1: \documentclass[paper, 12pt, letterpaper]{JHEP} 

2: \usepackage{amsmath}

3: \usepackage{amssymb}

4: 

5: %1. Math symbols:

6: 

7: %1.1 AMS fonts:

8: 

9: \def\Bbb{\mathbb} \def\C{{\Bbb C}} \def\R{{\Bbb R}} \def\Z{{\Bbb Z}}

10: \def\H{{\Bbb H}} \def\Q{{\Bbb Q}} \def\P{{\Bbb P}} \def\ii{{\Bbb

11:     i}}\def\j{{\Bbb j}}\def\k{{\Bbb k}} \def\I{{\Bbb I}}\def\J{{\Bbb

12:     J}}\def\K{{\Bbb K}} \def\boldphi{\mbox{\boldmath $\phi$}}

13: \def\boldpsi{\mbox{\boldmath $\psi$}} \def\boldzeta{\mbox{\boldmath

14:     $\zeta$}}

15: 

16: 

17: 

18: %1.2 double line:

19: 

20: \font\mybbb=msbm10 at 8pt \font\mybb=msbm10 at 12pt

21: \def\bbb#1{\hbox{\mybbb#1}} \def\bb#1{\hbox{\mybb#1}}

22: 

23: \def\pRe{\bbb{R}} \def\Re {\bb{Re}} \def\S {\bb{S}} \def\T {\bb{T}}

24: %\def\P{\bb{P}}

25: %\def\Z {\bb{Z}}

26: 

27: 

28: 

29: %1.3 operatorname:

30: 

31: \def\Hom{\operatorname{Hom}} \def\Tors{\operatorname{Tors}}

32: \def\Ker{\operatorname{Ker}} \def\Spec{\operatorname{Spec}}

33: \def\Area{\operatorname{Area}} \def\Vol{{\rm Vol}} \def\vol {{\rm

34:     vol}} \def\ad{\operatorname{ad}} \def\tr{\operatorname{tr}}\def\Tr{\operatorname{Tr}}

35: \def\str{\operatorname{str}}

36: \def\Str{\operatorname{Str}}

37: \def\Pic{\operatorname{Pic}} \def\disc{\operatorname{disc}}

38: \def\cpl{\operatorname{cpl}} \def\Img{\operatorname{Im}}

39: \def\Rea{\operatorname{Re}} \def\Gr{\operatorname{Gr}}

40: \def\SO{\operatorname{SO}} \def\Sl{\operatorname{SL}}

41: \def\GO{\operatorname{O{}}} \def\SU{{\rm SU}}

42: \def\GU{\operatorname{U{}}} \def\Sp{\operatorname{Sp}}

43: \def\Spin{\operatorname{Spin}} \def\rank{\operatorname{rank}}

44: \def\Aff{\operatorname{Aff}} \def\diag{\operatorname{diag}}

45: \def\su{\rm{su(2)}}

46: 

47: 

48: %1.4 roman:

49: 

50: \def\ker{{\rm ker}} \def\coker{{\rm coker}} \def\rank{{\rm rank}}

51: \def\im{{\rm im\,}} \def\Im{{\rm Im\,}} \def\re{{\rm re\,}}

52: \def\Re{{\rm Re\,}} \def\tr{{\rm tr\, }} \def\dim{{\rm dim}}

53: \def\codim{{\rm codim}} \def\Tr{{\rm Tr\, }} \def\d{{\rm d }}

54: \def\Card{{\rm Card}} \def\li{{\rm linearly independent}} \def\ld{{\rm

55:     linearly dependent}} \def\deg{{\rm deg}} \def\rk{{\rm rk}} \def\det{{\rm det}}

56: \def\Div{{\rm Div}} \def\supp{{\rm supp}} \def\Gr{{\rm Gr}}

57: \def\End{{\rm End}} \def\Aut{{\rm Aut}}

58: \def\Spec{{\rm Spec}}

59: 

60: %1.5. script:

61: 

62: \def\cR{{\Scr R}} \def\cM{{\Scr M}} \def\cA{{\Scr A}} \def\cB{{\Scr

63:     B}} \def\cK{{\Scr K}} \def\cD{{\Scr D}} \def\cH{{\Scr H}}

64: \def\cT{{\Scr T}} \def\cL{{\Scr L}} \def\cF{{\Scr F}}

65: 

66: %1.6 gothic:

67: 

68: \font\gothics=ygoth at 10pt \font\gothicl=ygoth at 12pt

69: \def\gg#1{\hbox{\gothics#1}} \def\gG{\hbox{\gothicl G}} \def

70: \g{\hbox{\gothics g}}

71: 

72: %1.7 misc:

73: 

74: \def\id{\protect{{1 \kern-.28em {\rm l}}}}

75: 

76: 

77: %3. New commands

78: 

79: \newcommand{\be}{\begin{equation}} \newcommand{\ee}{\end{equation}}

80: \newcommand{\bea}{\begin{eqnarray}} \newcommand{\eea}{\end{eqnarray}}

81: \newcommand{\beann}{\begin{eqnarray*}}

82:   \newcommand{\eeann}{\end{eqnarray*}}

83: \newcommand{\bfig}{\begin{figure}} \newcommand{\efig}{\end{figure}}

84: \newcommand{\nn}{\nonumber}

85: \newcommand{\ba}{\begin{array}}\newcommand{\ea}{\end{array}}

86: 

87: 

88: %4. New theorems and their commands

89: 

90: \newtheorem{Proposition}{Proposition}[section]

91: \newtheorem{Definition}{Definition}[section]

92: \newtheorem{Theorem}{Theorem}[section]

93: \newtheorem{Lemma}{Lemma}[section]

94: \newtheorem{Corrolary}{Corrolary}[section]

95: 

96: \newcommand{\bp}{\begin{Proposition}}

97:   \newcommand{\ep}{\end{Proposition}}

98: \newcommand{\bt}{\begin{Theorem}} \newcommand{\et}{\end{Theorem}}

99: \newcommand{\bl}{\begin{Lemma}} \newcommand{\el}{\end{Lemma}}

100: \newcommand{\bc}{\begin{Corrolary}} \newcommand{\ec}{\end{Corrolary}}

101: 

102: 

103: %5. Karl

104: 

105: \def\omegat{\tilde{\omega}} \def\ot{\tilde{\omega}} \def\om{\omega}

106: \def\Qt{\tilde{Q}} \def\Vt{\tilde{V}} \def\Rt{\tilde{R}} \def\ep{\eps}

107: \def\CC{{\cal C}} \def\hn{{ \hat N}} \def\hq{{\hat Q}} \def\hp{{\hat

108:     \Phi}} \def\hs{{ \hat S}} \def\ha{{\hat A}} \def\cs{{\cal S}}

109: 

110: \author{

111: Manfred Herbst${}^a$, Calin-Iuliu Lazaroiu${}^b$ \\}

112: \author{~\\

113:      ${}^a$Department of Physics, CERN\\

114:      Theory Division\\

115:      CH-1211 Geneva 23\\

116:      Switzerland\\

117:      Manfred.Herbst@cern.ch\\{~}\\

118:      ${}^b$5243 Chamberlin Hall\\

119:      University of Wisconsin at Madison\\

120:      1150 University Ave\\

121:      Madison, Wisconsin 53706, USA\\

122:      calin@physics.wisc.edu\\

123:     }

124: 

125: 

126: \title{Localization and traces in open--closed topological Landau--Ginzburg models}

127: 

128: 

129: \abstract{We reconsider the issue of localization in open-closed B-twisted

130: Landau-Ginzburg models with arbitrary Calabi-Yau target. Through careful

131: analsysis of zero-mode reduction, we show that the closed model allows for a

132: one-parameter family of localization pictures, which generalize the standard

133: residue representation. The parameter $\lambda$ which indexes these pictures

134: measures the area of worldsheets with $S^2$ topology, with the residue

135: representation obtained in the limit of small area. In the boundary sector, we

136: find a double family of such pictures, depending on parameters $\lambda$ and

137: $\mu$ which measure the area and boundary length of worldsheets with disk

138: topology. We show that setting $\mu=0$ and varying $\lambda$ interpolates

139: between the localization picture of the B-model with a noncompact target space

140: and a certain residue representation proposed recently. This gives a complete

141: derivation of the boundary residue formula, starting from the explicit construction of the

142: boundary coupling. We also show that the various localization pictures are

143: related by a semigroup of homotopy equivalences.}

144: 

145: 

146: \preprint{MAD-TH-04-3\\  

147:           CERN--PH--TH/2004-070}

148: 

149: 

150: \begin{document}

151: 

152: 

153: \tableofcontents

154: 

155: \pagebreak

156: 

157: \vskip .6in

158: 

159: \section{Introduction}

160: \label{intro}

161: 

162: 

163: Closed topological Landau-Ginzburg models \cite{Vafa_LG} have been a useful

164: testing ground for string theory. They make direct contact with

165: the topological sector of rational conformal field theories 

166: through the Landau-Ginzburg approach to minimal models, and arise as phases of

167: $N=2$ string compactifications \cite{Witten_phases}.

168: Moreover, they give explicit realizations of the 

169: WDVV equations and examples of Frobenius manifolds, and have interesting relations with 

170: singularity theory.

171: 

172: 

173: In a similar vein, one can expect to learn important lessons about open

174: string theory by studying topological Landau-Ginzburg models in the presence

175: of D-branes (see \cite{us, fractional} for some recent results in this direction). 

176: While basic D-brane constructions were considered by many authors

177: (see \cite{Hori} and references therein), a systematic study has been

178: hampered by the lack of a reasonably general description of the boundary

179: coupling (the "Warner problem" \cite{Warner}).

180: 

181: Progress in removing this obstacle was made recently in  \cite{Kontsevich, Kap1,  Lerche,

182:   Kap2, Kap3} (see also \cite{Orlov}). These papers proposed a solution of the Warner problem for

183:   B-twisted Landau-Ginzburg models with target space $\C^n$ and for particular 

184: families

185:   of D-branes described by superbundles whose rank is constrained to be a

186:   power of two. In a slightly modified form, this 

187:   solution was generalized in \cite{coupling} by removing unnecessary 

188:   assumptions, thus giving the general form of the

189:   relevant boundary coupling. 

190: 

191: 

192: As in the closed string case, it is natural to translate the

193: physical data of open Landau-Ginzburg models into the language of 

194: singularity theory. For the closed string sector, a crucial step in this regard is

195: the localization formula of \cite{Vafa_LG}, which relates topological field

196: theory correlators to residues (see, for

197: example, \cite{Griffiths}). An open string version of this formula was  proposed in \cite{Kap2}, 

198: though a complete derivation based on the microscopic boundary coupling was not given. 

199: 

200: Given the boundary coupling constructed in \cite{coupling}, we shall re-consider this issue in the 

201: more general set-up of \cite{Labastida}, and give a complete derivation of

202: this localization formula. 

203: Another purpose of this paper is to extend the open and closed localization

204: formulae in a manner which reflects the basic intuition 

205: \cite{coupling} that the B-branes of Landau-Ginzburg models are the

206: result of tachyon condensation between the elementary branes of the B-type

207: sigma model, with tachyon condensation driven by the Landau-Ginzburg superpotential. 

208: As we shall show somewhere else, this allows one to make contact with the

209: string field theory approach advocated in a different context in 

210: \cite{Witten_CS, CIL2, CIL3, CIL4, CIL5, CIL6, CIL7, CIL8, CIL9, CIL10, Diaconescu}.

211: 

212: Perhaps surprisingly, this is non-trivial to

213: achieve already for the closed string sector. Indeed, the usual on-shell descriptions

214: of the space of bulk  observables differ markedly

215: between the two models. In the B-twisted sigma model, this space is described

216: as the ${\bar \partial}$-cohomology of the algebra of $(0,p)$-forms on the

217: target space, valued in holomorphic polyvector fields. In the Landau-Ginzburg

218: case, the space of on-shell bulk observables can be identified with the Jacobi ring

219: of the Landau-Ginzburg superpotential. As we shall see below, these two

220: descriptions are related in a subtle manner, namely by a one-parameter

221: family of "localization pictures" which interpolates between them. The

222: existence of such a family will be established by refining 

223: the localization argument of \cite{Vafa_LG}. 

224: The different pictures are indexed by a parameter $\lambda$, 

225: which roughly measures the area of 

226: worldsheets with $S^2$ topology.

227: The B-model description of the algebra of observables

228: arises in the limit when the worldsheet is collapsed to a point, while the 

229: Jacobi realization is recovered for very large areas. More

230: precisely, one can construct an off-shell model for each localization 

231: picture, with a reduced BRST operator whose cohomology reproduces the space

232: of on-shell bulk observables. The various pictures 

233: are related by a "homotopy flow", i.e. a one-parameter semigroup of operators 

234: homotopic to the identity. 

235: This flow induces the trivial action on BRST cohomology, thus identifying 

236: on-shell data between different pictures. 

237: 

238: Extending this construction to the boundary sector, we find a similar

239: description. Namely, we shall 

240: construct a family of localization pictures indexed by {\em two}

241: parameters $\lambda$ and $\mu$, which -- when real -- measure the area of a worldsheet with disk

242: topology and the length of its boundary. 

243: Taking $\mu=0$, the standard realization of observables in the open B-model

244: arises  for $\lambda\rightarrow 0$, while the LG description and residue formula 

245: of \cite{Kap2} are obtained in the opposite limit $\lambda\rightarrow +\infty$. The parameter

246: $\mu$ plays an auxiliary role, related to a certain boundary term which was

247: not included in \cite{coupling} since it is not essential for the

248: topological model. Contrary to previous proposals, we show that this

249: parameter can be safely set to zero, without affecting the localization data. 

250: Physically, localization on the critical set of $W$ is controlled by the 

251: bulk parameter $\lambda$, and is completely independent of the choice of

252: $\mu$. In fact, one must set $\mu=0$ in order to recover

253: \footnote{The authors of \cite{Kap2} propose a different limit, namely

254: $\mu\rightarrow +\infty$ with $\lambda=0$. It does not seem possible to achieve

255: localization on the critical set of $W$ in this limit.} the residue representation of 

256: \cite{Kap2}.

257: 

258: 

259: The paper is organized as follows. In Section 2, we review the bulk Lagrangian

260: and some of its basic properties, following \cite{Labastida}. In Section 3, we

261: discuss localization in the bulk sector. Through careful 

262: analysis, we show that one can localize on the

263: zero modes of the {\em sigma model} action, namely that part of the bulk action

264: which is independent of the Landau-Ginzburg superpotential $W$. This gives the

265: one-parameter family of localization pictures. We also give 

266: the geometric realization of these pictures, and the homotopy flow connecting them. 

267: In Section 5, we recall the boundary coupling given in  \cite{coupling} and

268: adapt it by adding a supplementary term

269: also suggested in \cite{Kap1, Lerche, Kap2} for a special case. This assures that

270: the coupling preserves a full copy of the $N=2$ topological algebra, when the

271: model is considered on a flat strip. Since the second generator of this

272: algebra is gauged when considering the model on a curved Riemann surface, this

273: condition does not play a fundamental role for unintegrated amplitudes, 

274: but the modified coupling is useful for comparison with \cite{Kap2}.

275: Section 5 constructs the boundary observables and correlators, and explains

276: how our model for the boundary  BRST operator arises in this approach. In Section 6, we

277: discuss localization in the boundary sector. As for the bulk, we proceed by

278: reducing to zero modes of the sigma model action. This gives a family 

279: of representations depending on two parameters $\lambda$ and $\mu$.

280: The second of these weights the contribution arising from the supplementary

281: boundary term. After describing boundary homotopy flows and

282: the associated geometric realization, we construct an off-shell representative

283: for the bulk-boundary map of \cite{CIL1} (see also \cite{Moore_Segal, Moore}), 

284: and use it to recover (an extension

285: of) the localization formula proposed in \cite{Kap2}, by taking the limit

286: $\lambda\rightarrow +\infty$ with $\mu=0$\footnote{In this limit, the

287:   supplementary term introduced in Section 5 does not contribute, so 

288: one can use the simplified boundary coupling of \cite{coupling}.}. 

289: Section 7 presents our conclusions. The Appendix gives the boundary 

290: conditions for the general D-brane coupling. 

291: 

292: 

293: 

294: \section{The bulk action}

295: \label{bulk}

296: 

297: The general formulation of closed B-type topological Landau-Ginzburg models

298: was given in \cite{Labastida}, extending the work of

299: \cite{Vafa_LG}. We take the target space to be a Calabi-Yau manifold $X$,

300: with the Landau-Ginzburg potential a holomorphic function $W\in H^0({\cal

301: O}_X)$.  Since any holomorphic function on a compact complex manifold is

302: constant, we shall assume that $X$ is non-compact.  

303: In the on-shell formulation, 

304: the Grassmann even worldsheet fields are 

305: the components $\phi^i$, $\phi^{\bar i}$ of the map $\phi:\Sigma \rightarrow

306: X$, while the G-odd fields are sections $\eta$, $\theta$ and $\rho$ of the

307: bundles $\phi^*({\bar T}X)$, $\phi^*(T^*X)$ and $\phi^*(TX)\otimes {\cal

308: T}^*\Sigma$ over the worldsheet $\Sigma$. Here ${\cal T}^*\Sigma$ is the

309: complexified cotangent bundle to $\Sigma$, while $TX$ and ${\bar T}X$ are the

310: holomorphic and antiholomorphic components of the complexified tangent bundle

311: ${\cal T}X$ to $X$. This agrees with the on-shell field content of 

312: the B-twisted sigma model \cite{Witten_mirror}.

313: 

314: 

315: \subsection{Action and BRST transformations}

316: 

317: To write the bulk action, we introduce new fields $\chi, {\bar

318:   \chi} \in \Gamma(\Sigma, \phi^*({\bar T}X))$ by the relations:

319: \bea

320: \eta^{\bar i}&=&\chi^{\bar i}+{\bar \chi}^{\bar i}\\

321: \theta_i&=&G_{i{\bar j}}(\chi^{\bar j}-{\bar \chi}^{\bar j})~~.

322: \eea

323: We shall also use the quantity $\theta^{\bar i}=G^{{\bar i}j}\theta_j$.

324: 

325: As in \cite{Labastida}, it is convenient to use an off-shell

326: realization of the BRST symmetry. For

327: this, consider an auxiliary G-even field ${\tilde F}$ transforming as a section of

328: $\phi^*({\cal T}X)$. Then the BRST transformations are:

329: \bea

330: \label{BRST}

331: \delta \phi^i=0~~&,&~~\delta \phi^{\bar i}=\chi^{\bar i}+{\bar \chi}^{\bar

332:   i}=\eta^{\bar i}\nn\\

333: \delta \chi^{\bar i}={\tilde F}^{\bar i}-\Gamma^{\bar

334:   i}_{{\bar j}{\bar k}}{\bar \chi}^{\bar j}\chi^{\bar k}~~&,&~~\delta {\bar \chi}^{\bar

335:   i}=-{\tilde F}^{\bar i}+\Gamma^{\bar i}_{{\bar j}{\bar

336:   k}}{\bar \chi}^{\bar j}\chi^{\bar k}~~\nn\\

337: \delta \rho^i_\alpha&=&2\partial_\alpha \phi^i~~\\

338: \delta {\tilde F}^i=i\varepsilon^{\alpha\beta} \left[D_\alpha

339:   \rho^i_\beta+\frac{1}{4}R^i_{j{\bar l}k}(\chi^{\bar l}+{\bar \chi}^{\bar l})

340: \rho^j_\alpha\rho^k_\beta\right]~~&,&~~\delta

341: {\tilde F}^{\bar i}=\Gamma^{\bar i}_{{\bar j}{\bar k}}{\tilde F}^{\bar j}(\chi^{\bar k}+{\bar \chi}^{\bar k})~~.\nn

342: \eea

343: These transformations are independent of

344:   $W$. Moreover, the transformations of $\phi$, $\eta$ and $\rho$ 

345: do not involve the auxiliary fields. In particular, we have  $\delta \eta^{\bar i}=0$.

346: These observations will be used in Section 4. 

347: 

348: Let us pick a Riemannian metric $g$ on the  worldsheet. The bulk action

349: of \cite{Labastida} takes the form:

350: \be

351: \label{Sbulk_decomp}

352: S_{bulk}=S_B+S_W

353: \ee

354: where:

355: \bea

356: \label{S_B}

357: \!\!\!\!\!S_B&=&\int_\Sigma {d^2\sigma \sqrt{g}}{\Big[ G_{i{\bar

358: j}}\left(g^{\alpha\beta} \partial_\alpha \phi^i \partial_\beta \phi^{\bar j}

359: -i\varepsilon^{\alpha\beta}\partial_\alpha \phi^i\partial_\beta\phi^{\bar

360: j}-\frac{1}{2}g^{\alpha\beta}\rho_\alpha^i D_\beta \eta^{\bar

361: j}-\frac{i}{2}\varepsilon^{\alpha\beta} \rho_\alpha^i D_\beta \theta^{\bar

362: j}-{\tilde F}^i{\tilde F}^{\bar j}\right)}\nn\\

363: \!\!\!\!&+&\frac{i}{4}\varepsilon^{\alpha\beta}R_{i{\bar l}k{\bar

364: j}}\rho^i_\alpha{\bar \chi}^{\bar l}\rho^k_\beta\chi^{\bar j}\Big] 

365: \eea

366: is the action of the B-twisted sigma model and $S_W=S_0+S_1$ is the potential-dependent term, with:

367: \bea

368: S_0&=&-\frac{i}{2} \int_{\Sigma}{d^2\sigma \sqrt{g} \left[D_{\bar i}

369: \partial_{\bar j}{\bar W} \chi^{\bar i}{\bar \chi}^{\bar j}-(\partial_{\bar

370: i}{\bar W}){\tilde F}^{\bar i}\right]}\\

371: S_1&=&-\frac{i}{2}\int_{\Sigma}{d^2\sigma \sqrt{g} \left[(\partial_i W){\tilde

372: F}^i + \frac{i}{4} \varepsilon^{\alpha\beta}D_i\partial_jW

373: \rho_\alpha^i\rho_\beta^j\right]}~~.

374: \eea

375: The quantity $\varepsilon^{\alpha\beta}=\frac{\epsilon^{\alpha\beta}}{\sqrt{g}}$ is

376: the Levi-Civita tensor, while $\epsilon^{\alpha\beta}$ is the associated density. We

377: have rescaled the Landau-Ginzburg potential $W$ by a factor of $\frac{i}{2}$

378: with respect to the conventions of \cite{Labastida} (the conventions for the

379: target space Riemann tensor and covariantized worldsheet derivative $D_\alpha$

380: are unchanged).  In $S_W$, we separated the term depending on $W$ from that

381: depending on its complex conjugate.

382: 

383: As shown in \cite{Labastida}, the topological sigma model action 

384: (\ref{S_B}) is BRST exact on closed Riemann surfaces.  Since in this

385: paper we shall allow $\Sigma$ to have a nonempty boundary, we must be careful

386: with total derivative terms. Extending the computation of \cite{Labastida} to

387: this case, we find:

388: \be

389: \label{exactS_B}

390: S_B+s=\delta V_B

391: \ee

392: where:

393: \be

394: V_B:=\int_{\Sigma}{d^2\sigma \sqrt{g} G_{i{\bar

395:   j}}\left(\frac{1}{2}g^{\alpha\beta}\rho^i_\alpha\partial_\beta \phi^{\bar

396:   j}-\frac{i}{2}\varepsilon^{\alpha\beta}\rho^i_\alpha

397:   \partial_\beta\phi^{\bar j}-{\tilde F}^i\chi^{\bar j}\right)}

398: \ee

399: and:

400: \be

401: \label{s}

402: s:=i\int_{\Sigma}{d^2\sigma

403:       \sqrt{g}\varepsilon^{\alpha\beta}\partial_\alpha(G_{{\bar i}j}\chi^{\bar

404:       i}\rho^j_\beta)}=i\int_\Sigma{d(G_{{\bar i}j}\chi^{\bar i}\rho^j)}~~.

405: \ee

406: Since total derivative  terms do not change physics on

407: closed Riemann surfaces, we are free to redefine the bulk 

408: sigma-model action by adding (\ref{s}) to $S_B$:

409: \be

410: {\tilde S}_B:=S_B+s=\delta V_B~~.

411: \ee

412: Accordingly, we shall use the modified bulk Landau-Ginzburg action:

413: \be

414: \label{tilde_S_bulk}

415: {\tilde S}_{bulk}=S_{bulk}+s={\tilde S}_B+S_0+S_1~~.

416: \ee

417: It is not hard to check that the term $S_0$ is BRST exact:

418: \be

419: \label{exactS_0}

420: S_0=\delta V_0

421: \ee

422: where:

423: \be

424: V_0=\frac{i}{4}\int_{\Sigma}{d^2\sigma \sqrt{g}\theta^{\bar i} \partial_{\bar

425: i}{\bar W}}~~.

426: \ee

427: Equations (\ref{exactS_B}) and (\ref{exactS_0}) are local, i.e. they hold for

428: the associated Lagrange densities without requiring integration by parts. Thus

429: both of these relations can be applied to bordered Riemann surfaces.

430: 

431: Since the boundary term (\ref{s}) is independent of the worldsheet metric, 

432: the bulk stress energy tensor has the form given in \cite{Labastida}:

433: \bea

434: T_{\mu \nu}&=&\frac{1}{2\sqrt{g}}\frac{\delta {\tilde S}_{bulk}}{\delta g^{\mu

435:     \nu}}=\frac{1}{2}G_{i{\bar j}}\left[\partial_\mu \phi^i\partial_\nu \phi^{\bar j}+

436: \partial_\nu \phi^i\partial_\mu \phi^{\bar j}-\frac{1}{2}\left(\rho_\mu^iD_\nu

437: \eta^{\bar j}+\rho_\nu^i D_\mu \eta^{\bar j}\right)\right]-\\

438: &-&\frac{1}{2}g_{\mu \nu}\left[G_{i{\bar j}}g^{\alpha\beta}(\partial_\alpha\phi^i\partial_\beta\phi^{\bar

439:     j}-\frac{1}{2} \rho_\alpha^i D_\beta \eta^{\bar j}-{\tilde F}^i{\tilde

440:     F}^{\bar j})-\frac{i}{2}\partial_i W {\tilde

441:     F}^i+\frac{i}{2}\partial_{\bar i}{\bar W}{\tilde

442:     F}^{\bar i}-\frac{i}{4}D_{\bar i}\partial_{\bar j}{\bar W}\theta^{\bar

443:     i}\eta^{\bar j}\right]~~.\nn

444: \eea

445: As explained in \cite{Labastida}, $T_{\mu\nu}$ is BRST exact only modulo

446: the equations of motion for the auxiliary fields:

447: \bea

448: \label{F_eom}

449: {\tilde F}^i=\frac{i}{2}G^{i{\bar j}}\partial_{\bar j}{\bar W}~~,~~

450: {\tilde F}^{\bar i}=-\frac{i}{2}G^{{\bar i}j}\partial_j W~~.

451: \eea

452: Imposing these equations, one finds:

453: \bea

454: T^{os}_{\mu\nu}&=&\frac{1}{2}G_{i{\bar j}}\left[\partial_\mu \phi^i\partial_\nu \phi^{\bar j}+

455: \partial_\nu \phi^i\partial_\mu \phi^{\bar j}-\frac{1}{2}\left(\rho_\mu^iD_\nu

456: \eta^{\bar j}+\rho_\nu^i D_\mu \eta^{\bar j}\right)\right]-\\

457: &-&\frac{1}{2}g_{\mu \nu}\left[G_{i{\bar j}}g^{\alpha\beta}(\partial_\alpha\phi^i\partial_\beta\phi^{\bar

458:     j}-\frac{1}{2} \rho_\alpha^i D_\beta \eta^{\bar j})+\frac{1}{4}G^{i{\bar j}}\partial_i W\partial_{\bar

459:     j}{\bar W}-\frac{i}{4}D_{\bar i}\partial_{\bar j}{\bar W}\theta^{\bar

460:     i}\eta^{\bar j}\right]~~.\nn

461: \eea

462: This obeys the BRST exactness condition \cite{Labastida}:

463: \be

464: \label{T_exact}

465: T^{os}_{\mu \nu}=\delta G_{\mu \nu}

466: \ee

467: where\footnote{The formula given for $G_{\mu \nu}$ in \cite{Labastida} seems

468:   to be missing a global prefactor of $-\frac{1}{2}$.}:

469: \be

470: \label{G_mu_nu}

471: G_{\mu\nu}=\frac{1}{4}\Big[G_{i{\bar j}}(\rho_\mu^i\partial_\nu \phi^{\bar j}+

472: \rho_\nu^i\partial_\mu \phi^{\bar j})-g_{\mu

473:   \nu}(G_{i{\bar j}}g^{\alpha\beta}\rho_\alpha^i\partial_\beta\phi^{\bar

474:   j}+\frac{i}{2}\theta^{\bar i}\partial_{\bar i}{\bar W})\Big]~~.

475: \ee

476: On an infinite flat cylinder, the supercharges: 

477: \be

478: \label{G_mu}

479: G_\mu:=\int{d\sigma_1 G_{0\mu}} 

480: \ee

481: generate symmetries $\delta_\mu=\{G_\mu, \cdot\}_P$ which together with $\delta=\{Q,\cdot\}_P$

482: and a supplementary nilpotent transformation $\delta':=\{M,\cdot\}_P$ 

483: form the topological algebra of

484: \cite{Labastida} (here  $\{\cdot, \cdot\}_P$ is the Poisson bracket of the

485: Hamiltonian formulation). When placing the model on a flat strip, the boundary

486: conditions break the symmetries $\delta'$ and $\delta_1$, but preserve the subalgebra

487: generated by $\delta$ and $\delta_0$. In the untwisted model, this subalgebra

488: corresponds to the usual B-type supersymmetry considered, for example, in \cite{Hori}.

489: 

490: 

491: 

492: 

493: 

494: \subsection{BRST variation of the bulk action and the topological Warner term}

495: 

496: It is not hard to check that the BRST variation of ${\tilde S}_{bulk}$ produces

497: a boundary term:

498: \be

499: \label{deltaSbulk}

500: \delta {\tilde S}_{bulk} =\delta S_1=\frac{1}{2} \int_{\partial

501:   \Sigma}{\rho^i\partial_i W}~~.

502: \ee

503: The presence of a non-zero right hand side in (\ref{deltaSbulk}) is known as

504:   the Warner problem \cite{Warner}.

505: 

506: 

507: \subsection{$\delta_0$-variation of the bulk action on flat Riemann surfaces}

508: 

509: When considered on a flat Riemann surface, our model has an enlarged symmetry

510: algebra which was originally described in \cite{Labastida}. In this paper we

511: shall need only the subalgebra obtained by considering an additional odd

512: generator $\delta_0$ beyond the BRST generator of (\ref{BRST}). In the

513: notations of \cite{Labastida}, we have $\delta_0=\{G_0,\}$, where

514: $G_0=G_z+G_{\bar z}$.  Using the results of \cite{Labastida}, one

515: finds:

516: \bea

517: \label{delta_0}

518: \delta_0\phi^i=\frac{1}{2}\rho^i_0~~&,&~~\delta_0\phi^{\bar i}=0\nn\\

519: \delta_0\rho_0^i=0~~&,&~~\delta_0\rho_1^i=-i{\tilde F}^i\\ 

520: \delta_0\eta^{\bar i}=\partial_0\phi^{\bar i}~~&,&~~\delta_0\theta^{\bar i}=-i\partial_1

521: \phi^{\bar i}\nn\\ 

522: \delta_0 {\tilde F}^i=0~~&,&~~\delta_0{\tilde F}^{\bar

523: i}=\frac{1}{2}(\partial_0\theta^{\bar i}+i\partial_1\eta^{\bar i})~~.\nn

524: \eea

525: In this subsection, we are assuming that the worldsheet metric is flat, namely

526: $g_{\alpha\beta}=\delta_{\alpha\beta}$ in the real coordinates $\sigma^0$ and

527: $\sigma^1$.

528: 

529: If $\Sigma$ is a cylinder, one finds $\delta_0{\tilde S}_{bulk}=0$. However,

530: the $\delta_0$ variation of ${\tilde S}_{bulk}$ gives a boundary term when the

531: model is considered on the strip or on the disk.  Let us take $\Sigma$ to be

532: infinite the strip given by $(\sigma^0,\sigma^1)\in \R\times [0,\pi]$. We find:

533: \be \delta_0 {\tilde S}_{bulk}=-\frac{1}{4}\int_{\partial \Sigma}{d\tau

534: \eta^{\bar i}\partial_{\bar i}{\bar W}}~~.  \ee Here $d\tau=d\sigma^0$ is the

535: length element along the boundary of $\Sigma$.

536: 

537: 

538: \section{Localization formula for correlators on the sphere}

539: \label{loc_bulk}

540: 

541: Let us consider zero-form bulk observables ${\cal O}$ which are independent

542: on the auxiliary fields ${\tilde F}^i$ or ${\tilde F}^{\bar i}$.  We are interested in the

543: sphere correlator of such observables:

544: \be

545: \label{correlator}

546: \langle {\cal O}\rangle_{\rm sphere} =\int{{\cal D}[\phi]{\cal D}[{\tilde F}]

547: {\cal D}[\eta]{\cal D}[\theta]{\cal D}[\rho]{~e^{-{\tilde S}_{bulk}}{\cal

548: O}}}~~,

549: \ee

550: where we assume that ${\cal O}$ is BRST closed.

551: 

552: In this section, we re-consider the localization formula for such correlators. 

553: Extending the original argument of \cite{Vafa_LG}, we will extract a

554: one-parameter family of representations of (\ref{correlator}) as a

555: finite-dimensional integral. The basic point is a follows. Since the B-model

556: piece of the Landau-Ginzburg action is BRST exact off-shell, the standard

557: argument of \cite{Witten_mirror} implies that we can localize on the zero

558: modes of the associated {\em sigma model}. Thus we shall localize on constant

559: maps, without requiring that such maps send the worldsheet to the critical

560: points of $W$. After this reduction, one finds that the Lagrangian density

561: of the model becomes BRST exact, so the resulting integral representation 

562: is insensitive to multiplying the Lagrangian density by a prefactor

563: $\lambda$. Since the former appears multiplied by the worldsheet area,

564: this prefactor measures the scale of the underlying $S^2$ worldsheet. Thus we

565: obtain a one-parameter family of localization formulae for our

566: correlator. Each such "localization picture" allows us to give a geometric

567: representation of genus zero data, thus providing a geometric model for

568: the off-shell state space, BRST operator and bulk trace. 

569: 

570: In this approach, the residue representation of \cite{Vafa_LG} is recovered in

571: the limit $\Re\lambda\rightarrow +\infty$, which forces the point-like image of

572: the worldsheet to lie on the critical set of $W$. Intuitively, this is the

573: limit of {\em large} worldsheet areas, the opposite of the

574: "microscopic" limit $\lambda\rightarrow 0$. 

575: Varying $\lambda$ allows one to interpolate between these

576: limits, thus connecting the "sigma-model like" and "residue-like" 

577: models of genus zero data. 

578: 

579: \subsection{Localization on $B$-model zero modes}

580: 

581: Since ${\tilde S}_B$ is BRST exact, we can replace the bulk

582: action with:

583: \be

584: {\tilde S}_{bulk}=t {\tilde S}_B+ S_W=t\delta V_B+S_W~~,

585: \ee

586: where $t$ is a complex parameter with $\Re t >0$ (so that the integral is

587: well-defined).  BRST invariance of the path integral together with BRST

588: closure of ${\cal O}$ imply that the resulting correlator is independent of

589: $t$. This means that (\ref{correlator}) can be computed in the limit

590: $\Re~t\rightarrow +\infty$, where the integral localizes on the zero-modes of

591: ${\tilde S}_B$. Since $\rho$ has no zero-modes on the sphere, we must

592: consider configurations for which $\rho^i_\alpha=0$ while $\phi$, $\eta$,

593: $\theta$ and ${\tilde F}$ are constant on the worldsheet.  For such configurations,

594: ${\tilde S}_B$ reduces to:

595: \be

596: {\tilde S}_B|_{\rm zero~modes}=-A G_{i{\bar j}}{\tilde F}^i{\tilde F}^{\bar j}~~,

597: \ee

598: with $G_{i{\bar j}}=G_{i{\bar j}}(\phi)$. Here $A$ is the area of the

599: worldsheet.  The contribution $S_W$ becomes:

600: \be

601: S_W|_{\rm zero~modes}=-\frac{i}{2}A\left[ D_{\bar i} \partial_{\bar j}{\bar W}

602: \chi^{\bar i}{\bar \chi}^{\bar j}- (\partial_{\bar i}{\bar W}){\tilde F}^{\bar

603: i}+ (\partial_i W){\tilde F}^i\right]~~.

604: \ee

605: Combining these expressions, we find the zero-mode reduction of the worldsheet

606: action:

607: \be

608: {\tilde S}_{0}:={\tilde S}_{bulk}|_{\rm zero~modes}=-A G_{i{\bar j}}{\tilde

609: F}^i{\tilde F}^{\bar j}- \frac{i}{2}A\left[ \frac{1}{2}D_{\bar i}

610: \partial_{\bar j}{\bar W} \theta^{\bar i}\eta^{\bar j}- (\partial_{\bar

611: i}{\bar W}){\tilde F}^{\bar i}+ (\partial_i W){\tilde F}^i\right]~~,

612: \ee

613: where we wrote $\chi$ and ${\bar \chi}$ in terms of $\eta$ and $\theta$.

614: The correlator (\ref{correlator}) reduces to an ordinary integral over the

615: zero-modes $\phi^i, \phi^{\bar i}$, ${\tilde F}^i, {\tilde F}^{\bar i}$ and

616: $\eta^{\bar i},\theta_i$:

617: \be

618: \label{correlator_0}

619: \langle {\cal O}\rangle_{\rm sphere} =\int{d\phi d{\tilde F} d\eta d\theta

620: {~e^{-{\tilde S}_{0}}{\cal O}}}~~.

621: \ee

622: On zero modes, the BRST generator (\ref{BRST}) takes the form:

623: \bea

624: \label{BRST_0}

625: \delta \phi^i=0~~&,&~~\delta \phi^{\bar i}=\eta^{\bar i}\nn\\ \delta

626: \eta^{\bar i}=0~~&,&~~\delta \theta^{\bar i}=2{\tilde F}^{\bar i}+\Gamma^{\bar

627: i}_{{\bar j}{\bar k}}\theta^{\bar j}\eta^{\bar k}~~\\ \delta {\tilde

628: F}^i=0~~&,&~~\delta {\tilde F}^{\bar i}=\Gamma^{\bar i}_{{\bar j}{\bar

629: k}}{\tilde F}^{\bar j}\eta^{\bar k}~~.\nn

630: \eea

631: In particular, we have:

632: \be

633: \label{tildeS_B_0_ex}

634: G_{i{\bar j}}{\tilde F}^i{\tilde F}^{\bar j}=\frac{1}{2}\delta [G_{i{\bar j}}{\tilde

635:   F}^i\theta^{\bar j}]~~,

636: \ee

637: which is the zero-mode remnant of equation (\ref{exactS_B}).  One can also

638: check directly that the reduced action is BRST-closed.

639: 

640: The integral over ${\tilde F}$ can be cast into Gaussian form through the change

641: of variables:

642: \be

643: {\tilde F}^i={\hat F}^i+\frac{i}{2}G^{i{\bar j}}\partial_{\bar j}{\bar W}~~,~~

644: {\tilde F}^{\bar i}={\hat F}^{\bar i}-\frac{i}{2}G^{{\bar i}j}\partial_j W~~.

645: \ee

646: Then the reduced action becomes: 

647: \be

648: {\tilde S}_{0}=-AG_{i{\bar j}}{\hat F}^i{\hat F}^{\bar j}+

649: \frac{1}{4}A \left[

650: -iD_{\bar i} \partial_{\bar j}{\bar W} \theta^{\bar

651:       i}\eta^{\bar j}+G^{i{\bar j}}(\partial_i W)(\partial_{\bar j}{\bar W})\right]~~.

652: \ee

653: Integrating over ${\hat F}$, we find:

654: \be

655: \label{correlator_0intmd}

656: \langle {\cal O}\rangle_{\rm sphere}=N \int{d\phi d\eta d\theta

657: e^{-\frac{A}{4} {\tilde L}_0}{\cal O}}~~,

658: \ee

659: where:

660: \be

661: \label{tildeL_0}

662: {\tilde L}_0:=-i D_{\bar i} \partial_{\bar j}{\bar W} \theta^{\bar

663:       i}\eta^{\bar j}+G^{i{\bar j}}(\partial_i W)(\partial_{\bar j}{\bar

664:       W})~~

665: \ee

666: plays the role of zero-mode Lagrange density. The prefactor in 

667: (\ref{correlator_0intmd})  has the form:

668: \be

669: N=\frac{(2\pi)^n}{ A^n \det(G_{i{\bar j}})}~~,

670: \ee

671: where $n$ is the complex dimension of the target space $X$. Since we

672: integrated out the fields ${\hat F}$, the BRST generator on

673: zero-modes reduces to:

674: \bea

675: \label{BRST_os_0}

676: \delta \phi^i=0~~&,&~~\delta \phi^{\bar i}=\eta^{\bar i}\nn\\

677: \delta \eta^{\bar i}=0~~&,&~~\delta \theta^{\bar

678:   i}=-iG^{{\bar i}j}\partial_jW+ \Gamma^{\bar i}_{{\bar j}{\bar

679:   k}}\theta^{\bar j}\eta^{\bar k}~~,

680: \eea

681: which is obtained from  (\ref{BRST_0}) by imposing the equations of motion:

682: \be

683: {\tilde F}^i=\frac{i}{2}G^{i{\bar j}}\partial_{\bar j}{\bar W}~~,~~

684: {\tilde  F}^{\bar i}=-\frac{i}{2}G^{{\bar i}j}\partial_jW~~.

685: \ee

686: For later reference, notice that the last relation in (\ref{BRST_os_0}) is equivalent with:

687: \be

688: \label{delta_theta_os}

689: \delta \theta_i=-i\partial_i W~~.

690: \ee

691: 

692: Using this form of the BRST transformations, one finds that the zero-mode

693: Lagrange density (\ref{tildeL_0}) is BRST exact:

694: \be

695: \label{exact_tildeL_0}

696: {\tilde L}_0=\delta {\tilde v}_0~~,

697: \ee

698: where:

699: \be

700: \label{tildev_0}

701: {\tilde v}_0=i\theta^{\bar i}\partial_{\bar i}{\bar W}~~.

702: \ee

703: Thus we can replace  (\ref{correlator_0intmd}) by:

704: \be

705: \label{correlator_final}

706: \langle {\cal O}\rangle_{\rm sphere} 

707: =N\int{d\phi d\eta d\theta {~e^{-\lambda {\tilde L}_0}{\cal O}}}~~,

708: \ee

709: where $\lambda$ is a complex parameter with positive real part. The integral

710: (\ref{correlator_final}) is independent of its value.

711: 

712: 

713: \subsection{The space of bulk observables and its cohomology}

714: \label{bulk_observables}

715: 

716: The observables of interest have the form:

717: \be

718: \label{bulk_obs}

719: {\cal O}_\omega(\sigma):=\omega^{j_1\dots j_q}_{{\bar i}_1\dots {\bar

720:     i}_p}(\phi(\sigma))\eta^{{\bar

721:     i}_1}(\sigma) \dots \eta^{{\bar i}_p}(\sigma)

722: \theta_{j_1}(\sigma) \dots \theta_{j_q}(\sigma)~~,

723: \ee

724: where $\omega:=\omega^{j_1\dots j_q}_{{\bar i}_1\dots {\bar i}_p}dz^{{\bar

725:   i}_1}\wedge \dots \wedge dz^{{\bar i}_p}\wedge \partial_{j_1}\wedge \dots \wedge \partial_{j_q}$

726: is a section of the bundle 

727: $\Lambda^p{{\bar T}^*X}\wedge \Lambda^q TX$. After reduction to B-model 

728: zero modes, we are left with:

729: \be

730: {\cal O}_\omega=\omega^{j_1\dots j_q}_{{\bar i}_1\dots {\bar

731:     i}_p}(\phi)\eta^{{\bar

732:     i}_1} \dots \eta^{{\bar i}_p}

733: \theta_{j_1}\dots \theta_{j_q}~~,

734: \ee

735: which can be identified with the polyvector-valued form $\omega$ upon setting:

736: \be

737: \label{ids_geom}

738: \eta^{\bar j}\equiv dz^{\bar j}~~,~~\theta_j\equiv \partial_j~~. 

739: \ee

740: 

741: The reduced BRST operator (\ref{BRST_os_0})  becomes:

742: \be

743: \label{delta_geom}

744: \delta\equiv {\bar \partial}+i_{\partial W}

745: \ee

746: where $i_{\partial_W}$ is the odd derivation of $\Lambda^*TX$ uniquely determined

747: by the conditions:

748: \be

749: i_{\partial W}(\partial_j)=-i\partial W(\partial_j)=-i\partial_j W~~.

750: \ee

751: Thus the cohomology of the differential superalgebra 

752: $({\cal H}, \delta)$, where ${\cal H}:=\Gamma(\Lambda^*{\bar T}^*X\wedge \Lambda^*TX)$, models

753: the algebra of bulk observables. 

754: The obvious  relations:

755: \be

756: {\bar \partial}^2=(i_{\partial W})^2={\bar

757:   \partial}i_{\partial_W}+i_{\partial_W}{\bar \partial}=0~~

758: \ee

759: show that $({\cal H},\delta)$ is a bicomplex. Hence the BRST cohomology is

760: computed by a spectral sequence $E_*$ whose second term equals:

761: \be

762: \label{E}

763: E_2:=H_{i_{\partial_W}}(H_{\bar \partial}({\cal H}))~~.

764: \ee

765: Since the target space is non-compact, we must of course specify a growth condition at

766: infinity. We shall take ${\cal H}$ to consist of those sections of 

767: the bundle $\Lambda^*{\bar T}^*X\wedge \Lambda^*TX$ whose coefficients have at most

768: polynomial growth. When the spectral sequence collapses to its second term,

769: the BRST cohomology reduces to (\ref{E}). A standard example is the case

770: $X=\C^n$, with $W$ a polynomial function of $n$ variables. 

771: Then the $\bar{\partial}$-Poincare Lemma implies 

772: that $H_{\bar \partial}({\cal H})$ coincides with the space

773: $\Gamma_{poly}(\Lambda^*TX)$ of polyvector fields with polynomial

774: coefficients.  In this case, the BRST cohomology reduces to the Jacobi ring

775: $\C[x_1\dots x_n]/\langle \partial_1 W\dots \partial_n W\rangle$, thereby

776: recovering a well-known result.  

777: 

778: 

779: 

780: 

781: 

782: \subsection{The geometric model}

783: \label{bulk_geom}

784: 

785: Let us translate (\ref{correlator_final}) into

786: classical mathematical language. Using (\ref{ids_geom}), we find:

787: \be

788: \label{tildeL_0form_0}

789: {\tilde L}_0\equiv i{\bar H}^{i}_{~{\bar j}}

790: dz^{\bar j}\wedge \partial_{i}+

791: G^{i{\bar j}}(\partial_iW)(\partial_{\bar j} {\bar W}) \in {\cal H}~~.

792: \ee

793: Here $H^i_{~{\bar j}}:=G^{i{\bar k}}{\bar H}_{{\bar k}{\bar j}}$, where 

794: $H_{ij}:=D_i\partial_j W$ is the Hessian of $W$. Consider the Hessian

795: operator:

796: \be

797: H=H^{\bar i}_{~j}dz^j\otimes \partial_{\bar i}\in Hom(TX,{\bar T}X)=T^*X\otimes

798: {\bar T}X~~,

799: \ee

800: whose complex conjugate has the form:

801: \be

802: {\bar H}={\bar H}^i_{~{\bar j}}dz^{\bar j}\otimes \partial_i\in Hom({\bar

803:   T}X,TX)={\bar T}^*X\otimes TX~~.

804: \ee

805: The quantity

806: ${\bar H}_a:={\bar H}^i_{~{\bar j}}dz^{\bar j}\wedge \partial_i\in {\bar T}^*X\wedge TX$ 

807: appearing in (\ref{tildeL_0form_0}) is the antisymmetric part of

808: ${\bar H}$.  On the other hand, the second term of (\ref{tildeL_0form_0}) is the norm of the

809: differential form $\partial W=\partial_iW dz^{i}$. This gives the

810: coordinate-independent version of (\ref{tildeL_0form_0}):

811: \be

812: {\tilde L}_0=i{\bar H}_{a}+||\partial W||^2~~.

813: \ee

814: Also note the representation:

815: \be

816: {\tilde v}_0=iG^{i{\bar j}}\partial_i\wedge \partial_{\bar j} {\bar W}~~.

817: \ee

818: It is now easy to see that (\ref{correlator_final}) becomes:

819: \be

820: \label{correlator_geom}

821: \Tr\omega:=\langle {\cal O}_\omega\rangle_{\rm sphere}=N\int_{X}{\Omega\wedge \left[

822: \Omega \lrcorner \left(e^{-\lambda {\tilde L}_0}\wedge \omega\right)\right]}~~,

823: \ee

824: where $\lrcorner$ denotes the total contraction of a form with a

825: polyvector. The linear functional $\Tr$ realizes the bulk trace of \cite{CIL1}. 

826: 

827: 

828: \paragraph{Observation}

829: The integral representation (\ref{correlator_geom}) allows us 

830: to give another (and completely rigorous) 

831: proof of $\lambda$-independence for $\delta

832: \omega=0$, with the assumption $\Re \lambda>0$. 

833: For this, we have to show that the $\lambda$-derivative of

834: (\ref{correlator_geom}) vanishes.

835: Since ${\tilde L}_0=\delta {\tilde v}_0$ and $\delta \omega=0$, this

836: derivative takes the form:

837: \be

838: \label{der_geom}

839: \frac{d}{d\lambda}\Tr \omega =

840: -\lambda N\int_{X}{\Omega\wedge \left[\Omega \lrcorner 

841: \delta \left(e^{-\lambda {\tilde L}_0}\wedge {\tilde v}_0\wedge \omega\right)\right]}~~.

842: \ee

843: Thus it suffices to show that 

844: $\int_{X}{\Omega\wedge \left[\Omega \lrcorner \delta \alpha\right]}$

845: vanishes for any $\alpha\in {\cal H}$ which decays exponentially at infinity

846: on $X$ (the exponential  decay for $\alpha=e^{-\lambda {\tilde L}_0}{\tilde

847:   v}_0\wedge \omega$

848: in (\ref{der_geom}) is due to the second term in (\ref{tildeL_0form_0})). 

849: Notice further that $\int_{X}{\Omega\wedge \left[\Omega

850:   \lrcorner \delta \alpha\right]}$

851: vanishes for degree reasons unless $\delta \alpha\in \Gamma(\Lambda^n{\bar T}^*X\wedge \Lambda^n

852: TX)$. Hence it is enough to show vanishing of $\int_{X}{\Omega\wedge \left[

853: \Omega \lrcorner \delta \alpha\right]}$ for an exponentially decaying $\alpha$

854: such that

855: $\delta \alpha\in \Gamma(\Lambda^n{\bar T}^*X\wedge \Lambda^n TX)$. In this

856: case, we obviously have $\delta\alpha={\bar \partial}\beta$ for some 

857: exponentially decaying $\beta\in \Gamma(\Lambda^{n-1}{\bar T}^*X\wedge

858: \Lambda^n TX)$ (this follows by noticing that the

859: image of $i_{\partial W}$ has vanishing intersection with the subspace 

860: $\Gamma(\Lambda^n{\bar T}^*X\wedge \Lambda^n TX)$). Therefore, we only need to show

861: that $\int_{X}{\Omega\wedge \left[\Omega \lrcorner {\bar \partial} \beta\right]}$

862: vanishes. This last fact follows from 

863: $\Omega\wedge \left[\Omega \lrcorner {\bar \partial}

864:   \beta\right]={\bar \partial}\left(\Omega\wedge \left[\Omega

865:     \lrcorner \beta\right]\right)$, since the boundary term

866: vanishes due to the exponential decay of $\beta$. 

867: The assumption $\Re\lambda>0$ is crucial,

868: since otherwise we cannot rely on exponential decay to conclude that the

869: boundary term vanishes.

870: 

871: 

872: \subsection{Localization pictures and homotopy flows}

873: \label{bulk_flow}

874: 

875: Expression (\ref{correlator_geom}) admits the following interpretation.

876: Consider the one-parameter semigroup of operators $U(\lambda)$ acting on

877: ${\cal H}$ through wedge multiplication by $e^{-\lambda{\tilde L}_0}$:

878: \be

879: U(\lambda)\omega:=e^{-\lambda{\tilde L}_0}\wedge \omega~~{\rm~for~all}~~\omega\in {\cal H}~~.

880: \ee

881: The semigroup is defined on the half-plane $\Delta:=\{\lambda\in

882: \C|\Re\lambda>0\}$, so that $U(\lambda)$ maps ${\cal H}$ into a subspace of

883: itself. Then (\ref{correlator_geom}) takes the form: 

884: \be

885: \label{Tr_flow}

886: \Tr\omega:=\Tr^B(U(\lambda)\omega)~~,

887: \ee

888: where $\Tr^B$ is the bulk trace of the B-twisted sigma model: 

889: \be

890: \Tr^B\omega:=N\int_{X}{\Omega\wedge \left(

891: \Omega \lrcorner \omega\right)}~~.

892: \ee

893: 

894: Since ${\tilde L}_0$ is BRST closed (${\tilde L}_0=\delta {\tilde v}_0$), each

895: operator $U(\lambda)$ is homotopy equivalent with the identity in the complex $({\cal H},

896: \delta)$:

897: \be

898: U(\lambda)=1+[\delta, W_\lambda]~~,

899: \ee

900: for some operator $W_\lambda$. In particular, $U(\lambda)$ is an endomorphism of our

901: complex, i.e. the following relation holds: 

902: \be

903: \label{endomorphism}

904: U(\lambda)\circ \delta=\delta \circ U(\lambda)~~. 

905: \ee

906: Such a semigroup will be called a {\em homotopy

907: flow}. It is clear that each $U(\lambda)$ is a quasi-isomorphism, i.e. 

908: induces an automorphism $U_*(\lambda)$ on the BRST cohomology $H_\delta({\cal

909:   H})$. Following relation (\ref{Tr_flow}), we define the {\em localization

910:   picture $\lambda$} by associating $\omega_\lambda:=U(\lambda)(\omega)\in

911: {\cal H}$ to each $\omega\in {\cal H}$ (then $\omega_\lambda$ is the representative of the "state"

912: $\omega$ in the picture $\lambda$). The representatives of this

913: picture belong to the subspace ${\cal H}_\lambda:=U(\lambda)({\cal

914:   H})\subset {\cal H}$. As in quantum mechanics, we have a representative for

915: any operator $T\in End({\cal H})$ in the localization picture $\lambda$:

916: \be

917: T_\lambda:=U(\lambda)\circ T \circ U(-\lambda)\in End({\cal H}_\lambda)~~,

918: \ee

919: where $\Re\lambda>0$ and $U(-\lambda)$ is defined as an operator from ${\cal

920:   H}_\lambda$ to ${\cal H}$. Relation (\ref{endomorphism})

921: shows that the BRST operator is "picture-independent" in the following sense: 

922: \be

923: Q_\lambda=Q|_{{\cal H}_\lambda}~~,

924: \ee

925: where in the right hand side we restrict both the domain and image 

926: to ${\cal H}_\lambda$. Relation (\ref{Tr_flow}) becomes:

927: \be

928: \Tr\omega=\Tr^B\omega_\lambda~~. 

929: \ee

930: 

931: For $\lambda=W=0$, we have $U(0)=Id_{\cal H}$ and we recover the familiar data

932: of the B-twisted sigma model. Namely, ${\cal H}$ provides a geometric model for the

933: off-shell state space, the Dolbeault operator ${\bar \partial}$ models the "localized

934: BRST operator" and $\Tr^B$ models the bulk trace of \cite{CIL1}. 

935: Turning on the Landau-Ginzburg superpotential $W$ and

936: performing localization as above with "worldsheet area" $\lambda$  leads to a

937: geometric model given by the triplet $({\cal H}, \delta, \Tr)$. This is

938: related to the triplet describing the B-twisted sigma model by the modification

939: $\delta={\bar \partial}+i_{\partial W}$ of the BRST operator, followed by the

940: homotopy flow $U(\lambda)$. 

941: 

942: Because varying $\lambda$ along the real axis amounts to changing the area of the

943: worldsheet, the operators $U(\lambda)$ implement a sort of

944: "renormalization group flow" connecting the point-like (UV) limit $\lambda=0$

945: with the large area (IR limit) $\lambda=+\infty$. Since the model is

946: topological, such a flow "does nothing" at the level of BRST cohomology, but

947: acts non-trivially off-shell.

948: 

949: 

950: \subsection{The residue formula for sphere correlators}

951: 

952: Since the integral (\ref{correlator_final}) is independent of $\lambda$, we

953: can compute its value for $\Re \lambda\rightarrow +\infty$ (more specifically,

954: we shall take $\lambda\rightarrow +\infty$ with $\lambda \in \R$). In this limit,

955: the second term in (\ref{tildeL_0}) forces the integral to localize on the

956: critical points of $W$, and

957: the Gaussian approximation  around these points 

958: becomes exact. For simplicity, we shall assume that the critical points of $W$

959: are isolated (the general case can be incorporated by a continuity argument). 

960: For simplicity, we shall also assume that the spectral sequence of Subsection

961: \ref{bulk_observables} collapses to its second term. 

962: 

963: Taking $\lambda\rightarrow +\infty$ with $\lambda \in \R$, we find that the correlator 

964: (\ref{correlator_final}) vanishes unless $\omega=f$ with $f$ a complex-valued 

965: function defined on $X$ \footnote{For this, notice that the bosonic Gaussian integral over

966:   fluctuations of $\phi$ around

967:   each critical point of $W$ produces a factor which is weighted by

968:   $\frac{1}{\lambda^n}$. Thus the fermionic Gaussian integral over $\theta$

969:   and $\eta$ must produce $n$ powers of $\lambda$ if one is to obtain a

970:   non-vanishing result in the limit $\lambda\rightarrow \infty$. This

971:   obviously requires that ${\cal O}_\omega$ contain no $\eta$'s or $\theta$'s, 

972:   so that the highest ($n$-th order term) in the expansion of $e^{+i\lambda D_{\bar i}

973:     \partial_{\bar j}{\bar W} \theta^{\bar i}\eta^{\bar j}}$ survives when

974:   performing the integral over $\eta$ and $\theta$. }. In this case, we obtain:

975: \bea &&\langle {\cal O}_f \rangle_{\rm sphere}=\nn\\

976: &=&\lim_{\lambda\rightarrow +\infty} \left(N\sum_{p\in {\rm Crit}

977: W}{\left[n!^2 (i\lambda)^n (-1)^{n(n-1)/2}\det ({\bar H}_{{\bar i}{\bar

978: j}})\right]\left[\frac{(2\pi)^n}{\lambda^n \det (H_{ij}) \det ({\bar H}_{{\bar

979: i}{\bar j}})}\det (G_{i{\bar j}})\right]}f|_{p}+O(1/\lambda)\right)\nn\\

980: &=&n!^2 (2\pi)^n (-1)^{n(n+1)/2}N\det (G_{i{\bar j}})\sum_{p\in {\rm Crit} W}{

981: \frac{1}{\det (H_{ij}(p))} f(p)}~~.  \eea

982: Since $\sum_{p\in {\rm Crit} W}{\frac{1}{\det (H_{ij}(p))} f(p)}

983: \propto \int_{X}{\frac{f(z)dz^1\wedge\dots \wedge dz^n }{\partial_1W\dots

984:   \partial_n W}}$ by residue theory \cite{Griffiths}, 

985: one recovers the following generalization of the well-known result of \cite{Vafa_LG}:

986: \bea

987: \langle {\cal O}_\omega \rangle_{\rm sphere}&=&0~~{\rm unless}~~\omega=f \\

988: \langle {\cal O}_f\rangle_{\rm sphere}&=& C

989: \int_{X}{\Omega \frac{f(z) }{\partial_1W\dots  \partial_n W}}~~.

990: \eea

991: Here $C$ is an uninteresting normalization constant. 

992: 

993: 

994: 

995: 

996: 

997: 

998: 

999: 

1000: \section{The boundary coupling}

1001: \label{boundary_coupling}

1002: 

1003: In this section we discuss the boundary coupling of our models. The

1004: construction is based on \cite{coupling}, with a certain modification which

1005: will prove useful later on. After recalling the basics of 

1006: superconnections, we construct the coupling in the form of \cite{coupling},

1007: with the addition of a term which insures $\delta_0$-invariance on a flat

1008: strip. While this does not affect the target space equations of motion,

1009: it will help us make contact with previous work on the subject. We also give

1010: the target space reflection of the $\delta_0$-invariance constraint. 

1011: 

1012: 

1013: 

1014: \subsection{Mathematical preparations}

1015: 

1016: Consider a complex superbundle $E=E_+\oplus E_-$ over $X$, and a

1017: superconnection \cite{Quillen} ${\cal B}$ on $E$. We let $r_\pm:=\rk E_\pm$. 

1018: The bundle of endomorphisms $End(E)$ is endowed with the 

1019: natural $\Z_2$ grading, with even and odd components:

1020: \bea

1021: End_+(E)&:=&~~~~~End(E_+)\oplus End(E_-)\\

1022: End_-(E)&:=& Hom(E_+,E_-)\oplus Hom(E_-,E_+)~~.

1023: \eea

1024: The superconnection ${\cal B}$  can be viewed 

1025: as a section of $[{\cal T}^*X\otimes End_+(E)]\oplus End_-(E)$. 

1026: In a local frame of $E$ compatible with the grading, this is a matrix:

1027: \be

1028: {\cal B}=\left[\ba{cc} A^{(+)}& F\\G & A^{(-)}\ea\right]

1029: \ee

1030: whose diagonal entries $A^{(\pm)}$ are connection one-forms on $E_\pm$, while 

1031: $F,G$ are elements of  $Hom(E_-,E_+)$ and $Hom(E_+,E_-)$. We 

1032: require that the superconnection has type $(0,\leq 1)$, i.e. the one-forms

1033: $A^{(\pm)}$ belong to $\Omega^{(0,1)}(End(E_\pm))$. The morphism $F$ should not be

1034: confused with the curvature form used below.

1035: 

1036: When endowed with the ordinary composition of morphisms, the space of sections

1037: $\Gamma(End(E))$ becomes an associative superalgebra. The space

1038: ${\cal H}_b:=\Omega^{(0,*)}(End(E))$ also carries an associative superalgebra structure,

1039: which is induced from $(\Omega^{(0,*)}(X),\wedge)$ and 

1040: $(\Gamma(End(E)),\circ)$ via the tensor product decomposition:

1041: \be

1042: \Omega^{(0,*)}(End(E))=\Omega^{(0,*)}(X)\otimes_{\Omega^{(0,0)}(X)} \Gamma(End(E))~~.

1043: \ee

1044: For decomposable

1045: elements $u=\omega\otimes f$ and $v=\eta\otimes g$, with

1046: homogeneous  

1047: $\omega,\eta$ and $f,g$, the associative product on ${\cal H}_b$ takes the form:

1048: \be

1049: uv=(-1)^{ \deg f~\rk \eta}(\omega\wedge \eta)\otimes (f\circ g)~~,

1050: \ee

1051: where $\deg$ denotes the grading of the superalgebra $End(E)$:

1052: \be

1053: \deg(f)=0\in \Z_2~~{\rm~if~}f\in End_+(E)~~,~~\deg(f)=1\in \Z_2~~{\rm~if~}f\in End_-(E)~~.

1054: \ee

1055: The total degree on ${\cal H}_b$ is given by:

1056: \be

1057: |\omega\otimes f|=\rk \omega +\deg f~~(mod~2)~~.

1058: \ee

1059: We also recall the supertrace on $End(E)$:

1060: \be

1061: \label{str}

1062: \str(f)=\tr f_{++}-\tr f_{--}~~,

1063: \ee

1064: where $f=\left[\ba{cc}f_{++}&f_{-+}\\f_{+-}&f_{--}\ea\right]$ is an

1065: endomorphism of $E$ with components $f_{\alpha\beta}\in Hom(E_\alpha,E_\beta)$

1066: for $\alpha,\beta=+,-$. This has the property:

1067: \be

1068: \label{str_cyc}

1069: \str(f\circ g)=(-1)^{\deg f \deg g}\str(g\circ f)~~

1070: \ee

1071: for homogeneous elements $f,g$. 

1072: 

1073: The twisted Dolbeault operator:

1074: \be

1075: \label{sc}

1076: {\bar {\cal D}}={\overline \partial} +{\cal B}=\left[\ba{cc} 

1077: {\bar \partial}+A^{(+)}& F\\G &{\bar \partial}+A^{(-)}\ea\right]

1078: \ee

1079: induces an odd derivation ${\bar \partial}+[{\cal B},\cdot]$ 

1080: of the superalgebra ${\cal H}_b$, where

1081: $[u,v]:=uv-(-1)^{|u||v|}vu$ is the supercommutator. 

1082: 

1083: The $(0,\leq 2)$ part of the superconnection's curvature has the form:

1084: \be

1085: \label{curvature}

1086: {\cal F}^{(0,\leq 2)}={\bar {\cal D}}^2={\overline \partial}

1087: {\cal B}+\frac{1}{2}[{\cal B}, {\cal B}]={\overline \partial}

1088: {\cal B}+{\cal B}{\cal B}=

1089: \left[\ba{cc} F^{(+)}_{(0,2)}+FG& {\bar \nabla} F\\{\bar \nabla} G &F^{(-)}_{(0,2)}+GF\ea\right]

1090: \ee

1091: where $F^{(\pm)}_{(0,2)}$ are the $(0,2)$ pieces of the curvature forms

1092: $F^{(\pm)}$ of $A^{(\pm)}$ and:

1093: \bea

1094: {\bar \nabla}F&=&{\bar \partial}F+

1095: A^{(+)}F+FA^{(-)}={\bar \partial}F+A^{(+)}\circ F-F\circ A^{(-)}~~\nn\\

1096: {\bar \nabla}G&=&{\bar \partial}G+A^{(-)}G+G A^{(+)}

1097: ={\bar \partial} G+A^{(-)}\circ G-G\circ A^{(+)}~~.

1098: \eea

1099: 

1100: We will use the the notations:

1101: \be

1102: \label{AD}

1103: A:=A^{(+)}\oplus A^{(-)}=\left[\ba{cc} A^{(+)}& 0\\0 & A^{(-)}\ea\right]~~,~~

1104: D:=\left[\ba{cc} 0& F\\G & 0\ea\right]

1105: \ee

1106: for the diagonal and off-diagonal parts of ${\cal B}$. Then $A$ is an

1107: connection one-form on $E$ (compatible with the grading), while

1108: $D$ is an odd endomorphism. We have ${\cal B}=A+D$ and:

1109: \be

1110: \label{02curvature}

1111: {\cal F}^{(0,\leq 2)}=F^{(0,2)}+{\bar \nabla}_A D +D^2~~.

1112: \ee

1113: Here $F^{(0,2)}=F^{(+)}_{(0,2)}+F^{(-)}_{(0,2)}$ is the $(0,2)$ part of the curvature of $A$ and 

1114: ${\bar \nabla}_A={\bar \partial}+[A,\cdot]$ is the Dolbeault operator twisted by $A$.  

1115: 

1116: \subsection{The boundary coupling}

1117: 

1118: Following \cite{coupling}, we define the partition function on a bordered and

1119: oriented Riemann surface $\Sigma$ by:

1120: \be

1121: \label{Z}

1122: Z:=\int{{\cal D}[\phi]{\cal D}[{\tilde F}]{\cal D}[\theta]{\cal D}[\rho]{\cal D}[\eta]

1123:   e^{-{\tilde S}_{bulk}} {\cal U}_1\dots {\cal U}_h}~~,

1124: \ee

1125: where $h$ is the number of holes and the factors ${\cal U}_a$ have the form:

1126: \be

1127: \label{calU}

1128: {\cal U}_a:=\Str Pe^{-\oint_{C_a}{d\tau_a M}}~~.

1129: \ee

1130: We are assuming that the boundary of $\Sigma$ is a disjoint union of 

1131: smooth circles $C_a$, associated with holes labeled by $a$. 

1132: The symbol $\Str$ denotes the supertrace on $GL(r_+|r_-)$, while $d\tau_a$ 

1133: stands for the length element along $C_a$ induced by the metric on the

1134: interior of $\Sigma$. The quantity $M$ is given by:

1135: \be

1136: \label{M}

1137: M=\left[\ba{cc} 

1138: {\hat {\cal A}}^{(+)}+\frac{i}{2}(FF^\dagger+G^\dagger G) &

1139: \frac{1}{2}\rho^i_0 \nabla_i F+\frac{i}{2}\eta^{\bar i}\nabla_{\bar i}G^\dagger

1140: \\

1141: \frac{1}{2} \rho^i_0 \nabla_i G +\frac{i}{2}\eta^{\bar i}\nabla_{\bar

1142:   i}F^\dagger &

1143: {\hat {\cal A}}^{(-)} +\frac{i}{2}(F^\dagger F+GG^\dagger)

1144: \ea\right]~~.

1145: \ee

1146: Here $\rho_0^i d\tau_a$ is the pull-back of $\rho^i$ to $C_a$ and:

1147: \be

1148: \label{calApm} 

1149: {\hat {\cal A}}^{(\pm)}:=A_{\bar i}^{(\pm)}{\dot \phi}^{\bar i} +\frac{1}{2}\eta^{\bar i}  

1150: F^{(\pm)}_{{\bar i} j}\rho_0^j

1151: \ee

1152: are connections on the bundles ${\cal E}_\pm$ obtained by pulling back

1153: $E_\pm$ to the boundary of $\Sigma$. The dot in (\ref{calApm}) stands for the

1154: derivative $\frac{d}{d\tau_a}$. Notice that $\nabla_i F=\partial_i F$ and

1155: $\nabla_i G=\partial_i G$ since $A$ is a $(0,1)$-connection.

1156: 

1157: We have:

1158: \be

1159: \label{M_dec}

1160: M={\hat {\cal A}}+\Delta+K

1161: \ee

1162: where: 

1163: \be

1164: \Delta:=\frac{1}{2}\rho_0^i\partial_i D~~,

1165: \ee

1166: \be

1167: K:=\frac{i}{2}\left(\eta^{\bar i}\nabla_{\bar i}D^\dagger +[D,D^\dagger]_+ \right)

1168: \ee

1169: and:

1170: \be

1171: {\hat {\cal A}}={\dot \phi}^{\bar i}A_{\bar i}+\frac{1}{2}F_{{\bar

1172:     i}j}\eta^{\bar i}\rho^j_0~~.

1173: \ee

1174: Here  $A$ is the direct sum connection on $End(E)$ introduced in

1175: (\ref{AD}). The first two terms in (\ref{M_dec}) agree with \cite{coupling},

1176: while the last term $K$ is added for comparison with \cite{Kap2}. As we shall

1177: see below, this term preserves BRST-invariance of the partition function

1178: (which is already preserved by the sum of the first two terms

1179: \cite{coupling}). As for the open B-model,

1180: adding $K$ insures invariance of the boundary coupling with respect to 

1181: the second generator $\delta_0$ of the $N=2$ topological algebra, thereby

1182: fixing an ambiguity familiar form Hodge theory\footnote{

1183: The symmetry generators $\delta$ and $\delta_0$ can be viewed as analogues of

1184: the operators ${\bar \partial}$ and ${\bar \partial}^\dagger$ of Hodge theory, as already pointed out in \cite{Witten_mirror}

1185: in the context of twisted B-models. The boundary coupling of \cite{coupling}

1186: is chosen to preserve BRST invariance of the partition function. This is

1187: ambiguous up to addition of 'exact' terms, an ambiguity which we can

1188: fix by requiring $\delta_0$-invariance of the partition function. }. 

1189: This modification has minor effects which can be safely

1190: ignored for most purposes 

1191: \footnote{As we shall see in the next section, the extra-term in the boundary

1192:   coupling can be used to 

1193: introduce a parameter $\mu$ characterizing boundary localization pictures. 

1194: For most practical purposes, this parameter can be set to zero, which amounts

1195: to neglecting the last term in (\ref{M_dec}). In

1196: particular, one must set $\mu$ to zero in order to recover the trace

1197: formula of \cite{Kap2}. It is the {\em bulk} parameter $\lambda$ which must be

1198: taken to infinity in order to recover the proposal of \cite{Kap2}.}. 

1199: 

1200: \subsection{The target space equations of motion}

1201: 

1202: To insure BRST invariance of the partition function 

1203: (\ref{Z}), we must choose the background superconnection

1204: ${\cal B}$ such that:

1205: \be

1206: \label{invar}

1207: \delta {\cal U}_a = \frac{1}{2}\left[\int_{C_a}{d\tau

1208:     \rho^i_0\partial_i W} \right] {\cal U}_a~~.

1209: \ee

1210: In this paper, we also require $\delta_0$-invariance of the partition function on the flat

1211: strip:

1212: \be

1213: \label{invar_0}

1214: \delta_0 {\cal U}_a = -\frac{1}{4}\left[\int_{C_a}{d\tau

1215:     \eta^{\bar i}\partial_i {\bar W}} \right] {\cal U}_a~~.

1216: \ee

1217: It is not hard to check the relations:

1218: \be

1219: \label{Qhol}

1220: \delta {\cal U}_a=-\Str \left[~I_a(\delta M)

1221:   Pe^{-\oint_{C_a}{d\tau_a M}}\right]

1222: \ee

1223: where:

1224: \bea

1225: I_a(\delta M)&=&\oint_{C_a}d\tau_a 

1226:   U_a^{-1}\Big(F_{{\bar i}{\bar j}}\eta^{\bar i}{\dot \phi}^{\bar

1227:       j}-\frac{1}{4} \nabla_k F_{{\bar i}{\bar j}}\eta^{\bar i}\eta^{\bar

1228:       j}\rho_0^k-{\dot \phi}^{\bar

1229:       i}\nabla_{\bar

1230:       i}D-\frac{1}{2}\rho^i_0\nabla_i(D^2)+\frac{1}{2}\eta^{\bar

1231:       i}\rho^j_0\nabla_j\nabla_{\bar i}D+\nn\\

1232: &&+\frac{1}{2}\eta^{\bar i}\eta^{\bar j}[F_{{\bar i}{\bar

1233:   j}},D^\dagger]+\eta^{\bar i}[\nabla_{\bar i}D,D^\dagger]+[D^2,D^\dagger]

1234: \Big)U_a~,~~~~\nn

1235: \eea

1236: and:

1237: \be

1238: \label{delta_0hol}

1239: \delta_0 {\cal U}_a=-\Str \left[~I_a(\delta_0 M)

1240:   Pe^{-\oint_{C_a}{d\tau_a M}}\right]

1241: \ee

1242: where:

1243: \bea

1244: I_a(\delta_0 M)&=&\frac{1}{4}\oint_{C_a}{d\tau_a 

1245: U_a^{-1}\left(\eta^{\bar i}\nabla_{\bar i}(D^\dagger)^2+[D,(D^\dagger)^2]\right)U_a}\nn\\

1246: &+&\frac{i}{2}\oint_{C_a}{d\tau_a 

1247: U_a^{-1}\left( -{\dot \phi}^i \nabla_i D^\dagger +\frac{1}{2}\eta^{\bar i}\nabla_{\bar

1248:   i}\nabla_j D^\dagger \rho_0^j+\frac{1}{2}\rho_0^i [D,\nabla_i D^\dagger]

1249:   \right) U_a}~~.~~~~\nn

1250: \eea

1251: Here $U_a(\tau_a)\in GL(r_+|r_-)$ is a certain invertible

1252:   operator\footnote{This should not be confused with the homotopy flow of

1253:   Subsection \ref{bulk_flow} !}) which plays

1254: the role of `parallel transport' defined by $M$ along $C_a$ (see

1255: \cite{coupling} for details). Namely $U_a(\tau_a)=U_a(\tau_a, 0)$, where:  

1256: \be

1257: \label{parr_tr}

1258: U_a(\tau_2,\tau_1):=Pe^{-\int_{\tau_1}^{\tau_2} M(\tau)d\tau}~~

1259: \ee

1260: if $\tau_2>\tau_1$. The origin of the proper length coordinate $\tau_a$ 

1261: along $C_a$ is chosen arbitrarily, while the orientation on $C_a$ is

1262: compatible with that of $\Sigma$. 

1263: The quantities   $F_{{\bar i}{\bar j}}$ etc. are the 

1264: $(0,2)$-components of the curvature of the direct sum connection $A$ introduced in (\ref{AD}).

1265: Notice the relations: 

1266: \be

1267: {\cal U}_a=\Str H_a(\tau)~~

1268: \ee

1269: where: 

1270: \be

1271: \label{holonomy}

1272: H_a(\tau)=U(\tau+l_a,\tau)

1273: \ee 

1274: are the "superholonomy operators" (here $l_a$ the length of $C_a$).

1275: 

1276: 

1277: Relations (\ref{invar}, \ref{invar_0}) and (\ref{Qhol},\ref{delta_0hol}) 

1278: show that the BRST and $\delta_0$-invariance conditions amount to:

1279: \bea

1280: \label{invariance_eqs}

1281: &&F_{{\bar i}{\bar j}}=0\\

1282: &&\nabla_{\bar i} D=0\\

1283: &&\nabla_i(D^2)=\partial_iW\\

1284: &&[D^\dagger,D^2]=0~~.

1285: \eea

1286: The first relation says that $A$ is integrable, so it defines a complex

1287: structure on the bundle $E$. The second condition means that $D\in End(E)$ is

1288: holomorphic with respect to this complex structure. The third equation 

1289: requires $D^2=c+W {\rm id}_E$, with $c$ a covariantly-constant endomorphism. Comparing with

1290: (\ref{02curvature}), we find that these first three conditions are equivalent with:

1291: \bea

1292: \label{eom}

1293: {\cal F}^{(0,\leq 2)}=c+W{\rm id}_E\Longleftrightarrow {\cal {\bar D}}^2=c+W{\rm id}_E~~.

1294: \eea

1295: This is the target space equation of motion for our open string background

1296: \cite{coupling}. Notice that (\ref{eom}) admit solutions only when

1297: $r_+=r_-$.

1298: 

1299: 

1300: For backgrounds satisfying the equation of motion, 

1301: the last condition in (\ref{invariance_eqs}) reads:

1302: \be

1303: [D^\dagger,D^2]=0 \Longleftrightarrow [D^\dagger,c]=0~~. 

1304: \ee

1305: This can be viewed as a partial "gauge-fixing" constraint,

1306: which is fulfilled, for example, if one takes $c$ to be proportional to the

1307: identity endomorphism (in which case the proportionality constant can be absorbed into $W$). 

1308: For simplicity, we shall take $c=0$ for the remainder of this paper. 

1309: 

1310: \section{Boundary observables and correlators}

1311: 

1312: As we  shall see in Section \ref{loc_boundary}, 

1313: the boundary conditions derived from the partition function (\ref{Z}) constrain $\theta$ in terms

1314: of $\eta$ along the boundary of the worldsheet. Hence it suffices to consider

1315: boundary observables of the form:

1316: \be

1317: \label{boundary_observable}

1318: {\cal O}_\alpha(\tau)=\alpha_{{\bar i}_1\dots {\bar i}_p}(\phi(\tau))\eta^{{\bar

1319:     i}_1}(\tau)\dots \eta^{{\bar i}_p}(\tau)~~,

1320: \ee

1321: where $\tau$ is a point on $\partial \Sigma$. Here $\alpha:=\alpha_{{\bar

1322:   i}_1\dots {\bar i}_p}dz^{{\bar i}_1}\wedge \dots \wedge dz^{{\bar i}_p}$ is a $(0,p)$ form

1323: valued in $End(E)$.

1324: 

1325: 

1326: Consider a collection of $m$ topological D-branes described by superbundles

1327: $E_a$ endowed with superconnections ${\cal B}_a$, such that the

1328: target space equations of motion are satisfied. The index $a$ runs from

1329: $1$ to $m$. Let $\Sigma$ be a Riemann

1330: surface with $m$ circle boundary components $C_a$, which we

1331: endow with the orientation induced from $\Sigma$.

1332: Choosing forms

1333: $\alpha^{(a)}_j\in \Omega^{(0,p^{(a)}_j)}(End(E_a))$ and points

1334: $\tau^{(a)}_1\dots \tau^{(a)}_{k_a}$ arranged in increasing cyclic order 

1335: along $C_a$, we are interested in the correlator:

1336: \bea

1337: \label{boundary_corr}

1338: &&\langle \prod_{a=1}^{m}\prod_{j_a=k_a}^{1}

1339: {\cal O}_{\alpha_{j_a}^{(a)}}(\tau^{(a)}_{j_a}) \rangle_\Sigma:=

1340: \int{{\cal D}[\phi]{\cal  D}[F]{\cal D}[\rho]{\cal D}[\eta]{\cal D}[\theta]}

1341: e^{-{\tilde  S}_{bulk}}\\

1342: &&\prod_{a=1}^{m}{\Str\left[

1343: {\cal O}_{\alpha^{(a)}_{k_a}}(\tau_{k_a}^{(a)})

1344: U_a(\tau^{(a)}_{k_a},\tau^{(a)}_{k_a-1})

1345: {\cal O}_{\alpha^{(a)}_{k_a-1}}(\tau^{(a)}_{k_a-1})\dots

1346: {\cal O}_{\alpha^{(a)}_1}(\tau_1^{(a)})

1347: U_a(\tau^{(a)}_1,\tau^{(a)}_{k_a})\right]}~~~,\nn

1348: \eea

1349: where we used the "parallel supertransport" operators defined in

1350: (\ref{parr_tr}). The integration domain in  (\ref{boundary_corr}) is

1351:   specified by the appropriate boundary conditions on the worldsheet fields,

1352:   which will be discussed in more detail below. 

1353: 

1354: Let us first consider a single operator insertion 

1355: ${\cal O}_\alpha$ along a circle boundary component $C$. In this case, the

1356: relevant factor in (\ref{boundary_corr}) is:

1357: \be

1358: \Str[H(\tau){\cal O}_\alpha(\tau)]~~.

1359: \ee

1360: We wish to compute the BRST variation of this quantity. From the relation:

1361: \be

1362: \label{deltaH}

1363: \delta H(\tau)=

1364: \left[\frac{1}{2}\int_{C}{\rho^i\partial_i W}\right] H(\tau)+[H(\tau), D(\tau)+A_{\bar

1365: i}(\tau)\eta^{\bar i}(\tau)]~~

1366: \ee

1367: we obtain:

1368: \be

1369: \label{deltaStr}

1370: \delta \Str[H(\tau){\cal O}_\alpha(\tau)]=

1371: \left[\frac{1}{2}\int_{C}{\rho^i\partial_i W}\right]\Str[H(\tau){\cal

1372:   O}_\alpha(\tau)]+\Str[H(\tau)\delta_b {\cal

1373:   O}_\alpha (\tau)]~~,

1374: \ee

1375: where:

1376: \be

1377: \label{delta_b}

1378: \delta_b{\cal O}_\alpha:=\delta {\cal O}_\alpha+[D+A_{\bar

1379: i}\eta^{\bar i}, {\cal O}_\alpha]~~.

1380: \ee

1381: Using the target space equations of motion, one easily checks that

1382: \footnote{Indeed, one has:

1383: \be

1384: \delta_b^2{\cal O}=\frac{1}{2}[F_{{\bar i}{\bar j}},{\cal O}]+\eta^{\bar

1385:   i}[\nabla_{\bar i}D,{\cal O}]+[D^2,{\cal O}]~~.

1386: \ee}

1387: $\delta_b$ squares to zero, so that it plays the role of an `effective' BRST operator

1388: in the boundary sector. Notice that $\delta_b$ arises naturally due to the

1389: second term in the BRST variation (\ref{deltaH}) of $H(\tau)$. 

1390: Using (\ref{delta_b}), we find the relation:

1391: \be

1392: \label{delta_b_loc}

1393: \delta_b{\cal O}_\alpha={\cal O}_{{\bar D}\alpha}

1394: \ee

1395: where ${\bar D}={\bar D}_{\cal B}$ is the Dolbeault operator on $\Omega^{(0,*)}(End(E))$

1396: twisted by the superconnection ${\cal B}$. 

1397: 

1398: 

1399: It is not hard to generalize (\ref{deltaStr}) to the case of $k$ 

1400: insertions along $C$:

1401: \bea

1402: \label{delta_Str_2}

1403: &&\delta \Str[{\cal O}_{\alpha_k}(\tau_k) U(\tau_k, \tau_{k-1})

1404: \dots {\cal O}_{\alpha_1}(\tau_1)U(\tau_1,\tau_k)]=\\

1405: &&~~~=\left[\frac{1}{2}\int_{C}{\rho^i\partial_i W}\right]

1406: \Str[{\cal O}_{\alpha_k}(\tau_k) U(\tau_k, \tau_{k-1})

1407: \dots {\cal O}_{\alpha_1}(\tau_1)U(\tau_1,\tau_k)]+\nn\\

1408: && ~~~~~~~+\sum_{j=1}^{k}{\Str[

1409: {\cal O}_{\alpha_k}(\tau_k)\dots 

1410: U(\tau_{j+1},\tau_j)\delta_b{\cal O}_{\alpha_j}(\tau_j)U(\tau_j,\tau_{j-1})

1411: \dots {\cal O}_{\alpha_1}(\tau_1)U(\tau_1,\tau_k) ]}~~.\nn

1412: \eea

1413: Applying this to (\ref{boundary_corr}), we find that the BRST variation of

1414: $e^{-{\tilde S}_{bulk}}$ is canceled by the first contribution in

1415: (\ref{delta_Str_2}), summed over circle boundary components. This gives:

1416: \bea

1417: \label{deltaStr_final}

1418: && \delta \left(e^{-{\tilde  S}_{bulk}}\prod_{a=1}^{m}{\Str\left[

1419: {\cal O}_{\alpha^{(a)}_{k_a}}(\tau_{k_a}^{(a)})

1420: U(\tau^{(a)}_{k_a},\tau^{(a)}_{k_a-1})

1421: \dots {\cal O}_{\alpha^{(a)}_1}(\tau_1^{(a)})

1422: U(\tau^{(a)}_1,\tau^{(a)}_{k_a})\right]}\right)=~~~\\

1423: &=&e^{-{\tilde S}_{bulk}}\sum_{a=1}^m\sum_{j_a=1}^{k_a}

1424: \Str\left[{\cal O}_{\alpha^{(a)}_{k_a}}(\tau_{k_a}^{(a)})

1425: U(\tau^{(a)}_{k_a},\tau^{(a)}_{k_a-1})

1426: \dots \delta_b {\cal O}_{\alpha^{(a)}_{j_a}}(\tau_{j_a}^{(a)})

1427: \dots {\cal O}_{\alpha^{(a)}_1}(\tau_1^{(a)})

1428: U(\tau^{(a)}_1,\tau^{(a)}_{k_a})\right]~~.\nn

1429: \eea

1430: Equation (\ref{deltaStr_final}) replaces the more familiar formula known from the open

1431: topological sigma model. Unlike the sigma model case, the left hand side 

1432: includes the factor $e^{-{\tilde S}_{bulk}}$, because its BRST variation does

1433: not vanish separately. Equation (\ref{deltaStr_final}) implies that the 

1434: correlator of $\delta_b$-closed boundary observables only depends on their

1435: $\delta_b$-cohomology class, and in particular such a correlator vanishes if

1436: one of the boundary observables is $\delta_b$-exact. Remember that ${\tilde S}_{bulk}={\tilde

1437:   S}_B+S_W$.

1438: BRST closure of ${\tilde S}_B$ implies that (\ref{deltaStr_final}) is equivalent with:

1439: \bea

1440: \label{deltaStr_modified}

1441: && \delta \left(e^{-S_W}\prod_{a=1}^{m}{\Str\left[

1442: {\cal O}_{\alpha^{(a)}_{k_a}}(\tau_{k_a}^{(a)})

1443: U(\tau^{(a)}_{k_a},\tau^{(a)}_{k_a-1})

1444: \dots {\cal O}_{\alpha^{(a)}_1}(\tau_1^{(a)})

1445: U(\tau^{(a)}_1,\tau^{(a)}_{k_a})\right]}\right)=~~~\\

1446: &=&e^{-S_W}\sum_{a=1}^m\sum_{j_a=1}^{k_a}

1447: \Str\left[{\cal O}_{\alpha^{(a)}_{k_a}}(\tau_{k_a}^{(a)})

1448: U(\tau^{(a)}_{k_a},\tau^{(a)}_{k_a-1})

1449: \dots \delta_b {\cal O}_{\alpha^{(a)}_{j_a}}(\tau_{j_a}^{(a)})

1450: \dots {\cal O}_{\alpha^{(a)}_1}(\tau_1^{(a)})

1451: U(\tau^{(a)}_1,\tau^{(a)}_{k_a})\right]~,\nn

1452: \eea

1453: a fact which will be used in Section \ref{loc_boundary}.

1454: 

1455: 

1456: 

1457: \paragraph{Observation} It is easy to extend the discussion above by including

1458: boundary condition changing observables, which in the present context have the

1459: form (\ref{boundary_observable}), but with $\alpha$ an element of 

1460: $\Omega^{(0,p)}(Hom(E_a,E_b))$. In this case, the operator 

1461: (\ref{delta_b}) is replaced by:

1462: \be

1463: \delta_b{\cal O}_\alpha:=\delta {\cal O}_\alpha+

1464: (D^{(b)}+A^{(b)}_{\bar i}\eta^{\bar i}){\cal O}_\alpha

1465: -(-1)^{\rk\alpha}{\cal O}_\alpha (D^{(a)}+A^{(a)}_{\bar i}\eta^{\bar i})

1466: \ee

1467: and ${\bar D}$ in relation (\ref{delta_b_loc}) becomes the Dolbeault operator

1468: on $\Omega^{(0,*)}(Hom(E_a,E_b))$, twisted by the superconnections 

1469: ${\cal B}_a$ and ${\cal B}_b$.

1470: 

1471: 

1472: \section{Localization formula for boundary correlators on the disk}

1473: \label{loc_boundary}

1474: 

1475: 

1476: We next discuss localization in the boundary sector. As for the bulk, we will

1477: proceed by localizing on sigma model zero-modes, thereby extracting a

1478: two-parameter family of localization formulae. The first index of this 

1479: family is the bulk parameter $\lambda$ of Section 3, while second parameter

1480: $\mu$ is associated with the last term in (\ref{M_dec}). These two parameters 

1481: measure the area and circumference length of a worldsheet with disk topology.  Each pair

1482: $(\lambda,\mu)$ defines a localization picture, and a certain

1483: off-shell representation of the boundary trace of \cite{Moore_Segal, Moore,

1484:   CIL1}. As we shall see below, 

1485: the various pictures are again related by a homotopy flow, and in particular

1486: the various representations of the boundary trace agree when reduced to the cohomology of

1487: $\delta_b$. In this approach, the appropriate 

1488: generalization of the residue representation of \cite{Kap2}

1489: is recovered in the limit $\lambda\rightarrow

1490: +\infty$ with $\mu=0$. 

1491: 

1492: The boundary conditions induced by the coupling (\ref{Z}) can be

1493: extracted by studying the Euler-Lagrange variations of the non-local action 

1494: $S_{eff}={\tilde S}_{bulk}-\ln {\cal U}$. These conditions are given

1495: explicitly in Appendix \ref{BCS}, where we also show that they are BRST

1496: invariant modulo the equations of motion for the auxiliary fields $F$. 

1497: The disk correlator of a collection of boundary observables ${\cal

1498:   O}_{\alpha_1}(\tau_1)\dots {\cal O}_{\alpha_k}(\tau_k)$ (with $\tau_1\dots

1499: \tau_k$ arranged in increasing cyclic order along the boundary) 

1500: is obtained by performing the relevant path integral while imposing 

1501: the  boundary conditions, which  cut out a subset ${\cal C}$ in field

1502: configuration space:

1503: \bea

1504: \label{disk_correlator}

1505: &&\langle {\cal O}_{\alpha_k}(\tau_k)\dots {\cal O}_{\alpha_1}(\tau_1)\rangle_{\rm disk}=\\

1506: &&~~~=\int_{\cal C}{{\cal D}[\phi]{\cal D}[{\tilde F}]

1507: {\cal D}[\eta]{\cal D}[\theta]{\cal D}[\rho]~e^{-{\tilde

1508:     S}_{bulk}}\Str[{\cal O}_{\alpha_k}(\tau_k) U(\tau_k, \tau_{k-1})

1509: \dots {\cal O}_{\alpha_1}(\tau_1)U(\tau_1,\tau_k)]}~~.\nn

1510: \eea

1511: In this section, we assume that ${\cal O}_{\alpha_j}$ are $\delta_b$-closed:

1512: \be

1513: \label{alpha_closure}

1514: \delta_b{\cal O}_{\alpha_j}=0\Longleftrightarrow {\bar D}_{\cal B}\alpha_j=0~~.

1515: \ee

1516: 

1517: \subsection{Localization on $B$-model zero-modes}

1518: 

1519: Remember that ${\tilde S}_{bulk}={\tilde S}_B+S_W$, where ${\tilde S}_B=\delta

1520: V_B$ is BRST exact. As in Section \ref{loc_bulk}, this allows us to replace 

1521: ${\tilde S}_{bulk}$ by $t{\tilde S}_B+S_W$, without changing the value of the

1522: correlator (\ref{disk_correlator}). Here $t$ is a complex variable with

1523: positive real part. Invariance of (\ref{disk_correlator}) under changes in $t$ 

1524: follows by differentiation with respect to this parameter upon using 

1525: $\delta_b$-closure of ${\cal O}_{\alpha_j}$ and equation (\ref{deltaStr_modified})

1526: \footnote{The path integral over ${\cal C}$ of the BRST exact term involved in

1527:   this argument vanishes

1528: because the boundary conditions determining ${\cal C}$ are preserved by the BRST transformations 

1529: up to terms which vanish by the equations of motion for ${\tilde F}$ (see Appendix \ref{BCS}).}. 

1530: 

1531: We can now take the limit $\Re t\rightarrow +\infty$

1532: to localize on the zero modes of ${\tilde S}_B$. This gives:

1533: \be

1534: \label{disk_correlator_loc}

1535: \langle {\cal O}_{\alpha_k}\dots {\cal

1536:   O}_{\alpha_1}\rangle_{\rm disk}=\langle {\cal O}_\alpha \rangle_{\rm

1537:   disk}=N\int_{{\cal C}_0}{d\phi d\eta d\theta

1538: e^{-\frac{A}{4}{\tilde L}_0}\Str[H_0{\cal O}_\alpha]}~~.

1539: \ee

1540: To arrive at this formula, 

1541: we noticed that the dependence of $\tau_j$ disappears on zero-modes, we 

1542: set $\alpha:=\alpha_k\wedge \dots \wedge \alpha_1$ and 

1543: integrated out the auxiliary fields ${\tilde F}$. 

1544: Also notice that $\delta_b {\cal O}_\alpha=0$ due to relations (\ref{alpha_closure}).

1545: The symbol $H_0$ denotes the

1546: restriction of the superholonomy factor $H$ to zero-modes:

1547: \be

1548: H_0=e^{-\frac{il}{2}k_0}~~,

1549: \ee

1550: where $l$ is the length of the disk's boundary and:

1551: \be

1552: \label{k0}

1553: k_0:=\eta^{\bar i}\nabla_{\bar i}D^\dagger +[D,D^\dagger]=\delta_b D^\dagger~~.

1554: \ee

1555: The symbol ${\cal C}_0$ denotes the subset of the space of zero modes cut out

1556: by the boundary conditions. Since we integrated out the auxiliary fields, this 

1557: subset is strictly BRST invariant:

1558: \be

1559: \delta {\cal C}_0\subset {\cal C}_0~~.

1560: \ee

1561: Using this property as well as $\delta$-exactness (\ref{exact_tildeL_0}) of ${\tilde L}_0$

1562: and $\delta_b$-exactness (\ref{k0}) of $k_0$, one checks

1563: \footnote{The proof requires the identity $\delta \Str B=\Str\delta_b B$ for any

1564:   quantity $B$ built out of zero modes. This holds because the supertrace of any

1565:   supercommutator vanishes.}  that

1566: (\ref{disk_correlator_loc}) is insensitive to rescalings of these

1567: quantities, and hence can be replaced with:

1568: \be

1569: \label{disk_correlator_lambda_mu}

1570: \langle {\cal O}_\alpha\rangle_{\rm disk}=N\int_{{\cal C}_0}{d\phi d\eta d\theta

1571: e^{-\lambda{\tilde L}_0}\Str[e^{-\mu k_0}{\cal O}_\alpha]}~~,

1572: \ee

1573: where $\lambda$ and $\mu$ are complex numbers such that $\Re \lambda>0$. The

1574: quantity (\ref{disk_correlator_lambda_mu}) is independent of the values of

1575: these two parameters.

1576: 

1577: To make (\ref{disk_correlator_lambda_mu}) explicit, we must describe the

1578: restriction to ${\cal C}_0$. The relevant boundary condition

1579: takes the form (see Appendix \ref{BCS}):

1580: \be

1581: \label{eta_theta}

1582: i(G_{i{\bar j}}\eta^{\bar j}+\theta_i){\cal U}=\Str[H(\tau)(\partial_i

1583: D+F_{i{\bar j}}\eta^{\bar j})]\Longleftrightarrow \Str[H(\tau)(\theta_i+iV_i)]=0~~,

1584: \ee

1585: where:

1586: \be

1587: V_i:=\partial_iD+(F_{i{\bar j}}-iG_{i{\bar j}})\eta^{\bar j}~~.

1588: \ee

1589: Equation (\ref{eta_theta}) instructs us to replace $\theta_i$ by $-iV_i$ under

1590: the supertrace in order to produce the desired restriction. 

1591: To implement these constraints, we shall use the quantity:

1592: \be

1593: \label{Pi}

1594: \Pi:=\frac{1}{n!}\epsilon^{i_1\dots i_n} (\theta_{i_1}

1595: 1_{End(E)}+iV_{i_1})\dots 

1596: (\theta_{i_n} 1_{End(E)}+iV_{i_n})~~.

1597: \ee

1598: Consider an $End(E)$-valued function $f$ of $\theta_i$: 

1599: \be

1600: f(\theta_1\dots \theta_n)=\sum_{p=0}^n\sum_{1\leq i_1<\dots <i_p\leq

1601:   n}{\theta_{i_1}\dots \theta_{i_p}f^{i_1\dots

1602:     i_p}}=\sum_{p=0}^n\frac{1}{p!}\theta_{i_1}\dots \theta_{i_p}f^{i_1\dots i_p}~~,

1603: \ee

1604: where $f_{i_1\dots i_p}\in End(E)$ with $f^{i_{\sigma(1)}\dots

1605:   i_{\sigma(p)}}=\epsilon(\sigma)f^{i_1\dots i_p}$ for all $\sigma\in

1606: \Sigma_p$ and in the last equality we use implicit summation over $i_1\dots

1607: i_p=1\dots n$. Here $\Sigma_p$ is the group of permutations on $p$ elements, while

1608: $\epsilon(\sigma)$ is the signature of the permutation $\sigma$. Then one

1609: checks the identity:

1610: \be

1611: \label{f_identity}

1612: \int{d\theta_1\dots d\theta_n 

1613: \Pi f(\theta_1\dots \theta_n)}=f(-iV_1\dots -iV_n)~~,

1614: \ee

1615: where the right hand side is defined by:

1616: \bea

1617: f(-iV_1\dots -iV_n)&:=&\sum_{p=0}^n\sum_{1\leq i_1<\dots <i_p\leq

1618:   n}{\frac{1}{p!}\sum_{\sigma\in \Sigma_p}

1619: \epsilon(\sigma)(-iV_{i_{\sigma(1)}})\dots (-iV_{i_{\sigma(p)}}) f^{i_1\dots

1620:   i_p}}\nn\\

1621: &=&\sum_{p=0}^n{\frac{1}{p!}

1622: (-iV_{i_1})\dots (-iV_{i_p}) f^{i_1\dots

1623:   i_p}}~~.

1624: \eea

1625: For later reference, we note the case $f=e^{q^i\theta_i}$, with $q^i$ some

1626: Grassmann-odd quantities depending on $\phi$ and $\eta$. 

1627: Then we have  $f_{i_1\dots i_p}=(-1)^{p(p+1)/2}q^{i_1}\dots q^{i_p}$

1628: and find $f(-iV_1\dots -iV_n)=e^{-iq^i V_i}$, using the fact that

1629: $q_j$ are mutually anti-commuting. This gives:

1630: \be

1631: \label{exp_identity}

1632: \int{d\theta_1\dots d\theta_n e^{q^i\theta_i}\Pi }=\int{d\theta_1\dots d\theta_n 

1633: \Pi e^{q^i\theta_i}}=e^{-iq^iV_i}~~.

1634: \ee

1635: 

1636: 

1637: Relation (\ref{f_identity}) shows that $\Pi$ is a sort of  `Poincare dual' of

1638: ${\cal C}_0$ on the supermanifold of

1639: field configurations. Using (\ref{exp_identity}), this observation allows us to write

1640: (\ref{disk_correlator_lambda_mu}) as an unconstrained integral over the space

1641: of sphere zero-modes :

1642: \be

1643: \label{disk_correlator_loc_bulk}

1644: \langle {\cal O}_\alpha\rangle_{\rm disk}=N\int{d\phi d\eta d\theta

1645: e^{-\lambda{\tilde L}_0}\Str[e^{-\mu k_0} 

1646: \Pi {\cal O}_\alpha ]}~~.

1647: \ee

1648: Employing equation (\ref{exp_identity}), we find:

1649: \be

1650: \label{disk_correlator_loc_boundary}

1651: \langle {\cal O}_\alpha\rangle_{\rm disk}=N\int{d\phi d\eta

1652: \Str[e^{-\mu k_0}e^{-\lambda{\tilde L}_0^b}{\cal O}_\alpha]}~~,

1653: \ee

1654: where:

1655: \bea

1656: \label{tildeL_0b}

1657: {\tilde L}_0^b:={\tilde L}_0|_{\theta_i\rightarrow -iV_i}&=&-D_{\bar i}\partial_{\bar

1658:   j}{\bar W}V^{\bar i}\eta^{\bar j}+G^{i{\bar j}}(\partial_iW)(\partial_{\bar

1659:   j}{\bar W})=\\

1660: &=&{\bar H}^i_{~\bar j}\eta^{\bar j}\partial_iD+\eta^{\bar

1661:   i}\eta^{\bar j}{\bar H}^k_{~{\bar i}}F_{k{\bar j}}+G^{i{\bar j}}(\partial_iW)(\partial_{\bar

1662:   j}{\bar W})~~.\nn

1663: \eea

1664: It is easy to check the relation:

1665: \be

1666: \delta_b V_i=\partial_i W~~,

1667: \ee

1668: which shows that the on-shell BRST variation (\ref{delta_theta_os}) of 

1669: $\theta_i$ agrees with the $\delta_b$-variation of $-iV_i$:

1670: \be

1671: \label{delta_delta_b}

1672: \delta_b(\theta_i 1_{End(E)}+iV_i)=0\Longleftrightarrow (\delta

1673: \theta_i)1_{End(E)}=-i\delta_b V_i~~

1674: \ee

1675: (this in particular implies that $\delta \Pi=0$). Using this equation and

1676: relation (\ref{exact_tildeL_0}), one finds that ${\tilde

1677:   L}_0^b$ is $\delta_b$-exact:

1678: \be

1679: \label{tildeL_0_b_ex}

1680: {\tilde L}_0^b=\delta_b {\tilde v}_0^b~~,

1681: \ee

1682: where:

1683: \be

1684: {\tilde v}_0^b={\tilde v}_0|_{\theta_i\rightarrow -iV_i}=V^{\bar i}\partial_{\bar i}{\bar W}~~.

1685: \ee

1686: Together with $\delta_b$-exactness of $k_0$, this can be used to give a direct

1687: proof of independence of (\ref{disk_correlator_loc_boundary}) of $\lambda$ and $\mu$.

1688: 

1689: We end by mentioning some useful properties of $V_i$. It is

1690: easy to compute the anticommutator:

1691: \be

1692: \label{V_commutator}

1693: [V_i,V_j]_{-}=\partial_i\partial_jW-\delta_b[\partial_i\partial_j (D+A_{\bar k}\eta^{\bar k})]~~.

1694: \ee

1695: Moreover, it is not hard to check the identity:

1696: \be

1697: \label{VO}

1698: [V_i, {\cal O}]=\partial_i(\delta_b{\cal O})-\delta_b(\partial_i{\cal O})

1699: \ee

1700: for any boundary observable ${\cal O}$ (as usual,  the quantity on the left hand

1701: side is a supercommutator). In particular, a $\delta_b$-closed

1702: boundary observable supercommutes with $V_i$ up to a $\delta_b$-exact term. 

1703: 

1704: \subsection{The space of boundary observables}

1705: \label{boundary_geom}

1706: 

1707: After reduction to zero-modes, each boundary insertion ${\cal O}_\alpha$ can

1708: be identified with the superbundle-valued differential form $\alpha$. This

1709: amounts to setting

1710: $\eta^{\bar i}\equiv dz^{\bar i}$, so the superalgebra ${\cal

1711:   H}_b:=\Omega^{(0,*)}(End(E))$ provides an off-shell model for the space of boundary

1712: excitations. Moreover, equation (\ref{delta_b_loc})

1713: identifies the boundary BRST operator $\delta_b$ with the operator 

1714: ${\bar D}_{\cal B}={\bar \nabla}_A+{\cal D}$ acting on ${\cal H}_b$,  

1715: where ${\bar \nabla}_A$ acts in the adjoint representation and 

1716: ${\cal D}=[D,\cdot]$. The

1717: target space equations of motion imply that $\delta_b$ squares to zero. Thus 

1718: ${\cal H}_b$ is a differential superalgebra.  To be precise, 

1719: we take ${\cal H}_b$ to consist of bundle-valued differential forms with at

1720: most polynomial growth at infinity. 

1721:  

1722: The target space equations of motion imply the relations:

1723: \be

1724: {\bar \partial}_A^2={\cal D}^2={\bar \partial}_A\circ {\cal D}+{\cal D}\circ

1725: {\bar \partial}_A=0~~,

1726: \ee

1727: which show that ${\cal H}_b$ is a bicomplex. Thus the boundary BRST cohomology is 

1728: computed by a spectral sequence $E^b_*$ whose second term has the form:

1729: \be

1730: \label{Eb}

1731: E^b_2=H_{\cal D}(H_{{\bar \partial}_A}({\cal H}_b))~~.

1732: \ee

1733: 

1734: 

1735: In the simple case $X=\C^n$, the holomorphic bundle  $E$ is the trivial

1736: superbundle of type $(r_+,r_-)$ and the spectral 

1737: sequence collapses to its second term. Then the BRST cohomology 

1738: coincides with the cohomology of ${\cal D}$ taken in the space of square matrices of dimension

1739: $r_++r_-$ whose entries are polynomial functions of $n$ complex variables.

1740: This recovers the result of \cite{Kap1}.

1741: 

1742: 

1743: 

1744: \subsection{The boundary-bulk and bulk-boundary maps}

1745: \label{bulk_bdry}

1746: 

1747: 

1748: The equivalent expressions (\ref{disk_correlator_loc_bulk}) and 

1749: (\ref{disk_correlator_loc_boundary}) allow us to extract an off-shell

1750: version of the boundary-bulk map of \cite{CIL1}:

1751: \be

1752: f_\mu({\cal O}_\alpha)=\Str[e^{-\mu k_0}\Pi {\cal O}_\alpha]~~.

1753: \ee

1754: This maps ${\cal H}_b$ to ${\cal H}$ and obeys:

1755: \be

1756: \langle {\cal O}_\alpha\rangle_{\rm disk}=\langle f_\mu({\cal

1757:   O}_{\alpha })\rangle_{\rm sphere}\Longleftrightarrow \Tr_b \alpha =\Tr

1758: f_\mu(\alpha)~~,

1759: \ee

1760: where we identified $\alpha$ with ${\cal O}_\alpha$. Here 

1761: $\Tr$ and $\Tr_b$ are the bulk and boundary traces determined by the 

1762: localization formulae (\ref{correlator_final}) and (\ref{disk_correlator_loc_boundary}):

1763: \bea

1764: \Tr\omega~&=&\langle {\cal O}_\omega\rangle_{\rm sphere}~=\int{d\phi d\eta d\theta e^{-\lambda {\tilde L}_0}{\cal

1765:     O}_\omega}\label{Tr}\\

1766: \Tr_b\alpha &=&\langle  {\cal O}_\alpha\rangle_{\rm disk}=

1767: \int{d\phi d\eta \Str[e^{-\mu k_0}e^{-\lambda {\tilde L}^b_0}{\cal

1768:     O}_\alpha]}~~.\label{Tr_o}

1769: \eea

1770: As in \cite{CIL1}, we can also define a bulk-boundary map $e$ through the adjunction formula:

1771: \be

1772: \label{adjunction}

1773: \Tr({\cal O}_\omega f_\mu({\cal O}_\alpha))=\Tr_b(e({\cal O}_\omega) {\cal O}_\alpha)~~.

1774: \ee

1775: From the relations above, we find:

1776: \be

1777: e({\cal O}_\omega):=e^{\lambda {\tilde L}_0^b}\int{d\theta_1\dots d\theta_n

1778:   e^{-\lambda {\tilde L}_0}{\cal O}_\omega \Pi}~~.

1779: \ee

1780: This maps ${\cal H}$ to ${\cal H}_b$. 

1781: 

1782: Using (\ref{delta_delta_b}), we find that $f_\mu$ and $e$ are

1783: compatible with the bulk and boundary BRST operators:

1784: \bea

1785: \delta\circ f_\mu &=& (-1)^n f_\mu\circ \delta_b ~~\label{delta_f}\\

1786: \delta_b \circ e ~&=& e \circ \delta~~.\label{delta_e}

1787: \eea

1788: To prove the second equation, we used the identity:

1789: \be

1790: \delta\int{d\theta_1\dots d\theta_n f(\theta_1\dots

1791:   \theta_n)}=\int{d\theta_1\dots d\theta_n \delta f(\theta_1\dots \theta_n)}~~,

1792: \ee

1793: which follows from (\ref{delta_theta_os}). 

1794: Relations (\ref{delta_f}) and (\ref{delta_e}) show that 

1795: $e$ and $f_\mu$ descend to well-defined maps $e_*$ and $f_*$ between the bulk and

1796: boundary BRST cohomologies (the latter are the maps considered in \cite{CIL1}). 

1797: Since $k_0$ is $\delta_b$-exact, one easily checks that $f_*$ is independent of $\mu$.

1798: 

1799: \subsection{A geometric model for the boundary trace}

1800: 

1801: As in Section \ref{loc_bulk}, we can use the identifications $\eta^{\bar

1802:   i}\equiv dz^{\bar i}$ and $\theta_i\equiv \partial_i$ to represent our formulae in

1803:   terms of standard geometric objects. We find:

1804: \be

1805: {\tilde L}_0^b\equiv{\bar H}\lrcorner(\partial D+F)+||\partial W||^2=\delta_b

1806:   {\tilde v}_0^b~~

1807: \ee

1808: with:

1809: \be

1810: {\tilde v}_0^b={\rm grad} {\bar W}\lrcorner ~(\partial D+F)~~

1811: \ee

1812: and:

1813: \be

1814: k_0={\bar \nabla}_{\cal B}D^\dagger={\bar \nabla}_A D^\dagger+[D,D^\dagger]~~.

1815: \ee

1816: The disk localization formula (\ref{disk_correlator_loc_boundary}) becomes:

1817: \be

1818: \label{Tr_bdry_geom}

1819: \Tr_b \alpha =N\int_{X}{\Omega\wedge \str[e^{-\lambda {\tilde L}_0^b}\wedge e^{-\mu

1820:   k_0} \wedge \alpha]}~~, 

1821: \ee

1822: while the quantity (\ref{Pi}) takes the form:

1823: \be

1824: \Pi=\frac{1}{n!}(\partial_1+iV_1)\wedge \dots \wedge (\partial_n+iV_n)~~,

1825: \ee

1826: with:

1827: \be

1828: V_i=\partial_i D+(F_{i{\bar j}}-iG_{i{\bar j}})dz^{\bar j}~~.

1829: \ee

1830: Defining $V=dz^i \otimes V_i$, we obtain:

1831: \be

1832: V=\partial D+ F-iG~~,

1833: \ee

1834: where $G:=G_{i{\bar j}}dz^{i}\otimes dz^{\bar j}$. Notice that here and above,

1835:   $\partial D$ is defined by $\partial D=dz^i\otimes \partial_i D$ (the order

1836:   matters since $D$ is odd). 

1837: 

1838: 

1839: \subsection{Boundary localization pictures and the homotopy flow}

1840: 

1841: As for the bulk sector, one can define a two-parameter semigroup of operators

1842: acting on ${\cal H}_b$ through:

1843: \be

1844: U_b(\lambda, \mu)(\alpha):=e^{-\lambda {\tilde L}_0^b}\wedge e^{-\mu

1845:   k_0} \wedge \alpha~~. 

1846: \ee

1847: The pair $(\lambda, \mu)$ is taken inside the domain: 

1848: \be

1849: \Delta_b:=\{(\lambda,\mu)\in \C^2 | \Re \lambda>0 \}~~.

1850: \ee

1851: Since both ${\tilde L}_0$ and ${\tilde L}_0^b$ are BRST-exact, each

1852: $U_b(\lambda,\mu)$ is homotopy-equivalent with the identity so

1853: this defines a homotopy flow. We let: 

1854: \be

1855: \Tr_b^B(\alpha):=N\int_{X}{\Omega\wedge \str(\alpha)}

1856: \ee

1857: denote the boundary trace of the B-twisted sigma model (viewed as a linear

1858: functional on the off-shell state space ${\cal H}_b$). Then equation (\ref{Tr_bdry_geom})

1859: becomes: 

1860: \be

1861: \Tr_b(\alpha)=\Tr_b^B(U_b(\lambda,\mu)(\alpha))~~. 

1862: \ee

1863: Again one can define localization pictures indexed by $\lambda$ and

1864: $\mu$. The boundary BRST operator satisfies: 

1865: \be

1866: U_b(\lambda,\mu)\circ\delta_b=\delta_b \circ U_b(\lambda, \mu)~~. 

1867: \ee

1868: 

1869: 

1870: 

1871: \subsection{Residue formula for boundary correlators on the disk}

1872: 

1873: As in Section \ref{loc_bulk}, we can use equation (\ref{disk_correlator_loc_bulk})

1874: to express boundary correlators in terms of generalized residues. For

1875: simplicity, we shall assume that the spectral sequence of Subsection \ref{boundary_geom}

1876: collapses to its second term. Setting $\mu=0$ in (\ref{disk_correlator_loc_bulk}) gives:

1877: \be

1878: \label{step1}

1879: \langle {\cal O}_\alpha\rangle_{\rm disk}=N\int{d\phi d\eta d\theta

1880: e^{-\lambda{\tilde L}_0}\Str[\Pi {\cal O}_\alpha]}~~.

1881: \ee

1882: We next take the limit $\lambda\rightarrow +\infty$ with $\lambda \in \R_+$. 

1883: As in Section \ref{loc_bulk}, this forces the integral to localize on the

1884: critical points of $W$, while the Gaussian approximation around these points

1885: becomes exact. Counting the powers of $\lambda$ produced by the bosonic and

1886: fermionic Gaussian integrals, we find that the correlator vanishes unless

1887: $\alpha=f$ with $f$ a section of $End(E)$. In this case, $\delta_b$-closure of

1888: ${\cal O}_f$ amounts to the conditions ${\bar \nabla}_Af=0$ and $[D,f]=0$, 

1889: and counting powers of $\lambda$ shows that the only contributions which

1890: survive in the limit come from those pieces of the factor 

1891: $\Pi$

1892: which are independent of $\theta$ and $\eta$. This gives:

1893: \bea

1894: \langle {\cal O}_\alpha\rangle_{\rm disk}&=&0~~{\rm unless}~~\alpha=f\in

1895: End(E)~~\\

1896: \langle {\cal O}_f\rangle_{\rm disk}&=& 

1897: \frac{C}{n!}\int_{X}{

1898: \Omega \frac{\Str[(i\partial D)^{\wedge n}f(z)]}{\partial_1W\dots \partial_nW}}~~,

1899: \eea

1900: where $C$ is the constant introduced in Section \ref{loc_bulk}. These

1901: expressions generalize the residue formula proposed in \cite{Kap2}. Notice,

1902: however, that the residue formula of \cite{Kap2} arises for $\mu=0$ and in the

1903: limit $\Re \lambda\rightarrow +\infty$. The limit proposed in \cite{Kap2} 

1904: (namely $\Re\mu\rightarrow +\infty$ with $\lambda=0$) does not suffice to

1905: localize the model's excitations unto the critical set of $W$. 

1906: 

1907: 

1908: \section{Conclusions}

1909: 

1910: 

1911: We gave a detailed and general discussion of localization in the bulk and boundary sectors

1912: of B-type topological Landau-Ginzburg models. In the bulk sector, we showed

1913: that careful reconsideration of the localization argument of \cite{Vafa_LG}

1914: leads to an entire family of localization formulae, parameterized by a complex

1915: number $\lambda$ of positive real part. When real, this parameter 

1916: measures the area of worldsheets with $S^2$ topology.

1917: The various "localization pictures" are related by a "homotopy flow"

1918: (a semigroup of operators homotopic to the identity), which implements

1919: rescalings of this area. The generalized

1920: localization argument leads to a one-parameter family of off-shell models for

1921: the bulk trace, extending the well-know result of \cite{Vafa_LG}. 

1922: The later is recovered for $\Re \lambda\rightarrow +\infty$, a degenerate

1923: limit which leads to the standard residue representation. 

1924: 

1925: In the boundary sector, a similar argument gives a family of localization

1926: formulae parameterized by complex variables $\lambda$ and $\mu$ subject to the

1927: condition $\Re \lambda>0$. When real, these parameters describe the area of a worldsheet with

1928: disk topology, respectively the length of its boundary.

1929: The boundary localization pictures are once again related by 

1930: a semigroup of homotopy equivalences, which implements rescaling of the disk's

1931: area and of the length of its boundary. This leads to a two-parameter family

1932: of off-shell models for the boundary trace. We also showed that the residue

1933: formula proposed in \cite{Kap2} arises in the limit $\lambda\rightarrow

1934: +\infty$ with $\mu=0$, and generalizes to the set-up of \cite{Labastida,

1935:   coupling}, which does not require constraints on the target space or on the

1936: rank of the holomorphic superbundle describing the relevant D-brane. In

1937: particular, this proves and generalizes the proposal of

1938: \cite{Kap2}, though the residue representation we have found arises in a limit which differs

1939: from previous proposals. The argument required to establish this result 

1940: is rather subtle, due to the complicated form of the boundary conditions.

1941: 

1942: \acknowledgments{The authors thank W. Lerche for collaboration in a related

1943:   project. C.~I.~L. thanks A. Kapustin and K. Hori for stimulating

1944:   conversations, and A. Klemm for support and interest in his work. }

1945: 

1946: 

1947: 

1948: \appendix

1949: 

1950: 

1951: \section{Euler-Lagrange variations and boundary conditions}

1952: \label{BCS}

1953: 

1954: Let us consider the Euler-Lagrange variations for our model.

1955: It is not hard to compute the variations of the bulk action:

1956: \bea

1957: \delta_\theta {\tilde S}_{bulk}&=&\frac{i}{2}\int_{\Sigma}{d^2\sigma \sqrt{g}

1958: \left[\varepsilon^{\alpha\beta}D_\beta\rho^i_\alpha+\frac{1}{2}G^{i{\bar

1959:       k}}D_{\bar

1960:   k}\partial_{\bar j} {\bar W}\eta^{\bar j}\right]\delta \theta_i}~~\label{bulk_EL_vars1}

1961: \\

1962: \delta_\eta {\tilde S}_{bulk}&=&\frac{1}{2}\int_{\Sigma}{d^2\sigma \sqrt{g}

1963: \left[g^{\alpha\beta} G_{{\bar i}j}D_\alpha\rho^j_\beta-\frac{i}{2}D_{\bar

1964:   i}\partial_{\bar j} {\bar W}\theta^{\bar j}\right]\delta\eta^{\bar

1965: i}}-\frac{1}{2}\int_{\partial\Sigma}{d\tau G_{{\bar i}j}(\rho_n^j+i\rho_0^j)\delta\eta^{\bar

1966: i}}~~~~~~~~~~\label{bulk_EL_vars2}\\

1967: \delta_\rho {\tilde S}_{bulk}&=&\frac{1}{2}\int_{\Sigma}{d^2\sigma \sqrt{g}

1968: \left[G_{i{\bar j}}

1969: (i\varepsilon^{\alpha\beta}D_\beta\theta^{\bar

1970:     j}+g^{\alpha\beta}D_\beta\eta^{\bar j})

1971: -\frac{1}{2}\varepsilon^{\alpha\beta}D_i\partial_j W

1972:   \rho_\beta^j\right]\delta \rho_\alpha^i}~~\nn\\

1973: &+&\frac{i}{2}\int_{\partial\Sigma}{d\tau

1974: G_{i{\bar j}}(\eta^{\bar j}+\theta^{\bar j})\delta \rho_0^i}~~\label{bulk_EL_vars3}\\

1975: \delta_\phi {\tilde S}_{bulk}&=&

1976: \int_{\Sigma}{d^2\sigma \sqrt{g}\left[

1977: -G_{{\bar i}j}\Delta\phi^j-\frac{i}{4}D_{\bar i}D_{\bar

1978:       j}\partial_{\bar k}{\bar W}\theta^{\bar j}\eta^{\bar

1979:       k}+\frac{i}{2}D_{\bar i}\partial_{\bar j}{\bar W}{\tilde F}^{\bar j}

1980: \right]\delta \phi^{\bar i}}\nn\\

1981: &+&\int_{\Sigma}{d^2\sigma \sqrt{g}\left[

1982: -G_{i{\bar j}}\Delta\phi^{\bar j}-\frac{i}{2}D_i\partial_j W {\tilde F}^j

1983: +\frac{1}{8}\varepsilon^{\alpha\beta}D_iD_j\partial_k W \rho_\alpha^j\rho_\beta^k

1984: \right]\delta \phi^i}\nn\\

1985: &+&\int_{\partial\Sigma}{d\tau G_{{\bar i}j}(\partial_n\phi^j+i{\dot \phi}^j)\delta

1986:   \phi^{\bar i}}+\int_{\partial\Sigma}{d\tau G_{i{\bar

1987:       j}}(\partial_n\phi^{\bar j}-i{\dot \phi}^{\bar j})\delta

1988:   \phi^i}~~.\label{bulk_EL_vars4}

1989: \eea

1990: For the boundary coupling ${\cal U}=Str H(0)$, we find:

1991: \be

1992: \delta {\cal U}=-\Str[H(0)I_C(\delta M)]~~,

1993: \ee

1994: where:

1995: \be

1996: I_C(\delta M)=\int_{0}^l{d\tau U(\tau)^{-1}\delta M(\tau) U(\tau)}~~.

1997: \ee

1998: For $\delta_M$ we substitute the Euler-Lagrange variations:

1999: \bea

2000: \delta_\theta M&=&0\\

2001: \delta_\eta M&=&-\left(\frac{1}{2}F_{{\bar

2002:       i}j}\rho_0^j+\frac{i}{2}\nabla_{\bar i}D^\dagger\right)\delta\eta ^{\bar i}\\

2003: \delta_\rho M&=&-\frac{1}{2}(\partial_iD+F_{i{\bar j}}\eta^{\bar j})\delta \rho_0^i~~

2004: \eea

2005: and: 

2006: \be

2007: U^{-1}\delta_\phi M U=\frac{d}{d\tau}(U^{-1}A_{\bar i}\delta \phi^{\bar

2008:   i}U)+U^{-1}(S_i\delta \phi^i+S_{\bar i}\delta \phi^{\bar i})U

2009: \ee

2010: with:

2011: \bea

2012: S_i&=&F_{i{\bar j}}{\dot \phi}^{\bar j}+\frac{1}{2}\partial_iF_{{\bar

2013:     j}k}\eta^{\bar j}\rho_0^k+\frac{1}{2}\rho_0^j\partial_i\partial_j

2014: D+\frac{i}{2}\eta^{\bar j}[F_{i{\bar j}},D^\dagger]+\frac{i}{2}[\partial_iD,D^\dagger]\\

2015: S_{\bar i}&=&F_{{\bar i}j}{\dot \phi}^j+\frac{1}{2}\nabla_{\bar i}F_{{\bar

2016:     j}k}\eta^{\bar j}\rho_0^k+\frac{1}{2}\rho_0^j[F_{{\bar

2017:     i}j},D]+\frac{i}{2}\eta^{\bar j}\nabla_{\bar i}\nabla_{\bar j}D^\dagger

2018: +\frac{i}{2}[D,\nabla_{\bar i}D^\dagger]~~.

2019: \eea

2020: This gives:

2021: \bea

2022: \delta_\theta {\cal U}&=&0\\

2023: \delta_\eta {\cal

2024:   U}&=&\int_{\partial\Sigma}{d\tau \Str[H(\tau)(\frac{i}{2}\nabla_{\bar

2025:     i}D^\dagger+\frac{1}{2}F_{{\bar i}j}\rho_0^j)]\delta\eta^{\bar i}}\\

2026: \delta_\rho {\cal U}&=& \frac{1}{2}\int_{\partial \Sigma}{d\tau

2027:   \Str[H(\tau)(\partial_iD+F_{i{\bar j}}\eta^{\bar j})]\delta \rho_0^i}\\

2028: \delta_\phi{\cal U}&=&-\int_{\partial \Sigma}{d\tau 

2029: \left(\Str[H(\tau)S_i]\delta\phi^i+\Str[H(\tau)S_{\bar i}]\delta \phi^{\bar i}\right)}~~.

2030: \eea

2031: To extract the boundary conditions, we write:

2032: \be

2033: e^{-{\tilde S}_{bulk}}{\cal U}=e^{-S_{eff}}~~,

2034: \ee

2035: where $S_{eff}={\tilde S}_{bulk}-\ln {\cal U}$ is viewed as a (non-local) worldsheet action. 

2036: Since we desire local equations of motion, the boundary contributions to

2037: (\ref{bulk_EL_vars1}-\ref{bulk_EL_vars4}) must cancel the variation of $\ln {\cal U}$:

2038: \be

2039: {\cal U} \delta{\tilde S}_{bulk}=\delta {\cal U}~~.

2040: \ee 

2041: Imposing this requirement, we find the boundary conditions:

2042: \bea

2043: G_{{\bar i}j}(\rho_n^j+i\rho_0^j){\cal U}&=&-\Str H(\tau)(i\nabla_{\bar i}D^\dagger +

2044: F_{{\bar i}j}\rho_0^j)\label{bc1}\\

2045: i(G_{i{\bar j}}\eta^{\bar j}+\theta_i){\cal U}&=&

2046: \Str H(\tau)(\partial_i D+F_{i{\bar j}}\eta^{\bar j})\label{bc2}\\

2047: G_{{\bar i}j}(\partial_n\phi^j+i{\dot \phi}^j){\cal U} &=&

2048: -\Str [H(\tau)S_{\bar i}]\label{bc3}\\

2049: G_{i{\bar j}}(\partial_n\phi^{\bar j}-i{\dot \phi}^{\bar j}){\cal U}&=&-

2050: \Str [H(\tau)S_i]~~.\label{bc4}

2051: \eea

2052: The Euler-Lagrange equations can be read off from the bulk contributions to 

2053: (\ref{bulk_EL_vars1}-\ref{bulk_EL_vars4}):

2054: \bea

2055: \epsilon^{\alpha\beta}D_\alpha \rho^i_\beta&=&\frac{1}{2}G^{i{\bar j}}D_{\bar

2056:   j}\partial_{\bar k}{\bar W}\eta^{\bar k}\\

2057: g^{\alpha\beta}D_\alpha\rho_\beta^i&=&\frac{i}{2}G^{i{\bar j}}D_{\bar j}\partial_{\bar

2058:   k}{\bar W}\theta^{\bar k}\\

2059: i\epsilon^{\alpha\beta}D_\beta \theta_i+g^{\alpha\beta}G_{i{\bar j}}D_\beta\eta^{\bar

2060:   j}&=&\frac{1}{2}\epsilon^{\alpha\beta}D_i\partial_jW

2061: \rho_\beta^j~~.

2062: \eea

2063: 

2064: It is not hard to see that the boundary conditions are BRST invariant modulo

2065: the equations of motion for $F$. For simplicity, we explain this for 

2066: condition (\ref{bc2}), which is of interest in Section

2067: \ref{loc_boundary}. Starting with (\ref{bc2}),

2068: one easily computes the BRST variations of the left and right hand sides:

2069: \bea

2070: \delta(LHS)&=&2iG_{i{\bar j}}{\tilde F}^{\bar j}{\cal U}+i(G_{i{\bar j}}\eta^{\bar

2071:     j}+\theta_i)\left[\frac{1}{2}\int_{\partial\Sigma}\rho^i \partial_i W\right]{\cal

2072:     U}~~\\

2073: \delta(RHS)&=&(\partial_i W){\cal U}+

2074: \left[\frac{1}{2}\int_{\partial\Sigma}\rho^k \partial_k W\right]

2075: \Str[H(\tau)(\partial_iD+F_{i{\bar

2076:     j}}\eta^{\bar j})]~~.

2077: \eea

2078: The two variations obviously agree if one uses equation (\ref{bc2}),

2079:     provided that the equation of motion ${\tilde F}^{\bar

2080:     i}=-\frac{i}{2}G^{{\bar i}j}\partial_j W$ holds.

2081: 

2082: 

2083: 

2084: 

2085: 

2086: 

2087: 

2088: \begin{thebibliography}{100}

2089: \bibitem{Vafa_LG}{C.~Vafa,

2090: ``Topological Landau-Ginzburg Models",

2091: Mod.\ Phys.\ Lett.\ A {\bf 6}, 337 (1991).}

2092: %%CITATION = MPLAE,A6,337;%%

2093: \bibitem{Witten_phases}{E.~Witten,

2094: ``Phases of N = 2 theories in two dimensions,''Nucl.\ Phys.\ B {\bf 403}, 159 (1993)

2095: [arXiv:hep-th/9301042].}

2096: %%CITATION = HEP-TH 9301042;%%

2097: \bibitem{Hori}{K.~Hori, A.~Iqbal and C.~Vafa,

2098: ``D-branes and mirror symmetry,''arXiv:hep-th/0005247.}

2099: %%CITATION = HEP-TH 0005247;%%

2100: \bibitem{Warner}{N.~P.~Warner,

2101: ``Supersymmetry in boundary integrable models,''

2102: Nucl.\ Phys.\ B {\bf 450}, 663 (1995)[arXiv:hep-th/9506064].}

2103: %%CITATION = HEP-TH 9506064;%%

2104: \bibitem{Kontsevich}{M. Kontsevich, {\em unpublished}.}

2105: \bibitem{Kap1}{A.~Kapustin and Y.~Li,

2106: ``D-branes in Landau-Ginzburg models and algebraic geometry,''

2107: JHEP {\bf 0312}, 005 (2003)[arXiv:hep-th/0210296].}

2108: %%CITATION = HEP-TH 0210296;%%

2109: \bibitem{Orlov}{D. Orlov, 

2110: "Triangulated categories of singularities and D-branes in Landau-Ginzburg

2111:   models", math.AG/0302304.}

2112: \bibitem{Lerche}{I.~Brunner, M.~Herbst, W.~Lerche and B.~Scheuner,

2113: ``Landau-Ginzburg realization of open string TFT,''

2114: arXiv:hep-th/0305133.}

2115: %%CITATION = HEP-TH 0305133;%%

2116: \bibitem{Kap2}{A.~Kapustin and Y.~Li,

2117: ``Topological correlators in Landau-Ginzburg models with boundaries,''

2118: arXiv:hep-th/0305136.}

2119: %%CITATION = HEP-TH 0305136;%%

2120: \bibitem{Kap3}{A.~Kapustin and Y.~Li,

2121: ``D-branes in topological minimal models: The Landau-Ginzburg approach,''

2122: arXiv:hep-th/0306001.}

2123: %%CITATION = HEP-TH 0306001;%%

2124: \bibitem{fractional}

2125: S.~K.~Ashok, E.~Dell'Aquila and D.~E.~Diaconescu,

2126: ``Fractional branes in Landau-Ginzburg orbifolds,''

2127: arXiv:hep-th/0401135.

2128: %%CITATION = HEP-TH 0401135;%%

2129: \bibitem{coupling}{C.~I.~Lazaroiu,

2130: ``On the boundary coupling of topological Landau-Ginzburg models,'' arXiv:hep-th/0312286.}

2131: %%CITATION = HEP-TH 0312286;%%

2132: \bibitem{Griffiths}{P. Griffiths and J. Harris: {\em Principles of Algebraic

2133:     Geometry}, Wiley, 1994. }

2134: \bibitem{Labastida}{J.~M.~F.~Labastida and P.~M.~Llatas,

2135: ``Topological matter in two-dimensions,''

2136: Nucl.\ Phys.\ B {\bf 379}, 220 (1992)[arXiv:hep-th/9112051].}

2137: %%CITATION = HEP-TH 9112051;%%

2138: 

2139: 

2140: \bibitem{Witten_CS}{E.~Witten,

2141: ``Chern-Simons gauge theory as a string theory,''

2142: Prog.\ Math.\  {\bf 133}, 637 (1995) [arXiv:hep-th/9207094].}

2143: %%CITATION = HEP-TH 0104200;%%

2144: \bibitem{CIL2}{C.~I.~Lazaroiu,

2145: ``Generalized complexes and string field theory,''

2146: JHEP {\bf 0106}, 052 (2001)[arXiv:hep-th/0102122].}

2147: %%CITATION = HEP-TH 0102122;%%

2148: \bibitem{CIL3}{C.~I.~Lazaroiu,

2149: ``Unitarity, D-brane dynamics and D-brane categories,''

2150: JHEP {\bf 0112}, 031 (2001)[arXiv:hep-th/0102183].}

2151: %%CITATION = HEP-TH 0102183;%%

2152: \bibitem{CIL4}{C.~I.~Lazaroiu,

2153: ``String field theory and brane superpotentials,''JHEP {\bf 0110}, 018 (2001)

2154: [arXiv:hep-th/0107162].}

2155: %%CITATION = HEP-TH 0107162;%%

2156: \bibitem{CIL5}{C.~I.~Lazaroiu,

2157: ``D-brane categories,'' Int.\ J.\ Mod.\ Phys.\ A {\bf 18}, 5299 (2003)

2158: [arXiv:hep-th/0305095].}

2159: %%CITATION = HEP-TH 0305095;%%

2160: \bibitem{CIL6}{C.~I.~Lazaroiu,

2161: ``Graded Lagrangians, exotic topological D-branes and enhanced  triangulated

2162: categories,''JHEP {\bf 0106}, 064 (2001)[arXiv:hep-th/0105063].}

2163: %%CITATION = HEP-TH 0105063;%%

2164: \bibitem{CIL7}{C.~I.~Lazaroiu, R.~Roiban and D.~Vaman,

2165: ``Graded Chern-Simons field theory and graded topological D-branes,''

2166: JHEP {\bf 0204}, 023 (2002) [arXiv:hep-th/0107063].}

2167: %%CITATION = HEP-TH 0107063;%%

2168: \bibitem{CIL8}{C.~I.~Lazaroiu and R.~Roiban,

2169: ``Holomorphic potentials for graded D-branes,''

2170: JHEP {\bf 0202}, 038 (2002)[arXiv:hep-th/0110288].}

2171: %%CITATION = HEP-TH 0110288;%%

2172: \bibitem{CIL9}{C.~I.~Lazaroiu,

2173: ``An analytic torsion for graded D-branes,''

2174: JHEP {\bf 0209}, 023 (2002)[arXiv:hep-th/0111239].}

2175: %%CITATION = HEP-TH 0111239;%%

2176: \bibitem{CIL10}{C.~I.~Lazaroiu and R.~Roiban,

2177: ``Gauge-fixing, semiclassical approximation and potentials for graded

2178: Chern-Simons theories,'' JHEP {\bf 0203}, 022 (2002)[arXiv:hep-th/0112029].}

2179: %%CITATION = HEP-TH 0112029;%%

2180: \bibitem{Diaconescu}{D.~E.~Diaconescu,

2181: ``Enhanced D-brane categories from string field theory,''

2182: JHEP {\bf 0106}, 016 (2001) [arXiv:hep-th/0104200].}

2183: \bibitem{Moore_Segal}{G. Moore and G. Segal, unpublished; 

2184: see http://online.kitp.ucsb.edu/online/mp01/}

2185: \bibitem{Moore}{G.~W.~Moore,

2186: ``Some comments on branes, G-flux, and K-theory,''

2187: Int.\ J.\ Mod.\ Phys.\ A {\bf 16}, 936 (2001)

2188: [arXiv:hep-th/0012007].}

2189: %%CITATION = HEP-TH 0012007;%%}

2190: \bibitem{CIL1}{C.~I.~Lazaroiu,

2191:  ``On the structure of open-closed topological field theory in two dimensions,''

2192: Nucl.\ Phys.\ B {\bf 603}, 497 (2001) [arXiv:hep-th/0010269].}

2193: %%CITATION = HEP-TH 0010269;%%

2194: \bibitem{Witten_mirror}{E.~Witten,

2195: ``Mirror manifolds and topological field theory,'', in " Essays on mirror

2196: manifolds",  pp 120--158, Internat. Press, Hong Kong, 1992.  [arXiv:hep-th/9112056].}

2197: %%CITATION = HEP-TH 9112056;%%

2198: 

2199: 

2200: \bibitem{us}{M.~Herbst, C.~I.~Lazaroiu and W.~Lerche,

2201: ``Superpotentials, A(infinity) relations and WDVV equations for open

2202: topological strings,'' arXiv:hep-th/0402110.}

2203: %%CITATION = HEP-TH 0402110;%%

2204: 

2205: \bibitem{Quillen}{D. Quillen, "Superconnections and the Chern character",

2206: Topology {\bf 24} (1985), no. 1, 89--95.}

2207: \end{thebibliography}

2208: \end{document}

2209: