1: %\documentstyle[prl,aps,preprint,psfig]{revtex4}

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

3: 

4: \documentclass[floats,aps,showpacs,twocolumn]{revtex4}

5: 

6: \usepackage[dvips]{graphicx}

7: \usepackage{amsfonts}

8: \usepackage{amssymb}

9: \usepackage{float}

10: 

11: \newcommand{\PSImagx}[2]{\includegraphics[width=#2]{#1}} % command from Arnd

12: \newcommand{\tmin}{\tau_{\text{min}}}

13: \newcommand{\leqa}{\stackrel{<}{\scriptstyle \sim}}

14: \newcommand{\geqa}{\stackrel{>}{\scriptstyle \sim}}

15: \newcommand{\ef}{\epsilon_{\text {\tiny F}}}

16: \newcommand{\ecb}{\overline{\cal E}}

17: \newcommand{\ect}{\widetilde{\cal E}}

18: \newcommand{\rmb}{\rho_{\text {\tiny MB}}}

19: \newcommand{\rmbb}{\overline{\rho}_{\text {\tiny MB}}}	

20: \newcommand{\rmbt}{\widetilde{\rho}_{\text {\tiny MB}}}

21: \newcommand{\varkappab}{\overline{\varkappa}}

22: \newcommand{\rhob}{\overline{\rho}}

23: \newcommand{\rhot}{\widetilde{\rho}}

24: \newcommand{\sca}{{\cal S}}

25: \newcommand{\scab}{\overline{{\cal S}}}

26: \newcommand{\scat}{\widetilde{{\cal S}}}

27: \newcommand{\taut}{\tau_{\scriptscriptstyle T}}

28: 

29: \newlength{\figwidth}

30: \setcounter{equation}{0}

31: \setlength{\figwidth}{3.0in}

32: 

33: 

34: \begin{document}

35: 

36: \title{Level density of a Fermi gas: average growth and fluctuations}

37: 

38: \author{Patricio Leboeuf and J\'er\^ome Roccia}

39: 

40: \affiliation{Laboratoire de Physique Th\'eorique et Mod\`eles

41: Statistiques$^*$, B\^at. 100, \\ Universit\'e de Paris-Sud, 91405 Orsay Cedex,

42: France\\}

43: 

44: %\date{\today, \RightText}

45: \date{\today}

46: 

47: \begin{abstract}

48: We compute the level density of a two--component Fermi gas as a function

49: of the number of particles, angular momentum and excitation energy. The

50: result includes smooth low--energy corrections to the leading Bethe

51: term (connected to a generalization of the partition problem and

52: Hardy--Ramanujan formula) plus oscillatory corrections that describe shell

53: effects. When applied to nuclear level densities, the theory provides a

54: unified formulation valid from low--lying states up to levels entering the

55: continuum. The comparison with experimental data from neutron resonances gives

56: excellent results.

57: \end{abstract}

58: 

59: \pacs{03.75.Ss, 21.10.Ma, 24.60.-k}

60: 

61: \maketitle

62: 

63: Many physical properties of an interacting Fermi gas depend on the number of

64: available states at a given energy, like for instance the optical and

65: electrical response of solids, or the reaction rates in nuclear processes. The

66: most common framework to compute the many--body (MB) density of states (DOS)

67: is a mean--field approximation, where each (quasi-)particle moves

68: independently in an average self--consistent potential. In this case, the

69: energy of the Fermi gas is expressed as the sum of the occupied

70: single--particle (SP) energies. The computation of the MB DOS is thus reduced

71: to a combinatorial problem: to count the different ways into which the energy

72: can be distributed among the particles. A first answer to this question was

73: given by Bethe \cite{bethe}, who showed that at high excitation energies $Q$

74: (compared to the SP spacing at Fermi energy $\ef$) and for two types of

75: fermions (protons and neutrons), the MB density grows like $\rmb (Q) \sim

76: \exp\left(2 \sqrt{a Q}\right)/Q^{5/4}$. The Fermi gas parameter $a=\pi^2 \rhob

77: (\ef)/6$ depends only on the average SP DOS at $\ef$, $\rhob (\ef)$.

78: 

79: In practice the parameter $a$ is often used as a fitting parameter. For a

80: given excitation energy $Q$ and particle number $A$, $a(Q,A)$ is extracted

81: from the available experimental data. In this way, important deviations from

82: the independent particle model predictions are observed. Though there are

83: certainly effects that are beyond that model, our purpose here is to show that

84: a detailed treatment is able to describe features of individual systems with

85: good accuracy, therefore providing a solid theoretical basis for

86: extrapolations to unknown regions and for improvements. Generalizing the

87: results obtained in Ref.\cite{lmr} to a two--component system of given angular

88: momentum, we show that Bethe's result can be viewed as the first (smooth) term

89: of an expansion. The corrections to that term do not enter as corrections of

90: the $a$--parameter (as was often assumed in the past), but simply as

91: additional terms in the exponential. A first series of terms are smooth in $A$

92: and $Q$, and provide higher--order (in inverse powers of the excitation

93: energy) corrections. Keeping only the first correction generates a uniform

94: expression, which cancel the divergence produce by the $Q^{-5/4}$ term at low

95: energies, and therefore make unnecessary the use of composite models (\`a la

96: Gilbert--Cameron). On top of the smooth contributions are oscillatory terms,

97: that describe density fluctuations as $A$ varies. These are shell effects,

98: which turn out to be related to the fluctuations of the total energy of the

99: system. A detailed description of these fluctuations and of the relevant

100: energy scales is provided. Finally, a comparison of the results with the

101: nuclear level density at neutron threshold is made. With a few adjustable

102: parameters, a very good overall agreement is obtained, with a relative error

103: $\leqa$ 10\% for the logarithm of the density of the 295 nuclei analyzed.

104: 

105: From a theoretical point of view, Ref.\cite{lmr} and the present work may be

106: viewed as a generalization to MB systems of the theory developed to describe

107: the SP DOS \cite{bh,bb,gutz}. Compared to direct generalizations of the SP

108: Weyl's law and of semiclassical trace formulas \cite{trace}, the statistical

109: approach employed here has the advantage of leading to simple and physically

110: meaningful results.

111: 

112: The DOS at energy $E$ of a system composed of $Z$ protons, $N$ neutrons and

113: with projection $M$ of the angular momentum on some given axis is defined as

114: \begin{eqnarray} \label{rhodef}

115: \rmb (E,N,Z,M) = \sum_{\nu} \delta (E-E^{\nu}) \delta (N-A^{\nu}_N) \nonumber \\ 

116: \delta (Z-A^{\nu}_Z) \ \delta (M - M^{\nu}) \ .

117: \end{eqnarray}

118: The index $\nu$ denotes all the possible neutron and proton SP configurations

119: (of arbitrary number of particles), $A^{\nu}_\lambda = \sum_i

120: n^{\nu}_{\lambda,i}$ are the neutron and proton number of particles,

121: respectively ($n^{\nu}_{\lambda,i}=0, 1$ are the corresponding occupation

122: numbers of the $i$-th SP state, and $\lambda=N$ or $P$); $E^{\nu}=\sum_\lambda

123: \sum_i n^{\nu}_{\lambda,i} \epsilon_{\lambda,i}$ and $M^{\nu}=\sum_\lambda

124: \sum_i n^{\nu}_{\lambda,i} m_{\lambda,i}$ are the energy and angular momentum

125: projection, where $\epsilon_{\lambda,i}$ and $m_{\lambda,i}$ denote the SP

126: energies and angular momentum projections, respectively.

127: 

128: The conservation of the angular momentum projection is introduced in order to

129: deal with only a subset of states, those with given total angular momentum

130: $J$. A standard treatment \cite{bloch,ericson} of this degree of freedom leads

131: to a DOS of angular momentum $J$ of the form

132: \begin{equation} \label{rhoj}

133: \rmb (Q,N,Z,J) = \frac{2J+1}{2\sqrt{2\pi}\sigma^3} {\rm

134: e}^{-\frac{(J+1/2)^2}{2\sigma^2}} \ \rmb (Q,N,Z),

135: \end{equation}

136: where $\rmb (Q,N,Z)$ is the total MB DOS, $Q=E-E_0$ is the excitation energy

137: measured with respect to the ground state of the system, and $\sigma$ is

138: the spin cutoff parameter.

139: 

140: For an arbitrary SP spectrum the computation of the density of excited states

141: is a difficult combinatorial problem for which no exact solution exists. There

142: is, however, a particular case that can be worked out explicitly: when the SP

143: spectrum consists of equidistant levels separated by $\delta$ (we assume, for

144: simplicity, that the neutron and proton spacings are equal). The MB excitation

145: energies are then given by the sum of two integers corresponding to the total

146: energy of each of the components, $Q=(j+k) \ \delta=K \ \delta$. Each MB state

147: characterized by an integer $K$ has a nontrivial degeneracy. The computation

148: of the degeneracy reduces to the computation of the number of ways into which

149: the total energy may be distributed among the two components, and of the

150: different ways the partial energy of each component can be distributed among

151: its elements. This leads to compute the value of the function $p_2 (K) =

152: \sum_{j=1}^K p(j) \ p(K-j)$, where $p(j)$ is the partition of $j$ (the number

153: of ways into which the integer $j$ can be decomposed as a sum of integers). We

154: are assuming here, to avoid finite size effects, that the excitation energy is

155: small compared to the Fermi energy of each component. Based on the work of

156: Hardy and Ramanujan, an exact expression (written as a convergent series) for

157: $p(j)$ was obtained by Rademacher \cite{hrr}. We have adapted their method

158: (i.e., the circle method, cf Ref.\cite{andrews}) to obtain an exact formula

159: for $p_2 (K)$. Putting back the appropriate units, the MB density can be

160: expressed in terms of $p_2$ as $\rmb /\rhob = 2^{1/4} p_2 (K=\rhob \ Q /2)$,

161: where $\rhob=\rhob_P + \rhob_N = 2/\delta$ is the total (proton + neutrons) SP

162: average density. Then, expressing the exact result as an expansion in terms of

163: $\rhob \ Q$ valid in the range $\rhob^{-1} \ll Q < N \delta, Z \delta$, we

164: obtain

165: \begin{equation} \label{rhoeq1}

166: \rmb(Q,N,Z)/\rhob = \frac{6^{1/4}}{12 \ (\rhob \ Q)^{5/4}} \ {\rm e}^{\sca}

167: \end{equation}

168: where the ``entropy'' $\sca=\sca_{eq}$ of the equidistant spectrum is given by

169: \begin{equation} \label{seq}

170: \sca_{eq} = 2 \sqrt{\frac{\pi^2}{6} \rhob Q} - \left( \frac{\pi}{36} +

171: \frac{15}{16 \pi} \right) \frac{\sqrt{6}}{\sqrt{\rhob Q}} + \left(

172: \frac{35}{96} + \frac{\pi^2}{432} \right) \frac{1}{\rhob Q}

173: \end{equation}

174: plus ${\cal O}((\rhob \ Q)^{-3/2})$ corrections that can be computed but are

175: not given here. The prefactor in Eq.(\ref{rhoeq1}) and the first term of the

176: expansion (\ref{seq}) reproduce Bethe's formula \cite{bethe}. The additional

177: terms provide further smooth corrections of higher order in inverse powers of

178: the excitation energy. Though Eq.(\ref{seq}) represents an asymptotic

179: expansion, we find that an accurate uniform approximation is obtained by

180: keeping only up to the $1/\sqrt{\rhob Q}$ term that kills, when $Q \rightarrow

181: 0$, the divergence produced by the $(\rhob Q)^{-5/4}$ in the prefactor. It is

182: also interesting to note that the correct coefficients of the correction terms

183: in (\ref{seq}) are obtained through the expansion of the exact result, whereas

184: a saddle point approximation of the sum involved in $p_2 (K)$ leads to wrong

185: coefficients.

186: 

187: The previous expression describes in detail the MB DOS for a SP spectrum made

188: of equidistant levels. However, it is clearly unphysical in most situations. A

189: generic SP spectrum contains {\sl fluctuations}, which are manifested at the

190: scale of the average distance between levels, but also on much larger scales

191: (see Ref.\cite{seville} for a review). What is missing in Eq.(\ref{rhoeq1})

192: are the fluctuations in the MB density induced by the SP fluctuations. In this

193: respect, one may consider Eq.(\ref{rhoeq1}) as the MB analog of the Weyl or

194: Wigner-Kirkwood expansions.

195: 

196: It remains to compute the MB level density for an arbitrary SP spectrum,

197: including fluctuations. The way to do it was shown, for a single--component

198: gas, in Ref.\cite{lmr}. The method uses a saddle point approximation of the

199: inverse Laplace transform of the MB density. We have adapted that calculation,

200: following similar lines, to treat Eq.(\ref{rhodef}), that includes two

201: components and angular momentum conservation. The result may be written under

202: the form of Eqs.(\ref{rhoj}) and (\ref{rhoeq1}), but with the entropy in the

203: latter equation given by

204: \begin{equation} \label{sfin}

205: \sca = \sca_{eq} + \frac{1}{T} \left[ \ect (N,Z,0) - \ect (N,Z,T) \right] \ .

206: \end{equation}

207: The parameter $T$ is the temperature, connected to the excitation energy $Q$

208: through the usual relation $Q = \pi^2 \rhob T^2/6= a \ T^2$. $\ect (N,Z,T) =

209: \sum_\lambda \int d \epsilon \ \rhot_\lambda (\epsilon) \ \epsilon \

210: f(\epsilon,\mu,T)$ is the fluctuating part of the energy of the system at

211: temperature $T$ and chemical potential $\mu \sim \ef$ fixed, neglecting

212: temperature variations, by the particle--number conditions $N \sim Z =

213: \int^{\ef} d \epsilon \ \rho_{\lambda} (\epsilon)$. The function $\rho_

214: \lambda (\epsilon) = \sum_j \delta (\epsilon - \epsilon_{\lambda,j})$ is the

215: SP density of the component $\lambda$, and $\rhot_\lambda (\epsilon) =

216: \rho_\lambda (\epsilon) - \rhob_\lambda (\epsilon)$ its fluctuating part.

217: $\ect (N,Z,0)$ is thus the fluctuating part of the ground state energy of the

218: system. Finally, $\sca_{eq}$ in Eq.(\ref{sfin}) is given by Eq.(\ref{seq}),

219: with $\rhob$ the total average SP density of the system at Fermi energy. In

220: fact, for an arbitrary spectrum the saddle point technique does not allow to

221: derive the terms of order $(\rhob Q)^{-1/2}$ and higher in $\sca_{eq}$. The

222: corrections obtained from an equidistant spectrum are thus conjectured to

223: provide a good approximation to the corrections of the smooth part of an

224: arbitrary system, but the validity of this statement has to be confirmed. An

225: explicit numerical verification of its validity for a two--dimensional

226: one--component system was done in Ref.\cite{lmr}.

227: 

228: The function $\ect (N,Z,T)$ presents oscillations when $N$ or $Z$ are varied,

229: in contrast to the more gentle variations as a function of $T$ (a detailed

230: description of the fluctuations and of their $T$--dependence is given below).

231: The MB level density contains now two types of terms: some that vary smoothly,

232: and others that fluctuate as the number of particles varies. The result

233: presented above contains the dominant smooth and oscillatory terms. In the

234: derivation of Eqs.(\ref{rhoj}), (\ref{rhoeq1}) and (\ref{sfin}) we have

235: neglected other terms (for instance, the chemical potentials and $T$ have

236: small corrections that depend on $J$, and thus strictly speaking the

237: factorization (\ref{rhoj}) of the angular momentum is not exact, etc). A

238: detailed account of the derivation will be given elsewhere.

239: 

240: It is remarkable that the MB level density at excitation energy $Q$ depends

241: explicitly on the ground--state energy fluctuations $\ect (N,Z,0)$. A

242: convenient way to analyze the behavior of the fluctuating part of the entropy

243: $\scat = \left[ \ect (N,Z,0) - \ect (N,Z,T) \right]/T$ is through a

244: semiclassical theory. The result is an expression for $\scat$ written as a sum

245: over all the classical periodic orbits of the mean field potential. The main

246: conclusions that can be drawn from that expression are now listed. To be

247: specific, we consider the particular case of an atomic nucleus of $Z$ protons

248: and $N$ neutrons: (a) as the mass number $A=Z+N$ varies at a fixed excitation

249: energy $Q$, $\scat$ presents oscillations of characteristic period $\delta A

250: \approx (\pi/3) A^{2/3}$, that are independent of $Q$; (b) when $Q$ varies at

251: fixed particle number $A$, $\scat$ does not present similar oscillations, but

252: rather gentle variations; (c) the typical amplitude $\sigma_{\scat}$ of

253: $\scat$ at given $(Q,A)$ depends on the dynamical properties of the classical

254: dynamics (integrable or chaotic); (d) since the mean field dynamics of most

255: nuclei is well approximated by a regular motion \cite{seville}, then the

256: behavior of the typical amplitude of $\scat$ is given, to a first

257: approximation, by those of a regular dynamics, that we now detail; (e) using

258: the definition of the temperature $T=\sqrt{Q/a}$ with $a=\pi^2 \rhob/6 \approx

259: A/15$ MeV$^{-1}$ \cite{bm}, we find that there is only one relevant

260: temperature scale in the variation of $\sigma_{\scat}$ with $T$, given by $T_c

261: \approx 4/A^{1/3}$ MeV; for convenience we also introduce $T_{\delta} = (2

262: \pi^2 \rhob)^{-1}\approx 1.3/A$ MeV (the temperature associated to the SP mean

263: level spacing), and $g=T_c/T_{\delta} \approx 3 A^{2/3}$; (f) at low

264: temperatures, $\sigma_{\scat} \approx \sqrt{T/T_{\delta}}$; the typical

265: amplitude of shell effects in the MB DOS therefore increases from 0 at $T=0$

266: up to $\sim \sqrt{g}$ at $T \sim T_c$; (g) at temperatures of order $T_c$ the

267: amplitude is maximal, and starts to decrease for $T > T_c$; (h) in the limit

268: $T \gg T_c$ the typical amplitude tends to zero as $\sigma_{\scat} \approx

269: \sqrt{g/6} \ T_c/T$; using the previous values of $T_c$ and $g$ this gives

270: $\sigma_{\scat} \approx 2 \sqrt{2} /T$ MeV; we thus predict a slow power--law

271: decay of the amplitude of shell effects at high temperatures.

272: 

273: We now turn to a direct application of the previous results to experimental

274: data. Though it will be important to make a systematic analysis of the

275: validity of Eqs.(\ref{rhoj}), (\ref{rhoeq1}) and Eq.(\ref{sfin}) and of their

276: predictions at different energies and mass numbers, we restrict here to a

277: comparison with slow neutron resonances, which have been experimentally

278: studied for a large number of nuclei \cite{RIPL2}. The excitation energies of

279: neutron resonances coincide with the neutron binding energies, $Q = Q_n

280: = S_n (N,Z)$, whose values are in the range $6-8$ MeV for most nuclei.

281: This corresponds to a temperature $T_n \approx 8/\sqrt{A}$ MeV. According to

282: the previous results, the typical amplitude of the fluctuations depends on

283: temperature, with a maximum at $T \approx T_c$. At neutron resonances the

284: ratio $T_n /T_c \approx 2/A^{1/6}$. From $A=30$ to $A=250$, this ratio varies

285: from $1.13$ to $0.8$. We thus find that at excitation energies $Q \approx

286: Q_n$, the temperature is very close to $T_c$; shell effects are maximal. We

287: expect a typical value of $\scat (Q,N,Z)$ very close to its maximum $\sim

288: \sqrt{g} \approx \sqrt{3} A^{1/3}$ (this varies from 5.4 to 11 in the previous

289: range of $A$). In contrast, in the same particle--number range the first

290: correcting term (proportional to $(\rhob \ Q)^{-1/2}$) in the smooth expansion

291: (\ref{seq}) varies from 0.32 to 0.11. That term, and the following ones in the

292: expansion, can thus be neglected at $Q\approx Q_n$.

293: 

294: To make a comparison with experiments we need the different quantities

295: involved in the theoretical expressions. One possibility is to compute them

296: from a particular model. In our case, however, in order to avoid

297: model--dependent features and to make a direct test of our predictions we

298: prefer to extract as much information as possible from experimental data. For

299: each nucleus, the excitation energy at neutron threshold $Q = Q_n = S_n (N,Z)$

300: is taken from the experimental value of $S_n (N,Z)$, and $T_n (N,Z) =

301: \sqrt{Q_n/a}$. For $\ect (N,Z,T_n)$ we have no experimental data available. We

302: compute it as follows. Semiclassically, $\ect (N,Z,T_n)$ is written as a sum

303: over the periodic orbits $p$ (and repetitions) of the mean field potential

304: \cite{seville}. The analysis of the temperature dependence of that sum and of

305: the main contributing orbits leads to the approximation $\ect (N,Z,T_n)

306: \approx \overline{\kappa}_n \ \ect (N,Z,0)$, where $\overline{\kappa}_n$ is

307: the average over the shortest periodic orbits $p$ of the function $\kappa

308: (x_p) = x_p/\sinh (x_p)$, where $x_p = 3 \pi \ell_p A^{1/3} T_n/(4 \ef)$ and

309: $\ell_p$ is the length of the periodic orbit $p$ measured in units of the

310: nuclear radius (notice the mass number and temperature dependence of $x_p$).

311: For each nucleus, $A$ and $T_n$ are given and the average $\overline{\kappa}_n

312: = \overline{\kappa} (N,Z,T_n)$ is computed. In practice, the average is

313: estimated using the shortest periodic orbits of a spherical cavity of radius

314: $R = 1.2 A^{1/3}$fm.

315: 

316: The expression of the entropy $\sca$ in Eq.(\ref{rhoeq1}) takes now the form

317: \begin{equation} \label{sexp}

318: \sca (Q_n,N,Z) = 2 \sqrt{a \ Q_n} + (1 - \overline{\kappa}) \ \ect (N,Z,0) /T_n \ .

319: \end{equation}

320: Finally, $\ect (N,Z,0)$ and $a$ are required. $\ect (N,Z,0)$ can be obtained

321: from the experimental value of the ground--state energy. It corresponds to the

322: fluctuating part of the nuclear binding energy, that we compute by subtracting

323: from the 1995 Audi--Wapstra compilation \cite{aw} the liquid drop expression

324: $\ecb = a_v A - a_s A^{2/3} - a_c Z^2/A^{1/3} - a_A (N-Z)^2/A$, using the

325: parameters (from Ref.\cite{prsz}) $a_v=15.67, \ a_s=17.23, \ a_A=23.29$, and

326: $a_c=0.714$ (all in MeV; we have moreover excluded the pairing term). This

327: parametrization produces a fluctuating part whose average (over $A$) $\langle

328: \ect (N,Z,0) \rangle_A \approx 0$. However, the determination of the average

329: of the fluctuating part is a delicate question that deserves a careful

330: discussion. Due to the discrete variation of the chemical potential as the

331: mass number varies, one can verify that generically $\langle \ect (N,Z,0)

332: \rangle_A$ is non-zero. We have therefore added to the fluctuating part a term

333: $b A + c$, where $b$ and $c$ are two constants. Equation (\ref{sexp}) thus

334: depends on three constants, $a$, $b$, and $c$, that we fix by minimizing the

335: root mean square error with respect to the experimental value of the density,

336: $\sca_{exp}$ (obtained by computing $\sca$ from Eqs.(\ref{rhoj}) and

337: (\ref{rhoeq1}) when $\rmb (Q,N,Z,J)$ is the experimental DOS, $J$ the ground

338: state angular momentum and $\sigma^2 \approx 0.15 a A^{2/3} T_n$

339: \cite{RIPL2}). The result is $a=A/10.42$ MeV$^{-1}$, $b=-0.019$ MeV and

340: $c=7.9$ MeV. The comparison is made in Fig.1. The experimental values

341: $\sca_{exp}$ shown on the top part are to be compared with the ``theoretical''

342: entropies plotted in the middle part. A clear overall agreement is observed.

343: For most nuclei, the relative error in the lower panel is smaller than 10\%,

344: with some remaining structure as a function of $A$, and larger deviations for

345: closed shells (we suspect that this is due to our very schematic estimate of

346: $\overline{\kappa}_n$).

347: 

348: The precision of the present calculation, with only three adjusted parameters,

349: is comparable to the best fits obtained nowadays. We can in fact make the

350: comparison more precise by noticing that Eq.(\ref{sexp}) can be approximated,

351: using an effective value of $a$, by $\sca (Q_n,N,Z) \approx 2 \sqrt{a_{ef} \

352: Q_n}$, where $a_{ef}=a \ [1+ \ect (N,Z,0) (1-\overline{\kappa})/Q_n]$. Under

353: this form, Eq.(\ref{sexp}) is quite similar to one of the best

354: phenomenological formulas studied so far, proposed by Ignatyuk and

355: collaborators \cite{RIPL2,ist}.

356: 

357: To conclude, we have derived an explicit formula for the MB DOS of a two

358: component Fermi gas of fixed angular momentum. The results were applied to the

359: particular case of nuclear level densities, where precise predictions for the

360: smooth dependence and shell fluctuations as a function of excitation energy

361: and mass number were made. Good agreement between theory and experiment in the

362: region of neutron resonances is found. Although it was derived within an

363: independent particle model, the comparison with experiments shows that the

364: final result is probably of more general validity and includes, through the

365: energy fluctuations, effects like pairing. Going to high excitation energies,

366: the main prediction is the decay of shell effects when $Q \geqa Q_n$ (with a

367: power--law tail). However, that prediction is valid for closed systems. Before

368: proceeding in that direction, the theory should be improved to include finite

369: size effects (e.g. a finite number of nucleons) as well as the influence of

370: the continuum.

371: 

372: \noindent $^*$ Unit\'e de recherche associ\'ee au CNRS. We acknowledge fruitful

373: discussions with C. Schmit. This work was supported by grants ACI Nanoscience

374: 201, ANR NT05-2-42103, ANR-05-Nano-008-02 and the IFRAF Institute.

375: 

376: \begin{figure} [H]

377: \includegraphics[width=8.5cm,height=7.4cm]{lr1_fig1.eps} 

378: \caption{Entropy $\sca$ as a function of the mass number $A$ for nuclear level

379: densities at neutron threshold. (a) experimental values; (b) theoretical

380: prediction; (c) relative error.}

381: \end{figure}

382: 

383: \begin{thebibliography}{10}	

384: \bibitem{bethe} H. A. Bethe, Phys. Rev. {\bf 50} (1936) 332.

385: \bibitem{lmr} P. Leboeuf, A. G. Monastra, A. Rela\~no, Phys. Rev. Lett.

386: \textbf{94} (2005) 102502.

387: \bibitem{bh} H. P. Baltes and E. R. Hilf, {\it Spectra of Finite Systems},

388: Bibliographisches Institut-Wissenschaftsverlag, Mannheim, 1976.

389: \bibitem{bb} R. Balian and C. Bloch, Ann. Phys. (N.Y.) {\bf 69} (1972) 76.

390: \bibitem{gutz} M. C. Gutzwiller, J. Math. Phys. {\bf 10} (1969) 1004.

391: \bibitem{trace} H. M. Sommermann and H. A. Weidenm\'{u}ller, Europhys. Lett.

392: \textbf{23} (1993) 79; H. A. Weidenm\'{u}ller, Phys. Rev. A \textbf{48} (1993)

393: 1819 ; J. Sakhr and N. D. Whelan, Phys. Rev. E \textbf{67}(2003) 066213.

394: \bibitem{bloch} C. Bloch, Phys. Rev. \textbf{93} (1954) 1094. 

395: \bibitem{ericson} T. Ericson, Adv. Phys. \textbf{9} (1960) 425.

396: \bibitem{hrr} G. H. Hardy and S. Ramanujan, Proc. London Math. Soc. {\bf 17}

397: (1918) 75; H. Rademacher, Proc. London Math. Soc. {\bf 43} (1937) 241.

398: \bibitem{andrews} G. E. Andrews, {\sl The Theory of Partitions}, Cambridge

399: University Press, Cambridge, 1998.

400: \bibitem{seville}  P. Leboeuf, {\it Regularity and chaos in the nuclear masses},

401: Lect. Notes Phys. \textbf{652} (2005) 245; nucl-th/0406064.

402: \bibitem{bm} A. Bohr and B. R. Mottelson, {\it Nuclear Structure}, Vol.I,

403: Benjamin, Reading, Massachusetts, 1969.

404: \bibitem{RIPL2} Handbook for calculation of nuclear reaction data RIPL-2,

405: http://www-nds.iaea.org/ripl-2.

406: \bibitem{aw} G. Audi and A. H. Wapstra: Nucl. Phys. A \textbf{595}, 409 (1995)

407: \bibitem{prsz} B. Povh, K. Rith, C. Scholz, and F. Zetsche: \textit{Particles

408: and Nuclei: an introduction to the physical concepts}, Springer-Verlag, New

409: York, 1995.

410: \bibitem{ist} A. V. Ignatyuk, G. N. Smirenkin, A.S. Tishin, Sov. J. Nucl.

411: Phys. {\bf 21} (1975) 255.

412: 

413: \end{thebibliography}

414: 

415: \end{document}

416: 

417: