1: \documentclass[prd,aps,floats,

2: twocolumn,tightenlines,notitlepage]{revtex4}

3: 

4: %\documentclass[12pt]{article}

5: 

6: 

7: \usepackage{epsfig}

8: 

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

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

11: \def\beq{\begin{eqnarray}}

12: \def\eeq{\end{eqnarray}}

13: \def\a{\alpha}

14: \def\b{\beta}

15: \def\e{\epsilon}

16: \def\l{\lambda}

17: \def\g{\gamma}

18: \def\p{\partial}

19: \def\t{\tilde}

20: \def\vp{\varphi}

21: \def\G{\Gamma}

22: \def\N{{\cal N}}

23: \def\O{{\cal O}}

24: \def\R{{\cal R}}

25: \def\V{{\cal V}}

26: \def\RR{{\bf R^3}}

27: \def\({\left (}

28: \def\){\right )}

29: \def\[{\left [}

30: \def\[{\right ]}

31: \def\S{{\cal S}}

32: \def\pr{{(\phi_{,r})}}

33: \def\tr{\mathrm{tr}}

34: \def\ra{\rightarrow}

35: 

36: \begin{document}

37: \title{Designer Gravity and Field Theory Effective Potentials}

38: 

39: \author{Thomas Hertog and Gary T. Horowitz}

40: 

41: \affiliation{Department of Physics, UCSB, Santa Barbara, CA 93106}

42: 

43: 

44: \bibliographystyle{unsrt}

45: 

46: 

47: 

48: 

49: \begin{abstract}

50: Motivated by the AdS/CFT correspondence, we show that there is a

51: remarkable agreement between static supergravity solutions and

52: extrema of a field theory potential. For essentially  any function

53: $\V(\a)$ there are boundary conditions in anti de Sitter space so

54: that gravitational solitons exist precisely at the extrema of $\V$

55: and have masses given by the value of $\V$ at these extrema. Based

56: on this, we propose new positive energy conjectures. 

57: On the field theory side, each function $\V$ can be interpreted as the 

58: effective potential for a certain operator in the dual field theory.

59: 

60: \end{abstract}

61: %\end{titlepage}

62: \maketitle

63: 

64: {\bf Introduction.}

65: In theories of gravity coupled to matter, the theory is usually fully

66: determined by the action. The boundary conditions at

67: infinity are often not independent, but uniquely determined by basic

68: requirements such as finite total energy. This is

69: not the case, however, for certain theories of gravity in asymptotically

70: anti de Sitter (AdS) spacetimes. For the same action, there can be 

71: many possible boundary conditions, and changing the boundary

72: conditions changes the properties of the theory. In particular,

73: we will see that one can ``pre-order" the number and masses of 

74: solitons in some theories: There are boundary conditions which yield 

75: any desired result. For this reason, we call this phenomenon 

76: ``designer gravity".

77: 

78: Among the theories of gravity for which this is possible are

79: certain supergravity theories. In fact, although this result is

80: independent of string theory, it was discovered

81: while investigating the AdS/CFT correspondence \cite{Maldacena98}.

82: Furthermore, in cases where there is a field theory dual, the

83: gravitational solitons can be used to compute certain effective potentials 

84: in the field theory.

85: 

86: We will consider theories of gravity coupled to a scalar field

87: with potential $V(\phi)$. We require that $V$ has a negative

88: maximum, so that AdS is a solution and small scalar fluctuations

89: are tachyonic, $m^2<0$.   It has long been known that tachyonic

90: scalars in $d+1$ dimensional AdS spacetime  are stable provided

91: their mass is above the Breitenlohner-Freedman (BF) bound

92: \cite{Breitenlohner82} $m^2_{BF} = -d^2/4$ (in units of the AdS

93: radius). It has been shown much more recently that if

94: \be\label{bound} m_{BF}^2 \le m^2 <m_{BF}^2 +1 \ee then more

95: general boundary conditions are possible which still admit a

96: conserved finite total energy and preserve all the AdS symmetries

97: \cite{Hertog04,Henneaux04}.

98: 

99: For definiteness, we will  focus on the case of $\N=8$ gauged

100: supergravity in four dimensions, and comment on generalizations at

101: the end.

102:  This theory can be consistently truncated to

103:  include just gravity and a single scalar field with potential

104:  \cite{Duff99}

105:  \be\label{pot}

106:  V(\phi) = -2-\cosh(\sqrt{2}\phi)

107:  \ee

108:  so, setting $8\pi G=1$, our action is

109:  \be \label{4-action} S=\int d^4x\sqrt{-g}\left[\frac{1}{2}R

110: -\frac{1}{2}(\nabla\phi)^2 +2+\cosh(\sqrt{2}\phi) \right] \ee

111:  The potential (\ref{pot}) has a maximum at $\phi=0$ corresponding to 

112:  an $AdS_4$ solution with unit

113:  radius.  It is unbounded from below, but small fluctuations have

114:  $m^2 =-2$ which is above the BF bound, and satisfies (\ref{bound}).

115: 

116: In all asymptotically AdS solutions, the scalar $\phi$  decays

117: at large radius as

118:  \be\label{hair4d}

119:   \phi(r)=\frac{\alpha}{r}+\frac{\beta}{r^2}

120:   \ee

121: where $r$ is an asymptotic area coordinate, and $\a$ and $\b$ can

122: depend on the other coordinates. The standard boundary conditions

123: correspond to either $\a=0$ or $\b=0$ \cite{Breitenlohner82,Klebanov:1999tb}.

124:  It was shown in \cite{Hertog04} that $\b = k\a^2$

125: (with $k$ an arbitrary constant) 

126: was another possible boundary condition that preserves all the

127: asymptotic AdS symmetries. We now consider even

128: more general boundary

129: conditions $\b = \b(\a)$. Although these will generically break

130: some of the asymptotic AdS symmetries, they are invariant under

131: global time translations. Hence there is still a conserved total

132: energy, as we now show.

133: 

134: As discussed in \cite{Hertog04}, the usual definition of energy in

135: AdS diverges whenever $\a\ne 0$. This is because the 

136: backreaction of the scalar field causes certain 

137: metric components to fall off slower than usual. The

138: complete set of boundary conditions can be found in

139: \cite{Hertog04} but the main change is in $g_{rr}$:

140:  \be \label{4-grr}

141: g_{rr}=\frac{1}{r^2}-\frac{(1+\alpha^2/2)}{r^4}+

142: O(1/r^5)

143: \ee

144: 

145: The expression for the conserved

146: mass depends on the asymptotic behavior 

147: of the fields and is defined as follows.

148: Let $\xi^\mu$ be a timelike vector which asymptotically approaches

149: a (global)  time translation in AdS. The Hamiltonian takes the

150: form

151: \be

152:  H = \int_\Sigma \xi^\mu  C_\mu + {\rm surface \ terms}

153: \ee

154:  where $\Sigma$ is a spacelike surface, $C_\mu$ are the usual

155: constraints, and the surface terms should be chosen so that the

156: variation of the Hamiltonian is well defined. The variation of the

157: usual gravitational surface term is given by

158: \beq\label{gravch}

159: \delta Q_{G}[\xi]&=&\frac{1}{2}\oint dS_i

160: \bar G^{ijkl}(\xi^\perp \bar{D}_j \delta h_{kl}-\delta h_{kl}\bar{D}_j\xi^\perp)

161: \eeq

162: where $G^{ijkl}={1 \over 2} g^{1/2} (g^{ik}g^{jl}+g^{il}g^{jk}-2g^{ij}g^{kl})$,

163: $h_{ij}=g_{ij}-\bar{g}_{ij}$ is the deviation from the spatial metric 

164: $\bar{g}_{ij}$ of pure AdS,  $\bar{D}_i$ denotes covariant differentiation 

165: with respect to $\bar{g}_{ij}$ and $\xi^\perp = \xi \cdot n$ with $n$ the

166: unit normal to $\Sigma$.  Since our scalar

167: field is falling off more slowly than usual if $\a \ne 0$, 

168: there is an additional scalar contribution to the surface terms. 

169: Its variation is simply

170: \be\label{deltaQ}

171: \delta Q_\phi[\xi] =-\oint \xi^\perp \delta \phi D_i \phi dS^i \ee

172:  Using the asymptotic

173: behavior (\ref{hair4d})  this becomes

174: \be

175: \delta Q_\phi[\xi] = r\oint(\a\delta \a)d\Omega 

176: + \oint [\delta (\a\b) + \b\delta \a]d\Omega

177: \ee

178: Since there is a

179: term proportional to the radius of the sphere, this scalar surface

180: term diverges. However, this divergence is exactly canceled by the

181: divergence of the usual gravitational surface term (\ref{gravch}).

182: The total charge can therefore be integrated, yielding

183: \be\label{scalarst}

184: Q [\xi] = Q_{G}[\xi]+ r\oint {\a^2\over 2}d\Omega  + 

185: \oint [\a\b + W(\a)]d\Omega \ee

186: where we have defined

187: \be W(\a) = \int_0^\a \b(\tilde \a)d\tilde \a \ee

188: In addition to canceling the divergence in (\ref{scalarst}), the

189: gravitational surface term contributes a finite amount $M_0$. For

190: the spherically symmetric solutions we consider below, this is

191: just the coefficient of the $1/r^5$ term in $g_{rr}$. Since $\a$

192: and $\b$ are now independent of angles, the total mass becomes

193: \be\label{mass}

194: M= 4\pi(M_0 + \a\b + W)

195: \ee

196: (For $\b=k\a^2$, this agrees with the expression for the mass given in

197: \cite{Hertog04}.)

198: 

199: 

200: {\bf Gravitational Solitons.}

201: We want to study solitons in this theory. These are nonsingular,

202: static, spherically symmetric solutions. Writing the metric as \be

203: ds^2=-h(r)e^{-2\chi(r)}dt^2+h^{-1}(r)dr^2+r^2d\Omega \ee the

204: field equations read \be\label{hairy14d}

205: h\phi_{,rr}+\left(\frac{2h}{r}+\frac{r}{2}\phi_{,r}^2h+h_{,r}

206: \right)\phi_{,r}   =  V_{,\phi} \ee \be\label{hairy24d}

207: 1-h-rh_{,r}-\frac{r^2}{2}\phi_{,r}^2h =  r^2V(\phi) \ee \be

208: \chi_{,r} = -{r\phi_{,r}^2\over 2} \ee

209:  Regularity at the origin requires $h=1$ and $ h' = \phi'= \chi'=0$. Rescaling $t$ shifts

210:  $\chi$ by a constant, so its value at the origin is arbitrary. Thus solutions can be labeled by

211:  the value of  $\phi$ at the origin. For each $\phi(0)$,

212:  one can integrate these ODE's and get a soliton.

213: Asymptotically, $\phi$ behaves like (\ref{hair4d}), so we get a

214: point in the $(\a,\b)$ plane. Repeating for all  $\phi(0)$ yields

215: a curve $\b_s(\a)$ where the subscript indicates this is

216: associated with solitons. This curve is plotted in Fig.

217: 1. (Since the potential $V(\phi)$ is even, it suffices to

218: consider positive $\phi(0)$ which corresponds to positive $\a$.)

219: Note that solitons exist for arbitrarily small $\a$. 

220: When $\a\ll 1$, $\phi(r)$ is small everywhere, and one might have

221: thought a linearized approximation should be valid implying no solitons

222: could exist. This is incorrect since for any $\a\ne 0$, the backreaction 

223: is always large asymptotically as shown in (\ref{4-grr}).

224: Given a choice of boundary condition $\b(\a)$, the allowed

225: solitons are simply given by points where the soliton curve

226: intersects the boundary condition curve: $\b_s(\a) = \b(\a)$.

227: 

228: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

229: \begin{figure}[htb]

230: \begin{picture}(0,0)

231: \put(0,60){$\beta$} 

232: \put(210,140){$\alpha$}

233: \end{picture}

234: \psfig{file=solitonbversusa.eps,width=3.2in} 

235: \caption{The function $\beta_s$  obtained from the solitons.} 

236: \label{1}

237: \end{figure}

238: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

239: 

240: We can now state our prescription for choosing boundary conditions

241: to reproduce any prescribed set of solitons.  Set

242: \be\label{Wnot}

243: W_0(\a) = -\int_0^\a \b_s(\a)

244: \ee

245: This function is universal, in the sense that it is independent of our choice of

246: boundary

247: conditions. Now given any smooth function $\V(\a)$ with $\V(0)=0$,

248: we write $\V=W_0+W$ and take our

249: boundary conditions to be $\b=W'(\a)$. It follows immediately that the extrema

250: of $\V$ are in one-to-one correspondence with solitons that obey these 

251: boundary conditions:

252: \be

253: 0 = \V'=W_0' + W' = -\b_s + \b

254: \ee So the extrema of $\V$ are precisely the points where

255: $\b_s=\b$. Furthermore, the mass of each soliton is given by the

256: value of $\V$ at the corresponding extremum. To see this remember

257: that static solutions are extrema of the mass

258: \cite{Sudarsky:1992ty}.  Suppose we choose our boundary condition

259: to be $\b = \b_s(\a)$. For this special case, all the solitons are

260: allowed by the boundary conditions. Since we have a one parameter

261: family of static solutions, the mass must be constant, i.e., all

262: the solitons have the same mass. But this includes $\b=\a =0$

263: which is just AdS and has zero mass. So all the solitons have zero

264: mass. From (\ref{mass}), with boundary conditions $\b = \b_s(\a)$

265: we have \be 0 = M_0 + \a \b_s -W_0 \ee Therefore, for our general

266: boundary condition $\b=W'(\a)$, we have 

267: \be

268: M= 4\pi(M_0 + \a\b + W) =4\pi(W_0+W) =\oint \V d\Omega 

269: \ee

270: where we have used the fact that $\b = \b_s(\a)$ for

271: a soliton. Thus the mass of the soliton is indeed given by the

272: value of $\V$ at the corresponding extremum. Notice that the only 

273: restriction on $\V$ (that $\V(0)=0$) comes from the fact that we want the

274: total mass of pure AdS to be zero.

275: 

276: We have also studied the stability of these solitons. The most likely

277: mode to go unstable is a spherically symmetric scalar perturbation

278: like the one studied for hairy black holes in \cite{Maeda04}. We

279: have found numerically that this mode is indeed unstable if

280: $\V''<0$. We expect the solitons with $\V''> 0$ to be stable.

281: This leads to a new class of ``positive" energy conjectures

282: \footnote{This does not contradict our

283: earlier result (discussed in hep-th/0406134) that theories with solitons must 

284: have solutions with arbitrarily negative energy, since that 

285: assumed conventional boundary conditions that are scale invariant.

286:  The boundary conditions used here are generally

287: not invariant under rescaling $r$.}. For given boundary conditions, the

288: minimum energy solution is expected to be static, and hence one of

289: the solitons we have been discussing. If $\V$ has a global

290: minimum, then it seems likely that the energy of any

291: supergravity solution cannot be less than the minimum mass soliton.

292: In other words,

293: 

294: {\it Conjecture:  Given any smooth function $\V(\a)$  with $\V(0)=0$ 

295: and a global

296: minimum $\V_{min}$, consider  solutions to (\ref{4-action}) with

297: boundary condition $\b=W'$ where $W=\V-W_0$ and $W_0$ is given by

298: (\ref{Wnot}). Then the conserved mass (\ref{scalarst}) of

299: any nonsingular initial data set is bounded below by

300: $4\pi \V_{min}$.}

301: 

302: {\bf Field theory.}

303: We now turn to the dual field theory interpretation. String theory

304: on spacetimes which asymptotically approach $AdS_4\times S^7$ is

305: dual to the 2+1 conformal field theory (CFT)

306: describing the low energy excitations of a

307: stack of M two-branes. This theory is not well understood, but we

308: can learn something nontrivial using the gravitational solitons.

309: With $\b=0$ boundary conditions, the bulk scalar $\phi$ is dual to

310: a dimension one operator $\O$. One way of obtaining this CFT is by

311: starting with the field theory on a stack of D two-branes and

312: taking the infrared limit. In that description \cite{Aharony:1998rm},

313: \be \O = Tr T_{ij}\vp^i\vp^j \ee

314: where $T_{ij}$ is symmetric and traceless and

315: $\vp^i$ are the adjoint scalars.

316: 

317: Let $S_0$ denote the CFT lagrangian and  consider the deformation

318: \be\label{deform}

319: S= S_0 - k\int \O \ee

320: Using the standard AdS/CFT dictionary,

321: the vacuum expectation value of $\O$ in this deformed theory is

322: obtained by finding nonsingular static supergravity solutions with

323: $\b=-k$. But these are precisely our solitons. Given a soliton

324: with $\b=-k$, one has $\langle \O\rangle = \a$. Hence the function 

325: $\V(\alpha)=W_0-k\a$ can be interpreted as

326: the effective potential for $\langle \O\rangle$, where

327: $W_0$ is the function (\ref{Wnot}) computed earlier from the

328: soliton solutions.

329: From Fig. 1 we see that

330: there  are three qualitatively different regions. For small $k$,

331: there is a unique soliton and hence a unique nonzero value for 

332: $\langle \O\rangle$. For intermediate values of $k$ there are two

333: solitons indicating there are two vacua, and for large $k$ there

334: are no solitons indicating there is no vacuum at all.

335: 

336: Since our CFT lives on $S^2\times R$ (it is dual to a

337: bulk theory which approaches global AdS asymptotically) one might

338: have expected a mass term ${1\over 2}m^2 \vp^2$ coming from the

339: conformal coupling of the scalars to the curvature of the $S^2$.

340: The radius of the $S^2$ is equal to the AdS radius, so one

341: expects $m^2 = R/8 = 1/4$ in AdS units. Since $\O$ is quadratic in

342: $\vp$ the presence of a mass term would mean that for small 

343: $k<m^2/2$, the vacuum would be unchanged and $\langle \O\rangle=0$. 

344: But this is not what we find. Figure 1 shows that $\b_s$ is linear

345: in $\a$ for small $\a$, so that $W_0$ is quadratic in $\a$, hence the vacuum

346: expectation value $\langle \O\rangle$ is shifted even for small $k$.

347: This is illustrated in Figure 2 for $k=1/2$. For slightly larger $k$ a new 

348: maximum appears at larger $\a$ and the theory becomes nonperturbatively 

349: unstable.

350: 

351: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

352: \begin{figure}[htb]

353: \begin{picture}(0,0)

354: \put(30,140){$\V$} 

355: \put(210,50){$\alpha$}

356: \end{picture}

357: \psfig{file=singletrace.eps,width=3.2in} 

358: \caption{The effective potential $\V (\a)=W_0 -\frac{1}{2}\a$.} 

359: \label{2}

360: \end{figure}

361: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

362: 

363: 

364: Now suppose we replace $-k\int \O$ in (\ref{deform}) with $\int

365: W(\O)$ where $W$ is an arbitrary function of $\O$. Remarkably, the

366: expectation values $\langle \O \rangle$ in different vacua are again given by the 

367: extrema of $\V=W_0+W$, where $W_0$ is the same function 

368: as above, and $W$ is unchanged.

369: This is because the addition of $\int W(\O)$ to the CFT action

370: corresponds in the bulk to using the modified boundary conditions

371: $\b=W'$ \cite{Witten02}. But we have already seen that the extrema

372: of $\V$ correspond to solitons with  precisely these boundary

373: conditions. The fact that the function $W$ does not receive any

374: corrections in the effective potential is surprising and

375: reminiscent of a nonrenormalization theorem, but we are dealing

376: with configurations that are far from supersymmetric. Perhaps it

377: is related to taking the large $N$ limit or to properties of

378: the operators that are dual to scalars with masses in the range (\ref{bound}).

379: An example of the effective potential in the presence of a 

380: multitrace deformation $W$ that yields a 

381: nontrivial false vacuum is given in Figure 3.

382: 

383: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

384: \begin{figure}[htb]

385: \begin{picture}(0,0)

386: \put(75,135){$\V$} 

387: \put(210,30){$\alpha$}

388: \end{picture}

389: \psfig{file=doubletrace.eps,width=3.2in} 

390: \caption{The effective potential $\V = W_0 -\frac{1}{4}\a^2 +

391: \frac{1}{40}\a^3$.} 

392: \label{3}

393: \end{figure}

394: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

395: 

396: 

397: {\bf Discussion.}

398: In summary, we have seen that one can ``pre-order" solitons in supergravity,

399: in the following sense: 

400: Given essentially any function $\V(\a)$ there are boundary conditions such that gravitational

401: solitons exist precisely for each extremum of $\V(\a)$ and have masses given by the value of 

402: $\V$ at the corresponding extremum.  Furthermore, in supergravity theories with a field

403: theory dual, the function $\V$ can be interpreted as the effective potential for the dual

404: operator $\O$.

405: It would be interesting to perform an independent field theory calculation of the

406: effective potential, for instance in the case of a simple single trace deformation, since

407: this would provide a new test of AdS/CFT. 

408: One can also extend our results from solitons to black holes with scalar hair. 

409: One can pre-order black holes either in terms of their size or temperature. In the latter case,

410: a similar bulk analysis yields again a function $\tilde \V (\a)$ that can be interpreted as the finite 

411: temperature  effective potential  for $\O$ in the dual field theory.

412: This will be discussed in more detail in \cite{Hertog05}.

413: 

414: Although we have focused on a scalar field with $m^2=-2$ in four dimensional  $\N=8$ supergravity, the gravity side of the story can be generalized to other dimensions and all scalars with

415: masses in the range (\ref{bound}). For

416:  asymptotically $AdS_{d+1}$ spacetimes, a scalar field with mass $m$ asymptotically falls off like

417:  \be \label{asympt-sc}

418: \phi = {\alpha\over r^{\Delta_-}} + {\beta\over r^{\Delta_+}}

419: \ee

420: where

421: \be\label{roots}

422: \Delta_\pm = {d \pm \sqrt{d^2 + 4 m^2}\over 2}

423: \ee

424:  If the mass is

425:  in the range $m^2_{BF} \le m^2 < m^2_{BF} +1$, then a finite, conserved

426:  total energy

427:  can be defined for any boundary condition $\b(\a)$. The variation of the scalar

428: surface term is still (\ref{deltaQ}), so inserting this asymptotic

429: behavior of $\phi$ yields

430: \be \label{genmass}

431:  M=\mathrm{Vol}(S^{d-1})\left[ {d-1 \over 2} M_0 + \Delta_- \a\b +

432: (\Delta_+ - \Delta_-) W\right]

433: \ee

434: One can again construct the

435: soliton curve $\beta_s(\a)$ and find boundary conditions that admit

436: any desired soliton solutions.  If a dual field theory exists, then one

437: can again compute effective potentials for the operators dual to

438: the bulk scalar field.

439: \newline

440: {\bf Acknowledgments} It is a pleasure to thank O. Aharony,  J.

441: Maldacena, D. Marolf, J. Polchinski and N. Seiberg for discussions.

442: Part of this work was

443: done while G.H. was visiting the IAS in Princeton and he thanks

444: them for their hospitality. T.H. thanks the Solvay Institute at the

445: Universit\'e Libre de Bruxelles 

446: for its kind hospitality during the completion of this work.

447: This work was supported in part by NSF grant PHY-0244764.

448: 

449: \begin{thebibliography}{99}

450: 

451: \bibitem{Maldacena98}

452: J. M. Maldacena,

453: % ``The large N limit of superconformal field theories and supergravity,''

454:  Adv.\ Theor.\ Math.\ Phys.\  {\bf 2}

455: (1998) 231, hep-th/9711200.

456: 

457: \bibitem{Breitenlohner82}

458: P.~Breitenlohner and D.~Z.~Freedman,

459: % ``Stability In Gauged Extended Supergravity,''

460:  Annals Phys.\  {\bf 144} (1982) 249;

461: %``Positive Energy In Anti-De Sitter Backgrounds And Gauged Extended Supergravity,''

462:  Phys.\ Lett.\ B {\bf 115} (1982) 197.

463: 

464: \bibitem{Hertog04}

465: T. Hertog, K. Maeda,

466: %``Black Holes with Scalar Hair and Asymptotics in $N=8$ Supergravity,''

467: JHEP {\bf 0407} (2004) 051, hep-th/0404261.

468: 

469: \bibitem{Henneaux04}

470: M. Henneaux, C. Martinez, R. Troncoso, J. Zanelli,

471: %``Asymptotically Anti-de Sitter Spacetimes and Scalar Fields with a Logarithmic Branch,'' 

472: Phys. Rev. D {\bf 70} (2004) 044034, hep-th/0404236.

473: 

474: \bibitem{Duff99}

475: M. J. Duff, J. T. Liu,

476: %``Anti-de Sitter Black Holes in Gauged N=8 Supergravity,''

477: Nucl. Phys. {\bf B554} (1999) 237, hep-th/9901149.

478: 

479: %\cite{Klebanov:1999tb}

480: \bibitem{Klebanov:1999tb}

481: I.~R.~Klebanov and E.~Witten,

482: %``AdS/CFT correspondence and symmetry breaking,''

483: Nucl.\ Phys.\ B {\bf 556} (1999) 89,

484: hep-th/9905104.

485: %%CITATION = HEP-TH 9905104;%%

486: 

487: %\cite{Sudarsky:1992ty}

488: \bibitem{Sudarsky:1992ty}

489: D.~Sudarsky and R.~M.~Wald,

490: %``Extrema of mass, stationarity, and staticity, and solutions to the Einstein

491: %Yang-Mills equations,''

492: Phys.\ Rev.\ D {\bf 46} (1992) 1453.

493: %%CITATION = PHRVA,D46,1453;%%

494: 

495: %\cite{Maeda04}

496: \bibitem{Maeda04}

497: T.~Hertog and K.~Maeda,

498: %``Stability and thermodynamics of AdS black holes with scalar hair,''

499: hep-th/0409314.

500: %%CITATION = HEP-TH 0409314;%%

501: 

502: \bibitem{Aharony:1998rm}

503: O.~Aharony, Y.~Oz and Z.~Yin,

504: %``M-theory on AdS(p) x S(11-p) and superconformal field theories,''

505: Phys.\ Lett.\ B {\bf 430} (1998) 87,

506: hep-th/9803051;

507: %%CITATION = HEP-TH 9803051;%%

508: S.~Minwalla,

509: %``Particles on AdS(4/7) and primary operators on M(2/5) brane  worldvolumes,''

510: JHEP {\bf 9810} (1998) 002,

511: hep-th/9803053.

512: %%CITATION = HEP-TH 9803053;%%

513: 

514: \bibitem{Witten02}

515: E. Witten,

516: %``Multi-Trace Operators, Boundary Conditions, and AdS/CFTCorrespondence,'' 

517: hep-th/0112258.

518: 

519: \bibitem{Hertog05}

520: T. Hertog and G. T. Horowitz, to appear.

521: 

522: \end{thebibliography}

523: \end{document}

524: