1: \documentclass[paper, 12pt, letterpaper, epsf]{JHEP} 

2: \input{epsf.tex}

3: \usepackage{graphics} 

4: \usepackage{epsfig}

5: %\renewcommand{\theequation}{\thesubsection.\arabic{equation}}

6: 

7: %1. Math symbols:

8: 

9: %1.1 boldface:

10: 

11: \def\Bbb{\bf} 

12: 

13: \def\C{{\Bbb C}} 

14: \def\R{{\Bbb R}}

15: \def\Z{{\Bbb Z}} 

16: \def\H{{\Bbb H}} 

17: \def\Q{{\Bbb Q}}

18: \def\P{{\Bbb P}}

19: \def\i{{\Bbb i}}\def\j{{\Bbb j}}\def\k{{\Bbb k}}

20: \def\I{{\Bbb I}}\def\J{{\Bbb J}}\def\K{{\Bbb K}}

21: \def\boldphi{\mbox{\boldmath $\phi$}} 

22: \def\boldpsi{\mbox{\boldmath $\psi$}} 

23: \def\boldzeta{\mbox{\boldmath $\zeta$}}

24: 

25: 

26: %1.2 double line:

27: 

28: \font\mybbb=msbm10 at 8pt \font\mybb=msbm10 at 12pt

29: \def\bbb#1{\hbox{\mybbb#1}} \def\bb#1{\hbox{\mybb#1}} 

30: 

31: \def\pRe{\bbb{R}} 

32: \def\Re {\bb{R}}

33: \def\S {\bb{S}}

34: \def\T {\bb{T}} 

35: %\def\P{\bb{P}} 

36: %\def\Z {\bb{Z}}

37: 

38:               

39: 

40: %1.3 operatorname:

41: 

42: \def\Hom{\operatorname{Hom}} \def\Tors{\operatorname{Tors}}

43: \def\Ker{\operatorname{Ker}} \def\Spec{\operatorname{Spec}}

44: \def\Area{\operatorname{Area}} \def\Vol{\operatorname{Vol}}

45: \def\ad{\operatorname{ad}} \def\tr{\operatorname{tr}}

46: \def\Pic{\operatorname{Pic}} \def\disc{\operatorname{disc}}

47: \def\cpl{\operatorname{cpl}} \def\Img{\operatorname{Im}}

48: \def\Rea{\operatorname{Re}} \def\Gr{\operatorname{Gr}}

49: \def\SO{\operatorname{SO}} \def\Sl{\operatorname{SL}}

50: \def\GO{\operatorname{O{}}} \def\SU{\operatorname{SU}}

51: \def\GU{\operatorname{U{}}} \def\Sp{\operatorname{Sp}}

52: \def\Spin{\operatorname{Spin}} \def\rank{\operatorname{rank}}

53: \def\Aff{\operatorname{Aff}} \def\diag{\operatorname{diag}}

54: \def\su{\rm{su(2)}}

55: 

56: %1.4 roman:

57: 

58: \def\ker{{\rm ker}} \def\coker{{\rm coker}} \def\rank{{\rm rank}}

59: \def\im{{\rm im}} \def\dim{{\rm dim}} \def\codim{{\rm codim}}

60: \def\Card{{\rm Card}} \def\li{{\rm linearly independent}} \def\ld{{\rm

61: linearly dependent}} \def\deg{{\rm deg}} \def\det{{\rm det}}

62: \def\Div{{\rm Div}} \def\supp{{\rm supp}} \def\Gr{{\rm Gr}}

63: \def\End{{\rm End}} \def\Aut{{\rm Aut}}

64: 

65: %1.5. script:

66: 

67: \def\cR{{\Scr R}} \def\cM{{\Scr M}} \def\cA{{\Scr A}} \def\cB{{\Scr B}}

68: \def\cK{{\Scr K}} \def\cD{{\Scr D}} \def\cH{{\Scr H}} \def\cT{{\Scr

69: T}} \def\cL{{\Scr L}} \def\cF{{\Scr F}}

70: 

71: %1.6 gothic:

72: 

73: \font\gothics=ygoth at 10pt \font\gothicl=ygoth at 12pt

74: \def\gg#1{\hbox{\gothics#1}}

75: \def\gG{\hbox{\gothicl G}}

76: \def \g{\hbox{\gothics g}}

77: 

78: %1.7 misc:

79: 

80: \def\id{\protect{{1 \kern-.28em {\rm l}}}}

81: 

82: %3. New commands

83: 

84: \newcommand{\be}{\begin{equation}} \newcommand{\ee}{\end{equation}}

85: \newcommand{\bea}{\begin{eqnarray}} \newcommand{\eea}{\end{eqnarray}}

86: \newcommand{\beann}{\begin{eqnarray*}} \newcommand{\eeann}{\end{eqnarray*}}

87: \newcommand{\bfig}{\begin{figure}} \newcommand{\efig}{\end{figure}}

88: \newcommand{\nn}{\nonumber}

89: \newcommand{\ba}{\begin{array}}\newcommand{\ea}{\end{array}}

90: 

91: %4. New theorems and their commands

92: 

93: \newtheorem{Proposition}{Proposition}[section]

94: \newtheorem{Definition}{Definition}[section]

95: \newtheorem{Theorem}{Theorem}[section]

96: \newtheorem{Lemma}{Lemma}[section]

97: \newtheorem{Corrolary}{Corrolary}[section]

98: 

99: \newcommand{\bp}{\begin{Proposition}} \newcommand{\ep}{\end{Proposition}} 

100: \newcommand{\bt}{\begin{Theorem}} \newcommand{\et}{\end{Theorem}} 

101: \newcommand{\bl}{\begin{Lemma}} \newcommand{\el}{\end{Lemma}} 

102: \newcommand{\bc}{\begin{Corrolary}} \newcommand{\ec}{\end{Corrolary}} 

103: 

104: %EOF

105: 

106: 

107: \title{Enhanced gauge symmetry from `toric' $G_2$ cones}

108: 

109: 

110: \author{L. Anguelova, C. I. Lazaroiu

111: \\C.~N.~Yang Institute for Theoretical Physics\\

112: SUNY at Stony Brook, NY, 11794-3840,

113: U.S.A.\\anguelov, calin @insti.physics.sunysb.edu}

114: 

115: 

116: \abstract{We review our

117: recent work on M-theory compactifications on `toric' $G_2$ cones, 

118: a class of models which generalize those recently considered by 

119: Acharya and Witten 

120: and lead to chiral matter in four dimensions. We explain our criteria

121: for identifying the gauge group content of such theories and briefly 

122: discuss the associated metrics. }

123: 

124: 

125: \preprint{YITP-SB-02-46}

126: 

127: \begin{document}

128: 

129: 

130: %\tableofcontents

131: 

132: %\pagebreak 

133: 

134: \vskip .3in

135: 

136: \section{Introduction} 

137: 

138: M-theory suggests phenomenologically interesting constructions

139: based on compactification from eleven dimensions.  

140: Among these, compactification on 7-dimensional spaces of $G_2$ holonomy

141: deserves special attention, since it leads naturally to $N=1$ supersymmetric

142: field theories in four dimensions. 

143: 

144: It is well-known, however, that compactification on {\em smooth} $G_2$

145: spaces has major phenomenological drawbacks, as it fails to produce

146: nonabelian gauge groups and chiral matter \cite{PT}. As pointed

147: out in \cite{Witten_anom}, this can be overcome by considering {\em

148: singular} $G_2$ backgrounds.  In such cases, nonabelian gauge symmetry

149: arises through $M2$-branes wrapping some vanishing two-cycles

150: \cite{Achar}, while net chirality can occur for purely topological

151: reasons \cite{Witten_anom}.  This mechanism was illustrated in

152: \cite{Witten_Acharya} by considering a specific class of conical $G_2$

153: spaces, some of which admit a IIA description in terms of intersecting

154: D6-branes.  In particular models, the appearance of chiral matter can

155: also be inferred directly from IIA constructions \cite{CSU1,

156: CSU2}. Such type II systems consist of D6-branes/orientifold 6-planes,

157: thus admitting a lift to an $M$-theory background of $G_2$ holonomy.

158: 

159: The class of spaces studied in \cite{Witten_Acharya} is obtained as

160: follows \cite{BS, GP}. Starting with an Einstein self-dual space $M$

161: (of positive scalar curvature),

162: one considers its six-dimensional twistor space $Y$, which can be

163: written as the sphere bundle associated with the vector bundle

164: $\Lambda^{2,-}T^*M$ of antiselfdual two forms. The twistor space

165: carries a Kahler-Einstein metric: 

166: \be

167: d\rho^2 = d\sigma^2+|d_A{\vec u}|^2~~,

168: \ee

169: where $d\sigma^2$ is the metric on $M$, $A$ is the connection induced

170: on $\Lambda^{2,-}T^*M$ by the Levi-Civita connection of $M$, and

171: ${\vec u}=(u^1,u^2,u^3)$ are coordinates on the $S^2$ fiber, subject

172: to the constraint $|{\vec u}|^2=1$.  To obtain a $G_2$ space, one

173: takes the metric cone over $Y$, where the latter is endowed with the 

174: {\em modified} metric 

175: $d\rho'^2=\frac{1}{2}(d\sigma^2+\frac{1}{2}|d_A{\vec u}|^2)$. 

176: In fact, this cone admits a one-parameter family of $G_2$

177: deformations given by \cite{BS, GP}:

178: \be

179: \label{G2metric}

180: ds_{G_2}^2=\frac{1}{1-(r_0/r)^4}dr^2+\frac{r^2}{2}

181: (d\sigma^2+\frac{1}{2}(1-(r_0/r)^4)|d_A{\vec u}|^2)~~,~~r_0\geq 0~~.

182: \ee 

183: The conical limit is reached for $r_0=0$. It is clear from this 

184: expression that the

185: singularities of the conical $G_2$ metric are determined by the singularities

186: of $Y$: except for the apex, the former are obtained by taking 

187: the cone over the latter (and their singularity type coincides with that 

188: in $Y$).

189: The importance of studying such singularities is clear since 

190: they determine the essential physics of M-theory on our 

191: backgrounds \cite{Witten_Acharya}.

192: 

193: To specify the $G_2$ space completely, one must know the

194: (positive curvature) 

195: Einstein-selfdual metric on $M$.  If $M$ is smooth, then there are

196: only two choices, namely $S^4$ and $\C\P^2$. For these cases, M-theory

197: physics on the associated $G_2$ spaces was studied in

198: \cite{AW}. Allowing for singularities in $M$ leads to many other

199: examples. Among these are the models considered in

200: \cite{Witten_Acharya}, which correspond to $M = W\C\P^2_{p,q,r}$,

201: endowed with the Einstein-selfdual metric obtained implicitly in

202: \cite{GL}. In these examples, the singularities of the twistor space

203: can be determined by elementary methods \cite{Witten_Acharya}.

204: However, this becomes a rather difficult task if one wishes to

205: consider more general situations.

206: 

207: A characteristic feature of the examples of  \cite{Witten_Acharya} 

208: is the presence of a two-torus of isometries. It is by now well-known 

209: \cite{GL,CP, AG, BGMR, BG} that there exist infinitely

210: many inequivalent compact ESD orbifolds\footnote{By this we mean 

211: a {\em local} orbifold (V-manifold). The spaces under consideration are 

212: generally {\em not} global quotients of some manifold by a finite group.}  

213: admitting two commuting isometries. 

214: Such spaces provide a vast generalization of the models

215: discussed in \cite{Witten_Acharya}. In \cite{toric}, we developed a method

216: for analyzing the singularities of the twistor space 

217: (and thus of the associated $G_2$ cone and of its deformation 

218: (\ref{G2metric})) for this much larger class of `toric' ESD orbifolds. 

219: 

220: The approach of \cite{toric} relies on a correspondence between ESD

221: orbifolds and certain hyperkahler cones (a hyperkahler cone is a

222: hyperkahler space which is the metric cone over a compact Riemannian

223: space).  Recall that the ESD property of the metric amounts to the

224: quaternion-Kahler condition in four dimensions.  Now, it is well-known

225: that every quaternion-Kahler space $M$ has an associated hyperkahler

226: cone\footnote{The correspondence of \cite{Swann} holds irrespective of

227: the sign of the scalar curvature of $M$.  In the case of interest for

228: us (namely when $M$ has positive scalar curvature) the hyperkahler

229: metric on $X$ will be Riemannian (positive-definite).  Negative

230: curvature quaternion-Kahler spaces have pseudo-Riemannian hyperkahler

231: cones, a situation which is of interest in supergravity.}  $X$

232: \cite{Swann}, and this correspondence can be `inverted' i.e. $M$ can

233: be re-constructed given $X$. This allows one to translate problems in

234: quaternion-Kahler geometry into the language of hyperkahler spaces,

235: which tends to be more amenable to a solution.  In our case, the

236: hyperkahler cone is real eight-dimensional and will admit a $T^2$'s

237: worth of isometries, induced from the isometries of $M$.

238: $4n$-dimensional hyperkahler spaces with an $n$-torus of isometries

239: have been considered in the mathematics literature \cite{BD} and are

240: called {\it toric hyperkahler}. Our particular situation fits into

241: that theory for $n=2$. Such spaces can be described as hyperkahler

242: quotients by certain torus actions, much in the spirit of usual toric

243: geometry (which is concerned with {\em Kahler} toral quotients). This

244: allows one to reduce geometric questions for $X$ to problems in

245: integral linear algebra and combinatorial convex geometry.

246: 

247: The twistor space of $M$ can be obtained from $X$ by performing a

248: certain $U(1)$ {\em Kahler} quotient at a positive moment map

249: level. This description allows one to extract the singularities of $Y$

250: by first determining the singularities of $X$ and then studying the

251: effect of this Kahler reduction \cite{toric}.

252: The outcome of this analysis is a simple algorithm for identifying the

253: singularities of the twistor space, which we summarize in the next

254: section. In \cite{toric} we also used geometric arguments to describe

255: the type IIA reduction along one of the $U(1)$ isometries of the

256: associated $G_2$ space.  For a generic choice of the reduction

257: isometry, the resulting IIA background is strongly coupled, and does

258: not seem to admit a simple perturbative description. However, certain

259: models admit a `good isometry', which leads to a system of $D6$-branes

260: upon reduction.  Such an isometry (when it exists) can be found by a

261: simple criterion discussed in \cite{toric}.  Since the IIA metric

262: inherits a $U(1)$ isometry induced from the M-theory solution, one can

263: T-dualize along its direction to obtain a IIB background.  This will

264: correspond to a system of delocalized 5-branes if a `good isometry'

265: is used for IIA reduction.  While models with a good isometry admit a

266: simple IIA/IIB description, we stress that there is no obvious reason

267: to discard other models. From the M-theory perspective, models without

268: a good isometry are as good as any other.

269: 

270: 

271: The abstract arguments of \cite{toric} were confirmed in \cite{metrics} 

272: by direct calculation. Using the result of \cite{CP}, we wrote

273: down the $G_2$ metric obtained from a general ESD space of positive

274: scalar curvature and admitting a $T^2$ of isometries.  By performing

275: the reduction, we found the corresponding T-dual IIA and IIB

276: backgrounds. Guided by the analysis of \cite{toric} and the

277: topological arguments of \cite{BGMR}, we computed the explicit

278: asymptotics of relevant fields (metric, coupling constant, and

279: R-R/NS-NS forms) in the vicinity of the branes and determined the

280: relevant RR fluxes. The main results of \cite{toric} and

281: \cite{metrics} are briefly reviewed below.

282: 

283: 

284: 

285: \section{Singularities of the twistor space}

286: 

287: %\setcounter{equation}{0} 

288: 

289: The toric hyperkahler cones associated with our ESD spaces are

290: obtained as hyperkahler quotients of some affine quaternion space

291: $\H^n$ by an appropriate torus action. Namely, we have 

292: $X=\H^n///_0U(1)^{n-2}$, where

293: the subscript $0$ indicates that the hyperkahler quotient is taken

294: at zero moment map levels. The action of $U(1)^{n-2}$ on the quaternion

295: coordinates $u_1\dots u_n\in \H$ is given by: \be

296: \label{action}

297: u_k \rightarrow

298: \prod_{\alpha=1}^{n-2}{\lambda_\alpha^{q_k^{(\alpha)}}}u_k~~, \ee

299: where $\lambda_{\alpha}$ are complex numbers of unit modulus. The

300: $(n-2)\times n$ matrix\footnote{One has to impose some 

301: mild conditions on $Q$

302: to guarantee that the $U(1)^{n-2}$ action on $\H^n$ is

303: effective. See \cite{toric} and footnote 6 of this letter.}

304: $Q_{\alpha k}=q_k^{(\alpha)}$ determines the cone $X$.

305: Following toric geometry procedure, we introduce a 

306: $2\times n$ matrix $G$ whose  

307: rows form an integral basis for the kernel

308: of $Q$. The columns of $G$ are two-dimensional integral vectors $\nu_1

309: \dots \nu_n$, which we call {\it toric hyperkahler

310: generators}. Splitting every quaternion coordinate into its complex

311: components: \be u_k = w_k^{(+)} + \j w_k^{(-)} ~~, \ee (where

312: $\i,\j,\k$ are the imaginary quaternion units) we obtain 

313: the following re-writing of the torus action: \be

314: \label{complex_action}

315: w_k^{(+)}\rightarrow \prod_{\alpha=1}^{n-2} \lambda_\alpha

316: ^{q^{(\alpha)}_k}w_k^{(+)}~~,~~ w_k^{(-)}\rightarrow

317: \prod_{\alpha=1}^{n-2} \lambda_\alpha ^{-q^{(\alpha)}_k}w_k^{(-)}~~.

318: \ee The twistor space is obtained as a $U(1)$ Kahler quotient 

319: $Y=X//_{\zeta}U(1)$  of $X$ at a

320: positive level $\zeta$ (the precise choice of $\zeta$ 

321: fixes the overall scale of $Y$ and $M$). The $U(1)$ action is given by: 

322: \be

323: \label{U1proj}

324: w_k^{(\pm)}\rightarrow \lambda w_k^{(\pm)}~~  

325: \ee

326: and will be called `the projectivizing action'. For simplicity, 

327: we shall assume

328: \footnote{If this assumption does not hold, then the projectivizing 

329: $U(1)$ has a trivially acting $\Z_2$ subgroup. In this case, there 

330: are {\em two} toric hyperkahler cones associated with $Y$ and 

331: the results described below must be slightly modified. The required 

332: modifications are described in \cite{toric}.}

333: that this action is effective on $X$. 

334: 

335: In \cite{toric}, we show that the singular locus of $Y$ is a subset 

336: of the so-called 

337: {\it distinguished locus} $Y_D$. The latter consists of two kinds of

338: holomorphically embedded two-spheres, which are distinguished 

339: by their position with respect to the $S^2$ fibration $Y\rightarrow M$:

340: 

341: 1) $Y_D$ contains $n$ {\it vertical} spheres $Y_j$ ($j=1\dots n$), 

342: which are fibers of $Y$ over $M$.

343: 

344: 2) it also contains $2n$ {\it horizontal} spheres 

345: $Y_e$, which are lifts of spheres lying in $M$. 

346: 

347: \noindent The meaning of the index $e$ will become apparent in a 

348: moment (it corresponds to the edges of a certain polygon). The union 

349: $Y_V$ of all vertical spheres and the union $Y_H$ of horizontal spheres 

350: will be called the {\em vertical} and {\em horizontal} loci.

351: 

352: A basic observation (which goes back to \cite{HKLR}, see also

353: \cite{BD}) is that the hyperkahler moment map constraints are solved

354: by points $u\in \H^n$ whose complex coordinates are solutions of the

355: system: \be

356: \label{ab}

357: \frac{1}{2}(|w^{(+)}_k|^2-|w^{(-)}_k|^2)=\nu_k\cdot a~~,~~

358: w^{(+)}_kw^{(-)}_k=\nu_k\cdot b~~, ~{\rm~for~all~}k=1\dots n~~,

359: \label{X} \ee for some parameters $a=(x_1, y_1)\in \R^2$ and $b=(x_3+ix_2,

360: y_3+iy_2)\in \C^2$. Using this, one shows \cite{toric}

361: that the distinguished locus can be

362: described in terms of the {\it characteristic polygon}: \be

363: \label{polygon} \Delta=\{a\in \R^2|\sum_{k=1}^n{|\nu_k\cdot

364: a|}=\zeta\}~~.  \ee 

365: This is a planar convex polygon with $2n$ vertices, which is symmetric

366: under the point reflection $a \rightarrow -a$.  In particular,

367: $\Delta$ has $n$ {\em principal diagonals} $D_j=\{a\in \R^2|a\cdot

368: \nu_j=0\}$. These are the diagonals which pass through the origin and

369: thus connect opposite vertices.

370: 

371: Each vertical sphere $Y_j$ is defined by the condition $u_j = 0$

372: (i.e. $w_j^{(+)} = w_j^{(-)} = 0$) with $j=1\dots n$. The horizontal

373: spheres $Y_e$ are indexed by the edges $e$ of $\Delta$ and are

374: determined as follows. Given such an edge, define the signs

375: $\epsilon_j(e) = sign(\nu_j\cdot p_e)$, where $p_e$ is the two-vector

376: from the origin of the plane to the middle point of $e$.  Then $Y_e$

377: is the locus in $Y$ given by the equations $w_1^{(-\epsilon_1)} =

378: w_2^{(-\epsilon_2)} = \dots = w_n^{(-\epsilon_n)} = 0$. As discussed

379: in \cite{toric},  $Y_j$ is a circle fibration over

380: the principal diagonal $D_j$ (the circle fibers collapse to points

381: above the two vertices connected by the diagonal). Each horizontal

382: sphere $Y_e$ is a circle fibration over the edge $e$, whose circle

383: fibers collapse to points above the vertices connected by this

384: edge. The twistor space admits an antiholomorphic involution acting

385: along its $S^2$ fibers. Its restriction to $Y_H$ intertwines the

386: horizontal spheres $Y_e$ and $Y_{-e}$, while covering the point

387: reflection $\iota :a\rightarrow -a$ of $\Delta$. Finally, the two

388: horizontal spheres associated with adjacent edges of $\Delta$ and the

389: vertical sphere associated with the principal diagonal passing through

390: their common vertex have precisely one point in common. Such {\em

391: special points} of the distinguished locus are in one-one

392: correspondence with the vertices of $\Delta$. This situation is

393: summarized in figure \ref{dist}.

394: 

395: 

396: \begin{figure}[hbtp]

397: \begin{center}

398: \scalebox{0.6}{\input{dist.pstex_t}}

399: \end{center}

400: \caption{\label{dist} Fibration of the distinguished locus over the 

401: characteristic polygon 

402: $\Delta$. Since we attempt to represent this in two dimensions, it may 

403: appear that some spheres intersect at more than isolated points. This is 

404: {\em not} actually the case. }

405: \end{figure}

406: 

407: 

408: 

409: As mentioned above, the singular locus of the twistor space is a 

410: subset of its distinguished locus. Namely, some of the vertical and 

411: horizontal spheres will consist of cyclic singularities, 

412: while some of the special points will carry singularities of their own. 

413: The relevant singularity types are described by the following rules:

414: 

415: (a) Given a horizontal sphere $Y_e$, consider the integral vector

416: \footnote{With our assumptions on $Q$, this vector can never 

417: vanish.}: 

418: \be

419: \label{nu} \nu_e=\sum_{k=1}^n{\epsilon_k(e)\nu_k}~~, 

420: \ee 

421: where the signs $\{\epsilon_k(e)\}$ are defined as explained above.

422: Then $Y$

423: has a $\Gamma_e=\Z_{m_e}$ quotient singularity along $Y_e$, where 

424: $m_e$ is the greatest common divisor of the two entries of the vector 

425: $\nu_e$.  

426: The orbifold action on the coordinates transverse to $Y_e$ can 

427: be determined as explained in \cite{toric}. 

428: 

429: (b) Given a vertical sphere $Y_j$, consider the matrix ${\tilde Q}_j$

430: obtained by deleting the $j^{th}$ and $(j+n)^{th}$ columns of the

431: $(n-1)\times (2n)$ matrix ${\tilde Q} =

432: \left[\begin{array}{ccc}Q&~&-Q\\1&\dots&~1\end{array}\right]$.  Then

433: the singularity group $\Gamma_j$ of $Y$ along $Y_j$ coincides with

434: $\Z_{m_j}$ or $\Z_{2m_j}$, where $m_j$ is the greatest common divisor

435: of the components of $\nu_j$.  

436: To find which of these cases occurs,

437: one computes the integral Smith form\footnote{Given an integral

438: $r\times n$ matrix $F$ with $r\leq n$, one can find matrices $U\in

439: GL(r,\Z)$ and $V\in GL(n,\Z)$ such that the matrix

440: $F^{ismith}=U^{-1}FV$ (the {\em integral Smith form} of $F$) has

441: the form $[D,0]$. Here $D=diag(t_1 \dots t_r)$ with $t_1 \dots

442: t_r$ some non-negative integers satisfying the division relations

443: $t_1|t_2|\dots |t_r$. These integers are called the {\em invariant

444: factors} of $F$. The mild condition 

445: mentioned in footnote 3 is that 

446: all invariant factors of $Q$ be equal to $1$.} 

447: of the matrix ${\tilde Q}_j$: \be {\tilde Q}_j^{ismith}=[diag(1\dots 1,

448: t_j),0]~~, \ee where $t_j=m_j$ or $t_j=2m_j$. The singularity group

449: $\Gamma_j$ coincides with $\Z_{t_j}$. 

450: The orbifold action on the transverse coordinates can be determined 

451: as explained in \cite{toric}.

452: 

453: Note that some of the vertical and horizontal spheres will be smooth 

454: (this happens respectively when $m_e=1$ and $t_j=1$).

455: The singularity type at the special points (which correspond to the vertices of $\Delta$) can also be determined 

456: by simple criteria, which the interested reader 

457: can find in \cite{toric}. 

458: 

459: 

460: The criteria listed above determine the 

461: gauge groups produced at the singular loci of the $G_2$ cone: 

462: a $\Z_m$ singularity corresponds to an $SU(m)$ gauge factor. 

463: On the other hand, the second Betti number of our $G_2$ cone equals $n-1$. 

464: As in \cite{Witten_Acharya},  M-theory on such cones

465: produces an Abelian $U(1)^{n-1}$ factor, and 

466: chiral fermions charged under this group and 

467: localized at the cones' apex. The second Betti number reduces to $n-2$

468: (and the Abelian gauge factor is Higgsed 

469: down to $U(1)^{n-2}$) when deforming the cone to the solution with 

470: non-vanishing $r_0$. All of this is qualitatively identical with the 

471: behavior of the models discussed in \cite{Witten_Acharya}.

472: 

473: Dimensionally reducing M-theory through one of the $T^2$ isometries, 

474: one obtains a type IIA background with a non-vanishing RR one-form. 

475: Such an isometry is called {\em good} 

476: if its fixed point set coincides with the singular locus of the

477: $G_2$ cone. If a good isometry exists, then the associated type IIA 

478: reduction gives a system

479: of D6-branes. It is shown in \cite{toric} that a 

480: good isometry exists if all two-vectors $\nu_e,\nu_j$ are collinear; 

481: in this case, the good isometry corresponds to their common direction, 

482: viewed as a direction in the Lie algebra of the two-torus of isometries 

483: of the $G_2$ space.

484: 

485: Since the resulting IIA background inherits a 

486: $U(1)$ isometry, one can T-dualize to a IIB solution. 

487: It turns out that the orbits of the T-dualizing isometry

488: lie along the $S^1$ fibers of the loci $Y_j\rightarrow D_j$ 

489: and $Y_e\rightarrow e$, both of which descend to the IIA solution. 

490: In the good isometry 

491: case, it follows that T-duality acts along the worldvolume directions, 

492: thereby producing delocalized 5-branes in type IIB.

493: This abstract argument can be substantiated by a direct analysis of 

494: the relevant field configurations, which was carried out in \cite{metrics}

495: and is briefly reviewed below.  

496: 

497: \section{$G_2$ metrics and reduction to type II string theory}

498: 

499: As discussed above, the ESD space $M$ and its associated 

500: $G_2$ cone, as well as the hyperkahler 

501: cone $X$ admit a two-torus of isometries for 

502: our class of models. The most general ESD metric admitting two commuting 

503: isometries was given explicitly in recent work of Calderbank and Pedersen 

504: \cite{CP}. To describe the solution, one introduces 

505: coordinates $\phi$, $\psi$ along  the $T^2$ fibers of $M$ and 

506: $\rho, \eta$ along the base of its $T^2$ 

507: fibration. There exists an $SL(2,\Z)$ freedom in the choice of $\phi$ and 

508: $\psi$, which is related to modular transformations of the lattice $\Z^2$

509: containing the toric hyperkahler generators.

510: As explained in \cite{CP}, the base coordinates can be chosen 

511: such that $\rho\geq 0$ and

512: $\eta\in \R$, and parameterize the upper half plane (the hyperbolic

513: plane ${\cal H}^2$). The metric of \cite{CP} takes the form:

514: 

515: {\footnotesize \bea \!\!\!\!\!\!\!\!\!\!  d\sigma^2=\frac{ F^2 -

516: 4\rho^2(F_\rho^2 + F_\eta^2) }{4 F^2}\; \frac{d\rho^2 +

517: d\eta^2}{\rho^2} +\frac{ \left[ (F - 2 \rho F_\rho) \alpha - 2 \rho

518: F_\eta \beta \right]^2 + \left[ -2\rho F_\eta \alpha + (F + 2 \rho

519: F_\rho )\beta \right]^2 }{ F^2\left[F^2-4\rho^2(F_\rho^2 +

520: F_\eta^2)\right]}, \label{Mmetric} \eea} where \be

521: \label{F}

522: F=\sum_{k=1}^n \frac{\sqrt{(\nu_k^2)^2\rho^2

523: +(\nu_k^2\eta+\nu_k^1)^2}}{\sqrt\rho} \ee and

524: $\alpha=\sqrt\rho\,d\phi$, \,\, $\beta=(d\psi+\eta\,

525: d\phi)/\sqrt\rho$, and $F_{\rho}= \partial F/\partial \rho$, $F_\eta=

526: \partial F/\partial \eta$. The

527: metric on the $S^2$ fiber of the twistor space of $M$ takes the form:

528: \be \label{fmetric} |d_A {\vec u} |^2=(du^i + \epsilon^{ijk} A^j

529: u^k)^2~~, \ee where ${\vec u}=(u^1,u^2,u^3)$ are constrained 

530: coordinates $(\sum_{i=1}^3

531: (u^i)^2 = 1)$ on the fiber and the connection $A$ is given by 

532: \cite{CP}: \be A^1 = -

533: \frac{F_{\eta}}{F} d\rho + \left(\frac{1}{2 \rho} + \frac{F_{\rho}}{F}

534: \right) d\eta~~,~~A^2 = - \frac{\sqrt{\rho}}{F} d\phi~~,~~A^3 =

535: \frac{\eta}{F\sqrt{\rho}} d\phi + \frac{1}{F \sqrt{\rho}} d\psi \, .

536: \ee Substituting (\ref{Mmetric}) and (\ref{fmetric}) into

537: (\ref{G2metric}) we obtain the $G_2$ metric of our models.

538: As expected, this leads to rather complicated formulae, which 

539: can be found in \cite{metrics} and will not be reproduced here. 

540: 

541: As explained in \cite{metrics}, the 

542: singular loci of $Y$ project onto the boundary of 

543: ${\cal H}^2$. The latter is obtained by adding the point at infinity

544: to the upper half plane: this results in 

545: a disk whose boundary corresponds to $\rho = 0$. In agreement with the

546: topological analysis of \cite{BGMR}, the vertical loci project to

547: points $P_1\dots P_n$ lying on this boundary, while the horizontal loci

548: project to the segments $P_k P_{k+1}$ connecting them.

549: The points $P_k$ correspond to $\rho=0$ and $\eta=\eta_k$, 

550: where $\eta_k=-\frac{\nu_k^1}{\nu_k^2}$.

551: The circular polygon (with vertices $P_k$) obtained in this manner 

552: can be identified topologically with the polygon $\Delta_M$

553: defined as the quotient of the characteristic polygon 

554: $\Delta$ (see (\ref{polygon})) through the 

555: point reflection $\iota : a \rightarrow -\,a$ of the plane: \be

556: \Delta_M=\Delta/\iota~~.  \ee 

557: The polygon $\Delta_M$ has $n$ vertices. Opposite edges $e$ and $-e$ 

558: of $\Delta$ cover the same edge of $\Delta_M$ (figure \ref{DDM}). 

559: 

560: \begin{figure}[hbtp]

561: \begin{center}

562: \scalebox{0.6}{\input{DDM.pstex_t}}

563: \end{center}

564: \caption{\label{DDM} Examples of the 

565: polygons $\Delta$ and $\Delta_M$ for $n=4$. For the edge $e$ of $\Delta$ 

566: drawn as a bold line, we show the vector $p_e$ used in the definition of

567: the signs $\epsilon_j(e)$. Note that the principal diagonals 

568: $D_j$ need not lie in trigonometric order.}

569: \end{figure}

570: 

571: 

572: Starting with the purely geometric background $ds_{11}^2 =

573: ds^2(\R^{3,1}) + ds_{G_2}^2$ of 11d supergravity (see

574: (\ref{G2metric})), one can reduce along one of the $T^2$ circles 

575: to obtain a IIA solution with non-constant dilaton and non-vanishing 

576: RR 1-form. Modulo a common modular transformation of the vectors $\nu_j$, 

577: it suffices to consider the reduction along the $\phi$-circle. 

578: The explicit form of the resulting IIA  solution 

579: can be found in \cite{metrics} and will not be reproduced here.

580: 

581: Parameterizing the $S^2$ fibers of $Y$ by spherical

582: coordinates: \be u^3=\cos\theta \, , \qquad u^1=\sin\theta \cos\chi \,

583: , \qquad u^2=\sin\theta \sin \chi

584:  \label{sphcoord} \ee 

585:  with $\theta \in [0, \pi] \, , \, \chi \in [0, 2 \pi]$, one obtains

586: coordinates $(r, \rho, \eta, \chi, \theta,\psi)$ for the internal

587: 6-dimensional part of the resulting IIA background. For every fixed

588: value of $r$, one has a compact 5-manifold ${\cal N}^5$ with

589: coordinates $(\rho,\eta,\chi,\theta,\psi)$. This is the $U(1)_\phi$

590: reduction of the twistor space $Y$. Under this reduction, the

591: distinguished locus of the twistor space corresponds to the locus

592: ${\cal N}_D$ in ${\cal N}^5$ defined by $\rho=0, \theta=0,\pi$ or

593: $\rho=0,\eta=\eta_k$ (for some $k$).  In the limit $\rho \rightarrow

594: 0$, the angular part of the metric turns out to depend only on

595: $\theta$ and $\xi \equiv \nu_{\epsilon}^1 \chi + \psi$, where

596: $\nu_{\epsilon}^1$ is the first component of a two-component vector

597: $\nu_{\epsilon}$ related to the edge of $\Delta_M$ covered by the

598: edges $e, -e$ of $\Delta$ ($\nu_{\pm e}$ of (\ref{nu}) equals $\pm

599: \nu_{\epsilon}$). Hence the locus $\rho=0$ in ${\cal N}^5$ is only

600: three-dimensional, while the sublocus ${\cal N}_D$ has dimension

601: two. The horizontal part $\rho=0,\theta=0,\pi$ of ${\cal N}_D$ turns

602: out to be a union of $2n$ components ${\cal N}_e$ indexed by the edges

603: of $\Delta$. The vertical part consists of the

604: components ${\cal N}_k$ defined by $\rho=0$ and $\eta=\eta_k$.

605: 

606: The $\rho\rightarrow 0$ asymptotics of the IIA fields (dilaton, RR 1-form and

607: metric) was analyzed in detail in \cite{metrics}. 

608: Fixing a component ${\cal N}_e$ or ${\cal N}_j$, the IIA background turns out

609: to be strongly coupled in its vicinity unless $\nu_e^1=0$ (resp. $\nu_j^1=0$), 

610: i.e. unless the direction $\phi$ used for reduction corresponds to the 

611: direction $\nu_e$ (resp. $\nu_j$) in the Lie algebra $\R^2$ of the

612: $T^2$ isometry group. Remember from the previous

613: section that ${\cal N}_e$ (resp. ${\cal N}_j$) corresponds to enhanced

614: gauge symmetry if $\nu_e$ (resp. $\nu_j$) is not primitive.  To insure

615: weak coupling along all such loci, one must be able to choose

616: $U(1)_\phi$ in the direction of all non-primitive $\nu_e$ and $\nu_j$.

617: This requires that all these vectors be proportional (over the reals),

618: which is the condition for a `good isometry' mentioned above. 

619: 

620: For simplicity, we shall restrict to the case when $U(1)_\phi$ is good.

621: In this situation, all interesting components of the locus ${\cal

622: N}_D$ are weakly coupled. Moreover, the computations of \cite{metrics}

623: show that the interesting loci ${\cal N}_e$, ${\cal N}_j$ carry

624: $|\nu_{e}^2|$ resp. $|\nu_j^2|$ units of RR flux.  Since

625: $\nu_e^1=\nu_j^1=0$, one has $|\nu_{e}^2|=m_e$ and $|\nu_j^2|=m_j$,

626: where $m_e=gcd(\nu_e^1,\nu_e^2)$ and $m_j=gcd(\nu_j^1,\nu_j^2)$.  This

627: confirms the prediction of \cite{toric} reviewed in the previous

628: section, and shows that the appearance of enhanced $SU(m_e)$ or

629: $SU(m_j)$ symmetry at these loci is due to the presence of coincident

630: D6-branes\footnote{The equivalence of the two mechanisms of

631: enhancement of gauge symmetry - coinciding D-branes in type IIA string

632: theory and M2-branes wrapping collapsing two-cycles at an $A-D-E$

633: singularity - was first pointed out in \cite{Sen}.}. 

634: The geometry at $\rho=0$ is depicted in

635: figure \ref{2Aboundary} for a model with $n=3$ which admits a good

636: isometry (for example the $W_{p,p,q}$ models of \cite{Witten_Acharya}).

637: 

638: \begin{figure}[hbtp]

639: \begin{center}

640: \mbox{\epsfxsize=10truecm \epsffile{2Aboundary.eps}}

641: \end{center}

642: \caption{\label{2Aboundary} The locus $\rho=0$ (the 3-dimensional

643: boundary of the space parameterized by $\rho, \eta, \psi, \theta,

644: \chi$) at a fixed value of $r$.  The figure shows the case $n=3$, for

645: a model admitting a good isometry.  $\Delta_M$ is the triangle with

646: vertices $\eta_1,\eta_2,\eta_3\in \partial{\overline {\cal

647: H}}^2=\R\cup \{\infty\}$; these 3 points are shown in the upper half

648: plane on the right figure.  The picture corresponds to a good isometry

649: given by $\nu_\epsilon^1=0$ along the edge $[\eta_1,\eta_2]$ (singular

650: horizontal locus giving a pair of D6-branes at $\theta = 0, \pi$ with

651: worldvolume containing $S_{\eta \xi}^2$) and $\nu_3^1=0$ at the vertex

652: $\eta_3$ (singular vertical locus again giving a D6-brane containing

653: $S_{\theta \xi}^2$).}

654: \end{figure}

655: 

656: T-dualizing along the remaining $U(1)$ isometry (parameterized by

657: $\psi$) one finds that the $\chi$-circle is always

658: collapsed to a point in the limit $\rho \rightarrow 0$ \cite{metrics}. 

659: Analyzing the asymptotic behavior of the IIB fields enables the computation

660: of the relevant NS-NS/R-R fluxes. 

661: When the $U(1)$ isometry along $\phi$ is good, one 

662: finds that the corresponding IIB loci support delocalized

663: five-branes, once again confirming the predictions of \cite{toric}.

664: 

665: 

666: \acknowledgments{

667: The present work was supported by the Research Foundation under NSF 

668: grant PHY-0098527.}

669: 

670: 

671: 

672: \begin{thebibliography}{150}

673: \bibitem{PT}{G.~Papadopoulos, P.~Townsend, {\em

674: Compactification of $d=11$ Supergravity on Manifolds of Exceptional

675: Holonomy}, Phys. Lett. {\bf B357} (1995) 300} 

676: \bibitem{Achar}{B.~Acharya, {\em M theory, Joyce Orbifolds and Super

677: Yang-Mills}, Adv. Theor. Math. Phys. 3 (1999) 227, hep-th/9812205}

678: \bibitem{Witten_anom}{E.~Witten, {\em Anomaly Cancellation On

679: Manifolds Of $G_2$ Holonomy}, hep-th/0108165.}

680: \bibitem{Witten_Acharya}{ B.~Acharya, E.~Witten, {\em Chiral Fermions

681: from Manifolds of $G_2$ Holonomy}, hep-th/0109152.}

682: \bibitem{CSU1}{M.~Cvetic, G.~Shiu and A.~M.~Uranga, {\em Chiral

683: four-dimensional $N = 1$ supersymmetric type IIA orientifolds from

684: intersecting D6-branes}, Nucl.Phys. {\bf B615} (2001) 3-32,

685: hep-th/0107166.}  

686: \bibitem{CSU2}{M.~Cvetic, G.~Shiu and A.~M.~Uranga,

687: {\em Three-Family Supersymmetric Standard-like Models from

688: Intersecting Brane Worlds}, Phys.Rev.Lett. {\bf 87} (2001) 201801,

689: hep-th/0107143.}  

690: \bibitem{BS}{R. L. Bryant, S. M. Salamon, {\em On

691: the construction of some complete metrics with exceptional holonomy},

692: Duke Math. J. {\bf 58} (1989)3, 829.}  

693: \bibitem{GP}{G.~W.~Gibbons,

694: D.~N.~Page, C.~N.~Pope, {\em Einstein metrics on $S^3$, $\R^3$ and

695: $\R^4$ bundles}, Commun. Math. Phys {\bf 127} (1990), 529-553.}

696: \bibitem{AW}{M.~Atiyah, E.~Witten, {\em M-theory dynamics on a

697: manifold of $G_2$ holonomy}, hep-th/0107177.}

698: \bibitem{GL}{K. Galicki, H. B. Lawson, Jr, {\em Quaternionic reduction

699: and quaternionic orbifolds}, Math. Ann. {\bf 282} (1988) 1-21.}

700: \bibitem{CP}{D. M. J. Calderbank, H. Pedersen, {\em Selfdual Einstein

701: metrics with torus symmetry}, math-DG/0105263.}

702: \bibitem{AG}{V.~Apostolov, P.~Gauduchon, {\em

703: Self-dual Einstein Hermitian four manifolds}, math.DG/0003162.}

704: \bibitem{BGMR}{C.~P.~Boyer, K.~Galicki, B.~M.~Mann, E.~G.~Rees, {\em

705: Compact $3$-Sasakian $7$-manifolds with arbitrary second Betti number}

706: Invent. Math.  {\bf 131} (1998), no. 2, 321--344.}  

707: \bibitem{BG}{

708: C.~P.~Boyer, K.~Galicki, {\em 3-Sasakian Manifolds}, Surveys in

709: differential geometry: Essays on Einstein manifolds, 123--184,

710: Surv. Differ. Geom., VI, Int. Press, Boston, MA, 1999,

711: hep-th/9810250.}  

712: \bibitem{Swann}{A. Swann, {\em Hyperk\"ahler and

713: quaternionic K\"ahler geometry}, Math. Ann. {\bf 289} (1991) 421.}

714: \bibitem{toric}{L.~Anguelova, C.~I.~Lazaroiu, {\em M-theory

715: compactifications on certain `toric' cones of $G_2$ holonomy},

716: hep-th/0204249.}  

717: \bibitem{metrics}{L. Anguelova, C. I. Lazaroiu, {\em

718: M-theory on `toric' $G_2$ cones and its type II reduction},

719: hep-th/0205070.}  

720: \bibitem{Sen}{A.~Sen, {\em A note on enhanced gauge

721: symmetries in M- and string theory}, JHEP 9709 (1997) 001,

722: hep-th/9707123} 

723: \bibitem{HKLR}{N. Hitchin, A. Karlhede, U. Lindstr\"{o}m,

724: M. Ro\v{c}ek, {\em Hyperk\"{a}hler Metrics and Supersymmetry}, 

725: Commun. Math. Phys. {\bf 108} (1987) 535.}

726: \bibitem{BD}{R. Bielawski, A. S. Dancer, {\em The

727: geometry and topology of toric hyperkahler manifolds},

728: Commun. Anal. Geom, vol 8, No. 4 (2000), 727-759.}

729: \end{thebibliography}

730: \end{document}