1: %\documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2: \documentclass[12pt,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: \usepackage{dcolumn}
4: \usepackage{epsfig}
5:
6:
7: \newcommand{\be}{\begin{equation}}
8: \newcommand{\ee}{\end{equation}}
9: \newcommand{\bea}{\begin{eqnarray}}
10: \newcommand{\eea}{\end{eqnarray}}
11: \begin{document}
12:
13: \title{Regular non-Abelian vacua in ${\cal N}=4$, SO(4) gauged supergravity }
14: \author{Ali H. Chamseddine$^\dagger$ and Mikhail S. Volkov$^{\dagger\dagger}$ }
15: \affiliation{$\dagger$ CAMS and Physics Department, American University of Beirut, LEBANON\\
16: $\dagger\dagger$ LMPT CNRS-UMR 6083,
17: Universit\'e de Tours, Parc de Grandmont,
18: 37200 Tours, FRANCE
19: }
20:
21:
22:
23:
24: \begin{abstract}
25: {We present a family of globally regular ${\cal N}=1$ vacua in the D=4, ${\cal N}=4$
26: gauged supergravity of Gates and Zwiebach. These solutions
27: are labeled by the ratio $\xi$ of the two gauge couplings,
28: and for $\xi=0$ they reduce to
29: the supergravity monopole previously used for constructing the gravity dual
30: of ${\cal N}=1$
31: super Yang-Mills theory.
32: For $\xi>0$ the
33: solutions are asymptotically anti de Sitter, but with
34: an excess of the solid angle, and they reduce exactly to anti de Sitter for $\xi=1$.
35: Solutions with $\xi<0$ are topologically $R^1\times S^3$, and for $\xi=-2$ they become
36: $R^1\times S^3$ geometrically.
37: All solutions with $\xi\neq 0$ can be promoted to D=11 to become vacua of M-theory.}
38: \end{abstract}
39: \pacs{04.65.+e, 11.25.Mj, 11.25.Yb, 11.27.+d }
40: \maketitle
41:
42: Regular supersymmetric backgrounds in gauged supergravities (SUGRA) play an important role
43: in the context of the AdS/CFT correspondence (see \cite{ADSCFT} for a review).
44: Upon uplifting to higher dimensions they become vacua of string/M theory
45: and can be used for the dual description of
46: strongly coupled gauge field theories. In this way,
47: for example, the monopole solution \cite{CV} of the ${\cal N}=4$
48: gauged SUGRA has given rise to the holographic
49: interpretation of confining
50: ${\cal N}=1$ super Yang-Mills (SYM) theory
51: \cite{MN}.
52: Constructing such solutions, however, is rather involved. This is why,
53: despite their importance, very few regular vacua of gauged
54: SUGRA's are known.
55:
56:
57: In this Letter we present a family of globally regular ${\cal
58: N}=1$ vacua that contains the monopole solution of Ref. \cite{CV}
59: as special case. We work in the context of the ${\cal N}=4$ gauged
60: SUGRA in four dimensions. This theory exists in two inequivalent
61: versions: the SU(2)$\times$SU(2) model of Freedman and Schwarz
62: (FS) \cite{FS}, whose solutions were studied in \cite{CV},
63: and the SO(4)
64: model of Gates and Zwiebach (GZ) \cite{GZ}. Both models contain in the
65: bosonic sector the graviton $g_{\mu\nu}$, dilaton $\phi$, axion
66: ${\bf a}$, and two non-Abelian gauge fields $A^a_\mu$ and
67: $B^a_\mu$ with gauge couplings $e_A$ and $e_B$ and with gauge
68: group SU(2)$\times$SU(2). The important difference between the two
69: models is that in the FS model the dilaton potential
70: has no stationary points, while in the GZ model one has
71: (when ${\bf a}=0$) \be \label{0} {\rm U}(\phi)=-\frac{e_A^2}{8}\,
72: (e^{-2\phi}+\xi^2e^{2\phi}+4\xi). \ee This potential does have
73: stationary points, and,
74: depending on the sign of $\xi\equiv e_B/e_A$,
75: its extremal value -- the cosmological constant --
76: %, $-\frac{e_A^2}{4}(2\xi+|\xi|)$,
77: can be positive or negative. If one sets ${\bf a}=B^a_\mu=\xi=0$,
78: then the FS and GZ models coincide and admit as a solution the
79: ${\cal N}=1$ vacuum of Ref. \cite{CV} -- the Chamseddine-Volkov
80: (CV) monopole. If ${\bf a}=B^a_\mu=0$ but $\xi\neq 0$, then the
81: two models are no longer the same, and we find that within the GZ
82: model the CV monopole admits generalizations for any $\xi\neq 0$.
83: These solutions are topologically different from the CV monopole,
84: although approach the latter pointwise as $\xi\to 0$. They can be
85: uplifted to D=11, which may suggest a holographic interpretation
86: for them.
87:
88:
89:
90: We consider the ${\bf a}=B^a_\mu=0$ truncation of the GZ model
91: whose bosonic sector is described by the Lagrangian \bea
92: \label{lag} {\cal L} =\frac{1}{4}R
93: -\frac12\partial_\mu\phi\partial^\mu\phi
94: -\frac{1}{4}\,e^{2\phi}F^{a}_{\mu\nu}F^{a\mu\nu} -{\rm U}(\phi).
95: \eea Here $F^a_{\mu\nu}=\partial_\mu A^a_\nu-\partial_\nu A^a_\mu
96: +\epsilon_{abc}A^b_\mu A^c_\nu$ with $a=1,2,3$, the scale is
97: chosen such that $e_A=1$, and U$(\phi)$ is given by (\ref{0}).
98: Consistency of setting the axion to zero requires that $\ast
99: F^{a}_{\mu\nu}{F}^{a\mu\nu}=0$. The theory contains also fermions:
100: the gaugino $\chi$ and gravitino $\psi_\mu$, whose SUSY variations
101: for a purely bosonic background are \bea \label{SUSY}
102: \delta{\chi}&=& \frac{1}{\sqrt{2}}\,\gamma^\mu
103: \partial_\mu\phi\,\epsilon+ \frac{1}{2}\,e^{\phi}{\cal F}\epsilon
104: +\frac{1}{4}\,(e^{-\phi}-\xi e^{\phi})\epsilon, \\
105: \delta{\psi}_\mu&=&
106: {\cal D}_\mu\epsilon +
107: \frac{1}{2\sqrt{2}}\,e^{\phi}{\cal F}\gamma_\mu \epsilon
108: +\frac{1}{4\sqrt{2}}\,(e^{-\phi}+\xi e^{\phi})\gamma_\mu\epsilon. \nonumber
109: \eea
110: Here ${\cal F}=\frac12\alpha^aF^a_{\alpha\beta}\gamma^\alpha\gamma^\beta$ and
111: $
112: {\cal D}_{\mu}=
113: {\partial}_{\mu}
114: +\frac14\,\omega_{\alpha\beta,\mu}\gamma^\alpha\gamma^\beta
115: -\frac{1}{2}A^a_{\mu}\alpha^a$;
116: the late ($\mu,\nu$) and early $(\alpha,\beta)$ Greek letters correspond
117: to the spacetime and tangent space indices, respectively.
118: The gamma matrices are subject to
119: $\frac12(\gamma_{{\alpha}}\gamma_{{\beta}}
120: +\gamma_{{\beta}}\gamma_{{\alpha}})=
121: \eta_{\alpha\beta}\equiv
122: {\rm diag}(-1,1,1,1)$.
123: Introducing Pauli matrices of four different types,
124: $\sigma^a$, $\underline{\sigma}^b$, $\tau^c$, $\underline{\tau}^d$,
125: which act in four different spaces, respectively
126: (such that, for example, $[\sigma^a,\underline{\sigma}^b]=0$),
127: one can choose $\gamma^\alpha\equiv
128: (\gamma^0,\gamma^a)=(i\underline{\sigma}^1,\underline{\sigma}^2\sigma^a)$.
129: The gauge group SU(2)$\times$SU(2) is generated by the antihermitean matrices
130: $\alpha^a$ and $\beta^b$, $[\alpha^a,\beta^b]=0$,
131: $\alpha^a\alpha^b=-\epsilon_{abc}\alpha^c-\delta_{ab}$,
132: and similarly for $\beta^a$. One can choose
133: $\alpha^a=i\tau^a$, $\beta^a=i\underline{\tau}^a$.
134: The generators $\beta^a$ correspond to the field $B^a_\mu$
135: that is truncated to zero.
136:
137:
138: We wish to study fields that preserve some of the supersymmetries,
139: in which case $\delta\chi=\delta\psi_\mu=0$ for certain $\epsilon\neq 0$.
140: We restrict to
141: the static and spherically
142: symmetric sector parameterized by coordinates $(t,\rho,\vartheta,\varphi)$
143: with
144: \bea \label{sol}
145: ds^2_{(4)}&=&-e^{2V(\rho)}dt^2+e^{2\lambda(\rho)}d\rho^2+r^2(\rho)d\Omega^2, \\
146: \tau^a A^a_\mu dx^\mu&=&\frac{i}{2}(1-w(\rho))[T,dT],~~~~~\phi=\phi(\rho). \label{sol1}
147: \eea
148: Here, with $n^a=(\sin\vartheta\cos\varphi,\sin\vartheta\sin\varphi,\cos\vartheta)$, one has
149: $d\Omega^2=dn^a dn^a$ and $T=\tau^a n^a$. Imposing the isotropic
150: gauge condition, $r=\rho e^{\lambda}$, the spatial part of the metric becomes
151: conformally flat, $ds_{(3)}^2=e^{2\lambda}dx^a dx^a$, with $x^a=\rho n^a$.
152: Choosing the
153: tetrad $\theta^\alpha=(e^{V}dt,e^{\lambda}dx^a)$,
154: the spin connection is obtained from
155: $d\theta_\alpha+\omega_{\alpha\beta}\wedge\theta^\beta=0$.
156: Setting in (\ref{SUSY}) $\delta\chi=\delta\psi_\mu=0$
157: gives then the equations for the
158: SUSY Killing spinors $\epsilon$:
159: \bea \label{d0}
160: 0&=&{2\sqrt{2}}\,{\rm e}^{-\lambda}\phi'\,
161: \underline{\sigma}^2\,
162: (\vec{n}\vec{\sigma})\epsilon+
163: 2\,{\rm e}^{\phi}\,
164: {\cal F} \epsilon
165: +
166: ({\rm e}^{-\phi}-\xi e^\phi)
167: \epsilon \, , \nonumber \\
168: 0&=&2i\sqrt{2}e^{-V}\underline{\sigma}^1\partial_t\epsilon+
169: {\sqrt{2}}\,e^{-\lambda}(V-\phi)'\underline{\sigma}^2(\vec{n}\vec{\sigma}) \epsilon
170: +\xi e^\phi\epsilon, \nonumber \\
171: 0&=&
172: \vec{\nabla}\epsilon+
173: \frac{i}{2}\,\lambda'\,
174: (\vec{n}\times\vec{\sigma})\epsilon
175: +\frac{i}{2}\frac{w-1}{\rho}\,
176: (\vec{n}\times\vec{\tau})\epsilon
177: \nonumber \\
178: &+&
179: \frac{e^{\lambda}}{4\sqrt{2}}(
180: 2{\rm e}^{\phi}{\cal F}+
181: {\rm e}^{-\phi}+\xi e^\phi)\underline{\sigma}^2
182: \vec{\sigma}\epsilon.
183: \eea
184: Here
185: ${\cal F}=-r^{-2}\, [
186: \rho w'(\vec{\sigma}\vec{\tau})+
187: (w^2-1-\rho w')
188: (\vec{n}\vec{\sigma})(\vec{n}\vec{\tau})]$ and $^{\prime}\equiv \frac{d}{d\rho}$;
189: also the usual operations for Euclidean 3-vectors
190: are assumed, for example
191: $\vec{n}\vec{\sigma}\equiv n^a\sigma^a$ and
192: $\vec{\nabla}\equiv \partial/\partial x^a$.
193: Eqs. (\ref{d0}) comprise
194: an overdetermined system of 80 equations for 16 components of $\epsilon$,
195: whose consistency conditions we shall now study.
196:
197: Let $\psi_A$, $\chi_A$, $\underline{\psi}_A$,
198: $\underline{\chi}_A$ be eigenspinors of $\sigma^3$, $\tau^3$,
199: $\underline{\sigma}^1$, $\underline{\tau}^2$, respectively,
200: with the eigenvalues $(-1)^A$, $A=1,2$.
201: We make the ansatz
202: $\epsilon=\epsilon_{AB}$ with $A,B=1,2$ and
203: \be \label{spinor}
204: \epsilon_{AB}=
205: {\cal U}\exp(i\Psi t)
206: [\Phi_{-}(\rho)+
207: \Phi_{+}(\rho)\,\underline{\sigma}^2
208: (\vec{n}\vec{\sigma})]\,
209: \epsilon_0\,\underline{\psi}_A\underline{\chi}_B.
210: \ee
211: Here
212: $\Psi=\frac{1}{2\sqrt{2}}(-1)^{A+B}{\xi Q\, }\,\underline{\tau}^2$
213: where $Q$ is a real constant,
214: ${\cal U}=\exp(-\frac{i}{2}\tau_3\varphi)\exp(-\frac{i}{2}\tau_2\vartheta)$,
215: and
216: $\epsilon_0=\psi_1\,\chi_2-\psi_{2}\chi_1$. In fact, $ \epsilon_{AB}$
217: is the most general spinor whose total angular momentum,
218: including the orbital part plus spin plus isospin, is zero.
219: Inserting (\ref{spinor}) into (\ref{d0}), the variables decouple, and the
220: system reduces to six linear algebraic and two ordinary
221: differential equations for $\Phi_{\pm}$:
222: \bea \label{dif}
223: {A}^{\pm}_{(m)}\Phi_\pm=B^{\mp}_{(m)}\Phi_\mp,~~~~
224: \Phi_{\pm}'-{\cal A}\Phi_\pm ={\cal B}\Phi_\mp~.
225: \eea
226: Here the
227: coefficients ${A}^{\pm}_{(m)}$, ${B}^{\pm}_{(m)}$ ($m=1,2,3$),
228: ${\cal A}$ and ${\cal B}$ are functions of the background
229: amplitudes $V,\lambda,r,w,\phi$ and their derivatives. The
230: algebraic equations can have a nontrivial solution if only their
231: coefficients fulfill the conditions $A^{+}_{(m)}
232: A^{-}_{(n)}=B^{+}_{(m)} B^{-}_{(n)}$, of which only 5 are
233: independent. Introducing $N=\rho\lambda'+1$, these 5 conditions
234: are equivalent to the first 5 of the following 6 relations: \bea
235: &&V'-\phi'=\xi\,\frac{P}{\sqrt{2}N}e^{\phi+\lambda},~~~~~
236: Q=e^{V+\phi}\frac{w}{N}, ~~~~~\label{1} \\
237: &&\phi'=\sqrt{2}\,\frac{BP}{N}\, e^{\lambda},~~~~~~
238: w'=-\frac{rwB}{N} e^{-\phi+\lambda}, \label{2} \\
239: &&N=\sqrt{w^2+P^2},~~~~~~
240: r'=Ne^\lambda. \label{3}
241: \eea
242: Here
243: $P=e^\phi\frac{1-w^2}{\sqrt{2}r}+\frac{r}{2\sqrt{2}}(e^{-\phi}+\xi e^\phi)$
244: and $B=-\frac{P}{\sqrt{2}r}+\frac12e^{-\phi}$.
245: These relations impose nonlinear differential constraints on the
246: background functions $\phi$, $w$, $V$, $\lambda$ and the parameter $Q$.
247: Remarkably, although we have in (\ref{1}) two equations for the
248: same function $V(\rho)$, the first of these equations is in fact a differential
249: consequence of the second one, and so the system is not overdetermined.
250: The last equation in (\ref{3}), added for the later convenience,
251: is the identity (in the isotropic gauge used) implied by the definition
252: of $N$.
253: If Eqs.(\ref{1})--(\ref{3})
254: are fulfilled, the algebraic equations in (\ref{dif})
255: are consistent with each other and express $\Phi_{-}$ in terms of $\Phi_{+}$.
256: Inserting this to the first
257: differential constraint in (\ref{dif}) gives a linear differential equation
258: for $\Phi_{+}(\rho)$, whose solution can be expressed in quadratures.
259: The second differential constraint in (\ref{dif}) then turns out to be fulfilled
260: { identically}, by virtue of Eqs. (\ref{1})--(\ref{3}).
261:
262: The Bogomolnyi equations (\ref{1})--(\ref{3}) therefore provide
263: the full set of consistency conditions that guarantee the
264: existence of SUSY Killing spinors. One can now pass in these
265: equations to an arbitrary gauge by treating $\lambda(\rho)$ as a
266: free function subject to a gauge condition, while considering the
267: second relation in (\ref{3}) as the dynamical equation for the
268: Schwarzschild radial function $r(\rho)$.
269: Finding then $\Phi_{\pm}$ gives the spinor $\epsilon_{AB}$ for each choice of $A,B$,
270: which finally corresponds to four independent SUSY Killing spinors, that is to
271: ${\cal N}=1$.
272:
273: Introducing $y^1=w$, $y^2=\phi$, $y^3=V$, $y^4=V+\ln r$, the
274: Bogomolnyi equations can also be written as \be \label{B}
275: Y^n\equiv \frac{dy^n}{d\rho}-G^{nm}\frac{\partial {\cal
276: W}}{\partial y^m}=0,~~~~~ m,n=1,2,3,4, \ee where the target space
277: metric is defined by $G_{mn}=r^2\,e^{V-\lambda}{\rm
278: diag}(2e^{2\phi}/r^2,1,1,-1)$ and the superpotential is ${\cal
279: W}=-r e^V N$ with $N$ given by (\ref{3}). We note also that
280: inserting the ansatz (\ref{sol}),(\ref{sol1}) to the Lagrangian
281: (\ref{lag}) and integrating over the angles gives $\int {\cal
282: L}\sqrt{-g}\,d\vartheta d\varphi=4\pi G_{mn}Y^mY^n+$total
283: derivative. It then follows that solutions of Eqs. (\ref{B}) are
284: stationary points of the action.
285:
286: To integrate the Bogomolnyi equations (\ref{1})--(\ref{3}), the
287: problem actually reduces to studying the closed subsystem
288: (\ref{2})--(\ref{3}) for $\phi,w,r$, since $V$ can be obtained
289: afterwards from (\ref{1}). It seems that these equations can be
290: resolved analytically only for some special values of $\xi$, and
291: we shall therefore resort to numerical analysis to study the
292: generic case. First of all, we notice the following symmetry of
293: the equations: if $\xi$, $\phi(\rho)$, $w(\rho)$, $r(\rho)$,
294: $\lambda(\rho)$ is a solution for some value of $\xi$, then, for
295: any $\epsilon$,
296: \be \label{sym}
297: e^{-2\epsilon}\xi,~~\phi(\rho)+\epsilon,~~w(\rho),~~e^\epsilon
298: r(\rho), ~~\lambda(\rho)+\epsilon
299: \ee
300: is also a solution. To fix
301: this symmetry, we impose the condition $\phi_0=0$, with $\phi_0$
302: being the value of $\phi$ at $r=0$. Since the sign of $\xi$ is
303: invariant under (\ref{sym}), there are three separate cases to
304: study: $\xi>0$, $\xi<0$, and $\xi=0$.
305:
306: \underline{$\xi=0$.} Eqs. (\ref{2})--(\ref{3}) reduce in this case
307: to the system previously studied in the context of the
308: half-gauged FS model \cite{CV}. Its solution is the CV monopole:
309: \bea \label{CV}
310: w=\frac{\rho}{\sinh\rho}, ~~~~~
311: \frac{re^{-\phi}}{\sqrt{2}}=\frac{\rho
312: e^{-2\phi}}{w}=\sqrt{2\rho\coth\rho-w^2-1},~~
313: \eea and
314: $e^\lambda=\sqrt{2}e^\phi$. In this case one has $\phi\sim\rho$ as
315: $\rho\to\infty$.
316:
317: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
318: \begin{figure}
319: \epsfxsize=13cm
320: \epsfysize=8cm
321: \centerline{\epsffile{ads.eps}}
322: \caption{Globally regular solutions with $\xi>0$. For $\xi=0.001$ and $\xi=0$
323: the amplitudes $w$ are almost identical.}
324: \label{fig1}
325: \end{figure}
326: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
327: \underline{$\xi>0$.}
328: Choosing the Schwarzschild gauge, $\rho=r$,
329: the essential equations are given by (\ref{2}),(\ref{3})
330: with $e^\lambda=1/N$.
331: They determine $\phi(r)$, $w(r)$,
332: while (\ref{1}) gives $e^{2V}=Q^2{N^2}e^{-2\phi}/w^2$.
333: We are interested in everywhere regular solutions, in which case
334: $\phi=O(r^2)$, $w=1+O(r^2)$, $N=1+O(r^2)$
335: for $r\to 0$. For $r\to\infty$ one has
336: \bea \label{infinity}
337: \phi&=&-\frac{\ln\xi}{2}+\frac{b}{r} +O(r^{-2}),~~~
338: w=w_\ast-\frac{w_\ast b}{r} +O(r^{-2}), \nonumber \\
339: N^2&=&1+\frac12\,\xi b^2-\frac{2b(w_\ast^2-1)}{r}+\frac{\xi r^2}{2}+O(r^{-2}),
340: \eea
341: where $b$ and $w_\ast$ are integration constants.
342: The numerical integration of the equations reveals
343: for every value of $\xi>0$ a global solution $\phi(r)$, $w(r)$ with such boundary
344: conditions in the interval
345: $r\in[0,\infty)$; see Fig.I. For all these solutions the dilaton varies in the
346: finite range and runs into the stationary point of U$(\phi)$ for $\rho\to\infty$.
347: As $\xi\to 0$, the asymptotic value of $\phi$ tends to infinity and
348: the solutions approach pointwise
349: the CV monopole (\ref{CV}).
350:
351: For $\xi=1$ the solution can be obtained
352: analytically: $\phi(r)=0$, $w(r)=1$.
353: Choosing $Q=1$ (the value of $Q$ can be adjusted
354: by rescaling the time),
355: the metric assumes the
356: standard anti de Sitter (AdS) form,
357: $ds^2=-N^2dt^2
358: +dr^2/N^2
359: +r^2d\Omega^2$
360: with $N^2=1+\frac{r^2}{2}$,
361: while the gauge field vanishes.
362: This solution actually has ${\cal N}=4$ supersymmetry,
363: since in this case there are additional SUSY Killing spinors
364: not contained in the ansatz (\ref{spinor}).
365:
366: Solutions with $\xi\neq 1$ describe globally regular ${\cal N}=1$
367: deformations of the AdS.
368: Their asymptotic form is determined by (\ref{infinity}). Choosing the new radial
369: coordinate $\tilde{r}=r/\sqrt{1+\delta}$ with $\delta=\frac12\,\xi b^2$
370: and setting $Q=w_\ast/\sqrt{\xi(1+\delta)}$,
371: the metric asymptotically approaches
372: \be
373: ds^2=-N^2 dt^2
374: +\frac{d\tilde{r}^2}{N^2}
375: +(1+\delta)\,\tilde{r}^2d\Omega^2,
376: \ee
377: where $N^2=1-\frac{2M}{\tilde{r}}+\frac{\xi\tilde{r}^2}{2}$ and
378: $M=b(w_\ast^2-1)/(1+\delta)^{3/2}$.
379: This is the Schwarzschild-AdS metric with an {\it excess} of the solid angle --
380: the area of the 2-sphere of constant $\tilde{r}$ being $4\pi(1+\delta)\tilde{r}^2$
381: in this geometry. The excess parameter $\delta$ and the `mass' $M$
382: vanish only for $\xi=1$, and they tend to infinity as $\xi\to 0$.
383:
384: \underline{$\xi<0$.} Solutions in this case are of the `bag of gold' type, since
385: they have compact spatial sections with the $S^3$ topology.
386: The range of the Schwarzschild function $r(\rho)$ is finite:
387: it starts from zero at $\rho=0$ (`north pole'), increases up to a maximal
388: value at some $\rho_e>0$ (`equator'), and then decreases down to zero at some
389: $\rho_\ast>\rho_e$ (`south pole'). Since
390: $N\sim r'=0$ at the equator, Eqs. (\ref{1})--(\ref{3})
391: become singular at this point. To desingularize them, we set
392: $P=wS$ thus obtaining $N=w\sqrt{1+S^2}=we^{V+\phi}$
393: (having chosen $Q=1$ in (\ref{1})). Using this in (\ref{1})--(\ref{3}) reduces
394: the system to
395: \bea \label{reg}
396: r'&=&we^{V+\phi+\lambda},~~~~~\phi'=\sqrt{2}BSe^{\lambda-V-\phi}, \\
397: w'&=&-rBe^{\lambda-V-2\phi},~~~~~V'-\phi'=\frac{\xi}{\sqrt{2}}Se^{\lambda-V},
398: \nonumber
399: \eea
400: with $S=\sqrt{e^{2V+2\phi}-1}$
401: and $B=-\frac{wS}{\sqrt{2}r}+\frac12 e^{-\phi}$. In addition, the relation
402: $P=wS$ with $P$ defined after Eqs. (\ref{3}) gives the first integral
403: for these equations.
404: Eqs. (\ref{reg}) are completely regular at the equator,
405: whose position coincides with zero of $w$.
406: Imposing the gauge condition
407: $\lambda=0$ and demanding the solution to be regular at the `north pole' gives
408: $r=\rho+O(\rho^3)$, $w=1+O(\rho^2)$, $\phi=O(\rho^2)$, $V=O(\rho^2)$
409: for small $\rho$. At the `south pole' we find the formal
410: power series solution to be generically
411: \bea \label{pole}
412: r=3w_\ast\beta_\ast\,x+O(x^3),~
413: w=w_\ast + \frac{w_\ast \kappa}{8} \,x^4+O(x^6),~ \\
414: e^{-\phi}=\frac{|\kappa| x}{3\sqrt{2}}
415: +O(x^3), ~
416: e^{V-\phi}=\nu_\ast +\frac{\xi\beta_\ast\kappa}{4}\,x^2+O(x^4),
417: \nonumber
418: \eea
419: with $x=(\rho-\rho_\ast)^{1/3}$. Here $\rho_\ast$, $w_\ast<0$,
420: $\beta_\ast>0$ and $\nu_\ast$ are integration constants, and
421: $\kappa=(1-w_\ast^2)/(w_\ast^2 \beta_\ast^2)$.
422:
423: Solutions of Eqs. (\ref{reg}) in the interval $\rho\in[0,\rho_\ast]$
424: comprise a one-parameter family labeled by $\xi$.
425: These solutions are regular for $\rho<\rho_\ast$,
426: while at $\rho=\rho_\ast$
427: the dilaton diverges and the curvature is singular too. For any $\xi$, the
428: profile of these solutions is qualitatively similar to the one shown in Fig. 2.
429: As $\xi\to 0$, one has $\rho_\ast\to\infty$, $r(\rho_e)\to\infty$,
430: and the solutions approach pointwise
431: the CV monopole.
432: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
433: \begin{figure}
434: \epsfxsize=13cm
435: \epsfysize=8cm
436: \centerline{\epsffile{comp.eps}}
437: \caption{The compact solutions with $\xi<0$. They are generically
438: singular at the `south pole' where $r$ vanishes and $\phi$ diverges.
439: }
440: \label{fig2}
441: \end{figure}
442: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
443:
444: For one special value, $\xi=-2$,
445: one has $w_\ast=-1$ and the expansions (\ref{pole}) are no longer valid.
446: However, the solution can then be obtained analytically:
447: $\phi=V=P=S=0$, $w=\cos\rho$,
448: $r=\sqrt{2}\sin\rho$, $e^{\lambda}=\sqrt{2}$.
449: This solution is {\it globally} regular, also at the south pole,
450: the spatial geometry being
451: that of the round $S^3$.
452: One can write down the metric and gauge field as
453: \be \label{-2}
454: ds^2=-dt^2+2\theta^a\theta^a,~~~~~~A^a=\theta^a,
455: \ee
456: where $\theta^a$ are invariant forms
457: on $S^3$, $d\theta^a+\epsilon_{abc}\theta^b\wedge\theta^c=0$.
458: However, there is no SUSY enhancement in this case, and so
459: ${\cal N}=1$.
460:
461: We have thus obtained the generalizations of the CV monopole (\ref{CV})
462: that comprise a two-parameter family labeled by $\xi$
463: and $\phi_0$. Although we have described explicitly only
464: solutions with $\phi_0=0$, those with $\phi_0\neq 0$ can be obtained by using the
465: symmetry (\ref{sym}). The solutions generically have ${\cal N}=1$, while for
466: $\xi=1$ the sypersymmetry is enhanced up to ${\cal N}=4$.
467: We know that the $\xi=0$ solution can be uplifted to D=10
468: to become a vacuum of string theory \cite{CV}.
469: It turns out that solutions with $\xi\neq 0$ can be
470: uplifted to $D=11$ to become vacua of M-theory.
471:
472: The derivation of the GZ model via dimensional reduction of D=11 SUGRA
473: was considered in Ref. \cite{Pope}. Using formulas given in there combined
474: with the symmetry (\ref{sym}),
475: every D=4 vacuum $(ds_4^2,A^a,\phi)$ considered above
476: maps to the M-theory solution $(ds_{(11)}^2,F_{[4]})$.
477: The metric in D=11 is given by
478: $ds^2_{(11)}=|\xi|\Delta^\frac{2}{3}ds_4^2+8\Delta^{-\frac13}ds_{(7)}^2$
479: with
480: \be \label{s7}
481: ds^2_{(7)}=
482: \Delta\, d\eta^2
483: +\frac{{\bf c}^2}{X} \sum_a(\theta^a_{(1)}-\frac12 A^a)^2
484: +{{\bf s}^2}X\sum_a(\theta_{(2)}^a)^2.
485: \ee
486: Here
487: $\Delta={\bf s}^2/X+{\bf c}^2 X$ with $1/X=\sqrt{|\xi|}e^{\phi}$, and
488: $\theta^a_{(\iota)}$ $(\iota=1,2)$ are invariant forms on two different
489: 3-spheres,
490: $d\theta^a_\iota+\epsilon_{abc}\theta^b_\iota\wedge \theta^c_\iota=0$.
491: The case $\xi>0$ corresponds to the reduction on $S^7$,
492: one has then ${\bf c}=\cos\eta$, ${\bf s}=\sin\eta$ with $\eta\in[0,\pi/2]$,
493: while for $\xi<0$ one reduces on the ${\cal H}^{(2,2)}$ hyperbolic space \cite{Pope},
494: in which case ${\bf c}=\cosh\eta$, ${\bf s}=\sinh\eta$ and $\eta\in[0,\infty)$.
495: The 4-form in D=11 reads
496: \be \label{F4}
497: F_{[4]}=\sqrt{2}{\bf sc} \ast d\phi\wedge
498: d\eta -\left(
499: \frac{\xi}{|\xi|}\,{{\bf c}^2}X^2+\frac{{\bf s}^2}{X^2}+2
500: \right)\frac{\epsilon_{(4)}}{\sqrt{2}}, \ee
501: where $\epsilon_{(4)}$ is the 4-volume form and $\ast$ is the Hodge dual in the
502: 4-space.
503:
504: These formulas suggest a holographic interpretation for our solutions.
505: For $\xi=0$, according to \cite{MN}, the dual theory is
506: $D=4$, ${\cal N}=1$ SYM. For $\xi=1$
507: we have $\phi=A^a=0$, and the D=11 geometry
508: is $AdS_4\times S^7$. This is the near-horizon limit of the M2 brane,
509: and the dual theory is therefore $D=3$, ${\cal N}=4$ SYM. This suggests that other solutions
510: with positive $\xi\neq 1$ may describe some ${\cal N}=1$ deformations of this theory.
511: It would also be interesting to work out an interpretation for the compact solutions
512: with $\xi<0$,
513: especially for the one given by Eq. (\ref{-2}).
514:
515: M.S.V. thanks Michaela Petrini for discussions. Research of A.H.C.
516: is supported in part by National Science Foundation Grant No.
517: Phys-0313416.
518: \begin{thebibliography}{10}
519:
520:
521:
522:
523: \bibitem{ADSCFT}
524: O.~Aharony, S.~Gubser, J.~Maldacena,
525: H.~Ooguri and Y.~Oz, {\it Phys.Rep.}, {\bf 232} (2000) 183.
526:
527:
528: \bibitem{CV}
529: A.H.~Chamseddine and M.S.~Volkov, {\it Phys.Rev.Lett.}, {\bf 79} (1997) 3343;
530: {\it Phys.Rev.}, {\bf D57} (1998) 6242.
531:
532: \bibitem{MN}
533: J.~Maldacena and C.~Nunez, {\it Phys.Rev.Lett.}, {\bf 86} (2001) 588.
534:
535:
536:
537: \bibitem{FS}
538: D.Z.~Freedman and J.H.~Schwarz,
539: {\it Nucl.Phys.}, {\bf B137} (1978) 333.
540:
541: \bibitem{GZ}
542: S.J.~Gates and B.~Zwiebach,
543: {\it Phys.Lett.}, {\bf B123} (1983) 200;
544: B.~Zwiebach, {\it Nucl.Phys.}, {\bf B238} (1984) 367.
545:
546:
547: \bibitem{Pope}
548: M.~Cvetic, H.~Lu and C.N.~Pope, {\it Nucl.Phys.}, {\bf B574} (2000) 761;
549: M.~Cvetic, G.W.~Gibbons and C.N.~Pope, {\tt hep-th/0401151.}
550:
551: \end{thebibliography}
552:
553: \end{document}
554: