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: %\documentclass[12pt]{article}

8: %\documentclass[11pt]{article}

9: %\usepackage{epsf,amsfonts,amssymb,epsfig}

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

11: %\addtolength{\topmargin}{-2cm}

12: %\addtolength{\textheight}{3.5cm}

13: %\addtolength{\oddsidemargin}{-1cm}

14: %\addtolength{\textwidth}{1.5cm}

15: %\addtolength{\footskip}{0.7cm}

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

17: 

18: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

19: %%%%%%%%%%%%%%%%%%%%%%%%%% Macros %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

20: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

21: 

22: %%%%%%%%%%%%%%%%%%%%%%%% Structure definitions %%%%%%%%%%%%%%%%%%%%%%%%%%%

23: 

24: \newcommand{\bref}[1]{(\ref{#1})}

25: 

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

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

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

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

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

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

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

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

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

35: \newcommand{\nn}{\nonumber \\}

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

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

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

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

40: 

41: \newtheorem{theorem}{Theorem}

42: 

43: %%%%%%%%%%%%%%%%%%%% References %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

44: 

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

46: %\newcommand{\cmp}[3]{{Commun. Math. Phys.} {\bf #1} {(#2)} #3}

47: %\newcommand{\ijmpa}[3]{{Int. J. Mod. Phys.} {\bf A #1} {(#2)} #3}

48: %\newcommand{\jhep}[3]{{J. High Energy Phys.} {\bf #1} {(#2)} #3}

49: %\newcommand{\mpla}[3]{{Mod. Phys. Lett.} {\bf A #1} {(#2)} #3}

50: %\newcommand{\npb}[3]{{Nucl. Phys.} {\bf B #1} {(#2)} #3}

51: %\newcommand{\plb}[3]{{Phys. Lett.}{\bf B #1} {(#2)} #3}

52: %\newcommand{\prd}[3]{{Phys. Rev.} {\bf D #1} {(#2)} #3}

53: %\newcommand{\prep}[3]{{Phys. Rep.} {\bf #1} {(#2)} #3}

54: %\newcommand{\prl}[3]{{Phys. Rev. Lett.} {\bf #1} {(#2)} #3}

55: 

56: %\newcommand{\hepth}[1]{{\tt hep-th/#1}}

57: 

58: 

59: %%%%%%%%%%%%%%%%%%%%%%%% Greek Letters %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

60: 

61: \def\l {\lambda}

62: \def\a {\alpha}

63: %\def\ap {\alpha'}

64: \def\b {\beta}

65: \def\g {\gamma}

66: \def\G {\Gamma}

67: \def\d {\delta}

68: \def\s {\sigma}

69: \def\e {\epsilon}

70: \def\vt {\vartheta}

71: \def\vp {\varphi}

72: \def\T {\Theta}

73: 

74: 

75: \renewcommand{\t}{\theta}

76: 

77: 

78: % Shortcuts added by Toni:

79: 

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

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

82: \def\G{\Gamma}

83: \def\g{\gamma}

84: 

85: \def\e{\epsilon}

86: 

87: \def\a{\alpha}

88: \def\b{\beta}

89: 

90: 

91: %%%%%%%%%%%%%%%%%% Calligraphic Letters %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

92: 

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

119: 

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

121: 

122: 

123: %%%%%%%%%%%%%%%%% Boldmath Letters %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

124: 

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

140: 

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

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

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

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

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

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

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

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

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

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

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

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

153: 

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

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

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

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

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

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

160: 

161: %%%%%%%If you do not have the msbm fonts, delete the following 5 lines

162: \font\mybb=msbm10 at 8pt

163: \def\bb#1{\hbox{\mybb#1}}

164: %\def\bZ {\bb{Z}}

165: \def\bR {\bb{R}}

166: %\def\bE {\bb{E}}

167: 

168: 

169: 

170: %%%%%%%%%%%%%%%%%%%%%% Miscellaneous  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

171: 

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

173: 

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

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

176: \newcommand{\para}{\parallel}

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

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

179: 

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

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

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

183: 

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

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

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

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

188: 

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

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

191: 

192: \newcommand{\un}{\underline}

193: \newcommand{\pa}{\partial}

194: \newcommand{\fc}{\frac}

195: \newcommand{\w}{\wedge}

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

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

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

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

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

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

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

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

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

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

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

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

208: \newcommand{\ep}{\ensuremath{{\eta'}}}

209: \newcommand{\ratio}{\ensuremath{{\nf/\nc}}}

210: 

211: \newcommand{\cone}{\ensuremath{{C_{\it 1}}}}

212: \newcommand{\ftwo}{\ensuremath{{F_{\it 2}}}}

213: \newcommand{\csev}{\ensuremath{{C_{\it 7}}}}

214: \newcommand{\feig}{\ensuremath{{F_{\it 8}}}}

215: \newcommand{\osev}{\ensuremath{{\omega_{\it 7}}}}

216: 

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

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

219: \newcommand{\ve}{\varepsilon}

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

221: 

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

223: \def\ls{\ell_s}

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

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

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

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

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

229: \def\ukk {U_\mt{KK}}

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

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

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

233: \def\half{{1\over 2}\,}

234: \def\shalf{\mbox{${1\over 2}\,$}}

235: 

236: 

237: 

238: 

239: 

240: \newcommand{\ra}{\rightarrow}

241: 

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

243: 

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

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

246: 

247: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

248: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

249: %%%%%%%%%%%%%%%%%%%%%% TITLEPAGE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

250: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

251: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

252: 

253: 

254: \title{\LARGE The Holographic Life of the $\eta'$}

255: 

256: \author{Jos\'e L. F. Barb\'on,$^{a}$ Carlos Hoyos,$^{b}$

257:   David Mateos,$^{c}$ and Robert C. Myers$\,^{c,d}$ \\

258:   $^a$ Theory Division, CERN, CH-1211, Geneva 23, Switzerland \\

259:   $^b$ Instituto de F\'{\i}sica Te\'orica UAM/CSIC,  C-XVI \\

260:        and Departamento de F\'{\i}sica Te\'orica, C-XI \\

261:        Universidad Aut\'onoma de Madrid, E-28049--Madrid, Spain \\

262:   $^c$ Perimeter Institute for Theoretical Physics \\

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

264:   $^d$ Department of Physics, University of Waterloo  \\

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

266: 

267: E-mail: \email{barbon@cern.ch, c.hoyos@uam.es,

268:   dmateos@perimeterinstitute.ca, rmyers@perimeterinstitute.ca}}

269: 

270: \abstract{In the string holographic dual of large-$\nc$ QCD with

271: $\nf$ flavours of \cite{KMMW03}, the $\eta'$ meson is massless at

272: infinite $\nc$ and dual to a collective fluctuation of $\nf$

273: D6-brane probes in a supergravity background. Here we

274: identify the string diagrams responsible for the generation

275: of a mass of order $\nf/\nc$, consistent with the

276: Witten-Veneziano formula, and show that the supergravity limit

277: of these diagrams corresponds to mixings with pseudoscalar glueballs.

278: We argue that the  dependence on the

279: $\theta$-angle in the supergravity description

280:  occurs only through the combination

281: $\theta + 2\sqrt{\nf} \, \eta' / f_\pi$, as dictated by the $\ua$

282: anomaly. We provide a quantitative test by computing the linear term

283: in the $\eta'$ potential in two independent ways, with perfect

284: agreement.}

285: 

286: \keywords{D-branes, Supersymmetry and Duality, AdS/CFT, QCD}

287: 

288: \preprint{}

289: 

290: \begin{document}

291: 

292: {\vskip 1cm}

293: 

294: 

295: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

296: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

297: \section{Introduction}

298: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

299: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

300: 

301: In QCD with three light flavours of quark,

302: $m_\mt{u}, m_\mt{d}, m_\mt{s} \ll \Lambda_\mt{QCD}$,

303: there is a very succesful model

304: of light meson phenomenology in terms of the spontaneous breaking of

305: the chiral $SU(3)_\mt{L} \times SU(3)_\mt{R}$  flavour symmetry down

306: to the diagonal subgroup.  In the same context, the spontaneous

307: breaking of the axial  $\ua$  group would imply the existence of a

308: neutral pseudoscalar meson with the quantum numbers of the $\ep$ meson

309: and mass $m_\ep <\sqrt{3}\, m_\pi$. The measured mass of the $\ep$

310: meson, close to $1$ GeV, exceeds this bound by a large amount, leading

311: to the so-called `$U(1)$ problem' \cite{GMOR68, Weinberg75}.

312: 

313: Quantum mechanically, the $\ua$ symmetry is broken by the anomaly,

314: proportional to $\tr\, F\wedge F$, which in turn means that the $U(1)$

315: problem is tied to the dependence of physical quantities on the

316: $\theta$-angle of QCD. In particular, the $\ep$ meson can only be lifted by

317: non-perturbative effects, since the anomaly itself is a total derivative, and

318: thus inocuous in perturbation theory.

319: 

320: Because of the anomaly, the effective CP-violating phase is the combination

321: $\theta + {\rm arg}\,(\,{\rm det}\,\mq \,)$, where $\mq$ denotes the

322: quark mass matrix for $\nf$ flavours. Hence, normalizing the would-be

323: $\ua$ Goldstone boson  by the global phase $e^{i\phi}$ of the

324: $U(\nf)_\mt{A}$ Goldstone-boson matrix $\Sigma$, the anomaly constrains

325: the low-energy effective potential of the phase field to depend on the

326: combination  $\theta + \nf \,\phi$ in the chiral limit, $\mq =0$.

327: For example, a dilute gas of instantons generates a potential of the

328: form (c.f. \cite{thooft})

329: \be

330: \label{potinst}

331: V(\Sigma)_\mt{inst} = A \,e^{i\theta} \,\det \,\Sigma + {\rm h.c.} \,,

332: \ee

333: where $A \sim \exp(-8\pi^2 /\gym^2)$. In the large-$\nc$ limit, this potential

334: is exponentially supressed. However, it was shown by Witten \cite{Witten79} (see

335: also \cite{Veneziano79, clasicos}) that a non-trivial $\theta$-dependence

336: within the $1/\nc$ expansion of the pure Yang--Mills (YM) theory implies

337: a potential  of the form

338: \be

339: \label{scau}

340: V(\Sigma)_\mt{WV} = {1\over 2} \,\chi_\mt{YM}\,

341: \left(\theta -i\,\log\,{\rm det}\,\Sigma\, \right)^2

342: \ee

343: to first non-trivial order in the $1/\nc$ expansion (generated by a

344: non-perturbative resummation of OZI-supressed quark annihilation

345: diagrams \cite{RGG75, Witten79, Veneziano79}).

346: The constant $\chi_\mt{YM}$ is the topological susceptibility of the

347: {\it pure} YM theory,

348: \be

349: \label{topsus}

350: \chi_\mt{YM} = {d^2 \,\cale_\mt{vac} \over d\,\theta^2}

351: \Big |_{\nf =0, \;\theta=0} \,,

352: \ee

353: to leading order in the $1/\nc$ expansion. More generally, the large-$\nc$

354: scaling of the vacuum energy density in the pure YM theory is

355: \be

356: \label{pg}

357: \cale_\mt{vac} = \nc^2 \,F(\theta/\nc) \,,

358: \ee

359: where the function $F(y)$ has a Taylor expansion with coefficients of

360: $O(1)$ in the large-$\nc$ limit, and it  should be multivalued under

361: $\theta\rightarrow \theta+2\pi$ in order for the $\theta$-angle to be

362: defined with $2\pi$ periodicity.  Then, applying the substitution

363: $\theta \rightarrow \theta + \nf \,\phi$ dictated by the anomaly, we

364: find a potential of the general form

365: \be

366: \label{pot}

367: V(\phi) = \nc^2 \, F\left({\theta + \nf \,\phi \over \nc}\right) \,.

368: \ee

369: Notice that the multivalued nature of $\theta$-dependence in the large-$\nc$

370: limit of pure YM theory is tied to an analogous `multibranched' nature of the

371: $\ep$ potential, already apparent by the contrast between (\ref{potinst}) and

372: (\ref{scau}). The $\ep$ mass is obtained by selecting the quadratic term and

373: introducing the canonically normalized $\ep$ field:\footnote{Note that the present

374: normalization is consistent with \cite{Witten79}, however, this differs from that

375: used in \cite{KMMW03}: $f_\pi$\cite{KMMW03}$=f_\pi$\cite{Witten79}$/2$.}

376: \be

377: \label{cann}

378: \phi (x)= {2 \over f_\pi \,\sqrt{\nf}} \,\ep (x) \,,

379: \ee

380: where $f_\pi$ is the pion decay constant; since

381: $f_\pi = f_\ep + O(1/\nc)$, we will not distinguish between the two.

382: This results in the famous Witten--Veneziano formula

383: \be

384: \label{wv}

385: m_\ep^2 = {4 \nf \over f_\pi^2} \,\chi_\mt{YM} \,.

386: \ee

387: Since $f_\pi \sim \sqrt{\nc}$, we get a mass-squared of $O(\nf / \nc)$.

388: 

389: In the same fashion, one can also derive soft-$\ep$ amplitudes by

390: applying the substitution

391: $\theta \rightarrow \theta + 2\sqrt{\nf} \,\eta' /f_\pi$ to the

392: $\theta$-dependence of pure-glueball amplitudes. We can specify

393: not only the low-energy effective action of the pseudo-Goldstone

394: field $\ep$,  but also a large-$\nc$ effective Lagrangian featuring

395: glueballs and mesons with masses of $O(1)$ in the large-$\nc$ limit,

396: together with a light $\eta'$ meson with mass of $O(1/\nc)$.

397: 

398: In string descriptions of large-$\nc$ gauge theories, such as AdS/CFT

399: models, it should be possible to verify this scenario by direct

400: inspection of the low-energy effective action of the

401: string theory in the AdS-like background, either at the level of the

402: classical supergravity approximation (glueball-meson spectrum) or at

403: the level of string loop corrections. In particular, one should find

404: the potential (\ref{pot}) as part of the effective action in the

405: background geometry.

406: 

407: As we will review below, the first part of this check was carried out

408: by Witten \cite{Witten98b}, who studied the $\theta$-dependence of an

409: AdS-like model \cite{Witten98a} dual to a non-supersymmetric,

410: confining cousin of pure YM theory. Introducing  $\theta$-dependence

411: through Ramond--Ramond (RR) fields, Witten derived the analog of

412: (\ref{pg}) for this model, with the result

413: \be

414: \label{tow}

415: \cale_\mt{vac}^{(k)} = \nc^2 \,F_k(\theta/\nc) =

416: {1\over 2} \,\chi_g \,(\theta + 2\pi k)^2 + O(1/\nc)

417: \ee

418: to leading order in the $1/\nc$ expansion,

419: where the integer $k$ labels the $k$-th stable `vacuum'.

420: Minimizing over $k$ for a given value of $\theta$ selects the true

421: vacuum and restores the $2\pi$ periodicity. The $O(1)$ constant

422: $\chi_g$ is the topological susceptibility in this model.

423: 

424: In order to complete the check we need a generalization of this setup

425: that incorporates flavour degrees of freedom in the chiral limit.

426: In the large-$\nc$ limit it  should also incorporate a {\it massless},

427: pseudoscalar Goldstone boson that can be identified with the $\eta'$ field.

428: Following the general ideas of \cite{flavour}, a model with exactly

429: these properties was constructed in \cite{KMMW03} by introducing

430: flavour degrees of freedom via D6-brane probes embedded in the

431: previous background.\footnote{Following the ideas of \cite{flavour},

432:   meson physics has been studied in the context of AdS/CFT in

433:   \cite{mesons}.} In this note we investigate the $\ep$ physics in

434: this model.

435: 

436: We first argue that the introduction of D6-branes corresponding to

437: massless quarks allows the dependence of the supergravity

438: description on the microscopic

439: $\theta$-angle to be shifted away, precisely as expected on field

440: theory grounds. We then discuss the kind of string loop corrections

441: that must be responsible for the generation of the anomaly-induced

442: potential (\ref{scau}), in a string analog of the old Isgur-de

443: R\'ujula-Georgi-Glashow mechanism

444: \cite{RGG75}. Although we are unable to provide an independent

445: stringy calculation of the $\ep$ mass, we show that, in the

446: supergravity limit, the leading Wess--Zumino coupling of the

447: D6-brane probes to the RR background fields induces the right

448: structure of mixings between the $\ep$ meson and pseudoscalar

449: glueballs. In section \ref{quanti} we present a non-trivial

450: quantitative check of this scenario by computing the linear term of

451: the potential (\ref{scau}) in two independent ways, with precise

452: agreement.

453: 

454: In order for this paper to be self-contained, we have included, in

455: section \ref{summary}, a summary of the aspects of

456: \cite{KMMW03, Witten98b, Witten98a} that are needed in the rest of

457: the paper. Readers who are familiar with these can go directly to

458: section \ref{anomaly}.

459: 

460: 

461: 

462: 

463: 

464: 

465: 

466: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

467: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

468: \section{The Model} \label{summary}

469: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

470: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

471: 

472: A proposal to realize a holographic dual of four-dimensional,

473: non-supersymmetric, pure $SU(\nc)$ YM theory was made in \cite{Witten98b}.

474: One starts with $\nc$ D4-branes in the type IIA Minkowski vacuum

475: $\bbr{9} \times S^1$. The D4-branes wrap the compact direction, of

476: radius $M_\mt{KK}^{-1}$, and anti-periodic boundary

477: conditions are imposed for the worldvolume fermions on this circle.

478: Before compactification, the D4-brane theory is a five-dimensional,

479: supersymmetric $SU(\nc)$ gauge theory whose field content

480: includes fermions and scalars in the  adjoint representation of

481: $SU(\nc)$, in addition to the gauge fields. At energies much below

482: the compactification scale, $M_\mt{KK}$, the theory is effectively

483: four-dimensional. The anti-periodic boundary conditions break all of

484: the supersymmetries and give a tree-level mass to the fermions, while

485: the scalars also acquire a mass through one loop-effects. Thus, at

486: sufficiently low energies, the dynamics is that of four-dimensional,

487: massless gluons.

488: 

489: If the type IIA vacuum is such that there is a non-trivial holonomy

490: around the circle for the RR one form, $\cone$, then the Wess-Zumino

491: %R footnote **

492: coupling on the D4-branes,\footnote{We adopt a nonstandard

493: convention where the field components $(\cone)_\mu$ have

494: dimensions of length$^{-1}$, \ie,

495: $\cone$\cite{dielectric}$=g_s\ell_s\,\cone$[present]. Hence as

496: forms, $\cone$ and $\ftwo$ are both dimensionless which will

497: simplify various expressions in the following. Note that with

498: these conventions, the forms $\csev$ and $\feig$, defined by the

499: usual duality relation $\feig = *\,\ftwo$ in subsequent sections,

500: both have dimensions of length$^{6}$.}

501: %R too lazy to say anything about F4 in the background

502: \be

503: \label{thec}

504: {1\over 8\pi^2} \int_{\bR^4 \times S^1} \cone \wedge

505: \tr \, F\wedge F \,,

506: \ee

507: induces a $\t$-term in the gauge theory with

508: \be

509: \label{ett}

510: \theta = \int_{S^1} \cone \,.

511: \ee

512: 

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

514: string theory in the near-horizon region of the associated

515: (non-supersymmetric) supergravity background.

516: Using this description, Witten showed \cite{Witten98b}

517: that the $\t$-dependence of the vacuum energy of the YM theory has

518: precisely the form expected on field theory grounds, as reviewed in

519: the Introduction.

520: 

521: In order to explore the new physics associated to the $\eta'$

522: particle, we need to extend Witten's construction in such a way that,

523: in the limit in which the KK modes would decouple, the only additional

524: degrees of freedom would be $\nf$ flavours of fundamental, massless

525: quarks.\footnote{As usual

526: in AdS/CFT-like dualities, this limit is not fully realisable within

527: the supergravity approximation; see \cite{KMMW03} for a more

528: detailed discussion.} Such an extension was proposed in

529: \cite{KMMW03}, following the general strategy of adding fundamental

530: matter to AdS/CFT by adding D-brane probes \cite{flavour}. The

531: construction is as follows.

532: 

533: Consider adding $\nf$ D6-branes to the original system, oriented as

534: described by the array

535: \be

536: \begin{array}{rccccccccccl}

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

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

539: \ea

540: \label{intersection}

541: \ee

542: The original gauge fields and adjoint matter on the D4-branes

543: arise from the light modes of the 4-4 open strings, and propagate in

544: five dimensions. In contrast, the light modes of the 4-6 open strings

545: give rise to $\nf$ hypermultiplets in the fundamental

546: representation of $SU(\nc)$ that propagate only along the four

547: directions common to both branes.\footnote{We emphasize that these

548: fields are intrinsically four-dimensional, \ie, they do {\it not} propagate

549: along the circle direction.} Each hypermultiplet consists of one

550: Dirac fermion, $\psi = \psi_\mt{L} + \psi_\mt{R}$, and two complex

551: scalars. The addition of the D6-branes leaves $\caln =2$ unbroken

552: supersymmetry (in four-dimensional language). This ensures that there is

553: no force between the D4- and the D6-branes, and hence that they can be

554: separated in the 89-plane. The bare mass of the hypermultiplets,

555: $\mq$, is proportional to this separation. If the D6-branes lie at the

556: origin in the 89-plane, then the system enjoys

557: a $\ua$ symmetry associated to rotations in this plane. A crucial

558: fact in the construction of \cite{KMMW03} is that,

559: in the gauge theory, this symmetry acts on the

560: fundamental fermions as a {\it chiral} symmetry, since it rotates

561: $\psi_\mt{L}$ and $\psi_\mt{R}$ with opposite phases. Hence the

562: $\ua$ symmetry acts on the relevant fields as

563: \be

564: \label{rsim}

565: X_8 + iX_9 \ra e^{i\alpha}\,(X_8+iX_9) \sac

566: \psi_\mt{L} \ra e^{i\alpha/2} \, \psi_\mt{L} \sac

567: \psi_\mt{R} \ra e^{-i\alpha/2} \, \psi_\mt{R} \,.

568: \ee

569: 

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

571: $2\pi/M_\mt{KK}$, and with anti-periodic boundary conditions for

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

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

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

575: massive. Similarly, we expect loop effects to induce a mass for the

576: scalars in the fundamental representation. Generation of a mass for

577: the fundamental fermions is, however, forbidden (in the strict

578: large-$\nc$ limit) by the existence of the chiral $\ua$ symmetry

579: above. Therefore, at low energies, we expect to

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

581: $\nf$ flavours of fundamental quark.

582: 

583: In the so-called `probe limit', $\nf \ll \nc$, a holographic

584: description of this theory is obtained by replacing the D4-branes

585: by their supergravity background. The condition $\nf \ll \nc$ ensures

586: that the backreaction of the D6-branes on this background is

587: negligible, and hence that they can be treated as probes.

588: The D6-brane worldvolume fields (and, more generally, all open string

589: excitations on the D6-branes) are dual to gauge-invariant field

590: theory operators constructed with at least two hypermultiplet fields,

591: that is, meson-like operators; of particular importance here will be

592: the quark bilinear operator, $\bar{\psi} \psi \equiv \bar\psi_i \psi^i$,

593: where $i=1, \ldots , \nf$ is the flavour index.

594: 

595: Having reviewed the general construction, we now provide some of the

596: details from \cite{KMMW03} that will be needed in the following sections.

597: 

598: The supergravity background dual to the $\nc$ D4-branes

599: takes the form

600: \begin{eqnarray}

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

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

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

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

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

606: \sac F_{\it 4} = \frac{\nc}{\Omega_{\it 4}} \, \varepsilon_{\it 4} \sac

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

608: \label{metric1}

609: \end{eqnarray}

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

611: $\bbr{4}$, and correspond to the four non-compact directions along

612: the D4-branes, as in \eqn{intersection}, whereas $\tau$

613: parametrizes the circular 4-direction on which the branes are

614: compactified. $d\Omega_{\it 4}^2$ and $\varepsilon_{\it 4}$ are

615: the $SO(5)$-invariant line element and volume form on a unit

616: four-sphere, respectively, and $\Omega_{\it 4}=8\pi^2/3$ is its

617: volume. $U$ has dimensions of length and may be thought of as a

618: radial coordinate in the 56789-directions transverse to the

619: D4-branes. Since the $\tau$-circle shrinks to zero size at

620: $U=\ukk$, to avoid a conical singularity $\tau$ must be identified

621: with period

622: \be

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

624: \label{deltatau}

625: \ee

626: Under these circumstances the supergravity solution above is regular

627: everywhere. $U$ and $\tau$ parametrize a `cigar' (as opposed to a

628: cylinder). That is, the surface parametrized by these coordinates

629: is topologically a plane.

630: The solution is specified by the string coupling constant, $g_s$,

631: the Ramond--Ramond flux quantum (\ie, the number of D4-branes),

632: $\nc$, and the constant $\ukk$. (The remaining parameter is given by

633: $R^3 =  \pi g_s \nc\,\ell_s^3$, with $\ell_s$ the string length.)

634: If $\ukk$ is set to zero, the solution (\ref{metric}, \ref{metric1})

635: reduces to the extremal,

636: 1/2-supersymmetric D4-brane solution, so we may say that $\ukk$

637: characterizes the deviation from extremality.

638: %R equations

639: The relation between these parameters and those of the $SU(\nc)$

640: dual gauge theory, namely, the compactification scale,

641: $\mkk=2\pi/\d\tau$, and the four-dimensional coupling constant

642: {\it at} the compactification scale, $\gym$, is \cite{KMMW03}:

643: \be

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

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

646: \ukk = {2\over9} \gym^2 \nc \, \mkk \ell_s^2 \,.

647: \label{inverse}

648: \ee

649: 

650: In the gravity description, the defining equation \eqn{ett} for the

651: $\t$-angle must be understood as an asymptotic boundary condition for the RR

652: one-form at $U \ra \infty$. In other words, we must impose

653: \be

654: \label{deft}

655: \theta + 2\pi k = \lim_{U\to \infty} \int_{S^1} \cone =

656: \int_{\rm Cigar} \ftwo \,,

657: \ee

658: where the $S^1$ is parametrized by $\tau$ and lies at

659: $U=\mbox{constant}$, as well as at constant positions in $\bbr{4}$ and

660: $S^4$, and $\ftwo=d\cone$.

661: Notice that the asymptotic holonomy of $\cone$ is measured over a

662: contractible cycle of the background geometry. Under these circumstances,

663: the right-hand side of (\ref{deft}) defines an arbitrary real number,

664: and we must specify the integer $k$ to respect the angular nature

665: of $\theta$.

666: 

667: To leading order in $1/\nc$, the solution of the supergravity

668: equations that obeys the constraint \eqn{deft} is obtained

669: \cite{Witten98b} simply by adding to \eqn{metric} and

670: \eqn{metric1} the RR two-form \be \label{wittf} \ftwo = {C \over

671: U^4} \, (\t + 2\pi k) \,dU\wedge d\tau \,, \ee where $C= 3\,

672: \ukk^3 /\d \tau$. Inserting this expression into the kinetic

673: action of the RR forms we get Witten's result for the energy

674: density

675: \be

676: \label{wr}

677: \cale_\mt{vac}^{(k)}

678: = {1\over 2 (2\pi)^7\ls^{\,\, 6} V_{\it 4}} \int \ftwo\wedge *

679: \ftwo = {1\over 2} \,\chi_g \,(\theta+ 2\pi k)^2 \,,

680: \ee

681: %R ls reinstated above

682: where $V_{\it 4} = \int d^4 x$. The topological susceptibility is

683: thus given by (c.f. \cite{HO98})

684: \be

685: \label{tosu}

686: \chi_g = {(\gym^2 \nc)^3  \over   4 \cdot (3\pi)^6} \, \mkk^4 \,.

687: \ee

688: The generation of a topological susceptibility of

689: $O(1)$ constrasts with naive expectations based on an instanton gas

690: picture. In this model, one can explicitly check that the

691: semiclassical approximation based on a dilute instanton gas does not

692: commute with the large-$\nc$ resummation provided by the

693: supergravity approximation \cite{BP99}.

694: 

695: 

696: The study of the embedding of the D6-brane probes is greatly

697: simplified by working in isotropic coordinates in the

698: 56789-directions. Towards this end, we first define a new

699: radial coordinate, $\rho$, related to $U$ by

700: \be

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

702: \frac{\ukk^3}{4\rho^{3/2}}\right)^{2/3} \,,

703: \label{isomer}

704: \ee

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

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

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

708: \eqn{metric} becomes

709: \be

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

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

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

713: \label{isometric1}

714: \ee

715: where

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

717: \ee

718: Here $U$ is now thought of as a function of $\rho$. To

719: exploit the symmetries of the D6-brane embedding, we finally

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

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

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

723: \be

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

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

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

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

728: \ee

729: where $\rho^2 = \l^2 + r^2$. The $\ua$ symmetry corresponds here to

730: shifts of the $\phi$ coordinate.

731: 

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

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

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

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

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

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

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

739: respectively. We also set $\tau=\mbox{constant}$, as

740: corresponds to D6-branes localized in the circle direction.

741: 

742: The function $r(\l)$ is determined by the requirement that the

743: equations of motion of the D6-brane in the D4-brane background be

744: satisfied. In the supersymmetric limit, $\ukk=0$, $r(\l)=2\pi\ls^{\,2}\,\mq$ is a

745: solution for any (constant) quark mass $\mq$, as depicted in figure

746: \ref{d6-embedding}(a); this reflects the BPS nature of the system.

747: If the quarks are massive then the D6-brane

748: embedding is not invariant under rotations in the 89-plane and the

749: $\ua$ symmetry is explicitly broken. If instead $\mq=0$ then the $\ua$

750: symmetry is preserved.

751: \FIGURE{\epsfig{file=d6-embedding.eps, height=8cm}

752: \caption{(a) D6-brane embedding if $\ukk=0$, for some non-zero value

753:   of $\mq$. (b) D6-brane embedding for $\ukk \neq 0$ and $\mq=0$.}

754: \label{d6-embedding}}

755: 

756: %R a 2\pi\ls^2 inserted above and below -- not quite the notation

757: %R of KMMW03 but we did NOT rescale r. do we need to warn the reader?

758: If $\ukk \neq 0$ supersymmetry is broken and $r(\l)=\mbox{constant}$

759: is no longer a solution. The new solution is found as follows.

760: For large $\l$, the equation of motion linearizes, and its general

761: solution is

762: \be

763: r(\l) \simeq 2\pi\ls^{\,2}\,\mq + \fc{c}{\l} + O(\lambda^{-2}) \,.

764: \label{rt}

765: \ee

766: As explained in \cite{KMMW03}, the field $r(\l)$ is dual to the quark

767: bilinear operator $\bar{\psi} \psi$, so the constants $\mq$ and

768: $c$ are dual to the quark mass and the chiral condensate,

769: respectively. The requirement that the solution be regular everywhere

770: imposes a constraint between these two constants, that is, determines

771: $c=c(\mq)$. This is exactly as expected on field theory grounds, since the

772: chiral condensate should be dynamically determined once the quark mass

773: is specified.

774: 

775: The solution for massless quarks is depicted in figure

776: \ref{d6-embedding}(b).

777: We see that, although the D6-branes align asymptotically with

778: the $\l$-axis, they develop a `bump' in the 89-plane as

779: $\l \ra 0$, that is, $r(0) \neq 0$. The D6-brane embedding is

780: therefore not invariant under rotations in the 89-plane, and hence the

781: $\ua$ symmetry is {\it spontaneously} broken. The reason why this

782: breakng is spontaneous is that both the boundary condition,

783: $r(\infty)=0$, and the D6-brane equation of motion, are $\ua$-invariant,

784: yet the lowest-energy solution breaks the $\ua$ symmetry.

785: On gauge theory grounds, we expect this breaking to be caused by a

786: non-zero chiral condensate, $\cc \neq 0$. This is confirmed in the

787: gravity description by the fact that $c(\mq)$ approaches a non-zero

788: constant in the limit $\mq \ra 0$ \cite{KMMW03}.

789: 

790: The D6-brane embedding described above must be thought as the `vacuum

791: state' of the D6-branes in the D4-brane background. By studying

792: fluctuations around this embedding, the spectrum of (a certain class

793: of) scalar and pseudoscalar mesons was computed in \cite{KMMW03}.

794: In particular, for $\nf=1$, a massless, pseudoscalar meson was

795: found. This is the Goldstone boson expected from the spontaneous

796: breaking of $\ua$ symmetry, that is, the $\ep$. The corresponding

797: mode in the gravity description is the zero mode associated to

798: rotations of the D6-brane in the 89-plane, that is, it corresponds

799: to fluctuations of the D6-brane worldvolume field $\phi$.\footnote{

800: The odd-parity nature of these fluctuations is due to the fact that

801: a gauge-theory parity transformation acts on $X_8+ i X_9=r e^{i\phi}$

802: by complex conjugation. See \cite{KMMW03} for a detailed discussion.}

803: 

804: 

805: 

806: 

807: 

808: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

809: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

810: \section{The Anomaly-induced Potential and Glueball Mixings}

811: \label{anomaly}

812: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

813: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

814: 

815: In this section we discuss the general structure of $1/\nc$

816: corrections responsible for the generation of a potential that lifts

817: the $\ep$ meson. We first show that the introduction of D6-branes

818: corresponding to massless quarks allows the

819: $\theta$-dependence of the supergravity description to be shifted

820: away, as expected on field theory grounds. We then isolate the

821: relevant string diagrams and study their main properties in the

822: supergravity approximation.

823: 

824: 

825: 

826: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

827: \subsection{The anomaly relation in the ultraviolet regime}

828: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

829: 

830: At very high energies, the string model based on $\nc$ D4-branes and

831: $\nf$ D6-branes realizes the anomalous $\ua$ symmetry of QCD as an R-symmetry

832: on their common $\bbr{4}$ worldvolume. Since this symmetry is

833: anomalous, the

834: $\ua$ rotation of the D6-brane fields by an angle

835: $\alpha$, as specified in (\ref{rsim}), must be equivalent to a

836: shift of the effective

837: $\theta$-angle in (\ref{thec}) by

838: \be

839: \label{shift}

840: \int_{S^1} \cone \rightarrow \int_{S^1} \cone + \nf\, \alpha \,,

841: \ee

842: so that the dependence on the microscopic $\t$-angle can be eliminated

843: by a phase rotation  of the

844: $X_8+iX_9$ field, as argued

845: in the Introduction.

846: 

847: In the dual gravity description, the microscopic coupling (\ref{thec})

848: and the elementary quark fields $\psi_\mt{L,R}$ are not directly

849: visible, since the D4-branes are replaced by the background

850: \eqn{metric} and the effective action only contains colour-singlet degrees of

851: freedom. However, the fact that the dependence on the microscopic

852: $\t$-angle can be eliminated, as implied by the anomaly,

853: still follows from topological properties of the RR fluxes induced

854: by the D6-branes, as we now show.

855: 

856: In the gravity description, the microscopic $\t$-angle is defined by

857: the boundary condition \eqn{deft}. The key observation is that the

858: D6-branes' contribution to this integral has precisely the form

859: (\ref{shift}). To see this, we recall that, by definition, the

860: D6-branes are magnetic sources for the RR two-form, such that the

861: flux through any two-sphere that links the D6-branes is

862: \be

863: \label{links}

864: \int_{S^2} \ftwo = 2\pi\nf \;.

865: \ee

866: The D6-branes are localized in the $\tau$-direction, and, in the chiral

867: limit, they are also asymptotically localized at the origin of the

868: $89$-plane, \ie, $\lim_{\lambda\to \infty} r(\lambda)=0$. A two-sphere

869: surrounding the D6-branes in this region is shown in figure

870: \ref{flux}. Since $\tau$ is periodically identified, this two-sphere

871: can be continously deformed to a torus, $T^2$, parametrized by

872: $\tau$ and $\phi$ at fixed $r$ and (large) $\l$. Since $\ftwo$ is a

873: closed form, the captured flux is the same, \ie,

874: \be

875: \label{ff}

876: \int_{T^2} \ftwo = 2\pi\nf \;.

877: \ee

878: Since a translation in $\phi$ is an isometry of the background, it

879: follows that the flux through any strip defined by two angles

880: $\phi_1$ and $\phi_2$, as in the figure, must be proportional to the

881: area of the strip, that is,

882: \be

883: \int_{\mt{Strip}} \ftwo = \nf (\phi_2 - \phi_1) \;.

884: \ee

885: Note that this result relies crucially on the fact that all

886: integrals above are evaluated in the UV, \ie, in the limit

887: $\l \ra \infty$, as appropriate to the definition of the {\it

888: microscopic} $\t$-angle. In this limit the D6-branes lie at the

889: origin of the 89-plane and the integrals above are insensitive to

890: the deformation of the D6-branes in the region $\l \ra 0$.

891: \FIGURE{\epsfig{file=flux.eps, height=7cm}

892: \caption{Asymptotically, the D6-branes lie at the origin of the

893: 89-plane and are localized in the $\tau$-direction.}

894: \label{flux}}

895: Finally, since locally we have $\ftwo=d\cone$, we can use Stokes'

896: theorem to write

897: \be

898: \int_{\mt{Strip}} \ftwo =

899: \int_{S^1_{\phi_\mt{2}}} \cone - \int_{S^1_{\phi_\mt{1}}} \cone \,,

900: \ee

901: where $S^1_{\phi_i}$ is parametrized by $\tau$ at $\phi=\phi_i$.

902: Combining these results we deduce that the Wilson line of $\cone$

903: at a given angle $\alpha$, as induced by the D6-branes, is

904: \be

905: \label{ind}

906: \int_{S^1_{\alpha}} \cone = \nf \, \alpha \,,

907: \ee

908: where we have set to zero a possible additive constant by choosing

909: the origin of the polar angle $\alpha$ appropriately.

910: If, in addition, there is a background value for this Wilson line

911: (an asymptotically flat connection defining the $\theta$-angle) then

912: the total value of the Wilson line is

913: \be

914: \label{tot}

915: \int_{S^1_\alpha} \cone = \theta + \nf \,\alpha \;.

916: \ee

917: Under a rotation by

918: angle $\alpha$ in the $89$-plane of the background, the `Dirac

919: sheet' singularity that is used to define $\cone$ (extending as a

920: string in the plane $(r,\phi)$ at $\phi=0$) rotates by minus this

921: same angle and shifts the theta angle according to (\ref{tot}).

922:  Since the position of this Dirac sheet is a gauge artefact, we see

923: explicitly how the microscopic $\t$-angle can be shifted away by a

924: $\ua$ transformation.

925: 

926: This supergravity argument proves that the physics is independent of

927: the microscopic $\t$-angle when the D6-branes are asymptotically

928: located at the origin of the $89$-plane, \ie, in the chiral limit.

929: Supersymmetry breaking at a scale $\mkk$ implies that a shift

930: $\delta \theta$ of the $\t$-angle by a change of the RR two-form $\ftwo$

931: costs energy $\chi_g \,\t \, \delta \theta$, to linear order in

932: $\delta \theta$. At the same time, chiral symmetry breaking implies that

933: a linear potential $\chi_g \,\t\, \nf \, \phi$ for the D6-brane

934: coordinate $\phi$ must be somehow generated, so that the complete

935: potential energy is only a function of the $\ua$-invariant

936: combination $\t + \nf \phi$. This is  checked in section

937: \ref{quanti} by an explicit computation.

938: 

939:  Since $\phi$ starts life in ten dimensions as a

940: gauge field, the mass term $\half \chi_g \,\nf^2 \phi^2$

941:  looks very much like a Green--Schwarz

942: correction to the field-strength of the $\cone$ axion field. It

943: would be interesting to confirm this by  finding a more geometrical

944: construction in ten-dimensional notation.

945: 

946: 

947: 

948: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

949: \subsection{String contributions to the potential}

950: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

951: 

952: The $\theta$-dependence computed by Witten in the pure-glue sector, plus

953: the above anomaly argument, constrain the leading potential of the $\ep$

954: field in the $k$-th branch to be

955: \be\label{potl}

956: V(\phi)^{(k)} = \half \chi_g\,(\theta + 2\pi k+  \nf\,\phi\,)^2 \,.

957: \ee

958: Mimicking the field theory arguments of

959: \cite{Witten79, Veneziano79} we can identify

960: the candidate string diagrams that generate the mass term by considering

961: string contributions to the two-point function of the {\it total} topological

962: susceptibility $\chi_{\rm total}$, which vanishes because of the anomalous

963: $\ua$ symmetry. In the string loop expansion, the pure-glue contribution

964: calculated in (\ref{tosu}) must be cancelled by contributions from meson

965: diagrams.

966: \FIGURE{\epsfig{file=disc.eps, height=3.5cm, width=6.5cm}

967: \caption{The leading open-string correction to the two-point function

968: $U_1(p)$ of topological charge operators ${\cal Q} = \tr\,F\wedge F$.}

969: \label{disc}

970: }

971: The leading such diagram is depicted in figure \ref{disc}

972: and features a single open-string boundary attached to the D6-branes,

973: together with two closed-string vertex operators

974: dual to the anomaly operator $\calq = \tr\,F\wedge F$. This diagram is the

975: string counterpart of the OZI-suppressed quark annhilation diagrams

976: considered in \cite{RGG75, Witten79, Veneziano79}.

977: 

978: A spectral decomposition of this diagram yields

979: \be

980: \label{spec}

981: U_1 (p) = {\nf \over \nc} \sum_n {|C_n|^2 \over p^2 + m_n^2}\;,

982: \ee

983: where $C_n = O(1)$ in the large-$\nc$ limit and the meson spectrum

984: $m_n$, calculated from the fluctuations of the D6-brane, is also of

985: $O(1)$, except for the lowest excitation, the $\ep$, which is massless.

986: The contribution to the topological susceptibility arises from the

987: formal $p \rightarrow 0$ limit, which of course is infrared-divergent

988: because of the massless $\ep$ meson.

989: \FIGURE{\epsfig{file=sum.eps, height=3.5cm, width=6.5cm}

990: \caption{Diagram $U_h (p)$ with $h$ open-string boundaries.}

991: \label{sum}

992: }

993: 

994: A standard procedure to resolve this infrared divergence is to resum

995: a chain of highly divergent diagrams, $U_h (p)$, of the form

996: depicted in figure \ref{sum}, where the index $h$ stands for the

997: number of open-string boundaries. Isolating the massless meson in

998: $h$ intermediate propagators, we see that $U_h (k)$ diverges in the

999: infrared as $(\nf /  \nc p^2 )^h$. Summing up the geometric series

1000: of such terms induces a 1PI self-energy contribution, of order $\nf

1001: /\nc$, given by the cylinder diagram in figure \ref{cylinder}(a).

1002: The same diagram with the other possible inequivalent insertion of

1003: the $\eta'$ field (contributing of course self-energy corrections of

1004: the same order) is depicted in figure \ref{cylinder}(b).

1005: 

1006: The closed and open string interpretations\footnote{By this we mean

1007:   those obtained by cutting the

1008:   diagrams in such a way that the intermediate states are closed or open

1009:   strings, respectively. Of course, in both cases the external states

1010:   are an open string state, namely, the $\eta'$.} of these diagrams is

1011: given in figures \ref{closed} and \ref{open}, respectively. Note

1012: that the indices carried by the double lines are not $SU(\nc)$ colour

1013: indices (there are only $SU(\nc)$ singlets in the gravity description)

1014: but flavour indices of $SU(\nf)$,\footnote{This is a global symmetry

1015:   of the boundary field theory, and a gauge symmetry on the

1016:   worldvolume of the D6-branes in the dual gravity description.}

1017: under which the pion fields transform in the adjoint representation

1018: but the $\eta'$ is inert.

1019: 

1020: As shown by its open string representation, the diagram in figure

1021: \ref{cylinder}(b) is equivalent to a standard one-loop correction in

1022: the effective meson theory, and this contributions is common to

1023: singlet and non-singlet mesons. In contrast, the diagram in figure

1024: \ref{cylinder}(a) will only couple

1025: to the flavour singlet mesons and so distinguishes the behaviour

1026: of the $\eta'$ meson from the rest of the `Goldstone' modes.

1027: This would suggest that, at a quantitative level, this diagram gives

1028: the most important contribution to the mass of the $\eta'$.

1029: 

1030: In order to contribute to the $\ep$ mass, either of these self-energy

1031: corrections must shift the zero-momentum pole of the large-$\nc$

1032: meson propagator. Unfortunately, direct computation of the full string

1033: diagrams is not possible in the background in question, since we are

1034: restricted to the supergravity approximation. It is then interesting to

1035: separate the part of figure \ref{cylinder}(a) corresponding to the

1036: exchange of supergravity modes from a stringy `contact term' coming

1037: from the infinite tower of closed string modes and possible

1038: contributions at the boundary of worldsheet moduli space.

1039: The contribution of a {\it finite} number of low-lying glueball modes

1040: with mass $M_n$ shifts the $\eta'$ pole mass-squared by

1041: \be\label{mixshift}

1042: \delta m_\ep^2 = - \sum_n {g_n (0)^2 \over M_n^2}

1043: \;,

1044: \ee

1045: where $g_n (0)$ stands for the zero-momentum limit of the

1046: glueball-$\ep$ coupling, which must be non-vanishing for this

1047: contribution to be non-trivial.  The shift (\ref{mixshift}) has

1048: the `wrong' sign though, so the stringy contact term (the high-energy

1049: part of the full string diagram) must be positive and all-important

1050: at the quantitative level.

1051: 

1052: We will elaborate further on these issues in the last section.

1053: 

1054: \FIGURE{\epsfig{file=string-cylinders.eps, width=12cm}

1055: \caption{Basic cylinder diagram of order $\nf/\nc$, with the two possible

1056: inequivalent insertions of the $\eta'$ field.}

1057: \label{cylinder}}

1058: \FIGURE{\epsfig{file=closed-interp.eps, width=12cm}

1059: \caption{Closed string interpretation of the cylinder diagrams of

1060:   figure \ref{cylinder}. The representation

1061:   (a) exhibits the fact that diagram \ref{cylinder}(a) contributes

1062:   low-lying glueball mixing at tree level (supergravity fields)

1063:   plus a high-energy contact term coming from the infinite tower of closed

1064:   string

1065:   states. The $\eta'$-glueball coupling in (a) is of order $\sqrt{\nf/\nc}$.

1066:   The strength of the glueball tadpole in (b) is of order $\nf$,

1067:   whereas the cubic $\eta'$$\eta'$-glueball vertex is of order $1/\nc$

1068:   --- see Appendix A.}

1069: \label{closed}}

1070: \FIGURE{\epsfig{file=open-interp.eps, width=12cm} \caption{Open

1071: string interpretation of the cylinder diagrams of figure

1072: \ref{cylinder}. (b) is the standard one-loop meson self-energy; each

1073: of the two vertices is of order $1/\sqrt{\nc}$, and the internal

1074: loop yields a factor of $\nf$. The internal meson propagators in (a)

1075: are both twisted. The factor of $\nf$ now comes from the fact that

1076: the flavour of the incoming lines need not be the same as that of

1077: the outgoing lines.}

1078: \label{open}}

1079: 

1080: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1081: \subsection{Meson-glueball mixing}

1082: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1083: 

1084: %R add subscript c to closed string field

1085: In this subsection we show that $g_n (0) \neq 0$ by analysing the

1086: glueball-$\ep$ mixing at the supergravity level.

1087: Quite generally, any closed-string field $\Phi_c$ that

1088: is sourced by the D6-branes and has non-trivial wave-function with respect

1089: to the $\phi$ angle is subject to mixing with the $\ep$ meson.

1090: Expanding $\Phi_c$ in Fourier modes one has

1091: \be

1092: \label{expp}

1093: \Phi_c(\phi) = \sum_n {\cal G}_n \,e^{-in\phi}

1094: \;,\ee

1095: where the normalizable modes ${\cal G}_n$, when pulled back to the

1096: $\bbr{4}$ factor in the D6 world-volume, represent  glueballs of

1097: $\ua$ charge $n$. If $\Phi_c$ enters linearly the world-volume theory

1098: on the D6-branes, equation (\ref{expp}) gives the required non-derivative

1099: couplings to the $\ep$ meson. The prototypical example is the dilaton

1100: term in the Born-Infeld action:

1101: %R \ls factor inserted; gs removed as it lives in our e^Phi

1102: \be

1103: \label{bi}

1104: {1 \over (2\pi)^6 \,\ls^{\,\,7} } \sum_{j=1}^{\nf}

1105:  \int_{\Sigma_7} e^{-\Phi} \,\sqrt{-\det G_\mt{ind} (\partial \phi_j)}

1106: \;,

1107: \ee

1108: where $\Sigma_7$ is the D6-branes worldvolume,

1109: \be

1110: \label{ex}

1111: \sqrt{-\det G_\mt{ind} \,(\partial \phi_j)} =

1112: 1 + O\left[(\partial \phi_j)^2 \right]

1113: \ee

1114: and the corresponding pullbacks are understood to each of the coinciding

1115: $\nf$ branes. Selecting the non-derivative term in the expansion of

1116: the square root we have couplings of the form (\ref{expp}).

1117: In chiral-lagrangian notation, assembling the collective coordinates

1118: of the D6-branes in a Goldstone-boson diagonal matrix

1119: $\Sigma = {\rm diag}\,(e^{i\phi_j})$, we have terms of the form

1120: \be\label{cl}

1121: \nc \sum_n \int_{\bR^4} {\cal G}_n \,\tr\,\Sigma^n + {\rm h.c.}

1122: \ee

1123: These glueballs, being charged with respect to the $\ua$ group, are

1124: `Kaluza--Klein artefacts', not present in real QCD. In fact, the couplings

1125: (\ref{cl}) respect the $\ua$ symmetry and cannot induce a potential that

1126: breaks it upon integrating out the glueballs. For example, at tree level

1127: we generate terms proportional to $\tr\,\Sigma^n \cdot \tr\,\Sigma^{-n}$,

1128: because the glueball propagator couples ${\cal G}_n$ and ${\cal G}_{-n}$

1129: ($n$ being a Kaluza--Klein momentum). The global phase $e^{i\phi}$ drops

1130: from these expressions and we see that such couplings do not generate

1131: a potential for the $\ep$ particle. This is just as well, since such

1132: contributions seem completely independent of the $\t$-dependence, as dictated

1133: by the Witten--Veneziano formula.

1134: 

1135: In fact, the candidate glueballs are selected by the general arguments

1136: in the Introduction. First, we expect the required couplings to show the

1137: characteristic multivaluedness of $\t$-dependence at large $\nc$, \ie, we

1138: expect the coupling to be a function of $-i\,\log\,{\rm det}\,\Sigma \sim

1139: \nf\,\phi$, precisely linear in the angular coordinate, so that the

1140: angular periodicity of the effective action would require an explicit sum over

1141: different branches.  The supposed relation to $\t$-dependence suggests

1142: that we investigate the glueballs in the RR sector of the closed-string theory.

1143: 

1144: Natural candidates are the normalizable modes of the RR potential

1145: $\cone$, or, equivalently, of its Hodge dual $\csev$, since these give

1146: rise to {\it pseudo}-scalar glueballs. In the absence of D6-brane

1147: sources, it is truly equivalent to work with $\cone$ or $\csev$. The

1148: D6-branes, however, couple minimally to $\csev$ through the

1149: Wess--Zumino term

1150: %R ls factor inserted

1151: \be

1152: \label{wzt}

1153: S_\mt{WZ} = {\nf \over (2\pi\ls)^6} \int_{\Sigma_7} \csev \,.

1154: \ee

1155: In terms of $\csev$, this coupling is both local and can be defined

1156: for off-shell values of the RR seven-form potential. For on-shell configurations

1157: this coupling can be reexpressed in terms of $\cone$ at the expense of

1158: introducing non-locality. However, since we wish to exhibit

1159: $\eta'$-glueball couplings at zero momentum, which are necessarily

1160: off-shell, we must work with $\csev$.

1161: 

1162: Now we wish to demonstrate that \eqn{wzt} contains a linear coupling to

1163: the $\phi$ field when reduced to the $\bbr{4}$ factor of the space-time.

1164: Towards this end, let us consider the following ansatz for fluctuations:

1165: $\csev = (\phi+\phi_0) W_{\it 7}$,

1166: where $\phi_0$ is a constant and $W_{\it 7}$ is the $\phi$-independent

1167: seven-form

1168: \bea

1169: W_{\it 7} = &&- G(x) \, h(U) \, r \l^2 \, (r d\l - \l dr) \w

1170: d\Omega_{\it 2} \w dV_{\it 4} \nn

1171: && + \tilde{h}(U) \, r\l^2 \, d\l \w dr \w d\Omega_{\it 2} \w

1172: i_{N(x)} dV_{\it 4} \,.

1173: \label{W}

1174: \eea

1175: Here $d\Omega_{\it 2}$ and $dV_{\it 4}$ are the volume forms on the $S^2$ wrapped

1176: by the D6-branes and on the $\bbr{4}$ factor, respectively.

1177: $G(x)$ is a pseudoscalar field and $N^\mu(x)$ is a vector field with

1178: 

1179: \be

1180: i_{N(x)} dV_{\it 4} =

1181: \fc{1}{3!} N^\mu (x) \, \e_{\mu\nu\a\b} \, dx^\nu \w dx^\a \w dx^\b \,.

1182: \ee

1183: Finally, $h(U)$ and $\tilde{h}(U)$ are radial profiles to be determined.

1184: Note that $\csev$, not being gauge-invariant, is

1185: allowed to be multivalued in $\phi$,\footnote{An average over the action

1186:   $\phi_0 \ra \phi_0 +2\pi$ can restore the angular character of

1187:   $\phi$ that is lost in the expression for $\csev$, and is the

1188:   counterpart of the average over large-$\nc$ branches of

1189:   $\t$-dependence.}

1190: but its gauge-invariant field strength, $F_{\it 8}=d\csev$, must be

1191: single-valued. This restriction

1192: forces $W_{\it 7}$ to be closed, which in turn implies

1193: \be

1194: G(x) \, \left[ 5h(U) + \rho \, \fc{dU}{d\rho} \, h'(U) \right] +

1195: \pa_\mu N^\mu (x) \, \tilde{h}(U)=0 \,,

1196: \ee

1197: and therefore

1198: \be

1199: \tilde{h} = 5h + \rho \, \fc{dU}{d\rho} \, h' \sac \pa_\mu N^\mu = - G \,.

1200: \label{cons}

1201: \ee

1202: We see that closure of $W_{\it 7}$ relates the two radial functions,

1203: as well as the scalar and the vector. Under these conditions

1204: $F_{\it 8} = d\phi \w W_{\it 7}$, which is, of course, single-valued.

1205: 

1206: We may regard the second equation above as a constraint on $N$,

1207: and $G$ as a yet totally unconstrained pseudoscalar glueball field.

1208: In fact, $N$ is not an independent field

1209: on-shell, but is completely determined by $G$. Indeed, the equations

1210: of motion for $G$ and $N$ come from $d * F_{\it 8}=0$, or

1211: equivalently the Bianchi identity

1212: $d F_{\it 2} =0$. A straightforward calculation yields

1213: \be

1214: F_{\it 2} = * F_{\it 8} = G(x) \, \tilde{H}(U) \, \rho

1215: \left(\fc{dU}{d\rho} \right)^{-1} \, d\tau \w dU -

1216: H(U) \, N_\mu \, d\tau \w dx^\mu \,,

1217: \ee

1218: where

1219: \bea

1220: \tilde{H}(U) &=& - \left( \fc{U}{R} \right)^{-9/4} \, K(\rho)^{-3/2} \,

1221: f(U)^{1/2} \, h(U) \,, \nn

1222: H(U) &=& \left( \fc{U}{R} \right)^{-3/4} \, K(\rho)^{-5/2} \,

1223: f(U)^{1/2} \, \tilde{h}(U) \,,

1224: \eea

1225: and we have made use of the fact that

1226: \be

1227: dU = \fc{dU}{d\rho} \left( \fc{\l}{\rho} \, d\l +

1228: \fc{r}{\rho} \, dr \right) \,.

1229: \ee

1230: Closure of $F_{\it 2}$ then implies

1231: \be

1232: H' = M^2 \, \tilde{H} \, \rho \, \left(\fc{dU}{d\rho} \right)^{-1}

1233: \sac N_\mu = -\fc{1}{M^2} \, \pa_\mu G

1234: \label{all}

1235: \ee

1236: for some constant $M^2$. As anticipated, the second equation above

1237: determines $N$ in terms of $G$. Combined with the constraints

1238: \eqn{cons}, it imposes  the on-shell condition for $G$:

1239: \be

1240: \pa_\mu \pa^\mu \, G = M^2 \, G \,.

1241: \ee

1242: Further combining the first constraint in \eqn{cons} with

1243: the first equation in \eqn{all} yields a second-order ODE for

1244: the radial profile. This equation provides an eigenvalue problem

1245: that determines the pseudoscalar glueball mass spectrum, $M_n^2$,

1246: as well as the corresponding normalizable radial profiles,

1247: $h_n, \tilde{h}_n$. Once these profiles are known, the

1248: non-derivative $\phi$--$G_n$ couplings arise from the Wess--Zumino term

1249: by pulling back $\csev$ onto the D6-branes worldvolume and reducing

1250: the result along the $S^2$ and along the radial direction down to

1251: four-dmensions. The coordinate $\phi$ is pulled back into a field

1252: $\phi_0+\phi(x)$ that depends {\it only} on the $\bbr{4}$ coordinates.

1253: The form $W_{\it 7}$ is pulled-back on the ground state of the

1254: D6-branes embedding, $\tau=0, r=r(\l)$, since we are only interested

1255: in the couplings of the $\eta'$ and not the rest of the mesons.

1256: The non-derivative couplings originate from the first summand in

1257: $W_{\it 7}$, and take the form (setting $\phi_0=0$)

1258: \be

1259: \label{mixc}

1260: S_\mt{WZ} \ra {\nf f_\pi \over 2\sqrt{\nc}} \,

1261: g_n \,\int_{{\bR}^4} \phi(x) \, G(x)

1262: =\sqrt{\nf \over \nc} \, g_n \int_{{\bR}^4} \eta'(x) \,

1263: G(x) \,,

1264: \ee

1265: where

1266: \be

1267: \label{intf}

1268: {f_\pi \over 2\sqrt{\nc} } \;g_n = {{\rm vol}(S^2)

1269:  \over (2\pi\ls)^6} \int_0^\infty d\lambda\;

1270: h_n (U(\lambda)) \, r^2 \l \,\left( r-\lambda {\dot r} \right) \,,

1271: \ee

1272: and we recall that $\phi$ and $\eta'$ are related as in \eqn{cann}.

1273: In principle, these couplings can be evaluated numerically, given the

1274: embedding $r(\l)$ and the eigenmode profiles $h_n(U)$. However, the

1275: fact that they are in general non-vanishing is already

1276: an important result, for it confirms that the

1277: cylinder diagram in figure \ref{cylinder}(a) is capable of generating a

1278: potential for the $\ep$ with the right properties.

1279: 

1280: 

1281: 

1282: 

1283: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1284: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1285: \section{A Quantitative Check  to Order $1/\sqrt{\nc}$}

1286: \label{quanti}

1287: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1288: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1289: 

1290: 

1291: We have argued in the previous sections that certain quantum corrections to

1292: the supergravity model with probe D6-branes \cite{KMMW03} generate a

1293: potential for the $\ep$ meson of the form

1294: \be

1295: \label{thep}

1296: V(\ep\,) = \half\,\chi_g\,\left(\theta +

1297: {2\sqrt{\nf} \over f_\pi}\,\ep\,\right)^2

1298: \ee

1299: to leading order in the $1/\nc$ expansion. In this expression, we have considered

1300: the $k=0$ branch of the vacuum energy and we have fixed the additive normalization

1301: of the $\ep$ field so that $V(\ep =0)$ equals the pure-glue vacuum energy

1302: derived in equation \eqn{wr}. With these conventions, taking into account that

1303: $f_\pi = O(\sqrt{\nc})$, we can expand the square and separate the pure-glue

1304: term of $O(1)$, the Witten--Veneziano mass term of $O(\nf/\nc)$, and a

1305: cross term of $O(\sqrt{\nf /\nc})$ which acts as a tadpole upon expanding

1306: the potential around the wrong vacuum, $\ep=0$.

1307: In this section we present a calculation of this linear term by two independent

1308: methods, one based on a closed-string calculation plus the anomaly argument,

1309: and the other based on a direct open-closed string coupling.

1310: 

1311: In terms of the $\phi$ field, the `tadpole' term can be identified as

1312: \be\label{tadpole}

1313: {\rm tadpole} ={\cal T}= \chi_g\,\t\,\nf\, \phi\;,

1314: \ee

1315: and we may evaluate it in two independent ways. First, we can use the

1316: explicit supergravity calculation \eqn{tosu} of the pure-glue

1317: topological susceptibility and introduce the $\phi$-dependence

1318: via the anomaly argument $\t \rightarrow \t + \nf\,\phi$. We find

1319: \be

1320: \label{tuno}

1321: {\cal T} = {C \, \theta \,\nf \over 3\cdot 2^4 \cdot \pi^5\cdot\ls^{\,\,6}}

1322: \, \phi \,,

1323: \ee

1324: %R ls factor and comment added

1325: where we remind the reader that $C= 3\, \ukk^3 /\d \tau$ from

1326: (\ref{wittf}). We emphasize that this calculation only uses the

1327: closed-string sector, plus the microscopic anomaly argument.

1328: 

1329: On the other hand, we may read the linear term directly from the

1330: Wess--Zumino action (\ref{wzt}) for the particular seven-form

1331: $\csev$ that is induced by the $\theta$-angle background \eqn{wittf}.

1332: Setting $k=0$ in this equation, a straight-forward calculation shows

1333: that the dual seven-form potential is given (locally) by

1334: $\csev = \phi \, \osev$, where

1335: \be

1336: \label{ome}

1337: \osev = {C \, \theta \over U^4} \,B(U) \,(r\,d\lambda - \lambda\,dr)

1338: \wedge d\Omega_{\it 2} \wedge dV_{\it 4} \,,

1339: \ee

1340: where

1341: \be

1342: B(U) = \fc{1}{\rho} \, \fc{dU}{d\rho} \,

1343: \left( \fc{U}{R} \right)^{9/4} \, \fc{K(\rho)^{3/2}}{f(\rho)^{1/2}}

1344: \l^2 r \,,

1345: \ee

1346: and we have set to zero the additive normalization of $\phi$, as well as

1347: the discrete $2\pi$-shift implementing the large-$\nc$ branches of vacua.

1348: Calculating the pull-back of $\csev$ on the D6 world-volume,

1349: as in the previous section, we obtain

1350: \be

1351: \label{calta}

1352: {\nf \over (2\pi\ls)^6} \int_{\Sigma_7} \phi \,\osev =

1353: \int_{\bR^4} {\cal T} \,,

1354: \ee

1355: where

1356: \be

1357: \label{tadd}

1358: {\cal T} = C\, \t \, \nf \, \phi \,

1359: {{\rm vol}(S^2) \over (2\pi\ls)^6} \int_0^\infty d\lambda

1360: {H(U(\lambda)) \over U(\lambda)^4} \,\left(r-\lambda {\dot r}\right)

1361: \ee

1362: or, using $\rho^2 (\lambda) = r^2 (\lambda) + \lambda^2$,

1363: \be

1364: \label{ct}

1365: {\cal T}= {C \, \theta \, \nf \over 2^4 \cdot \pi^5\cdot\ls^{\,\,6}} \, \phi \,

1366: \int_0^\infty d\lambda {\lambda^2 r \over \rho^5} (r-\lambda {\dot r})

1367: \,.

1368: \ee

1369: %R ls factors in above three eq.'s

1370: Agreement with \eqn{tuno} requires the last integral to equal

1371: $1/3$. Remarkably, this is so, for the integral can be transformed into

1372: \be

1373: \int_0^\infty d\l {\l^2 \over \rho^4} \,(\rho-\l {\dot \rho}) =

1374: \fc{1}{3} \int_0^\infty d\l \, \fc{d}{d\l} \left( \fc{\l^3}{\rho^3}\right)

1375: = \fc{1}{3} \,,

1376: \ee

1377: where we have used in the last step the fact that $r(\lambda)$ remains

1378: bounded as $\lambda \rightarrow \infty$ in the D6-brane embedding.

1379: 

1380: A more geometrical version of this calculation  can be given as

1381: follows. We have argued that, locally in the $\phi$-direction,

1382: $\csev = \phi \,\osev$. The tadpole comes just from the integration of

1383: $\osev$ over $\Sigma_7$, the equilibrium worldvolume of the D6-branes

1384: at $\phi=0$. Now, let $\Sigma_8$ be the hypersurface that results from

1385: rotating the worldvolume around the angle $\phi$. Since

1386: $\partial_\phi\, \osev =0$ for the $\theta$-induced form \eqn{ome},

1387: we can write

1388: \be

1389: \int_{\Sigma_7} \osev = {1\over 2\pi} \int_{\Sigma_8} d\phi \wedge \osev

1390: = {1\over 2\pi} \int_{\Sigma_8} \feig \,.

1391: \ee

1392: In addition, $\Sigma_8$ is topologically equivalent to the $S^4$

1393: at fixed $U$ coordinate, times the spacetime $\bbr{4}$ factor. Since

1394: $\feig$ is a closed form, we can use Stokes' theorem to write

1395: \be

1396: \int_{\Sigma_7} \osev= {1\over 2\pi} \int_{\bR^4 \times S^4} \feig \,.

1397: \ee

1398: The latter integral is trivially evaluated by computing $\feig$

1399: directly in the original coordinate system, in which it takes the

1400: simple form

1401: \be

1402: \feig = *\,\ftwo = C\,\theta\,d\Omega_{\it 4} \w dV_{\it 4} \,.

1403: \ee

1404: It follows that the prediction for the tadpole is

1405: \be

1406: \label{tadp}

1407: {\cal T} =  \phi\cdot {\nf \over (2\pi\ls)^6} \cdot

1408: {C\theta \over 2\pi} \cdot {\rm vol}(S^4) =

1409: {C\, \theta\,\nf \over 3\cdot 2^4 \cdot \pi^5\cdot\ls^{\,\,6}} \, \phi\,,

1410: \ee

1411: %R ls factors added above

1412: again in perfect numerical agreement with (\ref{tuno}). We regard this

1413: check as highly non-trivial, since the kinetic RR term only knows

1414: about closed strings (glueballs) and the Wess--Zumino term specifies the

1415: direct coupling to the open strings (mesons). The exact agreement for

1416: the tadpole is an indication that the basic physical picture is right.

1417: 

1418: 

1419: 

1420: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1421: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1422: \section{Concluding Remarks}

1423: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1424: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1425: 

1426: In \cite{KMMW03} a string dual of large-$\nc$ QCD with $\nf$

1427: flavours, based on $\nf$ D6-brane probes in a fixed supergravity

1428: background, was studied in detail. It was found that the string

1429: description captures some of the low-energy physics expected on

1430: field theory grounds. In particular, for $\nf=1$, it exhibits

1431: spontaneous chiral symmetry breaking of an $\ua$ symmetry, and the

1432: mesonic spectrum contains a pseudoscalar that is exactly massless at

1433: infinite $\nc$. This is the analog of the $\ep$ meson of large-$\nc$

1434: QCD, and is dual to the zero-mode associated to the motion of the

1435: D6-brane.

1436: 

1437: %R changes in follow para.

1438: As discussed in the Introduction, the importance of the

1439: Witten-Veneziano formula (\ref{wv}) is in producing a qualitative

1440: understanding of the mass splitting of the $\eta'$ meson from the

1441: other light mesons, namely the pseudo-Goldstone modes associated

1442: with the spontaneous breaking of the chiral flavour symmetry. With

1443: multiple flavours, that is, multiple D6-branes, our AdS-like model

1444: produces $\nf^2$ massless pseudoscalars \cite{KMMW03}. However,

1445: only the diagonal mode corresponding to the collective

1446: center-of-mass motion of all the D6-branes is obviously a

1447: Goldstone mode. In Appendix B, we argue that, when the analysis is

1448: taken beyond tree-level, the other $\nf^2-1$ modes acquire masses

1449: of order $(\lambda \nf / \nc)^{1/2} \mkk$, where $\lambda =

1450: \gym^2 \nc$ is the 't Hooft coupling.

1451: As further shown in Appendix B, this precisely matches the

1452: mass of the $\eta'$ at the level of their parametric dependences.

1453: We interpret the masses of the off-diagonal modes as arising from

1454: closed string interactions between the individual D6-branes and

1455: so, to leading order, they are generated by the same string

1456: diagrams as illustrated in figure

1457: \ref{cylinder}(b) -- recall only the diagonal mode couples in

1458: figure \ref{cylinder}(a). This result points to a qualitative

1459: distinction between the physics of our model and QCD: even in the

1460: limit of vanishing quark masses, our entire multiplet of $\nf^2$

1461: light mesons acquires a mass squared of the same order as the

1462: $\ep$, while in QCD only the $\ep$ becomes massive. The additional

1463: mass terms in the present case are natural as the off-diagonal

1464: scalars are not Goldstone modes --- in fact, it is the

1465: masslessness of these modes in the large-$\nc$ limit that was

1466: surprising \cite{KMMW03}. However, it would still be interesting

1467: to refine the estimates made in Appendix B, to see if there is any

1468: dramatic difference in the numerical values of the off-diagonal

1469: and the $\ep$ masses for our model.

1470: 

1471: On general grounds, the identification by Witten \cite{Witten98b}

1472: of an $O(1)$ contribution to the pure-glue topological susceptibility,

1473: plus a microscopic anomaly argument, implies the generation of a potential

1474: for the $\eta'$ with a mass term of $O(\nf/\nc)$, along the lines

1475: of the Witten--Veneziano argument. We have argued that the

1476: cylinder diagram of figure \ref{cylinder}(a) is the relevant stringy

1477: correction responsible for the generation of the $\ep$ mass, in a

1478: string analog of the Isgur-de R\'ujula-Georgi-Glashow mechanism

1479: \cite{RGG75}. We have shown that, in the supergravity approximation,

1480: this diagram induces non-derivative mixings between the glueballs and

1481: the $\ep$ that shift the zero-momentum pole of the $\ep$. However,

1482: this shift by itself would make the $\ep$ tachyonic, and hence we

1483: argued that there is an important contact term

1484: coming from the stringy completion of the

1485: glueball-exchange diagrams in figure \ref{cylinder}(a). This

1486: discussion is natural when one thinks of the worldsheet cylinder

1487: as being long in comparison to its circumference. Coming from the

1488: opposite end of the worldsheet moduli space ({\it i.e.,} a short

1489: cylinder with a large circumference), this diagram has a natural

1490: interpretation, depicted in figure \ref{open}(a), in terms of

1491: an open string, and hence meson, loop,

1492: where the internal meson propagators are now both twisted. From

1493: this point of view, which naturally figures as the UV completion of

1494: the glueball sum, these loop contributions do not have a definite

1495: sign and so certainly allow for the necessary shift with a

1496: positive sign. Of course, string duality tells us that that the

1497: sums over all tree-level glueball exchanges and over one-loop

1498: meson graphs are the same and should not be computed separately.

1499: 

1500: Similarly, the open string or meson loop of figure \ref{open}(b)

1501: also has an interpretation as a

1502: closed string coupling with two $\eta'$ mesons and being absorbed

1503: by the D6-brane, as displayed in figure \ref{closed}(b).

1504: It might be emphasized here that both diagrams in figure \ref{closed}

1505: play an important role in the $\eta'$ physics. This is

1506: particularly evident from the discussion at the beginning of

1507: section 3.3, where we argue that the Neveu-Schwarz glueballs will {\it not}

1508: generate a potential for the $\eta'$. To see how the cancellation

1509: presented there occurs order by order in $\phi$, one would make a

1510: Taylor expansion of the exponentials in equation (\ref{expp}). At

1511: order $\phi^2$, this reveals that the vanishing mass arises

1512: precisely as a cancellation between glueball exchange as in figure

1513: \ref{closed}{a} and tadpole contributions of figure \ref{closed}(b).

1514: 

1515: More generally, the anomaly argument implies that the dependence of

1516: the theory on the $\t$-angle occurs only through the combination

1517: $\t + 2 \sqrt{\nf} \, \ep / f_\pi$, so that the microscopic $\t$-dependence

1518: can be eliminated by a $\ua$ transformation.

1519:  We have verified this statement in

1520: the ultraviolet regime by analizing the RR flux sourced by the

1521: D6-branes.

1522: 

1523: We have strenghened the physical picture by performing a

1524: quantitiative check of the $\eta'$-potential at order

1525: $1/\sqrt{\nc}$. We have computed this potential in two independent

1526: ways. One method employs only the closed string sector, that is, the

1527: pure-glue sector, together with the anomaly argument. The second

1528: method involves the open string sector, that is, the mesonic sector.

1529: We regard the perfect agreement between the two results as a

1530: non-trivial check that the right physics is captured.

1531: 

1532: The `master substitution' $\t \ra \t + 2\sqrt{\nf} \,\ep/f_\pi$

1533: applied to the pure-glue effective Lagrangian generates all

1534: soft-$\ep$ amplitudes. In the supergravity formalism, this implies

1535: precise correlations between the effective couplings of the closed

1536: string sector, and those of the closed string sector to the open

1537: string sector. The simplest of these correlations was checked in

1538: section \ref{quanti}, but it would be interesting to investigate the

1539: more complicated ones, even at a qualitative level.

1540: 

1541: Since the pure-glue $\theta$-dependence comes from the energy of RR

1542: fluxes, the stringy mechanism is akin to a Green--Schwarz modification

1543: of the RR field strengths.\footnote{See \cite{adi} for a study of

1544:   this interpretation in a slightly different model.}

1545: It would be interesting to sharpen the anomaly argument in the

1546: supergravity regime by identifying a ten dimensional anomaly polynomial

1547: that yields the substitution $\t \ra \t + 2 \sqrt{\nf} \, \ep / f_\pi$

1548: as a standard Green--Schwarz modification of the $\ftwo$ field strength,

1549: after appropriate reduction on the D6-branes worldvolume. In this case,

1550: the basic stringy calculation of the ultraviolet contact terms could

1551: be performed locally in the flat limit of the

1552: ten-dimensionsional string theory.

1553: 

1554: 

1555: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1556: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1557: \acknowledgments

1558: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1559: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1560: 

1561: We would like to thank  Margarita Garc\'{\i}a P\'erez,

1562: C\'esar G\'omez and Sean Hartnoll for helpful

1563: discussions, and especially Adi Armoni

1564: for many discussions on the subject of this paper and \cite{adi}.

1565: The work of J.L.F.B. was partially supported by MCyT

1566: and FEDER under grant BFM2002-03881 and the European RTN network

1567: HPRN-CT-2002-00325. The work of C.H. was partially supported by

1568: European Commission (HPRN-CT-200-00148) and CICYT (Spain) and

1569: by the MECD (Spain) through a FPU grant. Research at the Perimeter

1570: Institute is supported in part by funds from NSERC of Canada.

1571: RCM is further supported by an NSERC Discovery grant.

1572: 

1573: \appendix

1574: 

1575: \section{Large-$\nc$ Scalings from Supergravity and DBI}

1576: 

1577: In the presence of $\nf$ D6-branes, the low-energy dynamics on the

1578: gravity side is described by the action

1579: $S=S_\mt{Sugra}+S_\mt{D6}$. Let $\Phi$ and $\phi$ denote collectively

1580: fluctuations of the supergravity and the D6-branes' worldvolume fields,

1581: respectively. Schematically, the action for these fields takes then

1582: the form

1583: \be

1584: S = \nc^2 \, \int_{M_\mt{10}} (\pa \Phi)^2 + \sum_{\ell\geq 2} \Phi^\ell

1585: + \nf \nc \int_{\Sigma_7} (\pa \phi)^2 +

1586: \sum_{m \geq 0, n \geq 1} \Phi^m \, \phi^n \,.

1587: \ee

1588: The coefficients in front of each integral arise from the scalings

1589: with $g_s$ of Newton's constant, $G_N \sim g_s^2$, and the fact that

1590: $g_s \sim 1/ \nc$ in the 't Hooft limit. Further the

1591: D6-branes' tension scales as $T \sim \nf /g_s$ --- note that we are

1592: assuming that $\phi$ is a $U(1)$ field in the $U(\nf)$ gauge group

1593: on the D6-branes, as is the $\eta'$.

1594: 

1595: In terms of canonically normalized fields, defined through

1596: $\Phi \ra \nc \Phi$ and $\phi \ra \sqrt{\nf \nc} \phi$, we find

1597: \be

1598: S = \int_{M_\mt{10}} (\pa \Phi)^2 + \sum_{\ell\geq 2} \nc^{2-\ell} \Phi^\ell

1599: + \int_{\Sigma_7} (\pa \phi)^2 + \sum_{m \geq 0, n \geq 1}

1600: \nf^{1-\fc{n}{2}} \nc^{1-m-\fc{n}{2}} \Phi^m \, \phi^n \,.

1601: \ee

1602: The strength of the different couplings can now be directly read

1603: off. For example, a glueball tadpole (\ie, a closed string one-point

1604: function) is of order $\nf$ while the $\eta'$-glueball coupling is of order

1605: $\sqrt{\nf/\nc}$. These results are used in section 3.2.

1606: 

1607: %R new section

1608: \section{Pseudoscalar Masses Revisited}

1609: 

1610: In this paper, we have argued that the $\ep$ meson, in the model

1611: of \cite{KMMW03}, acquires a mass consistent with the

1612: Witten-Veneziano formula:

1613: %

1614: \be \label{wv2} m_\ep^2 = {4 \nf \over f_\pi^2} \,\chi_g \,. \ee

1615: %

1616: We now wish to evaluate this mass in terms of the microscopic

1617: parameters of the field theory. In the following, we will only

1618: determine the parametric dependence but drop numerical factors.

1619: First, equation (\ref{tosu}) gives the topological susceptibility

1620: as

1621: %

1622: \be \label{sus2} \chi_g\sim\l^3\mkk^4 \ee

1623: %

1624: where $\l=\gym^2\nc$ is the 't Hooft coupling. From (4.18) of

1625: \cite{KMMW03}, one deduces that, up to numerical factors, the pion

1626: decay constant is given by

1627: %

1628: \be \label{fpi} f_\pi^2\sim T_\mt{D6} \ukk^3/\mkk^2 \sim

1629: \nc\l^2\mkk^2\,. \ee

1630: %

1631: The second expression above was determined using (\ref{inverse})

1632: and $T_\mt{D6}\sim1/g_s \ls^7$. Combining these results, we find

1633: %

1634: \be \label{wv3} m_\ep^2 \sim {\nf\over\nc}\,\l\,\mkk^2\,. \ee

1635: %

1636: 

1637: With a collection of $\nf$ D6-branes in the holographic model of

1638: \cite{KMMW03}, the dual theory contains $\nf$ quark flavors. If

1639: all of the quark masses vanish, it was shown to leading order in

1640: the large-$\nc$ expansion that the spectrum contains $\nf^2$

1641: massless pseudoscalar mesons. Of these, only the $\ep$ meson,

1642: which corresponds to the collective motion of all of the D6-branes

1643: together, appears to be a true Goldstone mode. Again the latter

1644: only applies in the large-$\nc$ limit, as we have argued here that

1645: a mass appears through $1/\nc$ effects. Similarly it was argued in

1646: \cite{KMMW03} that the remaining $\nf^2-1$ pseudoscalars will

1647: acquire masses at this order. These mesons can be interpreted as

1648: modes which separate the individual D6-branes and so the masses

1649: can be understood as arising from closed string interactions

1650: between the separated branes. We will now give a concrete (albeit

1651: crude) estimate of the masses which are produced in this way. In

1652: the following, we will be working with the background D4-brane

1653: metric in the form given in eq.~(\ref{isometric}).

1654: 

1655: \FIGURE{\epsfig{file=figure8.eps, width=12cm} \caption{D6-brane

1656: embedding for $\ukk \neq 0$ and $\mq=0$, but separating two groups

1657: of $\nf/2$ D6-branes by a small angle $\delta\phi$ in the

1658: 89-plane.} \label{figure8}}

1659: 

1660: Imagine we have embedded $\nf$ D6-branes with $\mq=0$ but they

1661: have been divided into two groups of $\nf/2$ branes separated by a

1662: small angle $\delta\phi$ in the 89-plane, as depicted in figure

1663: \ref{figure8}. We will assume this depicts a canonical mode and so

1664: its mass is typical of that for all of the off-diagonal modes. We

1665: will also assume that the interaction can be approximated by

1666: integrating the `Newtonian' potential between volume elements on

1667: the separated (sets of) branes. Now as all of the branes have the

1668: 123-space in common, it is simplest to consider integrating over

1669: slices which extend through this subspace. These three-dimensional

1670: volume elements are then only infinitesimal in their extent in the

1671: $z^i$ subspace where the embedding is nontrivial, and will

1672: interact with a $1/\rho^4$ potential. We express the interaction

1673: energy as the energy density per unit volume (in the 123-space),

1674: which will roughly take the form:

1675: %

1676: \be \delta{\cal E} \simeq \nf^2\,G_{\it 10}\, h(\Omega_1,\Omega_2)

1677: {(T_{\mt{D6}}\,\delta^3A)_1\,(T_{\mt{D6}}\,\delta^3A)_2\over

1678: |\vec{z}_1-\vec{z}_2|^4}\ ,\label{pot3}\ee

1679: %

1680: where we have introduced $h(\Omega_1,\Omega_2)$ to indicate that

1681: the strength of the interaction depends on the details of the

1682: relative orientation of the two area elements.\footnote{Implicitly

1683: we will assert below that this interaction strength is positive.

1684: This intuition comes from considering D-branes in flat space where

1685: static parallel branes are supersymmetric. Hence a relative

1686: rotation between elements of the brane can only increase the

1687: energy.}

1688: 

1689: Now if we examine eq.~(2.17) of \cite{KMMW03} which determines the

1690: embedding profile of the D6-branes in the background geometry, we

1691: see that $\ukk$ is the only scale which enters this equation and

1692: so this is the scale of the interesting deformation in figure 8.

1693: Hence, we expect that the appropriate integrations yield

1694: %

1695: \be \left\langle h(\Omega_1,\Omega_2)

1696: {(\delta^3A)_1\,(\delta^3A)_2\over |\vec{z}_1-\vec{z}_2|^4}

1697: \right\rangle\sim \ukk^2\,\delta\phi^2\ , \ee

1698: %

1699: up to an overall purely numerical factor. That is, $\ukk$ and

1700: $\delta\phi$ are the only relevant scale and

1701: angle\footnote{Symmetry would rule out the appearance of a single

1702: power of $\delta\phi$ and hence the leading contribution must be

1703: $\delta\phi^2$.}, respectively, in the geometry of the D6-brane

1704: configuration. However, we must also account for the fact that all

1705: proper distances in the $z^i$ subspace are contracted by a factor

1706: of $K^{1/2}$, as seen in eq.~(\ref{isometric}). Hence the energy

1707: density acquires an additional factor of

1708: %

1709: \be

1710: K={R^{3/2}U^{1/2}\over\rho^2}\sim\left({R\over\ukk}\right)^{3/2}\

1711: , \ee

1712: %

1713: where in the last expression we have again used the approximation

1714: that $\ukk$ is the only scale relevant for the embedding. Next, we

1715: note that the final expression may be simplified using

1716: %

1717: \be G_{\it 10}\, T_\mt{D6}^2\simeq

1718: g_s^2\ls^8\left(1/g_s\ls^7\right)^2=1/\ls^6\ .\ee

1719: %

1720: Hence our approximation for the total interaction energy density

1721: becomes

1722: %

1723: \bea {\cal E}&\sim& \nf^2

1724: {R^{3/2}\ukk^{1/2}\over\ls^6}\,\delta\phi^2\nonumber\\

1725: %

1726: &\sim&\nf^2{\lambda\over\ls^4}\,\delta\phi^2 \eea

1727: %

1728: where we simplified the second line with (\ref{inverse}).

1729: 

1730: Now one might find the appearance of $\ls$ in this expression

1731: disturbing, however, one must realize that this is not a field

1732: theory quantity rather it is a proper energy density, ${\cal

1733: E}_{proper}$. To convert our expression to the energy density in

1734: the dual field theory, we must include additional metric factors

1735: for the $x^\mu$ directions relating proper bulk space distances to

1736: simple coordinate distances, which are relevant for the field

1737: theory, \ie, from eq.~(\ref{isometric}), $\Delta x_{proper}\sim

1738: \left(\ukk/R\right)^{3/4}\Delta x_{coord}$ again making the

1739: approximation that the only scale relevant for the embedding is

1740: $\ukk$. Hence we have

1741: %

1742: \be {\cal E}_{field}\sim \left({\ukk\over R}\right)^3{\cal

1743: E}_{proper}\sim \nf^2\lambda^3\mkk^4\,\delta\phi^2 \ee

1744: %

1745: where the latter expression is simplified using (\ref{inverse}).

1746: Finally to identify the mass, we must normalize the coefficient of

1747: $\delta\phi^2$ above by that in front of the corresponding kinetic

1748: term. Following the discussion of \cite{KMMW03}, it is

1749: straightforward to see that the kinetic term for this off-diagonal

1750: mode has an overall factor of $\nf f_\pi^2$. Hence, using

1751: eq.~(\ref{fpi}), our final expression for the mass is

1752: %

1753: \be \label{wv4} m_{pseudo}^2 \sim {\nf\over\nc}\,\l\,\mkk^2\,. \ee

1754: %

1755: Hence the parametric dependence of these off-diagonal pseudoscalar

1756: masses precisely matches that of the $\ep$ in eq.~(\ref{wv3}). It

1757: may be interesting to perform a more detailed analysis to

1758: determine the numerical coefficients in these two mass formulae.

1759: 

1760: \begin{thebibliography}{99}

1761: 

1762: \bibitem{KMMW03}

1763: M.\ Kruczenski, D.\ Mateos, R.\ C.\ Myers and D.\ J.\ Winters,

1764: {\sl Towards a holographic dual of large-$N_c$ QCD},

1765: \jhep{05}{2004}{041}, \hepth{0311270}.

1766: 

1767: \bibitem{GMOR68}

1768: M.\ Gell-Mann, R.\ J.\ Oakes and B.\ Renner,

1769: {\sl Behavior of current divergences under $SU(3) \times SU(3)$},

1770: \pr{175}{1968}{2195}.

1771: 

1772: \bibitem{Weinberg75}

1773: S.\ Weinberg, {\sl The $U(1)$ Problem}, \prd{11}{1975}{3583}.

1774: 

1775: \bibitem{thooft}

1776: G.\ 't Hooft, {\sl Symmetry Breaking through Bell-Jackiw Anomalies},

1777: \prl{37}{1976}{8}; {\sl Computation of the Quantum Effects due to a

1778:   Four-dimensional pseudo-particle}, \prd{14}{1976}{3432};

1779: {\sl How Instantons Solve the $U(1)$ Problem},

1780: \pr{142}{1986}{357}.

1781: %\bibitem{thooft} Original paper on U(1) problem and Physics Reports.

1782: 

1783: \bibitem{Witten79}

1784: E.\ Witten, {\sl Current Algenra Theorems for the $U(1)$ Goldstone Boson},

1785: \npb{156}{1979}{269}.

1786: %\bibitem{wi} Old Witten on the eta prime mass

1787: 

1788: \bibitem{Veneziano79}

1789: G.\ Veneziano, {\sl $U(1)$ without Instantons}, \npb{159}{1979}{213}.

1790: %\bibitem{veneziano} Veneziano on the same

1791: 

1792: \bibitem{clasicos}

1793: E.\ Witten, {\sl Large-$N$ Chiral Dynamics}, \ap{128}{1980}{363};

1794: P.\ Di Vecchia and G.\ Veneziano, {\sl Chiral Dynamics in the large N limit},

1795: \npb{171}{1980}{253};

1796: C. Rosenzweig, J. Schechter and G. Trahern, {\sl Is the Effective Lagrangian

1797: for QCD a Sigma Model?}, \prd{21}{1980}{3388};

1798:  K.\ Kawarabayashi and N. Ohta,

1799: {\sl The $\eta$ Problem in the Large-$N$ Limit:  Effective Lagrangian

1800: Approach}, \npb{175}{1980}{477}; H.\ Hata, T.\ Kugo and N.\ Ohta,

1801: {\sl Skew-Symmetric Tensor Gauge Field Theory Dynamically

1802: Realized in the QCD $U(1)$ Channel}, \npb{178}{1981}{527};

1803: P. Nath and R. Arnowitt, {\sl The $U(1)$ Problem:  Current Algebra and

1804: the Theta Vacuum}, \prd{23}{1981}{473}.

1805: 

1806: \bibitem{RGG75}

1807: A.\ De R\'ujula, H.\ Georgi and S.\ L.\ Glashow,

1808: {\sl Hadron Masses in a Gauge Theory}, \prd{12}{1975}{147};

1809: N. Isgur, {\sl Mass Formula for Nonets}, \prd{13}{1976}{122}.

1810: %\bibitem{agg} Old paper on the quark model by Alvaro, Georgi, Glashow.

1811: 

1812: \bibitem{Witten98b}

1813: E.\ Witten, {\sl Theta-dependence in the Large-$N$

1814: Limit of Four-dimensional Gauge Theories}, \prl{81}{1998}{2862},

1815: \hepth{9807109}.

1816: 

1817: \bibitem{Witten98a}

1818: E.\ Witten, {\sl Anti-de Sitter space, thermal phase transition,

1819: and confinement in gauge theories}, \atmp{2}{1998}{505},

1820: \hepth{9803131}.

1821: 

1822: %R extra ref

1823: \bibitem{dielectric}

1824: R.\ C.\ Myers, {\sl Dielectric-branes}, \jhep{12}{1999}{022},

1825: \hepth{9910053}.

1826: 

1827: \bibitem{flavour}

1828: O.\ Aharony, A.\ Fayyazuddin and J.\ M.\ Maldacena,

1829: {\sl The large-N limit of ${\cal N} =2,1$ field theories

1830: from threebranes in F-theory}, \jhep{07}{1998}{013},

1831: \hepth{9806159};

1832: %

1833: A.\ Karch and L.\ Randall,

1834: {\sl Open and closed string interpretation of SUSY CFT's on

1835: branes with boundaries}, \jhep{06}{2001}{063}, \hepth{0105132};

1836: %

1837: M.\ Bertolini, P.\ Di Vecchia, M.\ Frau, A.\ Lerda, and R.\ Marotta,

1838: {\sl N=2 Gauge theories on systems of fractional D3/D7 branes},

1839: \npb{621}{2002}{157}, \hepth{0107057};

1840: %

1841: M.\ Bertolini, P.\ Di Vecchia, G.\ Ferretti, and R.\ Marotta,

1842: {\sl Fractional Branes and N=1 Gauge Theories},

1843: \npb{630}{2002}{222}, \hepth{0112187};

1844: %

1845: A.\ Karch and E.\ Katz, {\sl Adding flavor to AdS/CFT},

1846: \jhep{06}{2002}{043}, \hepth{0205236};

1847: 

1848: \bibitem{mesons}

1849: A.\ Karch, E.\ Katz and N.\ Weiner, {\sl Hadron masses and screening

1850: from AdS Wilson loops}, \prl{90}{2003}{091601}, \hepth{0211107};

1851: %

1852: M.\ Kruczenski, D.\ Mateos, R.\ C.\ Myers and D.\ J.\ Winters,

1853: {\sl Meson spectroscopy in AdS/CFT with flavour},

1854: \jhep{07}{2003}{049}, \hepth{0304032};

1855: %

1856: T.\ Sakai and J.\ Sonnenschein,

1857: {\sl Probing flavored mesons of confining gauge theories by

1858: supergravity}, \jhep{09}{2003}{047}, \hepth{0305049};

1859: %

1860: J.\ Babington, J.\ Erdmenger, N.\ Evans, Z.\ Guralnik and I.\

1861: Kirsch, {\sl Chiral symmetry breaking and pions in

1862: non-supersymmetric gauge/gravity duals}, \prd{69}{2004}{066007},

1863: \hepth{0306018};

1864: %

1865: X.\ J.\ Wang and S.\ Hu, {\sl Intersecting branes and adding flavors

1866: to the Maldacena-Nunez background}, \jhep{09}{2003}{017},

1867: \hepth{0307218};

1868: %

1869: P.~Ouyang, {\sl Holomorphic D7-branes and flavored N = 1 gauge

1870: theories}, \hepth{0311084};

1871: %

1872: C.\ Nunez, A.\ Paredes and A.\ V.\ Ramallo,

1873: {\sl Flavoring the gravity dual of $\caln=1$ Yang-Mills with probes},

1874: \jhep{12}{2003}{024}, \hepth{0311201};

1875: %

1876: S.\ Hong, S.\ Yoon and M.\ J.\ Strassler, {\sl Quarkonium from the

1877: Fifth Dimension}, \jhep{04}{2004}{046}, \hepth{0312071};

1878: %

1879: E.\ Schreiber,

1880: {\sl Excited Mesons and Quantization of String Endpoints},

1881: \hepth{0403226};

1882: %

1883: N.\ Evans and J.\ P.\ Shock, {\sl Chiral Dynamics From AdS Space},

1884: \hepth{0403279}.

1885: 

1886: \bibitem{HO98}

1887: A.\ Hashimoto and Y.\ Oz, {\sl Aspects of QCD Dynamics from String Theory},

1888: \npb{548}{1999}{167}, \hepth{9809106}.

1889: 

1890: \bibitem{BP99}

1891: J.L.F. Barb\'on and A. Pasquinucci, {\sl Aspects of Instanton Dynamics

1892: from AdS/CFT duality}, \plb{458}{1999}{288}, \hepth{9904190}.

1893: 

1894: \bibitem{adi}

1895: A.\ Armoni, {\sl Witten--Veneziano from Green--Schwarz},

1896: \jhep{06}{2004}{024}, \hepth{0404248}.

1897: 

1898: \end{thebibliography}

1899: 

1900: 

1901: \end{document}

1902: