1: \documentclass[12pt,letter,hyper]{JHEP3}
2: \usepackage{amsmath}
3: \usepackage{amscd}
4: \usepackage{amsfonts}
5: \usepackage{amssymb}
6: \usepackage{amsthm}
7:
8: \usepackage{cite}
9: \usepackage{epsfig}
10:
11: %\renewcommand{\textfraction}{0.20}
12: %\renewcommand{\topfraction}{0.95}
13: %\renewcommand{\bottomfraction}{0.95}
14: %\renewcommand{\floatpagefraction}{0.95}
15: %\setcounter{totalnumber}{7}
16:
17: % SETUP THE PAGE
18:
19: % left 30, right 10, up 20, low 25 210-30
20: %\textwidth 500mm
21: %\textheight 220mm
22: %\topmargin 0mm
23: %\oddsidemargin 0mm
24: %\evensidemargin 0mm
25: %\topskip 0mm
26: %\headsep 0mm
27: %\headheight 0mm
28: %\normalsize
29: %\renewcommand{\baselinestretch}{1.2}
30:
31:
32: %%% BRACKETS
33: \newcommand {\la} {\left \langle}
34: \newcommand {\ra} {\right \rangle}
35: \newcommand {\lb} {\left (}
36: \newcommand {\rb} {\right )}
37: \newcommand {\lfb} {\left \{}
38: \newcommand {\rfb} {\right \}}
39: \newcommand {\lsb} {\left [}
40: \newcommand {\rsb} {\right]}
41: \newcommand{\br}[1]{\left( #1 \right)}
42: \newcommand{\vev}[1]{\left\langle #1 \right\rangle}
43: \newcommand{\ket}[1]{\left | #1 \right \rangle}
44:
45: %%% CAL LETTERS
46: \newcommand {\CalA} {\mathcal A}
47: \newcommand {\CalD} {\mathcal D}
48: \newcommand {\CalE} {\mathcal E}
49: \newcommand {\CalF} {\mathcal F}
50: \newcommand {\CalH} {\mathcal H}
51: \newcommand {\CalI} {\mathcal I}
52: \newcommand {\CalJ} {\mathcal J}
53: \newcommand {\CalO} {\mathcal O}
54: \newcommand {\CalZ} {\mathcal Z}
55: \newcommand {\CalN} {\mathcal N}
56: \newcommand {\CalC} {\mathcal C}
57: \newcommand {\CalL} {\mathcal L}
58: \newcommand {\CalS} {\mathcal S}
59: \newcommand {\CalK} {\mathcal K}
60: \newcommand {\CalM} {\mathcal M}
61: \newcommand {\CalP} {\mathcal P}
62: \newcommand {\CalU} {\mathcal U}
63: \newcommand {\CalX} {\mathcal X}
64: \newcommand {\CalW} {\mathcal W}
65:
66: %%% BOLD LETTERS
67: \newcommand {\BH} {\mathbb H}
68: \newcommand {\BR} {\mathbb R}
69: \newcommand {\BZ} {\mathbb Z}
70: \newcommand {\BC} {\mathbb C}
71: \newcommand {\BQ} {\mathbb Q}
72: \newcommand {\RP} {\mathbb R \mathrm P}
73: \newcommand {\CP} {\mathbb C \mathrm P}
74: \newcommand {\BI} {\mathbb I}
75: \newcommand {\Unity}{\mathbf 1}
76:
77: %%% GREEK LETTERS
78: \newcommand {\al} {\alpha}
79: \newcommand {\ald}{\dot \alpha}
80: \newcommand {\be} {\beta}
81: \newcommand {\bed} {\dot \beta}
82: \newcommand {\de} {\delta}
83: \newcommand {\ve} {\varepsilon}
84: \newcommand {\ep} {\epsilon}
85: \newcommand {\ka} {\kappa}
86: \newcommand {\si} {\sigma}
87: \newcommand {\sib} {\bar \sigma}
88: \newcommand {\thb} {\bar \theta}
89: \newcommand {\vphi} {\varphi}
90: \newcommand {\ze} {\zeta}
91: \newcommand {\ro} {\rho}
92: \newcommand {\om} {\omega}
93:
94: %%% FRAK LETTERS
95: \newcommand{\g}{\mathfrak{g}}
96: \newcommand{\h}{\mathfrak{h}}
97: \newcommand{\frakH}{\mathfrak{H}}
98: \newcommand{\frakd}{\mathfrak{d}}
99: \newcommand{\frakk}{\mathfrak{k}}
100: \newcommand{\frakl}{\mathfrak{l}}
101:
102:
103: %%% DERIVATIVES
104: \newcommand {\p} {\partial}
105:
106:
107: %%% OPERATORS
108: \DeclareMathOperator{\Ker}{Ker} \DeclareMathOperator{\Img}{Img}
109: \DeclareMathOperator{\End}{End} \DeclareMathOperator{\Hom}{Hom}
110: \DeclareMathOperator{\Div}{div} \DeclareMathOperator{\tr} {tr}
111: \DeclareMathOperator{\rk} {rk} \DeclareMathOperator{\HH} {HH}
112: \DeclareMathOperator{\I} {Im} \DeclareMathOperator{\R} {Re}
113: \DeclareMathOperator{\Pf} {Pf}
114:
115:
116:
117: %%% GROUPS
118: \newcommand{\Symp}{\mathop{\rm Sp}\nolimits}
119: \newcommand{\SU}{\mathop{\rm SU}\nolimits}
120: \newcommand{\SO}{\mathop{\rm SO}\nolimits}
121: \newcommand{\Spin}{\mathop{\rm Spin}\nolimits}
122: \newcommand{\GL}{\mathop{\rm GL}\nolimits}
123: \newcommand{\SL}{\mathop{\rm SL}\nolimits}
124:
125:
126: \newcommand{\ti} {\tilde}
127:
128: \title{Black hole entropy and topological strings on generalized CY manifolds }
129:
130: \author{Vasily Pestun\thanks{On leave of absence from ITEP, 117259, Moscow, Russia}\\
131: Physics Department, Princeton University, Princeton NJ, 08544 }
132:
133:
134:
135:
136: \preprint{ ITEP-TH-110/05 \\PUPT-2188
137: \\ }
138:
139: \abstract{
140: The H. Ooguri, A. Strominger and C. Vafa conjecture $Z_{BH}=|Z_{top}|^2$ is
141: extended for the topological strings on generalized CY manifolds.
142: It is argued that the classical black hole entropy
143: is given by the \emph{generalized Hitchin functional}, which defines by critical points
144: a generalized complex structure on $X$. This geometry differs
145: from an ordinary geometry if $b_1(X) \neq 0$.
146: In a critical point
147: the generalized Hitchin functional equals to Legendre transform
148: of the free energy of generalized topological string. The examples of $T^6$ and $T^2 \times K3$
149: are considered in details.
150:
151: }
152:
153:
154:
155: \begin{document}
156: \section{Introduction}
157:
158: In~\cite{Ooguri:2004zv} H. Ooguri, A. Strominger and C. Vafa (OSV) suggested
159: a relation between the black hole entropy and the topological string
160: partition function symbolically written as $Z_{BH}=|Z_{top}|^2$.
161: In~\cite{Nekrasov:2004vv} N. Nekrasov and in~\cite{Dijkgraaf:2004te} R. Dijkgraaf et. al
162: explained that at the classical level the black hole entropy and
163: the topological strings partition function
164: are related to a certain Hitchin functional~\cite{Hitchin} for real three-forms,
165: which defines by critical points a CY structure on a real six-dimensional
166: manifold~$X$, so $Z_{hit} = Z_{BH} =|Z_{top}|^2$.
167: The important relation between Hitchin functional~\cite{Hitchin}
168: and the quantization of the topological $B$-model~\cite{Bershadsky:1993cx}
169: was shown in~\cite{Gerasimov:2004yx} by A.~Gerasimov and S.~Shatashvili.
170: Moreover, in~\cite{Gerasimov:2004yx}
171: was also suggested to use the generalized Hitchin functional~\cite{Hitchin:2}, whose
172: degrees of freedom are extended by one-forms and five-forms on $X$.
173: The necessity to turn on forms of all ranks was proposed in~\cite{Nekrasov:2004vv}
174: in the perspective of a certain seven-dimensional topological theory.
175: See~\cite{Pestun:2005gs} for topological strings in generalized complex space~\cite{Hitchin:2,GCS}.
176:
177:
178:
179: The whole construction is about compactifications with (at least) $\CalN=2$ supersymmetry.
180: The usual ones are compactifications on Calabi-Yau manifolds. Then one
181: has the $A$-model and the $B$-model~\cite{Witten:1991zz},
182: parameterized by symplectic or complex structures. They are each
183: examples of generalized complex
184: structures~\cite{Hitchin:2,GCS} and each can be described by a suitable
185: Hitchin functional~\cite{Hitchin,Hitchin:2}.
186: The black hole entropy in the
187: supergravity approximation is equal to
188: the Hitchin functional~\cite{Ooguri:2004zv,Nekrasov:2004vv,Dijkgraaf:2004te}.
189:
190: At the one-loop level, however, to reproduce the first quantum
191: correction to the Hitchin functional,
192: one needs to use the generalized Hitchin functional, as was shown in~\cite{Pestun:2005rp}.
193: In other words, for Calabi-Yau compactification, at tree level, only the modes of
194: the ordinary Hitchin functional
195: are turned on, but at one-loop level, to get the right quantum
196: correction, one needs to allow the extra
197: fields of the generalized Hitchin functional to run around in the loop.
198:
199: The present paper is devoted to answering the following question: Is it
200: possible to find a situation in which
201: it is necessary to use the generalized Hitchin functional at tree level in
202: order to reproduce the supergravity
203: approximation to the black hole entropy? This will happen if the
204: extra fields of the generalized Hitchin
205: functional have expectation values at tree level.
206:
207: In other words, it will happen in the case of a compactification with at
208: least $\CalN=2$ supersymmetry that
209: cannot be described as compactification on a complex manifold ($B$-model)
210: or a symplectic manifold ($A$-model)
211: -- compactification that requires the language of a generalized complex
212: structure.
213: Concretely, this will happen in compactifications of $\CalN\geq 2$
214: supersymmetry on a manifold X with $b_1(X)$ nonzero.
215:
216: There is a consistent supergravity analysis of $\CalN=2$ supersymmetric
217: compactifications with $b_1(X)$ nonzero
218: and using generalized complex geometry~\cite{Witt:2005oct,Grana:2005jc,Grana:2005ny,Grana:2004bg,Chuang:2005qd,Lindstrom:2004iw,Fidanza:2003zi}. In this paper, we will show
219: that in this situation, the generalized Hitchin functional reproduces
220: the supergravity approximation to the black
221: hole entropy, generalizing the results of OSV for the Kahler case.
222:
223: However, actual examples of the framework of~\cite{Grana:2005jc,Witt:2005oct,Grana:2005ny,Grana:2004bg,Chuang:2005qd,Lindstrom:2004iw,Fidanza:2003zi} are apparently not yet known.
224: To give concrete examples of our calculations, therefore, we will
225: consider the examples of
226: $X=T^6$ and $X=T^2 \times K3$, which certainly do have $b_1(X)$ nonzero. They have
227: more than $\CalN=2$ supersymmetry,
228: so it may be the case that at the loop level, they cannot be
229: described by the generalized Hitchin
230: functional (but require some further extension of it with additional
231: fields related to the higher supersymmetry).
232: However, at tree level compactification on $T^6$ or $T^2 \times K3$ has a
233: consistent truncation with $\CalN=2$ supersymmetry
234: that includes deformations best described by generalized complex
235: geometry. In this paper, we will
236: show that in this subspace of the $T^6$ and $T^2 \times K3$ moduli space, the
237: black hole entropy at tree level
238: is described by the generalized Hitchin functional. We hope that in
239: the future examples will be found
240: illustrating the ideas of the framework of~\cite{Grana:2005jc,Witt:2005oct,Grana:2005ny,Grana:2004bg,Chuang:2005qd,Lindstrom:2004iw,Fidanza:2003zi,Gurrieri:2002wz} with $\CalN=2$ supersymmetry.
241:
242:
243:
244: % In~\cite{Pestun:2005rp} the relation was tested at one-loop level and
245: % it was found that the agreement between $|Z_{top}|^2$
246: % and $Z_{Hit}$ requires
247: % introduction of additional degrees of freedom, captured by
248: % the \emph{generalized} Hitchin functional~\cite{Hitchin:2},
249: % which defines by critical points a generalized complex structure~\cite{GCS}.
250: % However, the relation in~\cite{Pestun:2005rp} was tested only
251: % at the points of the moduli space corresponding to an ordinary CY structure.
252:
253: % For simply connected CY manifolds there are no extra deformations
254: % of generalized complex structure in addition to the deformations
255: % of an ordinary complex structure.
256: % Therefore, if $H^{1}(X)=0$
257: % it was consistent in~\cite{Pestun:2005rp} to consider
258: % critical points of generalized Hitchin functional to be an ordinary
259: % CY manifold and make one-loop quantization in that background.
260:
261: % However, if $H^{1}(X) \neq 0$, generalized Hitchin functional and generalized
262: % complex structure are inevitable on a tree level.
263: % The examples include $T^6$ and $T^2 \times K3$ cases.
264:
265:
266: % In the present note we make a simple
267: % observation that the relation $Z_{Hit} = Z_{BH} = |Z_{top}|^2$
268: % also holds in the generalized case in arbitrary point of the generalized moduli space.
269:
270: For generalized complex space, the necessary formalism
271: of topological strings -- topological $\CalJ$-model -- is presented
272: in~\cite{Pestun:2005gs}, see also~\cite{MR1609624,MR1919435,MR1432574,Kapustin:2004gv,Kapustin:2003sg,Li:2005tz,Zabzine:2005qf,Lindstrom:2004hi,Lindstrom:2004iw,Zucchini:2005cq,Zucchini:2005rh,Zucchini:2004ta}.
273: The generalized Hitchin functional in~\cite{Hitchin:2},
274: and the compactifications of type II string theory on generalized CY
275: manifolds are studied in~\cite{Grana:2005ny,Grana:2005jc,Grana:2004bg,Chuang:2005qd,Lindstrom:2004iw,Fidanza:2003zi}.
276: For recent developments on black hole
277: entropy see~\cite{Dabholkar:2004yr,Dabholkar:2005dt,Dabholkar:2005by,Parvizi:2005aa,Sen:2005ch,Sen:2005wa,Sen:2005iz,Verlinde:2004ck,Strominger:1996kf,Strominger:1996sh,Horowitz:1996fn, Maldacena:1996gb,Breckenridge:1996sn,Kaplan:1996ev,Maldacena:1997de,Maldacena:1999bp,Ferrara:1995ih} and on generalized
278: complex structures in string theory
279: see~\cite{Lindstrom:2004hi,Zucchini:2005rh,Zucchini:2005cq,Lindstrom:2004eh,
280: Zucchini:2004ta,Kapustin:2004gv,Chiantese:2004pf,Zabzine:2004dp,Gates:1984nk}.
281:
282:
283:
284:
285: In sec.~\ref{sec:gen} we briefly review the standard logic,
286: in sec.~\ref{sec:main} we show that it is easily generalized. In sec.~\ref{sec:ex}
287: we illustrate an emergence of the generalized Hitchin
288: functional for $T^6$ and $T^2 \times K3$ compactifications.
289: The sec.~\ref{sec:con} concludes the note.
290:
291:
292: \section{A review}\label{sec:gen}
293:
294: The relation between $Z_{BH}$ and $Z_{top}$ comes from considering
295: a compactification of the physical type II string on a Calabi-Yau threefold $X$.
296: See~\cite{Mohaupt:2000mj} for a comprehensive review of the subject and the complete list of references.
297:
298:
299: The resulting low energy effective theory is $\CalN=2$
300: four-dimensional supergravity. It contains the $\CalN=2$
301: gravitational multiplet, the universal hypermultiplet and a number
302: of vector and hyper multiplets depending on the geometry of $X$.
303: For type IIA string $h^{1,1}$ vector multiplets correspond to
304: the complexified Kahler moduli of $X$, and $h^{2,1}$ hyper
305: multiplets correspond to the complex moduli of $X$. For type IIB
306: string the structure is reversed. The low energy effective action
307: for vector multiplets is fully specified by a single
308: holomorphic\footnote{In homogeneous special coordinates, $\CalF$ is
309: a homogeneous function of weight 2.} function, the prepotential
310: $\CalF(X^I)$. The $X^I$ are scalar components of the vector
311: multiplets, they describe moduli of the CY manifold $X$. The
312: prepotential $\CalF(X)$ defines a structure of the special Kahler
313: geometry on the corresponding moduli space.
314:
315: On the one hand, the physical string amplitudes on $X$, which compute $\CalF(X)$, can be formulated
316: in the language of topological strings~\cite{Antoniadis:1993ze,Bershadsky:1993cx}. Namely,
317: $\CalF(X)$ is just the classical free energy of the topological string. The higher
318: genus amplitudes give the terms
319: \begin{align}
320: \label{I_g} I_g = \int d^4 \theta \, W^{2g}F_g(X^I),
321: \end{align}
322: where $X^k$ are the $\CalN=2$ chiral superfields constructed
323: from the vector multiplets, and $W$ is the $\CalN=2$ chiral superfield for the Weyl
324: multiplet $W^{ij}_{\mu\nu} = T^{ij}_{\mu\nu} -
325: R_{\mu\nu\l\rho}\theta^i \sigma_{\l\rho} \theta^j + \cdots $ (with
326: $T$ being the graviphoton field, so the expansion in components
327: of~\eqref{I_g} gives terms $R^2T^{2g-2}$).
328:
329: On the other hand, the four-dimensional $\CalN=2$ supergravity
330: admits BPS black hole solutions~\cite{Ferrara:1995ih,Strominger:1996kf,Sabra:1997dh}.
331: These BPS black hole solutions are generalizations of the extremal
332: Reissner-Nordstrom black holes in Einstein-Maxwell theory with $M = |Q|$.
333: The Reissner-Nordstrom black hole has the metric
334: \begin{align}
335: ds^2 = -dt^2 (1 - 2M/r + Q^2/r^2) + dr^2 (1-2M/r + Q^2/r^2)^{-1} + r^2
336: d\Omega_2^2.
337: \end{align}
338: There is a bound $|Q| \leq M$. When the bound $M=Q$ is
339: reached, the solution becomes BPS solution. The BPS solution preserves $N=1$ supersymmetry.
340: Moreover, the near horizon geometry of such a solution is given by the Bertotti-Robinson
341: metric $AdS_2 \times S^2$. In coordinates, where horizon is located at $r=0$, the metric
342: is given by
343: \begin{align}
344: ds^2 = -\frac {r^2} {Q^2} dt^2 + \frac{Q^2} {r^2} (dr^2 + r^2
345: d\Omega^2).
346: \end{align}
347: The radius of this black hole is $r_0 = M = |Q|=|Z|$, where we also introduced
348: the central charge of $\CalN=2$ algebra for such a BPS object.
349: The Bekenstein-Hawking-Wald
350: entropy~\cite{Bardeen:1973gs,Bekenstein:1974ax,Bekenstein:1973ur,Wald:1993nt} is given by
351: the familiar formula
352: \begin{align}
353: S = \frac {1} {4} Area = \pi r_{0}^2 = \pi |Q|^2.
354: \end{align}
355:
356: In the full $\CalN=2$ supergravity we turn on the abelian vector multiplets.
357: Each one has a complex scalar $X^I$ and magnetic and electric fields $F^+_{\mu\nu}{}^I,
358: G^+_{\mu\nu}{}^I$. In the language of special Kahler geometry, it is convenient
359: to organize the fields into pairs $(X^I, F_I := \p_I \CalF)$ and $(F^+,G^+)$
360: that transform linearly under $Sp(2n+2,\BR)$ duality group,
361: actually broken to $Sp(2n+2,\BZ)$:
362: \begin{align}
363: \left(\begin{array}[2]{c}
364: X^I \\
365: F_I
366: \end{array}\right) \quad \quad
367: \left(
368: \begin{array}[2]{c}
369: F^+_{\mu\nu}{}^I \\
370: G^+_{\mu\nu}{}_I
371: \end{array}\right).
372: \end{align}
373: The black hole can carry magnetic and electric charges $(p^I, q_I)$, and the solution
374: is given by the usual $\frac 1 {r^2}$ law in the metric $ds^2 = -e^{2g(r)} dt^2 + e^{2f(r)}[dr^2+r^2 d\Omega_2^2]$
375: \begin{align}
376: i F_{23}^I = i \frac {e^{-2f(r)}} {r^2} p^I, \quad i G_{23I} = i \frac {e^{-2f(r)}} {r^2} q_I.
377: \end{align}
378: It is convenient to introduce the central charge field $Z = e^{K/2} (p^I F_I - q_I X^I)$,
379: where the Kahler potential $ e^{-K} = 2 \I (X^I \bar F_I)$.
380: The supersymmetry condition gives the solution for the
381: metric in terms of $Z$, so we have $e^{2g(r)} = e^{-2f(r)} = e^{-K}\frac{r^2}{|Z|^2}$.
382: From the Wald formula for the entropy one obtains again
383: \begin{align}
384: \label{eq:BH}
385: S = \pi |Z|^2.
386: \end{align}
387: Since $Z$ is expressed in terms of the scalars $(X^I,F_I)$ and
388: the charges $(p^I,q_I)$ we still need to find $(X^I,F_I)$
389: in terms of $(p^I,q_I)$ and then plug into~\eqref{eq:BH}. The relation is given
390: by the so called attractor equations\footnote{See~\cite{Moore:1998pn} for studies of their relation to number theory.}~\cite{Ferrara:1996um,Ferrara:1996dd,Ferrara:1995ih,Strominger:1996kf,LopesCardoso:1998wt,LopesCardoso:1999cv}
391: \begin{align}
392: \bar Z \left(\begin{array}[2]{c}
393: X^I \\
394: F_I
395: \end{array}\right) - Z
396: \left(
397: \begin{array}[2]{c}
398: \bar X^I \\
399: \bar F_J
400: \end{array}\right)
401: = i e^{-K/2}
402: \left( \begin{array}[2]{c}
403: p^I \\
404: q_J
405: \end{array}\right).
406: \end{align}
407: So we have the formula for the entropy $S(p,q)$
408: \begin{align}
409: \label{eq:Spq}
410: S(p^I,q_I) = \pi \frac{|pF - qX|^2}{ 2 \I (X \bar F)}.
411: \end{align}
412: where $\R (CX^I) = p^I$, $\R (CF_I) = q_I$ (we suppress index $I$ in contractions like $X^I \bar F_I$).
413: The formula~\eqref{eq:Spq} is invariant under a homogeneous complex dilatation,
414: so we can put $C=1$. The attractor equations $\R (CX^I) = p^I, \R (CF_I) = q_I$ can be also
415: obtained minimizing~\eqref{eq:Spq} by $X^I$ for the fixed charges $(p^I,q_I)$
416: with $F_I = \p_I \CalF$.
417:
418: Let us decompose $X^I$ into the imaginary
419: and the real part\footnote{In the following we often omit the index $I$,
420: assuming the natural contraction in products.} $X' + i X'', F = F' + iF''$.
421: Then we plug $p = X', q = F'$ and compute $S(p,q) = S(X',F')$
422: \begin{align}
423: S_{BH}(X',F') = \pi \frac{ |X'(F'+iF'') - F'(X'+iX'')|^2 }{2( X'' F' - X' F'')} = \frac {\pi}{2} (X'' F' -
424: X'F'').
425: \end{align}
426:
427: Now compare the function $S_{Hit}$, whose rationale will become clear in a moment,
428: $S_{Hit}(X',F') = \frac {1} {\pi} S_{BH}(X',F')$
429: with the imaginary part or the prepotential $\CalF= \frac 1 2 X^I F_I = \CalF' + i \CalF''$
430: \begin{align}
431: S_{Hit}(X',F') = \frac 1 2 (X'' F' - X' F'') \\
432: \CalF''(X',X'') = \frac 1 2 ( X''F' + X' F'').
433: \end{align}
434: We see that
435: \begin{align}
436: \frac 1 {\pi} S_{BH}(X',F') = S_{Hit}(X',F') = X''F' - \CalF''(X',X'').
437: \end{align}
438: Moreover, since $\CalF = \frac 1 2 X^I F_I$ and $\CalF$ is holomorphic we have the relation on the derivative
439: \begin{align}
440: F'_I = \frac{ \p \CalF''} {\p X^I{}''}.
441: \end{align}
442: Therefore $\frac 1 {\pi} S_{BH} (X',F') = S_{Hit}(X',F')$ is
443: the Legendre transform of the imaginary part
444: of the topological string free energy $ \CalF''(X',X'')$ in the imaginary part $X''\equiv \I
445: X$~\cite{Ooguri:2004zv,Dijkgraaf:2004te,Nekrasov:2004vv}
446: \begin{align}
447: \label{eq:HitLeg}
448: \boxed{
449: S_{Hit} (X',F') = \mathrm{Legendre} [\CalF''(X',X''), F' =\p_{X''} \CalF'' ]}
450: \end{align}
451:
452: \section{A generalization of the OSV conjecture}
453: \label{sec:main}
454: Before going to generalization, let us recall the meaning of the Hitchin functional $S_{Hit}$
455: in the formulas above. Let $\Omega$ be the holomorphic $(3,0)$ form on the CY manifold $X$.
456: As usual we have $X^I = \int_{A_I} \Omega, F_I = \int_{B^I} \Omega$ for some canonical basis
457: of cycles $A_I, B^I$. The Hitchin functional in its critical point is the
458: integral of the volume form defined by $\Omega$
459: \begin{align}
460: S_{Hit} (X', F') = - \frac {i} {4} \int \Omega \wedge \bar \Omega = \frac {1} {4 i} (X \bar F - \bar XF)=
461: \frac 1 2 \I X\bar F.
462: \end{align}
463: The reason why we write $S_{Hit}$ as a function of real part of periods $X',F'$ is that
464: it is actually a function of them by the construction~\cite{Hitchin}
465: \begin{align}
466: S_{Hit}[\rho] = \frac {1} {4i} \int (\rho + i\hat \rho ) \wedge (\rho -i \hat \rho) =
467: -\frac {1} {2} \int ( \rho \wedge \hat \rho) = \int \sqrt{I_4(\rho)} = \int vol.
468: \end{align}
469: Here $\rho$ is a stable real three-form, and $\hat \rho$ is a
470: certain non-linear function of $\rho$, such that $\rho+i \hat
471: \rho$ is the decomposable almost holomorphic $(3,0)$ form with
472: respect to the complex structure also defined by $\rho$. The
473: integrability of the complex structure can be cast in the form $d
474: (\rho + i \hat \rho) =0$. The field theory is defined by
475: restricting $\rho$ to some cohomology class in $H^3(X,\BR)$, so $d\rho = 0$.
476: In a critical point of $S_{Hit}[\rho]$ we have $d \hat \rho =
477: 0$, and the complex structure is integrable.
478: We see that at the classical level the relation~\eqref{eq:HitLeg} holds: the Hitchin
479: functional, proportional to the black hole entropy, is the
480: Legendre transform of the imaginary part of the holomorphic
481: prepotential~$\CalF$~\cite{Ooguri:2004zv,Dijkgraaf:2004te,Nekrasov:2004vv}.
482: The relation between Hitchin functional and topological
483: string was also studied
484: classically in~\cite{Gerasimov:2004yx} and at the one-loop in~\cite{Pestun:2005rp}.
485: For micro/macroscopical tests of the OSV conjecture see~\cite{Dabholkar:2004yr,Dabholkar:2005dt,Dabholkar:2005by,Verlinde:2004ck,Strominger:1996kf,Strominger:1996sh,Horowitz:1996fn, Maldacena:1996gb,Breckenridge:1996sn,Kaplan:1996ev,Maldacena:1997de,Maldacena:1999bp,Ferrara:1995ih}.
486:
487:
488: In the case of generalized complex structures the whole construction works exactly in the same
489: way. For a generalized complex structure, an analogue
490: of the holomorphic $(3,0)$-form will be a mixed differential form
491: in complex $H^{odd} = H^1 + H^3 + H^5$
492: or $H^{even} = H^0 + H^2 + H^4 + H^6$,
493: which is at the same time a pure spinor $\Omega=\rho + i \hat \rho$ of $SO(6,6)$~\cite{Hitchin:2,GCS}.
494: The off-shell generalized Hitchin functional is defined by the
495: real part $\rho$ of the pure spinor~$\Omega$
496: \begin{align}
497: \label{eq:HitGen}
498: S_{Hit}^G = -\frac {1} {2} \int (\rho,\hat\rho) = \int
499: \sqrt{I_4(\rho)},
500: \end{align}
501: where $\hat \rho$ is a certain nonlinear function of $\rho$, and $(,)$ is an appropriate bilinear
502: form on the space of mixed differential forms~\cite{Hitchin:2}.
503: A mixed differential form $\rho$ in $\Omega^1+\Omega^3+\Omega^5$
504: or $\Omega^0+\Omega^2+\Omega^4+\Omega^6$, according to its chirality,
505: transforms as a spinor of $SO(6,6)$, and $I_4(\rho)$
506: is the singlet in the tensor product of four $SO(6,6)$ spinors.
507:
508: The moduli space of ordinary CY structures
509: locally is
510: \[(H^{3,0}\oplus H^{2,1})(X,\BC),\]
511: or $H^3(X,\BR)$ by Hitchin construction.
512: The moduli space of
513: generalized CY structures locally near the point of an ordinary
514: complex structure is
515: \[(H^{1,0} \oplus H^{2,1} \oplus H^{3,2} \oplus H^{3,0})(X,\BC),\]
516: or $H^1(X,\BR) \oplus H^3(X,\BR) \oplus H^5(X,\BR)$ by Hitchin
517: construction.
518:
519: The $even/odd$ cases of generalized complex structure in six real dimensions
520: correspond to the type A/B
521: strings. They are distinguished by the chirality of the canonical pure spinor $\Omega$
522: that defines the corresponding generalized complex structure. In real six dimensions
523: a usual complex structure is of odd type, and a usual symplectic structure is of even type.
524:
525: In \cite{Hitchin:2} Hitchin shows that the moduli
526: space of generalized complex structures has \emph{a special Kahler geometry}.
527: Since $\CalN=2$ supergravity is fully defined by an appropriate special Kahler structure
528: on the target manifold for the scalar fields from vector multiplets,
529: all $\CalN=2$ computations for the black hole entropy can be done in
530: the generalized complex case, as long as one includes the extra multiplets.
531:
532:
533:
534: The outcome of $\CalN=2$ supergravity is the
535: formula~\eqref{eq:HitLeg},
536: which tells us that $S_{BH}$ is the Legendre transform of $\I \CalF$. Here $\CalF$ is the
537: prepotential of the special geometry of the moduli space of generalized complex structures.
538: It can be defined in a similar way. We pick up a basis of $A_I,B^I$ cycles in
539: $H_{odd}=H_1 \oplus H_3 \oplus H_5$ or in $H_{even}=H_0 \oplus H_2 \oplus H_4 \oplus H_6$,
540: which is canonical with respect to the \emph{sign twisted} wedge product\footnote{
541: This sign twisted wedge
542: product for two forms $\alpha, \beta$ is defined as $(\alpha, \beta) = \alpha \wedge \beta$
543: for $\deg \beta = 4k+0,1$ and $(\alpha,\beta) = -\alpha \wedge \beta$ for $\deg \beta = 4k+2,3$ \cite{Hitchin:2}.}
544: that agrees with the bilinear form on spinors of $Spin(TX,TX)$~\cite{Hitchin:2}.
545: Then
546: \begin{align}
547: X^I = \int_{A_I} \Omega,\quad F_I \int_{B^I} \Omega,
548: \end{align}
549: where now $A_I,B^I$ runs over all degrees in $H_{odd}$ of $H_{even}$.
550: For example, let us consider an ordinary symplectic structure $\omega$ as a generalized complex structure.
551: Then $\Omega = e^{i\omega}$, or
552: \[\Omega = 1 + i\omega -\frac 1 2 \omega^2 - \frac 1 6 i \omega^3.\]
553: We have the zero-cycle and a number of two-cycles of $A$ type, and a number four-cycles and the
554: six-cycle of $B$ type.
555: The sign twisted wedge product is antisymmetric and defines
556: a symplectic structure on~$H^{even}(X)$. Then we recover the standard formulas
557: \begin{align}
558: X^0 =1 \quad \quad F_0 = \int_X -i \frac 1 6 \omega^3\\
559: X^I = \int_{A_I} i \omega \quad \quad F_I = \int_{B^I} - \frac 1 2 \omega^2\\
560: \quad \CalF = \left(\frac{-i} {4}+ \frac {i} {12} \right) \int_X \omega^3 = - \frac {i} {6} \int_X\omega^3.
561: \end{align}
562:
563: The topological string in a generalized complex space -- topological $\CalJ$-model\footnote{$\CalJ$
564: stands for a generalized complex structure, which can in particular be an
565: ordinary symplectic (A) or an ordinary complex (B).} --- is described
566: in~\cite{Pestun:2005gs}, see there a complete list of references on the related works.
567: In agreement with~\cite{Hitchin:2} it is explained in~\cite{Pestun:2005gs},
568: that in the case $\dim_{\BC}X = 3$ the
569: moduli space of geometrical deformations of a generalized complex
570: structure is a special Kahler manifold. It is also shown that the topological
571: string three-point function is the third derivative $C_{IJK}= \p_I \p_J \p_K \CalF$ of the
572: holomorphic prepotential $\CalF$ of that special geometry. The manifold
573: of the geometrical deformations of a generalized complex structure is
574: a holomorphic Lagrangian submanifold
575: inside the total extended moduli space of deformations of the associated
576: special differential BV algebra.
577: The outcome of~\cite{Pestun:2005gs} is that the genus zero topological
578: string free energy without instanton corrections is given by the
579: same formula $\CalF = \frac 1 2 X^I F_I$, where $X_I$ and $F^I$ are
580: periods the canonical pure spinor that defines a generalized complex structure
581: over extended set of cycles on $X$.
582: Therefore the relation
583: \begin{align}
584: S_{Hit} (X',F') = \mathrm{Legendre} [\CalF''(X',X''), F' =\p_{X''} \CalF'' ]
585: \end{align}
586: holds in the generalized complex case, and $\CalF''$ is the imaginary part of the free energy
587: of the topological $\CalJ$-model~\cite{Pestun:2005gs}.
588:
589: What about the black hole entropy? On the one hand, given a special Kahler
590: geometry,
591: we can formally write down an appropriate $\CalN=2$ four-dimensional supergravity, and then
592: the relation~\eqref{eq:HitLeg} automatically holds due to the special Kahler geometry relations.
593: But can it physically be related to topological strings in generalized complex space?
594: The answer seems to be yes, and the connection is again realized by the
595: ten-dimensional type II string
596: theory compactified on the given generalized CY manifold $X$.
597: Recently non Calabi-Yau compactifications
598: were studied in much details in\cite{Grana:2004bg,Grana:2005jc,Chuang:2005qd,Fidanza:2003zi,Lindstrom:2004iw}.
599: We expect that the type II ten-dimensional string theory compactified on a generalized
600: CY manifold $X$ is related to the topological string on $X$ exactly in the same fashion
601: like in the usual case.
602: In~\cite{Witt:2005oct,Grana:2005ny} the direct relation between Hitchin functionals for generalized complex
603: geometry in $\CalN=2$ supergravity and the type II string compactification was described.
604:
605: \section{Examples: $T^6$ and $T^2\times K3$}
606: \label{sec:ex}
607: Here we consider a simple example when the generalized Hitchin
608: functional differs from the ordinary Hitchin functional at tree level.
609: This is possible only when $X$ has $b_1(X) \neq 0$, so $T^6$ and $T^2 \times K3$
610: are natural examples to see explicitly how the generalized Hitchin functional works.
611:
612: First of all, one shall note that the physical type II string
613: compactified on $T^6$ or $T^2 \times K3$ space gives rise to $\CalN=8$
614: or $\CalN=4$ supergravity. Of course, the structure of these gravity theories differs from
615: $\CalN=2$. The usual, or generalized like in~\cite{Pestun:2005gs}
616: topological string, as well as attractor equations,
617: deals only with $\CalN=2$ terms.
618:
619: The additional massless vector multiplets of $\CalN=4$ or $\CalN=8$ gravities
620: are not among observables of the topological string, which couples to variations
621: of (generalized) complex or symplectic structure on $X$.
622: We consider $\CalN=2$ truncation of the $\CalN=4,8$ theories and leave only those vector
623: multiplets, whose scalars come from (generalized) complex or symplectic moduli of $X$.
624:
625: In $T^6$ case the $\CalN=8$ supergravity
626: multiplet~\cite{} contains the following $\CalN=2$ multiplets.
627: There is 1 $\CalN=2$ gravity multiplet, 6 $\CalN=2$ gravitini multiplets,
628: 15 $\CalN=2$ vector multiplets
629: and 10 $\CalN=2$ hypermultiplets. Each gravitini multiplet has two gauge fields, so there are in
630: total $1+12+15=28$ gauge fields for the $T^6$ compactification. We throw away the gravitini multiplets
631: and stay with $1+15=16$ gauge fields coming from the $\CalN=2$ supergravity sector.
632:
633:
634: In $T^2 \times K3$ case, after decomposition under $\CalN=2$,
635: the gauge fields are counted as follows. There is one $\CalN=4$ supergravity multiplet
636: and 22 $\CalN=2$ vector multiplets. The $\CalN=4$ supergravity multiplet is decomposed
637: into 1 $\CalN=2$ gravity multiplets, 2 $\CalN=2$ gravitini multiplets and 1 $\CalN=2$ vector multiplet. It has
638: $1+4+1 = 6$ gauge fields. In total there are $22+6=28$ gauge fields with corresponding
639: 28 magnetic and 28 electric charges. Again we throw away the $\CalN=2$ gravitini multiplets
640: and stay with $1+22+1=24$ gauge fields coming from $\CalN=2$ supergravity sector.
641:
642: The corresponding black hole solution carries magnetic and electric charges only for these $\CalN=2$ vector
643: multiplets. The solution is $1/4$ BPS for $T^6$ compactifications and
644: $1/2$ BPS for $T^2 \times K3$, so it preserves $\CalN=1$ four-dimensional supersymmetry.
645: The truncation is consistent classically, and we will work here only at the classical level.
646:
647: Though in derivation of the Legendre transform we closely follow~\cite{Pioline:2005vi},
648: the novelty is the relation of the result with the generalized Hitchin functional~\cite{Hitchin:2}
649: and with the generalized topological $\CalJ$-model~\cite{Pestun:2005gs}.
650:
651: The simplest case is the 1/8 BPS black hole for the IIA on $T^6$ with the charges corresponding
652: to $D0,D2,D4$ and $D6$ branes\cite{Shih:2005uc,Shih:2005qf,Shih:2005he,Pioline:2005vi,Dabholkar:2005by,Dabholkar:2005dt,Dabholkar:2004yr,Parvizi:2005aa}.
653: The IIA corresponds to the topological $A$-model.
654: The genus zero topological string free energy is given by
655: \begin{align}
656: \label{eq:TF}
657: \CalF = - \frac 1 6 \frac {C_{IJK} {X^I X^J X^A}} {X^0},
658: \end{align}
659: where $X^I = \int_{A^I} \omega$ are integrals of the complexified Kahler class over two cycles $I=1\dots 15$,
660: and $C_{IJK}$ is the intersection matrix for the two-cycles on $T^6$. We will consider $A_I$ cycles
661: to be the $0$-cycle and all $2$-cycles, the dual $B^I$-cycles are all $4$-cycles and the $6$-cycle.
662: The $2$-cycles on $T^6$ are labelled by pairs $1 \leq i < j \leq 6$, which we can organize into
663: labels of the components of $6 \times 6$ antisymmetric matrix.
664: The periods $X^I$, $I=1\dots 15$ are entries of this matrix. The non-zero intersection of three 2-cycles
665: correspond to the choice of three pairs of indexes $(i,j)$ such that all of them are different,
666: with an appropriate sign coming from parity
667: of permutation $(i_1,i_2), (i_3,i_4), (i_5,i_6)$ into $(1,2,3,4,5,6)$.
668: Therefore the expression~\eqref{eq:TF} can be reorganized into the Pfaffian of the
669: antisymmetric $6\times 6$ matrix $X$ with entries $X^I$
670: \begin{align}
671: \label{eq:TF2}
672: \CalF = - \frac {\Pf(X)} {X^0}.
673: \end{align}
674:
675:
676: Now we need to find the Legendre transform of~\eqref{eq:TF2} in imaginary part $X^I{}''$ for
677: $X^I = X^I{}' + i X^I{}''$, $I=0\dots 15$.
678: In order to do that for a general cubic prepotential of $n$ variables
679: one has to solve a system of $n$ quadratic equations, which generally speaking
680: does not have a closed algebraic solution~\cite{Shmakova:1996nz}. The key property of~\eqref{eq:TF2} that allows
681: to explicitly find its Legendre transform in $( X^0{}'', X^I{}'')$ is its extremely simple behavior
682: under the full complex Legendre transform for all variables $(X^I, X^0)$ at once.
683: There exist a very distinguished set of cubic prepotentials $ C_{IJK} X^I X^J X^K /X^0$ that
684: are invariant under the Legendre transform in all variables.
685: They were all algebraically classified in~\cite{Etingof:0003009} with even stronger condition.
686: The exponents of these functions are invariant under
687: the Fourier transform. In the $T^2 \times K3$ case, the invariance is easy to see, and we will
688: demonstrate it below. As for the $T^6$ case, see~\cite{Pioline:2005vi,Pioline:2003uk,Etingof:0003009}.
689: (The semiclassical evaluation of the Fourier transform
690: reduces to the Legendre transform. In other words,
691: integrals of exponents of such cubic functions are
692: exactly localized on the their critical points.\footnote{The same distinguished types of
693: cubic prepotentials were also classified much earlier in~\cite{Gunaydin:1983bi} studies
694: of $\CalN=2$ supergravity.They can appear as $\CalN=2$ four dimensional prepotentials of dimensional
695: reduction $\CalN=2$ five-dimensional supergravity.} Such nice prepotentials are labelled
696: by $B_n,D_n, E_6, E_7, E_8, F_4, G_2$ algebraic types~\cite{Pioline:2005vi}, and the case with Pfaffian
697: of an antisymmetric $6\times 6$ matrix is the $E_7$ case.)
698:
699: Given such an invariant function $I_3(X^I)/X^0$, Pioline~\cite{Pioline:2005vi} computes
700: its Legendre transform in $(X^{0}{}'',X^{I}{}'')$.
701: The idea is to shift variables $x^I = X^{I}{}'' - \frac{X^0{}''} {X^0{}'} X^I{}'$ in
702: such a way to kill the quadratic term in $X^{I}{}''$
703: in the expansion $I_3(X^I{}'+iX^I{}'')$. Then the Legendre transform is computed using the invariance
704: of $I_3(X^{I}{}'') / X^0{}''$.
705: In the notations $p=X',q=F',\phi = X''$ the Pioline result~\cite{Pioline:2005vi} is
706: \begin{multline}
707: S_{Hit} = \mathrm{Legendre}[-\I \frac {I_3(p^I+i\phi^I)}{p^0+i\phi^0},\phi^I] = \\
708: =\sqrt{ 4p^0 I_3(q) - 4 q_0 I_3(p) + 4 \p^I I_3(q) \p_I I_3(p) - (p^0 q_0 + p^A q_A)^2} =: \sqrt{I_4(p,q)}.
709: \label{eq:HitTr}
710: \end{multline}
711:
712: Specializing to the $T^6$ case,
713: Pioline~\cite{Pioline:2005vi} obtains quartic $SO(6,6)$ invariant functional
714: $I_4(p^I,q_I)$ of 32 charges $p^I,q_I, I=0\dots 16$.
715: The charge vector $p^I, q_I$ of $T^6$ transforms as a spinor under $SO(6,6)$, and $I_4(p_I,q^I)$
716: is the singlet in the symmetric tensor product of four $SO(6,6)$ spinors.
717:
718: Now recall the definition~\eqref{eq:HitGen} of the generalized Hitchin functional~\cite{Hitchin:2}.
719: Specializing to the case of $T^6$, where in the critical point $\Omega = \rho + i \hat \rho$ is constant,
720: one immediately recognizes the agreement with the Legendre transform~\eqref{eq:HitTr} of the
721: topological string free energy~\eqref{eq:TF2}. In the framework of the generalized topological
722: strings~\cite{Pestun:2005gs}, the
723: periods $(X^I, F_I)$ are defined by integrals of the canonical pure spinor $\Omega=\rho + i \hat \rho$ of
724: $SO(TX,T^*X)$, equivalently it is a mixed differential form on $X$. After the Legendre
725: transform the charges $(p^I,q_I)$ are identified with the periods of the real part $\rho$ of $\Omega$.
726: In the case of $A$-model, $\Omega = e^{i \omega + b}$, which gives the claimed correspondence.
727:
728: What about the generalized $B$-model on $T^6$? An ordinary complex structure is defined by
729: a holomorphic $(3,0)$ form. A generalized complex structure is defined by
730: a pure $SO(6,6)$ spinor of odd chirality, which can be represented as a mixed differential
731: form\footnote{In this correspondence gamma matrices of $SO(TX,T^*X)$ are organized into
732: creation and annihilation operators $a^{i+},a_j$, $\{a^i{}^+,a_j\} =\delta^{i}_j$.
733: Then $a^i{}^+ \simeq dx^{i}\wedge$ corresponds to the wedge product with $dx^i$,
734: and $a_i{} \simeq \p_i$ corresponds to the contraction with the vector field $\p_i$.} $\Omega = \Omega^{(1)} + \Omega^{(3)} + \Omega^{(5)}$. The condition `pure'
735: for the $SO(6,6)$ spinor $\Omega$ in the generalized complex case is an analogue
736: of the $(3,0)$ type condition for the form $\Omega$ in the ordinary complex case.
737: Deformations of an ordinary complex structure are parameterized
738: by Beltrami differentials $\mu_{\bar j}^i$, so that $\p_{\bar j} \to \p_{\bar j} + \mu_{\bar j}^i \p_i$ and
739: $\Omega \to e^{-\mu} \Omega$. In the generalized complex case deformations are given\footnote{
740: At an arbitrary reference point the geometrical
741: deformations in the topological $\CalJ$-model
742: are given by $\Lambda^{2}(L^*)$, and all extended deformations are given
743: by $\Lambda^{\bullet}(L^*)$, where $L$ is the $+i$-eigenbundle of the generalized complex structure
744: $\CalJ \in \End(TX\oplus T^*X)$.}
745: by $\mu^{ij} + \mu^i_{\bar j} + \mu_{\bar i \bar j}$, which can be viewed as a section of
746: $\Lambda^2(TX^{10} \oplus T^*X^{01}) =: \Lambda^{2}(L^*)$, that is a subalgebra of $so(6,6,\BC)$.
747: A deformation $\Omega \to e^{-\mu} \Omega$ is a rotation of a spinor by an
748: element $\mu$ of $\Lambda^{2}(L^*) \subset so(6,6,\BC)$. We restrict $so(6,6,\BC)$ to $\Lambda^{2}(L^*)$
749: to keep the spinor pure.
750: Let us introduce indexes $(a,b)$ which run over upper holomorphic ${}^{123}$ indexes and
751: lower antiholomorphic ${}_{\bar 1 \bar 2 \bar 3}$ indexes.
752: Then an element $\mu_{ab}$ of $ \Lambda^{2}(L^*) \subset so(6,6,\BC)$ defines a rotation of the spinor $\Omega$ by the formula
753: \begin{align}
754: \Omega = e^{-\mu} \Omega_0 = e^{-\frac 1 2 \mu_{ab} \Gamma^{a} \Gamma^{b}} \Omega_{0}.
755: \end{align}
756:
757: The entries $\mu_{ab} = (\mu^{ij}, \mu^{i}_{\bar j}, \mu_{\bar i \bar j})$ are organized
758: into an antisymmetric $6\times 6$ matrix
759: \begin{align}
760: \mu_{ab} = \left( \begin{array}{cc}
761: \mu^{ij} & \mu^i_{\bar j} \\
762: -(\mu_{\bar j}^i)^{T} & \mu_{\bar i \bar j}
763: \end{array} \right ).
764: \end{align}
765:
766: In the case of $T^6$ deformations, $\mu$ is a constant matrix,
767: and the general Chern-Simons like cubic formula~\cite{Pestun:2005gs} for the tree level free energy of $\CalJ$-model
768: is reduced to
769: \begin{align}
770: \CalF(\mu|\Omega_0) = -\frac 1 6 ((\mu_{ab} \Gamma^a \Gamma^b)^{3} \Omega_0, \Omega_0),
771: \end{align}
772: which in turn gives
773: \begin{align}
774: \CalF{(\mu|\Omega_0)} = -\Pf(\mu).
775: \end{align}
776:
777: We see that in the canonical coordinates, the free energy of the $B$-model on $T^6$ is also given
778: by the cubic polynomial, namely Pfaffian of an antisymmetric $6\times 6$ matrix. We can
779: also write the formula in terms of periods $X^I = \int_{A_I}( \mu \cdot \Omega)$,
780: where 15 $A_I$ cycles in $(H_{1} \oplus H_{3} \oplus H_{5})(X,\BC)$ are dual to the forms $\mu \cdot \Omega$ as follows.
781: There are 3 one-cycles for $dz^i$, 9 three-cycles $dz^{\bar i}
782: \wedge dz^j \wedge dz^k$,
783: and 3 five-cycles $dz^1\wedge dz^2\wedge dz^3 \wedge dz^{\bar i} \wedge dz^{\bar j}$.
784: In addition there is one distinguished cycle $A_0$, which is dual to $dz^1 \wedge dz^2 \wedge dz^3 $.
785: In terms of these periods $X_I = \int_{A^I} \ (\mu \cdot \Omega)$ we obtain
786: \begin{align}
787: \CalF= -\frac{ \Pf(X)} {X^0}.
788: \end{align}
789: Then one proceeds in a similar way to the $A$-model considered above.
790: For an illustration let us look at the Hodge diamond of $T^6$.
791: The spaces of $\Omega$ and $(\mu \cdot \Omega)$, which describe
792: deformations of generalized complex structure with a reference point being
793: an ordinary complex structure, are underlined. They are mirror to $H^0 \oplus H^2$ in the $A$-model
794: by 90 degree rotation of the Hodge diamond
795: \begin{align}
796: \begin{array}{ccccccc}
797: & & & h^{00} & & & \\
798: & & h^{10} & & h^{01} & & \\
799: & h^{20} & & h^{11} & & h^{02} & \\
800: h^{30} & & h^{21} & & h^{12} & & h^{03} \\
801: & h^{31} & & h^{22} & & h^{13} & \\
802: & & h^{32} & & h^{23} & & \\
803: & & & h^{33} & & & \\
804: \end{array} =
805: \begin{array}{ccccccc}
806: & & & 1\,\, & & & \\
807: & & \underline{3}\,\, & & 3\,\, & & \\
808: & 3\,\, & & 9\,\, & & 3\,\, & \\
809: \underline{1}\,\, & & \underline{9}\,\, & & 9\,\, & & 1\,\, \\
810: & 3\,\, & & 9\,\, & & 3\,\, & \\
811: & & \underline{3}\,\, & & 3\,\, & & \\
812: & & & 1 & & & \\
813: \end{array}.
814: \end{align}
815:
816:
817:
818: Let us remark however, that such a simple cubic formula for $\CalF$ of
819: the generalized $B$-model
820: is obtained only in the so called canonical coordinates $\mu$,
821: in terms of periods $X^I$ over carefully chosen set of cycles by the condition that
822: $X^I$ are linear functions of $\mu$. And the fact that $\CalF$ of the $B$-model on $T^6$ does not
823: have corrections to the cubic term by mirror symmetry means the well-known fact
824: that the topological $A$-model on $T^6$
825: does not have instanton contributions, so the formula~\eqref{eq:TF2} is exact in genus zero\footnote{Actually,
826: the higher genus contributions also vanish.}.
827: The function $\CalF = \frac 1 2 X^I F_I $ is not $Sp(2N)$ invariant under
828: a change of basis of cycles, but $(X^I,F_I \equiv \p_I \CalF)$ transforms as a fundamental of $Sp(2N)$.
829: One can also compare the present computation with computation
830: of ordinary deformations of complex structure parameterized by $H^{2,1}(T^6)$
831: in~\cite{Moore:1998pn}.
832:
833:
834: Let us turn to the type II string on $K3\times T^2$~\cite{Cvetic:1995bj,Cvetic:1995uj,Shih:2005uc,Shih:2005he,Shih:2005qf,Dabholkar:2005by,Dabholkar:2004yr,Dabholkar:2005dt,Parvizi:2005aa,Pioline:2005vi}.
835: As we explained above,
836: we consider the truncation of the spectrum to $1+23=24$ gauge fields with $24$
837: electric and $24$ magnetic charges. The gauge fields come from reduction of RR $(p+1)$-forms on $p$-cycles on $X$. In type IIA
838: $p$ is even, and in type IIB $p$ is odd.
839: The Hodge diamond for $\dim H^{p,q}$ of $T^2 \times K3$ has the following
840: form
841: \begin{align}
842: \begin{array}{ccccccc}
843: & & & h^{00} & & & \\
844: & & h^{10} & & h^{01} & & \\
845: & h^{20} & & h^{11} & & h^{02} & \\
846: h^{30} & & h^{21} & & h^{12} & & h^{03} \\
847: & h^{31} & & h^{22} & & h^{13} & \\
848: & & h^{32} & & h^{23} & & \\
849: & & & h^{33} & & & \\
850: \end{array} =
851: \begin{array}{ccccccc}
852: & & & 1 & & & \\
853: & & \underline{1} & & 1 & & \\
854: & 1 & & 21 & & 1 & \\
855: \underline{ 1} & & \underline{21} & & 21 & & 1 \\
856: & 1 & & 21 & & 1 & \\
857: & &\underline{1} & & 1 & & \\
858: & & & 1 & & & \\
859: \end{array}.
860: \end{align}
861:
862: Again we underlined spaces of generalized deformations with a reference point being
863: an ordinary complex structure (the $B$-model). There are 22 ordinary CY moduli (20+1 for complex
864: structures on $K3$ and $T^2$, and 1 for an overall dilatation of the holomorphic $(3,0)$-form)
865: and 2 generalized extra moduli coming from deformations by a holomorphic bivector $\beta^{ij}$
866: and $B$-field $B_{\bar i \bar \j}$. After contraction with the holomorphic
867: $(3,0)$ form, the $\beta^{ij}$ and $B_{\bar i\bar j}$ generalized deformations
868: sit in $\Omega^{10}$ and $\Omega^{32}$ entries of the Hodge diamond.
869: We can decompose this deformation over the following basis in $H^{odd} = H^{1} + H^{3} + H^{5}$.
870:
871: There is 1 deformation of complex structure on $T^2$, which after contraction with the holomorphic $(1,0)$ form on
872: $T^2$ is mapped to the $(0,1)$ form on $T^2$ times the holomorphic $(2,0)$ form on $K3$.
873: The coefficient at the form $(\overline {dz})_{T^2} \wedge \Omega_{K3}$ is called $X^1$.
874:
875: There are 20 deformations of complex structure on $K3$ $\mu_{\bar j}^i$, which after contraction with
876: the holomorphic $(2,0)$ form are mapped into $(1,1)$ forms $(\mu \cdot \Omega_{K3})_{i \bar j}$ on
877: $K3$ times the holomorphic $(1,0)$ form on $T^2$. The corresponding coefficients
878: are called $X^I$, $I=2..21$.
879:
880: There is 1 generalized deformation by holomorphic bivector on $K3$, which after contraction with the $(3,0)$
881: holomorphic form $\Omega$ is mapped to the $(1,0)$ holomorphic form on $T^{2}$.
882: The corresponding coefficient is $X^{22}$.
883:
884: There is 1 generalized deformation by $B_{\bar i \bar j}$ field, which is mapped to the
885: space spanned by $(\overline{dz})_{T^2} \wedge \Omega_{K3} \wedge \bar \Omega_{K3}$.
886: The corresponding coefficient is $X^{23}$.
887:
888: There is 1 dilatation of $\Omega$, which is mapped to the same space of $(3,0)$ forms $\Omega$.
889: The corresponding coefficient is $X^0$.
890:
891: In total we have $24$ periods $X^I$, $I=0,\dots 23$
892: corresponding to the $24$ gauge fields in $\CalN=2$ vector multiplets.
893:
894: Using formalism of~\cite{Pestun:2005gs} it is not difficult to see
895: the topological string free energy is given by
896: \begin{align}
897: \label{eq:FK3}
898: \CalF = \frac 1 2 \frac {X^1 C_{ab} X^a X^b}{X^0}
899: \end{align}
900: where $C_{ab}$ is the intersection matrix in $H^{2}(K3)$, $a=2\dots 23$. Again
901: the $B$-model answer is a simple cubic expression\footnote{
902: The solution of the Kodaira-Spencer equation for $\bar \p (a+x) + \frac 1 2 \{ (a+x),(a+x)\}$
903: for $\mu = x+a$, gives a nonzero correction $a$ to the harmonic
904: representative $x$ of cohomology class $H^1(TX)$.
905: However the correlation $\int_{T^2\times K3} ((\mu \cdot)^3 \Omega, \Omega)$ decouples into
906: $\int_{T^2}$ and $\int_{K3} ((\mu \cdot)^2 \Omega_{K3}, \Omega_{K3}) = \int_{K3} (\mu \cdot \Omega) \wedge (\mu \cdot \Omega)$. That differs from $ (x\cdot \Omega) \wedge (x \cdot \Omega)$ by an integral
907: of $\p$-exact times $\p$-closed term, which vanishes.},
908: exactly in agreement with the mirror symmetry ($T^2 \times K3$ is mirror symmetric to itself)
909: and the fact that $A$-model does not have
910: any worldsheet instanton corrections in genus zero.
911:
912: The full Legendre transform of the function
913: $\CalF = \frac 1 2 X^1 C_{ab} X^{a} X^{b}/X^{0}$ in all complex variables $X^I$
914: is given simply by $-\frac 1 2 F_1 C^{ab} F_a F_b/F_{0}$. (In other words, for the bilinear form
915: that satisfies $C=C^{-1}$, the function~\eqref{eq:FK3} is invariant under the full Legendre transform
916: and fall into the classification of~\cite{Etingof:0003009}).
917: Explicitly, we need to solve $\p_0 \CalF = F_0, \p_1 \CalF = F_1$ and $\p_a \CalF = F_a$, so we have
918: \begin{align}
919: -\frac 1 2 \frac {X^1} {(X^0)^2} C_{ab} X^{a} X^{b} = F_0 \\
920: \frac 1 2 \frac 1 {X^0} {C_{ab} X^{a} X^{b}} = F_1\\
921: \frac {X^1} {X^0} C_{ab} X^{b} = F_a.
922: \end{align}
923: Dividing the first line over the second, we have $\frac {X^{0}}{X^{1}} = -\frac {F_1} {F_0}$.
924: From the third line we have $X^{b} = \frac {X^0}{X^1} C^{ba} F_{a} = -\frac{F_1} {F_0} C^{ba} F_{a}$.
925: Then $C_{ab} X^{a} X^{b} = (\frac{F_1} {F_0})^2 C^{ba}F_a F_b$. Then from the second line
926: we have $X^{0} = \frac 1 2 \frac {F_1} {(F_0)^2} C^{ba}F_{a} F_{b}$, and,
927: using $F_0/F_1 = -X^{0}/X^{1}$ we get $X^{1} = -\frac 1 2 \frac 1 {F^0} {C^{ba}F_a F_b}$.
928: We see that $X^I$ are expressed in terms of $F_I$ in the same way as $F_I$ in terms of $X^I$
929: (up to the minus sign).
930:
931: For any homogeneous function of weight two $\CalF = \frac 1 2 F_I X^{I}$, the Legendre
932: transform is given by $\tilde \CalF = F_I X^I - \CalF = \frac 1 2 F_I X^I(F_J)$.
933: We plug the expressions for $X^{I}$ and obtain
934: \begin{align}
935: \tilde \CalF = -\frac 1 2 \frac {F_1 C^{ab} F_a F_b} {F_{0}}.
936: \end{align}
937:
938:
939: Then we can use Pioline~\cite{Pioline:2005vi} formula \eqref{eq:HitTr} for
940: the Legendre transform in imaginary part of $X$ to find
941: \begin{align}
942: S_{BH} = \pi S_{Hit} = \pi \sqrt{ p^2 q^2 - (p \cdot q)^2 },
943: \end{align}
944: where the charge vectors $(p^I,q_I)$ are identified with real part of $(X^I,F_I)$,
945: and the scalar product is taken in the $(20,4)$ signature lattice.
946: This is the truncation of the full $(22,6)$ lattice for type II on $T^{2} \times K3$
947: to the charges of $\CalN=2$ vector multiplets
948: in agreement with~\cite{Cvetic:1995bj,Cvetic:1995uj,Shih:2005uc,Shih:2005he,Dabholkar:2005by,Dabholkar:2004yr,Dabholkar:2005dt,Parvizi:2005aa,Pioline:2005vi}.
949:
950:
951: \section{Conclusion}\label{sec:con}
952: In this note it was argued that the OSV conjecture~\cite{Ooguri:2004zv} is applicable to the case
953: of generalized complex structures~\cite{Hitchin:2,GCS}. If $b_1(X)=0$ one has to use
954: generalized Hitchin functional at quantum level~\cite{Pestun:2005rp}; classically
955: the generalized and the ordinary geometry does not differ. However, if $b_1(X) \neq 0$,
956: like in the case of $T^6$ or $T^2 \times K3$, the emergence of the generalized Hitchin functional
957: is inevitable at tree level.
958:
959: Deformations of a generalized complex structure on a three-fold $X$ are parameterized by the half
960: of all even/odd cycles in type $A/B$.
961: The extra moduli exist at the classical level if $H^1(X)$ is not trivial.
962: For example, the extra deformations of complex structure include $H^{1,0}$ and $H^{3,2}$ in
963: addition to the standard $H^{2,1}$. The classical black hole entropy in this case
964: is given by the \emph{generalized} Hitchin functional of the form $\int \sqrt{I_{4}(\rho)}$,
965: where $\rho$ is the real part of the canonical pure $SO(6,6)$ spinor (mixed differential form) on $X$.
966: The scalar fields in $\CalN=2$ multiplets come from all such generalized deformations,
967: and the corresponding gauge fields come from reduction of all odd/even RR forms $C_{p}$ in type II $A/B$
968: on all even/odd cycles in $X$.
969:
970: We did not touch extremely interesting subjects of higher genus and nonperturbative
971: corrections to the relation, but suggest that the generalized geometry must be an appropriate
972: framework for study of the subject.
973: Especially this is interesting in the context
974: of non CY background compactifications~\cite{Grana:2005ny,Grana:2005jc,Grana:2004bg,Chuang:2005qd,Lindstrom:2004iw,Fidanza:2003zi}. The microscopical counting of black hole entropy could illuminate
975: non-perturbative structure of the topological $\CalJ$-model~\cite{Pestun:2005gs}.
976:
977:
978: \acknowledgments
979: I would like to especially thank E. Witten for very useful comments and
980: suggestions. I also learned much from communication with N.~Nekrasov, to whom I am very grateful.
981: I thank M.~Grana, F.~Denef, A.~Kapustin, A.~Losev, A.~Neitzke, S.~Shatashvili and D.~Shih for interesting discussions.
982: Part of this research was done during my visits
983: to Institut des Hautes Etudes Scientifiques, Bures-sur-Yvette, France and
984: the 3rd Simons Workshop in Mathematics and Physics at Stony Brook University, NY.
985: I thank these institutions for their kind hospitality.
986: The work was supported in part by grant RFBR 04-02-16880 and grant NSF 245-6530.
987:
988:
989: \bibliographystyle{myunsrt}
990:
991: \bibliography{bsample}
992:
993: \end{document}
994: