1: \documentclass[12pt]{article}

2: 

3: \usepackage{cite}

4: \usepackage{color}

5: \usepackage{graphicx}

6: \usepackage{amsmath}

7: \usepackage{amssymb}

8: \usepackage{xspace}

9: 

10: 

11: \makeatletter

12: \@addtoreset{equation}{section}

13: \renewcommand{\theequation}{\thesection.\arabic{equation}}

14: 

15: \makeatletter

16: \renewcommand\section{\@startsection {section}{1}{\z@}%

17:                                    {-3.5ex \@plus -1ex \@minus -.2ex}%nn

18:                                    {2.3ex \@plus.2ex}%

19:                                    {\normalfont\large\bfseries}}

20: \renewcommand\subsection{\@startsection{subsection}{2}{\z@}%

21:                                      {-3.25ex\@plus -1ex \@minus -.2ex}%

22:                                      {1.5ex \@plus .2ex}%

23:                                      {\normalfont\bfseries}}

24: 

25: 

26: 

27: \def\baselinestretch{1.2}

28: \parskip 6 pt

29: 

30: \marginparwidth 0pt

31: \oddsidemargin  0pt

32: \evensidemargin  0pt

33: \marginparsep 0pt

34: \topmargin   -0.5in

35: \textwidth   6.5in

36: \textheight  9.0 in

37: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

38: %macros here:

39: 

40: \def\Reals{{\mathbb{R}}}

41: 

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

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

44: \newcommand{\beq}{\begin{eqnarray}}

45: \newcommand{\eeq}{\end{eqnarray}}

46: \newcommand{\oo}{\"o}

47: \newcommand{\tr}{{\rm Tr}}

48: \newcommand{\gone}[1]{{}}

49: \newcommand{\sty}[1]{{{\rm #1}}}

50: \newcommand{\bl}{\noindent $\bullet\ $}

51: 

52: % New definitions from 12/6/2006

53: \def\anglephi{{\phi}} % a periodic coordinate 

54: \def\dilaton{{\Phi}} % dilaton field

55: 

56: % coordinates that describe directions transverse to D3-brane

57: % in various solutions

58: \def\fullu{{u}}  % u of the full SUGRA solution of a D3-brane

59: \def\fully{{y}} % y of the full SUGRA solution of a D3-brane

60: \def\rescaledU{{U}} % U for the near horizon limit

61: \def\rescaledY{{Y}} % Y for the near horizon limit

62: \def\hotrho{{\rho}} % \sqrt{U^2+Y^2} for nonzero temperature 

63: \def\hotangle{{\varphi}} % for nonzero temperature

64: 

65: \def\basephi{{\varphi}} % spherical coordinate on the Hopf base

66: \def\basetheta{{\theta}} % spherical coordinate on the Hopf base

67: \def\fiberphi{{\phi}} % a periodic coordinate on the Hopf fiber

68: 

69: \def\anglepsi{{\psi}} % angle between U and |Y|.

70: \def\angleY{{\xi}} % angle between Y_8 and Y_9

71: \def\Mcircle{{y_{10}}} % M-theory circle

72: 

73: \def\radialV{{V}}

74: 

75: \newcommand\dsBase[1]{ds_{B({#1})}} % \sqrt{metric} on base

76: \def\connHopf{{{\cal A}}} % Hopf connection

77: \def\augmentedh{{k}} % H + \eta^2 u^2

78: 

79: \def\rescaledH{{H}} % h rescaled by $\alpha'^2$

80: \def\rescaledAugH{{K}} % \augmentedh rescaled

81: 

82: \def\gIIA{g_{\text{\tiny IIA}}}

83: \def\tgIIA{\tilde{g}_{\text{\tiny IIA}}}

84: \def\gIIB{g_{\text{\tiny IIB}}}

85: \def\tgIIB{\tilde{g}_{\text{\tiny IIB}}}

86: 

87: \def\gYM{g_{\text{\tiny YM}}} % YM coupling constant

88: \newcommand\gYMDim[1]{g_{\text{\tiny YM{{#1}}}}} % in other Dims

89: \def\Ricci{{\cal R}} % Ricci scalar

90: \def\RCurvature{{\cal R}} % Curvature tensor

91: 

92: \def\lst{{\ell_s}} % string scale

93: \def\lPlanck{{\ell_p}} % Planck scale

94: 

95: \def\dualphi{{\tilde{\phi}}} % T-dual coordinate

96: \def\BfieldNS{{B^{\text{\tiny (NSNS)}}}} % NSNS 2-form field

97: \newcommand\AfieldRR[1]{{A^{\text{\tiny (RR)}}_{{#1}}}} % RR 

98: 

99: 

100: \newcommand\Op[2]{{{\cal O}^{({#1})}_{{#2}}}} % deformation operator

101: \newcommand\SUSY[1]{${\cal N}={#1}$}  % SUSY N=#1

102: 

103: % definitions for appendix (\ref{app:Operator}):

104: \def\hM{{B}} % twist matrix

105: \def\bPhi{{\overline{\Phi}}}   % complex conjugate

106: \def\bD{{\overline{D}}} % superderivative

107: \def\bW{{\overline{W}}} % anti-chiral superfield strength

108: 

109: % spinors indices

110: \def\spa{{\alpha}}

111: \def\dtspa{{\dot{\alpha}}}

112: \def\spb{{\beta}}

113: \def\dtspb{{\dot{\beta}}}

114: 

115: % spinors

116: \def\th{{\theta}}

117: \def\bth{{\overline{\theta}}}

118: 

119: \def\Spin{{\text{Spin}}} % spin group

120: 

121: \def\fThermal{{f}} % thermal factor

122: 

123: \def\Area{{A}} % horizon area

124: \def\Volume{{\cal V}} % volume

125: \def\tVolume{{\tilde{\Volume}}} % tilde volume

126: 

127: \def\VExpression{{V}} % V(E)

128: 

129: \def\dualizedRescaledU{{\tilde{\rescaledU}}} % dualized U = U/d

130: \def\dualizedRescaledH{{\tilde{\rescaledH}}} % dualized H

131: \def\dualizedRescaledAugH{{\tilde{\rescaledAugH}}} % dualized H

132: \def\dualizedDelta{{\tilde{\Delta}}} % = \Delta/d

133: 

134: \def\tz{{\tilde{z}}} % z/d

135: 

136: \def\angularForm{{\omega}}

137: \def\RCurrent{{\cal J}} % R-current

138: 

139: 

140: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

141: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

142: 

143: \begin{document}

144: \begin{titlepage}

145: \begin{flushright}

146: hep-th/0702030\\

147: MAD-TH-07-01\\

148: UCB-PTH-07/02\\

149: LBNL-62285

150: \end{flushright}

151: %\vspace{12 mm}

152: 

153: \vfil\

154: %vfil

155: 

156: \begin{center}

157: 

158: 

159: {\Large{\bf Aspects of Puff Field Theory}}

160: 

161: 

162: \vfil

163: 

164: Ori J. Ganor,$^{1,2}$ Akikazu Hashimoto,$^3$

165: Sharon Jue,$^1$\\

166: Bom Soo Kim,$^{1,2}$ and

167: Anthony Ndirango$^{1}$

168: 

169: \vfil

170: 

171: ${}^1$Department of Physics, University of California, Berkeley, CA 94720\\

172: ${}^2$Theoretical Physics Group, Lawrence Berkeley National Laboratory, Berkeley, CA 94720\\

173: ${}^3$Department of Physics, University of Wisconsin, Madison, WI 53706

174: \vfil

175: 

176: \end{center}

177: 

178: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

179: \begin{abstract}

180: \noindent 

181: We describe some features of the recently constructed ``Puff Field

182: Theory,'' and present arguments in favor of it being a field theory

183: decoupled from gravity.  We construct its supergravity dual and

184: calculate the entropy of this theory in the limit of large 't Hooft

185: coupling.  We also determine the leading irrelevant operator that

186: governs its deviation from ${\cal N}=4$ super Yang-Mills theory.

187: 

188: 

189: \end{abstract}

190: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

191: \vspace{0.5in}

192: 

193: 

194: \end{titlepage}

195: \renewcommand{\baselinestretch}{1.05}  %Line spacing

196: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

197: 

198: \section{Introduction}

199: The Melvin Universe is an exact solution of 

200: Einstein gravity coupled with

201: gauge fields \cite{Melvin:1963qx}. It describes a consistent

202: gravitational backreaction when one attempts to support a uniform

203: magnetic field in the background. (Of course, the magnetic field is no

204: longer uniform when the gravitational backreaction is taken into

205: account.)

206: Melvin Universes are especially  natural  in the context of

207: Kaluza-Klein theory. Simply consider twisting an angular coordinate so that the space-time, where a plane is expressed in cylindrical coordinates, has the form

208: %

209: \be ds^2 = -dt^2 + dx^2 + dr^2 

210: + r^2 (d \anglephi + \eta dz)^2 + dz^2 \ . \label{twist}\ee

211: %

212: Here $z \sim z + 2 \pi R$ is periodic, 

213: and $\eta$ therefore cannot be trivially eliminated by

214: a change of coordinates

215: (since a redefinition $\anglephi\rightarrow\anglephi+\eta z$

216: would modify the periodicity conditions on $z$ and $\anglephi$).

217: For $\eta = 0$, the space reduces to flat space in 4+1

218: dimensions. Kaluza-Klein reduction along the $z$ coordinate 

219: gives rise to a space-time with background magnetic field 

220: and some background scalar field configuration.

221: 

222: This type of space-time has a natural embedding in string theory. Simply embed (\ref{twist}) in 10 or 11 dimensional supergravity. One concrete realization is to embed (\ref{twist}) in type IIA supergravity. In this case, one can find an interesting type IIB supergravity solution by T-dualizing along the $z$ direction. The type IIB supergravity solution is of the form

223: %

224: \beq ds^2 & = & -dt^2 + d \vec x^2 + dr^2 

225: + {{r^2 d \anglephi^2} \over {1 + \eta^2 r^2}}

226: + {1 \over 1 + \eta^2 r^2} d\tilde z^2 \cr

227: B & = & 

228:   {\eta r^2 \over {1 + \eta^2 r^2}} 

229:      d \anglephi \wedge d \tilde z \cr

230: e^{\dilaton} & = & \sqrt{1 \over 1 + \eta^2 r^2} \cr

231: \tilde z &\sim& \tilde z + 2 \pi \tilde R, 

232: \qquad 

233: \tilde R  = {\alpha' \over R} \ .  

234: \label{melvinbg}

235: \eeq

236: %

237: This background is not supersymmetric. However, it is straightforward

238: to twist in more than one plane in such a way that some fraction of

239: supersymmetry is preserved.

240: 

241: These background space-times, from the point of view of string theory,

242: are special in that their world sheet sigma model is exactly solvable

243: \cite{

244: Russo:1994cv,Tseytlin:1994ei,Russo:1995tj,Tseytlin:1995fh,Russo:1995aj,Russo:1995ik}. This

245: follows, in essence, from the fact that they are dual to flat space

246: with some periodic identifications.

247: 

248: Melvin universes are also useful for constructing non-local quantum

249: field theories as a decoupling limit. Simply consider adding a

250: D3-brane in the background (\ref{melvinbg}) and take the appropriate

251: scaling limit for the parameters $\eta$, $r$, and $R$ as one sends

252: $\alpha'$ to zero. In enumerating distinct embeddings of D3 branes in this background, 

253: it is useful to note that there are essentially two

254: special spatial coordinates, $\tilde z$ and $\anglephi$. 

255: Taking both

256: $\tilde z$ and $\anglephi$ to be along the brane gives rise to

257: non-commutative geometry with a position dependent non-commutativity

258: parameter \cite{Hashimoto:2004pb,Hashimoto:2005hy} for which the

259: deformation quantization formula of Kontsevich

260: \cite{Kontsevich:1997vb} becomes relevant.

261: Taking $\tilde z$ to be along the brane but $\anglephi$ 

262: to be transverse gives rise to a dipole

263: deformation, which was introduced in

264: \cite{Bergman:2000cw,Bergman:2001rw}.  

265: The construction by Lunin and

266: Maldacena of $\beta$-deformed \SUSY{4} superconformal theory 

267: \cite{Lunin:2005jy}

268: can also be viewed as an example of this construction where both 

269: $\tilde z$ and $\anglephi$ are transverse to the brane. 

270: Other constructions of a similar kind were studied in 

271: \cite{Hashimoto:2002nr} and \cite{Alishahiha:2003ru}.

272: There are

273: generalizations of these constructions that can be obtained

274: by modifying the embeddings of

275: the $\anglephi$ and the $\tilde z$ coordinates in the full geometry, which

276: were classified and tabulated in \cite{Hashimoto:2004pb}. 

277: Most of these

278: constructions give rise to a non-local field theory as a decoupling

279: limit of string theories with solvable world sheet sigma model.

280: 

281: Recently, one of us proposed an example of a non-local field theory

282: not included in \cite{Hashimoto:2004pb}, which arises from a novel

283: embedding of a D-brane in a Melvin universe \cite{Ganor:2006ub}. 

284: The construction proceeds as follows. 

285: %

286: \begin{enumerate}

287: \item Start with $N$ D0-branes in flat 10

288: dimensional background of type IIA string theory. 

289: The M-theory lift of

290: this configuration corresponds to a Kaluza-Klein wave 

291: traveling along the M-theory circle. 

292: Let us parameterize the coordinates of the background

293: geometry in 11 dimensions using coordinates

294: %

295: \be ds^2 = -dt^2 + \sum_{i=1}^3 dx_i^2

296: + \sum_{i=1}^3 (dr_i^2 + r_i^2 d \anglephi_i^2) + dz^2,

297: \qquad 

298: z \sim z + 2 \pi R \ . \ee

299: %

300: \item Twist the angular coordinate $\anglephi_i$ by a deformation parameter $\beta_i$ so that the metric becomes

301: %

302: \be ds^2 = -dt^2 + \sum_{i=1}^3 dx_i^2

303: + \sum_{i=1}^3 \left\lbrack

304: dr_i^2 + r_i^2 (d \anglephi_i + \beta_i dz)^2\right\rbrack

305: + dz^2\ . 

306: \ee

307: %

308: \item Reduce back to type IIA along the $z$ coordinate, giving rise to a D0-brane in a Melvin universe 

309: supported by a flux of magnetic RR 2-form field strength.

310: 

311: \item T-dualize along the $x_1$, $x_2$, and $x_3$ directions. This will give rise to  a configuration of D3-branes embedded in a Melvin universe with  RR 5-form field strength in the background in type IIB theory.

312: \end{enumerate}

313: %

314: The construction described in \cite{Ganor:2006ub} started with a KK

315: wave in type II theory but leads to the same geometry. 

316: 

317: By taking a suitable scaling limit involving $\alpha'$, $\beta_i$,

318: and

319: the compactification radii as we will describe in more detail

320: below, one arrives at a presumably decoupled system of 

321: non-local quantum field theory, 

322: similar in many regards to non-commutative Yang-Mills theory,

323: dipole theory, and NCOS. This theory was named the ``Puff Field

324: Theory'' (PFT) in \cite{Ganor:2006ub}, because the light degrees of

325: freedom ``puff up'' in all three dimensions. The distinguishing

326: feature of PFT is the fact that it leaves unbroken a spatial

327: subgroup of the Lorentz group $SO(3) \subset SO(1,3)$, 

328: unlike the more familiar

329: non-commutative/dipole field theories. Such a theory could possibly be

330: of phenomenological interest when applied to maximally isometric

331: cosmological scenarios of the Freedman-Robertson-Walker type.

332: 

333: The goal of this article is to describe the features of PFT

334: and to provide more evidence in support of the conjecture

335: that PFT is decoupled from gravity.

336: This is facilitated by the explicit construction 

337: of the supergravity dual.

338: This paper is organized as follows. 

339: In section~\ref{sec:SupergravityDual}, we

340: describe the construction of the supergravity dual itself. 

341: In section~\ref{sec:Thermodynamics}, 

342: we construct the finite temperature case and describe the

343: thermodynamics of PFT.

344: In section~\ref{sec:Renormalization},

345: we describe the RG flow of PFT in greater detail.

346: In section~\ref{sec:Deformation},

347: we identify the leading irrelevant

348: operator responsible for deforming \SUSY{4} SYM to PFT.  

349: We conclude in section~\ref{sec:Concluding}.

350: 

351: % ==================================================

352: \section{Supergravity dual of PFT

353: \label{sec:SupergravityDual}}

354: 

355: In this section, we describe the construction of the supergravity solution that is dual to PFT. 

356: In what follows, we present the steps leading 

357: to the solution (\ref{eqn:dualSG}) in some detail.

358: After that, we will analyze the regime of validity of

359: the classical supergravity solution.

360: 

361: To obtain the supergravity dual of PFT,

362: we start from type IIA supergravity compactified along the $x_i$ directions and consider $N$ D0-branes smeared along the compact directions. In anticipation of the $T$-duality, we will denote the period of compactification of $x_i$ by 

363: ${\alpha' /  R_i}$. 

364: This geometry can be written explicitly in the form

365: %

366: \beq ds^2 &=& -h^{-1/2} dt^2 + h^{1/2} 

367: \left(\sum_{i=1}^3 dx_i^2 + \sum_{i=4}^9 d y_i^2 \right) \cr

368: A & = & h^{-1} dt \cr

369: e^{2 \dilaton} &=& h^{3/2}\ .  \label{IIA} \eeq

370: %

371: where

372: %

373: \be h = 1 + 

374: {60 \pi^3 \gIIA N \alpha'^{7/2} \over (\|x\|^2 + \|y\|^2)^{7/2}}

375: \ee

376: %

377: for the ``non-smeared'' solution (for which directions

378: $1,2,3$ are noncompact and $-\infty<x_1, x_2, x_3<\infty$),

379: and

380: %

381: \be h = 1 + {4 \pi \gIIA N R_1 R_2 R_3   \alpha'^{1 / 2} \over 

382: \|y\|^4} \ee

383: %

384: for the ``smeared'' solution (which is obtained

385: from the ``non-smeared'' one

386: by replacing the second term in $h$ with its integral

387: over the full range of $x_1, x_2, x_3$ and dividing by the

388: total volume $(2\pi\alpha')^3/R_1 R_2 R_3$ of the $T^3$).

389: Here 

390: \be

391: \|x\|^2\equiv\sum_{i=1}^3 x_i^2,

392: \qquad

393: \|y\|^2\equiv\sum_{i=4}^9 y_i^2\ .

394: \ee

395: Now we perform the M-theory lift, twist, and reduction back to type

396: IIA.  The M-theory lift of (\ref{IIA}) is purely geometric

397: %

398: \be ds^2 =  - h^{-1} dt^2 + h (dz - h^{-1} dt)^2 +  

399: \left(\sum_{i=1}^3 dx_i^2 + \sum_{i=4}^9 d y_i^2 \right)\ . \ee

400: %

401: Here $z$ is a periodic coordinate (the ``M-circle'')

402: with period $z\sim z+2\pi \gIIA\lst$ 

403: (where $\lst\equiv{\alpha'}^{1/2}$ is the string scale).

404: One can in principle consider letting all $\beta_i$'s take

405: independent values in performing the twist.  We will, however,

406: concentrate on the case where $\beta_1 = \beta_2 = \eta$ and

407: $\beta_3 =0$, which leaves half of the supersymmetries unbroken,

408: and will at the end give us a theory with ${\cal N}=2$ in 4D.

409: Then, the metric after the twist has the form

410: %

411: \be ds^2 =  - h^{-1} dt^2 + h (dz - h^{-1} dt)^2 +  

412: \sum_{i=1}^3 dx_i^2 + \sum_{i=1}^2

413: \left\lbrack dr_i^2  

414: + r_i^2 (d \anglephi_i + \eta dz)^2\right\rbrack

415:  + \sum_{i=8}^9 d y_i^2\ . 

416: \label{eqn:twisted}

417: \ee

418: %

419: Before we proceed, we make another convenient change of variables.

420: We replace the four coordinates $r_1, r_2, \anglephi_1, \anglephi_2$

421: with the radial variable

422: \be

423: \fullu\equiv \sqrt{r_1^2 + r_2^2},

424: \qquad

425: (0\le\fullu<\infty)\ ,

426: \ee

427: and three new angular coordinates

428: $\fiberphi, \basetheta, \basephi$ 

429: defined as follows,

430: \be

431: \basephi\equiv\anglephi_1-\anglephi_2,

432: \qquad

433: \sin\basetheta\equiv \frac{2 r_1 r_2}{r_1^2 + r_2^2},

434: \qquad

435: \fiberphi\equiv\anglephi_1\ .

436: \ee

437: For fixed $t,x_1,x_2,x_3,y_8,y_8,$ and $\fullu$,

438: the variables $\fiberphi, \basetheta, \basephi$

439: describe an $S^3$ in the form of a Hopf fibration:

440: $(\basetheta,\basephi)$  are spherical coordinates on the

441: $S^2$ base, and $\fiberphi$ is a periodic coordinate

442: on the $S^1$ fiber with period $2\pi.$

443: Presenting $S^3$ as a Hopf fibration is convenient

444: because the twist is in the direction of the fiber $\fiberphi.$

445: In order to save space,

446: we denote the (Fubini-Study)

447: metric on the base of the Hopf fibration by

448: \be

449: \dsBase{2}^2\equiv\frac{1}{4}

450: (d\basetheta^2 + \sin^2\basetheta\,d\basephi^2)\ ,

451: \label{eqn:dsBase}

452: \ee

453: and we denote the connection of the Hopf fibration by

454: \be

455: \connHopf\equiv

456: -\frac{1}{2}(1-\cos\basetheta)d\basephi\ .

457: \label{eqn:defconnHopf}

458: \ee

459: The metric can now be expressed in the form

460: %

461: \beq ds^2 &=&  

462: - h^{-1} dt^2 + h (dz - h^{-1} dt)^2 

463: + \sum_{i=1}^3 dx_i^2 

464: + d\fullu^2

465: \cr &&\qquad

466: + \fullu^2 \left\lbrack

467: \dsBase{2}^2 

468: + (d \fiberphi +  \eta dz + \connHopf)^2\right\rbrack 

469: + \sum_{i=8}^9 d y_i^2\ .

470: \label{Mtheory} 

471: \eeq

472: %

473: Reducing along $dz$ to IIA then gives

474: %

475: %

476: \beq ds^2 & = & 

477: \augmentedh^{1/2}\Bigl( - h^{-1} dt^2

478: +\sum_i dx_i^2 

479: + d\fullu^2  + \fullu^2 \dsBase{2}^2 

480: +\sum_{i=8,9} dy_i^2 \Bigr)

481: \cr && \qquad

482: + \augmentedh^{-1/2}h\fullu^2

483: (d \fiberphi + \connHopf + \eta h^{-1} dt)^2\ ,

484: \cr

485: A & = &  

486: \augmentedh^{-1}

487:   \left( -  dt 

488:  + \fullu^2 \eta (d \fiberphi  + \connHopf) \right)\ , \cr

489: e^\dilaton & = & \gIIA \augmentedh^{3/4}\ ,

490: \eeq

491: where

492: \be

493: \augmentedh\equiv h + \eta^2 u^2 

494: =1 + {4 \pi \gIIA N R_1 R_2 R_3 \alpha'^{1/2} 

495:    \over (u^2 + \|y\|^2)^2} + \eta^2 u^2\ .

496: \ee

497: T-dualizing along $x_1$, $x_2$, and $x_3$ then gives rise to the supergravity solution

498: \beq ds^2 & = & 

499: \augmentedh^{1/2} \left( - h^{-1} dt^2

500: + d\fullu^2  + \fullu^2 \dsBase{2}^2 

501: +\sum_{i=8,9} dy_i^2 \right) 

502: \cr && \qquad

503: +\augmentedh^{-1/2} \left(

504: \sum_i dx_i^2 

505: + h  \fullu^2(d \fiberphi + \connHopf + \eta h^{-1} dt)^2

506: \right)\ ,

507: \cr

508: A & = &  \augmentedh^{-1}

509: \left( -  dt + \fullu^2 \eta (d \fiberphi+\connHopf) \right) 

510: \wedge dx_1 \wedge dx_2 \wedge dx_3\ ,

511: \cr

512: e^\dilaton & = & \gIIB =  R_1 R_2  R_3 \alpha'^{-3/2} \gIIA\ .

513: \label{eqn:FullSolution}

514: \eeq

515: %

516: Here $A$ is not quite the full Ramond-Ramond gauge 4-form field,

517: but the full Ramond-Ramond 5-form field strength is

518: the self-dual part of $dA.$

519: In the new variables, $h$ and $\augmentedh$ take the forms

520: %

521: \be h = 1 + {4 \pi \gIIB N \alpha'^2 \over 

522: (\fullu^2 + \|y\|^2)^2} \ ,\qquad

523: \augmentedh = 1 + {4 \pi \gIIB N \alpha'^2 \over 

524: (\fullu^2 + \|y\|^2)^2} +\eta^2 \fullu^2\ .

525: \label{eqn:hh} \ee

526: %

527: Note that if we set $N=0$ we get $h=1$

528: and the supergravity solution reduces to

529: a Melvin universe with background RR 5-form flux.

530: For $N\neq 0$, 

531: The warping due to $h(\fullu,y)$ describes the gravitational 

532: back-reaction of the D3-branes.

533: 

534: Now, we can take the decoupling limit following the procedure of \cite{Maldacena:1997re}. 

535: As usual, in order to decouple the gauge theory from

536: the rest of the string theory modes, we take the zero slope

537: limit $\alpha'\rightarrow 0$,

538: and we need to specify how to scale the coordinates

539: $y_8, y_9,\fullu$ and the twist parameter $\eta$ in this

540: limit.

541: It turns out that the appropriate scaling is to keep finite

542: the following rescaled coordinates 

543: \be

544: \rescaledU\equiv {\alpha'}^{-1}\fullu,

545: \qquad

546: \rescaledY_i\equiv {\alpha'}^{-1}y_i\quad (i=8,9)

547: \ ,

548: \ee

549: while scaling the twist parameter $\eta$ so that

550: %

551: \be \Delta^3 \equiv \eta \alpha'^2 = \text{fixed}\ , 

552: \label{eqn:defDelta}

553: \ee

554: %

555: and keeping $\gIIB = 2 \pi \gYM^2$, $R_1$, $R_2$,

556: and $R_3$ fixed. 

557: This scaling is chosen so that in the dual supergravity

558: solution the effects of the deformation will be finite

559: in the scaling limit. For example, the second and

560: third terms in $\augmentedh$ in (\ref{eqn:hh}) are

561: comparable in this scaling limit.

562: Our scaling limit also

563: turns out to be the one suggested in \cite{Ganor:2006ub}

564: using different arguments.

565: 

566:  Note in particular that this is the

567: scaling that keeps the angle of twist 

568: per radius of the M-circle,

569: %

570: \be \chi \equiv \gIIA\lst\eta = 

571: {\eta \gIIB \alpha'^{2} \over  R_1  R_2  R_3} 

572: = {2\pi \gYM^2 \Delta^3 \over  R_1  R_2  R_3}\ ,\ee

573: %

574: finite. We will see below that an integer shift in $\chi$ leads to an equivalent theory up to a certain duality, similarly to the structure of Morita equivalence encountered in the context of ordinary non-commutative field theories.

575: We are now almost ready to take the $\alpha'\rightarrow 0$ limit.

576: We define the scaled harmonic functions 

577: (with the notation 

578: \textit{$\|\rescaledY\|^2\equiv \rescaledY_8^2 + \rescaledY_9^2$})

579: %

580: \be 

581: \rescaledH\equiv\lim_{\alpha'\rightarrow 0}\alpha'^2 h 

582: ={4 \pi \gIIB N \over (\rescaledU^2 + \|\rescaledY\|^2)^2}\ ,

583: \qquad

584: \rescaledAugH\equiv\lim_{\alpha'\rightarrow 0}\alpha'^2\augmentedh

585: ={4 \pi \gIIB N \over (\rescaledU^2 + \|\rescaledY\|^2)^2} 

586: + \Delta^6 \rescaledU^2\ ,

587: \label{eqn:defrescaledHAugH}

588: \ee

589: %

590: which captures the asymptotic behavior of the harmonic functions

591: $h$ and $\augmentedh$ in the decoupling limit, 

592: and depend only on the PFT parameters

593: \textit{$\gIIB\equiv 2\pi\gYM^2$} and $\Delta.$

594: In terms of $\rescaledH$ and $\rescaledAugH$ we can write

595: %

596: \beq {ds^2\over \alpha'} & = & 

597: \rescaledAugH^{1/2} \left( - \rescaledH^{-1} dt^2

598: + d\rescaledU^2  + \rescaledU^2 \dsBase{2}^2 

599: +\sum_{i=8,9} d\rescaledY_i^2 \right) 

600: \cr && \qquad

601: +\rescaledAugH^{-1/2} \left(

602: \sum_i dx_i^2 

603: + \rescaledH\rescaledU^2

604: (d \fiberphi + \connHopf + \Delta^3\rescaledH^{-1} dt)^2

605: \right)\ ,

606: \cr

607: {A \over \alpha'^2} & = &  

608: \rescaledAugH^{-1}

609: \left( -  dt + \rescaledU^2 \Delta^3 

610: (d \fiberphi + {\cal A}) \right) 

611: \wedge dx_1 \wedge dx_2 \wedge dx_3\ ,

612: \cr

613: e^\dilaton & = & \gIIB = 2 \pi \gYM^2\ . \label{eqn:dualSG}

614: \eeq

615: %

616: This is an exact solution of the classical

617: equations of motion of type IIB supergravity,

618: and we will interpret it as the supergravity dual of PFT.  

619: Much of the conclusions we draw regarding the nature of PFT 

620: will be based on this supergravity solution, 

621: which is one of the main results we are

622: reporting in this paper.

623: 

624: PFT depends on a dimensionful parameter $\Delta$,

625: which according to its definition in (\ref{eqn:defDelta}),

626: has dimensions of length.

627: For generic $\Delta$, the solution (\ref{eqn:dualSG})

628: is invariant under Poincar\'e transformations in

629: the $t, x_1, x_2, x_3$ directions, under $SU(2)$

630: rotations of the base $B(2)$ of the Hopf fibration

631: (acting on the spin structure and the fiber direction

632: $\fiberphi$ as well), under $U(1)$ translations generated by

633: the vector field $\partial/\partial\fiberphi$, 

634: and under 8 supersymmetries.

635: For $\Delta=0$ the solution (\ref{eqn:dualSG})

636: reduces to $AdS_5\times S^5$ --

637: the coordinates $t, x_1, x_2, x_3$ and 

638: \be

639: \radialV\equiv\sqrt{\rescaledU^2 +\|\rescaledY\|^2}

640: \label{eqn:defV}

641: \ee

642: parameterize the $AdS_5$ part.

643: ($\radialV$ can be traced back to the radial direction

644: transverse to the D3-brane.)

645: For later use, it is also convenient to define

646: the angle $0\le\anglepsi\le\pi/2$ by

647: \be

648: \rescaledU=\radialV\cos\anglepsi,\qquad

649: \|\rescaledY\|=\radialV\sin\anglepsi\ .

650: \label{eqn:defchi}

651: \ee

652: Up to factors of certain powers of 

653: $\lambda = 2 \pi \gYM^2 N = \gIIB N$

654: that will be discussed later,

655: $\Delta^{-1}$ sets the interesting energy scale for PFT.

656: This is the scale above which

657: PFT becomes appreciably different from \SUSY{4} SYM. 

658: This can be seen directly from (\ref{eqn:dualSG}):

659: for $\rescaledU \ll \Delta^{-1}$ (and fixed $\lambda$), the

660: solution asymptotes to $AdS_5 \times S_5$, indicating that the

661: infra-red fixed point of this theory is \SUSY{4} SYM.

662: On the other hand, for $\rescaledU \gg \Delta^{-1}$, 

663: the solution deviates strongly from the

664: $AdS_5\times S_5$ background. The supergravity duals of

665: other nonlocal field theories such as 

666: non-commutative Yang-Mills theory

667: \cite{Hashimoto:1999ut,Maldacena:1999mh} and dipole theory

668: \cite{Bergman:2001rw} also exhibit similar features.

669: 

670: % - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

671: \subsubsection*{Regime of validity}

672: 

673: Let us comment on the region of validity of the dual supergravity description. 

674: The 't Hooft coupling constant

675: $\lambda = 2 \pi \gYM^2 N = \gIIB N$ must be large in order for the curvature to be weak.

676: We assume that the Yang-Mills coupling constant $\gYM$ itself

677: is kept finite.

678: (Of course, $\gYM$ has to be small if one wishes to

679: extend the discussion beyond the classical supergravity

680: description, for example, to include the excited string spectrum.)

681: 

682: If $\lambda\gg 1$,

683: the curvature is small everywhere

684: provided $\anglepsi\neq \pi/2.$

685: More specifically,

686: the invariant square of the curvature tensor,

687: as calculated from the string-frame metric (\ref{eqn:dualSG}), is

688: \beq

689: \RCurvature_{\mu\nu\sigma\tau}\RCurvature^{\mu\nu\sigma\tau} &=&

690: {\alpha'}^{-2}

691: \bigl(

692: 4\pi\lambda

693: +\Delta^6 \rescaledU^2\radialV^4

694: \bigr)^{-5}

695: \Bigl\lbrack

696: 80(4\pi\lambda)^4

697: -80(4\pi\lambda)^3\Delta^6\radialV^4 (3\rescaledU^2 + 5\radialV^2)

698: \cr &&

699: +24(4\pi\lambda)^2\Delta^{12}\radialV^8

700:   (136\rescaledU^4 + 29\rescaledU^2\radialV^2 + 5\radialV^4)

701: \cr &&

702: +32\pi\lambda\Delta^{18}\radialV^{12}\rescaledU^2

703:   (15\radialV^4 + 7\rescaledU^2\radialV^2 -72\rescaledU^4)

704: +65\Delta^{24} \rescaledU^4\radialV^{20}

705: \Bigr\rbrack\ .

706: \eeq

707: So, if $\anglepsi\neq\pi/2$ we see that for

708: \textit{$\rescaledU\Delta\ll\lambda^{1/6}$} the curvature

709: scale is of order \textit{${\alpha'}^{-1/2}\lambda^{-1/4}$},

710: while for \textit{$\rescaledU\Delta\gg\lambda^{1/6}$}

711: the curvature scale is of order

712: \textit{${\alpha'}^{-1/2}(\rescaledU\Delta)^{-3/2}$}. 

713: Both of these

714: quantities are small for $\lambda \gg 1$.

715: If $\anglepsi=\pi/2$, on the other hand, the curvature is small

716: only for  \textit{$\Delta\|\rescaledY\|\ll\lambda^{3/4}$}. Therefore, 

717: observables that sensitively probe the $\anglepsi=\pi/2$ region might receive corrections due to stringy effects.

718: 

719: Another requirement for the classical supergravity analysis

720: to be applicable is that the proper size of the various

721: compact directions be large compared to the string scale.

722: For the $\fiberphi$ direction we get

723: \be

724: \rescaledAugH^{-1/2}\rescaledH\rescaledU^2 \gg 1.

725: \ee

726: Assuming again that $\anglepsi\neq\pi/2$,

727: we find that the $\fiberphi$ direction is large if

728: \textit{$\rescaledU\Delta\ll\lambda^{1/3}$}.

729: On the other hand, if \textit{$\rescaledU\Delta\gg\lambda^{1/3}$}

730: the $\fiberphi$-circle is smaller than string scale

731: and its radius is of order

732: \textit{${\alpha'}^{1/2}\lambda^{1/2}(\rescaledU\Delta)^{-3/2}$}.

733: In this regime 

734: the supergravity dual (\ref{eqn:dualSG}) cannot be trusted.

735: For $\anglepsi\neq\pi/2$, the radius of the $\fiberphi$-circle

736: shrinks to zero, but the solution is not singular.

737: To see this, note that for fixed nonzero $\|\rescaledY\|$ we

738: have $\radialV\neq 0$, and as 

739: $\anglepsi\neq\pi/2$ (and therefore $\rescaledU\rightarrow 0$)

740: the metric on the base \textit{$\dsBase{2}^2$} and

741: the fiber $\fiberphi$ combine to a metric on $S^3$, and together

742: with the $\rescaledU$ direction we get the metric on a ball.

743: 

744: Next, we need to discuss directions $x_1, x_2, x_3.$

745: The proper size of each of these compact directions

746: needs to be large in comparison to string scale. 

747: One way to achieve this is to simply take the 

748: decompactification limit $R_i\rightarrow\infty$ ($i=1,2,3$)

749: and formulate PFT on $\Reals^{3,1}$, so to speak.

750: Alternatively, we can keep the compactification radii

751: $R_1, R_2, R_3$ finite.

752: Taking this approach, as we shall see in 

753: section~\ref{sec:Renormalization}, yields a richer

754: structure of energy scales in the theory,

755: but then in order 

756: for the classical supergravity solution (\ref{eqn:dualSG})

757: to be valid, we need the additional requirements

758: \be

759: R_i\gg\rescaledAugH^{1/4},\qquad i=1,2,3.

760: \label{eqn:RiA}

761: \ee

762: Assuming that $\anglepsi\neq\pi/2$, 

763: we find that \textit{$\rescaledAugH$}

764: is of the order of 

765: \textit{$4\pi\lambda\rescaledU^{-4}+\Delta^6\rescaledU^2$}.

766: This expression is never smaller than

767: \textit{$3(\pi\lambda)^{1/3}\Delta^4$}, and therefore 

768: (\ref{eqn:RiA}) will not be satisfied unless

769: %

770: \be  R_i \gg \lambda^{1/12} \Delta\ ,

771: \qquad (i=1,2,3). \label{eqn:assump} \ee

772: %

773: Assuming (\ref{eqn:assump}) now,

774: the condition (\ref{eqn:RiA}) sets the following range

775: requirement for $\rescaledU$:

776: \be {\lambda^{1/4} \over  R_i} \ll 

777: \rescaledU \ll { R_i^2  \over \Delta^3} \ . \label{eqn:Urange} \ee

778: The validity conditions that we found so far can be recast in

779: terms of the energy scale.

780: For $AdS_5\times S^5$,

781: the holographic energy/distance relation  \cite{Peet:1998wn}

782: takes the form

783: %

784: \be E = {\rescaledU \over \sqrt{\lambda}} \ . \label{uvir} \ee

785: %

786: In order of magnitude,

787: this form is also applicable to our metric, at least 

788: if we assume that $\cos\anglepsi$ is of order $O(1)$,

789: so that $\rescaledU$ and $\radialV$ are comparable.

790: We will demonstrate this later in (\ref{eqn:UT}) of

791: section~\ref{sec:Thermodynamics}, 

792: when we discuss PFT at nonzero temperature.

793: 

794: The conditions  (\ref{eqn:Urange}) can now be written

795: as a range of energy scales

796: \be {\lambda^{-1/4} \over  R_i} \ll E \ll {\lambda^{-1/2}  R_i^2  \over \Delta^3} \ , \label{eqn:Erange} \ee

797: %

798: and the condition about the $\fiberphi$-circle

799: that was found above becomes

800: \be

801: E\ll\frac{\lambda^{-1/6}}{\Delta}

802: \label{eqn:ERangeFiber}.

803: \ee

804: Combining (\ref{eqn:ERangeFiber}) and (\ref{eqn:Erange}) we get

805: \be 

806: {\lambda^{-1/4} \over  R_i} \ll E \ll 

807: \min\bigl\{

808: \frac{\lambda^{-1/6}}{\Delta},

809: {\lambda^{-1/2}  R_i^2  \over \Delta^3}

810: \bigr\} \ . \label{eqn:ERangeAll} 

811: \ee

812: Similar sets of bounds on the region of validity for the case of non-commutative Yang-Mills theory were 

813: pointed out in \cite{Hashimoto:1999yj}.

814: Note also that, assuming (\ref{eqn:assump}) and $\lambda\gg 1$,

815: we have the inequality

816: \be 

817: {\lambda^{-1/4} \over  R_i}

818: \ll

819: \frac{\lambda^{-1/3}}{\Delta}

820: \ll 

821: \min\bigl\{

822: \frac{\lambda^{-1/6}}{\Delta},

823: {\lambda^{-1/2}  R_i^2  \over \Delta^3}

824: \bigr\} \ . 

825: \label{eqn:InterestingScale}

826: \ee

827: The energy scale \textit{$\lambda^{-1/3}/\Delta$} is 

828: important because it corresponds to 

829: \textit{$\rescaledU\Delta = \lambda^{1/6}$},

830: which is the scale at which the metric starts to deviate

831: markedly from $AdS_5\times S^5.$ 

832: For example, below that scale $\rescaledH\approx\rescaledAugH.$

833: This is therefore the scale at which PFT effects enter into

834: play, and we see from (\ref{eqn:InterestingScale})

835: that it is inside the range of validity (\ref{eqn:ERangeAll}).

836: 

837: In (\ref{eqn:ERangeAll}),

838: the lower bound \textit{$\lambda^{-1/4}/R_i \ll E$} is

839: independent of the non-locality and applies 

840: just as well to the case of ordinary $AdS_5 \times S_5$

841: compactified on a circle. The bound simply

842: indicates the presence of finite size effects cutting off the spectrum in the IR. 

843: The order of magnitude of the

844: size of a typical excitation with energy $E$ can be

845: estimated as the Compton length $L = 1/E$, 

846: but this estimate fails

847: when the size of the excitation gets bigger 

848: than the size of the box.

849: This explains why the lower bound on $E$ is proportional to

850: $1/R_i.$

851: The factor of \textit{$\lambda^{-1/4}$} in the bound

852: is the effect of strong coupling.

853: For energies below the bound,

854: \textit{$E\ll\lambda^{-1/4}/R_i$}, one should

855: look for a description in terms of the near horizon 

856: geometry of a lower dimensional brane. Readers are referred to

857: \cite{Itzhaki:1998dd,Brandhuber:1998er} for explanations concerning

858: the correct cross-over behavior and the correspondence 

859: principle at work around this scale.

860: 

861: The upper bound on $E$ in

862: (\ref{eqn:Erange}) implies that the size of a typical excitation

863: starts to grow with energy according to 

864: \textit{$L\sim\sqrt{\lambda^{1/2}\Delta^3 E}$},

865: so that the upper bound is reached when \textit{$L\sim R_i$}.

866: This is a characteristic feature of non-local field theories. 

867: The size of an object grows both 

868: in the extreme IR and in the extreme UV.

869: When the size of the

870: object becomes larger than the size of the box, 

871: one must adopt an alternative description. 

872: We will comment further on this issue in

873: section~\ref{sec:Renormalization}.

874: 

875: 

876: % - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

877: \subsubsection*{The high energy regime 

878: \textit{$E\gg \lambda^{-1/6}/\Delta$}}

879: 

880: So far, we discussed the upper bound on energy coming from

881: (\ref{eqn:Erange}), but we also have another upper bound

882: from (\ref{eqn:ERangeFiber}).

883: The latter suggests another interesting 

884: length-scale in the problem,

885: namely \textit{$\lambda^{1/6}\Delta$},

886: at least for strong `t Hooft coupling $\lambda\gg 1.$

887: As we reach the corresponding range 

888: \textit{$\rescaledU\sim\lambda^{1/3}/\Delta$}

889: in the supergravity solution (\ref{eqn:dualSG})

890: the ten-dimensional description loses its classical

891: interpretation.

892: If the compactification radii $R_1,$ $R_2$ and $R_3$

893: are all much bigger than the length scale

894: \textit{$\lambda^{1/6}\Delta$}, then

895: \be

896: \frac{\lambda^{1/3}}{\Delta}

897: \ll

898: \frac{R_i^2}{\Delta^3},

899: \ee

900: and the energy scale $E$ (or equivalently, $\rescaledU$)

901: at which the $\fiberphi$-circle becomes comparable to

902: the string length-scale is lower than the energy scale

903: at which the radii of the $x_1$, $x_2$, $x_3$ fall below

904: the string scale.

905: In particular, this is the case in the decompactification limit

906: $R_i\rightarrow\infty$.

907: There is then a range of $\rescaledU$ for which,

908: even though the $\fiberphi$-circle is small,

909: we can still dimensionally reduce

910: along it to get a valid nine-dimensional

911: classical supergravity description, as long as we keep

912: away from the $\anglepsi=\pi/2$ locus, near which

913: the size of the fiber varies rapidly.

914: 

915: In the extreme regime 

916: \textit{$\rescaledU\gg\lambda^{1/3}/\Delta$}

917: the $\fiberphi$-circle is smaller than string scale,

918: and it therefore makes sense to apply a T-duality transformation,

919: at least away from the $\anglepsi=\pi/2$ locus.

920: Using the formulas of \cite{Bergshoeff:1995as} we arrive

921: at a background with the following NSNS fields:

922: %

923: \beq {ds^2\over \alpha'} & = & 

924: \rescaledAugH^{1/2} \left( - \rescaledH^{-1} dt^2

925: + d\rescaledU^2  + \rescaledU^2 \dsBase{2}^2 

926: +\sum_{i=8,9} d\rescaledY_i^2 \right) 

927: \cr && \qquad

928: +\rescaledAugH^{-1/2}\sum_i dx_i^2 

929: +\rescaledAugH^{1/2}\rescaledH^{-1}\rescaledU^{-2}d \dualphi^2\ ,

930: \cr

931: %\alpha'^{-3/2}\AfieldRR{3} & = & 

932: %\frac{8}{3}

933: %\rescaledAugH^{-1}

934: %\rescaledU^2 \Delta^3 

935: %dx_1 \wedge dx_2 \wedge dx_3

936: %+\Bigl(\cdots\Bigr)\rescaledY_8 d\rescaledY_9\wedge d\connHopf,

937: %\cr

938: \frac{1}{\alpha'}\BfieldNS &=& d\dualphi\wedge

939: (\connHopf + \Delta^3\rescaledH^{-1} dt)\ ,

940: \cr

941: e^\dilaton & = & 

942: 2 \pi \gYM^2 \rescaledAugH^{1/4}\rescaledH^{-1/2}\rescaledU^{-1}\ . 

943: \label{eqn:dualSGIIA}

944: \eeq

945: %

946: Here $\dualphi$ is a periodic variable with period $2\pi$

947: that parameterizes the T-dual circle, and

948: there are also nonzero RR fields that 

949: have not been written down here, for simplicity.

950: (See also \cite{Duff:1998us} for a related discussion

951: where T-duality has been applied to $AdS_5\times S^5$

952: by viewing the $S^5$ as a Hopf fibration over a base 

953: \textit{$CP^2$}.)

954: There are, however, at least three extra complications:

955: \begin{enumerate}

956: \item

957: The physics at the 

958: vicinity of the locus $\anglepsi=\pi/2$ is not 

959: captured properly by (\ref{eqn:dualSGIIA}).

960: As we will now explain, the strongly curved metric in that

961: region should be replaced by an NS5-brane.

962: To see this, first note that for fixed $t$,

963: $x_1$, $x_2$, $x_3$, $\dualphi$ and fixed $\radialV\neq 0$,

964: the remaining parts of the metric

965: \be

966: \rescaledAugH^{1/2} \left( 

967: d\rescaledU^2  + \rescaledU^2 \dsBase{2}^2 

968: +\sum_{i=8,9} d\rescaledY_i^2 \right) 

969: \label{eqn:pieceds}

970: \ee

971: describe a space that is topologically equivalent to

972: an \textit{$S^4$}.

973: In fact, defining a new periodic coordinate

974: $0\le\angleY<2\pi$ by

975: \be

976: \rescaledY_8 = \|\rescaledY\|\cos\angleY,

977: \qquad

978: \rescaledY_8 = \|\rescaledY\|\sin\angleY,

979: \ee

980: our piece of the metric reduces to

981: \be

982: \rescaledAugH^{1/2}

983: \radialV^2 \bigl(

984: d\anglepsi^2 

985: +\frac{1}{4}\cos^2\anglepsi\,

986: (d\basetheta^2 + \sin^2\basetheta\,d\basephi^2)

987: +\sin^2\anglepsi\,d\angleY^2\bigr)\ ,

988: \label{eqn:almostS4}

989: \ee

990: where we used (\ref{eqn:dsBase}).

991: And were it not for the explicit dependence of 

992: $\rescaledAugH$ on $\anglepsi$ and 

993: for the factor of $1/4$ in front of the second term,

994: the metric (\ref{eqn:almostS4}) would describe an $S^4$ exactly.

995: We are now ready to analyze the region near $\anglepsi=\pi/2.$

996: The locus $\anglepsi=\pi/2$ is an $S^1$ 

997: (that can be parameterized by $\angleY$), and

998: the $(\basetheta,\basephi)$ variables describe an \textit{$S^2$}

999: that shrinks to zero as $\anglepsi\rightarrow \pi/2.$

1000: The radius of the $\dualphi$-circle, on the other hand,

1001: increases indefinitely.

1002: As we approach $\anglepsi=\pi/2$, however, the flux of the NSNS 

1003: 3-form field-strength through the \textit{$S^2\times S^1$}

1004: (generated by $\basetheta$, $\basephi$ and $\dualphi$)

1005: remains finite:

1006: $$

1007: \frac{1}{\alpha'}\int_{S^2\times S^1}d\BfieldNS =

1008:  d\dualphi\wedge d\connHopf =  4\pi^2.

1009: $$

1010: This indicates the presence of one unit of NS5-brane charge

1011: at $\anglepsi=\pi/2$.

1012: 

1013: \item

1014: When applying T-duality in superstring theory to a background

1015: that is a circle fibration, one has to be careful about

1016: the boundary conditions for fermions along the fiber direction.

1017: Specifically, before the T-duality, consider

1018: the holonomy for a closed path that wraps the fiber over

1019: a fixed point in the base. Assuming that the fiber varies

1020: slowly over the base, the geometrical holonomy is close to

1021: the identity in $SO(10)$, but when lifted to spinors

1022: the holonomy could be close to either $(+1)$ or $(-1)$

1023: in $\Spin(10).$

1024: We can determine which case corresponds to our metric

1025: by noting that the fiber of the Hopf fibration $S^1$

1026: (the $\fiberphi$-circle) over $S^2$ 

1027: (the $\basetheta,\basephi$ base) is contractible,

1028: and its spin holonomy can therefore be calculated unambiguously,

1029: and it is easy to see that it is $(-1).$

1030: Note, however, that the full holonomy of fermions along the fiber

1031: is $(+1),$ because our solution preserves supersymmetry.

1032: The minus sign from the geometrical spin holonomy

1033: is canceled by the nongeometrical contribution to the 

1034: holonomy due to the RR 5-form field strength.

1035: 

1036: \item

1037: Finally, we note that the dilaton in (\ref{eqn:dualSGIIA})

1038: gets large for

1039: \be

1040: \rescaledU\sim\frac{\lambda^{1/3}}{\gIIB^{2/3}\Delta}

1041:  \sim\frac{N^{2/3}}{\lambda^{1/3}\Delta}\ ,

1042: \label{eqn:StrongIIA}

1043: \ee

1044: assuming $\anglepsi\neq\pi/2$, as usual.

1045: This scale of $\rescaledU$ is larger than the bound

1046: \textit{$\lambda^{1/3}/\Delta$},

1047: because we are always taking $N$ to be very large.

1048: Thus, (\ref{eqn:dualSGIIA}) is likely to be valid in a range

1049: \be

1050: \lambda^{1/3}/\Delta\ll

1051: \rescaledU\ll\frac{N^{2/3}}{\lambda^{1/3}\Delta}\ .

1052: \ee

1053: The upper bound (\ref{eqn:StrongIIA}) is smaller than

1054: the upper bound of (\ref{eqn:Urange}) if

1055: \be

1056: \Delta < N^{-1/3}\lambda^{1/6}R_i\ .

1057: \label{eqn:ConditionDelta}

1058: \ee

1059: For finite $R_i$, this is never the case, but

1060: in the decompactification limit $R_i\rightarrow\infty$ this holds.

1061: Then, for \textit{$\rescaledU\gg N^{2/3}\lambda^{-1/3}/\Delta$}

1062: the dilaton becomes large, and a proper description

1063: requires 11-dimensional supergravity.

1064: Lifting the solution (\ref{eqn:dualSGIIA}) to M-theory,

1065: we get the metric

1066: \beq

1067: \frac{ds^2}{\lPlanck^2} &=&

1068: N^{2/3}\lambda^{-2/3}\Bigl\lbrack

1069: \rescaledAugH^{1/3}\rescaledH^{1/3}\rescaledU^{2/3}

1070: \bigl( - \rescaledH^{-1} dt^2

1071: + d\rescaledU^2  + \rescaledU^2 \dsBase{2}^2 

1072: +\sum_{i=8,9} d\rescaledY_i^2 \bigr) 

1073: \cr && \qquad

1074: +\rescaledAugH^{-2/3}\rescaledH^{1/3}\rescaledU^{2/3}

1075: \sum_i dx_i^2 

1076: +\rescaledAugH^{1/3}\rescaledH^{-2/3}\rescaledU^{-4/3}

1077: d \dualphi^2\Bigr\rbrack

1078: \cr && \qquad

1079: +N^{-4/3}\lambda^{4/3}

1080: \rescaledAugH^{1/3}\rescaledH^{-2/3}\rescaledU^{-4/3} 

1081: d\Mcircle^2

1082: \ ,

1083: \eeq

1084: where $0\le \Mcircle<2\pi$ is a new periodic coordinate.

1085: For fixed $\anglepsi\neq\pi/2$ and 

1086: $\rescaledU\gg\lambda^{1/6}/\Delta$, we may approximate

1087: \be

1088: \rescaledH\approx 4\pi\lambda\rescaledU^{-4}\cos^4\anglepsi,

1089: \qquad

1090: \rescaledAugH\approx\Delta^6\rescaledU^2\ ,

1091: \ee

1092: and

1093: \beq

1094: \frac{ds^2}{\lPlanck^2} &\approx&

1095: \left(\frac{N}{4\pi\lambda^2\cos^4\anglepsi}\right)^{2/3}

1096: \Bigl\lbrack

1097: -\rescaledU^{4}\Delta^2 dt^2

1098: +4\pi\lambda\cos^4\anglepsi\,

1099:   \rescaledU^{-2}\Delta^{-4}\sum_i dx_i^2 

1100: +\rescaledU^2\Delta^2 d\dualphi^2

1101: \cr &&

1102: + 4\pi\lambda\Delta^2\cos^4\anglepsi

1103: \Bigl(d\rescaledU^2  + \rescaledU^2 \dsBase{2}^2 

1104: +\sum_{i=8,9} d\rescaledY_i^2 

1105: \Bigr) 

1106: %\cr && \quad

1107: +N^{-2}\lambda^2

1108: \Delta^2\rescaledU^2 d\Mcircle^2

1109: \Bigr\rbrack

1110: \ .

1111: \label{eqn:Mlift}

1112: \eeq

1113: There is also a nonzero 3-form that we will not write down.

1114: It is interesting to note that as

1115: $\rescaledU\rightarrow\infty$,

1116: the metric (\ref{eqn:Mlift}) becomes more and more flat

1117: (for $\anglepsi\neq\pi/2$), and this suggests that

1118: the ultrahigh energy regime of noncompact PFT

1119: (all $R_i=\infty$) is holographically dual to a weakly coupled

1120: M-theory background.

1121: On the other hand, if $R_i$ is finite

1122: the size of the $x_i$ direction in (\ref{eqn:Mlift})

1123: becomes comparable to the Planck scale for

1124: \be

1125: \rescaledU\sim\frac{\lambda^{-1/6}N^{1/3}R_i}{\Delta^2}\ .

1126: \ee

1127: Beyond that scale, the lift to M-theory is insufficient,

1128: and more complicated duality transformations are in order.

1129: There is in fact an intricate phase structure depending

1130: sensitively on the rationality of $\chi$ 

1131: (or how well it is approximated by a rational number

1132: with a given denominator), which we will study in detail in

1133: section~\ref{sec:Renormalization}.

1134: 

1135: \end{enumerate}

1136: 

1137: 

1138: % ==================================================

1139: \section{Thermodynamics of Puff Field Theory}

1140: \label{sec:Thermodynamics}

1141: 

1142: A simple observable one can compute from the supergravity dual is the entropy. 

1143: The entropy as a function of temperature can be extracted from

1144: the finite temperature

1145: generalization of the dual supergravity solution by

1146: applying the Beckenstein-Hawking formula

1147: \cite{Callan:1996dv,Breckenridge:1996is}.

1148: The finite temperature solution is also easy to construct for PFT.

1149: One simply starts with the smeared non-extremal D0-brane solution instead of (\ref{IIA}) which has the following form:

1150: %

1151: %

1152: \beq ds^2 &=& -\fThermal h^{-1/2} dt^2 

1153:  + h^{1/2} \sum_{i=1}^3 dx_i^2

1154:  + h^{1/2} \fThermal^{-1} d \rho^2 \cr

1155: && \qquad 

1156: + \rho^2 \bigl(d \anglepsi^2 + \sin^2\anglepsi\,d\angleY^2 

1157: + \cos^2\anglepsi \lbrack 

1158: \dsBase{2}^2 + (d\fiberphi + \connHopf)^2

1159: \rbrack\bigl) 

1160: \cr

1161: A & = & h^{-1} dt \cr

1162: e^{2 \dilaton} &=& h^{3/2}  \eeq

1163: %

1164: where

1165: %

1166: \be \fThermal\equiv 1 - {\rho_0^4 \over \rho^4}\ee  

1167: %

1168: is the ``thermal factor,''

1169: and \textit{$\dsBase{2}^2$} and $\connHopf$ were defined 

1170: in (\ref{eqn:dsBase}) and (\ref{eqn:defconnHopf}), respectively.

1171: Applying the same set of transformations, we arrive at a solution

1172: %

1173: \beq {ds^2\over \alpha'} & = & 

1174: \rescaledAugH^{1/2} \left( -\fThermal\rescaledH^{-1} dt^2

1175: + \fThermal^{-1}d\radialV^2  

1176: + \radialV^2

1177: \bigl(\cos^2\anglepsi\,\dsBase{2}^2 

1178: +\sin^2\anglepsi\,d\angleY^2 +d\anglepsi^2\bigr) \right) 

1179: \cr && \qquad

1180: +\rescaledAugH^{-1/2} \left(

1181: \sum_i dx_i^2 

1182: + \rescaledH\radialV^2\cos^2\anglepsi\,

1183: (d \fiberphi + \connHopf + \Delta^3\rescaledH^{-1} dt)^2

1184: \right)\ ,

1185: \cr

1186: {A \over \alpha'^2} & = &  

1187: \rescaledAugH^{-1}

1188: \left( -  dt + \radialV^2 \Delta^3 \cos^2\anglepsi\,

1189: (d \fiberphi + {\cal A}) \right) 

1190: \wedge dx_1 \wedge dx_2 \wedge dx_3\ ,

1191: \cr

1192: e^\dilaton & = & \gIIB = 2 \pi \gYM^2\ , \label{eqn:ThermalSUGRA}

1193: \eeq

1194: %

1195: where

1196: %

1197: \be \fThermal = 1 - {\radialV_0^4 \over \radialV^4}\ , 

1198: \ee

1199: and

1200: \be

1201: \rescaledH = \frac{4\pi\gIIB N}{\radialV^4}\ ,

1202: \qquad

1203: \rescaledAugH = 

1204:   \rescaledH + \cos^2\anglepsi\,\Delta^6\radialV^2\ ,

1205: \ee

1206: in accordance with (\ref{eqn:defrescaledHAugH}).

1207: Here \textit{$\radialV_0$} is a free parameter,

1208: and the background (\ref{eqn:ThermalSUGRA}) reduces

1209: to (\ref{eqn:dualSG}) for \textit{$\radialV_0=0$}.

1210: %

1211: Similar constructions of non-extremal solutions in

1212: asymptotically non-trivial geometries have also appeared in

1213: \cite{Gimon:2003xk}.

1214: 

1215: The $\radialV=\radialV_0$

1216: hypersurface corresponds to the horizon in this geometry.

1217: In order to extract the thermodynamic behavior of entropy

1218: $S(T)$ as a function of temperature,

1219: it is useful to first determine the temperature $T$ 

1220: and the horizon area $\Area$ as a function of the horizon

1221: radius \textit{$\radialV_0$}.

1222: As usual, the Hawking temperature can be inferred from 

1223: the condition that the Euclidean continuation of this solution

1224: be singularity-free. This gives

1225: %

1226: \be \radialV_0 = 2\pi\sqrt{\pi\lambda} T\ ,

1227: \label{eqn:UT}

1228: \ee

1229: %

1230: which is also one of the standard derivations of the 

1231: energy-distance relation (\ref{uvir}).

1232: 

1233: In order to apply the Beckenstein-Hawking formula, we need

1234: the area of the horizon in the Einstein frame.

1235: In string frame, we need 

1236: %

1237: \be A = {1 \over \alpha'^4} e^{-2 \dilaton} \sqrt{g_s} \ , \ee

1238: %

1239: where $g_s$ is the determinant of the induced metric

1240: on the horizon.

1241: This formula, applied to  (\ref{eqn:ThermalSUGRA}), gives 

1242: %

1243: \be S = {1 \over \gIIB^2} 

1244: \sqrt{ \gIIB N} (2\pi)^3\Volume \radialV_0^3 = 

1245: N^2 (2\pi)^3\Volume T^3 , \label{entropy}\ee

1246: %

1247: where \textit{$\Volume =  R_1  R_2  R_3$}. This is our main

1248: conclusion concerning the entropy. 

1249: Note that the final expression is

1250: independent of the ``puffness'' $\Delta$. It should be emphasized, however, that this result is reliable only

1251: in the range of temperatures

1252: %

1253: \be {\lambda^{-1/4} \over  R_i} \ll T \ll {\lambda^{-1/2}  R_i^2  \over \Delta^3} \ , \ee

1254: %

1255: which does depend on the puffness.  This is similar to what was found in non-commutative gauge theory

1256: \cite{si99,Maldacena:1999mh,Hashimoto:1999yj}. 

1257: We will comment further

1258: on the implication of the range of validity in 

1259: section~\ref{sec:Renormalization}.

1260: 

1261: The metric (\ref{eqn:ThermalSUGRA}) also contains

1262: information about the chemical potential conjugate

1263: to R-charge.

1264: Like PFT itself,

1265: the metric  preserves an

1266: $SU(2)\times U(1)$ subgroup of the R-symmetry group $SU(4).$

1267: The $U(1)$ component is generated by $\partial/\partial\fiberphi$,

1268: and the chemical potential conjugate to the corresponding

1269: R-charge can easily be read-off from

1270: the solution (\ref{eqn:ThermalSUGRA}).

1271: It is the angular velocity of the event-horizon:

1272: %

1273: \be

1274: \mu_R \equiv \Omega_{\text{horizon}} = 

1275: \Delta^3 \rescaledH^{-1}\Bigr\rvert_{\text{horizon}} = 

1276: \frac{\eta {\alpha'}^2}{4\pi \gIIB N}\radialV_0^4 

1277:  = 4 \pi^5 \lambda \Delta^3 T^4 \ . 

1278: \ee

1279: The chemical potential depends on the puffness $\Delta$,

1280: and it would be interesting to find the holographic dual

1281: for zero chemical potential, for which

1282: the entropy might also depend on $\Delta.$

1283: We hope to report on this in a separate paper.

1284: 

1285: % ==================================================

1286: \section{Renormalization Group Flow and Hierarchy of PFT}

1287: \label{sec:Renormalization}

1288: 

1289: The supergravity solution (\ref{eqn:dualSG})

1290: is reliable in the range (\ref{eqn:Urange}).

1291: It is natural to contemplate what alternative

1292: description takes over as a reliable description outside this range.

1293: Precisely such an issue, in the context of 

1294: non-commutative gauge theory,

1295: was investigated in \cite{Hashimoto:1999yj}. 

1296: We will see below that

1297: the PFT case is quite similar.

1298: 

1299: The infra-red boundary

1300: of the region of validity (\ref{eqn:Urange})

1301: has a simple interpretation: 

1302: at sufficiently low energies the higher

1303: dimensional operators deforming the theory become irrelevant, 

1304: and one simply undergoes dimensional 

1305: reduction below the scale of the size

1306: of the compactification. Let us assume for simplicity that 

1307: \textit{$R_1$}, \textit{$R_2$} and \textit{$R_3$}

1308: are of the same order of magnitude.

1309: {}From the string theory dual point of view,

1310: T-duality along the compact directions maps the D3-branes

1311: to D0-branes.

1312: One expects the Gregory-Laflamme instability to localize the

1313: smeared supergravity solution, simply giving rise to the near horizon

1314: geometry of the D0-branes as the effective description beyond the

1315: region of validity (\ref{eqn:Urange}) on the IR side

1316: \cite{Itzhaki:1998dd,Brandhuber:1998er}.

1317: 

1318: The proper size of the compact direction also becomes

1319: sub-stringy at the other end of the region of validity

1320: (\ref{eqn:Urange}), i.e., at the upper bound on $\rescaledU.$ 

1321: One does not expect the same three T-dualities to transform this

1322: background to a description which is effective. 

1323: The lesson from non-commutative gauge theories and NCOS

1324: prompts us to look for more complicated U-duality transformations

1325: that can make that region of the background weakly coupled.

1326: In the case of

1327: non-commutative gauge theories and NCOS, 

1328: the appropriate transformations

1329: are elements of $SL(2,Z)$, namely T- and S-duality transformations,

1330: respectively \cite{Hashimoto:1999yj,Chan:2001gs}. 

1331: For PFT, we propose the following $SL(2,Z)$ transformation:

1332: %

1333: \begin{enumerate}

1334: \item First,  T-dualize along $x_1$, $x_2$, and $x_3$ and lift to M-theory. This brings us back to (\ref{Mtheory}) where $z$ is a periodic coordinate with radius

1335: %

1336: \be R = \gIIA \lst\ . 

1337: \label{eqn:defR}\ee

1338: %

1339: \item Now, perform a coordinate transformation

1340: %

1341: \be \left(\begin{array}{c} d \fiberphi  \\ {d z \over  R} \end{array} \right) 

1342: \rightarrow \left( \begin{array}{cc} \sty{a} & \sty{b} \\ \sty{c} & \sty{d} \end{array}\right)

1343: \left(\begin{array}{c} d \fiberphi  \\ {d z \over  R} \end{array} \right),

1344: \qquad

1345: \left( \begin{array}{cc} \sty{a} & \sty{b} \\ \sty{c} & \sty{d} \end{array}\right)

1346: \in SL(2,Z)\ ,

1347: \ee

1348: %

1349: which makes the metric take the form

1350: %

1351: \beq ds^2 &=&  - h^{-1} dt^2  

1352: +   h (\sty{d}\,d  z + \sty{c}  R d \fiberphi - h^{-1} dt)^2  

1353: +\sum_{i=1}^3 dx_i^2 + d r^2 

1354: + r^2  \dsBase{2}^2 \cr

1355: &&   +  r^2\left( {(  \sty{d} \eta  R  + \sty{b})d  z \over  R} 

1356: + {(\sty{a} + \sty{c}   \eta  R) d \fiberphi 

1357: + \connHopf}\right)^2

1358:  +\sum_{i=8}^9 d  y_i^2\ .

1359: \eeq

1360: 

1361: \item Reduce to IIA along the $z$ direction. There are several subtleties in performing this step.  

1362: At this point it is convenient to set

1363: %

1364: \be z = {1 \over \sty{d}} \tz\ , \ee

1365: %

1366: where $\tz$ has the periodicity

1367: $\tz \sim \tz + 2 \pi \sty{d} R$, and

1368: %

1369: \beq ds^2 &=&  - h^{-1} dt^2  

1370: +   h (d  \tz + c  R d \fiberphi - h^{-1} dt)^2  

1371: +\sum_{i=1}^3 dx_i^2 + d r^2 + r^2  \dsBase{2}^2 \cr

1372: &&   +  r^2\left( {( \eta  R  + {\sty{b} \over \sty{d}} )d  \tz \over  R} + {(\sty{a} + \sty{c}  \eta  R ) d \fiberphi 

1373: + \connHopf}\right)^2

1374:  +\sum_{i=8}^9 d  y_i^2\ .

1375: \eeq

1376: %

1377: The IIA solution after the reduction then has the form

1378: %

1379: \beq ds^2 & = & \sqrt{

1380: h + {(\sty{b}+ \sty{d} \eta R)^2 r^2 \over \sty{d}^2 R^2}} \Bigl( - h^{-1} dt^2+

1381: \sum_i dx_i^2 + dr^2  + r^2\dsBase{2}^2  \cr

1382: && + 

1383:  \frac{r^2 ( \sty{d} h R \, \connHopf + h R \, d \fiberphi +

1384:   \left( \sty{b} + \sty{d} \eta R \right) dt

1385: )^2}{h \left( \sty{d}^2 h R^2 + r^2(\sty{b} + \sty{d} \eta R)^2 \right) }

1386:  +\sum_{i=8,9} dy_i^2 \Bigr)\ , \cr

1387: A & = &  \sty{c} R d \fiberphi + {R (\sty{b}+\sty{d} \eta R) r^2 (d \fiberphi + \sty{d} \connHopf) - \sty{d}^2 R dt \over \sty{d}^2 R^2 h+ (\sty{b} + \sty{d} \eta R)^2 r^2} \cr

1388: e^\dilaton & = & \tgIIA \left(h + {(\sty{b}+ \sty{d} \eta R)^2 r^2 \over \sty{d}^2 R^2}\right)^{3/4}\ ,

1389: \eeq

1390: %

1391: for which the string coupling constant and the tension change to

1392: %

1393: \be \tgIIA = \sty{d}^{3/2} \gIIA, \qquad 

1394: \tilde \alpha' = {1 \over \sty{d}} \alpha'\ , 

1395: \label{eqn:tgta}\ee

1396: %

1397: because the choice of M-theory circle is different.

1398: 

1399: \item T-dualize along $x_1$, $x_2$, and $x_3$. This brings the background to the form

1400: %

1401: \beq ds^2 & = & \sqrt{

1402: h + {(\sty{b}+ \sty{d} \eta R)^2 r^2 \over \sty{d}^2 R^2}} 

1403: \Bigl( - h^{-1} dt^2+

1404: {1 \over

1405: h + {(\sty{b}+ \sty{d} \eta R)^2 r^2 \over \sty{d}^2 R^2}}

1406: \sum_i dx_i^2 + dr^2  + r^2\dsBase{2}^2  \cr

1407: && + 

1408:  \frac{r^2 ( \sty{d} h R \, \connHopf + h R \, d \fiberphi +

1409:   \left( \sty{b} + \sty{d} \eta R \right) dt

1410: )^2}{h \left( 

1411: \sty{d}^2 h R^2 + r^2(\sty{b} + \sty{d} \eta R)^2

1412: \right) }

1413:  +\sum_{i=8,9} dy_i^2 \Bigr)\ , \cr

1414: A & = &  \left(\sty{c} R d \fiberphi + {R (\sty{b}+\sty{d} \eta R) r^2 (d \fiberphi + \sty{d}\connHopf) - \sty{d}^2 R dt \over \sty{d}^2 R^2 h+ (\sty{b} + \sty{d} \eta R)^2 r^2}\right) \wedge dx_1 \wedge dx_2 \wedge dx_3\ , \cr

1415: e^\dilaton & = & \tgIIB\ .

1416: \eeq

1417: %

1418: The radii of the $x_i$ coordinates are

1419: %

1420: \be \tilde R_i = {\tilde \alpha'  \over \alpha'} R_i 

1421: =  {R_i \over \sty{d}}\ .  \ee

1422: %

1423: We also find 

1424: %

1425: \be \tgIIB = \tgIIA \tVolume \tilde\alpha'^{-3/2} =  

1426: \gIIB \  , \qquad 

1427: \tVolume = \tilde R_1 \tilde R_2 \tilde R_3\ . 

1428: \label{eqn:deftgIIB}\ee

1429: %

1430: In terms of $\tilde\alpha'$ and $\tgIIB=\gIIB$,

1431: %

1432: \be h = 1 + {4 \pi \gIIB N \sty{d}^2 \tilde\alpha'^2 \over r^4}\ ,

1433: \ee

1434: %

1435: indicating that the number of D3-branes has become $\sty{d}^2 N$. However, we see that near $r=0$, the $d\fiberphi^2$ component of the metric has the form

1436: %

1437: \be {1 \over \sty{d}^2} r^2 (\sqrt{h}) d \fiberphi^2 \ee

1438: %

1439: indicating that there is a  $Z_\sty{d}$ orbifold singularity. 

1440: The total D3-brane charge is therefore $\sty{d}N$.

1441: 

1442: \item Now, we take the decoupling limit sending 

1443: $\tilde \alpha'\rightarrow 0$ keeping 

1444: $\dualizedRescaledU \equiv r/\tilde \alpha'$, 

1445: $\tgIIB = \sty{d} R \tVolume /\tilde \alpha'^2$

1446: [using

1447: (\ref{eqn:defR}),(\ref{eqn:tgta}) and (\ref{eqn:deftgIIB})],

1448: and $\chi = \eta R.$

1449: This brings the SUGRA solution to the form

1450: %

1451: \beq {ds^2\over \alpha'} & = & 

1452: \dualizedRescaledAugH^{1/2} \left( 

1453: - \dualizedRescaledH^{-1} dt^2

1454: + d\dualizedRescaledU^2  + \dualizedRescaledU^2 \dsBase{2}^2 

1455: +\sum_{i=8,9} d\rescaledY_i^2 \right) 

1456: \cr && \qquad

1457: +\dualizedRescaledAugH^{-1/2} \left(

1458: \sum_i dx_i^2 

1459: + \dualizedRescaledH\dualizedRescaledU^2

1460: ({d \fiberphi \over \sty{d}}

1461:  + \connHopf + \Delta^3\rescaledH^{-1} dt)^2

1462: \right)\ ,

1463: \cr

1464: {A \over \alpha'^2} & = &  

1465: \left\{ {\sty{c} \over \sty{d}} \tgIIB\tVolume^{-1} d \fiberphi 

1466: +\dualizedRescaledAugH^{-1}

1467: \left( -  dt + \dualizedRescaledU^2 \dualizedDelta^3 

1468: ({d \fiberphi \over \sty{d}} + {\cal A}) \right)\right\} 

1469: \wedge dx_1 \wedge dx_2 \wedge dx_3\ ,

1470: \cr

1471: e^\dilaton & = & \tgIIB\ ,  \label{sl2zsol}

1472: \eeq

1473: %

1474: where

1475: \be \dualizedDelta^3 \equiv 

1476: {(\sty{b}+ \sty{d} \chi) \tVolume \over \tgIIB}\ ,

1477: \qquad

1478: \dualizedRescaledH \equiv

1479: {4 \pi \tgIIB N \over 

1480: (\dualizedRescaledU^2 + \|\rescaledY\|^2)^2}\ ,

1481: \qquad

1482: \dualizedRescaledAugH \equiv 

1483: \dualizedRescaledH + \dualizedRescaledU^2\dualizedDelta^6\ .

1484: \ee

1485: \end{enumerate}

1486: %

1487: This is the SUGRA dual of the $SL(2,Z)$ transform of PFT.  It has the same form as the SUGRA dual of PFT (\ref{eqn:dualSG}),

1488: except that the $\fiberphi$ coordinate has a deficit angle, 

1489: and there is an extra constant term in the RR 4-form potential

1490: (which cannot be gauged away because the $x_1, x_2, x_3$

1491: and $\fiberphi$ directions are compact).

1492: 

1493: The $SL(2,Z)$ transformation also acts non-trivially on the D3-brane charge, the volume of the torus, and the puffness. Specifically

1494: %

1495: \be N \rightarrow \tilde N = \sty{d}^2 N, \qquad 

1496: R_i \rightarrow

1497: \tilde R_i = R_i/\sty{d}, \qquad 

1498: \chi \rightarrow \tilde \chi =

1499: \sty{b} + \sty{d} \chi\ , \ee

1500: %

1501: which is the analogue of the Morita transformation formula 

1502: of NCYM \cite{Hashimoto:1999yj,Pioline:1999xg}.

1503: Note that the UV/IR relation

1504: %

1505: \be E = {\dualizedRescaledU \over \sqrt{\tilde \lambda}} 

1506:    = {\rescaledU \over \sqrt{\lambda}} \ee

1507: %

1508: gives rise to a consistent holographic embedding. The constant part of

1509: the RR 4-form in PFT is the analogue of the ``$\Phi$ parameter'' 

1510: in NCYM.

1511: 

1512: We can immediately infer the range of validity of this solution 

1513: %

1514: \be {\tilde \lambda^{-1/4} \over  \tilde R_i} \ll E \ll {\tilde \lambda^{-1/2}  \tilde R_i^2  \over \dualizedDelta^3}  = 

1515: {\tgIIB \tilde \lambda^{-1/2} \over  \tilde\chi} 

1516: {1 \over \tilde R_i}  \ . \label{Erange2} \ee

1517: %

1518: Computing the entropy from the near extremal generalization of (\ref{sl2zsol})

1519: yields

1520: %

1521: \be S = {1 \over \sty{d}} (\sty{d}^2 N)^2 (2\pi)^3\tVolume T^3 

1522: = (2\pi)^3 N^2\Volume T^3 \ . \ee

1523: %

1524: In other words, the functional form of the entropy formula

1525: (\ref{entropy}) appears to extend beyond its naive range of

1526: applicability, as long as there is some dual description in terms 

1527: of one of the $SL(2,Z)$ duals listed in (\ref{sl2zsol}).

1528: 

1529: If the value of $\chi$ of PFT we start with is rational, say,

1530: %

1531: \be \chi = {\sty{r} \over \sty{s}} \ee

1532: %

1533: then there exists an element of $SL(2,Z),$

1534: \be \left(\begin{array}{cc}\sty{a} & \sty{b} \\\sty{c} & \sty{d}\end{array}\right) = 

1535:  \left(\begin{array}{cc}\sty{p} & -\sty{r} \\-\sty{q} & \sty{s}\end{array}\right)\ ,  \ee

1536: %

1537: for which $\dualizedDelta = 0,$ and the solution becomes

1538: %

1539: \beq {ds^2\over \tilde \alpha'} & = & 

1540: - \dualizedRescaledH^{-1/2} dt^2

1541: + \dualizedRescaledH^{-1/2} \sum_i dx_i^2 

1542: + \dualizedRescaledH^{1/2} \left( d\dualizedRescaledU^2 

1543: + \dualizedRescaledU^2 \dsBase{2}^2  + 

1544: \dualizedRescaledU^2

1545: \left({d \fiberphi \over \sty{s}}+  \connHopf \right)^2\right)

1546:  \cr

1547: {A \over \tilde \alpha'^2}  & = &  \left( -{ \sty{q} \over \sty{s}} \tgIIB\tVolume^{-1} d \fiberphi 

1548: - \dualizedRescaledH^{-1}  dt \right)

1549:    \wedge dx_1 \wedge dx_2 \wedge dx_3 \cr

1550: e^\dilaton & = & \tgIIB  \label{localsg}

1551: \eeq

1552: %

1553: which is essentially $AdS_5 \times S_5 / Z_\sty{s}$ with constant RR 4-form potential.

1554: The fact that this solution is  anti de-Sitter 

1555: provides further evidence that gravity is decoupled from the

1556: dynamics of PFT.

1557: 

1558: The region of validity of (\ref{localsg}) is

1559: %

1560: \be {\tilde \lambda^{-1/4} \over \tilde R} < E  

1561: \label{Erange3} \ . \ee

1562: %

1563: Provided $\chi$ is chosen such that

1564: %

1565: \be 1 \ll {\sqrt{\sty{s}} \lambda^{1/4} \over \gIIB \chi}\ , \ee

1566: %

1567: which is easy to arrange since we assumed $\lambda \gg 1$ and 

1568: $\gIIB \ll 1$,

1569:  the upper bound of (\ref{Erange2}) is smaller than the lower bound of (\ref{Erange3}):

1570: %

1571: \be {\lambda^{-1/2} R^2 \over \Delta^3} \ll {\tilde \lambda^{-1/4} \over \tilde R}\ . \ee

1572: %

1573: As the supergravity dual (\ref{localsg}) does not have an upper bound on its region of applicability (\ref{Erange2}), 

1574: one can conclude that any PFT with rational value of $\chi$ 

1575: is described in terms of it.

1576: Since rational $\chi$'s form a dense subset of the set of real values of $\chi$, 

1577: we conclude that for arbitrary values of $\chi$, the entropy

1578: formula (\ref{entropy}) is valid for all energies,

1579: %

1580: \be {\lambda^{-1/4} \over R_i} < E \ , \label{Erange4}\ee

1581: %

1582: assuming that the entropy is a continuous function of $\chi.$

1583: The specific $SL(2,Z)$ element which gives the most effective

1584: description at a given energy $E$ in the range (\ref{Erange4}) does,

1585: however, depend sensitively on the rationality of $\chi$.  

1586: To determine which $SL(2,Z)$ is most effective, one looks for 

1587: a pair $(\sty{b},\sty{d})$ that maximizes the proper size

1588: of the $x_i$ circle, or equivalently the expression

1589: %

1590: \be \VExpression(E) \equiv

1591: {\tilde R^4 \over \dualizedRescaledH^2 

1592:   + \dualizedDelta^6\dualizedRescaledU^2} 

1593: =

1594: {\gIIB^2  R^2  \lambda^2 E^4 \over 4 \pi \sty{d}^2 \gIIB^2 

1595:  +  (\sty{b}+ \sty{d}\chi)^2  \lambda^2 R^6 E^6} \ . 

1596: \label{volume} 

1597: \ee

1598: %

1599: For example,

1600: take $\gIIB = 1/3$, $\lambda = 9$, and $\chi = 2/1023.$

1601: We find that $(\sty{b},\sty{d}) = (0,1)$, 

1602: $(\sty{b},\sty{d}) = (-1,511)$, 

1603: and $(\sty{b},\sty{d}) = (-2,1023)$ give rise to a 

1604: $\VExpression(E)$ that is illustrated in

1605: figure \ref{figa}. 

1606: Similar structures were encountered in the case of

1607: non-commutative gauge theory \cite{Hashimoto:1999yj} and NCOS

1608: \cite{Chan:2001gs}, where a self-similar structure,

1609: closely related to

1610: the continued fraction expansion for the appropriate counterparts 

1611: of the dimensionless non-locality parameter $\chi$,

1612: characterizes the phase diagram.

1613: 

1614: \begin{figure}

1615: \centerline{\includegraphics[width=4in]{cascade}}

1616: \caption{Log-log plot of  (\ref{volume})  for $(\sty{b},\sty{d}) = (0,1)$, $(\sty{b},\sty{d}) = (-1,511)$, and $(\sty{b},\sty{d}) =

1617: (-2,1023)$ for a PFT with parameters $2 \pi \gYM^2 = \gIIB = 1/3$, $\lambda = 9$, and $\chi =

1618: 2/1023$.

1619: \label{figa}}

1620: \end{figure}

1621: 

1622: Here, the fact that the range of validity (\ref{Erange2}) depends on

1623: two large dimensionless parameters $\tilde\lambda$ and $1/\tgIIB$, in addition to $\chi$, makes the full phase structure

1624: somewhat more cumbersome to determine.  For example, 

1625: for $\dualizedDelta\ll\tgIIB^{1/3} {\tilde\lambda}^{-1/6} R$

1626: [cf. (\ref{eqn:ConditionDelta}),

1627: which is stronger than (\ref{eqn:assump})

1628: for $\lambda^{1/4}\gg\gIIB$],

1629: we need to go to the M-theory description as we explained 

1630: at the end of section~\ref{sec:SupergravityDual}.

1631: This description

1632: does not drastically alter the form of the entropy 

1633: as a function of temperature, because

1634: the entropy formula (\ref{entropy}),

1635: which is based on the area of the horizon 

1636: in Einstein frame, is generally 

1637: unaffected by T-dualities and by the M-theory lift.

1638: Strictly speaking, we have not ruled out the possibility

1639: of some exotic thermodynamic behavior in the range of energies for

1640: which the supergravity description is not effective, along the lines

1641: of what was observed in \cite{Chan:2001gs}. Nonetheless, one expects

1642: some specific dual description to be effective for any range of

1643: parameters and energies. In these duality cascades, the fact that

1644: there is a PFT in the far IR and (\ref{localsg}) in the far UV

1645: for any rational value of $\chi$ appears to be a robust feature.

1646: 

1647: This also highlights the point that the decompactification 

1648: limit $R_i\rightarrow \infty$ is a tricky limit to take even if one concentrates on the UV.

1649: This is because making $R_i$ large while keeping $\Delta$ fixed

1650: changes the rationality of $\chi$ in a chaotic way. 

1651: While physical observables, such as the entropy, 

1652: have a smooth limit, the phase

1653: structure in the UV region evolves erratically. Such interference

1654: between flows to the UV and decompactification is a typical feature of non-local field theories 

1655: \cite{Hashimoto:1999yj,Chan:2001gs}.

1656: 

1657: 

1658: % ==================================================

1659: \section{Deformation Operator of Lowest Dimension}

1660: \label{sec:Deformation}

1661: 

1662: In the limit $\rescaledU\rightarrow 0$

1663: the supergravity dual (\ref{eqn:dualSG}) becomes 

1664: $AdS_5\times S^5$,

1665: which corresponds to PFT flowing to \SUSY{4} SYM in the IR.

1666: The supergravity dual (\ref{eqn:dualSG})

1667: can also be used to read off the

1668: lowest dimension operator responsible 

1669: for deforming the \SUSY{4} theory. 

1670: We see in (\ref{eqn:dualSG}) that a linear combination of the

1671: metric and the RR 4-form potential,  polarized partly along the brane and partly transverse to the brane, are deformed. 

1672: The deformation of $AdS_5\times S^5$

1673: that (\ref{eqn:dualSG}) describes

1674: has been arranged to preserve half of the supersymmetries, 

1675: and therefore the corresponding operator

1676: has to be a descendant of a chiral primary operator.

1677: 

1678: We denote the \SUSY{4} SYM gauge field strength by

1679: $F_{\mu\nu}$, the scalars by $X^I$ ($I=4,\dots,9$ for convenience),

1680: and the spinors by $\lambda$ and $\bar{\lambda}$.

1681: (We will not need to specify the indices on the spinors.)

1682: The descendents of chiral primary operators of \SUSY{4} SYM

1683: are listed in table~7 of \cite{D'Hoker:2002aw}.

1684: In their notation,

1685: our requisite descendant takes the form

1686: %

1687: \be 

1688: \Op{17}{k} \sim

1689: \tr F_+ F_- \lambda \bar \lambda X^k \ee

1690: %

1691: for $k=0.$

1692: The schematic notation here is as follows:

1693: $F_{+}$ ($F_{-}$) stands for the self-dual (anti-self-dual)

1694: part of the field-strength, $X^k$ stands for

1695: a product of $k$ scalar fields, there is

1696: an unspecified index contraction, and terms involving derivatives

1697: and commutators have been suppressed.

1698: $\Op{17}{k=0}$

1699: is an operator of dimension 7 in the representation

1700: ${\bf 15}$ of the

1701: $SO(6)$ $R$-symmetry group. This $SO(6)$ multiplet accounts for

1702: distinct ways in which the space $\Reals^4$, which we twist, can be

1703: embedded into the $\Reals^6$ space transverse to the D3-brane.

1704: 

1705: Let us note, in contrast, that the leading irrelevant operators

1706: that deform \SUSY{4} SYM into SYM on a noncommutative

1707: $\Reals^4$ (NCYM)

1708: and the noncommutative open string theory (NCOS) are,

1709: respectively,  

1710: the real and the imaginary parts of the dimension 6 operator 

1711: %

1712: \be 

1713: \Op{16}{k=0}\sim

1714: \tr F_+ F_-^2 X^{k=0}\ ,  \ee

1715: %

1716: whereas the dipole deformation and its S-dual

1717: are generated by the real and imaginary parts of the

1718: dimension 5 operator

1719: %

1720: \be

1721: \Op{10}{k=0}\sim

1722: \tr F_+ \lambda \bar \lambda X^{k=0} \ . \ee

1723: %

1724: These dimensions fit well with the fact that the parameters

1725: characterizing the dipole, the non-commutative, and the 

1726: puff field theories have dimensions 1, 2, and 3, respectively.

1727: 

1728: PFT can also be defined as the decoupled field-theory

1729: that describes $N$ D3-brane probes in

1730: the strongly-coupled type-IIB background

1731: obtained from (\ref{eqn:FullSolution}) by setting $N=0.$

1732: This is the Melvin background that can be written as

1733: \beq ds^2 & = & 

1734: (1+\eta^2\fullu^2)^{1/2} \left( - dt^2

1735: + d\fullu^2  + \fullu^2 \dsBase{2}^2 

1736: +\sum_{i=8,9} dy_i^2 \right) 

1737: \cr && \qquad

1738: +(1+\eta^2\fullu^2)^{-1/2} \left(

1739: \sum_i dx_i^2 

1740: + \fullu^2(d \fiberphi + \connHopf + \eta dt)^2

1741: \right)\ ,

1742: \cr

1743: A & = &  \frac{1}{1+\eta^2\fullu^2}

1744: \left( -  dt + \fullu^2 \eta (d \fiberphi+\connHopf) \right) 

1745: \wedge dx_1 \wedge dx_2 \wedge dx_3\ ,

1746: \cr

1747: e^\dilaton & = & \gIIB\ .

1748: \label{eqn:FullSolutionZ}

1749: \eeq

1750: It is strongly coupled in the limit $\alpha'\rightarrow 0$

1751: keeping (\ref{eqn:defDelta}).

1752: The operator $\Op{17}{k=0}$ can be interpreted as follows.

1753: Expand (\ref{eqn:FullSolutionZ}) formally in powers of $\eta$,

1754: and keep only terms up to order $O(\eta)$.

1755: Using the notation

1756: $$

1757: \angularForm\equiv

1758: \fullu^2(d\fiberphi + \connHopf) 

1759: = y_4 dy_5 - y_5 dy_4 + y_6 dy_7 - y_7 dy_6\ ,

1760: $$

1761: we can write (\ref{eqn:FullSolutionZ}) as

1762: \beq ds^2 & = & 

1763: -dt^2 + \sum_i dx_i^2 

1764: + \sum_{i=4}^9 dy_i^2+2\eta\omega dt + O(\eta^2)\ ,

1765: \cr

1766: A_4^{\text{\tiny (full)}} & = &

1767: (-dt +\eta\omega) 

1768: \wedge dx_1 \wedge dx_2 \wedge dx_3

1769: +\eta\omega\wedge dt\wedge dy_8\wedge dy_9 + O(\eta^2)\ ,

1770: \cr

1771: e^\dilaton & = & \gIIB\ ,

1772: \label{eqn:SpecialMelvinOrderEta}

1773: \eeq

1774: where we have completed the RR 4-form so that

1775: $dA_4^{\text{\tiny (full)}}$ is self-dual.

1776: The bosonic part of $\Op{17}{k=0}$ can now be deduced

1777: from the Dirac-Born-Infeld (DBI) action and

1778: the Wess-Zumino (WZ) term,

1779: $$

1780: {\cal S}_{\text{\tiny DBI+WZ}} = 

1781: \frac{1}{{\alpha'}^2\gIIB}\int_{D3}\left(

1782: \sqrt{-\det{G + \alpha' F}}\,

1783: +A_4^{\text{\tiny (full)}}\right)\ ,

1784: $$

1785: where the induced metric $G$ is given by

1786: $$

1787: G_{\mu\nu} = \eta_{\mu\nu} 

1788: +{\alpha'}^2 \sum_{i=4}^9 \partial_\mu X^i\partial_\nu X^i

1789: +\eta {\alpha'}^2 (\delta_{\mu 0} \RCurrent_\nu 

1790: + \delta_{\nu 0}\RCurrent_\mu) + O(\eta^2)\ ,

1791: $$

1792: $\RCurrent_\mu$ is the R-current:

1793: $$

1794: \RCurrent_\mu\equiv

1795:  X^4\partial_\mu X^5 - X^5\partial_\mu X^4

1796: +X^6\partial_\mu X^7 - X^7\partial_\mu X^6\ ,

1797: $$

1798: and we used the standard relation 

1799: $y_i = \alpha' X^i$ ($i=4\dots 9$) between

1800: the transverse coordinates of the D3-brane and the

1801: scalar fields of the effective field theory on the brane.

1802: 

1803: Expanding the DBI action to order $O(\eta)$ we find, for $N=1$,

1804: \beq

1805: \Op{17}{k=0} &=& T^{0\mu}\RCurrent_\mu

1806: +\epsilon^{0\mu\nu\sigma}

1807: \partial_\mu X^8\partial_\nu X^9\RCurrent_\sigma

1808: +\text{fermions}\ ,

1809: \label{eqn:Op17}

1810: \eeq

1811: where

1812: $$

1813: T^{\mu\nu} = \sum_{i=4}^9\partial^\mu X^i\partial^\nu X^i

1814: -\frac{1}{2}\eta^{\mu\nu}\sum_{i=4}^9

1815:   \partial_\tau X^i\partial^\tau X^i

1816: +{F^\mu}_{\tau}F^{\tau\nu}

1817: +\frac{1}{4}\eta^{\mu\nu}F_{\sigma\tau}F^{\sigma\tau}

1818: +\text{fermions}\ ,

1819: $$

1820: is the stress-energy tensor.

1821: For $N>1$, (\ref{eqn:Op17}) is missing an overall trace

1822: and additional commutator terms.

1823: 

1824: 

1825: 

1826: % ==================================================

1827: \section{Concluding remarks}

1828: \label{sec:Concluding}

1829: 

1830: In this paper we inferred a number of basic features 

1831: of Puff Field Theory by analyzing its supergravity dual.

1832: In particular, we computed

1833: the thermodynamic entropy, studied its range of validity, and

1834: identified the leading irrelevant operator deforming the \SUSY{4}

1835: theory.

1836: These results lend more credence to the conjecture that PFT

1837: is decoupled for gravity. In fact, the mere existence of

1838: a (geodesically complete) near-horizon limit

1839: of the background (\ref{eqn:FullSolution}) implies decoupling.

1840: The finite entropy (\ref{entropy}) suggests that the

1841: spectrum is discrete (for appropriate boundary conditions

1842: that eliminate the zero modes of the low-energy scalar fields).

1843: Furthermore, we have seen that for rational $\chi$

1844: the supergravity dual can be transformed into

1845: an orbifold of $AdS_5\times S^5$ 

1846: with extra RR flux (\ref{localsg}), which certainly

1847: describes a decoupled theory.

1848: 

1849: It would, of course, be interesting if a microscopic definition 

1850: of PFT can be found. 

1851: Non-commutative Yang-Mills theory and dipole theories

1852: can be formulated in terms of a concrete action, and NCOS can be

1853: defined as a strong coupling limit of NCYM. It would be nice if PFT

1854: can be defined at the same level of specificity.

1855: 

1856: Lessons from NCYM and dipole theories suggest that a good 

1857: starting point might be to study

1858: PFT on $T^3$ with a rational parameter $\chi.$

1859: One approach might be to identify the field theory dual of

1860: (\ref{localsg}). This rather innocent looking supergravity solution

1861: contains a closed RR 4-form potential which, combined with the

1862: orbifold, is responsible for all the non-trivial IR physics. 

1863: We are currently investigating this issue and 

1864: we hope to report our findings in the near future.

1865: 

1866: PFT arose as the decoupling limit of D0-branes in a Melvin universe

1867: supported by an RR 1-form potential in the type IIA theory.

1868: It is also natural to consider what happens for

1869: other type-IIA D$p$-branes.

1870: 

1871: For the case of D2-branes, there are two possible choices of embedding:

1872: the twisted $\fiberphi$ coordinate could either

1873: be along or transverse to the D2-brane.

1874: If it is along the brane, one ends up with an NCOS, which is

1875: S-dual to the non-commutative gauge theory of

1876: \cite{Hashimoto:2004pb,Hashimoto:2005hy},

1877: dimensionally reduced to 2+1 dimensions.

1878: The supergravity dual of the NCOS (prior to the

1879: dimensional reduction) was discussed in 

1880: \cite{Huang:2005mi,Cai:2006td}.

1881: If the $\fiberphi$ direction is transverse to the brane,

1882: we end up with the S-dual of dipole theory.

1883: In both of these constructions, we are

1884: dimensionally reducing along the non-local direction from the

1885: NCYM/dipole point of view, but the non-locality of the S-dual

1886: survives dimensional reduction.

1887: 

1888: The case of D4-branes does not appear to have any interesting non-local field theory in the decoupling limit, because

1889: when  D4-branes are lifted to M-theory they 

1890: are extended along the M-theory circle.

1891: The case of NS5-branes appears to lead to a non-local 

1892: deformation of little string theory (LST),

1893: and neither D6-branes nor D8-branes support 

1894: any decoupled field theory, so we will not pursue them further.

1895: 

1896: We will elaborate on the details of the twisted decoupling of type

1897: IIA D2, D4, and NS5-branes in appendix \ref{appb}. As NCOS arising

1898: from 2+1 was already known, and LST arising from NS5 is already a

1899: non-local theory, the PFT based on D0-branes appears to be rather

1900: special in giving rise to a novel non-local deformation of a local

1901: field theory. 

1902: 

1903: % ==========================================================

1904: \section*{Acknowledgements}

1905: %

1906: We would like to thank

1907: %

1908: Lisa~Dyson, 

1909: Ian~Ellwood,

1910: Eric~Gimon, 

1911: Thomas~Grimm,

1912: Petr~Ho\v{r}ava,

1913: Christopher~Hull,

1914: Nissan~Itzhaki,

1915: Yasunori~Nomura

1916: and Jesse~Thaler

1917: %

1918: for discussions.

1919: %

1920: This work was supported in part by

1921: the Director, Office of Science, Office of High Energy and Nuclear

1922: Physics, of the U.S. Department of Energy under

1923: Contract DE-AC03-76SF00098 and under Contract

1924: DE-FG02-95ER40896, in part by

1925: the NSF under grant PHY-0098840,

1926: and in part by

1927: the Center of Theoretical Physics at UC Berkeley.

1928: OJG also wishes to thank the organizers

1929: of the conference ``M-theory in the City,'' which took place

1930: at Queen Mary University of London in November 2006,

1931: and AH thanks UC Berkeley, where this work was initiated,

1932: for their warm hospitality.

1933: 

1934: 

1935: 

1936: \appendix

1937: 

1938: \section*{Appendix}

1939: 

1940: 

1941: % ==================================================

1942: \section{Melvin twists of D2, D4, and NS5 branes\label{appb}}

1943: 

1944: In this article, we primarily focused on the decoupled 

1945: field theory on D0-branes embedded in a Kaluza-Klein 

1946: Melvin universe with the M-theory

1947: circle playing the role of the Kaluza-Klein circle.  Such a

1948: construction naturally extends to other branes 

1949: in type-IIA string theory.

1950: In this appendix, we elaborate on the cases of D2, D4, and NS5

1951: branes. In all of these cases, the appropriate scaling of the 

1952: Melvin flux can be inferred from requiring the 

1953: dimensionless parameter $\chi$ to be finite.

1954: 

1955: \subsection{D2-brane}

1956: 

1957: The decoupled theory on D2-branes turns out to be a known non-local

1958: field theory. In order to identify this field theory, let us analyze

1959: the supergravity dual explicitly.

1960: 

1961: Let us follow the construction of 

1962: section~\ref{sec:SupergravityDual}.

1963: Start with the

1964: supergravity solution of D2

1965: %

1966: \beq  ds^2 & = & h^{-1/2} (-dt^2 + \sum_{i=1}^2 dx_i^2) 

1967: +  h^{1/2} \sum_{i=3}^9 dx_i^2 \ ,\cr

1968: A & = & h^{-1} dt \wedge dx_1 \wedge dx_2\ , \cr

1969: e^{\dilaton} & = & \gIIA h^{1/4}\ , \cr

1970: h & = & 1 + {6 \pi^2 \gIIA N \alpha'^{5/2}\over r^5} \ . \eeq

1971: %

1972: Lifting  to M-theory gives

1973: %

1974: \beq  ds^2 & = & h^{-2/3} (-dt^2 + \sum_{i=1}^2 dx_i^2) 

1975: +  h^{1/3} \sum_{i=3}^9 dx_i^2 + h^{1/3} dz^2\ ,\cr

1976: A & = & h^{-1} dt \wedge dx_1 \wedge dx_2\ ,  \cr

1977: z & \sim & z + 2 \pi \gIIA \lst \ . \eeq

1978: %

1979: Twisting along the $(x_1,x_2)$ plane gives

1980: %

1981: \beq  ds^2 & = & h^{-2/3} (-dt^2 + d\rho^2 

1982: + \rho^2 (d \fiberphi+ \eta dz)^2) 

1983: +  h^{1/3} \sum_{i=3}^9 dx_i^2 + h^{1/3} dz^2\ ,\cr

1984: A & = & h^{-1} \rho\,  dt \wedge d\rho \wedge 

1985: ( d \fiberphi  + \eta dz)\ , \cr

1986: z & \sim & z + 2 \pi \gIIA \lst \ . \eeq

1987: %

1988: Now, reduce to IIA on $z$ to find

1989: %

1990: \beq  ds^2 & = & 

1991: \left(h + \eta^2 \rho^2 \over h\right)^{1/2} 

1992: \left\lbrack

1993: h^{-1/2} \left(-dt^2 + d\rho^2 

1994: + {h \rho^2 \over h + \eta^2 \rho^2} d \fiberphi^2\right) 

1995: +  h^{1/2} (dr^2 + r^2 d \Omega_6^2)\right\rbrack\ ,\cr

1996: A_1 & = & {\eta \rho^2 \over h + \eta^2 \rho^2} d \fiberphi\ , \cr

1997: A_3 & = & h^{-1} dt \wedge dx_1 \wedge dx_2\ ,  \cr

1998: B_2 & = & \eta h^{-1} r\,  dt \wedge dr\ , \cr

1999: e^{\dilaton} & = &  

2000: \gIIA h^{1/4}\left({h + \eta^2 \rho^2 \over h} \right)^{3/4} \ .

2001: \eeq

2002: %

2003: Finally, taking the $\alpha' \rightarrow 0$ decoupling limit, keeping $U = r/\alpha'$, $\gYMDim{2}^2 = \gIIA \lst^{-1}$

2004: (the YM coupling constant of the 2+1D theory) 

2005: and $\chi = \eta R$ fixed, gives

2006: %

2007: \beq  {ds^2 \over \alpha'} & = & 

2008: \left(1 + {\chi^2 \rho^2 \over \gYMDim{2}^4 H}\right)^{1/2}

2009: \left\lbrack H^{-1/2} 

2010: \left(-dt^2 + d\rho^2 + {\rho^2 d \fiberphi^2 \over 

2011: 1 + {\chi^2 \rho^2 \over \gYMDim{2}^4 H} }\right) 

2012: +  H^{1/2} (dU^2 + U^2 d \Omega_6^2)\right\rbrack\ ,\cr

2013: {A_1 \over \alpha'}  & = & 

2014: {\chi \gYMDim{2}^2 \rho^2  \over 

2015: \gYMDim{2}^4 H + \chi^2 \rho^2} d \fiberphi\ ,\cr

2016: {A_3 \over \alpha'^2} & = & H^{-1} dt \wedge dx_1 \wedge dx_2\ ,

2017: \cr

2018: {B_2 \over \alpha'}  & = & 

2019: {1 \over \gYMDim{2}^2}\chi H^{-1} \rho\,  dt \wedge d\rho\ , \cr

2020: e^{\dilaton} & = &  \gYMDim{2}^2  H^{1/4} 

2021: \left({\gYMDim{2}^4 H 

2022: + \chi^2 \rho^2 \over \gYMDim{2}^4 H} \right)^{3/4}\ , \cr

2023: H & = & {6 \pi^2 \gYMDim{2}^2 N \over U^5} \label{a5} \ .

2024: \eeq

2025: %

2026: This is a non-local deformation of a strongly coupled SYM with 16

2027: supercharges in 2+1 dimensions.

2028: 

2029: In order to bring this theory into context, it is useful to compactify $x_3$ on a circle of radius $\alpha'/R_3$ and smear, so that (\ref{a5}) becomes

2030: %

2031: \beq  {ds^2 \over \alpha'} & = & 

2032: \left(1 + {\chi^2 \rho^2 \over \gYMDim{2}^4 H}\right)^{1/2}

2033: \left\lbrack

2034: H^{-1/2} \left(-dt^2 + d\rho^2 

2035: + {\rho^2 d \fiberphi^2 \over 1 

2036: + {\chi^2 \rho^2 \over \gYMDim{2}^4 H} }\right)\right. 

2037: \cr &&  \qquad\qquad\qquad\qquad\left.

2038: % The following command is not part of the equation

2039: % but rather makes the \right\rbrack the same size

2040: % of \left\lbrack by letting LaTeX know what is the largest

2041: % construct size that it needs to match the parenthesis with: 

2042: \vphantom{\left({\rho^2 d \fiberphi^2 \over 1 %

2043: + {\chi^2 \rho^2 \over \gYMDim{2}^4 H}}\right)} %

2044: % END OF \vphantom command

2045: %

2046: + {H^{1/2} \over \alpha'^2}dx_3^2   

2047: + H^{1/2} (dU^2 + U^2 d \Omega_5^2)\right\rbrack\ ,\cr

2048: {A_1 \over \alpha'}  & = & 

2049: {\chi  \rho^2  \over \gYMDim{2}^2 H

2050: (1 + {\chi^2 \rho^2 \over \gYMDim{2}^4 H})} d \fiberphi

2051: \ , \cr

2052: {A_3 \over \alpha'^2} & = & H^{-1} dt \wedge dx_1 \wedge dx_2

2053: \ ,  \cr

2054: {B_2 \over \alpha'}  & = & 

2055: {1 \over \gYMDim{2}^2}\chi H^{-1} \rho\,  dt \wedge d\rho

2056: \ , \cr

2057: e^{\dilaton} & = &  \gYMDim{2}^2  H^{1/4} 

2058: \left({\gYMDim{2}^4 H + \chi^2 \rho^2 \over 

2059: \gYMDim{2}^4 H} \right)^{3/4}\ , \cr

2060: H &\equiv& {8 \pi^2 \gYMDim{2}^2 R N \over U^4}  \ .

2061: \eeq

2062: %

2063: T-dualizing along $x_3$ brings this to the form

2064: %

2065: \beq  {ds^2 \over \alpha'} & = & 

2066: \left(1 + {(2 \pi R_3 \chi) \rho^2 \over 

2067: \gYMDim{3}^4 H}\right)^{1/2}

2068: \left\lbrack

2069: H^{-1/2} \left(-dt^2 + d\rho^2 

2070: + {\rho^2 d \fiberphi^2 + dx_3^2 \over 

2071: 1 + {(2 \pi R_3\chi \rho)^2 \over \gYMDim{3}^4 H} }\right) 

2072: \right.

2073: \cr &&\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad

2074: \left.

2075: % The following command is not part of the equation

2076: % but rather makes the \right\rbrack the same size

2077: % of \left\lbrack by letting LaTeX know what is the largest

2078: % construct size that it needs to match the parenthesis with: 

2079: \vphantom{

2080: \left(-dt^2 + d\rho^2 %

2081: + {\rho^2 d \fiberphi^2 + dx_3^2 \over %

2082: 1 + {(2 \pi R_3\chi \rho)^2 \over \gYMDim{3}^4 H} }\right) %

2083: } %

2084: % END OF \vphantom command

2085: %

2086: +   H^{1/2} (dU^2 + U^2 d \Omega_5^2)

2087: \right\rbrack\ ,\cr

2088: {A_2 \over \alpha'}  & = & 

2089: {2 \pi R_3 \chi  \rho^2  \over \gYMDim{3}^2 H 

2090: (1 + {(2 \pi R_3 \chi)^2 \rho^2 \over \gYMDim{3}^4 H})} 

2091: d \fiberphi \wedge dx_3\ , \cr

2092: {A_4 \over \alpha'^2} & = & 

2093: H^{-1} dt \wedge dx_1 \wedge dx_2  \wedge dx_3\ , \cr

2094: {B_2 \over \alpha'}  & = & 

2095: {1 \over \gYMDim{3}^2}(2 \pi R_3 \chi) 

2096: H^{-1} \rho\,  dt \wedge d\rho\ , \cr

2097: e^{\dilaton} & = &  \gYMDim{3}^2  

2098: \left(1+{(2 \pi R_3 \chi)^2 \rho^2 \over 

2099: \gYMDim{3}^4 H} \right)^{1/2}\ , \cr

2100: H & = & {4 \pi \gYMDim{3}^2  N \over U^4}\ ,  \cr

2101: \gYMDim{3}^2 & = & 2 \pi R_3 \gYMDim{2}^2 \ .

2102: \eeq

2103: %

2104: (Here $A_4$ is not the complete RR 4-form,

2105: but is such that the 5-form RR field-strength is the

2106: self-dual part of $dA_4$.)

2107: This solution is the S-dual of the solution of

2108: \cite{Hashimoto:2004pb,Hashimoto:2005hy},

2109: which was also discussed in \cite{Huang:2005mi,Cai:2006td}.

2110:  In other words, (\ref{a5}) can be viewed as a

2111: dimensional reduction of NCOS from 3+1 to 2+1 dimensions.

2112: 

2113: Twisting instead along a direction transverse to the 

2114: D2-brane gives rise to the S-dual of dipole theories 

2115: \cite{Bergman:2000cw,Bergman:2001rw},

2116: dimensionally reduced from 3+1 to 2+1 dimensions along similar lines.

2117: 

2118: \subsection{D4-brane}

2119: 

2120: A similar construction applied to the case of a D4-brane turns out 

2121: {\it not} to give rise to any interesting non-local field theory. 

2122: One reason for this is the fact that a D4-brane, 

2123: when lifted to M-theory,

2124: wraps the M-theory circle unlike the D0 and the D2-branes.  

2125: Also, the radius of the M-theory circle remains finite in the decoupling limit, as can be seen from the relation

2126: %

2127: \be R = \gIIA \alpha'^{1/2} = \gYMDim{4}^2 = \text{finite.} \ee

2128: %

2129: The decoupled theory turns out to be nothing more than a twisted

2130: compactification of the decoupled M5 superconformal field theory.

2131: 

2132: To see this more explicitly, start with the supergravity solution of the 

2133: D4-brane\footnote{We abbreviate the RR 2-form potential as it plays no significant role here.}

2134: %

2135: \beq  ds^2 & = & h^{-1/2} (-dt^2 + \sum_{i=1}^4 dx_i^2) 

2136: +  h^{1/2} \sum_{i=5}^9 dx_i^2\ , \cr

2137: e^{\dilaton} & = & \gIIA h^{-1/4}\ , \cr

2138: h & = & 1 + {g N \alpha'^{3/2}\over r^3} \ . \eeq

2139: %

2140: This solution lifts to M-theory as follows

2141: %

2142: \be  ds^2  =  h^{-1/3} (-dt^2 + \sum_{i=1}^4 dx_i^2 + dz^2) 

2143: +  h^{2/3} \sum_{i=5}^9 dx_i^2 \ . \ee

2144: %

2145: Now, twist along the world volume,

2146: %

2147: \be  ds^2  =   h^{-1/3} \left(-dt^2 +   d\rho^2 

2148: + \rho^2 \left\{

2149: \dsBase{2}^2 +  (d \fiberphi + \connHopf+\eta dz)^2 \right\}

2150: + dz^2\right) +  h^{2/3} \sum_{i=5}^9 dx_i^2 \ , \ee

2151: %

2152: and reduce to IIA,

2153: %

2154: \be  ds^2  =  {1 \over \sqrt{1 + \eta^2 r^2}} 

2155: \left(h^{-1/2} \left\lbrack-dt^2 + d\rho^2 

2156: + \rho^2 \left\{\dsBase{2}^2 + {1 \over 1+\eta^2 \rho^2} 

2157: (d \fiberphi + \connHopf)^2 \right\}\right\rbrack

2158:  +  h^{1/2} \sum_{i=5}^9 dx_i^2\right) \ . \ee

2159: %

2160: In the decoupling limit where $R = \gIIA \lst = g_{YM4}^2$ is kept

2161: fixed, $\eta$ also stays finite. This leads to a relatively boring

2162: field theory which is nothing more than 4+1 SYM in a Kaluza-Klein

2163: Melvin universe, which lifts to a M5 SCFT on flat 5+1 dimensional

2164: space-time with twisted compactification

2165: %

2166: \be ds^2 = -dt^2 +   d\rho^2 + \rho^2 \left\{\dsBase{2}^2 

2167: +  (d \fiberphi + \connHopf+\eta dz)^2\right\}

2168: + dz^2, \qquad z \sim z + 2 \pi R \ . \ee

2169: 

2170: \subsection{NS5-brane}

2171: 

2172: The Kaluza-Klein Melvin background also gives rise to a UV deformation

2173: for the decoupled theory on NS5-branes. To see this, start with the

2174: supergravity solution for NS5-branes\footnote{We ignore the

2175: NSNS 2-form potential here as well.}

2176: %

2177: \beq

2178: ds^2 & = & -dt^2 + \|d\vec x\|^2

2179: + h(r) \left(dr^2 + r^2\left\{\dsBase{2}^2 

2180: + (d \fiberphi + \connHopf)^2\right\}\right), \cr

2181: % H_3 & = &  N d\connHopf\wedge (d \fiberphi + \connHopf), \cr

2182: e^\dilaton & = & g_s h(r)^{1/2}, \cr

2183: h(r) &=& 1 + {m \alpha' \over r^2}\ . \label{ns5bg}

2184: \eeq

2185: %

2186: Lifting to M-theory, we find

2187: %

2188: \be 

2189: ds^2  =  h^{-1/3} (-dt^2 + \|d\vec x\|^2) 

2190: + h^{2/3}(r) \left(dr^2 + r^2\left\{\dsBase{2}^2 

2191: + (d \fiberphi + \connHopf)^2 + dz^2\right\}\right)\ ,\ee

2192: %

2193: with $z \sim z + 2 \pi R$.  Twisting gives

2194: %

2195: \be 

2196: ds^2  =  -h^{-1/3} (-dt^2 + \|d\vec x\|^2) 

2197: + h^{2/3}(r) \left(dr^2 + r^2\left\{\dsBase{2}^2 

2198: + (d \fiberphi + \connHopf+ \eta dz)^2 + dz^2\right\}\right)\ . \ee

2199: %

2200: Reducing to IIA then gives

2201: %

2202: \be ds^2  =  \sqrt{1 + \eta^2 r^2} \left[ -dt^2 + \|d\vec x\|^2

2203:  + h(r) \left(dr^2 + r^2\left\{\dsBase{2}^2 + 

2204: {(d \fiberphi + \connHopf)^2\over 1+\eta^2 r^2} 

2205: \right\}\right)\right] \ . \ee 

2206: 

2207: In terms of $\chi = \eta \gIIA \lst$, which we are keeping finite, 

2208: %

2209: \be ds^2  =  \sqrt{1 + \frac{\chi^2 r^2}{\gIIA^2 \alpha'}} 

2210: \left[ -dt^2 + \|d\vec x\|^2 

2211: + \left(1 + { N \alpha' \over r^2}\right) 

2212: \left(dr^2 + r^2\left\{\dsBase{2}^2 + {(d \fiberphi + \connHopf)^2\over 1+\chi^2 r^2/\gIIA^2 \alpha'} 

2213: \right\}\right)\right] \ . \ee 

2214: %

2215: If we now let $r = \gIIA\rho$, and send $\gIIA\rightarrow 0$,

2216: keeping $\alpha'$ fixed,

2217: to derive the dual of the decoupling limit

2218: %

2219: \be ds^2  =  \sqrt{1 + \chi^2 \rho^2 \alpha'} 

2220: \left[ -dt^2 + \|d\vec x\|^2 

2221: + { N \alpha' \over \rho^2} 

2222: \left(d\rho^2 + \rho^2\left\{\dsBase{2}^2 

2223: + {(d \fiberphi + \connHopf)^2\over 1+\chi^2 \rho^2 / \alpha'} 

2224: \right\}\right)\right]\ . \ee 

2225: %

2226: This is a UV deformation of little string theory in type IIA with

2227: string tension $\alpha'$.  This solution reduces to the

2228: Callan-Harvey-Strominger solution \cite{Callan:1991ky} in the limit

2229: $\chi \rightarrow 0$. However, if the deformation parameter $\chi$ is rational, there exists an $SL(2,Z)$ 

2230: transformation along the lines 

2231: of what is described in section~\ref{sec:Renormalization}

2232: that brings the supergravity solution to the form of a discrete

2233: orbifold of CHS with a constant RR 1-form potential.

2234: 

2235: 

2236: \bibliography{PFT}\bibliographystyle{utphys}

2237: \end{document}

2238: