}$ in the other.

1715: Namely, if ${\mathcal H}_1$ and ${\mathcal H}_2$ are the Hilbert spaces

1716: of states of the two theories and $\Psi$ is the isomorphism between them,

1717: the following diagram commutes

1718: \beq

1719: \begin{array}{ccc}

1720: {\mathcal H}_1&\mapr{\Psi}&{\mathcal H}_2\\

1721: \mapdl{W\otimes W'\,\,}&&\mapdr{\omega(W)\otimes W'}\\

1722: {\mathcal H}_1&\mapr{\Psi}&{\mathcal H}_2

1723: \end{array}

1724: \eeq

1725: On each ${\mathcal A}\otimes{\mathcal A}$-irreducible subspace,

1726: the isomorphism $\Psi$ acts as $V_{\omega}^{-1}\otimes {\rm id}$ times a

1727: constant.

1728: The two basic modular invariants

1729: --- the charge conjugation modular invariant

1730: ${\mathcal H}^{\mathcal C}=\oplus_i\Scr{H}_i\otimes \Scr{H}_{\bi}$

1731: and the diagonal modular invariant

1732: ${\mathcal H}^{\mathcal D}=\oplus_i\Scr{H}_i\otimes\Scr{H}_i$ ---

1733: are mirror to each other.

1734: 

1735: A typical example of mirror symmetry is T-duality.

1736: The sigma model on the circle of radius $R=\sqrt{k_1/k_2}$

1737: and the model of radius $1/R=\sqrt{k_2/k_1}$, with $k_1, k_2$

1738: integers, are both

1739: RCFTs with chiral algebra $U(1)_{k_1k_2}$.

1740: T-duality between them is a mirror symmetry.

1741: Another example is the level $k$ $SU(2)/U(1)$ gauged WZW model,

1742: which is the charge-conjugate modular invariant of the level $k$

1743: parafermion algebra, and its orbifold by a $\Z_k$ symmetry group,

1744: which is the diagonal modular invariant.

1745: These examples will be studied in detail later in this paper.

1746: A related example is the level $k$ $SU(2)/U(1)$ supersymmetric

1747: gauged WZW model (Kazama-Suzuki model),

1748: which is the charge-conjugate modular invariant of the

1749: level $k$ superparafermion algebra, and its

1750: orbifold by a $\Z_{k+2}\times \Z_2$ symmetry group,

1751: which is the diagonal modular invariant.

1752: This last example will be studied in detail in \cite{BH2}.

1753: In fact, in this example, the two theories are mirror in the standard

1754: sense: the isomorphism of the Hilbert spaces

1755: acts on the $(2,2)$ supersymmetry algebra via

1756: the standard mirror automorphism.

1757: 

1758: 

1759: 

1760: 

1761: \subsubsection{A-branes/B-branes and A-parities/B-parities}

1762: 

1763: 

1764: In what follows, D-branes and parities that preserve

1765: the ordinary diagonal symmetry

1766: ${\mathcal A}\subset {\mathcal A}\otimes {\mathcal A}$

1767: shall be referred to as

1768: {\it A-branes} and {\it A-parities}.

1769: Cardy branes and PSS parities are therefore A-branes and A-parities.

1770: For a mirror automorphism $\omega$,

1771: D-branes and parities that preserve

1772: the $\omega$-diagonal

1773: symmetry shall be referred to as

1774: {\it B-branes} and {\it B-parities}.

1775: A-branes and B-branes are exchanged under mirror symmetry,

1776: and so are A- and B-parities.

1777: Let $\Psi:{\mathcal H}_1\to{\mathcal H}_2$ be a mirror isomorphism as above.

1778: If $|\Scr{B}\rangle_2$ and $|\Scr{C}\rangle_2$ are

1779: the boundary and crosscap states corresponding

1780: to an A-brane and an A-parity in ``theory 2'',

1781: then $\Psi^{-1}|\Scr{B}\rangle_2$ and

1782: $\Psi^{-1}|\Scr{C}\rangle_2$ correspond to

1783: a B-brane and a B-parity in ``theory 1''.

1784: (The terminology of ``A-type'' and ``B-type'' is the extension

1785: of the one used for ${\mathcal N}=2$ supersymmetric theories \cite{OOY,BH2}.

1786: Mirror symmetry for orientifolds is used in \cite{AAHV}

1787: in that context.)

1788: 

1789: An RCFT ${\mathcal C}$ is sometimes

1790: mirror to one of its orbifold models, ${\mathcal C}/G$, as in

1791: the three examples mentioned above --- rational

1792: $U(1)$, $SU(2)/U(1)$ coset model, and supersymmetric $SU(2)/U(1)$ model.

1793: In such a case, one can construct B-branes/B-parities

1794: in the model  ${\mathcal C}$

1795: by applying the mirror isomorphism

1796: $\Psi^{-1}$ to the A-branes/A-parities of the orbifold model,

1797: which are in turn obtained by applying the orbifold technique

1798: developed in the literature and in Section~\ref{subsec:orb}.

1799: 

1800: 

1801: To be specific, let ${\mathcal C}$ be the charge-conjugate modular invariant

1802: and ${\mathcal C}/G$ be the mirror diagonal modular invariant.

1803: ${\mathcal C}/G$

1804: has A-parities $P_{g'}^{\theta}$ with crosscap (\ref{orbcross2}),

1805: labelled by the solutions $\theta$ to (\ref{thetaEq}) and $g'\in\Scr{G}/G$.

1806: Thus, ${\mathcal C}$ has B-parities $P_B^{\theta,g'}$

1807: whose crosscap states are given by

1808: \beq

1809: |\Scr{C}_{P_B^{\theta,g'}}\rangle

1810: =\Psi^{-1}|\Scr{C}_{P_{g'}^{\theta}}\rangle^{{\mathcal C}/G}.

1811: \eeq

1812: ($\Psi^{-1}$

1813: acts as $V_{\omega}\otimes 1$, up to a phase multiplication.)

1814: B-parities obtained this way are not in general involutive.

1815: We recall from (\ref{SqP}) that the square of $P_{g'}^{\theta}$

1816: is the multiplication by

1817: $\e^{2\pi i (\theta(g)-Q_{g'}(g))}$ on

1818: ${\mathcal H}_{g}=\oplus_{i}\Scr{H}_{i}\otimes\Scr{H}_{g(\bi)}$.

1819: Since the orbifold model is the diagonal modular invariant,

1820: only the subspaces with $g(\bi)=i$ remain in the spectrum

1821: of ${\mathcal C}/G$.

1822: Thus, $(P_{g'}^{\theta})^2=\e^{2\pi i (\theta(g)-Q_{g'}(g))}$

1823: on $\Scr{H}_i\otimes \Scr{H}_i$ such that $g(\bi)=i$,

1824: or on $\Scr{H}_{\bi}\otimes \Scr{H}_{\bi}$ such that $g(i)=\bi$.

1825: Since the mirror isomorphism $\Psi^{-1}$ maps

1826: $\Scr{H}_{\bi}\otimes \Scr{H}_{\bi}$ to

1827: $\Scr{H}_i\otimes\Scr{H}_{\bi}$, we find

1828: \beq

1829: (P_B^{\theta,g'})^2

1830: =\e^{2\pi i (\theta(g)-Q_{g'}(g))}\quad

1831: \mbox{on $\Scr{H}_i\otimes\Scr{H}_{\bi}\subset {\mathcal H}^{\mathcal C}$

1832: such that $g(i)=\bi$.}

1833: \label{sqformula}

1834: \eeq

1835: Here we are assuming that $g(i)=\bi$ uniquely fixes $g$.

1836: However, this is not always the case

1837: if there are simple current fixed points.

1838: In such a case, we need to trace back in order to see

1839: from which twisted sector comes the subspace $\Scr{H}_{\bi}\otimes\Scr{H}_{\bi}$ in the orbifold theory. 

1840: 

1841: 

1842: 

1843: 

1844: 

1845: 

1846: \section{Circle of Radius $R$}\label{sec:arbR}

1847: 

1848: 

1849: 

1850: \newcommand{\talpha}{\widetilde{\alpha}}

1851: \newcommand{\tilg}{\widetilde{g}}

1852: \newcommand{\RP}{{\bf R}{\rm P}}

1853: \newcommand{\mDx}{{\mit\Delta} x}

1854: \newcommand{\mDa}{{\mit\Delta} a}

1855: 

1856: 

1857: The sigma model whose target space is

1858: $S^1$ of radius $R$ is described by a periodic scalar field

1859: $X\equiv X+2\pi R$.

1860: The algebra of oscillator modes

1861: $\alpha_n$ and $\talpha_n$ of $X$ acts on the space of states

1862: \beq

1863: {\mathcal H}=\bigoplus_{l,m\in \Z}{\mathcal H}_{l,m},

1864: \eeq

1865: where the labels $l$ and $m$ on the Fock space ${\mathcal H}_{l,m}$ correspond

1866: to the momentum and winding number, respectively. 

1867: We denote by $|l,m\rangle\in {\mathcal H}_{l,m}$ the lowest energy state 

1868: annihilated by the modes $\alpha_n$ and $\talpha_n$ with $n>0$. The energy of

1869: this state is 

1870: ${1\over 2}(({l\over R})^2+(Rm)^2)

1871: -{1\over 12}$.

1872: There are two $U(1)$ symmetries

1873: \beq

1874: g_{{}_{\mDx} }:|l,m\rangle\mapsto \e^{-il\mDx/R}|l,m\rangle,

1875: \quad

1876: \tilg_{{}_{\mDa}}:|l,m\rangle\mapsto \e^{-imR\mDa}|l,m\rangle.

1877: \label{actgg}

1878: \eeq

1879: We interpret $g_{{}_{\mDx}}$  as the rotation of the circle,

1880: $X\to X+\mDx$.

1881: Under T-duality, the sigma model on the circle of radius $R$

1882: is mapped to the model on the circle of radius $1/R$.

1883: The states and operators are mapped as follows

1884: \beq

1885: |l,m\rangle\to |m,l\rangle,\quad

1886: \alpha_n\to -\alpha_n,\quad

1887: \talpha_n\to\talpha_n.

1888: \label{Tduality}

1889: \eeq

1890: The operation $\tilg_{{}_{\mDa}}$ is interpreted as the rotation of

1891: the T-dual circle $X'\to X'+\mDa$.

1892: 

1893: 

1894: \subsection{D-Branes}

1895: 

1896: There are two kinds of D-branes ---

1897: D1-branes and D0-branes associated with the Neumann and Dirichlet

1898: boundary conditions

1899: on $X$ respectively.

1900: We denote by $N_a$  the D1-brane with Wilson line $a$, and

1901: by $D_{x_{}}$

1902: the D0-brane located at $X=x_{}$.

1903: 

1904: 

1905: The Heisenberg algebra $[\alpha_r,\alpha_{r'}]=r\delta_{r+r',0}$

1906: acts on open string states, where $r\in \Z$ for N--N and D--D strings

1907: and $r\in \Z+{1\over 2}$ for D--N and N--D strings.

1908:  The $N_{a_1}$--$N_{a_2}$ string states are labelled

1909: by the momentum $l\in \Z$ and the state $|l\rangle_{a_1,a_2}$ annihilated

1910: by $\alpha_{n>0}$ has the lowest energy

1911: $({l\over R}+{a_2-a_1\over 2\pi})^2-{1\over 24}$.

1912: We assume the identification

1913: $|l\rangle_{a_1,a_2}

1914: =|l+p_1-p_2\rangle_{a_1+{2\pi\over R}p_1,a_2+{2\pi\over R}p_2}$

1915: for integers $p_1,p_2$.

1916: The $D_{x_1}$--$D_{x_2}$ string states are labelled by the winding number

1917: $m\in\Z$ and the state

1918: $|m\rangle_{x_1,x_2}$ annihilated by $\alpha_{n>0}$ has

1919: the lowest energy $(Rm+{x_2-x_1\over 2\pi})^2-{1\over 24}$.

1920: We assume the identification

1921: $|m\rangle_{x_1,x_2}=|m+q_1-q_2\rangle_{x_1+2\pi Rq_1,x_2+2\pi Rq_2}$

1922: for $q_1,q_2 \in \Z$.

1923: 

1924: 

1925: Computing the partition function and performing the modular transform,

1926: we find the boundary states for these branes:

1927: \beqa

1928: &&

1929: |N_a\rangle=\sqrt{R\over \sqrt{2}}

1930: \sum_{m\in\Z}\e^{-i Ram}\exp\left(-\sum_{n=1}^{\infty}

1931: {1\over n}\alpha_{-n}\talpha_{-n}\right)|0,m\rangle,

1932: \label{Nina}

1933: \\

1934: &&

1935: |D_{x_{}}\rangle

1936: =\sqrt{1\over R\sqrt{2}}\sum_{l\in \Z}\e^{-i{x_{}\over R}l}

1937: \exp\left(\sum_{n=1}^{\infty}{1\over n}\alpha_{-n}\talpha_{-n}\right)

1938: |l,0\rangle

1939: \label{Din}

1940: \eeqa

1941: 

1942: The D1-brane wrapped on a circle and the D0-brane in the dual circle

1943: are mapped to

1944: each other under T-duality (\ref{Tduality}),

1945: where the Wilson line of a D1-brane is mapped to the location of the D0-brane.

1946: 

1947: The rotation symmetries (\ref{actgg}) act on the branes and

1948: the open string states as

1949: \beqa

1950: &&g_{{}_{\mDx}}:\left\{

1951: \begin{array}{l}

1952: N_a\to N_a;\quad

1953: |l\rangle_{a_1,a_2}\mapsto

1954: \e^{-i\mDx({l\over R}+{a_2-a_1\over 2\pi})}|l\rangle_{a_1,a_2},\\

1955: D_x\to D_{x+\mDx};\quad

1956: |m\rangle_{x_1,x_2}\mapsto |m\rangle_{x_1+\mDx,x_2+\mDx},

1957: \end{array}\right.

1958: \\

1959: &&\tilg_{{}_{\mDa}}:\left\{

1960: \begin{array}{l}

1961: N_a\to N_{a+\mDa};\quad

1962: |l\rangle_{a_1,a_2}\mapsto

1963: |l\rangle_{a_1+\mDa,a_2+\mDa},

1964: \label{gdX}\\

1965: D_x\to D_x;\quad

1966: |m\rangle_{x_1,x_2}\mapsto

1967: \e^{-i\mDa(Rm+{x_2-x_1\over 2\pi})}

1968: |m\rangle_{x_1,x_2}.

1969: \end{array}\right.

1970: \eeqa

1971: The extra phases, such as $\e^{-i{a_2-a_1\over 2\pi}\mDx}$ in

1972: (\ref{gdX}), come from the parallel transport

1973: of the open string boundary.

1974: Note that $g_{2\pi R}$ and $\tilg_{2\pi\over R}$ act as the identity

1975: on the closed string states, but not on the open string states

1976: for generic values of Wilson lines and positions.

1977: As a consequence, the symmetry group is no longer $U(1)\times U(1)$ but $\R\times\R$.

1978: 

1979: 

1980: \subsection{$\Z_2$ Orientifolds}

1981: 

1982: 

1983: Let us consider the parities

1984: \beqa

1985: &&\Omega:X(t,\s)\to X(t,-\s),\\

1986: &&s\Omega:X(t,\s)\to X(t,-\s)+\pi R.

1987: \eeqa

1988: These act on the states and branes as

1989: \beqa

1990: &&\Omega:\left\{

1991: \begin{array}{l}

1992: |l,m\rangle\to |l,-m\rangle\\

1993: N_a\to N_{-a};\quad |l\rangle_{a_1,a_2}\to |l\rangle_{-a_2,-a_1}\\

1994: D_{x_{}}\to D_{x_{}};

1995: \quad

1996: |m\rangle_{x_1,x_2}\to |-m\rangle_{x_2,x_1}

1997: \end{array}\right.

1998: \label{acOm}

1999: \\

2000: &&s\Omega:\left\{

2001: \begin{array}{l}

2002: |l,m\rangle\to (-1)^l|l,-m\rangle\\

2003: N_a\to N_{-a};\quad |l\rangle_{a_1,a_2}\to

2004: \e^{-\pi i (l+{R(a_1+a_2)\over 2\pi})}|l\rangle_{-a_2,-a_1}\\

2005: D_{x_{}}\to D_{x_{}+\pi R};

2006: \quad

2007: |m\rangle_{x_1,x_2}\to |-m\rangle_{x_2+\pi R,x_1+\pi R}

2008: \end{array}\right.

2009: \label{acsOm}

2010: \eeqa

2011: The parity $s\Omega$ acts in the same way as $g_{\pi R}\Omega$

2012: on the closed string states as well as the DD string states,

2013: but differs from it by an overall phase $\e^{-iRa_1}$

2014: in the action on the NN string states.

2015: This is to make $s\Omega$ involutive

2016: (note that $(g_{\pi R}\Omega)^2=g_{2\pi R}\ne 1$ on NN string states).

2017: Computing the partition functions and making the modular transform,

2018: we find the following expressions for the crosscap states

2019: \beqa

2020: &&|\Scr{C}_{\Omega}\rangle

2021: =\sqrt{R\sqrt{2}}\sum_{m'\in\Z}\exp\left(-\sum_{n=1}^{\infty}

2022: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|0,2m'\rangle,

2023: \label{Om}

2024: \\

2025: &&|\Scr{C}_{s\Omega}\rangle

2026: =\sqrt{R\sqrt{2}}\sum_{m'\in\Z}\exp\left(-\sum_{n=1}^{\infty}

2027: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|0,2m'+1\rangle.

2028: \label{sOm}

2029: \eeqa

2030: 

2031: 

2032: 

2033: 

2034: 

2035: Applying T-duality to (\ref{acOm}),(\ref{acsOm}) and

2036: (\ref{Om}),(\ref{sOm})

2037: in the system of radius $1/R$,

2038: we find two other parity symmetries

2039: \beqa

2040: &&I\Omega:\left\{

2041: \begin{array}{l}

2042: |l,m\rangle\to |-l,m\rangle\\

2043: N_a\to N_{a};\quad |l\rangle_{a_1,a_2}\to |-l\rangle_{a_2,a_1}\\

2044: D_{x_{}}\to D_{-x_{}};

2045: \quad

2046: |m\rangle_{x_1,x_2}\to |m\rangle_{-x_2,-x_1}

2047: \end{array}\right.

2048: \\

2049: &&I'\Omega:\left\{

2050: \begin{array}{l}

2051: |l,m\rangle\to (-1)^m|-l,m\rangle\\

2052: N_a\to N_{a+{\pi\over R}};

2053: \quad |l\rangle_{a_1,a_2}\to |-l\rangle_{a_2+{\pi\over R},a_1+{\pi\over R}}\\

2054: D_{x_{}}\to D_{-x_{}};

2055: \quad

2056: |m\rangle_{x_1,x_2}\to

2057: \e^{-\pi i(m+{x_1+x_2\over 2\pi R})}|m\rangle_{-x_2,-x_1}

2058: \end{array}\right.

2059: \eeqa

2060: with the crosscap states

2061: \beqa

2062: &&|\Scr{C}_{I\Omega}\rangle

2063: =\sqrt{\sqrt{2}\over R}\sum_{l'\in\Z}\exp\left(\sum_{n=1}^{\infty}

2064: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|2l',0\rangle,

2065: \\

2066: &&|\Scr{C}_{I'\Omega}\rangle

2067: =\sqrt{\sqrt{2}\over R}\sum_{l'\in\Z}\exp\left(\sum_{n=1}^{\infty}

2068: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|2l'+1,0\rangle.

2069: \eeqa

2070: We see that they both correspond to the involution

2071: \beq

2072: X(t,\s)\to -X(t,-\s).

2073: \eeq

2074: There are two

2075: orientifold fixed points of $X\to -X$; one at $X=0$ and

2076: another one at $X=\pi R$.

2077: The difference between $I\Omega $ and $I'\Omega$ arises, for instance,

2078: in the $\RP^2$ diagram:

2079: $\langle 0|\Scr{C}_{I\Omega}\rangle=\sqrt{R\sqrt{2}}$

2080: whereas $\langle 0|\Scr{C}_{I'\Omega}\rangle=0$.

2081: In a full string model that contains the circle as one of the

2082: compactified dimensions, taking this overlap corresponds to

2083: determining the  total tension of the orientifold planes. 

2084: The fact that the overlap for $I'\Omega$ vanishes means that the

2085: tensions of the orientifold planes located at the two fixed points

2086: have opposite signs and cancel out. In string theory language,

2087: the orientifold located at one of the fixed points 

2088: is a ${\cal O}^+$-plane, the one at the other a

2089: ${\cal O}^-$-plane. For $I\Omega$ the tensions add up and the

2090: two orientifold planes are of the same type.

2091: 

2092: This can be confirmed by comparing

2093: the action on the open strings stretched between D0-branes.

2094: We consider D0-branes at the fixed  points. 

2095: The state $|0\rangle_{0,0}$ is a state of the

2096: string ending on the D0-brane at $X\equiv 0$ without winding,

2097: whereas $|1\rangle_{\pi R,-\pi R}$ is a state of

2098: the string ending on the D0-brane at $X\equiv \pi R$ without winding.

2099: The action of the two parities on these open string states are

2100: \beq

2101: I\Omega:\left\{\begin{array}{l}

2102: |0\rangle_{0,0}\to |0\rangle_{0,0},\\

2103: |1\rangle_{\pi R,-\pi R}\to

2104: |1\rangle_{\pi R,-\pi R}

2105: \end{array}\right.

2106: \quad

2107: I'\Omega:\left\{\begin{array}{l}

2108: |0\rangle_{0,0}\to |0\rangle_{0,0},\\

2109: |1\rangle_{\pi R,-\pi R}\to

2110: -|1\rangle_{\pi R,-\pi R}

2111: \end{array}\right.

2112: \eeq

2113: We see that the action of $I\Omega$ on the two string states is identical,

2114: while $I'\Omega$ acts on them with opposite signs.

2115: For $I\Omega$ orientifold, the orientifold planes at $X=0$ and $X=\pi R$

2116: are both of $SO$-type.

2117: For the $I'\Omega$ orientifold,

2118: the orientifold plane at $X=0$ is of $SO$-type and the one at $X=\pi R$

2119: is of $Sp$-type.

2120: 

2121: 

2122: Since $I'\Omega$

2123: is obtained from T-duality applied to $s\Omega$,

2124: we have proved that the 

2125: $s\Omega$ orientifold is T-dual to the orientifold with $Sp/SO$ mixture.

2126: This was observed in the context of superstring theory in \cite{nvs}.

2127: 

2128: 

2129: 

2130: 

2131: \subsection{Combination with Rotations I: Involutive Parities}

2132: 

2133: One can combine  the parity symmetries considered above

2134: and the rotation symmetries $g_{{}_{\mDx}}$ and $\tilg_{{}_{\mDa}}$.

2135: We first consider the combinations of the form

2136: $gPg^{-1}$.

2137: They are involutive

2138: so that the crosscap states belong to the ordinary space

2139: of states ${\mathcal H}$.

2140: In fact, in order to find the crosscap state we can use the recipe

2141: given in Appendix~\ref{app:some}:

2142: $g|\Scr{C}_P\rangle=|\Scr{C}_{gPg^{-1}}\rangle$.

2143: 

2144: 

2145: Let us first consider the

2146: parity $g_{{}_{\mDx\over 2}}I\Omega g_{{}_{\mDx\over 2}}^{-1}$,

2147: which is actually the same as $g_{{}_{\mDx}}I\Omega$.

2148: The crosscap state is obtained by applying

2149: $g_{{}_{\mDx/2}}$ to $|\Scr{C}_{I\Omega}\rangle$:

2150: \beq

2151: |\Scr{C}_{g_{{}_{\mDx}}I\Omega}\rangle

2152: =\sqrt{\sqrt{2}\over R}\sum_{l'\in\Z}

2153: \e^{-il'\mDx /R}\exp\left(\sum_{n=1}^{\infty}

2154: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|2l',0\rangle.

2155: \label{CrdI}

2156: \eeq

2157: One can also consider $g_{{}_{\mDx\over 2}}I'\Omega g_{{}_{\mDx\over 2}}^{-1}$.

2158: It is the same as $g_{{}_{\mDx}}I'\Omega$

2159: in the action on the closed string and N--N strings,

2160: but differs from it in the action on the D--D strings.

2161: We therefore denote it  as

2162: $\widetilde{g_{{}_{\mDx}}I'\Omega}$.

2163: \beq

2164: |\Scr{C}_{\widetilde{g_{{}_{\mDx}}I'\Omega}}\rangle

2165: =\sqrt{\sqrt{2}\over R}\sum_{l'\in\Z}

2166: \e^{-i(l'+{1\over 2})\mDx /R}\exp\left(\sum_{n=1}^{\infty}

2167: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|2l'+1,0\rangle,

2168: \label{CrdIp}

2169: \eeq

2170: Both $g_{{}_{\mDx}}I\Omega$

2171: and 

2172: $\widetilde{g_{{}_{\mDx}}I'\Omega}$

2173: act on the free boson as

2174: $$

2175: X(t,\s)\to -X(t,-\s)+\mDx.

2176: $$

2177: This action has two fixed points

2178: at $X={\mDx\over 2}$ and $X={\mDx\over 2}+\pi R$.

2179: If we move $\mDx$ from $0$ to $2\pi R$,

2180: under which the two fixed points are exchanged,

2181: the crosscap

2182: for $I\Omega$ comes back to itself but the one for $I'\Omega$ comes back

2183: with a sign flip.

2184: This is because the two orientifold planes are of the same type

2185: for $I\Omega$ (both $SO$-type),

2186: while they are of the opposite type for $I'\Omega$ (one is $SO$-type and

2187: the other is $Sp$-type).

2188: 

2189: 

2190: 

2191: We next consider the parity symmetries

2192: $\tilg_{{}_{\mDa}}\Omega=\tilg_{{}_{\mDa\over 2}}\Omega

2193: \tilg_{{}_{\mDa\over 2}}^{-1}$

2194: and $\widetilde{\tilg_{{}_{\mDa}}s\Omega}:=

2195: \tilg_{{}_{\mDa\over 2}}s\Omega

2196: \tilg_{{}_{\mDa\over 2}}^{-1}$.

2197: The crosscap states for these parities are

2198: \beqa

2199: &&

2200: |\Scr{C}_{\tilg_{{}_{\mDa}}\Omega}\rangle

2201: =\sqrt{R\sqrt{2}}\sum_{m'\in\Z}\e^{-im'R\mDa}

2202: \exp\left(-\sum_{n=1}^{\infty}

2203: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|0,2m'\rangle,

2204: \\

2205: &&

2206: |\Scr{C}_{\widetilde{\tilg_{{}_{\mDa}}s\Omega}}\rangle

2207: =\sqrt{R\sqrt{2}}\sum_{m'\in\Z}\e^{-i(m'+{1\over 2})R\mDa}

2208: \exp\left(-\sum_{n=1}^{\infty}

2209: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|0,2m'+1\rangle.\qquad

2210: \eeqa

2211: We note that $\tilg_{{}_{\mDa}}\Omega$ as well as

2212: $\widetilde{\tilg_{{}_{\mDa}}s\Omega}$ change the Wilson line as

2213: $a\to -a+\mDa$.

2214: 

2215: 

2216: The symmetry $\tilg_{{}_{\mDa}}$ commutes with the parities

2217: $I\Omega$ and $I'\Omega$, whereas

2218: $g_{{}_{\mDx}}$ commutes with

2219: $\Omega$ and $s\Omega$.

2220: Thus, there is no dressing of the form $gPg^{-1}$

2221: other than the ones considered above.

2222: 

2223: 

2224: 

2225: \subsection{Combination with Rotation II: Non-involutive Parities}

2226: 

2227: We can also consider parities that are not involutive.

2228: An example is $g_{{}_{\mDx}}\Omega: X(t,\s)\to X(t,\s)+\mDx$,

2229: where $(g_{{}_{\mDx}}\Omega)^2=g_{{}_{2\mDx}}$.

2230: The crosscap states for such a parity $P$ are not in the ordinary

2231: space of states but in the space of states with twisted boundary condition

2232: determined by $P^2$.

2233: We note that the effect of the twisting by

2234: $g_{{}_{\mDx}}$ and $\tilg_{{}_{\mDa}}$

2235: is just to shift the momentum and winding number

2236: by $-{R\mDa\over 2\pi}$ and $-{\mDx\over 2\pi R}$:

2237: \beq

2238: {\mathcal H}_{g_{{}_{\mDx}}\tilg_{{}_{\mDa}}}

2239: =\bigoplus_{l\in \Z-

2240: {R\mDa\over 2\pi},

2241: m\in\Z-{\mDx\over 2\pi R}}

2242: {\mathcal H}_{l,m}.

2243: \eeq

2244: 

2245: 

2246: The Wilson line of a D1-brane is preserved

2247: under the rotation $g_{{}_{\mDx}}$, while

2248: the position of a D0-brane is preserved by

2249: $\tilg_{{}_{\mDa}}$. Thus one can consider the $g_{{}_{\mDx}}$-twisted

2250: boundary state for $N_a$ and the $\tilg_{{}_{\mDa}}$-twisted

2251: boundary state for $D_x$.

2252: Computing the twisted partition function and making the modular transform,

2253: we obtain the following expressions for these twisted boundary states:

2254: \beqa

2255: &&

2256: |N_a\rangle_{g_{{}_{\mDx}}}

2257: =\sqrt{R\over \sqrt{2}}

2258: \sum_{m\in\Z}

2259: \e^{-i Ram}\exp\left(-\sum_{n=1}^{\infty}

2260: {1\over n}\alpha_{-n}\talpha_{-n}\right)

2261: |0,m-\mbox{$\mDx\over 2\pi R$}\rangle,

2262: \\

2263: &&

2264: |D_{x_{}}\rangle_{\tilg_{{}_{\mDa}}}

2265: =\sqrt{1\over R\sqrt{2}}\sum_{l\in\Z}

2266: \e^{-i{x_{}\over R}l}

2267: \exp\left(\sum_{n=1}^{\infty}{1\over n}\alpha_{-n}\talpha_{-n}\right)

2268: |l-\mbox{$R\mDa\over 2\pi$},0\rangle.

2269: \eeqa

2270: Note that these states change by a phase under $\mDx\to\mDx+2\pi R$

2271: and $\mDa\to\mDa+{2\pi\over R}$

2272: because of the parallel transport involved in the action of

2273: $g_{{}_{\mDx}}$ and $\tilg_{{}_{\mDa}}$.

2274: 

2275: 

2276: 

2277: 

2278: The crosscap state for $g_{{}_{\mDx}}\Omega$ consists of states in

2279: ${\mathcal H}_{g_{{}_{2\mDx}}}$ and is given by

2280: \beq

2281: |\Scr{C}_{g_{{}_{\mDx}}\Omega}\rangle

2282: =\sqrt{R\sqrt{2}}\sum_{m'\in\Z}

2283: \exp\left(-\sum_{n=1}^{\infty}

2284: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|0,2m'-

2285: \mbox{${\mDx\over \pi R}$}

2286: \rangle.

2287: \label{CgdX}

2288: \eeq

2289: The crosscap state for $\tilg_{{}_{\mDa}}I\Omega$

2290: is a sum of states in ${\mathcal H}_{\tilg_{{}_{2\mDa}}}$:

2291: \beq

2292: |\Scr{C}_{\tilg_{{}_{\mDa}}I\Omega}\rangle

2293: =\sqrt{\sqrt{2}\over R}\sum_{l'\in\Z}

2294: \exp\left(\sum_{n=1}^{\infty}

2295: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)

2296: |2l'-\mbox{${R\mDa\over \pi}$},0\rangle.

2297: \label{Ctgda}

2298: \eeq

2299: (\ref{CgdX}) interpolates between the ${\Omega}$ crosscap

2300: and the $s\Omega$ crosscap. Recall that

2301: $g_{\pi R}\Omega$ and $s\Omega$ are the same except for the difference

2302: in the action on N--N string states by

2303: an overall phase.

2304: The same can be said of (\ref{Ctgda}).

2305: 

2306: 

2307: One can also consider the parities

2308: $g_{{}_{\mDx}}\tilg_{{}_{\mDa}}\Omega$

2309: or $g_{{}_{\mDx}}\tilg_{{}_{\mDa}}I\Omega$.

2310: The crosscap states are

2311: given by

2312: \beqa

2313: &&|\Scr{C}_{g_{{}_{\mDx}}\tilg_{{}_{\mDa}}\Omega}\rangle

2314: =\sqrt{R\sqrt{2}}\sum_{m'\in\Z}

2315: \e^{-iR\mDa m'}

2316: \exp\left(-\sum_{n=1}^{\infty}

2317: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)|0,2m'-

2318: \mbox{${\mDx\over \pi R}$}

2319: \rangle,\qquad

2320: \label{CxaO}\\

2321: &&|\Scr{C}_{g_{{}_{\mDx}}\tilg_{{}_{\mDa}}I\Omega}\rangle

2322: =\sqrt{\sqrt{2}\over R}\sum_{l'\in\Z}

2323: \e^{-i{\mDx\over R}l'}

2324: \exp\left(\sum_{n=1}^{\infty}

2325: {(-1)^n\over n}\alpha_{-n}\talpha_{-n}\right)

2326: |2l'-\mbox{${R\mDa\over \pi}$},0\rangle.

2327: \label{CxaIO}

2328: \eeqa

2329: The crosscap for $g_{{}_{\mDx}}\tilg_{{}_{\mDa}}\Omega$

2330: is obtained by applying

2331: $\tilg_{{}_{\mDa\over 2}}$ to

2332: $|\Scr{C}_{g_{{}_{\mDx}}\Omega}\rangle$

2333: and multiplying by the phase $\e^{-i{\mDa\mDx\over 2\pi}}$.

2334: The application of

2335: $\tilg_{{}_{\mDa\over 2}}$ is because

2336: $g_{{}_{\mDx}}\tilg_{{}_{\mDa}}\Omega

2337: =\tilg_{{}_{\mDa\over 2}}g_{{}_{\mDx}}\Omega

2338: \tilg_{{}_{\mDa\over 2}}^{-1}$.

2339: The extra phase

2340: arises because 

2341: $\tilg_{{}_{\mDa}}|N_a\rangle_{g_{{}_{\mDx}}}$

2342: is not just $|N_{a+\mDa}\rangle_{{g_{{}_{\mDx}}}}$

2343: but has an  extra phase

2344: $\e^{i{\mDa\mDx\over 2\pi}}\,\,$

2345: \footnote{In fact,

2346: ${}_{g_{{}_{2\mDx}}}\langle N_a|q^{H_c}

2347: |\Scr{C}_{g_{{}_{\mDx}}\tilg_{{}_{\mDa}}\Omega}\rangle$

2348: should be the same as

2349: ${}_{g_{{}_{2\mDx}}}\langle N_{a-{\mDa\over 2}}|q^{H_c}

2350: |\Scr{C}_{g_{{}_{\mDx}}\Omega}\rangle$.

2351: The claimed construction

2352: of $|\Scr{C}_{g_{{}_{\mDx}}\tilg_{{}_{\mDa}}\Omega}\rangle$

2353: follows because

2354: ${}_{g_{{}_{2\mDx}}}\langle N_{a-{\mDa\over 2}}|

2355: =\e^{-i{\mDa\mDx\over 2\pi}}

2356: {}_{g_{{}_{2\mDx}}}\langle N_{a}|\tilg_{{}_{\mDa\over 2}}$.}.

2357: The same can be said of

2358: $|\Scr{C}_{g_{{}_{\mDx}}\tilg_{{}_{\mDa}}I\Omega}\rangle$.

2359: 

2360: 

2361: 

2362: \section{Rational $U(1)$}\label{sec:ratR}

2363: 

2364: \newcommand{\wtn}{\widetilde{n}}

2365: 

2366: 

2367: Let us consider the case $R=\sqrt{k}$ for a positive integer $k$.

2368: We can now use two approaches to construct D-branes and orientifolds:

2369: on the one hand, we can insert the special value of $R$ in the formulae

2370: worked out in the previous section.

2371: On the other hand, the boson at this

2372: particular radius is described by a rational conformal theory, so

2373: that  the methods developed in Section \ref{general} can be applied. 

2374: Needless to say,

2375: the two approaches lead to the same results.

2376: 

2377: Let us briefly review the basic structure of the rational conformal field

2378: theory description, and in particular collect the ingredients for the

2379: construction of Section \ref{general}. 

2380: At the radius $R=\sqrt{k}$,

2381: the system has two copies of chiral algebra ${\mathcal A}=U(1)_k$.

2382: One copy is generated by

2383: the spin 1 and spin $k$ currents 

2384: $$J=\sqrt{k}(\partial_t-\partial_{\s})X\,\,\,\,

2385: {\rm and}\,\,\,\,

2386: W_{\pm}=\e^{\pm 2i\sqrt{k} X_R},$$

2387: while the other copy

2388: is generated by 

2389: $$\widetilde{J}=-\sqrt{k}(\partial_t+\partial_{\s})X\,\,\,\,

2390: {\rm and}\,\,\,\, \widetilde{W}_{\pm}=\e^{\mp 2i\sqrt{k}X_L}.$$

2391: Since $W_{\pm}$ has $J$-charge $\pm 2k$, the representation of $U(1)_k$

2392: is labelled by a modulo $2k$ integer, $n$, and the representation space

2393: is denoted by $\Scr{H}_n$.

2394: Note that the state $|l,m\rangle$ of momentum $l$ and winding number $m$

2395: has $(J,\widetilde{J})$-charge $(l-km,-l-km)$.

2396: Thus, one may relabel the states as

2397: $$

2398: |l,m\rangle=|l-km\rangle\otimes |-l-km\rangle

2399: =|n+2kp\rangle\otimes |-n-2k\tilde{p}\rangle,

2400: $$

2401: where we have made the

2402: reparametrization

2403: $l=n+k(p+\tilde{p})$ and $m=-p+\tilde{p}$.

2404: If $l$ and $m$ run over integers,

2405: then $(n,p,\tilde{p})$ runs over $\Z_{2k}\oplus\Z\oplus\Z$.

2406: Thus, the space of states is given by

2407: $$

2408: {\mathcal H}=\bigoplus_{l,m}

2409: {\mathcal H}_{l,m}

2410: =\bigoplus_{n\in\Z_{2k}}\Scr{H}_n\otimes\Scr{H}_{-n}.

2411: $$

2412: The primary states are labelled by a mod $2k$ integer, $n\in\Z_{2k}$,

2413: and the fusion rules are simply given by the addition modulo $2k$.

2414: We choose the range $n=-k+1,\dots, k$

2415: as a fundamental domain for $\Z_{2k}$.

2416: For any integer $n$, we denote by $\hat{n}$ the representative

2417: in this fundamental domain.

2418: We also denote the addition of

2419: two labels $n,m$ mod $2k$ by $\hat{+}$, such that

2420: $n\hat{+}m \in \{-k+1,\dots, k\}$.

2421: The conformal weight of the primary field with label $n$ is

2422: $$

2423: h_n=\frac{\hat{n}^2}{4k}.

2424: $$

2425: 

2426: \subsubsection*{\it Modular Matrices}

2427: 

2428: The modular $T$ and $S$ matrices are

2429: \beq

2430: T_{nn'} = \delta_{n,n'}\e^{\frac{\pi i n^2}{2k}- \frac{i\pi}{12}},\qquad

2431: S_{nn'} = \frac{1}{\sqrt{2k}} e^{-\frac{i\pi nn'}{k}}.

2432: \eeq

2433: For orientifold constructions one needs in addition the

2434: $P=\sqrt{T}ST^2S\sqrt{T}$ and

2435: $Y$ matrices.

2436: In order to compute them, it is convenient to introduce related matrices

2437: that do not involve $\sqrt{T}$:

2438: \beq

2439: Q=ST^2S

2440: \label{defSTS}

2441: \eeq

2442: and

2443: \beq

2444: \tilde{Y}_{ab}^c = \sum_d \frac{S_{ab} Q_{bd}Q^*_{cd}}{S_{0d}}

2445: =\sqrt{\frac{T_c}{T_b}} \ Y_{ab}^c.

2446: \label{defYtilde}

2447: \eeq

2448: The absence of $\sqrt{T}$ makes the computation easier, and

2449: we find

2450: \beqa

2451: &&Q_{nn'} =  \frac{1}{\sqrt{k}} \e^{\pi i\over 12}

2452: \e^{-\frac{i\pi}{4k}(n+n')^2} \ \delta^{(2)}_{n+n'+k},

2453: \nn\\

2454: &&\tilde{Y}_{nn'}^{n''} = \e^{-\frac{\pi i}{4k}(-{n''}^2+{n'}^2)} \ 

2455: \delta^{(2)}_{n'+n''} \

2456: \left( \delta^{(2k)}_{n+\frac{n'-n''}{2}} + (-1)^{n'+k} \

2457: \delta^{(2k)}_{n+\frac{n'-n''}{2}+k} \right).

2458: \nn

2459: %\label{Ytilde}

2460: \eeqa

2461: The $P$-matrix and $Y$-tensor are then found to be

2462: \beqa

2463: \label{u1p}

2464: &&P_{nn'}=\delta_{k+n+n'}^{(2)}{1\over \sqrt{k}}

2465: \e^{-\pi i {\hat{n}\hat{n}'\over 2k}},\\

2466: &&Y_{nn'}^{n''} = 

2467: \delta^{(2)}_{n'+n''} \

2468: \left( \delta^{(2k)}_{n+\frac{\hat{n}'-\hat{n}''}{2}} + (-1)^{n'+k} \

2469: \delta^{(2k)}_{n+\frac{\hat{n}'-\hat{n}''}{2}+k} \right).

2470: \label{Yt}

2471: \eeqa

2472: 

2473: \subsubsection*{\it Discrete Symmetry}

2474: 

2475: The group of simple currents is the group of primaries itself,

2476: $\Scr{G}=\Z_{2k}$. The charge $Q_n(n')$ is given by

2477: $$

2478: Q_n(n')={n^2\over 4k}+{(n')^2\over 4k}-{(n+n')^2\over 4k}

2479: =-{nn'\over 2k}\qquad\mbox{mod 1}.

2480: $$

2481: Thus, we find a discrete symmetry group $\Z_{2k}$ generated by an element

2482: $g$ that acts on $\Scr{H}_n\otimes\Scr{H}_{-n}$ by phase multiplication

2483: $\e^{-\pi i {n\over k}}\times$.

2484: In terms of the symmetries $g_{\mDx}$ and $\widetilde{g}_{\mDa}$,

2485: this generator can be expressed as

2486: \beq

2487: g=g_{\pi R\over k}\widetilde{g}_{\pi\over R},

2488: \eeq

2489: where $R=\sqrt{k}$ is understood.

2490: One can also show that among $g_{\mDx}\widetilde{g}_{\mDa}$

2491: the symmetries that commute with the algebra

2492: $U(1)_k\otimes U(1)_k$ are of the form

2493: $g_{\pi Rn\over k}\widetilde{g}_{\pi n\over R}=g^n$ for some $n\in\Z_{2k}$.

2494: 

2495: \subsubsection*{\it T-duality}

2496: 

2497: The T-dual model has radius $1/R=1/\sqrt{k}=R/k$, and can actually be

2498: regarded as the orbifold by the group

2499: $G=\Z_k$ generated by $g_{2\pi R/k}=g^2$.

2500: As a representation of the

2501: $U(1)_k\otimes U(1)_k$ algebra, the space of states

2502: is given by the diagonal modular invariant

2503: $$

2504: {\mathcal H}^{\rm T-dual}

2505: =\bigoplus_n\Scr{H}_n\otimes\Scr{H}_n.

2506: $$

2507: T-duality induces the mirror automorphism $M:\alpha_n\to -\alpha_n$, and

2508: the map of states $\Psi:{\mathcal H}\to

2509: {\mathcal H}^{\rm T-dual}$ is given by $|l,m\rangle_R\mapsto

2510: |m,l\rangle_{1\over R}$, which reads in the RCFT language as

2511: $$

2512: \Psi=V_M\otimes 1:|q\rangle\otimes |\widetilde{q}\rangle

2513: \mapsto |-q\rangle\otimes |\widetilde{q}\rangle.

2514: $$

2515: 

2516: 

2517: \subsection{A-parities}

2518: 

2519: 

2520: 

2521: A-branes and A-parities correspond to 

2522: the Cardy states and the PSS crosscaps

2523: \beqa

2524: &&

2525: |\Scr{B}_n\rangle

2526: ={1\over (2k)^{1\over 4}}

2527: \sum_{n'\in\Z_{2k}}\e^{-\pi i{nn'\over k}}\kket{n'},\\

2528: &&

2529: |\Scr{C}_n\rangle

2530: ={(2k)^{1\over 4}\over\sqrt{k}}\sum_{n'\in\Z_{2k}}

2531: \e^{-\pi i{\hat{n}\hat{n}'\over 2k}}\delta^{(2)}_{n+n'+k}\cket{n'}.

2532: \eeqa

2533: 

2534: To find out the geometrical meaning of these branes and parities,

2535: we express these states in

2536: terms of the basis $|l,m\rangle$ labelled by momentum

2537: and winding number.

2538: We first re-express Ishibashi states:

2539: \beqa

2540: &&\kket{n}=

2541: \sum_{p\in\Z}\e^{\sum{1\over m}\alpha_{-m}\talpha_{-m}}

2542: |n+2kp\rangle\otimes |-n-2kp\rangle

2543: =\sum_{p\in\Z}\e^{\sum{1\over m}\alpha_{-m}\talpha_{-m}}

2544: |n+2kp,0\rangle,\quad

2545: \nn\\

2546: &&\cket{n}=\e^{\pi i (L_0-h_n)}\kket{n}

2547: =\e^{-\pi i h_n}\sum_{p\in\Z}

2548: \e^{\sum{(-1)^m\over m}\alpha_{-m}\talpha_{-m}}\e^{\pi i {(n+2kp)^2\over 4k}}

2549: |n+2kp,0\rangle.

2550: \nn

2551: \eeqa

2552: Then, the Cardy states are expressed as

2553: \beqa

2554: |\Scr{B}_n\rangle

2555: &=&

2556: {1\over (2k)^{1\over 4}}

2557: \sum_{n'\in\Z_{2k}}\e^{-\pi i{nn'\over k}}\sum_{p\in\Z}

2558: \e^{\sum{1\over m}\alpha_{-m}\talpha_{-m}}|n'+2kp,0\rangle

2559: \nn\\

2560: &=&{1\over (2k)^{1\over 4}}

2561: \sum_{l\in\Z}\e^{-\pi i {nl\over k}}\e^{\sum{1\over m}\alpha_{-m}\talpha_{-m}}

2562: |l,0\rangle

2563: \nn\\

2564: &=&|D_{\pi R{n\over k}}\rangle.

2565: \eeqa

2566: Thus, the $n$-th Cardy state is identified as

2567: the D0-brane located at the point $X=2\pi R{n\over 2k}$

2568: of the circle.

2569: To express the PSS crosscaps, it is convenient to use the

2570: $Q$-matrices, 

2571: \beqa\label{PSSU1}

2572: |\Scr{C}_n\rangle

2573: &=&

2574: \e^{\pi i (h_n-{1\over 12})}\sum_{n'}{Q_{nn'}\over\sqrt{S_{0n'}}}

2575: \e^{\pi i L_0}\kket{n'}

2576: \nn\\

2577: &=&

2578: \e^{\pi i (h_n-{n^2\over 4k})}

2579: \left({2\over k}\right)^{1\over 4}

2580: \sum_{l''\in\Z}\e^{-\pi i {n\over k}(l''+{n+k\over 2})}

2581: \e^{\sum{(-1)^m\over m}\alpha_{-m}\talpha_{-m}}

2582: |2l''+n+k,0\rangle

2583: \nn\\

2584: &=&

2585: \left\{

2586: \begin{array}{ll}

2587: |\Scr{C}_{g_{\pi Rn\over k}I\Omega}\rangle&\mbox{$(n+k)$ even}

2588: \\

2589: |\Scr{C}_{\widetilde{g_{\pi R\hat{n}\over k}I'\Omega}}\rangle

2590: &\mbox{$(n+k)$ odd}

2591: \end{array}\right.

2592: \eeqa

2593: In the last step, we used the crosscap formulae

2594: (\ref{CrdI}) and (\ref{CrdIp}) for the parity symmetries

2595: studied

2596: in Section~\ref{sec:arbR}.

2597: We see that the PSS parities are associated with reflections of the circle.

2598: The $n$-th PSS parity has orientifold fixed points

2599: at the diametrically opposite points,

2600: $X=2\pi R\frac{n}{4k}$ and  $X=2\pi R\frac{n}{4k}+\pi R$.

2601: For $n$ even, the location of the orientifold points 

2602: coincides with the location of the Cardy branes

2603: $\Scr{B}_{n/2}$ and $\Scr{B}_{n/2+k}$. 

2604: For $n$ odd, the fixed points are halfway

2605: between possible locations of Cardy branes. 

2606: Furthermore, the two

2607: orientifold points are both of the same type for $n+k$ even,

2608: whereas they are of the opposite type for $n+k$ odd. 

2609: 

2610: We will now see how this information is encoded in the $Y$ tensor of the RCFT.

2611: Using (\ref{Yt}) it is straightforward to write down the 

2612: A-type M\"obius strips:

2613: \beqa

2614: \label{MAU1}

2615: \langle\Scr{C}_n|q_t^{H}|\Scr{B}_m\rangle

2616: &=&\sum_{n'} Y_{mn'}^n \widehat\chi_{n'}(\tau)

2617: =\pm\widehat{\chi}_{n-2m}(\tau);\nn\\

2618: \pm &=&\left\{\begin{array}{ll}

2619: 1&\mbox{if $\widehat{n-2m}=\hat{n}-2m$ mod $4k$}\\

2620: (-1)^{n+k}&

2621: \mbox{if $\widehat{n-2m}=\hat{n}-2m+2k$ mod $4k$}.

2622: \end{array}

2623: \right.

2624: \eeqa

2625: Since $\langle \Scr{B}_{m'}|q_t^H|\Scr{B}_{m}\rangle

2626: =\sum_{n'}N_{n'm}^{m'}\chi_{n'}(\tau)=\chi_{m'-m}(\tau)$,

2627: we see that the $P_n$-image of the Cardy brane

2628: $\Scr{B}_{m}$ is $\Scr{B}_{n-m}$.

2629: (Actually, this also follows from the general rule

2630: (\ref{Pgi}).)

2631: In particular, for $n$ even, the Cardy branes $\Scr{B}_{n/2}$ 

2632: and $\Scr{B}_{n/2+k}$ are

2633: left invariant, confirming that these branes are located at the orientifold

2634: fixed points.

2635: The two cases in

2636: (\ref{MAU1}) are interchanged under the shift

2637: $m\to m+k$.

2638: In particular,

2639: the amplitude

2640: flips its sign under the exchange $\Scr{B}_m\leftrightarrow \Scr{B}_{m+k}$

2641: if and only if $n+k$ is odd.

2642: We note that the Cardy branes $\Scr{B}_m$ and $\Scr{B}_{m+k}$

2643: are located at diametrically opposite points.

2644: In this way, the RCFT data encode the fact that the crosscaps

2645: with $n+k$ odd lead to orientifold projections of different types

2646: at the two fixed points, whereas crosscaps for $n+k$ even give rise

2647: to the same projection.

2648: 

2649: For completeness, let us also write the Klein bottles, which are

2650: \beq

2651: \langle\Scr{C}_n\ket{\Scr{C}_l}

2652: = \sum_m \delta^{(2)}_{n+l} \

2653: \left( \delta^{(2k)}_{m+\frac{n-l}{2}} + (-1)^{k+n} \ 

2654: \delta^{(2k)}_{m+\frac{n-l}{2} +k}

2655: \right) \chi_m

2656: \eeq

2657: 

2658: 

2659: 

2660: \subsection{B-parities}

2661: 

2662: We next study B-parities.

2663: To find B-crosscaps in our model, we first find A-crosscaps

2664: in the mirror

2665: $\Z_k$-orbifold model, and then bring them back by the mirror map.

2666: 

2667: To find the A-crosscaps in the orbifold model, we apply

2668: the method of Section~\ref{subsec:orb}.

2669: The bilinear form $q$ of the group $G=\Z_k$

2670: is uniquely fixed by the requirement

2671: $q(n,n)=-h_n=-{n^2\over 4k}$ (mod 1)

2672: and is given by

2673: \beq

2674: q(n,m)=-{nm\over 4k}, \quad

2675: n,m\,\,\mbox{even}.

2676: \eeq

2677: Note that it is well-defined, namely

2678: invariant (mod 1) under $2k$ shifts of $n$ and $m$ 

2679: since both of them are even.

2680: Note also that

2681: $2q(n,m)=-nm/2k=Q_n(m)$ (mod 1) as required.

2682: To write down the Eq.~(\ref{thetaEq}) for $\theta$,

2683: we first note that

2684: \beqa

2685: -\hat{Q}_n(m)+2q(n,m)

2686: &=&-{\hat{n}^2\over 4k}-{\hat{m}^2\over 4k}+{(n\hat{+}m)^2\over 4k}

2687: -{nm\over 2k}

2688: \nn\\

2689: &=&{(n\hat{+}m)^2\over 4k}-{(\hat{n}+\hat{m})^2\over 4k}

2690: ={n\hat{+}m\over 2}-{\hat{n}+\hat{m}\over 2}\quad\,\, \mbox{mod 2}.

2691: \,\,\,\,\,\,\,

2692: \label{Qcomp}

2693: \eeqa

2694: Thus, the equation is $\theta(n+m)=\theta(n)+\theta(m)+{n\hat{+}m\over 2}

2695: -{\hat{n}+\hat{m}\over 2}$ and the solutions are

2696: \beq

2697: \theta_l(n)

2698: ={\hat{n}\over 2}+{nl\over k},\qquad

2699: l\in \Z/k\Z.

2700: \eeq

2701: Then, the crosscap states (\ref{orbcross1}) and (\ref{orbcross2})

2702: are given by

2703: \beqa

2704: &&|\Scr{C}_{P^{\theta_l}_{0}}\rangle

2705: ={\e^{i\omega_0}\over \sqrt{k}}\sum_{n:{\rm even}}

2706: \e^{-\pi i ({\hat{n}\over 2}+{nl\over k})}

2707: |\Scr{C}_n\rangle

2708: =\e^{i\omega_0}(2k)^{1\over 4}\cket{-2l-k}.

2709: \nn\\

2710: &&|\Scr{C}_{P^{\theta_l}_{1}}\rangle

2711: ={\e^{i\omega_1}\over \sqrt{k}}\sum_{n:{\rm even}}

2712: \e^{-\pi i ({\hat{n}\over 2}+{nl\over k}-Q_1(n))}

2713: |\Scr{C}_{n+1}\rangle

2714: =\e^{i\omega_1+\pi i{\widehat{2l+1+k}\over 2k}}(2k)^{1\over 4}\cket{-2l-1-k}.

2715: \nn

2716: \eeqa

2717: In the latter we have chosen $n'=1$ as the representative of the

2718: non-trivial element of $\Scr{G}/G=\Z_{2k}/\Z_k=\Z_2$.

2719: The phases $\e^{i\omega_0}$ and $\e^{i\omega_1}$ can be tuned

2720: so that no phases appear in front of the Ishibashi states.

2721: 

2722: 

2723: B-parities in the original model are obtained from these by the mirror map.

2724: We denote the mirror images of

2725: $P_0^{\theta_l}$ and $P_1^{\theta_l}$

2726: by $P_B^{2l+k}$ and $P_B^{2l+1+k}$ respectively.

2727: Since the mirror map

2728: $\Psi^{-1}$ sends the Ishibashi states $\cket{-n}$ to

2729: B-type Ishibashi states $\cket{n}_B$,

2730: we find that the crosscap states are given by

2731: \beq

2732: |\Scr{C}_{P_B^n}\rangle

2733: =(2k)^{1\over 4}\cket{n}_B.

2734: \label{CPB}

2735: \eeq

2736: These parities are not necessarily involutive.

2737: Applying  formula (\ref{sqformula})

2738: we find that

2739: \beq

2740: (P_B^{n})^2=g^{2n}.

2741: \label{PBsqu}

2742: \eeq

2743: The crosscap state (\ref{CPB}) belongs to

2744: the space $\Scr{H}_{n}\otimes\Scr{H}_{n}$.

2745: Since $n=2n+(-n)=g^{2n}(\bar n)$,

2746: $\Scr{H}_{n}\otimes\Scr{H}_{n}

2747: =\Scr{H}_{n}\otimes\Scr{H}_{g^{2n}(\bar n)}$

2748: is a space of states with $g^{2n}$-twisted boundary condition.

2749: Namely, $|\Scr{C}_{P_B^{n}}\rangle$ is a $g^{2n}$-twisted

2750: state, which is consistent with (\ref{PBsqu}).

2751: 

2752: 

2753: Next, let us  examine the geometrical interpretation of these

2754: parity symmetries.

2755: To this end, we express the Ishibashi states in terms of the

2756: $|l,m\rangle$ basis.

2757: \beqa

2758: &&\kket{n}_B=(V_M\otimes 1)\kket{-n}

2759: =(V_M\otimes 1)\sum_{p\in\Z}\e^{\sum{1\over m}\alpha_{-m}\talpha_{-m}}

2760: |-n+2kp\rangle\otimes |n-2kp\rangle

2761: \nn\\

2762: &&\qquad\qquad

2763: =\sum_{p\in\Z}\e^{-\sum{1\over m}\alpha_{-m}\talpha_{-m}}

2764: |n-2kp\rangle\otimes |n-2kp\rangle

2765: =\sum_{p\in \Z}\e^{-\sum{1\over m}\alpha_{-m}\talpha_{-m}}

2766: |0,2p-\mbox{$n\over k$}\rangle,

2767: \nn\\

2768: &&\cket{n}_B

2769: =\e^{\pi i (L_0-h_n)}\kket{n}_B

2770: =\e^{\pi i({n^2\over 4k}-h_n)}

2771: \sum_{p\in\Z}

2772: \e^{-\pi i(n+k)p}

2773: \e^{-\sum{(-1)^m\over m}\alpha_{-m}\talpha_{-m}}

2774: |0,2p-\mbox{$n\over k$}\rangle.

2775: \nn

2776: \eeqa

2777: Comparing the latter with the formula (\ref{CxaO}), we find that

2778: \beq

2779: |\Scr{C}_{\tilg_{{\pi(n+k)\over R}}g_{\pi Rn\over k}\Omega}\rangle

2780: =\e^{-\pi i({n^2\over 4k}-h_n)}

2781: (2k)^{1\over 4}\cket{n}_B.

2782: \label{Ccc}

2783: \eeq

2784: The crosscap state (\ref{CPB}) is equal to

2785: (\ref{Ccc}), up to

2786: an overall sign, which is $+$ if we choose $n=\hat{n}$.

2787: Thus, we conclude that the RCFT parities are interpreted as

2788: \beq

2789: P_B^{n}

2790: =g_{\pi R\hat{n}\over k}\tilde{g}_{\pi(\hat{n}+k)\over R}\Omega.

2791: \eeq

2792: If $n+k$ is even, $P_B^n$ is equal to $g_{\pi R\hat{n}\over k}\Omega$,

2793: which is simply the worldsheet orientation reversal

2794: $\Omega$ followed by the $n/2k$ rotation of the circle.

2795: If $n+k$ is odd, $\widetilde{g}_{\pi (n+k)\over R}$

2796: is non-trivial, and $P_B$ is not just

2797: $\Omega$ followed by the $n/2k$ rotation, but it acts by extra sign

2798: multiplication on odd-winding states.

2799: Note that 

2800: $g_{\pi Rn\over k}\tilde{g}_{\pi(n+k)\over R}

2801: =g^n\tilde{g}_{\pi k\over R}$ and hence

2802: $P_B^n=g^{\hat{n}}\tilde{g}_{\pi k\over R}\Omega$.

2803: Since $(\tilde{g}_{\pi k\over R}\Omega)^2=1$,

2804: this also explains (\ref{PBsqu}).

2805: 

2806: \subsubsection{Klein Bottles}

2807: 

2808: We record here the Klein bottle amplitudes:

2809: \beqa

2810: \Tr\!\mathop{}_{{\mathcal H}}\!

2811: P_B^n q^H&=&

2812: \langle\Scr{C}_{P_B^n}|\e^{-{\pi i\over 2\tau}H}|\Scr{C}_{P_B^n}\rangle

2813: =\sqrt{2k}\chi_n(-1/2\tau)

2814: \nn\\

2815: &=&\sum_m\e^{-\pi i {mn\over k}}\chi_m(2\tau)

2816: =\sum_m\e^{2\pi i Q_n(m)}\chi_m(2\tau).

2817: \eeqa

2818: This indeed shows that $P_B^{n}=g^nP_B^0$ on the closed string states.

2819: One could also consider

2820: $\langle\Scr{C}_{P_B^{n+k}}|q_t^H|\Scr{C}_{P_B^{n}}\rangle$,

2821: which is interpreted as

2822: $\Tr_{{\mathcal H}_{g^k}}P_B^nq^H$. This vanishes.

2823: 

2824: 

2825: 

2826: \subsubsection{M\"obius Strips}

2827: 

2828: 

2829: Let us compute the M\"obius strip amplitudes.

2830: Since $P_B^n$ are not in general involutive (\ref{PBsqu}),

2831: we need to find the boundary states on a circle with $g^{2n}$-twisted 

2832: boundary condition.

2833: They are obtained via mirror symmetry from

2834: the twisted boundary states for the A-branes in the orbifold model

2835: ${\mathcal C}/G$.

2836: 

2837: To find the A-brane boundary states in the orbifold model, we use the

2838: method developed in Section~\ref{subsubsec:Borb}.

2839: The symmetry $g^{2n}$ in the original model is mapped to the quantum symmetry

2840: in the orbifold model associated with the character

2841: $\rho_n$ of $G$ defined by $\rho_n(m)=mn/2k$ ($m$ even).

2842: Applying  formula (\ref{bir}) with $i=n'=0,1$, we find

2843: \beq

2844: |\Scr{B}_{[n']}\rangle^{{\mathcal C}/G}_{\rho_n}

2845: =\e^{i\lambda+{\pi inn'\over k}}\left({k\over 2}\right)^{1\over 4}

2846: (\kket{-n}+(-1)^{n'}\kket{-n-k}).

2847: \eeq

2848: We choose the phase $\lambda$ so that $\e^{i\lambda+{\pi inn'\over k}}=1$.

2849: The (twisted) boundary states

2850: for B-branes in the original model

2851: are obtained by applying the mirror map

2852: $\Psi^{-1}$ to these states:

2853: \beq

2854: |\Scr{B}^B_{[n']}\rangle_{g^{2n}}

2855: =

2856: \left({k\over 2}\right)^{1\over 4}

2857: (\kket{n}_B+(-1)^{n'}\kket{n+k}_B),\qquad

2858: n'=0,1.

2859: \eeq

2860: 

2861: 

2862: To find the geometrical meaning, we express them

2863: in terms of the $|l,m\rangle$ basis.

2864: Using the expression for $\kket{n}_B$ obtained above, we find

2865: \beqa

2866: &&|\Scr{B}^B_{[0]}\rangle_{g^{2n}}

2867: =\sqrt{\sqrt{k}\over\sqrt{2}}

2868: \sum_{r\in \Z}\e^{-\sum{1\over m}\alpha_{-m}\talpha_{-m}}

2869: |0,r-\mbox{${n\over k}$}\rangle

2870: =|N_0\rangle_{g_{2\pi Rn\over k}},\\

2871: &&|\Scr{B}^B_{[1]}\rangle_{g^{2n}}

2872: =\sqrt{\sqrt{k}\over\sqrt{2}}\sum_{r\in \Z}(-1)^r

2873: \e^{-\sum{1\over m}\alpha_{-m}\talpha_{-m}}

2874: |0,r-\mbox{${n\over k}$}\rangle

2875: =|N_{\pi\over R}\rangle_{g_{2\pi Rn\over k}}.

2876: \eeqa

2877: Thus, $\Scr{B}^B_{[0]}$ is the D1-brane wrapped on $S^1$ with trivial

2878: Wilson line, while $\Scr{B}^B_{[1]}$ is the

2879: D1-brane with Wilson line $\pi$ along the $S^1$.

2880: Using the action of $\Omega$, $g_{\mDx}$ and $\widetilde{g}_{\mDa}$

2881: on the D-branes studied in Section~\ref{sec:arbR},

2882: we see that

2883: $P_B^n=g_{\pi R\hat{n}\over k}\tilde{g}_{\pi(\hat{n}+k)\over R}\Omega$

2884: maps the brane $N_a$ to $N_{-a+{\pi (\hat{n}+k)\over R}}$. In particular,

2885: the B-branes $\Scr{B}^B_{[0]}$ and $\Scr{B}^B_{[1]}$ are invariant

2886: under $P_B^n$ with even $n+k$,

2887: while they are exchanged with each other under

2888: $P_B^n$ with odd $n+k$.

2889: 

2890: Let us see how the RCFT data encode this information.

2891: It is straightforward to compute the open string partition function

2892: $\Tr_{ab}(g^{2n}q^H)

2893: ={}_{g^{2n}}\langle \Scr{B}_{[a]}^B|q_t^H|\Scr{B}_{[b]}^B\rangle_{g^{2n}}$:

2894: \beqa

2895: &&

2896: \Tr_{00}(g^{2n}q^H)

2897: =\Tr_{11}(g^{2n}q^H)

2898: =\sum_{m:\,{\rm even}}\e^{-\pi i {mn\over k}}\chi_m(\tau)

2899: \nn\\

2900: &&

2901: \Tr_{01}(g^{2n}q^H)

2902: =\Tr_{10}(g^{2n}q^H)

2903: =\sum_{m:\,{\rm odd}}\e^{-\pi i {mn\over k}}\chi_m(\tau).

2904: \nn

2905: \eeqa

2906: This shows that 0--0 and 1--1 string states have even charges under $U(1)_k$

2907: and 1-0 and 0-1 string states have odd charges.

2908: (Also, $g^{2n}$ acts on the charge $m$ representation as

2909: the phase multiplication by $\e^{-\pi i {nm\over k}}$.)

2910: On the other hand, the M\"obius strip amplitudes are

2911: \beq

2912: {}_{g^{2n}}\langle \Scr{B}_{[a]}^B|q_t^H|\Scr{C}_{P_B^n}\rangle

2913: =\left\{\begin{array}{ll}

2914: {\displaystyle \sum_{m:\,{\rm even}}\e^{-\pi i {\hat{n}\hat{m}\over 2k}}

2915: \widehat{\chi}_m(\tau)}&

2916: \mbox{$n+k$ even},\\[0.5cm]

2917: {\displaystyle \sum_{m:\,{\rm odd}}\e^{-\pi i {\hat{n}\hat{m}\over 2k}}

2918: \widehat{\chi}_m(\tau)}&

2919: \mbox{$n+k$ odd},

2920: \end{array}

2921: \right.

2922: \eeq

2923: for both $a=0,1$.

2924: They are to be identified with the twisted open string partition functions

2925: $\Tr_{[a],P_B^n[a]}(P_B^nq^H)$.

2926: For $n+k$ even, only even charge states propagate,

2927: which means that the brane $\Scr{B}_{[a]}^B$ is preserved under

2928: the parity $P_B^n$.

2929: For odd $n+k$, only odd charge states propagate,

2930: which means that $\Scr{B}^B_{[0]}$ is exchanged with $\Scr{B}^B_{[1]}$.

2931: 

2932: 

2933: 

2934: 

2935: 

2936: \section{Parity Symmetry of (gauged) WZW Models}\label{sec:gWZW}

2937: 

2938: 

2939: In this section, we study Parity symmetry of

2940: the WZW model on a group manifold $G$,

2941: and the $G$ mod $H$ gauged WZW model in which 

2942: the vectorial rotation $g\mapsto h^{-1}gh$

2943: is gauged.

2944: We will focus on the case $G=SU(2)$, the group of $2\times 2$

2945: unitary matrices of determinant 1, and its

2946: diagonal subgroup $H=U(1)$.

2947: We denote the Lie algebras of these groups as

2948: $\fg$ and $\fh$.

2949: 

2950: 

2951: \subsection{The Models}\label{subsec:Model}

2952: 

2953: 

2954: Let $\Sigma$ be a $1+1$ dimensional worldsheet.

2955: The level $k$ WZW action \cite{Wnab}

2956: for a map $g:\Sigma\to SU(2)$

2957:  in a background gauge field $A\in \Omega^1(\Sigma,\fg)$ is given by

2958: \beqa

2959: S_k(A,g)&=&{k\over 8\pi}\int_{\Sigma}

2960: \tr\left(g^{-1}D^{\mu}g\,g^{-1}D_{\mu}g\right)\dd^2x

2961: \nn\\

2962: &&+{k\over 12\pi}\int_B\tr\left(\tilde{g}^{-1}\dd \tilde{g}\right)^3

2963: -{k\over 4\pi}\int_{\Sigma}\tr\left\{A(g^{-1}\dd g+\dd g g^{-1})

2964: +Ag^{-1}Ag\right\},

2965: \label{WZWaction}

2966: \eeqa

2967: where $B$ is a 3-manifold bounded by the worldsheet,

2968: $\partial B=\Sigma$, $\tilde{g}$ is an extension to it,

2969: and

2970: $g^{-1}D_{\mu}g:=g^{-1}\partial_{\mu}g+g^{-1}A_{\mu}g-A_{\mu}$.

2971: Let us consider the {\it chiral}

2972: gauge transformation

2973: \beq

2974: \begin{array}{l}

2975: g\to h_1^{-1}g h_2,\\

2976: A_+\to h_1^{-1}A_+h_1+h_1^{-1}\partial_+h_1,\\

2977: A_-\to h_2^{-1}A_-h_2+h_2^{-1}\partial_-h_2,\\

2978: \end{array}

2979: \eeq

2980: for any $h_i$ with values in $G$ or its complexification $G_c$. 

2981: (Here we used the light-cone coordinate $x^{\pm}=t\pm \s$, with

2982: $\partial_{\pm}={\partial\over \partial x^{\pm}}

2983: ={1\over 2}(\partial_t\pm\partial_{\s})$.)

2984: This changes the action according to

2985: the Polyakov--Wiegmann (PW) identity \cite{PW,Gaw}:

2986: \beq

2987: S_k(A,g)\to S_k(A^{h_1,h_2},h_1^{-1}gh_2)=S_k(A,g)-S_k(A,h_1h_2^{-1}).

2988: \label{PWid}

2989: \eeq

2990: In particular, it is invariant under

2991: the vectorial transformations

2992: $g\to h^{-1}gh,

2993: A\to h^{-1}Ah+h^{-1}\dd h$.

2994: The action $S_k(A,g)$ can also be defined

2995: when $A$ is a connection of a topologically

2996: non-trivial $G/Z_G$-bundle on $\Sigma$ (where $Z_G$ is the center

2997: of $G$) and $g$ is a section of the

2998: associated adjoint bundle, so that the PW-identity still holds \cite{FIH}.

2999: 

3000: 

3001: The level $k$ WZW model is the theory of the variable $g$

3002: with the action $S_k(g)=S_k(0,g)$.

3003: As a consequence of the PW identity (\ref{PWid}),

3004: the action $S_k(g)$ is invariant under

3005: \beq

3006: g\to h_1(x^-)^{-1}gh_2(x^+).

3007: \eeq

3008: The corresponding currents

3009: ($X\in\fg$),

3010: \beqa

3011: &&J_n(X)={-k\over 2\pi i}\int_0^{2\pi}

3012: \tr\left(\partial_- g g^{-1}X\right)\e^{in(t-\s)}\dd\s,

3013: \label{Jint}\\

3014: &&\wtJ_n(X)={k\over 2\pi i}\int_0^{2\pi}

3015: \tr\left(g^{-1}\partial_+ g X\right)

3016: \e^{in(t+\s)}\dd\s,

3017: \label{wtJint}

3018: \eeqa

3019: obey the $SU(2)\times SU(2)$ current algebra relations at level $k$.

3020: %\beqa

3021: %&&[J_n(X),J_m(Y)]=J_{n+m}([X,Y])+kn\delta_{n+m,0}\tr(XY),\\

3022: %&&[\wtJ_n(X),\wtJ_m(Y)]=\wtJ_{n+m}([X,Y])+kn\delta_{n+m,0}\tr(XY).

3023: %\eeqa

3024: The Hilbert space of states decomposes into the irreducible representations

3025: of this algebra $\whfg\oplus\whfg$.

3026: Only the integrable representations

3027: $\V_j=\V_j^{G,k}$ appear, where the spin $j$ ranges

3028: over the set ${\rm P}_k=\{0,{1\over 2},1,...,{k\over 2}\}$:

3029: \beq

3030: {\cal H}^{G,k}=\bigoplus_{j\in {\rm P}_k}

3031: \V_{j}\otimes \wtV_{j}.

3032: \label{HilbG}

3033: \eeq

3034: %The currents $J_n(X)$ and $\wtJ_n(X)$ act

3035: %on the first and the second factors of $\V_j\otimes \wtV_j$ respectively.

3036: The system is a conformal field theory with

3037: $c={3k\over k+2}$.

3038: %The partition function is given by

3039: %$

3040: %Z^{G,k}

3041: %=\sum_{j\in{\rm P}_k}|\chi^{}_{j}(q)|^2,

3042: %$

3043: %where $\chi^{}_{j}(q)=\Tr_{\V_{j}}q^{L_0-c/24}$.

3044: The spin $j\in {\rm P}_k$ representation $V_j$ of $SU(2)$ is included in $\V_j$

3045: as the space of Virasoro primary states

3046: with $h_j={j(j+1)\over k+2}$, and the matrix elements

3047: of $g$ in $V_j$ are the Virasoro primary fields corresponding to

3048: the states in

3049: $V_j\otimes V_j\subset\V_{j}\otimes \wtV_{j}$.

3050: In particular, for spin ${1\over 2}$ representation,

3051: we have the relation

3052: \beq

3053: g=\left(\begin{array}{cc}

3054: g_{11}&g_{12}\\

3055: g_{21}&g_{22}

3056: \end{array}

3057: \right)

3058: \leftrightarrow

3059: \left(\begin{array}{cc}

3060: -|{1\over 2},-\half\rangle\otimes |{1\over 2},\half\rangle &

3061: -|{1\over 2},-\half\rangle\otimes |{1\over 2},-\half\rangle\\

3062: |{1\over 2},\half\rangle\otimes |{1\over 2},\half\rangle&

3063: |{1\over 2},\half\rangle\otimes |{1\over 2},-\half\rangle

3064: \end{array}

3065: \right),

3066: \label{FS}

3067: \eeq

3068: where $|j,m\rangle$ is the basis of $V_j$ with

3069: $\sigma_3/2=m$.

3070: The minus signs in (\ref{FS})

3071: originate in the relation $v^a=\epsilon^{ab}v_{b}$

3072: defining the isomorphism

3073: $V_{1\over 2}\cong V_{1\over 2}^{\vee}$.

3074: The relation for the higher spin representations can be obtained from this

3075: by using the realization $V_j={\rm Sym}^{2j}V_{1\over 2}$.

3076: 

3077: \subsubsection*{\it Gauged WZW Models}

3078: 

3079: 

3080: We next consider the $SU(2)$ mod $U(1)$ gauged WZW model.

3081: This is the model with the action $S_k(A,g)$ where

3082: $A$ also varies over $U(1)$ gauge fields.

3083: To be precise, the gauge group $H\cong U(1)$ is the diagonal

3084: subgroup of $SU(2)$ divided by the center $\Z_2=\{\pm 1\}$.

3085: %,

3086: %which is generated by $i\sigma_3/2$ via the exponential mapping

3087: %of periodicity $2\pi$.

3088: The model is a conformal field theory with the central charge

3089: $c={3k\over k+2}-1$, which is known as the $\Z_k$ parafermion system.

3090: Let us decompose the representation $\V_{j}$ of $\whfg$

3091: into the irreducible representations of the subalgebra

3092: $\whfh$ generated by $J_n(\sigma_3)$;

3093: \beq

3094: \V_{j}=\bigoplus_{n} \B_{j,n}\otimes \V_{-n}.

3095: \eeq

3096: The sum is over the eigenvalue of $-J_0(\sigma_3)$,

3097: which are integers such that $2j+n$ is even.

3098: The space $\B_{j,n}$ can be identified as the subspace of

3099: $\V_j$ consisting of states obeying $J_m(\sigma_3)=0$, $m\geq 1$, and

3100: $J_0(\sigma_3)=-n$.

3101: The physical states of the $G$ mod $H$ model are

3102: the gauge invariant states of the WZW model,

3103: which satisfy

3104: \beqa

3105: &&

3106: J_m(v)|{\rm phys}\rangle=\wtJ_m(v)|{\rm phys}\rangle=0,

3107: \,\,\,m\geq 1,\,v\in \fh,

3108: \label{Hinv1}\\

3109: &&

3110: \bigl(J_0(v)+\wtJ_0(v)\bigr)|{\rm phys}\rangle=0,

3111: \,\,\,v\in \fh.

3112: \label{Hinv2}

3113: \eeqa

3114: The subspace of $\V_{j}\otimes\wtV_{j}$

3115: obeying this condition can be identified as

3116: $\oplus_{n}\B_{j,n}\otimes

3117: \wtB_{j,-n}$.

3118: However, the space of states is not the whole sum of these spaces.

3119: Since the gauge group has a non-trivial fundamental group

3120: $\pi_1(U(1))=\Z$, there are

3121: large gauge transformations which relate the physical states, acting on

3122: the labels as $(j,n)\to ({k\over 2}-j,n+k)\to (j,n+2k)\to\cdots$.

3123: The space of states is found by

3124: selecting one member from each orbit,

3125: \beq\label{parahilbert}

3126: {\cal H}^{G/H,k}=\bigoplus_{(j,n)\in {\rm PF}_k}\,

3127: \B_{j,n}\otimes

3128: \wtB_{j,-n},

3129: \eeq

3130: where the sum is over 

3131: $$

3132: {\rm PF}_k={\{(j,n)\in {\rm P}_k\times \Z;2j+n\,{\rm even}\}\over\pi_1(U(1))}

3133: ={\{(j,n)\in {\rm P}_k\times\Z_{2k};

3134: 2j+n\,{\rm even}\}\over (j,n)\equiv ({k\over 2}-j,n+k)}.

3135: $$

3136: 

3137: 

3138: The $H$-valued chiral rotations $g\to h_1(x^-)^{-1}gh_2(x^+)$

3139: commute with the gauge group and 

3140: shift the action according to the PW-identity (\ref{PWid}):

3141: \beq

3142: S_k(A,h_1^{-1}(x^-)gh_2(x^+))

3143: =S_k(A,g)+{k\over 2\pi}\int_{\Sigma}\tr(F_A\log(h_1h_2^{-1})),

3144: \eeq

3145: where $F_A=\dd A$ is the curvature of the gauge potential $A$.

3146: For the constant $h_1,h_2$ with $h_1^{-1}=h_2=\exp(i\alpha\sigma_3/2)$,

3147: the shift is $-k\alpha\times {i\over 2\pi}\int\tr(F_A\sigma_3)$.

3148: Since $\tr((\sigma_3/2)\sigma_3)=1$,

3149: the integral ${i\over 2\pi}\int\tr(F_A\sigma_3)$ is an integer

3150: on a compact space. Thus, the path-integral weight $\e^{iS_k(A,g)}$

3151: is invariant if $k\alpha\in 2\pi \Z$.

3152: Therefore, the system has a symmetry generated by

3153: \beq\label{Defa}

3154: a:g\to \e^{\pi i\sigma_3/k}g\e^{\pi i\sigma_3/k}.

3155: \label{genZk}

3156: \eeq

3157: This is an order $k$ symmetry since $(\e^{\pi i\sigma_3/k})^k=

3158: \e^{\pi i\sigma_3}=-1$ acts trivially on $g$. 

3159: Thus the system has an axial $\Z_k$ symmetry.

3160: The ``axial anomaly'' $U(1)\to\Z_k$ can also be seen in the operator

3161: formulation.

3162: The axial rotation $g\to \e^{i\alpha\sigma_3/2}g\e^{i\alpha\sigma_3/2}$

3163: acts on the $\B_{j,n}\otimes\wtB_{j,-n}$ subspace

3164: as a multiplication by

3165: $\e^{-i\alpha (-n)/2}\times\e^{i\alpha n/2}=\e^{i \alpha n}$.

3166: However, it is consistent with the field identification

3167: $(j,n)\equiv ({k\over 2}-j,n+k)$ only if $\alpha k\in 2\pi \Z$.

3168: The rotation (\ref{genZk}) acts on this subspace

3169: by multiplication by $\e^{2\pi i n/k}$, which is indeed well-defined.

3170: The two reasonings are of course related since the field identification

3171: originates from large gauge transformations that produce

3172: topologically non-trivial gauge field configurations.

3173: 

3174: 

3175: 

3176: A geometric picture can be given to the model.

3177: We parametrize the group element and the gauge field by

3178: \beq

3179: g=\e^{i(\phi+t)\sigma_3/2}\e^{i\theta \sigma_1}\e^{i(\phi-t)\sigma_3/2}

3180: =\left(\begin{array}{cc}

3181: \e^{i\phi}\cos\theta&i\e^{it}\sin\theta,\\

3182: i\e^{-it}\sin\theta&\e^{-i\phi}\cos\theta

3183: \end{array}\right),

3184: \label{parametrization}

3185: \eeq

3186: and $A={i\over 2}\sigma_3a_{\mu}\dd x^{\mu}$.

3187: The gauge transformation $h=\e^{i\lambda\sigma_3/2}$

3188: acts on these variables as $t\to t-\lambda$,

3189: $a_{\mu}\to a_{\mu}+\partial_{\mu}\lambda$.

3190: %The action (\ref{WZWaction}) can be written as

3191: %\beqa

3192: %S&=&{k\over \pi}\int_{\Sigma}

3193: %\dd^2x\Bigl\{\,\partial_-\theta\partial_+\theta

3194: %+\cot^2\theta\partial_-\phi\partial_+\phi\\

3195: %&&\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,

3196: %+\sin^2\theta(a_-+\partial_-t+\cot^2\theta\partial_-\phi)

3197: %(a_++\partial_+t-\cot^2\theta\partial_+\phi)

3198: %\,\Bigr\}

3199: %\nonumber

3200: %\eeqa

3201: Integrating out the gauge field $a_{\mu}$, we obtain the sigma model

3202: on the space with metric

3203: \beq

3204: \dd s^2=k\Bigl[\,(\dd\theta)^2+\cot^2\theta(\dd\phi)^2\,\Bigr].

3205: \label{discmetric}

3206: \eeq

3207: In terms of the complex coordinate $z=\e^{i\phi}\cos\theta$,

3208: it is the disk $|z|\leq 1$ with the metric $\dd s^2=k|\dd z|^2/(1-|z|^2)$.

3209: As discussed in \cite{MMS}, the string coupling appears to diverge

3210: at the boundary $|z|=1$, but it is simply because of the choice of

3211: variables.

3212: The $\Z_k$ symmetry (\ref{genZk}) acts on the coordinates as the

3213: shift $\phi\to \phi+2\pi/k$, or equivalently $z\to \e^{2\pi i/k}z$

3214: --- the rotation of the disk with angle

3215: $2\pi/k$.

3216: 

3217: 

3218: 

3219: 

3220: 

3221: 

3222: 

3223: 

3224: 

3225: 

3226: \subsection{Parity Symmetry of WZW Models}

3227: \label{subsec:PbWZW}

3228: 

3229: 

3230: The WZW action $S_k(g)$

3231: is not invariant under the simple Parity transformation

3232: \beq

3233: \Omega: t\to t,\,\,\,\s\to -\s,

3234: \eeq

3235: since the WZ term $\int_B\tr(\tilde{g}^{-1}\dd\tilde{g})^3$ flips its sign.

3236: However, if $\Omega$ is combined with the transformation

3237: \beq

3238: {\cal I}: g\to g^{-1},

3239: \eeq

3240: it is invariant because $g^{-1}\dd g\to g\dd g^{-1}\to-g(g^{-1}\dd g)g^{-1}$

3241: yields an extra minus sign to the WZ term.

3242: The kinetic term is of course invariant under both $\Omega$ and ${\cal I}$.

3243: Thus, WZW model has a Parity symmetry $P={\cal I}\Omega$.

3244: Under this symmetry,

3245: the currents (\ref{Jint}) and (\ref{wtJint})

3246: transform as

3247: \beq

3248: \begin{array}{l}

3249: J_n(X)\to \wtJ_n(X),\\

3250: \wtJ_n(X)\to J_n(X).

3251: \end{array}

3252: \label{Pcurr}

3253: \eeq

3254: In particular, the right-moving highest weight state of spin $j$

3255: is mapped to a left-moving highest weight state of spin $j$,

3256: and vice-versa.

3257: This shows that the Parity symmetry acts on the states

3258: as the right--left exchange

3259: $P:u\otimes v \mapsto \pm v\otimes u$,

3260: up to the sign that may depend on the spin $j$ of the state.

3261: To fix the sign, we recall the field-state correspondence

3262: (\ref{FS}). Since $g\to g^{-1}$ sends the spin ${1\over 2}$

3263: matrix elements as $g_{11}\leftrightarrow g_{22},

3264: g_{12}\to -g_{12}$ and $g_{21}\to -g_{21}$, we find that the sign is $-1$ for

3265: $j={1\over 2}$.

3266: The sign for higher $j$ is $(-1)^{2j}$, since $V_j$ is realized as

3267: the symmetric product of $2j$ copies of $V_{1\over 2}$.

3268: Thus, we find that the action of $P$ is given by

3269: \beq

3270: P:u\otimes v\in \V_j\otimes\wtV_j\longmapsto

3271: (-1)^{2j}v\otimes u\in \V_j\otimes\wtV_j.

3272: \eeq

3273: The partition function with $P$-twist is

3274: \beq

3275: \Tr^{}_{\cal H}(P\e^{2\pi i\tau H})

3276: =\sum_{j\in {\rm P}_k}(-1)^{2j}\chi^{}_{j}(2\tau),

3277: \label{Ppf}

3278: \eeq

3279: where $\tau$ is a positive imaginary number.

3280: 

3281: Variants of the above involution ${\cal I}$ can be considered.

3282: One is an involution

3283: \beq

3284: {\cal I}^{-}:g\to -g^{-1}.

3285: \eeq

3286: The model is invariant under $P^{-}={\cal I}^{-}\Omega$, and

3287: the currents transform in the same way as (\ref{Pcurr}).

3288: Since $P^{-}$ is the composition of $P$ and multiplication by

3289: the center $-{\bf 1}_2$, which is represented as $(-1)^{2j}$ on

3290: $\V_j$,

3291: it acts on the states as

3292: \beq

3293: P^{-}:

3294: u\otimes v\in \V_j\otimes\wtV_j\longmapsto

3295: v\otimes u\in \V_j\otimes\wtV_j.

3296: \eeq

3297: The $P^{-}$-twisted partition function is given by

3298: \beq

3299: \Tr^{}_{\cal H}(P^{-}\e^{2\pi i\tau H})

3300: =\sum_{j\in{\rm P}_k} \chi^{}_{j}(2\tau).

3301: \label{zPpf}

3302: \eeq

3303: More general involutions are

3304: \beq

3305: {\cal I}^{\pm}_{g_0}:g\to \pm g_0g^{-1}g_0,

3306: \eeq

3307: for any element $g_0$ of $G$.

3308: $P_{g_0}={\cal I}_{g_0}^{\pm}\Omega$ is also a Parity symmetry of the model.

3309: The currents transform as

3310: \beq

3311: \begin{array}{l}

3312: J_n(X)\to \wtJ_n(g_0Xg_0^{-1}),\\

3313: \wtJ_n(X)\to J_n(g_0^{-1}Xg_0).

3314: \end{array}

3315: \eeq

3316: Since $P_{g_0}$ is the composition of $P^{\pm}$

3317: and  $g\to g_0gg_0$ (where $P^+:=P$), it acts on the states as

3318: \beq

3319: P_{g_0}^{\pm}:u\otimes v\in \V_j\otimes\wtV_j\longmapsto

3320: (\mp 1)^{2j} g_0^{-1}v\otimes g_0u\in

3321: \V_j\otimes\wtV_j.

3322: \eeq

3323: The twisted partition function is independent of $g_0$

3324: and reduces to (\ref{Ppf}) for $P^+_{g_0}$

3325: and (\ref{zPpf}) for $P^-_{g_0}$.

3326: 

3327: 

3328: 

3329: 

3330: 

3331: 

3332: 

3333: 

3334: 

3335: 

3336: 

3337: \subsection{Parity Symmetry of Gauged WZW Models}

3338: \label{subsec:PbgWZW}

3339: 

3340: Now, we study Parity symmetries of

3341: the $SU(2)$ mod $U(1)$ gauged WZW model.

3342: 

3343: 

3344: Under the involution $g\to g^{-1}$,

3345: the covariant derivative is transformed

3346: as $g^{-1}D_{\mu}g\to -g (g^{-1}D_{\mu}g)g^{-1}$ with

3347: the gauge field $A$ fixed.

3348: Thus, the kinetic term is invariant under

3349: \beq

3350: {\cal I}_A:(A,g)\to (A,g^{-1}).

3351: \eeq

3352: On the other hand,

3353: the WZ term --- second line of (\ref{WZWaction}) --- flips its sign

3354: under ${\cal I}_A$.

3355: Thus, $P_A={\cal I}_A\Omega$

3356: is a Parity symmetry of the gauged WZW model.

3357: Note that $\Omega$ exchanges the $\pm$ components of the gauge

3358: field: $(\Omega A)_{\pm}(t,\s)=A_{\mp}(t,-\s)$.

3359: Another Parity is $P_B={\cal I}_B\Omega$

3360: where 

3361: \beq

3362: {\cal I}_B:(A,g)\to(g_*^{-1}Ag_*,g_*^{-1}g^{-1}g_*),

3363: \eeq 

3364: \beq

3365: g_*:=i\sigma_1=\left(\begin{array}{cc}

3366: 0&i\\

3367: i&0

3368: \end{array}\right).

3369: \eeq

3370: Conjugation by $g_*$ preserves the $U(1)$ subgroup,

3371: acting as $g_*^{-1}hg_*=h^{-1}$.

3372: Thus, a $U(1)$ bundle with connection $A$ is mapped to

3373: another $U(1)$ bundle with connection $-A$.

3374: By PW-identity (\ref{PWid}), we have $kS(g_*^{-1}Ag_*,g_*^{-1}gg_*)=kS(A,g)$.

3375: Thus, $g_*$-conjugation is a symmetry of the system.

3376: The Parity $P_B$ is obtained by combining $P_A$ with this symmetry.

3377: Note that $g_*$ can be replaced by $g_*h_1$ for any $h_1\in H$

3378: by a gauge transformation.

3379: The two involutions ${\cal I}_A$ and ${\cal I}_B$ act on the

3380: disk coordinate $z=\e^{i\phi}\cos\theta$ as

3381: \beqa

3382: &&{\cal I}_A:z\to\bz,

3383: \\

3384: &&{\cal I}_B:z\to z.

3385: \eeqa

3386: In fact, up to the $\Z_k$ axial rotations, these are the only

3387: Parity symmetry obtained by starting from the involutions of $SU(2)$

3388: of the type $g\to\pm g_0g^{-1}g_0$.

3389: 

3390: 

3391: These Parity symmetries act on the states as

3392: \beqa

3393: &&P_A:u\otimes v\in \B_{j,n}\otimes\wtB_{j,-n}\longmapsto

3394: (-1)^{2j}v\otimes u\in \B_{j,-n}\otimes \wtB_{j,n},

3395: \label{bPA}\\[0.1cm]

3396: &&P_B:u\otimes v\in \B_{j,n}\otimes\wtB_{j,-n}\longmapsto

3397: (-1)^{2j}g_*v\otimes g_*u\in \B_{j,n}\otimes\wtB_{j,-n}.

3398: \label{bPB}

3399: \eeqa

3400: Here $g_*u\in \B_{j,-n}$ for $u\in \B_{j,n}$ is defined by considering

3401: $u$ as an element of

3402: $\V_j$: if $u$ is a charge $n$ highest weight state with respect to

3403: $\whfh$, then $g_*u$ is a charge $-n$ highest weight state

3404: which can be regarded as an element of $\B_{j,-n}$.

3405: 

3406: 

3407: 

3408: One can check that (\ref{bPA}) and (\ref{bPB})

3409: are consistent with the field identification.

3410: Let us start with $P_A$. The problem is trivial if

3411: $(j,n)$ is not equivalent to $(j,-n)$ since one can choose the phase of the

3412: states so that

3413: $P_A$ is compatible with the field identification.

3414: The cases with

3415: $(j,n)\equiv (j,-n)$ consist of

3416: $(j,n)=(j,0)$ and $(j,n)=({k\over 4},\pm{k\over 2})$.

3417: The case $(j,0)$ is trivial for an obvious reason.

3418: Non-trivial is the latter case.

3419: Let $u_{\pm}=|{k\over 4},\mp{k\over 4}\rangle\in V_{k\over 4}\subset

3420: \V_{k\over 4}$ be the vector representing the primary state of

3421: $\B_{{k\over 4},\pm {k\over 2}}$.

3422: The field identification identifies the states

3423: $u_+\otimes u_-

3424: \in\B_{{k\over 4},{k\over 2}}\otimes \B_{{k\over 4},-{k\over 2}}$

3425: and

3426: $u_-\otimes u_+

3427: \in \B_{{k\over 4},-{k\over 2}}\otimes \B_{{k\over 4},{k\over 2}}$,

3428: up to a constant;

3429: \beq

3430: u_+\otimes u_-=\epsilon u_-\otimes u_+.

3431: \label{FIeg}

3432: \eeq

3433: Then, $P_A$ maps these states as

3434: $$

3435: \begin{array}{l}

3436: u_+\otimes u_-\mapsto (-1)^{k\over 2} u_-\otimes u_+

3437: =(-1)^{k\over 2}\epsilon^{-1}u_+\otimes u_-,\\

3438: u_-\otimes u_+\mapsto (-1)^{k\over 2} u_+\otimes u_-

3439: =(-1)^{k\over 2}\epsilon u_-\otimes u_+.

3440: \end{array}

3441: $$

3442: These are indeed the same map, provided

3443: $\epsilon^{-1}=\epsilon$, or $\epsilon=\pm 1$.

3444: Let us next consider the action of $P_B$

3445: on the ground state $|0,0\rangle\otimes |0,0\rangle$

3446: in $\B_{0,0}\otimes \wtB_{0,0}$

3447: which is identified with the state

3448: $|{k\over 2},{k\over 2}\rangle\otimes |{k\over 2},-{k\over 2}\rangle$

3449: in $\B_{{k\over 2},-k}\otimes \wtB_{{k\over 2},k}$,

3450: up to some phase,

3451: say $c$.

3452: Now, $g_*$ sends $|j,m\rangle\to i^{2j}|j,-m\rangle$.

3453: Thus, $P_B$ maps these states as

3454: \beqa

3455: |0,0\rangle\otimes |0,0\rangle&\mapsto&

3456: |0,0\rangle\otimes |0,0\rangle,

3457: \nn\\

3458: \mbox{$c|{k\over 2},{k\over 2}\rangle\otimes

3459: |{k\over 2},-{k\over 2}\rangle$}

3460: &\mapsto&\mbox{$c(-1)^{k}i^{k}|{k\over 2},{k\over 2}\rangle\otimes

3461: i^{k}|{k\over 2},-{k\over 2}\rangle$}.\nn

3462: \eeqa

3463: Since $(-1)^{k}i^{k}i^{k}=1$, it is indeed compatible with

3464: the field identification.

3465: 

3466: 

3467: 

3468: 

3469: Using (\ref{bPA})-(\ref{bPB}),

3470: one can compute  the twisted partition function.

3471: For $P_A$, only  representations with $(j,-n)\equiv

3472: (j,n)$ contribute. As we have seen above, these are

3473: $(j,0)$ with even $j$ and $({k\over 4},{k\over 2})$

3474: (the latter is possible only when $k$ is even).

3475: It is then easy to see that

3476: \beq

3477: \Tr(P_A\e^{2\pi i \tau H})=

3478: \sum_{j:{\rm integer}}

3479: \chi^{}_{j,0}(2\tau)+\delta_k^{(2)}\epsilon(-1)^{k\over 2}

3480: \chi_{{k\over 4},{k\over 2}}(2\tau),

3481: \eeq

3482: where $\epsilon$ is the constant that appears in the field identification

3483: (\ref{FIeg}), which we learned to be a sign $\pm 1$.

3484: For $P_B$, all the representations contribute.

3485: The trace on the subspace $\Scr{H}_{j,n}\otimes\Scr{H}_{j,-n}$ is

3486: \beqa

3487: \Tr_{j,n}(P_B\e^{2\pi i\tau H})

3488: &=&\sum_{N,M}

3489: \langle N|\otimes \langle M|q^H(-1)^{2j}g_*|M\rangle

3490: \otimes g_*|N\rangle

3491: \nn\\

3492: &=&\sum_{N,M}q^{E_N}(-1)^{2j}\langle N|g_*|M\rangle

3493: \langle M|g_*|N\rangle

3494: =\Tr\!\!\mathop{}_{\Scr{H}_{j,n}}q^{2(L_0-{c\over 24})}(-1)^{2j}g_*^2,

3495: \nn

3496: \eeqa

3497: where $\{|N\rangle\}$ and $\{|M\rangle\}$ are the basis vectors of 

3498: $\Scr{H}_{j,n}$ and $\Scr{H}_{j,-n}$.

3499: We note here that $g_*^2$ is equal to $-{\bf 1}_2$ and thus acts on

3500: the spin $j$ representation as $(-1)^{2j}$.

3501: Thus, this contribution is just $\chi_{j,n}(2\tau)$

3502: and the total trace is the sum over $(j,n)\in {\rm PF}_k$.

3503: 

3504: 

3505: One could also consider parities combined with the axial rotation symmetry

3506: $a^{\ell}$.

3507: Such parities map the state

3508: $u\otimes v\in \Scr{H}_{j,n}\otimes \Scr{H}_{j,-n}$

3509: as $a^{\ell}P_A:

3510: u\otimes v\mapsto (-1)^{2j}\e^{-{2\pi i \ell n\over k}}v\otimes u$

3511: and $a^{\ell}P_B:u\otimes v\mapsto (-1)^{2j}\e^{2\pi i\ell n\over k}

3512: g_*v\otimes g_*u$.

3513: Note that all $a^{\ell}P_A$ are involutive but

3514: $a^{\ell}P_B$ are not;

3515: \beq

3516: (a^{\ell}P_A)^2=1,\qquad

3517: (a^{\ell}P_B)^2=a^{2\ell}.

3518: \eeq

3519: For $a^{\ell}P_B$, only the one with $\ell=0$ and $\ell={k\over 2}$

3520: are involutive (the latter applies only when $k$ is even).

3521: This can also be understood from the geometrical point of view,

3522: $a^{\ell}{\mathcal I}_A:z\to \e^{2\pi i\ell\over k}\bz$

3523: and $a^{\ell}{\mathcal I}_B:z\to \e^{2\pi i\ell \over k}z$. 

3524: The twisted partition functions are

3525: \beqa

3526: &&

3527: \Tr(a^{\ell}P_A\e^{2\pi i \tau H})=

3528: \sum_{j:{\rm integer}}

3529: \chi^{}_{j,0}(2\tau)+\delta_k^{(2)}\epsilon(-1)^{k\over 2}

3530: (-1)^{\ell}\chi_{{k\over 4},{k\over 2}}(2\tau),

3531: \label{TrPAe}\\

3532: &&

3533: \Tr(a^{\ell}P_B\e^{2\pi i\tau H})

3534: =\sum_{(j,n)\in {\rm PF}_k}\e^{2\pi i\ell n\over k}\chi^{}_{j,n}(2\tau).

3535: \label{TrPBe}

3536: \eeqa

3537: We recall that $\epsilon$ is some sign which has not been determined yet.

3538: 

3539: 

3540: 

3541: 

3542: %Let us next examine the involution $g\to g_0g^{-1}g_0$.

3543: %For this to keep $g^{-1}D_{\mu}g$ invariant up to sign or

3544: %similarity transformations, $A_{\mu}$ needs to be transformed to

3545: %$g_0A_{\mu}g_0^{-1}$ and also to $g_0^{-1}A_{\mu}g_0$.

3546: %This is possible only if

3547: %the subgroup $H=U(1)$ is

3548: %invariant under ${\rm ad}(g_0):h\ti g_0hg_0^{-1}$ and that

3549: %${\rm ad}(g_0^2)=1$ on $H$. This requires $g_0=h_0$ or

3550: %$i\sigma_1h_1$ for some elements $h_0,h_1$ of $H$.

3551: %In fact, $h_0$ does the axial rotation while

3552: % $h_1$ can be absorbed by a gauge transformation.

3553: %Thus, we have found two classes of involutions

3554: %\beqa

3555: %{\cal I}_A:g\to g^{-1},\\

3556: %{\cal I}_B:g\to g_*^{-1}

3557: 

3558: 

3559: \section{Parafermions: RCFT versus geometry}\label{sec:paraf}

3560: 

3561: In this section, we describe the crosscap states

3562: of the $SU(2)/U(1)$ coset model

3563: following the general procedure given in Section~\ref{general}.

3564: Comparison with some of the results in Section~\ref{sec:gWZW}

3565: will provide the geometric interpretation of the PSS and other

3566: parity symmetries. This is also confirmed by using localized wave packets.

3567: 

3568: 

3569: 

3570: As a warm-up and for later use, we briefly review orientifolds of

3571: $SU(2)_k$ \cite{PSS,PSSII,HOcs,BiSt} and their geometrical interpretation. 

3572: 

3573: 

3574: \subsection{Orientifolds of $SU(2)$: Summary of the RCFT}

3575: 

3576: 

3577: Following Section \ref{general}, we collect 

3578: the basic RCFT data. The $S$-matrix of the

3579: $SU(2)$ theory is given by the well known expression

3580: \beq

3581: S_{jj'} = \sqrt{\frac{2}{k+2}} \ \sin \pi \frac{(2j+1)(2j'+1)}{k+2}.

3582: \eeq

3583: From this and $\Delta_j=\frac{j(j+1)}{k+2}$ one computes the

3584: $P$-matrix

3585: \beq\label{su2p}

3586: P_{jj'} = \frac{2}{\sqrt{k+2}} \ \sin \pi \frac{(2j+1)(2j'+1)}{2(k+2)}.   \

3587: \delta^{(2)}_{2j+2j'+k}

3588: \eeq

3589: With this, $k+1$ Cardy boundary states can be constructed,

3590: labelled by the integral highest weight representations. 

3591: It is by now well known \cite{AlSc,ASR,FFFS} 

3592: that the Cardy boundary state $\ket{\Scr{B}_J}$

3593: corresponds geometrically to a brane wrapping the conjugacy class $C_J$ 

3594: of $SU(2)$

3595: containing the group element $e^{2\pi i J\sigma_3/k}$. 

3596: 

3597: The simple current

3598: group of the model is $\Z_2$, and is generated by the sector labelled $k/2$.

3599: Fusing $k/2$ with itself yields the identity representation. Hence, one

3600: can construct two different crosscap states, the standard PSS state

3601: $\ket{\Scr{C}}$ and a simple current induced state

3602: $|\Scr{C}_\frac{k}{2}\rangle$.

3603: The geometrical interpretation

3604: of those crosscap states has been given in \cite{HSSc,BCW,ib}: the standard

3605: PSS crosscap corresponds to the involution ${\cal I}: g \to g^{-1}$,

3606: whereas the simple current induced crosscap corresponds to 

3607: ${\cal I}^- : g\to - g^{-1}$. 

3608: 

3609: In terms of the RCFT data, the Klein bottle  amplitudes are

3610: obtained as

3611: \beqa

3612: \Tr^{}_{\cal H}(P\e^{2\pi i\tau H})

3613: &=&  \sum_{j\in {\rm P}_k} Y_{j0}^0 \ \chi_j(2\tau) =

3614: \sum_{j\in {\rm P}_k}(-1)^{2j}\chi^{}_{j}(2\tau) \\ \no

3615: \Tr^{}_{\cal H}(P^-\e^{2\pi i\tau H})

3616: &=& \sum_{j\in {\rm P}_k} Y_{j\frac{k}{2}}^{\frac{k}{2}}

3617: \chi^{}_{j}(2\tau) 

3618: =\sum_{j\in {\rm P}_k}\chi^{}_{j}(2\tau)

3619: \eeqa

3620: where we have used the fact that $Y_{j0}^0 = (-1)^{2j}$

3621: and that $Y_{j\frac{k}{2}}^{\frac{k}{2}}=1$ for all $j$.

3622: 

3623: To construct and interpret the M\"obius strips, one needs to know that

3624: $Y_{J 0}^j= (-1)^j (-1)^{2J} N_{JJ}^j $ 

3625: and $Y_{J \frac{k}{2}}^j = N_{J \frac{k}{2}-J}^j$.

3626: With the help of these identities, one obtains

3627: \beqa

3628: \langle \Scr{C} |\e^{-{\pi i\over 4\tau}H}\ket{\Scr{B}_{J}} 

3629: &=& \sum_j Y_{J0}^{j} \ \widehat\chi_j(\tau)

3630: = \sum_j N_{JJ}^j \ (-1)^{2J} \ (-1)^j \ \widehat\chi_j(\tau) \\ \no

3631: \langle \Scr{C}_{k\over 2} | \e^{-{\pi i\over 4\tau}H}\ket{\Scr{B}_{J}} 

3632: &=& \sum_j Y_{J\frac{k}{2}}^{j} \ \widehat\chi_j(\tau)

3633: = \sum_j N_{J \frac{k}{2}-J}^j \ \widehat\chi_j (\tau)

3634: \eeqa

3635: To make the connection to geometry for the first line, recall that

3636: ${\cal I}: g\to g^{-1}$ acts as the anti-podal map on the $S^2$ wrapped

3637: by the brane. In the classical limit, the primary fields living on the

3638: brane become functions on $S^2$. More precisely, the algebra of functions

3639: on $S^2$ is spanned by the spherical harmonics $Y_{j,m}$, $j \in \Z$. Under

3640: reflection they transform as $Y_{j,m} \to (-1)^j Y_{j,m}$. This is

3641: exactly the action of $P$ that one reads off from the M\"obius amplitude.

3642: 

3643: For the second line, observe that the spectrum of open string states

3644: in the M\"obius amplitude is exactly that of a brane $J$ and its

3645: image $\frac{k}{2} - J $ under ${\cal I}^-$. 

3646: 

3647: In this way, the CFT data encodes the geometry of the orientifold.

3648: 

3649: 

3650: \subsection{Parafermions}

3651: 

3652: 

3653: We first review some basic facts on the $SU(2)/U(1)$ coset model, as

3654: a rational conformal field theory.

3655: The Hilbert space

3656: $$

3657: {\mathcal H}=\bigoplus_{(j,n)\in PF_k}\Scr{H}_{j,n}\otimes\Scr{H}_{j,-n},

3658: $$

3659: decomposes into  irreducible representations of

3660: the parafermion algebra.

3661: The conformal weight of the primary field with label

3662: $(j,n)$

3663: is given by

3664: \beq\label{paraweight}

3665: h_{j,m} = \frac{j(j+1)}{(k+2)} - \frac{n^2}{4k}

3666: \eeq

3667: if $(j,n)$ is in the range

3668: $j=0,\dots, k/2$ and $-2j \leq n \leq 2j$.

3669: We shall call the latter

3670: {\it the standard range} (abbreviated by S.R.).

3671: Any label $(j,n)$ can be reflected to the standard range by

3672: field identification $(j,n)\to ({k\over 2}-j,n+k)$.

3673: 

3674: 

3675: 

3676: 

3677: \subsubsection*{\it Modular matrices}

3678: 

3679: The $S$ and $T$-matrices of the coset model have the factorized form

3680: \beq

3681: S_{(j,n)(j',n')} = 2 \ S_{jj'} \ S^*_{nn'},~~~

3682: T_{(j,n),(j',n')}=T_{jj'}T^*_{nn'},

3683: \label{factorPF}

3684: \eeq

3685: where it is understood that the matrices with

3686: pure $j$ labels are those of the $SU(2)_k$

3687: WZW model, and matrices with pure $n$ labels are

3688: those of $U(1)_k$.

3689: Using this factorization property, we find

3690: \beq

3691: N_{(j,n)(j',n')}^{\,\,\,(j'',n'')}

3692: =N_{jj'}^{\,\,j''}\delta_{n+n',n''}^{(2k)}

3693: +N_{jj'}^{\,\,{k\over 2}-j''}\delta_{n+n',n''+k}^{(2k)}.

3694: \label{paraN}

3695: \eeq

3696: We also need to determine $P$ and $Y$.

3697: As a first step, it is useful to consider instead the quantities

3698: $Q=ST^2S$ and $\tilde{Y}$ defined by (\ref{defSTS}) and (\ref{defYtilde}).

3699: The computation is easy since $\sqrt{T}$ is not involved

3700: and one can use the factorization (\ref{factorPF}).

3701: The result is

3702: \beqa

3703: &&

3704: Q_{(j,n)(j',n')} = Q_{jj'} Q_{nn'}^* + Q_{\frac{k}{2}-j, j'} Q_{n+k,n'}^*,

3705: \\

3706: &&

3707: \tilde{Y}_{(j,n)(j',n')}^{(j'',n'')} = \tilde{Y}_{jj'}^j \

3708: \overline{\tilde{Y}}_{nn'}^{n''} + \tilde{Y}_{jj'}^{\frac{k}{2}-j} \

3709: \overline{\tilde{Y}}_{nn'}^{n''+k}.

3710: \eeqa

3711: From this, one can compute $P=\sqrt{T}Q\sqrt{T}$ and

3712: $Y_{ab}^{c}=\sqrt{T_b/T_c}\tilde{Y}_{ab}^c$.

3713: 

3714: \subsubsection*{\it Discrete Symmetries}

3715: 

3716: The group of simple currents of the model is $\Z_k$

3717: generated by $(0,2)$.

3718: The monodromy charge of the field in the representation $(j,n)$

3719: under the simple current $(0,2\ell)$ is

3720: \beq

3721: Q_{(0,2\ell)}(j,n) = \frac{\ell n}{k}.

3722: \eeq

3723: Accordingly, there is a symmetry group $\Z_k$

3724: acting on the states as

3725: \beq

3726: a^{\ell}: \psi_{(j,n)} \to

3727: \e^{\frac{2\pi i \ell n}{k}} \ \psi_{(j,n)}

3728: \quad {\rm for} \quad \psi_{(j,n)} \in \Scr{H}_{j,n} \otimes \Scr{H}_{j,-n},

3729: \eeq

3730: where we denote the generator of the group by $a$.

3731: This is in fact equivalent to the $\Z_k$ axial rotation symmetry

3732: of the gauged WZW model (\ref{Defa}).

3733: 

3734: \subsubsection*{\it Mirror symmetry}

3735: 

3736: The orbifold by the full symmetry group $\Z_k$

3737: has the Hilbert space of states

3738: \beq

3739: {\cal H}^M = \bigoplus_{(j,n)\in PF_k}

3740: \Scr{H}_{j,n} \otimes \Scr{H}_{j,n},

3741: \eeq

3742: and can be regarded as the mirror of the original model.

3743: The mirror map $\Psi: {\cal H} \to {\cal H}^M$ 

3744: acts on states as

3745: $\Psi=V_M \otimes 1: 

3746: \ket{j,n} \otimes \ket{j,-n} \to  \ket{j,-n} \otimes \ket{j,-n}$.

3747: 

3748: \subsection{A-type parities: RCFT and geometry}

3749: 

3750: According to Section~\ref{general},

3751: there are Cardy branes (A-branes) $\Scr{B}_{j,n}$

3752: labelled by the representation and PSS parities (A-parities) $P_{\ell}$

3753: labelled by the simple currents.

3754: The branes are transformed under the parities as (\ref{Pgi}),

3755: which reads

3756: \beq

3757: P_{\ell}:\Scr{B}_{j,n}\to\Scr{B}_{j,2\ell-n}.

3758: \label{paraBT}

3759: \eeq

3760: The crosscap state for the parity $P_{\ell}$ is

3761: \beq

3762: \ket{\Scr{C}_{\ell}} = \sum_{(j,n) \in PF_k} 

3763: \frac{P_{(0,2\ell)(j,n)}}{\sqrt{S_{(0,0)(j,n)}}} \ \cket{j,n}.

3764: \eeq

3765: Explicit expressions for the A-type crosscap states can be found

3766: in the appendix.

3767: Of particular interest

3768: is the coefficient of the identity $(0,0)$, since in a full string

3769: model this coefficient would give rise to a contribution to the total tension

3770: of the orientifold plane. It is given by

3771: \beq

3772: T_{O^A_{\ell}}=\left\{

3773: \begin{array}{ll}

3774: \frac{1}{[(k+2)k]^{\frac{1}{4}}} \ 

3775: \cot^{\half}\!\left[\frac{\pi}{2(k+2)}\right]

3776: & k \ {\rm odd} \\

3777: \frac{1}{[(k+2)k]^{\frac{1}{4}}} \ 

3778: \left( \cot^{\frac{1}{2}}\!\left[\frac{\pi}{2(k+2)}\right]

3779: + (-1)^{\ell}

3780: \tan^{\frac{1}{2}}\!\left[\frac{\pi}{2(k+2)}\right] \right)

3781: &k \ {\rm even}

3782: \end{array}\right.

3783: \eeq

3784: For $k$ odd, it is manifestly independent of $\ell$

3785: --- all the PSS orientifolds have the same tension.

3786: For $k$ even, there is an additional term that depends on $\ell$ mod $2$.

3787: The latter is consistent with the action of the

3788: $\Z_k$ generator $a$  on the crosscap states

3789: $$

3790: a: \ket{\Scr{C}_{\ell}} \to \ket{\Scr{C}_{\ell+2}};$$

3791: which implies that orientifolds related by symmetry operations 

3792: have the same mass, as required.

3793: In the geometric

3794: limit of infinite $k$, the $\ell$ dependence drops out and  we get 

3795: equal masses also for orientifolds that are not related by symmetry.

3796: 

3797: The Cardy formula yields the boundary states

3798: \beq

3799: \ket{\Scr{B}_{j,n} } = \sum_{(j',n')\in PF_k} 

3800: \frac{S_{(j,n)(j',n')}}{\sqrt{S_{(0,0)(j',n')}}} \ \kket{(j',n')}.

3801: \eeq

3802: whose tension is

3803: \beq

3804: T_{\Scr{B}_{j,n}} = \frac{\sqrt{2}}{[k(k+2)]^{\frac{1}{4}}} \, 

3805: \frac{\sin \frac{\pi(2j+1)}{k+2}}{\sqrt{\sin \pi \frac{1}{k+2}}},

3806: \eeq

3807: which is $n$-independent.

3808: 

3809: 

3810: \subsubsection{The one-loop amplitudes}

3811: 

3812: We next compute the cylinder, MS and KB amplitudes.

3813: Details are recorded in Appendix~\ref{app:detai}.

3814: The cylinder amplitude are

3815: \beq

3816: \langle \Scr{B}_{j_1,n_1} | \e^{-{\pi i\over \tau}H}\ket{\Scr{B}_{j_2,n_2}}

3817: = \sum_{2j+n \,{\rm even}}

3818: N_{j_1j_2}^{j} \delta_{n_2-n_1+n}^{(2k)}\chi_{j,n}(\tau).

3819: \label{AparaBB}

3820: \eeq

3821: The M\"obius strip with boundary condition $\Scr{B}_{j_2,n_2}$ is

3822: \beq

3823: \langle\Scr{C}_{\ell}|\e^{-{\pi i\over 4\tau}H} |\Scr{B}_{j_2,n_2}\rangle 

3824: =\sum_{2j+n\,{\rm even}}

3825: N_{j_2j_2}^j\delta_{2n_2-2\ell+n}^{(2k)}

3826: \epsilon_{j,n}\widehat{\chi}_{j,n}(\tau),

3827: \label{AparaBC}

3828: \eeq

3829: where

3830: $\epsilon_{j,n}$ is a sign factor which is

3831: $(-1)^{2j+n\over 2}, 1, (-1)^n$ if

3832: $(j,n),({k\over 2}-j,n+k),({k\over 2}-j,n-k)$ is

3833: in the standard range, respectively.

3834: Comparison of (\ref{AparaBC}) with (\ref{AparaBB}) implies

3835: that the image brane of $\Scr{B}_{j_2,n_2}$

3836: is $\Scr{B}_{j_2,2\ell-n_2}$, which is indeed correct

3837:  (\ref{paraBT}).

3838: The Klein bottle amplitudes are

3839: \beq

3840: \langle \Scr{C}_{\ell_1}|\e^{-{\pi i\over 2\tau}H}

3841: |\Scr{C}_{\ell_2}\rangle

3842: =\sum_{2j+\ell_1-\ell_2\,{\rm even}}

3843: \chi_{j,\ell_1-\ell_2}(2\tau)

3844: +\delta_{k}^{(2)}\delta_{\ell_1,\ell_2}^{(2)}

3845: (-1)^{\ell_1}

3846: \chi_{{k\over 4},{k\over 2}+\ell_1-\ell_2}(2\tau).

3847: \label{AparaCC}

3848: \eeq

3849: In particular,

3850: for $\ell_1=\ell_2$ we have

3851: \beq

3852: \langle \Scr{C}_{\ell}|\e^{-{\pi i\over 2\tau}H}

3853: |\Scr{C}_{\ell}\rangle

3854: =\sum_{j\,{\rm integer}}

3855: \chi_{j,0}(2\tau)

3856: +\delta_{k}^{(2)}(-1)^{\ell}

3857: \chi_{{k\over 4},{k\over 2}}(2\tau).

3858: \label{AparaKB}

3859: \eeq

3860: For $k$ odd, this is independent of 

3861: $\ell$, whereas for $k$ even there is an $\ell$ dependent term, which

3862: plays only a role at finite $k$.

3863: This behavior was observed before, when we computed the tensions

3864: of the orientifolds.

3865: 

3866: 

3867: \subsubsection{Geometrical interpretation}

3868: 

3869: As noted in Section~\ref{sec:gWZW},

3870: the model has a $\sigma$-model interpretation,

3871: where the target space is a disk $|z|\leq 1$.

3872: A geometrical interpretation of the Cardy boundary states in that

3873: geometry was provided by \cite{MMS}. 

3874: It was found  that the Cardy states with $j=0$

3875: correspond to D0-branes distributed at the $k$ symmetric points at

3876: the boundary of the disk:

3877: $\Scr{B}_{0,n}$ ($n$ even)

3878: is the D0-brane at the boundary point

3879: $z=\e^{\pi i n\over k}$.

3880: The $\Z_k$ symmetry rotations act on the boundary states by shifting

3881: $n$.

3882: The branes

3883: with higher $j$ are D1 branes

3884: stretched between two special points separated by

3885: the angle $4\pi j/k$:

3886: $\Scr{B}_{j,n}$ ($2j+n$ even)

3887: is the D1-brane along the straight line

3888: connecting the points $z=\e^{{\pi i\over k}(n+2j)}$

3889: and $z=\e^{{\pi i\over k}(n-2j)}$.

3890: Branes of a given $j$ are related by $\Z_k$ symmetry, just as in

3891: the case $j=0$.

3892: 

3893: 

3894: The parity $P_A$ found in the gauged WZW model analysis

3895: acts on the disk as $z\to\bz$, a reflection with respect to the

3896: diameter ${\rm Im}(z)=0$.

3897: Since it maps the special points as $z=\e^{{\pi i\over k}(n\pm 2j)}\mapsto

3898: \e^{-{\pi i\over k}(n\pm 2j)}$, it transform the Cardy branes as

3899: $P_A:\Scr{B}_{j,n}\to\Scr{B}_{j,-n}$.

3900: Also, the combination with the axial rotation symmetry $a^{\ell}$

3901: would act

3902: on the branes as $a^{\ell}P_A:\Scr{B}_{j,n}\to\Scr{B}_{j,2\ell-n}$.

3903: Comparing with the rule (\ref{paraBT}), we find the relation of gauged

3904: WZW parities to the PSS parities:

3905: \beq

3906: P_{\ell}=a^{\ell}P_A.

3907: \eeq

3908: Indeed, under this identification,

3909: the Klein bottle amplitude (\ref{AparaKB}) is consistent with

3910: the partition function (\ref{TrPAe}) in the gauged WZW model.

3911: The sign $\epsilon$ of the field identification

3912: is now determined to be $\epsilon=(-1)^{k\over 2}$.

3913: It would be interesting to examine this using the functional integral method.

3914: 

3915: 

3916: \subsubsection{Sketch of a shape computation}

3917: 

3918: We now give an independent argument to determine the location of

3919: the orientifold planes.

3920: In \cite{MMS} the location and geometry of D-branes was tested by

3921: scattering graviton wave packets from the

3922: D-branes

3923: and taking the classical limit. These computations have

3924: been repeated  for orientifolds in \cite{HSS,hikida,ib}. Suitable

3925: graviton wave packets are localized $\delta$-functions, which are

3926: written down for parafermions in \cite{MMS} appendix D. Since closed string

3927: states with $j\sim k$  are not well-localized, one only uses  states

3928: with $j\ll k$ as part of the test wave function. This means that the

3929: shape of the brane is encoded in the couplings of the brane

3930: to bulk fields with $j \ll k$. From the above argument, it is expected 

3931: that the orientifold planes in the parafermion theory are located

3932: along diameters of the disk. On the other hand, it is known that

3933: the D-branes  $\Scr{B}_{\frac{k}{4},n}$ are also located along diameters. To

3934: compute the shape it is therefore not required to repeat the computation

3935: of \cite{MMS}, but to merely compare the coefficients of the crosscap

3936: state with the coefficients of the boundary states

3937: $|\Scr{B}_{\frac{k}{4},n}\rangle$.

3938: If the asymptotic behavior of the boundary and crosscap

3939: coefficients is the same, it can  be concluded that their

3940: locations coincide.

3941: The D-brane  couplings of the Cardy state 

3942: $|\Scr{B}_{\frac{k}{4},n}\rangle$ to the ground states $\ket{j,m}$ 

3943: are given by

3944: \beq

3945: B_{(\frac{k}{4}, n)(j,m)} = \frac{\sqrt{2}}{[k(k+2)]^{\frac{1}{4}}} \

3946: \frac{(-1)^{2j} \delta^{(2)}_{2j}}{\sqrt{\sin \pi\frac{2j+1}{k+2}}} \,

3947: e^{\frac{\pi i n m}{k}}

3948: \eeq

3949: and the couplings of the PSS-crosscap state (for $k$ even) are

3950: \beq

3951: \Gamma_{j,m} = \frac{\sqrt{2}}{[k(k+2)]^{\frac{1}{4}}} \ \delta^{(2)}_{2j} \, 

3952: \frac{ \sin \pi \frac{2j+1}{2(k+2)} \, + (-1)^{\frac{m+2j}{2}}

3953: \cos \pi \frac{2j+1}{2(k+2)}}{\sqrt{\sin \pi\frac{2j+1}{k+2}}}.

3954: \eeq

3955: In the large $k$ limit, the contribution of the second term is

3956: dominant. In that limit, the orientifold-couplings behave 

3957: exactly like those of the

3958: boundary state $\ket{\Scr{B}_{k/4, k/2}}$ and the conclusion is that

3959: the PSS crosscap is located along

3960: the same diameter as that brane. 

3961: This matches with the earlier

3962: conclusion

3963: based on the brane transformation rule (\ref{paraBT}).

3964: The other

3965: crosscap states differ only in the

3966: $m$-dependence and hence correspond to rotated diameters.

3967: 

3968: 

3969: \subsection{B-type Parities: RCFT and geometry}

3970: 

3971: As before, we construct B-type crosscap states by constructing

3972: A-type crosscaps in the mirror $\Z_k$ orbifold followed by an

3973: application of the mirror map.

3974: 

3975: To construct

3976: the crosscap states in the orbifold theory, we first need the

3977: bilinear form $q$ for the group $G=\Z_k$.

3978: It is uniquely fixed by the requirement $q(g,g)=-h_g$

3979: and is given by

3980: \beq

3981: q((0,n),(0,m)) := \frac{nm}{4k},\qquad\mbox{$n$, $m$ even}.

3982: \eeq

3983: Note that it is well-defined (invariant under

3984: $2k$ shifts of $n$ and $m$)

3985: and obeys $2q(g_1,g_2) = Q_{g_1}(g_2)$. 

3986: We also need $\hat{Q}_{g}(h)$ defined modulo 2.

3987: For this, we note that

3988: \beq

3989: h_{(0,n)} = -\frac{n^2}{4k} +\frac{|n|}{2},\qquad\mbox{for $-k\leq n\leq k$}.

3990: \eeq

3991: Then, we have (for $n$, $m$ even)

3992: \beqa

3993: \hat{Q}_{(0,n)}((0,m)) 

3994: &=& -\frac{n^2}{4k} -\frac{m^2}{4k} + \frac{(n\hat{+}m)^2}{4k}

3995: +\frac{|n|}{2} +\frac{|m|}{2} - 

3996: \frac{|n\hat{+}m|}{2}

3997: \nn\\

3998: &=& \frac{n\hat{+}m-|n\hat{+}m|}{2}

3999: -\frac{n-|n|}{2} -\frac{m-|m|}{2} +\frac{nm}{2k}

4000: =\frac{nm}{2k} \quad {\rm mod} \ 2.

4001: \nn

4002: \eeqa

4003: In the second step we used (\ref{Qcomp}).

4004: In the last step we have used that ${n-|n|\over 2}$ is

4005: an even integer if $n$ is even.

4006: Thus, we arrive at the conclusion that $\hat{Q}=2q$ mod $2$. 

4007: Therefore, Eq.~(\ref{thetaEq}) is homogeneous,

4008: $\theta(gh)=\theta(g)+\theta(h)$, and the general solution is given by

4009: \beq

4010: \theta_r(\ell)=-2\frac{r\ell}{k},\qquad r\in \Z/k\Z.

4011: \eeq

4012: 

4013: 

4014: Following the procedure given in

4015: Section~\ref{general},

4016: we find B-parities $P^r_B$ parametrized by a mod-$k$ integer $r$,

4017: whose crosscaps are given by

4018: \beq

4019: |\Scr{C}^B_r\rangle

4020: ={1\over\sqrt{k}}\sum_{\ell}\e^{2\pi i r\ell\over k}(V_M\otimes 1)

4021: |\Scr{C}_{\ell}\rangle,

4022: \label{CBr}

4023: \eeq

4024: where $V_M:\Scr{H}_{j,n}\to \Scr{H}_{j,-n}$ is the map induced from the

4025: mirror automorphism.

4026: More explicitly, they are expressed as

4027: \beqa

4028: \ket{\Scr{C}_r^B}&=&

4029: \frac{k^{\frac{1}{4}}}{(k+2)^{\frac{1}{4}}}\Biggl[

4030: \sum_{j,(j,-2r)\in S.R.} 

4031: (-1)^{j+r}

4032: \sqrt{\cot\mbox{$\frac{\pi(2j+1)}{2(k+2)}$}}

4033: \ \cket{(j,2r)}_B \nn\\

4034: && ~~+ \sum_{j,(j,-2r)\notin S.R.} 

4035: \sqrt{\cot\mbox{$\frac{\pi(2j+1)}{2(k+2)}$}}

4036: \cket{(j,2r)}_B \Biggr].

4037: \eeqa

4038: The details of the computation are summarized in Appendix~\ref{subapp:Bcc}.

4039: The square of the B-type involutions $P^r_B$ can be computed to be

4040: \beq

4041: (P^r_B)^2 = a^{2r}.

4042: \eeq

4043: This is consistent with the crosscap

4044: $|\Scr{C}_r^B\rangle$ being a $a^{2r}$-twisted state

4045: (it belongs to

4046: $\oplus_j\Scr{H}_{j,2r}\otimes\Scr{H}_{j,2r}$, which is a subspace of

4047: $\oplus_{j,n}\Scr{H}_{j,n}\otimes\Scr{H}_{j,4r-n}={\mathcal H}_{a^{2r}}$).

4048: As before, the tensions of the orientifold planes can  be determined as

4049: overlaps with the ground state. They are only non-vanishing for the

4050: involutive crosscaps $P_B^0$ and $P_B^{\frac{k}{2}}$ (the latter exists

4051: only for $k$ even). The result is

4052: \beqa

4053: T_{\Scr{C}_0^B} &=& \frac{k^{\frac{1}{4}}}{(k+2)^{\frac{1}{4}}}

4054: \cot^{\frac{1}{2}}\!\left[\frac{\pi}{2(k+2)}\right]

4055: \label{TB0}\\

4056: T_{\Scr{C}_{\frac{k}{2}}^B} &=& \frac{k^{\frac{1}{4}}}{(k+2)^{\frac{1}{4}}}

4057: \tan^{\frac{1}{2}}\!\left[\frac{\pi}{2(k+2)}\right]

4058: \label{TBk2}  

4059: \eeqa

4060: Note the different behavior of the tension as $k\to \infty$: the

4061: tension of the orientifold plane for$P^0_B$ becomes infinite, whereas 

4062: that for $P^{\frac{k}{2}}_B$ goes to zero. 

4063: 

4064: 

4065: \subsubsection*{\it Boundary states for B-branes}

4066: 

4067: In this section we write down the B-type boundary states, which are obtained

4068: by taking the average over the $\Z_k$-orbit of the A-type boundary states,

4069: followed by the action of the mirror map $\Psi^{-1}$.

4070: Since there is only one orbit for each $j$, they are parametrized by

4071: just $j$ (with the identification

4072: $j\equiv {k\over 2}-j$):

4073: \beq

4074: \ket{\Scr{B}_j^B}

4075: ={1\over \sqrt{k}}\sum_{\ell}(V_M\otimes 1)\ket{\Scr{B}_{j,n+2\ell}}

4076: =(2k)^{1\over 4}\sum_{j'\,{\rm integer}}{S_{jj'}\over\sqrt{S_{0j'}}}

4077: \cket{(j',0)}_B,

4078: \eeq

4079: where $n$ is an arbitrary integer such that $2j+n$ is even.

4080: For $j={k\over 4}$, which is possible only when $k$ is even,

4081: the $\Z_k$ action has a fixed point,

4082: $\ell={k\over 2}:({k\over 4},n)\mapsto ({k\over 4},n+k)\equiv ({k\over 4},n)$,

4083: and special care is needed in the construction of the boundary states.

4084: In fact, it splits into two B-branes distinguished by $\eta=\pm 1$

4085: \cite{MMS}, with the boundary states

4086: \beq

4087: \ket{\Scr{B}_{\frac{k}{4},\eta}^B}= \frac{1}{2} \

4088: (2k)^{\frac{1}{4}} \sum_{j\,{\rm integer}} 

4089: \frac{S_{{k\over 4}j}}{\sqrt{S_{0j}}}

4090: \kket{(j,2r)}_B + \frac{\eta}{2} \ [k(k+2)]^{\frac{1}{4}} \ 

4091: \kket{\frac{k}{4}, \frac{k}{2}}_B

4092: \eeq

4093: Under the symmetry $a^{\ell}$, these boundary states

4094: are transformed as $|\Scr{B}_j^B\rangle\to|\Scr{B}_j^B\rangle$

4095: and $|\Scr{B}^B_{{k\over 4},\eta}\rangle\to

4096: |\Scr{B}^B_{{k\over 4},(-1)^{\ell}\eta}\rangle$.

4097: Thus, each of the ordinary B-branes $\Scr{B}_j^B$ ($j\not\equiv {k\over 4}$)

4098: is invariant under $\Z_k$, but the special B-branes

4099: $\Scr{B}^B_{{k\over 4},+}$ and $\Scr{B}^B_{{k\over 4},-}$ are exchanged

4100: under odd elements of $\Z_k$.

4101: 

4102: 

4103: One can also consider boundary states on a circle with  $\Z_k$

4104: twisted boundary

4105: conditions, which can be used, for example, when we compute

4106: the M\"obius strip amplitudes with respect to non-involutive

4107: parities.

4108: The twist adds an appropriate

4109: phase factor in the $\Z_k$-average, as explained

4110: in Section~\ref{subsubsec:Borb}.

4111: For an ordinary B-brane $\Scr{B}_j^B$, 

4112: since it is invariant under any element of $\Z_k$,

4113: one can consider the boundary state with any twist $a^r$.

4114: The symmetry $a^r$ is mapped under mirror symmetry to

4115: the quantum symmetry of the orbifold model, associated with the character

4116: $a^{\ell}\to\e^{-{2\pi i r \ell\over k}}$. Therefore, the relevant average is

4117: \beq

4118: \ket{\Scr{B}_j^B}_{a^{r}}

4119: ={\e^{\pi i rn\over k}

4120: \over \sqrt{k}}\sum_{\ell}\e^{2\pi i r\ell\over k}

4121: (V_M\otimes 1)\ket{\Scr{B}_{j,n+2\ell}}

4122: =(2k)^{1\over 4}\sum_{j'\,{\rm integer}}{S_{jj'}\over\sqrt{S_{0j'}}}

4123: \cket{(j',r)}_B.

4124: \label{BjB}

4125: \eeq

4126: The overall phase $\e^{\pi i rn\over k}$ is chosen so that $n$-dependence

4127: disappears in the final expression.

4128: The special ones $\Scr{B}_{{k\over 4},\pm}$ are invariant only under even

4129: elements $a^{2r}$. Thus we can only consider even twists:

4130: \beq

4131: \ket{\Scr{B}_{\frac{k}{4},\eta}^B}_{a^{2r}}= \frac{1}{2} \

4132: (2k)^{\frac{1}{4}} \sum_{j\,{\rm integer}} 

4133: \frac{S_{{k\over 4}j}}{\sqrt{S_{0j}}}

4134: \kket{(j,2r)}_B + \frac{\eta}{2} \ [k(k+2)]^{\frac{1}{4}} \ 

4135: \kket{\mbox{$(\frac{k}{4}, \frac{k}{2}+2r)$}}_B

4136: \eeq

4137: 

4138: 

4139: 

4140: 

4141: 

4142: 

4143: \subsubsection{The one loop amplitudes}

4144: 

4145: We present here the cylinder, MS and KB amplitudes.

4146: When the average formulae

4147: (\ref{CBr}), (\ref{BjB}), etc, are available,

4148: the computation is easily done using

4149: the results (\ref{AparaBB}), (\ref{AparaBC}) and (\ref{AparaCC})

4150: for the A-type amplitudes.

4151: In this way we obtain

4152: \beq

4153: {}_{a^{r}}\langle \Scr{B}_{j_1}^B|\e^{-{\pi i\over\tau}H}

4154: |\Scr{B}_{j_2}^B\rangle_{a^{r}}

4155: =\sum_{2j+n\,{\rm even}}

4156: N_{j_1j_2}^{\,\,j}\e^{\pi i rn\over k}\chi_{j,n}(\tau).

4157: \label{BparaBB}

4158: \eeq

4159: \beq

4160: \langle \Scr{C}_r^B|\e^{-{\pi i\over 4\tau}H}

4161: |\Scr{B}_{j'}^B\rangle_{a^{2r}}

4162: =\sum_{2j+n\,{\rm even}}

4163: N_{j'j'}^{\,\,j}\e^{\pi i rn\over k}\epsilon_{j,-n}\widehat{\chi}_{j,n}(\tau).

4164: \label{BparaBC}

4165: \eeq

4166: \beq

4167: \langle \Scr{C}_r^B|\e^{-{\pi i\over 2\tau}H}

4168: |\Scr{C}_r^B\rangle

4169: =\sum_{(j,n)\in PF_k}\e^{2\pi i rn\over k}\chi_{j,n}(2\tau).

4170: \label{BparaCC}

4171: \eeq

4172: Those involving the special B-branes may be computed independently.

4173: \beqa

4174: {}_{a^{2r}}\langle \Scr{B}^B_{\frac{k}{4},\eta} | \e^{-{\pi i\over \tau}H} 

4175: |\Scr{B}^B_{\frac{k}{4},\eta'}\rangle_{a^{2r}}

4176: &=& \frac{1}{4} \sum_{2j+n\,{\rm even}} \delta^{(2)}_{2j} \

4177: (1+\eta\eta'(-1)^{\frac{2j-n}{2}}) \ \e^{\frac{2\pi i rn}{k}} \ \chi_{j,n}

4178: \nn\\

4179: &=& \sum_{(j,n)\in PF_k\atop j\,{\rm integer}}

4180: {1\over 2}(1+\eta\eta'(-1)^{2j+n\over 2})

4181: \e^{2\pi i rn\over k}\chi_{j,n}(\tau).

4182: \label{BparaSBB}

4183: \eeqa

4184: The last step would fail if $2r$ were formally replaced by an odd integer,

4185: which is consistent with

4186: the boundary state for $\Scr{B}_{{k\over 4},\eta}$

4187: not admitting odd twists.

4188: The M\"obius strip with special boundary conditions is

4189: \beqa

4190: \langle \Scr{C}_r^B|\e^{-{\pi i\over 4\tau}H}

4191: |\Scr{B}_{\frac{k}{4},\eta}^B\rangle_{a^{2r}}

4192: &=&\frac{1}{2} \sum_{2j+n\,{\rm even}} \e^{\frac{\pi irn}{k}}

4193: \epsilon_{j,-n}\widehat{\chi}_{j,n}(\tau)

4194: \nn\\

4195: &=&\sum_{(j,n)\not\in S.R.\atop j\,{\rm integer}}

4196: {1\over 2}(1+(-1)^r(-1)^{2j+n\over 2})\e^{\pi i rn\over k}

4197: \widehat{\chi}_{j,n}(\tau).

4198: \label{BparaSBC}

4199: \eeqa

4200: $\epsilon_{j,-n}$ are evaluated in the last step.

4201: 

4202: 

4203: Let us examine how the B-parities $P_B^r$ transform the B-branes.

4204: Comparing (\ref{BparaBC}) and (\ref{BparaBB}), one realizes that the

4205: $P_B^r$-image of $\Scr{B}_j^B$ is $\Scr{B}_j^B$ itself.

4206: Comparing (\ref{BparaSBB}) and (\ref{BparaSBC}),

4207: we see that the propagating modes in the loop channel are

4208: the same if $\eta\eta'=(-1)^r$, namely if $r$ even and $\eta=\eta'$

4209: or $r$ odd and $\eta=-\eta'$.

4210: This means that $P_B^r$ with even $r$ preserves each of

4211: the two special branes,

4212: $\Scr{B}_{{k\over 4},+}^B$ and $\Scr{B}_{{k\over 4},-}^B$,

4213: while $P_B^r$ with odd $r$ exchanges them.

4214: 

4215: 

4216: 

4217: \subsubsection{Geometrical interpretation}

4218: 

4219: To find the geometrical meaning of the parity symmetries,

4220: we first look at the Klein bottle amplitudes (\ref{BparaCC}).

4221: Comparing this with the result (\ref{TrPBe}) in the gauged WZW model,

4222: we find that the B-parities $P_B^r$ are identified as

4223: \beq

4224: P_B^r=a^rP_B.

4225: \label{PBint}

4226: \eeq

4227: This means that the RCFT parity $P_B^r$ is the orientation reversal 

4228: $\Omega$ followed by the

4229: rotation $z\to \e^{2\pi i r\over k} z$ of the disk.

4230: 

4231: 

4232: Let us next see how the

4233: M\"obius amplitudes fit with this interpretation.

4234: The geometrical interpretation of B-type boundary states has already

4235: been given in \cite{MMS}: the brane $\Scr{B}_{j=0}^B$

4236: corresponds to a D0 sitting at the center of the disk,

4237: and the branes of higher $j< {k\over 4}$ are $D2$ branes on a disk whose

4238: radius depends on $j$.

4239: Each of these are invariant under the rotation $z\to\e^{2\pi i r\over k}z$,

4240: and hence also by $P_B^r$.

4241: This agrees with the conclusion from the cylinder and MS amplitudes.

4242: Moreover, the $r$-dependence of

4243: the MS amplitudes (\ref{BparaBC})

4244: resides only in the phase factor $\e^{\pi irn\over k}$,

4245: which is identified as the effect of the $a^r$-action,

4246: as can be seen from the

4247: cylinder amplitude (\ref{BparaBB}).

4248: This supports the structure of (\ref{PBint}).

4249: The special B-branes $\Scr{B}_{{k\over 4},\pm}$ are interpreted

4250: as D2-branes covering the whole disc. How the two are distinguished

4251: is not yet understood in a geometric way, but the boundary states

4252: show that they are exchanged

4253: under the unit axial rotation $a$ (or other

4254: odd rotations).

4255: The interpretation (\ref{PBint}) is therefore consistent

4256: with the conclusion from the cylinder/MS comparison

4257: that the two are exchanged under $P_B^r$ with odd $r$

4258: and preserved under $P_B^r$ with even $r$.

4259: 

4260: 

4261: 

4262: 

4263: Finally we examine the tension formulae (\ref{TB0}) and (\ref{TBk2})

4264: from the geometric point of view.

4265: The orientifold fixed point set of the parity $P_B^0$ is the whole disk. 

4266: This fits with the divergence of its tension in the geometric limit.

4267: The orientifold fixed point set of the other involutive parity

4268: $P_B^{k\over 2}$ (present only for even $k$)

4269: is just one point, the center of the disk.

4270: This is consistent with the fact that it becomes light in the geometric limit.

4271: 

4272: 

4273: \subsubsection{Sketch of a shape computation}

4274: 

4275: The localized wave packets one uses to determine the shape of

4276: branes are naturally  elements of the Hilbert space 

4277: ${\cal H}$ (\ref{parahilbert}). Hence, the scattering experiment

4278: makes sense only for crosscap states containing Ishibashi states

4279: in that space. These are the crosscap states leading to involutive

4280: parities.

4281: 

4282: From our previous considerations, we already know that the orientifold

4283: plane corresponding to $P_B^0$ extends over the whole disk. So does

4284: the D-brane $\Scr{B}_{\frac{k}{4},\eta}$, which exists for $k$ even.

4285: We can therefore independently confirm the location of $\Scr{C}_{P_B^0}$ by

4286: comparing its couplings $\Gamma_{(j,m)}$ to the closed string

4287: sector with the couplings of the brane $B_{\frac{k}{4},(j,m)}$.

4288: Explicitly

4289: \beqa

4290: B_{\frac{k}{4}(j,m)} &=& \frac{\sqrt{2}k^{\frac{1}{4}}}{(k+2)^{\frac{1}{4}}} \

4291: \delta^{(2k)}_m \ \delta^{(2)}_{2j} \

4292: \frac{(-1)^j}{\sqrt{\sin \pi \frac{2j+1}{k+2}}} \\ 

4293: \Gamma_{(j,m)} &=& \frac{\sqrt{2}k^{\frac{1}{4}}}{(k+2)^{\frac{1}{4}}} \

4294: \delta^{(2k)}_m \ \delta^{(2)}_{2j} \ 

4295: \frac{(-1)^j\, \cos \pi \frac{2j+1}{2(k+2)}}{\sqrt{\sin \pi \frac{2j+1}{k+2}}} .

4296: \eeqa

4297: In the large $k$ limit, these couplings do indeed agree, confirming that

4298: the crosscap and boundary state are located at the same place.

4299: 

4300: Similarly, we know that $P_B^{\frac{k}{2}}$ corresponds to a fixed point

4301: set consisting of the center of the disk. That is exactly the location of 

4302: the boundary state with $J=0$. The closed string couplings of the

4303: boundary and crosscap state are given by

4304: \beqa

4305: B_{0 ,(j,m)} &=& \frac{\sqrt{2}k^{\frac{1}{4}}}{(k+2)^{\frac{1}{4}}} \

4306: \delta^{(2k)}_m \ \delta^{(2)}_{2j} \

4307: \frac{\sin \pi \frac{2j+1}{k+2}}{\sqrt{\sin \pi \frac{2j+1}{k+2}}} \\ 

4308: \Gamma_{(j,m)} &=&   \frac{\sqrt{2}k^{\frac{1}{4}}}{(k+2)^{\frac{1}{4}}} \

4309: \delta^{(2k)}_m \ \delta^{(2)}_{2j} \

4310: \frac{ \sin \pi \frac{2j+1}{2(k+2)}}{\sqrt{\sin \pi \frac{2j+1}{k+2}}} .

4311: \eeqa

4312: In the large $k$ limit we can (for small $j$) replace the $\sin$ by the angle,

4313: and notice that the two couplings have the same large $k$ behaviour. This

4314: confirms that in this case the orientifold is located at the center

4315: of the disk, just as the boundary state.

4316: 

4317: 

4318: \section*{Acknowledgements}

4319: 

4320: We would like to thank B.~Acharya, A.~Adem, M.~Aganagic,

4321: C.~Angelantonj, J.~Maldacena, R.~Myers, 

4322: A.~Sagnotti, E.~Sharpe and C.~Vafa.

4323: KH is supported in part by NSF DMS 0074329 and NSF PHY 0070928.

4324: 

4325: 

4326: \newpage

4327: 

4328: \appendix{Conventions on Modular Matrices}\label{app:PSY}

4329: 

4330: 

4331: Here we collect the conventions on the modular matrices that are used

4332: throughout the paper.

4333: For a representation $\Scr{H}_i$ of the chiral algebra ${\mathcal A}$

4334: we define the character by

4335: \beq

4336: \chi_i(\tau,u)

4337: :=\Tr\!\mathop{}_{\Scr{H}_i}\left(\e^{2\pi i\tau (L_0-{c\over 24})}

4338: \e^{2\pi i J_0(u)}\right),

4339: \eeq

4340: where $\tau$ is in the upper half-plane

4341: and $J_0(u)$ is the zero mode of a spin 1 current

4342: (or sum of commuting spin 1 currents) in ${\mathcal A}$.

4343: We note that

4344: \beq

4345: \chi_i(\tau,u)=\chi_{\bi}(\tau,-u),

4346: \label{Cc}

4347: \eeq

4348: and therefore, $i$ and its conjugate $\bi$ can be distinguished

4349: by the introduction of $u$.

4350: We also note

4351: $\overline{\chi_i(\tau,u)}=\chi_i(-\overline{\tau},-\overline{u})

4352: =\chi_{\bi}(-\overline{\tau},\overline{u})$.

4353: The parameter $u$ is usually suppressed in the main text, but its presence

4354: is always borne in mind.

4355: We encountered its importance already in the discussion of

4356: D-branes \cite{zuber}.

4357: 

4358: 

4359: 

4360: For an element $({a\,b\atop c\,d})$ of $SL(2,\Z)$, the characters

4361: transform as

4362: \beq

4363: \chi_i\Bigl({b+d\tau\over a+c\tau},{u\over a+c\tau}\Bigr)

4364: =\sum_j\chi_j(\tau,u)M_{ji}(\mbox{${a\,b\atop c\,d}$}).

4365: \eeq

4366: $M({a\, b\atop c\,d})$ is $\tau$-independent at $u=0$.

4367: We define $S$, $T$, $C$ as $M({a\, b\atop c\,d})|_{u=0}$

4368: for the $SL(2,\Z)$ elements

4369: \beq

4370: {\mathcal S}=\left(\begin{array}{cc}

4371: 0&-1\\

4372: 1&0

4373: \end{array}\right),\qquad

4374: {\mathcal T}=\left(\begin{array}{cc}

4375: 1&1\\

4376: 0&1

4377: \end{array}\right),\qquad

4378: {\mathcal C}=\left(\begin{array}{cc}

4379: -1&0\\

4380: 0&-1

4381: \end{array}\right),

4382: \label{STC}

4383: \eeq

4384: respectively. 

4385: $S,T$ and $C$ are all unitary and

4386: obey the same algebraic relations as the above $SL(2,\Z)$ elements, such as

4387: $[S,C]=[T,C]=0$,

4388: $S^2=C$,

4389: $(ST)^3=C$, $STS=T^{-1}ST^{-1}$.

4390: $C$ is the charge conjugation matrix, $C_{ij}=\delta_{j,\bi}$,

4391: because of relation (\ref{Cc}).

4392: $T$ is a diagonal matrix $T_{ij}

4393: =\delta_{i,j}\e^{2\pi i(h_i-{c\over 24})}=:\delta_{i,j}T_i$.

4394: More non-trivial is the fact that

4395: $S$ is a symmetric matrix, $S_{ij}=S_{ji}$ \cite{DV}.

4396: 

4397: The modular transformation of the M\"obius strip involves

4398: \beqa

4399: \widehat{\chi}_i(\mbox{$-{1\over 4\tau},{u\over 2\tau}$})

4400: &=&\sqrt{T_i}^{-1}\chi_i(\mbox{$-{1\over 4\tau}+{1\over 2},{u\over 2\tau}$})

4401: =\chi_i(\mbox{$-{1\over 4\tau}-{1\over 2},{u\over 2\tau}$})\sqrt{T_i}

4402: \nn\\

4403: &=&

4404: \sum_k\chi_k(\mbox{${4\tau\over 1+2\tau},{2u\over 1+2\tau}$})S_{ki}\sqrt{T_i}

4405: =

4406: \sum_k\chi_k(\mbox{$-{2\over 1+2\tau},{2u\over 1+2\tau}$})T_k^2S_{ki}\sqrt{T_i}

4407: \nn\\

4408: &=&\sum_{k,l}\chi_l(\mbox{$\tau+{1\over 2},u$})

4409: S_{lk}T_k^2S_{ki}\sqrt{T_i}

4410: =

4411: \sum_l\widehat{\chi}_l(\tau,u)(\sqrt{T}ST^2S\sqrt{T})_{li}.

4412: \nn

4413: \eeqa

4414: This introduces the new matrix

4415: \beq

4416: P=\sqrt{T}ST^2S\sqrt{T}.

4417: \eeq

4418: Using the properties of $S,T$ and $C$, we find

4419: that $P$ is a unitary, symmetric matrix such that $P^2=C$.

4420: 

4421: 

4422: 

4423: 

4424: 

4425: 

4426: 

4427: 

4428: \appendix{Some properties of $Q_g(i)$}\label{app:Q}

4429: 

4430: 

4431: A simple current $g$ is a representation of the chiral algebra

4432: ${\mathcal A}$ such that the fusion product of $g$ with any representation

4433: $i$ is just one representation, which we denote by $g(i)$.

4434: The set of simple currents forms an abelian group $\Scr{G}$.

4435: Let us introduce the number (defined modulo $1$)

4436: \beq

4437: Q_g(i):=h_g+h_i-h_{g(i)}\qquad\mbox{mod 1}.

4438: \eeq

4439: They obey the following properties:

4440: 

4441: \noindent

4442: (i) If $N_{ij}^{\,\,\,k}\ne 0$, then $Q_g(i)+Q_g(j)=Q_g(k)$.\\

4443: (ii) $Q_{g_1}(i)+Q_{g_2}(i)=Q_{g_1g_2}(i)$.\\

4444: (iii) $S_{g(i)j}=\e^{2\pi i Q_g(j)}S_{ij}$.\\

4445: (iv) $g\mapsto \e^{2\pi i Q_g(i)}$

4446: is a homomorphism $\Scr{G}\to U(1)$, as a consequence of (ii).\\

4447: (v) $Q_{\bar g}(\bi)=Q_g(i)$.\\

4448: (vi) $Q_g(\bi)=-Q_g(i)$, as a consequence of (ii) and (v).\\

4449: (vii) $\e^{2\pi i Q_g(i)}=\pm 1$ if $i=\bi$, as a consequence of (vi).\\

4450: (viii) $Q_g(g)=-2h_g$.\\

4451: (ix) $Q_{g_1}(g_2(i))-Q_{g_1}(i)=Q_{g_1}(g_2)$, as a consequence of

4452: (i).

4453: 

4454: \noindent

4455: (i) follows from the operator product expansion of

4456: $g$, $i$, and $j$. Since $N_{g_2i}^{\,\,g_2(i)}\ne 0$

4457: we find $Q_{g_1}(g_2)+Q_{g_1}(i)=Q_{g_1}(g_2(i))$ by (i).

4458: This is in fact equivalent to (ii).

4459: (iii) is shown in \cite{ScYa,SYQ,Intri}.

4460: (v) is because $h_{\bi}=h_i$.

4461: (viii) is because $Q_g(g)=-Q_{g^{-1}}(g)=-h_{g^{-1}}-h_g+h_1=-2h_g$.

4462: 

4463: 

4464: Let us consider $Q_{g_1}(g_2)=h_{g_1}+h_{g_2}-h_{g_1g_2}$,

4465: which is symmetric under the exchange $g_1\leftrightarrow g_2$.

4466: By the property (ii) above,

4467: $(g_1,g_2)\mapsto Q_{g_1}(g_2)$ is a symmetric bilinear form

4468: of $\Scr{G}$ with values in $\R/\Z$.

4469: There is not always  a symmetric bilinear form

4470: $q(g_1,g_2)$ of $\Scr{G}$ such that

4471: \beq

4472: Q_{g_1}(g_2)=2q(g_1,g_2)

4473: \qquad

4474: \mbox{mod $1$}.

4475: \eeq

4476: However, for some subgroup

4477: $G$ of $\Scr{G}$, there can be such a symmetric bilinear form.

4478: The existence of such a form is the condition for the

4479: existence of an $G$-orbifold with modular invariant partition function

4480: (see Appendix~\ref{app:Twist}).

4481: For example let us consider a simple current $g$ of order $N$,

4482: $g^N=1$.

4483: Let us note from (ii) and (viii) that

4484: $Q_{g^n}(g^m)=-2nmh_g$.

4485: For the group $G$ generated by $g$,

4486: a candidate bilinear form is thus

4487: $q(g^n,g^m)=-nmh_g$. However, this is well-defined

4488: (as a function with values in $\R/\Z$)

4489: if and only if $Nh_g$ is an integer.

4490: 

4491: \subsubsection*{\it Formulae involving $S$, $T$ and $P$}

4492: 

4493: We record some formulae

4494: involving $S,T,P$.

4495: We first quote from \cite{ScYa,SYQ,Intri} that

4496: \beq

4497: S_{g(i),j} = e^{2\pi i Q_g(j)} \, S_{ij} \quad, \quad 

4498: T_{g(i)} = e^{2\pi i (h_g-Q_g(i))} \, T_i.

4499: \eeq

4500: One can derive a similar relation for the $P$-matrix \cite{HSSc,FHSSW}:

4501: \beq

4502: P_{g^{2m}(i),j} = \phi (2m,i) \, e^{2\pi i m Q_g(j)} P_{ij},

4503: \eeq

4504: where 

4505: \beq

4506: \phi (2m,i) := e^{\pi i ( h_i-h_{g^{2m}i} - 2Q_{g^m}(g^m(i)))}.

4507: \eeq

4508: We are particularly interested in the case that $i=0$. In that case,

4509: the expression in brackets in the exponent becomes

4510: $-h_{g^{2m}}-2Q_{g^m}(g^m) = h_{g^{2m}}-4h_{g^m}$ mod $1$. This can be

4511: rewritten as $-Q_{g^{-m}}(g^{-m}) - Q_{g^m}(g^{-m})$. Applying property

4512: (ii) above, one obtains that this is $0$ mod $1$.

4513: The conclusion is that $\phi(2m,0)$ is just a sign, and therefore

4514: \beq\label{Pvac}

4515: P_{g^{2m},j} = \pm e^{2\pi i m Q_g(l)} \, P_{0,j}

4516: \eeq

4517: 

4518: 

4519: 

4520: 

4521: 

4522: 

4523: \appendix{Orbifolds}\label{app:Twist}

4524: 

4525: We consider the model ${\mathcal C}$ with the Hilbert space of states

4526: ${\mathcal H}^{\mathcal C}=\oplus_i\Scr{H}_i\otimes\Scr{H}_{\bi}$.

4527: Simple currents act on ${\mathcal H}^{\mathcal C}$ by

4528: \beq

4529: g:v\mapsto 

4530: \e^{2\pi i Q_g(i)}v,\qquad v\in\Scr{H}_i\otimes\Scr{H}_{\bi}.

4531: \eeq

4532: This is a discrete symmetry of the system.

4533: We record and explain some known facts on orbifold by a group of

4534: simple currents.

4535: 

4536: 

4537: 

4538: 

4539: \subsection{$g$-Twisted Hilbert Space}\label{subap:gtw}

4540: 

4541: Let us quantize the system on the circle $x\equiv x+1$

4542: with $g$-twisted boundary condition.

4543: Namely, we impose the boundary condition

4544: $\phi(x)=g\phi(x+1)$ on the fields.

4545: We are interested in the wavefunctions of such a system,

4546: the $g$-twisted states.

4547: We show that the space of such states is given by

4548: \beq

4549: {\mathcal H}_g=\bigoplus_i\Scr{H}_i\otimes \Scr{H}_{g(\bi)}.

4550: \label{Hg}

4551: \eeq

4552: 

4553: 

4554: The partition function

4555: $\Tr_{{\mathcal H}_g}q^{L_0-c/24}\overline{q}^{L_0-c/24}$

4556: is realized as the path-integral on the torus

4557: $(x,y)\equiv (x+1,y)\equiv (x,y+1)$ (with complex coordinate $z=x+\tau y$)

4558: on which the fields are periodic in ``time'',

4559: $\phi(x,y)=\phi(x,y+1)$, but obey the $g$-twisted boundary condition in

4560: ``space'' $\phi(x,y)=g\phi(x+1,y)$.

4561: One can switch the role of time and space by the coordinate transformation

4562: $(x,y)^t={\mathcal S}^{-1}(x',y')^t$ where

4563: ${\mathcal S}$ is the $SL(2,\Z)$ matrix (\ref{STC}).

4564: Then, we have $\phi(x',y')=\phi(x'+1,y')=g\phi(x',y'+1)$.

4565: Thus, the partition function can be written as

4566: \beqa

4567: \Tr\!\!\mathop{}_{{\mathcal H}_g}q^{L_0-c/24}\overline{q}^{L_0-c/24}

4568: &=&\Tr\!\!\mathop{}_{{\mathcal H}}g{q'}^{L_0-c/24}\overline{q'}^{L_0-c/24}

4569: =\sum_i\e^{2\pi i Q_g(i)}\chi_i(\tau')\overline{\chi_i(\tau')}

4570: \nn\\

4571: &=&\sum_{ijk}\e^{2\pi i Q_g(i)}\chi_j(\tau)S^{-1}_{ji}

4572: \overline{\chi_k(\tau)S^{-1}_{ki}}

4573: \nn\\

4574: &=&\sum_{ijk}S^{-1}_{g^{-1}(j)i}\overline{S^{-1}_{ki}}

4575: \chi_j(\tau)\overline{\chi_k(\tau)},

4576: \nn

4577: \eeqa

4578: where we have used $S_{g^{-1}(j)i}=\e^{-2\pi i Q_g(i)}S_{ji}$.

4579: Since $\sum_iS_{g^{-1}(j)i}^{-1}\overline{S^{-1}_{ki}}=\delta_{k,g^{-1}(j)}$

4580: we find that

4581: the partition function equals

4582: \beq

4583: \Tr_{{\mathcal H}_g}q^{L_0-c/24}\overline{q}^{L_0-c/24}

4584: =\sum_{j}\chi_j(\tau)\overline{\chi_{g^{-1}(j)}(\tau)}.

4585: \nonumber

4586: \eeq

4587: Since $\overline{g^{-1}(j)}=g(\bj)$, this proves Eq.~(\ref{Hg}).

4588: 

4589: 

4590: 

4591: \subsection{$g$ action on ${\mathcal H}_{g^n}$}

4592: 

4593: We next show that

4594: the discrete symmetry $g^m$ acts on ${\mathcal H}_{g^n}$

4595: by the phase multiplication

4596: \beq

4597: \e^{-2\pi i(Q_{g^m}(g^n(\bi))+nmh_g)}\times \quad

4598: \mbox{on the subspace}\quad

4599: \Scr{H}_i\otimes \Scr{H}_{g^n(\bi)}.

4600: \label{actQgg}

4601: \eeq

4602: %One can also show that this phase equals

4603: %$\pm\e^{\pi i (Q_g(i)-Q_g(g^n(\bi)))}$. In particular, $g^2$ acts by

4604: %phase multiplication $\e^{2\pi i (Q_g(i)-Q_g(j))}\times$

4605: %on $\Scr{H}_i\otimes\Scr{H}_j$.

4606: %Let $G$ be a group of simple currents such that each element $g$ can

4607: %be represented as $h^2$ for some simple current $h$

4608: %(which may not be in $G$).

4609: %Suppose $\e^{2\pi i Q_h(i)}=\e^{2\pi i Q_h(j)}$

4610: %for any $h^2\in G$ implies $i=j$. Then, $G$-invariant

4611: %subspace of ${\mathcal H}_g$ is given by

4612: %\beq

4613: %{\mathcal H}_g^G=\bigoplus_{i=g(\bi)}\Scr{H}_i\otimes \Scr{H}_i.

4614: %\eeq

4615: %In particular, if each $i$ obey $i=g(\bi)$ for some $g\in G$,

4616: %then the space of states of $G$-orbifold theory is given by

4617: %\beq

4618: %{\mathcal H}^{{\mathcal C}/G}=\bigoplus_i\Scr{H}_i\otimes \Scr{H}_i.

4619: %\eeq

4620: 

4621: 

4622: 

4623: To find the action of $g$ on ${\mathcal H}_{g^n}$ we compute

4624: the partition function

4625: $\Tr_{{\mathcal H}_{g^n}}gq^{L_0-c/24}\overline{q}^{L_0-c/24}$.

4626: This is realized as the path-integral on the torus

4627: with boundary condition

4628: $\phi(x,y)=g^n\phi(x+1,y)=g\phi(x,y+1)$.

4629: Under the change of coordinates $(x,y)^t={\mathcal S}

4630: {\mathcal T}^n{\mathcal S}^{-1}(x',y')^t$, the boundary

4631: condition becomes $\phi(x',y')=\phi(x'+1,y')=g\phi(x',y'+1)$.

4632: Thus the partition function

4633: can be written as

4634: \beqa

4635: \Tr\!\!\mathop{}_{{\mathcal H}_{g^n}}gq^{L_0-c/24}\overline{q}^{L_0-c/24}

4636: &=&\Tr\!\!\mathop{}_{{\mathcal H}}g{q'}^{L_0-c/24}\overline{q'}^{L_0-c/24}

4637: =\sum_i\e^{2\pi i Q_g(i)}\chi_i(\tau')\overline{\chi_i(\tau')}

4638: \nn\\

4639: &=&\sum_{ijk}

4640: \e^{2\pi i Q_g(i)}\chi_j(\tau)(ST^nS^{-1})_{ji}

4641: \overline{\chi_k(\tau)(ST^nS^{-1})_{ki}}

4642: \nn\\

4643: &=&\sum_{jkl}\chi_j(\tau)S_{jg(l)}T_{g(l)}^n

4644: \overline{\chi_k(\tau)S_{kl}T_l^n}

4645: \eeqa

4646: We note here that $T_{g(l)}^n\overline{T_l^n}=\e^{2\pi i n(h_{g(l)}-h_l)}

4647: =\e^{2\pi i Q_{g^{-n}}(g(l))}\e^{-2\pi i nh_g}$.

4648: Using $S_{jg(l)}\e^{2\pi i Q_{g^{-n}}(g(l))}=S_{g^{-n}(j)g(l)}

4649: =\e^{2\pi i Q_g(g^{-n}(j))}S_{g^{-n}(j)l}$, we find that the partition

4650: function is given by

4651: \beq

4652: \Tr\!\!\mathop{}_{{\mathcal H}_{g^n}}gq^{L_0-c/24}\overline{q}^{L_0-c/24}

4653: =\sum_j\e^{2\pi i (Q_g(g^{-n}(j))-nh_g)}

4654: \chi_j(\tau)\overline{\chi_{g^{-n}(j)}(\tau)}.

4655: \eeq

4656: This shows the action of $g$ on ${\mathcal H}_{g^n}$.

4657: $g^m$ action (\ref{actQgg}) is obtained by iteration.

4658: 

4659: 

4660: %Finally, let us show $2Q_g(g^{-n}(i))-2nh_g=Q_g(i)-Q_g(g^n(\bi))$ mod $\Z$.

4661: %We first note that $Q_{g^{-n}}(g)=nQ_{g^{-1}}(g)=2nh_g$, and therefore

4662: %$2nh_g=h_{g^{-n}}+h_g-h_{g^{1-n}}$. Thus, we have

4663: %$Q_g(g^{-n}(i))-2nh_g

4664: %=h_{g^{-n}(i)}+h_g-h_{g^{1-n}(i)}-(h_{g^{-n}}+h_g-h_{g^{1-n}})

4665: %=Q_{g^{1-n}}(i)-Q_{g^{-n}}(i)

4666: %=(1-n)Q_g(i)-(-n)Q_g(i)=Q_g(i)$.

4667: %By this we see $2Q_g(g^{-n}(i))-2nh_g =Q_g(g^{-n}(i))-2nh_g-Q_g(g^n(\bi))

4668: %=Q_g(i)-Q_g(g^n(\bi))$ mod $\Z$.

4669: 

4670: 

4671: \noindent

4672: {\bf Remark.}

4673: If $g$ is an order $N$ simple current, we find $Q_{g^N}(-)=0$ mod 1,

4674: by the property (ii) in Appendix~\ref{app:Q}.

4675: Thus, $g^N$ acts on ${\mathcal H}_{g^n}$ by 

4676: phase multiplication $\e^{-2\pi i nNh_g}\times$.

4677: If we want $g$ to be order $N$ also in the action on the twisted states,

4678: we need $Nh_g$ to be an integer.

4679: This explains why we need $Nh_g\in\Z$ in order to have

4680: the orbifold of ${\mathcal C}$

4681: by the group $G=\{g^n\}_{n=0}^{N-1}$.

4682: 

4683: \subsection{$g_1$ action on ${\mathcal H}_{g_2}$}

4684: 

4685: Suppose a group $G$ of simple currents has

4686: a symmetric bilinear form $q(g_1,g_2)\in\R/\Z$

4687: obeying $Q_{g_1}(g_2)=2q(g_1,g_2)$ mod $1$

4688: such that $q(g,g)=-h_g$.

4689: Then, one can define a $G$-orbifold of ${\mathcal C}$

4690: with the modular-invariant partition function

4691: \beq

4692: Z^{{\mathcal C}/G}

4693: ={1\over |G|}\sum_{i,g_1,g_2}

4694: \e^{2\pi i(Q_{g_2}(i)-q(g_2,g_1))}

4695: \chi_i(\tau)\overline{\chi_{g_1^{-1}(i)}(\tau)}.

4696: \label{MIPF}

4697: \eeq

4698: (Note that the phase can also be written as

4699: $\e^{-2\pi i(Q_{g_2}(g_1(\bi))-q(g_2,g_1))}$.)

4700: This shows that a $g_2$ action on ${\mathcal H}_{g_1}$

4701: (for $g_1,g_2\in G$)

4702: is given by

4703: \beq

4704: \e^{-2\pi i(Q_{g_2}(g_1(\bi))-q(g_2,g_1))}\times

4705: \quad\mbox{on the subspace}\quad

4706: \Scr{H}_i\otimes\Scr{H}_{g_1(\bi)}.

4707: \label{TTT}

4708: \eeq

4709: 

4710: It is straightforward to show modular invariance of (\ref{MIPF}).

4711: Under $\tau\to \tau+1$, the extra phase

4712: $h_i-h_{g_1^{-1}(i)}=Q_{g_1^{-1}}(i)-h_{g_1}$ appears.

4713: At this point, we use $h_{g_1}=-q(g_1,g_1)=q(g_1^{-1},g_1)$.

4714: Then, we see that

4715: the expression for $Z(\tau+1)$ is the same as (\ref{MIPF}) with $g_2$

4716: replaced by $g_2g_1^{-1}$.

4717: On the other hand

4718: \beq

4719: Z \left( -{1\over\tau} \right)

4720: ={1\over |G|}\sum\e^{2\pi i (Q_{g_2}(i)-q(g_2,g_1))}

4721: \chi_j(\tau)S_{ji}\overline{\chi_k(\tau)}S^{-1}_{kg_1^{-1}(i)}.

4722: \nn

4723: \eeq

4724: We now use

4725: $\e^{2\pi iQ_{g_2}(i)}S_{ji}=S_{g_2(j)i}$ and

4726: $S^{-1}_{kg_1^{-1}(i)}=\e^{2\pi i Q_{g_1}(k)}S^{-1}_{ki}$,

4727: so that the sum over $i$ can be performed,

4728: $\sum_iS_{g_2(j)i}S^{-1}_{ki}=\delta_{k,g_2(i)}$.

4729: Noting $Q_{g_1}(g_2(j))-q(g_2,g_1)=Q_{g_1}(j)+q(g_1,g_2)$, we find that

4730: it is the same as (\ref{MIPF}) with $(g_1,g_2)$

4731: replaced by $(g_2^{-1},g_1)$.

4732: 

4733: 

4734: 

4735: The complete set of all simple current modular invariant partition functions

4736: is obtained in \cite{KrSc}.

4737: There are other modular invariants

4738: associated with the ``discrete torsion''

4739: \cite{Vafa:DT}.

4740: Addition of a discrete torsion corresponds to

4741:  shifting $q(g_2,g_1)$ in (\ref{TTT}) by

4742: an antisymmetric bilinear form of $G$ with values in $\R/\Z$

4743: that vanishes on the diagonals $(g,g)$.

4744: 

4745: 

4746: \subsection{Quantum Symmetry}\label{subap:quantums}

4747: 

4748: For each character $g\mapsto \e^{2\pi i \rho(g)}$ of $G$,

4749: there is a symmetry $g_{\rho}$ of

4750: the orbifold theory ${\mathcal C}/G$

4751: that acts on the $g$-twisted sector

4752: states by the multiplication by the phase

4753: $\e^{2\pi i\rho(g)}$. This is called a {\it quantum symmetry}.

4754: The group of quantum symmetries is the character group $G^{\vee}$, which

4755: is isomorphic to $G$ itself.

4756: 

4757: Let us find out what  the $g_{\rho}$-twisted states are

4758: in the orbifold theory.

4759: We recall that the untwisted

4760: states of the orbifold model are $G$-invariant states in

4761: $\oplus_{h\in G}{\mathcal H}_{h}$, the states obeying $g=1$,

4762: $\forall g\in G$.

4763: We claim that the $g_{\rho}$-twisted states

4764: are the states in $\oplus_{h\in G}{\mathcal H}_{h}$

4765: obeying $g=\e^{2\pi i\rho(g)}$, $\forall g\in G$.

4766: Namely

4767: \beq

4768: ({\mathcal H}^{{\mathcal C}/G})_{g_{\rho}}

4769: =\bigoplus_{j,h}

4770: \left.

4771: \Scr{H}_i\otimes\Scr{H}_{h(\bi)}

4772: \right|_{g=\e^{2\pi i \rho(g)},\forall g\in G}.

4773: \eeq

4774: Since the action of $g$ on $\Scr{H}_i\otimes\Scr{H}_{h(\bi)}$

4775: is given in (\ref{TTT}), it is

4776: the sum of $\Scr{H}_i\otimes\Scr{H}_{h(\bi)}$ over those $(i,h)$ such that

4777: $\e^{-2\pi i (Q_g(h(\bi))-q(g,h))}=\e^{2\pi i \rho(g)}$ for any $g\in G$.

4778: 

4779: 

4780: This is shown as in Appendix \ref{subap:gtw}.

4781: The partition function on the $g_{\rho}$-twisted circle

4782: is the same as the partition function on the untwisted circle,

4783: but with a $g_{\rho}$ insertion in the evolution operator:

4784: \beq

4785: Z_{\rho}^{{\mathcal C}/G}

4786: =

4787: {1\over |G|}

4788: \sum_{i,g_1,g_2}

4789: \e^{2\pi i\rho(g_1)}\e^{2\pi i (Q_{g_2}(i)-q(g_2,g_1))}

4790: \chi_i(\tau')\overline{\chi_{g_1^{-1}(i)}(\tau')}.

4791: \eeq

4792: After a manipulation similar to

4793: Appendix~\ref{subap:gtw}, we find that it is equal to

4794: \beq

4795: {1\over |G|}\sum_{j,g_1,g_2}

4796: \e^{2\pi i\rho(g_1)}

4797: \e^{2\pi i (Q_{g_1}(g_2(\bj))-q(g_2,g_1))}\chi_j(\tau)

4798: \overline{\chi_{g_2^{-1}(j)}(\tau)}.

4799: \eeq

4800: This shows the claim.

4801: 

4802: 

4803: 

4804: \appendix{Alternative Way of Dressing}\label{app:some}

4805: 

4806: For a crosscap state $|\Scr{C}_P\rangle$ and a symmetry

4807: $g$, what is $g|\Scr{C}_P\rangle$?

4808: We first note that

4809: \beqa

4810: \langle \Scr{B}_{\alpha}|\e^{-{L\over 2} H_c(2\beta)}g|\Scr{C}_P\rangle

4811: &=&\langle \Scr{B}_{g^{-1}(\alpha)}|\e^{-{L\over 2} H_c(2\beta)}

4812: |\Scr{C}_P\rangle

4813: \nn\\

4814: &=&\Tr\!\!\mathop{}_{{\mathcal H}_{g^{-1}(\alpha),Pg^{-1}(\alpha)}}

4815: P\e^{-\beta H_o(L)}

4816: \nn\\

4817: &=&\Tr\!\!\mathop{}_{{\mathcal H}_{\alpha,gPg^{-1}(\alpha)}}

4818: gPg^{-1}\e^{-\beta H_o(L)}.

4819: \eeqa

4820: This suggests that $g|\Scr{C}_P\rangle$ is the crosscap state

4821: for the parity $gPg^{-1}$.

4822: Under this interpretation,

4823: $\langle\Scr{C}_{P'}|q^{H_c}g|\Scr{C}_P\rangle$ can be regarded as

4824: $\langle\Scr{C}_{P'}|q^{H_c}|\Scr{C}_{gPg^{-1}}\rangle$ as well as

4825: $\langle\Scr{C}_{g^{-1}P'g}|q^{H_c}|\Scr{C}_P\rangle$.

4826: The two indeed agree because

4827: \beq

4828: \Tr\!\!\mathop{}_{{\mathcal H}_{P'gP^{-1}g^{-1}}}gPg^{-1}

4829: \e^{-\beta H_c}

4830: =\Tr\!\!\mathop{}_{{\mathcal H}_{g^{-1}P'gP^{-1}}}P

4831: \e^{-\beta H_c},

4832: \eeq

4833: where we have used that

4834: $g$ maps ${\mathcal H}_{h}$ to ${\mathcal H}_{ghg^{-1}}$

4835: (we have in mind $h=g^{-1}P'gP^{-1}$).

4836: Thus, we conclude that

4837: \beq

4838: g|\Scr{C}_P\rangle=|\Scr{C}_{gPg^{-1}}\rangle.

4839: \label{gC}

4840: \eeq

4841: If $P$ and $gP$ are both involutive (or more weakly

4842: if $(gP)^2=P^2$), then we find $gPg=P$ and hence

4843: \beq

4844: gPg^{-1}=g^2P.

4845: \eeq

4846: 

4847: 

4848: \newcommand{\wtP}{\widetilde{P}}

4849: \newcommand{\wtg}{\widetilde{g}}

4850: 

4851: 

4852: Let us apply this to the case of $P=P_0$ and $g$ the symmetry

4853: associated with a simple current:

4854: The crosscap state for the parity

4855: $\wtP_g:=gP_0g^{-1}$ is $g|\Scr{C}_{P_0}\rangle$ with the coefficients

4856: \beq

4857: \widetilde{\gamma}_{gi}=\e^{2\pi i Q_g(i)}{P_{0i}\over\sqrt{S_{0i}}}.

4858: \eeq

4859: Since $(gP_0)^2=P_0^2=1$ on ${\mathcal H}$, we find

4860: $\wtP_g=g^2P_0$ on ${\mathcal H}$.

4861: % just like $P_{g^2}$.

4862: %and therefore it maps the Cardy branes as

4863: %\beq

4864: %\wtP_g:i\mapsto g^2(\bi).

4865: %\eeq

4866: %It is easy to see that the corresponding numbers

4867: %$k_{\wtg_1\wtg_2}^i=Y_{g_1^{-1}g_2(i)0}^0$,

4868: %$k_{g_1\wtg_2}^i=Y_{g_2(i)0}^{g_1}$

4869: %and $m_{i\wtg}^j=Y_{g(\bi)0}^{\bj}$ obey the condition (\ref{Ca4}).

4870: Since the action of $\wtP_g$ and $P_{g^2}$ agree on ${\mathcal H}$,

4871: it is a natural question to ask whether they are the same parity symmetry.

4872: In fact, we found in Appendix~\ref{app:Q} that

4873: \beq

4874: P_{g^2i}=\pm \e^{2\pi i Q_g(i)}P_{0i},

4875: \eeq

4876: where the sign $\pm$ depends only on $g$ but not on $i$.

4877: Thus, we indeed see that the crosscap states

4878: $|\Scr{C}_{P_{g^2}}\rangle$ and $|\Scr{C}_{\wtP_{g}}\rangle$

4879: agree up to a sign.

4880: 

4881: 

4882: 

4883: %For them to be the same,

4884: %the crosscap coefficients have to agree, which requires

4885: %$P_{g^2i}=\e^{2\pi i Q_g(i)}P_{0i}$.

4886: %This holds in many of the examples as we will see.

4887: 

4888: 

4889: \appendix{Partition Functions of the Circle Sigma Model}

4890: 

4891: We record here the cylinder, Klein bottle and M\"obius strip amplitudes

4892: of the circle sigma model.

4893: This can be used to justify the formula for the boundary and crosscap states

4894: used in Section~\ref{sec:arbR}.

4895: To perform the modular transformation, we will make use of the 

4896: Poisson resummation formula

4897: $$

4898: \sum_{n\in \Z}

4899: \e^{-\pi \alpha n^2-2\pi i\beta n}

4900: ={1\over \sqrt{\alpha}}\sum_{m\in\Z}

4901: \e^{-{\pi\over\alpha}(m+\beta)^2},

4902: $$

4903: as well as the relations

4904: $$

4905: f_1(\e^{-\pi/T})=\sqrt{T}f_1(\e^{-\pi T}),

4906: \quad\,

4907: f_3(\e^{-\pi/T})=f_3(\e^{-\pi T}),

4908: \quad\,

4909: f_2(\e^{-\pi/T})=f_4(\e^{-\pi T}),

4910: $$

4911: among the functions

4912: \beqa

4913: &&\mbox{$f_1(q)=q^{1\over 12}\prod_{n=1}^{\infty}(1-q^{2n}),\qquad

4914: f_2(q)=\sqrt{2}q^{1\over 12}\prod_{n=1}^{\infty}(1+q^{2n})$,}

4915: \nn\\

4916: &&\mbox{$f_3(q)=q^{-{1\over 24}}\prod_{n=1}^{\infty}(1+q^{2n-1}),\qquad

4917: f_4(q)=q^{-{1\over 24}}\prod_{n=1}^{\infty}(1-q^{2n-1})$.}

4918: \nn

4919: \eeqa

4920: Note that 

4921: $f_1(q)=\eta(q^2)$.

4922: 

4923: The cylinder amplitudes are ($q=\e^{-2\pi T}, q_t=\e^{-\pi/T}$):

4924: \beqa

4925: &&\Tr\!\!\mathop{}_{N_{a_1}N_{a_2}}(q^H)

4926: ={\sum_{l}q^{({l\over R}+{\mDa\over 2\pi})^2}

4927: \over

4928: \eta(q)}

4929: ={R\over\sqrt{2}}{\sum_m q_t^{{1\over 2}(Rm)^2}

4930: \e^{-iR\mDa m}\over\eta(q_t^2)}

4931: =\langle N_{a_1}|q_t^H|N_{a_2}\rangle,

4932: \qquad

4933: \\

4934: &&

4935: \Tr\!\!\mathop{}_{D_{x_1}D_{x_2}}(q^H)

4936: ={\sum_{m}q^{(Rm+{\mDx\over 2\pi})^2}

4937: \over

4938: \eta(q)}

4939: ={1\over R\sqrt{2}}{\sum_l q_t^{{1\over 2}({l\over R})^2}

4940: \e^{-i{\mDx\over R} l}\over\eta(q_t^2)}

4941: =\langle D_{x_1}|q_t^H|D_{x_2}\rangle,

4942: \\

4943: &&

4944: \Tr\!\!\mathop{}_{DN}(q^H)={q^{1\over 48}\over

4945: \prod_{n=1}^{\infty}(1-q^{n-{1\over 2}})}

4946: ={1\over

4947: \sqrt{2}q_t^{1\over 12}\prod_{n=1}^{\infty}(1+q_t^{2n})}

4948: =\langle D|q_t^H|N\rangle.

4949: \eeqa

4950: Here $\mDa=a_2-a_1$ and $\mDx=x_2-x_1$.

4951: 

4952: The Klein bottle amplitudes are ($q=\e^{-2\pi T}, q_t=\e^{-\pi/2T}$):

4953: \beqa

4954: &&

4955: \Tr\!\!\mathop{}_{\mathcal H}(\Omega q^H)

4956: ={\sum_l q^{l^2\over 2R^2}\over \eta(q^2)}

4957: =R\sqrt{2}{\sum_m^{\rm even} q_t^{R^2m^2\over 2}\over\eta(q_t^2)}

4958: =\langle\Scr{C}_{\Omega}|q_t^H|\Scr{C}_{\Omega}\rangle,\\

4959: &&

4960: \Tr\!\!\mathop{}_{\mathcal H}(s\Omega q^H)

4961: ={\sum_l (-1)^lq^{l^2\over 2R^2}\over \eta(q^2)}

4962: =R\sqrt{2}{\sum_m^{\rm odd} q_t^{R^2m^2\over 2}\over\eta(q_t^2)}

4963: =\langle\Scr{C}_{s\Omega}|q_t^H|\Scr{C}_{s\Omega}\rangle.

4964: \eeqa

4965: 

4966: 

4967: 

4968: 

4969: The M\"obius strip amplitudes are:

4970: ($q=\e^{-2\pi T}, q_t=\e^{-\pi/4T}$)

4971: \beqa

4972: &&

4973: \Tr\!\!\mathop{}_{N_{a}N_{-a}}(\Omega q^H)

4974: ={\sum_{l}q^{({l\over R}+{\mDa\over 2\pi})^2}

4975: \over

4976: q^{1\over 24}\prod (1-(-1)^nq^n)}

4977: =R{\sum_{m}^{\rm even} q_t^{{1\over 2}(Rm)^2}

4978: \e^{-iR\mDa m}

4979: \over

4980: q_t^{1\over 12}\prod (1-(-1)^nq_t^{2n})}

4981: =\langle N_{a}|q_t^H|\Scr{C}_{\Omega}\rangle,

4982: \qquad

4983: \\

4984: &&

4985: \Tr\!\!\mathop{}_{N_{a}N_{-a}}(s\Omega q^H)

4986: ={\sum_{l}(-1)^lq^{({l\over R}+{\mDa\over 2\pi})^2}

4987: \over

4988: q^{1\over 24}\prod (1-(-1)^nq^n)}

4989: =R{\sum_{m}^{\rm odd} q_t^{{1\over 2}(Rm)^2}

4990: \e^{-iR\mDa m}

4991: \over

4992: q_t^{1\over 12}\prod (1-(-1)^nq_t^{2n})}

4993: =\langle N_{a}|q_t^H|\Scr{C}_{s\Omega}\rangle,

4994: \qquad\quad

4995: \\

4996: &&

4997: \Tr\!\!\mathop{}_{D_xD_x}(\Omega q^H)

4998: ={1

4999: \over

5000: q^{1\over 24}\prod (1+(-1)^nq^n)}

5001: ={1

5002: \over

5003: q_t^{1\over 12}\prod (1+(-1)^nq_t^{2n})}

5004: =\langle D_x|q_t^H|\Scr{C}_{\Omega}\rangle,

5005: \qquad

5006: \\

5007: &&

5008: \Tr\!\!\mathop{}_{D_xD_{x-\pi R}}(s\Omega q^H)=0

5009: =\langle D_x|q_t^H|\Scr{C}_{s\Omega}\rangle.

5010: \eeqa

5011: Here $\mDa=(-a)-a=-2a$.

5012: The last partition function vanishes because

5013: $s\Omega$ maps $|m\rangle_{x,x-\pi R}$

5014: to $|-m\rangle_{x,x+\pi R}=|1-m\rangle_{x,x-\pi R}$, which cannot be the same

5015: as $|m\rangle_{x,x-\pi R}$ for integer $m$.

5016: 

5017: The cylinder with $g_{\mDx}$-twist is ($q=\e^{-2\pi T},q_t=\e^{-\pi/T}$)

5018: \beqa

5019: &&

5020: \Tr\!\!\mathop{}_{N_{a_1}N_{a_2}}(g_{\mDx}q^H)

5021: ={\sum_{l}q^{({l\over R}+{\mDa\over 2\pi})^2}

5022: \e^{-i\mDx({l\over R}+{\mDa\over 2\pi})}

5023: \over

5024: \eta(q)}

5025: ={R\over\sqrt{2}}

5026: {\sum_m q_t^{{1\over 2}(Rm-{\mDx\over 2\pi})^2}

5027: \e^{-iR\mDa m}\over\eta(q_t^2)}

5028: \nn\\

5029: &&\qquad\qquad\qquad\qquad

5030: ={}_{g_{\mDx}}\langle N_{a_1}|q_t^H|N_{a_2}\rangle_{g_{\mDx}},

5031: \eeqa

5032: where $\mDa=a_2-a_1$.

5033: M\"obius strip with $g_{\mDx}$-twist is

5034: ($q=\e^{-2\pi T},q_t=\e^{-\pi/4T}$)

5035: \beqa

5036: &&

5037: \Tr\!\!\mathop{}_{N_{a}N_{-a}}(g_{\mDx}q^H)

5038: ={\sum_{l}q^{({l\over R}+{(-2a)\over 2\pi})^2}

5039: \e^{-i\mDx({l\over R}+{(-2a)\over 2\pi})}

5040: \over

5041: q^{1\over 24}\prod_{n=1}^{\infty}(1-(-1)^nq^n)}

5042: =R

5043: {\sum_m^{\rm even} q_t^{{1\over 2}(Rm-{\mDx\over \pi})^2}

5044: \e^{iRa m}\over

5045: q_t^{1\over 12}\prod_{n=1}^{\infty}(1-(-1)^nq_t^{2n})}

5046: \nn\\

5047: &&\qquad\qquad\qquad\qquad

5048: ={}_{g_{2\mDx}}\langle N_{a}|q_t^H|\Scr{C}_{g_{\mDx}\Omega}\rangle.

5049: \eeqa

5050: 

5051: 

5052: \appendix{ Formulae for $SU(2)/U(1)$}

5053: 

5054: \subsection{A-type crosscaps}

5055: 

5056: We compute the explicit coefficients of the A-type crosscaps

5057: \beq

5058: \ket{\Scr{C}_{\ell}} = \sum_{(j,n) \in PF_k} 

5059: \frac{P_{(0,2\ell)(j,n)}}{\sqrt{S_{(0,0)(j,n)}}} \ \cket{(j,n)}

5060: \eeq

5061: using the formula

5062: \beq

5063: P_{(j,n)(j',n')} = T^{\half}_{j,n} \left(

5064: Q_{jj'} Q_{nn'}^* + Q_{\frac{k}{2}-j, j'} Q_{n+k,k'}^* \right)

5065: T^{\half}_{j',n'}.

5066: \eeq

5067: The subtlety is that $T_{j,n}^{\half}$ does not usually

5068: factorize as $T_j^{\half}T_n^{-\half}$ except in the standard range

5069: (henceforth $S.R.$) where (\ref{paraweight}) holds.

5070: Using

5071: \beqa

5072: &&

5073: T^{\half}_{j,n} = T^{\half}_j \ T^{-\half}_n,\qquad (j,n)\in S.R.,

5074: \nn\\

5075: &&

5076: T^{\half}_{j,n} = T^{\half}_{\frac{k}{2}-j} \ T^{-\half}_{n+k} 

5077: = T^{\half}_j \ T^{-\half}_n (-1)^{\frac{2j+n}{2}},

5078: \qquad (\mbox{${k\over 2}$}-j,n+k)\in S.R.,

5079: \nn\\

5080: &&

5081: T^{\half}_{j,n} = T^{\half}_{\frac{k}{2}-j} \ T^{-\half}_{n-k} 

5082: = T^{\half}_j \ T^{-\half}_n (-1)^{\frac{2j-n}{2}},

5083: \qquad (\mbox{${k\over 2}$}-j,n-k)\in S.R.,

5084: \eeqa

5085: the required $P$-matrix elements are computed to be

5086: \beq

5087: P_{(0,2\ell)(j,n)} = (-1)^{\ell} P^{SU(2)}_{0,j} \

5088: \left( P^{U(1)}_{2\ell,n} \right)^* +  P^{SU(2)}_{\frac{k}{2},j} \ 

5089: \left( P^{U(1)}_{2\ell+k,n}\right)^*,

5090: \eeq

5091: for $(j,n)$ in the standard range and $\ell$ in the range

5092: $-k\leq 2\ell\leq k$. If $(k/2-j,n\pm k)$ is in the standard range,

5093: one needs the extra sign factor $(-1)^{\frac{2j\pm n}{2}}$.

5094: The explicit expression for the crosscap is

5095: \beqa

5096: \lefteqn{\ket{\Scr{C}_{\ell}} = {1\over[k(k+2)]^{\frac{1}{4}}}\times}

5097: \nn\\

5098: && \sum_{(j,n)\in S.R.}

5099: \e^{\frac{\pi i \ell n}{k}}

5100: \left(\delta^{(2)}_{n+k}

5101: (-1)^{\ell} \sqrt{\tan\mbox{$\frac{\pi(2j+1)}{2(k+2)}$}}

5102: +\delta^{(2)}_{n}

5103:  (-1)^{\frac{2j+n}{2}} 

5104: \sqrt{\cot\mbox{$\frac{\pi(2j+1)}{2(k+2)}$}}

5105: \right) \cket{(j,n)},

5106: \nn\\

5107: &&

5108: \label{Aexpl}

5109: \eeqa

5110: where, on the RHS, we need to bring $\ell$ in the range

5111: $-k\leq 2\ell\leq k$.

5112: %In case one wants to use a summation over field identified states

5113: %(not in the standard range), the appropriate signs have to be

5114: %included.

5115: 

5116: \subsection{B-type crosscaps}\label{subapp:Bcc}

5117: 

5118: We first construct A-type crosscaps in the orbifold, and then

5119: apply the mirror map. The crosscaps of the orbifold are

5120: \beq

5121: \ket{\Scr{C}_{P^{\theta_r}}} = \frac{1}{\sqrt{k}} \ \sum_{\ell}

5122: \e^{-\pi i \theta_r (\ell)} \ \ket{\Scr{C}_{\ell}},

5123: \eeq

5124: where $\theta_r(\ell)= -2r\ell/k$, as explained in the main text.

5125: We also have set $\omega_1=0$.

5126: Inserting the states (\ref{Aexpl}), one sees that the following

5127: summations over $\ell$ are relevant:

5128: $$

5129: \sum_{\ell}(-1)^{\ell} \e^{\frac{\pi i \ell}{k}(2r+n)}

5130: = k \delta^{(2k)}_{2r+n+k},\qquad

5131: \sum_{\ell} \e^{\frac{\pi i \ell}{k}(2r+n)}

5132: = k \delta^{(2k)}_{2r+n}.

5133: $$

5134: The first sum plays a role when one sums up $\ell$ in the first term

5135: ($\sim \sqrt{\tan}$)

5136: in the parenthesis in (\ref{Aexpl}), and the second sum is relevant

5137: for the second term ($\sim \sqrt{\cot}$) in the parenthesis. 

5138: $n$ is projected on either $n=-2r-k$ or $n=-2r$. Only one term

5139: gives a contribution, since in (\ref{Aexpl}) we are summing over

5140: $(j,m)$ in the standard range.

5141: This leads to the following expressions

5142: \beqa

5143: \ket{\Scr{C}_{P^{\theta_r}}}&=&

5144: \frac{k^{\frac{1}{4}}}{(k+2)^{\frac{1}{4}}}\Biggl[

5145: \sum_{j,(j,-2r)\in S.R.} 

5146: (-1)^{j} \ (-1)^r \ 

5147: \sqrt{\cot\mbox{$\frac{\pi(2j+1)}{2(k+2)}$}}

5148: \ \cket{(j,-2r)} \nn\\

5149: && ~~+ \sum_{j,(j,-2r)\notin S.R.} 

5150: \sqrt{\cot\mbox{$\frac{\pi(2j+1)}{2(k+2)}$}}

5151: \cket{(j,-2r)} \Biggr].

5152: \eeqa

5153: Applying the mirror map, one obtains the B-type

5154: crosscaps

5155: \beqa

5156: \ket{\Scr{C}_r^B}&=&

5157: \frac{k^{\frac{1}{4}}}{(k+2)^{\frac{1}{4}}}\Biggl[

5158: \sum_{j,(j,-2r)\in S.R.} 

5159: (-1)^{j} \ (-1)^r \ 

5160: \sqrt{\cot\mbox{$\frac{\pi(2j+1)}{2(k+2)}$}}

5161: \ \cket{(j,2r)}_B \nn\\

5162: && ~~+ \sum_{j,(j,-2r)\notin S.R.} 

5163: \sqrt{\cot\mbox{$\frac{\pi(2j+1)}{2(k+2)}$}}

5164: \cket{(j,2r)}_B \Biggr]

5165: \eeqa

5166: For the standard crosscap state with $r=0$, only the first term contributes,

5167: since all states $(j,0)$ are in the standard range. For the state

5168: with $r=k/2$ ($k$ even), only the second term contributes.

5169: The respective crosscap states can be rewritten as

5170: \beqa

5171: &&

5172: \ket{\Scr{C}_{0}^B}

5173: ={k^{1\over 4}\over (k+2)^{1\over 4}}\sum_{j\,{\rm integer}}

5174: (-1)^j\sqrt{\cot\mbox{${\pi(2j+1)\over 2(k+2)}$}}

5175: \cket{(j,0)}_B,

5176: \\

5177: &&

5178: \ket{\Scr{C}_{\frac{k}{2}}^B}

5179: ={k^{1\over 4}\over (k+2)^{1\over 4}}\sum_{j\,{\rm integer}}

5180: \sqrt{\tan\mbox{${\pi(2j+1)\over 2(k+2)}$}}

5181: \cket{(j,0)}_B.

5182: \eeqa

5183: 

5184: 

5185: 

5186: 

5187: \subsection{Computation of one-loop amplitudes}\label{app:detai}

5188: 

5189: Here we record some detail of the computation of

5190: the one-loop amplitudes

5191: (\ref{AparaBB}), (\ref{AparaBC}), and (\ref{AparaCC}).

5192: For (\ref{AparaBB}):

5193: \beqa

5194: \langle \Scr{B}_{J,M} | q_t^H\ket{\Scr{B}_{J'M'}}

5195: &=& \sum_{(j,m)\in PF_k} N_{(J,-M)(J',M')}^{(j,-m)}  \chi_{j,m}(\tau)

5196: \nn\\

5197: &=&{1\over 2}\sum_{2j+m \,{\rm even}}N_{(J,-M)(J',M')}^{(j,-m)}

5198: \chi_{j,m}(\tau)

5199: =\sum_{2j+m \,{\rm even}}

5200: N_{JJ'}^{j} \delta^{(2k)}_{M'-M+m} \chi_{j,m}(\tau),

5201: \nn

5202: \eeqa

5203: where we have used (\ref{paraN}) in the last step.

5204: For (\ref{AparaBC}), we first note

5205: $$

5206: \langle\Scr{C}_{\ell}|q_t^H\ket{\Scr{B}_{(J,M)}} 

5207: =\sum_{(j,m)\in PF_k} Y_{(J,M) \ (j,m)}^{(0,2\ell)} 

5208: \widehat{\chi}_{j,m}(\tau) 

5209: = \sum_{2j+m\,{\rm even}} \tilde{Y}_{Jj}^{\frac{k}{2}} 

5210: \overline{\tilde{Y}}_{Mm}^{2\ell+k} 

5211: T^{-\frac{1}{2}}_{0,2\ell} \,\chi_{j,m}(\tau+\frac{1}{2}).

5212: $$

5213: Inserting the known $Y$-tensors from the $U(1)$ theory,

5214: we see that this is equal to

5215: \beqa

5216: &&\sum_{2j+m\,{\rm even}} Y_{Jj}^{\frac{k}{2}} \ \delta^{(2)}_{m+k} \ 

5217: \left( \delta^{(2k)}_{M-\ell+\frac{m-k}{2}}+ (-1)^{m+k} \ 

5218: \delta^{(2k)}_{M-\ell+\frac{m+k}{2}} \right)

5219: e^{-\pi i (h_j-\frac{m^2}{4k} - \frac{c}{24})} \ \chi_{j,m} (\tau+\half)

5220: \nn\\

5221: &&=

5222: \sum_{2j+m\,{\rm even}}

5223: Y_{Jj}^{\frac{k}{2}} \ \delta^{(2k)}_{2M-2\ell-k+m} \ 

5224: e^{-\pi i (h_j-\frac{m^2}{4k} - \frac{c}{24})} \chi_{j,m}(\tau+\half).

5225: \nn

5226: \eeqa

5227: Replacing $(j,m)\to ({k\over 2}-j,m+k)$ in the sum, and using

5228: $Y_{J,{k\over 2}-j}^{k\over 2}=N_{JJ}^j$, we find

5229: this to be equal to

5230: $$

5231: \sum_{2j+m\,{\rm even}}

5232: N_{JJ}^j \delta^{(2k)}_{2M-2\ell+m} \ 

5233: e^{\pi i (h_{j,m}-h_{{k\over 2}-j}+\frac{(m+k)^2}{4k})}

5234: \widehat{\chi}_{j,m}(\tau).

5235: $$

5236: It is straightforward to see that

5237: $$

5238: \epsilon_{j,m}:=

5239: \e^{\pi i (h_{j,m}-h_{{k\over 2}-j}+{(m+k)^2\over 4k})}

5240: =\left\{\begin{array}{ll}

5241: 1&({k\over 2}-j,m+k)\in S.R.\\

5242: (-1)^{2j+m\over 2}&(j,m)\in S.R.\\

5243: (-1)^m&({k\over 2}-j,m-k)\in S.R.

5244: \end{array}\right.

5245: $$

5246: This shows (\ref{AparaBC}).

5247: Computation of (\ref{AparaCC}) is similarly straightforward.

5248: It is convenient to use

5249: $Y_{j0}^0=(-1)^{2j}$ and $Y_{j0}^{k\over 2}=N_{j,{{k\over 2}-j}}^{\,\,0}

5250: =\delta_{j,{k\over 2}-j}=\delta_{j,{k\over 4}}$ (the latter is possible only

5251: for $k$ even).

5252: 

5253: 

5254: 

5255: 

5256: 

5257: \begin{thebibliography}{99}

5258: 

5259: \small

5260: \parskip=0pt plus 2pt

5261: 

5262: 

5263: %\cite{Sagnotti:1987tw}

5264: \bibitem{Stalk}

5265: A.~Sagnotti,

5266: ``Open strings and their symmetry groups'',

5267: ROM2F-87-25,

5268: {\it Talk presented at the Carg\`ese Summer Institute on Non-Perturbative Methods in Field Theory, Carg\`ese, 1987}

5269: [arXiv:hep-th/0208020].

5270: 

5271: %\cite{Bianchi:1990yu}

5272: \bibitem{BSI}

5273: M.~Bianchi and A.~Sagnotti,

5274: ``On the systematics of open string theories'',

5275: Phys.\ Lett.\ B {\bf 247}, 517 (1990).

5276: %%CITATION = PHLTA,B247,517;%%

5277: 

5278: 

5279: %\cite{Bianchi:1990tb}

5280: \bibitem{BSII}

5281: M.~Bianchi and A.~Sagnotti,

5282: ``Twist symmetry and open string Wilson lines'',

5283: Nucl.\ Phys.\ B {\bf 361}, 519 (1991).

5284: %%CITATION = NUPHA,B361,519;%%

5285: 

5286: %\cite{Horava:1989vt}

5287: \bibitem{HOI}

5288: P.~Horava,

5289: ``Strings on world sheet orbifolds'',

5290: Nucl.\ Phys.\ B {\bf 327}, 461 (1989).

5291: %%CITATION = NUPHA,B327,461;%%

5292: 

5293: %\cite{Horava:1989ga}

5294: \bibitem{HOII}

5295: P.~Horava,

5296: ``Background duality of open string models'',

5297: Phys.\ Lett.\ B {\bf 231}, (1989) 251.

5298: %%CITATION = PHLTA,B231,251;%%

5299: 

5300: %\cite{Angelantonj:2002ct}

5301: \bibitem{ASreview}

5302: C.~Angelantonj and A.~Sagnotti,

5303: ``Open strings'',

5304: arXiv:hep-th/0204089.

5305: %%CITATION = HEP-TH 0204089;%%

5306: 

5307: \bibitem{CLNY}

5308: C.~G.~Callan, C.~Lovelace, C.~R.~Nappi and S.~A.~Yost,

5309: ``Adding holes and crosscaps to the superstring'',

5310: Nucl.\ Phys.\ B {\bf 293} 83 (1987).

5311: %%CITATION = NUPHA,B293,83;%%

5312: 

5313: \bibitem{PC}

5314: J.~Polchinski and Y.~Cai,

5315: ``Consistency of open superstring theories'',

5316: Nucl.\ Phys.\ B {\bf 296} 91 (1988).

5317: %%CITATION = NUPHA,B296,91;%%

5318: 

5319: \bibitem{GP}

5320: E.~G.~Gimon and J.~Polchinski,

5321: ``Consistency conditions for orientifolds and D-manifolds'',

5322: Phys.\ Rev.\ D {\bf 54} 1667 (1996)

5323: [arXiv:hep-th/9601038].

5324: 

5325: %\cite{Pradisi:1988xd}

5326: \bibitem{PS-GP}

5327: G.~Pradisi and A.~Sagnotti,

5328: ``Open string orbifolds'',

5329: Phys.\ Lett.\ B {\bf 216}, 59 (1989).

5330: %%CITATION = PHLTA,B216,59;%%

5331: 

5332: \bibitem{torus}

5333: 

5334: C.~Vafa and E.~Witten,

5335: ``Dual string pairs with N = 1 and N = 2 supersymmetry in four  dimensions'',

5336: Nucl.\ Phys.\ Proc.\ Suppl.\  {\bf 46}, 225 (1996)

5337: [arXiv:hep-th/9507050].

5338: %%CITATION = HEP-TH 9507050;%%

5339: 

5340: C.~Angelantonj, M.~Bianchi, G.~Pradisi, A.~Sagnotti and Y.~S.~Stanev,

5341: ``Chiral asymmetry in four-dimensional open- string vacua'',

5342: Phys.\ Lett.\ B {\bf 385}, 96 (1996)

5343: [arXiv:hep-th/9606169].

5344: %%CITATION = HEP-TH 9606169;%%

5345: 

5346: M.~Berkooz and R.~G.~Leigh,

5347: ``A D = 4 N = 1 orbifold of type I strings'',

5348: Nucl.\ Phys.\ B {\bf 483}, 187 (1997)

5349: [arXiv:hep-th/9605049].

5350: %%CITATION = HEP-TH 9605049;%%

5351: 

5352: Z.~Kakushadze and G.~Shiu,

5353: ``A chiral N = 1 type I vacuum in four dimensions and its heterotic dual'',

5354: Phys.\ Rev.\ D {\bf 56}, 3686 (1997)

5355: [arXiv:hep-th/9705163].

5356: %%CITATION = HEP-TH 9705163;%%

5357: 

5358: Z.~Kakushadze and G.~Shiu,

5359: ``4D chiral N = 1 type I vacua with and without D5-branes'',

5360: Nucl.\ Phys.\ B {\bf 520}, 75 (1998)

5361: [arXiv:hep-th/9706051].

5362: %%CITATION = HEP-TH 9706051;%%

5363: 

5364: G.~Zwart,

5365: ``Four-dimensional N = 1 Z(N) x Z(M) orientifolds'',

5366: Nucl.\ Phys.\ B {\bf 526}, 378 (1998)

5367: [arXiv:hep-th/9708040].

5368: %%CITATION = HEP-TH 9708040;%%

5369: 

5370: 

5371: G.~Aldazabal, A.~Font, L.~E.~Ibanez and G.~Violero,

5372: ``D = 4, N = 1, type IIB orientifolds'',

5373: Nucl.\ Phys.\ B {\bf 536}, 29 (1998)

5374: [arXiv:hep-th/9804026].

5375: %%CITATION = HEP-TH 9804026;%%

5376: 

5377: R.~Blumenhagen, L.~G\"orlich and B.~K\"ors,

5378: ``Supersymmetric 4D orientifolds of type IIA with D6-branes at angles'',

5379: JHEP {\bf 0001}, 040 (2000)

5380: [arXiv:hep-th/9912204].

5381: %%CITATION = HEP-TH 9912204;%%

5382: 

5383: R.~Blumenhagen, L.~G\"orlich, B.~K\"ors and D.~L\"ust,

5384: ``Noncommutative compactifications of type I strings on tori with  magnetic background flux'',

5385: JHEP {\bf 0010}, 006 (2000)

5386: [arXiv:hep-th/0007024].

5387: %%CITATION = HEP-TH 0007024;%%

5388: 

5389: S.~F\"orste, G.~Honecker and R.~Schreyer,

5390: ``Supersymmetric Z(N) x Z(M) orientifolds in 4D with D-branes at angles'',

5391: Nucl.\ Phys.\ B {\bf 593}, 127 (2001)

5392: [arXiv:hep-th/0008250].

5393: %%CITATION = HEP-TH 0008250;%%

5394: 

5395: M.~Cvetic, A.~M.~Uranga and J.~Wang,

5396: ``Discrete Wilson lines in N = 1 D = 4 type IIB orientifolds: A  systematic exploration for Z(6) orientifold'',

5397: Nucl.\ Phys.\ B {\bf 595}, 63 (2001)

5398: [arXiv:hep-th/0010091].

5399: %%CITATION = HEP-TH 0010091;%%

5400: 

5401: 

5402: M.~Cvetic, G.~Shiu and A.~M.~Uranga,

5403: ``Chiral four-dimensional N = 1 supersymmetric type IIA orientifolds from  intersecting D6-branes'',

5404: Nucl.\ Phys.\ B {\bf 615}, 3 (2001)

5405: [arXiv:hep-th/0107166].

5406: %%CITATION = HEP-TH 0107166;%%

5407: 

5408: S.~B.~Giddings, S.~Kachru and J.~Polchinski,

5409: ``Hierarchies from fluxes in string compactifications'',

5410: [arXiv:hep-th/0105097].

5411: %%CITATION = HEP-TH 0105097;%%

5412: 

5413: S.~Kachru, M.~B.~Schulz and S.~Trivedi,

5414: ``Moduli stabilization from fluxes in a simple IIB orientifold'',

5415: [arXiv:hep-th/0201028].

5416: %%CITATION = HEP-TH 0201028;%%

5417: 

5418: 

5419: 

5420: %\cite{Blumenhagen:1998tj}

5421: \bibitem{BlWi}

5422: R.~Blumenhagen and A.~Wisskirchen,

5423: ``Spectra of 4D, N = 1 type I string vacua on non-toroidal CY threefolds'',

5424: Phys.\ Lett.\ B {\bf 438}, 52 (1998)

5425: [arXiv:hep-th/9806131].

5426: %%CITATION = HEP-TH 9806131;%%

5427: 

5428: 

5429: %\cite{Angelantonj:1996mw}

5430: \bibitem{ABPSS}

5431: C.~Angelantonj, M.~Bianchi, G.~Pradisi, A.~Sagnotti and Y.~S.~Stanev,

5432: ``Comments on Gepner models and type I vacua in string theory'',

5433: Phys.\ Lett.\ B {\bf 387}, 743 (1996)

5434: [arXiv:hep-th/9607229].

5435: %%CITATION = HEP-TH 9607229;%%

5436: 

5437: \bibitem{BH2} I.~Brunner and K.~Hori, to appear.

5438: 

5439: \bibitem{PSS}

5440: G.~Pradisi, A.~Sagnotti and Y.~S.~Stanev,

5441: ``Planar duality in SU(2) WZW models'',

5442: Phys.\ Lett.\ B {\bf 354} 279 (1995) 

5443: [arXiv:hep-th/9503207].

5444: 

5445: %\cite{Huiszoon:1999xq}

5446: \bibitem{HSS}

5447: L.~R.~Huiszoon, A.~N.~Schellekens and N.~Sousa,

5448: ``Klein bottles and simple currents'',

5449: Phys.\ Lett.\ B {\bf 470}, 95 (1999)

5450: [arXiv:hep-th/9909114].

5451: %%CITATION = HEP-TH 9909114;%%

5452: 

5453: %\cite{Fuchs:2000cm}

5454: \bibitem{FHSSW}

5455: J.~Fuchs, L.~R.~Huiszoon, A.~N.~Schellekens, C.~Schweigert and J.~Walcher,

5456: ``Boundaries, crosscaps and simple currents'',

5457: Phys.\ Lett.\ B {\bf 495}, 427 (2000)

5458: [arXiv:hep-th/0007174].

5459: %%CITATION = HEP-TH 0007174;%%

5460: 

5461: 

5462: %\cite{Huiszoon:2000ge}

5463: \bibitem{HuSc}

5464: L.~R.~Huiszoon and A.~N.~Schellekens,

5465: ``Crosscaps, boundaries and T-duality'',

5466: Nucl.\ Phys.\ B {\bf 584}, 705 (2000)

5467: [arXiv:hep-th/0004100].

5468: %%CITATION = HEP-TH 0004100;%%

5469: 

5470: 

5471: %\cite{Huiszoon:1999jw}

5472: \bibitem{HSSI}

5473: L.~R.~Huiszoon, A.~N.~Schellekens and N.~Sousa,

5474: ``Open descendants of non-diagonal invariants'',

5475: Nucl.\ Phys.\ B {\bf 575}, 401 (2000)

5476: [arXiv:hep-th/9911229].

5477: %%CITATION = HEP-TH 9911229;%%

5478: 

5479: %\cite{Ooguri:1996ck}

5480: \bibitem{OOY}

5481: H.~Ooguri, Y.~Oz and Z.~Yin,

5482: ``D-branes on Calabi-Yau spaces and their mirrors'',

5483: Nucl.\ Phys.\ B {\bf 477}, 407 (1996)

5484: [arXiv:hep-th/9606112].

5485: %%CITATION = HEP-TH 9606112;%%

5486: 

5487: 

5488: \bibitem{MMS}

5489: J.~Maldacena, G.~W.~Moore and N.~Seiberg,

5490: ``Geometrical interpretation of D-branes in gauged WZW models'',

5491: JHEP {\bf 0107} 046 (2001) 

5492: [arXiv:hep-th/0105038].

5493: 

5494: %\cite{Hikida:2002ws}

5495: \bibitem{hikida}

5496: Y.~Hikida,

5497: ``Orientifolds of SU(2)/U(1) WZW models'',

5498: [arXiv:hep-th/0201175].

5499: %%CITATION = HEP-TH 0201175;%%

5500: 

5501: \bibitem{Cardy}

5502: J.~L.~Cardy,

5503: ``Boundary conditions, fusion rules And the Verlinde formula'',

5504: Nucl.\ Phys.\ B {\bf 324} 581 (1989).

5505: %%CITATION = NUPHA,B324,581;%%

5506: 

5507: 

5508: %\cite{Ishibashi:1988kg}

5509: \bibitem{Ishibashi}

5510: N.~Ishibashi,

5511: ``The boundary and crosscap states in conformal field theories'',

5512: Mod.\ Phys.\ Lett.\ A {\bf 4}, 251 (1989).

5513: %%CITATION = MPLAE,A4,251;%%

5514: 

5515: %\cite{Pradisi:1995pp}

5516: \bibitem{PSSII}

5517: G.~Pradisi, A.~Sagnotti and Y.~S.~Stanev,

5518: ``The Open descendants of nondiagonal SU(2) WZW models'',

5519: Phys.\ Lett.\ B {\bf 356}, 230 (1995)

5520: [arXiv:hep-th/9506014].

5521: %%CITATION = HEP-TH 9506014;%%

5522: 

5523: \bibitem{bantay}

5524: P.~Bantay,

5525: ``The Frobenius-Schur indicator in conformal field theory'',

5526: Phys.\ Lett.\ B {\bf 394},  89 (1997) 

5527: [arXiv:hep-th/9610192].

5528: 

5529: \bibitem{bantay2}

5530: P.~Bantay,

5531: ``Frobenius-Schur indicators, the Klein-bottle amplitude,

5532: and the  principle of orbifold covariance'',

5533: Phys.\ Lett.\ B {\bf 488} 207 (2000) 

5534: [arXiv:hep-th/0001173].

5535: 

5536: %\cite{Fuchs:1997fu}

5537: \bibitem{FSbranes}

5538: J.~Fuchs and C.~Schweigert,

5539: ``Branes: from free fields to general backgrounds'',

5540: Nucl.\ Phys.\ B {\bf 530}, 99 (1998)

5541: [arXiv:hep-th/9712257].

5542: %%CITATION = HEP-TH 9712257;%%

5543: 

5544: 

5545: \bibitem{verlinde}

5546: E.~Verlinde,

5547: ``Fusion rules and modular transformations in 2-D conformal field theory'',

5548: Nucl.\ Phys.\ B {\bf 300}, 360 (1988).

5549: 

5550: %\cite{Fuchs:1999zi}

5551: \bibitem{FSbreakI}

5552: J.~Fuchs and C.~Schweigert,

5553: ``Symmetry breaking boundaries. I: General theory'',

5554: Nucl.\ Phys.\ B {\bf 558}, 419 (1999)

5555: [arXiv:hep-th/9902132].

5556: %%CITATION = HEP-TH 9902132;%%

5557: 

5558: %\cite{Fuchs:1999xn}

5559: \bibitem{FSbreakII}

5560: J.~Fuchs and C.~Schweigert,

5561: ``Symmetry breaking boundaries. II: More structures, examples'',

5562: Nucl.\ Phys.\ B {\bf 568}, 543 (2000)

5563: [arXiv:hep-th/9908025].

5564: %%CITATION = HEP-TH 9908025;%%

5565: 

5566: \bibitem{BFS}

5567: L.~Birke, J.~Fuchs and C.~Schweigert,

5568: ``Symmetry breaking boundary conditions and WZW orbifolds'',

5569: Adv.\ Theor.\ Math.\ Phys.\  {\bf 3}, 671 (1999)

5570: [arXiv:hep-th/9905038].

5571: %%CITATION = HEP-TH 9905038;%%

5572: 

5573: 

5574: \bibitem{Vafa:DT}

5575: C.~Vafa,

5576: ``Modular invariance and discrete torsion on orbifolds'',

5577: Nucl.\ Phys.\ B {\bf 273} 592 (1986).

5578: 

5579: \bibitem{GCbook}

5580: G.~Karpilovsky,

5581: {\it Projective Representations of Finite Groups},

5582: Monographs and textbooks in pure and applied mathematics 94

5583: (Marcel Dekker, 1985).

5584: 

5585: \bibitem{GSW} 

5586: M.~Green, J.~Schwarz and E.~Witten,

5587: {\it Superstring theory}, 

5588: Cambridge University Press,1987.

5589: 

5590: \bibitem{SS}

5591: A.~Sen and S.~Sethi,

5592: ``The mirror transform of type I vacua in six dimensions'',

5593: Nucl.\ Phys.\ B {\bf 499}, 45 (1997)

5594: [arXiv:hep-th/9703157].

5595: %%CITATION = HEP-TH 9703157;%%

5596: 

5597: %\cite{AAADS}

5598: \bibitem{AAADS}

5599: C.~Angelantonj, I.~Antoniadis, G.~D'Appollonio, E.~Dudas and A.~Sagnotti,

5600: ``Type I vacua with brane supersymmetry breaking'',

5601: Nucl.\ Phys.\ B {\bf 572}, 36 (2000)

5602: [arXiv:hep-th/9911081].

5603: %%CITATION = HEP-TH 9911081;%%

5604: %\cite{Klein:2000tf}

5605: 

5606: \bibitem{KlRa}

5607: M.~Klein and R.~Rabadan,

5608: ``Orientifolds with discrete torsion'',

5609: JHEP {\bf 0007}, 040 (2000)

5610: [arXiv:hep-th/0002103].

5611: %%CITATION = HEP-TH 0002103;%%

5612: 

5613: 

5614: 

5615: %\cite{Recknagel:1997sb}

5616: \bibitem{RSGep}

5617: A.~Recknagel and V.~Schomerus,

5618: ``D-branes in Gepner models'',

5619: Nucl.\ Phys.\ B {\bf 531}, 185 (1998)

5620: [arXiv:hep-th/9712186].

5621: %%CITATION = HEP-TH 9712186;%%

5622: 

5623: %\cite{Acharya:2002ag}

5624: \bibitem{AAHV}

5625: B.~Acharya, M.~Aganagic, K.~Hori and C.~Vafa,

5626: ``Orientifolds, mirror symmetry and superpotentials'',

5627: arXiv:hep-th/0202208.

5628: %%CITATION = HEP-TH 0202208;%%

5629: 

5630: \bibitem{nvs}

5631: E.~Witten,

5632: ``Toroidal compactification without vector structure'',

5633: JHEP {\bf 9802} 006 (1998) 

5634: [arXiv:hep-th/9712028].

5635: 

5636: \bibitem{Wnab}

5637: E.~Witten,

5638: ``Nonabelian bosonization in two dimensions'',

5639: Commun.\ Math.\ Phys.\  {\bf 92}, 455 (1984).

5640: %%CITATION = CMPHA,92,455;%%

5641: 

5642: \bibitem{PW}

5643: A.~M.~Polyakov and P.~B.~Wiegmann,

5644: ``Goldstone fields in two-dimensions with multivalued actions'',

5645: Phys.\ Lett.\ B {\bf 141}, 223 (1984).

5646: %%CITATION = PHLTA,B141,223;%%

5647: 

5648: \bibitem{Gaw}

5649: K.~Gawedzki and A.~Kupiainen,

5650: ``Coset construction from functional integrals'',

5651: Nucl.\ Phys.\ B {\bf 320}, 625  (1989).

5652: %%CITATION = NUPHA,B320,625;%%

5653: 

5654: \bibitem{FIH}

5655: K.~Hori,

5656: ``Global aspects of gauged Wess-Zumino-Witten models'',

5657: Commun.\ Math.\ Phys.\  {\bf 182} (1996) 1

5658: [arXiv:hep-th/9411134].

5659: 

5660: 

5661: 

5662: 

5663: %\cite{Horava:1994ee}

5664: \bibitem{HOcs}

5665: P.~Horava,

5666: ``Chern-Simons gauge theory on orbifolds: Open strings from  three dimensions'',

5667: J.\ Geom.\ Phys.\  {\bf 21}, 1 (1996)

5668: [arXiv:hep-th/9404101].

5669: %%CITATION = HEP-TH 9404101;%%

5670: 

5671: %\cite{Bianchi:1997gt}

5672: \bibitem{BiSt}

5673: M.~Bianchi and Y.~S.~Stanev,

5674: ``Open strings on the Neveu-Schwarz pentabrane'',

5675: Nucl.\ Phys.\ B {\bf 523}, 193 (1998)

5676: [arXiv:hep-th/9711069].

5677: %%CITATION = HEP-TH 9711069;%%

5678: 

5679: %\cite{Alekseev:1998mc}

5680: \bibitem{AlSc}

5681: A.~Y.~Alekseev and V.~Schomerus,

5682: ``D-branes in the WZW model'',

5683: Phys.\ Rev.\ D {\bf 60}, 061901 (1999)

5684: [arXiv:hep-th/9812193].

5685: %%CITATION = HEP-TH 9812193;%%

5686: 

5687: %\cite{Alekseev:1999bs}

5688: \bibitem{ASR}

5689: A.~Y.~Alekseev, A.~Recknagel and V.~Schomerus,

5690: ``Non-commutative world-volume geometries: Branes on SU(2) and fuzzy  spheres'',

5691: JHEP {\bf 9909}, 023 (1999)

5692: [arXiv:hep-th/9908040].

5693: %%CITATION = HEP-TH 9908040;%%

5694: 

5695: \bibitem{FFFS}

5696: G.~Felder, J.~Fr\"ohlich, J.~Fuchs and C.~Schweigert,

5697: ``The geometry of WZW branes'',

5698: J.\ Geom.\ Phys.\  {\bf 34}, 162 (2000)

5699: [arXiv:hep-th/9909030].

5700: %%CITATION = HEP-TH 9909030;%%

5701: 

5702: %\cite{Brunner:2001fs}

5703: \bibitem{ib}

5704: I.~Brunner,

5705: ``On orientifolds of WZW models and their relation to geometry'',

5706: JHEP {\bf 0201}, 007 (2002)

5707: [arXiv:hep-th/0110219].

5708: %%CITATION = HEP-TH 0110219;%%

5709: 

5710: \bibitem{HSSc}

5711: L.~R.~Huiszoon, K.~Schalm and A.~N.~Schellekens,

5712: ``Geometry of WZW orientifolds'',

5713: Nucl.\ Phys.\ B {\bf 624}, 219 (2002)

5714: [arXiv:hep-th/0110267].

5715: %%CITATION = HEP-TH 0110267;%%ION = NUPHA,B327,673;%%

5716: 

5717: %\cite{Bachas:2001id}

5718: \bibitem{BCW}

5719: C.~Bachas, N.~Couchoud and P.~Windey,

5720: ``Orientifolds of the 3-sphere'',

5721: JHEP {\bf 0112}, 003 (2001)

5722: [arXiv:hep-th/0111002].

5723: %%CITATION = HEP-TH 0111002;%%

5724: 

5725: 

5726: 

5727: \bibitem{zuber}

5728: R.~E.~Behrend, P.~A.~Pearce, V.~B.~Petkova and J.~B.~Zuber,

5729: ``Boundary conditions in rational conformal field theories'',

5730: Nucl.\ Phys.\ B {\bf 570} (2000) 525

5731: [Nucl.\ Phys.\ B {\bf 579} (2000) 707]

5732: [arXiv:hep-th/9908036].

5733: %%CITATION = HEP-TH 9908036;%%

5734: 

5735: %\cite{Dijkgraaf:1988tf}

5736: \bibitem{DV}

5737: R.~Dijkgraaf and E.~Verlinde,

5738: ``Modular invariance and the fusion algebra'',

5739: Nucl.\ Phys.\ B (Proc. Suppl.) {\bf 5B}, 87 (1988),

5740: 

5741: \bibitem{ScYa}

5742: A.~N.~Schellekens and S.~Yankielowicz,

5743: ``Extended chiral algebras and modular invariant partition functions'',

5744: Nucl.\ Phys.\ B {\bf 327} (1989) 673.

5745: %%CITAT

5746: 

5747: 

5748: 

5749: %\cite{Schellekens:1989dq}

5750: \bibitem{SYQ}

5751: A.~N.~Schellekens and S.~Yankielowicz,

5752: ``Modular invariants from simple currents: an explicit proof'',

5753: Phys.\ Lett.\ B {\bf 227}, 387 (1989).

5754: %%CITATION = PHLTA,B227,387;%%

5755: 

5756: %\cite{Intriligator:1989zw}

5757: \bibitem{Intri}

5758: K.~A.~Intriligator,

5759: ``Bonus symmetry in conformal field theory'',

5760: Nucl.\ Phys.\ B {\bf 332}, 541 (1990).

5761: %%CITATION = NUPHA,B332,541;%%

5762: 

5763: %\cite{Kreuzer:tf}

5764: \bibitem{KrSc}

5765: M.~Kreuzer and A.~N.~Schellekens,

5766: ``Simple currents versus orbifolds with discrete torsion: a complete classification'',

5767: Nucl.\ Phys.\ B {\bf 411}, 97 (1994)

5768: [arXiv:hep-th/9306145].

5769: %%CITATION = HEP-TH 9306145;%%

5770: 

5771: 

5772: 

5773: 

5774: \end{thebibliography}

5775: 

5776: 

5777: \end{document}

5778: 

5779: 

5780: 

5781: 

5782: 

5783: 

5784: