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

2: % 

3: % Gravothermal Catastrophe and Tsallis' Generalized 

4: % Entropy of Self-Gravitating Systems III. 

5: %      ---- quasi-equilibrium structure using normalized q-values ----

6: % 

7: % 

8: %                                  Atsushi Taruya \& Masa-aki Sakagami

9: %

10: %                                        UTAP-428, RESCEU-17/02, KUCP0223

11: %

12: %                 Modified version according to the referee's suggestion 

13: %                                                            2002/12/30

14: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%  

15: \documentclass[12pt]{elsart}

16: %\documentclass[12pt,doublespacing]{elsart}

17: 

18: % Use the option doublespacing or reviewcopy to obtain double line spacing

19: % \documentclass[doublespacing]{elsart}

20: 

21: % if you use PostScript figures in your article

22: % use the graphics package for simple commands

23: % \usepackage{graphics}

24: % or use the graphicx package for more complicated commands

25: % \usepackage{graphicx}

26: % or use the epsfig package if you prefer to use the old commands

27: % \usepackage{epsfig}

28: 

29: % The amssymb package provides various useful mathematical symbols

30: \usepackage{amssymb}

31: \usepackage{graphics}

32: \usepackage{graphicx}

33: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

34: \newcommand{\xx}{\mbox{\boldmath$x$}}

35: \newcommand{\yy}{\mbox{\boldmath$y$}}

36: \newcommand{\vv}{\mbox{\boldmath$v$}}

37: \newcommand{\pp}{\mbox{\boldmath$p$}}

38: \newcommand{\re}{r_e}

39: \newcommand{\xxie}{\xi_e}

40: \newcommand{\Zq}{N_{q}}

41: \newcommand{\unitm}{m_{\scriptscriptstyle 0}}

42: \newcommand{\unitl}{l_{\scriptscriptstyle 0}}

43: \newcommand{\unitv}{v_{\scriptscriptstyle 0}}

44: \newcommand{\bftau}{\mbox{\boldmath$\tau$}}

45: \newcommand{\Tphys}{T_{\rm phys}}

46: \newcommand{\Cv}{C_{\rm\scriptscriptstyle V}}

47: \newcommand{\drho}{\delta\rho}

48: \newcommand{\dPhi}{\delta\Phi}

49: \newcommand{\ue}{u_e}

50: \newcommand{\ve}{v_e}

51: \newcommand{\xie}{\xi_e}

52: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

53: \begin{document}

54: 

55: \begin{frontmatter}

56: 

57: % Title, authors and addresses

58: 

59: % use the thanksref command within \title, \author or \address for footnotes;

60: % use the corauthref command within \author for corresponding author footnotes;

61: % use the ead command for the email address,

62: % and the form \ead[url] for the home page:

63: % \title{Title\thanksref{label1}}

64: % \thanks[label1]{}

65: % \author{Name\corauthref{cor1}\thanksref{label2}}

66: % \ead{email address}

67: % \ead[url]{home page}

68: % \thanks[label2]{}

69: % \corauth[cor1]{}

70: % \address{Address\thanksref{label3}}

71: % \thanks[label3]{}

72: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

73: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

74: \title{\large Gravothermal Catastrophe and Tsallis' Generalized 

75: Entropy of Self-Gravitating Systems III. \\

76: quasi-equilibrium structure using normalized $q$-values

77: }

78: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

79: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

80: % use optional labels to link authors explicitly to addresses:

81: % \author[label1,label2]{}

82: % \address[label1]{}

83: % \address[label2]{}

84: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

85: \author[taruya]{Atsushi Taruya}

86: \address[taruya]{Research Center for the Early Universe(RESCEU), 

87: School of Science, University of Tokyo, Tokyo 113-0033, Japan}

88: \ead{ataruya@utap.phys.s.u-tokyo.ac.jp}

89: 

90: % ---------------------------

91: 

92: \author[sakagami]{Masa-aki Sakagami}

93: \address[sakagami]{Department of Fundamental Sciences, FIHS, 

94: Kyoto University, Kyoto 606-8501, Japan}

95: \ead{sakagami@phys.h.kyoto-u.ac.jp}

96: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

97: \begin{abstract}

98: We revisit the issues on the thermodynamic property of 

99: stellar self-gravitating system arising from Tsallis' non-extensive 

100: entropy. Previous papers (Physica A 307 (2002) 185; 

101: {\it ibid}. (2003) in press (cond-mat/0204315)) have revealed that the 

102: extremum-state of Tsallis entropy, so-called {\it stellar polytrope}, 

103: has consistent thermodynamic structure, which predicts 

104: the thermodynamic instability due to the negative specific heat. However, 

105: their analyses heavily relies on the old Tsallis formalism using standard 

106: linear mean values. In this paper, extending our previous study, we focus 

107: on the quasi-equilibrium structure based on the standard framework by means 

108: of the normalized $q$-expectation values. It then turns out that the new 

109: extremum-state of Tsallis entropy essentially remains unchanged 

110: from the previous result, i.e., the stellar quasi-equilibrium distribution 

111: can be described by the stellar polytrope. While the thermodynamic 

112: stability for a system confined in an adiabatic wall completely agrees 

113: with the previous study and thereby the stability/instability criterion 

114: remains unchanged, the stability analysis reveals a new 

115: equilibrium property for the system surrounded by a thermal bath. 

116: In any case, the stability/instability criteria are consistently 

117: explained from the presence of negative specific heat and within 

118: the formalism, the stellar polytrope is characterized as a 

119: plausible non-extensive meta-equilibrium state. 

120: \end{abstract}

121: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

122: \begin{keyword}

123: % keywords here, in the form: keyword \sep keyword

124: non-extensive entropy \sep self-gravitating system 

125: \sep gravothermal instability \sep negative specific heat 

126: \sep stellar polytrope 

127: % PACS codes here, in the form: \PACS code \sep code

128: \PACS 05.20.-y, 05.90.+m, 95.30.Tg

129: \end{keyword} 

130: \end{frontmatter}

131: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

132: %

133: %

134: %

135: %

136: %

137: %

138: %

139: % main text

140: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

141: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

142: \section{Introduction}

143: \label{sec: intro}

144: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

145: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

146: %

147: %

148: %

149: %

150: %

151: %

152: %

153: %%%%%%%%%%%%%%%%%%%%%

154: In many astrophysical problems involving self-gravitating many-body 

155: system, thermodynamic or statistical mechanical treatment 

156: usually loses its power due to the peculiar feature of long-range 

157: attractive force. In some restricted cases with the long-term 

158: evolution of stellar self-gravitating systems, 

159: however, thermodynamic stability criterion or statistical mechanical analysis 

160: recovers the physical relevance and plays an essential role 

161: in predicting the fate of such system. In fact, it is well-known that 

162: late-time stellar dynamical evolution of the globular clusters as a 

163: real astrophysical system is driven by 

164: so-called {\it gravothermal catastrophe}, i.e., thermodynamic instability 

165: arising from the negative specific heat, which is widely accepted as a 

166: fundamental astrophysical process \cite{BT1987,EHI1987,MH1997}. 

167: Historically, the gravothermal 

168: catastrophe has been investigated in detail considering the very idealized 

169: situation, i.e., stellar self-gravitating system consisting of 

170: many particles confined within a cavity of hard sphere 

171: \cite{Antonov1962,LW1968,HS1978,Padmanabhan1989,Padmanabhan1990}. 

172: In particular, special attention to the statistical mechanical approach 

173: using the Boltzmann-Gibbs entropy has been paid. 

174: 

175: 

176: %%%%%%%%%%%%%%%%%%%%%

177: Recently,  this classic issue was re-analyzed by present authors 

178: on the basis of the non-extensive 

179: thermostatistics with Tsallis' generalized entropy \cite{T1988} 

180: (for comprehensive review of Tsallis' non-extensive formalism and 

181: its application to the other subject of physics, see 

182: Refs.\cite{T1999,AO2001}). 

183: In contrast to the Boltzmann-Gibbs entropy, the quasi-equilibrium 

184: distribution\footnote{ Here and in what follows, we use 

185: the term {\it quasi-equilibrium} rather than the (thermal) equilibrium,  

186: since there is no strict thermal equilibrium in a self-gravitating 

187: system. The quasi-equilibrium state means that the system is, at least,   

188: stable in a dynamical equilibrium state.  

189: } 

190: 

191: 

192: characterized by the Tsallis entropy can be reduced to the stellar 

193: polytropic system\cite{PP1993,PP1999}. Evaluating the second variation of 

194: entropy around the quasi-equilibrium state, we have developed the stability 

195: analysis and discussed the criterion for onset of gravothermal 

196: catastrophe (Ref.\cite{TS2002a}, hereafter paper I). Further, 

197: to clarify the origin of this instability,  

198: thermodynamic properties has been investigated in detail 

199: calculating the specific heat of the stellar polytrope 

200: (Ref.\cite{TS2002b}, hereafter paper II). 

201: The most noticeable thing in their papers is that the existence of 

202: thermodynamic instability indicated from the second variation of 

203: entropy or free-energy is completely explained from the presence of 

204: negative specific heat. 

205: As a consequence, the gravothermal instability appears at the 

206: polytrope indices $n>5$ for a system confined in an adiabatic wall 

207: and at $n>3$ for a system surrounded by a thermal bath. 

208: 

209: 

210: %%%%%%%%%%%%%%%%%%%%%

211: While the results in papers I and II are indeed satisfactory and the 

212: physical interpretation of the instability is fully consistent with previous 

213: early works, from a standard viewpoint of the Tsallis non-extensive formalism, 

214: several important issues still remain unresolved. 

215: Among these, the most crucial problem is the choice of the 

216: statistical average in non-extensive thermostatistics. 

217: The previous analyses have been all investigated utilizing the 

218: {\it old} Tsallis formalism with the 

219: standard linear mean values, however, 

220: a more sophisticated framework by means of 

221: the normalized $q$-values has been recently presented 

222: \cite{TMP1998,MNPP2000}. As several authors 

223: advocated, the analysis using normalized $q$-values is thought to be 

224: essential, since the undesirable divergences in some physical systems 

225: can be eliminated safely when introducing the normalized $q$-values. 

226: Further, non-uniqueness of the Boltzmann-Gibbs theory has been 

227: shown using the normalized $q$-values \cite{AR2000}. 

228: Of course, this does not imply that 

229: all the analyses with standard linear means or un-normalized $q$-values 

230: lose the physical significance\footnote{Indeed, even the old formalism 

231: consistently recovers the standard Legendre transform structure leading to 

232: the usual thermodynamic relations \cite{CT1991,PP1997}.}, 

233: however, in order to 

234: pursue the physical reality of the non-extensive thermodynamics,   

235: thermodynamic structure of the stellar self-gravitating system still 

236: needs to be investigated and the relation between the old and the new 

237: formalism must be clarified.  

238: 

239: 

240: %%%%%%%%%%%%%%%%%%%%%

241: This paper especially focuses on this matter starting from the construction 

242: of the extremum-entropy state by means of the normalized $q$-values. 

243: Within a mean-field treatment, we investigate the thermodynamic 

244: property of the quasi-equilibrium distribution 

245: surrounded by an adiabatic and a thermally conducting wall. 

246: In particular, we discuss the existence or the absence of thermodynamic 

247: instability evaluating the specific heat of the quasi-equilibrium systems. 

248: Further, in order to check the consistency of the thermodynamic structure,  

249: the stability/instability criteria obtained from the thermodynamic property 

250: are re-analyzed from the second variation of entropy and free-energy.   

251: We found that the extremum state of the Tsallis entropy essentially 

252: remains unchanged and is described by the stellar polytrope. While 

253: the stability of the system in an adiabatic wall exactly 

254: coincides with the results in paper I, the new formalism using 

255: the normalized $q$-values reveals a new thermodynamic structure 

256: for a stellar system surrounded by a 

257: thermal bath. This point will be discussed in detail comparing it with 

258: the previous results.

259: 

260: 

261: %%%%%%%%%%%%%%%%%%%%%

262: The paper is organized as follows. In Section \ref{sec: polytrope}, 

263: employing the standard Tsallis formalism, 

264: we revisit the issue on the most probable 

265: state of stellar quasi-equilibrium distribution on the basis of maximum 

266: entropy principle. While the resultant extremum state of Tsallis entropy 

267: reduces to the same stellar polytropic distribution as previously found,  

268: the new equilibrium distribution has several distinct structures.

269: Taking fully account of this fact, 

270: in Section \ref{sec: thermodynamics}, the thermodynamic 

271: properties of stellar polytropic system are investigated in detail. 

272: Thermodynamic temperature in stellar system is 

273: identified through the {\it modified} Clausius relation, 

274: which is indeed consistent with the recent claim based on the thermodynamic 

275: zero-th law. Then, we evaluate the specific heat and discuss the existence or 

276: absence of thermodynamic instability. In Section 

277: \ref{sec: stability_variation}, the stability/instability criteria are 

278: re-considered by means of the second variation of entropy and free-energy.  

279: In contrast to the previous analysis using the old Tsallis formalism, 

280: the second variation of free-energy shows a distinct thermodynamic 

281: structure. Nevertheless, 

282: the zero-eigenvalue problem in each case exactly 

283: recovers the marginal stability condition inferred from the specific heat. 

284: Thus, within the new formalism using normalized $q$-values, 

285: all the analyses are consistent and the stellar polytrope can be 

286: regarded as a plausible non-extensive thermal state. 

287: Finally, Section \ref{sec: conclusion} is devoted to the conclusion and 

288: the discussion.   

289: %

290: %

291: %

292: %

293: %

294: %

295: %

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

297: \section{Maximum entropy principle revisited} 

298: \label{sec: polytrope}

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

300: %

301: %

302: %

303: %

304: %

305: %

306: %

307: %

308: %%%%%%%%%%%%%%%%%%%%%

309: Throughout the paper, we pursue to investigate the equilibrium property of 

310: the stellar self-gravitating system consisting of $N$ particles 

311: confined in a spherical cavity of radius $\re$. For simplicity, 

312: each particle has the same mass $m_0$ and interacts via Newton gravity 

313: only. Then, the total mass becomes $M=Nm_0$. 

314: In this situation, owing to the maximum entropy principle, we seek 

315: the most probable quasi-equilibrium distribution 

316: in an adiabatic treatment. That is, 

317: we consider the quasi-equilibrium structure as the extremum-entropy 

318: state in which 

319: the particles elastically bounce from the wall, keeping the mass $M$ 

320: and energy $E$ constant. 

321: 

322: 

323: 

324: %%%%%%%%%%%%%%%%%%%%%

325: Following the papers I and II, 

326: we treat this issue employing the mean-field approach that 

327: the correlation between particles is smeared out and the system can 

328: be simply described by the 

329: one-particle distribution function $f(\xx,\vv)$, defined in six-dimensional 

330: phase-space $(\xx,\vv)$. In this treatment, the one-particle distribution 

331: is regarded as a fundamental quantity characterizing the stellar system.  

332: Let us denote the phase-space element as $h^3=(l_0v_0)^3$ with 

333: unit length $l_0$ and unit velocity $v_0$ and define the integral measure 

334: $d^6\bftau\equiv d^3\xx d^3\vv/h^3$. Regarding the function 

335: $f(\xx,\vv)$ as a fundamental statistical quantity, the energy and 

336: the mass are respectively expressed as follows: 

337: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

338: \begin{eqnarray}

339: &&   E= K+U = \int \left\{ \frac{1}{2}\,v^2 + \frac{1}{2}\,

340:   \Phi(\xx) \right\}\,\,f(\xx,\vv)\,\,d^6\bftau, 

341: \label{eq: def_E} \\

342: &&    M = \int \,\, f(\xx,\vv)\,\,d^6\bftau, 

343: \label{eq: def_M}

344: \end{eqnarray}

345: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

346: where the quantity $\Phi(\xx)$ is the gravitational potential 

347: given by 

348: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

349: \begin{eqnarray}

350:   \Phi(\xx) = -G\int \frac{f(\xx',\vv')}{|\xx-\xx'|}\,d^6\bftau'.

351: \label{eq: def_Phi}

352: \end{eqnarray}

353: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

354: 

355: 

356: 

357: %%%%%%%%%%%%%%%%%%%%%

358: On the other hand, in the new framework of Tsallis' non-extensive 

359: thermostatistics, all the macroscopic observables of the quasi-equilibrium 

360: system can be characterized by the escort distribution, but the 

361: escort distribution itself is not thought to be fundamental. 

362: Rather, there exists a more fundamental probability function 

363: $p(\xx,\vv)$ that quantifies the phase-space structure. 

364: With a help of this function, the escort distribution is defined and 

365: the macroscopic observables are expressed as the normalized 

366: $q$-expectation value as follows (e.g., Refs.\cite{TMP1998,MNPP2000}): 

367: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

368: \begin{eqnarray}

369: &  \mbox{escort distribution}~~ &: 

370: P_q(\xx,\vv)= \frac{\displaystyle \left\{p(\xx,\vv)\right\}^q}

371:   {\displaystyle \int d^6\bftau \left\{p(\xx,\vv)\right\}^q}\,\,\,, 

372: \label{eq: def_of_escort}   

373: \\

374: &  \mbox{normalized $q$-value}~~ &: 

375: \langle {\it O}_i\rangle_q = \int d^6 \bftau \,\,{\it O}_i\,\,

376: P_q(\xx,\vv)\,\,.

377: \label{eq: def_of_q-values} 

378: \end{eqnarray}

379: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

380: And based on the fundamental probability $p(\xx,\vv)$, 

381: the Tsallis entropy is given by

382: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

383: \begin{equation}

384:         S_q=-\frac{1}{q-1}\,\int d^6\bftau \,

385: \left[\left\{p(\xx,\vv)\right\}^q-p(\xx,\vv)\right].

386: \label{eq: Tsallis entropy}

387: \end{equation}

388: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

389: Note that the probability $p(\xx,\vv)$ satisfies the 

390: normalization condition:  

391: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

392: \begin{equation}

393:   \label{eq: normarization_p(x,v)}

394:   \int\,\,d^6\bftau\,\,p(\xx,\vv) = 1.

395: \end{equation}

396: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

397: 

398: 

399: 

400: %%%%%%%%%%%%%%%%%%%%% 

401: To  apply the above Tsallis formalism to the present problem 

402: without changing the definition of energy and mass 

403: (\ref{eq: def_E}) and (\ref{eq: def_M}), we identify the 

404: one-particle distribution with the escort distribution $P_q$, not 

405: the probability function $p(\xx,\vv)$:  

406: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

407: \begin{equation}

408:   \label{eq: escort_dist}

409:   f(\xx,\vv) = M\,\,\frac{\{p(\xx,\vv)\}^q}{\Zq}\,\,; 

410: \,\,\,\,\,\,\,\,\,\,\,\,

411:   \Zq = \int d^6\bftau\,\,\{p(\xx,\vv)\}^q

412: \end{equation}

413: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

414: so as to satisfy the mass conservation (\ref{eq: def_M}). 

415: With this identification (\ref{eq: escort_dist}), the normalized 

416: $q$-values are naturally incorporated into our mean-field treatment. 

417: As a consequence, later analyses become almost parallel to the 

418: previous study, in a form-invariant manner. 

419: 

420: 

421: %%%%%%%%%%%%%%%%%%%%%

422: Now, adopting the relation (\ref{eq: escort_dist}), let us seek the 

423: extremum-entropy state under the constraints (\ref{eq: def_E}) and 

424: (\ref{eq: normarization_p(x,v)}). As has been 

425: discussed recently, there are two approaches that extremize 

426: the entropy under certain constraints, i.e., the method developed by 

427: Tsallis, Mendes \& Plastino (TMP)\cite{TMP1998} and the optimal 

428: Lagrange multiplier(OLM) method by Mart\'inez et al. \cite{MNPP2000}. 

429: Here, we specifically apply the TMP procedure and find the extremum 

430: entropy state. The alternative derivation using OLM method is presented 

431: in Appendix A. The only differences in the final expressions 

432: between the TMP and the OLM method are the dependence of the quantity 

433: $\Zq$, which can be summarized in a unified form (see 

434: Eq.(\ref{eq: extremum_state})  with (\ref{eq: A_Phi0})).  

435: 

436: 

437: %%%%%%%%%%%%%%%%%%%%%

438: The variational problem in the TMP method is given by the 

439: following equation: 

440: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

441: \begin{equation}

442:  \delta\left[ S_q- \alpha\left\{\int d^6\bftau p-1\right\}-

443: \beta\left\{ \int d^6\bftau 

444: \left(\frac{1}{2}v^2+\frac{1}{2}\Phi\right)f -  E\right\}\right] = 0,

445: \label{eq: Max.Ent.Prin}

446: \end{equation}

447: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

448: where the variables $\alpha$ and $\beta$ denote the Lagrange multipliers. 

449: The variation with respect to the probability $p(\xx,\vv)$ leads to 

450: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

451: \begin{equation}

452: \int d^6\bftau\left[-\frac{1}{q-1}\left(q\,p^{q-1}-1\right)\delta p 

453: -\alpha\,\, \delta p - \beta\, \left(\frac{1}{2}v^2+\Phi\right) 

454: \delta f\right]= 0. 

455: \label{eq: Max.Ent.Prin_2}

456: \end{equation}

457: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

458: Here, we used the fact that 

459: $\int d^6\bftau \delta\Phi f= \int d^6\bftau \Phi\delta f$. 

460: The above equation includes the variation $\delta f$, which can be expressed  

461: with a help of the relation (\ref{eq: escort_dist}): 

462: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

463: \begin{equation}

464: \delta f(\xx,\vv) = q \,\,f(\xx,\vv) \left\{ \frac{\delta p}{p(\xx,\vv)} 

465: -\frac{1}{M}\,\int d^6\bftau' f(\xx',\vv') 

466: \frac{\delta p'(\xx',\vv')}{p(\xx',\vv')} \right\}.

467: \label{eq: delta_f}

468: \end{equation}

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

470: Then, substituting the above expression into (\ref{eq: Max.Ent.Prin_2}) 

471: becomes 

472: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

473: \begin{eqnarray}

474: && \int d^6\bftau \left[-\frac{1}{q-1}\left(q\,p^{q-1}-1\right)

475: \right.

476: \nonumber\\

477: &&\left. ~~~~~~~~~~~~~~~~~~~~~~~~~~~

478: -\alpha -\beta\, M\,\,q\,\, \frac{p^{q-1}}{\Zq}\,

479: \left(\frac{1}{2}v^2+\Phi - \varepsilon \right) \right] \delta p =0.

480: \label{eq: Max.Ent.Prin_3}

481: \end{eqnarray}

482: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

483: with the quantity $\varepsilon$ being 

484: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

485: \begin{equation}

486: \varepsilon = \frac{1}{M}\,\int d^6\bftau 

487: \left(\frac{1}{2}v^2 + \Phi\right) f. 

488: \label{eq: epsilon}

489: \end{equation}

490: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

491: In arriving at the equation (\ref{eq: Max.Ent.Prin_3}), we have exchanged 

492: the role of the variables $(\xx,\vv)\leftrightarrow(\xx',\vv')$. 

493: Since the above equation must be satisfied independently of the choice of 

494: the variation $\delta p$, we obtain

495: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

496: \begin{equation}

497: - \frac{1}{q-1}  \left( q\, p^{q-1} -1 \right) 

498: - \alpha -\beta \,\,M \,q\,\frac{p^{q-1}}{\Zq}\,

499: \left(\frac{1}{2}\,v^2+\Phi-\varepsilon\right)= 0,   

500: \end{equation}

501: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

502: which reduces to the same power-law distribution 

503: of $\varepsilon$ as has been derived in papers I and II. That is, 

504: even in the new formalism, the extremum-entropy state remains unchanged and 

505: can be described by the so-called {\it stellar polytrope}. Together with 

506: the result by OLM method (Appendix A), it can be summarized as follows: 

507: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

508: \begin{equation}

509: f(x,v) = M\,\,\frac{\{p(x,v)\}^q}{\Zq}= 

510: A \left[\Phi_0 - \frac{1}{2}v^2 - \Phi(x) \right]^{q/(1-q)}, 

511: \label{eq: extremum_state}

512: \end{equation}

513: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

514: where we define the constants $A$ and $\Phi_0$: 

515: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

516: \begin{eqnarray}

517: A = \frac{M}{\Zq}\,\, \left\{ \frac{q(1-q)}{\alpha(1-q)+1} 

518:         \frac{\beta M }{\Zq'}\right\}^{q/(1-q)},~~~~~~\Phi_0

519:  = \frac{\Zq'}{\beta M(1-q)} + \varepsilon.

520: \label{eq: A_Phi0}

521: \end{eqnarray}

522: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

523: Note that the quantity $\Zq'$ means $\Zq'=\Zq$ for the 

524: TMP method and $\Zq'=1$ for the result using OLM procedure. 

525: 

526: 

527: %%%%%%%%%%%%%%%%%%%%%

528: For later analysis, we define the density $\rho(r)$ and the 

529: isotropic pressure $P(r)$ at the radius $r=|\xx|$ as    

530: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

531: \begin{eqnarray}

532: \rho(r) &\equiv& \int\frac{d^3\vv}{h^3}\,\,\, f(\xx,\vv),

533: \nonumber \\

534: &=& 4\sqrt{2}\pi \,B\left(\frac{3}{2},\frac{1}{1-q}\right)

535:   \,\frac{A}{h^3}\,\,[\Phi_0-\Phi(r)]^{1/(1-q)+1/2}

536: \label{eq: density}

537: \end{eqnarray}

538: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

539: and 

540: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

541: \begin{eqnarray}

542: P(r) &\equiv& \int\frac{d^3\vv}{h^3} \,\,\,\frac{1}{3}\, v^2 \,

543: f(\xx,\vv),

544: \nonumber \\

545: &=& \frac{8\sqrt{2}\pi }{3}\,B\left(\frac{5}{2},\frac{1}{1-q}\right)

546:   \,\frac{A}{h^3}\,\,[\Phi_0-\Phi(r)]^{1/(1-q)+3/2}.  

547: \label{eq: pressure}

548: \end{eqnarray}

549: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

550: Here, the function $B(a,b)$ denotes the beta function. 

551: These two equations lead to the following polytropic relation: 

552: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

553: \begin{equation}

554:   \label{eq: polytrope}

555:   P(r) = K_n\rho ^{1+1/n}(r), 

556: \end{equation}

557: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

558: with the polytrope index $n$ given by 

559: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

560: \begin{equation}

561:   \label{eq: index}

562:   n = \frac{1}{1-q} \,+ \,\frac{1}{2},

563: \end{equation}

564: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

565: and with the dimensional constant $K_n$: 

566: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

567: \begin{equation}

568:   \label{eq: K_n}

569:   K_n = \frac{1}{n+1}\,

570: \left[ \frac{4\sqrt{2}\,\pi}{h^3}\,\,

571: B\left(\frac{3}{2},n-\frac{1}{2}\right)\,\,

572: \left\{\frac{q(1-q)}{\alpha(1-q)+1}\,\,\frac{\beta\,M}{\Zq'}

573: \right\}^{n-3/2}\,\frac{M}{\Zq}

574: \right]^{-1/n}.

575: \end{equation}

576: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

577: Using these quantities, the one-particle distribution can be 

578: rewritten as follows: 

579: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

580: \begin{eqnarray}

581: f(x,v)&=& \frac{1}{4\sqrt{2}\pi\,\,B(3/2,n-1/2)}\,\,\,

582: \frac{\rho\,\,h^3}{ \{ (n+1)\,K_n\,\rho^{1/n}\}^{3/2}}  

583: \nonumber \\

584: &&~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

585: \times\,\,\,\left\{1-\frac{v^2/2}{(n+1)\,K_n\,\rho^{1/n}}\right\}^{n-3/2},    

586:   \label{eq: poly_dist}

587: \end{eqnarray}

588: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

589: which agrees with the previous result 

590: (see Eq.(16) with the identification $(n+1) K_n = (n-3/2)T$ 

591: in paper I). 

592: 

593: 

594: %%%%%%%%%%%

595: While the resultant form of the quasi-equilibrium distribution 

596: (\ref{eq: poly_dist}) turns out to be invariant irrespective of the 

597: choice of the statistical averages, we should be aware of the 

598: two important differences between the old and the new Tsallis 

599: formalism, which we shall describe below.   

600: 

601: 

602: %%%%%%%%%%%

603: First, the relation between the polytrope 

604: index $n$ and Tsallis parameter $q$ in the polytropic relation 

605: (\ref{eq: polytrope}) 

606: differs from the one obtained previously, but is related to 

607: it through the {\it duality transformation}, $q\leftrightarrow 1/q$ 

608: (see Eq.(14) in paper I or Eq.(12) in paper II). This property has been 

609: first addressed in Ref.\cite{TMP1998} in more general context, 

610: together with the changes in Lagrangian multiplier $\beta$.  

611: The duality relation implies that all of the thermodynamic properties 

612: in the new formalism can also be translated into those obtained in the old 

613: formalism. As shown in section \ref{sec: thermodynamics} and 

614: \ref{sec: stability_variation}, this is indeed true in the system confined 

615: in an adiabatic wall(micro-canonical ensemble case), however,  

616: the duality of thermodynamic structure cannot hold in the system 

617: surrounded by a thermal bath(canonical ensemble case). 

618: 

619: 

620: %%%%%%%%%%%

621: Second, notice that the quasi-equilibrium distribution (\ref{eq: extremum_state}) 

622: with (\ref{eq: A_Phi0}) contains the new quantities $\Zq$ and $\varepsilon$,    

623: which implicitly depend on the distribution function itself. In marked contrast to 

624: the result in old Tsallis formalism,    

625: this fact gives rise to the non-trivial thermodynamic relations as follows. 

626: Using the definitions of density and pressure (\ref{eq: density}) and 

627: (\ref{eq: pressure}),  the quantity $\varepsilon$ becomes

628: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

629: \begin{eqnarray}

630:   \varepsilon &=& \frac{1}{M}\,

631: \left\{\frac{3}{2}\,\int d^3x\,P(x) + \int d^3x\,\rho(x)\,\Phi(x)\,\right\}

632: \nonumber \\

633: &=& \frac{1}{M}\,

634:  \left\{\frac{3}{2}\,\int d^3x\,P(x) 

635:         - \int d^3x\,\rho(x)\,[\Phi_0-\Phi(x)]\,\right\} + \Phi_0.

636: \nonumber

637: \end{eqnarray}

638: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

639: Further using the relation $\Phi_0-\Phi(x)= (n+1)(P/\rho)$ from 

640: (\ref{eq: density}) and (\ref{eq: pressure}) and substituting the equation 

641: (\ref{eq: A_Phi0}) into the above expression,  the variable $\varepsilon$ 

642: is cancelled and the equation reduces to 

643: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

644: \begin{equation}

645: \frac{\Zq'}{\beta}\,=\,\int d^3x\,P(x).

646: \label{eq: beta-P_relation}

647: \end{equation}

648: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

649: As for the dependence of $\Zq$, the normalization 

650: condition (\ref{eq: normarization_p(x,v)}) implies that

651: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

652: \begin{eqnarray}

653:   \Zq\, = \,\left[ 

654: \int d^6\bftau\,\,\left\{\,f(\xx,\vv)\,\right\}^{1/q}\,\,

655: \right]^{-q}.

656: \nonumber

657: \end{eqnarray}

658: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

659: Substituting the distribution function (\ref{eq: poly_dist}) into the above 

660: equation and integrating over the velocity space, after some manipulation, 

661: one obtains 

662: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

663: \begin{equation}

664:   \Zq = \,\,c_n\,\,K_n^{(3/2)/(n-1/2)}\,\,

665: \left\{\int d^3x\,\,\rho^{1+1/n}(x)\,\,\right\}^{-(n-3/2)/(n-1/2)}

666:   \label{eq: Zq-rho_relation}

667: \end{equation}

668: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

669: with the constant $c_n$ given by 

670: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

671: \begin{equation}

672: c_n \,\,= \,\, \frac{4\sqrt{2}\,\pi\,B(3/2,n-1/2)}

673: {\left\{4\sqrt{2}\,\pi\,B(3/2,n+1/2) \right\}^{(n-3/2)/(n-1/2)}}\,\,

674: \left\{\frac{(n+1)^{3/2}}{h^3}\right\}^{1/(n-1/2)}.

675: \end{equation}

676: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

677: Equations (\ref{eq: beta-P_relation}) 

678: and (\ref{eq: Zq-rho_relation}) play a crucial role 

679: in determining the thermodynamic temperature of stellar polytrope 

680: in section \ref{sec: thermodynamics}, 

681: as well as the onset of gravothermal instability in section 

682: \ref{sec: stability_variation}.  

683: In particular, the equation (\ref{eq: beta-P_relation}) 

684: yields the {\it radius-mass-temperature relation} characterizing the 

685: quasi-equilibrium structure of the system confined in a thermal bath.

686: 

687: 

688: %%%%%%%%%%%

689: Keeping the above remarks in mind, hereafter, we will specifically 

690: focus on the spherically symmetric case with the 

691: polytrope index $n>3/2\,(q>0)$, in which the quasi-equilibrium distribution 

692: is at least dynamically stable (see Chap.5 of Ref.\cite{BT1987}). 

693: In this case, despite the above detailed 

694: differences,  the stellar quasi-equilibrium distribution 

695: can be characterized by the so-called {\it Emden solutions} 

696: (e.g., \cite{Chandra1939,KW1990}) and all the physical 

697: quantities are expressed in terms of the homology invariant 

698: variables $(u,v)$, which are subsequently used in later analysis. 

699: In Appendix B, together with some 

700: useful integral formulae, we summarize the relation between the Emden 

701: solution and the stellar polytropic distribution.

702: %

703: %

704: %

705: %

706: %

707: %

708: %

709: %

710: %

711: %

712: %

713: %

714: %

715: %

716: %

717: %

718: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

719: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

720: \section{Thermodynamic properties of stellar polytrope}

721: \label{sec: thermodynamics}

722: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

723: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

724: %

725: %

726: %

727: %

728: %

729: %

730: %

731: %

732: Having established the quasi-equilibrium distribution, we now 

733: investigate the thermodynamic properties of stellar polytropic system. 

734: To do this, we adopt the same procedure as in paper II.   

735: That is, we examine the Clausius relation under the quasi-static variation 

736: for the new equilibrium system in section \ref{subsec: clausius_relation}. 

737: In contrast to the previous analysis, 

738: the thermodynamic temperature can be identified with a help of 

739: (\ref{eq: beta-P_relation}), consistently with the recent claim based on 

740: the thermodynamic zero-th law. Using this temperature, in section 

741: \ref{subsec: negative_specific_heat}, presence or absence of thermodynamic 

742: instability is discussed evaluating the specific heat of stellar system. 

743: %

744: %

745: %

746: %

747: %

748: %

749: %

750: % ///////////////////////////////////////////////////////////// %

751: \subsection{On the definition of thermodynamic temperature}

752: \label{subsec: clausius_relation}

753: % ///////////////////////////////////////////////////////////// %

754: %

755: %

756: %

757: %

758: %

759: %

760: %

761: %

762: %

763: %

764: %

765: %

766: %%%%%%%%%%%

767: As usual, the thermodynamic instability in stellar system is 

768: intimately related to the presence of negative specific heat 

769: \cite{LW1968,LyndenBell1999}. 

770: The evaluation of specific heat is thus necessary for clarifying the 

771: thermodynamic property. In this regard, the identification of 

772: temperature in stellar system is the most essential task.  

773: 

774: 

775: %%%%%%%%%%%

776: In the new framework of Tsallis' non-extensive thermostatistics, 

777: the physically plausible thermodynamic temperature, $\Tphys$, can be 

778: defined from the zero-th law of thermodynamics 

779: \cite{AMPP2001,Abe2001,MPP2001}. Then, the thermodynamic 

780: temperature in non-extensive system differs from 

781: the usual one, i.e., the inverse of the Lagrange multiplier, $\beta$. 

782: Depending on the methods extremizing the entropy, one has 

783: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

784: \begin{eqnarray}

785: \Tphys=\left\{

786: \begin{array}{cr}

787: \beta^{-1}                      & ;\,\,\,\,\,\,(\mbox{OLM method}), 

788: \\

789: \,[1+(1-q)\,S_q]\,\,\beta^{-1}  & ;\,\,\,\,\,\,(\mbox{TMP method}). 

790: \end{array}

791: \right.

792: \label{eq: def_Tphys}

793: \end{eqnarray}

794: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

795: 

796: 

797: %%%%%%%%%%%

798: Translating the above result into our notation immediately yields 

799: that the physical temperature in stellar polytropic system is given by 

800: $\Tphys=\Zq'/\beta$. However, the relation (\ref{eq: def_Tphys}) 

801: should be carefully applied to the present case, since the  

802: verification of thermodynamic zero-th law is very difficult 

803: in stellar equilibrium system with long-range interaction. 

804: Furthermore, even using the new formalism, the energy $E$ 

805: still keeps non-extensive due to the self-referential form of the 

806: potential energy (see Eqs.(\ref{eq: def_E})(\ref{eq: def_Phi})). 

807: In order to validate the use of the definition (\ref{eq: def_Tphys}), 

808: we therefore adopt a rather simple procedure as examined in paper II. 

809: That is, we consider the relation between heat transfer and entropy change 

810: in the quasi-static treatment under keeping the total mass constant. 

811: According to the definition (\ref{eq: def_Tphys}), the Clausius relation 

812: is appropriately modified and is expressed as follows 

813: \cite{AMPP2001,Abe2001,MPP2001}: 

814: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

815: \begin{equation}

816:   \label{eq: modified_clausius}

817:   dS_q = \frac{1}{\Tphys}\{ 1+(1-q)\,S_q\}\,d'Q. 

818: \end{equation}

819: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

820: This, in turn, implies that the modified Clausius relation 

821: can be used as a consistency check of the physical 

822: temperature (\ref{eq: def_Tphys}) in stellar system. 

823: 

824: 

825: %%%%%%%%%%%

826: Let us first write down the entropy of extremum state:  

827: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

828: \begin{equation} 

829: S_q = \left(n-\frac{1}{2}\right)\,\,(\Zq-1).

830: \label{eq: S_q_extreme}

831: \end{equation} 

832: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

833: >From equation (\ref{eq: Zq-rho_relation}) with the polytropic relation 

834: (\ref{eq: polytrope}), the quantity $\Zq$ in equation 

835: (\ref{eq: S_q_extreme}) is rewritten as 

836: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

837: \begin{eqnarray}

838:   \Zq &=& c_n \, K_n^{n/(n-1/2)}\,

839:   \left\{\,\int d^3x\,\,P(x)\,\right\}^{-(n-3/2)(n-1/2)}

840: \nonumber \\

841:  &=& c_n \, K_n^{n/(n-1/2)}\, \left\{ \,\frac{1}{n-5}\,\frac{GM^2}{r_e}\,

842:  \left(\frac{n+1}{\ve}-2\frac{\ue}{\ve}-1\right)\right\}^{-(n-3/2)/(n-1/2)}

843: \end{eqnarray}

844: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

845: with a help of the integral formula (\ref{eq: formula_1}). 

846: Here, the variables $(\ue,\ve)$ 

847: denote the homology invariants evaluated at the boundary $r=\re$ 

848: (see definitions (\ref{eq: def_u})(\ref{eq: def_v})).  

849: Note also the fact that the constant $K_n$ is 

850: expressed in terms of $(\ue,\ve)$-variables: 

851: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

852: \begin{eqnarray}

853:   K_n = \frac{P_e}{\rho_e^{1+1/n}} = 

854:    \frac{GM}{\re}\,\frac{1}{\ve}\,\left(\frac{4\pi\re^3}{M}\,

855:   \frac{1}{\ue}\right)^{1/n}.

856: \end{eqnarray}

857: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

858: Using these expressions, the variation of the quantities 

859: $\Zq$ and $K_n$ respectively becomes 

860: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

861: \begin{eqnarray}

862: &&\left(n-\frac{1}{2}\right)\,\frac{d\Zq}{\Zq}\, =

863: \,n\,\frac{dK_n}{K_n}\,+\,\left(n-\frac{3}{2}\right)\,\frac{d\re}{\re}

864: \nonumber \\

865: &&~~~~~~~~~~~~  -\frac{n-3/2}{n-5}\,\frac{GM^2\beta}{\re\Zq'}\,

866: \left\{-\frac{n+1}{\ve}\,

867:   \frac{d\ve}{\ve}-2\left(\frac{d\ue}{\ue}-\frac{d\ve}{\ve}\right)

868: \right\},

869: \label{eq: dZ_q}

870: \end{eqnarray}

871: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

872: and 

873: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

874: \begin{eqnarray}

875: &&\frac{dK_n}{K_n} = -\frac{n-3}{n}\,\frac{d\re}{\re}\,-\frac{d\ve}{\ve}

876:   -\frac{1}{n}\,\frac{d\ue}{\ue}.

877: \label{eq: dK_n}

878: \end{eqnarray}

879: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

880: In the last line of equation (\ref{eq: dZ_q}),   

881: we have used the relation (\ref{eq: beta-P_relation}).  

882: Collecting these results, 

883: the entropy change $dS_q$ can be expressed as the variations of both

884: the homology invariants $(\ue,\ve)$ and the wall radius $\re$ 

885: as follows:

886: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

887: \begin{eqnarray}

888: &&  dS_q = \left(n-\frac{1}{2}\right)\,d\Zq 

889:   = \Zq\,\,\left[ \frac{3}{2}\,\frac{d\re}{\re}\,-

890:     \,\frac{d\ue}{\ue}\,-\,n\,\frac{d\ve}{\ve} 

891: \right.

892: \nonumber\\

893: &&~~~~~~~~~~~~~~\left. -\frac{n-3/2}{n-5}\,\frac{GM^2\beta}{\re\Zq'}\,

894: \left\{-\frac{n+1}{\ve}\,

895:   \frac{d\ve}{\ve}-2\left(\frac{d\ue}{\ue}-\frac{d\ve}{\ve}\right)

896: \right\}

897: \right], 

898: \end{eqnarray}

899: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

900: In the above equation, the term $GM^2\beta/(\re\Zq')$ can be 

901: factorized out using the relation (\ref{eq: beta-P_relation}) and 

902: we have 

903: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

904: \begin{eqnarray}

905: &&  dS_q = \Zq\,\,\frac{GM^2\beta}{\re\,\Zq'}

906: \,\,\left[\,-\,\frac{3}{2}\,\frac{1}{n-5}\,

907: \left( 2\,\frac{\ue}{\ve}-\frac{n+1}{\ve} + 1\right)\frac{d\re}{\re}

908: \right.

909: \nonumber\\

910: &&~~~~~~~~~~~~

911: +\frac{1}{n-5}\,\frac{1}{2\ve}\,

912: \left\{4\left(n-\frac{1}{2}\right)\ue+2\ve-2(n+1)\right\}\frac{d\ue}{\ue}

913: \nonumber\\

914: &&\left.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

915: +\left\{6\ue+2n\ve-3(n+1)\right\}\frac{d\ve}{\ve}

916: \right]. 

917: \end{eqnarray}

918: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

919: Further notice the fact that the variation of homology invariants is 

920: expressed as the variation of dimensionless quantity $d\xxie$ 

921: (see Eq.(\ref{eq: d(u,v)/dxi})).  

922: The entropy change is finally reduced to the following expression: 

923: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

924: \begin{eqnarray}

925:   &&  dS_q = \Zq\,\,\frac{GM^2\beta}{\re\,\Zq'}

926: \,\,\left[\,-\,\frac{3}{2}\,\frac{1}{n-5}\,

927: \left( 2\frac{\ue}{\ve}-\frac{n+1}{\ve} + 1\right)\frac{d\re}{\re}

928: -\frac{n-2}{n-5}\,\frac{1}{2\ve}\right.

929: \nonumber\\

930: &&\left.\times\left\{

931: 4\ue^2 + 2\ue\ve-\left(8+3\frac{n+1}{n-2}\right)\ue -

932: \frac{3}{n-2}\ve+3\left(\frac{n+1}{n-2}\right)\right\}

933: \frac{d\xi_e}{\xi_e}\,\right]. 

934: \label{eq: entropy_change} 

935: \end{eqnarray}

936: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

937: 

938: 

939: %%%%%%%%%%%

940: The Clausius relation in a quasi-static variation relates 

941: the variation of entropy with the heat change. According to the 

942: first-law of thermodynamics, the heat change in a quasi-static variation 

943: is estimated as follows (see Eq.(26) in paper II): 

944: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

945: \begin{eqnarray}

946:   d'Q \,\,= \,\,dE+ P_e dV 

947: &=& d\left(-\lambda\,\frac{

948: GM^2}{r_e}\right)+ 4\pi\re^2\,P_e\, d\re

949: \nonumber \\

950:   &=& \frac{GM^2}{\re}\,\,

951: \left\{\,\left(\lambda+\frac{\ue}{\ve}\right)\frac{d\re}{\re}-

952: \xxie\,\frac{d\lambda}{d\xxie}\frac{d\xxie}{\xxie}\right\},

953: \label{eq: heat_change}

954: \end{eqnarray}

955: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

956: where the dimensionless quantity $\lambda$ related to the energy 

957: and its derivative $d\lambda/d\xxie$ are respectively given by 

958: (see Eq.(\ref{eq: energy_uv})): 

959: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

960: \begin{eqnarray}

961: &&  \lambda\equiv -\frac{\re\,E}{GM^2}\,=

962: -\frac{1}{n-5}\,\left\{ \frac{3}{2}\,\left(1-\frac{n+1}{\ve}\right)

963: +(n-2)\frac{\ue}{\ve}\right\},

964: \label{eq: def_lambda}

965: \end{eqnarray}

966: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

967: and 

968: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

969: \begin{eqnarray}

970: &&\,\,\xxie\frac{d\lambda}{d\xxie} = \frac{n-2}{n-5}\,

971: \frac{g(\ue,\ve)}{2\ve}\,\,;\,\,

972: \nonumber\\

973: && g(u,v) = 4u^2 +2uv-\left\{8+3\left(\frac{n+1}{n-2}\right)

974: \right\}u-\frac{3}{n-2}\,v+3\left(\frac{n+1}{n-2}\right).

975: \label{eq: g(u,v)}

976: \end{eqnarray}

977: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

978: 

979: 

980: %%%%%%%%%%%

981: Hence, the comparison between (\ref{eq: entropy_change}) and 

982: (\ref{eq: heat_change}) with (\ref{eq: def_lambda})(\ref{eq: g(u,v)}) 

983: leads to the modified Clausius relation corresponding to 

984: the equation (\ref{eq: modified_clausius}): 

985: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

986: \begin{eqnarray}

987:   dS_q = \frac{\beta}{\Zq'}\,\Zq\,(dE + P_e\,dV)

988: = \frac{\beta}{\Zq'}\,\,\left\{1+(1-q)\,S_q\right\}d'Q.   

989: \end{eqnarray}

990: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

991: %

992: %

993: %

994: %

995: %

996: %

997: %

998: %

999: Therefore, with this relation, the thermodynamic temperature can be 

1000: consistently identified along the line of the argument in 

1001: Ref.\cite{AMPP2001} and the plausible physical temperature is now 

1002: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1003: \begin{equation}

1004:   \label{eq: T_phys}

1005:   \Tphys= \frac{\Zq'}{\beta}. 

1006: \end{equation}

1007: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1008: %

1009: %

1010: %

1011: %

1012: %

1013: %

1014: %

1015: %

1016: %

1017: %

1018: %

1019: %

1020: %

1021: %

1022: %

1023: %

1024: %

1025: %

1026: % ///////////////////////////////////////////////////////////// %

1027: \subsection{Thermodynamic instability arising from 

1028: the negative specific heat} 

1029: \label{subsec: negative_specific_heat}

1030: % ///////////////////////////////////////////////////////////// %

1031: %

1032: %

1033: %

1034: %

1035: %

1036: %

1037: %

1038: %

1039: %

1040: %

1041: %

1042: %

1043: %

1044: %%%%%%%%%%%

1045: Once adopting the definition (\ref{eq: T_phys}), we immediately 

1046: obtain the {\it radius-mass-temperature relation} in terms of the 

1047: homology invariants as follows. Equation (\ref{eq: beta-P_relation}) 

1048: in section \ref{sec: polytrope} leads to 

1049: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1050: \begin{eqnarray}

1051: \Tphys = \int d^3x\,P(x) = -\frac{1}{n-5}\,\frac{GM^2}{\re}\,

1052:         \left(2\,\frac{\ue}{\ve}-\frac{n+1}{\ve}+1\right)

1053: \label{eq: Tphys_P}

1054: \end{eqnarray}

1055: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1056: from (\ref{eq: formula_1}). 

1057: Thus, defining the dimensionless quantity $\eta=GM^2/(\re\Tphys)$, we obtain 

1058: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1059:   \begin{equation}

1060: \label{eq: eta}

1061: \eta\equiv\frac{GM^2}{r_e\Tphys} = \frac{(n-5)\,\ve}{n+1-2\ue-\ve}.    

1062:   \end{equation}

1063: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1064: This is in marked contrast to the result using the standard linear 

1065: mean (c.f. Eq.(31) of paper II).  While the radius-mass-temperature relation 

1066: in previous paper includes the residual dimensional parameter $h=(l_0v_0)^3$, 

1067: the expression (\ref{eq: eta}) has no such parameter dependence and 

1068: is quite similar to the result in Boltzmann-Gibbs case (e.g.,  Eq.(29) 

1069: in Ref.\cite{LW1968} or Eq.(25) in Ref.\cite{Chavanis2002a})). 

1070: The nice form of the radius-mass-temperature relation implies that the 

1071: specific heat can be determined independently of the residual parameter 

1072: $h$, which is indeed a desirable property for a rigid theoretical 

1073: prediction without any uncertainty. 

1074: 

1075: 

1076: %%%%%%%%%%%

1077: By definition, the specific heat at constant volume $\Cv$ is given by  

1078: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1079: \begin{equation}

1080:   \label{eq: specific_heat}

1081:   \Cv \equiv \left(\frac{d E}{d\Tphys}\right)_e

1082:   =\frac{\displaystyle \left(\frac{d E}{d\xi}\right)_e}

1083:   {\displaystyle \left(\frac{d\Tphys}{d\xi}\right)_e}. 

1084: \end{equation}

1085: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1086: The numerator and the denominator in the last expression are 

1087: respectively rewritten as follows:

1088: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1089: \begin{eqnarray}

1090:   \label{eq: stability_1}

1091: & \left(\frac{d E }{d\xi}\right)_e =&-\frac{GM^2}{r_e}\frac{d\lambda}{d\xie}

1092: =-\frac{GM^2}{r_e}\,\frac{n-2}{n-5}\,\,\frac{g(\ue,\ve)}{2\ue\xie},

1093: \end{eqnarray}

1094: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1095: and

1096: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1097: \begin{eqnarray}

1098: &\left(\frac{d\Tphys}{d\xi}\right)_e=& 

1099: \frac{GM^2}{r_e}\frac{d}{d\xie}\,\eta^{-1}

1100: =\frac{GM^2}{r_e}\,\frac{1}{n-5}\,\,\frac{k(\ue,\ve)}{\ue\xie}.

1101:   \label{eq: stability_2}

1102: \end{eqnarray}

1103: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1104: In the above equations, the functions $g(\ue,\ve)$ is already 

1105: given by (\ref{eq: g(u,v)})  

1106: and the function $k(\ue,\ve)$ is expressed as 

1107: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1108: \begin{eqnarray}

1109:   \label{eq: h(u,v)}

1110: &&  k(\ue,\ve) = 4\ue^2 +2\ue\ve - (n+9)\ue-\ve + n+1,   

1111: \end{eqnarray}

1112: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1113: from (\ref{eq: d(u,v)/dxi}). Thus, the expression of specific heat 

1114: can be reduced to the following simple form: 

1115: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1116: \begin{equation}

1117:   \label{eq: C_v}

1118:   \Cv = -\frac{n-2}{2}\,\,\,\,\frac{g(\ue,\ve)}{k(\ue,\ve)}. 

1119: \end{equation}

1120: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1121: Provided the homology invariants at the boundary, 

1122: the specific heat in the new formalism 

1123: is uniquely determined, free from the residual parameter $h$ 

1124: (c.f. Eq.(42) in paper II). 

1125: In this sense, the result (\ref{eq: C_v}) can be regarded 

1126: as a successful outcome of the new Tsallis formalism using 

1127: the normalized $q$-values. 

1128: 

1129: 

1130: %%%%%%%%%%%

1131: Now, we focus on the thermodynamic instability inferred from 

1132: the qualitative behavior of specific heat. Recall that the 

1133: thermodynamic instability appears when the specific heat of the 

1134: system changes its sign. From (\ref{eq: C_v}), we readily expect 

1135: the two possibilities. One is the case when the function $g(\ue,\ve)$ 

1136: changes its sign. In this case, the condition for marginal stability, 

1137: $\Cv=0$, becomes 

1138: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1139: \begin{equation}

1140:   \label{eq: marginal_stability(1)}

1141:   g(\ue,\ve)=0.

1142: \end{equation}

1143: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1144: The other cases appear when the sign of the function $k(\ue,\ve)$ 

1145: is changed. In this case, the marginal stability leads to 

1146: the divergent behavior, $\Cv\to\pm\infty$ and the condition is given by  

1147: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1148: \begin{equation}

1149:   \label{eq: marginal_stability(2)}

1150:   k(\ue,\ve)=0.

1151: \end{equation}

1152: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1153: 

1154: 

1155: %%%%%%%%%%%

1156: To see how and when the instability develops, 

1157: in Fig. \ref{fig: eta_lambda}, we plot a family of Emden solutions in 

1158: the $(\eta,\lambda)$-plane. Since the dimensionless parameters 

1159: $\lambda$ and $\eta$ are respectively proportional to $-E$ and 

1160: $\Tphys^{-1}$, the signature of the specific heat can be easily deduced 

1161: from the slope of the curve. Note that each point along the trajectory 

1162: represents an Emden solution evaluated at the different value 

1163: of the radius $\re$. 

1164: >From the boundary condition (\ref{eq: boundary}), 

1165: all the trajectories start from $(\eta,\lambda)=(0,-\infty)$, corresponding 

1166: to the limit $\re\to0$. As increasing the radius, the trajectories 

1167: first move to the upper-right direction as marked by the arrow, and they 

1168: suddenly change their direction to upper-left. This means that the divergent 

1169: behavior of specific heat eventually appears and beyond that point, 

1170: the signature of specific heat changes from positive to negative. 

1171: That is, the potential energy dominates the kinetic energy($\lambda>0$) 

1172: and the quasi-equilibrium state ceases to exist for a system in contact with a 

1173: thermal bath (paper II). 

1174: Notice that even in this case, the stable quasi-equilibrium state  

1175: still exists for a system surrounded by an 

1176: adiabatic wall. On the other hand, for more larger radius, 

1177: while the curves with index $n<5$ abruptly terminate, 

1178: the trajectories with $n>5$ next reach at another 

1179: critical point $d\eta/d\lambda=0$, i.e., $\Cv=0$. Further, 

1180: they progressively change their direction and finally spiral 

1181: around a fixed point. 

1182: The appearance of the critical point $C_{\rm V} = 0$ is explained as follows. 

1183: While the inner part of the system keeps the 

1184: specific heat negative, the outer part seems to have positive 

1185: one. Thus, the heat current from inner to outer part causes 

1186: the raise of the temperature at both parts. For a system with the radius 

1187: $r_e$ smaller than certain critical value, the amount of the heat capacity 

1188: at outer part is  small so that the outer part easily catches up with the 

1189: increase of the inner-part temperature. 

1190: As increasing $r_e$, the fraction of the outer normal part grows up 

1191: and it eventually balances with the inner gravothermal part. 

1192: Thus, beyond the point characterized by the condition $C_{\rm V} = 0$, 

1193: no thermal balance is attainable and 

1194: the system becomes gravothermally unstable. This is even true in the 

1195: system surrounded by an adiabatic wall. 

1196: 

1197: 

1198: 

1199: 

1200: %%%%%%%%%%%

1201: >From these discussions, one can immediately verify that the 

1202: condition (\ref{eq: marginal_stability(1)})  represents the 

1203: marginal stability for a system confined in an adiabatic 

1204: wall, which exactly coincides with the previous result using standard 

1205: linear means. Hence, one concludes that the quasi-equilibrium structure obtained 

1206: from the new formalism does not alter the thermal properties in 

1207: micro-canonical ensemble case. On the other hand, the condition 

1208: (\ref{eq: marginal_stability(1)}) indicates the onset of thermodynamic 

1209: instability for a system in contact with a thermal bath (i.e., canonical 

1210: ensemble case), 

1211: which significantly differs from the results in old Tsallis formalism. 

1212: Of course, 

1213: this might be a natural consequence of the different choice of 

1214: statistical average, leading to the different definition $\Tphys$, 

1215: however, the appearance of instability in present case 

1216: might seem somewhat curious.  

1217: While the previous results indicate 

1218: the unstable state at the indices 

1219: $n>3$, consistent with the suggestion by Chavanis 

1220: \cite{Chavanis2002b}, 

1221: Fig. \ref{fig: eta_lambda} implies that 

1222: thermodynamic instability appears for stellar polytrope with 

1223: any value of the index $n$. 

1224: One might worry about whether the 

1225: present results rigorously match the stability analysis from the 

1226: variational problem. In next section, 

1227: to check the consistency of new Tsallis formalism, we develop the 

1228: stability analysis  based on the second variation of entropy and 

1229: free-energy. 

1230: 

1231: 

1232: %%%%%%%%%%%

1233: Finally, in Fig. \ref{fig: c_v}, varying the radius $\re$, 

1234: the specific heat $\Cv$ is plotted as a function of density 

1235: contrast, $D\equiv\rho_c/\rho_e$ for typical polytrope indices 

1236: with $n\geq 3/2$. Clearly, the critical point $|\Cv|\to\infty$ 

1237: marked by crosses exists in each case, while the marginal stability 

1238: $\Cv=0$ only appears when $n>5$ at a 

1239: certain high density contrast (c.f. Fig. 3 in paper II). 

1240: The numerical values indicated by arrows represent the 

1241: critical values $D_{\rm crit}$ evaluated at the point $\Cv=0$, which are 

1242: the same results as in Table 1 of paper I (see also Fig. 

1243: \ref{fig: lambda_crit}). 

1244: These behaviors can also be deduced from the energy-radius-mass relation 

1245: and the radius-mass-temperature relation. In Figs. \ref{fig: lambda_d} 

1246: and \ref{fig: eta_d}, using the expressions (\ref{eq: eta}) and 

1247: (\ref{eq: def_lambda}), the dimensionless values $\lambda$ and $\eta$ 

1248: for various polytrope indices are evaluated and plotted as a function of 

1249: density contrast, respectively.\footnote{

1250: Fig. \ref{fig: lambda_d} is essentially the same result as 

1251: in Fig. 2 of paper I.} 

1252: The Emden trajectories in $(\eta,\lambda)$ plane 

1253: (see Fig.\ref{fig: eta_lambda}) state that the marginal stability for 

1254: the system 

1255: in contact with a thermal bath, $|\Cv|\to\infty$ implies  

1256: the first turning point $d\eta/d\xxie=0$, or equivalently $d\eta/dD=0$. 

1257: Similarly, the marginal stability $\Cv=0$ represents the condition 

1258: $d\lambda/dD=0$. 

1259: >From Figs. \ref{fig: lambda_d} and \ref{fig: eta_d}, we readily estimate 

1260: the critical density contrast at the first turning points in each case, 

1261: which exactly coincide with the points marked by the arrows and the 

1262: crosses, respectively. Consistently, the turning point disappears 

1263: when $n<5$ in Fig. \ref{fig: lambda_d}, while it does always exist 

1264: independently of the polytrope index in Fig. \ref{fig: eta_d}. 

1265: %

1266: %

1267: %

1268: %

1269: %

1270: %

1271: %

1272: %

1273: %

1274: %

1275: %

1276: %

1277: %

1278: %

1279: %

1280: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1281: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1282: \section{Stability/instability criteria from the variational problems} 

1283: \label{sec: stability_variation}

1284: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1285: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1286: %

1287: %

1288: %

1289: %

1290: %

1291: %

1292: %

1293: %

1294: %

1295: %

1296: %

1297: %

1298: %

1299: %

1300: %

1301: %%%%%%%%%%%

1302: Previous section reveals the existence of two types of thermodynamic 

1303: instability.  

1304: Then, the marginal stability conditions (\ref{eq: marginal_stability(1)}) 

1305: and (\ref{eq: marginal_stability(2)}) are obtained for 

1306: a system confined in an adiabatic wall and for a system in contact with 

1307: a thermal bath, respectively. 

1308: In this section, in order to check the consistency of these results, 

1309: we reconsider the marginal stability criteria based on the variational 

1310: problem. 

1311: 

1312: 

1313: %%%%%%%%%%%

1314: According to the maximum entropy principle, 

1315: the stable quasi-equilibrium state for a system confined in an adiabatic wall 

1316: is only possible when the second variation of entropy around the 

1317: extremum state is negative, i.e.,  $\delta^2S_q<0$.  Similarly, the 

1318: stable quasi-equilibrium distribution surrounded by a thermal bath should 

1319: have minimum free-energy, indicating the positive value of the 

1320: second variation of free-energy, i.e., $\delta^2F_q>0$. Thus, 

1321: the condition $\delta^2S_q=0$ or $\delta^2F_q=0$ readily implies 

1322: the marginal stability in each case. 

1323: 

1324: 

1325: %%%%%%%%%%%%%%%%

1326: To begin with, let us write down the entropy of quasi-equilibrium state. 

1327: >From (\ref{eq: Zq-rho_relation}) and (\ref{eq: S_q_extreme}), 

1328: we have 

1329: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1330: \begin{equation}

1331:   S_q = \left(n-\frac{1}{2}\right)\,\left\{

1332:     c_n\,\,\left(\Tphys^{-(3/2)/(n-3/2)}\,\,W\right)^{-(n-3/2)/(n-1/2)}-1

1333: \,\,\right\}.  

1334: \label{eq: S_q_max}

1335: \end{equation}

1336: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1337: Here, just for convenience, we introduced the quantity $W$: 

1338: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1339: \begin{equation}

1340:   \label{eq: def_of_W}

1341:   W\,=\,\int\,d^3 x \,\,\rho^{1+1/n}. 

1342: \end{equation}

1343: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1344: Note that in deriving the equation (\ref{eq: S_q_max}), we used the 

1345: relation $\Tphys=K_n\,W$ from (\ref{eq: beta-P_relation}). 

1346: 

1347: 

1348: %%%%%%%%%%%

1349: Using the above expression (\ref{eq: S_q_max}), 

1350: we compute the variation of entropy up to the second order terms. 

1351: To be specific, we consider the density perturbation $\delta\rho(r)$ 

1352: around the quasi-equilibrium configuration, under keeping the total mass $M$ 

1353: and the radius $\re$ constant. Then the variation of entropy becomes  

1354: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1355: \begin{eqnarray}

1356: &  \delta S_q =& \left(n-\frac{1}{2}\right)\,\,\delta \Zq

1357: \nonumber\\

1358: &    =& \Zq\,\,\left[\frac{3}{2}\frac{\delta\Tphys}{\Tphys}-

1359:     n\frac{\delta W}{W}-

1360: \frac{3}{4}\,\frac{n-2}{n-1/2}\,\left(\frac{\delta\Tphys}{\Tphys}\right)^2

1361: \right.\nonumber\\

1362: &&~~~~~~~~~~~~~\left.

1363: +\frac{n(n-1/4)}{n-1/2}\,\left(\frac{\delta W}{W}\right)^2

1364: -\frac{3}{2}\frac{n}{n-1/2}\,\frac{\delta\Tphys}{\Tphys}\frac{\delta W}{W}

1365: \,\right]

1366: \label{eq: d2S_q}

1367: \end{eqnarray}

1368: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1369: As for the variation $\delta W$, we have

1370: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1371: \begin{eqnarray}

1372:   \delta W = \delta\left(\int d^3x\,\,\rho^{1+1/n}\right)

1373:   =\frac{n+1}{n}\,\int d^3x \left\{\,\delta\rho+\frac{1}{2n}\,

1374: \frac{(\delta\rho)^2}{\rho}\,\right\}\rho^{1/n}.

1375: \label{eq: d2W}

1376: \end{eqnarray}

1377: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1378: Also, we write down the variation of energy: 

1379: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1380: \begin{eqnarray}

1381: &&  \delta E = \delta\,\left(\frac{3}{2}\,\int d^3 x \,\,P + 

1382: \frac{1}{2}\,\int d^3 x\,\, \rho\,\Phi\right)

1383: \nonumber \\

1384: &&~~~~~~~~~~~~~~~~~~~~~=\frac{3}{2}\,\delta\Tphys\,+\,

1385: \frac{1}{2}\,\int d^3x\,\,\left(2\,\Phi\,\delta \rho\,+\,

1386: \delta\rho\,\delta\Phi\,\right), 

1387: \label{eq: d_E}

1388: \end{eqnarray}

1389: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1390: where we used the fact that 

1391: $\int d^3x\,\rho\delta\Phi = \int d^3x\,\Phi\delta\rho$. 

1392: 

1393: 

1394: %%%%%%%%%%%%%%%%

1395: Below, we separately analyze the variational problem in each case.   

1396: %

1397: %

1398: %

1399: %

1400: %

1401: %

1402: %

1403: %

1404: %

1405: %

1406: % ///////////////////////////////////////////////////////////// % 

1407: \subsection{Stability/instability criterion from the second 

1408: variation of entropy}

1409: \label{subsec: entropy}

1410: % ///////////////////////////////////////////////////////////// %

1411: %

1412: %

1413: %

1414: %

1415: %

1416: %

1417: %

1418: %

1419: %%%%%%%%%%%%

1420: Let us first consider the stability/instability condition for a 

1421: system confined in an adiabatic wall. In this case, 

1422: the conservation of total energy $E$ is always guaranteed and   

1423: we further put another constraint $\delta E=0$, which yields  

1424: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1425: \begin{eqnarray}

1426: \delta\Tphys = -\frac{1}{3}\,\,\int d^3x\, 

1427:         (2\Phi\,\delta\rho \,+\,\delta\rho\,\delta\Phi)

1428: \nonumber

1429: \end{eqnarray}

1430: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1431: from (\ref{eq: d_E}).

1432: Substituting the above equation into (\ref{eq: d2S_q}), 

1433: a straightforward calculation leads to 

1434: the variation of entropy up to the second order, 

1435: summarized as follows: 

1436: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1437: \begin{eqnarray}

1438:   \delta S_q = \delta^{(1)}S_q + \delta^{(2)}S_q\,\,\, ;

1439: \nonumber 

1440: \end{eqnarray}

1441: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1442: \begin{eqnarray}

1443:   \delta^{(1)} S_q \,&= &\,-\, \frac{\Zq}{\Tphys}\,\,

1444: \int d^3x\left\{\, \Phi + 

1445: (n+1)\,\frac{\Tphys}{W}\,\rho^{1/n}(x)\,\right\}\delta\rho,

1446: \label{eq: del(1)S_q}   

1447: \end{eqnarray}

1448: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1449: \begin{eqnarray}

1450:   \delta^{(2)} S_q \,&=&\,-\Zq\, \left[\,\,

1451: \int d^3x\left\{ \frac{1}{2\,\Tphys}\,\delta\rho\,\delta\Phi + 

1452: \frac{1}{2W}\,\frac{n+1}{n}\,\rho^{1/n-1}(\delta\rho)^2\,\right\}

1453: \right.

1454: \nonumber \\

1455: &+&\frac{1}{3\Tphys^2}\,\frac{n-2}{n-1/2}\,

1456: \left(\int d^3 x\,\Phi\,\delta\rho\right)^2 

1457: \nonumber \\

1458: && ~~-\frac{1}{W^2}\,\frac{(n+1)^2(n-1/4)}{n(n-1/2)}\,

1459: \left(\int d^3 x\,\delta\rho^{1/n}\,\delta\rho\,\right)^2 

1460: \nonumber \\

1461: && \left.~~~~~-\frac{1}{W\,\Tphys}\,\frac{n+1}{n-1/2}\,

1462: \left(\,\int d^3x\,\rho^{1/n}\delta\rho\,\right)

1463: \left(\,\int d^3x\,\Phi\,\delta\rho\,\right)\,\,\right].

1464: \label{eq: del(2)S_q}   

1465: \end{eqnarray}

1466: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1467: 

1468: 

1469: %%%%%%%%%%%%

1470: The resultant form of the second variation of entropy $\delta^{(2)} S_q$ 

1471: seems rather complicated, however, recalling  the fact that 

1472: the background solution of stellar polytropic distribution always satisfies 

1473: the condition $\delta^{(1)}S_q=0$, the last three terms in right-hand-side 

1474: of equation (\ref{eq: del(2)S_q}) can be rewritten in more compact form. 

1475: After some algebra, we obtain

1476: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1477: \begin{eqnarray}

1478: &&  \delta^{(2)}S_q = -\,\Zq\,\left[ \,\,\int d^3x \,

1479:   \left\{ \frac{1}{2\Tphys}\,\delta\rho\,\delta\Phi\,+\,

1480:   \frac{1}{2W}\,\frac{n+1}{n}\,\rho^{1/n-1}\,(\delta\rho)^2\right\}

1481: \right.

1482: \nonumber \\

1483: &&~~~~~~~~~~~~~

1484: \left. +\frac{W}{3\Tphys^2}\,\,\frac{n}{n-3/2}\,

1485: \left(\int d^3x\,\Phi\,+\,\frac{3}{2}\frac{n+1}{n}\,\frac{\Tphys}{W}\,\,

1486: \delta\rho\right)^2\,\,\right]. 

1487: \end{eqnarray}

1488: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

1489: Apart from the over-all positive constant, the above expression is 

1490: indeed the same equation as previously obtained in paper I (see Eq.(36) 

1491: in paper I, with the identification $\Tphys/W=\{(n-3/2)/(n+1)\}T$). 

1492: Thus, just following the same calculation as in paper I, 

1493: the stability/instability criterion from $\delta^{(2)}S_q=0$ can be 

1494: obtained, from which one can rigorously prove that the marginal stability 

1495: condition 

1496: for a system confined within an adiabatic wall exactly reproduces the equation 

1497: (\ref{eq: marginal_stability(1)}). As a result, for certain critical 

1498: values of dimensionless energy and density, $\lambda_{\rm crit}$ and 

1499: $D_{\rm crit}$, the onset of the gravothermal instability appears when 

1500: the stellar polytrope with index $n>5$ (see Figs.\ref{fig: lambda_d} 

1501: and \ref{fig: lambda_crit}).

1502: %

1503: %

1504: %

1505: %

1506: %

1507: %

1508: %

1509: %

1510: %

1511: %

1512: % ///////////////////////////////////////////////////////////// %

1513: \subsection{Stability/instability criterion from the second variation of 

1514: free-energy}

1515: \label{subsec: free-energy}

1516: % ///////////////////////////////////////////////////////////// %

1517: %

1518: %

1519: %

1520: %

1521: %

1522: %

1523: %

1524: %

1525: %

1526: %

1527: %%%%%%%%%%%%

1528: Now, turn to focus on the second variation of free-energy. 

1529:       In paper II, usual definition of free-energy 

1530: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1531: \begin{equation}

1532: F'_q = E - \beta^{-1} S_q 

1533: \label{eq: old_free-energy}

1534: \end{equation}

1535: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1536: has been used to investigate the marginal stability for a system surrounded 

1537: by a thermal bath. As several authors recently pointed out, the usual 

1538: definition 

1539: of free-energy becomes inadequate when the physical temperature cannot be 

1540: identified with the inverse of Lagrange multiplier $\beta$. 

1541: According to the modification of Clausius 

1542: relation (\ref{eq: modified_clausius}), the free-energy must be generalized 

1543: as follows \cite{AMPP2001,Abe2001}: 

1544: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1545: \begin{equation}

1546: F_q =  E - \Tphys\,\frac{1}{1-q}\ln\left\{1+(1-q)S_q\right\}.

1547: \label{eq: generalized_free-energy}

1548: \end{equation}

1549: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1550: %%%%%%%%%%%%

1551: Using this {\it generalized} free-energy,  we consider the 

1552: variation up to the second order, keeping the thermodynamic temperature 

1553: $\Tphys$ constant, not the total energy $E$:    

1554: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1555: \begin{eqnarray}

1556: \delta F_q = \delta E - \Tphys\,\frac{1}{1-q}\,

1557: \left\{

1558: \frac{\delta \Zq}{\Zq}-\frac{1}{2}\left(\frac{\delta \Zq}{\Zq}\right)^2

1559: \right\}.

1560: \nonumber

1561: \end{eqnarray}

1562: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1563: Substituting the equations (\ref{eq: d2S_q}), (\ref{eq: d2W}) 

1564: and (\ref{eq: d_E}) 

1565: into the above equation and keeping the terms up to the second order, 

1566: after some manipulation, one obtains

1567: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1568: \begin{eqnarray}

1569:   \delta F_q = \delta^{(1)}F_q + \delta^{(2)}F_q\,\,\,;

1570: \nonumber

1571: \end{eqnarray}

1572: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1573: \begin{eqnarray}

1574: \delta^{(1)} F_q &=& \int d^3x\left\{\Phi(x) + 

1575: (n+1)\frac{\Tphys}{W}\rho^{1/n}(x)\right\}\delta\rho,

1576: \label{eq: del(1)F_q}   

1577: \end{eqnarray}

1578: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1579: \begin{eqnarray}

1580: \delta^{(2)}F_q &=& 

1581: \frac{1}{2} \int d^3x \left\{ \delta\rho \delta\Phi + 

1582: \frac{n+1}{n}\, \frac{\Tphys}{W}\,\rho^{1/n-1}(\delta\rho)^2 \right\} 

1583: \nonumber\\

1584: &&

1585: ~~~~~~~~~~~~~~~~~~~~~~~

1586: -\frac{1}{2}\,\frac{\Tphys}{n}\,\left(\frac{n+1}{W}\right)^2\,

1587: \left(\int d^3x \rho^{1/n}\delta\rho\right)^2.

1588: \label{eq: del(2)F_q}   

1589: \end{eqnarray}

1590: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1591: Note that apart from the over-all factor, the first 

1592: variation $\delta^{(1)}F_q$ exactly coincides 

1593: with the first variation of entropy, $\delta^{(1)}S_q$, which ensures the 

1594: fact that extremum state of generalized free-energy is the same 

1595: quasi-equilibrium distribution as obtained from the maximum entropy 

1596: principle. Using the fact $\delta^{(1)}F_q=0$, equation (\ref{eq: del(1)F_q}) 

1597: is now rewritten with 

1598: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1599: \begin{eqnarray}

1600: &&\delta^{(2)}F_q = 

1601: \frac{1}{2} \int d^3x \left\{ \delta\rho \delta\Phi + 

1602: \frac{n+1}{n} \,\frac{\Tphys}{W}\, \rho^{1/n-1}(\delta\rho)^2 \right\} 

1603: \nonumber \\

1604: &&~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1605:  -\frac{1}{n\Tphys}\,

1606: \left(\int d^3x \Phi\delta\rho\right)^2.

1607: \label{eq: reduced_del(2)F_q}

1608: \end{eqnarray}

1609: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1610: Obviously, the above equation differs from the old result based on the 

1611: usual definition of free-energy (\ref{eq: old_free-energy}) (c.f. Eq.(51) 

1612: in paper II). This fact readily implies that the thermodynamic stability 

1613: has nothing to do with the dynamical stability, which apparently contradicts 

1614: with the recent claim by Chavanis \cite{Chavanis2002a,Chavanis2002b}. 

1615: Thus, in present case using normalized $q$-values, it seems non-trivial 

1616: whether the above equation indeed leads to the marginal stability condition 

1617: (\ref{eq: marginal_stability(2)}). 

1618: 

1619: 

1620: %%%%%%%%%%%%

1621: In what follows, restricting our attention to the radial 

1622: mode of density and potential 

1623: perturbations, we discuss the existence or the absence of perturbation 

1624: mode satisfying the equation $\delta^{(2)}F_q=0$. 

1625: Following the papers I and II, we introduce the new perturbed quantity: 

1626: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1627: \begin{equation}

1628:   \label{eq: Q}

1629:   \delta\rho(r)\equiv \frac{1}{4\pi r^2}\frac{dQ(r)}{dr}. 

1630: \end{equation}

1631: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1632: Then the mass conservation $\delta M=0$ 

1633: implies the boundary condition $Q(0)=Q(r_e)=0$. 

1634: Substituting (\ref{eq: Q}) into (\ref{eq: reduced_del(2)F_q}) 

1635: and repeating the integration by parts, 

1636: the second variation of generalized free-energy 

1637: is reduced to the following quadratic form: 

1638: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1639: \begin{equation}

1640:   \label{eq: del_2_F_}

1641:   \delta^{(2)}F_q = -\frac{1}{2} \int_{0}^{r_e} dr_1\,\int_{0}^{r_e} dr_2\,\,

1642: Q(r_1)\, \hat{K}(r_1, r_2) \, Q(r_2),

1643: \end{equation}

1644: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1645: where $\hat{K}(r_1, r_2)$ stands for the operator given by 

1646: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1647: \begin{eqnarray}

1648:   \label{eq: operator_K}

1649:   \hat{K}(r_1, r_2) &\equiv& \left\{ \frac{n+1}{n}\,\frac{d}{dr_1}

1650: \left(\frac{1}{4\pi r_1^2\rho(r_1)}\frac{P(r_1)}{\rho(r_1)}\frac{d}{dr_1}\right)+

1651: \frac{G}{r_1^2}\right\}\delta_D(r_1-r_2)

1652: \nonumber\\

1653: &&~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1654: +\frac{1}{n\Tphys}\,\frac{d\Phi(r_1)}{dr_1}\frac{d\Phi(r_2)}{dr_2}.

1655: \end{eqnarray}

1656: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1657: 

1658: 

1659: %%%%%%%%%%%%

1660: Thus, the problem reduces to the eigenvalue problem and 

1661: the marginal stability condition just corresponds to 

1662: the zero-eigenvalue problem: 

1663: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1664: \begin{eqnarray}

1665: \int_0^{\re}dr' \,\hat{K}(r,r')\,\,Q(r') =  0, 

1666: \end{eqnarray}

1667: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1668: which yields

1669: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1670: \begin{eqnarray}

1671:   \label{eq: zero-eigenvalue_eq}

1672: &&  \hat{L}[Q]\equiv 

1673: \left[\frac{n+1}{n}\,\frac{d}{dr}\left(\frac{1}{4\pi\,r^2\,\rho}\,\,

1674: \frac{P}{\rho}\,\frac{d}{dr}\right)+\frac{G}{r^2}\right]Q

1675: \nonumber\\

1676:   &&~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1677: =-\frac{1}{n\Tphys}\,\frac{G\,m(r)}{r^2}\,

1678:         \int_0^{r_e}dr' \frac{Gm(r')}{r'^2}\,Q(r').

1679: \end{eqnarray}

1680: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1681: 

1682: 

1683: %%%%%%%%%%%%

1684: To obtain the solution of the above zero-eigenvalue problem, 

1685: we follow the similar procedure in paper I. First recall the fact 

1686: that the following equations are satisfied:  

1687: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1688: \begin{equation}

1689: \hat{L}[4\pi\,r^3\,\rho(r)]=\frac{n-3}{n}\,\frac{Gm(r)}{r^2},

1690: ~~~~~~~~~~

1691: \hat{L}[m(r)]=\frac{n-1}{n}\,\frac{Gm(r)}{r^2}.

1692: \end{equation}

1693: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1694: Then, these relations allow us to put the ansatz of the solution, 

1695: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1696: \begin{equation}

1697:   Q(r) = c_1 \,4\pi r^3 \rho(r) + c_2 \, m(r), 

1698:   \label{eq: ansatz}

1699: \end{equation}

1700: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1701: and to determine the coefficients $c_1$ and $c_2$ by 

1702: substituting the ansatz into (\ref{eq: zero-eigenvalue_eq}):   

1703: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1704: \begin{equation}

1705:   \frac{n-3}{n}\,\,c_1 + \frac{n-1}{n}\,\, c_2  = \Lambda_n, 

1706:         \label{eq: condition_1}

1707: \end{equation}

1708: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1709: where we define 

1710: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1711: \begin{equation}

1712:   \label{eq: Lambda_n}

1713:   \Lambda_n \equiv -\frac{1}{n\Tphys}\,\int_0^{r_e}\,dr\,\,

1714:   \frac{Gm(r)}{r^2}\,\,Q(r).

1715: \end{equation}

1716: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1717: In addition to the above condition, we must further consider 

1718: the boundary conditions $Q(0)=Q(\re)=0$. 

1719: While the condition $Q(0)=0$ is automatically fulfilled,  

1720: the remaining condition $Q(r_e)=0$ requires 

1721: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1722: \begin{equation}

1723: \label{eq: condition_2}

1724:   c_1 \,4\pi r^3 \rho_e + c_2 \, M = 0.

1725: \end{equation}

1726: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1727: Hence, from (\ref{eq: condition_1}) and (\ref{eq: condition_2}),  

1728: the coefficients $c_1$ and $c_2$ are determined and are expressed 

1729: in terms of homology invariants:  

1730: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1731: \begin{equation}

1732:   \label{eq: c_1,c_2}

1733:   c_1 = \frac{n\,\,\Lambda_n}{n-3-(n-1)\,\ue},

1734: ~~~~~

1735:   c_2 = -\frac{n\,\,\ue\,\Lambda_n}{n-3-(n-1)\,\ue}.

1736: \end{equation}

1737: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1738: As noted in paper I, the above coefficients are not specified completely 

1739: because of the quantity $\Lambda_n$ in the coefficients, which depends on 

1740: the solution (\ref{eq: ansatz}) itself.  To eliminate this self-referential 

1741: structure, we directly evaluate the non-local term $\Lambda_n$, 

1742: leading to the consistency condition for the existence of solution 

1743: (\ref{eq: ansatz}). Substitution of the solution (\ref{eq: ansatz}) with 

1744: the coefficients (\ref{eq: c_1,c_2}) into the expression (\ref{eq: Lambda_n}) 

1745: yields

1746: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1747: \begin{eqnarray}

1748:   1= -\,\frac{\Tphys^{-1}}{n-3-(n-1)\ue}  \,\,\int_0^{r_e}\,dr\,\,

1749: \frac{Gm(r)}{r^2}\,\,\left\{4\pi r^3\rho(r)-\ue\,m(r)\right\}.

1750: \nonumber 

1751: \end{eqnarray}

1752: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1753: In the above equation, the integrals in the right-hand-side 

1754: can be evaluated using 

1755: the formulae (\ref{eq: formula_2}) and (\ref{eq: formula_3}) 

1756: listed in Appendix B. Further, the radius-mass-temperature relation  

1757: (\ref{eq: eta}) eliminates the dependence of $\Tphys$, leading to   

1758: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1759: \begin{eqnarray}

1760: &&  1 = \frac{\ve}{2\ue+\ve-(n+1)}\,\,\frac{1}{n-3-(n-1)\ue}

1761: \nonumber\\

1762: &&~~~~~~~~~~~~~\times \left[(n-5)\ue + (2\ue-1)

1763: \left\{3+(n+1)\left(\frac{\ue}{\ve}-\frac{3}{\ve}\right)\right\}\right].

1764: \nonumber

1765: \end{eqnarray}

1766: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1767: After some algebra, the above equation is rewritten with 

1768: the following quadratic form: 

1769: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1770: \begin{eqnarray}

1771:   0&=&4\ue^2 + 2\ue\ve-(n+9)\ue-\ve+n+1 

1772: \nonumber \\

1773: &=& k(\ue,\ve),

1774: \end{eqnarray}

1775: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1776: which coincides with the marginally stability criterion 

1777: (\ref{eq: marginal_stability(2)}) derived from 

1778: the condition for specific heat, $\Cv\to\infty$. 

1779: 

1780: 

1781: %%%%%%%%%%%%

1782: Therefore, unlike the first impression, we reach at 

1783: the fully satisfactory conclusion that the thermodynamic instability 

1784: inferred from the specific heat is consistently explained from the 

1785: variational problems. In other words, the application of 

1786: new Tsallis formalism to the stability of stellar quasi-equilibrium 

1787: system reveals a consistent thermodynamic structure of 

1788: self-gravitating system, as well as the existence of thermodynamic 

1789: instability. 

1790: %

1791: %

1792: %

1793: %

1794: %

1795: %

1796: %

1797: %

1798: %

1799: %

1800: %

1801: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1802: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1803: \section{Discussion \& conclusion}

1804: \label{sec: conclusion}

1805: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1806: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1807: %

1808: %

1809: %

1810: %

1811: %

1812: %%%%%%%%%%%%%

1813: In this paper, 

1814: we revisited the issues on the thermodynamic properties of 

1815: stellar self-gravitating system arising from the Tsallis entropy, 

1816: with a particular emphasis to the standard framework using 

1817: the normalized $q$-values. It then turns out that 

1818: the new extremum-entropy state essentially remains unchanged from the 

1819: previous study and is characterized by the 

1820: stellar polytrope, although the distribution function shows several 

1821: distinct properties. Taking these facts carefully, 

1822: the thermodynamic temperature of the extremum state was identified 

1823: through the modified Clausius relation and the specific heat was 

1824: evaluated explicitly. The detailed discussion on the behavior of 

1825: specific heat finally leads to the conclusion that the stability of the 

1826: system surrounded by a thermal wall(canonical case) is drastically 

1827: changed from previous result, 

1828: while the onset of gravothermal instability remains unchanged 

1829: for a system confined in an adiabatic wall(micro-canonical case).  

1830: The existence of these thermodynamic instabilities can also be deduced 

1831: from the variation of entropy and free-energy rigorously.  

1832: As a result, above the certain critical values 

1833: of $\lambda$ or $D$, 

1834: the thermodynamic instability appears at $n>5$ for a system confined in 

1835: an adiabatic wall. As for the system in contact with a thermal bath, 

1836: the onset of thermodynamic instability appears at 

1837: $(\eta_{\rm crit},D_{\rm crit})$ with any value of $n$. In 

1838: Figs. \ref{fig: lambda_crit} and \ref{fig: eta_crit}, 

1839: for the sake of the completeness, the 

1840: critical values evaluated from the conditions 

1841: (\ref{eq: marginal_stability(1)}) and (\ref{eq: marginal_stability(2)}) 

1842: are summarized, respectively. Note also the fact that the critical values in 

1843: Fig. \ref{fig: lambda_crit} exactly coincide with the previous results 

1844: (c.f. Table 1 in paper I).  

1845: 

1846: 

1847: %%%%%%%%%%%%%

1848: The one noticeable point in the present result is that the 

1849: macroscopic relation such as the radius-mass-temperature relation 

1850: or the specific heat as a function of density contrast can be 

1851: solely determined from the Emden solutions without any uncertainty. 

1852: Indeed, previous study using the standard linear means 

1853: seriously suffer from the residual dimensional parameter 

1854: $h=(l_0v_0)^3$, which must be practically disappeared from the 

1855: macroscopic description. In paper II, 

1856: the origin of this residual dependence has been addressed in 

1857: connection with the non-extensivity of the entropy. Although the 

1858: scaling relation appeared in the old results is simply 

1859: deduced from the asymptotic behavior of the Emden solutions, 

1860: the explicit $h$-dependence itself has originated from the 

1861: radius-mass-temperature relation. By contrast, in present case,  

1862: the radius-mass-temperature relation was 

1863: derived from the non-trivial relation (\ref{eq: beta-P_relation}), 

1864: in which no such $h$-dependence appears. 

1865: The resultant specific heat is thus obtained free from the residual 

1866: dependence, which is a natural outcome of the 

1867: new framework using the normalized $q$-values. Therefore,  

1868: it seems likely that the new formalism provides a better 

1869: characterization for non-extensive meta-equilibrium state. 

1870: 

1871: 

1872: %%%%%%%%%%%%%

1873: In fact, for several application of the new formalism, 

1874: the equation of state for non-extensive system has been 

1875: found to be similar to the result from the ordinary extensive 

1876: thermodynamics. For instance, the equation of state for 

1877: the classical gas model without interaction reduces to the 

1878: ideal gas sate even in the power-law nature of velocity 

1879: distribution \cite{AMPP2001}. Indeed, this fact can be clearly seen 

1880: in our case neglecting the gravity. Recalling that 

1881: both the density and the pressure become {\it homogeneous} 

1882: in the limit $G\to0$,  the equation (\ref{eq: Tphys_P}) immediately 

1883: reduces to the relation $\Tphys=PV$, i.e., $P\propto \rho\Tphys $, 

1884: where $V$ denotes the volume of the system. 

1885: This result apparently seems contradiction with the polytropic relation 

1886: (\ref{eq: polytrope}), however, 

1887: it turns out that the dimensional 

1888: constant $K_n$ immediately yields the relation $K_n=P/\rho^{1+1/n}$ 

1889: from the definition (\ref{eq: K_n}) and thereby 

1890: the polytropic relation makes no sense in the limit $G\to0$. 

1891: Thus, the stellar polytropic system using normalized $q$-values 

1892: successfully recovers the ideal gas limit, in contrast to the old results 

1893: (see Appendix B in paper II). In other words, if the interaction is turned 

1894: on, the equation of state for the ideal gas is no longer valid and 

1895: instead the polytropic relation holds. 

1896: In Sec.\ref{sec: polytrope},  the polytropic relation was 

1897: derived without assuming any specific choice of the gravitational 

1898: potential. This in turn suggests that at least in the mean-field treatment, 

1899: the polytropic relation is a common feature in presence of long-range 

1900: interaction and it generally holds for non-extensive quasi-equilibrium system.   

1901: 

1902: 

1903: %%%%%%%%%%%%%%

1904: Finally, we have firmly confirmed that the stellar 

1905: polytropic system can be consistently characterized as a plausible 

1906: meta-equilibrium state in the new framework of the non-extensive 

1907: thermostatistics, as well as in the old formalism. At present, however, 

1908: one cannot rigorously discriminate the correct and the applicable formalism 

1909: among them (there might be another possibility that both the formalism become 

1910: indeed correct depending on the situations). While the new framework 

1911: by means of the normalized $q$-values has theoretically desirable 

1912: properties as mentioned above, the results in old formalism might 

1913: provide an interesting connection with dynamical instability of 

1914: gaseous system, as has been suggested by Chavanis 

1915: \cite{Chavanis2002a,Chavanis2002b}. In paper II, 

1916: the thermodynamic instability of stellar polytropic system in the canonical 

1917: ensemble case is shown to exist at the polytrope indices $n>3$, which 

1918: is exactly the same condition as derived from the gaseous system.  

1919: On the other hand, in the new formalism, there is no such 

1920: correspondence, since the instability is completely different from 

1921: that of the gaseous system (see Sec.\ref{subsec: free-energy}). 

1922: This means that the choice of the statistical average is crucial for 

1923: the thermodynamic properties and the physical reason for this discrepancy 

1924: should be further clarified. In any case, in order to pursue 

1925: the physical reality of the non-extensive thermostatistics, 

1926: the limitation of thermodynamical approach is now apparent and the detailed 

1927: kinematical study based on the Boltzmann or the Fokker-Planck equation 

1928: provides an deep insight into the thermal transport property. 

1929: In the light of this, the long-term stellar dynamical evolution by $N$-body 

1930: simulation also becomes useful and the analysis is now in progress. 

1931: The results will be presented elsewhere. 

1932:   

1933: 

1934: %

1935: %

1936: %

1937: %

1938: %

1939: %

1940: %

1941: 

1942: 

1943: \bigskip

1944: We are grateful to Prof. S. Abe for invaluable discussions and comments. 

1945: This work is supported in part by the Grant-in-Aid for Scientific Research 

1946: of Japan Society for the Promotion of Science (No.$1470157$). 

1947: %

1948: %

1949: %

1950: %

1951: %

1952: %

1953: %

1954: %

1955: %

1956: %

1957: %

1958: %

1959: %

1960: \clearpage

1961: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1962: %%%%%%%%%%%%%%%%%%%%%%%%%%   Appendix   %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1963: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1964: %

1965: %

1966: %

1967: %

1968: %

1969: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1970: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1971: \section*{Appendix A: Extremum-state of Tsallis entropy from the OLM method}

1972: \label{appen_A}

1973: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1974: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1975: %

1976: %

1977: %

1978: %

1979: %

1980: Here, owing to the maximum Tsallis entropy principle, 

1981: we derive an extremum state of the stellar quasi-equilibrium distribution 

1982: by means of the OLM method. According to Ref.\cite{MNPP2000}, 

1983: with a help of the relation (\ref{eq: escort_dist}), 

1984: the variational problem that extremizes the entropy under the 

1985: energy constraint becomes 

1986: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1987: \begin{equation}

1988:   \delta\left[S_q\,-\alpha\left\{\int d^6\bftau\,\,p-1\right\}

1989: -\beta \,\int d^6\bftau\left\{ M\left(

1990: \frac{1}{2}v^2+\frac{1}{2}\Phi\right)-E\right\}\,p^q\,

1991: \right]=0.

1992: \end{equation}

1993: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1994: Then the variation with respect to probability function $p(\xx,\vv)$ 

1995: leads to 

1996: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1997: \begin{eqnarray}

1998: &&  \int d^6\bftau\left[-\frac{1}{q-1}(q\,p^{q-1}-1)\,

1999:     -\alpha-\beta \,M\left( 

2000: \frac{1}{2}\,v^2-\frac{E}{M}\right)\,q\,p^{q-1}\,\right]\delta p

2001: \nonumber

2002: \\

2003: && ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

2004: -\frac{1}{2}\,\beta\,

2005: \delta\left(\Zq\,\int d^6\bftau\,\Phi\,f\right)=0.

2006: \label{appendix_A: variation_OLM_1}

2007: \end{eqnarray}

2008: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2009: In the above equation, the last term in the left-hand-side is 

2010: rewritten with 

2011: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2012: \begin{eqnarray}

2013: \delta\left( \Zq\int d^6\bftau \,\Phi\, f\right)&=&

2014: \delta\Zq

2015:  \int d^6\bftau\,\Phi\,f+2\,\Zq\int d^6\bftau\,\Phi\delta f

2016: \nonumber \\

2017: &=& q\int d^6\bftau\left\{2M\,\Phi-\int d^6\bftau'\,\Phi(\xx')

2018: f(\xx',\vv')\right\}p^{q-1}\,\delta p, 

2019: \label{appendix_A: variation_phi_f}

2020: \end{eqnarray}

2021: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2022: with a help of the equation (\ref{eq: delta_f}). In the last line, 

2023: we have exchanged the role of the variables 

2024: $(\xx,\vv)\leftrightarrow(\xx',\vv')$.  Substituting 

2025: (\ref{appendix_A: variation_phi_f})  into 

2026: (\ref{appendix_A: variation_OLM_1}), we arrive at 

2027: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2028: \begin{eqnarray}

2029:   \int d^6\bftau \left[-\frac{1}{q-1}\left(q\,p^{q-1}-1\right)

2030: -\alpha-\beta \,M\,q\,p^{q-1}\left(\frac{1}{2}\,v^2+\Phi-\varepsilon 

2031: \right)\right]\delta p = 0, 

2032: \nonumber

2033: \end{eqnarray}

2034: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2035: where the quantity $\varepsilon$ is given by (\ref{eq: epsilon}). 

2036: Recalling the fact that the above equation holds independently of the 

2037: choice of the variation  $\delta p$, we obtain 

2038: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2039: \begin{eqnarray}

2040: -\frac{1}{q-1}\left(q\,p^{q-1}-1\right)

2041: -\alpha-\beta \,M\,q\,p^{q-1}\left(\frac{1}{2}\,v^2+\Phi-\varepsilon 

2042: \right)=0, 

2043: \nonumber

2044: \end{eqnarray}

2045: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2046: which leads to the power-law distribution. The resultant 

2047: expressions for the one-particle distribution function 

2048: (escort distribution) is summarized in 

2049: equations (\ref{eq: extremum_state}) and (\ref{eq: A_Phi0}), 

2050: with the identification $\Zq'=1$.  

2051: %

2052: %

2053: %

2054: %

2055: %

2056: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2057: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2058: \section*{Appendix B: Stellar polytropic system characterized by 

2059:         Emden solutions}

2060: \label{appen_B}

2061: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2062: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2063: %

2064: %

2065: %

2066: %

2067: %

2068: %

2069: %

2070: %

2071: %

2072: In this appendix, we briefly describe the equilibrium 

2073: configuration of stellar polytrope in the spherically symmetric case,   

2074: together with some useful formulae which has been used in 

2075: the main analysis of 

2076: section \ref{sec: thermodynamics} and \ref{sec: stability_variation}. 

2077: 

2078: 

2079: First notice that the one-particle distribution function 

2080: (\ref{eq: extremum_state}) does not yet completely specify 

2081: the equilibrium configuration, 

2082: due to the presence of gravitational potential which implicitly 

2083: depends on the distribution function itself. Hence, we need to 

2084: specify the gravitational potential or density profile.

2085: >From the gravitational potential (\ref{eq: def_Phi}), it reads 

2086: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2087: \begin{equation}

2088: \label{eq: poisson_eq}  

2089:  \frac{1}{r^2}\frac{d}{dr}\left\{r^2\frac{d\Phi(r)}{dr}\right\}=

2090:   4\pi G \rho(r).

2091: \end{equation}

2092: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2093: Combining the above equation with (\ref{eq: density}), we 

2094: obtain the ordinary differential equation for $\Phi$. 

2095: Alternatively, a set of equations which represent 

2096: the hydrostatic equilibrium are derived using 

2097: (\ref{eq: poisson_eq}), (\ref{eq: density}) and (\ref{eq: pressure}):  

2098: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2099: \begin{eqnarray}

2100: & \frac{dP(r)}{dr}\,=&\,\,-\frac{Gm(r)}{r^2}\,\rho(r), 

2101: \label{eq: hydro_1}

2102: \\

2103: & \frac{dm(r)}{dr}\,=&\,\,4\pi\rho(r)\,r^2.  

2104: \label{eq: hydro_2}

2105: \end{eqnarray}

2106: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2107: The quantity $m(r)$ denotes the mass evaluated at the radius $r$ 

2108: inside the wall. Denoting the central density and pressure by 

2109: $\rho_c$ and $P_c$, we then introduce the dimensionless quantities: 

2110: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2111: \begin{equation}

2112: \label{eq: dimensionless}

2113:  \rho=\rho_c\,\left[\theta(\xi)\right]^n,\,\,\,\,\,\,

2114: r=\left\{\frac{(n+1)P_c}{4\pi G\rho_c^2}\right\}^{1/2}\,\xi, 

2115: \end{equation}

2116: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2117: which yields the following ordinary differential equation: 

2118: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2119: \begin{equation}

2120:  \theta''+\frac{2}{\xi}\theta'+\theta^n=0,

2121: \label{eq: Lane-emden_eq}

2122: \end{equation}

2123: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2124: where prime denotes the derivative with respect to $\xi$. 

2125: The quantities $\rho_c$ and $P_c$ in (\ref{eq: dimensionless}) 

2126: are the density and the pressure at $r=0$, respectively. 

2127: To obtain the physically relevant solution of (\ref{eq: Lane-emden_eq}), 

2128: we put the following boundary condition:

2129: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2130: \begin{equation}

2131:  \theta(0)=1, \,\,\,\,\,\,\,\theta'(0)=0.    

2132: \label{eq: boundary}

2133: \end{equation}

2134: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2135: A family of solutions satisfying (\ref{eq: boundary}) is referred to 

2136: as the {\it Emden solution}, which is well-known in the subject of 

2137: stellar structure (e.g., see Chap.IV of Ref.\cite{Chandra1939}).  

2138: 

2139: 

2140: %%%%%%%%%%%%%

2141: To characterize the equilibrium properties of Emden solutions,  

2142: it is convenient to introduce the following 

2143: set of variables, referred to as homology invariants 

2144: \cite{Chandra1939,KW1990}: 

2145: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2146: \begin{eqnarray}

2147:  u &\equiv& \frac{d\ln m(r)}{d\ln r}=

2148: \frac{4\pi r^3\rho(r)}{m(r)}=-\frac{\xi\theta^n}{\theta'},

2149: \label{eq: def_u}

2150: \\

2151: \nonumber\\

2152:  v &\equiv&  - \frac{d\ln P(r)}{d\ln r}=

2153: \frac{\rho(r)}{P(r)}\,\,\frac{Gm(r)}{r}

2154: =-(n+1)\frac{\xi\theta'}{\theta},  

2155: \label{eq: def_v}

2156: \end{eqnarray}

2157: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2158: which reduce the degree of equation (\ref{eq: Lane-emden_eq}) 

2159: from two to one. The derivative of these variables with 

2160: respect to $\xi$ becomes

2161: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2162: \begin{equation}

2163: \frac{du}{d\xi} = \left(3-u-\frac{n}{n+1}\,v \right)\,\frac{u}{\xi},

2164: ~~~~~~~~

2165: \frac{dv}{d\xi} = \left(-1+u+\frac{1}{n+1}\,v \right)\,\frac{v}{\xi}.

2166: \label{eq: d(u,v)/dxi}

2167: \end{equation}

2168: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2169: Equations (\ref{eq: Lane-emden_eq}) can thus be re-written with 

2170: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2171: \begin{equation}

2172:  \label{eq: uv_eqn}

2173:   \frac{u}{v}\,\frac{dv}{du}=\frac{(n+1)(u-1)+v}{(n+1)(3-u)-nv}.   

2174: \end{equation}

2175: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2176: The corresponding boundary condition to (\ref{eq: boundary}) 

2177: becomes  $(u,v)=(3,0)$.  

2178: 

2179: 

2180: Now, utilizing the above homology invariants, 

2181: we list some useful formulae which can be derived 

2182: from the hydrostatic equations (\ref{eq: hydro_1}) and (\ref{eq: hydro_2}) 

2183: (see also Appendix A in paper I): 

2184: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2185: \begin{eqnarray}

2186: \int_0^{r_e}dr \,4\pi r^2\,P(r)  &=& - \frac{1}{n-5}

2187: \left\{8\pi\,r_e^3P_e-(n+1)\frac{MP_e}{\rho_e}+\frac{GM^2}{r_e}\right\}

2188: \nonumber \\

2189: &=& -\frac{1}{n-5}\,\,\frac{GM^2}{\re}\,\,

2190: \left(2\,\frac{\ue}{\ve}-\frac{n+1}{\ve}+1\right),

2191: \label{eq: formula_1}

2192: \end{eqnarray}

2193: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2194: \begin{eqnarray}

2195: \int_0^{r_e} dr\,\frac{Gm(r)}{r^2}\,4\pi r^3 \rho(r) 

2196: &=& -4\pi r_e^3 P_e + 3 \int_0^{r_e}dr \,4\pi r^2\,P(r), 

2197: \nonumber \\

2198: &=& -\frac{n+1}{n-5}\,\,\frac{GM^2}{\re}\,\,

2199: \left\{ \frac{\ue}{\ve}-\frac{3}{\ve}+\frac{3}{n+1}\,\right\},

2200: \label{eq: formula_2}

2201: \end{eqnarray}

2202: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2203: \begin{eqnarray}

2204: \int_0^{r_e}  dr\,\frac{G\,m^2(r)}{r^2} 

2205: &=& -\frac{GM^2}{r_e} \, - \,8\pi r_e^3P_e \,+ \,6 \int_0^{r_e}dr \,

2206: 4\pi r^2\,P(r),

2207: \nonumber \\

2208: &=& -\frac{n+1}{n-5}\,\,\frac{GM^2}{\re}\,\,

2209: \left\{ 2\left(\frac{\ue}{\ve}-\frac{3}{\ve}\right)+1\,\right\}.

2210: \label{eq: formula_3}

2211: \end{eqnarray}

2212: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2213: 

2214: 

2215: Finally, using these formulae, we evaluate the total energy of the 

2216: stellar system in terms of the variables at the edge $\re$: 

2217: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2218: \begin{eqnarray}

2219:   E= K+U &=& \frac{3}{2}\,\int_{0}^{\re}dr\,4\pi r^2\,P(r) 

2220:   -\int_0^{\re}\,dt\,\frac{Gm(r)}{r}\,\frac{dm}{dr}

2221: \nonumber \\

2222: &=& -\frac{1}{n-5}\,

2223: \left[\,\frac{3}{2}\left\{ \frac{GM^2}{\re}-(n+1)\frac{MP_e}{\rho_e}\right\}

2224: +(n-2)\,4\pi\,\re^3\,P_e\,\right], 

2225: \nonumber 

2226: \end{eqnarray}

2227: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2228: which can be re-expressed in terms of the homology invariants: 

2229: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2230: \begin{eqnarray}

2231: E &=& \frac{1}{n-5}\,\frac{GM^2}{\re}\,\,

2232: \left[\,\frac{3}{2}\left\{1-\frac{n+1}{\ve}\right\}+(n-2)\frac{\ue}{\ve}

2233: \,\right].

2234:   \label{eq: energy_uv}

2235: \end{eqnarray}

2236: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2237: %

2238: %

2239: %

2240: %

2241: %

2242: %

2243: %

2244: %

2245: %

2246: \clearpage

2247: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2248: %%%%%%%%%%%%%%%%%%%%%%%%%       References     %%%%%%%%%%%%%%%%%%%%%%%%%%%%

2249: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2250: %

2251: %

2252: %

2253: %

2254: %

2255: %

2256: \begin{thebibliography}{00}

2257: %

2258: %

2259: %

2260: \bibitem{BT1987} J. Binney, S. Tremaine, {\it Galactic Dynamics} 

2261: (Princeton Univ. Press, Princeton, 1987). 

2262: \bibitem{EHI1987} R. Elson, P. Hut and S. Inagaki, Ann. Rev. Astron. 

2263: Astrophys. 25 (1987) 565.

2264: \bibitem{MH1997} G. Meylan, D.C. Heggie, Astron.Astrophys.Rev. 8 (1997) 1.

2265: \bibitem{Antonov1962} V.A. Antonov, {\it Vest. Leningrad Gros. Univ.}, 

2266: 7 (1962) 135 (English transl. in {\it IAU Symposium 113, Dynamics of 

2267: Globular Clusters}, ed. J. Goodman and P. Hut [Dordrecht: Reidel], 

2268: pp. 525--540 [1985]) 

2269: \bibitem{LW1968} D. Lynden-Bell, R. Wood, Mon.Not.R.Astr.Soc. 138 (1968) 495.

2270: \bibitem{HS1978} I. Hachisu, D. Sugimoto, Prog. Theor. Phys. 60 (1978) 393.

2271: \bibitem{Padmanabhan1989} T. Padmanabhan, Astrophys.J.Suppl. 71 (1989) 651. 

2272: \bibitem{Padmanabhan1990} T. Padmanabhan, Phys.Rep. 188 (1990) 285.

2273: \bibitem{T1988} C. Tsallis, J.Stat.Phys. 52 (1988) 479.

2274: \bibitem{T1999} C. Tsallis, Braz. J. Phys. 29 (1999) 1.

2275: \bibitem{AO2001} S. Abe, Y. Okamoto (Eds.), {\it Nonextensive Statistical Mechanics and Its Applications} (Springer, Berlin, 2001) 

2276: \bibitem{PP1993} A. R. Plastino, A. Plastino, Phys.Lett. A 174 (1993) 384. 

2277: \bibitem{PP1999} A. R. Plastino, A. Plastino, Braz. J. Phys. 29 (1999) 79.  

2278: \bibitem{TS2002a} A. Taruya, M. Sakagami, Physica A 307 (2002) 185 (paper I).

2279: \bibitem{TS2002b} A. Taruya, M. Sakagami, Physica A (2003) in press 

2280: (cond-mat/0204315, paper II).

2281: \bibitem{TMP1998} C. Tsallis, R.S. Mendes, A. R. Plastino, 

2282: Physica A 261 (1998) 534.

2283: \bibitem{MNPP2000} S. Mart\'inez, F. Nicol\'as, F. Pennini, A. Plastino, 

2284: Physica A 286 (2000) 489.

2285: \bibitem{AR2000} S. Abe, A. K. Rajagopal, Phys.Lett. A 272 (2000) 345; 

2286:   J. Phys. A 33 (2000) 8733; Europhys. Lett. 52 (2000) 610. 

2287: \bibitem{CT1991} E. M. F. Curado, C. Tsallis, J.Phys.A 24 (1991) L69. 

2288: \bibitem{PP1997} A. Plastino, A. R. Plastino, Phys.Lett. A 226 (1997) 257.

2289: \bibitem{Chandra1939} S. Chandrasekhar, {\it Introduction to the Study of

2290:         Stellar Structure} (New York, Dover, 1939) 

2291: \bibitem{KW1990} R. Kippenhahn, A. Weigert, {\it Stellar Structure 

2292: and Evolution} (Springer, Berlin, 1990)

2293: \bibitem{LyndenBell1999} D. Lynden-Bell, Physica A 263 (1999) 293.

2294: \bibitem{AMPP2001} S. Abe, S. Mart\'inez, F. Pennini, A. Plastino, 

2295: Phys.Lett. A 281 (2001) 126.

2296: \bibitem{Abe2001} S. Abe, Physica A 300 (2001) 417.

2297: \bibitem{MPP2001} S. Mart\'inez, F. Pennini, A. Plastino, 

2298: Physica A 295 (2001) 416. 

2299: \bibitem{Chavanis2002a} P. H. Chavanis, Astron. \& Astrophys. 381 (2002) 340.

2300: \bibitem{Chavanis2002b} P. H. Chavanis, Astron. \& Astrophys. 386 (2002) 732.

2301: %

2302: %

2303: %

2304: \end{thebibliography}

2305: %

2306: %

2307: %

2308: %

2309: %

2310: %

2311: %

2312: %

2313: %

2314: %

2315: %

2316: %

2317: %

2318: %

2319: %

2320: %

2321: %

2322: %

2323: %

2324: %

2325: \clearpage

2326: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2327: %%%%%%%%%%%%%%%%%%%%       Figures        %%%%%%%%%%%%%%%%%%%%%%%%%%%%

2328: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2329: %

2330: %

2331: %

2332: %

2333: %

2334: %

2335: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

2336: \begin{figure}

2337:   \begin{center}

2338:     \includegraphics*[width=14cm]{eta_lambda.eps}

2339:     \caption{Trajectory of Emden solutions in $(\eta,\lambda)$-plane. }

2340:     \label{fig: eta_lambda}

2341:   \end{center}

2342: \end{figure}

2343: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

2344: %

2345: %

2346: %

2347: %

2348: %

2349: %

2350: %

2351: %

2352: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

2353: \begin{figure}

2354:   \begin{center}

2355:   \includegraphics*[width=14cm]{c_v.eps}

2356:   \end{center}

2357:     \caption{Specific heat as a function of density contrast 

2358:         $D(=\rho_c/\rho_e)$ for various polytrope indices. }

2359:     \label{fig: c_v}

2360: \end{figure}

2361: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

2362: %

2363: %

2364: %

2365: %

2366: %

2367: %

2368: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

2369: \begin{figure}

2370:   \begin{center}

2371:     \includegraphics*[width=10.5cm]{lambda_d.eps}

2372:     \caption{Energy-radius-mass relation as a function of density 

2373:       contrast.}

2374:     \label{fig: lambda_d}

2375: 

2376: %\vspace*{0.8cm}

2377: 

2378:     \includegraphics*[width=10.5cm]{eta_d.eps}

2379:     \caption{Radius-mass-temperature relation as a function of 

2380:       density contrast.}

2381:     \label{fig: eta_d}

2382:   \end{center}

2383: \end{figure}

2384: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

2385: %

2386: %

2387: %

2388: %

2389: %

2390: %

2391: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

2392: \begin{figure}

2393:   \begin{center}

2394: \vspace*{-1.0cm}

2395: \hspace*{0.6cm}

2396: \vspace*{-1.0cm}

2397:     \includegraphics*[width=12.5cm]{lambda_crit_embed.eps}

2398:     \caption{Critical values $(D_{\rm crit}, \lambda_{\rm crit})$ 

2399:       for a system confined in an adiabatic wall.} 

2400:     \label{fig: lambda_crit}

2401: 

2402: \vspace*{+0.8cm}

2403: 

2404:     \includegraphics*[width=10cm]{eta_crit.eps}

2405:     \caption{Critical values $(D_{\rm crit}, \eta_{\rm crit})$ 

2406:       for a system in contact with a thermal bath. }

2407:     \label{fig: eta_crit}

2408:   \end{center}

2409: \end{figure}

2410: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

2411: %

2412: %

2413: %

2414: %

2415: %

2416: %

2417: %

2418: %

2419: %

2420: %

2421: %

2422: %

2423: %

2424: %

2425: %

2426: %

2427: %

2428: %

2429: %

2430: %

2431: %

2432: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2433: \end{document}

2434: