1: \documentclass[nohyper,11pt,letterpaper]{JHEP3}

2: \usepackage[dvips]{epsfig}

3: %\usepackage{showkeys} % for cool labelling!

4: \usepackage{epsf,amsfonts,amssymb}

5: %\usepackage[active]{srcltx}

6: 

7: 

8: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

9: %%%%%%%%%%%%%%%%%%%%%%%%%% Macros %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

10: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

11: 

12: %%%%%%%%%%%%%%%%%%%%%%%% Structure definitions %%%%%%%%%%%%%%%%%%%%%%%%%%%

13: 

14: \newcommand{\eqn}[1]{(\ref{#1})}

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

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

17: \newcommand{\ben}{\begin{displaymath}}

18: \newcommand{\een}{\end{displaymath}}

19: \newcommand{\bea}{\begin{eqnarray}}

20: \newcommand{\eea}{\end{eqnarray}}

21: \newcommand{\bean}{\begin{eqnarray*}}

22: \newcommand{\eean}{\end{eqnarray*}}

23: \newcommand{\nn}{\nonumber \\}

24: \newcommand{\ba}{\begin{array}}

25: \newcommand{\ea}{\end{array}}

26: \newcommand{\bi}{\begin{itemize}}

27: \newcommand{\ei}{\end{itemize}}

28: 

29: \newtheorem{theorem}{Theorem}

30: 

31: %%%%%%%%%%%%%%%%%%%%%%%% Greek Letters %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

32: 

33: \def\l {\lambda}

34: \def\a {\alpha}

35: \def\ap {\alpha'}

36: \def\b {\beta}

37: \def\g {\gamma}

38: \def\G {\Gamma}

39: \def\d {\delta}

40: \def\s {\sigma}

41: \def\e {\epsilon}

42: \def\vt {\vartheta}

43: \def\vp {\varphi}

44: \def\T {\Theta}

45: 

46: \renewcommand{\O}{\Omega}

47: \renewcommand{\L}{\Lambda}

48: \renewcommand{\t}{\theta}

49: 

50: %%%%%%%%%%%%%%%%%% Calligraphic Letters %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

51: 

52: \newcommand{\cala}{\mbox{${\cal A}$}}

53: \newcommand{\calb}{\mbox{${\cal B}$}}

54: \newcommand{\calc}{\mbox{${\cal C}$}}

55: \newcommand{\cald}{\mbox{${\cal D}$}}

56: \newcommand{\cale}{\mbox{${\cal E}$}}

57: \newcommand{\calf}{\mbox{${\cal F}$}}

58: \newcommand{\calg}{\mbox{${\cal G}$}}

59: \newcommand{\calh}{\mbox{${\cal H}$}}

60: \newcommand{\cali}{\mbox{${\cal I}$}}

61: \newcommand{\calj}{\mbox{${\cal J}$}}

62: \newcommand{\calk}{\mbox{${\cal K}$}}

63: \newcommand{\call}{\mbox{${\cal L}$}}

64: \newcommand{\calm}{\mbox{${\cal M}$}}

65: \newcommand{\caln}{\mbox{${\cal N}$}}

66: \newcommand{\calo}{\mbox{${\cal O}$}}

67: \newcommand{\calp}{\mbox{${\cal P}$}}

68: \newcommand{\calq}{\mbox{${\cal Q}$}}

69: \newcommand{\calr}{\mbox{${\cal R}$}}

70: \newcommand{\cals}{\mbox{${\cal S}$}}

71: \newcommand{\calt}{\mbox{${\cal T}$}}

72: \newcommand{\calu}{\mbox{${\cal U}$}}

73: \newcommand{\calv}{\mbox{${\cal V}$}}

74: \newcommand{\calw}{\mbox{${\cal W}$}}

75: \newcommand{\calx}{\mbox{${\cal X}$}}

76: \newcommand{\caly}{\mbox{${\cal Y}$}}

77: \newcommand{\calz}{\mbox{${\cal Z}$}}

78: 

79: \newcommand{\bfcalc}{\mbox{\boldmath ${\cal C}$}}

80: 

81: 

82: %%%%%%%%%%%%%%%%% Boldmath Letters %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

83: 

84: \newcommand{\bfe}{\mbox{\boldmath $E$}}

85: \newcommand{\bfb}{\mbox{\boldmath $B$}}

86: \newcommand{\bfpi}{\mbox{\boldmath $\Pi$}}

87: \newcommand{\bfp}{\mbox{\boldmath $P$}}

88: \newcommand{\bfna}{\mbox{\boldmath $\nabla$}}

89: \newcommand{\bfd}{\mbox{\boldmath $D$}}

90: \newcommand{\bfeh}{\mbox{\boldmath $\hat{E}$}}

91: \newcommand{\bfbh}{\mbox{\boldmath $\hat{B}$}}

92: \newcommand{\bfah}{\mbox{\boldmath $\hat{A}$}}

93: \newcommand{\bx}{\mbox{\boldmath $x$}}

94: \newcommand{\by}{\mbox{\boldmath $y$}}

95: \newcommand{\bX}{\mbox{\boldmath $X$}}

96: \newcommand{\bY}{\mbox{\boldmath $Y$}}

97: \newcommand{\bV}{\mbox{\boldmath $V$}}

98: \newcommand{\bU}{\mbox{\boldmath $U$}}

99: 

100: \newcommand{\bfK}{\mbox{\boldmath $K$}}

101: \newcommand{\bfz}{\mbox{\boldmath $\theta$}}

102: \newcommand{\bfZ}{\mbox{\boldmath $Z$}}

103: \newcommand{\bfr}{\mbox{\boldmath $\rho$}}

104: \newcommand{\bfxi}{\mbox{\boldmath $\xi$}}

105: \newcommand{\bfs}{\mbox{\boldmath $\sigma$}}

106: \newcommand{\bft}{\mbox{\boldmath $t$}}

107: \newcommand{\bfT}{\mbox{\boldmath $T$}}

108: \newcommand{\bfss}{\mbox{\boldmath $s$}}

109: \newcommand{\bfS}{\mbox{\boldmath $S$}}

110: \newcommand{\bfG}{\mbox{\boldmath $\Gamma$}}

111: \newcommand{\bfo}{\mbox{\boldmath $0$}}

112: 

113: \newcommand{\bbe}[1]{\mbox{${\mathbb E}^{#1}$}}

114: \newcommand{\bbr}[1]{\mbox{${\mathbb R}^{#1}$}}

115: %\newcommand{\bbz}[1]{\mbox{${\mathbb Z}^{#1}$}}

116: \newcommand{\bbz}[1]{\mbox{${\mathbb Z}_{#1}$}}

117: \newcommand{\bbi}[1]{\mbox{${\mathbb I}_{#1}$}}

118: \newcommand{\bbo}[1]{\mbox{${\mathbb O}_{#1}$}}

119: 

120: %%%%%%%%%%%%%%%%%%%%%% Miscellaneous  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

121: 

122: \newcommand{\nef}{{$\caln =4$} }

123: \newcommand{\mkk}{M_\mt{KK}}

124: 

125: \newcommand{\dd}[3]{\mbox{$( #1 | \mbox{D} #2 \perp \mbox{D} #3)$}}

126: \newcommand{\ds}[3]{\mbox{$( #1 | D #2 \perp \calt #3)$}}

127: \newcommand{\para}{\parallel}

128: \newcommand{\inter}{\, \cap \,}

129: \newcommand{\su}[1]{$\calt #1\,$}

130: 

131: \newcommand{\ads}[1]{\mbox{${AdS}_{#1}$}}

132: \newcommand{\adss}[2]{\mbox{$AdS_{#1}\times {S}^{#2}$}}

133: \newcommand{\rn}{Reissner-Nordstr\"{o}m }

134: 

135: \newcommand{\bra}[1]{\mbox{$\langle #1 |$}}

136: \newcommand{\ket}[1]{\mbox{$| #1 \rangle$}}

137: \newcommand{\braket}[2]{\mbox{$\langle #1  | #2 \rangle$}}

138: \newcommand{\proj}[1]{\ket{#1}\!\bra{#1}}

139: 

140: \newcommand{\Gu}[1]{\Gamma_{\underline{#1}}}

141: \newcommand{\Gn}{\Gamma_\natural}

142: 

143: \newcommand{\un}{\underline}

144: \newcommand{\pa}{\partial}

145: \newcommand{\fc}{\frac}

146: \newcommand{\w}{\wedge}

147: \newcommand{\trace}{\mbox{Tr}}

148: \newcommand{\sac}{\ , \qquad}

149: \newcommand{\eg}{{\it e.g.}}

150: \newcommand{\ie}{{\it i.e.}}

151: \newcommand{\sgn}[1]{\mbox{sgn}(#1)}

152: \newcommand{\com}[1]{[[[{\bf #1}]]]}

153: \newcommand{\ad}{\dot{a}}

154: \newcommand{\bd}{\dot{b}}

155: \newcommand{\ik}{{\it k}}

156: \newcommand{\ione}{{\it 1}}

157: \newcommand{\itwo}{{\it 2}}

158: \newcommand{\ifive}{{\it 5}}

159: 

160: \newcommand{\ph}{\hat{P}}

161: \newcommand{\xh}{\hat{X}}

162: \newcommand{\ve}{\varepsilon}

163: \newcommand{\tr}{\mbox{Tr}}

164: 

165: \newcommand{\ra}{\rightarrow}

166: 

167: \newcommand{\comment}[1]{{\bf [[[#1]]]}}

168: 

169: %\newcommand{\atmp}[3]{{\it Adv. Theor. Math. Phys.} {\bf #1} {(#2)} #3}

170: %\newcommand{\ijtp}[3]{{\it Int. J. Theor. Phys.} {\bf #1} {(#2)} #3}

171: 

172: 

173: %%%%%%%%%%%%%%%%% Equation Numbering %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

174: 

175: \newcommand{\sect}[1]{\setcounter{equation}{0}\section{#1}}

176: \renewcommand{\theequation}{\arabic{section}.\arabic{equation}}

177: 

178: %%%%%%%%%%%%%%%%%%%% Page formatting %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

179: 

180: %\renewcommand{\baselinestretch}{1.4}

181: %\hoffset -0.5in % moves text horizontally

182: %\textwidth 165mm

183: %\textheight 220mm

184: %\evensidemargin 0mm

185: %\topmargin -15mm

186: %\headsep 1.3cm

187: %\footskip 20mm

188: %\headheight 0pt

189: 

190: 

191: %%%%%%%%%%%%%%%%%% For this paper %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

192: 

193: 

194: \newcommand{\beq}{\begin{equation}}

195: \newcommand{\eeq}{\end{equation}}

196: \newcommand{\beqr}{\begin{displaymath}}

197: \newcommand{\eeqr}{\end{displaymath}}

198: \newcommand{\beqa}{\begin{eqnarray}}

199: \newcommand{\eeqa}{\end{eqnarray}}

200: \newcommand{\beqar}{\begin{eqnarray*}}

201: \newcommand{\eeqar}{\end{eqnarray*}}

202: %\newcommand{\htx}{s} %\hat{t}}

203: %\renewcommand{\a}{\alpha}

204: %\newcommand{\ap}{\alpha'}

205: %\renewcommand{\b}{\beta}

206: %\renewcommand{\d}{\delta}

207: %\newcommand{\D}{\Delta}

208: %\newcommand{\e}{\epsilon}

209: %\newcommand{\g}{\gamma}

210: %\newcommand{\G}{\Gamma}

211: \renewcommand{\k}{\kappa}

212: %\renewcommand{\l}{\lambda}

213: %\renewcommand{\L}{\Lambda}

214: \newcommand{\m}{\mu}

215: \newcommand{\n}{\nu}

216: \newcommand{\p}{\phi}

217: %\renewcommand{\o}{\omega}

218: %\renewcommand{\O}{\Omega}

219: %\renewcommand{\P}{\Phi}

220: %\newcommand{\p}{\phi}

221: %\newcommand{\Ps}{\Psi}

222: \renewcommand{\r}{\rho}

223: \newcommand{\vr}{\varrho}

224: %\newcommand{\s}{\sigma}

225: %\renewcommand{\t}{\theta}

226: %\newcommand{\vt}{\vartheta}

227: %\newcommand{\ve}{\varepsilon}

228: %\newcommand{\z}{\zeta}

229: %\newcommand{\x}{\xi}

230: %\newcommand{\cM}{{\cal M}}

231: %\newcommand{\cA}{{\cal A}}

232: %\newcommand{\cH}{{\cal H}}

233: %\newcommand{\cG}{{\cal G}}

234: %\newcommand{\cS}{{\cal S}}

235: \newcommand{\cN}{{\cal N}}

236: %\newcommand{\cD}{{\cal D}}

237: %\newcommand{\cO}{{\cal O}}

238: %\newcommand{\cF}{{\cal F}}

239: \newcommand{\cL}{{\cal L}}

240: \newcommand{\ssc}{\scriptscriptstyle}

241: %%\newcommand{\eg}{{\it e.g.}}

242: %%\newcommand{\ie}{{\it i.e.}}

243: %\newcommand{\etal}{{\it et al.~}}

244: \newcommand{\labell}[1]{\label{#1}\qquad_{#1}} %{\label{#1}}

245: %\newcommand{\labels}[1]{\label{#1}} %{\vskip-.6cm${#1}$\label{#1}}

246: \newcommand{\reef}[1]{(\ref{#1})}

247: \newcommand{\non}{\nonumber}

248: \newcommand{\pf}{\partial}

249: %\renewcommand{\ni}{\vspace{0.2cm}\noindent}

250: %\newcommand{\dir}[1]{\overline{#1}}

251: %\newcommand{\ZZ}{\mathbb{Z}}

252: %\newcommand{\ket}[1]{|#1\rangle}

253: %\newcommand{\bra}[1]{\langle #1|}

254: %\newcommand{\sR}{\textrm{\tiny R}}

255: %\newcommand{\sNS}{\textrm{\tiny NS}}

256: %\newcommand{\tr}{\textrm{Tr}\;}

257: %\newcommand{\til}[1]{\widetilde{#1}}

258: %\newcommand{\ti}[1]{\tilde{#1}}

259: \newcommand{\df}{\textrm{d}}

260: %\newcommand{\sla}[1]{#1\hspace{-0.16cm}/}

261: \newcommand{\mt}[1]{\textrm{\tiny #1}}

262: %\newcommand{\ph}{\phantom{1}}

263: %\newcommand{\com}[1]{{\bf (#1)}}

264: \newcommand{\cF}{{\cal F}}

265: \newcommand{\cA}{{\cal A}}

266: \newcommand{\cB}{{\cal B}}

267: \newcommand{\ls}{\ell_s}

268: 

269: \newcommand{\zD}{\ensuremath{z_{D7}}}      % Position of D7 brane

270: \newcommand{\tzD}{\ensuremath{\zeta_m}}      % Position of D7 brane

271: \newcommand{\trho}{\ensuremath{\tilde{\rho}}}

272: % rescaled rho can also be \tilde{\rho}

273: \newcommand{\tz}{\ensuremath{\tilde{z}}}

274: % rescaled $z$, can also be \tilde{z}

275: \newcommand{\Rl}{\ensuremath{(R/l_s)^2}}% prefactor for Wilson loops

276: \newcommand{\tc}{\ensuremath{\sqrt{g_s \nc}}} %'t Hooft coupling

277: \newcommand{\stc}{\ensuremath{(g_sN)^{\frac{1}{4}}}}

278: % sqrt of 't Hooft coupling

279: \newcommand{\mq}{\ensuremath{m_\mt{q}}}      % Quark mass.

280: \newcommand{\eff}{\ensuremath{\mbox{\small eff.}}}

281: \newcommand{\calN}{\mbox{${\cal N}$}}

282: \newcommand{\cY}{\ensuremath{{\cal Y}}}

283: \newcommand{\vY}{\ensuremath{{\cal Y}^{\ell,\pm}}}

284: \newcommand{\ellh}{\ensuremath{\frac{\ell}{2}}}

285: \newcommand{\suLR}{\ensuremath{SU(2)_R\times SU(2)_L}}

286: \newcommand{\br}{\ensuremath{\bar{\rho}}}

287: \newcommand{\sun}{\ensuremath{{SU(N)}} }

288: \newcommand{\half}{\ensuremath{\frac{1}{2}}}

289: \newcommand{\mpi}{\ensuremath{M_{\pi}}}

290: \newcommand{\Lqcd}{\ensuremath{\Lambda_{\mt{QCD}}}}

291: \newcommand{\fphi}{\ensuremath{f_{\phi}}}

292: \newcommand{\cc}{\langle \bar{\psi} \psi \rangle}

293: \newcommand{\fpi}{f_\pi}

294: 

295: \def\lt {\lambda}

296: \def\rt {r}

297: \def\rhot {\rho}

298: \def \rvac{r_\mt{vac}}

299: \def\nc {N_\mt{c}}

300: \def\nf {N_\mt{f}}

301: \def\ua {U(1)_\mt{A}}

302: \def\t6 {T_\mt{D6}}

303: \def\ut {U_\mt{KK}}

304: \def\uh {U_\mt{T}}

305: \def\gym {g_\mt{YM}}

306: \newcommand{\nonsol}{D4-soliton}

307: \newcommand{\te}{t_\mt{E}}

308: \newcommand{\ct}{T_\mt{deconf}}

309: \newcommand{\tcc}{T_\mt{fund}}

310: \newcommand{\tb}{\bar{M}}

311: \newcommand{\msusy}{M_\mt{susy}}

312: \newcommand{\lm}{\l_\mt{match}}

313: 

314: 

315: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

316: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

317: %%%%%%%%%%%%%%%%%%%%%% TITLEPAGE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

318: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

319: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

320: 

321: 

322: \title{\LARGE Towards a holographic dual of large-$N_\mathrm{\large c}$ QCD}

323: 

324: \author{Mart\'\i n Kruczenski,$^{a}$

325:   David Mateos,$^{b}$ Robert C. Myers$\,^{b,c}$ and

326:   David J. Winters$\,^{b,d}$ \\

327:   $^a$ Department of Physics, Brandeis University \\

328:        Waltham, MA 02454, USA \\

329:   $^b$ Perimeter Institute for Theoretical Physics \\

330:        Waterloo, Ontario N2J 2W9, Canada \\

331:   $^c$ Department of Physics, University of Waterloo  \\

332:        Waterloo, Ontario N2L 3G1, Canada \\

333:   $^d$ Department of Physics, McGill University \\

334:        Montr\'eal, Qu\'ebec H3A 2T8, Canada

335: 

336: E-mail: \email{martink@brandeis.edu,

337:   dmateos@perimeterinstitute.ca, rmyers@perimeterinstitute.ca,

338:   winters@physics.mcgill.ca}}

339: 

340: \abstract{We study $\nf$ D6-brane probes in the supergravity

341:   background dual to $\nc$ D4-branes compactified on

342:   a circle with supersymmetry-breaking boundary conditions.

343:   In the limit in which the resulting Kaluza--Klein modes decouple,

344:   the gauge theory reduces to non-supersymmetric,

345:   four-dimensional QCD with $\nc$ colours and $\nf \ll \nc$ flavours.

346:   As expected, this decoupling is not fully realised within the

347:   supergravity/Born--Infeld approximation.

348:   For \mbox{$\nf=1$} and massless quarks, $\mq =0$,

349:   we exhibit spontaneous chiral symmetry breaking by a quark

350:   condensate, $\cc \neq 0$, and find the associated massless `pion' in the

351:   spectrum. The latter becomes massive for $\mq >0$, obeying the

352:   Gell-Mann--Oakes--Renner relation: $\mpi^2 = - \mq \cc/ \fpi^2$.

353:   In the case $\nf >1$ we provide a holographic version of the

354:   Vafa--Witten theorem, which states that the $U(\nf)$ flavour symmetry

355:   cannot be spontaneously broken. Further, we find $\nf^2-1$ unexpectedly light

356:   pseudo-scalar mesons in the spectrum. We argue that these are not

357:   (pseudo-)Goldstone bosons and speculate on the string mechanism

358:   responsible for their lightness. We then study the theory at finite

359:   temperature and exhibit a phase transition associated with a

360:   discontinuity in $\cc (T)$. D6/$\overline{\mbox{D6}}$ pairs are also

361:   briefly discussed.}

362: 

363: 

364: 

365: 

366: \keywords{D-branes, Supersymmetry and Duality}

367: 

368: %\preprint{BRX TH-258}

369: \preprint{BRX TH-258 \\ 

370: \tt{hep-th/0311270}}

371: 

372: \begin{document}

373: 

374: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

375: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

376: %%%%%%%%%%%%%%%%%%%%%% SECTION  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

377: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

378: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

379: \section{Introduction and summary of results} \label{intro}

380: 

381: The AdS$_5$/CFT$_4$ correspondence relates string theory on the

382: near-horizon background of $\nc$ D3-branes to the conformal field

383: theory living on their worldvolume \cite{Maldacena97}. Inspired by

384: this correspondence, Witten proposed a construction of the

385: holographic dual of four-dimensional, pure $SU(\nc)$ Yang--Mills

386: (YM) theory \cite{Witten98b}. One starts with $\nc$ D4-branes

387: compactified on a circle of radius $M_\mt{KK}^{-1}$, and further

388: imposes anti-periodic boundary conditions for the worldvolume

389: fermions on this circle. Before compactification, the D4-brane

390: theory is a five-dimensional, supersymmetric $SU(\nc)$ gauge

391: theory whose field content includes fermions and scalars in the 

392: adjoint representation of $SU(\nc)$, in addition to the gauge

393: fields. At energies much below the compactification scale,

394: $M_\mt{KK}$, the theory is effectively four-dimensional. The

395: anti-periodic boundary conditions break all of the supersymmetries

396: and give a tree-level mass to the fermions, while the scalars also

397: acquire a mass through one loop-effects. Thus at sufficiently low

398: energies the dynamics is that of four-dimensional, massless

399: gluons.

400: 

401: The D4-brane system above has a dual description in terms of

402: string theory in the near-horizon region of the associated (non-supersymmetric) 

403: supergravity background.

404: Unfortunately, as observed in \cite{Witten98b}, the Kaluza--Klein

405: (KK) modes on the D4-brane do not decouple within the supergravity

406: approximation. For example, the mass of the lightest glueball is

407: of the same order as the strong coupling scale. In this sense,

408: there is no energy region, $\Lambda_\mt{QCD} \ll E \ll M_\mt{KK}$,

409: in which a description in terms of weakly-coupled gluons is

410: appropriate. Nevertheless, the qualitative features of the

411: glueball spectrum agree with lattice calculations \cite{COOT98},

412: and the supergravity description suffices, for example, to exhibit

413: an area law for Wilson loops \cite{Witten98b}.

414: 

415: Since most of our phenomenological understanding of low-energy QCD

416: comes from the study of mesons and baryons, it is interesting to

417: construct a holographic dual of a gauge theory whose low-energy

418: degrees of freedom consist not only of gluons but also of

419: fundamental quarks. In the context of AdS/CFT, fundamental matter

420: can be added to the gauge theory by introducing D-brane probes in

421: the dual supergravity background

422: \cite{AFM98,KR01,BDFLM01,BDFM01,KK02, KKW02, WH03, Ouyang03}. This has

423: been recently exploited to study mesons holographically in several

424: examples of gauge/gravity duals \cite{KMMW03, SS03, BEEGK03, NPR03}. 

425: The goal of this paper is to study a gauge/gravity dual in which the

426: gauge theory reduces to non-supersymmetric, four-dimensional QCD

427: in the limit in which the KK modes decouple. The construction is

428: as follows.

429: 

430: Consider the D4/D6 system with the branes oriented as described

431: by the following array:

432: \be

433: \begin{array}{rccccccccccl}

434: \nc \,\, \mbox{D4:}\,\,\, & 0 & 1 & 2 & 3 & 4 & \_ & \_ & \_ & \_ & \_ & \, \\

435: \nf \,\, \mbox{D6:}\,\,\, & 0 & 1 & 2 & 3 & \_ & 5 & 6 & 7 & \_ & \_ & \, .

436: \ea

437: \label{intersection}

438: \ee

439: Note that the D4- and the D6-branes may be separated from each other

440: along the 89-directions. This system is T-dual to the D3/D5

441: intersection, and in the decoupling limit for the D4-branes

442: \cite{IMSY98} provides a non-conformal version of the AdS/dCFT

443: correspondence \cite{KR01, DFO01}. On the gauge theory side one has

444: a supersymmetric, five-dimensional $SU(\nc)$ gauge theory coupled to a

445: four-dimensional defect. The entire system is invariant under

446: eight supercharges, that is, $\caln =2$ supersymmetry in

447: four-dimensional language. The degrees of freedom localized on

448: the defect are $\nf$ hypermultiplets in the fundamental representation

449: of $SU(\nc)$, which arise from the open strings connecting the D4- and

450: the D6-branes. Each hypermultiplet consists of two Weyl fermions

451: of opposite chiralities, $\psi_\mt{L}$ and $\psi_\mt{R}$,

452: and two complex scalars.

453: 

454: As discussed above, identifying the 4-direction with period

455: $2\pi/M_\mt{KK}$, and with anti-periodic boundary conditions for

456: the D4-brane fermions, breaks all of the supersymmetries and

457: renders the theory effectively four-dimensional at energies $E \ll

458: M_\mt{KK}$. Further, the adjoint fermions and scalars become

459: massive. Now, the bare mass of each hypermultiplet, $\mq$, is

460: proportional to the distance between the corresponding D6-brane

461: and the D4-branes. Even if these bare masses are zero, we expect

462: loop effects to induce a mass for the scalars in the fundamental

463: representation. Generation of a mass for the fundamental fermions

464: is, however, forbidden by a chiral, $\ua$ symmetry that rotates

465: $\psi_\mt{L}$ and $\psi_\mt{R}$ with opposite

466: phases.\footnote{This symmetry is broken by instanton

467: effects, but these vanish in the large-$\nc$ limit in which

468: we work.} Therefore, at low energies, we expect to

469: be left with a four-dimensional $SU(\nc)$ gauge theory coupled to

470: $\nf$ flavours of fundamental quark.

471: 

472: In the dual string theory description, the D4-branes are again

473: replaced by their supergravity background. In the so-called `probe

474: limit', $\nf \ll \nc$, the backreaction of the D6-branes on this

475: background is negligible and hence they can be treated as probes.

476: The D6-brane worldvolume fields are dual to

477: gauge-invariant field theory operators constructed with at least two

478: hypermultiplet fields, that is, meson-like operators; of

479: particular importance here will be the quark bilinear operator,

480: $\bar{\psi} \psi \equiv \bar\psi_i \psi^i$, where 

481: $\psi^i = \psi_\mt{L}^i + \psi_\mt{R}^i$ and $i=1, \ldots , \nf$ 

482: is the flavour index. In

483: the string description the $\ua$ symmetry is nothing but the

484: rotation symmetry in the 89-plane. In addition to acting on the

485: fermions as explained above, this symmetry also acts on the

486: adjoint scalar $X=X^8 + i X^9$ by a phase rotation.

487: 

488: Having presented the general construction, we now summarise our main

489: results. 

490: 

491: We begin in section \ref{broken} by considering the case of a single 

492: D6-brane.  For $\mq=0$, we expect the chiral $\ua$ symmetry to be 

493: spontaneously broken by a chiral condensate, \mbox{$\cc \neq 0$}, and there 

494: to be an associated massless, pseudo-scalar Goldstone boson in the 

495: spectrum, as in QCD with one massless flavour. In QCD, this is the 

496: $\eta'$ (which becomes massless in the large-$\nc$ limit), but, in an 

497: abuse of language, we will refer to it as a `pion'. 

498: We are indeed able to show that the string description provides a

499: holographic realisation of this physics, by showing that the $\ua$ 

500: symmetry is spontaneously broken by the brane embedding, as in

501: \cite{BEEGK03}. Moreover, we numerically determine the chiral condensate 

502: for an arbitrary quark mass. We find \mbox{$\cc(\mq=0)\neq 0$}, as expected,

503: and $\cc(\mq) \propto 1/\mq$ for $\mq \ra \infty$, again in agreement 

504: with field theory expectations (which we briefly review).

505:  

506: In sections \ref{fluct} and \ref{pseudo} we study the meson spectrum. 

507: We first show analytically, in section \ref{fluct}, that at $\mq=0$ 

508: there is exactly one normalisable, massless pseudo-scalar 

509: (in the four-dimensional sense) fluctuation of the D6-brane. 

510: This open-string mode corresponds 

511: precisely to the Goldstone mode of the D6-brane embedding. If $\mq>0$, 

512: this pion becomes a pseudo-Goldstone boson. We are able to show, 

513: analytically, that its squared mass scales linearly with the quark

514: mass, $\mpi^2 \propto \mq$, in the limits of small (section

515: \ref{pseudo}) and large (section \ref{fluct}) quark mass. With 

516: numerical calculations we then confirm that, in fact, such a linear 

517: relation holds for all quark masses. 

518: For small $\mq$ this linear relation is in perfect agreement with the 

519: Gell-Mann--Oakes--Renner (GMOR) relation \cite{GMOR68},

520: \be 

521: \mpi^2 = - \fc{\mq \, \cc}{\fpi^2} \,,

522: \label{GMOR} 

523: \ee

524: which gives the first term in the expansion of the pion mass around 

525: $\mq=0$. In section \ref{pseudo}, we compute the chiral condensate and the 

526: pion decay constant analytically (at $\mq=0$), and with these results 

527: we are able to verify that, for small quark mass, the mass of our pion

528: precisely satisfies the GMOR relation. In the opposite limit, 

529: $\mq \ra \infty$, we show in section \ref{fluct} that 

530: \be

531: \mpi^2 = a \, \fc{\mq \mkk}{\gym^2 \nc} \,,

532: \label{linear} 

533: \ee

534: where $a$ is a pure number; for the lightest mesons, we find $a \sim 20$.

535: 

536: The remaining mesons (D6-brane excitations) are studied in section

537: \ref{fluct}. For $\mq \lesssim \msusy$, where 

538: \be

539: \msusy = \gym^2 \nc \mkk \,,

540: \label{msusy}

541: \ee 

542: they all have masses of order the compactification scale, $\mkk$, and 

543: so the (pseudo-)Goldstone boson dominates the infrared physics in the

544: regime of small $\mq$. Of course, this result signals the lack of

545: decoupling between the KK and the QCD scales within the 

546: supergravity/Born--Infeld approximation, as expected. For 

547: $\mq \gtrsim \msusy$, supersymmetry is approximately restored, 

548: the spectrum exhibits the corresponding degeneracy, and all meson

549: masses obey a formula like \eqn{linear} with appropriate values of

550: $a$. The fact that supersymmetry is restored at $\mq \sim \msusy$ 

551: suggests that the effective mass of some of the microscopic degrees 

552: of freedom must be at least of order $\msusy$. We close section 

553: \ref{fluct} with an examination of certain stability issues for the

554: D6-brane embeddings. 

555: 

556: In section \ref{vafa}, for the case of multiple flavours, $\nf >1$,

557: we provide a holographic version of the Vafa--Witten theorem

558: \cite{VW84}, which states that the $U(\nf)$ flavour symmetry

559: cannot be spontaneously broken if $\mq >0$. In the holographic

560: description this is realised by the fact that the $\nf$ D6-branes

561: must be coincident in order to minimise their energy.

562: 

563: In section \ref{fintem}, we examine the theory at finite

564: temperature. If $\mq> \msusy$, the theory exhibits two

565: phase transitions: the first one is the well-known

566: confinement/deconfinement transition at $T = \ct \sim \mkk$

567: \cite{Witten98b}, and the second one is a phase transition at 

568: \be

569: T = \tcc  \sim \sqrt{\fc{\mq \mkk}{\gym^2 \nc}} > \ct \,, 

570: \ee

571: characterised by a discontinuity in the chiral condensate, 

572: $\cc (T)$, and in the specific heat. If $\mq < \msusy$ 

573: the second transition does not occur as a separate phase transition. 

574: 

575: In section \ref{Dbar} we discuss six-brane embeddings 

576: that, from the brane-construction viewpoint, correspond not to

577: D4/D6 intersections but to intersections of D4-branes with

578: D6/$\overline{\mbox{D6}}$ pairs. From the gauge theory viewpoint these

579: configurations correspond to having a defect/anti-defect pair. 

580: 

581: We close with a brief discussion of our results in section

582: \ref{discus}. In particular, one result of the analysis of the

583: $\nf>1$ case in section \ref{vafa} is the existence of $\nf^2$

584: massless, pseudo-scalar mesons in the gauge theory spectrum if $\mq=0$.

585: (We expect the inclusion of sub-leading effects in the $1/\nc$

586: expansion to generate small masses for these particles.) Although

587: it is tempting to interpret these particles as Goldstone bosons of

588: a putative, spontaneously-broken $U(\nf)_\mt{A}$ chiral symmetry,

589: we argue that this interpretation cannot be correct: even at large

590: $\nc$, only one of them is a true Goldstone boson. 

591: Thus we are unable to find any obvious reasons, from the viewpoint of

592: the four- or five-dimensional gauge theory, for these

593: scalars to be light. We speculate that, from the string

594: viewpoint, ten-dimensional $U(\nf)$ gauge invariance (as opposed

595: to just seven-dimensional gauge invariance on the D6-branes) is

596: responsible for their lightness.

597: 

598: 

599: 

600: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

601: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

602: %%%%%%%%%%%%%%%%%%%%%% SECTION  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

603: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

604: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

605: \section{Chiral symmetry breaking from D6-brane embeddings}

606: \label{broken}

607: 

608: In this section we will show how the spontaneous breaking of the

609: $\ua$ chiral symmetry expected from the field theory side is

610: realized in the string description.

611: 

612: 

613: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

614: \subsection{The D4-soliton background}

615: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

616: 

617: The type IIA supergravity background dual to $\nc$ D4-branes

618: compactified on a circle with anti-periodic boundary conditions

619: for the fermions takes the form

620: \begin{eqnarray}

621: ds^{2} &=& \left(\frac{U}{R}\right)^{3/2} \left( \eta_{\mu \nu} \,

622: dx^\mu dx^\nu + f(U) d\tau^{2} \right) + \left(

623: \frac{R}{U}\right)^{3/2} \frac{dU^{2}}{f(U)} +

624: R^{3/2} U^{1/2} \, d\Omega_{\it 4}^{2} \,, \label{metric} \\

625: e^{\phi} &=& g_s \left( \frac{U}{R}\right)^{3/4}

626: \sac F_{\it 4} = \frac{\nc}{V_{\it 4}} \, \varepsilon_{\it 4} \sac

627: f(U) = 1-\frac{\ut^{3}}{U^{3}} \,.

628: \label{metric1}

629: \end{eqnarray}

630: The coordinates $x^\mu=\{ x^0, \ldots , x^3\}$ parametrize the

631: four non-compact directions along the D4-branes, as in

632: \eqn{intersection}, whereas $\tau$ parametrizes the circular

633: 4-direction on which the branes are compactified. $d\Omega_{\it

634: 4}^2$ and $\varepsilon_{\it 4}$ are the $SO(5)$-invariant line

635: element and volume form on a unit four-sphere, respectively, and

636: $V_{\it 4}=8\pi^2/3$ is its volume. $U$ has dimensions of length

637: and may be thought of as a radial coordinate in the

638: 56789-directions transverse to the D4-branes. To avoid a conical

639: singularity at $U=\ut$, $\tau$ must be identified with period

640: \be 

641: \d \tau = \fc{4 \pi}{3} \, \fc{R^{3/2}}{\ut^{1/2}} \,.

642: \label{deltatau} 

643: \ee

644: Following the nomenclature of \cite{soliton}, we refer to this

645: solution as the \nonsol.

646: 

647: This supergravity solution above is regular everywhere and is

648: completely specified by the string coupling constant, $g_s$, the

649: Ramond--Ramond flux quantum (\ie, the number of D4-branes), $\nc$, and 

650: the constant $\ut$. The remaining

651: parameter, $R$, is given in terms of these quantities and the string

652: length, $\ell_s$, by

653: \be 

654: R^3 =  \pi g_s \nc\,\ell_s^3\,. \label{R} 

655: \ee

656: If $\ut$ is set to zero, the solution (\ref{metric},

657: \ref{metric1}) reduces to the extremal, 1/2-supersymmetric

658: D4-brane solution. Hence we may say that $\ut$ characterises the

659: deviation of the \nonsol\ from extremality.

660: 

661: The $SU(\nc)$ field theory dual to (\ref{metric}, \ref{metric1}) is

662: defined by the compactification scale, $\mkk$, below which the

663: theory is effectively four-dimensional, and the four-dimensional

664: coupling constant {\it at} the compactification scale, $\gym$. These are

665: related to the string parameters by

666: \be 

667: \mkk = \fc{3}{2}\fc{\ut^{1/2}}{R^{3/2}} =

668: \fc{3}{2 \sqrt{\pi}}\fc{\ut^{1/2}}{{( g_s \nc)}^{1/2} \ell_s^{3/2}} 

669: \sac \gym^2 =3 \sqrt{\pi} \left( \fc{g_s\ut}{\nc \ell_s} \right)^{1/2} \,.

670: \label{parameters} 

671: \ee

672: The first equation follows from the fact that $\tau$ is 

673: directly identified with the compact direction in the gauge theory,

674: so $\mkk \equiv {2\pi/\d\tau}$. The second equation follows from 

675: the fact that $g_{\it{5}}^2 = (2\pi)^2 g_s \ell_s$ and the relation between 

676: the four- and five-dimensional coupling 

677: constants, $\gym^2 = g_{\it{5}}^2 / \d \tau$, which can be seen  

678: by expanding the Born--Infeld action of the D4-brane and adopting 

679: the standard normalization for the gauge field kinetic term, 

680: $-F^2 / 4\,g_{\it{5}}^2$.

681: 

682: It is useful to invert these relations to express the string

683: parameters in terms of the gauge theory ones:

684: \be

685: R^3 = {1\over2} \fc{\gym^2 \nc \, \ell_s^2}{\mkk} \sac

686: g_s = {1\over2\pi} \fc{\gym^2}{\mkk \ell_s} \sac

687: \ut = {2\over9} \gym^2 \nc \, \mkk \ell_s^2 \,.

688: \label{inverse}

689: \ee

690: The string length will cancel in any calculation of a physical

691: quantity in the field theory. For example, the QCD string tension 

692: is\footnote{This is the tension of a string lying at $U=\ut$

693: and extending along $x$. It can also be computed by projecting such a

694: string onto the boundary and integrating $\langle F^2\rangle$ across

695: a transverse section of the string \cite{adsprop}.}

696: \be

697: \sigma =  \frac{1}{2\pi\ell_s^2} \left. \sqrt{-G_{tt} G_{xx}}

698: \right|_{U=\ut} = \frac{1}{2\pi \ell_s^2}

699: \left(\frac{\ut}{R}\right)^{3/2} = {2\over27\pi} \gym^2 \nc \, \mkk^2 \,.

700: \ee

701: Now we may ask in what situation the \nonsol\  provides a reliable

702: background in which one can study the dual field theory using

703: only classical supergravity. First, we must require that the curvature is

704: everywhere small compared to the fundamental string tension. This

705: ensures that higher derivative string corrections to the low

706: energy equations of motion are negligible. The maximum curvature

707: in the geometry given by \reef{metric} occurs at precisely $U=\ut$,

708: where the curvatures are of the order $(\ut R^3)^{-1/2}$. Hence we

709: require

710: \be

711: {\ut^{1/2} R^{3/2}\over\ell_s^2}  \simeq \gym^2 \nc  \gg 1\,,

712: \label{hoof}

713: \ee

714: where we have applied the results in eq.~\reef{inverse}. Therefore

715: the restriction to small curvatures corresponds to a

716: large 't Hooft coupling in the effective four-dimensional gauge

717: theory, precisely as in the conventional AdS/CFT correspondence

718: \cite{Maldacena97}. Further, to suppress string loop effects, we

719: must also require that the local string coupling, $e^\phi$, be small. Using

720: eqs.~\reef{metric1} and \eqn{inverse} we see that, for finite values 

721: of the gauge theory parameters, the inequality $e^\phi \ll 1$ can

722: only be satisfied up to some critical radius

723: \be

724: U_\mt{crit} \simeq \fc{\nc^{1/3} \mkk \ell_s^2}{\gym^2} \,.

725: \label{crit}

726: \ee

727: Beyond this radius, the M-theory circle opens up to

728: reveal an \adss{7}{4} background (with identifications).

729: Similarly, at the corresponding high-energy scales, the

730: five-dimensional Yang--Mills theory reveals a UV completion in

731: terms of the (2,0) theory compactified on a circle. In the present

732: context, we naturally demand that $U_\mt{crit}\gg\ut$, which,

733: using \eqn{crit} and \eqn{inverse}, reduces to

734: \be 

735: \gym^4\ll{1\over\gym^2\nc}\ll1\,, 

736: \label{small} 

737: \ee

738: where the second inequality follows from eq.~\eqn{hoof}. 

739: Eqs.~\eqn{hoof} and \eqn{small} imply that the supergravity analysis 

740: in the \nonsol\ background is reliable in precisely the

741: strong-coupling regime of the 't Hooft limit of the four-dimensional 

742: gauge theory: $\gym \ra 0$, $\nc \ra \infty$, $\gym^2 \nc$ fixed and

743: large.

744: 

745: 

746: 

747: 

748: 

749: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

750: \subsection{D6-brane embeddings}

751: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

752: 

753: We are now ready to study the embedding of a D6-brane probe in the

754: \nonsol\ geometry. Asymptotically (as $U\rightarrow \infty$), the

755: D6-brane is embedded as described by the array \eqn{intersection}.

756: The analysis is greatly simplified by introducing isotropic

757: coordinates in the 56789-directions. Towards this end, we first

758: define a new radial coordinate, $\rho$, related to $U$ by

759: \beq 

760: U(\rho) = \left(\rho^{3/2} +

761: \frac{\ut^3}{4\rho^{3/2}}\right)^{2/3} \,, 

762: \label{isomer}

763: \eeq

764: and then five coordinates $\vec{z}=(z^5, \ldots, z^9)$ such that

765: $\rho = |\vec{z}|$ and $d\vec{z} \cdot d\vec{z} = d\rho^2 + \rho^2

766: \, d\Omega_{\it 4}^2$. In terms of these coordinates the metric

767: \eqn{metric} becomes

768: \be

769: ds^{2} = \left(\frac{U}{R}\right)^{3/2}

770: \left( \eta_{\mu \nu} \, dx^\mu dx^\nu + f(U) d\tau^{2} \right) +

771: K(\rho) \, d\vec{z} \cdot d\vec{z} \,,

772: \ee

773: where

774: \be K(\rho) \equiv \fc{R^{3/2} U^{1/2}}{\rho^2}\,. 

775: \ee

776: Here $U$ is now thought of as a function of $\rho$. Finally, to

777: exploit the symmetries of the D6-brane embedding we seek, we

778: introduce spherical coordinates $\l, \Omega_{\it 2}$ for the

779: $z^{5,6,7}$-space and polar coordinates $r, \phi$ for the

780: $z^{8,9}$-space. The final form of the D4-brane metric is then

781: \be

782: ds^{2} = \left(\frac{U}{R}\right)^{3/2} \left( \eta_{\mu \nu} \,

783: dx^\mu dx^\nu + f(U) d\tau^{2} \right) + K(\rho) \, \left(

784: d\lambda^2 + \lambda^2 \, d\Omega_{\it 2}^2 + dr^2 + r^2 \,

785: d\phi^2 \right) \,, \label{isometric}

786: \ee

787: where $\rho^2 = \l^2 + r^2$.

788: 

789: In these coordinates the D6-brane embedding takes a particularly

790: simple form. We use $x^\mu$, $\l$ and $\Omega_{\it 2}$ (or $\s^a$,

791: $a=0,\ldots,6$, collectively) as worldvolume coordinates.  The

792: D6-brane's position in the 89-plane is specified as $r=r(\l)$, $\phi

793: = \phi_0$, where $\phi_0$ is a constant. Note that $\l$ is the

794: only variable on which $r$ is allowed to depend, by translational

795: and rotational symmetry in the 0123- and 567-directions,

796: respectively. We also set $\tau=\mbox{constant}$ in the following,

797: which corresponds to a single D6-brane localized in the circle

798: direction. We will study configurations in which this condition is

799: relaxed in section \ref{Dbar}.

800: 

801: With this ansatz for the embedding, the induced metric on the

802: D6-brane, $g_{ab}$, takes the form

803: \be 

804: ds^2(g) = \left(\frac{U}{R}\right)^{3/2} \, \eta_{\mu \nu} \,

805: dx^\mu dx^\nu + K(\rho) \, \left[ \left( 1+\dot{r}^2 \right)

806: d\lambda^2 + \lambda^2 \, d\Omega_{\it 2}^2 \right] \,,

807: \label{induced} 

808: \ee

809: where $\dot{r}\equiv\partial_\lambda r$. The D6-brane action

810: becomes

811: \be

812: S_{D6} = -\fc{1}{(2\pi)^6 \ell_s^7} 

813: \int \df^7\s\, e^{-\phi} \sqrt{-\det g} =

814: -\t6 \int \df^7\s\, \sqrt{h} 

815: \left( 1 + \fc{\ut^3}{4\rho^3} \right)^2 \,

816: \l^2 \sqrt{1 + \dot{r}^2} \,,

817: \label{eq:D6action}

818: \ee

819: where $T_\mt{D6}= 2\pi/g_s(2\pi \ell_s)^7$ is the six-brane

820: tension and $h$ is the determinant of the metric on the round unit

821: two-sphere. Recall that $\rho$ is a function of $\l$ both

822: explicitly and through its dependence on $r(\l)$. Hence the

823: equation of motion for $r(\l)$ is

824: \beq

825: \frac{d}{d\lambda} \left[\left(1+\frac{\ut^3}{4\rho^3}\right)^2

826:   \lambda^2 \frac{\dot{r}}{\sqrt{1+\dot{r}^2}} \right]

827: = -\frac{3}{2} \frac{\ut^3}{\rho^5}

828: \left(1+\frac{\ut^3}{4\rho^3}\right)

829: \lambda^2 \, r \, \sqrt{1+\dot{r}^2} \,.

830: \label{embedeq}

831: \eeq

832: Note that  $r(\l)=r_0$, where $r_0$ is a constant, is a solution in 

833: the supersymmetric limit ($\ut=0$), as in

834: \cite{KK02,KMMW03}. This reflects the BPS nature of the system,

835: which implies that there is no force on the D6-brane regardless

836: of its position in the 89-plane. In particular, then, the solution

837: with $r_0 =0$ preserves the $\ua$ rotational symmetry in the

838: 89-directions. If $\ut\neq 0$ the force on the D6-brane no longer

839: vanishes and causes it to bend as dictated by the equation of

840: motion above. We will see below that in this case there are no

841: (physical) solutions that preserve the $U(1)_\mt{A}$ symmetry.

842: 

843: If $\ut\neq 0$, the analysis is facilitated by rescaling to {\it

844: dimensionless} variables as follows:

845: \beq

846: \l \to \ut\l \sac r \to \ut r \sac \rho \to

847: \ut\rho \,, 

848: \label{tildes}

849: \eeq

850: in terms of which equation \eqn{embedeq} becomes

851: \beq

852: \frac{d}{d\lt} \left[\left(1+\frac{1}{4\rhot^3}\right)^2 \lt^2

853: \frac{\dot{\rt}}{\sqrt{1+\dot{\rt}^2}} \right] = -\frac{3}{2}

854: \frac{1}{\rhot^5} \left(1+\frac{1}{4\rhot^3}\right) \lt^2 \, \rt

855: \, \sqrt{1+\dot{\rt}^2} \,. 

856: \label{resc}

857: \eeq

858: We seek solutions such that the asymptotic separation of the

859: D6-brane and the D4-branes, $L$, is finite. That is, as

860: $\lt\rightarrow\infty$, $\rt(\lt) \rightarrow \rt_\infty$ with

861: \be

862: \rt_\infty = \fc{L}{\ut} \,.

863: \label{L}

864: \ee

865: In the region $\lt\rightarrow\infty$ we have $\dot{\rt}\rightarrow 0$

866: and $\rhot \simeq \lt$. Under these conditions we can linearize equation

867: \eqn{resc} with the result

868: \beq

869: \frac{d}{d\lt} \left[ \left. \lt \right.^2 \dot{\rt} \right]

870: \simeq -\frac{3}{2} \frac{1}{\left.\lt \right.^3}  \rt \,,

871: \eeq

872: the solution to which is

873: \beq

874: \rt(\lt) = A \, \frac{1}{\sqrt{\lt}} J_{-1/3} \left(

875: \frac{2^{1/2}3^{-1/2}}{\left. \lt \right.^{3/2}} \right) + B \,

876: \frac{1}{\sqrt{\lt}} J_{1/3} \left( \frac{2^{1/2}3^{-1/2}}{\left.

877: \lt \right.^{3/2}} \right) \,,

878: \eeq

879: where $J_{\nu}$ are Bessel functions and $A$ and $B$ are arbitrary

880: constants. For large $\lt$, the function multiplied by $A$ tends

881: to a constant, whereas that multiplied by $B$ behaves as

882: $1/\lt$. This means that for large $\l$ we have

883: \be

884: \rt(\l) \simeq \rt_\infty + \fc{c}{\lt} \,,

885: \label{rt}

886: \ee

887: where the constants $\rt_\infty$ and $c$ are related to the quark mass

888: and the chiral condensate, as we now show.

889: 

890: The bare quark mass is given by the string tension times the

891: {\it asymptotic} distance between the D4- and the 

892: D6-brane.\footnote{This mass is trivially derived for the 

893: brane array \reef{intersection} in asymptotically flat space. 

894: With supersymmetry ($\ut =0$), this bare mass persists in the 

895: decoupling limit and is then inherited by the compactified theory 

896: with supersymmetry-breaking boundary conditions, since setting 

897: $\ut \neq 0$ does not alter the asymptotic properties of the model

898: that, as usual in AdS/CFT-like dualities, determine the gauge theory

899: parameters.} 

900: Taking \eqn{L} into account, we have

901: \beq

902: \mq = \frac{L}{2\pi\ell_s^2} = \frac{\ut \, \rt_\infty}{2\pi\ell_s^2} \,.

903: \label{quarkmass}

904: \eeq

905: The quark condensate can now be computed using a simple argument.

906: The Hamiltonian density of the theory can be written as

907: \beq

908: \calh = \calh_0 + \mq \int \df^2\theta \, \tilde{Q}Q \,,

909: \eeq

910: where $\calh_0$ is $\mq$-independent, and $\tilde{Q}$, $Q$ represent

911: the hypermultiplet superfields in $\cN=1$ notation. It follows that

912: \beq

913: \frac{\delta \cale}{\delta \mq} =

914: %\frac{\delta}{\delta \mq} \langle \calh \rangle =

915: \langle  \int \df^2\theta \, \tilde{Q}Q \, \rangle = \cc \,,

916: \eeq

917: where $\cale = \langle \calh \rangle$ is the vacuum energy density. 

918: In the last equality, we have assumed that the vacuum expectation 

919: value of the fundamental scalars vanishes, since they are massive 

920: (even with $\mq=0$ after supersymmetry breaking), that is, the energy 

921: density increases if they acquire a non-zero expectation value.

922: 

923: Since in the string description the quark mass is given by the

924: asymptotic position of the D6-brane, we need to evaluate the change in

925: energy of the D6-brane associated to a change of the boundary

926: condition $\rt_\infty$. In view of (\ref{eq:D6action}) this is given by

927: \beq

928: \delta \cale = - \int \df\lambda \df\Omega_{\it 2} \, \delta \cL =

929: \t6 \ut^3 \int \df\lambda \df\Omega_{\it 2} \, \delta

930: \left[\sqrt{h} \left( 1 + \fc{1}{4\rho^3} \right)^2 \, \l^2 \sqrt{1 +

931: \dot{r}^2} \right] \,,

932: \eeq

933: where $\cL$ is the D6-brane Lagrangian density rescaled in accordance

934: with \reef{tildes}. Using the equations of motion this becomes

935: \beq

936: \delta \cale = -4\pi \left.\left(\d r\frac{\partial}{\partial\dot{r}}

937: \frac{\cL}{\sqrt{h}}\right)\right|^{\l=\infty}_{\l=0} =

938: 4\pi \t6 \ut^3\left. \left(1+\frac{1}{4\rho^3}\right)^2

939: \lambda^2\frac{\dot{r}\, \d r}{\sqrt{1+\dot{r}^2}}

940: \right|^{\l=\infty}_{\l=0} =

941: - 4\pi \t6 \ut^3 \, c\, \delta r_\infty  \,,

942: \eeq

943: where we used \eqn{rt} and the fact that $\dot{r}|_{\l =0}=0$,

944: by rotational symmetry in the 567-directions. In view of

945: \eqn{quarkmass} we finally obtain

946: \beq

947: \cc = \frac{\delta \cale}{\delta m_\mt{q}} =

948: -8 \pi^2 \ell_s^2 \t6 \ut^2 \, c \,.

949: \label{quarkcond}

950: \eeq

951: As anticipated, the quark condensate is directly related to

952: $c$; we will see below that $c$ is determined by $\rt_\infty$.

953: Note that, in terms of gauge theory quantities,

954: \be \fc{1}{\nc} \cc \simeq \gym^2 \nc \mkk^3 c\,. 

955: \ee

956: The normalization on the left-hand side is that expected in the 't

957: Hooft limit.

958: 

959: We now return to the full equation of motion (\ref{resc}), which we

960: will solve numerically. In order to understand the nature of the 

961: solutions better, however, it is

962: helpful to first consider the limit $\rt_\infty \gg 1$, in which

963: (approximate) analytical solutions can be found.

964: 

965: If $\rt_\infty \gg 1$, the entire D6-brane should lie far

966: away from the `bolt' at the center of the \nonsol, hence we 

967: expect the solution to be a small perturbation around the solution

968: for $\ut=0$. In other words, we set $\rt(\lt)=\rt_\infty + \d \rt

969: (\lt)$ and assume that $\d \rt \ll 1$. Substituting this into

970: \eqn{resc} and expanding to linear order in $\d \rt$ we find

971: \beq

972: \frac{d}{d\lt}\left(\left. \lt \right. ^2 \delta\dot{\rt}\right) \simeq

973: -\frac{3}{2} \lt^2 \frac{\rt_\infty}{\left( \left. \lt \right.^2+ \left.

974: \rt \right._\infty^2 \right)^{5/2}} \,,

975: \eeq

976: which is easily integrated with the result

977: \beq

978: \d \dot{\rt} \simeq -\frac{3}{2}\frac{\rt_\infty}{\left. \lt

979: \right.^2} \int_0^{\lt} \df x\, \frac{x^2}

980: {\left( x^2+ \left. \rt \right._\infty^2 \right)^{5/2}} =

981: - \frac{1}{2 \rt_\infty} \, \frac{\lt}

982: {\left( \left. \lt \right. ^2 +

983: \left. \rt \right._\infty^2 \right)^{3/2}} \,.

984: \eeq

985: Here we have imposed the boundary condition $\dot{\rt} = 0$ at

986: $\lt=0$, as required by regularity of the solution at the origin of

987: the 567-space. Integrating once more, with the boundary condition

988: $\d \rt|_{\lt=\infty}=0$,  we find

989: \beq

990: \rt(\lt) \simeq \rt_\infty + \frac{1}{2 \rt_\infty} \,

991: \frac{1}{\sqrt{\left. \lt \right. ^2 + \left. \rt \right._\infty^2}} \,.

992: \label{asymptotic}

993: \eeq

994: From this large-$\rt_\infty$ solution, we see  that the

995: D6-brane bends `outwards', that is, it is `repelled' by the

996: D4-branes. This repulsion persists for arbitrary values

997: of $\rt_\infty$, as is confirmed by numerical

998: analysis. Eq.~\eqn{resc} can be integrated numerically for

999: any value of $\rt_\infty$, and we have plotted solutions for several

1000: values of $\rt_\infty$ in Figure \ref{fig:D6embedding}.

1001: 

1002: \FIGURE{\epsfig{file=profile.eps, height=10cm}

1003: \caption{The D6-brane embedding for several values of the quark

1004:   mass $\mq \propto \rt_\infty$. The interior of the `circle' in the center is

1005:   the region $U<\ut$, which is actually not part of the space.

1006:   The `2' and `3' embeddings

1007:   correspond to the same value of $\rt_\infty$ but have opposite-sign

1008:   $c$'s; the `3' embedding has lower energy.}

1009: \label{fig:D6embedding} }

1010: 

1011: Figure \ref{condensate} displays the function $c(\rt_\infty)$

1012: found by numerical integrations. As mentioned above, the value of

1013: $\rt_\infty$ together with the requirement of regularity at

1014: $\lt=0$ (\ie, $\dot{\rt}|_{\lt=0}=0$) determines the solution

1015: completely. Hence choosing $\rt_\infty$ fixes $c=c(\rt_\infty)$.

1016: Recalling eqs.~\reef{quarkmass} and \reef{quarkcond}, which relate

1017: $\rt_\infty$ and $c$ to the quark mass and condensate,

1018: respectively, we see that this corresponds precisely to what is

1019: expected on field-theoretic grounds: once the quark mass is specified, 

1020: the infrared dynamics determines the chiral condensate.  In the field

1021: theory we expect the $\ua$ symmetry to be spontaneously broken by a

1022: non-zero condensate in the limit $\mq \ra 0$. This is indeed confirmed

1023: by the numerical results for the D6-brane embedding, since we

1024: see that, in figure \ref{condensate}, $c(\rt_\infty)$ approaches a

1025: finite constant in the limit $\rt_\infty \ra 0$. In this limit,

1026: the asymptotic boundary condition on the D6-brane respects the

1027: rotational symmetry in the 89-plane, but it is energetically favourable

1028: for the D6-brane to bend out in the 89-plane. 

1029: Hence the embedding spontaneously breaks this rotational symmetry, 

1030: which matches the $\ua$ symmetry-breaking in the

1031: field theory.

1032: 

1033: We see from \eqn{asymptotic} that $c(\rt_\infty) \sim 1/ (2\rt_\infty)$

1034: for large $\rt_\infty$, which is confirmed by the numerical

1035: results. This implies that the chiral condensate scales as $1/\mq$ for

1036: large $\mq$, as expected on field theory grounds \cite{SVZ79}.

1037: A quick, somewhat heuristic argument for this in QCD is as

1038: follows.\footnote{We thank D.\ T.\ Son for explaining this point to us.}   

1039: The trace of the energy-momentum tensor reads 

1040: \be

1041: T^\mu_{\,\,\,\, \mu} = \mq \bar{\psi} \psi -

1042: \fc{\alpha_s (11 \nc -2 \nf)}{24 \pi} \, \mbox{Tr} F^2 \,,

1043: \ee

1044: where we have explicitly written down the classical contribution, as

1045: well as the anomalous one coming from the one-loop beta-function. 

1046: In the limit $\mq \ra \infty$ the theory resembles QCD with no quarks,

1047: for which 

1048: \be

1049: T^\mu_{\,\,\,\, \mu} = -\fc{11 \nc \alpha_s}{24 \pi} \,\mbox{Tr} F^2  \,.

1050: \ee

1051: Taking vacuum expectation values and equating the two expressions we find 

1052: \be

1053: \cc = \fc{\alpha_s \nf}{12 \pi \mq} \, \langle \mbox{Tr} F^2 \rangle \,.

1054: \ee

1055: Assuming that $\langle \mbox{Tr} F^2 \rangle \neq 0$ we conclude that 

1056: $\cc \propto 1/\mq$.

1057: 

1058: For each value of $\rt_\infty>0$, the numerical analysis actually

1059: finds two regular solutions, which have $c$'s of opposite sign. 

1060: The curves labelled `2' and `3' in figure

1061: \ref{fig:D6embedding} are a representative example of such a pair

1062: of solutions.\footnote{ To describe the second solution (curve 2),

1063: we allow $\rt(\lt)<0$ and restrict $\phi$ to the range $[0,\pi)$.}

1064: We have confirmed that the solutions with positive $c$, for which

1065: $\rt(\lt)>0$ for all $\lt$ (\eg, curve 3), have the lower energy

1066: density. The negative-$c$ solutions bend around the opposite side of

1067: the bolt at the center of the \nonsol\ geometry. Naively, these

1068: solutions should be unstable, since the D6-brane is free to `slide

1069: off the bolt' by moving out of the $\phi=\phi_0$ plane, and relax to the 

1070: lower-energy, positive-$c$ solution. We will

1071: return to this issue in the next section, where we study

1072: fluctuations of the D6-brane worldvolume fields around embedding

1073: solutions of each type --- in fact, we will find tachyonic

1074: fluctuations for the negative-$c$ solutions, confirming the

1075: intuition given above.

1076: 

1077: \FIGURE{\epsfig{file=condensate2.eps, height=10cm}

1078:  \caption{The quark condensate

1079:    $\langle \bar{\psi} \psi \rangle \propto c$ as a function of the

1080:    quark mass $\mq \propto \rt_\infty$.}

1081: \label{condensate}}

1082: 

1083: Note that $r(\l)=0$ is an exact mathematical solution of

1084: eq.~\eqn{embedeq} even if $\ut\neq 0$. Therefore one might be

1085: tempted to conclude that it provides an alternative,

1086: $\ua$-preserving solution whose energy should be shown to be

1087: higher than that of the $\ua$-breaking solution above in order to

1088: establish spontaneous chiral symmetry breaking. However, the

1089: solution $r(\l)=0$ is not physically acceptable if $\ut \neq 0$

1090: because it corresponds to an open D6-brane. That is, this solution

1091: yields a D6-brane with a boundary, which is forbidden by charge

1092: conservation. The boundary is easily seen as follows. With

1093: $r(\l)=0$ we have $\rho =\l$, so this solution terminates at the origin

1094: of the $U\tau$-plane, that is, at $U=\ut$. Since, at this

1095: point, the $S^2$ in the metric \eqn{induced} still has a finite

1096: radius, $R^{3/4}\ut^{1/4}$, the D6-brane would have an 

1097: $\bbr{1,3} \times S^2$-boundary if it terminated at $U=\ut$. Note that if

1098: $\ut=0$ then the two-sphere radius shrinks to zero-size and the

1099: six-brane closes off at $U=0$. Hence the $\ua$-preserving

1100: solution $r(\l)=0$ is physically sensible in the supersymmetric

1101: case. This result is in agreement with the fact that unbroken

1102: supersymmetry forbids a non-zero chiral condensate. Note, however,

1103: that this is only a formal argument, since in the limit $\ut \ra 0$ the

1104: curvature at $U=\ut$ diverges, meaning that the supergravity description

1105: cannot be trusted in this region \cite{IMSY98}.

1106: 

1107: In section \ref{Dbar}, we will see that the solution $r(\l)=0$ can

1108: play a role in the case $\ut \neq 0$ if it is simply extended past

1109: the origin of the $U\tau$-plane. The physical interpretation of

1110: such a solution is not that of a D4/D6 intersection, but rather

1111: that of a D4/D6/$\overline{\mbox{D6}}$ intersection.

1112: 

1113: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1114: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1115: %%%%%%%%%%%%%%%%%%%%%%% SECTION  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1116: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1117: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1118: \section{Meson spectroscopy from D6-brane fluctuations ($\nf =1$)} 

1119: \label{fluct}

1120: 

1121: Here, we consider a certain class of fluctuations of the

1122: D6-brane around the embeddings described in the previous section.

1123: We consider first the positive-$c$ (negative-condensate)

1124: embeddings, which were argued to be stable and so should correspond

1125: to the true `vacuum state' for a given quark mass. For simplicity

1126: we restrict ourselves to fluctuations of the fields $r$ and

1127: $\phi$, as these are sufficient to illustrate the physics we wish

1128: to exhibit. We are therefore considering embeddings of the D6-brane 

1129: of the form

1130: \beq 

1131: \phi=0+\d\phi \sac r=\rvac(\l)+\d r \sac \tau=\mbox{constant}

1132: \,, \label{fluctansatz} 

1133: \eeq

1134: where the fluctuations $\d \phi$ and $\d r$ are functions of 

1135: {\it all} of the worldvolume coordinates and $\rvac(\l)$ is the

1136: numerically-determined vacuum embedding of the previous section.

1137: 

1138: In corroboration of our previous argument for the stability of

1139: these embeddings, we find, in particular, that the spectra of

1140: these fluctuations are non-tachyonic. We also briefly consider

1141: fluctuations around the negative-$c$ (positive-condensate)

1142: embeddings and show their spectra do contain tachyonic modes,

1143: making manifest the instability of these configurations.

1144: 

1145: In the dual gauge theory, the $r$- and the $\phi$-fluctuations

1146: correspond to a class of scalar and pseudo-scalar mesons, 

1147: respectively. To see this, consider the complex scalar

1148: field parametrising the 89-plane, $X = X_8 + i X_9 = r e^{i\phi}$. 

1149: Gauge theory Lorentz transformations act only on the

1150: 0123-directions, under which $X$ is inert, so $X$-fluctuations 

1151: clearly correspond to spin-zero mesons. However, a parity 

1152: transformation in the gauge theory corresponds to a ten-dimensional

1153: transformation that not only reverses the sign of (say) $X_3$, 

1154: but also replaces $X$ by its complex conjugate, $\bar{X}$; this

1155: leaves $r$ invariant but reverses the sign of $\phi$. 

1156: From the gauge theory viewpoint, this is clear from the

1157: presence in the microscopic Lagrangian of a coupling of the form 

1158: $\psi_\mt{R}^\dagger X \psi_\mt{L} + 

1159: \psi_\mt{L}^\dagger \bar{X} \psi_\mt{R}$: the transformation 

1160: $X_3 \ra -X_3$ exchanges $\psi_\mt{R}$ and 

1161: $\psi_\mt{L}$, so this term is only invariant if, simultaneously,

1162: $X \ra \bar{X}$. From the string theory viewpoint, the fact that only

1163: the combination of both transformations is a symmetry can be seen 

1164: by examining the fermionic spectrum of the D4-D6 open strings.

1165: This arises from the zero modes of the world-sheet fermions 

1166: in the Ramond sector, that is, in the NN and DD directions: 

1167: the 23-  and 89-directions (in the light-cone gauge) in 

1168: \eqn{intersection}. Each of these modes is labelled by its weight

1169: under rotations in each of the 23- and 89-planes, $(s_1, s_2), s_i=\pm 1/2$. 

1170: The GSO projection requires $s_1=-s_2$. Since under $X_3 \ra -X_3$ we

1171: have $(s_1, s_2) \ra (-s_1, s_2)$, and under $X_9 \ra -X_9$ we have 

1172: $(s_1, s_2) \ra (s_1, -s_2)$, we see that the spectrum is only

1173: invariant under the combination of both transformations.

1174: 

1175: 

1176: 

1177: 

1178: 

1179: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1180: \subsection{Analysis of the spectrum}

1181: \label{anal}

1182: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1183: 

1184: Using the background metric \eqn{isometric} with the rescalings

1185: \eqn{tildes}, the pullback of \eqn{isometric} to the embedding

1186: given in \eqn{fluctansatz} is

1187: %

1188: \beqa

1189: ds^2&=&\left(\frac{U}{R}\right)^{3/2}\eta_{\mu\nu}dx^\mu dx^\nu+

1190: K[(1+\dot{r}_\mt{vac}^2)d\l^2 + \l^2d\O_{\it 2}^2+

1191: 2\dot{r}_\mt{vac}\pf_a(\d r) d\l dx^a]\non\\

1192: %

1193: &+&K[\pf_a(\d r) \pf_b(\d r) dx^a dx^b+ (r_\mt{vac}+\d

1194: r)^2\pf_a(\d \phi) \pf_b(\d \phi) dx^a dx^b]\ , \label{indfluc}

1195: \eeqa

1196: %

1197: where $a$ and $b$ run over {\it all} of the worldvolume

1198: directions. To quadratic order in fluctuations, the D6-brane

1199: Lagrangian density is then

1200: %

1201: \beqa \call =\call_0&-&\t6 \ut^3 \l^2 \sqrt{h}

1202: \sqrt{1+\dot{r}_\mt{vac}^2}

1203: \Bigg\{\left(\frac{3(7\r_\mt{vac}^2-\l^2)}{16\rho_\mt{vac}^{10}}+

1204: \frac{3(4r_\mt{vac}^2-\l^2)}{4\rho_\mt{vac}^7}\right)(\d r)^2\non\\

1205: %

1206: &&\hspace{2cm}-\frac{3r_\mt{vac}}{4\rho_\mt{vac}^5}\left(1+

1207: \frac{1}{4\rho_\mt{vac}^3}\right) \frac{\dot{r}_\mt{vac} \pf_\l(\d

1208: r^2)}{1+\dot{r}_\mt{vac}^2}\non\\

1209: &&+\left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^2\, \sum_a

1210: \frac{K}{2g_{aa}} \left(\frac{[\pf_a(\d

1211: r)]^2}{1+\dot{r}_\mt{vac}^2}+ r_\mt{vac}^2[\pf_a(\d

1212: \phi)]^2\right)\Bigg\}\ , \label{expandlag} \eeqa

1213: %

1214: where $\call_0$ is the Lagrangian density evaluated for the vacuum

1215: embedding, as in eq.~\eqn{eq:D6action}. Here,

1216: $\rho_\mt{vac}^2=\l^2+r_\mt{vac}^2$ and the $g_{aa}$ in the last

1217: line correspond to the metric coefficients from eq.~\eqn{induced}.

1218: In particular then, they contain no dependence on the

1219: fluctuations. Of course, integration by parts and the equation of

1220: motion \eqn{embedeq} for $r_\mt{vac}$ allowed the terms linear in

1221: $\d r$ to be eliminated.

1222: 

1223: The linearised equations of motion that follow are, for $\d \phi$:

1224: \beqa

1225: &&\fc{9}{4\mkk^2} \frac{\rvac^2}{\rho_\mt{vac}^3}

1226: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^{4/3}\pf_\mu \pf^\mu (\d\phi)

1227: +\frac{\rvac^2}{\l^2}

1228: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^2\nabla^2 (\d\phi) \non\\

1229: &&\hspace{2cm}+\frac{1}{\l^2\sqrt{1+\dot{r}_\mt{vac}^2}}

1230: \frac{d}{d\l}\left[\frac{\l^2 \rvac^2}{\sqrt{1+\dot{r}_\mt{vac}^2}}

1231: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^2\pf_\l (\d\phi)\right]=0 \,,

1232: \label{phi}

1233: \eeqa

1234: and, for $\d r$:

1235: \beqa

1236: &&\frac{\l^2}{\sqrt{1+\dot{r}_\mt{vac}^2}}

1237: \left[\fc{9}{4\mkk^2} \frac{1}{\rho_\mt{vac}^3}

1238: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^{4/3}\pf_\mu \pf^\mu (\d r)

1239: +\frac{1}{\l^2}\left(

1240: 1+\frac{1}{4\rho_\mt{vac}^3}\right)^2\nabla^2 (\d r)\right] \non\\

1241: && \hspace{1cm}+\frac{d}{d\l}\left[\frac{\l^2}{(1+\dot{r}_\mt{vac}^2)^{3/2}}

1242: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^2\pf_\l(\d r)\right]

1243: -\frac{d}{d\l}\left[\frac{3\l^2\rvac\dot{r}_\mt{vac}}

1244: {2\rho_\mt{vac}^5\sqrt{1+\dot{r}_\mt{vac}^2}}

1245: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)\right] \d r \non\\

1246: &&\hspace{4cm}-\l^2\sqrt{1+\dot{r}_\mt{vac}^2}\left(\frac{3(7\rvac^2-\l^2)}

1247: {8\rho_\mt{vac}^{10}}+\frac{3(4\rvac^2-\l^2)}{2\rho_\mt{vac}^7}\right)\d r =0 \,.

1248: \label{r}

1249: \eeqa

1250: %

1251: In these expressions, $\nabla^2$ is the Laplacian on the

1252: two-sphere.

1253: 

1254: From the form of the equations of motion, it is clear that we may

1255: separate variables to write

1256: \beq 

1257: \d \phi=\calp(\l) e^{i k_\phi \cdot x}\, Y_{\ell_\phi

1258: m_\phi}(S^2) \sac \d r=\calr(\l) e^{i k_r \cdot x}\, Y_{\ell_r

1259: m_r}(S^2) \,, 

1260: \eeq

1261: where $Y_{\ell m}$ are the conventional spherical harmonics. We

1262: then look for normalizable solutions with definite $S^2$-angular

1263: momentum $\ell_{\phi,r}$ and four-dimensional mass

1264: $M_{\phi,r}^2=-k_{\phi,r}^2$. For the sake of simplicity in the

1265: following, we only focus on the case $\ell_{\phi,r}=0$ and find

1266: the lowest-mass modes for each fluctuation.

1267: 

1268: We employ the shooting method to solve the linearized equations

1269: of motion. First we use the asymptotic form of the equations  for

1270: $\l\to\infty$ and the requirement of normalizability to determine

1271: the asymptotic boundary conditions for the fluctuations. Then we

1272: vary the four-dimensional mass parameters until, by numerically

1273: integrating the equation of motion, we find solutions that are

1274: also regular at the origin. Recall\footnote{Also recall that we

1275: are working with dimensionless variables $r$ and $\l$, rescaled as

1276: in eq.~\eqn{tildes}.} that the asymptotic behaviour of the vacuum

1277: profile is $\rvac\simeq r_\infty + c/\l$, where $r_\infty$ is

1278: related to the quark mass by equation \eqn{quarkmass}. In this

1279: asymptotic region we have that $\rho_\mt{vac}(\l) \simeq\l$, so

1280: the equation of motion for $\d\phi$ becomes, approximately,

1281: \beq

1282: \frac{9M_\phi^2}{4\mkk^2}\frac{\rvac^2}{\l}(\d\phi)+

1283: \frac{d}{d\l}\left[ \l^2\rvac^2\pf_\l(\d\phi)\right] = 0\,,

1284: \eeq

1285: and similarly that for $\d r$ becomes

1286: \beq

1287: \frac{9M_r^2}{4\mkk^2}\frac{1}{\l}(\d r)+\frac{d}{d\l}

1288: \left[ \l^2 \pf_\l(\d r)\right] = 0\,.

1289: \eeq

1290: It follows

1291: that the asymptotic behaviour of $\d r(\l)$ does not depend on

1292: that of $\rvac(\l)$, whereas that of $\d \phi(\l)$ depends

1293: crucially on whether or not $r_\infty=0$, that is, whether or not

1294: the quarks are massless. We will now show that if $r_\infty=0$

1295: then there exists a normalizable $\d \phi$ mode with $M_\phi=0$,

1296: whereas if $r_\infty \neq 0$ there is a mass gap in the spectrum.

1297: 

1298: \FIGURE{\epsfig{file=spectrum2.eps, height=10cm} 

1299: \caption{Squared masses for the four lowest-lying $\d\phi$ modes 

1300: (dashed lines), dual to pseudo-scalar mesons, and the four 

1301: lowest-lying $\d r$ modes (solid lines), dual to scalar mesons, 

1302: as functions of the quark mass $\mq \propto r_\infty$.}  

1303: \label{squaredscalarmasses} }

1304: 

1305: If we set $r_\infty=0$, we find that the asymptotic equation for

1306: $\d \phi$ becomes

1307: \beq

1308: \frac{9M_\phi^2}{4\mkk^2}\frac{1}{\l^3}(\d\phi)+\pf_\l^2(\d\phi) = 0 \,.

1309: \eeq

1310: The first term is sub-leading for large $\l$, so we may drop it to

1311: find that the asymptotic form of $\d \phi$, regardless of

1312: the value of $M_\phi$, is

1313: \beq

1314: \d\phi\simeq b + a \l \,,

1315: \label{m=0}

1316: \eeq

1317: where $a$ and $b$ are arbitrary constants. For $r_\infty\ne 0$,

1318: however, we find that

1319: \beq

1320: \frac{9M_\phi^2}{4\mkk^2}\frac{1}{\l^3}(\d\phi)+\frac{2}{\l} \pf_\l (\d\phi) +

1321: \pf_\l^2(\d\phi) = 0\,.

1322: \eeq

1323: Again the first term is sub-leading for large $\l$, so the solution is

1324: \beq

1325: \d\phi\simeq \tilde{a}+\tilde{b}/\l \,.

1326: \label{mne0}

1327: \eeq

1328: Normalizable solutions correspond to the boundary conditions

1329: $a=0$ and $\tilde{a}=0$ in the cases $r_\infty=0$ and $r_\infty \neq 0$,

1330: respectively, since the terms with coefficients $b$ and $\tilde{b}$

1331: are the ones that

1332: fall off more rapidly at infinity. Since the differential equations we

1333: are solving are linear, we may choose $b=\tilde{b}=1$ without loss of

1334: generality, so we are left with the boundary conditions

1335: $\{\d\phi=1 \ ,\ \d\phi'=0\}|_{\l\to\infty}$ if $r_\infty=0$, and

1336: $\{\d\phi=1/\l\ ,\  \d\phi'=-1/\l^2\}|_{\l\to\infty}$ if $r_\infty\neq 0$.

1337: A similar analysis for the $\d r$ fluctuations shows that they obey

1338: the same asymptotic equation of motion as the $\d\phi$ fluctuations

1339: in the $r_\infty\ne 0$ case, so the appropriate boundary conditions are

1340: $\{\d r=1/\l\ ,\ \d r'=-1/\l^2\}|_{\l\to \infty}$.

1341: 

1342: \FIGURE{\epsfig{file=linearanddegenerate.eps, height=7cm}

1343: \caption{(a) Linear fit of the lowest-lying $\d\phi$ mode 

1344:   for different values of $\mq \sim \rt_\infty$.

1345:   (b) Degeneracy of the lowest-lying $\d\phi$ (dashed) and $\d r$

1346:   (solid) modes in the supersymmetric limit $r_\infty\gg 1$.}

1347: \label{linearanddegenerate} }

1348: 

1349: Now note that $\d \phi = e^{i k_\phi \cdot x}$, with $k_\phi^2=0$,

1350: is an exact, regular solution of \reef{phi} with zero

1351: four-dimensional mass (and $\nabla^2(\d\phi)=0$), which is

1352: normalisable {\it only} if $r_\infty=0$. This means that with

1353: massless quarks there is a massless, pseudo-scalar meson in the 

1354: spectrum of the

1355: gauge theory, whose dual is the zero-mode fluctuation of the

1356: D6-brane field $\phi$. The mathematical form of these

1357: fluctuations shows that they correspond to rotations of the vacuum

1358: embedding in the 89-plane. Of course, this form is precisely as

1359: expected for the Goldstone mode associated with the spontaneous

1360: breaking of this rotational symmetry.

1361: 

1362: The shooting technique allows us to verify this numerically, and

1363: to find the values of $M_{\phi,r}$ for which other

1364: normalizable, regular solutions exist. The results are the

1365: low-lying (pseudo)scalar meson spectra, which are displayed in Figure

1366: \ref{squaredscalarmasses} for different values of the quark mass.

1367: As anticipated, there is no mass gap in the $\phi$-meson spectrum

1368: when the quark is massless. There is a smooth transition to a

1369: theory with a mass gap as the quark mass is increased and, in

1370: fact, a best-fit (shown in figure \ref{linearanddegenerate}a) 

1371: yields a linear relationship,\footnote{ 

1372: An analogous relationship was observed in \cite{BEEGK03}.} 

1373: $M_\phi^2 \simeq 0.73\, \mkk^2\, r_\infty$, for

1374: small $r_\infty\propto m_\mt{q}$, as expected from the GMOR

1375: relationship \eqn{GMOR}, which we will reproduce analytically 

1376: in the next section.

1377: 

1378: As shown in figure \ref{squaredscalarmasses}, there are no

1379: normalisable, regular solutions of equation \eqn{r} with

1380: $M_{r} =0$, so there is a mass gap in the $r$-meson spectrum

1381: regardless of the value of the quark mass. For small quark mass

1382: the mass scale of all of the mesons, except the lowest-lying $\d

1383: \phi$ mode, is\footnote{This also sets the scale of the glueball

1384: mass spectrum \cite{COOT98}.}

1385: \be

1386: M^2 \sim \mkk^2 \sim \frac{U_\mt{KK}}{R^3}\,.

1387: \label{region1}

1388: \ee

1389: This is another reflection of the lack of decoupling between the QCD

1390: scale and the compactification scale.

1391: 

1392: As suggested by Figure \ref{squaredscalarmasses}, the $r$- and

1393: $\phi$-mesons occur in pairs that become degenerate for large enough

1394: a quark mass. The reason is that, in the supersymmetric case

1395: ($\ut=0$), these mesons belong to the same supermultiplet and hence have

1396: equal masses, as in \cite{KMMW03}. If $\ut \neq 0$ this degeneracy is

1397: only approximate for quark masses much larger than the

1398: supersymmetry-breaking scale, $\msusy$, becoming exact in the limit

1399: $\mq \ra \infty$. Geometrically, this corresponds to the fact

1400: that the distortion of the D6-brane away from

1401: the flat, supersymmetric profile decreases the farther away from the

1402: D4-branes we embed it. Since this distance scale is measured by

1403: $L \propto m_\mt{q}$, and $\ut$ characterises the supersymmetry

1404: breaking, we expect that, for $L \gg \ut$, the characteristic

1405: mass scale of the mesons should become independent of $\ut$ and

1406: depend, instead, on some ratio of $L$ and $R$. To see that this is

1407: indeed the case, we approximate the vacuum profile by the flat

1408: solution in equations \reef{phi} and \reef{r}, that is, we set

1409: $\rvac(\l) = r_\infty$, and we expand these equations assuming that

1410: $r_\infty \gg 1$. In terms of a new variable $y= \l / L$ we obtain

1411: (for zero $S^2$-angular momentum)

1412: \beq

1413: \frac{R^3 M_\phi^2}{L (y^2+1)^{3/2}} \d \phi +

1414: \frac{1}{y^2}\frac{d}{dy}[y^2\pf_y\d\phi] = 0 \ ,

1415: \label{phitrans}

1416: \eeq

1417: and

1418: \beq

1419: \frac{R^3 M_r^2}{L (y^2+1)^{3/2}} \d r +

1420: \frac{1}{y^2}\frac{d}{dy}[y^2\pf_y\d r] = 0 \,.

1421: \label{rtrans}

1422: \eeq

1423: The fact that the two types of fluctuations satisfy the same equation 

1424: confirms that the spectrum becomes degenerate in this limit. It is

1425: also clear that the mass scale for the mesons in this limit is now

1426: \be

1427: M^2 \sim \fc{L}{R^3} \sim \fc{\mq \mkk}{\gym^2 \nc} \,,

1428: \label{region2}

1429: \ee

1430: since this is the only scale that appears in eq. \eqn{phitrans},

1431: \eqn{rtrans}.

1432: 

1433: The transition between the regimes \eqn{region1} and \eqn{region2}

1434: occurs at $L \sim \ut$, and is illustrated for the

1435: lightest $r$- and $\phi$-mesons in figure \ref{linearanddegenerate}b, 

1436: which was generated by solving equations \reef{phi} and \reef{r} 

1437: numerically. The figure clearly illustrates the degeneracy of the

1438: modes in the expected regime. The condition $L \sim \ut$

1439: translates into $\mq \sim \msusy$, with $\msusy$ as defined in

1440: \eqn{msusy}, so $\msusy$ is indeed the scale at which supersymmetry

1441: is restored, as suggested by our choice of notation. Note that at the 

1442: transition point, at which of course eqs. \eqn{region1} and \eqn{region2}

1443: coincide, we have $M \sim \mq / (\gym^2 \nc)$. As we are working in 

1444: the 't Hooft limit of the gauge theory, these mesons are very deeply bound, 

1445: \ie, $M \ll 2\mq$. The suppression of these meson masses by the 

1446: 't Hooft coupling is even stronger than that found in a similar 

1447: context in \cite{KMMW03}, where $M \sim \mq / \sqrt{\gym^2 \nc}$.

1448: For large quark mass there is an extra suppression implicit in

1449: eq. \eqn{region2} due to the fact that $M^2$ scales linearly in $\mq$,

1450: as opposed to quadratically. 

1451: 

1452: As we are working in a range where we expect the gauge theory is

1453: five-dimensional, it is perhaps more appropriate to re-express

1454: the transition scale in terms of the five-dimensional gauge coupling;

1455: it then reads $\mq \sim g_{\it 5}^2\nc \mkk^2$. Similarly, the mass scale

1456: typical of the mesons becomes $M^2 \sim \mq/(g_{\it 5}^2\nc)$. In any

1457: event, the transition scale suggests that the effective mass squared of

1458: some of the microscopic degrees of freedom must be at least of the

1459: order $\msusy \gg \mkk$, rather than just 

1460: $\mkk$. Note that this mass scale is a factor of $\sqrt{\gym^2 \nc}$

1461: larger than that coming from a simple one-loop calculation for the

1462: scalar masses, which yields $\sqrt{\gym^2 \nc} \mkk$. 

1463: This factor seems to be ubiquitous in extrapolations from weak to

1464: strong 't Hooft coupling in the context of AdS/CFT.

1465: 

1466: As we saw above, the squared masses of all

1467: mesons scale linearly in the quark mass as $\mq \ra \infty$. 

1468: For the lightest meson, the results displayed in figure 

1469: \ref{linearanddegenerate}b seem to indicate that, in fact, this linear 

1470: relationship is valid not only in the limits of small and large $\mq$,

1471: but for all quark masses. This result has been 

1472: confirmed to within the best accuracy of our numerical analysis. 

1473: Unfortunately, eq.~\eqn{phitrans} has four regular singular points 

1474: and so no closed-form analytic solution is available, but its

1475: numerical solution gives, for the lowest-mass modes,

1476: \beq

1477: M^2 \simeq 1.66\, \frac{L}{R^3} \simeq 

1478: 20.91\, \frac{\mkk}{\gym^2\nc} \mq \simeq 0.74\, \mkk^2 r_\infty\ .

1479: \eeq 

1480: The final form of this expression agrees with a best-fit of the data plotted

1481: in the region of figure \ref{linearanddegenerate}b with $r_\infty\gg 1$ and, 

1482: in principle, describes the extension of this figure to arbitrarily high values

1483: of $r_\infty$. We therefore conclude that, in our model, the GMOR

1484: linear relation between $M_\pi^2$ and $\mq$ extrapolates to all quark masses.

1485:  

1486: 

1487: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1488: \subsection{Stability Issues}

1489: \label{stable}

1490: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1491: 

1492: Recall that, at the end of section \ref{broken}, we remarked that

1493: our numerical integrations had revealed two classes of six-brane

1494: embeddings, those with positive $c$ and those with negative $c$.

1495: From the numerical analysis, we could show that, for a given

1496: $\rt_\infty$, the positive-$c$ or negative-condensate solutions

1497: had the lower energy density. Further we argued that the

1498: negative-$c$ solutions should be unstable.

1499: 

1500: The stability of the negative-condensate (positive-$c$) vacuum embedding is

1501: reflected in the fact that the spectrum of fluctuations does not

1502: contain tachyonic modes. While numerical searches did not reveal

1503: any tachyonic fluctuations, one can formulate a general argument

1504: that no such modes can exist in the spectrum of

1505: $\d\phi$.\footnote{We expect that these arguments extend to $\d r$

1506: fluctuations but we did not explicitly consider these modes a

1507: possible decay channel between the two D6-brane embeddings, due to

1508: the geometric obstruction provided by the `bolt' at the centre of

1509: the \nonsol.} Consider the equation of motion \eqn{phi} (with zero

1510: angular momentum on the two-sphere, for simplicity). Multiplying

1511: by $\d\phi$, integrating over $\l$ and integrating by parts,

1512: yields

1513: %

1514: \beqa &&\int_0^\infty\df \l\,

1515: \left\{\frac{9M_\phi^2}{4\mkk^2}\frac{\l^2\rvac^2}{\rho_\mt{vac}^3}

1516: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^{4/3}

1517: \sqrt{1+\dot{r}_\mt{vac}^2}\, \d\phi^2

1518: -\frac{\l^2\rvac^2}{\sqrt{1+\dot{r}_\mt{vac}^2}}

1519: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^2

1520: [\pf_\l(\d\phi)]^2\right\}\non\\

1521: &&\hspace{4cm}=\left[\frac{\l^2\rvac^2}{\sqrt{1+\dot{r}_\mt{vac}^2}}

1522: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^2\d\phi\,

1523: \pf_\l(\d\phi)\right]_0^\infty=0\ . \label{integrostability} \eeqa

1524: %

1525: Note that we were able to set the term on the right-hand side to

1526: zero using the behaviour of the normalizable $\d\phi$ modes at

1527: $\l=0$ and $\l=\infty$. On the left-hand side, the second term in

1528: the integrand is manifestly negative or zero, while the sign of

1529: the first term depends on that of $M_\phi^2$. Therefore, for

1530: tachyonic modes it is clear that the integral will be negative-definite,  

1531: which is inconsistent with the vanishing of the

1532: right-hand side. The argument extends to include

1533: nonvanishing $\ell_\phi$, which simply introduces an additional

1534: negative-definite term under the integral. Hence we conclude that

1535: the negative-condensate embeddings are free from tachyonic

1536: instabilities in this sector. As this was the sector with the

1537: lowest lying mode, we might have expected it to be the most

1538: potentially problematic with regards to stability. Hence we are

1539: confident that these positive-$c$ solutions are stable.

1540: 

1541: Now we turn our attention to the negative-$c$ or

1542: positive-condensate embeddings. Recall that our intuition was that

1543: these solutions should be unstable to `sliding off the bolt.'

1544: Since the embedding coordinate ($r$, here) always vanishes at some

1545: value of $\l$ for this case, we cannot address the issue of such

1546: an instability using the $\d\phi$ equation of motion, since $\phi$

1547: is not everywhere well-defined. 

1548: Hence, we change to Cartesian

1549: coordinates $X$ and $Y$ in the $r\phi$-plane. With the choice that

1550: the D6-brane lies at $\phi_\mt{vac}(\l)=\phi_0=0$ (\ie,

1551: $Y_\mt{vac}(\l)=0$), it follows that the embedding profile in $X$

1552: satisfies the equation of motion \reef{resc} and we may set

1553: $X_\mt{vac}(\l)=\rvac(\l)$. Fluctuations in $Y(\l)$ are now the

1554: modes relevant for stability and have the benefit of being

1555: well-defined for all $\l$. One finds the following equation of

1556: motion for these fluctuations:

1557: \beqa

1558: &&\frac{9}{4\mkk^2}\frac{1}{\rho_\mt{vac}^3}

1559: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^{4/3}

1560: \pf_\mu\pf^\mu(\d Y)

1561: +\frac{1}{\l^2}\left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^2

1562: \nabla^2(\d Y)\non \\

1563: &&+\frac{1}{\l^2\sqrt{1+\dot{X}_\mt{vac}^2}}\frac{d}{d\l}

1564: \left[\frac{\l^2}{\sqrt{1+\dot{X}_\mt{vac}^2}}

1565: \left(1+\frac{1}{4\rho_\mt{vac}^2}\right)^2\pf_\l(\d Y)\right]+

1566: \frac{3(1+4\rho_\mt{vac}^3)}{8\rho_\mt{vac}^8}\d Y=0\ ,

1567: \label{Y}

1568: \eeqa

1569: where, now, $\rho_\mt{vac}^2=\l^2+X_\mt{vac}^2$. Making a

1570: stability argument along the lines of that given above for

1571: $\d\phi$ is no longer possible, due to the presence of the final

1572: term in \reef{Y}. This term introduces into the integrand

1573: analogous to that in eq.~\reef{integrostability} a contribution that is

1574: manifestly positive. Hence this approach does not yield a definite

1575: conclusion in this case.

1576: \FIGURE{{\epsfig{file=lowestmodes2.eps, height=7cm}}

1577: \caption{The lowest-lying $\d Y$ mode for some negative-condensate

1578: embeddings (upper line), and the lowest-lying $\d Y$ mode for the

1579: corresponding positive-condensate embeddings (lower line) --- the

1580: $\d Y$ modes are tachyonic.}

1581: \label{tachfig}}

1582: 

1583: However, it is still possible to look directly for tachyonic modes

1584: in the spectrum of fluctuations around the negative-$c$ embeddings

1585: using our numerical techniques. From eq.~\reef{Y} one can deduce

1586: the asymptotic behaviour of the $\d Y$ modes to be

1587: \beq 

1588: \d Y\simeq \tilde{a}+\tilde{b}/\l\ , 

1589: \eeq

1590: independent of the vacuum embedding. Using the shooting method as

1591: before, the spectrum of each positive-condensate embedding

1592: considered numerically is found to contain a tachyonic mode and,

1593: furthermore, these modes become `more tachyonic' as the quark mass

1594: increases --- see figure \ref{tachfig}. As is suggested in the

1595: figure, the squared mass of the lowest-lying $\d Y$ mode has a

1596: linear dependence on the quark mass. In fact, a best-fit of the

1597: data gives $M_Y^2\simeq -0.72\, \mkk^2 r_\infty$.

1598: 

1599: An examination of the next-to-lowest-lying $\d Y$ mode (not shown

1600: in the figure) indicates that the mass of this mode, which is real

1601: and positive for $m_\mt{q}=0$, also decreases as a function of

1602: $r_\infty$. So for sufficiently large quark mass the spectrum may

1603: contain more than one tachyon. However, we could not confirm this

1604: directly because for large $r_\infty$ the negative-$c$ embeddings

1605: approach extremely close to the bolt, and our computer code

1606: behaved erratically in this situation.

1607: 

1608: 

1609: 

1610: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1611: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1612: %%%%%%%%%%%%%%%%%%%%%% SECTION  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1613: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1614: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1615: \section{The pseudo-Goldstone boson} \label{pseudo}

1616: 

1617: As discussed in the previous section, if $\mq=0$ the spectrum contains

1618: a massless, pseudo-scalar mode that can be understood as the Goldstone boson

1619: of the spontaneously broken $\ua$ symmetry. This is akin to the

1620: $\eta'$ meson in QCD, which is massless in the large-$\nc$ limit. In an

1621: abuse of language, however, we will refer to this Goldstone boson as a

1622: `pion'. For non-zero quark mass, the $\ua$ symmetry is explicitly broken

1623: and the pion becomes a pseudo-Goldstone boson with a mass $\mpi$

1624: that, according to our numerical analysis, grows as $\mpi^2 \sim \mq$.

1625: This means that at low energies, $E\ll \Lqcd$, the dynamics is

1626: dominated by this particle. In this section we will provide an

1627: analytic proof, using the string description, that the pion's mass

1628: actually obeys the well-known Gell-Mann--Oakes--Renner relation

1629: \eqn{GMOR}.

1630: 

1631: Recall that the gauge theory pion is dual to a $\phi$-fluctuation

1632: of the D6-brane of the form $\delta \phi = e^{ik \cdot x} \vp (\l)$

1633: that solves equation \eqn{phi} with $\nabla^2 (\d \phi)=0$, that is,

1634: \beq

1635: \frac{d}{d\l} \left[ p(\lambda)\dot{\vp} \right] =

1636: -M^2 \mu(\lambda) \vp\ ,

1637: \label{eq:eqvp}

1638: \eeq

1639: where $M^2 = -k^2$ is the four-dimensional mass,

1640: \beqa

1641: p(\lambda) &=& \lambda^2 \rvac^2

1642: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^2

1643: \frac{1}{\sqrt{1+\dot{r}_{\mt{vac}}^2}} \,,

1644: \label{pdef}

1645: \\ \mu(\lambda) &=& \lambda^2 \frac{9}{4\mkk^2}\frac{\rvac^2}{\rho_\mt{vac}^3}

1646: \left(1+\frac{1}{4\rho_\mt{vac}^3}\right)^{4/3}

1647: \sqrt{1+\dot{r}_{\mt{vac}}^2} \,,

1648: \label{pmu}

1649: \eeqa

1650: and $\rho_\mt{vac}^2 = \lambda^2 +\rvac^2$. 

1651: The allowed solutions

1652: are those that both satisfy $\dot{\vp}|_{\l =0} = 0$

1653: and are normalizable with respect to the scalar product

1654: \beq

1655: \langle \vp_1 |  \vp_2 \rangle = \int_0^\infty \df\l\, \mu(\lambda)

1656: \vp^*_1(\lambda) \vp_2(\lambda) \,.

1657: \eeq

1658: This defines an eigenvalue problem whose solution is

1659: a complete set of orthogonal

1660: eigenfunctions $\vp_n$ with eigenvalues $M^2_n$. Notice that

1661: if $\vp(\l)= \mbox{constant}$ then the eigenvalue equation

1662: is satisfied with $M^2=0$, but the mode is normalizable only if

1663: $\rvac(\l) \ra 0$ as $\lambda\rightarrow\infty$.

1664: If $\vp(\l)$ is not constant, multiplying eq.(\ref{eq:eqvp}) by $\vp$,

1665: integrating both sides with respect to $\l$ and performing an integral by

1666: parts in the left-hand side, we arrive at

1667: \beq

1668: M^2 = \frac{\int_0^{\infty} \df\l\, p(\l) \dot{\vp}^2}

1669: {\int_0^{\infty} \df\l\, \mu(\l) \vp^2} >  0\ .

1670: \eeq

1671: Since from (\ref{pdef}) and (\ref{pmu}) we have $p(\l)>0$ and

1672: $\mu(\l)>0$, it follows that $M^2$ is positive, as indicated.\footnote{

1673: This is not valid when $\rvac$ vanishes at some point (which

1674: is not the case studied here) --- see the more detailed discussion

1675: after equation (\ref{integrostability}).}

1676: 

1677: Having said that, we now want to compute the mode with lowest

1678: eigenvalue $M^2>0$ when $m_\mt{q}$ is non-vanishing but small, that

1679: is, let us suppose that $\rvac\rightarrow r_\infty$ as

1680: $\lambda\rightarrow\infty$, with $r_\infty$ small but non-zero. In

1681: this case we expect that the normalizable eigenmode can be obtained

1682: from the massless-quark zero-mode, namely $\vp(\l)=\mbox{constant}$,

1683: by using perturbation theory. To this end, it is convenient to define

1684: \beq

1685: \psi(\lambda) = \sqrt{p(\lambda)} \, \vp (\l) \,,

1686: \eeq

1687: which satisfies the new eigenvalue equation

1688: \beq

1689: \ddot{\psi} - \frac{\ddot\Psi}{\Psi} \psi = -M^2\, \nu \, \psi \,,

1690: \label{psieqn}

1691: \eeq

1692: where $\Psi=\sqrt{p}$, and $\nu=\mu/p$ is the new integration

1693: measure. If $r_\infty=0$ then $\psi(\l)=\Psi(\l)$ is a normalizable

1694: solution of \eqn{psieqn} with zero eigenvalue. Making a small change in

1695: the boundary condition, $r_\infty\to\d r_\infty\gtrsim 0$, induces a small

1696: change in the vacuum profile $\rvac(\l)$,

1697: which in turn induces a small change in the functions $\Psi(\l)$ and

1698: $\nu(\l)$ appearing in equation \eqn{psieqn}. Standard

1699: quantum mechanics perturbation theory can then be applied to obtain the

1700: new lowest eigenvalue.

1701: 

1702: Let us denote with a `bar' the quantities corresponding to

1703: $r_\infty=0$, so that for $r_\infty\neq 0$ we have

1704: %

1705:  \beqa

1706: \Psi &=& \bar{\Psi} + \delta \Psi \,, \\

1707: \nu &=& \bar{\nu} + \delta \nu \,,

1708: \eeqa

1709: %

1710: where $\delta \Psi$ and $\delta \nu$ are the differences induced

1711: by the change in $\rvac(\l)$. The new lowest eigenvalue and its

1712: associated wave-function, $M^2 \gtrsim 0$ and

1713: $\psi =\bar{\Psi} + \delta \psi$, obey, to leading order,

1714: \beq

1715: \delta \ddot\psi -\frac{\ddot{\bar{\Psi}}}{\bar{\Psi}} \, \delta \psi

1716: -\delta\left(\frac{\ddot\Psi}{\Psi}\right) \bar{\Psi} =

1717: - M^2 \, \bar{\nu} \, \bar{\Psi} \,.

1718: \eeq

1719: Decomposing $\delta \psi $ as a linear combination

1720: of massless-quark eigenfunctions,

1721: $\delta\psi = \sum_{n=0}^\infty \alpha_n \bar{\psi}_n $, we obtain

1722: \beq

1723: \sum_{n=1}^\infty \alpha_n \bar{M}^2_n \bar{\nu} \bar{\psi}_n +

1724: \delta\left(\frac{\ddot\Psi}{\Psi}\right) \bar{\Psi} =

1725: M^2 \, \bar{\nu} \, \bar{\Psi} \,,

1726: \eeq

1727: %

1728: where we have dropped the first term in the series, using

1729: $\bar{M}_0=0$. Multiplying by $\bar{\Psi}$, integrating over

1730: $\lambda$ and using the orthogonality condition $\int_0^{\infty}

1731: \df\l\, \bar\nu \bar{\Psi} \bar{\psi}_n =0$ for $n>0$ (since

1732: $\bar{\Psi}$ is the zero-mode eigenfunction and is real) we deduce

1733: that

1734: %

1735: \beq

1736: M^2 \int_0^{\infty} \df\l \, \bar{\nu}\, \bar{\Psi}^2 =

1737: \int_0^\infty \df\l \, 

1738: \delta\left(\frac{\ddot\Psi}{\Psi}\right) \bar{\Psi}^2 =

1739: \int_0^\infty \df\l \, 

1740: (\bar{\Psi}\delta\ddot{\Psi} -\ddot{\bar{\Psi}}\delta\Psi) =

1741: \left[ \bar{\Psi} \delta \dot\Psi - \dot{\bar{\Psi}} \delta \Psi

1742: \right]_{\l=0}^{\l=\infty} \,.

1743: \label{m2}

1744: \eeq

1745: Recalling that

1746: \be

1747: \bar{\Psi} = \lambda \bar{r}_\mt{vac}

1748: \left(1+\frac{1}{4\bar{\rho}_\mt{vac}^3}\right)

1749: (1+\dot{\bar{r}}_\mt{vac}^2)^{-1/4} \,,

1750: \ee

1751: which implies

1752: \bea

1753: \delta \Psi &=& \lambda \, \delta r_\mt{vac}

1754: \left(1+\frac{1}{4\bar{\rho}_\mt{vac}^3}\right)

1755: (1+\dot{\bar{r}}_\mt{vac}^2)^{-1/4} -

1756: \frac{3\lambda \bar{r}^2_{\mt{vac}}}{4\bar{\rho}_\mt{vac}^5}

1757: (1+\dot{\bar{r}}_\mt{vac}^2)^{-1/4} \, \delta \rvac - \nonumber \\

1758: && - \l\, \bar{r}_\mt{vac} \left(1+\frac{1}{4\bar{\rho}_\mt{vac}^3}\right)

1759: \frac{1}{2}(1+\dot{\bar{r}}_\mt{vac}^2)^{-5/4} \dot{\bar{r}}_\mt{vac} \,

1760: \delta\dot{r}_\mt{vac} \,,

1761: \eea

1762: we see that the only contribution to the last term in equation

1763: \eqn{m2} comes from $\l=\infty$. This is easily evaluated, since for

1764: large $\l$ we have

1765: \be

1766: \bar{r}_\mt{vac} \simeq \frac{\bar{c}}{\lambda} +

1767: \calo \left( \frac{1}{\lambda^2}\right) \sac

1768: \bar{\Psi} \simeq \l\bar{r}_\mt{vac} \simeq \bar{c} +

1769: \calo \left( \frac{1}{\l}\right) \,.

1770: \ee

1771: It follows that the new eigenvalue is given by

1772: \beq

1773: M^2 = \frac{\bar{c}\, \d r_\infty}

1774: {\int_0^\infty \df\l\, \bar{\nu} \bar{\Psi}^2} =

1775: \frac{\bar{c} \, \d r_\infty}{\int_0^\infty \df\l\, \bar{\mu}} \,.

1776: \label{GMORsugra}

1777: \eeq

1778: To show that this formula is precisely the GMOR relation

1779: \eqn{GMOR}, we need to identify the pion decay constant in the

1780: string description. In this description, a shift of the chiral

1781: angle that parametrizes the space of vacua related to each other

1782: by the $\ua$ symmetry corresponds to a rigid rotation of the

1783: D6-brane field $\phi$ (both take values between 0 and $2\pi$).

1784: Similarly, the pion field is identified with the normalizable,

1785: massless mode $\d\phi$, described in section \ref{anal}. The pion

1786: decay constant can be read off from the normalization of the

1787: kinetic term in the four-dimensional low-energy effective

1788: Lagrangian for this mode. That is, we integrate the $\d\phi$-terms

1789: in eq.~\eqn{expandlag} over $\l$ and the coordinates on $S^2$ and

1790: as a result find a four-dimensional action of the form

1791: %

1792: \be S =  -\fc{\fpi^2}{2}\int \df^4x\, \pf_\mu (\d{\phi})

1793: \pf^\mu(\d{\phi})\ , \ee

1794: %

1795: where $f_\pi$ is the pion decay constant. From

1796: eq.~\reef{expandlag}, then we have

1797: %

1798:  \beqa

1799:  \fpi^2 &=& \t6 \ut^3

1800: \int\df\Omega_{\it 2}\, \sqrt{h} \int_0^\infty \df\l\,

1801: \l^2\frac{9}{4\mkk^2}\frac{\bar{r}^2_\mt{vac}}{\bar{\rho}^3_\mt{vac}}

1802: \left(1+\frac{1}{4\bar{\rho}^3_\mt{vac}}\right)^{4/3}

1803: \sqrt{1+\dot{\bar{r}}^2_\mt{vac}}\non\\

1804: & =& 4\pi \t6 \ut^3 \int_0^\infty

1805: \df\l \, \bar{\mu}(\l)\ . \label{fpi} 

1806: \eeqa

1807: %

1808: Combining this result with eqs.~\eqn{quarkmass} and

1809: \eqn{quarkcond}, we see that eq.~\eqn{GMORsugra} is precisely the

1810: GMOR relation \eqn{GMOR}, as anticipated.

1811: 

1812: We close this section by verifying the agreement between the

1813: analytical expression \eqn{GMORsugra} above and the results obtained 

1814: purely by numerical methods and displayed in figure 

1815: \ref{linearanddegenerate}a. The integral in equation \eqn{GMORsugra} 

1816: can be evaluated numerically with the result

1817: \beq

1818: \int_0^\infty \df\l \, \bar{\mu}(\l) = \frac{9}{4\mkk^2} \int_0^\infty \df\l \,

1819: \frac{\lambda^2\bar{r}_\mt{vac}^2}{\bar{\rho}_\mt{vac}^3}

1820: \left(1+\frac{1}{4\bar{\rho}_\mt{vac}^3}\right)^{4/3}

1821: \sqrt{1+\dot{\bar{r}}_\mt{vac}^2} \simeq \frac{0.97}{\mkk^2} \,.

1822: \label{muint}

1823: \eeq

1824: Similarly, the numerical result for the constant $c$ for zero

1825: quark mass (see Figure \ref{condensate}) is $\bar{c} \simeq 0.71$.

1826: Substituting these two values in equation \eqn{GMORsugra} we get

1827: $M^2 \simeq 0.73\, \mkk^2 \d\rt_\infty$, in exact agreement with a best-fit

1828: of the points in figure \ref{linearanddegenerate}a.

1829: 

1830: Recall from section \ref{broken} that for nonvanishing quark mass

1831: we found that there are two possible values of the quark

1832: condensate, differing in sign, but that only the negative condensate is

1833: stable. In section \ref{stable} the positive-condensate solution

1834: was explicitly shown to be unstable by showing that the

1835: pseudo-Goldstone mode $\d Y$ was tachyonic for this configuration.

1836: Since the field-theoretic arguments establishing the GMOR relation

1837: apply regardless of the sign of the quark condensate, the squared 

1838: mass of this mode should also obey the GMOR relation, at least

1839: approximately, with an appropriate sign change. Recall that 

1840: eq. \reef{GMOR} represents the first term in an expansion

1841: around $\mq=0$. Now in the string description at $\mq=0$, the 

1842: D6-brane configurations for positive and negative $c$ are

1843: identical, except for a rotation of $\phi$ by $\pi$. Hence up to a

1844: crucial sign, the numerical coefficients in eq.~\reef{GMORsugra}

1845: are unchanged and hence the tachyonic mode should satisfy

1846: $M_Y^2 \simeq -0.73\, \mkk^2 \d r_\infty$. From Figure

1847: \ref{tachfig}, it does indeed appear that the slopes of the lowest-lying 

1848: $\d\phi$ and $\d Y$ modes are equal, but opposite in sign. More precisely, 

1849: our best-fit of the numerical results for negative $c$ gave 

1850: $M_Y^2\simeq -0.72\, \mkk^2 r_\infty$, which gives agreement to within 

1851: about 1.4\%.

1852: 

1853: 

1854: 

1855: 

1856: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1857: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1858: %%%%%%%%%%%%%%%%%%%%%% SECTION  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1859: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1860: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1861: \section{Multiple flavours ($\nf >1$) and a holographic Vafa--Witten theorem}

1862: \label{vafa}

1863: 

1864: Consider a vector-like gauge theory with vanishing $\theta$-angle

1865: (like QCD) and $\nf$ flavours, all with identical mass $\mq>0$.

1866: The Vafa--Witten theorem states that the vector-like $U(\nf)$

1867: flavour symmetry cannot be spontaneously broken \cite{VW84}. This

1868: theorem does not exclude the possibility that, at $\mq =0$, the

1869: $U(\nf)$-preserving vacuum becomes degenerate with one or more

1870: $U(\nf)$-breaking vacua. In this section we discuss the

1871: holographic realization of the Vafa--Witten theorem, within the

1872: approximations implied by the validity of the

1873: supergravity/Born--Infeld theory. In particular, we are

1874: considering $\nc \ra \infty$ and $\nf \ll \nc$. Note that,

1875: strictly speaking, the Vafa--Witten theorem need not apply to the

1876: gauge theory on the D4/D6-brane system because of the presence of

1877: Yukawa-like couplings between the fundamental fermions and the

1878: adjoint scalars. However, the latter scalar fields are massive and

1879: the theorem should certainly apply in a limit where the scalar

1880: masses are sent to infinity. Further, one expects the same result

1881: for large but finite masses since the vacuum should still not be

1882: sensitive to the scalars. In the present case, the scalar masses

1883: are likely of order $\mkk \sim \Lqcd$ and so these couplings would

1884: not be suppressed. Hence, from the gauge theory viewpoint, 

1885: it is not clear a priori whether the effect of these fields can 

1886: change the vacuum structure to the extent of violating the Vafa--Witten

1887: theorem.\footnote{We thank A.\ Nelson for a discussion on this

1888: point.} The results of this section answer this question

1889: in the negative.

1890: 

1891: The $U(\nf)$ global symmetry of the gauge theory corresponds, in

1892: the string description, to the $U(\nf)$ gauge symmetry on the

1893: worldvolume of $\nf$ D6-brane probes in the geometry \eqn{metric}.

1894: Their dynamics is described by the so-called non-Abelian

1895: Dirac--Born--Infeld (DBI) action \cite{Myers99}. This is a

1896: generalisation of the single-brane Abelian theory

1897: \eqn{eq:D6action}, in which both the gauge fields and the

1898: transverse scalars are promoted to Hermitian, $U(\nf)$ matrices.

1899: This means, in particular, that the positions of the D6-branes are

1900: no longer commuting quantities in general. In the particular cases

1901: in which the transverse scalar matrices can be simultaneously

1902: diagonalised, each of their $\nf$ eigenvalues can be interpreted

1903: as the position of one of the D6-branes along the corresponding

1904: transverse direction.

1905: 

1906: For the $U(\nf)$ global symmetry to be realized in the gauge

1907: theory, the mass matrix for the fundamental quarks must be

1908: diagonal with all eigenvalues equal to $\mq$. In our string

1909: picture, this translates into the boundary condition that,

1910: asymptotically, all the D6-branes should have the same asymptotic

1911: boundary conditions. That is, they are all aligned with each other

1912: in the ($z^5,\ldots,z^9$) subspace and all lie at a distance 

1913: $2\pi \ell_s^2 \mq$ from the origin. We would like to show that

1914: the configuration (which obviously satisfies the boundary

1915: condition above) in which all the D6-branes are exactly

1916: coincident, and are embedded exactly as in the case of a single

1917: D6-brane, is the minimum-energy solution of the equations of

1918: motion that follow from the non-Abelian DBI action.

1919: 

1920: To avoid problems with the coordinate singularity at $r=0$, we

1921: adopt the Cartesian coordinates $X$ and $Y$ introduced in section

1922: \ref{stable} to describe the transverse positions of the D6-branes

1923: in the $r\phi$-plane. Then the configuration for the $\nf$

1924: coincident six-branes may be written as:

1925: \be 

1926: X(\l) = \rvac (\l) \cdot \bbi{} \sac Y(\l) = 0 \sac \tau =

1927: \tau_0 \cdot \bbi{} \,, 

1928: \label{conf} 

1929: \ee

1930: where $\rvac (\l)$ is the profile for a single D6-brane determined

1931: in section 2 and $\bbi{}$ is the $\nf \times \nf$ identity matrix.

1932: Since all the transverse scalars commute, there is no ambiguity in the

1933: interpretation of this solution as $\nf$ overlapping D6-branes

1934: lying along the curve $X(\l) = \rvac (\l)$.

1935: 

1936: Let us first argue that this configuration is a stable solution of

1937: the non-Abelian DBI equations of motion. We start by expanding the

1938: non-Abelian DBI action to quadratic order in fluctuations around

1939: the above configuration. Since there is an overall single trace in

1940: front of the action, to linear order only variations proportional

1941: to the identity generator couple to the $U(1)$ background fields,

1942: as given above. Hence the resulting expression for these fields is

1943: exactly that obtained from the linear variation of the action for a

1944: single D6-brane. It thus follows that the configuration \eqn{conf}

1945: is a solution of the non-Abelian equations provided $\rvac (\l)$

1946: solves the corresponding Abelian equations.

1947: 

1948: We further observe that, to quadratic order in fluctuations, the

1949: non-Abelian DBI action reduces to the sum of $\nf^2$ identical

1950: copies of the Abelian action, one for each of the $\nf^2$

1951: independent fluctuations. The reasons for this are two-fold:

1952: First, given that the background fields in \eqn{conf} are

1953: proportional to the identity, the quadratic fluctuations must

1954: again contribute a term in the overall $U(1)$. Second, the

1955: difference between the non-Abelian and the Abelian DBI actions

1956: consists entirely of terms that involve commutators of fields.

1957: Hence, with the given background, the only possible nonvanishing

1958: commutators will involve two fluctuations, but such expressions

1959: are necessarily $SU(\nf)$-valued, \ie, traceless. It follows that 

1960: upon tracing over the gauge indices (and integrating by parts), the

1961: quadratic action will take the form

1962: \be 

1963: \int \df^7\s\,\sum_{\alpha=1}^{\nf^2} \d Z^{i\a}\, {\cal O}_{ij}\,

1964: \d Z^{j\a}\ , 

1965: \label{house} 

1966: \ee

1967: where $Z^i=X, Y$, $\a$ is the gauge index and ${\cal O}_{ij}$ 

1968: is the same operator as appears in the

1969: quadratic expansion of the Abelian action. Thus the linearized

1970: equations of motion, and hence the spectrum, are just $\nf^2$ copies

1971: of those in the Abelian theory describing a single D6-brane

1972: embedding. It follows that the solution \eqn{conf} is stable

1973: provided its Abelian analogue is stable, which we know to be the

1974: case from the analysis of section \ref{fluct}.

1975: 

1976: Note that these $\nf^2$ fluctuations of the D6-branes, which give

1977: rise to the $\nf^2$ copies of the spectrum, are distinct

1978: physical excitations and {\it not} gauge-equivalent: each of them

1979: is associated to a different generator of $U(N_f)$, so there is no

1980: small $U(N_f)$ gauge transformation (that is, one that tends to

1981: the identity at infinity in the 0123-directions) that takes one

1982: mode into another. In particular, this implies that in the limit

1983: $\mq \ra 0$ there are $\nf^2$ massless, pseudo-scalar particles! 

1984: We will come back to this point in the last section.

1985: 

1986: The arguments presented so far show that \eqn{conf} provides a stable

1987: extremum of the non-Abelian DBI action, and therefore a local minimum

1988: of the energy of the $\nf$ D6-branes. Strictly speaking, this

1989: does not exclude the possibility that there exist other minima. If this was

1990: the case, then the true vacuum (or vacua) would be that (or those)

1991: with the lowest energy. On physical grounds, however, the existence of

1992: these other vacua is unlikely. The reason is as follows. If we ignore the

1993: interactions of the D6-branes with each other, then it is clear that

1994: \eqn{conf} is the minimum-energy solution amongst those satisfying the

1995: boundary conditions described above, since in the absence of

1996: interactions each D6-brane must minimise its own energy. If we

1997: include open string tree-level interactions by describing the

1998: D6-branes' dynamics with the non-Abelian DBI action,

1999: then the only obvious configuration that satisfies the appropriate

2000: boundary conditions is \eqn{conf}, which we have shown to

2001: be a stable solution. Inclusion of one-loop (or higher) open string

2002: interactions between the D6-branes (such as exchange of closed

2003: strings) would presumably just strengthen the conclusion that

2004: \eqn{conf} is the preferred solution. To see this, consider the simple

2005: case of $\nf=2$ D6-branes that overlap with each

2006: other asymptotically, but that are slightly misaligned in some

2007: small region. Locally, this situation is analogous to that of two

2008: planar D6-branes in flat space that are almost but not exactly

2009: parallel, in which we know that there is a net force that tends to

2010: align the branes. It seems plausible that a similar force

2011: will act on the branes in the situation of interest here, and

2012: therefore that the preferred configuration will be \eqn{conf},

2013: in which all the branes are perfectly aligned with each other.

2014: 

2015: This conclusion provides a holographic realisation of the

2016: Vafa--Witten theorem, since the $U(\nf)$ gauge symmetry on the

2017: D6-branes is unbroken if and only if they are all exactly

2018: coincident.

2019: 

2020: By continuity, the arguments above imply that, in the limit $\mq

2021: \ra 0$, the $U(\nf)$-preserving configuration has an energy no

2022: greater than that of any other configuration. Just like in the

2023: Vafa--Witten theorem, however, this does not rule out the

2024: possibility that in this limit some other state becomes degenerate

2025: with that in eq.~\eqn{conf}. In fact, the appearance of

2026: $\nf^2$ massless modes in the linearized spectrum at $\mq=0$ suggests 

2027: that the D6-branes might be free to rotate independently in the

2028: 89-plane.\footnote{Note that this is not

2029:   guaranteed by the existence of massless modes, since this does not 

2030:   exclude the appearance of higher order potential terms for the 

2031:   fluctuations, \eg\ $\delta Z^4$.}

2032: To see that this is indeed the case, consider, for simplicity, the 

2033: $\nf=2$ configuration 

2034: \be 

2035: X(\l) = \rvac (\l) \cos\phi_0 \cdot \bbi{} \sac  

2036: Y(\l) = \rvac (\l)

2037: \sin\phi_0 \cdot \sigma_{\it 3} \sac \tau = \tau_0 \cdot \bbi{}

2038: \,, 

2039: \label{confbis} 

2040: \ee

2041: where $\sigma_{\it 3} = \mbox{diag} (1, -1)$ and $\phi_0$ is some 

2042: fixed angle. The radial vacuum profile above is that 

2043: corresponding to $r_\infty=0$. This configuration consists of two 

2044: D6-branes with the same radial profiles but separated an angle 

2045: $2\phi_0$ in the 89-plane, and obviously satisfies the boundary

2046: condition that the branes overlap at $\l \ra \infty$. To see that it solves

2047: the non-Abelain equations of motion, consider expanding the 

2048: non-Abelian DBI action to linear order in fluctuations around \eqn{confbis},

2049: as before. 

2050: Since commutator terms are irrelevant at this order, and $X$ and

2051: $Y$ are diagonal, the non-Abelian action again reduces to the sum of

2052: two identical Abelian actions. In fact, the same argument shows that not

2053: only is the configuration \eqn{confbis} a solution of the non-Abelian 

2054: equations, but also that its energy is, for all $\phi_0$, equal to

2055: twice that of a single D6-brane. It follows that \eqn{confbis} is

2056: stable for all $\phi_0$. To see this, suppose that it is

2057: not, \ie, that there is some fluctuation around it that is tachyonic.

2058: Presumably, this would mean that this solution can decay to another

2059: one of lower energy. However, the latter would also have lower energy

2060: than the solution with $\phi_0=0$, in contradiction with our

2061: continuity argument above. 

2062: 

2063: Note that the previous reasoning applies equally well to any

2064: distribution of $\nf$ D6-branes around the $\phi$-circle, since it

2065: only relies on $X$ and $Y$ being diagonal. Thus, at $\mq=0$, all these 

2066: configurations are degenerate, lowest-energy, stable solutions. 

2067: Although it may be an interesting exercise to verify their stability 

2068: explicitly by expanding the non-Abelian DBI action to quadratic order in

2069: fluctuations around \eqn{confbis}, one should keep in mind

2070: that this conclusion is only expected to hold at leading order in the

2071: $1/\nc \sim g_s$ expansion, since, as discussed above, we expect 

2072: higher-order effects to lead to an attractive force between the branes,

2073: which would tend to align them.

2074: 

2075: 

2076: 

2077: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2078: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2079: %%%%%%%%%%%%%%%%%%%%%%%%%% SECTION %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2080: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2081: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2082: \section{Finite temperature physics} \label{fintem}

2083: 

2084: The gauge theory we have so far been describing is at zero

2085: temperature. To study the theory at a finite temperature $T$, we

2086: may follow the standard prescription: analytically continue the

2087: time coordinate $t\to t_\mt{E}=it$, periodically identify

2088: $t_\mt{E}$ with period $\delta \te = 1/T$, and impose

2089: anti-periodic boundary conditions on the fermions around the

2090: $\te$-circle. This prescription applies equally well in the gauge

2091: theory and in the dual string description. In the latter, however,

2092: one must consider all possible spacetime solutions with the

2093: appropriate boundary conditions, that is, possible saddle points 

2094: of the Euclidean path integral over supergravity (or rather, string)

2095: configurations \cite{Witten98b}.

2096: 

2097: In the present case, there are two\footnote{In analogy with

2098: \cite{fairies}, there may be an infinite family of Euclidean

2099: solutions labelled by two integers specifying which cycle of the

2100: $\te\tau$-torus shrinks to zero-size in the bulk. However, given

2101: that the $\te$- and $\tau$-axes are orthogonal, here either the

2102: solutions in eq.~\eqn{d4nut} or \eqn{bhmetric} will dominate the

2103: path integral.} Euclidean supergravity solutions which must be

2104: considered. The first is simply the Euclideanised version of that

2105: appearing in eq.~\eqn{metric},

2106: \beq 

2107: ds^2=\left(\frac{U}{R}\right)^{3/2} \left(d\te^2 +

2108: \sum_{i=1}^3 dx^i dx^i + f(U) d\tau^2\right)+

2109: \left(\frac{R}{U}\right)^{3/2}\frac{dU^2}{f(U)}

2110: +R^{3/2}U^{1/2}d\Omega^2_{\it 4}\,,

2111: \label{d4nut} 

2112: \eeq

2113: which dominates at low temperatures. Recall that 

2114: $f = 1 -\ut^3 / U^3$. The second is the Euclidean black hole 

2115: coming from a nonextremal D4-brane throat

2116: %

2117: \beq

2118: ds^2=\left(\frac{U}{R}\right)^{3/2}\left(\tilde{f}(U)dt_\mt{E}^2+

2119: \sum_{i=1}^3 dx^i dx^i +d\tau^2\right)+

2120: \left(\frac{R}{U}\right)^{3/2}\frac{dU^2}{\tilde{f}(U)}

2121: +R^{3/2}U^{1/2}d\Omega^2_{\it 4} \,, \label{bhmetric} \eeq

2122: %

2123: which dominates at high temperatures. In this case, $\tilde{f}=1-

2124: \uh^3 / U^3$. In the Lorentzian-signature solution, the event

2125: horizon is located at $U=\uh$.

2126: 

2127: Of course, these Euclidean metrics are identical up to

2128: interchanging the role of $\te$ and $\tau$ (and replacing $\ut$

2129: with $\uh$). To match the boundary conditions set by the 

2130: finite-temperature gauge theory, we identify these coordinates with

2131: periods

2132: \be 

2133: \d\te  = \fc{1}{T} \sac \d\tau = \fc{2\pi}{\mkk} \,,

2134: \label{periods}

2135: \ee

2136: in both solutions. Since we wish to avoid conical

2137: singularities in either spacetime at $U=\ut, \uh$, the

2138: boundary conditions fix the metric parameters as

2139: \be 

2140: \ut =\left({4\pi\over3\,\d\tau}\right)^2R^3\sac \uh=

2141: \left({4\pi\over3\,\d\te}\right)^2R^3\,. 

2142: \label{parmu}

2143: \ee

2144: 

2145: As we commented above, the metric \reef{d4nut} dominates the path

2146: integral at low temperatures, whereas that given by eq.~\reef{bhmetric} 

2147: dominates at high temperatures. One determines the

2148: transition point by comparing the Euclidean action of these two

2149: solutions \cite{Witten98b}. As the metrics describe the same

2150: geometry, it is easy to see that the transition occurs precisely when 

2151: $\d \tau=\d \te$. That is, a phase transition occurs at the critical

2152: temperature

2153: \be 

2154: \ct=\mkk/2\pi\,.

2155: \label{phtemp}

2156: \ee

2157: From the gauge theory viewpoint, this is a confinement/deconfinement 

2158: phase transition. This can be seen by the fact that the temporal 

2159: Wilson loop (around $\te$) vanishes for the low-temperature 

2160: background \reef{d4nut}, whereas it is

2161: nonvanishing for the Euclidean black hole \reef{bhmetric}

2162: \cite{Witten98b}. Alternatively, on the Lorentzian sections, the

2163: Wilson loop as computed from the solution \eqn{metric} exhibits an

2164: area law behaviour, while that computed for the Lorentzian black

2165: hole does not because the string can break in two pieces, each

2166: having an endpoint at the horizon \cite{break}. One might also

2167: observe that, in the low-temperature phase, one has a discrete

2168: spectrum of glueballs \cite{COOT98}, but this gives way to a

2169: continuum of excitations in the high-temperature phase dual to the

2170: black hole background.\footnote{Note that the spectrum would again

2171: be calculated with Lorentzian signature.}

2172: The analogous phase transition in 3+1 dimensions was discussed 

2173: in ref.~\cite{phases}. Finally, note that the phase transition takes 

2174: place at a temperature, \reef{phtemp}, where the gauge theory is

2175: starting to behave five-dimensionally, rather than four-dimensionally.

2176: 

2177: The preceding discussion refers only to the physics of the

2178: (compactified) five-dimensional gauge theory and is independent of

2179: the presence of fundamental matter fields. In previous sections we

2180: have exhaustively analysed the physics of dynamical quarks in the

2181: confining phase. We can extend the discussion to the deconfining

2182: phase by considering probe D6-branes in the black hole background

2183: \eqn{bhmetric}. In reference \cite{BEEGK03} the study of D7-branes

2184: in an asymptotically \ads{5} black hole background revealed the

2185: existence of two sets of regular embeddings, parametrised by the

2186: quark mass and chiral condensate of the dual gauge theory, and

2187: distinguished by whether or not the D7-brane intersects the

2188: horizon. It was suggested that the gauge theory undergoes a phase

2189: transition, signalled by a discontinuity in $dc/dm_\mt{q}$, when

2190: the D7-brane undergoes its `geometric' transition. Here we

2191: consider similar physics in our four-dimensional gauge theory.

2192: 

2193: As before, the D6-branes span (Euclidean) time and the spatial

2194: 123-directions in common with the D4-brane worldvolume, and lie at

2195: a fixed value of $\tau$. In parallel with the steps described by

2196: eqs.~(\ref{isomer}--\ref{isometric}), we introduce isotropic

2197: coordinates $\{\l, \Omega_{\it 2}, r, \phi\}$ for the Euclidean

2198: black hole \reef{bhmetric}. Furthermore,  we rescale to dimensionless

2199: coordinates

2200: \beq 

2201: \l \to \uh\l\ ,\qquad r\to \uh r\ ,\qquad \r\to \uh\r\ , 

2202: \eeq

2203: in analogy with eq.~\reef{tildes}. Now the embedding equation for

2204: $r(\l)$ becomes

2205: \beq

2206: \frac{d}{d\l}\left[\left(1-\left(\frac{1}{4\r^3}\right)^2\right)\l^2

2207: \frac{\dot{r}}{\sqrt{1+\dot{r}^2}}\right]

2208: =\frac{3}{8}\frac{r\l^2}{\r^8}\sqrt{1+\dot{r}^2}\,.

2209: \label{bhembed}

2210: \eeq

2211: Comparing this result with eq.~\reef{resc}, we see that the sign

2212: of the right-hand side has changed. As we will see below, this

2213: change of sign indicates that the brane now bends towards $U=\uh$:

2214: as expected, the black hole exerts an attractive force on

2215: the D6-brane, in contrast to the repulsive force in the 

2216: \nonsol\ background.

2217: 

2218: If we look for asymptotically constant embeddings such that

2219: $r(\l)\to r_\infty=L/\uh$ as $\l\to\infty$, we find the same

2220: long-distance behaviour as before:

2221: %

2222: \beq r\simeq

2223: r_\infty+\frac{c}{\l}\ .\label{asymprof} \eeq

2224: %

2225: These asymptotic parameters are related to the quark mass and

2226: chiral condensate as before, except for the substitution of $\uh$

2227: for $\ut$:

2228: %

2229: \be \mq = \fc{\uh r_\infty}{2\pi\ell_s^2}\sac \cc =  -8 \pi^2

2230: \ell_s^2 \t6 \uh^2 \, c \,. \label{magain} \ee

2231: %

2232: Again, the embedding equation can be solved numerically for

2233: any value of $r_\infty$, with the asymptotic boundary conditions

2234: implicit in \reef{asymprof}. Inspection of the results reveals

2235: that, just as in reference \cite{BEEGK03}, two qualitatively

2236: different types of profile are possible: those for which the probe

2237: brane intersects the horizon, and those for which it does not. As

2238: we vary the boundary condition $r_\infty$, we find that there is

2239: a unique embedding for sufficiently large or small values. In the

2240: case of small $r_\infty$, the unique embedding is such that the

2241: D6-brane falls into the horizon, whereas in the large-$r_\infty$

2242: case it does not, despite, of course, still being attracted by the

2243: black hole. For an intermediate range with

2244: $r_\infty\sim 1$, we find that more than one embedding is

2245: possible. In particular, for a given $r_\infty$, we find both

2246: embeddings which intersect the horizon and those where the

2247: D6-brane smoothly closes off before reaching the horizon.

2248: Furthermore, we also find that there are typically

2249: multiple solutions of at least one of these varieties of

2250: embedding; figure \ref{bhprof} displays some representative

2251: examples. These results imply that in varying from large to

2252: small $r_\infty$, there will be a phase transition in the

2253: behaviour of the quarks. This transition will occur in the

2254: intermediate regime, where, for a fixed temperature and quark mass,

2255: the condensate in the gauge theory can have multiple values. Of

2256: course, the true ground-state condensate will be selected by

2257: minimizing the energy density.

2258: 

2259: \FIGURE{ \epsfig{file=bhprofiles.eps, height=6cm} 

2260: \caption{Some representative D6-brane embeddings from the region 

2261: in which $c$ is multi-valued. The three profiles in each figure 

2262: have the same asymptotic value of $r_\infty$; in (a) two of them 

2263: avoid the horizon whereas in (b) two fall into the horizon.}

2264: \label{bhprof} }

2265: 

2266: Combining various results --- see eqs.~\eqn{inverse}, \eqn{periods}, 

2267: \reef{parmu} and \eqn{magain} --- the relation between the temperature

2268: and $r_\infty$ is found to be

2269: \be 

2270: T = \fc{\tb}{\sqrt{r_\infty}} \sac 

2271: {\rm with}\ \tb^2 \equiv

2272: \frac{9}{4\pi}\frac{m_\mt{q}M_\mt{KK}}{g^2_\mt{YM}N_\mt{c}} \,.

2273: \label{Tfromr} 

2274: \ee

2275: Here $\tb$ can be interpreted, for a given quark mass, as the 

2276: typical meson mass scale in the regime where supersymmetry is 

2277: restored in the zero-temperature gauge theory --- see the 

2278: discussion below eq.~\reef{region2}. In the zero-temperature 

2279: analysis of previous sections we regarded the variation of 

2280: $r_\infty$ as a variation of the quark mass, while $\ut$ or, 

2281: equivalently, the KK scale, was fixed. In this section we 

2282: wish to study the phase structure of the gauge theory as the

2283: temperature is varied, keeping the microscopic gauge theory

2284: parameters, and in particular the quark mass, fixed.

2285: For this reason, since in this phase the physics will only 

2286: depend on the ratio $T/\tb$, as suggested by eq. \eqn{Tfromr}, 

2287: we will think of the variation of $r_\infty$ as a 

2288: variation of the temperature.

2289: 

2290: \FIGURE{ \epsfig{file=ecvsT.eps, height=15cm} \caption{The quark

2291: condensate $\cc \propto c$ (plots (a) and (b)),

2292: and the energy density $\tilde{\cale}$ (plots (c) and (d)) as

2293: functions of temperature. The solid (dashed) lines describe the

2294: family of embeddings that do (not) intersect the horizon. Plots

2295: (b) and (d) show in more detail the regions of plots (a) and (c)

2296: where the two branches intersect. The vertical line lies at

2297: $T=\tcc$, the temperature at which the phase transition in the

2298: behaviour of the quarks takes place.}

2299:  \label{cvsT} }

2300: 

2301: The behaviour of the condensate (represented by $c$) as a function

2302: of the temperature is shown in figures \ref{cvsT}a and \ref{cvsT}b. 

2303: These plots

2304: display some interesting features. First note that $c\ra0$ both as

2305: $T\ra0$ and as $T\ra\infty$. In the high-temperature regime, the

2306: unique embedding is such that the D6-brane falls into the

2307: horizon. By following the solid line in the figure, we see

2308: that these embeddings only exist down to some some minimum

2309: temperature, $T\simeq 0.9869 \tb$. Similarly there is a unique

2310: embedding at sufficiently low temperatures such that the D6-brane

2311: does not reach the horizon, and following the dashed line shows

2312: that embeddings of this form only exist up to $T\simeq1.0202\tb$.

2313: As shown most clearly in figure \ref{cvsT}b, multiple solutions

2314: exist in the intermediate temperature range between the two values

2315: given above.

2316: 

2317: Naturally, where more than one solution to the embedding equation

2318: exists, the ground-state profile will be that selected out by

2319: lowest energy density. Since we are considering static

2320: configurations, the energy density is given by

2321: \beq

2322: \cale=-\int \df\l\df\Omega_{\it 2} {\cal L}= 4\pi

2323: T_\mt{D6}\uh^3 \,\int_{\l_\mt{min}}^\infty\df\l\

2324: \left(1-\left(\frac{1}{4\r^3}\right)^2\right)

2325: \l^2\sqrt{1+\dot{r}^2}\ ,

2326: \label{bhenergy}

2327: \eeq

2328: where $\l_\mt{min}$ is either zero, or the point where the brane

2329: meets the horizon, depending on which variety of embedding we are

2330: considering. This integral is divergent, so to compare the energy

2331: densities of different embeddings we regularize $\cale$ by using as a 

2332: reference the

2333: embedding $r(\l)=0$, which runs from the horizon (at $\l=\l_\mt{H}$, say)

2334: to infinity. Since this is a numerical calculation, we

2335: also introduce a cut-off, $\lm$, in $\l$ at  the point where, by

2336: imposing the appropriate boundary conditions, we match the

2337: numerical profile to the asymptotic solution, $r\simeq r_\infty+c/\l$

2338: . 

2339: There is an

2340: infinite contribution to $\cale$ from the remainder of the profile, but

2341: when regularized the leading behaviour of this correction

2342: can be shown to be

2343: \beq

2344: \cale_\mt{corr}=\frac{c^2}{2\lm}+\calo(\lm^{-3})\ .

2345: \eeq

2346: Therefore, to be precise, we compare the energy densities by calculating

2347: \beqa

2348: \tilde{\cale}=\frac{\cale_\mt{reg}}{4\pi

2349: T_\mt{D6}\uh^3}&\equiv&\int_{\l_\mt{min}}^{\lm} \df\l\

2350: \left(1-\left(\frac{1}{4\r^3}\right)^2\right)\l^2\sqrt{1+\dot{r}^2}\non\\

2351: && \hspace{1cm} -\int_{\l_\mt{H}}^{\lm} \df\l\

2352: \left(\l^2-\frac{1}{16\l^4}\right)+

2353: \frac{c^2}{2\lm} \,.

2354: \eeqa

2355: We then trust that this provides a valid basis for the comparison of

2356: different energy densities when the differences in $\tilde{\cale}$ are

2357: greater than $\calo(\lm^{-3})$. This is the case here, as we use

2358: $\lm=100$.

2359: 

2360: The energy density $\tilde{\cale}$ is plotted as a function of temperature

2361: in figures \ref{cvsT}c and \ref{cvsT}d. Starting from low temperatures, figure

2362: \ref{cvsT}d shows that, as $T$ is increased to that in the

2363: intermediate region, the physical embedding remains that which

2364: does not reach the horizon. At $T\simeq 1.0202 \tb$, the energy

2365: density for this embedding and that which falls into the horizon

2366: match. For $T > 1.0202 \tb$, the energetically favoured embedding is

2367: one which falls into the horizon. Therefore, at

2368: \be 

2369: T\equiv\tcc\simeq 1.0202\tb\,, 

2370: \ee

2371: there is a phase transition in the behaviour of the fundamental

2372: quarks. This phase transition is signalled by a discontinuity in

2373: $\langle \bar{\psi} \psi \rangle(T)$, as shown in Figure

2374: \ref{cvsT}b, as well as a discontinuity in the specific heat

2375: $d\tilde{\cale}(T)/dT$, as can be seen in Figure \ref{cvsT}d. Note

2376: that, to within the precision of our numerical calculations, $\tcc$ 

2377: cannot be distinguished from the maximum temperature for which 

2378: a horizon-avoiding embedding is possible.  

2379: 

2380: From the string description with the D4 throat geometry, the basic

2381: physics behind this transition is straightforward. Increasing the

2382: temperature increases both the radial position of the horizon and

2383: energy density of the black hole. For  small enough a

2384: temperature, the D6-brane is pulled towards the horizon but

2385: its tension balances the gravitational attraction. However, above

2386: the critical temperature $\tcc$, the gravitational force has

2387: increased to the point where it overcomes the tension and the

2388: brane is drawn into the horizon.

2389: 

2390: Another feature that characterizes the present phase transition is

2391: that the meson spectrum changes at $T=\tcc$. The discrete spectrum of

2392: mesons of the low-temperature phase becomes a continuum 

2393: of excitations in the high-temperature phase where the D6-brane 

2394: embedding extends to the horizon. This behaviour is similar to that of 

2395: the glueball spectrum at the deconfinement phase transition in the 

2396: gauge theory, discussed previously. 

2397: Note, however, that for temperatures above $\ct$ we are, by 

2398: assumption, in the deconfined phase of the gauge theory, and so it is 

2399: possible to introduce free quarks; these correspond to open strings 

2400: extending from the D6-brane down to the horizon. If $\ct < \tcc$ there 

2401: is then a range of temperatures, $\ct < T < \tcc$, in which there are

2402: free quarks but also a finite mass gap, $m_\mt{eff}(T)>0$, 

2403: associated with the introduction of such quarks into the 

2404: system. In this regime,

2405: there still exists a (discrete) spectrum of bound states with 

2406: $M<2m_\mt{eff}$. Note also that the deconfined adjoint degrees of

2407: freedom still generate a finite screening length beyond which the force 

2408: between a quark and an anti-quark vanishes \cite{break}; in the 

2409: string description this happens because the string that joins the 

2410: quarks together snaps into two pieces, each of them going from the

2411: brane to the horizon, that can be separated with no cost in energy. 

2412: Above $\tcc$, however, $m_\mt{eff}(T)=0$ and the spectrum of

2413: fundamental matter field excitations becomes continuous. 

2414: 

2415: Up to this point, we have discussed two independent phase

2416: transitions: the deconfining phase transition at $T=\ct$, for

2417: which the holographic dual description involved a discrete change

2418: of the bulk spacetime geometry, and the quark phase transition at

2419: $T=\tcc$, for which the holographic description involved a

2420: discrete jump in the embedding geometry of the D6-brane. However,

2421: if the latter is to be realized as a separate phase transition,

2422: then it must be true that $\tcc>\ct$.\footnote{Note that, in this

2423: regime, the fundamental fields still `feel' the deconfining

2424: transition at $T=\ct$. For example, there is a discontinuity in

2425: the chiral condensate $\cc$, which jumps from a negative to a

2426: positive value at this phase transition.} 

2427: In terms of the microscopic parameters, this condition translates into

2428: \beq 

2429: m_\mt{q} > {0.98\over 9\pi}g^2_\mt{YM} N_\mt{c} M_\mt{KK}\ . 

2430: \label{phtemp2}

2431: \eeq

2432: In the zero-temperature theory, this is roughly the quark mass

2433: scale at which supersymmetry is restored --- see the discussion below

2434: eq.~\reef{region2}.

2435: 

2436: If $\tcc<\ct$, the relevant spacetime geometry at $T=\tcc$ is the

2437: \nonsol{} \reef{d4nut} and the ambiguity in the D6-brane embedding

2438: discussed above is not relevant. Alternatively, the gauge theory

2439: is still in a confined phase and so the chemical potential to

2440: introduce a free quark remains infinite; in the dual description this

2441: is because, since there is no horizon, an open string must have both

2442: ends on the D6-brane, and so represents a quark/anti-quark bound

2443: state. In this case, the two phase transitions are realized

2444: simultaneously at $T=\ct$. An important example of this behaviour is 

2445: when $\mq=0$. In this case, the low-temperature phase is characterized 

2446: by spontaneous chiral symmetry breaking, as in section \ref{broken}.

2447: In the high-temperature phase, $\mq \ra 0$ is equivalent to 

2448: $T \ra \infty$ (see eq. \eqn{Tfromr}), in which limit the chiral

2449: condensate vanishes (see figure \ref{cvsT}a). Hence the deconfining 

2450: phase transition is also accompanied by the restoration of chiral 

2451: symmetry in this model.

2452: 

2453: To close the section, we wish to stress to the reader that while

2454: these phase transitions are of inherent interest in understanding the

2455: model under study here, they take place in a temperature regime

2456: where the gauge theory is intrinsically five-dimensional, rather

2457: than behaving in a four-dimensional way, as can be seen by

2458: comparing the transition temperatures given in eqs.~\reef{phtemp}

2459: and \reef{phtemp2} to the compactification scale $\mkk$.

2460: 

2461: 

2462: 

2463: 

2464: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2465: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2466: %%%%%%%%%%%%%%%%%%%%%% SECTION  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2467: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2468: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2469: \section{D4/D6/$\overline{\mbox{D6}}$-intersections}

2470: \label{Dbar}

2471: 

2472: In the discussion of embeddings in section \ref{broken}, we noted

2473: that the solution $\rt (\lt) =0$ of eq.~\eqn{resc} is not a

2474: physical one (for $\ut \neq 0$) if it terminates at the origin of the

2475: $U\tau$-plane because it would correspond to an open D6-brane.

2476: However, this solution can be extended through this point to

2477: construct a physically acceptable embedding. To see this, recall

2478: that the full D6-brane embedding is further specified by the

2479: conditions $\phi=\phi_0$ and $\tau=\tau_0$, where $\phi_0$ and

2480: $\tau_0$ are arbitrary constants. We can therefore join together a

2481: solution with $\tau=\tau_0$ to another one with $\tau=\tau_0+ \d

2482: \tau/2$ (see eq.~\eqn{deltatau}) to obtain a complete regular

2483: solution that describes a D6-brane whose projection on the

2484: $U\tau$-plane is a straight line that passes through the origin

2485: and intersects the boundary at two antipodal values of $\tau$.

2486: This means that this solution describes an intersection of $\nc$

2487: D4-branes with two D6-branes, or, more precisely, with one

2488: D6-brane and one anti-D6-brane.\footnote{Technically, the reason

2489: is that the orientations on the intersections of the D6-brane with

2490: the boundary at $\tau=\tau_0$ and $\tau=\tau_0+ \d \tau/2$,

2491: induced from a given orientation on the D6-brane, are opposite to

2492: each other.} Note that these embeddings bear a striking resemblance to

2493: those discussed in \cite{SS03}. The parallel includes the facts that

2494: both constructions involve two boundary defects and both preserve

2495: the chiral symmetry.

2496: 

2497: 

2498: To better understand the defect theory corresponding to these

2499: D4/D6/$\overline{\mbox{D6}}$-intersections, we consider a family

2500: of more general embeddings that describe D6-branes that lie at

2501: $\rt=0$ (or alternatively, $z^8=0=z^9$) and stretch along some

2502: curve $U=U(\tau)$ in the $U\tau$-plane, intersecting the boundary

2503: at two, not necessarily antipodal, values of $\tau$. Such

2504: D6-branes wrap a maximal $S^2$ within the $S^4$ in the metric

2505: \eqn{metric}, so the induced metric on their worldvolume is

2506: \be 

2507: ds^2(g) = \left( \fc{U}{R} \right)^{3/2} \, 

2508: \eta_{\mu\nu} dx^\mu dx^\nu +

2509: R^{3/2} U^{1/2} \, d\Omega_{\it 2}^2 + \left[ \left( \fc{U}{R}

2510: \right)^{3/2} \, f(U) + \fc{U'^2}{\left( \fc{U}{R} \right)^{3/2}

2511: \, f(U)} \right] \, d\tau^2 \,. 

2512: \ee

2513: Since the Lagrangian density on the D6-brane

2514: \be 

2515: \call_{\mt{D6}} = - \fc{1}{(2\pi)^6 \ell_s^7} \, 

2516: e^{-\phi} \, \sqrt{-\det g_{\mt{ind}}} = 

2517: \t6 \, U^2 \, \sqrt{\left( \fc{U}{R} \right)^3 f(U) + \fc{U'^2}{f(U)}} 

2518: \ee

2519: (where $U' = dU/d\tau$) does not depend explicitly on $\tau$, 

2520: it follows that

2521: \be 

2522: U' \, \fc{\pa \call}{\pa U'} - \call = \fc{U^2 \left( \fc{U}{R}

2523: \right)^3 f(U)} {\sqrt{\left( \fc{U}{R} \right)^3 f(U) +

2524: \fc{U'^2}{f(U)}}} 

2525: \ee

2526: is also $\tau$-independent. Let $U_0 \geq \ut$ be the minimum

2527: value of $U(\tau)$. By symmetry we can assume that this value is

2528: reached at $\tau=0$, that is, that we have $U(0)=U_0$ and

2529: $U'(0)=0$. Integrating the equation above we then find

2530: \be

2531: \tilde{\tau} (u) = \fc{3}{2} \, b^{7/2} \, g^{1/2}(b) \, \int_b^u

2532: dx \fc{1}{x^{3/2} g(x) \sqrt{x^7 g(x) - b^7 g(b)}} \,, \label{tau}

2533: \ee

2534: where we have introduced

2535: \be 

2536: \tilde{\tau} \equiv \fc{3}{2} \,

2537: \fc{\ut^{1/2}}{R^{3/2}} \, \tau \sac u \equiv \fc{U}{\ut} \sac b

2538: \equiv \fc{U_0}{\ut} \geq 1 \sac g(x) \equiv 1 - \fc{1}{x^3} \,.

2539: \ee

2540: Note that the rescaled angle $\tilde{\tau}$ has period $2\pi$.

2541: Eq.~\eqn{tau} specifies the D6-brane profile in the $U\tau$-plane.

2542: The D6-brane intersects the boundary at $\tilde{\tau}_\pm (b) =

2543: \pm \tilde{\tau}(u \ra \infty)$. These two points are implicitly

2544: fixed by $U_0$, the `lowest' point of the D6-brane in the

2545: $U\tau$-plane. In particular, their separation is a

2546: monotonically decreasing function of $b$. It is easy to show that

2547: $\tilde{\tau}_\pm (b \ra 1) = \pm \pi/2$, so the two points

2548: become antipodal as the lowest point of the D6-brane approaches

2549: the origin of the $U\tau$-plane. In this limit the projection of

2550: the D6-brane on this plane is just a straight line, as discussed

2551: above. In the opposite limit in which the (lowest point of)

2552: D6-brane approaches the boundary, the separation between the

2553: intersection points becomes smaller and smaller. In the strict

2554: limit $b\ra \infty$ the two coincide and the D6-brane disappears.

2555: 

2556: Now, these embeddings seem to portend fascinating physics for the

2557: dual five-dimensional gauge theory coupled to two defects or, more

2558: precisely, to a defect and an anti-defect. It appears, however, that 

2559: this physics is intrinsically five-dimensional. 

2560: Since the focus of the present

2561: paper is the four-dimensional QCD-like physics of the original

2562: model, we have not fully explored this system. However, we offer

2563: some preliminary remarks in the following.

2564: 

2565: The deflection of the D6-brane in the $\tau$-direction, which

2566: corresponds to a worldvolume direction of the D4-branes, is

2567: reminiscent of certain D5-brane embeddings in \ads{5} studied in

2568: ref.~\cite{ST02}. These authors found (non-supersymmetric)

2569: embeddings in which the D5-brane began at large radius running

2570: along $r$, but were deflected to run parallel to the \ads{5}

2571: horizon at some finite radius. At large $r$, these configurations can

2572: be understood as D3/D5 intersections that are T-dual to that

2573: represented in \eqn{intersection}. The dual gauge theory involved

2574: coupling $\caln=4$ SYM to a ($2+1$)-dimensional defect which broke 

2575: all supersymmetries. The deflection was interpreted as giving an 

2576: expectation value to an operator in the defect theory. 

2577: 

2578: In the present case, the \nonsol\ background breaks the

2579: supersymmetry but with both defects, supersymmetry would still not

2580: be restored in the limit $\ut\ra0$;\footnote{Note that the

2581: supersymmetry breaking of the background is not essential to

2582: constructing the general embeddings, \ie, one can simply replace

2583: $g(x)$ by 1 everywhere in eq.~\reef{tau}.} 

2584: one way to understand this in the field theory is to note that the half 

2585: of the supersymmetries of the ambient theory preserved (broken) 

2586: by the defect are precisely those that are broken (preserved) by the

2587: anti-defect.\footnote{Defect/anti-defect configurations, and more

2588:   general multiple defect configurations, were discussed in \cite{MNT02}.} 

2589: The embeddings described by eq.~\reef{tau} would correspond

2590: to a state where a symmetric pair of operators in each defect

2591: theory simultaneously acquire identical expectation values.

2592: Indirectly, the D6-brane embedding would also induce different

2593: expectation values of gauge theory operators, \eg, Tr $F^2$, in the

2594: two regions on the $\tau$-circle divided by the defects.

2595: 

2596: At the microscopic level, there are two sets of fundamental matter 

2597: fields in the field theory, one associated with each defect.

2598: This is realised in the string description by the fact that,

2599: despite there being a single D6-brane, its intersection with a

2600: constant-$U > U_0$ slice consists of two disconnected pieces, 

2601: corresponding to two independent sets of active degrees of freedom 

2602: at the corresponding energy scale. These degrees of freedom seem to

2603: disappear for energy scales associated to radial positions $U < U_0$, 

2604: as in \cite{ST02}. As discussed above, the D6-brane disappears in the limit 

2605: $\tilde{\tau}_+-\tilde{\tau}_- \ra 0$. Consistently, in this limit 

2606: $U_0 \ra \infty$ and the degrees of freedom above are completely

2607: removed. In the gauge theory, this corresponds to the limit in which

2608: the defect and the anti-defect are brought on top of each other.

2609: 

2610: This situation may be compared with that where both

2611: defects are associated with a `standard' D6-brane embedding as

2612: described in section \ref{broken}; one at 

2613: $\tilde{\tau}= \tilde{\tau}_+$  and the other at 

2614: $\tilde{\tau}= \tilde{\tau}_-$. Of course, in this case there are 

2615: also two sets of independent degrees of freedom associated with open 

2616: string excitations on both the D6- and 

2617: $\overline{\mbox{D6}}$-brane.\footnote{Note that in this case one 

2618: may also realise a defect/defect (as opposed to a defect/anti-defect) 

2619: theory by considering two D6-branes.} However, one difference is that

2620: the spectrum of these excitations will now be independent of the 

2621: defect/anti-defect separation,

2622: $\tilde{\tau}_+-\tilde{\tau}_-$.\footnote{In the approximation in

2623:   which interactions between the branes, which are suppressed in the

2624:   large-$\nc$ limit, are neglected.}  

2625: In addition, one would also find excitations corresponding to open 

2626: strings stretching between the two separate branes. 

2627: In contrast, in the case of a single D6-brane, the two defects can 

2628: interact through open string modes that propagate along the brane from 

2629: one defect to the other. 

2630: 

2631: Of course, one should ask the question which embedding minimizes

2632: the energy density of the system. That is, does energetics favour

2633: the smooth embedding which joins the two boundary defects or that

2634: in which there are two independent branes? As a preliminary step

2635: in this direction, we compared the energy density for the

2636: `standard' embeddings of the D6- and $\overline{\mbox{D6}}$-brane

2637: pair with $r_\infty=0$ to that of the smooth embedding above with

2638: $U_0=\ut$ (and hence $\tilde{\tau}_+-\tilde{\tau}_- =\pi$). Our

2639: numerical results indicate that the smooth embedding produces a

2640: lower energy density and hence describes the true vacuum

2641: configuration. Note then that with both defects corresponding to

2642: the D6- and $\overline{\mbox{D6}}$-branes, the vacuum preserves

2643: chiral symmetry.

2644: 

2645: Presumably if $U_0$ is increased above $\ut$, the smooth

2646: embeddings described above would continue to be energetically

2647: favored as it seems the energy density for these solutions must

2648: decrease as $\tilde{\tau}_+-\tilde{\tau}_-$ decreases while that

2649: for the standard embeddings remains unchanged. Note, however, that

2650: the above are just two particular cases of a more general family

2651: of embeddings in which the brane bends simultaneously in the 89-

2652: and the $U\tau$-planes, that is, embeddings with nontrivial

2653: functions $r=r(\l)$ and $\tau=\tau(\l)$ (using the notation of

2654: section 2). In general, we expect the minimum-energy two-defect

2655: embedding is of this type.

2656: 

2657: These defect theories might be explored in other ways as well. For

2658: example, one might also consider varying the parameter $\r_\infty$

2659: for the two defects, which may be done independently for each of

2660: the defects. Finally, of course, one might also consider the

2661: effects of a finite temperature, as in section 6.

2662: 

2663: 

2664: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2665: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2666: %%%%%%%%%%%%%%%%%%%%%% SECTION  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2667: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2668: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2669: \section{Discussion} \label{discus}

2670: 

2671: We have analyzed in detail the case of a single D6-brane, which

2672: corresponds to a single flavour in the gauge theory.

2673: In this case the gauge theory enjoys (at large $\nc$) an exact

2674: $U(1)_\mt{V} \times \ua$ symmetry. The first factor is just the flavour

2675: symmetry, which in the string description corresponds to the gauge

2676: symmetry on the worldvolume of the D6-brane. The second factor is

2677: a chiral symmetry that rotates the left- and right-handed

2678: fermions, $\psi_\mt{L}$ and $\psi_\mt{R}$, with opposite phases;

2679: it also rotates by a phase the adjoint scalar $X=X^8 + i X^9$,

2680: as is clear from the fact that in the string description the

2681: $\ua$ symmetry is just the rotational symmetry in the 89-plane.

2682: 

2683: Like in QCD, we expect the $\ua$ symmetry to be spontaneously

2684: broken by a chiral condensate, $\cc \neq 0$, and hence the

2685: existence of the corresponding Goldstone boson in the spectrum;

2686: this is the analog of the QCD $\eta'$ particle (which is massless in

2687: the $\nc \ra \infty$ limit), but, in an abuse of language, we have

2688: referred to it as a `pion'. The string description indeed confirms

2689: the field theory expectation: for zero quark mass, there is precisely

2690: one massless pseudo-scalar (in the four-dimensional sense) fluctuation of the

2691: D6-brane. If $\mq> 0$ this pion becomes a pseudo-Goldstone boson. Its

2692: mass, as well as the chiral condensate and the pion decay constant,

2693: can be computed in the string description, and we have shown that

2694: these three quantities and the quark mass obey the

2695: Gell-Mann--Oakes--Renner relation \eqn{GMOR} \cite{GMOR68}.

2696: Although this is a reassuring result, one should keep in mind that,

2697: just like in field theory, it is essentially a consequence of the

2698: {\it symmetry}-breaking pattern.

2699: 

2700: In the case $\nf >1$ we provided a holographic version of the

2701: Vafa--Witten theorem, which states that, in a vector-like theory

2702: with zero $\theta$-angle (like QCD) and $\nf$ quark flavours of

2703: identical masses $\mq >0$, the global, $U(\nf)$ flavour

2704: symmetry cannot be spontaneously broken. As we discussed,

2705: it is a priori unclear, on field theory grounds, that the Vafa--Witten

2706: theorem does apply to the actual gauge theory on the D4-branes, because

2707: of the presence of Yukawa-like couplings to scalar fields

2708: of masses $\mkk \sim \Lqcd$. The result from the string analysis

2709: therefore established that the presence of these fields does not alter

2710: the vacuum structure of the theory to the extent of violating the

2711: Vafa--Witten theorem.

2712: 

2713: The global $U(\nf)$ symmetry of the gauge theory becomes the gauge

2714: symmetry on the worldvolume of $\nf$ D6-branes in the string

2715: description. The holographic version

2716: of the Vafa--Witten theorem is the statement that the minimum-energy

2717: configuration for the $\nf$ D6-branes is a $U(\nf)$-preserving one in

2718: which all branes lie on top of each other. It should be emphasised

2719: that, unlike the Gell-Mann--Oakes--Renner relation, the Vafa--Witten

2720: theorem is not based on a symmetry(-breaking) argument, but depends on the

2721: {\it dynamics} of the theory. In the field theory, the proof relies

2722: on the reality and positivity of the fermion determinant that arises

2723: in the path integral formulation; in the holographic version, it

2724: relies on details of the dynamics of multiple D6-branes.

2725: 

2726: One result of our analysis of the $\nf>1$ case is the existence of

2727: $\nf^2$ independent, massless fluctuations of the D6-branes if $\mq=0$.

2728: It is tempting to interpret the associated $\nf^2$ massless pseudo-scalar

2729: particles in the spectrum of the gauge theory as the Goldstone bosons

2730: of a spontaneously broken $U(\nf)_\mt{A}$ symmetry, to which the

2731: $\ua$ symmetry would putatively be enhanced for $\nf >1$. However,

2732: only one of the $\nf^2$ states above is a true Goldstone boson,

2733: for two reasons. First, the five-dimensional theory contains a

2734: term of the form $\bar\psi_i \, X \, \psi^i$, where $i=1, \ldots, \nf$

2735: is the flavour index. Since $X$ only transforms under the $\ua$

2736: subgroup of $U(\nf)_\mt{A}$, the theory is only invariant under

2737: $\ua \times U(\nf)_\mt{V}$, which is spontaneously broken to

2738: $U(\nf)_\mt{V}$. Although $X$ is a massive field that can

2739: be integrated out, its mass $M_X \sim M_\mt{KK} \sim \Lambda\mt{QCD}$

2740: is of the same order as the strong coupling scale, so its integration

2741: should not lead to a real suppression of the symmetry-breaking operator

2742: above.

2743: 

2744: The second reason that makes implausible the interpretation of all

2745: but one of the massless modes above as Goldstone bosons is that

2746: they have non-derivative interactions. Indeed, the effective,

2747: interacting Lagrangian for these modes is obtained by expanding

2748: the non-Abelian Dirac--Born--Infeld action for the $\nf$ D6-branes

2749: \cite{Myers99} beyond quadratic order and then dimensionally

2750: reducing it to four dimensions, as we did in Section 4 in the

2751: $\nf=1$ case. Since the action for the D6-branes contains

2752: non-derivative interactions, such as commutator terms squared, so

2753: does the effective four-dimensional Lagrangian. These terms are

2754: only absent for the mode that multiplies the identity generator of

2755: $U(\nf)$. The caveat as to why this argument makes the Goldstone boson

2756: interpretation for the other modes implausible, but not strictly

2757: impossible, is that the coefficients in the effective Lagrangian

2758: might conspire so that, once all contributing diagrams are

2759: included, only momentum-dependent parts survive in any scattering

2760: amplitude (as in the linear sigma-model for the pion Lagrangian

2761: \cite{DGH92}). This possibility seems rather unlikely.

2762: 

2763: Thus we conclude that the string description predicts the existence

2764: of $\nf^2$ massless pseudo-scalar particles in the spectrum of the

2765: quantum gauge theory in the large-$\nc$ limit. Once sub-leading

2766: effects in the $1/\nc$ expansion are included, we expect these

2767: particles to acquire a small mass. The $\ua$ symmetry is anomalous at

2768: finite $\nc$, so its associated Goldstone boson is not exactly

2769: massless at finite $\nc$; presumably, this anomaly can be analysed in

2770: the supergravity description along the lines of 

2771: \cite{BDFLM01,KOW02,BDFLM02,GHP03}.

2772: The other $\nf^2 -1$ modes are likely to acquire a mass due to closed

2773: string interactions between the D6-branes, which are a

2774: $g_s \sim 1/\nc$ effect. From the gauge theory viewpoint, we have found

2775: no obvious reason for the masslessness (or lightness, at finite $\nc$)

2776: of these $\nf^2 -1$ pseudo-scalars, since we have argued that they are not

2777: Goldstone bosons.\footnote{Other examples with `unexpectedly' light scalars in

2778: the context of AdS/CFT have been discussed in \cite{Strassler03}.}

2779: Recall that, from the string viewpoint, the reason for this is that

2780: the only difference between the non-Abelian and the Abelian DBI

2781: actions consists exclusively of terms that involve commutators of

2782: fields. This can be regarded as a consequence of the fact that the

2783: seven-dimensional D6-brane action is T-dual to the ten-dimensional

2784: action of D9-branes; all commutators of scalar fields in the D6-brane

2785: action originate from commutators of gauge fields in the non-Abelian field

2786: strength on the D9-branes. Thus, it is {\it ten}-dimensional $U(\nf)$

2787: gauge invariance that seems to be responsible, from the string

2788: viewpoint, for the lightness of the gauge theory scalars.\footnote{

2789: This observation is due to M.\ Strassler.} Given the

2790: importance in high energy physics of mechanisms that naturally allow

2791: small scalar masses, it would be interesting to understand how to

2792: rephrase the constraints imposed by ten-dimensional $U(\nf)$

2793: gauge invariance purely in four-dimensional field theory terms.

2794: 

2795: 

2796: 

2797: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2798: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2799: %%%%%%%%%%%%%%%%%%%%  SECTION   %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2800: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2801: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2802: 

2803: 

2804: %\section*{Acknowledgements}

2805: \acknowledgments

2806: 

2807: We are especially grateful to J.\ Brodie for helpful

2808: conversations. We also wish to thank J.\ Donoghue, J.\ Erdmenger,

2809: J.\ I.\ Latorre, A.\ E.\ Nelson, D.\ T.\ Son, M.\ J.\ Strassler,

2810: S.\ Thomas, D.\ Tong and A.\ Uranga for discussions and comments. 

2811: Research at

2812: the Perimeter Institute is supported in part by funds from NSERC

2813: of Canada. RCM is further supported by an NSERC Discovery grant,

2814: DJW by Fonds FCAR du Qu\'ebec and by a McGill

2815: Major Fellowship, while MK is supported in part by NSF grants

2816: PHY-0331516, PHY99-73935 and DOE grant DE-FG02-92ER40706. 

2817: DJW wishes to thank the University of Waterloo Physics

2818: Department for their ongoing hospitality and MK would like to thank

2819: the Perimeter Institute for Theoretical Physics for partial

2820: support and hospitality while this work was being performed.

2821: 

2822: 

2823: \begin{thebibliography}{99}

2824: 

2825: \bibitem{Maldacena97}

2826: J.\ M.\ Maldacena, {\it The large N limit of superconformal field

2827: theories and supergravity}, \atmp{2}{1998}{231}

2828: [\ijtp{38}{1998}{1113}], \hepth{9711200}.

2829: 

2830: \bibitem{Witten98b}

2831: E.\ Witten, {\it Anti-de Sitter space, thermal phase transition,

2832: and confinement in gauge theories}, \atmp{2}{1998}{505},

2833: \hepth{9803131}.

2834: 

2835: \bibitem{COOT98}

2836: C.\ Csaki, H.\ Oouguri, Y.\ Oz and J.\ Terning,

2837: {\it Glueball mass spectrum from supergravity},

2838: \jhep{01}{1999}{017}, \hepth{9806021};

2839: R.\ de Mello Koch, A.\ Jevicki, M.\ Mihailescu and J.\ P.\ Nunes,

2840: {\it Evaluation of glueball masses from supergravity},

2841: \prd{58}{1998}{105009}, \hepth{9806125};

2842: H.\ Ooguri, H.\ Robins and J.\ Tannenhauser,

2843: {\it Glueballs and their Kaluza-Klein cousins},

2844: \plb{437}{1998}{77}, \hepth{9806171};

2845: J.\ A.\ Minahan,

2846: {\it Glueball mass spectra and other issues for supergravity duals

2847: of QCD models}, \jhep{01}{1999}{020}, \hepth{9811156};

2848: N.\ R.\ Constable and R.\ C.\ Myers,

2849: {\it Spin-two glueballs, positive energy theorems and

2850: the AdS/CFT correspondence}, \jhep{10}{1999}{037},

2851: \hepth{9908175};

2852: R.\ C.\ Brower, S.\ D.\ Mathur and C.\ Tan,

2853: {\it Glueball spectrum for QCD from AdS supergravity duality},

2854: \npb{587}{2000}{249}, \hepth{0003115}.

2855: 

2856: \bibitem{AFM98}

2857: O.\ Aharony, A.\ Fayyazuddin and J.\ M.\ Maldacena,

2858: {\it The large-N limit of ${\cal N} =2,1$ field theories

2859: from threebranes in F-theory}, \jhep{07}{1998}{013},

2860: \hepth{9806159}.

2861: 

2862: \bibitem{KR01}

2863: A.\ Karch and L.\ Randall,

2864: {\it Open and closed string interpretation of SUSY CFT's on

2865: branes with boundaries}, \jhep{06}{2001}{063}, \hepth{0105132}.

2866: 

2867: \bibitem{BDFLM01}

2868: M.\ Bertolini, P.\ Di Vecchia, M.\ Frau, A.\ Lerda, and R.\ Marotta,

2869: {\it N=2 Gauge theories on systems of fractional D3/D7 branes},

2870: \npb{621}{2002}{157}, \hepth{0107057}.    

2871: 

2872: \bibitem{BDFM01}

2873: M.\ Bertolini, P.\ Di Vecchia, G.\ Ferretti, and R.\ Marotta,

2874: {\it Fractional Branes and N=1 Gauge Theories},

2875: \npb{630}{2002}{222}, \hepth{0112187}.

2876: 

2877: \bibitem{KK02}

2878: A.\ Karch and E.\ Katz, {\it Adding flavor to AdS/CFT},

2879: \jhep{06}{2002}{043}, \hepth{0205236}.

2880: 

2881: \bibitem{KKW02}

2882: A.~Karch, E.~Katz and N.~Weiner, {\it Hadron masses and screening

2883: from AdS Wilson loops}, \prl{90}{2003}{091601}, \hepth{0211107}.

2884: 

2885: \bibitem{WH03}

2886: X.~J.~Wang and S.~Hu, {\it Intersecting branes and adding flavors

2887: to the Maldacena-Nunez background}, \jhep{09}{2003}{017},

2888: \hepth{0307218}.

2889: 

2890: \bibitem{Ouyang03}

2891: P.~Ouyang, {\it Holomorphic D7-branes and flavored N = 1 gauge

2892: theories}, \hepth{0311084}.

2893: 

2894: \bibitem{KMMW03}

2895: M.\ Kruczenski, D.\ Mateos, R.\ C.\ Myers and D.\ J.\ Winters,

2896: {\it Meson spectroscopy in AdS/CFT with flavour},

2897: \jhep{07}{2003}{049}, \hepth{0304032}.

2898: 

2899: \bibitem{SS03}

2900: T.\ Sakai and J.\ Sonnenschein,

2901: {\it Probing flavored mesons of confining gauge theories by

2902: supergravity}, \jhep{09}{2003}{047}, \hepth{0305049}.

2903: 

2904: \bibitem{BEEGK03}

2905: J.\ Babington, J.\ Erdmenger, N.\ Evans, Z.\ Guralnik and I.\ Kirsch,

2906: {\it Chiral symmetry breaking and pions in non-supersymmetric

2907: gauge/gravity duals}, \hepth{0306018}.

2908: 

2909: \bibitem{NPR03}

2910: C.\ Nunez, A.\ Paredes and A.\ V.\ Ramallo,

2911: {\it Flavoring the gravity dual of $\caln=1$ Yang-Mills with probes},

2912: \jhep{12}{2003}{024}, \hepth{0311201}.

2913: 

2914: \bibitem{IMSY98}

2915: N.\ Itzhaki, J.\ M.\ Maldacena, J.\ Sonnenschein and S.\ Yankielowicz,

2916: {\it Supergravity and the large N limit of theories with sixteen

2917: supercharges}, \prd{58}{1998}{046004}, \hepth{9802042}.

2918: 

2919: \bibitem{DFO01}

2920: O.\ DeWolfe, D.\ Z.\ Freedman and H.\ Ooguri,

2921: {\it Holography and Defect Conformal Field Theories},

2922: \prd{66}{2002}{025009}, \hepth{0111135}.

2923: 

2924: \bibitem{GMOR68}

2925: M.\ Gell-Mann, R.\ J.\ Oakes and B.\ Renner,

2926: {\it Behavior of current divergences under $SU(3) \times SU(3)$},

2927: \pr{175}{1968}{2195}.

2928: 

2929: \bibitem{VW84}

2930: C.\ Vafa and E.\ Witten, {\it Restrictions on symmetry breaking

2931: in vector-like gauge theories}, \npb{234}{1984}{173}.

2932: 

2933: \bibitem{SVZ79}

2934: M.\ A.\ Shifman, A.\ I.\ Vainshtein and V.\ I.\ Zakharov,

2935: {\it QCD and resonance physics. Sum rules.},

2936: \npb{147}{1979}{385}.

2937: 

2938: \bibitem{soliton}

2939: G.~T.~Horowitz and R.~C.~Myers, {\it The AdS/CFT correspondence

2940: and a new positive energy conjecture for  general relativity},

2941: Phys.\ Rev.\ {\bf D 59} (1999) 026005, \hepth{9808079}.

2942: 

2943: \bibitem{adsprop}

2944: U.~H.~Danielsson, E.~Keski-Vakkuri and M.~Kruczenski, {\it Vacua,

2945: propagators, and holographic probes in AdS/CFT}, JHEP {\bf 9901}

2946: (1999) 002, \hepth{9812007}.

2947: 

2948: \bibitem{Myers99}

2949: R.\ C.\ Myers, {\it Dielectric-Branes}, \jhep{12}{1999}{022},

2950: \hepth{9910053}.

2951: 

2952: \bibitem{fairies}

2953: R.\ Dijkgraaf, J.\ M.\ Maldacena, G.\ W.\ Moore and E.\ Verlinde,

2954: {\it A black hole farey tail}, \hepth{0005003}.

2955: 

2956: \bibitem{phases}

2957: R.\ C.\ Myers, {\it Stress tensors and Casimir energies

2958: in the AdS/CFT correspondence}, \prd{60}{1999}{046002},

2959: \hepth{9903203}; 

2960: S.\ Surya, K.\ Schleich and D.\ M.\ Witt,

2961: {\it Phase transitions for flat adS black holes},

2962: \prl{86}{2001}{5231}, \hepth{0101134}.

2963: 

2964: \bibitem{break}

2965: S.\ J.\ Rey, S.\ Theisen and J.\ T.\ Yee,

2966: {\it Wilson-Polyakov loop at finite temperature in large 

2967: N gauge theory and anti-de Sitter supergravity}, 

2968: \npb{527}{1998}{171}, \hepth{9803135}.

2969: 

2970: \bibitem{ST02}

2971: K.\ Skenderis and M.\ Taylor,

2972: {\it Branes in AdS and pp-wave spacetimes},

2973: \jhep{06}{2002}{025}, \hepth{0204054}.

2974: 

2975: \bibitem{MNT02}

2976: D.\ Mateos, S.\ Ng and P.\ K.\ Townsend,

2977: {\it Supersymmetric Defect Expansion in CFT from AdS Supertubes},

2978: \jhep{07}{2002}{048}, \hepth{0207136}.

2979: 

2980: \bibitem{DGH92}

2981: J.\ F.\ Donoghue, E.\ Golowich and B.\ R.\ Holstein,

2982: {\it Dynamics of the standard model},

2983: Cambridge University Press, 1992.

2984: 

2985: \bibitem{KOW02}

2986: I.\ R.\ Klebanov, P.\ Ouyang and E.\ Witten,

2987: {\it A gravity dual of the chiral anomaly},

2988: \prd{65}{2002}{105007}, \hepth{0202056}.

2989: 

2990: \bibitem{BDFLM02}

2991: M.\ Bertolini, P.\ Di Vecchia, M.\ Frau, A.\ Lerda and R.\ Marotta,

2992: {\it More Anomalies from Fractional Branes}, \plb{540}{2002}{104},

2993: \hepth{0202195}.

2994: 

2995: \bibitem{GHP03}

2996: U.\ Gursoy, S.\ A.\ Hartnoll and R.\ Portugues,

2997: {\it The chiral anomaly from M theory},

2998: \hepth{0311088}.

2999: 

3000: \bibitem{Strassler03}

3001: M.\ J.\ Strassler, {\it Non-supersymmetric theories with light scalar

3002: fields and large hierarchies}, \hepth{0309122}.

3003: 

3004: 

3005: \end{thebibliography}

3006: \end{document}

3007: