1: %\documentclass[12pt,draft]{JHEP} % 10pt is ignored!

2: \documentclass[12pt]{JHEP} % 10pt is ignored!

3: \usepackage{epsfig}

4: \textwidth=420pt

5: \newcommand{\ttbs}{\char'134}           % \backslash for \tt (Nucl.Phys. :)%

6: \newcommand\fverb{\setbox\pippobox=\hbox\bgroup\verb}

7: \newcommand\fverbdo{\egroup\medskip\noindent%

8: 			\fbox{\unhbox\pippobox}\ }

9: \newcommand\fverbit{\egroup\item[\fbox{\unhbox\pippobox}]}

10: \newbox\pippobox

11: 

12: %%%%%%%%%%%%%%%%%%%%%%%%%%%%

13: %   ...	definitions    %%%%%

14: 

15: %% symbols  %%%%%%%%%%%%%%

16: 

17: \def\wgnc{\bar{\wedge}}

18: \def\del{\partial}

19: \def\der{\overline \del}

20: \def\wg{\wedge}

21: \def\gap#1{\vspace{#1 ex}}

22: \def\tgap{\vspace{3ex}}

23: \def\sgap{\vspace{5ex}}

24: \def\lgap{\vspace{20ex}}

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

26: \def\pto{\vfill\eject}

27: \def\gst{g_{\rm st}}

28: \def\tC{{\widetilde C}}

29: \def\x{{\bar x}}

30: \def\n{\mathtt{n}}

31: \def\p{p^{\bar x}}

32: \def\I{{\mathbf I}}

33: \def\K{{\mathbf K}}

34: \def\o{{\cal O}}

35: \def\J{{\cal J}}

36: \def\S{{\cal S}}

37: \def\X{{\cal X}}

38: \def\N{{\cal N}}

39: \def\M{{\cal M}}

40: \def\A{{\cal A}}

41: \def\H{{\cal H}}

42: \def\V{\mathtt{V}}

43: \def\ddelta{\mathtt{\delta}}

44: \def\pht{{\tilde \phi}}

45: \def\D{D}

46: \def\U{{\cal U}}

47: \def\V{{\cal V}}

48: \def\W{{\cal W}}

49: %\def\u{\mathtt{u}}

50: \def\u{u}

51: \def\d{{\cal D}}

52: \def\myitem#1{\gap2\noindent\underbar{#1}\gap2}

53: 

54: %%%% equations, paragraphs %%%%%

55: 

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

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

58: \def\ba{\begin{array}{l}}

59: \def\ea{\end{array}}

60: \def\bea{\begin{eqnarray}}

61: \def\eea{\end{eqnarray}}

62: \def\eq#1{(\ref{#1})}

63: \def\fig#1{Fig \ref{#1}} 

64: \def\re#1{{\bf #1}}

65: \def\bull{$\bullet$}

66: \def\ub{\underbar}

67: \def\nn{\nonumber\\}

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

69: \def\ni{\noindent}

70: \def\bibi{\bibitem}

71: %\def\c#1{{\hat{#1}}}

72: \def\eps{{\epsilon}}

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

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

75: \def\mat#1#2#3{ \langle #1 | #2 | #3 \rangle}

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

77: 

78: %%%% phrases   %%%%%%

79: 

80: \def\ads{$AdS_5 \times S^5$}

81: 

82: 

83: \pretolerance=1000000

84: 

85: 

86: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

87: 

88: \title{Fermions from Half-BPS Supergravity}

89: \author{Gautam Mandal$^*$  \\

90: {Perimeter Institute of Theoretical Physics,\\

91: 31 Caroline Street North\\ 

92: Ontario, Canada N2L 2Y5\\ 

93: ($^*$On leave from Tata Institute of Fundamental Research)

94: }

95: \\~~\\

96: \email{gmandal@perimeterinstitute.ca, mandal@theory.tifr.res.in

97: }}

98: %\received{\today} 		%%

99: %\accepted{\today}		%% These are for published papers.

100: %\JHEP{12(2001)999}		%% 

101: 

102: \preprint{\hepth{0502104}\\

103: \\

104: %TIFR/TH/04-16

105: }

106: 	% OR: \preprint{Aaaa/Mm/Yy\\Aaa-aa/Nnnnnn}

107: 			  	% Use \hepth etc. also in bibliography.  

108: 

109: \abstract{We discuss collective coordinate quantization of the

110: half-BPS geometries of Lin, Lunin and Maldacena (hep-th/0409174). The

111: LLM geometries are parameterized by a single function $u$ on a

112: plane. We treat this function as a collective coordinate. We arrive at

113: the collective coordinate action as well as path integral measure by

114: considering D3 branes in an arbitrary LLM geometry. The resulting

115: functional integral is shown, using known methods (hep-th/9309028), to

116: be the classical limit of a functional integral for free fermions in a

117: harmonic oscillator. The function $u$ gets identified with the

118: classical limit of the Wigner phase space distribution of the fermion

119: theory which satisfies $u * u = u$. The calculation shows how

120: configuration space of supergravity becomes a phase space (hence

121: noncommutative) in the half-BPS sector. Our method sheds

122: new light on counting supersymmetric configurations in supergravity.}

123: 

124: \keywords{AdS-CFT, matrix model, string theory, supergravity}

125: 

126: \begin{document}

127: 

128: \section{Introduction} 

129: 

130: Recently it has been shown in \cite{LLM} that the half-BPS IIB

131: supergravity solutions, which are asymptotically $AdS_5 \times S^5$

132: and preserve an $O(4) \times O(4)$ symmetry of the asymptotic isometry

133: group, are in one-to-one correspondence with semiclassical

134: configurations of free fermions in a harmonic oscillator potential.

135: This result is yet another striking evidence of the AdS/CFT

136: correspondence \cite{Aharony:1999ti}, since the free fermions are

137: equivalent to \cite{Berenstein:2004kk} the half-BPS sector of the

138: super Yang-Mills theory. Related work can be found in

139: \cite{Corley:2001zk,Berenstein:2004kk,deMelloKoch:2004ws,Buchel:2004mc,Filev:2004yv,Bena:2004td,Suryanarayana:2004ig,Gauntlett:2004hs,Liu:2004ru,Martelli:2004xq,Burrington:2004hf,Chong:2004ce,Liu:2004hy,Sheikh-Jabbari:2005mf,Ebrahim:2005uz}.

140: 

141: The correspondence between the supergravity configurations and

142: semiclassical fermion configurations is based on a proposed

143: identification between a supergravity mode $u(x_1, x_2)$ with the

144: phase space density $u(q,p)$ of the free fermions, where $x_1,x_2$ are

145: two of the coordinates of the LLM geometry and $q,p$ are coordinates

146: of the phase space of the free fermions.  The present work began with

147: the questions (a) how two coordinates of space time can become phase

148: space (noncommutative) coordinates and (b) whether one can derive the

149: noncommutative dynamics directly from supergravity.

150: 

151: The plan of the paper is as follows. In Section 2, we mention a few

152: results of \cite{LLM} to identify the moduli space of half-BPS

153: vacua. The moduli space is parameterized by a single function

154: $u(x_1,x_2)$ (discussed in the previous paragraph) subject to two

155: constraints. In Section 3 we quantize the half-BPS configurations by

156: identifying $u$ as the collective coordinate. We provide a

157: parameterization of the generic function $u$ subject to the

158: constraints and identify them with D3 branes coupled to LLM

159: geometries. The collective coordinate actions are then calculated by

160: computing the D3 brane actions.  We use the formalism of phase space

161: path integrals to demonstrate how the phase space dimensions get

162: reduced by half under the BPS constraint and the configuration space

163: itself becomes a phase space. In Section 4 we collect the results and

164: rewrite the action as well as the measure in terms of the

165: $u$-variable. In Section 5 we identify the $u$-functional integral

166: with the classical limit of a functional integral describing free

167: fermions in a harmonic oscillator. In Section 6 we discuss a first

168: principles approach to derivation of the $u$-functional integral using

169: the general formalism of collective coordinates in the presence of BPS

170: constraints, using Kirillov's symplectic form. Section 7 contains a

171: summary and some open questions.  In Appendix A we present some

172: details concerning identification of the collective coordinate action

173: of Section 4 with the D3-brane actions of Section 3. Appendix B makes

174: a qualitative identification between gravitons and collective

175: excitations in the form of ripples.

176: 

177: Transformation of configuration space into a phase space under BPS

178: conditions has been considered in \cite{DJM} in the case of a giant

179: graviton probe in \ads. Supertubes have been discussed in somewhat

180: related contexts in \cite{Palmer:2004gu,Bak:2004kz}.  Rather appealing

181: similarities with parts of the present work can be found in

182: discussions on topological string/field theories

183: \cite{Dijk-Vafa,Aganagic:2003db,Vafa-talk}. Related ideas have also

184: appeared in the context of quantum hall systems in

185: \cite{Susskind:2001fb,Sheikh-Jabbari:2001au}.

186: 

187: \section{The moduli space of 1/2-BPS Supergravity }

188: 

189: As shown in \cite{LLM}, the half-BPS geometries (with $O(4) \times

190: O(4)$ symmetry) are characterized by a single function $z_0(x_1,x_2)

191: \equiv z(x_1,x_2,y=0)$ (see eqs. (2.5)-(2.15) of \cite{LLM}).  The

192: moduli space of these solutions is the space of $z_0$'s, subject to the

193: following regularity and topological constraints.

194: 

195: \myitem{The regularity constraint}

196: 

197: The constraint of regularity on the half-BPS geometries 

198: implies that $z_0$ can only be either $1/2$ or $-1/2$, that is

199: \footnote{$\chi_R(x)$ denotes the

200: characteristic function  of a region $R \subset {\bf R}^2$: 

201: \be

202: \chi_R(x) = 1 \ \hbox{if} \ x \in R, = 0\  \hbox{otherwise}.

203: \ee}

204: \be

205: z_0(x_1,x_2) = - \frac12\sum_i \chi_{R_i} + \frac12 \sum_j \chi_{\tilde R_j} 

206: \label{z-char}

207: \ee

208: where the $x_1,x_2$ plane is tessellated by the regions $R_i,\tilde R_j$,

209: with $z_0=-1/2$ in $R_i$ and  $z_0=1/2$ in the $\tilde R_j$. 

210: 

211: It is useful to define the function 

212: \be

213: \u(x_1, x_2) \equiv 1/2 - z_0(x_1, x_2)

214: \ee

215: The regularity constraint now reads $ \u(x_1, x_2) = 0 \ \hbox{or} \ 1

216: $, equivalently\footnote{$0 < u < 1$ gives rise to singular

217: solutions; e.g.\cite{Caldarelli:2004mz} identified the superstar solution 

218: \cite{Myers:2001aq,Cvetic:1999xp,Behrndt:1998jd} with

219: $0<u<1$. \cite{Caldarelli:2004mz} also showed in specific examples  

220: that the geometries with $u>1$ develop closed timelike curves.}

221: \be

222: \left(\u(x_1, x_2)\right)^2 = \u(x_1, x_2)

223: \label{quad-const}

224: \ee

225: The equation \eq{z-char} becomes 

226: \be

227: \u = \sum_i \chi_{R_i}(x_1,x_2)

228: \label{u-char}

229: \ee

230: where $R_i$ now denote regions with $u=1$.

231: 

232: \myitem{The topological constraint}

233: 

234: The topological constraint becomes \cite{LLM} 

235: \bea

236: \int_{R_i}\frac{dx_1dx_2}{2\pi \hbar} 

237: &&= N_i

238: \nn 

239: \int_{-\infty}^\infty \frac{dx_1dx_2}{2\pi \hbar} 

240: \u 

241: &&= \sum_i  N_i = N

242: \label{int-const}

243: \eea

244: where

245: \be

246: \hbar = 2\pi g_s \alpha'^2

247: \label{hbar-llm}

248: \ee

249: The condition that the geometries are asymptotically $AdS_5 \times

250: S^5$  implies that $R= \cup R_i$ is a

251: bounded region of the $x_1,x_2$ plane.

252: 

253: The functions $u(x_1,x_2)$ subject to the constraint

254: equations \eq{quad-const} and \eq{int-const} characterize

255: all regular half-BPS solutions of the system with $O(4) \times

256: O(4)$ symmetry and \ads\ asymptotics. 

257: 

258: \section{Quantization of half-BPS vacua}

259: 

260: We will treat the function $u$ as the collective coordinate of the

261: space of half-BPS configurations (with $O(4) \times O(4)$

262: symmetry). The space of $u$'s can be discussed in terms of orbits of a

263: specific $u_0$ under the action of the group of area-preserving

264: diffeomorphisms in two dimensions (see Section \ref{remarks} for this

265: description). Alternatively, $u$ can be parameterized as in

266: \eq{u-char}. By choosing generic enough regions $R_i$, we can describe

267: all functions $u$ subject to the constraints. This is the description

268: we will use in this and the following two sections to quantize the

269: space of $u$'s.

270: 

271: Let us choose

272: the regions as follows (see figure \ref{checkerboard.fig}):

273: \begin{figure}[ht]

274: %or [htb]

275: \vspace{0.5cm}

276: \hspace{-0.5cm}

277: \centerline{

278:     \epsfxsize=8.5cm

279:    \epsfysize=8cm

280:    \epsffile{checkerboard.eps}

281:  }

282: \caption{\sl Checkerboard parameterization. The white rectangles

283: inside the circle represent the regions $H_j$ in \eq{checkerboard},

284: while the black rectangles outside the circle denote the regions

285: $P_i$. A small number of isolated cells represents

286: giant gravitons in $S^5$ or in $AdS_5$. When the number of cells

287: is large, each additional cell (black or white) can be regarded

288: as a D3 brane in an arbitrary background LLM geometry defined by

289: the rest of the pattern.}

290: \label{checkerboard.fig}

291: \end{figure}

292: \be

293: u(x_1,x_2) = u_0(x_1,x_2) - \sum_{j=1}^m \chi_{H_j}(x_1, x_2)

294: + \sum_{i=1}^n \chi_{P_i}(x_1, x_2)

295: \label{checkerboard}

296: \ee

297: Here $u_0$ represents a filled circle of radius $r_0$:

298: \be

299: u_0 = \theta(r_0 - r)

300: \label{u0}

301: \ee

302: and the regions $H_j, P_i$ are non-intersecting rectangular

303: cells, with $H$'s (holes) inside the circle of radius $r_0$ and 

304: $P$'s outside the circle.  

305: 

306: The constraint \eq{quad-const} is obviously satisfied. The other

307: constraint \eq{int-const} can also be easily satisfied, by choosing

308: the area of each of the cells $H_j$ or $P_i$ to be integral (in units

309: of $2\pi \hbar$) and by choosing the radius $r_0$ in \eq{u0} so as to

310: keep the total area equal to $N$. Clearly the minimum area of the

311: cells $H_j$ or $P_i$ is $2\pi \hbar$. In the limit of a large number

312: of such cells, arbitrarily scattered, we can recover a rather 

313: general\footnote{See footnote \ref{tiling}.} representation of the type

314: \eq{u-char}, subject to \eq{quad-const} and \eq{int-const}.

315: 

316: Thus, in \eq{checkerboard} we will choose the $H_j$ to be minimum area

317: cells (we will take them to be squares without loss of generality,

318: with each side equal to $\sqrt{2\pi \hbar} \equiv \eps $), with

319: centres denoted by $(x_1^j, x_2^j), j=1,...,m$. Similarly we will take

320: $P_i$'s to be squares of the same minimal size, with centres denoted

321: by $(x_1^i, x_2^i), i=1,...,n$.

322: 

323: The specific rectangular shape of the cells is not important for our

324: discussions (except for visualizing a simple tiling\footnote{

325: \label{tiling} The tiling is only in an approximate sense since

326: we will regard the cell boundaries as separated by distances $>

327: O(\sqrt{\hbar})$ to prevent high curvatures arising from droplets that are too

328: close; such inter-cell separations can be interpreted in terms of

329: fuzzy $u$-configurations satisfying \eq{u-star-u} in the finite

330: $\hbar$ theory (see Sections \ref{nc} and \ref{remarks}). Droplets

331: closer than this distance can be assumed to merge, leading to

332: ``ripples''.  These are proposed in \cite{LLM} to correspond to

333: gravitons; we briefly explore the correspondence between gravitons and

334: the collective action for ripples in Appendix \ref{graviton}.}).  The

335: same results could be derived, e.g. by using cells with sides along

336: the $r$ and $\phi$ directions.

337: 

338: \subsection{\label{correspondence}Correspondence between checkerboard 

339: configurations and IIB geometries}

340: 

341: The correspondence with  IIB geometries, following 

342: \cite{LLM}, is described below:  

343: 

344: (a) When there are no $H$'s or $P$'s, the circle of radius $r_0$

345: represents \ads, where $r_0$ is given by \eq{valu-r0}.

346: 

347: (b) A configuration \eq{checkerboard} with a small number of

348: non-intersecting minimum-area cells $P_i$ and $ H_j$ represents giant

349: gravitons wrapping the three-spheres of $AdS_5$ or $S^5$, so that

350: the background configuration is essentially the same as

351: in the case (a) above.  The cell

352: $P_i$ will represent the $i$-th giant graviton extending in $AdS_5$

353: (such giant gravitons are called ``dual giant gravitons''

354: \cite{Grisaru:2000zn,Hashimoto:2000zp}).  The centre of mass of the

355: giant graviton will be identified with the centre $(x_1^i, x_2^i)$ of

356: the cell $P_i$. Similarly, the cell $H_j$ will represent the $j$-th

357: giant graviton extending in $S^5$ \cite{McGreevy:2000cw}. The centre

358: of mass of the giant graviton will be identified with the centre

359: $(x_1^j, x_2^j)$ of the cell $H_j$.

360: 

361: (c) A single minimum-area cell $H_j$ (hole) inside the filled part of a

362: generic $u$-configuration (representing an arbitrary LLM geometry) will

363: be identified as a D3-brane wrapping the three-sphere $\tilde S^3$ of

364: that geometry\footnote{We will not consider collective excitations

365: corresponding to multiple D3 branes, except to remark that 

366: two D3 branes which are classically on top of each other are

367: described in \cite{LLM} as a spread-out $u$ configuration occupying

368: twice the area, to be consistent with the constraints

369: \eq{quad-const} and \eq{int-const}. This accords  with

370: the fermionic description \eq{z-nc} or \eq{z-f} which we

371: ultimately arrive at.} (see \eq{giant-arbit}).

372: 

373: (d) Similarly, a minimum area cell $P_i$ in the unfilled part of an

374: arbitrary $u$-configuration will be identified as a D3-brane wrapping

375: the three-sphere $S^3$ of the corresponding geometry (see

376: \eq{giant-arbit-1}).

377: 

378: \subsection{\label{recipe}Recipe for the collective coordinate action}

379: 

380: We will derive the collective coordinate action\footnote{An

381: independent derivation, more directly from supergravity, based on

382: Kirillov's symplectic form, is briefly sketched in Section

383: \ref{remarks} (see point (2) and the references therein for details.)}

384: based on the above correspondences.  For example, for configurations

385: (b), the collective coordinate action for the $u$-fluctuation

386: represented by a cell $H_j$ or $P_i$ will be identified with the

387: action of the corresponding giant (or dual giant) graviton, subject to

388: the half-BPS constraint.  Similarly, for configurations (c) and (d),

389: the collective coordinate action will be identified with the action of

390: the corresponding D3-branes in an arbitrary LLM geometry, subject to

391: the half-BPS conditions.

392: 

393: To describe our method, let us consider the example of the case (c),

394: where we create a `hole' (a white pixel) at the position $(\bar x_1,

395: \bar x_2)$.  This changes the initial $u$-configuration

396: from an arbitrary initial value  $u_0$ to  $ u_0 - \delta u$

397: (where $\delta u$ is given by \eq{single-u} for a

398: rectangular hole). As mentioned above, this deformation $\delta u$

399: should be identified with a BPS D3 brane which wraps the 3-sphere

400: $\tilde S^3$ of the LLM geometry $u_0$ and is located

401: at $(\bar x_1, \bar x_2)$. The collective coordinate action $S[u]$

402: that we are looking for should, therefore, satisfy the property that

403: $\delta S = S[u_0 - \delta u] - S[u_0]$ should be identical to the action

404: $ S_{D3}^{BPS} $ (DBI + CS) of the above-mentioned  D3 brane%%%%

405: \footnote{\label{compensating} Note that 

406:  the configuration $u_0 - \delta

407: u$ does not preserve the area constraint \eq{int-const}. So we must

408: create another deformation $+ \delta u'$ by adding a ``particle'', or

409: inflating the periphery of one of the droplets comprising $u_0$. In principle

410: $\delta S$ could depend on the choice of  $+ \delta u'$; however,

411: it is easy to show that the effect is subleading in $1/N$ and

412: we will ignore it. This is consistent with the fact that the 

413: action $S[u]$ we will arrive at agrees with the  fermion action

414: in the semiclassical limit. We will discuss this further in

415: Section 7, point (6).}.

416: 

417: Similar considerations apply to the case (d), where one adds a

418: `particle' (a black pixel) at the position $(\bar x_1, \bar x_2)$ so

419: that $u_0 \to u_0 + \delta u$, with $\delta u$ given by \eq{single-u}.  In

420: this case one demands that $S[u]$ should satisfy the property that the

421: change in $S[u]$ should be equal to the action of a half-BPS D3-brane

422: at that position, wrapping $S^3$. The case (b) is of course simpler

423: where the background geometry is \ads\ and the D3-branes are the usual

424: giant or dual giant gravitons.

425: 

426: With the above understanding of terms, the classical action

427: $S[u]$ should satisfy the property 

428: \be

429: \delta S = S_{D3}^{BPS} 

430: \label{main-eqn}

431: \ee  

432: for an arbitrary choice of the fluctuation $\pm \delta u$,

433: around any background $u_0$.

434: 

435: We will find that such an $S[u]$ indeed exists (same as the one obtained

436: using the Kirillov form, Section \ref{remarks}).

437:   

438: Besides a classical action $S[u]$, we

439: will also find a measure $D[u]$ such that the measure for the

440: fluctuation $D[\delta u]_{\tilde u_0}$ agrees with the path integral

441: measure of the D3-brane dynamics.

442: 

443: Note that we are making the identification of the D3 brane

444: degrees of freedom with the collective coordinates of the

445: supergravity background. We are assuming this, as in \cite{LLM}.

446: This is similar in spirit with the original identification

447: by Polchinski \cite{Polchinski-D,Rey-collective} of 

448: Dirichlet branes as collective coordinates of supergravity

449: backgrounds carrying Ramond-Ramond charges.

450:  

451: We will discuss a more first principles approach in a later

452: section (Section \ref{remarks}).

453: 

454: Let us now consider, in turn, the D3-branes corresponding to

455: configurations (b), (c) and (d) of Section \ref{correspondence}.

456: 

457: \subsection{Single giant graviton in \ads}

458: 

459: In this and the next subsections we will describe the calculation

460: of the right hand side of \eq{main-eqn} in the cases (b), (c) and (d)

461: respectively. In Section 4 the action $S[u]$ and the calculation of

462: $\delta S$ in the left hand side of \eq{main-eqn} will be discussed.

463: 

464: We will first consider a giant graviton extending in $S^5$

465: \cite{McGreevy:2000cw}. As discussed above, this corresponds to a 

466: hole $H$ with each side equal to $\eps = \sqrt{2 \pi \hbar}$. We will

467: denote the centre of $H$ as $(\bar x_1, \bar x_2)$. The change in the

468: $u$-function corresponding to creation of the hole is $-\delta u$

469: where

470: \bea

471: \delta u &&= \chi_{\bar x_1, \bar x_2}(x_1, x_2) \nn

472: {}&&\equiv 

473: \theta(\bar x_1 + \eps/2 - x_1)

474: \theta(-\bar x_1 + \eps/2 + x_1) \theta(\bar x_2 + \eps/2 - x_2)

475: \theta(-\bar x_2 + \eps/2 + x_2)

476: \nn

477: \label{single-u}

478: \eea

479: 

480: We will now discuss the calculation of the

481: right hand side of \eq{main-eqn}, namely the giant graviton

482: action. Half-BPS configurations of a giant graviton extending in 

483: $S^5$ of \ads have been discussed in \cite{DJM}. The giant graviton

484: is a D3-brane with the embedding (in static gauge) 

485: \be

486: t = \tau, \theta = \theta(\tau), \tilde\phi = \tilde\phi(\tau),

487: \tilde\Omega_i= \sigma_i, \rho =0

488: \label{giant-embed}

489: \ee

490: where we have used global coordinates of \ads, defined by the metric

491: \be

492: ds^2 = r_0\Big[

493: -\cosh^2\rho~ dt^2 + d\rho^2 + \sinh^2\rho~ d\Omega_3^2

494: + \cos^2\theta ~d\pht^2 + d\theta^2 + \sin^2\theta~ d \tilde \Omega_3^2 

495: \Big]

496: \label{global}

497: \ee

498: Here 

499: \be

500: r_0^2 = R_{AdS}^4 = 4\pi N l_p^4 = 4 \pi N g_s \alpha'^2

501: \label{valu-r0}

502: \ee 

503: The relation to the LLM coordinates   is

504: \bea

505: r &&= r_0 \cosh\rho \cos\theta

506: \nn

507: y &&= r_0 \sinh\rho \sin\theta

508: \label{coord-trans}

509: \eea

510: and 

511: \be

512: \phi = \tilde\phi + t

513: \label{phi-trans}

514: \ee 

515: For $y=0$, we have 

516: \be

517: r = r_0 \cos\theta

518: \label{r-for-giant}

519: \ee

520: We have used the notation 

521: $(r,\phi)$ as polar coordinates for the $(x_1,x_2)$ plane. 

522: The D3 brane action is given by \footnote{$\tilde\phi$ here is

523: $-\phi$ of \cite{DJM}.}

524: \be

525: S= N \int d\tau\ \Big[-\sin^3\theta\sqrt{1 - \cos^2\theta\dot\pht^2

526: - \dot\theta^2} - \sin^4\theta~\dot\pht\Big]

527: \label{giant-action}

528: \ee

529: The factor $N$ in front arises as 

530: \be

531: N = T_3 \omega_3 r_0^2,

532: \label{genesis-of-n}

533: \ee 

534: where

535: $T_3 = 1/(8\pi^3 \alpha'^2 g_s)$ is the D3-brane tension,

536: $\omega_3 = 2\pi^2$ is the volume of the unit $S^3$ and $r_0^2$

537: is given in \eq{valu-r0}.

538: 

539: The configuration space of the giant graviton is given by

540: $\theta(\tau), \pht(\tau)$.

541:  This corresponds to a

542: four-dimensional phase space $\theta(\tau), p_\theta(\tau),

543: \tilde\phi(\tau), p_\pht(\tau)$. 

544: It is easy to see that for BPS configurations

545: we must have \cite{DJM}

546: \be

547: \dot\theta =0, \dot{\tilde \phi}= -1

548: \label{vel-const}

549: \ee

550: or, equivalently,

551: \be

552: p_\theta=0, p_{\tilde\phi}= - N \sin^2\theta

553: \label{mom-const}

554: \ee

555: In \cite{DJM} the BPS constraints 

556: \eq{mom-const} were imposed as

557: Dirac constraints on the four dimensional phase space.

558: The result was a two dimensional phase space which

559: could be coordinatized by $\theta,\tilde\phi$ which

560: satisfied the following Dirac bracket:

561: \be

562: \{ \theta, \pht \}_{DB} = {1\over 2 N \sin\theta \cos\theta},

563: \quad {\rm or}\ \{\sin^2 \theta, \phi \}_{DB} = 1/N

564: \ee

565: The Hamiltonian in the reduced phase space is given by

566: \footnote{\label{moving}If we use the ``moving coordinate'' $\phi$, the 

567: Hamiltonian becomes $H = \tilde H + p_\pht = \tilde H + p_\phi =0$.

568: This is a reflection of the relation 

569: $ \del/\del t|_{\phi} = \del/\del t|_{\pht} + \del/\del \pht|_{t} $.

570: See also remarks below equation \eq{x1-x2-pht}.

571: }

572: \be

573: \tilde H = - p_\pht = N \sin^2\theta

574: \label{h-red}

575: \ee

576: Another way of stating the above result is that the

577: unconstrained  path integral

578: for the system

579: \be

580: Z_{full}= \int \D\theta(\tau)\D p_\theta(\tau)\D\pht(\tau)

581: \D p_\pht(\tau) \exp

582: \Big[i\int d\tau \Big(\dot\pht p_\pht + \dot\theta p_\theta - H_{full}\Big) 

583: \Big]

584: \label{z-full}

585: \ee

586: reduces, under the BPS constraints, to the following path

587: integral

588: \be

589: Z_{BPS}= \int \D[\sin^2\theta(\tau)]\D[\pht(\tau)]

590: \exp\Big[i \int d\tau \left( -N \sin^2\theta\dot\pht - \tilde H \right)\Big]

591: \label{path-for-giant}

592: \ee

593: where $\tilde H$ is given by \eq{h-red}.

594: We will show in Section \ref{action} how the above functional

595: integral can be written in terms of the $u$-variable

596: in the sense of Section \ref{recipe} (in

597: particular \eq{main-eqn}).

598: 

599: \gap{2}

600: 

601: The treatment of the dual giant graviton, extending into $AdS_5$

602: \cite{Grisaru:2000zn,Hashimoto:2000zp} of \ads, is very similar. The

603: corresponding $u$-configuration consists of a single cell $P_i$ outside

604: of $u_0$. Thus, 

605: \be

606: u(x_1, x_2) = u_0 + \delta u

607: \label{single-u-1}

608: \ee

609: where $\delta u$ is again given by the expression in

610: \eq{single-u}.

611: 

612: The D3 brane embedding for the dual giant graviton  is

613: \be

614: t=\tau, \rho=\rho(\tau),\pht= \pht(\tau),\Omega_i=\sigma_i,\theta=0

615: \ee

616: For this embedding, $r$ gets related to $\rho$ as follows:

617: \be

618: r=r_0 \cosh\rho

619: \label{r-for-dual}

620: \ee

621: The BPS constraints are:

622: \be

623: p_\rho=0, p_\pht = -N \sinh^2 \rho

624: \ee

625: The constrained path integral (the analog of \eq{path-for-giant})

626: now is 

627: \bea

628: Z_{BPS} &&= \int \D[\sinh^2\rho(\tau)]\D[\pht(\tau)]

629: \exp\Big[i\int d\tau \left(-N \sinh^2\rho\ \dot\pht - 

630: \tilde H \right)\Big]

631: \nn

632: \tilde H &&= - p_\pht =  N \sinh^2 \rho

633: \label{path-for-giant-1}

634: \eea

635: We will show in Section \ref{action} that this is also

636: a special case of the same $u$-path integral

637: as the earlier example was.

638: 

639: \subsection{\label{case-c} D3 brane in arbitrary LLM geometry}

640: 

641: Let us first consider configuration (c) of Section

642: \ref{correspondence}, where we have a single cell $H$ (hole) inside a

643: filled (black) region of an arbitrary $u$-configuration,

644: which we will write as 

645: (see Section \ref{recipe})

646: \be

647: u(x_1, x_2)=  u_0 -  \delta u 

648: \label{single-u-arbit}

649: \ee

650: where $\delta u$ is again as in \eq{single-u}, but

651: $u_0$ represents an arbitrary background $u$-configuration,

652: satisfying the constraints \eq{int-const}, \eq{quad-const}.

653: We will ignore here the area-compensating change $\delta u'$ as discussed

654: in footnote \ref{compensating}.

655: 

656: The D3 brane corresponding to the fluctuation

657: \eq{single-u-arbit} is described by the following embedding

658: (using the LLM coordinates, see \eq{llm-metric}):

659: \be

660: t = \tau, x_1 = \bar x_1(\tau), x_2 = 

661: \bar x_2(\tau), y=0, \tilde \Omega_m = \sigma_m, ~ m=1,2,3

662: \label{giant-arbit}

663: \ee

664: Let us discuss the geometry corresponding to $u_0$.

665: Recall that the LLM metric is of the form \cite{LLM}

666: \be

667: ds^2 = g_{tt} (dt + V_i dx_i)^2 + g_{yy}(dx_i dx_i + dy^2)

668: + g_{\Omega\Omega}~ d\Omega_3^2 +

669: g_{\tilde\Omega \tilde\Omega}~d\tilde\Omega_3^2

670: \label{llm-metric}

671: \ee

672: where $d\Omega_3^2, d\tilde\Omega_3^2$ represent metric

673: on two unit 3-spheres $S^3$ and $\tilde S^3$ respectively

674: (the two 3-spheres

675: are distinguished by the fact that $S^3$ has vanishing radius

676: in the $u=1$ region of the $\vec x$-plane (see

677: \eq{black}), whereas $\tilde S^3$

678:  has vanishing radius

679: in the $u=0$ region of the $\vec x$-plane (see

680: \eq{white}).

681: The parts of the metric and RR background which are

682: important for us are near $y=0$:  

683: \bea 

684: u_0 &&= 1- y^2 f 

685: \nn 

686: V_i && = v_i, ~i=1,2 

687: \nn 

688: -g_{tt} &&= 1/g_{yy} = f^{-1/2} 

689: \nn

690: g_{\Omega\Omega} &&= y^2 \sqrt{f} 

691: \nn 

692: g_{\tilde\Omega \tilde\Omega} &&=f^{-1/2} 

693: \nn 

694: B_t &&= - {1\over 4} y^4 f 

695: \nn 

696: \tilde B_t && = -{1\over 4f} 

697: \nn 

698: d \hat B &&= - {1\over 4}y^3 *_3 df 

699: \nn 

700: d{\tilde {\hat B}} 

701: && = -\frac12 dx_1 \wedge dx_2 = -\frac14 d (x_1~dx_2 - x_2~dx_1) 

702: \label{black}

703: \eea 

704: Here $*_3$ is the flat space epsilon symbol in the three

705: dimensions parameterized by $y,x_1,x_2$.

706: All expressions on the right hand sides are understood to be multiplied by

707: $(1 + O(y^2))$. $f(x_1,x_2), v_i(x_1,x_2)$ are both obtainable from

708: $u_0(x_1,x_2)$. Explicitly,

709: \bea

710: f(\vec x) &&= {\rm Limit}_{y \to 0}[\frac1{y^2} - \frac1\pi

711: \int_{D}\frac{d^2\vec x'}{[(\vec x - \vec x')^2 + y^2]^2}]

712: \nn

713: v_i(\vec x) && =

714: {\eps_{ij}\over 2\pi}\oint_{\del D} \frac{dx'_j}{ (\vec

715: x - \vec x')^{2}}

716: \label{def-f-v}

717: \eea

718: Here $D$ denotes the support of $u$.

719: The limit for $f$ is well-defined since the explicit $1/y^2$ cancels

720: with a $1/y^2$ coming from the $\vec x = \vec x'$ region

721: of the integral. It is easy to calculate explicit forms for $f$,

722: for example, for ring configurations of $\tilde u_0$. 

723: 

724: Under the approximations \eq{black} the  metric

725: and the RR 4-form field are given, upto $(1 + O(y^2))$, by

726: \bea

727: ds^2 && = [-( dt + v_i dx_i)^2 + f(dx_1^2 + dx_2^2

728: + d\vec y^2) +  d\tilde\Omega^2]/\sqrt{f}

729: \nn

730: C^{(4)} && = -\frac14 \Big[ \frac{dt + v_i dx_i}f + r^2 d\phi \Big]

731: \wedge d^3\tilde \Omega

732: \label{fields}

733: \eea

734: where $d\vec y^2= dy^2 + y^2 d\Omega^2$, and $d^3\tilde \Omega$

735: is the volume form of the three-sphere $\tilde S^3$ (see

736: \eq{llm-metric}).

737: 

738: The D3 brane action is given by\footnote{Note the appearance in the

739: second line of the

740: $\hbar$ of \eq{int-const},\eq{hbar-llm} through the

741: equality $T_3 w_3 = N/r_0^2 \equiv 1/(2 \hbar)$, cf.

742: \eq{genesis-of-n}.} (dropping the bar's on $x_i(t)$

743: in \eq{giant-arbit})

744: \bea

745: S && = T_3 \omega_3 \int d\tau \Big[

746: -\frac1f \sqrt{(1 + v_r \dot r + v_\phi \dot \phi)^2  - 

747: f(\dot r^2 + r^2 \dot \phi^2)}

748: + r^2 \dot \phi + \frac1f (1 + v_r \dot r + v_\phi \dot \phi) \Big]  

749: \nn

750: &&= \frac{1}{2 \hbar} \int dt \Big[

751: -\frac1f \sqrt{(1 + v_r \dot r + v_\phi (\dot \pht +1))^2  - 

752: f(\dot r^2 + r^2 (\dot \pht +1)^2)} \Big.

753: \nn

754: &&~~~~~~~~~~~~~~~~~

755: \Big.  + r^2 (\dot\pht +1) + \frac1f (1 + v_r \dot r + v_\phi(\dot \pht +1)) 

756: \Big]  

757: \label{action-arbit}

758: \eea

759: The BPS conditions can be obtained by the constraint

760: $\tilde H = - p_\pht$, which gives

761: \be

762: \dot\pht = -1, \dot r=0

763: \ee

764: In the $\phi, t$ coordinates

765: \be

766: \dot\phi = 0, \dot r=0

767: \label{vel-const-arbit}

768: \ee

769: The Hamiltonian $H$ in the LLM frame is $H=0$ (see footnote

770: \ref{moving}). It should be possible to derive 

771: these equations from an analysis

772: of the Killing spinor and world-volume kappa-symmetry, but

773: another way of seeing the validity of equations \eq{vel-const-arbit}

774: is that it is equivalent to time-independence of $\delta u$ in 

775: \eq{single-u-arbit}. Any such time-independent

776: $u$-configuration is half-BPS, as shown in

777: \cite{LLM}; indeed the half-BPS condition 

778: does not allow

779: any time-dependence of $u$. Hence \eq{vel-const-arbit} is

780: equivalent to the Killing spinor condition. 

781: 

782: The remaining analysis is similar to the case of the giant

783: gravitons in \ads. On the constrained surface \eq{vel-const-arbit}

784: we have

785: \be

786: p_r =0, p_\pht = \frac1{2 \hbar} r^2

787: \ee

788: The Hamiltonian is given by 

789: \be

790: \tilde H= - p_\pht = -  \frac1{2 \hbar} r^2,

791: \label{ham-giant-arbit}

792: \ee

793: the negative sign reflecting the energy of a hole. 

794: 

795: The constrained path integral, the analog of

796: \eq{path-for-giant}, now becomes

797: \bea

798: Z_{BPS} &&= \int \D[r^2(\tau)]\D[\pht(\tau)]

799: \exp[i S_{BPS}]

800: \nn

801: S_{BPS} &&= \int d\tau \left( \frac1{2 \hbar} r^2 \dot\pht - \tilde H 

802: \right)

803: \label{path-for-giant-arbit}

804: \eea

805: where $\tilde H$ is given by \eq{ham-giant-arbit}. To compare 

806: with \eq{path-for-giant}, note that on \eq{r-for-giant}

807: $r^2/(2\hbar)= N \cos^2\theta = N - N \sin^2\theta$. The extra $N$

808: is explained in the paragraph following \eq{delta-s-ham}.

809: 

810: \gap{2}

811: 

812: Let us now consider configuration (d), where we have a single 

813: (black) cell $P$ 

814: in a white   region  of an arbitrary $u$-configuration. The full

815: u-configuration, including contribution from $P$ is given by 

816: \be

817: u(x_1, x_2) =  u_0 + \delta u

818: \label{single-u-arbit-2}

819: \ee

820: where $\delta u$ is given by \eq{single-u}. 

821: 

822: As in \eq{black}, the important parts of the metric and RR background are near

823: $y=0$. These are now given by

824: \bea

825: u_0 &&= y^2 f

826: \nn

827: V_i && = v_i 

828: \nn

829: -g_{tt} &&= 1/g_{yy} = f^{-1/2}

830: \nn

831: g_{\Omega\Omega} &&= f^{-1/2}

832: \nn

833: g_{\tilde\Omega  \tilde\Omega} &&= y^2 \sqrt{f}

834: \nn

835: B_t &&=  - {1\over 4f}

836: \nn

837: \tilde B_t && =- {1\over 4} y^4 f

838: \nn

839: d \hat B &&= 

840: \frac12 dx_1 \wedge dx_2 = \frac14 d (x_1~ dx_2 - x_2~ dx_1)

841: \nn

842: d B && =  {1\over 4}y^3 * df

843: \label{white}

844: \eea

845: All expressions on the right hand sides are understood to be 

846: multiplied by

847: $(1 + O(y^2))$. $v_i$ are again given by \eq{def-f-v},

848: while $f= (1/\pi) \int_D d^2 \vec x'~(\vec x - \vec x')^{-4}$.

849: 

850: The metric and the RR form are given by  

851: \bea

852: ds^2 && = [-( dt + v_i dx_i)^2 + f (dx_1^2 + dx_2^2

853: + d\vec y^2) + d \Omega^2]/\sqrt{f}

854: \nn

855: C^{(4)} && = -\frac14 \Big[ \frac{dt + v_i dx_i}f - r^2 d\phi \Big]

856: \wedge d^3\Omega

857: \label{fields-1}

858: \eea

859: where $d\vec y^2= dy^2 + y^2 d\tilde \Omega^2$ and $d^3\Omega$

860: represents the volume-form on $S^3$ (see \eq{llm-metric}). 

861: 

862: Let us consider the D3 brane represented by $\delta u$ in

863: \eq{single-u-arbit-2}.  Its embedding is given by

864: (using, again, the LLM coordinates of \eq{llm-metric})

865: \be

866: t=\tau, x_1= \bar x_1(\tau), x_2= \bar x_2(\tau), y=0,

867: \Omega_m = \sigma_m, ~m=1,2,3

868: \label{giant-arbit-1}

869: \ee

870: The D3 brane action, analogous to \eq{action-arbit}, is

871: given by  (dropping the bar's on $x_i(\tau)$)

872: \bea

873: S && = T_3 \omega_3 \int d\tau \Big[

874: -\frac1f \sqrt{(1 + v_r \dot r + v_\phi \dot \phi)^2  - 

875: f(\dot r^2 + r^2 \dot \phi^2)}

876: - r^2 \dot \phi + \frac1f (1 + v_r \dot r + v_\phi \dot \phi) \Big]  

877: \nn

878: &&= \frac{1}{2 \hbar} \int d\tau \Big[

879: -\frac1f \sqrt{(1 + v_r \dot r + v_\phi (\dot \pht +1))^2  - 

880: f(\dot r^2 + r^2 (\dot \pht +1)^2)} \Big.

881: \nn

882: &&~~~~~~~~~~~~~~~~~

883: \Big.  - r^2 (\dot\pht +1) + \frac1f (1 + v_r \dot r + v_\phi(\dot \pht +1)) 

884: \Big]  

885: \label{action-arbit-2}

886: \eea

887: The BPS condition $H= - p_\pht$, once again equivalent to

888: $\dot \pht = -1, \dot r =0$, implies that the BPS dynamics

889: is described by the path integral (analog of \eq{path-for-giant-arbit})

890: \bea

891: Z_{BPS} &&= \int \D[r^2(\tau)]\D[\pht(\tau)]

892: \exp[i S_{BPS}]

893: \nn

894: S_{BPS} &&=  \int d\tau \left(

895: -\frac{r^2}{2\hbar} \dot\pht - \tilde H \right)

896: \nn

897: \tilde H  && = - p_\phi = \frac1{2\hbar} r^2

898: \label{path-for-giant-arbit-1}

899: \eea

900: Note that the Hamiltonian for the filled cell is

901: positive this time. For comparison with \eq{path-for-giant-1}, remarks

902: similar to the ones below \eq{path-for-giant-arbit} apply here as

903: well (note that according to \eq{r-for-dual} $r^2/(2\hbar) = N \cosh^2\rho

904: = N + N \sinh^2 \rho$).

905: 

906: \section{\label{action}Collective coordinate action}

907: 

908: We will now show that all the path integrals 

909: \eq{path-for-giant},\eq{path-for-giant-1},\eq{path-for-giant-arbit}

910: and \eq{path-for-giant-arbit-1} are 

911: equivalent to the  following path integral in terms

912: of the $u$-variable:

913: \bea

914: Z &&= \int \D\u \exp[i S_{BPS}]

915: \nn

916: S_{BPS}

917: && = \int {dx_1 dx_2 \over 2\pi \hbar}

918: \hbar \int_{\tilde \Sigma} d\tau\ 

919: ds\ \u(x_1,x_2,\tau,s)\{\del_\tau \u, \del_s \u \}_{PB} -  \int_\Sigma d\tau

920: \tilde H 

921: \label{u-action}

922: \nn

923: \tilde H &&= \int {dx_1 dx_2 \over 2\pi \hbar}

924: u(x_1,x_2,\tau) \frac{x_1^2 + x_2^2}{2 \hbar}

925: \label{path-for-u}

926: \eea

927: Here $\Sigma$ denotes a 

928: curve $\tau\! \mapsto\! u(x_1,x_2,\tau)$

929: in the $u$-configuration space and  $\tilde \Sigma$ 

930: denotes a one-parameter  extension of $\Sigma$

931: to the map $(\tau,s)\! \mapsto \! u(x_1,x_2,\tau,s), s < s_0,$ such that

932: $u(x_1,x_2,\tau,s_0)= u(x_1,x_2,\tau)$. Although in order

933: to write the action we need to introduce the $s$-extension,

934: it can be easily shown that the

935: extension does not affect the path integral as long as the boundary

936: value (at $s=s_0$) remains $u(x_1,x_2,\tau)$ (this

937: follows from the fact that the symplectic form 

938: appearing in \eq{u-action} is closed). In this and the

939: following section we use 

940: \be

941: (x_1, x_2) = (r \cos\pht, r \sin\pht)

942: \label{x1-x2-pht}

943: \ee

944: (see eq. \eq{phi-trans}). The $\pht$ coordinate, rather than $\phi$,

945: is the more natural angle to use for comparison with the

946: boundary theory, because, e.g., the time-derivative in the boundary

947: theory is the operator $\del/\del t|_\pht$ appearing in 

948: footnote \ref{moving}. In terms of $(r,\phi)$ coordinates

949: the Hamiltonian is zero (see footnote \ref{moving}).  

950: 

951: The notation $\{ \}_{PB}$ is defined here as

952: \[

953: \{ f, g \}_{PB} \equiv \frac{\del f}{\del x_1}\frac{\del g}{\del x_2}

954: - \frac{\del g}{\del x_1}\frac{\del f}{\del x_2}

955: \]

956: 

957: We will see later (see Section \ref{remarks} and references

958: therein) that the action \eq{u-action}

959: is Kirillov's coadjoint orbit action for the group of area-preserving

960: diffeomorphisms. 

961: 

962: The measure $Du$, described in Sections \ref{measure}

963: and \ref{remarks}, incorporates the constraints \eq{quad-const}

964: and \eq{int-const}.

965: The equation of motion

966: for $u(x_1,x_2,\tau)$ that follows from \eq{path-for-u} is

967: (see \cite{DMW-classical,DMW-nonrel}):

968: \be

969: \del_\tau u - (x_1 \del_2 - x_2 \del_1) u=0

970: \label{eom-u}

971: \ee

972: 

973: \subsection{Action}

974: 

975: We will show that the action \eq{u-action} gives rise to

976: the various D3-brane actions in

977: \eq{path-for-giant},\eq{path-for-giant-1},\eq{path-for-giant-arbit}

978: and \eq{path-for-giant-arbit-1}  in the sense of 

979: \eq{main-eqn}. 

980: Consider, for example,  configuration (d), \eq{single-u-arbit-2},

981: \eq{giant-arbit-1}.

982: It is easy to see that if  $\delta u$ does not

983: intersect with $u_0$, 

984: then the left hand side of \eq{main-eqn}

985: is given by local properties of the cell $\delta u$, viz.

986: $\delta S[u] = S[\delta u]$. Thus we get 

987: \bea

988: \delta S &&= \delta S_{kin} - \delta S_{ham}

989: \nn

990: \delta S_{kin} && = \int {dx_1 dx_2 \over 2\pi \hbar}

991: \hbar \int_{\tilde \Sigma} d\tau 

992: ds\ \delta u\{\del_\tau \delta u, \del_s \delta u \}_{PB} 

993: \nn

994: \delta S_{ham} &&= \int d\tau \int {dx_1 dx_2 \over 2\pi \hbar}

995: \delta u(x_1,x_2,\tau) \frac{x_1^2 + x_2^2}{2 \hbar}

996: \label{action-for-delta-u}

997: \eea

998: We need to show that the above action is equal to the

999: action $S_{BPS}$ appearing in  \eq{path-for-giant-arbit-1}.

1000: 

1001: Let us consider first the Hamiltonian term:

1002: \bea

1003: \delta S_{ham} && = 

1004: \int d\tau \langle \frac{x_1^2 + x_2^2}{2 \hbar} \rangle 

1005:  \int {dx_1 dx_2 \over 2\pi \hbar}

1006: \delta u(x_1,x_2,\tau) 

1007: \nn

1008: &&= \int d\tau \frac{{\bar x_1}^2 + {\bar x_2}^2}{2 \hbar}

1009: \nn

1010: && = \int d \tau \frac{r^2}{2 \hbar}

1011: \label{delta-s-ham}

1012: \eea

1013: which matches with the Hamiltonian term in \eq{path-for-giant-arbit-1}. 

1014: In the first step we have taken the integrand out of the cell $\delta u$

1015: since its size is small, in the second step we have used the fact that

1016: $\delta u$ has area $2\pi \hbar$ and also equated the average values of

1017: $x_1, x_2$ with the coordinates of the centre of mass 

1018: $\bar x_1, \bar x_2$ (see \eq{giant-arbit-1}) 

1019: which satisfies ${\bar x_1}^2 + {\bar x_2}^2 = r^2$.

1020: 

1021: The analysis of the Hamiltonian term for configuration (c)

1022: [\eq{single-u-arbit},\eq{giant-arbit},\eq{path-for-giant-arbit}] is

1023: similar. It is interesting to note that in the special cases

1024: \eq{path-for-giant} and \eq{path-for-giant-1} the Hamiltonian by

1025: convention measures the energy of the fluctuation $\delta u$ together

1026: with that of a compensating fluctuation $\delta u'$

1027: (see footnote \ref{compensating}) defined by

1028: adjusting the radius $r_0$ (this, again, corresponds to a choice of

1029: gauge for $C^{(4)}$ different from that in \eq{fields},

1030: \eq{fields-1}). Thus, e.g. the energy \eq{h-red} includes the energy

1031: of the hole $-N \cos^2 \theta$ as well as the energy $+N$ of the

1032: compensating outer circular strip $+\delta u'$, extending

1033: between $r_0$ and $r_0 + \delta r_0$ such that the latter

1034: radius has an area $N+1$. In the generic case it is more natural

1035: to keep the two effects separate, which is possible to do in 

1036: the semiclassical limit.

1037: 

1038: The analysis of the kinetic term $\delta S_{kin}$

1039: is more complicated and is presented in Appendix A. It is,

1040: however, somewhat simpler to match the equation of motion that follows

1041: from \eq{action-for-delta-u} with the equations of motion following

1042: from \eq{path-for-giant-arbit-1}. The latter are

1043: \be

1044: \dot{\bar x_1}= \bar x_2, \dot{\bar x_2} = - \bar x_1

1045: \label{sho}

1046: \ee

1047: The equation of motion following from the action

1048: \eq{action-for-delta-u} can be read off from

1049: \eq{eom-u} and is given by

1050: \be

1051: \dot{\delta u} - (x_1 \del_2 - x_2 \del_1)\delta u=0

1052: \label{eom-delta-u}

1053: \ee

1054: Using the expression \eq{single-u} for $\delta u$, 

1055: one can show that \eq{eom-delta-u} is satisfied to leading

1056: order in $\hbar$, provided \eq{sho} is valid. 

1057: 

1058: \subsection{\label{measure}Measure}

1059: 

1060: The measure $\D u$ is defined as the group-invariant measure where $u$

1061: is parameterized as an orbit of some specific field configuration

1062: $u_0$ under the group of area-preserving diffeomorphisms (see

1063: \cite{DMW-classical,DMW-nonrel} and Section \ref{remarks}). The

1064: reference configuration $u_0$ satisfies $u_0^2 = u_0$ and 

1065: $\int dx_1 dx_2~ u_0/(2\pi\hbar)= N$ so that 

1066: the measure $Du$ incorporates the two

1067: constraints \eq{quad-const} and \eq{int-const}.

1068: 

1069: When $g$ acts on $\delta u$ (see \eq{single-u}), the action gets

1070: transmitted to the centres of mass of $\delta u$ as a canonical

1071: transformation on $\bar x_1, \bar x_2$ (cf. \eq{group-action}). The

1072: invariant measure under canonical transformations is the one already

1073: used in \eq{path-for-giant-arbit-1}.  We find, therefore, that the

1074: measures also agree.

1075: 

1076: \section{\label{nc} Equivalence to Fermion path integral}

1077: 

1078: Ref. \cite{DMW-path} discussed 

1079: the  following path integral which

1080: represented a path integral for the phase space density

1081: $u(q,p,t)$ for free fermions moving in one

1082: dimension under a Hamiltonian $h(q,p)$

1083: \bea

1084: Z_{NC} &&= \int [Du(q,p,t)]_{u_0} \exp[i S[u]]

1085: \nn

1086: S[u]

1087: && = \int {dq~dp \over 2\pi \hbar}

1088: \hbar \int_{\tilde \Sigma} dt~ ds~ 

1089: \u(q,p,t,s) * \{\del_t \u, \del_s \u \}_{MB} -  

1090: \int_\Sigma dt

1091: \tilde H 

1092: \nn

1093: \tilde H &&= \int {dq~dp \over 2\pi \hbar}

1094: u(q,p,t)* {h(q,p)\over \hbar}

1095: \label{z-nc} 

1096: \eea  

1097: For free fermions moving in a harmonic oscillator potential

1098: \be

1099: h(q,p) = {p^2 + q^2\over 2}

1100: \label{harmonic}

1101: \ee

1102: The star product in \eq{z-nc} is defined as

1103: \be

1104: a * b (q,p) 

1105: = \left[\exp\Big({i\hbar\over 2}\left(

1106: \del_{q}\del_{p'} -  \del_{q'}\del_{p}\right)\Big)

1107: \left(a (q,p) b(q',p') \right)\right]_{q'=q,p'=p}

1108: \label{star}

1109: \ee

1110: The Moyal Bracket is defined as

1111: \be

1112: \{ a, b\}_{MB} = \frac{a * b - b * a}{i\hbar}

1113: \label{Moyal}

1114: \ee

1115: The measure $Du$ is defined as the group-invariant measure

1116: under the symmetry group $W_\infty$ of the 

1117: fermion configuration space \cite{DMW-path,DMW-nonrel}.

1118: The space of $u$'s is the $W_\infty$ orbit of a

1119: reference configuration $u_0$ which we can take

1120: to be the expectation value of

1121: the Wigner phase space distribution \eq{wigner}

1122: in the Filled Fermi sea.

1123: The measure incorporates the constraint

1124: \be

1125: u * u = u

1126: \label{u-star-u}

1127: \ee

1128: and

1129: \be

1130: \int \frac{dq dp}{2\pi \hbar} u = N

1131: \label{int-const2}

1132: \ee

1133: The operator definition of the Wigner distribution $\hat u(q,p,t)$

1134: is given in \eq{wigner}.

1135: 

1136: The equation of

1137: motion following from this path integral is

1138: \bea

1139: \del_t u(q,p, t) && = \{h(q,p), u(q,p, t)\}_{MB}

1140: \nn

1141: {}&& = \{h(q,p), u(q,p, t)\}_{PB}

1142: \nn

1143: {}&& = (q\del_p - p \del_q) u(q,p,t)

1144: \label{eom-old}

1145: \eea

1146: $h(q,p)$ is the single particle Hamiltonian appearing

1147: in \eq{z-nc}.

1148: The second step follows for any quadratic Hamiltonian. 

1149: For the  $c=1$ matrix model, one takes $h = 

1150: (p^2 - q^2)/2$, but the analysis in \cite{DMW-path} is true for

1151: any Hamiltonian and in particular for $h =(p^2 + q^2)/2 $. 

1152: The third line follows from this latter Hamiltonian.

1153: Although the equation of motion \eq{eom-old} coincides with its classical

1154: limit \eq{eom-u}, the finite $\hbar$ dynamics differs significantly 

1155: from its classical limit because the constraint \eq{u-star-u} involves

1156: star products, involving fuzzy solutions for $u$ 

1157: \cite{DMW-nonrel,DMW-instanton}, unlike the constraint

1158: \eq{quad-const} whose solutions are characteristic functions

1159: \eq{u-char}. This is discussed further in the next two sections.

1160: 

1161: In \cite{DMW-path} it was shown that \eq{z-nc} is exactly

1162: equal to a  path integral for $N$ free fermions moving in a simple

1163: harmonic oscillator potential, defined as follows:

1164: \bea

1165: Z_{NC}= Z_F &&= \int D[\Psi]_{\ket{F_0}} \exp[i S_F/\hbar]

1166: \nn

1167: S_F &&= \int dt\ dx \ [\Psi^\dagger(x,t)

1168: (i \hbar \del_t - h(x, \del_x)) \Psi(x,t)]

1169: \nn

1170: h &&=\frac12(-\hbar^2 \frac{\del^2}{\del x^2} + x^2)

1171: \label{z-f}

1172: \eea  

1173: Here $\Psi(x,t), \Psi^\dagger(x,t)$ are the second

1174: quantized annihilation and creation operators (respectively)

1175: for the fermions.

1176: The subscript ${\ket{F_0}}$ in the measure implies that

1177: the functional integral is over states 

1178: obtained from the reference Fock space state $\ket{F_0}$

1179: under $W_\infty$ transformations.  

1180: These in fact span all states with the same

1181: fermion number as $\ket{F_0}$, which we take to be $N$.

1182: 

1183: \myitem{Wigner phase space distribution}

1184: 

1185: The Wigner phase space  distribution $u(q,p,t)$ which appears in

1186: \eq{z-nc} as a path integral variable, can be defined

1187: as an operator (second quantized, see, e.g. 

1188: \cite{DMW-nonrel,DMW-instanton}) 

1189: as follows:

1190: \be

1191: \hat u(q,p,t) = 

1192: \int d\eta \Psi^\dagger(q+ \eta/2,t)\Psi(q-\eta/2,t) \exp[i{p

1193: \over \hbar}\eta ] 

1194: \label{wigner}

1195: \ee

1196: Salient properties of this quantity as well those of

1197: its expectation values in various states have been

1198: listed in \cite{DMW-nonrel,DMW-instanton, DMW-classical}.

1199: 

1200: \myitem{The correspondence}

1201: 

1202: It is clear that \eq{path-for-u} is simply the $\hbar\to 0$ limit of

1203: \eq{z-nc}, provided one identifies $u(x_1, x_2)$ of Section 4

1204: with $u(q,p)$ of this section. This is the advertised

1205: transformation of configuration  space into phase space. 

1206: The constraints \eq{quad-const}

1207: and \eq{int-const} also follow from \eq{u-star-u} and 

1208: \eq{int-const2}. Note that the equation $u * u =u$

1209: reduces to $u^2 = u$ in the semiclassical limit, a fact

1210: which has been extensively exploited in \cite{DMW-nonrel,

1211: DMW-instanton, DMW-classical}.

1212: 

1213: Hence the collective coordinate quantization of LLM

1214: geometries gives rise to the $\hbar \to 0$ limit of free fermions in a

1215: harmonic oscillator potential. This is of course what we expect from

1216: the AdS/CFT correspondence \cite{LLM}, but we arrived at this result here

1217: starting from D-branes in supergravity. How to elevate this result to finite

1218: $\hbar$ remains an interesting issue. Some possible

1219: subtleties are mentioned in the next section. In the next section

1220: we also briefly discuss a more direct derivation of the semiclassical

1221: correspondence from supergravity using Kirillov's symplectic

1222: form.  

1223: 

1224: \section{\label{remarks}Remarks on collective coordinate 

1225: method with BPS constraint}

1226: 

1227: In this section we will briefly discuss a first principles approach to

1228: the collective coordinate quantization of half-BPS geometries without

1229: using the D3 brane actions.

1230: 

1231: We begin by noting that the group G of  time-independent area-preserving

1232: diffeomorphisms (SDiff) is a symmetry of the constraints \eq{int-const} and

1233: \eq{quad-const}, as well as of the equations of motion of the type IIB

1234: theory (since the geometries corresponding to various $u$'s

1235: all satisfy IIB equations of motion).

1236: The Lie Algebra $\bar G$ is the algebra of symplectic vector fields. Thus,

1237: elements $g= 1 + X_f$ near identity of $G$,  act on a function $u(x_1,x_2)$ 

1238: as 

1239: \bea 

1240: u\to u^g &&= u + X_f.u = u + \{f, u\}_{PB} 

1241: \nn 

1242: X_f &&=

1243: \epsilon_{ij} {\del f\over \del x_i}{\del \over \del x_j} 

1244: \eea 

1245: This action can also be regarded as induced by the motion

1246: of points on the plane under a Hamiltonian $f$:

1247: \bea

1248: u^g(x)&& = u(x^{g^{-1}}), {\rm where}

1249: \nn

1250: (x_1, x_2)^g &&\equiv (x_1 + \del f/\del x_2, x_2 - \del f/\del x_1)

1251: \label{group-action}

1252: \eea

1253: Finite group elements $g \in {\rm SDiff}$ can be dealt with  by

1254: exponentiation. 

1255: Now, since the function $u$ completely determines the supergravity fields

1256: (collectively denoted below as $\Phi$):

1257: $ \Phi = \Phi[u]$, the  group $G$ of area-preserving diffeomorphisms

1258: has a natural action on  supergravity fields:

1259: \be

1260: \Phi^g = \Phi[u^g]

1261: \ee

1262: The choice of any given function $u_0$, and the corresponding

1263: $\Phi_0$  breaks the

1264: symmetry $G \to H$, where $H$ denotes the subgroup generated by

1265: functions which have zero Poisson bracket with $u_0$. 

1266: 

1267: The collective  coordinate method

1268: \cite{Gervais:1975yg,Gervais:1976ca} consists of making a change of

1269: variable $\Phi(t) \to \{ g(t), \tilde \Phi(t)

1270: \equiv \Phi^{g(t)}(t) \}$, where $\tilde \Phi(t)$ represents

1271: motion in the body-fixed frame which is over and above the

1272: collective motion. The dynamics of the collective

1273: coordinate  is obtained by implementing the change of variable

1274: in the field theory functional integral.

1275: 

1276: A first principles derivation of the collective coordinate action

1277: (without using the identification with D3 branes) would involve

1278: implementing the above procedure in the case of IIB supergravity.

1279: We will not attempt to do this here in detail, but give

1280: a brief discussion:

1281: 

1282: \begin{enumerate}

1283: 

1284: \item 

1285: Since the IIB Lagrangian is second order in time derivatives, the low

1286: energy action for $g(t)$, is expected to be quadratic in $\dot g$

1287: (before implementing the BPS condition).  This corresponds, for

1288: example, to \eq{giant-action}, \eq{action-arbit} or

1289: \eq{action-arbit-2}, which are second order in time at low

1290: velocities. The phase space of the collective degree of freedom $g(t)$

1291: at this stage involves $g(t)$ as well as

1292: $\pi_g(t)$ where $ \pi_g(t)$ is the ``momentum''

1293: for $g(t)$.

1294: 

1295: \item 

1296: In case of the D3 brane dynamics one can explicitly see how

1297: \eq{z-full} changes to \eq{path-for-giant} with the imposition of the

1298: BPS constraint \eq{mom-const}. One would similarly expect that, 

1299: if one implements the change of variable $\Phi \to

1300: \{ g(t), \tilde \Phi \}$ in the IIB functional

1301: integral ``in the presence of a BPS constraint'',  

1302: the dynamics of the collective variable $g(t)$ will be

1303: described by a first-order action and $g(t)$'s themselves would become

1304: a phase space. The most natural such an action on a

1305: $G$-orbit of a configuration $u_0$ is 

1306: given by Kirillov's method of coadjoint orbits 

1307: \cite{DMW-nonrel,DMW-classical,Kirillov}

1308: (see, e.g  eq. (68) of \cite{DMW-classical})  

1309: \be

1310: S_{BPS} 

1311: =  \int dt \langle X_t, u_0 \rangle

1312: - \int dt \langle g^{-1} X_h g, u_0 \rangle  

1313: \label{action-g}

1314: \ee 

1315: where $\langle X_f, u_0 \rangle \equiv \int \frac{dx_1 dx_2}{2\pi

1316: \hbar} (f(x_1, x_2) u_0(x_1, x_2))$. The notation $X_t$ denotes the

1317: Lie algebra element $g^{-1} \dot g$ and $X_h \equiv

1318: g^{-1} h g$ denotes the $g$-transported Lie algebra element

1319: corresponding to the   

1320: Hamiltonian $h = (x_1^2 + x_2^2)/2$.

1321: This action exactly coincides with \eq{u-action}

1322: \cite{DMW-nonrel,DMW-classical}.  Indeed the measure also coincides

1323: with the measure of \eq{u-action}. As a matter of fact 

1324: we initially arrived at the

1325: action \eq{u-action} by considering the Kirillov action

1326: \cite{Mandal-talk};

1327: from this viewpoint the D3-brane method can be viewed 

1328: as additional evidence in support of the Kirillov action. 

1329: 

1330: \item 

1331: Evaluation of a functional integral ``in the presence of a BPS

1332: constraint'' involves insertion of an appropriate projection operator.

1333: It is possible that the resulting functional integral is an index, as

1334: in \cite{Maldacena:1999bp,Cecotti:1992qh}, which are natural tools for

1335: counting geometries satisfying a specific number of supersymmetries.

1336: 

1337: \item

1338: If all $\Phi[u]$'s can be generated by the collective motion

1339: $\Phi[u^g]$, clearly the other degrees of freedom $\tilde \Phi$ are to

1340: be omitted from the functional integral under the half-BPS

1341: constraint. In this sense, the collective coordinate functional

1342: integral would appear to be the entire supergravity functional

1343: integral when subjected to the half-BPS constraint (see the next

1344: point, however).

1345: 

1346: \item

1347: There is an important subtlety regarding the number of connected

1348: components of the $u$-configurations. Although SDiff acts on the LLM

1349: geometries, it is not clear how it can change the number of connected

1350: components of a given $u$-configuration. Of course, under the

1351: $W_\infty$ group mentioned in Section \ref{nc}, such a transformation

1352: can happen (a fuzzy droplet can split into two fuzzy

1353: droplets). However, $W_\infty$ is the symmetry group of the equation

1354: $u * u =u$ and is not naturally associated with the LLM constraint

1355: $u^2 = u$. It remains unclear to us at the moment how to describe the

1356: entire space of LLM geometries as the orbit of a given configuration

1357: under a certain group $G$.

1358: 

1359: \end{enumerate}

1360: 

1361: \section{Conclusion}

1362: 

1363: In this paper we considered collective coordinate quantization of LLM

1364: geometries identifying the function $z(x_1,x_2,0)\equiv 1/2 -

1365: u(x_1,x_2)$ of \cite{LLM} as the collective coordinate. The explicit

1366: form of the collective coordinate action (and measure) is derived by

1367: identifying the collective degree of freedom as that of a D3 brane

1368: coupled to an arbitrary LLM geometry.  The D3 brane functional

1369: integral, subject to the BPS constraint, can be written directly in

1370: terms of the $u$-variable. We show that the resulting functional

1371: integral is the $\hbar\to 0$ limit of a functional integral describing

1372: free fermions in a harmonic oscillator potential.  We discuss a first

1373: principles approach towards derivation of the $u$-integral using the

1374: general method of collective coordinates subject to a BPS constraint.

1375: 

1376: We note a few important points:

1377: 

1378: \begin{enumerate}

1379: 

1380: \item 

1381: We find that supergravity configuration space becomes a phase space

1382: (hence noncommutative, with a noncommutativity parameter given by a

1383: certain $\hbar$), when constrained to configurations preserving a

1384: certain number of supersymmetries. Although we found this phenomenon

1385: in a specific case here (half-BPS IIB supergravity solutions with

1386: $O(4) \times O(4)$ symmetry), it is clear that this phenomenon should

1387: be generic. In particular the appearance of a first order action,

1388: discussed in Section \ref{remarks}, is related to the fact that the

1389: BPS equations are first order. The formalism of phase space path

1390: integrals employed in this paper makes it rather apparent how a

1391: configuration space path integral with second order action becomes a

1392: phase space path integral with first order action under the imposition

1393: of the BPS constraint.  It appears to be possible, using this, to

1394: count supersymmetric configurations within low energy field theories

1395: including supergravity. This observation clearly has implications for

1396: counting entropy of supersymmetric black holes and other related

1397: configurations. 

1398: 

1399: \item

1400: As we mentioned in the previous section, functional integrals

1401: preserving a certain number of supersymmetries have earlier been

1402: treated in, for example, \cite{Maldacena:1999bp,Cecotti:1992qh}, where

1403: the partition function is a `twisted' one involving insertion of

1404: operators related to $(-1)^F$.  It would be interesting to see if this

1405: is the case for half-BPS supergravity solutions treated in this

1406: paper. One would imagine defining such path integrals in terms of

1407: projection operators in the Hilbert space enforcing the supersymmetry

1408: conditions; it is of interest to explore the connection between this

1409: definition and the `twisted' partition function mentioned above.

1410: Another related way of understanding ``BPS functional integrals''

1411: would be to use topological twisting so that the relevant

1412: supersymmetry operators become BRST operators and the desired path

1413: integral becomes the normal path integral in the topological

1414: theory.

1415:   

1416: \item

1417: It is entirely possible, as in the context of the $c=1$ matrix models,

1418: that the semiclassical collective excitation approach misses important

1419: subtle points of the fermion theory. In the case of $c=1$ this was

1420: discussed in great detail in \cite{Sengupta-Wadia,

1421: Mandal:1991ua,DMW-nonrel,DMW-instanton,DMW-beta,DMW-discrete}. One

1422: important effect missed by classical collective excitations

1423: (corresponding to the massless `tachyons') is the unstable D0 brane of

1424: the two-dimensional string theory

1425: \cite{McGreevy:2003kb,Klebanov:2003km} (this viewpoint is explained in

1426: \cite{Mandal:2003tj}).  In the present case, the semiclassical

1427: collective excitations consist of  ripples (corresponding to

1428: gravitons, see Appendix B) as well as D3 branes (roughly analogous to the

1429: tachyons and D0 branes, respectively, of  two dimensional string

1430: theory). However, we might discover other important effects related to

1431: the non-perturbative description \eq{z-nc} possibly missed by the

1432: semiclassical treatment of the collective excitations.

1433: 

1434: \item 

1435: We have used D3 branes coupled to LLM geometries to find

1436: noncommutative dynamics in the configuration space.  It is interesting to

1437: note that in the limit of LLM geometries  which describes D3

1438: branes in the Coulomb branch \cite{LLM}, the value of $\hbar$ scales to zero

1439: causing the noncommutativity to disappear, as one would expect.

1440: 

1441: \item

1442: As seen in Section 5, the phase space density action obtained from the

1443: fermion theory has an additional degree of noncommutativity reflected

1444: in the appearance of star products, over and above the

1445: noncommutativity mentioned in points (1) and (4). The latter is

1446: already evident in the semiclassical limit itself where the Moyal

1447: brackets get reduced to Poisson brackets and reflects a phase space

1448: structure of the classical configuration space. Clearly the former is

1449: related to the issue of finite $\hbar$ correspondence between the

1450: half-BPS geometries and the fermion theory. Of particular importance

1451: is whether the generalization to the constraint $u * u = u$ (instead

1452: of $u^2 = u$) allows some insight into $g_{st}$ effects in string

1453: theory. Some aspects of the effect of finite $g_s$ have been discussed

1454: at the end of the previous section. 

1455: 

1456: \item

1457: A specific subleading $1/N$ correction briefly mentioned in this paper

1458: is the effect of the compensating fluctuations $\delta u'$ (see

1459: footnote \ref{compensating}). This effect is indeed calculable in the right

1460: hand side of \eq{main-eqn} for various choices of $\delta u'$ and it

1461: is an interesting question whether the corresponding modification

1462: in the left hand side arises correctly by taking into account

1463: interaction between $\delta u$ and $\delta u'$ coming from

1464: the star product structure of $S[u]$.

1465: 

1466: \item 

1467: Most of this paper dealt with collective excitations

1468: identified as D3 branes. We discuss gravitons briefly

1469: in Appendix B; it would be interesting to quantitatively

1470: reproduce the graviton fluctuations from our collective action.

1471:  

1472: \end{enumerate}

1473: 

1474: \section{Acknowledgments}

1475: 

1476: I would like to thank David Berenstein, Avinash

1477: Dhar, Jaume Gomis, Oleg Lunin, Alex Maloney, Liam McAllister, John

1478: McGreevy, Rob Myers, Hiroshi Ooguri, Joe Polchinski, Yasuhiro Sekino,

1479: Lenny Susskind and Spenta Wadia for discussions, Nemani Suryanarayana

1480: for discussions and initial collaboration and 

1481: Shahin Sheikh-Jabbari for comments.

1482: I would like to acknowledge the  hospitality at SLAC, Stanford and

1483: Caltech during the finishing stages of this work.

1484: 

1485: \appendix

1486: 

1487: \section{Phase space density action for a single cell}

1488: 

1489: In this Appendix we will evaluate $\delta S_{kin}$

1490: appearing in \eq{action-for-delta-u}, with

1491: $\delta u$ as in \eq{single-u}.

1492: For simplicity of notation, we will denote 

1493: \be

1494: x_1 =q, x_2 = p

1495: \label{simple}

1496: \ee

1497: Let us define 

1498: \be

1499: q^\pm  = \pm \bar q + \eps/2 \mp  q,  

1500: p^\pm  = \pm \bar p + \eps/2 \mp  p

1501: \ee

1502: Then 

1503: \be

1504: \delta u(q,p) = \theta(q^+)\theta(q^-)\theta(p^+)\theta(p^-)

1505: \ee

1506: It is easy to calculate

1507: \bea

1508: \dot{\delta u}

1509: &&= \dot{\bar p}\Big[

1510: \delta (p^+) \theta(p^-) - \theta(p^+)\delta(p^-)\Big]

1511: \theta(q^+) \theta(q^-)

1512: \nn

1513: &&+  \dot{\bar q}

1514: \Big[\delta (q^+) \theta(q^-) - \theta(q^+)\delta(q^-)

1515: \Big]

1516: \theta(p^+) \theta(p^-)

1517: \eea

1518: and

1519: \bea

1520: {\delta u}'

1521: &&= {\bar p}'

1522: \Big[ \delta (p^+) \theta(p^-) - \theta(p^+)\delta(p^-)

1523: \Big] 

1524: \theta(q^+) \theta(q^-)

1525: \nn

1526: &&+ {\bar q}'

1527: \Big[\delta (q^+) \theta(q^-) - \theta(q^+)\delta(q^-)

1528: \Big]\theta(p^+) \theta(p^-)

1529: \eea

1530: We define the Poisson bracket 

1531: \be

1532: \{f,g\}_{PB} = \del_q f \del_p g

1533: - \del_p f \del_q g

1534: \ee

1535: We get, after some simplification,

1536: \bea

1537: \delta S_{kin}

1538: &&=  \int d\tau ds\int \frac{dq dp}{2\pi \hbar} \hbar

1539: \delta u \{\dot{\delta u}, {\delta u}'\}_{PB} 

1540: \nn

1541: &&= \int \frac{3 dp\ dq}{8 \pi} \{

1542: \delta^2(q^+) +\delta^2(q^-)\}\{\delta^2(p^+) + \delta^2(p^-)\}

1543: \Big[ \int d\tau ds\ (\dot{\bar q}{\bar p}' -  \dot{\bar p}{\bar q}')

1544: \Big]

1545: \nn

1546: &&= A

1547: \int d\tau (- \frac{r^2}{2\hbar} \dot \pht)

1548: \nn

1549: A && = \frac{3\hbar}{\pi}\delta_q(0)\delta_p(0)

1550: \label{tmp}

1551: \eea

1552: In the last line we have used eqs. \eq{x1-x2-pht} and

1553: \eq{simple} and the equality 

1554: \be

1555: \dot{\bar q}{\bar p}' -  \dot{\bar p}{\bar q}' 

1556: = \del_s (\bar p \dot{\bar q} - \bar q \dot{\bar p} )

1557: = \del_s (- r^2 \dot \pht)

1558: \label{partial}

1559: \ee

1560: Thus $\delta S_{kin}$ appearing in \eq{action-for-delta-u}

1561: agrees with the corresponding term in \eq{path-for-giant-arbit-1}

1562: apart from a proportionality constant $A$. 

1563: 

1564: Let us discuss the constant $A$.  In the last line of \eq{tmp}

1565: $\delta_q(0)$ denotes $\delta(x_1 - x_1)$, similarly $\delta_p(0)$

1566: denotes $\delta(x_2 - x_2)$.  Clearly we need a regularization. It is

1567: natural to choose $\delta_q(0) = \delta_p(0) = a/\sqrt{\hbar}$. We get

1568: $A=1$ if $a^2 = \pi/3$. We do not believe that this regularization has

1569: a particular significance since the agreement at the level of the

1570: equation of motion, between \eq{sho} and \eq{eom-delta-u}, does not

1571: use any such regularization. In other words, the equation of motion

1572: \eq{eom-delta-u}, which can be derived from \eq{sho}, can be used to

1573: fix the relative coefficients between $\delta S_{kin}$ and $\delta

1574: S_{ham}$ in 

1575: \eq{action-for-delta-u}, thus 

1576: determining $A=1$ in \eq{tmp}. Such a method proves the desired result

1577: without the use of a regularization. 

1578: 

1579: \section{\label{graviton}Gravitons}

1580: 

1581: So far in this paper we have primarily 

1582: considered collective motions identified as D3

1583: branes. We found that (see \eq{action-arbit}) the $\hbar$ of 

1584: the collective action

1585: naturally corresponds to the D3-brane tension:

1586: \be

1587: \frac1{2\hbar} =  T_3 \omega_3

1588: \label{d3-tension-match}

1589: \ee

1590: This raises a puzzle about other collective motions

1591: such as 

1592: gravitons. Suppose we consider an equation

1593: analogous to \eq{main-eqn}, where the $\delta u$

1594: fluctuation corresponds to a ripple (see

1595: footnote \ref{tiling}) and the brane refers 

1596: now to a fundamental string. Since the left hand

1597: side of \eq{main-eqn} continues to have a prefactor $1/\hbar$

1598: (see, e.g \eq{action-for-delta-u}, 

1599: \eq{delta-s-ham}), while 

1600: the fundamental string tension does not

1601: involve $1/g_s$, we apparently have a puzzle here.

1602: 

1603: The resolution comes from the fact that $\delta u$

1604: now describes  ``ripples'' which are

1605: fluctuations extending from the original droplet(s) by distances

1606: $O(\sqrt\hbar)$.  Because of this, as we will  show below, 

1607: the collective action

1608: evaluates to $O(g_s)$ which cancels the $1/g_s$, 

1609: reproducing the fundamental string tension so far

1610: as $g_s$-counting is concerned.

1611: 

1612: The simplest parameterization \cite{Polchinski,Sengupta-Wadia}

1613: for the ripples is as in Figure \ref{ripple.fig}.  

1614: \begin{figure}[ht]

1615: %or [htb]

1616: \vspace{0.5cm}

1617: \hspace{-0.5cm}

1618: \centerline{

1619:     \epsfxsize=8.5cm

1620:    \epsfysize=8cm

1621:    \epsffile{ripple.eps}

1622:  }

1623: \caption{\sl The fluctuations $p^+(x_1)$ and $p^-(x_1)$

1624: extend from the original droplet to distances $O(\sqrt\hbar)$.

1625: The Lagrangian for these fluctuations evaluates to $O(g_s)$.

1626: This cancels the prefactor $1/g_s$ sitting outside the collective action

1627: \eq{action-for-delta-u}, consistent with fundamental

1628: string tension which is independent of $g_s$.

1629: }

1630: \label{ripple.fig}

1631: \end{figure}

1632: For simplicity we have considered the unperturbed 

1633: droplet to correspond to \ads, but similar arguments

1634: can be made with respect to ripples traveling

1635: in other backgrounds.

1636: 

1637: The precise form of $u(x_1, x_2)$ is

1638: \be

1639: u(x_1, x_2) = \theta([p^+(x_1) - x_2][x_2 - p^-(x_1)])

1640: \ee

1641: where $p^\pm(x_1)$ are to be chosen consistent

1642: with \eq{int-const}.  

1643: The fact that the amplitude of the fluctuations

1644: $\sim O(\sqrt\hbar)$ implies $\delta p^+, \delta p^-

1645: \sim O(\sqrt\hbar)$, where $\delta p^\pm =

1646: p^\pm(x_1) \mp p_0^\pm(x_1)$. $ p_0^\pm(x_1)$ denote

1647: the unperturbed profile. 

1648: Following steps similar

1649: to \cite{Polchinski,Sengupta-Wadia} 

1650: the action $\delta S$ for the fluctuation turns out to be quadratic

1651: in $\delta p^+, \delta p^-$ and hence $\sim O(\hbar)

1652: \propto O(g_s)$. 

1653: Thus, $g_s$ cancels from the left hand side of

1654: \eq{main-eqn} for ripples, consistent with their

1655: interpretation as  fundamental

1656: string modes.

1657: 

1658: We hope to come back to a quantitative derivation of the action (as well

1659: as path integral) for gravitons from the collective coordinate path

1660: integral \eq{path-for-u}.

1661: 

1662: %\section{Measure for $u$}

1663: 

1664: \bibliographystyle{utphys} 

1665: \bibliography{myrefs}

1666: 

1667: 

1668: \end{document}

1669: 

1670: 

1671: 

1672: