1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2: %                                                  

3: %     This is a revtex4 file to be processed twice

4: %                                                  

5: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

6: \documentclass[aps, twocolumn, superscriptaddress, floatfix, 

7: letterpaper, 

8: preprintnumbers]{revtex4}

9: \usepackage{graphicx}

10: 

11: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

12: % GENERAL DEFINITIONS

13: \newcommand{\be}{\begin{eqnarray}}

14: \newcommand{\ee}{\end{eqnarray}}

15: \newcommand\del{\partial}

16: \newcommand{\el}{\nonumber \hfill \\}

17: \newcommand{\Tr}{\mathrm{Tr}}

18: \newcommand{\nn}{\nonumber }

19: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

20: 

21: \setlength{\topmargin}{-1in}

22: 

23: \begin{document}

24: 

25: %\preprint{SUNY-NTG-04/}

26: 

27: \title{Diquark and Pion Condensation in Random Matrix Models for

28:   two-color QCD}

29: \author {B. Klein} 

30: \affiliation{Institute for Theoretical Physics, University of

31:   Heidelberg, Philosophenweg 19, 69120 Heidelberg, Germany}

32: \author{D. Toublan}

33: \affiliation{Department of Physics, University of Illinois at 

34: Urbana-Champaign,

35: Urbana, IL 61801-3080}

36: \author{J.J.M. Verbaarschot}

37: \affiliation{Department of Physics and Astronomy, State University of New 

38: York at Stony Brook,

39: Stony Brook, NY 11794-3800}

40: 

41: \date{\today}

42: \begin{abstract}

43: We introduce a random matrix model with the symmetries of QCD with

44: two colors at nonzero isospin and baryon 

45: chemical potentials and temperature. 

46: We analyze its phase diagram  

47:  and find phases with condensation of pion and diquark states in 

48: addition to the phases with spontaneously broken chiral symmetries. 

49: In the limit of small chemical potentials and quark masses, we

50: reproduce the mean field results obtained from chiral Lagrangians.

51: As in the case of QCD with three colors, the presence of two

52: chemical potentials breaks the flavor symmetry and leads to phases that are

53: characterized by different behaviors of the chiral condensates for each

54: flavor. In particular, the phase diagram we obtain 

55: is                 

56: similar to QCD with three colors and three flavors of quarks of equal masses

57: at zero baryon chemical potential and nonzero isospin and strange

58: chemical potentials. 

59: A tricritical point of the superfluid transitions found in lattice

60: calculations and from an analysis in terms of chiral Lagrangians does

61: not appear in the random matrix model. 

62: Remarkably,  at fixed isospin chemical potential,

63: for the regions outside of the superfluid phases,

64: the phase diagram in the

65: temperature -- baryon chemical potential plane  for two colors and

66: three colors are qualitatively the same.

67: 

68: 

69: \end{abstract}

70:  

71: \pacs{}

72: 

73: \maketitle

74: 

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

76: \setcounter{equation}{0}

77: 

78: \section{Introduction}

79: \label{nc2section-1}

80: \setcounter{equation}{0}

81: Recently, the phase diagram of QCD

82: at nonzero chemical potential for the baryon number has received

83: a great deal of attention. 

84: It has been found that a plethora of phase transitions are possible 

85: as a function of the baryon chemical potential, $\mu_B$, the isospin chemical

86: potential, $\mu_I$, and the temperature $T$. 

87: Among others, a superconducting phase

88: \cite{Alford:1997zt, Rapp:1997zu}, a pion condensate \cite{Migdal:jn,

89:   Son:2000xc} 

90: and a critical endpoint have been predicted \cite{Barducci:1989wi,

91:   Halasz:1998qr, Alford:1997zt} 

92: (see \cite{Rajagopal:2000wf} for a review). A systematic analysis of

93: the phase diagram is possible in the $\mu_B=0$ plane and, perturbatively,

94: for asymptotically large values of $\mu_B$ 

95: \cite{Son:1998uk}.

96: At finite, not asymptotically large baryon chemical potential, 

97: where the interactions between the 

98: quasi-particles that carry color charge are strong, the theoretical

99: basis of these predictions is rather weak.

100: So far, it has not been possible to confirm the

101: existence of the proposed phases in this region, 

102: neither from experimental results nor from lattice calculations. 

103: The reason for the failure of standard lattice

104: simulations is that in the presence of a chemical potential for baryons the

105: fermion determinant in the partition function becomes complex so that

106: a probabilistic interpretation of the partition function is not possible.

107: 

108: 

109: Thus, it is attractive to try to gain insight into the properties of such

110: condensed phases by studying QCD-like theories  

111: which avoid some of these complications. In this paper we study QCD with

112: two colors with baryons that are colorless diquark states. 

113: As in the case of

114: three colors, chiral symmetry is spontaneously broken and the theory

115: is confining at low temperatures and chemical potential. However, at

116: nonzero baryon number chemical potential,

117: two important simplifications occur in this theory.

118: First, the condensing diquark states are Goldstone bosons associated

119: with the spontaneous breaking of chiral symmetry with a critical chemical

120: potential equal to half the pion mass.

121: This makes it possible

122: to analyze the condensed phase by means of chiral Lagrangians

123: \cite{Kogut:1999iv, Kogut:2000ek, Splittorff:2001fy,

124:   Splittorff:2002xn, Splittorff:2000mm}.

125: Second, as a consequence of the pseudoreality of the $SU(2)$

126: gauge group, the fermion determinant remains real when a chemical potential is

127: introduced. Therefore, for an even number of flavors,

128: this theory can be simulated on the lattice without

129: encountering a sign problem. 

130: 

131: The phase diagram in two-color QCD has been studied using a variety of

132: methods. 

133: The appearance of a baryonic diquark condensate 

134: above a critical chemical potential of $\mu_{Bc} \simeq

135: m_\pi/2$ was first  observed in lattice simulations at strong

136: coupling \cite{Dagotto:1986gw, Dagotto:1986ms, Dagotto:xt}.     

137: More recently, the phase diagram was analyzed  in the framework of

138: chiral Lagrangians \cite{Kogut:1999iv, Kogut:2000ek,Splittorff:2001fy,

139: Splittorff:2002xn}, 

140: where it was found that the critical value of the

141: chemical potential is indeed given by the mass of the lightest state,

142: and that the transition is of second order at zero temperature.  

143: Lattice simulations confirm these results \cite{Aloisio:2000nr, 

144:   Aloisio:2000if, Aloisio:2000rb, Hands:1999md, Hands:2000ei,

145:   Hands:2000hi, Liu:2000in, Muroya:2001av, Muroya:2002ry,

146:   Kogut:2001if, Kogut:2001na, Kogut:2002tm, 

147:   Kogut:2002zg, Kogut:2002cm, Kogut:2003ju}.

148: In \cite{Kogut:2001if,Kogut:2002zg,Kogut:2002cm} 

149: it was observed that beyond a critical  value of the

150: baryon chemical potential and temperature, the transition to the

151: diquark condensation transition ceases to be of second order and becomes

152: a first order transition. This

153: implies the existence of a tricritical point.  In chiral

154: perturbation theory, such a tricritical point was found 

155: \cite{dunne,Splittorff:2001fy} at

156: values for the chemical potential and the temperature of the order of

157: the pion mass $m_\pi$.  

158: 

159: In physical applications, such as in neutron stars or

160: in heavy ion collisions, the chemical potentials for different quark

161: flavors are not necessarily equal. While the chemical potential for

162: baryon charge is the same for all quarks, an additional isospin

163: chemical potential parameterizes the difference between the chemical

164: potentials for two light quark flavors. Its effect on the phase

165: diagram  has been studied for $N_c=2$ in \cite{Splittorff:2000mm}

166: using effective chiral Lagrangians, and for $N_c=3$ in

167: \cite{Klein:2003fy} using a random matrix model, in

168: \cite{Toublan:2003tt, Frank:2003ve, Barducci:2004tt} by means of a

169: Nambu--Jona-Lasinio 

170: model, and in \cite{Barducci:2003un}

171: within the ladder approximation.  

172: 

173: In this article, we employ a random matrix model to study

174: QCD with two colors at finite 

175: chemical potentials for isospin and baryon density and finite

176: temperature. Originally, random matrix models have been introduced in  

177: QCD in order to describe correlations of low-lying eigenvalues of the

178: Dirac operator \cite{Shuryak:1992pi, Verbaarschot:1994qf}. These

179: random matrix models are equivalent to the static part of a chiral

180: Lagrangian which is uniquely determined by the symmetries of the

181: microscopic theory 

182: \cite{Osborn:1998qb, Damgaard:1998xy, Toublan:1999hi, Toublan:2000dn}. 

183: ~From this connection, it has been

184: shown that they provide an exact analytical description of 

185: correlations in the low-lying Dirac spectrum. 

186: In addition to their use as exact analytical models for the spectrum,

187: random matrix models can also serve as schematic models for the phase

188: transitions in QCD. The first of these models was introduced in

189: \cite{Jackson:1995nf, Wettig:wi}

190: to describe the chiral restoration transition at

191: finite temperature. At finite baryon chemical potential, they have

192: been successfully used to explain the failure of the quenched

193: approximation in lattice QCD \cite{Stephanov:1996ki}, and to predict a

194: tricritical point for the chiral restoration transition at finite

195: density and temperature \cite{Halasz:1998qr}. The static part of the 

196: effective chiral Lagrangian for QCD with the coupling of charged

197: Goldstone bosons to a chemical potential has also been derived from

198: a chiral random matrix model \cite{Toublan:1999hx}. More recently, 

199: closely related models with a random gauge potential have been

200: considered for the superconducting phases \cite{Vanderheyden:1999xp,

201:   Vanderheyden:2000ti} and for diquark condensation for $N_c=2$ 

202: \cite{Vanderheyden:2001gx}. Though these models have additional

203: symmetries beyond those considered in this article, they lead to the

204: same spectral density as the corresponding chiral random matrix

205: ensembles \cite{Vanderheyden:2002gz}. 

206: 

207: The random matrix theory approach rests on the idea that it

208: is the matrix equivalent of a Landau-Ginzburg potential. Both theories

209: are based on the symmetries of the microscopic partition function. 

210: Indeed, as we will also see in this paper, it is possible to derive

211: a Landau-Ginzburg functional from the random matrix theory. 

212: The presence of a diquark and a pion condensation 

213: phase raises several interesting questions, which can be 

214: addressed by studying a random matrix model. In particular, we hope

215: to investigate the robustness of the tricritical point that has been

216: found in lattice simulations \cite{Kogut:2002cm}

217: and in chiral perturbation theory

218: \cite{Splittorff:2001fy}.

219: In general, we wish to identify which features of the phase diagram 

220: are generic and due to the underlying symmetries and which features

221: are model dependent.

222: 

223: 

224: 

225: This article is organized as follows. In section~\ref{nc2section-2}, we give a

226: short review of the symmetries and phase diagram of 

227: the partition function of QCD with two colors.

228: Our random matrix model is introduced in

229: section~\ref{nc2section-3}. We show

230: that it can be rewritten exactly in terms of the static part of

231: a chiral Lagrangian. With an ansatz

232: based on  the structure of the expected condensates we obtain an

233: effective potential for these condensates. Our main results are

234: presented in section~\ref{nc2section-5}, where we analyze the phase

235: diagram in the $\mu_B$-$\mu_I$ chemical potential plane.  

236: Concluding remarks are made in section~\ref{nc2section-6}. 

237: 

238: 

239: \section{QCD with two colors}

240: \label{nc2section-2}

241: \setcounter{equation}{0}

242: The partition function for QCD with two colors 

243: is given by

244: \be

245: Z_{QCD} &=& \left \langle \prod_{f=1}^{N_f} \det(D+m_f + \mu_f 

246: \gamma_0)\right\rangle,

247: \label{ZQCDNc2article}

248: \ee

249: where the Euclidean Dirac operator is $D=\gamma_\mu

250: D_\mu=\gamma_\mu(\partial_\mu + iA_\mu)$, $\gamma_\mu$ are the

251: Euclidean $\gamma$-matrices, and $A_\mu$ is   

252: the $SU(2)$-valued gauge potential. For each of the $N_f$ quark flavors, the 

253: quark mass is given by $m_f$ and the chemical potential by $\mu_f$. The 

254: brackets denote an average over the Euclidean Yang-Mills action. In the case 

255: of only two flavors, the chemical potentials can be expressed in terms of 

256: isospin and baryon chemical potential, 

257: \be

258: \mu_B&=& \frac{1}{2} (\mu_1 + \mu_2) \el

259: \mu_I&=& \frac{1}{2} (\mu_1 - \mu_2).  

260: \ee

261: 

262: Due to the pseudoreality of the $SU(2)$ gauge group, the flavor

263: symmetry group is enlarged  to $SU(2N_f)$ from the 

264: usual $SU(N_f) \times SU(N_f)$ in

265: case of QCD with three or more colors.

266: In the vacuum, this symmetry

267: is spontaneously broken to $Sp(2N_f)$. Thus, in the chiral limit, 

268: one expects $N_f(2N_f-1) -1$ exactly massless

269: Goldstone bosons. Among these Goldstone bosons there are diquark

270: states which carry baryon charge, in addition to the mesonic pion

271: states which carry a charge with respect to isospin number. 

272: The mass term breaks the chiral flavor symmetry explicitly down to

273: $Sp(2N_f)$. A baryon chemical potential term by itself is invariant

274: under an $U(1)_B \times SU(N_f) \times SU(N_f)$ subgroup of the

275: original $SU(2N_f)$, and in the presence of a mass term this symmetry is

276: further broken down to $U(1)_B \times SU(N_f)$. 

277: 

278: The presence of an additional chemical potential coupling to light

279: bosonic states greatly increases the number of possible phases.

280: At low temperature and chemical potential we expect that chiral symmetry

281: is spontaneously broken. 

282: The phases with broken chiral symmetry are characterized by the

283: order parameters $\langle\bar u u\rangle$ and $\langle \bar d d \rangle$. 

284: As in the case of $N_c=3$ \cite{Klein:2003fy}, in the presence of two

285: different chemical potentials for the two flavors $u$ and $d$, there

286: is no reason to believe that $\langle\bar u u\rangle =\langle

287: \bar d d \rangle$, and in general, one expects that the chiral condensates

288: for both flavors are not equal. 

289: At $\mu_I \neq 0$, when the two chemical potentials are different, it cannot

290: be excluded that chiral symmetry

291: restoration will take place via two separate phase transitions.

292: 

293: In addition to the pion condensate $\rho=\frac{1}{2}(\langle \bar u

294: \gamma_5 d \rangle - \langle \bar d \gamma_5 u \rangle )$, already

295: known from the case $N_c=3$, which

296: appears above a critical isospin chemical potential, for QCD with two colors

297: we also have a

298: diquark condensate $\frac{1}{4} i ( \langle 

299: d^\dagger u^*\rangle - \langle u^\dagger  d^* \rangle + \langle d^T u

300: \rangle -  \langle u^T d \rangle ) $ above a critical baryon chemical

301: potential. ~From general arguments, one expects a phase transition 

302: when the chemical potential becomes

303: equal to the ratio of the mass and the  

304: charge for the lightest charged excitation.

305: Indeed, this has been shown explicitly by an analysis

306: in terms of chiral Lagrangians at low chemical

307: potential and temperature

308: \cite{Kogut:1999iv,Kogut:2000ek,Splittorff:2001fy, Splittorff:2002xn,

309:   Splittorff:2000mm}. 

310: It was also found that

311: the transition to a baryonic diquark condensation

312: phase or a pion condensation is of second order.

313: 

314: ~From the symmetries of the partition function alone, it is possible to

315: predict some of the properties  of the phase diagram. 

316: As has been shown in \cite{Kogut:1999iv, Kogut:2000ek}, the

317: determinantal term in the partition function (\ref{ZQCDNc2article}) can be

318: rewritten in such a way that the enlarged flavor symmetry becomes

319: manifest. Using a chiral representation for the Euclidean

320: $\gamma$-matrices, in which

321: $\gamma_0\gamma_\mu=\mathrm{diag}(i\sigma_\mu, -i\sigma_\mu^\dagger)$

322: with the spin matrices $\sigma_\mu=(-i, \sigma_k)$, and the

323: pseudoreality property of the generators of the gauge group, 

324: $-\tau_2 \tau_a \tau_2=\tau_a^{*}$, 

325: the fermionic part of the Lagrangian with the 

326: Dirac operator can be rewritten in terms of a new spinor basis.

327: In the chiral limit $m_f \to 0$, but including a nonzero chemical potential,  

328: it becomes,

329: \be

330: \left(\begin{array}{c} \psi^*_L \\ \widetilde{\psi}^{*}_R\\\end{array}\right)^T

331: \left(\begin{array}{cc} 

332: i\sigma_\mu D_\mu +\mu_f & 0 \\

333: 0 & i\sigma_\mu D_\mu - \mu_f \\

334: \end{array}\right)

335: \left(\begin{array}{c} \psi_L \\

336:     \widetilde{\psi}_R\\\end{array}\right),

337: \ee

338: where the spinors $\widetilde{\psi}_R=\sigma_2\tau_2 \psi^*_R$

339: transform in the same representation of the color group as the spinors

340: $\psi_L$. 

341: Clearly,

342: at $\mu_f=0$ and $m_f =0$, the flavor symmetry group is enlarged to

343: $SU(2N_f)$.

344: For two flavors, the fermion determinant in the partition

345: function factorizes into

346: \be

347: \lefteqn {

348: \det(m^2 -(iD_\mu\sigma_\mu + \mu_1)\sigma_2\tau_2

349: (i(D_\mu\sigma_\mu)^T - \mu_1)\sigma_2\tau_2) } \nn \\

350: &&\times 

351: \det(m^2 -(iD_\mu\sigma_\mu + \mu_2)\sigma_2\tau_2

352: (i(D_\mu\sigma_\mu)^T - \mu_2) \sigma_2\tau_2).\nn \\

353: \ee

354: The partition function is therefore invariant under the

355: interchanges $\mu_{f} \to -\mu_f$, $f=1, 2$, and $\mu_1

356: \leftrightarrow \mu_2$. 

357:  Translated into

358: the physical chemical potentials, the partition function is invariant

359: under the individual sign changes $\mu_I \to -\mu_I$ and $\mu_B \to

360: -\mu_B$, and under the interchange $\mu_I \leftrightarrow \mu_B$.    

361: The latter symmetry, which amounts to $\mu_2 \to

362: -\mu_2$, means the interchange of quarks and conjugate anti-quarks. Under

363: this transformation, a diquark condensate is transformed into a pion

364: condensate. 

365: Thus, overall one expects that the phase diagram is

366: symmetric about the lines 

367: $\mu_I^2=\mu_B^2$ and the lines $\mu_I=0$ and $\mu_B=0$. Therefore

368: from the symmetry of the QCD partition function we can restrict our

369: study of the phase diagram to one half quadrant in the

370: $\mu_B-\mu_I$-plane. 

371: 

372: 

373: 

374: \section{Random Matrix Model}

375: \label{nc2section-3}

376: \setcounter{equation}{0}

377: We will introduce a random matrix model for QCD with

378: two colors and study its phase diagram at finite chemical potentials

379: for isospin and baryon 

380: number and finite temperature. 

381: The idea of the random matrix model is to replace the matrix elements of the 

382: Dirac operator by Gaussian distributed random variables, while respecting 

383: all the global symmetries of the QCD partition function. The dependence on 

384: temperature and chemical potentials will enter through external fields, 

385: which are also subject to symmetry constraints.

386: 

387: Besides the chiral and flavor symmetries, the Dirac operator can also have 

388: additional anti-unitary symmetries. For an $SU(2)$ gauge group, the Dirac 

389: operator satisfies the commutation relation 

390: \be

391: [ iD, C \tau_2 K] = 0, \;\;\;\; (C \tau_2 K)^2 = 1,

392: \ee

393: where $C=\gamma_2\gamma_4$ is the charge conjugation matrix, $\tau_2$ is a 

394: generator of the gauge group, and $K$ is the complex conjugation operator. 

395: Due to this anti-unitary symmetry, it is possible to find a basis in which 

396: all matrix elements of $D$ are real. Thus, the matrices in the

397: corresponding random matrix ensemble will have real

398: entries, 

399: which is denoted by the  Dyson index $\beta=1$. 

400: 

401: 

402: In the absence of a temperature and chemical potential, the 

403: matrix elements of the Dirac operator are simply replaced by real random 

404: variables with a Gaussian distribution. The chemical potential breaks the 

405: flavor symmetry in the same way as in the QCD partition function.  

406: Just as the dependence on the chemical potentials, an effective temperature 

407: dependence enters through an external field \cite{Jackson:1995nf}. 

408: Because of 

409: the flavor symmetries and anti-unitary symmetries 

410: of the partition function the temperature field has to satisfy the 

411: following conditions

412: \begin{itemize}

413: \item The temperature term must be real.

414: \item It may not break the flavor symmetry.

415: \item Its eigenvalues are given by $\pm i n T$. In this article, we will 

416: consider only $n=1$. 

417: \end{itemize}

418: A real temperature term satisfying these constraints is given 

419: by the matrix 

420: \be

421: \omega(T)&=& \left(\begin{array}{cc} 0 & T \\ -T & 0 \end{array}\right).

422: \ee 

423: For vanishing chemical potential, such a temperature term leads to a second 

424: order phase transition \cite{Jackson:1995nf}. We wish to stress that

425: the temperature term here is   

426: different from the one for QCD with three colors. 

427: Naively taking the same temperature dependence as for the three color case 

428: breaks the flavor symmetry and leads to a very 

429: different phase diagram \cite{Vanderheyden:2001gx}.

430: 

431: 

432: The matrix representation for the Dirac operator in the random matrix model 

433: with two flavors is

434: \begin{widetext}

435: \be

436: D= \left(\begin{array}{cccc} 

437: m_1 & 0 &  W+ \omega(T)+ \mu_1 & 0\\

438: 0 & m_2 & 0 &  W+ \omega(T)+ \mu_2 \\

439: -W^T -\omega(T)^T+\mu_1 & 0 & m_1 & 0\\

440: 0 &  -W^T -\omega(T)^T+\mu_2 & 0 & m_2

441:  \end{array} \right),

442: \ee

443: \end{widetext}

444: where $W$ is a real $n \times n$ matrix. The probability distribution of its 

445: matrix elements is given by

446: \be

447: P(W)&=&\exp\left[ -\frac{n}{2}G^2 \Tr W^T W\right].

448: \ee

449: When we write the determinant as an integral over fermionic variables, the 

450: random matrix partition function becomes

451: \be 

452: Z_{RMT} &=& \int {\cal D}W \prod_f d \psi^f d \bar{\psi}^f P(W) \exp[ -

453:  \bar{\psi} D \psi ].\nn\\

454: \label{Zrmt}

455: \ee

456: We will also include source terms for a pion condensate, 

457: \be

458: \label{pionsource}

459: \lambda \frac{1}{2}\bar{\psi}\gamma_5 i \tau_2 \psi,

460: \ee

461: and for a diquark condensate, 

462: \be

463: \label{diquarksource}

464: j \frac{1}{4}( \psi^T \tau_2 \psi +  \bar{\psi}^T \tau_2 \bar{\psi}),

465: \ee

466: where the anti-symmetric Pauli matrix $\tau_2$ acts in flavor space. 

467: 

468: For a fixed number of flavors ($N_f=2$), the partition function is a 

469: function of  $\mu_1, \mu_2, T$ and $m_1, m_2, \lambda, j$. While the

470: masses act as source terms for the chiral condensates, they are also

471: parameters of the model. 

472: 

473: \section{Effective Partition Function}

474: \label{nc2section-4}

475: \setcounter{equation}{0}

476: Due to the unitary invariance of the random matrix partition 

477: function, it is possible to rewrite it in terms of effective degrees of 

478: freedom.  We will do this below, and then analyze the resulting effective 

479: partition function, first to lowest order in chiral perturbation theory

480: and then beyond this domain. 

481: 

482: Starting from the random matrix partition function (\ref{Zrmt}), we perform 

483: the Gaussian integrations over the elements of the random matrix. To 

484: decouple the resulting four-fermion terms, we perform a Hubbard-

485: Stratonovich-transformation and introduce an integration over the mesonic 

486: low-energy degrees of freedom. The integrations over the fermionic variables 

487: can then be performed exactly. 

488: The resulting partition function in terms of the new 

489: variables is

490: \be

491: Z^{\rm eff}&=& \int {\cal D} A \exp[-{\cal L}(A, A^\dag)], 

492: \ee

493: where

494: \be

495: {\cal L} &=&\frac{n}{2}G^2 \Tr A A^\dag -\frac{n}{4}\log \det Q'. 

496: \label{Leff}

497: \ee

498: The dependence on the chemical potentials and the temperature is contained 

499: in the $8 N_f \times 8 N_f$ matrix $Q'$

500: \begin{widetext}

501: \be

502: Q'&=&

503: \left(\begin{array}{cccc} 

504: A^\dagger + M  & 0 &\mu_I I_3 + \mu_B B & -T \\

505: 0 & A^\dagger + M & T & \mu_I I_3 + \mu_B B \\

506: -(\mu_I I_3 + \mu_B B) & -T & A + M^\dagger & 0 \\

507: T & -(\mu_I I_3 + \mu_B B) & 0 & A+ M^\dagger \\

508: \end{array}\right).

509: \ee

510: \end{widetext}

511: The matrix $A$ is a complex, antisymmetric $2N_f \times 2N_f$ matrix. 

512: All the source terms are contained in the $2N_f \times 2N_f$ matrix $M$, 

513: which for $N_f=2$ is 

514: \be

515: \label{sourcematrix}

516: M= \left(\begin{array}{cccc}

517:   0 & m & -ij & \lambda \\

518: -m & 0 & \lambda & -ij\\

519: ij & -\lambda & 0 & m \\

520: -\lambda & ij & -m & 0 \end{array}\right ).

521: \ee

522: where we have  set $m_1=m_2=m$. ~From here on we will only consider quarks 

523: with equal mass.

524: The chemical potentials enter as the coefficients of the matrices for baryon 

525: and isospin charges, 

526: \be

527: B = \mathrm{diag}(1, -1, 1, -1),\;\;\;\; I_3 = \mathrm{diag}(1, -1, -1, 1). 

528: \ee   

529: This particular form of the charge matrices is due to the appearance of 

530: quarks and conjugate anti-quarks 

531: in the spinors transforming under the enlarged 

532: $SU(2N_f)$ symmetry, which have opposite charges with respect to the 

533: chemical potentials.

534:  

535:  

536: This effective Lagrangian can be simplified by a unitary transformation of the 

537: off-diagonal blocks, which leaves the value of the determinant unchanged. 

538: Exploiting the block structure, the effective Lagrangian (\ref{Leff})

539: can be expressed as

540: \be

541: {\cal L}&=&\frac{n}{2}G^2 \Tr A A^\dag -\frac{n}{4}\log \det Q_{+} Q_{-},

542: \label{Leff4}

543: \ee

544: where $Q_\pm$ are

545: the $ 4 N_f \times 4 N_f $ matrices

546: \be

547: Q_\pm\! = \!\! \left(\begin{array}{cc} 

548: A^\dag + M & \mu_B B + \mu_I I_3 \pm iT \\

549: -(\mu_B B + \mu_I I_3) \pm iT &  A + M^\dag

550: \end{array}\right)\!\! .                                     

551: \ee

552: The effective partition function with Lagrangian (\ref{Leff4}) 

553: is invariant under

554: \be

555: Q_\pm\to  U Q_\pm U^T, \quad \mu_2 \to -\mu_2, \quad \lambda \to j,

556: j \to -\lambda,

557: \label{symIB}

558: \ee

559: with 

560: \be

561: U =\left ( \begin{array}{cccc} 

562: 1 & & & \\ & i\sigma_1 & & \\ & & 1 & \\ & & & -i\sigma_1 \end{array}

563: \right ) .

564: \ee 

565: This transformation thus corresponds to an interchange of $\mu_B$ and

566: $\mu_I$ and of the diquark condensate and the pion condensate. 

567: 

568: 

569: We stress that this effective Lagrangian is an exact mapping of the

570: original random matrix partition function.

571: 

572: \subsection{Observables}

573: We wish to study the partition function and obtain the phase diagram with 

574: regards to four order parameters: the chiral condensates $\langle \bar u u 

575: \rangle$ and $\langle \bar d d \rangle$ for the two quark flavors, the pion 

576: condensate $\frac{1}{2}\left(\langle \bar u \gamma_5 d \rangle - \langle 

577: \bar d \gamma_5 u \rangle \right)$ and the diquark condensate $\frac{1}{4} i 

578: \left( \langle  d^\dagger u^*\rangle - \langle u^\dagger d^* \rangle + 

579: \langle d^T u \rangle -  \langle u^T d \rangle \right)$. We identify these 

580: observables by taking the derivatives of the partition function with respect 

581: to the sources. After partial integration the expressions for 

582: the chiral condensates for the two flavors are given by

583: \be

584: \langle \bar u u \rangle &=& \frac{1}{2n} \partial_{m_1} \log 

585: Z^{\mathrm{eff}}\el

586: &=& G^2 \frac{1}{4} \left\langle A_{21} -A_{12}+A_{21}^* - A_{12}^* 

587: \right\rangle  \el 

588: &\equiv& G^2 \sigma_1\el

589: \langle \bar d d \rangle &=& \frac{1}{2n} \partial_{m_2} \log 

590: Z^{\mathrm{eff}}\el

591: &=& G^2 \frac{1}{4} \left\langle A_{43} -A_{34}+A_{43}^* - A_{34}^* 

592: \right\rangle  \el

593: &\equiv& G^2 \sigma_2.

594: \ee

595: For the pion condensate we find in terms of expectation values for the 

596: elements of the matrix $A$

597: \be

598: \lefteqn{\frac{1}{2}(\langle \bar u \gamma_5 d \rangle - \langle \bar d 

599: \gamma_5 u \rangle )}\el

600: &=& \frac{1}{4n} \partial_{\lambda} \log Z^{\mathrm{eff}} 

601: \Big|_{\lambda=0}  \el

602: &=& G^2 \frac{1}{8} \langle A_{32} - A_{23} +A_{41} - A_{14}\el

603: & &   +A_{32}^* - A_{23}^* + A_{41}^* - A_{14}^* \rangle \el

604: &\equiv& G^2 \rho   

605: \ee

606: The diquark condensate is

607: \be

608: \lefteqn{\frac{1}{4} i ( \langle  d^\dagger u^*\rangle - \langle u^\dagger 

609: d^* \rangle + \langle d^T u \rangle -  \langle u^T d \rangle )}\el

610: &=&\frac{1}{4n} \partial_{j} \log Z^{\mathrm{eff}} \Big|_{j=0} \el

611: &=& G^2 \frac{1}{8} i \langle A_{13} - A_{31} + A_{24} - A_{42}\el 

612: & &    - A_{13}^* + A_{31}^* - A_{24}^* +A_{42}^* \rangle \el

613: &\equiv& G^2 \Delta.

614: \ee

615: In our convention, expectation values of the diagonal blocks 

616: correspond to the chiral condensates, and expectation values of 

617: the off-diagonal 

618: blocks correspond to the diquark and pion condensates. 

619: 

620: \subsection{Chiral Lagrangian}

621: The chiral Lagrangian

622:  is completely determined by the flavor symmetries of the 

623: microscopic theory and the pattern of the chiral symmetry breaking. Thus, it 

624: must be in agreement with the Lagrangian of chiral perturbation theory. For 

625: vanishing chemical potential and zero temperature, it has been shown 

626: that the random matrix partition function and the static part of the chiral 

627: Lagrangian in QCD are indeed equivalent 

628: \cite{Osborn:1998qb, Damgaard:1998xy, Toublan:1999hi}.

629: 

630: To show this equivalence in the present case, we will expand the

631: Lagrangian (\ref{Leff4}) 

632: in terms of the quark mass and the chemical potentials around the saddle 

633: point solution at $m=\mu_I=\mu_B=0$. In accordance with the usual power 

634: counting rules \cite{Kogut:2000ek}

635: we use $m={\cal O}(\epsilon^2)$ and $\mu_{I, B}={\cal 

636: O}(\epsilon)$ and expand to second order in $\epsilon$.

637: In contrast to the case $N_c=3$, the baryon chemical potential does not drop 

638: out of the effective Lagrangian. The baryon chemical potential couples to 

639: low energy degrees of freedom in form of diquark states, which carry a 

640: baryon charge. 

641: 

642: For vanishing mass and chemical potentials, the saddle point equation 

643: corresponding to (\ref{Leff4})

644: is given by

645: \be

646: \frac{1}{G^2} A &=& (A^\dagger A + T^2) A,

647: \ee

648: which has the nontrivial solution

649: \be

650: A&=&\frac{1}{G}\bar\sigma(T) \Sigma= \frac{1}{G}(1-G^2T^2)^{1/2} \Sigma,

651: \ee

652: where $\Sigma$ is any complex, antisymmetric, unitary matrix.

653: Expanding the effective Lagrangian (\ref{Leff4}) 

654: around this saddle point to order ${\cal O}(\epsilon^2)$, we find

655: \be

656: \lefteqn{{\cal L}(\Sigma) = \frac{n}{2} \left( c_0(\mu_B, \mu_I, T)\right.}\el

657:  & & - G\bar\sigma(T)\Tr(\Sigma M + \Sigma^\dagger M^\dagger) \el

658:  & & -G^2\bar\sigma^2(T)\Tr( (\Sigma^\dagger (\mu_I I_3 + \mu_B B) \Sigma 

659: (\mu_I I_3 + \mu_B B) )\el

660:  & & \left. +{ \cal O}(\epsilon^2)\right).

661: \ee

662: We note that the temperature does not break any flavor 

663: symmetries and has no effect 

664: on the matrix structure of $\Sigma$, as we expect for a correctly 

665: implemented temperature dependence. Its only effect is to change the overall 

666: magnitude of the condensates.

667: This result agrees with the static part of the chiral Lagrangian 

668: derived in \cite{Splittorff:2000mm}, which is (in our notation):

669: \be

670: {\cal L}_{\mathrm{stat}}(\Sigma) &=& -\frac{F^2 m_{\pi}^2}{2}\Tr(\Sigma M + 

671: \Sigma^\dagger M^\dagger) \el

672: & & \hspace{-3.5em} - \frac{F^2}{4} \Tr ( \Sigma^\dagger (\mu_I I_3 + \mu_B 

673: B) \Sigma (\mu_I I_3 + \mu_B B)). \nonumber \\   

674: \ee

675: 

676: This Lagrangian has been analyzed in detail in \cite{Splittorff:2000mm}. In the

677: remainder of this subsection we give 

678: a brief review of the main results of this paper. 

679: Let us first discuss the symmetries of the Lagrangian.

680: In the vacuum, the $SU(2N_f)$ symmetry is spontaneously broken

681: according to $SU(2N_f) \to Sp(2N_f)$, in the same way as it is broken

682: by the mass term. At nonzero baryon chemical potential,

683:  the symmetry is $SU(N_f) \times SU(N_f)$ and is  broken by the

684: diquark condensate to $Sp(N_f)\times Sp(N_f)$.

685: If in addition the mass is nonzero, the symmetry at $\mu \ne 0$

686: is $ U(1) \times SU(N_f)$ which is spontaneously broken by the

687: diquark condensate to $Sp(N_f)$ \cite{Kogut:2000ek}. 

688: 

689: The mass term and the terms for baryon and isospin chemical potential are 

690: individually minimized by $\Sigma$-matrices corresponding to a chiral 

691: condensate, a diquark condensate and a pion condensate, respectively. 

692: To minimize the Lagrangian, we use an ansatz consisting of a linear 

693: combination of these individual minima,

694: \be

695: \Sigma &=& \cos \alpha \Sigma_\sigma + \sin \alpha (\cos \eta \Sigma_\Delta 

696: + \sin \eta \Sigma_\rho),  

697: \ee

698: which is again antisymmetric and unitary.

699: These matrices, which correspond to the source terms  

700: introduced in (\ref{pionsource}), (\ref{diquarksource}), and

701: (\ref{sourcematrix}), 

702: are given by 

703: \be

704: \Sigma_\sigma=\left(\begin{array}{cccc} 

705: 0 & -1 & 0 & 0 \\

706: 1 & 0 & 0 & 0 \\

707: 0 & 0 & 0 & -1 \\

708: 0 & 0 & 1 & 0 

709: \end{array} \right), \el

710: \Sigma_\Delta=\left(\begin{array}{cccc} 

711: 0 & 0 & -i & 0 \\

712: 0 & 0 & 0 & -i \\ 

713: i & 0 & 0 & 0 \\

714: 0 & i & 0 & 0 

715: \end{array} \right), \el

716: \Sigma_\rho=\left(\begin{array}{cccc} 

717: 0 & 0 & 0 & -1 \\

718: 0 & 0 & -1 & 0 \\

719: 0 & 1 & 0 & 0 \\

720: 1 & 0 & 0 & 0 

721: \end{array} \right). 

722: \ee

723: Although the baryon chemical potential term is invariant under the

724: $U_B(1)$ rotations generated by $B$, this symmetry is spontaneously

725: broken by the diquark condensate, violating baryon number conservation

726: and giving rise to a massless Goldstone boson \cite{Kogut:2001if}. In

727: the same way, the pion condensate breaks the $I_3$ generated symmetry,

728: which also leads to a violation of isospin number conservation and a

729: Goldstone boson \cite{Son:2000by, Son:2000xc}.

730: 

731: \noindent

732: Using the ansatz for $\Sigma$, the effective Lagrangian (\ref{Leff4}) 

733: becomes

734: \be

735: {\cal L} &=& nN_f \left[ 1 + \log G^2 -2 m G \cos \alpha \right.\el

736:          & & + G^2 (\cos^2 \alpha (\mu_I^2 + \mu_B^2) -\sin^2 \alpha \cos (2 

737: \eta) (\mu_B^2 -\mu_I^2)) \el

738:          & & + \left.{\cal O}(\epsilon^2)\right]. 

739: \ee

740: When this is minimized with respect to $\eta$, for $\alpha \neq 0$ and 

741: $\mu_B^2 \neq \mu_I^2$, we find

742: \be

743: \eta= & 0 \;\;\;\mathrm{for} & \mu_B^2 > \mu_I^2\el

744: \eta= & \frac{\pi}{2} \;\;\;  \mathrm{for} & \mu_B^2 < \mu_I^2,

745: \ee

746: where the first case corresponds to the presence of a diquark condensate, 

747: and the second to that of a pion condensate. Thus, there is a first order 

748: transition between these two phases on the lines $\mu_B^2=\mu_I^2$. 

749: Minimizing now with respect to $\alpha$, we find 

750: \be

751: \cos \alpha   = & 1 & \mathrm{for} \;\;\;\mu_B^2 < \frac{m}{2G} 

752: \;\mathrm{and}\; \mu_I^2 < \frac{m}{2G} \el

753: \cos \alpha   = & \frac{m}{2G}\frac{1}{\mu_B^2} & \mathrm{for}\;\;\; \mu_B^2 

754: > \frac{m}{2G} \;\mathrm{and}\; \mu_B^2 > \mu_I^2 \el

755: \cos \alpha   = &\frac{m}{2G}\frac{1}{\mu_I^2} & \mathrm{for}\;\;\; \mu_I^2 

756: > \frac{m}{2G} \;\mathrm{and}\; \mu_B^2 < \mu_I^2.

757: \label{chptcond}

758: \ee

759: Since the critical chemical potential for the condensation of Goldstone 

760: bosons in our convention is half the pion mass, $\mu_{c} = 

761: m_\pi/2$, we can identify to this order in the expansion parameter

762: \be

763: m_\pi &=& \sqrt{\frac{2m}{G}}.

764: \ee

765: 

766: ~From the analysis of the chiral Lagrangian we thus have obtained the following

767: picture of the phase diagram at zero temperature: 

768: below a chemical potential of half the pion 

769: mass, the chiral condensate is constant. Above this critical chemical 

770: potential, depending on the relative magnitude of isospin and baryon 

771: chemical potential, pion or diquark condensation sets in and the chiral 

772: condensate drops quickly. The phases of pion and diquark condensation phases 

773: are separated by first order transition lines at $\mu_I^2=\mu_B^2$. 

774: 

775: 

776: \subsection{Effective Potential}

777: We now wish  to analyze the random matrix model to all orders in the mass and 

778: chemical potentials. To this end, we make an ansatz for the matrix $A$, and 

779: obtain an effective potential for the condensates in our model.

780: As in the case for three colors, the ansatz for $A$ is guided by two 

781: observations: ~From chiral perturbation theory, we know that above a critical 

782: chemical potential pion and diquark states will condense. We also expect 

783: that the presence of two flavor dependent chemical potentials, which 

784: explicitly breaks the flavor symmetry, will lead to a different behavior of 

785: the chiral condensates for the two flavors.  This leads to the

786: following ansatz  for $A$,

787: \be

788: A &=& \left( \begin{array}{cccc} 

789: 0 & -\sigma_1 & -i\Delta &-\rho\\

790: \sigma_1 & 0 & -\rho & -i\Delta \\

791: i\Delta & \rho & 0 & -\sigma_2 \\

792: \rho & i\Delta & \sigma_2 &0 

793: \label{ansatz}

794: \end{array} \right).

795: \ee

796: With this ansatz, evaluating the determinant we obtain the effective

797: potential for the condensates: 

798: \begin{widetext}

799: \begin{eqnarray}

800: \frac{1}{n}{\cal L} 

801:  &=& G^2 (\sigma_1^2 +\sigma_2^2 +2 \Delta^2 + 2 \rho^2) -\frac 12 

802: {\rm Tr} \log Q_+ Q_-,

803: \label{pot}

804: \end{eqnarray}

805: where

806: \begin{eqnarray}

807: \det Q_{\pm} 

808: &=& \left \{

809:   [(\sigma_1+m+\mu_1\pm iT)(\sigma_2+m-\mu_2\mp

810:   iT)+\rho^2+\Delta^2]\right . \el 

811:  &&\hspace{2pt} \times \left .

812:   [(\sigma_1+m -\mu_1\pm iT)(\sigma_2+m+\mu_2 \mp iT)+\rho^2+\Delta^2] 

813: + 4 \Delta^2 \mu_1\mu_2 \right\} \el

814: && \hspace{0pt} \left\{

815:   [(\sigma_1+m-\mu_1\mp iT)(\sigma_2+m+\mu_2 \pm

816:   iT)+\rho^2+\Delta^2]\right. \el 

817:  &&\hspace{2pt} \times \left.

818:   [(\sigma_1+m +\mu_1\mp iT)(\sigma_2+m -\mu_2\pm iT)+\rho^2+\Delta^2] 

819: + 4 \Delta^2 \mu_1\mu_2 \right\},

820: \end{eqnarray}

821: \end{widetext}

822: Although $\det Q_+$ and $\det Q_-$ are complex conjugate to each other by

823: construction, they turn out to be real themselves. 

824: With the ansatz (\ref{ansatz}) the symmetry (\ref{symIB}) corresponds

825: to the interchange (for $j = \lambda =0$)

826: \be

827: \Delta \to \rho, \quad \rho \to -\Delta, \quad \mu_2 \to -\mu_2.

828: \ee

829: The effective potential (\ref{pot}) is indeed invariant under this

830: symmetry, as well as the interchange of the flavor indices, and 

831: $\mu_f \rightarrow -\mu_f$ for both

832: $f=1,2$. These are the symmetries of the QCD partition function, as

833: shown in section~\ref{nc2section-2}. They allow us to restrict our study to one

834: half quadrant of the $\mu_B$-$\mu_I$--plane. The other sectors of

835: the phase diagram  are easily

836: obtained by applying these symmetries. 

837: The transition between the 

838: phases with diquark and pion condensation must necessarily be of first 

839: order and take place on the lines $\mu_I^2=\mu_B^2$.  

840: 

841:  

842: \section{Phase Diagram}

843: \label{nc2section-5}

844: \setcounter{equation}{0}

845: In this section, we will  determine the complete phase diagram of the random 

846: matrix model in the space of  the chemical potentials, $\mu_B$ and $\mu_I$, 

847: and the temperature $T$. 

848: To this end, we solve the saddle point equations 

849: \be

850: \frac{\partial {\cal L}}{\partial \sigma_1}=0, \;\;\; 

851: \frac{\partial {\cal L}}{\partial \sigma_2}=0, \;\;\; 

852: \frac{\partial {\cal L}}{\partial \rho}=0, \;\;\; 

853: \frac{\partial {\cal L}}{\partial \Delta}=0 \;\;\;

854: \ee

855: and determine the expectation values of all the condensates in the saddle 

856: point approximation. For vanishing temperature,  we will obtain analytical 

857: solutions in the chiral limit as well as for finite quark mass. For nonzero 

858: temperature, we will present analytical results only in the chiral limit 

859: and study the potential numerically for finite quark mass. 

860: 

861: 

862: \subsection{Chiral limit at zero temperature}

863: \setcounter{paragraph}{0}

864: First, we investigate the phase diagram in the $\mu_1$-$\mu_2$ chemical 

865: potential plane at temperature $T=0$. 

866: In the chiral limit, the Goldstone bosons arising from the spontaneous 

867: breaking of the chiral symmetry are exactly massless. 

868: Since all Goldstone bosons carry a charge that 

869: couples to one of the chemical potentials, this has a direct 

870: bearing on the phase diagram. At zero temperature, the chiral

871: condensate will be rotated completely into a bosonic condensate 

872: as soon as one of the chemical 

873: potentials becomes nonzero. 

874: 

875: \noindent

876: \paragraph{Chiral condensates and chiral restoration transition.}

877: In the phase where the bosonic condensates vanish, $\Delta=\rho=0$, 

878: the effective potential 

879: greatly simplifies and becomes a sum over the contributions of the chiral 

880: condensates for the two flavors,

881: \be

882: \frac1n {\cal L} &=& \sum_{f=1, 2} G^2 \sigma_f^2 -\frac{1}{2} \log

883: (\sigma_f^2-\mu_f^2)^2. 

884: \ee

885: Exactly as in the case of three color QCD, the saddle point equations

886: \be

887: \sigma_f [ G^2 (\sigma_f^2 - \mu_f^2) -1 ] =0, \;\;\; f = 1, 2, 

888: \ee

889: have the solutions

890: \be

891: \sigma_f  &=& 0, \; \; f=1,2, \\

892: \sigma_f^2&=& \frac{1}{G^2} +\mu_f^2,\;\; f=1, 2, 

893: \ee

894: which correspond to a phase with restored and broken chiral symmetry,

895: in this order. 

896: The contributions to the free energy for one flavor are

897: \be

898: \Omega_f(\mu_f) &=& - \log \mu_f^2 \\

899: \Omega_f(\mu_f) &=& 1 + \log G^2 +\mu_f^2 G^2,

900: \ee

901: for each of these phases, respectively.

902: The chiral symmetry restoration phase transitions take place at the 

903: chemical potential for which the two free energies coincide.

904: It is given by the solution of

905: \be

906: \label{limitmuccondition}

907: 1+ \mu_f^2 G^2 + \log \mu_f^2 G^2 = 0. 

908: \ee

909: We note that this condition depends on only one of the two chemical 

910: potentials and is strictly constant in the other.  

911: The behavior of the two flavors in these phases is completely independent. 

912: This changes only in the presence of either a pion or diquark condensate, 

913: which couples both flavors.

914: As a result, the phases with one nonzero chiral condensate extend along the 

915: chemical potential axes. Only in the center of the phase diagram, where 

916: these strips overlap, could a phase exist for which both chiral condensates 

917: are nonzero.

918: 

919: \noindent\paragraph{Phases with bosonic condensates.}

920: At the point $\mu_1=\mu_2=0$, 

921: the 

922: solution of the saddle point equation is degenerate. It satisfies $\sigma_1 

923: \sigma_2 + \Delta^ 2 + \rho^2 = 1/G^2$, without a preference for a specific 

924: direction in the space of the condensates.  

925: Due to the absence of flavor symmetry breaking terms we expect 

926: $\sigma_1=\sigma_2$. This is the only point 

927: where all condensates can be simultaneously nonzero. 

928: 

929: In the chiral limit, the chiral condensates vanish at nonzero chemical 

930: potential in the presence of a bosonic condensate. As we argued above, the 

931: phases with diquark and pion condensation must be separated by a first order 

932: transition. This is confirmed by the saddle point equation: the only 

933: solution for which $\Delta \neq 0 $ \emph{and} $\rho \neq 0$ requires that 

934: either $\mu_1 = 0$ or $\mu_2 = 0$, which is equivalent to $\mu_I^2 = 

935: \mu_B^2$ and thus coincides with the expected first order transition line. 

936: Therefore, we always have either a pion or a diquark condensate, and the 

937: effective potential in these two cases is reduced to

938: \be

939: \frac{1}{n} {\cal L} &=& G^2 2 \rho^2 - \log(\rho^2 - \mu_1\mu_2)^2,\\

940: \frac{1}{n} {\cal L} &=& G^2 2 \Delta^2 - \log(\Delta^2 + \mu_1\mu_2)^2. 

941: \ee

942: The corresponding saddle point equations become

943: \be

944: G^2 \rho ( \rho^2 - \mu_1 \mu_2) - \rho &=& 0,\\

945: G^2 \Delta (\Delta^2 + \mu_1\mu_2) - \Delta &=& 0,

946: \ee

947: and have the respective solutions

948: \be

949: \rho &= 0, \;\;\; \rho^2 &=\frac{1}{G^2} + \mu_1\mu_2 \;\;\\

950: \Delta&=0, \;\;\; \Delta^2&=\frac{1}{G^2} - \mu_1 \mu_2. 

951: \ee

952: Second order phase transition lines are given by the conditions

953: $\mu_1\mu_2= -1/G^2$ and $\mu_1\mu_2= 1/G^2$. 

954: The free energies for the two solutions are

955: \be

956: \Omega_\rho &=& 2 ( 1 + \log G^2 + \mu_1\mu_2 G^2), \\

957: \Omega_\Delta&=& 2( 1 + \log G^2 - \mu_1 \mu_2 G^2).  

958: \ee

959: By comparing the free energies, we observe that in the quadrants with 

960: $\mu_1\mu_2 > 0$, the diquark condensate is nonzero,  

961: and that for $\mu_1\mu_2 < 

962: 0$ the solution with a pion condensate has the lower free energy. Equating 

963: the free energies, we confirm again that first order transition lines are 

964: given by $\mu_1\mu_2=0$. 

965: 

966: \noindent\paragraph{Phase diagram.}

967: To find the complete phase diagram, we use the results we obtained directly 

968: for the second order phase transitions, and we compare the free energies of 

969: all phases to determine the first order transitions. We conclude that, 

970: for massless quarks, 

971: phases with condensed bosons are always favored above a phase where both

972: chiral condensates are non-vanishing. 

973: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

974: \begin{figure}

975: \hspace*{-0.4cm}

976: \includegraphics[scale=0.60, clip=true, angle=0, draft=false]{fig1.eps}

977: \caption{\label{fig1} Phase diagram in the $\mu_1$-$\mu_2$-plane at $T=0$ 

978: and vanishing quark mass $mG=0$. First (second) order transitions are 

979: indicated by solid (dashed) lines. Phases are labeled by the nonvanishing 

980: condensates. All condensates vanish in the corner regions, where chiral 

981: symmetry is restored.}

982: \end{figure}

983: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

984: In addition to the first order transitions between the pion and diquark 

985: condensate phases, we find first order transitions from these to the phases 

986: where only one of the chiral condensates is different from zero. 

987: 

988: The phase diagram is shown in Fig.~\ref{fig1}.

989: At the origin, the solution to the saddle point equation is degenerate, and 

990: only the magnitude of the condensates is fixed. ~From the origin, a pion 

991: condensation phase extends into the quadrants where $\mu_1\mu_2 < 0$ 

992: ($\mu_I^2 > \mu_B^2$), and a diquark condensation phase extends 

993: into the quadrants 

994: where $\mu_1 \mu_2 > 0$ ($\mu_I^2 < \mu_B^2$). The domain of these phases is 

995: limited by a line of second order transitions (see dashed curve in 

996: Fig.~\ref{fig1}), which 

997: in each quadrant is given by the solutions of 

998: \be

999: 1 \pm G^2 \mu_1\mu_2 =0.  

1000: \ee

1001: These transitions can be interpreted as saturation

1002: transitions which also occur in lattice QCD because of

1003: the finite number of available fermionic states \cite{Kogut:2001na}.

1004: Our saturation transition is somewhat different. Because we are working

1005: in the thermodynamic limit, the number of available states is infinite

1006: as well. However, since the level spacing of the relevant matrices

1007: $\sim 1/n$, the support of the spectrum remains compact. It is the

1008: non-analyticity that occurs because of the extreme edges of the spectrum

1009: that gives rise to the saturation transition.

1010: The phases with broken chiral symmetry and only {\it one} of the chiral 

1011: condensate nonzero appear along both chemical potential axes. They are 

1012: separated from the Goldstone condensation phases by first order transition 

1013: lines.   

1014: 

1015: The phase diagram exhibits a fourfold symmetry, along both the

1016: $\mu_B$- and the $\mu_I$- axis. This is in contrast to the twofold

1017: symmetry with respect to the $\mu_B$-axis that we found in the phase

1018: diagram of three color QCD, where only a pion condensate appears

1019: \cite{Klein:2003fy}. Fundamentally, this is due to the flavor

1020: symmetry in the two color case which is enlarged beyond the symmetry in

1021: the three color case. 

1022: However, the upper-left quadrant in the $\mu_B$-$\mu_I$--plane is

1023: strictly identical in our random matrix model for both two and three

1024: colors \cite{Klein:2003fy}.

1025: The important point is that both

1026: chemical potentials couple to Goldstone bosons of the theory.

1027: Therefore, the two chemical potential terms in the present case are 

1028: on equal footing. They lead to two different possible symmetry breaking 

1029: patterns. For three colors, there is only one chemical potential term, since 

1030: only the isospin chemical potential couples to the Goldstone modes, and thus 

1031: there is only one way to break the flavor symmetry. 

1032: 

1033: \subsection{Nonzero quark mass at $T=0$}

1034: \setcounter{paragraph}{0}

1035: 

1036: The phase diagram  in the presence of a finite quark differs from the one  

1037: obtained in the chiral limit. 

1038: As in the case of three colors, this is 

1039: due to two effects: 

1040: First, the chiral condensate becomes nonzero 

1041: because of the explicit breaking of  chiral symmetry. Because of this, phase

1042: transitions may turn into crossover transition.

1043: Second, previously exactly massless 

1044: Goldstone bosons become massive. This primarily affects 

1045: the low energy regions of the phase diagram. In addition, the Goldstone 

1046: condensation phases become qualitatively different and are now mixed phases 

1047: with nonzero chiral condensates. In spite of this, at zero temperature we 

1048: can still solve the saddle point equations exactly, and we will give the 

1049: results below. 

1050: 

1051: \noindent

1052: \paragraph{Chiral condensation phases.}

1053: For a small quark mass, the behavior of the chiral condensate is still 

1054: dominated by the spontaneous breaking of the chiral symmetry. 

1055: Thus, a natural way to treat the chiral condensates in this case is to 

1056: expand in the quark mass about the solutions for $m=0$.  

1057:   

1058: The effective potential that describes the phases without bosonic

1059: condensates is given by

1060: \be

1061: \frac{1}{n}{\cal L} &=& \sum_{f=1, 2} G^2 \sigma_f^2 - \frac{1}{2} \log 

1062: [(\sigma_f+m)^2 - \mu_f^2]^2. \nonumber \\

1063: \ee

1064: For each flavor we have to solve the saddle point equation

1065: \be

1066: G^2 \sigma_f [(\sigma_f+m)^2 - \mu_f^2] - (\sigma_f +m )=0.

1067: \ee

1068: As expected, the solution $\sigma_f=0$ no longer exists, and we find by 

1069: expanding in $m$ (without assumptions about the magnitude of $\mu_f$) the 

1070: new solution

1071: \be

1072: \sigma_f &=& - \frac{1}{1+\mu_f^2 G^2} m + {\cal O}(m^3).

1073: \ee

1074: Expanding in $mG$ in the phase with spontaneously broken chiral symmetry 

1075: around the second solution with $G^2 \sigma_f = {\cal O}(G)$, we find the 

1076: solutions

1077: \be

1078: G\sigma_f &=&\pm \sqrt{1+\mu_f^2G^2} 

1079: +\Big(\frac{1}{2(1+\mu_f^2G^2)} -1\Big) mG \el

1080: && +{\cal O}(m^2G^2).

1081: \ee 

1082: The respective free energies for one flavor of these two phases with 

1083: spontaneously broken and restored chiral symmetry are 

1084: to lowest order in $m$ 

1085: given by

1086: \be

1087: \Omega_f&=& 1 + \mu_f^2G^2 + \log G^2 \mp 2 mG \sqrt{1+\mu_f^2 G^2}, \\

1088: \Omega_f&=& - \log \mu_f^2. 

1089: \ee 

1090: If these two phases are not separated by a Bose-con\-den\-sa\-tion phase and  

1091: meet directly, the chiral restoration transition is of first order. The 

1092: chemical potential at the transition is obtained by matching the free 

1093: energies of the two phases. We find

1094: \be

1095: \mu^\prime_{f, c} G &=& \mu_{c}G\left(1 + \frac{mG}{1+\mu_c^2 G^2} + {\cal 

1096: O} (m^2G^2) \right), 

1097: \label{crshift}

1098: \ee

1099: where $\mu_{c}G$ is the critical chemical potential in the chiral

1100: limit 

1101: determined  from (\ref{limitmuccondition}). 

1102:  The shift in the chiral restoration transition is proportional to the 

1103: quark mass $m$.

1104: 

1105: \noindent

1106: \paragraph{Goldstone boson condensation phases.}

1107: For finite quark mass, the critical chemical potential for condensation of 

1108: Goldstone bosons becomes nonzero due to their finite mass. The symmetry with 

1109: regards to the chemical potentials  $\mu_I$ and $\mu_B$ remains unaffected. 

1110: In particular, above the critical chemical potential, we still expect to see 

1111: either a phase with diquark condensation or one with pion 

1112: condensation depending on the relative magnitude of these two potentials, 

1113: and a first order transition separating them. 

1114: Since the results for both Goldstone condensation phases are related by this 

1115: symmetry and are thus very similar, we will give explicit results for

1116: one half quadrant only. 

1117: 

1118: In both phases, due to the finite quark mass, the chiral condensates do not 

1119: vanish. Therefore, we have to solve three saddle point equations 

1120: simultaneously, for the chiral condensates of both flavors and for either 

1121: $\rho$ or $\Delta$. 

1122: In the region in which $|\mu_I| > |\mu_B|$, we thus have $\Delta =0$.

1123: The solutions of the saddle point equations are given by:

1124: \be

1125: \sigma_1 &=& -m + \frac{m}{2}\frac{1}{G^2}\frac{1}{\mu_I^2 -m^2} 

1126: + \frac12 m  (\mu_B + \mu_I) \frac{\mu_B}{\mu_I^2} \el

1127: \sigma_2 &=& -m + \frac{m}{2}\frac{1}{G^2}\frac{1}{\mu_I^2 -m^2} 

1128: + \frac12 m (\mu_B -\mu_I) \frac{ \mu_B}{\mu_I^2} \el

1129: \rho^2 + \sigma_1\sigma_2 &=& \frac{1}{G^2} +\mu_B^2-\mu_I^2 +m^2 -m^2 

1130: \frac{\mu_B^2}{\mu_I^2}. 

1131: \ee  

1132: ~From the relation between the chiral and the pion condensate, we find

1133: the explicit result for $\rho$,

1134: \be

1135: \rho^2 &=& \mu_B^2 -\mu_I^2 + \mu_I^2 \frac{1}{G^2}\frac{1}{\mu_I^2 -m^2 } - 

1136: \frac{m^2 }{4}\frac{1}{G^4}\frac{1}{(\mu_I^2 -m^2)^2} \el

1137:         & & -m^2 \frac{\mu_B^2}{\mu_I^2}\left( \frac{1}{4 \mu_I^2}(\mu_B^2-

1138: \mu_I^2) + \frac{1}{2}\frac{1}{G^2}\frac{1}{\mu_I^2 - m^2} \right).

1139: \nonumber \\

1140: \label{pc}   

1141: \ee

1142: The difference between the two chiral condensates depends on the ratio of 

1143: $\mu_B$ and $\mu_I$: 

1144: \be

1145: \sigma_1-\sigma_2 &=& m \frac{\mu_B}{\mu_I}.

1146: \ee

1147: The free energy of this phase is given by

1148: \be

1149: \Omega_{\rho}(m, \mu_B, \mu_I) &=& 2 \left( 1 + \log G^2 + (\mu_B^2-

1150:                 \mu_I^2)G^2 +m^2 G^2 \right) \el

1151: & & - m^2 G^2 \frac{ \mu_B^2}{\mu_I^2} 

1152:        -\log \left(\frac{\mu_I^2}{\mu_I^2 -m^2 } \right). \nonumber \\

1153: \ee

1154: 

1155: 

1156: The condition $\rho^2=0$ 

1157: describes a line of second order phase 

1158: transitions at which the bosonic condensate vanishes. For finite quark mass, 

1159: this condition has more than one solution. For $\mu_B=0$, the two solutions 

1160: for $\mu_I^2$ are 

1161: \be

1162: \mu_{I\,c}^{2} G^2 &=& \frac{m G}{2} + \frac{5}{8} m^2 G^2 +{\cal 

1163: O}(m^3 G^3)\\

1164: \mu_{I\,c}^{2} G^2 &=& 1 + \frac{3}{4} m^2 G^2 + {\cal O}(m^4 

1165: G^4). 

1166: \ee 

1167: We note that the solution for which $\mu_{I\, c}^2 G^2 = mG/2 + 

1168: {\cal O}(m^2G^2)$ agrees to leading order with the result 

1169: which was  

1170: obtained from  

1171: chiral perturbation theory \cite{Kogut:1999iv, Kogut:2000ek}. 

1172: For chemical potentials below the critical value of $\mu_B$,

1173: at which diquark condensation occurs, the pion mass is expected to be

1174: independent from $\mu_B$ \cite{Splittorff:2000mm}. Thus, the critical value 

1175: for $\mu_I$ should be exactly constant as

1176: a function of $\mu_B$. In our random matrix model we find that

1177: $\mu_{I\,c}$ is 

1178: only

1179: approximately constant as a function of $\mu_B$.

1180: This is due to the fact that, in random

1181: matrix models, the chemical potential affects the condensates already

1182: below the threshold, even at zero temperature \cite{Stephanov:1996ki,

1183:   Halasz:1998qr, Vanderheyden:2001gx}. 

1184: In \cite{Halasz:1998qr, Vanderheyden:2001gx} it was 

1185: thus argued that the random matrix

1186: result should be interpreted as the critical contribution to the free

1187: energy, on top of a noncritical part not contained in the model. 

1188: 

1189: For small  baryon chemical potential, small isospin 

1190: chemical potential and quark mass,  

1191: the chiral condensates in the pion condensation phase become

1192: \be

1193: \sigma_1=\sigma_2&=&\sigma=\frac{m}{2G^2 \mu_I^2} + {\cal O}(m)\\

1194: \sigma^2 + \rho^2 &=& \frac{1}{G^2} +{\cal O}(m),

1195: \ee

1196: which agrees with the results (\ref{chptcond})

1197: obtained in chiral perturbation theory. 

1198: In this region of the phase diagram the effects of a finite quark mass are 

1199: most significant. For large values of the chemical 

1200: potentials both the pion and diquark condensates vanish 

1201: for any value

1202: of the quark mass. 

1203: The second order transition to this phase is to leading order

1204: independent of the quark mass.  

1205: 

1206: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1207: \begin{figure}

1208: \hspace*{-0.4cm}\includegraphics[scale=0.60, clip=true, angle=0, 

1209: draft=false]{fig2.eps}

1210: \caption{\label{fig2} Phase diagram in the $\mu_1$-$\mu_2$-plane at $T=0$ 

1211: for a finite quark mass  $mG=0.1$. First (second) order transitions are 

1212: indicated by solid (dashed) lines. The different phases are labeled by the 

1213: non-vanishing condensates or the condensates which have large

1214: expectation values. The finite quark mass leads to a phase with chiral 

1215: condensation only in the center of the phase diagram. In the corner regions, 

1216: both the diquark condensate and the pion condensate vanish, 

1217: and the chiral condensates are of 

1218: ${\cal O}(m)$. }

1219: \end{figure}

1220: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1221: 

1222: %{$\bullet$ \bf Phase diagram}

1223: \paragraph{Phase diagram.}

1224: 

1225: As in the chiral limit, in order 

1226: to obtain the complete phase diagram we combine the 

1227: results for the second order transitions with an analysis of the free 

1228: energies of the different phases. The resulting phase diagram in the 

1229: $\mu_1$-$\mu_2$--plane is shown in Fig.~\ref{fig2} for a quark mass of 

1230: $mG=0.1$. The four-fold symmetry remains the same as in the case of $mG=0$. In 

1231: particular, the transitions between the phases with the different Goldstone 

1232: boson condensates are still of first order. The main difference from the 

1233: chiral limit is the appearance of a phase in the center of the phase diagram 

1234: in which the chiral condensates for both flavors are nonzero. This is a 

1235: consequence of the finite critical chemical potential for the

1236: condensation of bosons with a finite mass.  

1237: 

1238: A second effect of the finite quark masses is that 

1239: the chiral restoration transitions 

1240: outside the regions of Bose condensation are shifted slightly from their 

1241: positions in the chiral limit. As can be seen 

1242: from (\ref{crshift}), the  correction is of order $mG$. 

1243: Thus, while changing the nature of the phase

1244: transition considerably in the regions of low chemical potential 

1245: in the center of the phase diagram, a finite quark mass has only a

1246: small effect on the phase diagram at large chemical potential.     

1247: 

1248: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1249: \begin{figure}

1250: \hspace*{-0.4cm}\includegraphics[scale=0.60, clip=true, angle=0, 

1251: draft=false]{fig3.eps}

1252: \caption{\label{fig3} Phase diagram in the $\mu_1$-$\mu_2$-plane at $T=0$ 

1253: for a finite quark mass  $mG=0.2$. First (second) order transitions are 

1254: indicated by solid (dashed) lines. The different phases are labeled by the 

1255: condensates which are non-vanishing or have a large expectation value. 

1256: As described in the text, for a quark mass larger than $mG  \sim

1257: 0.184$ the pion and diquark condensation phases no longer

1258: have a boundary in common. In contrast to the situation shown in

1259: Fig.~\ref{fig2}, 

1260: the chiral first order transition lines therefore appear also if one

1261: of the chemical potentials $\mu_f$ is small.}

1262: \end{figure}

1263: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1264: For leading order in the mass expansion, we find that the first

1265: order transition between the phases with $\rho\neq0$ and $\Delta\neq0$

1266: disappears at the point $1+\log2mG=0$. Therefore, for masses larger than

1267: $mG\sim0.184$, the pion and diquark condensation phases are not

1268: contiguous anymore. 

1269: In this case, if either $\mu_1$ or $\mu_2$ is kept small and the

1270: chemical potential for the other quark flavor is increased, one

1271: encounters the chiral restoration  

1272: transition without passing through either the pion or the diquark

1273: condensation phase.

1274: 

1275: 

1276: \subsection{Nonzero $T$ at $m=0$}

1277: \subsubsection{The $\mu_I=0$-- and $\mu_B=0$-planes}

1278: \setcounter{paragraph}{0}

1279: Since the presence of a temperature term does not change the 

1280: symmetries, even at finite temperature 

1281: the two phases with the different Goldstone condensates are 

1282: still related by an interchange of the two chemical 

1283: potentials. We can again treat both the pion and diquark condensation phases 

1284: in parallel. In either the $\mu_I=0$ or the $\mu_B=0$ plane, the

1285: presence of only one of these chemical potentials restores the symmetry between

1286: the two quark flavors. Consequently, there is only one chiral condensate

1287: $\sigma=\sigma_1=\sigma_2$. 

1288: The results for the chiral phase transition

1289: were first obtained for $\mu_I=0$ in \cite{Halasz:1998qr}.

1290: The results in the plane $\mu_B=0$

1291: coincide with  those for $N_c=3$ at nonzero 

1292: isospin chemical potential \cite{Klein:2003fy} 

1293: where the same effective potential

1294: was found. Therefore, and mainly for completeness, we briefly

1295: review the main results.

1296: The result for the diquark condensate can be

1297: obtained by the simple substitution $\mu_I \to \mu_B$ and $\rho \to

1298: -\Delta$. 

1299: 

1300: In the chiral limit the critical chemical potential for pion condensation 

1301: of this model 

1302: is independent of the temperature. For any nonzero

1303: value $|\mu_I| >0$, the system will be immediately in the pion

1304: condensation phase (via a first order transition from the degenerate

1305: point $\mu_I=0$). The effective potential for this phase is

1306: \be

1307: \frac{1}{n} {\cal L}&=& 2 G^2 \rho^2 -\log[ (\rho^2+\mu_I^2 +T^2)^2] ,

1308: \ee

1309: and the saddle point equations are solved by 

1310: \be

1311: \rho^2&=& \frac{1}{G^2} -\mu_I^2 -T^2, \\

1312: \rho&=&0.

1313: \ee

1314: The phase diagram is now

1315: simple: 

1316: A second order transition

1317: line is  given by  the circle $G^2 (\mu_I^2+T^2)=1$. The free

1318: energy of this phase with $\sigma_1=\sigma_2 = 0$, given by

1319: \be

1320: \Omega_{\rho}&=& 2[ 1+\log G^2 - G^2(\mu_I^2+T^2)],

1321: \ee

1322: is always lower than that of the chiral broken phase, and thus in the

1323: chiral limit, the chiral condensates do not appear in this 

1324: plane 

1325: (see the upper panel of Fig.~\ref{fig4}). Therefore, the 

1326: tricritical point of the chiral restoration transition found in the

1327: $\mu_B$-$T$--plane 

1328: at $\mu_I=0$ 

1329: for $N_c=3$ \cite{Halasz:1998qr} does not appear

1330: at all in the $\mu_I$-$T$--plane

1331: given by $\mu_B=0$. 

1332: Using the symmetry of the phase diagram,

1333: we see that, in contrast to the case $N_c=3$, for $N_c=2$

1334: this tricritical point

1335: is not present in the $\mu_B$-$T$--plane 

1336: with $\mu_I=0$, 

1337: either.  

1338: 

1339: \subsubsection{The complete $\mu_I$-$\mu_B$--plane at $T \neq 0$} 

1340: \setcounter{paragraph}{0}

1341: In the chiral limit, 

1342: we can still solve

1343: the saddle point equations analytically. 

1344: Since the first order phase transition

1345: between the pion and diquark condensation phases is not affected by the

1346: temperature, we continue to have phases of pure diquark and pure pion

1347: condensation which meet at the first order transition 

1348: lines $\mu_1=0$ or $\mu_2=0$

1349: (i.e. $\mu_I=\pm \mu_B$).

1350: There are no phases in which a pion

1351: condensate or a diquark condensate coexists with a chiral condensate

1352: of any flavor. 

1353: 

1354: 

1355: \noindent\paragraph{Chiral condensation phases.}

1356: The phase with only chiral

1357: condensation occurs along strips parallel to the $\mu_1 =0$ or $\mu_2 =0$

1358: axis (see Fig.~\ref{fig1}). In these phases only one of the chiral 

1359: condensates is non-vanishing. 

1360: The effective potential simplifies to a sum over 

1361: two independent contributions of both flavors, 

1362: \be

1363: \lefteqn{\frac{1}{n} {\mathcal L} = \sum_{f=1, 2} \big\{ G^2 \sigma^2_f} \el

1364: && -\frac{1}{2}

1365: \log\big[ (\sigma_f^2-\mu_f^2 + T^2)^2 + 4 \mu_f^2 T^2\big]\big\}, \el  

1366: \ee

1367: and coincides with the effective potential found for QCD with three

1368: colors.

1369: For each of the two flavors, the chiral condensate depends only

1370: on the temperature and the chemical potential for that flavor. 

1371: In the chiral limit, the magnitude of the chiral condensate is 

1372: given by the solution of the saddle point equation

1373: \be

1374: \lefteqn{G^2 \sigma_f \Bigg[ \sigma_f^4-2 \left(\frac{1}{2G^2} +\mu_f^2 -T^2

1375:     \right) \sigma_f^2} \el

1376: && + \frac{1}{G^2}(\mu_f^2 -T^2)+ (\mu_f^2 +T^2)^2\Bigg]= 0. 

1377: \label{chspT}

1378: \ee

1379: Its  solutions are given by \cite{Halasz:1998qr}

1380: \be 

1381: \sigma_f &=& 0 \\

1382: \sigma_f^2 &=& \frac{1}{2G^2} +\mu_f^2 -T^2 \pm \frac{1}{2G^2}

1383: \sqrt{1-(4G^2\mu_f T)^2}, \el

1384: \ee

1385: and due to the solution $\sigma_f=0$ a first order

1386: transition to the phase with restored chiral symmetry is possible. 

1387: For any one flavor $f=1, 2$, this first order transition takes place

1388: where the free energies of the two solutions, given by

1389: \be

1390: \Omega_f &=& -\log (\mu_f^2 +T^2) \\

1391: \Omega_f &=& \frac{1}{2} + \log G^2 + G^2 (\mu_f^2 -T^2) \pm

1392: \frac{1}{2} \sqrt{1-(4G^2 \mu_f T)^2} \el

1393:  & & -\frac{1}{2} \log \left (\frac{1}{2}

1394: \pm \frac{1}{2} \sqrt{1-(4G^2 \mu_f T)^2} \right),  

1395: \ee

1396: respectively, become equal.

1397: The solution $\sigma_f \ne 0$ with the negative square root is not a global

1398: minimum of the free energy, and is in fact never realized. We find

1399: that the solution $\sigma_f \ne 0$ approaches the solution $\sigma_f = 0$

1400: at the  second order phase transition

1401: line in the $\mu_f$-$T$--plane given by 

1402: \be

1403: (\mu_f^2 -T^2) + G^2 (\mu_f^2+T^2)^2 =0. 

1404: \ee

1405: The tricritical point, where the phase transition line changes from

1406: first to second order, appears where the coefficients of $\sigma_f$

1407: and $\sigma_f^3$ in (\ref{chspT}) vanish. We find \cite{Halasz:1998qr}

1408: \be 

1409: \mu_{f,3}^2 G^2 &=& \frac{\sqrt{2}-1}{4},\quad f=1, 2 \el

1410: T_3^2 G^2 &=& \frac{\sqrt{2}+1}{4}.

1411: \label{nc2tricrit}

1412: \ee

1413: Since these results are independent of the chemical potential

1414: for the other flavor, all transition lines are straight lines in the

1415: $\mu_1$-$\mu_2$--plane at fixed temperature. In fact, the transition

1416: lines for one flavor in this plane have to be perpendicular to the

1417: corresponding lines of the other flavor. 

1418: 

1419: \noindent\paragraph{Goldstone boson condensation phases.} 

1420: Once again relying on the symmetry of the 

1421: effective potential (\ref{pot}), it is sufficient to analyze the 

1422: pion condensate. Results for the diquark condensate

1423: simply follow from  interchanging $\mu_I$ and $\mu_B$.  

1424: The effective potential in the pion condensation phase at finite

1425: temperature is given by  

1426: \be

1427: \lefteqn{\frac{1}{n}{\cal L} = 2 G^2 \rho^2 }\el

1428: &&-\log[(\rho^2-\mu_B^2 +\mu_I^2+T^2)^2 + 4 \mu_B^2 T^2].

1429: \ee

1430: To explore the structure of the boundaries of the domain of this phase, 

1431: we analyze the saddle point equation

1432: \be

1433: \lefteqn{G^2\rho \Big[ \rho^4 - 2\left(\frac{1}{2G^2} +\mu_B^2 -\mu_I^2 -T^2

1434:   \right) \rho^2} \el

1435: &&+(\mu_B^2-\mu_I^2 -T^2)^2 + 4 \mu_B^2 T^2

1436:   +\frac{1}{G^2}(\mu_B^2-\mu_I^2-T^2)\Big] =0.\el

1437: \label{nc2sprhot}

1438: \ee  

1439: Besides the trivial solution $\rho=0$, the saddle point equation has

1440: the solutions

1441: \be

1442: \rho^2&=& \frac{1}{2G^2} +\mu_B^2 -\mu_I^2 -T^2 \pm \frac{1}{2G^2}

1443: \sqrt{1-(4G^2\mu_B T)^2},\el  

1444: \ee

1445: of which only the one with the positive sign in front of the

1446: square root is a minimum of the free energy. 

1447: A second order transition line is given by the condition that this solution

1448: coincides with the trivial solution,

1449: \be

1450: \lefteqn{(\mu_B^2-\mu_I^2-T^2)^2 + 4 \mu_B^2 T^2} \el

1451: &&+ \frac{1}{G^2} (\mu_B^2-\mu_I^2-T^2)=0.

1452: \label{nc2rhosectran}

1453: \ee

1454: There is always the

1455: possibility of a first order transition to a phase with $\rho=0$.

1456: Indeed,  for a certain range of

1457: parameters, where the coefficients of

1458: both $\rho^3$ and $\rho$ in eq. (\ref{nc2sprhot}) vanish, there exists a

1459: tricritical point at which the transition changes order. But when the

1460: complete phase diagram is considered, including all other condensates,

1461: the solutions that lead to this tricritical point do not appear: 

1462: at this point the free energy

1463: of one of the chiral condensation phases is lower than that of the

1464: pion condensation phase, and a first order transition 

1465: takes place before the point is reached.

1466: The free energy in the pion condensation phase is given by

1467: \be

1468: \Omega_{\rho}&=& 2\bigg\{ \frac{1}{2} + \log G^2 + G^2(\mu_B^2 -

1469:   \mu_I^2-T^2)\el

1470: &&  +\frac{1}{2} \sqrt{1-(4 G^2 \mu_B T)^2} \el

1471: &&-\frac{1}{2}\log\bigg[\frac{1}{2}  +\frac{1}{2} \sqrt{1-(4 G^2 \mu_B

1472:       T)^2} \bigg] \bigg\}, \el

1473: \ee

1474: and we determine the complete phase diagram by comparing the free

1475: energies of the different phases. Using the symmetry (\ref{symIB}),

1476: all the above results also apply to  the diquark

1477: condensate.

1478: 

1479: 

1480: \noindent\paragraph{Phase diagram.}

1481: At finite temperature, the phase diagram in the $\mu_1$-$\mu_2$ plane

1482: still has the fourfold symmetry as seen at $T=0$ in 

1483: Fig.~\ref{fig1}. 

1484: As can be seen from the second

1485: order transition condition (\ref{nc2rhosectran}), the boundaries of

1486: the regions with diquark condensation or pion condensation 

1487: move gradually towards the origin  $\mu_B=\mu_I=0$ with increasing

1488: temperature, and shrink rapidly to a point when $TG$ approaches $TG=1$. 

1489: The regions with a nonzero chiral condensate also shrink gradually

1490: with increasing temperature. When the temperature reaches the

1491: tricritical point  given by

1492: eq. (\ref{nc2tricrit}), the transition to the chirally restored region

1493: becomes second order. As the temperature approaches $TG=1$ this region

1494: rapidly collapses onto the axes $\mu_1 =0$ and $\mu_2=0$. 

1495: All condensates vanish and the symmetries are completely restored for

1496: $TG \ge 1$. 

1497: 

1498: \subsection{Finite quark mass at nonzero $T$}

1499: 

1500: In general, for $m\neq 0$ the  saddle point equations cannot be solved

1501: analytically when $\mu_I$, $\mu_B$ and $T$ are all different from zero. 

1502: However, as outlined in the previous section, the symmetries of the

1503: partition function under an interchange of the chemical potentials for

1504: the two flavors, or a change in their sign, are unaffected by the

1505: presence of the temperature term. Thus, the four symmetry axes $\mu_I=0$,

1506: $\mu_B=0$ and $\mu_I=\pm

1507: \mu_B$ remain at finite 

1508: temperature, and the first order transitions between the condensates

1509: $\Delta$ and $\rho$ at  $\mu_I=\pm \mu_B$ remain. 

1510: The structure of the phase diagram in this case can be deduced from

1511: the analytical results at $T=0$ and from the planes where one of the

1512: chemical potentials vanishes.

1513: 

1514: \subsubsection{The $\mu_B=0$-- plane}

1515: \setcounter{paragraph}{0}

1516: 

1517: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1518: \begin{figure}[h!]

1519: \hspace*{-0.9cm}\includegraphics[scale=0.85, clip=true, angle=0,

1520:   draft=false]{fig4.eps}

1521: \caption{\label{fig4} Phase diagram in $\mu_I$ and $T$ in the

1522:   $\mu_B=0$-plane, both for the chiral limit (top)  

1523: and a finite quark mass $mG=0.1$ (bottom). First (second) order

1524:   transitions are  

1525: indicated by solid (dashed) lines, the dotted lines in the case of

1526:   finite quark mass mark the crossover.} 

1527: \end{figure}

1528: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1529: 

1530: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1531: \begin{figure*}[h!]

1532: \hspace*{-0.9cm}\includegraphics[scale=0.94, clip=true, angle=0,

1533:   draft=false]{fig5.eps} 

1534: \caption{\label{fig5} Phase diagram in the $\mu_1$-$\mu_2$--plane at

1535:   fixed values of $TG$ 

1536: and for a finite quark mass of $mG=0.1$. Values for the temperature $TG$ are

1537:   given in the panels. First (second) order transitions are 

1538: indicated by solid (dashed) lines, the dotted lines indicate a

1539:   crossover. Phases are labeled by the non-vanishing 

1540: condensates, resp. the condensates with large expectation values. In

1541:   the corner regions, the pion and diquark condensates vanish, and

1542:   both chiral condensates become small.} 

1543: \end{figure*}

1544: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1545: 

1546: Although an analytical solution of the saddle point equations is not

1547: possible for both chemical potentials and the temperature nonzero, in

1548: order to have a complete picture of the phase diagram it is

1549: instructive to consider the plane $\mu_B=0$. Here, of course, the

1550: chiral condensates for both flavors coincide again.

1551: This region of the phase diagram has been studied at zero temperature in

1552: \cite{Son:2000xc} and \cite{Splittorff:2000mm} by means of a

1553: chiral Lagrangian for small values 

1554: of $\mu_I$.

1555: Finite temperature effects have been included in a one-loop

1556: calculation in chiral perturbation theory in \cite{Splittorff:2002xn} and

1557: in the random gauge model of \cite{Vanderheyden:2001gx}. 

1558: 

1559: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1560: \begin{figure*}

1561: \hspace*{-0.9cm}\includegraphics[scale=0.92, clip=true, angle=0,

1562:   draft=false]{fig6.eps} 

1563: \caption{\label{fig6} Phase diagram in the $\mu_B$-$T$--plane at

1564:   fixed values of the isospin chemical potential $\mu_I G$ 

1565: for a finite quark mass of $mG=0.1$. Values for $\mu_I G$ are

1566:   given in the panels. First (second) order transitions are 

1567: indicated by solid (dashed) lines, the dotted lines indicate a

1568:   crossover. Phases are labeled by the non-vanishing 

1569: condensates, or the condensates with large expectation values.}

1570: \end{figure*}

1571: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1572: 

1573: \paragraph{Chiral condensate.}

1574: Below the critical value of $\mu_I$, chiral symmetry is broken by 

1575: the chiral condensate, and at high temperature it is restored after a

1576: crossover transition. 

1577: For $m \neq 0$, it is most transparent to solve the saddle point

1578: equations for the chiral condensate by expanding in the quark

1579: mass $m$.

1580: In the broken phase, the chiral condensate $\sigma=\sigma_1=\sigma_2$

1581: is given by 

1582: \be

1583: \sigma=\sigma_0 + m \frac{1}{2\sigma_0^2}

1584: \frac{\sigma_0^2-\mu_I^2+T^2-8 G^2 T^2 \mu_I^2}{1-(4G^2\mu_I T)^2} -m

1585: +{\cal O}(m^2)  

1586: \nonumber \\

1587: \ee 

1588: where $\sigma_0$ is the solution of the saddle point equation for

1589: $m=0$, 

1590: \be

1591: \sigma_0^2&=&\frac{1}{2G^2} +\mu_I^2-T^2 \el

1592: &&+\frac{1}{2G^2} \sqrt{1-(4G^2\mu_I T)^2}.

1593: \ee

1594: With increasing temperature, the chiral condensate decreases. In the

1595: high temperature phase, the chiral

1596: condensate can be expanded about the trivial solution (5.39) as

1597: \be

1598: \sigma=-m\frac{\mu_I^2-T^2}{G^2 (\mu_I^2+T^2)^2

1599:   +\mu_I^2-T^2} + {\cal O}(m^3).

1600: \ee

1601: 

1602: \paragraph{Pion condensate.}

1603: As in the case $T=0$, for finite quark mass the chiral condensate does not

1604: vanish in the pion condensation phase. In this mixed phase, we 

1605: can solve the saddle point

1606: equations  analytically. The solution given by

1607: \be

1608: \sigma&=&-m+\frac{m}{2G^2} \frac{1}{\mu_I^2-m^2}\\

1609: \sigma^2+\rho^2&=& \frac{1}{G^2} +m^2 -\mu_I^2 -T^2. 

1610: \label{nc2rotate}

1611: \ee

1612: was first  obtained for diquark condensation \cite{Vanderheyden:2001gx}.

1613: We notice that the chiral condensate is completely independent of the

1614: temperature, and only the total magnitude $\sigma^2+\rho^2$ of the

1615: condensates decreases with increasing temperature. At zero temperature

1616: these equations describe the rotation of a chiral condensate into

1617: a pion condensate for increasing $\mu_I$.

1618: The explicit solution for the pion condensate obtained from

1619: (\ref{nc2rotate}) is given by

1620: \be

1621: \rho^2&=& \frac{1}{G^2} \frac{\mu_I^2}{\mu_I^2-m^2} \el

1622: && -\frac{1}{4G^4}

1623: \frac{m^2}{(\mu_I^2-m^2)^2} -\mu_I^2 -T^2.

1624: \ee

1625: The phase with $\rho \ne 0$ is 

1626: bounded by a second order line at which $\rho$ vanishes:

1627: \be

1628: \lefteqn{\mu_I^2(\mu_I^2-m^2) -\frac{1}{4}m^2} \el

1629: &&-(\mu_I^2+T^2)(\mu_I^2-m^2)^2G^4=0.

1630: \ee

1631: An important observation is that in this case, as for $N_c=3$, the

1632: boundary of the pion condensation phase is always a second order

1633: transition, and there is no tricritical point. 

1634: The complete phase diagram for $\mu_B=0$ is pictured in Fig.~\ref{fig4}. 

1635: In the $\mu_I$-$T$--plane 

1636: the critical endpoint of the first order chiral restoration

1637: transition does not appear, because its position is in the region where

1638: the pion condensation phase has a smaller free energy than the chiral

1639: condensation phase.

1640: We have shown  \cite{Klein:2003fy} that in this plane even for large

1641: quark masses, the critical endpoint remains masked by the pion

1642: condensation phase.  

1643: 

1644: 

1645: \subsubsection{The complete $\mu_I$-$\mu_B$--plane at $T\neq0$}

1646: \setcounter{paragraph}{0}

1647: To obtain the full picture of the phase diagram at nonzero values for

1648: the two chemical potentials, the temperature and the quark mass, we

1649: solve the saddle point equations % j4 for this case 

1650: numerically. The

1651: results for the phase diagram are presented in Fig.~\ref{fig5}.

1652: It is in complete agreement with the expectations from our discussion

1653: of the phase diagram at either $m=0$ or $T=0$: The finite quark mass

1654: mainly affects the phase diagram at low values of the

1655: chemical potentials, and the regions with either a diquark or a pion

1656: condensate shrink with increasing temperature and finally vanish for a

1657: temperature of order one, $TG={\mathcal O}(1)$.

1658: 

1659: We wish to draw the attention to one particular detail of the phase

1660: diagram in our model.

1661: ~From Fig.~\ref{fig3}, 

1662: the diquark and pion condensation phases are no longer contiguous at the

1663: lines $\mu_1=0$ and $\mu_2=0$, and

1664: there is no first order transition between them. Instead, this region

1665: of the phase diagram is occupied once again by 

1666: the chirally broken phases, and the chiral phase transition line reemerges. 

1667: While the first order chiral transition line does not

1668: appear for $\mu_B=0$ (and by symmetry not for $\mu_I=0$, either), we

1669: thus find that for both $\mu_B$ and $\mu_I$ different from zero and an

1670: appropriate value of the quark mass this first order line and its

1671: critical endpoint can be observed. The phase diagram in the

1672: $\mu_B$-$T$ plane at an isospin chemical potential that is chosen

1673: to avoid the superfluid phases at low $\mu_B$ shares striking

1674: similarities with the phase diagram of QCD with three colors at

1675: nonzero baryon and isospin chemical potentials. 

1676: 

1677: 

1678: \subsubsection{The $\mu_B$-$T$--plane at constant $\mu_I$}

1679: 

1680: To study the chiral phase transitions for both $\mu_B$ and $\mu_I$

1681: nonzero in some more detail, we turn to

1682: the phase diagram in planes described by constant isospin

1683: chemical potential.

1684: Fig.~\ref{fig6} shows the phase diagram in these $\mu_B$-$T$--planes for fixed

1685: values of $\mu_I$. The corresponding cuts through the phase diagram in

1686: the three 

1687: parameters $\mu_I$, $\mu_B$ and $T$ are perpendicular to the cuts at

1688: constant $TG$ in Fig.~\ref{fig5}. Since

1689: $\mu_1=\mu_B+\mu_I$ and $\mu_2=\mu_B-\mu_I$, planes at constant

1690: $\mu_I$ intersect the planes with constant $T$ along lines

1691: parallel to the diagonal from the lower  

1692: left-hand to the upper right-hand corners of the panels in Fig.~\ref{fig5}.  

1693: 

1694: In our analysis of the plane $\mu_B=0$ and in

1695: Fig.~\ref{fig4} above, we have seen that the first order

1696: chiral phase transition and its 

1697: critical endpoint are not present for any value of the

1698: quark mass, and by virtue of the symmetries this also holds for the

1699: plane given by $\mu_I=0$, shown in the first panel of

1700: Fig.~\ref{fig6}. But with increasing value of the isospin chemical

1701: potential the phase 

1702: diagram in the $\mu_B$-$T$--plane changes considerably: first, the

1703: first order chiral transition for $\sigma_1 \sim \langle \bar u

1704: u\rangle$ at large temperature and its critical endpoint reappear.

1705: With a further increase of $\mu_I$, the chiral transition for

1706: $\sigma_2$ at large values of $\mu_B$ appears as well (see the second and

1707: following panels of Fig.~\ref{fig6}). We have to caution, though, that

1708: for the second chiral condensate $\sigma_2$ this is possible only

1709: because of the 

1710: saturation transition of the diquark condensate, which exists on the

1711: lattice as well as in our random matrix model. If $\mu_I$ is made

1712: larger still, the critical endpoint of the transition for $\sigma_1$

1713: is once again swallowed by the pion condensation phase (see last panel

1714: of Fig.~\ref{fig6}, where this is about to take place). 

1715: 

1716: 

1717: For a quark mass $mG>0.184$ and at any temperature, the diquark and

1718: pion superfluid phases are not contiguous. Therefore for a

1719: sufficiently large quark mass ($mG>0.184$) and an 

1720: isospin chemical potential that is

1721: small enough so that the pion superfluid phase is never reached

1722: and large enough so that the diquark superfluid phase is reached only

1723: at large $\mu_B$, the structure of the phase diagram is as in 

1724: Fig.~\ref{fig7}.  

1725:  The resulting phase diagram is strikingly similar to

1726: the phase diagram of the   Random Matrix Model for QCD with three

1727: color at nonzero baryon and isospin chemical potentials

1728: \cite{Klein:2003fy}. 

1729: 

1730: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1731: \begin{figure}

1732: \hspace*{-0.4cm}

1733: \includegraphics[scale=1.00, clip=true, angle=0,

1734: draft=false]{fig7.eps} 

1735: \caption{\label{fig7} Phase diagram in the $\mu_B$-$T$--plane at

1736:   fixed values of the isospin chemical potential $\mu_I G=0.3$ 

1737: for a finite quark mass of $mG=0.2$. First (second) order transitions are 

1738: indicated by solid (dashed) lines, the dotted lines indicate a

1739:   crossover. Phases are labeled by the non-vanishing 

1740: condensates, or the condensates with large expectation values. This

1741: phase diagram is very similar for either QCD with two colors or for

1742: QCD with three colors.}

1743: \end{figure}

1744: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

1745: 

1746: \section{Conclusion} 

1747: \label{nc2section-6} 

1748: \setcounter{equation}{0} 

1749: In this article, we have introduced a random matrix model with the

1750: symmetries of QCD with two colors and two flavors and nonzero

1751: baryon chemical potential, isospin chemical potential and temperature. 

1752: Contrary to work elsewhere in the literature, our temperature term

1753: respects the flavor symmetry of the partition function as it should.

1754:  

1755: The random matrix partition function has been 

1756: rewritten exactly in terms of an effective Lagrangian.

1757: To lowest order in the chiral perturbation expansion, the effective Lagrangian 

1758: agrees with the one derived in chiral

1759: perturbation theory \cite{Kogut:1999iv, Kogut:2000ek,

1760:   Splittorff:2000mm}.

1761: We have analyzed the phase diagram of this model 

1762: for values of the mass and chemical potential outside of the domain

1763: of chiral perturbation theory. 

1764: It shows a rich phase structure 

1765: with phases

1766: characterized by the diquark condensate, the pion condensate and the chiral

1767: condensates, $\langle\bar u u\rangle$ and $\langle \bar d d \rangle$.

1768: In the presence of two chemical potentials,

1769: the chiral condensates are not necessarily equal. For a nonzero quark

1770: mass, in regions

1771: without pion or baryonic condensate at small values for the

1772: chemical potentials, the restoration of the spontaneously broken

1773: symmetry takes place via two separate crossovers, across which the

1774: chiral condensates for the two flavors rapidly change in value.  

1775: The vanishing of the superfluid phases at large values of

1776: the relevant chemical potentials, which is also observed on the

1777: lattice, has been explained as a saturation effect.

1778: On the lattice, this is due to the finite number of lattice sites

1779: \cite{Kogut:2001na}, and for random matrix models, this is   

1780: due the compactness of the Dirac operator. 

1781: 

1782: 

1783: The tricritical point of the diquark

1784: and pion condensation transitions observed in lattice calculations

1785: \cite{Kogut:2001if, Kogut:2002tm, Kogut:2002zg, Kogut:2002cm} is

1786: absent in the random matrix models. 

1787: Since the Random Matrix partition

1788: function is equivalent to a mean field description of the phase

1789: transition, this is not entirely surprising. Symmetries alone

1790: do not determine the presence of a tricritical point.

1791: In chiral perturbation

1792: theory, a tricritical point is obtained from a one-loop effective

1793: potential \cite{Splittorff:2002xn,dunne}.

1794: Ultimately, the question of the

1795: structure of the phase diagram in this regard must be answered by

1796: lattice calculations \cite{Kogut:2002cm}.

1797: 

1798: Finally, we would like to compare our results with the 

1799: random matrix study of the

1800: phase diagram for three-color QCD at nonzero temperature, baryon and

1801: isospin chemical potentials \cite{Klein:2003fy}. 

1802: First we notice that the upper-left quadrant of

1803: the phase diagram in the $\mu_B$-$\mu_I$--plane is identical in both

1804: cases. This is due to the fact that in both cases, the isospin

1805: chemical potential is related to Goldstone bosons of the microscopic

1806: theory. The upper-right quadrants are obviously different in the two

1807: theories, since there

1808: is no Goldstone boson that carries baryon number in three-color

1809: QCD. However, the phase diagram for three-color QCD with three quark

1810: flavors of equal masses at nonzero isospin and strange chemical

1811: potential should be very similar to the one we have presented here. 

1812: 

1813: Second, and more importantly, we notice that it is possible to find

1814: a quark mass and an isospin chemical potential for which the diquark

1815: and pion condensation phases appear only in certain regions of the

1816: phase diagram. Namely, they are present only at low temperature and

1817: either very small or 

1818: very large values of the baryon chemical potential. As shown in

1819: Figs.~\ref{fig6} and \ref{fig7}, we see that the first order

1820: transition lines of the 

1821: chiral transitions and their critical endpoints are not at all affected

1822: by the phases with bosonic condensates. Indeed, the transition lines

1823: are identical to those that are found in the random matrix model for

1824: three-color QCD. This observation implies that the critical endpoint

1825: of three-color QCD can be studied in two-color QCD. 

1826: For an appropriate choice of quark mass and isospin chemical potential,

1827: the phase diagram in the $\mu_B$-$T$ plane are almost identical for

1828: QCD with two colors and for QCD with three colors.

1829: In particular, the doubling of the critical lines due to a nonzero

1830: isospin chemical potential can be studied in QCD

1831: with two colors.

1832: 

1833: \begin{acknowledgments}

1834: J. Kogut, K. Splittorff and B. Vanderheyden are acknowledged for

1835: useful discussions. D.T. is supported in part by the

1836: ``Holderbank''-Stiftung. This work was partially supported by the

1837: U.S. DOE Grant No. DE-FG-88ER40388 and by the NSF under Grant

1838: No. NSF-PHY-0102409.  

1839: \end{acknowledgments}

1840: 

1841: % thebibliography is in file nc2bib.tex

1842: 

1843: \input nc2bib

1844: 

1845: \end{document}

1846: