1: %\documentstyle[prl,aps,amssymb,epsfig,graphicx,tabularx,multicol]{revtex}

2: \documentclass[11pt]{article}

3: 

4: \usepackage{graphicx}

5: \usepackage{amsmath}

6: \usepackage{amssymb}

7: \usepackage[english,francais]{babel}

8: 

9: %\usepackage{showkeys}

10: 

11: %Projet d'article : Alain Comtet, Jean Desbois et Christophe Texier

12: %24 Juin 2004

13: %update: 05/04/2005

14: %Commande pour le bulletin de liaison du laboratoire de probabilites

15: %                                             (envoye le 07/04/2005)

16: %

17: %Apres rapports des referees...

18: %update: 27/07/2005

19: %

20: %Dernier nettoyage pour nouvelle version sur cond-mat: 06/09/2005

21: %

22: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

23: %%%%%%%%%%%%%  Journal of Physics A: Mathematical and General  %%%%%%%%%%%%%%%%

24: %

25: % Reference number: 198380 Surname: comtet

26: %

27: % Title: Functionals of the Brownian motion, localization and metric graphs

28: %

29: % Date      Status

30: %01 Aug 2005    Accepted for publication

31: %22 Jul 2005    Resubmission

32: % 6 Jul 2005    We ask to extend the deadline to the 27/07 

33: %       (I was in Russia from 24/06 to 05/07 and Alain in vacations afterwards)

34: %22 Jun 2005    Reports send to author (deadline: 20/07)

35: %03 Jun 2005    Manuscript for adjudication 

36: %03 Jun 2005    Referee report received

37: %03 Jun 2005    Article sent to Adjudicator

38: %27 May 2005    Referee report received

39: %25 Apr 2005    Referee did not report

40: %25 Apr 2005    Article sent to referee

41: %21 Apr 2005    Referee did not report

42: %21 Apr 2005    Article sent to referee

43: %19 Apr 2005    Article sent to referee

44: %19 Apr 2005    Article sent to referee

45: %18 Apr 2005    New submission received and acknowledged

46: %

47: % Reference: J. Phys. A: Math. Gen. {\bf38}, R341-R383 (2005)

48: %

49: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

50: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

51: \textwidth= 16cm

52: \oddsidemargin= 0.5cm

53: \evensidemargin=-0.5cm

54: \topmargin=-2cm

55: \textheight= 24cm

56: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% AMS Fonts %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

57: \newfont{\ensmathquatorze}{msbm10 scaled 1400}

58: \newfont{\ensmathonze}{msbm10 scaled 1100}

59: \newfont{\ensmathdix}{msbm10}

60: \newfont{\ensmathneuf}{msbm10 scaled 833}

61: \newfont{\ensmathhuit}{msbm10 scaled 694}

62: \newfam\ensmathfam                        

63: \textfont\ensmathfam=\ensmathonze        

64: \scriptfont\ensmathfam=\ensmathdix       

65: \scriptscriptfont\ensmathfam=\ensmathhuit

66: \def\ensmf{\fam\ensmathfam\ensmathonze}         % Usage: {\ensmf texte...}

67: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

68: \def\be{\begin{equation}}

69: \def\ee{\end{equation}}

70: \def\benn{$$}

71: \def\eenn{$$}

72: \def\bea{\begin{eqnarray}}

73: \def\eea{\end{eqnarray}}

74: \def\beann{\begin{eqnarray*}}

75: \def\eeann{\end{eqnarray*}}

76: 

77: \def\typ{\mbox{\tiny typ}}

78: 

79: \renewcommand{\leq}{\leqslant}

80: \renewcommand{\geq}{\geqslant}

81: 

82: \def\eqdef{\stackrel{\mbox{\tiny\rm def}}{=}}     % egal a ... par definition

83: \def\eqloi{\stackrel{\mbox{\tiny\rm (loi)}}{=}}     % egal en loi

84: \def\eqlaw{\stackrel{\mbox{\tiny\rm (law)}}{=}}     % egal en loi

85: \newcommand{\ket}[1]{|\kern.3ex#1\kern.3ex\rangle}

86: \newcommand{\bra}[1]{\langle\kern.3ex #1 \kern.3ex|}

87: \newcommand{\APPROX}[1]{                % approximativement egal a .. si ..

88:    {{\raisebox{-.3cm}{$\textstyle\simeq$}} \atop {\scriptstyle{#1}}}}

89: \newcommand{\APPROXs}[1]{               % similaire a ... si ...

90:    {{\raisebox{-.3cm}{$\textstyle\sim$}} \atop {\scriptstyle{#1}}}}

91: \newcommand{\leadto}[1]{                

92:    {{\raisebox{-.3cm}{$\textstyle\longrightarrow$}} \atop {\scriptstyle{#1}}}}

93: \newcommand{\mean}[1]{\left\langle #1 \right\rangle} % moyenne statistique

94: \newcommand{\smean}[1]{\langle #1 \rangle} % moyenne statistique

95: \newcommand{\cum}[1]{\langle\langle #1 \rangle\rangle} % cumulant

96: 

97: \newcommand{\EXP}[1]{{\mbox{\large e}}^{#1}}         % exponentielle

98: %\renewcommand{\cosh}[1]{\mathop{\mathrm{ch}}\nolimits #1} % cosinus hyperbolique

99: %\renewcommand{\sinh}[1]{\mathop{\mathrm{sh}}\nolimits #1} % sinus hyperbolique

100: %\renewcommand{\tanh}[1]{\mathop{\mathrm{th}}\nolimits #1} % tangente hyperbolique

101: \newcommand{\argcosh}{\mathop{\mathrm{argch}}\nolimits}

102: \newcommand{\argsinh}{\mathop{\mathrm{argsh}}\nolimits}

103: 

104: \newcommand{\re}{\mathop{\mathrm{Re}}\nolimits}      % partie reelle

105: \newcommand{\im}{\mathop{\mathrm{Im}}\nolimits}      % partie imaginaire

106: \newcommand{\tr}[1]{\mathop{\mathrm{Tr}}\nolimits\left\{ #1 \right\}}  % Trace

107: \newcommand{\cotg}{\mathop{\mathrm{cotg}}\nolimits}  % cotangente

108: \newcommand{\sign}{\mathop{\mathrm{sign}}\nolimits}  % signe

109: \renewcommand{\min}[2]{\mathop{\mathrm{min}}\nolimits\left( #1 , #2\right)}  

110: \renewcommand{\max}[2]{\mathop{\mathrm{max}}\nolimits\left( #1 , #2\right)}  

111: 

112: \newcommand{\proba}{\mathop{\mathrm{Prob}}\nolimits}  

113: \newcommand{\heav}{\mathop{\mathrm{\rm Y}}\nolimits}  

114: 

115: \newcommand\Dslash{D\hspace{-0.25cm}/}               % D slash de Dirac

116: 

117: \def\NN{{\ensmf N}}                 % entiers 

118: \def\ZZ{{\ensmf Z}}                 % entiers relatifs

119: \def\QQ{{\ensmf Q}}                 % rationnels 

120: \def\RR{{\ensmf R}}                 % ensemble des reels

121: \def\CC{{\ensmf C}}                 % ensemble des complexes

122: 

123: \def\I{{\rm i}}                  % le i mathematique 

124: \def\D{{\rm d}}                  % la differenciation

125: \def\Dc{{\rm D}}                 % dérivée covariante

126: 

127: \newcommand{\dz}{\partial_{z}}

128: \newcommand{\dzb}{\partial_{\bar z}}

129: \newcommand{\drond}[2]{\frac{\partial #1}{\partial #2}} % derivee partielle

130: 

131: \newcommand\dphi{{\cal D}\phi{\cal D}\bar\phi\:} % mesure pour l'integrale de 

132:                                                  % chemin

133: \newcommand\ab{{\alpha\beta}}

134: \newcommand\ba{{\beta\alpha}}

135: \newcommand\mb{{\mu  \beta}}

136: \newcommand\bm{{\beta\mu}}

137: \newcommand\lab{l_{\alpha\beta}}

138: 

139: \newcommand\ide{\mbox{\bf 1}}%\hspace{-0.1cm}{\rm I}} % matrice identite.

140: 

141: % Pour ajuster la hauteur des diagrammes:

142: %=======================================

143: \newcommand{\diagram}[3]{\raisebox{#3}{\includegraphics[scale=#2]{#1}}}

144: 

145: \def\sw{S_{\rm w}}

146: 

147: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

148: %%%%%%%%%%%%%%%%%%%%%%%%  MACRO JEAN %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

149: %\newcommand{\lab}{l_{\alpha\beta}}

150: 

151: \newcommand{\xab}{x_{\alpha\beta}}

152: \newcommand{\xba}{x_{\beta\alpha}}

153: 

154: \newcommand{\xaa}{x_{ab}}

155: \newcommand{\xbb}{x_{ba}}

156: \newcommand{\Gab}{G_{(\alpha\beta )}}

157: \newcommand{\Gba}{G_{(\beta\alpha )}}

158: \newcommand{\Gaa}{G_{(ab )}}

159: \newcommand{\Gbb}{G_{(ba )}}

160: \newcommand{\Gpaa}{G'_{(ab )}}

161: \newcommand{\Gpbb}{G'_{(ba )}}

162: 

163: \newcommand{\Gpab}{G'_{(\alpha\beta )}}

164: \newcommand{\Gpba}{G'_{(\beta\alpha )}}

165: 

166: \newcommand{\dg}{\partial_{\gamma}}

167: \newcommand{\pal}{\psi_{\alpha\beta}}

168: \newcommand{\pbe}{\psi_{\beta\alpha}}

169: \newcommand{\ppal}{\psi'_{\alpha\beta}}

170: \newcommand{\ppbe}{\psi'_{\beta\alpha}}

171: \newcommand{\fal}{\phi_{(\alpha\beta )}}

172: \newcommand{\fbe}{\phi_{(\beta\alpha )}}

173: \newcommand{\fpal}{\phi'_{(\alpha\beta )}}

174: \newcommand{\fpbe}{\phi'_{(\beta\alpha )}}

175: 

176: \newcommand{\cab}{\chi_{\alpha\beta}}

177: \newcommand{\cba}{\chi_{\beta\alpha}}

178: \newcommand{\cpab}{\chi'_{\alpha\beta}}

179: \newcommand{\cpba}{\chi'_{\beta\alpha}}

180: 

181: \newcommand{\caa}{\chi_{ab}}

182: \newcommand{\cbb}{\chi_{ba}}

183: 

184: \newcommand{\cabi}{\chi_{\alpha\beta_i}}

185: \newcommand{\cbia}{\chi_{\beta_i\alpha}}

186: \newcommand{\cpabi}{\chi'_{\alpha\beta_i}}

187: \newcommand{\cpbia}{\chi'_{\beta_i\alpha}}

188: \newcommand{\sg}{\sqrt {\gamma }}

189: \newcommand{\pab}{\psi_{ab}}

190: \newcommand{\pba}{\psi_{ba}}

191: \newcommand{\ppab}{\psi'_{ab}}

192: \newcommand{\ppba}{\psi'_{ba}}

193: 

194: \newcommand{\dpl}{ \frac{ \D \psi_{\alpha\beta}}{  \D x_{\alpha\beta}} }

195: \newcommand{\dpt}{ \frac{ \D \psi_{\beta\alpha}}{ \D x_{\alpha\beta}} }

196: \newcommand{\dpa}{ \frac{ \D \psi_{ab}} {  \D x_{ab}} }

197: \newcommand{\dpb}{ \frac{ \D \psi_{ba}} { \D x_{ab}} }

198: \newcommand{\dpi}{ \frac{ \D \psi_{\alpha\beta_i}}{ \D x_{\alpha\beta_i}} }

199: \newcommand{\dpj}{ \frac{ \D \psi_{\beta_i\alpha}}{ \D x_{\alpha\beta_i}} }

200: \newcommand{\wal}{W_{\alpha\beta}}

201: \newcommand{\wla}{W_{\beta\alpha}}

202: \newcommand{\wab}{W_{ab}}

203: \newcommand{\wba}{W_{ba}}

204: \newcommand{\Wal}{{\cal W}_{\alpha\beta}}

205: \newcommand{\Wab}{{\cal W}_{ab}}

206: \newcommand{\Wla}{{\cal W}_{\beta\alpha}}

207: \newcommand{\Wba}{{\cal W}_{ba}}

208: 

209: %\def\fp{\displaystyle}

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

211: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

212: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

213: 

214: \begin{document}

215: \selectlanguage{english}

216: 

217: %Alain: Brownian functionals in quantum physics~: localization and graphs

218: 

219: \title{Functionals of the Brownian motion,\\ localization and metric graphs}

220: 

221: \author{Alain Comtet$^{(a,b)}$, Jean Desbois$^{(a)}$ and 

222:         Christophe Texier$^{(a,c)}$}

223: 

224: %\date{\today}

225: %\date{April 14, 2005}

226: \date{July 22, 2005}

227: 

228: \maketitle

229: 

230: \noindent

231: \hspace{1cm}

232: \begin{minipage}[t]{13cm}

233: {\small

234: 

235:   \noindent

236:   $^{(a)}$

237:   \begin{minipage}[t]{12.5cm}

238:     Laboratoire de Physique Th\'eorique et Mod\`eles Statistiques,

239:     UMR 8626 du CNRS.\\

240:     Universit\'e Paris-Sud,

241:     B\^at. 100, F-91405 Orsay Cedex, France.

242:   \end{minipage}

243: 

244:   \vspace{0.15cm}

245: 

246:   \noindent

247:   $^{(b)}$

248:   \begin{minipage}[t]{12.5cm}

249:     Institut Henri Poincar\'e, 11 rue Pierre et Marie Curie, F-75005 Paris,

250:     France.

251:   \end{minipage}

252: 

253:   \vspace{0.15cm}

254: 

255:   \noindent

256:   $^{(c)}$

257:   \begin{minipage}[t]{12.5cm}

258:     Laboratoire de Physique des Solides,

259:     UMR 8502 du CNRS.\\

260:     Universit\'e Paris-Sud,

261:     B\^at. 510, F-91405 Orsay Cedex, France.

262:   \end{minipage}

263:   }

264: \end{minipage}

265: 

266: \begin{abstract}

267: We review several results related to the problem of a quantum

268: particle in a random environment.

269: 

270: In an introductory part, we recall how several functionals of the Brownian

271: motion arise in the study of electronic transport in weakly disordered

272: metals (weak localization).

273: 

274: Two aspects of the physics of the one-dimensional strong localization are

275: reviewed~:  some properties of the scattering by a random potential (time

276: delay distribution) and a study of the spectrum of a random potential on a

277: bounded domain (the extreme value statistics of the eigenvalues).

278: 

279: Then we mention several results concerning the diffusion on graphs, and more

280: generally the spectral properties of the Schr\"odinger operator on graphs. The

281: interest of spectral determinants as generating functions characterizing the

282: diffusion on graphs is illustrated.

283: 

284: Finally, we consider a two-dimensional model of a charged particle coupled to

285: the random magnetic field due to magnetic vortices. We recall the connection

286: between spectral properties of this model and winding functionals of the

287: planar Brownian motion.

288: \end{abstract}

289: 

290: \noindent

291: PACS numbers~: 72.15.Rn~; 73.20.Fz~; 02.50.-r~; 05.40.Jc.

292: 

293: %72.   Electronic transport in condensed matter

294: %72.10.-d   Theory of electronic transport; scattering mechanisms

295: %72.10.Bg   General formulation of transport theory 

296: %72.15.Rn Localization effects (Anderson or weak localization)

297: 

298: %73.   Electronic structure and electrical properties of surfaces, interfaces,

299: %      thin films, and low-dimensional structures 

300: %73.23.-b     Electronic transport in mesoscopic systems 

301: %73.20.Fz Weak or Anderson localization

302: 

303: %02.50.-r Probability theory, stochastic processes, and statistics

304: %05.40.Jc Brownian motion

305: 

306: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

307: 

308: \tableofcontents

309: 

310: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

311: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

312: 

313: \section{Introduction}

314: 

315: \subsection{Weak and strong localization}

316: 

317: At low temperature, the electric conductivity $\sigma$ of metals and weakly

318: disordered semiconductors is determined by the scattering of electrons on

319: impurities. It is given by the Drude formula

320: \be

321: \sigma_0=\frac{n_e e^2 \tau_e}{m},

322: \ee

323: where $e$ and $m$ are the charge and mass of the electron respectively. $n_e$

324: is the electronic density and $\tau_e$ the elastic scattering time\footnote{

325:   The elastic scattering time $\tau_e$ is the time characterizing the 

326:   relaxation of the direction of the momentum of the electron. The 

327:   conductivity is proportional to $\tau_e$ for isotropic scattering by 

328:   impurities. For anisotropic scattering the conductivity involves a 

329:   different time  $\tau_{\rm tr}$ called the ``transport'' time

330:   \cite{AshMer76}

331:   (see ref. \cite{AkkMon04} for a discussion within the perturbative approach).

332: }. 

333: This purely classical formula is only valid in a regime where quantum

334: mechanical effects can be neglected. This is the case if the elastic mean free

335: path of electrons $\ell_e=v_F\tau_e$ is large compared with the De Broglie

336: wave length $\lambda_F= 2\pi\hbar /mv_F$ corresponding to  the

337: Fermi energy ($v_F$ is the Fermi velocity).

338: 

339: \vspace{0.25cm}

340: 

341: \noindent%{\bf(1)} 

342: {\it Strong localization.--} When these two length scales

343: are of the same order $\ell_e\approx\lambda_F$ (strong disorder) the fact that

344: the electrons are quantum objects must be taken into account and the wave like

345: character of these particles is of primary importance. It is indeed this wave

346: character which is responsible for the localization phenomenon. The multiple

347: scattering on impurities distributed randomly in space creates random phases

348: between these different waves which can interfere destructively. These

349: interference effects reduce the electronic conductivity. In the extreme case

350: of very strong disorder the waves do not propagate anymore and the system

351: becomes insulating. This is the {\it strong localization} phenomenon which was

352: conjectured by Anderson in 1958.

353: 

354: \vspace{0.25cm}

355: 

356: \noindent%{\bf(2)} 

357: {\it Weak localization.--} In the 80's it was realized that

358: even far from the strong localization regime the quantum transport is

359: affected by the disorder. Diagrammatic techniques used in the weak 

360: disorder limit  $\lambda_F\ll\ell_e$, were initiated by

361: the works of Al'tshuler, Aronov, Gor'kov, Khmel'nitzki{\u\i}, Larkin and Lee

362: \cite{GorLarKhm79,AltKhmLarLee80} (see ref.~\cite{AltLee88} for an 

363: introduction and ref.~\cite{AkkMon04} for a recent presentation). 

364: In this regime, called the {\it weak localization} regime, the

365: Drude conductivity gets a small sample dependent correction whose

366: average, denoted $\smean{\Delta\sigma}$, is called the ``weak localization

367: correction''. $\smean{\cdots}$ denotes averaging with respect to the 

368: random potential.

369: From the experimental side, this phenomenon is well 

370: established and has been the subject of many studies (see 

371: refs.~\cite{Ber84,AroSha87} for review articles).

372: 

373: 

374: \vspace{0.25cm}

375: 

376: \noindent%{\bf(3)} 

377: {\it Phase coherence and dimensional reduction.--} The localization phenomenon

378: comes from the interplay between the wave nature of electronic transport and

379: the disorder. This manifestation of quantum interferences requires that the

380: phase of the electronic wave is well defined, however several mechanisms limit

381: the phase coherence of electrons in metals, among which are the effect of the

382: vibrations of the crystal (electron-phonon interaction) or the

383: electron-electron interaction. We introduce a length scale $L_\varphi$, the

384: phase coherence length, that characterizes the length over which phase

385: breaking phenomenon becomes effective. The lack of phase coherence in real

386: systems is the reason why the strong localization regime has not been

387: observed in experiments on metals. In dimension $d=3$ the strong localization

388: regime is only expected to occur for sufficiently strong

389: disorder\footnote{

390:   Following the scaling ideas initiated by Thouless \cite{LicTho75} and

391:   Wegner \cite{Weg76,Weg79} it was shown in ref.~\cite{AbrAndLicRam79} that 

392:   the localization-delocalization transition exists only for dimension 

393:   $d\geq3$.

394:   In $d=1$ and $d=2$, the fully coherent system is always strongly localized,

395:   whatever the strength of the 

396:   disorder is.\label{Wegner}

397: }. 

398: However, strong localization can also be observed in the weak disorder

399: limit ($\lambda_F\ll\ell_e$) by reducing the dimensionnality\footnote{

400:   The effective dimension of the system is obtained by comparing the sample 

401:   size with the phase coherence length $L_\varphi$. For example, a long 

402:   wire of length $L$ and of transverse dimension $W$ is effectively in 

403:   a 1d regime if $W\ll L_\varphi\ll L$.

404: }. It has been shown in the framework of random matrix

405: theory that the localization length of a weakly disordered quasi-1d wire

406: behaves as $\lambda\sim N_c\ell_e$, where $N_c$ is the number of

407: conducting channels\footnote{

408:   The localization length predicted by the random matrix theory (RMT) can 

409:   be found in ref.~\cite{Bee97}~: 

410:   $\lambda_{\rm RMT}=[\beta(N_c-1)+2]\ell_e$, 

411:   where $\beta=1,2,4$ is the Dyson index describing orthogonal, unitary

412:   and symplectic ensembles, respectively. 

413:   Note that one must be careful with the coefficient involved in 

414:   this relation since the definitions of $\lambda_{\rm RMT}$ and $\ell_e$

415:   differ slightly in RMT and in perturbation theory.

416: } \cite{Dor82,Dor88,MelPerKum88,Bee97}.

417: Therefore the strong localization regime is expected to occur when coherence is

418: kept at least over a scale $\lambda$. At low temperature, in the absence of

419: magnetic impurity, phase breaking mechanisms are dominated by

420: electron-electron interaction, which leads to a divergence of the phase

421: coherence length $L_\varphi(T)\propto(N_c/T)^{1/3}$ predicted in

422: ref.~\cite{AltAroKhm82} and verified in several experiments like

423: \cite{WinRooChaPro86,ThoPepAhmAndDav86,EchGerBozBogNil93b}

424: (see ref.~\cite{PieGouAntPotEstBir03} for recent measurements down to 

425: $40\:$mK). 

426: Therefore the

427: temperature below which strong localization might be observable is given by

428: $L_\varphi(T_*)\sim\lambda$, which leads to $T_*\sim1/(N_c^2d\,\tau_e)$.

429: %(this equality also gives

430: %the threshold beyond which perturbation theory breaks down \cite{AltAro85}).

431: The crossover temperature in metals is out of the experimental range, however

432: it becomes reachable in wires etched at the interface of two semiconducting

433: materials, when the number of conducting channels is highly reduced, and the

434: manifestation of the strong localization has been observed in 

435: ref.~\cite{GerKhaMikBozBog97}.

436: %,KhaGerBog98b}.

437: 

438: \vspace{0.25cm}

439: 

440: \noindent%{\bf(4)} 

441: {\it Strictly one-dimensional case.--} In a weakly disordered and

442: coherent quasi-1d wire, the weak localization regime only occurs for length

443: scales intermediate between the elastic mean free path and the localization

444: length $\ell_e\ll L\ll\lambda$. In the strictly one dimensional case,

445: since $\lambda\simeq4\ell_e$ for weak disorder\footnote{

446:   The elastic mean free path $\ell_e=v_F\tau_e$ is given by the self 

447:   energy $1/(2\tau_e)=-\im\Sigma^{\rm R}(E)$. For example, for a Gaussian

448:   disorder with local correlations, $\smean{V(x)V(x')}={w}\delta(x-x')$, 

449:   we obtain $1/\tau_e\simeq2\pi\rho_0{w}$ for a weak disorder, where 

450:   $\rho_0$ is the free density of states.

451:   In one dimension $\rho_0=1/(\pi v_F)$, therefore 

452:   $\ell_e\simeq v_F^2/(2{w})$, 

453:   which coincides with the {\it high energy} (weak disorder) expansion of the 

454:   localization length 

455:   $\lambda\simeq2v_F^2/{w}$~\cite{AntPasSly81,LifGrePas88}.

456: }, 

457: such a regime does not exist and the system is

458: either ballistic ($\ell_e\gg L$) or strongly localized ($\ell_e\ll L$).

459: 

460: \vspace{0.25cm}

461: 

462: \noindent%{\bf(5)} 

463: Anderson localization in one dimension has been studied by

464: mathematicians and mathematical physicists from the view point of spectral

465: analysis and in connection with limit theorems for products of random

466: matrices \cite{BouLac85}. A breakthrough was the proof of wave localization 

467: in one dimension

468: by Gol'dshtein, Molchanov \& Pastur \cite{GolMolPas77}. Although some

469: progress has been made, the multidimensional case is still out of reach 

470: and the subject of weak localization has almost not been touched in the

471: mathematical literature. One of the main fields of interest in the last twenty

472: years is the investigation of random Schr\"odinger operators in the presence

473: of magnetic fields (see for example the 

474: refs.~\cite{Weg83,BreGroItz84,DesFurOuv95,DesFurOuv96,Fur97,Fur00}~; 

475: for recent results on Lifshitz tails with magnetic field see 

476: refs.~\cite{Fur00,LesWar04}).

477: From the physics side significant advances have been realized. 

478: Field theoretical methods based on supersymmetry provide a general framework

479: for disordered systems and also allow establishing some links with quantum

480: chaos~\cite{Efe97}.

481: 

482: \vspace{0.25cm}

483: 

484: While writing this review, we have tried to collect a large list of references

485: which is however far from being exhaustive.  Ref.~\cite{LifGrePas88}

486: provides a reference book on strong localization, mostly focused on spectrum

487: and localization properties (see also ref.~\cite{Luc92} for a review on 1d

488: discrete models and Lifshitz tails). A recent text about disorder and random

489: matrix theory is ref.~\cite{Efe97}.  Many excellent reviews have been written

490: on weak localization among which \cite{Ber84,ChaSch86} (for the role of

491: disorder and electron-electron interaction, see the book \cite{EfrPol85} 

492: or \cite{LeeRam85}). A recent reference is~\cite{AkkMon04}.

493: 

494: 

495: 

496: \subsection{Overview of the paper}

497: 

498: Section \ref{sec:wl} shows that several functionals of the Brownian motion

499: arise in the study of electronic transport in weakly disordered metals or

500: semiconductors (weakly localized). 

501: The brief presentation of weak localization given in

502: sections~\ref{sec:brownian1} and~\ref{sec:brownian2} follows the heuristic

503: discussion of refs.~\cite{AltLee88,ChaSch86}, which is based on the picture

504: proposed by Khmel'nitzki{\u\i} \& Larkin. In spite of its heuristic

505: character, it allows drawing suggestive connections with well known

506: functionals of the Brownian motion.

507: 

508: The sections \ref{sec:expfun} and \ref{sec:evss} deal with problems of strong

509: localization in one dimension. 

510: A powerful approach to handle such problems is the phase

511: formalism\footnote{

512:   The phase formalism is a continuous version of the Dyson-Schmidt method

513:   \cite{Dys53,Sch57}. A nice presentation can be found

514:   in ref.~\cite{Luc92}.

515: } (presented in refs.~\cite{AntPasSly81,LifGrePas88} for instance). This

516: formalism leads to a broad variety of stochastic processes.

517: Section~\ref{sec:expfun} discusses scattering properties of a random potential

518: and section~\ref{sec:evss} studies spectral properties of a Schr\"odinger

519: operator defined on a finite interval.

520: 

521: In section \ref{sec:graphs}, we review some results obtained for networks of

522: wires (graphs), which can be viewed as systems of intermediate dimension

523: between one and two. We will put the emphasis on spectral determinants, which

524: appear to be an efficient tool to construct several generating functions

525: characterizing the diffusion on graphs (or its discrete version, the random

526: walk).

527: 

528: Finally, in section \ref{sec:magimp}, we show that the physics of a two

529: dimensional quantum particle submitted to the magnetic field of an assembly of

530: randomly distributed magnetic vortices involves fine properties of the

531: planar Brownian motion.

532: 

533: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

534: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

535: 

536: \section{Functionals of the Brownian motion in the context of 

537:          weak localization\label{sec:wl}}

538: 

539: \subsection{Feynman paths, Brownian motion and weak localization

540:             \label{sec:brownian1}}

541: 

542: In the path integral formulation of quantum mechanics, each trajectory is

543: weighted with a phase factor $\EXP{\I S/\hbar}$ where $S$ is the classical

544: action evaluated along this trajectory. The superposition principle states

545: that the amplitude of propagation of a particle between two different points

546: is given by the sum of amplitudes over all paths connecting these two points.

547: This formulation of quantum mechanics is most useful when there is a small

548: parameter with respect to which one can make a quasiclassical expansion. One

549: encounters a similar situation in the derivation of geometrical optics starting

550: from wave optics. In this case the small parameter is the ratio of the wave

551: length to the typical distances which are involved in the problem and the

552: classical paths are light rays. In the context of weak localization the small

553: parameter is $\lambda_F/\ell_e$ and the quasiclassical approximation amounts

554: to sum over a certain subset of Brownian paths. The fact that Brownian paths

555: come into the problem is not so surprising if one goes back to our previous

556: physical picture of electrons performing random walk due to the scattering by

557: the impurities. In the continuum limit, describing the physics at length

558: scales much larger than $\ell_e$, this random walk may be described as a

559: Brownian motion. It can be shown that $\smean{\Delta\sigma}$ is related to the

560: time integrated probability for an electron to come back to its initial

561: position (see ref.~\cite{ChaSch86} for a heuristic derivation or 

562: ref.~\cite{AkkMon04} for a recent derivation in space representation)~:

563: %

564: %A probability is the square modulus of a quantum amplitude. This latter 

565: %is given by a sum of contributions associated to all possible paths.

566: %In the limit $\lambda_F\ll\ell_e$, the phases of these amplitudes are

567: %large random variables, therefore the time reversed paths play a special 

568: %role~:

569: %

570: %Since a probability is the square modulus of a sum of quantum amplitudes,

571: %whose phases are large random variables in the regime $\lambda_F\ll\ell_e$,

572: %the time reversed paths play a special role~: because the weak localization is

573: %an average quantity, reversed paths survive disorder averaging thanks to phase

574: %cancellation and give a contribution that can be written

575: %

576: \be\label{WL}

577: \smean{\Delta\sigma}= -\frac{2e^2 D}{\pi\hbar} \int_{\tau_e}^\infty \D t\, 

578: {\cal P}(\vec r,t|\vec r,0)\,\EXP{-t/\tau_\varphi}

579: \:.\ee

580: where $D$ is the diffusion constant and the factor $2$ accounts for spin 

581: degeneracy. 

582: ${\cal P}(\vec r,t|\vec r\,',0)$ is the Green's function of the diffusion 

583: equation

584: \be

585: \left(\drond{}{t}-D\Delta\right)\,{\cal P}(\vec r,t|\vec r\,',0)

586: = \delta(\vec r-\vec r\,')\,\delta(t)

587: \:.\ee

588: In eq.~(\ref{WL}) the exponential damping at large time describes the lack of

589: phase coherence due to inelastic processes. The phase coherence time

590: $\tau_\varphi$ can be related to the phase coherence length by

591: $L_\varphi^2=D\tau_\varphi$. We have also introduced a cut-off at short time

592: in eq.~(\ref{WL}), that takes into account the fact that the diffusion

593: approximation is only valid for times larger than~$\tau_e$.

594: 

595: 

596: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

597: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

598: 

599: \subsection{Planar Brownian motion~: stochastic area, winding and

600:             magnetoconductance\label{sec:brownian2}}

601: 

602: Weak localization corrections are  directly related to the

603: behaviour of the probability of return to the origin and thus to recurrence

604: properties of the diffusion process. It therefore follows that dimension $d=2$

605: plays a very special role. Consider for instance a thin film whose thickness

606: $a$ is much less than $L_\varphi$. The sample is effectively two dimensional 

607: and the correction to the conductivity is given by

608: \be

609: \smean{\Delta\sigma}= - \frac{e^2}{\pi^2\hbar} \ln(L_\varphi/\ell_e).

610: \ee

611: On probabilistic grounds the logarithmic scaling is of course not unexpected

612: here, it is the same logarithm which occurs in asymptotic laws of the planar

613: Brownian motion \cite{PitYor86,LeG92}. 

614: The neighbourhood recurrence of the planar

615: Brownian motion favors the quantum interference effects and leads to a

616: reduction of the electrical conductivity. A more striking effect is predicted

617: if one applies a constant and homogeneous magnetic field ${\cal B}$ over the

618: sample. In this case the classical action contains a coupling to the magnetic

619: field $S=e{\cal B}A$ where $A$ is a functional of the path given by the line

620: integral

621: \be

622: A= \frac12 \int (x \D y -y \D x) 

623: \:.\ee

624: Properly interpreted, this line integral is nothing but the stochastic area

625: of planar Brownian motion, whose distribution was

626: first computed by P.~L\'evy before the discovery of

627: the Feynman path integral. The weak localization correction reads

628: \be

629: \smean{\Delta\sigma({\cal B})}-\smean{\Delta\sigma(0)}= 

630: \frac{e^2}{2\pi^2\hbar}\int_0^\infty \frac{\D t}{t}\,

631: \EXP{-t/\tau_\varphi}\,

632: \left(1-{\rm E}\!\left[\EXP{2\I e{\cal B}A/ \hbar}\right]\right)

633: \:.\ee

634: The coupling to the magnetic field now appears with an additional factor $2$

635: coming from the fact that the weak localization describes quantum

636: interferences of reversed paths. The expectation ${\rm E}[\cdots]$ is taken

637: over Brownian loops for a time $t$. The magnetoconductivity is given by

638: \cite{AltKhmLarLee80,Ber84,ChaSch86}

639: \be\label{B}

640: \smean{\Delta\sigma({\cal B})}-\smean{\Delta\sigma(0)}= 

641: \frac{e^2}{2\pi^2\hbar}\left[

642:   \psi\left(\frac12+\frac{\phi_0}{8\pi {\cal B} L_\varphi^2} \right)

643:   - \ln\left(\frac{\phi_0}{8\pi {\cal B} L_\varphi^2} \right)

644: \right]

645: \simeq\frac43 \frac{e^2}{\hbar} 

646: \left(\frac{{\cal B}L_\varphi^2}{\phi_0}\right)^2

647: \:,\ee

648: where $\psi(z)$ is the digamma fonction and $\phi_0=h/e$ is the quantum flux.

649: The {\it r.h.s} corresponds to the weak field limit ${\cal

650: B}\ll\phi_0/L_\varphi^2$. This increase of the conductivity with magnetic

651: field is opposite to the behaviour expected classically. This phenomenon is

652: called ``positive (or anomalous) magnetoconductance''. Experimentally this

653: effect is of primary importance~: the magnetic field dependence allows one to

654: distinguish the weak localization correction from other contributions and

655: permits to extract the phase coherent length $L_\varphi$. This expression fits

656: the experimental results remarkably well\footnote{

657:   Note that in materials with strong spin orbit scattering, like gold, the

658:   effect is reversed and a negative magnetoconductance is observed at small

659:   magnetic fields~\cite{HikLarNag80,Ber84}. 

660: } \cite{Ber84}.

661: 

662: Strictly speaking eq.~(\ref{B}) only holds for an infinite sample. In the case

663: of bounded domains the variance depends on the geometry of the system and can

664: be computed explictly for rectangles and strips \cite{DesCom92}. The case of

665: inhomogeneous magnetic fields can be treated along the same line. Consider for

666: instance a magnetic vortex carrying a flux $\phi$ and threading the sample at

667: point $0$. In this case the functional which is involved is not the stochastic

668: area but the index of the Brownian loop with respect to $0$ (winding number).

669: The probability distribution of the index has been computed independently by

670: Edwards \cite{Edw67} in the context of polymer physics and Yor \cite{Yor80} in

671: relation with the Hartman-Watson distribution. For a planar Brownian motion

672: started at $r$ and conditioned to hit its starting point at time 1, the

673: distribution of the index when $n$ goes to infinity is

674: \be

675: {\rm Proba}\left[{\rm Ind}=n\right] \simeq

676: \frac{1}{2\pi^2 n^2}\, {\rm K}_0 (r^2)\,\EXP{-r^2}

677: \:,\ee

678: where ${\rm K}_0(x)$ is the modified Bessel function of second kind (MacDonald 

679: function).

680: The fact that the even moments of this law are infinite is reflected in the

681: non analytic behaviour of the conductivity

682: \be\label{eq9}

683: \smean{\Delta\sigma(\phi)}-\smean{\Delta\sigma(0)}\propto \left|\phi \right|

684: \:.\ee

685: It is interesting to

686: compare with eq.~(\ref{B}) where the quadratic behaviour is given by the

687: second moment of the stochastic area \cite{Mon95}. The behaviour given

688: by eq.~(\ref{eq9}) has been observed experimentally \cite{Gei89,BenKliPlo90}

689: and a theoretical interpretation is provided in ref.~\cite{RamShe87}.

690: 

691: 

692: 

693: \subsection{Dephasing due to electron-electron interaction and 

694:                 functionals of Brownian bridges}

695: 

696: \noindent

697: {\it Dephasing in a wire~: relation with the area below a Brownian bridge.--}

698: Another interesting example of a non trivial connection between a physical

699: quantity and a functional of the Brownian motion occurs in the study of

700: electron-electron interaction and weak localization

701: correction in a quasi-1d metallic wire. 

702: In ref.~\cite{AltAroKhm82}, Al'tshuler, Aronov \& Khmel'nitzki{\u\i} (AAK)

703: proposed to model the effect of the interaction between an electron and its

704: surrounding environment by the interaction with a fluctuating classical field.

705: The starting point is a path integral representation of the probability

706: ${\cal P}(\vec r,t|\vec r\,',0)$ 

707: in which is included the effect of the fluctuating 

708: field that brings a random phase.

709: After averaging over Gaussian fluctuations of the field, given by the

710: fluctuation-dissipation theorem, AAK obtained

711: \be\label{AAK}

712: \smean{\Delta\sigma}=-2\frac{e^2D}{\pi\sw}

713: \int_0^\infty\D t\:\EXP{-\gamma t}

714:   \int_{x(0)=x}^{x(t)=x}{\cal D}x(\tau)\,

715:   \EXP{-\frac1{4D}\int_0^t\D\tau\,\dot x(\tau)^2

716:        -\frac{e^2T}{\sigma_0\sw}\int_0^t\D\tau\,|x(\tau)-x(t-\tau)|}

717: \:,\ee

718: where $T$ is the temperature (the Planck and Boltzmann constants are 

719: set equal to unity $\hbar=k_B=1$), and $\sw$ is the area of the section of the 

720: quasi-1d wire. 

721: In contrast with eq.~(\ref{WL}) where the loss of phase coherence was

722: described phenomenologically by an exponential damping\footnote{

723:   Note that the introduction of an exponential damping describes rigorously 

724:   several effects~: the loss of phase coherence due to spin-orbit 

725:   scattering or spin-flip \cite{HikLarNag80}, the penetration of a

726:   weak perpendicular magnetic field in a quasi 1d wire \cite{AltAro81}.

727:   In this latter case the effect of the magnetic field is taken into 

728:   account through $\gamma=\frac13(e{\cal B}W)^2$ where $W$ is the width of the 

729:   wire of rectangular section.

730: } with the parameter

731: $\tau_\varphi$, here the electron-electron interaction affects the weak

732: localization correction through the introduction in the action of the 

733: functional of the Brownian bridge\footnote{

734:   A Brownian bridge, $(x(\tau),\,0\leq\tau\leq1\,|\,x(0)=x(1))$, is a

735:   Brownian path conditioned to return to its starting point.

736: }

737: \be\label{Atilde}

738: \widetilde{\cal A}_t=\int_0^t\D\tau\,|x(\tau)-x(t-\tau)|

739: \:.\ee

740: The additional damping $\exp{-\gamma t}$ in eq.~(\ref{AAK}) describes the 

741: loss of phase coherence due to other phase breaking mechanisms.

742: By using a trick which makes the path integral local in time, AAK have 

743: computed explicitely the path integral and obtained~:

744: \be\label{AAKres}

745: \smean{\Delta\sigma}=\frac{e^2}{\pi\sw}\,L_N\,

746: \frac{{\rm Ai}(\gamma\tau_N)}{{\rm Ai}'(\gamma\tau_N)}

747: \:,\ee

748: where ${\rm Ai}(z)$ is the Airy function.

749: The Nyquist time $\tau_N=(\frac{\sigma_0\sw}{e^2 T\sqrt{D}})^{2/3}$, gives the

750: time scale over which electron-electron interaction is effective and therefore

751: plays the role of a phase coherence time. We have also introduced the

752: corresponding length $L_N=\sqrt{D\tau_N}$. 

753: Note that the $T$ dependence of $\tau_N\propto T^{-2/3}$ directly reflects the

754: scaling of the area with time~:  

755: $\widetilde{\cal A}_t\eqlaw t^{3/2}\widetilde{\cal A}_1$. We stress that the 

756: AAK theory makes a quantitative prediction for the dependence of the phase

757: coherence length $L_N$ as a function of the temperature, which has been

758: verified experimentally for a wide range of parameters~: on metallic (Gold)

759: wires \cite{EchGerBozBogNil93b,PieGouAntPotEstBir03} 

760: (the behaviour of $L_N\propto(\sw/T)^{1/3}$ was

761: observed before in Aluminium and Silver wires \cite{WinRooChaPro86}), and on

762: wires etched at the interface of two semiconductors \cite{ThoPepAhmAndDav86}.

763: 

764: 

765: 

766: \begin{figure}[!ht]

767: \begin{center}

768: \includegraphics[scale=0.9]{bridge.eps}

769: \hspace{1cm}

770: \includegraphics[scale=0.9]{bridgesym.eps}

771: \end{center}

772: \caption{

773:          Left~: {\it The functional ${\cal A}_t$ gives the absolute area 

774:          below a Brownian bridge $x(\tau)$.}

775:          Right~: {\it  The functional $\widetilde{\cal A}_t$ measures

776:          the area between the Brownian bridge $x(\tau)$ and its 

777:          time reversed counterpart $x(t-\tau)$.

778:          The two functionals are equal in law [eq.~(\ref{AAtilde})].}

779:          \label{fig:bridge}}

780: \end{figure}

781: 

782: It is interesting to point out that the result of AAK can be interpreted as 

783: the Laplace transform of the distribution of the functional (\ref{Atilde}), 

784: ${\rm E}[\exp{-p \widetilde{\cal A}_t}]$. This functional represents the area

785: between a Brownian bridge and its time reversed  counterpart ({\it cf.}

786: figure~\ref{fig:bridge}b), where ${\rm E}[\cdots]$ describes averaging over

787: Brownian bridges. The conjugate parameter $p$ is played in (\ref{AAK}) by the

788: temperature~:  $p=\frac{e^2T}{\sigma_0\sw}$.

789: The result (\ref{AAKres}) has also been derived in the probability 

790: literature. 

791: Let us consider the functional

792: \be\label{A}

793: {\cal A}_t=\int_0^t\D\tau\,|x(\tau)|

794: \:,\ee

795: giving the absolute area below the Brownian bridge starting from the origin,

796: $x(0)=x(t)=0$ ({\it cf.} figure~\ref{fig:bridge}a). The double Laplace

797: transform of the distribution of ${\cal A}_t$ has been computed first by

798: Cifarelli and Regazzini \cite{CifReg75} in the context of economy and

799: independently by Shepp \cite{She82}, and led to

800: $

801: \int_0^\infty\frac{\D t}{\sqrt{t}}\EXP{-\gamma t}

802: {\rm E}[\EXP{-\sqrt{2}{\cal A}_t}]

803: =-\sqrt\pi\,\frac{{\rm Ai}(\gamma)}{{\rm Ai}'(\gamma)}

804: $, 

805: which is equivalent\footnote{

806:   Note that the $1/\sqrt{t}$ in the integral computed by Shepp corrects the 

807:   fact that the averaging over Brownian curves in (\ref{AAK}) is not 

808:   normalized to unity, whereas ${\rm E}[\cdots]$ is.

809: } to the result (\ref{AAKres}).

810: The connection between the two results is clear from the equality in 

811: law\footnote{

812:   For a given process $x(\tau)$ the two functionals ${\cal A}_t$ and 

813:   $\widetilde{\cal A}_t$ are obviously different, however they are 

814:   distributed according to the same  probability distribution. 

815:   They are said ``equal in law''.

816:   \label{footnoteEL}

817: }~:

818: \be\label{AAtilde}

819: {\cal A}_t \eqlaw\widetilde{\cal A}_t

820: \ee

821: which follows from\footnote{

822:   The relation (\ref{anicerel}) is easily proved by using that a Brownian

823:   bridge $(x(\tau),\,0\leq\tau\leq1\,|\,x(0)=x(1)=0)$ can be written in terms 

824:   of a free Brownian motion  $(B(\tau),\,\tau\geq0\,|\,B(0)=0)$ as~:

825:   $x(\tau)=B(\tau)-\tau\,B(1)$.

826: }

827: \begin{equation}

828:  \label{anicerel}

829:  x(\tau)-x(t-\tau) \eqlaw x(2\tau) 

830:  \hspace{0.5cm} \mbox{ for } \hspace{0.5cm}

831:  \tau\in[0,t/2]

832:  \:.

833: \end{equation}

834: The relation (\ref{AAtilde}) allows us to understand more deeply the trick 

835: used by AAK to make the path integral (\ref{AAK}) local in time.

836: The distribution of the area ${\cal A}_t$ has been also studied by Rice in 

837: ref.~\cite{Ric82}. 

838: Interestingly, the inverse Laplace transform of eq.~(\ref{AAKres}) obtained by

839: Rice has been rederived recently independently in ref.~\cite{MonAkk05} in 

840: order to analyze

841: the loss of phase coherence due to electron-electron interaction in a time

842: representation.

843: 

844: The study of statistical properties of the absolute area below a Brownian

845: motion was first addressed by Kac \cite{Kac46} for a free Brownian motion,

846: long before the case of the Brownian bridge studied by Cifarelli \&

847: Regazzini and by Shepp. Later on, it has been extended to other

848: functionals of excursions and meanders\footnote{

849:   An excursion is a part of a Brownian path between two consecutive zeros

850:   and a meander is the part of the path after the last zero.

851: %  A Brownian excursion~: 

852: %  $(e(\tau),\,0\leq\tau\leq1\,|\,e(\tau)>0,\,e(0)=e(1)=\varepsilon)$.

853: %  A Brownian meander~:

854: %  $(m(\tau),\,0\leq\tau\leq1\,|\,m(\tau)>0,\,m(0)=\varepsilon)$.

855: %  Where $\varepsilon\to0$ is a regulator.

856: } \cite{JeaPitYor97,PerWel96}, 

857: which arise in a number of seemingly unrelated problems such as computer 

858: science, graph theory \cite{FlaLou01}  and statistical physics

859: \cite{MajCom04,MajCom05}.

860: 

861: \vspace{0.25cm}

862: 

863: 

864: \noindent{\it Dephasing in a ring.--}

865: Very recently, the question of dephasing due to electron-electron interaction

866: in a weakly disordered metal has been raised again in ref.~\cite{LudMir04}. In 

867: particular it has been shown that the effect of the geometry

868: of the ring and the effect of electron-electron interaction combine 

869: in a nontrivial way leading to a behaviour of the 

870: harmonics of the magnetoconductance that differs from the one predicted by

871: eq.~(\ref{WL}) in ref.~\cite{AltAroSpi81}.

872: For the ring, the functional describing the effect of electron-electron 

873: interaction on weak localization is now given by

874: \begin{equation}

875:   \label{functring}

876:   {\cal R}_t = \int_0^t\D\tau\,

877:   |x(\tau)|\, \left( 1 - \frac{|x(\tau)|}{L} \right) 

878: \end{equation}

879: instead of the area defined by eq.~(\ref{A}). 

880: $x(\tau)$ is a Brownian path\footnote{

881:   As for the case of the wire, the functional involved in the study of

882:   dephasing in the ring ~\cite{TexMon05b} involves the difference

883:   $x(\tau)-x(t-\tau)$ instead of the bridge $x(\tau)$. We have used

884:   the equality in law (\ref{anicerel}) which implies the following property~:

885:   given a Brownian bridge $(x(\tau),\,0\leq\tau\leq{t}\,|\,x(0)=x(t)=0)$,

886:   for any even function $f(x)$ we have

887:   $\int_0^t\D\tau\,f(x(\tau)-x(t-\tau))\eqlaw\int_0^t\D\tau\,f(x(\tau))$.

888: } on a ring of perimeter $L$ such that $x(0)=x(t)=0$.

889: The question has been re-examined in more detail in ref.~\cite{TexMon05b},

890: where the double Laplace transform of the distribution

891: $

892: \int_0^\infty\D t\,\EXP{-\gamma t}\,

893: \frac{1}{\sqrt{t}}\EXP{-\frac{(nL)^2}{2t}}\,{\rm E}_n[\EXP{-{\cal R}_t}]

894: $

895: has been derived. ${\rm E}_n[\cdots]$ denotes averaging over 

896: Brownian bridges defined on a circle, with winding number $n$. Note that the 

897: Laplace transform of the distribution ${\rm E}_n[\EXP{-{\cal R}_t}]$ 

898: has been also studied in ref.~\cite{TexMon05b}.

899: 

900: 

901: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

902: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

903: 

904: \section{Exponential functionals of Brownian motion and Wigner time delay

905:          \label{sec:expfun}}

906: 

907: \subsection{Historical perspective}

908: 

909: Exponential functionals of the form,

910: \begin{equation}

911: \label{deffe}

912: A_t^{(\mu)}= \int_0^t \D s\, \exp{-2(B(s) +\mu s)}

913: \end{equation}

914: where $(B(s),\,s\geq0,\,B(0)=0)$ is an ordinary Brownian motion have been the 

915: object of many studies in mathematics \cite{Yor00}, mathematical finance and 

916: physics \cite{ComMonYor98}. 

917: In the physics of classical disordered systems, the starting point was the

918: analysis of the series (Kesten variable)

919: \be

920: Z= z_1 +z_1z_2+z_1z_2z_3+\cdots

921: \ee

922: where the $z_i$ are independent and identical random variables.

923: Several papers have been devoted to the study of this random variable in the

924: physics \cite{CalLucNiePet85,DerHil83} and mathematics \cite{Kes73,Ver79}

925: literature.

926: It was  realized that $Z$ is the discretized version of $ A_\infty^{(\mu)}$

927: defined in eq.~(\ref{deffe}),

928: which may be interpreted as a trapping time in the context of classical

929: diffusion in a random medium. The tail of the

930: probability distribution $P(Z)$ controls the anomalous diffusive behaviour of a

931: particle moving in a one-dimensional random force field \cite{BouComGeoLeD90}. 

932: The functional

933: $A_t^{(\mu)}$ also arises in the study of the transport properties of

934: disordered samples of finite length \cite{MonCom94,OshMogMor93}.

935: In the context of one dimensional localization the fact that the norm of the

936: wave function is distributed\footnote{

937:   When $\mu>0$, $A_t^{(\mu)}$ possesses a limit distribution for $t\to\infty$.

938:   Moreover, we have the equality in law (see footnote \ref{footnoteEL})~:

939:   $A_\infty^{(\mu)}\eqlaw 1/\gamma^{(\mu)}$, where 

940:   $\gamma^{(\mu)}$ is distributed according to a $\Gamma$-law~:

941:   $P(\gamma)=\frac1{\Gamma(\mu)}\gamma^{\mu-1}\EXP{-\gamma}$.

942:   \label{foot:gammalaw}

943: } as $ A_\infty^{(1)}$ is mentioned in the book of Lifshitz {\it et al}

944: \cite{LifGrePas88}.

945: 

946: 

947: 

948: \subsection{Wigner time delay\label{sec:wtd}}

949: 

950: As discussed in the introduction, the localization of quantum states in one

951: dimension is well understood. However since real systems are not infinite,

952: asking about the nature of the states which are not in the bulk is a perfectly

953: legitimate question. It was pointed out by Azbel in ref.~\cite{Azb83} 

954: that, when a disordered region is connected to a region free of disorder,

955: the localized states acquire a finite lifetime which

956: shows up in transport through sharp resonances,

957: refered as ``Azbel resonances''. This picture was used in

958: ref.~\cite{JayVijKum89} where it was argued that these resonances could be

959: probed in scattering experiments and lead to an energy dependent random time

960: delay of the incident electronic wave. Motivated by this result and by

961: developments in random matrix theory we consider the scattering problem for

962: the one dimensional Schr\"odinger equation (in units $\hbar=2m=1$)

963: \be

964: -\frac{\D^2}{\D x^2}\psi(x)+V(x)\psi(x)=E\,\psi(x)

965: \ee

966: defined on the half line $x\geq 0$  with the Dirichlet boundary conditions

967: $\psi(0)=0$.

968: The potential $V(x)$ has its support on the interval $[0,L]$.

969: Outside of this interval, the scattering state of energy $E=k^2$ is given by

970: \be\label{scattstate}

971:   \psi_E(x)=\frac1{\sqrt{hv_E}}

972:   \left( \EXP{-\I k(x-L)} +\EXP{\I k(x-L)+\I\delta} \right)

973: \:,\ee 

974: where $h=2\pi$ is the Planck constant (in unit $\hbar=1$) and 

975: $v_E=\D E/\D k=2k$ the group velocity.

976: Eq.~(\ref{scattstate})

977: represents the superposition of an incoming plane wave incident from

978: the right and a reflected

979: plane wave caracterized by its phase shift $\delta(E)$. 

980: The Wigner time delay, defined by the relation $\tau=\D\delta/\D{E}$,

981: can be understood as the time spent by the wave packet of energy $E$ 

982: in the disordered region (this interpretation is only valid 

983: at high energy~; see 

984: refs.~\cite{HauSto89,But90,LanMar94,CarNus02} for review articles on time 

985: delay and traversal times).

986: This representation of the time delay has been used in many papers both

987: in the mathematics and physics literature. Starting from this representation

988: one can derive a system of stochastic differential equations which can be

989: studied in certain limiting cases. However, to understand the universality 

990: of the statistical properties of the Wigner time in the high energy limit,

991: we follow a different approach below.

992: 

993: 

994: 

995: 

996: \subsubsection*{Universality of time delay distribution at high energy}

997: 

998: Using the expression of the scattering state (\ref{scattstate}) and the

999: definition $\tau=\D\delta/\D{E}$, we can obtain the so-called Smith formula

1000: \cite{Fri58,Smi60} relating the time delay to the wave function in the bulk~:

1001: \be\label{Smith}

1002:   \tau(E)= 2\pi\int_0^L \D x\, |\psi_E(x)|^2 

1003:   -\frac{1}{2E}\sin\delta(E)

1004: \:.\ee

1005: Following

1006: ref.~\cite{AntPasSly81} one can parametrize the wave function in the bulk in

1007: terms of its phase and modulus. 

1008: For this purpose we rewrite the Schr\"odinger equation as a set of two 

1009: coupled first order differential equations, $\D\psi_E/\D x=\psi'_E$

1010: and $\D\psi'_E/\D x=(V(x)-E)\psi_E$, and perform the change of variables~:

1011: $\psi_E(x)=\EXP{\xi(x)}\sin\theta(x)$ and 

1012: $\psi'_E(x)=k\EXP{\xi(x)}\cos\theta(x)$. One obtains a new set of 

1013: first order differential equations $\D\theta/\D x=k-\frac1kV(x)\sin^2\theta$

1014: and $\D\xi/\D x=\frac1{2k}V(x)\sin2\theta$. Up till now everything is exact. 

1015: If we now consider the high energy limit\footnote{

1016:   To define precisely the high energy limit, let us introduce the integral

1017:   of the correlation function of the disorder

1018:   ${w}=\int\D x\,\smean{V(x)V(0)}$. The high energy limit corresponds to

1019:   $k\gg{w}^{1/3}$.

1020: }

1021: we can neglect the second term on the {\it r.h.s} of eq.~(\ref{Smith})

1022: and integrate out the phase which is a fast variable~; we obtain

1023: \be\label{timedelay0}

1024: \tau = \frac{1}{k}\int_0^L \D x \:\EXP{2(\xi(x)-\xi(0))}

1025: \:.

1026: \ee

1027: This representation of the time delay holds for any realization of the

1028: disordered potential. Moreover, one can prove under rather mild conditions

1029: on the correlations of the random potential

1030: that $\xi(x)$ is a Brownian motion with drift 

1031: $\xi(x)=x/\lambda+\sqrt{1/\lambda}\,B(x)$ where

1032: $B(x)$ is an ordinary Brownian motion and $\lambda$ the localization length. 

1033: Using the scaling properties of the Brownian motion gives the following

1034: identity in law

1035: \be\label{timedelay}

1036: \tau \eqlaw \frac{\lambda}{k}\int_0^{L/\lambda} \D x\: \EXP{-2(B(x)+x)}

1037: = \frac{\lambda}{k}\, A_{L/\lambda}^{(1)}

1038: \:,

1039: \ee

1040: where $A_L^{(1)}$ has been defined in eq.~(\ref{deffe}).

1041: This representation of the time delay as an exponential functional of the

1042: Brownian motion, first established in ref.~\cite{FarTsa94} by a different

1043: method, allows us to obtain a number of interesting results

1044: \cite{ComTex97,TexCom99}~:

1045: 

1046: \vspace{0.25cm}

1047: 

1048: \noindent

1049: ({\it i}) Existence of a limit distribution for fixed\footnote{

1050:   From the remark of the footnote \ref{foot:gammalaw}, we notice that 

1051:   $1/\tau$ is distributed according to an exponential law.

1052: } $\tau$ and $ L\rightarrow\infty$

1053: \be

1054: P(\tau)=\frac{\lambda}{2k\tau^2}\, \EXP{- \frac{\lambda}{2k\tau}} 

1055: \:.\ee

1056: This result is reminiscent of the random matrix theory prediction in

1057: spite of the fact that this theory does not apply to systems that are

1058: strictly one-dimensional\footnote{

1059:   The random matrix theory describes the regime of weak localization or 

1060:   systems whose classical dynamics is chaotic. The distribution of 

1061:   time delay in this framework has the same functional form 

1062:   $P(\tau)\propto\frac1{\tau^{2+\mu}}\EXP{-\tau_0/\tau}$ 

1063:   \cite{FyoSom96a,GopMelBut96,BroFraBee97,OssFyo05} with two differences~: 

1064:   ({\it i}) the exponent $2+\mu>2$,

1065:   ({\it ii}) the time scale $\tau_0$.

1066: }. 

1067: 

1068: \vspace{0.25cm}

1069: 

1070: \noindent

1071: ({\it ii}) Linear divergence of the first moment, $\smean{\tau}=L/k$, and 

1072: exponential divergence of the higher moments 

1073: $\smean{\tau^n}\propto\EXP{2n(n-1)L/\lambda}$.

1074: This divergence reflects a log-normal tail of the distribution for a finite

1075: length $L$.

1076: 

1077: \subsubsection*{Time delay and density of states}

1078: 

1079: The relation (\ref{Smith}) also provides another interpretation to these 

1080: results. In the situation under consideration, the scattering state is 

1081: directly related to the local density of states (LDoS) by 

1082: $\rho(x;E)=\bra{x}\delta(E-H)\ket{x}=|\psi_E(x)|^2$. Therefore the time delay

1083: can be interpreted as the DoS of the disordered region\footnote{

1084:   The relation between time delay and DoS is sometimes refered as Krein-Friedel

1085:   relation \cite{Fri52,Kre53}. This relation has been originally introduced in

1086:   ref.~\cite{BetUhl37} in the context of statistical physics 

1087:   (see also ref.~\cite{DasMaBer69} and \S77 of ref.~\cite{LanLif66e}). 

1088:   Note that it has been recently

1089:   discussed in the context of graphs in refs.~\cite{Tex02,TexBut03,TexDeg03}.

1090: }. 

1091: This establishes a relation between the results given above and the work of

1092: Al'tshuler \& Prigodin~\cite{AltPri89} where the distribution of the LDoS

1093: was studied for a white noise potential using the method of

1094: Berezinski{\u\i}\footnote{

1095:   The method of Berezinski{\u\i} blocks has been introduced to study

1096:   specifically the case of white noise disordered potential in 

1097:   one dimension \cite{Ber74}.

1098:   This powerful approach has been widely used and has allowed to derive

1099:   numbers of important results like in refs.~\cite{GogMelRas76,GorDorPri83} 

1100:   for example.

1101: }.

1102: 

1103: 

1104: \subsubsection*{Time delay for Dirac Hamiltonian at the threshold energy}

1105: 

1106: The representation (\ref{timedelay}) of the time delay as an exponential

1107: functional of the Brownian motion only holds in the weak disorder ({\it i.e.}

1108: high energy) limit. A similar representation can be derived for the random

1109: mass Dirac model at the middle of the spectrum. 

1110: The Dirac Hamiltonian~is 

1111: \be\label{HDirac}

1112: H_{\rm D}=\sigma_2\,\I\frac{\D}{\D x} + \sigma_1\phi(x)

1113: \:,\ee

1114: where $\sigma_i$ are the Pauli matrices. $\phi(x)$ can be interpreted as a

1115: mass\footnote{

1116:   The Hamiltonian (\ref{HDirac}) with $\phi(x)$ a white noise was introduced

1117:   by Ovchinnikov \& Erikmann \cite{OvcEri77} as a model of 

1118:   one-dimensional semiconductor with a narrow fluctuating gap.

1119:   It is interesting to point out that the Dirac equation $H_{\rm D}\psi=k\psi$ 

1120:   also has an interpretation in the context of superconductivity, as 

1121:   linearized Bogoliubov-de~Gennes equations for a real random superconducting

1122:   gap $\phi(x)$.

1123:   Finally it is worth mentioning that more general 1d Dirac Hamiltonians 

1124:   with several kinds of disorder (mass term, potential term and magnetic 

1125:   field) have been studied in refs.~\cite{Boc99,Boc00}. 

1126: }. 

1127: The Dirac equation $H_{\rm D}\psi=k\,\psi$ possesses a

1128: particle-hole symmetry reflected in the symmetry of the spectrum with respect

1129: to $k=0$ (the dispersion relation of the Dirac equation is linear 

1130: in absence of the mass term and energy is equal to momentum).

1131: The middle of the spectrum is an interesting point where the divergence in the

1132: localization length signals the existence of a delocalized state

1133: (see footnote \ref{footnoteSUSY}). These

1134: properties should therefore show up in the probability distribution of the 

1135: Wigner time delay. 

1136: Since the dispersion relation of the Dirac equation is linear, the time 

1137: delay is now defined as $\tau=\D\delta/\D k$.

1138: The following representation has been derived 

1139: in ref.~\cite{SteCheFabGog99}~:

1140: \be \label{tdsusy}

1141: \tau= 2\int_0^L \D x\, \EXP{2\int_0^x \phi(y)\D y}

1142: \:.\ee

1143: In contrast with

1144: eqs.~(\ref{timedelay0},\ref{timedelay}), which have been obtained after

1145: averaging over the fast phase variable, the representation (\ref{tdsusy}) is

1146: exact (for $k=0$). If $\phi(x)$ is a white noise, then

1147: \begin{equation}

1148: \label{ChenTD}

1149: \tau = \frac{2}{g}\,A_{g L}^{(0)}

1150: \:.\end{equation}

1151: In this case since the drift vanishes, it is known \cite{MonCom94,Yor00} that

1152: there is no limiting distribution when $L\rightarrow\infty$. The physical

1153: meaning of the lognormal tail

1154: \be

1155: P(\tau)\sim \frac{1}{2\tau\sqrt{2\pi gL}}\, \EXP{-\frac{1}{8gL} \ln^2(g\tau)}

1156: \ee

1157: in the limit $\tau\rightarrow\infty$ is discussed in detail in

1158: ref.~\cite{SteCheFabGog99}.

1159: 

1160: 

1161: 

1162: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1163: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1164: 

1165: \section{Extreme value spectral statistics\label{sec:evss}}

1166: 

1167: The density of states (DoS) and the localization length of a one-dimensional

1168: Hamiltonian with a random potential can be studied by the phase formalism

1169: \cite{AntPasSly81,LifGrePas88} (see also ref.~\cite{Luc92} for discrete

1170: models).

1171: In the previous section we pointed out

1172: that the physics of the strong localization in one-dimension can also be

1173: probed from a different angle by considering the scattering of a plane wave by

1174: the random potential. This has led to the study of the Wigner time delay. In

1175: this section we show that the phase formalism also allows describing finer

1176: properties of the spectrum such as the probability distribution of the $n$-th

1177: eigenvalue. This is an instance of an {\it extreme value statistics} ~: given

1178: a ranked sequence of ${\cal N}$ random variables

1179: $x_1 \leq x_2 \leq \ldots\leq x_{\cal N}$, 

1180: the problem is to find the distribution of the $n$-th of these variables in a

1181: given interval. In the particular case of uncorrelated random variables, the

1182: extreme value distributions have been studied by E.~Gumbel

1183: \cite{Gum35,Gum54,Gum58}. This problem becomes much more complicated when the

1184: random variables are correlated, a case which has recently attracted a lot of

1185: attention. Such extreme value problems have appeared recently in a variety of

1186: problems ranging from disordered systems \cite{BouMez97,CarLeD01,DeaMaj01} to

1187: certain computer science problems such as growing search trees \cite{MajKra03}.

1188: In the case of the spectrum of a random Hamiltonian, the eigenvalues are in

1189: general correlated variables, apart from the case of strongly localized

1190: eigenstates \cite{Mol81}. Below we show that, in the one dimensional case,

1191: this problem is related to studying the first exit time distribution of a 

1192: one-dimensional diffusion process.

1193: 

1194: \mathversion{bold}

1195: \subsection{Distribution of the $n$-th eigenvalue~: relation with a first 

1196:             exit time distribution}

1197: \mathversion{normal}

1198: 

1199: We consider a Schr\"odinger equation $H\varphi(x)=E\varphi(x)$ on a finite

1200: interval $[0,L]$. The spectral (Sturm-Liouville) problem is further defined by

1201: imposing suitable boundary conditions. We choose the Dirichlet boundary

1202: conditions $\varphi(0)=\varphi(L)=0$. The spectrum of $H$ is denoted by 

1203: ${\rm Spec}(H)=\{E_0<E_1<E_2<\cdots\}$. Our purpose is to compute the

1204: probability 

1205: \be

1206: W_n(E) = \smean{\delta(E-E_n)}

1207: \ee

1208: for the eigenvalue $E_n$ to be at energy $E$ (the bracket $\smean{\cdots}$

1209: means averaging over the random potential).

1210: Note that the sum of these distributions 

1211: $\frac1L\sum_{n}W_n(E)=\rho(E)$ is the average DoS per unit length.

1212: We now show how the calculation of $W_n(E)$ can be cast into  a first exit time

1213: problem.

1214: 

1215: \subsubsection*{Random Schr\"odinger Hamiltonian}

1216: 

1217: We consider the Hamiltonian 

1218: \be\label{hamil}

1219: H=-\frac{\D^2}{\D x^2} + V(x)

1220: \:,\ee

1221: where $V(x)$ is a Gaussian white noise random potential~: $\smean{V(x)}=0$ and

1222: $\smean{V(x)V(x')}={w}\,\delta(x-x')$. 

1223: 

1224: We replace the Sturm-Liouville problem by a Cauchy problem~:

1225: let  $\psi(x;E)$ be the solution of the Schr\"odinger equation

1226: $H\psi(x;E) = E\psi(x;E)$ with the boundary conditions $\psi(0;E)=0$

1227: and $\frac{\D}{\D x}\psi(0;E)=1$. 

1228: %It follows that $\psi(x;E)$ is real.

1229: The boundary condition $\psi(L;E)=0$ is fulfilled whenever the energy $E$ 

1230: coincides with an eigenvalue $E_n$ of the Hamiltonian. In this case,

1231: the wave function 

1232: $\varphi_n(x)=\psi(x;E_n)/\left[\int_0^L\D{x'}\,\psi(x';E_n)^2\right]^{1/2}$ 

1233: has $n$ nodes in the interval $]0,L[$, and two nodes at the boundaries. 

1234: 

1235: \begin{figure}[!ht]

1236: \hspace{3cm}

1237: \begin{tabular}{ll}

1238: \underline{$E<E_0$}           &  \diagram{psi1b.eps}{1}{-0.75cm}  

1239: \\

1240: \underline{$E=E_0$}           &  \diagram{psi2b.eps}{1}{-0.75cm}  

1241: \\

1242: \underline{$E_0<E<E_1$}       &  \diagram{psi3b.eps}{1}{-1.40cm}  

1243: \\[0.5cm]

1244: \hspace{1.5cm}{\huge$\vdots$} & \hspace{1cm}{\huge$\vdots$\hspace{2cm}$\vdots$}

1245: \end{tabular}

1246: \caption{\it

1247:          The probability for the $n$-th level $E_n$ of the spectrum to be 

1248:          at $E$ is also the probability for the $n+1$-th node of $\psi(x;E)$ 

1249:          to be at $L$.\label{fig:psideE}}

1250: \end{figure}

1251: 

1252: Let us denote by $\ell_m$ ($m\geq1$), the length between two consecutive nodes.

1253: We consider the Ricatti variable

1254: \be

1255: z(x;E) = \frac{\D}{\D x} \ln|\psi(x;E)|

1256: \:,\ee

1257: which obeys the following equation:

1258: \be\label{Riceq}

1259: \frac{\D}{\D x} z = -E - z^2 + V(x)

1260: \:,\ee

1261: with initial condition $z(0;E)=+\infty$.

1262: This equation may be viewed as a Langevin equation for a particle located at

1263: $z$ submitted to a force $-\partial U(z)/\partial z$ deriving from the 

1264: unbounded potential 

1265: \be\label{potU}

1266: U(z)=Ez+\frac{z^3}{3}

1267: \ee 

1268: and to a random white noise $V(x)$. 

1269: 

1270: Each node of the wave function corresponds to $|z(x)|=\infty$. At ``time''

1271: $x=0$ the ``particle'' starts from $z(0)=+\infty$ and eventually ends at

1272: $z(\ell_1-0^+)=-\infty$ after a ``time'' $\ell_1$. Just after the first node

1273: it then starts again from  $z(\ell_1+0^+)=+\infty$, due to the continuity of

1274: the wave function. It follows from this picture that the distance

1275: $\ell_m$ between two consecutive nodes may be viewed as the ``time'' needed

1276: by the particle to go through the interval $]-\infty,+\infty[$ (the

1277: ``particle'' is emitted from $z=+\infty$ at initial ``time'' and absorbed

1278: when it reaches $z=-\infty$).

1279: The distances $\ell_m$ are random variables, interpreted as times needed

1280: by the process $z$ to go from $+\infty$ to $-\infty$.

1281: These random variables are statistically independent because each time the

1282: variable $z$ reaches $-\infty$, it loses the memory of

1283: its earlier history since it is brought back to the same initial condition

1284: and $V(x)$ is $\delta$-correlated. This remark is a crucial point for the

1285: derivation of $W_n(E)$.

1286: Interest in these random ``times'' $\ell_m$ lies in their relation

1287: with the distribution of the eigenvalues. Indeed the probability that

1288: the energy $E_n$ of the $n$-th excited state is at $E$ is also

1289: the probability that the sum of the $n+1$ distances between the nodes is

1290: equal to the length of the system: $L=\sum_{m=1}^{n+1}\ell_m$

1291: (this is illustrated on figure~\ref{fig:psideE}).

1292: Since the $\ell_m$ are independent and identically distributed

1293: random variables   

1294: $\proba\left[L=\sum_{m=1}^{n+1}\ell_m\right]$ is readily obtained from the

1295: distribution $P(\ell)$ of one of these variables. 

1296: 

1297: We introduce the intermediate variable ${\cal L}(z)$ giving the time needed by

1298: the process starting at $z$ to reach $-\infty$~: therefore we have 

1299: $\ell={\cal L}(+\infty)$. The Laplace transform of the distribution

1300: $

1301: h(\alpha,z) = 

1302: \smean{ \EXP{-\alpha{\cal L}} \:|\: z(0)=z;\:z({\cal L})=-\infty }

1303: $ 

1304: obeys \cite{Gar89}~:

1305: \be

1306: G_z \, h(\alpha,z) = \alpha \, h(\alpha,z)

1307: \:,\ee

1308: where the backward Fokker-Planck generator is

1309: \be\label{bfpegen}

1310: G_z = -U'(z) \partial_z + \frac{{w}}{2} \partial_z^2

1311: \:.\ee

1312: The boundary conditions are

1313: $\partial_z h(\alpha,z)|_{z=+\infty}=0$ and $h(\alpha,-\infty)=1$.

1314: By Laplace inversion we can derive $P(\ell)$ since

1315: $h(\alpha,+\infty)=\int_0^\infty\D\ell\,P(\ell)\EXP{-\alpha\ell}$.

1316: 

1317: The solution of this problem is given in ref.~\cite{Tex00}.

1318: Here we only consider the limit $E\to-\infty$, which corresponds to 

1319: the bottom of the spectrum. 

1320: In this regime the dynamics of the Ricatti variable $z$ can be read off 

1321: from the shape of the potential (see figure~\ref{fig:potUneg}).

1322: 

1323: \begin{figure}[!ht]

1324: \begin{center}

1325: \includegraphics{potUneg.eps}

1326: \end{center}

1327: \caption{\it The potential (\ref{potU}) related to the deterministic force 

1328:          to which the Ricatti variable is submitted in eq.~(\ref{Riceq}).

1329:          \label{fig:potUneg}}

1330: \end{figure}

1331: 

1332: The particle falls rapidly in the well from which it can only escape by a

1333: fluctuation of the random force. Therefore the time needed to go from

1334: $+\infty$ to $-\infty$ is dominated by the time spent in the well, which is

1335: distributed according to the Arrhenius formula. It follows that the average

1336: ``time'', which is related to the inverse of the integrated density of states

1337: (IDoS) per unit length $N(E)=\int_0^E\D E'\,\rho(E')$, behaves as

1338: $

1339: \mean{\ell} = N(E)^{-1}

1340: \simeq \frac{\pi}{\sqrt{-E}} \exp\frac{8}{3{w}}(-E)^{3/2}

1341: $.

1342: This simple picture, first provided by Jona-Lasinio \cite{Jon83} allows

1343: recovering the exponential tail obtained by several methods in 

1344: refs.~\cite{FriLlo60,Hal65,AntPasSly81,LifGrePas88,ItzDro89}\footnote{

1345:   It is interesting to quote ref.~\cite{KucSad98} where it was shown that 

1346:   this non perturbative exponential tail can be obtained by a simple counting

1347:   method of ``skeleton'' diagrams assuming that they all have the same value

1348:   (this approach was also applied to the 3d case).

1349: }.

1350: The distribution of the length $\ell$ is a Poisson law (see \cite{Gar89}

1351: or appendix of ref.~\cite{Tex00}):

1352: \be\label{dol-}

1353: P(\ell) = N(E)\,\exp{-\ell\,N(E)}

1354: \:.\ee

1355: Using this result we can show that \cite{Tex00} 

1356: \be\label{TheResult}

1357: W_n(E) = L\rho(E) \, \frac{(L\,N(E))^n}{n!} \, \EXP{-L\,N(E)} 

1358: \:.\ee

1359: This result has a clear meaning~: $L\rho(E)$ gives the probability 

1360: to find any level at $E$ and the factor $\frac{x^n}{n!}\EXP{-x}$ ``compels''

1361: the number of states below $E$, $x=LN(E)$, to be close to $n$.

1362: We may go further and write

1363: \be

1364: W_n(E) = \frac{1}{\delta E_n} \ 

1365:          \omega_n\left(\frac{E - E_n^{\typ}}{\delta E_n}\right)

1366: \:,\ee

1367: where the typical value of the energy is 

1368: $E_n^{\typ}(L)=-(\frac{3{w}}{8}\ln\tilde L)^{2/3}$,

1369: while the scale of the fluctuations reads 

1370: $

1371: \delta E_n =\frac{ {w}^{2/3} }{2\sqrt{n+1}}(3\ln\tilde L)^{-1/3}

1372: $, 

1373: where $\tilde L=\frac{L\,{w}^{1/3}}{2\pi(n+1)}$.

1374: The function

1375: \be\label{TheResultbis}

1376:   \omega_n(X)=\frac{(n+1)^{n+\frac12}}{n!}

1377:   \exp{\left(\sqrt{n+1}\:X -(n+1)\,\EXP{{X}/{\sqrt{n+1}}}\right)}

1378: \:\ee

1379: has the form of a Gumbel law for {\it uncorrelated} random variables.

1380: The fact that the eigenvalues are uncorrelated is a consequence of the 

1381: strong localization of the eigenfunctions \cite{Mol81}.

1382: 

1383: The work summarized in the above paragraph, which appeared in

1384: ref.~\cite{Tex00}, generalizes the result of McKean \cite{McK94} for the

1385: ground state.

1386: 

1387: A similar result was obtained by Grenkova {\it et al} \cite{GreMolSud83} for

1388: the model of $\delta$-impurities with random positions, in the limit of low

1389: impurity density which has no counterpart in the model we considered 

1390: here\footnote{

1391:   See footnote \ref{FrischLloyd}.

1392: }.

1393: However both models describe the same physics of strongly localized

1394: eigenstates.

1395: 

1396: 

1397: 

1398: \subsubsection*{Supersymmetric random Hamiltonian}

1399: 

1400: Random Schr\"odinger operators have been investigated through a wide range of

1401: models. Depending on the physical context, there are indeed many ways to model 

1402: the

1403: disorder. In the previous paragraph we have assumed that the potential $V(x)$

1404: is a white noise. We have also mentioned that if the potential is a 

1405: superposition of $\delta$-potentials randomly distributed along the line,

1406: although the spectra of the two models are quite different\footnote{

1407:   The model of $\delta$-impurities with random positions was introduced 

1408:   and studied by Schmidt \cite{Sch57} but often refered as the 

1409:   Frisch \& Lloyd model \cite{FriLlo60}. Compared to the model for 

1410:   white noise potential, which is characterized by one parameter, the Frisch \&

1411:   Lloyd model is characterized by two parameters~: the 

1412:   strength of the $\delta$-potential, and their density.

1413:   In the limit of high density of impurities this model is equivalent to 

1414:   the white noise potential model. The limit of low density presents 

1415:   different spectral singularities (Lifshitz singularity and, for 

1416:   negative weight of $\delta$-potentials, an additional Halperin singularity

1417:   in the negative part of the spectrum).\label{FrischLloyd}

1418: }, 

1419: the extreme spectral statistics are the same. The localization properties and

1420: the statistics of the time delay at high energy are also similar for both

1421: models \cite{Tex99,TexCom99}. These two models belong to the same class of

1422: random Hamiltonians with a random scalar potential with short range

1423: correlations. They are both continuous versions of discrete

1424: tight binding models with on-site random potential. This is the case of

1425: so-called {\it diagonal disorder}, since the random potential appears on the

1426: diagonal matrix elements of the tight binding Hamiltonian in the basis of

1427: localized orbitals\footnote{

1428:   Note that the continuum limit of tight binding Hamiltonian with diagonal 

1429:   disorder can lead to different continous models. 

1430:   Let us consider the discrete model

1431:   $H_{i,j} = -\delta_{i,j+1}-\delta_{i,j-1}+\delta_{i,j}V_i$. 

1432:   For $V_i=0$ the spectrum is $E_{k}=-2\cos(k)$, with $k\in]-\pi,\pi]$.

1433:   (A) If the continuous limit is taken by considering the band 

1434:   edge ($k\sim0$), one is led to the continuous model (\ref{hamil}).

1435:   (B) If the band center is considered instead ($k\sim\pi/2$), the 

1436:   spectrum can be linearized and one is led to a Dirac Hamiltonian,

1437:   like in refs.~\cite{Kel64,OvcEri77}.

1438:   This point has been recently rediscussed in ref.~\cite{SchTit03}.

1439:   \label{footnoteDD}

1440: }.

1441: 

1442: Other interesting models can be constructed by introducing disorder in the

1443: hoppings, instead. We refer to such models as {\it off-diagonal disorder}. The

1444: Dirac Hamiltonian (\ref{HDirac}) introduced above provides a continuum limit

1445: of such a model\footnote{

1446:   The continuum limit of a tight binding Hamiltonian with random hoppings

1447:   has been discussed in ref.~\cite{TakLinMak80} (see also the review in 

1448:   ref.~\cite{Mon95}).

1449:   As pointed in footnote \ref{footnoteDD}, random Dirac Hamiltonians 

1450:   can also appear as continuum limit of the band center of a 

1451:   discrete models with diagonal disorder. 

1452: }.

1453: Dirac Hamiltonians appear naturally in several contexts of condensed matter

1454: physics. The existence of symmetries in the Dirac Hamiltonian\footnote{

1455:   A complete classification of symmetries of disordered Hamiltonians 

1456:   extending the famous Wigner-Dyson ensembles of random matrix theory has 

1457:   been provided in refs.~\cite{Ver94,Zir96,AltZir97}

1458:   (see also the recent review article \cite{HeiHucZir05}). 

1459: }  

1460: (particle-hole, chiral,...) can lead to interesting features in the presence

1461: of disorder, which has attracted some attention

1462: (see for example

1463: refs.~\cite{NerTsvWen94,AltZir97,BroMudFur00,BroMudFur03,BocSerZir00,Boc00,EvaKat03},

1464: the last of these references gives a brief overview).

1465: The square of the Dirac Hamiltonian, 

1466: $H_{\rm D}^2=-{\D_x^2} + \phi(x)^2 +\sigma_3 \phi'(x)$,

1467: is related to the pair of supersymmetric Schr\"odinger isospectral 

1468: Hamiltonians $H_\pm=-{\D_x^2} + \phi(x)^2 \pm \phi'(x)$. When 

1469: $\mean{\phi(x)}=0$ we may forget the sign and simply consider~: 

1470: \be\label{Hsusy}

1471: H_S=-\frac{\D^2}{\D x^2} + \phi(x)^2 + \phi'(x)

1472: \:.\ee

1473: Such Hamiltonians appear in a variety of problems ranging from

1474: the 1d classical diffusion in a random force \cite{BouComGeoLeD90,IglMon05},

1475: electronic structure of polyacetylene \cite{TakLinMak80}, 

1476: spin Peierls chains \cite{FabMel97,FabMel97b,SteFabGog98} and also in a 

1477: continuum limit of the random field XY model \cite{GurCha03}

1478: (see ref.~\cite{ComTex98} for a short review on supersymmetric disordered

1479: quantum mechanics). Spectral and localization properties 

1480: have been studied in detail when $\phi(x)$ is a white noise 

1481: \cite{OvcEri77,LifGrePas88,BouComGeoLeD87,BouComGeoLeD90} and also

1482: when $\phi(x)$ has a finite correlation length as in a random telegraph process

1483: \cite{ComDesMon95} (this last case has found some application in the context

1484: of spin Peierls chains).

1485: 

1486: In the high energy limit, the localization properties and the statistics of

1487: the time delay do not show any difference with respect to the case of diagonal

1488: disorder. However the low energy properties are quite different. This is

1489: easily understood by noticing that, due to its relation to the Dirac 

1490: Hamiltonian, the supersymmetric Hamiltonian can be

1491: factorized as $H_S=Q^\dagger Q$ where $Q=-\D_x+\phi(x)$ and

1492: $Q^\dagger=\D_x+\phi(x)$. In particular, such a structure enforces a positive

1493: spectrum~:  ${\rm Spec}(H_S)\subset\RR^+$. 

1494: When $\phi(x)$ is white noise of zero mean, $\smean{\phi(x)}=0$ and

1495: $\smean{\phi(x)\phi(x')}=g\,\delta(x-x')$, the case considered below, the

1496: DoS presents a logarithmic Dyson singularity~: the integrated density 

1497: of states (IDoS) reads

1498: $N(E)\sim1/\ln^2E$ \cite{OvcEri77,BouComGeoLeD87,BouComGeoLeD90}, 

1499: while the localization length diverges logarithmically\footnote{

1500:   A more complete picture of the properties of this model at $E=0$ is given in

1501:   several works~:

1502:   (A) moments and correlations of the zero mode are studied in

1503:   refs.~\cite{SheTsv98,ComTex98}. 

1504:   (B) Additionally to the statistical properties of the time delay 

1505:   , eq.~(\ref{ChenTD}),

1506:   (C) the distribution of the transmission probability is derived in

1507:   ref.~\cite{SteCheFabGog99}. In particular it was shown that the 

1508:   average transmission through an interval of length $L$ decreases like 

1509:   $\smean{T}\propto1/\sqrt{L}$, which is slower than in the diffusive regime.

1510:   (D) The existence of a finite conductivity was

1511:   demonstrated in ref.~\cite{GogMel77}.

1512:   \label{footnoteSUSY}

1513: } $\lambda(E)\sim\ln(1/E)$.

1514: 

1515: It is interesting to investigate the extreme value statistics of the spectrum

1516: in the low energy regime, where we expect properties quite different from the

1517: one obtained for the diagonal disorder. The derivation follows closely 

1518: that in the previous case, however the relevant 

1519: random processes and the approximations are 

1520: different. The first step is to decouple the Schr\"odinger equation

1521: $H_S\varphi(x)=k^2\varphi(x)$ into two first order differential equations 

1522: ({\it i.e.} go back to the Dirac equation)~:

1523: \bea

1524: Q^\dagger \chi(x) &=& k \varphi(x) \\

1525: Q \varphi(x)      &=& k \chi(x)

1526: \:.\eea

1527: Then we may use the phase formalism by introducing a phase variable

1528: and an envelope variable~: 

1529: $\varphi(x)=\EXP{\xi(x)} \sin\vartheta(x)$ and 

1530: $\chi(x)=-\EXP{\xi(x)} \cos\vartheta(x)$.

1531: The phase variable obeys a stochastic differential equation with 

1532: a noise multiplying a trigonometric function of the phase.

1533: For convenience we introduce an additive process $\zeta(x)$ defined

1534: as $\zeta(x)=\pm\frac12\ln|\tan\vartheta(x)|$.

1535: The sign $+$ is chosen for 

1536: $(\vartheta\ {\rm mod}\ \pi)\in[0,\pi/2]$ and the sign $-$ for 

1537: $(\vartheta\ {\rm mod}\ \pi)\in[\pi/2,\pi]$. This new process obeys 

1538: the stochastic differential equation

1539: \be \label{eqzeta}

1540: \frac{\D}{\D x}\zeta = k\cosh2\zeta \pm \phi(x)

1541: \:.\ee

1542: Between two nodes of the wave function, the variable $\zeta$ crosses twice the

1543: interval $]-\infty,+\infty[$. Note that when $\phi(x)$ is a white noise of

1544: zero mean, the sign $\pm$ can be disregarded. The study of the time required

1545: by the process to cross the interval can be performed by the same method as

1546: above~:  we introduce the ``time'' $\tilde\Lambda(\zeta)$ needed 

1547: to go from $\zeta$ to $+\infty$.

1548: The Laplace transform of the distribution

1549: $h(\alpha,\zeta) = 

1550: \smean{ \EXP{-\alpha\tilde\Lambda} \:|\:\zeta(0)=\zeta;\ \zeta(\tilde\Lambda)

1551: =+\infty }$

1552: obeys a diffusion equation 

1553: $

1554: (k\cosh2\zeta\:\partial_\zeta + \frac{g}{2}\partial_\zeta^2)

1555: h(\alpha,\zeta) = \alpha \: h(\alpha,\zeta)

1556: $

1557: that involves the backward Fokker-Planck operator related to the 

1558: Langevin equation (\ref{eqzeta}).

1559: 

1560: In the low energy limit $k\ll g$ we expect that most of the ``time''

1561: $\Lambda\equiv\tilde\Lambda(-\infty)$ is spent in the region where the

1562: potential is almost flat (see figure~\ref{fig:potsusy}), therefore we replace

1563: the diffusion equation for $h(\alpha,\zeta)$ by the free diffusion equation

1564: $

1565: \frac{g}{2}\partial_\zeta^2h(\alpha,\zeta) = \alpha \: h(\alpha,\zeta)

1566: $ 

1567: on the finite interval $\zeta\in[\zeta_-,\zeta_+]$, with a reflecting boundary

1568: condition at one side $\partial_\zeta h(\alpha,\zeta_-)=0$ and 

1569: $h(\alpha,\zeta_+)=1$ at the other side (which corresponds actually to 

1570: the absorption at $\zeta=\zeta_+$). The coordinates $\zeta_\pm$ are

1571: the points where the deterministic force and the white noise have equal 

1572: strengths.

1573: 

1574: Now, we can obtain $h(\alpha,\zeta)$ straightforwardly.

1575: \be

1576: \smean{\EXP{-\alpha\Lambda}} \simeq h(\alpha,\zeta_-) 

1577: = \frac{1}{\cosh\sqrt{\alpha/N(E)}}

1578: \:,\ee

1579: where $N(E)={g}/{2\ln^2(g/k)}$ is the IDoS per unit length.

1580: 

1581: 

1582: \begin{figure}[!ht]

1583: \begin{center}

1584: \includegraphics{potUsusyb.eps}

1585: \end{center}

1586: \caption{\it The potential related to the deterministic force in 

1587:          eq.~(\ref{eqzeta}) felt by the  

1588:          variable $\zeta$.\label{fig:potsusy}}

1589: \end{figure}

1590: 

1591: %Strictly speaking there should be a $\pm$ sign in front of the $\phi(x)$, but

1592: %since $\phi(x)$ is a white noise of zero mean the sign plays no role and is

1593: %disregarded. eq.~(\ref{Riceq}) is a Langevin equation for a ``particle'' of

1594: %position $\zeta$ traveling from $-\infty$ to $+\infty$ in a potential

1595: %$U(\zeta)=-\frac{k}{2}\sinh2\zeta$.

1596: %As in the case of diagonal disorder, we study the distribution of the 

1597: %``time'' required by the variable $\zeta$ to go from $-\infty$ to $+\infty$.

1598: %We introduce the characteristic function for the  ``time'' 

1599: %$\tilde\Lambda(\zeta)$ 

1600: %needed to go from $\zeta$ to $+\infty$~:

1601: %$h(\alpha,\zeta) = 

1602: %\smean{ \EXP{-\alpha\tilde\Lambda} \:|\:\zeta(0)=\zeta;\ \zeta(\tilde\Lambda)

1603: %=+\infty }$

1604: %with boundary conditions 

1605: %$\partial_\zeta h(\alpha,\zeta)\big|_{-\infty}=0$ and 

1606: %$h(\alpha,+\infty)=1$.

1607: %It may be obtained by solving

1608: %$

1609: %(k\cosh2\zeta\:\partial_\zeta + \frac{g}{2}\partial_\zeta^2)

1610: %h(\alpha,\zeta) = \alpha \: h(\alpha,\zeta)

1611: %$.

1612: %We study this equation in the low energy limit $k\ll g$ by noticing that t

1613: %he effect of the random ``force'' $\phi(x)$ and  the deterministic ``force'' 

1614: %$-\partial_\zeta U(\zeta)$ decouple. Denoting by

1615: %$\zeta_\pm$  the two points where the deterministic force $k\cosh2\zeta$

1616: %is of the same order of magnitude as the Langevin force we then have the

1617: %following picture. The force

1618: %$-\partial_\zeta U(\zeta)$ brings the particle initially at $-\infty$ to 

1619: %$\zeta_-=-\frac12\ln(g/k)$ in a short time (of order $1/g$).

1620: %Then the particle diffuses normally to $\zeta_+=\frac12\ln(g/k)$

1621: %due to the Langevin ``force'' $\phi(x)$. At $\zeta_+$ the particle is 

1622: %brought back

1623: %to $+\infty$ mostly by the deterministic force again.

1624: %On $[\zeta_-,\zeta_+]$ the dynamics of the particle can be replaced 

1625: %by a normal diffusion with a reflecting 

1626: %boundary at $\zeta_-$ and an absorbing boundary at $\zeta_+$.

1627: 

1628: If $\Lambda\equiv\tilde\Lambda(-\infty)$ is the ``time'' needed to cross

1629: the interval, 

1630: the distance between two nodes of the wave function is a sum of two such

1631: (independent) random variables $\ell=\Lambda_1+\Lambda_2$. Therefore its 

1632: distribution is given by inverse Laplace transformation of 

1633: $\smean{\EXP{-\alpha\ell}}=1/\cosh^2\sqrt{\alpha/N(E)}$.

1634: We obtain~:

1635: \be\label{dol0susy2}

1636: P(\ell) = N(E) \, \varpi_0(N(E) \ell)

1637: \:,\ee

1638: where $\varpi_0(x)$ is the inverse Laplace transform of $\cosh^{-2}\sqrt{s}$~:

1639: \bea\label{pi0susy}

1640: \varpi_0(x) &=&  \heav(x) \sum_{m=0}^\infty

1641: \left[\pi^2(2m+1)^2x - 2\right] \EXP{ -\frac{\pi^2}{4}(2m+1)^2 x } 

1642: \APPROX{x\to\infty} \pi^2 x \EXP{ -\frac{\pi^2}{4} x }\\

1643: &=& \frac{4}{\sqrt{\pi}}\:\frac{\heav(x)}{x^{3/2}} 

1644: \sum_{m=1}^\infty(-1)^{m+1} m^2\EXP{-m^2/x}

1645: \APPROX{x\to0} \frac{4}{\sqrt{\pi}}\:\frac{\heav(x)}{x^{3/2}}\:\EXP{-1/x}

1646: \:,\eea

1647: where $\heav(x)$ is the Heaviside function.

1648: 

1649: We find the distribution of the ground state energy~:

1650: \be

1651: W_0(E) = L\rho(E)\,\varpi_0(LN(E))

1652: \:,\ee

1653: with the following limiting behaviour~:

1654: \bea \label{W0lim1}

1655: W_0(E) &\simeq& \frac{8}{\sqrt{2\pi gL}} \frac1E 

1656: \exp-\frac{\ln^2(g^2/E)}{2gL} 

1657: \hspace{1.2cm}{\rm for} \hspace{0.5cm} E\ll g^2\EXP{-\sqrt{2gL}}\\

1658: \label{W0lim2}

1659: &\simeq& \frac{8\pi^2g^2L^2}{E\ln^5(g^2/E)}

1660: \exp-\frac{\pi^2gL}{2\ln^2(g^2/E)} 

1661: \hspace{0.5cm} {\rm for} \hspace{0.5cm} g^2\EXP{-\sqrt{2gL}}\ll E\ll g^2

1662: \:.\eea

1663: In ref.~\cite{Tex00}, an integral representation of $W_n(E)$ is also given

1664: together with an explicit expression for $W_1(E)$.

1665: 

1666: It is also interesting to point out that the ground state distribution is 

1667: characterized by a typical value $E_0^{\typ} \simeq g^2 \EXP{-g L}$, 

1668: a median value $E_0^{\rm med} \sim g^2 \EXP{-\sqrt{gL}}$,

1669: and a mean value

1670: $\smean{E_0} \sim g^{2}\, (gL)^{1/2}\, \EXP{-C(gL)^{1/3}}$

1671: where $C$ is a numerical constant (see ref.~\cite{Tex00}).

1672: This latter expression has also been obtained in ref.~\cite{MonOshComBur96} 

1673: where upper and lower bounds were found using a perturbative 

1674: expression for the ground state energy as a functional of $\phi(x)$.

1675: 

1676: The distribution (\ref{pi0susy}) was obtained in ref.~\cite{LeDMonFis99} in

1677: the context of the classical diffusion by using a real space renormalization

1678: group method. In this case the distribution is interpreted as the distribution

1679: of the smallest relaxation time.

1680: 

1681: In summary, these two examples illustrate the fact that extreme value

1682: spectral statistics provides an information on correlations of eigenvalues~:

1683: the extreme value statistics of independent and identically distributed 

1684: random variables have been

1685: classified by Gumbel \cite{Gum54}. Therefore extreme value distribution for

1686: eigenvalues that differ from one of the three Gumbel's laws indicates level

1687: correlations.

1688: 

1689: 

1690: 

1691: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1692: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1693: 

1694: \section{Trace formulae, spectral determinant and diffusion on graphs\label{sec:graphs}}

1695: 

1696: \subsection{Introduction}

1697: 

1698: Up to now we have discussed several questions related to the physics of 

1699: weak and strong localization. In the previous section we have considered

1700: spectral properties while in the sections \ref{sec:wl} and \ref{sec:expfun}

1701: we mostly discussed transport properties.

1702: 

1703: In the present section we are going to review several results related to the

1704: study of the Laplace operator on metric graphs.  An object at the core of our

1705: discussion is the spectral determinant of the Laplace operator, formally

1706: defined as $S(\gamma)=\det(\gamma-\Delta)$, where $\gamma$ is a spectral

1707: parameter. This quantity encodes the information on the spectrum of the

1708: Laplace operator on the graph. The subsection \ref{sec:description} recalls

1709: the basic conventions required to describe metric graphs. The subsections

1710: \ref{sec:tf} and \ref{sec:sd} review general results on trace formulae and

1711: spectral determinants. These subsections will appear at first sight quite

1712: technical and unrelated to the previous sections. However we will see that

1713: there exists a close relation with the question of quantum transport~: the

1714: Laplace operator is the generator of the diffusion on the graph and its

1715: spectral properties play a central role in the study of transport. This

1716: connection, already evoked in the introduction and section~\ref{sec:wl}, will

1717: be emphasized again in subsection~\ref{sec:wlrg} where the explicit relation

1718: between quantum transport and spectral determinant is recalled.

1719: 

1720: One interest for spectral determinants is that they can be used as generating

1721: functions for various quantities characterizing the diffusion on the graph.

1722: This will be illustrated by using the connection with trace formulae

1723: (section~\ref{sec:tf}) and further exploited in subsection~\ref{sec:wlrg} in

1724: the context of quantum transport.  The efficiency of the method stems from the

1725: fact that, even though $S(\gamma)$ seems to be a complicated object at first

1726: sight, involving an infinite number of eigenvalues, it can be expressed as the

1727: determinant of a finite size matrix. This relation, established by Pascaud \&

1728: Montambaux \cite{Pas98,PasMon99}, allows computing easily and systematically

1729: the spectral determinant for arbitrary graphs.

1730: 

1731: The study of the Laplace operator on metric graphs (or {\it quantum graphs})

1732: is not restricted to transport and

1733: appears in many physical contexts ranging from 

1734: organic molecules \cite{RudSch53}, superconducting networks \cite{Ale83},

1735: phase coherent transport in networks of weakly disordered wires

1736: \cite{DouRam85,PasMon99,AkkComDesMonTex00,TexMon04}, 

1737: transport in mesoscopic networks 

1738: \cite{Sha82,Sha83,ButImrAzb84,GefImrAzb84,AvrSad91,VidMonDou00,TexMon01} 

1739: or metallic agregates \cite{CarAku04}.

1740: We are not going to review this history and refer the interested reader

1741: to the references~\cite{Ale83,AvrRavZur88,Col98,KotSmi99,AkkComDesMonTex00,KotSmi03} 

1742: (for a recent review see the appendix of Pavel Exner in 

1743: ref.~\cite{AlbGesHoeHol04}).  

1744: Mathematical aspects of quantum graphs are discussed in the recent issues of 

1745: Waves Random Media {\bf14} (2004) and Journal of Physics A~{\bf38} (n$^o$22) 

1746: (June 2005). See in particular the refs.~\cite{Kuc04,Kuc05}.

1747: 

1748: 

1749: \subsection{Description of metric graphs\label{sec:description}}

1750: 

1751: We collect some background material from the theory of graphs and establish

1752: our notations and conventions (illustrated on figure~\ref{fig:exampleqds}).

1753: 

1754: \noindent{\bf Vertices, bonds, arcs.--}

1755: Let us consider a network of $B$ wires (bonds) connected at $V$ vertices. The

1756: latter are labelled with Greek indices $\alpha$, $\beta$,... Therefore, the

1757: bonds of the graph can be denoted by a couple of Greek indices $(\ab)$. The

1758: oriented bonds, denoted as {\it arcs}, will also play an important role. The

1759: two arcs related to the bond $(\ab)$ will be labelled by $\ab$ and $\ba$ or

1760: more simply with Roman letters $i$, $j$,...

1761: 

1762: \noindent{\bf Adjacency matrix.--}

1763: The basic object is the adjacency

1764: $V\times V$~matrix $a_{\alpha\beta}$ characterizing the topology of the

1765: network~:  $a_{\alpha\beta}=1$ if $\alpha$ and $\beta$ are connected by a

1766: bond~; $a_{\alpha\beta}=0$ otherwise\footnote{

1767:   Note that this definition assumes that two vertices are connected by 

1768:   at most one bond and that a bond never forms a loop.

1769:   We emphasize that {\it this does not imply any particular restriction on 

1770:   the topology of the graph}~: one can always introduce a vertex on 

1771:   a bond, which separates the bond into two bonds, without modifying 

1772:   the properties of the graph. The numbers of vertices and bonds can 

1773:   always be made arbitrary large and this is partly a matter of choice. 

1774:   The case of closed bonds (forming a loop) or vertices connected by 

1775:   multiple bonds requires a simple generalization of the formalism

1776:   presented here (see for example ref.~\cite{Pas98} and appendix C of 

1777:   ref.~\cite{AkkComDesMonTex00}). This generalization allows in 

1778:   particular to minimize $B$ and $V$, which makes the computation 

1779:   sometimes easier.

1780: }.

1781: The connectivity of the vertex $\alpha$, denoted by $m_\alpha$, is 

1782: related to the adjacency matrix by $m_\alpha=\sum_\beta a_{\alpha\beta}$.

1783: Summed over the remaining index, the adjacency matrix gives  the 

1784: number of arcs~: $\sum_{\alpha,\beta} a_{\alpha\beta}=2B$.

1785: 

1786: \noindent{\bf Orbits.--}

1787: A path is an ordered set of arcs such that the end of an arc coincides with

1788: the begining of the following arc. The equivalence class of all closed paths

1789: equivalent by cyclic permutations is called an {\it orbit}.

1790: An orbit is said to be {\it primitive} when it can not be decomposed as 

1791: a repetition of a shorter orbit.

1792: 

1793: \begin{figure}[htbp]

1794:   \centering

1795:   \includegraphics[scale=1.25]{exampleqds.eps}

1796:   \caption{{\it An example of graph with 11 vertices and 13 bonds.

1797:            The arrows  show the orientations of the arc $i$ 

1798:            and the reversed arc $\bar i$.

1799:            The blue curve is an example of (primitive) orbit with one 

1800:            backtracking.}}

1801:   \label{fig:exampleqds}

1802: \end{figure}

1803: 

1804: \noindent{\bf Scalar functions.--}

1805: The graphs we consider here are not simply topological objects but have also

1806: some metric properties~: the bond $(\alpha\beta)$ is characterized by its 

1807: length $l_{\alpha\beta}$ and identified with the interval 

1808: $[0,l_{\alpha\beta}]$ of $\RR$. A scalar function $\psi(x)$ is defined

1809: by  its  components $\varphi_{\ab}(x_{\ab})$ on each bond,

1810: where $x_\ab$ is the coordinate measuring the distance along the bond 

1811: from the vertex $\alpha$ (note that $x_\ab+x_\ba=l_\ab$).

1812: 

1813: \noindent{\bf Continuous boundary conditions.--}

1814: When studying the Laplace operator $\Delta$ acting on scalar functions, 

1815: boundary conditions at the vertices must be specified in order to 

1816: ensure self adjointness of the operator.

1817: Let us introduce the notations~:

1818: \bea

1819: \label{phiab}

1820: \varphi_\ab &\equiv& \varphi_\ab(x_\ab=0) \\

1821: \label{phiprimeab}

1822: \varphi'_\ab &\equiv& \frac{\D\varphi_\ab}{\D x_\ab}(x_\ab=0)

1823: \:,\eea

1824: for the value of the function and its derivative at the vertex $\alpha$, along

1825: the arc $\ab$. 

1826: The continuous boundary conditions assume~:\\

1827: ({\it i}) continuity of the function at each vertex~: all the 

1828: components $\varphi_{\alpha\beta}$ for all $\beta$ neighbours of $\alpha$

1829: are equal. The value of the function at the vertex is denoted 

1830: $\varphi(\alpha)$.\\

1831: ({\it ii})

1832: $

1833: \sum_\beta a_{\alpha\beta} \varphi'_{\alpha\beta}=

1834: \lambda_\alpha\varphi(\alpha)

1835: $

1836: where the adjacency matrix in the sum constrains it to run over the

1837: neighbouring vertices of $\alpha$. Therefore the sum runs over all wires

1838: issuing from the vertex. The real parameter $\lambda_\alpha$ allows 

1839: describing several boundary conditions~: $\lambda_\alpha=\infty$ enforces the

1840: function to vanish, $\varphi(\alpha)=0$, and corresponds to the 

1841: Dirichlet boundary

1842: condition. $\lambda_\alpha=0$ corresponds to the Neumann boundary condition.

1843: We refer to the case of a finite $\lambda_\alpha$  as ``mixed boundary 

1844: conditions''.

1845: In the problem of classical diffusion, the Dirichlet condition describes the 

1846: connection to a reservoir that absorbs particles, while the Neumann condition

1847: ensures conservation of the probability current and describes an internal 

1848: vertex.

1849: 

1850: \noindent{\bf General boundary conditions.--}

1851: Thanks to the continuity hypothesis, the simple boundary conditions we have 

1852: just described allow us to introduce vertex variables $\varphi(\alpha)$ 

1853: (this is convenient since the number of vertices $V$ is  usually smaller than 

1854: the number of arcs $2B$).

1855: However these are not the most general boundary conditions, and in general the

1856: different components associated to the arcs issuing from a given vertex do not

1857: have {\it a priori} the same limit at the vertex. The most general boundary

1858: conditions can be written as~:

1859: \be

1860: C\,\varphi + D\,\varphi' =0

1861: \ee

1862: where $\varphi$ and $\varphi'$ are the column vectors of dimension $2B$

1863: collecting the variables (\ref{phiab}) and (\ref{phiprimeab}), respectively.

1864: $C$ and $D$ are two square $2B\times2B$ matrices. 

1865: In the arc formulation, the information on the topology of the graph is 

1866: encoded in these two matrices.

1867: The self adjointness of the Laplace operator is ensured if these two 

1868: matrices satisfy the following conditions \cite{KosSch99}~:

1869: ({\it i}) $CD^\dagger$ is self-adjoint.

1870: ({\it ii}) The  $2B\times4B$ matrix $(C,D)$ must have maximal rank.

1871: Note that the choice of $C$ and $D$ is not unique.

1872: The introduction of boundary conditions without continuity at the vertices has 

1873: been motivated physically in 

1874: refs.~\cite{Sha83,GefImrAzb84,ButImrAzb84,AvrExnLas94,Exn95,Exn96a}.

1875: 

1876: \noindent{\bf Scattering theory interpretation.--}

1877: We can give a more clear physical meaning to these conditions in a

1878: scattering setting. Let us consider the equation 

1879: $-\Delta\varphi(x)=E\varphi(x)$

1880: for a positive energy $E=k^2$. We decompose the component on the bond as the

1881: superposition of an incoming and an outgoing plane wave~:

1882: $\varphi_\ab(x_\ab)=I_\ab\,\EXP{-\I kx_\ab}+O_\ab\,\EXP{\I kx_\ab}$. It is

1883: convenient to collect the amplitudes in column vectors $I$ and $O$. The vector

1884: $I$ contains the incoming plane wave amplitudes and the vector $O$ 

1885: the outgoing plane wave amplitudes. Both are related by a vertex scattering

1886: matrix~:  $O=Q\,I$. The self-adjointness of the Schr\"odinger operator is now

1887: ensured by imposing the unitarity of the scattering matrix~: $Q^\dagger Q=1$.

1888: The relation between the two formulations yields~:

1889: \be\label{relQDC}

1890: Q=(\sqrt{\gamma}\,D-C)^{-1}(\sqrt{\gamma}\,D+C)

1891: \:.\ee 

1892: For the discussion below it is convenient to introduce the parameter

1893: $\gamma$ related to the energy by $E=-\gamma=k^2+\I0^+$ (the matrix $Q$ is

1894: unitary for $E>0$ only). 

1895: An example of a matrix $Q$ for one vertex without continuity of the wave

1896: function was given in refs.~\cite{Sha83,GefImrAzb84,ButImrAzb84}. This

1897: particular choice has become popular in mesoscopic physics (see also 

1898: ref.~\cite{TexMon01} where a more convenient parametrization was provided). 

1899: 

1900: \noindent

1901: Since we are considering here compact graphs, we are dealing only with discrete

1902: spectrum. The study of noncompact graphs (with some wires of infinite length),

1903: which have continuous spectra, requires a scattering theory approach,

1904: initiated by the work of Shapiro \cite{Sha82,Sha83} and discussed in 

1905: refs.~\cite{GerPav88,AvrSad91,Ada92,KosSch99,KotSmi00,BarGas01,KotSmi03,TexMon01,Tex02,TexBut03,TexDeg03}.

1906: 

1907: 

1908: \noindent{\bf Magnetic fluxes.--}

1909: The Laplace operator arises in the context of diffusion equation but also 

1910: in quantum mechanics. In this last case a natural generalization is to 

1911: introduce a magnetic field. This is achieved by introducing a 1-form 

1912: $A(x)\D x$ along the wires (the derivative must then be replaced by

1913: a covariant derivative~: $\D_x\to\Dc_x=\D_x-\I A(x)$). We denote by 

1914: $\theta_\ab=\int_\alpha^\beta\D x\,A(x)$ the corresponding line integral

1915: along the arc $\ab$.

1916: In the context of classical diffusion, magnetic fluxes and winding numbers are

1917: conjugated variables.

1918: 

1919: 

1920: \subsection{Trace formulae and zeta functions\label{sec:tf}}

1921: 

1922: Trace formulae play an important role in spectral theory. A famous example is

1923: the Selberg trace formula which may be viewed as an extension of the Poisson

1924: summation formula to non commutative groups \cite{Sel56}. An analogous formula

1925: in physics is the Gutzwiller trace formula \cite{Gut90} that has been

1926: extensively used in the context of quantum chaos and mesoscopic physics.

1927: Although one is exact and the other only a semiclassical approximation, both

1928: of them express the partition function (or the density of states, whose

1929: Laplace transform gives the partition function) as a sum over closed

1930: geodesics. They provide a connection between quantum properties (spectrum) and

1931: classical properties (classical trajectories). Below we discuss two examples

1932: of exact trace formulae that have been derived for graphs and their relation

1933: to spectral determinants.

1934: 

1935: 

1936: \subsubsection*{Roth's trace formula}

1937: 

1938: It expresses the trace of the heat kernel (partition function

1939: $Z(t)=\tr{\EXP{t\Delta}}$) as an infinite series of contributions of periodic

1940: orbits on the graph. This remarkable formula, due to Roth \cite{Rot83,Rot83a},

1941: applies to graphs with continuous boundary conditions with $\lambda_\alpha=0$.

1942: It it easy to include magnetic fluxes additionally~:

1943: \be\label{Roth}

1944: Z(t)=

1945: \frac{\cal L}{2\sqrt{\pi t}} + \frac{V-B}{2} + \frac{1}{2\sqrt{\pi t}}

1946: \sum_{{\cal C}} l(\widetilde{\cal C}) \alpha({\cal C})

1947: \EXP{-\frac{l({\cal C})^2}{4t}+\I\theta({\cal C})}

1948: \:.\ee

1949: ${\cal L}$ is the ``volume'' of the graph, {\it i.e.} the total length

1950: ${\cal L}=\sum_{(\ab)}l_\ab$.

1951: The sum runs over all orbits ${\cal C}=(i_1,i_2,\cdots,i_n)$ constructed in

1952: the graph. 

1953: $l({\cal C})=l_{i_1}+\cdots+l_{i_n}$ is the total length of the orbit,

1954: and $\theta({\cal C})$ the magnetic flux enclosed by it.

1955: $\widetilde{{\cal C}}$ designates the primitive orbit associated with a 

1956: given orbit ${\cal C}$.

1957: The weight $\alpha({\cal C})$ depends on the connectivity of the vertices 

1958: visited by the orbit~:

1959: $\alpha({\cal C})=\epsilon_{i_1i_2}\epsilon_{i_2i_3}\cdots\epsilon_{i_ni_1}$.

1960: The matrix $\epsilon$ couples the arcs of the graph\footnote{

1961:   Note that the parameter $\epsilon_{ij}$ has a simple interpretation in 

1962:   the scattering formulation \cite{TexMon01}~: it is the 

1963:   probability amplitude to be transmitted from arc $i$ to arc $j$.

1964:   The arc matrices $\epsilon$ and $Q$ are related by 

1965:   $\epsilon_{ij}=Q_{i\bar j}$ where $\bar j$ denotes the reversed arc.

1966: }~:\\

1967: $\bullet$ if $i$ ends at vertex $\alpha$ and $j$ starts from it, we have 

1968: $\epsilon_{ij}=2/m_\alpha$, where 

1969: $m_\alpha$ is the connectivities of the vertex.\\

1970: $\bullet$ If moreover $i$ and $j=\bar i$ are the reversed arcs

1971: $\epsilon_{i\bar i}=2/m_\alpha-1$.\\

1972: $\bullet$ Otherwise $\epsilon_{ij}=0$.

1973: 

1974: It is worth mentioning that the Roth trace formula has found recently some

1975: pratical applications to analyze magnetoconductance measurements on large

1976: square networks\footnote{   

1977:   More precisely, the expansion given below by eq.~(\ref{traceformulaS}) for

1978:   the Laplace transform of the partition function is relevant in this case.

1979: }~\cite{FerAngRowGueBouTexMonMai04}.

1980: 

1981: 

1982: \subsubsection*{Ihara-Bass trace formula}

1983: 

1984: Instead of considering metric graphs we now turn to graphs viewed as purely 

1985: combinatorial structures consisting of vertices connected by bonds 

1986: of equal lengths ($l_\ab=1$). 

1987: All the information is therefore encoded in the adjacency matrix $A$.

1988: In this setting, the Ihara $\zeta$-function is defined as

1989: \be

1990: \zeta(u)^{-1}  

1991: = \prod_{\widetilde{\cal C}_B} \left( 1 - u^{l(\widetilde{\cal C}_B)} \right)

1992: \:,\ee

1993: where the infinite product extends over all primitive backtrackless orbits

1994: $\widetilde{\cal C}_B$.

1995: The Ihara-Bass trace formula relates this infinite product to the determinant

1996: of a finite size matrix \cite{Bas92,StaTer96}~:

1997: \be\label{IharaZfct}

1998: \zeta(u)^{-1}  = (1-u^2)^{B-V}\,\det((1-u^2)\ide - u\,A + u^2\,Y)

1999: \ee

2000: where $\ide$ is the identity matrix, 

2001: $A$ is the adjacency matrix ($A_\ab\equiv a_\ab$) and $Y$ is

2002: the diagonal matrix encoding all connectivities~: $Y_\ab=\delta_\ab m_\alpha$.

2003: This relation was derived by Ihara \cite{Iha66} for regular graphs (all

2004: vertices with same connectivity) and later generalized to arbitrary graphs by

2005: Bass~\cite{Bas92}.

2006: The formalism that we have developed for metric graphs is flexible enough

2007: to describe these combinatorial structures in the same setting.

2008: We will see below that the $\zeta$-function is in fact directly related to the 

2009: spectral determinant. Moreover the Ihara-Bass trace formula can be further 

2010: generalized to include backtrackings.

2011: 

2012: 

2013: \subsection{Spectral determinant\label{sec:sd}}

2014: 

2015: In the physics literature spectral determinants arise in evaluating path

2016: integrals which are quadratic in the fluctuation around a given background

2017: field. A well known technique for regularizing such quadratic path integrals

2018: is the $\zeta$-function regularization. Given a certain operator ${\cal O}$

2019: whose eigenvalues $E_n$ are known, one defines the following $\zeta$-function~:

2020: $\zeta(s)=\sum_n E_n^{-s}$. This expression, which converges for $s$

2021: sufficiently large, can be analytically extended to a meromorphic function

2022: regular at the origin. The corresponding regularized determinant is then

2023: $\det{\cal O}=\exp(-\zeta'(0))$ ($\equiv\prod_nE_n$ formally). 

2024: One can find a general

2025: discussion on functional determinants in ref.~\cite{For87}.

2026: 

2027: %The success of functional methods and path integration has led to generalize

2028: %the concept of determinant to space of infinite dimensions. In this case

2029: %it is not possible to perform the infinite product over the eigenvalues

2030: %$\{E_n\}$. A possible way to define a determinant is to introduce the 

2031: %$\zeta$-function $\zeta(s)=\sum_n E_n^{-s}$, the determinant being defined 

2032: %as $\exp(-\zeta'(0))$ (see ref.~\cite{For87} for a general discussion on 

2033: %functional determinants).

2034: 

2035: In the context of graphs another regularization of the determinant of the

2036: Laplace operator has been used and has proved to be directly related to several

2037: physical quantities. If we introduce the trace of the resolvent

2038: $g(\gamma)=\sum_n(\gamma+E_n)^{-1}$, the spectral determinant is defined as

2039: $S(\gamma)=\exp(\int^\gamma\D\gamma'\,g(\gamma'))$ (it can be formally written

2040: as\footnote{

2041:   The limit $\gamma\to0$ of $S(\gamma)$ has been

2042:   discussed in ref.~\cite{AkkComDesMonTex00}. 

2043: }

2044: $S(\gamma)=\prod_n(\gamma+E_n)$). Moreover, the spectral parameter $\gamma$

2045: has in some cases a physical meaning (see section~\ref{sec:wlrg}).

2046: 

2047: \mathversion{bold}

2048: \subsubsection*{Laplace operator $\Delta$ with continuous boundary conditions}

2049: \mathversion{normal}

2050: 

2051: Pascaud \& Montambaux have shown in refs.~\cite{Pas98,PasMon99} that 

2052: $S(\gamma)=\det(\gamma-\Delta)$ can be 

2053: related to the determinant of a $V\times V$-matrix~:

2054: \be\label{spedet}

2055: S(\gamma) =

2056: \gamma^{\frac{V-B}2}

2057: \prod_{(\alpha\beta)}\sinh(\sqrt\gamma l_{\alpha\beta})\:

2058: \det M

2059: \:,\ee

2060: where the product runs over all bonds of the network.

2061: The $V\times V$-matrix $M$ is defined as~:

2062: \be\label{defM}

2063: M_\ab=\delta_\ab

2064: \left(\frac{\lambda_\alpha}{\sqrt\gamma} + \sum_\mu a_{\alpha\mu}

2065:       \coth(\sqrt{\gamma}l_{\alpha\mu})\right)

2066: -a_\ab\frac{\EXP{-\I\theta_\ab}}{\sinh(\sqrt{\gamma}l_\ab)}

2067: \:,\ee

2068: where the adjacency matrix constrains the sum to run over all vertices

2069: $\mu$ connected to $\alpha$. The matrix $M$ encodes all information about the

2070: network~: topology (matrix $a_{\alpha\beta}$), lengths of the wires ($l_\ab$),

2071: magnetic fluxes ($\theta_\ab$), boundary conditions ($\lambda_\alpha$). Below,

2072: we  give several examples which show how $S(\gamma)$ is related to the

2073: characteristic function of various interesting functionals of Brownian

2074: curves on a graph.

2075: 

2076: The expression (\ref{spedet}) was originally derived in ref.~\cite{PasMon99}

2077: by constructing the Green's function in the graph and eventually integrating

2078: it~:

2079: $\int\D x\,\bra{x}\frac1{\gamma-\Delta}\ket{x}=\drond{}{\gamma}\ln S(\gamma)$.

2080: A more direct derivation using path integral was later obtained  in 

2081: ref.~\cite{AkkComDesMonTex00}.

2082: 

2083: The relation between the Roth trace formula and the result (\ref{spedet})

2084: was addressed in ref.~\cite{AkkComDesMonTex00}. The main difficulty to

2085: establish this connection is to go from vertex variables to the arc language

2086: of eq.~(\ref{Roth}). A first step is to relate the determinant of the

2087: vertex-matrix, eq.~(\ref{spedet}), to the determinant of an arc-matrix~:

2088: \be\label{LfctS}

2089: S(\gamma)=\gamma^{\frac{V-B}{2}}\EXP{\sqrt{\gamma}\,{\cal L}}

2090: \,\det(\ide - QR)

2091: \:.\ee

2092: $R$ is the $2B\times2B$~matrix 

2093: $R_{ij}=\delta_{i\bar j}\,\EXP{-\sqrt{\gamma} l_i + \I\theta_i}$, where 

2094: $\bar j$ denotes the reversed arc.

2095: The matrix $Q$ was introduced above (eq.~(\ref{relQDC})) and is related to 

2096: $\epsilon$ by $Q_{ij}=\epsilon_{i\bar j}$. 

2097: For the continuous boundary conditions with $\lambda_\alpha=0$ it is given by

2098: $Q_{ii}=2/m_\alpha-1$, 

2099: $Q_{ij}=2/m_\alpha$ if $i$ and $j$ both issue from the vertex $\alpha$.

2100: $Q_{ij}=0$ in other cases.

2101: Eq.~(\ref{LfctS}) holds for the most simple boundary conditions~: 

2102: continuous with $\lambda_\alpha=0$. The general case is discussed below.

2103: Expanding the determinant by using 

2104: $\ln\det(\ide - QR)=- \sum_{n=1}^\infty \frac{1}{n}\tr{(QR)^n}$, 

2105: we eventually express the spectral determinant as an infinite product over the 

2106: primitive orbits~:

2107: \be\label{zetafctS}

2108: S(\gamma)=\gamma^{\frac{V-B}{2}}\EXP{\sqrt{\gamma}\,{\cal L}}

2109: \prod_{\widetilde{\cal C}}

2110: \left(  

2111:   1 - \alpha(\widetilde {\cal C})

2112:   \EXP{-\sqrt{\gamma}\,l(\widetilde {\cal C})+\I\theta(\widetilde {\cal C})}

2113: \right)

2114: \:.\ee

2115: This shows that the spectral determinant is a zeta function 

2116: (references on zeta functions on graphs are refs.~\cite{StaTer96,Che99}).

2117: The last step to connect this formula to Roth's trace formula is to 

2118: notice that 

2119: \be\label{traceformulaS}

2120: \drond{}{\gamma}\ln S(\gamma)=\frac{{\cal L}}{2\sqrt\gamma}+\frac{V-B}{2\gamma}

2121: +\frac{1}{2\sqrt\gamma}\sum_{{\cal C}} l(\widetilde {\cal C})

2122: \alpha({\cal C})\EXP{-\sqrt{\gamma}\,l({\cal C})+\I\theta({\cal C})}

2123: \:,\ee

2124: where the sum now runs over all orbits (if ${\cal C}$ is not primitive, 

2125: $\widetilde{\cal C}$ designates the related primitive orbit).

2126: Finally we perform an inverse Laplace transform of this expression,

2127: $\int_0^\infty\D t\,Z(t)\EXP{-\gamma t}=\drond{}{\gamma}\ln S(\gamma)$

2128: and eventually recover eq.~(\ref{Roth}).

2129: Examples of applications are studied in ref.~\cite{AkkComDesMonTex00}. 

2130: 

2131: 

2132: \mathversion{bold}

2133: \subsubsection*{Schr\"odinger operator $-\Delta+V(x)$ with general boundary conditions}

2134: \mathversion{normal}

2135: 

2136: The result (\ref{spedet}) of Pascaud \& Montambaux has been generalized by one

2137: of us. In refs.~\cite{Des00,Des00a} a similar formula was obtained for the

2138: spectral determinant of the Schr\"odinger operator $-\Delta+V(x)$ (the Hill

2139: operator), where $V(x)$ is a scalar potential defined on the graph. In

2140: ref.~\cite{Des01} the formula was further extended to describe general

2141: boundary conditions as well. As an illustration we construct the generating

2142: function of the number of closed orbits with a given number of backtrackings

2143: \cite{Iha66,WuKun99}.

2144: 

2145: \vspace{0.25cm}

2146: 

2147: The starting point is to introduce 

2148: two linearly independent solutions of the differential equation

2149: $(-\D_x^2+V_\ab(x)+\gamma)f(x)=0$ on $[0,l_\ab]$. We associate each 

2150: solution with an arc. Let us denote $f_\ab(x_\ab)$ the function

2151: satisfying 

2152: \be

2153: f_\ab(0)=1 

2154: \hspace{0.5cm} \mbox{ and } \hspace{0.5cm}

2155: f_\ab(l_\ab)=0

2156: \:.\ee 

2157: Therefore a second solution of the differential equation is naturally denoted 

2158: $f_\ba(x_\ba)=f_\ba(l_\ab-x_\ab)$.

2159: The Wronskian of these two solutions, defined as

2160: $W_\ab=f_\ab(x_\ab)\frac{\D f_\ba(x_\ba)}{\D x_\ab}

2161:       -\frac{\D f_\ab(x_\ab)}{\D x_\ab}f_\ba(x_\ba)$, 

2162: is constant along the bond~:

2163: $W_\ab=W_\ba=-f'_\ab(l_\ab)=-f'_\ba(l_\ab)$.

2164: If we consider the case $V(x)=0$, the solution is simply

2165: $

2166: f_\ab(x_\ab)

2167: =\frac{\sinh\sqrt\gamma(l_\ab-x_\ab)}{\sinh\sqrt\gamma l_\ab}

2168: \equiv\frac{\sinh\sqrt\gamma x_\ba}{\sinh\sqrt\gamma l_\ab}

2169: $.

2170: 

2171: All the required information about the potential is contained in the 

2172: $2B\times 2B$ arc-matrix $N$, defined as

2173: \be\label{n161}

2174: N_{\alpha\beta,\mu\eta}  = 

2175: \delta_{\alpha\mu}  \delta_{\beta\eta} 

2176: f'_\ab(0) 

2177: - \delta_{\alpha\eta}\delta_{\beta\mu} f'_\ab(l_\ab)

2178: \:.\ee

2179: This matrix couples a given arc to itself and to its reversed arc, only.

2180: If we assume that {\it the matrices $C$ and $D$ are independent on the 

2181: spectral parameter $\gamma$}, it was shown in ref.~\cite{Des01} that~:

2182: \be\label{n40} 

2183:  S(\gamma ) = \det(\gamma-\Delta+V(x)) = 

2184:  \prod_{(\alpha\beta)}  \frac{1}{\wal}\ \det(C+DN)

2185: \ee

2186: where the product runs over all bonds. 

2187: Functional determinants on a segment of $\RR$ with general boundary conditions

2188: at the boundaries have been studied by McKane \& Tarlie \cite{McKTar95}

2189: using the formalism developed by Forman~\cite{For87}.

2190: 

2191: It is also interesting to encode the information on the potential

2192: $V(x)$ in the matrix $R$ defined as\footnote{

2193:   Let us remark that, for the free case ($V(x)\equiv 0$),

2194:   we recover the expression of the matrix $R$ given above, coupling the arc 

2195:   $\ab$ to the reversed arc $\ba$ only~:

2196:   $

2197:   R_{\alpha\beta,\mu\eta} 

2198:   =\delta_{\alpha\eta}\delta_{\beta\mu} \EXP{-\sqrt{\gamma} \lab }

2199:   $. 

2200: }~:

2201: \be\label{n41} 

2202:   R \equiv  ( \sqrt{\gamma} \ide + N )( \sqrt{\gamma} \ide - N  )^{-1}

2203: \:.\ee

2204: Then

2205: \bea\label{n42} 

2206:  S(\gamma) =  \prod_{(\alpha\beta)} \frac{1}{ \wal } \:

2207:  \frac{1}{\det( \ide + R )} \: \det (C - \sqrt{\gamma} D ) \:

2208:  \det( \ide - Q R ) 

2209: \eea

2210: where the matrix $Q=(\sqrt{\gamma} D - C )^{-1}(\sqrt{\gamma} D + C )$

2211: was defined above by eq.~(\ref{relQDC}).

2212: Let us explain the structure of eq.~(\ref{n42})~: the term 

2213: $\det( \ide + R )\prod_{(\alpha\beta)}W_\ab$ contains informations on the 

2214: potential only\footnote{

2215:   The matrix $N$ encodes the information about the potential on the bonds 

2216:   through $f'_\ab(0)$ and $f'_\ab(l_\ab)$.

2217:   This information can also be introduced through transmission $t_\ab$ and 

2218:   reflection $r_\ab$ amplitudes by the potential $V_\ab(x)$. This relation

2219:   is developed in ref.~\cite{TexMon01}.

2220:   It is interesting to point out that 

2221:   $\det( \ide + R )\prod_{(\alpha\beta)}W_\ab

2222:   =2^B\gamma^{B/2}\prod_{(\alpha\beta)}t_\ab

2223:   =2^B\gamma^{B/2}\prod_{(\alpha\beta)}R_{\ab,\ba}$.

2224: }.

2225: The factor $\det (C -\sqrt{\gamma}D)$ contains only information on the 

2226: topology of the graph.

2227: The most interesting part is the last term $\det(\ide-QR)$ combining both

2228: informations. In particular this last part generates the infinite

2229: contributions of primitive orbits in (\ref{zetafctS}).

2230: 

2231: 

2232: 

2233: 

2234: \subsubsection*{Permutation-invariant boundary conditions.} 

2235: 

2236: In the previous section the matrices describing the graph are $2B\times2B$ arc

2237: matrices. A simplifiaction can be brought by passing to vertex variables. In

2238: the arc formulation, $C$ and $D$ define the topology of the graph~: two arcs

2239: coupled by $C$ and/or $D$ issue from the same vertex. Therefore it is possible

2240: to organize the basis of arcs in such a way that the matrices $C$ and $D$ have

2241: similar block diagonal structures. The matrices are made of $V$ square blocks,

2242: each corresponding to a vertex. A given block, of dimension

2243: $m_\alpha\times{m_\alpha}$ and denoted $C_{\alpha}$ (and $D_{\alpha}$),

2244: corresponds to the $m_\alpha$ arcs issuing from the vertex $\alpha$. If we

2245: assume that the boundary conditions are invariant under any permutation of the

2246: nearest neighbours of $\alpha$, then it is possible to introduce vertex

2247: variables. In this case we can write:

2248: \begin{eqnarray}

2249:    C_{\alpha } &=&  c_{\alpha } \ide  + t_{\alpha }  F_{\alpha }

2250:    \label{bci} \\ 

2251:    D_{\alpha } &=&  d_{\alpha } \ide  + w_{\alpha }  F_{\alpha }

2252:    \label{bci1} 

2253: \end{eqnarray}

2254: where 

2255: %$\ide$ is the unit matrix and 

2256: $F_{\alpha }$ is a matrix with all

2257: its elements equal to $1$. The boundary conditions at the vertex $\alpha$ are

2258: characterized by the four parameters $c_{\alpha }$, $d_{\alpha }$,

2259: $t_{\alpha }$ and $w_{\alpha }$ (note however that this choice

2260: is not unique).

2261: 

2262: \vspace{0.25cm}

2263: 

2264: Now, let us show  that, for boundary conditions

2265: given by eqs.~(\ref{bci},\ref{bci1})

2266: and $V(x)\ne 0$, the spectral determinant can be expressed in terms

2267: of the vertex $V \times V$-matrix.

2268: 

2269: We proceed as before but, this time, we consider, for each bond, two other

2270: independent solutions, $\cab(x_\ab)$ and $\cba(l_\ab-x_\ab)=\cba(x_\ba)$, of

2271: the equation $(-\D_{x_\ab}^2+V_\ab(x_\ab)+\gamma)\,\chi(x_\ab)=0$ that 

2272: satisfy the following conditions:

2273: \bea

2274: c_{\alpha}\, \chi_\ab(0)     + d_{\alpha}\,\chi'_\ab(0) &=& 1 \\

2275: c_{\beta} \, \chi_\ab(l_\ab) - d_{\beta} \,\chi'_\ab(l_\ab) &=& 0

2276: % c_{\alpha }\cab (\alpha ) + 

2277: % d_{\alpha }\frac{ \D \cab }{ \D \xab } (\alpha ) = 1

2278: %   \qquad &;& \qquad

2279: % c_{\beta }\cab (\beta ) + 

2280: % d_{\beta }\frac{ \D \cab }{ \D \xba } (\beta ) = 0     

2281: % \label{p1}  \\

2282: % c_{\alpha }\cba (\alpha ) + 

2283: % d_{\alpha }\frac{ \D \cba }{ \D \xab } (\alpha ) = 0

2284: %  \qquad &;& \qquad

2285: % c_{\beta }\cba (\beta ) + 

2286: % d_{\beta }\frac{ \D \cba }{ \D \xba } (\beta ) = 1        

2287: \label{p4}  

2288: \eea

2289: 

2290: 

2291: \noindent

2292: We denote by $\Xi_\ab$ the Wronskian of $\cab$ and $\cba$.

2293: Following the same steps as before, we  get 

2294: the spectral determinant (up to a multiplicative constant)~:

2295: \be\label{p19} 

2296:  S(\gamma) =  \prod_{(\alpha\beta)}\frac{1}{ \Xi_\ab }\ \det M

2297: \ee

2298: where $M$ is the $V \times V$-matrix~:

2299: %\bea 

2300: % M_{\alpha\alpha } &=& 1+ t_{\alpha }

2301: %  \sum_\mu a_{\alpha\mu} \chi_{\alpha\mu} (0)   +

2302: %  w_{\alpha } \label{p12} \sum_\mu a_{\alpha\mu} \chi_{\alpha\mu}'(0)  \\

2303: % M_{\alpha\beta } &=& ( c_{\alpha } w_{\alpha } -  t_{\alpha } d_{\alpha } )

2304: % \Xi_\ab   \qquad  \mbox{ \ if $(\alpha\beta )$ is a bond}   \label{p13}    \\

2305: %                &=& 0 \qquad \ \mbox{ otherwise} \nonumber

2306: %\eea

2307: \begin{eqnarray} \label{p12}

2308:   M_{\ab} = 

2309:   \delta_{\ab}

2310:   \left(

2311:     1 + \sum_\mu a_{\alpha\mu} 

2312:         \left[

2313:           t_{\alpha }\,\chi_{\alpha\mu}(0)+w_{\alpha }\,\chi_{\alpha\mu}'(0)

2314:         \right]

2315:   \right)

2316:   + a_\ab\, 

2317:   \left[ c_{\alpha } w_{\alpha } -  t_{\alpha } d_{\alpha } \right]\,\Xi_\ab 

2318: \:.

2319: \end{eqnarray}

2320: The eq.~(\ref{n40}), expressing the determinant in terms of arc matrices, and

2321: the eq.~(\ref{p19}), expressing it in terms of vertex matrix, have been

2322: derived up to multiplicative constants independent on $\gamma$. We can

2323: establish a precise relation by comparing their behaviour for

2324: $\gamma\to\infty$. We end up with~:

2325: \begin{equation}\label{p20} 

2326:  \prod_{(\alpha\beta)}\frac{1}{ \wal }\ \det(C+DN)

2327:  =  \prod_{(\alpha\beta)}\frac{1}{ \Xi_\ab }\ \det M

2328: \:.

2329: \end{equation}

2330: 

2331: 

2332: \mathversion{bold}

2333: \subsubsection*{Free case ($V(x)=0$). Applications~: counting backtrackings}

2334: \mathversion{normal}

2335: 

2336: In this subsection we show an application of generalized boundary conditions

2337: to count backtrackings (the figure~\ref{fig:exampleqds} shows an 

2338: orbit with one backtracking).

2339: 

2340: We now consider  the case $V(x)\equiv0$ still with 

2341:  permutation-invariant boundary conditions.

2342: 

2343: \noindent

2344: With the notations

2345: $$

2346:     \eta_{\alpha } = \frac{c_{\alpha }   +  \sg d_{\alpha } }

2347:                             { c_{\alpha }  -  \sg d_{\alpha } }

2348:          \qquad , \qquad                      

2349:      \rho_{\alpha } = \frac{\mu_{\alpha }^- - \mu_{\alpha }^+ }

2350:                             {1+ m_{\alpha } \mu_{\alpha }^- }

2351:     \qquad \mbox{and} \qquad

2352: \mu_{\alpha }^{\pm } = \frac{t_{\alpha }   \pm  \sg w_{\alpha } }

2353:       { c_{\alpha }  \pm  \sg d_{\alpha } }

2354: $$

2355: eqs.~(\ref{p20}, \ref{n41}, \ref{relQDC}) lead to~:

2356: \be\label{p804} 

2357: \det (\ide - Q R) = 2^{-V} \prod_{\alpha } \left( \rho_{\alpha }

2358: \eta_{\alpha } \right) \prod_{(\alpha\beta)}\left( 

2359:  1- \eta_{\alpha } \eta_{\beta } \EXP{- 2 \sg \lab}   \right)  \; 

2360:  \det{M} 

2361: \ee

2362: and the $V\times V$-matrix $M$ takes the form~:

2363: \bea

2364: \label{p805}

2365: {M}_{\alpha\beta} = 

2366: \delta_{\alpha\beta} \left(

2367:   \frac{2}{ \rho_{\alpha }\eta_{\alpha } } - 

2368:   \frac{ m_{\alpha }}{\eta_{\alpha } } +  \frac{1}{ \eta_{\alpha }} 

2369:   \sum_\mu  a_{\alpha\mu}

2370:   \frac{1+\eta_{\alpha }\eta_{\mu} \EXP{-2\sg l_{\alpha\mu}}  }

2371:        {1-\eta_{\alpha }\eta_{\mu } \EXP{-2\sg l_{\alpha\mu}} }

2372: \right)

2373: -a_{\alpha\beta}

2374:    \frac{2\,  \EXP{- \sg l_{\alpha\beta }}                         }

2375:         {1-\eta_{\alpha }\eta_{\beta }\EXP{-2\sg l_{\alpha\beta }} }

2376: \:.\eea

2377: Note that the expression (\ref{defM}) is recovered 

2378: for $\eta_\alpha=1$ and $\rho_\alpha=2/(m_\alpha+\lambda_\alpha/\sqrt\gamma)$.

2379: 

2380: For permutation-invariant boundary conditions, the matrices $C$, $D$ and

2381:  $Q$  (eq.~(\ref{relQDC})) are block-diagonal. The block $Q_{\alpha }$ 

2382: takes the simple form:

2383: \be\label{p900}

2384:      Q_{\alpha } = \eta_{\alpha } 

2385:      \left( -\ide + \rho_{\alpha }  F_{\alpha } \right)

2386: \ee

2387: The only non-vanishing elements of the matrix $QR$ are~:

2388: \be\label{p901}

2389:     (QR)_{\alpha\beta,\mu\alpha}= \left(

2390:     \rho_{\alpha }\eta_{\alpha } - \eta_{\alpha } \; \delta_{\beta\mu } 

2391:  \right) \; \EXP{- \sg l_{\alpha\mu }} 

2392: \ee

2393: 

2394: \vskip.3cm

2395: 

2396: \noindent

2397: We call 

2398: $\rho_{\alpha }\eta_{\alpha }$  the transmission factor

2399: at vertex $\alpha $ and $\rho_{\alpha}\eta_{\alpha}-\eta_{\alpha}$

2400: the reflection factor.

2401: By using the same expansion as above we write

2402: \be\label{p902}

2403:  \det(\ide - QR ) 

2404: = \prod_{\widetilde{\cal C}} \left( 1- \mu(\widetilde{\cal C} )

2405:  \EXP{-\sqrt\gamma l(\widetilde{\cal C})} \right)

2406: \:,\ee

2407: where the product is taken over all primitive orbits $\widetilde{C}$ whose

2408: lengths are denoted by $l(\widetilde{C})$.

2409: An orbit being a succession of arcs

2410: $\ldots  ,\tau\alpha,\alpha\beta,\ldots $  with, in $\alpha$,

2411: a reflection (if $\tau =\beta $) or a transmission  (if $\tau \ne \beta $),

2412: the weight $\mu(\widetilde{C})$, in eq.~(\ref{p902}),

2413: will be the product of

2414: all the reflection -- or transmission -- factors along~$\widetilde{C}$.  

2415: 

2416: 

2417: \vspace{0.25cm}

2418: 

2419: \noindent 

2420: {\it Graphs with wires of equal lengths.--}

2421: We consider the case of equal lengths $\lab=l$~; then we can choose

2422: $\gamma=1$ without loss of generality and introduce the notation

2423: $u\equiv\EXP{-l}$.

2424: It is clear from eq.~(\ref{p900}) that a backtracking at vertex $\alpha$ 

2425: brings a factor $\eta_{\alpha}$ in the weight $\mu(\widetilde{C})$. 

2426: In order to count backtrackings one has to choose the boundary conditions

2427: $\rho_{\alpha }\eta_{\alpha }=1$ and $\eta_{\alpha}=\eta$. Eq.~(\ref{p804})

2428: takes the simple form~:

2429: \be\label{p903}

2430: \prod_{\widetilde{C}_m} \left( 1 - (1-\eta )^{n_R(\widetilde{C}_m )} u^m

2431:  \right) = (1- \eta^2 u^2 )^{B-V} \det 

2432:  \left( ( 1- \eta^2 u^2  ) \ide +\eta u^2\, Y - u\, A   \right)

2433:  \equiv  Z^{-1}

2434: \ee

2435: $m$ is the number of arcs of the primitive orbit $\widetilde{C}_m$ and

2436: $n_R(\widetilde{C}_m)$ is the number of reflections (backtrackings) 

2437: occuring along  $\widetilde{C}_m$. 

2438: $Y$ is the $V\times V$-matrix 

2439: $Y_{\alpha\beta }=\delta_{\alpha\beta } \; m_{\alpha }$ and $A$ is the

2440: adjacency matrix.

2441: 

2442: Setting $\eta=1$ implies $n_R(\widetilde{C}_m)=0$ in the left-hand side of

2443: eq.~(\ref{p903})~: we recover Ihara-Bass formula \cite{Iha66,Bas92,StaTer96} 

2444: where

2445: only primitive orbits without tails and backtrackings are kept. (Ihara

2446: \cite{Iha66} established this formula for a regular graph~; the proof for a

2447: general graph is given in refs.~\cite{Bas92,StaTer96} using a direct counting

2448: technique).

2449: 

2450: %Note that some functional relation between generating functions of lengths and

2451: %number of backtrackings of open paths has been derived in ref.~\cite{Bar99}

2452: %for regular graphs. When applied to closed orbits the functional relation is a

2453: %simple consequence of the relation (\ref{p903}).

2454:  

2455: \vspace{0.25cm}

2456: 

2457: Now, let us consider closed random walks with a given number

2458: of backtrackings.

2459: 

2460: Eq.~(\ref{p903}) provides a non-trivial generalization of the Ihara-Bass

2461: formula (an independent derivation is also given in ref.~\cite{Bar99}).

2462: As an application let us consider the problem of enumerating $m$-steps

2463: random walks with $p$-backtracking steps~\cite{WuKun99}.

2464: Taking $Z$ in (\ref{p903}), we get: 

2465: \be\label{p1949}

2466: u \; \frac{ \D \ln Z}{ \D u} = \sum_{m=2}^{\infty } \; \sum_{p=0}^m \; 

2467: \sum_{\alpha =1}^V \; N_m^p(\alpha ) \; (1-\eta )^p \; u^m 

2468: \ee

2469: \noindent 

2470: where $ N_m^p(\alpha ) $ is the number of $m$-steps closed 

2471: random walks on the graph starting at $\alpha$, with $p$ backtrackings.

2472: 

2473: \vspace{0.25cm}

2474: 

2475: \noindent 

2476: {\it Example~: The complete graph.--}

2477: For the complete graph\footnote{

2478:   The complete graph $K_V$ with $V$ vertices is the $V-1$-simplex~: 

2479:   each vertex is connected to all other vertices (see figure~\ref{fig:k5}).

2480: } $K_V$, we get the results: 

2481: \bea

2482:   N_2^0(\alpha) &=& 0                              \nonumber \\

2483:   N_3^0(\alpha) &=&    (V-1)(V-2)                  \nonumber \\

2484:   N_4^0(\alpha) &=&    (V-1)(V-2)(V-3)             \nonumber \\

2485:   N_5^0(\alpha) &=&    (V-1)(V-2)(V-3)(V-4)        \nonumber \\

2486:   N_6^0(\alpha) &=&    (V-1)(V-2)(V^3-9V^2+29V-32) \label{p1950} 

2487: \eea

2488: \noindent 

2489: and also:

2490: \bea

2491:  N_2^1(\alpha) &=&  N_3^1(\alpha ) \ = \ N_4^1(\alpha ) \ = \ 0   \nonumber \\

2492:  N_5^1(\alpha) &=&  5(V-1)(V-2)(V-3)     \nonumber \\

2493:  N_6^1(\alpha) &=&  6(V-1)(V-2)(V-3)^2  \label{p1952} 

2494: \eea

2495: Note that these expressions have been obtained when vertices are all 

2496: characterized by the same parameter $\eta$. Introducing different parameters

2497: $\eta_\alpha$ allows counting the backtrackings at a given vertex.

2498: 

2499: \begin{figure}[!ht]

2500: \begin{center}

2501: \includegraphics[scale=0.6]{k5.eps}

2502: \end{center}

2503: \caption{\it The complete graph $K_5$ (the $4$-simplex)~: 

2504:          all vertices are connected to each 

2505:          other by bonds of equal lengths.\label{fig:k5}}

2506: \end{figure}

2507: 

2508: 

2509: 

2510: \subsection{Quantum chaos on metric graphs}

2511: 

2512: It has been recently realized that metric graphs are interesting 

2513: models for  quantum chaos. The paper of Kottos \& Smilansky

2514: \cite{KotSmi97} has stimulated several works on spectral statistics and

2515: level correlations \cite{KotSmi99,BarGas00,Tan01b}, and  

2516: renewed the interest in the Roth trace formula \cite{Rot83}.

2517: Progress in the understanding of universality of spectral statistics for

2518: generic quantum graphs has been achieved by Gnutzmann \& Altland

2519: \cite{GnuAlt04}. One of the input of this work is the observation that the

2520: spectral average for a given graph with incommensurate bond lengths is

2521: equivalent to an average over a certain ensemble of unitarity matrices. Star

2522: graphs \cite{BerKea99,KeaMarWin03} have provided simple examples of systems

2523: with intermediate level statistics similar to the one observed in {\v S}eba

2524: billiards and a precise connection has been established in

2525: ref.~\cite{BerBogKea01}. There is however an interesting class of graphs which

2526: do not enter in this category \cite{Tan05}. Chaotic scattering and transport

2527: properties in open graphs (with some infinitely long wires) have also been

2528: studied in refs.~\cite{KotSmi00,BarGas01,KotSmi03}. Since 

2529: quantum chaos is not the

2530: central subject of our review,  this list of papers is not exhaustive~: we

2531: refer the reader interested in this topic to the review

2532: papers~\cite{KotSmi99,KotSmi03,Tan05} and the PhD thesis~\cite{Win03}.

2533: 

2534: 

2535: \subsection{Weak localization and Brownian motion on graphs\label{sec:wlrg}}

2536: 

2537: Our initial physical motivation for the study of the spectral determinant on

2538: graphs was based on the observation that the weak localization correction to

2539: the conductivity is directly expressed in terms of the spectral determinant.

2540: We see from eq.~(\ref{WL}) that

2541: \be\label{WLsd} 

2542: \smean{\Delta\sigma}

2543: = - \frac{2 e^2}{\pi\,{\rm Volume}}

2544: \drond{}{\gamma} \ln S(\gamma)   

2545: \:,\ee

2546: where the spectral parameter is related to the phase coherence

2547: length $\gamma=1/L_\varphi^2$.

2548: The expressions (\ref{WLsd},\ref{spedet},\ref{defM}), due to Pascaud \&

2549: Montambaux \cite{Pas98,PasMon99,AkkComDesMonTex00}, improve the

2550: approach initiated by Dou{\c c}ot \& Rammal \cite{DouRam85,DouRam86}.

2551: 

2552: 

2553: 

2554: \subsubsection*{Nonlocality of the quantum transport in arbitrary networks}

2555: 

2556: We must stress that the formula (\ref{WLsd}) corresponds to a uniform

2557: integration of the cooperon, defined as 

2558: $P_c(x,x)=\bra{x}\frac1{\gamma-\Delta}\ket{x}$, on

2559: the graph $\smean{\Delta\sigma}\propto-\int\D x\,P_c(x,x)$. This approach is

2560: limited to the case of regular graphs, where all wires play the same role. In

2561: other terms, for a nonregular network, the quantity (\ref{WLsd}) does not

2562: correspond to a quantity measured in a transport experiment. For arbitrary

2563: networks, the cooperon must be integrated over the network with appropriate

2564: nontrivial weights that depend on the topology of the whole network and the

2565: way it is connected to external contacts. Such a generalization was provided in

2566: ref.~\cite{TexMon04}.

2567: 

2568: 

2569: \subsubsection*{Windings in a loop connected to a network} 

2570: 

2571: As we have mentioned in the section~\ref{sec:brownian2},  magnetoconductance 

2572: oscillations due to a magnetic flux and winding properties are

2573: closely related.

2574: For example, if we consider an isolated ring of perimeter $L$ pierced by a 

2575: magnetic flux $\phi$, the well-known behaviour of the harmonics

2576: \begin{equation}

2577:   \label{eq:AAS}

2578:   \smean{\Delta\sigma_n}=\int_0^{2\pi}\frac{\D\theta}{2\pi}

2579:   \smean{\Delta\sigma(\theta)}\,\EXP{-\I n\theta}

2580:   \propto\EXP{-|n|L/L_\varphi}

2581: \end{equation} 

2582: is a direct consequence of the fact that the winding around the ring scales 

2583: with time as $n_t\propto t^{1/2}$ (normal diffusion).

2584: $\theta=4\pi\phi/\phi_0$ is the reduced flux  and $\gamma=1/L_\varphi^2$.

2585: This effect was predicted by Al'tshuler, Aronov \& Spivak (AAS) in 

2586: ref.~\cite{AltAroSpi81} and observed in

2587: experiments on cylinder films~\cite{ShaSha82,AltAroSpiShaSha82}.

2588: 

2589: Recently it has been noticed that the fact that the ring is connected to arms,

2590: which is necessary to perform a transport experiment, can strongly affect the

2591: harmonics. If we consider a ring connected to $N_a$ long arms, in the limit

2592: $L_\varphi\ll L$, the harmonics are still given by the AAS behaviour

2593: (\ref{eq:AAS}), however when the perimeter is smaller than $L_\varphi$, the

2594: harmonics behave as

2595: $\smean{\Delta\sigma_n}\propto\EXP{-|n|\sqrt{N_a L/L_\varphi}}$.

2596: To clarify the origin of this behaviour, the winding of Brownian trajectories

2597: around a ring connected to another network has been recently examined in

2598: ref.~\cite{TexMon05}. This analysis is based on the fact that the winding

2599: number distribution can be expressed in terms of the spectral determinant. Let

2600: us briefly describe this approach.

2601: 

2602: \begin{figure}[!ht]

2603: \begin{center}

2604: \includegraphics[scale=1]{ringwitharm.eps}

2605: \end{center}

2606: \caption{\it A diffusive ring attached to a long arm. The parameter $\lambda_1$

2607:          describes the boundary condition at vertex 1~: $\lambda_1=\infty$

2608:          for the Dirichlet boundary and $\lambda_1=0$ for the Neumann boundary

2609:          condition.

2610:          \label{fig:ringwitharm}}

2611: \end{figure}

2612: 

2613: \begin{figure}[!ht]

2614: \begin{center}

2615: \includegraphics[scale=0.5]{ringmesh2.eps}

2616: \hspace{1cm}

2617: \includegraphics[scale=0.5]{ringbethe.eps}

2618: \end{center}

2619: \caption{\it A diffusive ring attached to {\rm (a)~:} an infinite square 

2620:          lattice,

2621:          {\rm (b)~:} an infinite Bethe lattice of connectivity $z=3$. 

2622:          To ensure that the Bethe lattice is regular, it must be embedded

2623:          in a constant negative curvature surface.

2624:          \label{fig:ringmesh}}

2625: \end{figure}

2626: 

2627: We consider a ring to which is attached an arbitrary network at vertex $0$

2628: (the figures \ref{fig:ringwitharm} and \ref{fig:ringmesh} give examples

2629: of such a situation). Our aim is to understand how the winding around the ring

2630: is affected by the presence of the network. To answer this question we

2631: introduce the probability to start from a point $x$ on the graph and come back

2632: to it after a time $t$, conditioned to wind $n$ times around the loop~:

2633: \be 

2634: {\cal P}_n(x,t|x,0) 

2635: = \int_{x(0)=x}^{x(t)=x}{\cal D}x(\tau)

2636: \:\EXP{-\frac14\int_0^t\D\tau\,\dot x(\tau)^2} 

2637: \delta_{n,{\cal N}[x(\tau)]} 

2638: \ee

2639: where ${\cal N}[x(\tau)]$ is the winding number of the path

2640: $x(\tau)$ around the loop.  For simplicity, we consider the case where the

2641: initial point is the vertex $0$.

2642: The computation of ${\cal P}_n(0,t|0,0)$ requires some local information

2643: (the eigenfunctions of the Laplace operator at point $x=0$). 

2644: On the other hand the spectral determinant encodes a global information

2645: since it results from a spatial integration of the Green's function

2646: of the Laplace operator over the network. 

2647: However we have shown in ref.~\cite{TexMon05} that if we consider mixed 

2648: boundary conditions at vertex $0$ described by a parameter $\lambda_0$

2649: (see the definition of the boundary conditions given above), 

2650: we can extract the probability of interest\footnote{

2651:   The fact that some ``local'' information, such as ${\cal P}_n(0,t|0,0)$,

2652:   can be extracted from a more ``global'' object, like the spectral 

2653:   determinant, by introducing the mixed boundary conditions with parameter

2654:   $\lambda_0$ has been also used in the context of scattering theory in

2655:   graphs in refs.~\cite{TexBut03,TexDeg03}. 

2656:   The derivative in eq.~(\ref{P0S}) can be understood as a functional

2657:   derivative since $\lambda_0$ plays a role similar to the weight of 

2658:   a $\delta$-potential at $x_0$. 

2659:   The use of functional derivatives in scattering theory for 

2660:   mesoscopic systems has been fruitfully used by

2661:   B\"uttiker and co-workers (see ref.~\cite{But02} and references therein).

2662: } as follows~:

2663: \be\label{P0S}

2664: \int_0^\infty\D t\,{\cal P}_n(0,t|0,0)\,\EXP{-\gamma t}

2665: =\int_{0}^{2\pi}\frac{\D\theta}{2\pi}\:\EXP{-\I n\theta}\:

2666: \frac{\D}{\D\lambda_0}\ln S^{(\lambda_0)}(\gamma)\Big|_{\lambda_0=0}

2667: \:,\ee

2668: where $S^{(\lambda_0)}(\gamma)$ is the spectral determinant for 

2669: mixed boundary condition at vertex $0$ and with a magnetic flux 

2670: $\theta$ piercing the ring.

2671: This formula allows us to express the probability as

2672: \be\label{Wn00}

2673:   \int_0^\infty\D t\:{\cal P}_n(0,t|0,0)\:\EXP{-\gamma t} = 

2674:   \frac{1}{2\sqrt\gamma}

2675:   \frac{\sinh\sqrt\gamma L}{\sinh\sqrt{\gamma} L_{\rm eff}(\gamma)}

2676:   \:\EXP{-|n|\sqrt{\gamma} L_{\rm eff}(\gamma)} 

2677: \:,\ee

2678: where all the information about the nature of the network attached to the 

2679: ring is contained in the effective perimeter $L_{\rm eff}(\gamma)$, defined 

2680: as~:

2681: \be\label{leffnet}

2682: \cosh\sqrt{\gamma} L_{\rm eff}(\gamma)

2683: =\cosh\sqrt{\gamma}L

2684: +\frac{\sinh\sqrt{\gamma}L}{2\,({M}_{\rm net}^{-1})_{00}}

2685: \:.\ee

2686: The matrix $M_{\rm net}$ describes the network in the absence of the loop

2687: (it is given by eq.~(\ref{defM})~; for the network of 

2688: figure~\ref{fig:ringmesh}a, ${M}_{\rm net}$ describes the infinite 

2689: square network without the ring).

2690: The matrix element $({M}_{\rm net}^{-1})_{00}$ has a clear

2691: meaning~: it is the Green function of the Laplace operator in the 

2692: network (without the loop) computed  at the position where 

2693: the ring is attached~:

2694: $\sqrt{\gamma}(M_{\rm net}^{-1})_{00}=\bra{0}\frac1{\gamma-\Delta}\ket{0}$. 

2695: 

2696: The effective perimeter $L_{\rm eff}(\gamma)$ probes the winding at time

2697: scale $t\sim1/\gamma$~: precisely, the winding number scales with time as

2698: $n_t\sim\sqrt{t}/L_{\rm eff}(1/t)$.

2699: 

2700: \vspace{0.25cm}

2701: 

2702: Two interesting examples are

2703: 

2704: \vspace{0.25cm}

2705: 

2706: \noindent$\bullet$ {\it Ring attached to an infinite wire 

2707: (figure~\ref{fig:ringwitharm}).--}

2708: When a wire of length $b$ with the Dirichlet boundary at one end is 

2709: attached to the ring, it is easy to see that eq.~(\ref{defM}) gives

2710: $({M}_{\rm net}^{-1})_{00}=\tanh\sqrt\gamma b\leadto{b\to\infty}1$.

2711: At large time $t\gg L^2$, the effective perimeter behaves 

2712: as $L_{\rm eff}\simeq\sqrt{L}\gamma^{-1/4}$. This behaviour is related 

2713: to a scaling of the winding number with  time

2714: \be\label{scalingwire}

2715: n_t \propto t^{1/4}

2716: \:.\ee

2717: The full distribution for the winding number $n$ is given in

2718: ref.~\cite{TexMon05}. The exponent $1/4$  that characterizes anomalously slow

2719: winding around the loop originates from the fact that the diffusive trajectory

2720: spends a long time in the infinite wire, which increases the effective

2721: perimeter at such time scales. This problem is also related to the anomalous

2722: diffusion along the skeleton of a comb, studied in

2723: refs.~\cite{WeiHav86,BalHavWei87} by different methods. It is interesting to

2724: use this picture. Let us consider a random walk along the sites of a line where

2725: the diffusive particle is trapped during a time $\tau$ on each site. The

2726: trapping time is distributed according to a broad distribution

2727: $P_1(\tau)\propto\tau^{-1-\mu}$ with $0<\mu<1$. It follows that the distance

2728: scales with time as $n_t\sim t^{\mu/2}$ \cite{BouGeo90}. If we go back to the

2729: problem of diffusion along the skeleton of a comb (or the winding in the ring

2730: connected to the long arm), the arm plays the role of the trap. The

2731: distribution of the trapping time is given by the first return probability of

2732: the one-dimensional diffusion~:  $P_1(\tau)\propto\tau^{-3/2}$ and we recover

2733: eq.~(\ref{scalingwire}).

2734: 

2735: \vspace{0.25cm}

2736: 

2737: \noindent$\bullet$ {\it Ring attached to a square network

2738: (figure~\ref{fig:ringmesh}a).--}

2739: When studying the winding around the loop, it is important to know

2740: whether the Brownian motion inside the network attached to the ring is

2741: recurrent or not. Let us consider the case where the network attached is a

2742: $d$-dimensional hypercubic network. For $d>2$ the Brownian motion is

2743: known to be transient where as, for $d=2$ it is neighbourhood recurrent

2744: (in $d=1$ the Brownian motion is pointwise recurrent). Therefore we

2745: expect the dimension 2 to play a special role. In the large time limit, when 

2746: $t\gg{L^2}$ and $t\gg{a^2}$ ($a$ is the lattice spacing), we find an 

2747: effective length

2748: $\sqrt\gamma\, L_{\rm eff}\simeq\sqrt{\frac{2\pi L}{a\ln(4/\sqrt{\gamma}a)}}$,

2749: that corresponds to a scaling of the winding around the loop

2750: \be\label{scalingsn}

2751: n_t \propto (\ln t)^{1/2}

2752: \:.\ee

2753: This result can be obtained in the same way as for the ring connected to 

2754: the arm. This time the plane acts as a trap. 

2755: The distribution of the trapping time is given by the first return

2756: probability on a square lattice, which is known to behave at large times 

2757: like $P_1(\tau)\propto1/(\tau\ln^2\tau)$ \cite{BarTac93}, from which

2758: we can recover eq.~(\ref{scalingsn}).

2759: 

2760: 

2761: \subsubsection*{Occupation time and local time distribution on a graph} 

2762: 

2763: Another set of problems concern occupation times~: {\it i.e.} time spent by a

2764: Brownian particle in a given region. An example of such a problem is provided

2765: by the famous arc-sine law for the 1d Brownian motion that gives the

2766: distribution of the time spent by a Brownian motion

2767: $(x(\tau),0\leq\tau\leq{t}\,|\,x(0)=0)$ on the half line $\RR^+$. This result

2768: was derived long ago by P.~L\'evy \cite{Lev48}. It has been extended by Barlow,

2769: Pitman \& Yor \cite{BarPitYor89} for a particular graph (star graph with arms

2770: of infinite lengths)~: instead of an infinite line, these authors consider $n$

2771: semi-infinite lines originating from the same point and study the joint

2772: distribution of the times spent on each branch. More recently, this problem

2773: has been reconsidered in the case of arbitrary graphs \cite{Des02}. It may be

2774: stated as follows~: consider a Brownian motion $x(\tau)$ on a graph, starting

2775: from a point $x_0$ at time $0$ and arriving at a point $x_1$ at time $t$. Let

2776: $T_{\ab}$ denotes the time spent on the wire $(\ab)$. This functional is

2777: defined as $T_{\ab}[x(\tau)]=\int_0^t\D\tau\,{\rm Y}_{\ab}(x(\tau))$ where the

2778: function ${\rm Y}_{\ab}(x)$ is $1$ for $x\in(\ab)$ and $0$ otherwise. Our aim

2779: is to compute the Laplace transforms of the joint distribution

2780: \begin{equation}

2781:   \label{eq:octimedef}

2782:   \mean{ \EXP{-\sum_{(\ab)} \xi_\ab\,T_{\ab}} }

2783:   = \int\D x_1\, {\cal F}(x_1,x_0;t;\{\xi_\ab\}) 

2784: \end{equation}

2785: with

2786: \begin{equation}

2787:   {\cal F}(x_1,x_0;t;\{\xi_\ab\})

2788:   =\int_{x(0)=x_0}^{x(t)=x_1} {\cal D}x(\tau)\,

2789:   \EXP{  -\frac14\int_0^t\D\tau\,\dot x^2

2790:          -\sum_{(\ab)} \xi_\ab\,T_{\ab}[x(\tau)]

2791:       }

2792: \:.\end{equation}

2793: The definition (\ref{eq:octimedef}) takes into account averaging over the

2794: final point $x_1$. A conjugate parameter $\xi_\ab$ is introduced for each

2795: variable $T_{\ab}$, {\it i.e.} each bond. A closed expression of the Laplace

2796: transform of the joint distribution (\ref{eq:octimedef}) has been derived in

2797: ref.~\cite{Des02}. The result is given as a ratio of two determinants,

2798: an expression reminiscent of the one of Leuridan \cite{Leu00}, although the 

2799: connection is not completely clear.

2800: 

2801: Similar methods have been also applied in ref.~\cite{ComDesMaj02} to study the

2802: distribution of the local time

2803: $T_{x_0}[x(\tau)]=\int_0^t\D\tau\,\delta(x(\tau)-x_0)$.

2804: 

2805: \vspace{0.25cm}

2806: 

2807: A simplification occurs when the final point coincides with the initial point.

2808: Then it is possible to relate the characteristic function to a single spectral

2809: determinant. We do not develop the general theory here but instead consider an

2810: example close to the one studied in the previous subsection.  Let us consider

2811: the graph in figure~\ref{fig:ringwitharm} and ask the following question~: for

2812: a Brownian motion starting from $x_0$ at time $0$ and coming back to it at 

2813: time $t$,

2814: what is the distribution of the time $T_{\rm arm}[x(\tau)]$ spent in the arm

2815: if in addition the Brownian motion is constrained to turn $n$ times around the

2816: ring~? It is natural to introduce the following function~:

2817: \begin{equation}

2818:   \label{eq:winoc}

2819:   {\cal F}_n(x_0,x_0;t,\xi) =

2820:   \int_{x(0)=x_0}^{x(t)=x_0} {\cal D}x(\tau)\,

2821:   \EXP{  -\int_0^t\D\tau\,

2822:         \left( 

2823:           \frac14\dot x^2 + \xi\,{\rm Y}_{\rm arm}(x)

2824:         \right)

2825:       }\:

2826:   \delta_{n,{\cal N}[x(\tau)]} 

2827: \end{equation}

2828: where ${\cal N}[x(\tau)]$ is the winding number around the ring.

2829: This function is related to the Laplace transform of the distribution

2830: of the functional $T_{\rm arm}[x(\tau)]$ 

2831: \begin{equation}

2832:   \label{disTbras}

2833:   \mean{\EXP{-\xi\, T_{\rm arm}[x]}}_{{\cal C}_n} 

2834:   = \frac{{\cal F}_n(x_0,x_0;t,\xi)}{{\cal F}_n(x_0,x_0;t,0)}

2835: \:,\end{equation}

2836: where $\mean{\cdots}_{{\cal C}_n}$ denotes averaging over curves of 

2837: winding~$n$. The denominator ensures normalization.

2838: The Laplace transform of (\ref{eq:winoc}) is given by a relation similar

2839: to eq.~(\ref{P0S})

2840: \begin{equation}

2841:   \int_0^\infty\D t\,{\cal F}_n(x_0,x_0;t,\xi)\,\EXP{-\gamma t}

2842:   =\int_{0}^{2\pi}\frac{\D\theta}{2\pi}\:\EXP{-\I n\theta}\:

2843:   \frac{\D}{\D\lambda_0}\ln S^{(\lambda_0)}(\gamma)\Big|_{\lambda_0=0}  

2844: \end{equation}

2845: where the appropriate spectral determinant is built as follows~:

2846: ({\it i}) since the starting point is fixed at $0$, we introduce mixed 

2847: boundary conditions at this point, with a parameter $\lambda_0$ that will be

2848: used to extract the ``local information''.

2849: ({\it ii}) A magnetic flux $\theta$ is introduced (conjugate to the winding

2850: number).

2851: ({\it iii}) The spectral parameter is shifted in the arm as 

2852: $\gamma\to\gamma+\xi$ to introduce the variable $\xi$ conjugate to the time 

2853: $T_{\rm arm}$.

2854: 

2855: We choose the vertex $0$ as initial condition and impose the Dirichlet

2856: boundary condition (which is achieved by setting $\lambda_1=\infty$) at the

2857: end of the arm of length $b$. The spectral determinant is found

2858: straightforwardly (for an efficient calculation of $S(\gamma)$ for a graph

2859: with loops, see ref.~\cite{Pas98} or appendix~C of

2860: ref.~\cite{AkkComDesMonTex00})~:

2861: \begin{equation}

2862:   S^{(\lambda_0)}(\gamma) =

2863:   \frac{\sinh\sqrt{\gamma} L\,\sinh\sqrt{\gamma+\xi}\,b}

2864:        {\sqrt{\gamma}\sqrt{\gamma+\xi}}

2865:   \left[

2866:     2\sqrt{\gamma}\frac{\cosh\sqrt{\gamma} L-\cos\theta}{\sinh\sqrt{\gamma} L}

2867:     +\lambda_0

2868:     +\sqrt{\gamma+\xi}\coth\sqrt{\gamma+\xi}\,b

2869:   \right]

2870: \end{equation}

2871: The terms in the brackets correspond to the matrix\footnote{

2872:   The introduction of the conjugate parameters $\{\xi_\ab\}$ corresponds to

2873:   shift the spectral parameter $\gamma$ on each wire as

2874:   $\gamma\to\gamma+\xi_\ab$.  The matrix to be generalized is not $M$, given

2875:   by eq.~(\ref{defM}), but ${\cal M}=\sqrt{\gamma}M$.

2876: } ${\cal M}$ (since $\lambda_1=\infty$ we 

2877: can consider only the element ${\cal M}_{00}$). The first term is the 

2878: contribution of the loop, the second comes from the boundary condition and 

2879: the last one comes from the arm. It immediately follows that 

2880: \begin{equation}

2881:   \label{relation1}

2882:   \int_0^\infty\D t\,{\cal F}_n(0,0;t,\xi)\,\EXP{-\gamma t}

2883:   =  

2884:   \frac{1}{2\sqrt\gamma}

2885:   \frac{\sinh\sqrt\gamma L}{\sinh\sqrt{\gamma} L_{\rm eff}}

2886:   \:\EXP{-|n|\sqrt{\gamma} L_{\rm eff}} 

2887: \end{equation}

2888: with

2889: \begin{equation}

2890:   \label{Leffwinoc}

2891:   \cosh\sqrt\gamma L_{\rm eff} = \cosh\sqrt\gamma L

2892:   +\frac12\sqrt{1+{\xi}/{\gamma}}\,\sinh\sqrt{\gamma} L\,

2893:    \coth\sqrt{\gamma+\xi}\,b

2894:   \:.

2895: \end{equation}

2896: Let us consider an infinitely long arm $b\to\infty$. If we are interested on

2897: time scales $t\gg\tau_L$, where $\tau_L=L^2$ is the Thouless time over which

2898: the ring is explored, eq.~(\ref{Leffwinoc}) gives 

2899: $\sqrt{\gamma}L_{\rm eff}\simeq(\gamma+\xi)^{1/4}\sqrt{L}$, therefore from

2900: eq.~(\ref{relation1}) we see that 

2901: ${\cal F}_n(0,0;t,\xi)\simeq\EXP{-\xi t}\,{\cal F}_n(0,0;t,0)$.

2902: The inverse Laplace transform of eq.~(\ref{disTbras}) leads to

2903: $\smean{\delta(T-T_{\rm arm}[x])}_{{\cal C}_n}\simeq\delta(T-t)$, which means

2904: that the Brownian motion spends almost all the time in the arm (this simple

2905: result confirms the picture presented to explain the scaling of the winding

2906: around the ring of the form $n_t\propto t^{1/4}$).

2907: 

2908: 

2909: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2910: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2911: 

2912: \section{Planar Brownian motion and charged particle in random magnetic field\label{sec:magimp}}

2913: 

2914: A model of random magnetic field describing a charged particle moving in a

2915: plane and subjected to the random magnetic field 

2916: ${\cal B}(\vec r)=\vec\nabla\times \vec A(\vec r)$ due to an ensemble of

2917: magnetic Aharonov-Bohm vortices is described by the following

2918: Hamiltonian~\cite{DesFurOuv95,Fur97}~:

2919: \bea\label{Him}

2920: H   = \frac12\left(\vec p - \vec A(\vec r) \right)^2 

2921:     + \frac12 \, {\cal B}(\vec r) 

2922:     = \frac12

2923: \left(

2924:   \vec p - \alpha\sum_i 

2925:   \frac{\vec u_z\times(\vec r-\vec r_i)}{(\vec r-\vec r_i)^2}

2926: \right)^2 

2927: +\pi\alpha\sum_i \delta(\vec r-\vec r_i)

2928: \eea

2929: where $0\leq\alpha<1$. We have set $\hbar=e=m=1$.

2930: $\alpha$ is the magnetic flux per vortex, in unit of the quantum flux 

2931: $\phi_0=h/e$~:  $\alpha=\phi/\phi_0=\phi/(2\pi)$.

2932: The model is periodic in $\alpha$ with a period $1$. 

2933: $\vec u_z$ is the unit vector perpendicular to the plane. $\vec r_i$ are the 

2934: positions of the vortices in the plane~: they are uncorrelated random  

2935: variables (Poisson distribution).

2936: The $\delta$ interactions (coupling to the magnetic field) are necessary 

2937: in order to define properly the model 

2938: \cite{JacPi90,McCOuv91,Ouv94,BerLoz94,ComMasOuv95}.

2939: 

2940: By comparison with scalar impurities discussed in the introduction, the

2941: magnetic nature of the scatterers gives rise to rather different

2942: properties\footnote{

2943:   For scalar impurities, the presence of a magnetic field also strongly

2944:   affects the spectral properties. The case of a Gaussian disorder projected

2945:   in the lowest Landau level (LLL) of a strong magnetic field has been studied

2946:   in ref.~\cite{Weg83}. Other disordered potentials have been considered

2947:   later.  In particular it was shown in ref.~\cite{BreGroItz84} 

2948:   that for a weak density of impurities, the spectrum can display power law

2949:   singularities. A physical interpretation of such power law singularities has

2950:   been discussed by Furtlehner in ref.~\cite{Fur97,Fur00}.  The relation

2951:   between the model of scalar impurities projected in the LLL of a strong

2952:   magnetic field and the model of magnetic vortices was discussed in

2953:   refs.~\cite{DesFurOuv96,Fur97}.  Finally we mention a recent work on

2954:   Lifshitz tails in the presence of magnetic field \cite{LesWar04}.

2955: }~: it was shown in ref.~\cite{DesFurOuv95} that the spectrum is

2956: reminiscent of a Landau spectrum with Landau levels broadened by disorder, in

2957: the limit of vanishing magnetic flux $\alpha\to0$ (see below). This analysis

2958: was performed using perturbative arguments supported by numerical

2959: simulations.

2960: The latter use a relation between the average DoS of

2961: this particular model and some winding properties of the Brownian motion.

2962: 

2963: 

2964: Let $Z(t)$ denotes the partition function for a given distribution of vortices

2965: and $Z_0(t)$ the partition function without fluxes. The ratio of partition

2966: functions can be written as a ratio of two path integrals

2967: \be\label{pi1}

2968: \frac{Z(t)}{Z_0(t)}=

2969: \frac{

2970:   \int \D \vec a 

2971:   \int_{\vec r(0)=\vec a}^{\vec r(t)=\vec a}{\cal D}\vec r(\tau) 

2972:   \:\EXP{\int_0^t (-\frac{1}{2} \dot{\vec r}^2 + \I\vec A\cdot\dot{\vec r}) 

2973:     \D\tau }

2974:      }

2975:      {

2976:   \int \D\vec a

2977:   \int_{\vec r(0)=\vec a}^{\vec r(t)=\vec a}{\cal D}\vec r(\tau) 

2978:   \:\EXP{-\int_0^t \frac{1}{2}\dot{\vec r}^2  \D\tau } 

2979:      }

2980: =

2981: \left\langle 

2982:   \EXP{ \I\oint_{\cal C}  \vec A \cdot \D\vec r}

2983: \right\rangle_{\cal C}

2984: \:,\ee

2985: where $Z_0(t)={V}/({2\pi t})$ is the free partition function ($V$ is the

2986: (infinite) area of the plane) and $\smean{\cdots}_{\cal C}$ stands for 

2987: averaging over all closed Brownian curves of the plane.

2988: 

2989: In order to average $Z(t)$ over the Poissonian distribution of vortices, let

2990: us consider, for a while, our problem on a square lattice with lattice spacing

2991: $a$. Let $N_i$ be the number of vortices in square $i$. 

2992: the magnetic flux through any closed random walk ${\cal C}$ on this lattice 

2993: can be written as~:

2994: \be\label{cl}

2995: \oint_{\cal C}  \vec A \cdot \D \vec r = \sum_i 2 \pi \alpha N_i n_i 

2996: \ee

2997: where $n_i$ is the number of times the square $i$ has been wound around by 

2998: ${\cal C}$.

2999: 

3000: Averaging $Z(t)$ with the Poisson distribution ($\mu$ being the mean density of

3001: vortices)

3002: \be\label{pois}

3003:  P(N_i) \; = \; \frac{ ( \mu a^2 )^{N_i} }{N_i!}\,\EXP{- \mu a^2 }  

3004: \ee

3005: gives

3006: \be\label{zbar}

3007: \mean{ Z(t) }_{ \{\vec r_i\} } 

3008: =Z_0(t) 

3009: \left\langle 

3010:   \EXP{ \I\oint_{\cal C}  \vec A \cdot \D\vec r}

3011: \right\rangle_{ {\cal C},\{\vec r_i\} }

3012: =

3013: Z_0(t) 

3014: \left\langle   

3015: \exp\left(     

3016:  \mu \sum_n S_n(\EXP{2\I\pi\alpha n} - 1)

3017:  \right) 

3018: \right\rangle_{\cal C}

3019: \ee

3020: where $\smean{\cdots}_{\{\vec r_i\}}$ denotes averaging over the positions

3021: $\{\vec r_i\}$ of the vortices.

3022: 

3023: The quantity $S_n$ denotes the area of the locus of points around which the

3024: curve ${\cal C}$ has wound $n$ times. This result was derived for a random

3025: walk on a lattice but is also obviously valid off lattice as well, for 

3026: Brownian curves. The winding sectors of a closed

3027: curve are displayed in figure~\ref{fig:jean1}.

3028: 

3029: \begin{figure}[!ht]

3030: \begin{center}

3031: \includegraphics[scale=.4,angle=0]{wind_sect.eps}

3032: %\hspace{1cm}

3033: %\includegraphics{winding_sn.eps}

3034: \caption{

3035:   \it A closed curve with its $n$-winding sectors. The label $n$

3036:   of a sector is  the winding number of any point inside 

3037:   this sector. \label{fig:jean1}}

3038: \end{center}

3039: \end{figure}

3040: 

3041: \noindent

3042: The random variable $S_n$ scales like $t$.  In particular, we know

3043: \cite{ComDesOuv90,GroFri03} its expectation $\smean{S_n}=\frac{t}{2\pi n^2}$.

3044: Moreover, it has been shown in  \cite{Wer94} that the variable

3045: $n^2S_n$ becomes more and more peaked when $n$ grows:

3046: $

3047: P(n^2 S_n = X) \leadto{n\to\infty} \delta(X-\frac{t}{2\pi })

3048: $ (the average area of the $n=0$-sector has been recently studied in 

3049: ref.~\cite{GarTru05}).

3050: 

3051: Thus, extracting   $t$, eq.~(\ref{zbar}) rewrites as

3052: \be\label{zbar1}

3053: \mean{ Z(t) }_{ \{\vec r_i\} }  = Z_0(t) 

3054: \int\D S\D A\:P(S,A) \:\EXP{- \mu t ( S + \I A )}

3055: \equiv 

3056: Z_0(t) 

3057: \left\langle \EXP{- \mu t ( S + \I A )}\right\rangle_{\cal C}

3058: \:,\ee

3059: where  $P(S,A)$ is the joint distribution of the 

3060: rescaled ($t$ independent) variables $S$ and $A$ defined as

3061: \bea

3062: \label{s} 

3063: S&=&\frac{2}{t} \sum_n S_n \sin^2 (\pi \alpha n)\\

3064: \label{a}   

3065: A&=&\frac{1}{t} \sum_n S_n \sin (2\pi \alpha n) 

3066: \:.\eea

3067: The averages of these two variables are given by~:

3068: \bea

3069:    \mean{S} &=& \pi\alpha (1 - \alpha) 

3070: \\

3071:    \mean{A} &=& 0

3072: \:.\eea

3073: With eq.~(\ref{zbar1}), we observe that $\frac1V\mean{Z(t)}_{\{\vec r_i\}}$

3074: has the scaling form ${F(\mu t)}/{t}$. Thus, its inverse Laplace transform,

3075: the average density of states per unit area, is a function of only $E/\mu$ and

3076: $\alpha$. Moreover, it is easy to realize that $\mean{Z(t)}_{\{\vec r_i\}}$ is

3077: even in $\alpha$ (each Brownian curve in $\{{\cal C}\}$ comes with its time

3078: reversed) and periodic in $\alpha$ with period 1. Thus, one can restrict to

3079: $0\leq\alpha\leq\frac{1}{2}$. Let us focus on the two limiting cases 

3080: $\alpha\to0$ and $\alpha={1}/{2}$.

3081: 

3082: \vspace{0.25cm}

3083: 

3084: \mathversion{bold}

3085: \noindent$\bullet$ {\bf Limit $\alpha \to 0$~: Landau spectrum.}\\

3086: \mathversion{normal}

3087: When  $\alpha \to 0$, a careful analysis shows that

3088: \be\label{atou0}

3089: \mean{Z(t)}_{ \{\vec r_i\} } 

3090: \simeq 

3091: Z_{\smean{{\cal B}}} \: \EXP{-\frac12\smean{{\cal B}}t }

3092: \ee

3093: where 

3094: $

3095: Z_{\smean{{\cal B}}}=Z_0(t)

3096: \frac{\smean{{\cal B}}t/2}{\sinh(\smean{{\cal B}}t/2)}

3097: $

3098: is the partition function of a charged particle in a uniform magnetic field

3099: $\mean{{\cal B}}=2\pi\mu\alpha$ (we recall that the impurity $i$ carries a

3100: magnetic field $2\pi\alpha\delta(\vec r-\vec r_i)$). The system of random

3101: vortices is, thus, equivalent to the uniform average magnetic field, albeit 

3102: with  an additional positive shift in the Landau spectrum~: the 

3103: inverse Laplace transform of the partition function (\ref{atou0})

3104: gives the Landau spectrum made of equally spaced infinitely degenerated

3105: levels. This corresponds to the oscillating behaviour shown on

3106: figure~\ref{fig:dosimpmag}.  The origin of the shift can be traced back to

3107: the presence of the repulsive $\delta$ interactions that have been added to

3108: the Hamiltonian to define properly the model.

3109:    

3110: \vspace{0.25cm}

3111: 

3112: \mathversion{bold}

3113: \noindent$\bullet$ 

3114: {\bf Half quantum flux vortices ($\alpha=1/2$) -- The spectral singularity at $E=0$.}\\

3115: \mathversion{normal}

3116: In this case the variable (\ref{a}) vanishes, $A\equiv0$, implying that

3117: $

3118: \mean{Z(t)}_{\{\vec r_i\}} = Z_0(t) 

3119: \smean{ \exp \left( - \mu t  S  \right) }_{\cal C}

3120: $ 

3121: where now

3122: \be\label{snodd}

3123:  S= \frac{2}{t} \sum_{n\ {\rm odd}} S_n  

3124: \:.\ee

3125: Performing  

3126: the inverse Laplace transform, we get the average density of states

3127:   per unit area:

3128: \be\label{dos1/2} 

3129: \left\langle \rho (E)

3130:  \right\rangle = \rho_0 (E)\ \int_0^{E/\mu} \D S\, P(S)

3131: \ee

3132: where $\rho_0(E)=\frac{1}{2\pi}$  is the free density of states per unit area

3133: and  $P(S)$ is the probability distribution of $S$.

3134: 

3135: \begin{figure}[!ht]

3136: \begin{center}

3137: \includegraphics[scale=.5,angle=0]{jean_im05fit.eps}

3138: \caption{\it 

3139:  The  average density of states at $\alpha =0.5$ (the free density of states

3140:  is constant $\rho_0=1/(2\pi)$). Circles are  simulation

3141:  results. The full line is the fit discussed in the  text.

3142:  \label{fig:jean2}}

3143: \end{center}

3144: \end{figure}

3145: 

3146: We may use this result to determine the nature of the singularity in the

3147: average DoS. As displayed in figure~\ref{fig:jean2} (where we have taken

3148: $\mu=1$), $\smean{\rho(E)}$ increases monotonically from $0$ to $\rho_0(E)$

3149: with a depletion of states at the bottom of the spectrum. Circles are the 

3150: result

3151: of numerical simulations (on a 2D square lattice, we have generated 10000

3152: closed random walks of 100000 steps each). The full line is a low-energy

3153: fit of the quantity $\smean{\rho(E)}/\rho_0$ by the function

3154: \begin{equation}

3155:   \label{specsing}

3156:   \frac{\rho_{\rm fit}(E)}{\rho_0} = a\, \EXP{-b(\mu/E)^2}

3157: \end{equation}

3158: with $a=2.8$ and $b=1.4$ (see figure~\ref{fig:jean2}).

3159: It is worth stressing that this behaviour is quite different from the one 

3160: expected for a disorder due to a scalar potential. The famous Lifshitz 

3161: argument, applicable to the low energy DoS for a low concentration of 

3162: {\it scalar} impurity, leads instead to\footnote{

3163:   The Lifshitz argument applies to the DoS for the random Hamiltonian

3164:   $H=-\Delta+\sum_i u(\vec r-\vec r_i)$ in dimension~$d$, 

3165:   where $u(\vec r)$ is a sharply peaked scalar potential.

3166:   For a low density of impurities $\mu$, the DoS behaves as 

3167:   $\rho_{\rm Lif.}(E) \sim \exp({-c_d\,\mu/E^{d/2}})$ at low energy.

3168:   \label{footnoteLif}

3169: }

3170: a behaviour 

3171: \begin{equation}

3172: \rho_{\rm Lif.}(E) \sim \EXP{-c_2\,\mu/E}\:,

3173: \end{equation}

3174: where $c_2$ is a constant. Our choice for this fit is motivated by a recent

3175: numerical work \cite{Ric03} concerning the area $\cal{A}$ of the outer

3176: boundary of planar random loops. In this work, the author suggests that the

3177: limit distribution of $\cal{A}$ is the Airy distribution implying, for small

3178: $\cal{A}$ values, a behaviour of the type 

3179: $\exp(-\mbox{const.}/{\cal A}^2)/{\cal A}^2$. Remarking that 

3180: ${\cal A}=\sum_{n\ {\rm even}}S_n+\sum_{n\ {\rm odd}}S_n$, it is natural to

3181: expect for the distribution of the random variable $S$, eq.~(\ref{snodd}), a

3182: behaviour at small $S$ that is roughly given by $\exp(-\mbox{const.}/S^2)$.

3183: Thus, we deduce the form $\rho_{\rm fit}(E)$ for the low-energy fit of

3184: $\smean{\rho(E)}$.

3185: 

3186: \vspace{0.25cm}

3187: 

3188: \mathversion{bold}

3189: \noindent$\bullet$ {\bf Transition between $\alpha=0.5$ and $\alpha\to0$.}\\

3190: \mathversion{normal}

3191: Finally, when $\alpha$ grows from $0$ to $0.5$, the oscillations in the

3192: spectrum must disappear at some critical value $\alpha_c$ (see

3193: figure~\ref{fig:dosimpmag}). Numerical simulations, specific heat

3194: considerations and, also, diagrammatic expansions \cite{DesFurOuv95} give

3195: $\alpha_c\sim0.3$.

3196: 

3197: 

3198: \begin{figure}[!h]

3199: \begin{center}

3200: 

3201: \includegraphics[scale=0.4]{dosimpmag1.eps}

3202: \hspace{0.5cm}

3203: \includegraphics[scale=0.4]{dosimpmag2.eps}

3204: 

3205: \vspace{0.25cm}

3206: 

3207: \includegraphics[scale=0.4]{dosimpmag3.eps}

3208: \hspace{0.5cm}

3209: \includegraphics[scale=0.4]{dosimpmag4.eps}

3210: 

3211: \caption{\it Average density of states of the Hamiltonian (\ref{Him}) 

3212:          for different values

3213:          of the flux per tube $\phi=\alpha\phi_0$. 

3214:          The average magnetic field reads $\mean{{\cal B}}=2\pi\mu\alpha$.

3215:          We recall that the corresponding Landau spectrum is given by 

3216:          $E_n=\mean{{\cal B}}(n+1)$, for $n\geq0$.

3217:          From ref.~\protect\cite{DesFurOuv95}.\label{fig:dosimpmag}}

3218: \end{center}

3219: \end{figure}

3220: 

3221: 

3222: 

3223: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3224: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3225: 

3226: %\newpage%\ \newpage

3227: 

3228: \section{Conclusion}

3229: 

3230: A non experienced reader may have the feeling that the two topics which have

3231: been covered in this review, namely one-dimensional disordered systems and

3232: quantum graphs, are essentially disjoint. In fact there are many interesting

3233: links between these two topics, both at a methodological and a conceptual

3234: level. The use of metric graphs for modelling quantum phenomena observed in

3235: disordered metals goes back to the pioneering work of Shapiro \cite{Sha82}

3236: and Chalker \& Coddington \cite{ChaCod88}. These systems have been used to

3237: study quantum localization and more recently as a model system for spectral

3238: statistics (for a recent review, see ref.~\cite{Tan05}).  Another set of

3239: similar questions is provided by the study of scattering properties of chaotic

3240: graphs \cite{KotSmi03} and disordered systems \cite{FyoSom97} (see also

3241: section~\ref{sec:wtd} and ref.~\cite{OssFyo05} for recent developments). We

3242: have also seen that apart from its interest to study the spectral properties

3243: of metric graphs, the spectral determinant also allows studying several

3244: properties of networks of quasi-one-dimensional weakly disordered 

3245: wires~\cite{Pas98,PasMon99,AkkComDesMonTex00}.

3246: 

3247: As a conclusion we would like to mention several open problems~:

3248: \\

3249: $\bullet$ A study of spectral statistics in the case of graphs with a random

3250: Schr\"odinger operator [there is still a factorised structure but the matrix

3251: $R$ is now given by eq.~(\ref{n41})].

3252: \\

3253: $\bullet$ A probabilistic understanding of the star graphs using the tools of

3254: excursion theory developed by Barlow {\it et al}. \cite{BarPitYor89} in the

3255: context of the Brownian spider. A first step would be to recover those

3256: probabilistic results (the joint law of the occupation time inside the

3257: branches) by a spectral approach. It is however not excluded that the

3258: probabilistic approach could provide a key to a deeper understanding of those

3259: quantum systems.  In the context of classical systems such probabilistic

3260: approaches have been very useful, {\it e.g.} recent studies of the Stochastic

3261: Loewner Equation have made enormous progress in understanding the statistical

3262: physics of a class of two-dimensional systems. This calls for new

3263: probabilistic techniques for quantum systems as well.

3264: \\

3265: $\bullet$ Several functionals of the Brownian motion (\ref{A},\ref{functring})

3266: appear when studying the important question of dephasing due to

3267: electron-electron interaction in networks of quasi-one-dimensional weakly

3268: disordered wires. The fact that such simple functionals appear relies on the

3269: translation invariance of the two particular problems studied (an infinite

3270: wire \cite{AltAroKhm82} and an isolated ring \cite{TexMon05}). For a network

3271: with arbitrary topology, the relevant functional of the Brownian bridge

3272: $x(\tau)$ is given by $\int_0^t\D\tau\,W(x(\tau),x(t-\tau))$ where

3273: $W(x,x')=\frac12[P_d(x,x)+P_d(x',x')]-P_d(x,x')$ with

3274: $-\Delta{}P_d(x,x')=\delta(x-x')$.  It now involves a nonlocal functional in

3275: time which is difficult to handle. Progress in this direction would allow 

3276: clarifying the interplay between the electron-electron interaction and the

3277: geometrical effect and help in analyzing recent experimental results

3278: \cite{FerAngRowGueBouTexMonMai04,BauMalMonSamSchTex05}.

3279: \\

3280: $\bullet$ The question of extreme value spectral statistics was addressed in

3281: the framework of random matrix theory \cite{TraWid93} and these studies have

3282: found several applications in the context of out-of-equilibrium statistical

3283: physics (see the review \cite{TraWid02}). However this question was first

3284: addressed in the context of one-dimensional disordered systems

3285: \cite{GreMolSud83}. The study of supersymmetric random Hamiltonian, for which

3286: the bottom of the spectrum plays a special role, has emphasized the interest

3287: in extreme value spectral statistics \cite{Tex00} (in particular it indicates

3288: level correlations). It would be interesting to extend such studies to other

3289: models and find some physical situations where these results would be

3290: applicable.

3291: \\

3292: $\bullet$ A beautiful heuristic argument was provided by Lifshitz to explain

3293: the spectral singularity for Hamiltonians with a weak concentration of

3294: localized scalar impurities (see for example the book \cite{LifGrePas88}).

3295: Other mechanisms should be invoked to explain the nature of the quantum states

3296: responsible for the low energy power law behaviour in the case of a uniform

3297: strong magnetic field with $\delta$-impurities \cite{Fur97,Fur00}.  For the

3298: model of randomly distributed magnetic fluxes, the numerical simulations for

3299: $\alpha=1/2$ have suggested the new type of singular behaviour

3300: (\ref{specsing}). It would be interesting to provide heuristic arguments to

3301: understand more deeply the origin of this singularity.

3302: 

3303: 

3304: 

3305: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3306: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3307: 

3308: \section*{Acknowledgments}

3309: 

3310: The article gives an overview of several joint works involving~: \'Eric

3311: Akkermans, Cyril Furtlehner, Satya Majumdar, Gilles Montambaux, C\'ecile

3312: Monthus, St\'ephane Ouvry and Marc Yor. We thank them for fruitful

3313: collaborations. Meanwhile we have also benefited from stimulating discussions

3314: with Marc Bocquet, Eug\`ene Bogomolny, Oriol Bohigas, H\'el\`ene Bouchiat,

3315: Markus B\"uttiker, David Dean, Richard Deblock, Meydi Ferrier, Sophie

3316: Gu\'eron, Jean-Marc Luck, Gleb Oshanin, Leonid Pastur and Denis Ullmo.

3317: We are pleased to acknowledge Satya Majumdar for his careful reading and 

3318: valuable remarks.

3319: 

3320: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3321: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

3322: 

3323: \newpage

3324: 

3325: %\bibliographystyle{chris4}

3326: %\bibliography{biblio}

3327: 

3328: \addcontentsline{toc}{section}{\protect\bibname}

3329: \begin{thebibliography}{200}

3330: 

3331: \bibitem{AbrAndLicRam79}

3332: E.~Abrahams, P.~W. Anderson, D.~C. Licciardello and T.~V. Ramakrishnan, Scaling

3333:   theory of localization: absence of quantum diffusion in two dimensions, Phys.

3334:   Rev. Lett. {\bf 42}(10), 673 (1979).

3335: 

3336: \bibitem{Ada92}

3337: V.~Adamyan, Scattering matrices for microschemes, Oper. Theory: Adv. \& Appl.

3338:   {\bf 59}, 1 (1992).

3339: 

3340: \bibitem{AkkComDesMonTex00}

3341: E.~Akkermans, A.~Comtet, J.~Desbois, G.~Montambaux and C.~Texier, On the

3342:   spectral determinant of quantum graphs, Ann. Phys. (N.Y.) {\bf 284}, 10--51

3343:   (2000).

3344: 

3345: \bibitem{AkkMon04}

3346: {\'E}.~Akkermans and G.~Montambaux, {\em Phy\-si\-que m\'esos\-co\-pi\-que des

3347:   \'elec\-trons et des pho\-tons\/}, EDP Sciences, CNRS \'editions (2004).

3348: 

3349: \bibitem{AlbGesHoeHol04}

3350: S.~Albeverio, S.~A. Gesztesy, R.~Hoegh-Krohn and H.~Holden, {\em Solvable

3351:   models in quantum mechanics\/}, A. M. S. Chelsea Publishing, second edition

3352:   (2004).

3353: 

3354: \bibitem{Ale83}

3355: S.~Alexander, Superconductivity of networks. A percolation approach to the

3356:   effects of disorder, Phys. Rev.~B {\bf 27}(3), 1541 (1983).

3357: 

3358: \bibitem{AltZir97}

3359: A.~Altland and M.~R. Zirnbauer, Nonstandard symmetry classes in mesoscopic

3360:   normal-superconducting hybrid structures, Phys. Rev.~B {\bf 55}(2), 1142

3361:   (1997).

3362: 

3363: \bibitem{AltAro81}

3364: B.~L. Al'tshuler and A.~G. Aronov, Magnetoresistance of thin films and of wires

3365:   in a longitudinal magnetic field, JETP Lett. {\bf 33}(10), 499 (1981).

3366: 

3367: \bibitem{AltAroKhm82}

3368: B.~L. Altshuler, A.~G. Aronov and D.~E. Khmelnitsky, Effects of

3369:   electron-electron collisions with small energy transfers on quantum

3370:   localisation, J.~Phys.~C: Solid St. Phys. {\bf 15}, 7367 (1982).

3371: 

3372: \bibitem{AltAroSpi81}

3373: B.~L. Al'tshuler, A.~G. Aronov and B.~Z. Spivak, The Aaronov-Bohm Effect in

3374:   disordered conductors, JETP Lett. {\bf 33}(2), 94 (1981).

3375: 

3376: \bibitem{AltAroSpiShaSha82}

3377: B.~L. Al'tshuler, A.~G. Aronov, B.~Z. Spivak, D.~{\relax{Yu}}. Sharvin and

3378:   {\relax{Yu}}.~V. Sharvin, Observation of the Aaronov-Bohm Effect in hollow

3379:   metal cylinders, JETP Lett. {\bf 35}(11), 588 (1982).

3380: 

3381: \bibitem{AltKhmLarLee80}

3382: B.~L. Al'tshuler, D.~E. Khmel'nitzki{\u\i}, A.~I. Larkin and P.~A. Lee,

3383:   Magnetoresistance and Hall effect in a disordered two-dimensional electron

3384:   gas, Phys. Rev.~B {\bf 22}(11), 5142 (1980).

3385: 

3386: \bibitem{AltLee88}

3387: B.~L. Al'tshuler and P.~A. Lee, Disordered Electronic Systems, Physics Today

3388:   {\bf 41}, 36 (december 1988).

3389: 

3390: \bibitem{AltPri89}

3391: B.~L. Altshuler and V.~N. Prigodin, Distribution of local density of states and

3392:   NMR line shape in a one-dimensional disordered conductor, Sov. Phys. JETP

3393:   {\bf 68}(1), 198 (1989).

3394: 

3395: \bibitem{AntPasSly81}

3396: T.~N. Antsygina, L.~A. Pastur and V.~A. Slyusarev, Localization of states and

3397:   kinetic properties of one-dimensional disordered systems, Sov. J. Low Temp.

3398:   Phys. {\bf 7}(1), 1--21 (1981).

3399: 

3400: \bibitem{AroSha87}

3401: A.~G. Aronov and {\relax{Yu}}.~V. Sharvin, Magnetic flux effects in disordered

3402:   conductors, Rev. Mod. Phys. {\bf 59}(3), 755 (1987).

3403: 

3404: \bibitem{AshMer76}

3405: N.~W. Ashcroft and N.~D. Mermin, {\em Solid State Physics\/}, Saunders College

3406:   (1976).

3407: 

3408: \bibitem{AvrExnLas94}

3409: J.~E. Avron, P.~Exner and Y.~Last, Periodic Schr{\"o}dinger operators with

3410:   large gaps and Wannier-Stark ladders, Phys. Rev. Lett. {\bf 72}(6), 896

3411:   (1994).

3412: 

3413: \bibitem{AvrRavZur88}

3414: J.~E. Avron, A.~Raveh and B.~Zur, Adiabatic quantum transport in multiply

3415:   connected systems, Rev. Mod. Phys. {\bf 60}, 873 (1988).

3416: 

3417: \bibitem{AvrSad91}

3418: J.~E. Avron and L.~Sadun, Adiabatic quantum transport in networks with

3419:   macroscopic components, Ann. Phys. (N.Y.) {\bf 206}, 440 (1991).

3420: 

3421: \bibitem{Azb83}

3422: M.~{\relax{Ya}}. Azbel, Resonance tunneling and localization spectroscopy,

3423:   Solid State Commun. {\bf 45}(7), 527 (1983).

3424: 

3425: \bibitem{BalHavWei87}

3426: R.~C. Ball, S.~Havlin and G.~H. Weiss, Non-Gaussian random walks, J.~Phys.~A:

3427:   Math. Gen. {\bf 20}, 4055 (1987).

3428: 

3429: \bibitem{BarPitYor89}

3430: M.~Barlow, J.~Pitman and M.~Yor, {\em Une extension multidimensionnelle de la

3431:   loi de l'arc sinus\/}, volume 1372 of {\em Lecture Notes in Maths\/}, p. 294,

3432:   Springer (1989).

3433: 

3434: \bibitem{BarGas00}

3435: F.~Barra and P.~Gaspard, On the level spacing distribution in quantum graphs,

3436:   J.~Stat. Phys. {\bf 101}, 283 (2000).

3437: 

3438: \bibitem{BarGas01}

3439: F.~Barra and P.~Gaspard, Transport and dynamics on open quantum graphs, Phys.

3440:   Rev.~E {\bf 65}, 016205 (2001).

3441: 

3442: \bibitem{Bar99}

3443: L.~Bartholdi, Counting paths in graphs, L'Enseignement Math\'ematique {\bf 45},

3444:   83 (1999).

3445: 

3446: \bibitem{BarTac93}

3447: A.~V. Barzykin and M.~Tachiya, Diffusion-influenced reaction kinetics on

3448:   fractal structures, J. Chem. Phys. {\bf 99}, 9591 (1993).

3449: 

3450: \bibitem{Bas92}

3451: H.~Bass, The Ihara-Selberg zeta function of a tree lattice, Internat. J. Math.

3452:   {\bf 3}, 717 (1992).

3453: 

3454: \bibitem{BauMalMonSamSchTex05}

3455: C.~Bau{\"e}rle, F.~Mallet, G.~Montambaux, L.~Saminadayar, F.~Schopfer and

3456:   C.~Texier, in preparation  (2005).

3457: 

3458: \bibitem{Bee97}

3459: C.~W.~J. Beenakker, Random-Matrix theory of quantum transport, Rev. Mod. Phys.

3460:   {\bf 69}(3), 731--808 (1997).

3461: 

3462: \bibitem{BenKliPlo90}

3463: S.~J. Bending, K.~{von Klitzing} and K.~Ploog, Weak localization in a

3464:   distribution of magnetic flux tubes, Phys. Rev. Lett. {\bf 65}(8), 1060

3465:   (1990).

3466: 

3467: \bibitem{Ber74}

3468: V.~L. Berezinski\u{i}, Kinetics of a quantum particle in a one-dimensional

3469:   random potential, Sov. Phys. JETP {\bf 38}(3), 620 (1974).

3470: 

3471: \bibitem{BerLoz94}

3472: O.~Bergman and G.~Lozano, Aharonov-Bohm scattering, contact interactions and

3473:   scale invariance, Ann. Phys. (N.Y.) {\bf 229}, 416 (1994).

3474: 

3475: \bibitem{Ber84}

3476: G.~Bergmann, Weak localization in thin films, Phys. Rep. {\bf 107}, 1 (1984).

3477: 

3478: \bibitem{BerBogKea01}

3479: G.~Berkolaiko, E.~B. Bogomolny and J.~P. Keating, Star graphs and {\v S}eba

3480:   billiards, J.~Phys.~A: Math. Gen. {\bf 34}, 335 (2001).

3481: 

3482: \bibitem{BerKea99}

3483: G.~Berkolaiko and J.~P. Keating, Two-point spectral correlations for star

3484:   graphs, J.~Phys.~A: Math. Gen. {\bf 32}, 7827 (1999).

3485: 

3486: \bibitem{BetUhl37}

3487: E.~Beth and G.~E. Uhlenbeck, Physica {\bf 4}, 915 (1937).

3488: 

3489: \bibitem{Boc99}

3490: M.~Bocquet, Some spectral properties of the one-dimensional disordered Dirac

3491:   equation, Nucl. Phys.~B [FS] {\bf 546}, 621 (1999).

3492: 

3493: \bibitem{Boc00}

3494: M.~Bocquet, {\em Cha\^{\i}nes de spins, fermions de Dirac et syst\`emes

3495:   d\'esordonn\'es\/}, Ph.D. thesis, \'Ecole Polytechnique (2000), available at

3496:   {\sf http://tel.ccsd.cnrs.fr/documents/archives0/00/00/15/60/}.

3497: 

3498: \bibitem{BocSerZir00}

3499: M.~Bocquet, D.~Serban and M.~R. Zirnbauer, Disordered 2d quasiparticles in

3500:   class D: Dirac fermions with random mass, and dirty superconductors, Nucl.

3501:   Phys.~B [FS] {\bf 578}, 628 (2000).

3502: 

3503: \bibitem{BouComGeoLeD87}

3504: J.-P. Bouchaud, A.~Comtet, A.~Georges and P.~{Le~Doussal}, The relaxation-time

3505:   spectrum of diffusion in a one-dimensional random medium: an exactly solvable

3506:   case, Europhys. Lett. {\bf 3}, 653 (1987).

3507: 

3508: \bibitem{BouComGeoLeD90}

3509: J.-P. Bouchaud, A.~Comtet, A.~Georges and P.~{Le~Doussal}, Classical diffusion

3510:   of a particle in a one-dimensional random force field, Ann. Phys. (N.Y.) {\bf

3511:   201}, 285--341 (1990).

3512: 

3513: \bibitem{BouGeo90}

3514: J.-P. Bouchaud and A.~Georges, Anomalous diffusion in disordered media:

3515:   Statistical mechanisms, models and physical applications, Phys. Rep. {\bf

3516:   195}, 267 (1990).

3517: 

3518: \bibitem{BouMez97}

3519: J.-P. Bouchaud and M.~M\'ezard, Universality classes for extreme value

3520:   statistic, J.~Phys.~A: Math. Gen. {\bf 30}, 7997 (1997).

3521: 

3522: \bibitem{BouLac85}

3523: P.~Bougerol and J.~Lacroix, {\em Products of random matrices with applications

3524:   to Schr\"odinger operators\/}, Birkh\"auser (1985).

3525: 

3526: \bibitem{BreGroItz84}

3527: {\'E}.~Br{\'e}zin, D.~J. Gross and C.~Itzykson, Density of states in the

3528:   presence of a strong magnetic field and random impurities, Nucl. Phys.~B [FS]

3529:   {\bf 235}, 24 (1984).

3530: 

3531: \bibitem{BroFraBee97}

3532: P.~W. Brouwer, K.~M. Frahm and C.~W. Beenakker, Quantum mechanical time-delay

3533:   matrix in chaotic scattering, Phys. Rev. Lett. {\bf 78}(25), 4737 (1997).

3534: 

3535: \bibitem{BroMudFur00}

3536: P.~W. Brouwer, C.~Mudry and A.~Furusaki, Density of States in Coupled Chains

3537:   with Off-Diagonal Disorder, Phys. Rev. Lett. {\bf 84}(13), 2913 (2000).

3538: 

3539: \bibitem{BroMudFur03}

3540: P.~W. Brouwer, C.~Mudry and A.~Furusaki, Universality of delocalization in

3541:   unconventional dirty superconducting wires with broken spin-rotation

3542:   symmetry, Phys. Rev.~B {\bf 67}, 014530 (2003).

3543: 

3544: \bibitem{But90}

3545: M.~B{\"u}ttiker, Traversal, reflection and dwell time for quantum tunneling, in

3546:   {\em Electronic properties of multilayers and low-dimensional semiconductors

3547:   structures\/}, edited by J.~M. {Chamberlain et al.}, p. 297, Plenum Press,

3548:   New York (1990).

3549: 

3550: \bibitem{But02}

3551: M.~B{\"u}ttiker, Charge densities and charge noise in mesoscopic conductors,

3552:   Pramana J. Phys. {\bf 58}, 241 (2002), (cond-mat/0112330).

3553: 

3554: \bibitem{ButImrAzb84}

3555: M.~B{\"u}ttiker, Y.~Imry and M.~{\relax{Ya}}. Azbel, Quantum oscillations in

3556:   one-dimensional normal-metal rings, Phys. Rev.~A {\bf 30}(4), 1982 (1984).

3557: 

3558: \bibitem{CarAku04}

3559: F.~Carlier and V.~M. Akulin, Quantum interference in nanofractals and its

3560:   optical manifestation, Phys. Rev.~B {\bf 69}, 115433 (2004).

3561: 

3562: \bibitem{CarLeD01}

3563: D.~Carpentier and P.~{Le~Doussal}, Glass transition of a particle in a random

3564:   potential, front selection and entropic phenomena in Liouville and

3565:   Sinh-Gordon models, Phys. Rev.~E {\bf 63}, 026110 (2001).

3566: 

3567: \bibitem{ChaSch86}

3568: S.~Chakravarty and A.~Schmid, Weak localization: the quasiclassical theory of

3569:   electrons in a random potential, Phys. Rep. {\bf 140}(4), 193 (1986).

3570: 

3571: \bibitem{ChaCod88}

3572: J.~T. Chalker and P.~D. Coddington, Percolation, quantum tunnelling and the

3573:   integer Hall-effect, J.~Phys.~C: Solid St. Phys. {\bf 21}(14), 2665 (1988).

3574: 

3575: \bibitem{Che99}

3576: L.~O. Chekhov, A spectral problem on graphs and $L$-functions, Russ. Math.

3577:   Surv. {\bf 54}(6), 1197 (1999).

3578: 

3579: \bibitem{CifReg75}

3580: D.~M. Cifarelli and E.~Regazzini, Contributi intorno ad un test per

3581:   l'homogeneita tra du campioni, Giornale Degli Economiste {\bf 34}, 233--249

3582:   (1975).

3583: 

3584: \bibitem{Col98}

3585: Y.~{Colin de Verdi\`ere}, {\em Spectres de Graphes\/}, Soci\'et\'e

3586:   Math\'ematique de France (1998).

3587: 

3588: \bibitem{ComDesMaj02}

3589: A.~Comtet, J.~Desbois and S.~N. Majumdar, The local time distribution of a

3590:   particle diffusing on a graph, J.~Phys.~A: Math. Gen. {\bf 35}, L687 (2002).

3591: 

3592: \bibitem{ComDesMon95}

3593: A.~Comtet, J.~Desbois and C.~Monthus, Localization properties in

3594:   one-dimensional disordered supersymmetric quantum mechanics, Ann. Phys.

3595:   (N.Y.) {\bf 239}, 312--350 (1995).

3596: 

3597: \bibitem{ComDesOuv90}

3598: A.~Comtet, J.~Desbois and S.~Ouvry, Winding of planar Brownian curves,

3599:   J.~Phys.~A: Math. Gen. {\bf 23}, 3563 (1990).

3600: 

3601: \bibitem{ComMasOuv95}

3602: A.~Comtet, S.~Mashkevich and S.~Ouvry, Magnetic moment and perturbation theory

3603:   with singular magnetic fields, Phys. Rev.~D {\bf 52}(4), 2594 (1995).

3604: 

3605: \bibitem{ComMonYor98}

3606: A.~Comtet, C.~Monthus and M.~Yor, Exponential functionals of Brownian motion

3607:   and disordered systems, J. Appl. Probab. {\bf 35}, 255 (1998).

3608: 

3609: \bibitem{ComTex97}

3610: A.~Comtet and C.~Texier, On the distribution of the Wigner time delay in

3611:   one-dimensional disordered systems, J.~Phys.~A: Math. Gen. {\bf 30},

3612:   8017--8025 (1997).

3613: 

3614: \bibitem{ComTex98}

3615: A.~Comtet and C.~Texier, One-dimensional disordered supersymmetric quantum

3616:   mechanics: a brief survey, in {\em Supersymmetry and Integrable Models\/},

3617:   edited by H.~Aratyn, T.~D. Imbo, W.-Y. Keung and U.~Sukhatme, Lecture Notes

3618:   in Physics, Vol. 502, pp. 313--328. Springer (1998), Proceedings of a

3619:   workshop held at Chicago, IL, USA, 12-14 June 1997 (also available as

3620:   cond-mat/9707313).

3621: 

3622: \bibitem{DasMaBer69}

3623: R.~Dashen, S.-K. Ma and H.~J. Bernstein, $S$-matrix formulation of statistical

3624:   mechanics, Phys. Rev. {\bf 187}(1), 345 (1969).

3625: 

3626: \bibitem{CalLucNiePet85}

3627: C.~de~Calan, J.-M. Luck, T.~M. Nieuwenhuizen and D.~P{\'e}tritis, On the

3628:   distribution of a random variable occuring in 1D disordered systems,

3629:   J.~Phys.~A: Math. Gen. {\bf 18}, 501 (1985).

3630: 

3631: \bibitem{CarNus02}

3632: C.~A.~A. de~Carvalho and H.~M. Nussenzveig, Time delay, Phys. Rep. {\bf 364},

3633:   83 (2002).

3634: 

3635: \bibitem{DeaMaj01}

3636: D.~S. Dean and S.~N. Majumdar, Extreme Value Statistics of Hierarchically

3637:   Correlated Variables: Deviation from Gumbel Statistics and Anomalous

3638:   Persistence, Phys. Rev.~E {\bf 64}, 046121 (2001).

3639: 

3640: \bibitem{DerHil83}

3641: B.~Derrida, Singular behaviour of certain infinite products of random

3642:   $2\times2$ matrices, J.~Phys.~A: Math. Gen. {\bf 16}, 2641 (1983).

3643: 

3644: \bibitem{Des00}

3645: J.~Desbois, Spectral determinant of Schr\"odinger operators on graphs,

3646:   J.~Phys.~A: Math. Gen. {\bf 33}, L63 (2000).

3647: 

3648: \bibitem{Des00a}

3649: J.~Desbois, Time-dependent harmonic oscillator and spectral determinant on

3650:   graphs, Eur. Phys. J.~B {\bf 15}, 201 (2000).

3651: 

3652: \bibitem{Des01}

3653: J.~Desbois, Spectral determinant on graphs with generalized boundary

3654:   conditions, Eur. Phys. J.~B {\bf 24}, 261 (2001).

3655: 

3656: \bibitem{Des02}

3657: J.~Desbois, Occupation times distribution for Brownian motion on graphs,

3658:   J.~Phys.~A: Math. Gen. {\bf 35}, L673 (2002).

3659: 

3660: \bibitem{DesCom92}

3661: J.~Desbois and A.~Comtet, Algebraic areas enclosed by Brownian curves on

3662:   bounded domains, J.~Phys.~A: Math. Gen. {\bf 25}, 3095 (1992).

3663: 

3664: \bibitem{DesFurOuv95}

3665: J.~Desbois, C.~Furtlehner and S.~Ouvry, Random magnetic impurities and the

3666:   Landau problem, Nucl. Phys.~B [FS] {\bf 453}, 759--776 (1995).

3667: 

3668: \bibitem{DesFurOuv96}

3669: J.~Desbois, C.~Furtlehner and S.~Ouvry, Random magnetic impurities and the

3670:   $\delta$ impurity problem, J.~Phys.~I (France) {\bf 6}, 641--648 (1996).

3671: 

3672: \bibitem{Dor82}

3673: O.~N. Dorokhov, Transmission coefficient and the localization length of an

3674:   electron in $N$ bound disordered chains, JETP Lett. {\bf 36}, 318 (1982).

3675: 

3676: \bibitem{Dor88}

3677: O.~N. Dorokhov, Solvable model of multichannel localization, Phys. Rev.~B {\bf

3678:   37}, 10526 (1988).

3679: 

3680: \bibitem{DouRam85}

3681: B.~Dou{\c c}ot and R.~Rammal, Quantum oscillations in normal-metal networks,

3682:   Phys. Rev. Lett. {\bf 55}(10), 1148 (1985).

3683: 

3684: \bibitem{DouRam86}

3685: B.~Dou{\c c}ot and R.~Rammal, Interference effects and magnetoresistance

3686:   oscillations in normal-metal networks: 1. weak localization approach, J.

3687:   Physique {\bf 47}, 973--999 (1986).

3688: 

3689: \bibitem{Dys53}

3690: F.~J. Dyson, The dynamics of a disordered linear chain, Phys. Rev. {\bf 92},

3691:   1331 (1953).

3692: 

3693: \bibitem{EchGerBozBogNil93b}

3694: P.~M. Echternach, M.~E. Gershenson, H.~M. Bozler, A.~L. Bogdanov and

3695:   B.~Nilsson, Nyquist phase relaxation in one-dimensional metal films, Phys.

3696:   Rev.~B {\bf 48}(15), 11516 (1993).

3697: 

3698: \bibitem{Edw67}

3699: S.~F. Edwards, Statistical mechanics with topological constraints, Proc. Phys.

3700:   Soc. {\bf 91}, 513 (1967).

3701: 

3702: \bibitem{Efe97}

3703: K.~Efetov, {\em Supersymmetry in disorder and chaos\/}, Cambridge University

3704:   Press (1997).

3705: 

3706: \bibitem{EfrPol85}

3707: A.~L. Efros and M.~Pollak (editors), {\em Electron-electron interactions in

3708:   disordered systems\/}, North-Holland (1985).

3709: 

3710: \bibitem{EvaKat03}

3711: S.~N. Evangelou and D.~E. Katsanos, Spectral statistics in chiral-orthogonal

3712:   disordered systems, J.~Phys.~A: Math. Gen. {\bf 36}, 3237 (2003).

3713: 

3714: \bibitem{Exn95}

3715: P.~Exner, Lattice Kronig-Penney Models, Phys. Rev. Lett. {\bf 74}(18), 3503

3716:   (1995).

3717: 

3718: \bibitem{Exn96a}

3719: P.~Exner, Contact interactions on graph superlattices, J.~Phys.~A: Math. Gen.

3720:   {\bf 29}(1), 87 (1996).

3721: 

3722: \bibitem{FabMel97}

3723: M.~Fabrizio and R.~M\'elin, Coexistence of antiferromagnetism and dimerization

3724:   in a disordered spin-Peierls model: exact results, Phys. Rev. Lett. {\bf

3725:   78}(17), 3382 (1997).

3726: 

3727: \bibitem{FabMel97b}

3728: M.~Fabrizio and R.~M\'elin, Enhanced magnetic fluctuations in doped

3729:   spin-Peierls systems: a single-chain-model analysis, Phys. Rev.~B {\bf

3730:   56}(10), 5996 (1997).

3731: 

3732: \bibitem{FarTsa94}

3733: W.~G. Faris and W.~J. Tsay, Time delay in random scattering, SIAM J. Appl.

3734:   Math. {\bf 54}, 443 (1994).

3735: 

3736: \bibitem{FerAngRowGueBouTexMonMai04}

3737: M.~Ferrier, L.~Angers, A.~C.~H. Rowe, S.~Gu{\'e}ron, H.~Bouchiat, C.~Texier,

3738:   G.~Montambaux and D.~Mailly, Direct measurement of the phase coherence length

3739:   in a GaAs/GaAlAs square network, Phys. Rev. Lett. {\bf 93}, 246804 (2004).

3740: 

3741: \bibitem{FlaLou01}

3742: P.~Flajeolet and G.~Louchard, Analytic variations on the Airy distribution,

3743:   Algorithmica {\bf 31}, 361 (2001).

3744: 

3745: \bibitem{For87}

3746: R.~Forman, Functional determinants and geometry, Invent. math. {\bf 88}, 447

3747:   (1987).

3748: 

3749: \bibitem{Fri52}

3750: J.~Friedel, The distribution of electrons round impurities in monovalent

3751:   metals, Phil. Mag. {\bf 43}, 153 (1952).

3752: 

3753: \bibitem{Fri58}

3754: J.~Friedel, Metallic alloys, Nuovo Cimento Suppl. {\bf 7}, 287 (1958).

3755: 

3756: \bibitem{FriLlo60}

3757: H.~L. Frisch and S.~P. Lloyd, Electron levels in a one-dimensional random

3758:   lattice, Phys. Rev. {\bf 120}(4), 1175 (1960).

3759: 

3760: \bibitem{Fur97}

3761: C.~Furtlehner, {\em \'Etude du spectre de Landau pour un champ magn\'etique

3762:   al\'eatoire en dimension 2\/}, Ph.D. thesis, Universit\'e Paris 6 (1997).

3763: 

3764: \bibitem{Fur00}

3765: C.~Furtlehner, Lifshitz-like argument for low-lying states in a strong magnetic

3766:   field, Eur. Phys. J.~B {\bf 18}, 297 (2000).

3767: 

3768: \bibitem{FyoSom96a}

3769: Y.~V. Fyodorov and H.-J. Sommers, Parametric correlations of scattering phase

3770:   shifts and fluctuations of delay times in few-channel chaotic scattering,

3771:   Phys. Rev. Lett. {\bf 76}, 4709 (1996).

3772: 

3773: \bibitem{FyoSom97}

3774: Y.~V. Fyodorov and H.-J. Sommers, Statistics of resonance poles, phase shift

3775:   and time delays in quantum chaotic scattering: Random matrix approach for

3776:   systems with broken time-reversal invariance, J. Math. Phys. {\bf 38}(4),

3777:   1918 (1997).

3778: 

3779: \bibitem{GarTru05}

3780: C.~Garban and J.~A. {Trujillo Ferreras}, The expected area of the filled planar

3781:   Brownian loop is $\pi/5$, unpublished  (2005), preprint math.PR/0504496.

3782: 

3783: \bibitem{Gar89}

3784: C.~W. Gardiner, {\em Handbook of stochastic methods for physics, chemistry and

3785:   the natural sciences\/}, Springer (1989).

3786: 

3787: \bibitem{GefImrAzb84}

3788: Y.~Gefen, Y.~Imry and M.~{\relax{Ya}}. Azbel, Quantum oscillations and the

3789:   Aharonov-Bohm effect for parallel resistors, Phys. Rev. Lett. {\bf 52}(2),

3790:   129 (1984).

3791: 

3792: \bibitem{Gei89}

3793: A.~K. Geim, Nonlocal magnetoresistance of bismuth films in nonuniform field of

3794:   Abrikosov vortices, JETP Lett. {\bf 50}(8), 389 (1989).

3795: 

3796: \bibitem{GerPav88}

3797: N.~I. Gerasimenko and B.~S. Pavlov, Scattering problems on noncompact graphs,

3798:   Theor. Math. Phys. {\bf 74}, 230 (1988).

3799: 

3800: \bibitem{GerKhaMikBozBog97}

3801: M.~E. Gershenson, {\relax{Yu}}.~B. Khavin, A.~Mikhalchuk, H.~M. Bolzer and

3802:   A.~L. Bogdanov, Crossover from Weak to Strong Localization in

3803:   Quasi-One-Dimensional Conductors, Phys. Rev. Lett. {\bf 79}(4), 725 (1997).

3804: 

3805: \bibitem{GnuAlt04}

3806: S.~Gnutzmann and A.~Altland, Universal Spectral Statistics in Quantum Graphs,

3807:   Phys. Rev. Lett. {\bf 93}, 194101 (2004).

3808: 

3809: \bibitem{GogMel77}

3810: A.~A. Gogolin and V.~I. Mel'nikov, Conductivity of one-dimensional metal with

3811:   half-filled band, Sov. Phys. JETP {\bf 46}, 369 (1977).

3812: 

3813: \bibitem{GogMelRas76}

3814: A.~A. Gogolin, V.~I. Mel'nikov and E.~I. Rashba, Conductivity in a disordered

3815:   one-dimensional system induced by electron-phonon interaction, Sov. Phys.

3816:   JETP {\bf 42}(1), 168 (1976).

3817: 

3818: \bibitem{GolMolPas77}

3819: I.~{\relax{Ya}}. Gol'dshtein, S.~A. Molchanov and L.~A. Pastur, A pure point

3820:   spectrum of the stochastic one-dimensional Schr\"odinger operator, Funct.

3821:   Anal. and App. {\bf 11}, 1 (1977).

3822: 

3823: \bibitem{GopMelBut96}

3824: V.~A. Gopar, P.~A. Mello and M.~B{\"u}ttiker, Mesoscopic capacitors: a

3825:   statistical analysis, Phys. Rev. Lett. {\bf 77}(14), 3005 (1996).

3826: 

3827: \bibitem{GorDorPri83}

3828: L.~P. Gor'kov, O.~N. Dorokhov and F.~V. Prigara, Structure of wave functions an

3829:   AC conductivity id disordered one-dimensional conductors, Sov. Phys. JETP

3830:   {\bf 58}(4), 852 (1983).

3831: 

3832: \bibitem{GorLarKhm79}

3833: L.~P. Gor'kov, A.~I. Larkin and D.~E. Khmel'nitzki{\u\i}, Particle conductivity

3834:   in a two-dimensional random potential, JETP Lett. {\bf 30}(4), 228 (1979).

3835: 

3836: \bibitem{GreMolSud83}

3837: L.~N. Grenkova, S.~A. Mol\v{c}anov and J.~N. Sudarev, On the Basic States of

3838:   One-Dimensional Disordered Structures, Commun. Math. Phys. {\bf 90}, 101

3839:   (1983).

3840: 

3841: \bibitem{GroFri03}

3842: A.~Grosberg and H.~Frisch, Winding angle distribution for planar random walk,

3843:   polymer ring entangled with an obstacle, and all that:

3844:   Spitzer-Edwards-Prager-Frisch model revisited, J.~Phys.~A: Math. Gen. {\bf

3845:   36}, 8955 (2003).

3846: 

3847: \bibitem{Gum35}

3848: E.~J. Gumbel, Les valeurs extr\^emes des distributions statistiques, Ann. de

3849:   l'Institut Henri Poincar\'e {\bf V}, 115 (1935).

3850: 

3851: \bibitem{Gum54}

3852: E.~J. Gumbel, Statistical Theory of Extreme Values and Some Practical

3853:   Applications, National Bureau of Standards Applied Mathematics Series {\bf

3854:   33} (1954), Issued February 12.

3855: 

3856: \bibitem{Gum58}

3857: E.~J. Gumbel, {\em Statistics of Extremes\/}, Columbia University Press, New

3858:   York (1958).

3859: 

3860: \bibitem{GurCha03}

3861: V.~Gurarie and J.~T. Chalker, Bosonic excitations in random media, Phys. Rev.~B

3862:   {\bf 68}, 134207 (2003).

3863: 

3864: \bibitem{Gut90}

3865: M.~G. Gutzwiller, {\em Chaos in Classical and Quantum Mechanics\/}, volume~1 of

3866:   {\em Interdisciplinary Applied Mathematics\/}, Springer Verlag, New York

3867:   (1990).

3868: 

3869: \bibitem{Hal65}

3870: B.~I. Halperin, Green's functions for a particle in a one-dimensional random

3871:   potential, Phys. Rev. {\bf 139}, A104 (1965).

3872: 

3873: \bibitem{HauSto89}

3874: E.~H. Hauge and J.~A. St{\o}vneng, Tunneling times: a critical review, Rev.

3875:   Mod. Phys. {\bf 61}(4), 917 (1989).

3876: 

3877: \bibitem{HeiHucZir05}

3878: P.~Heinzner, A.~Huckleberry and M.~R. Zirnbauer, Symmetry classes of disordered

3879:   fermions, Commun. Math. Phys. {\bf 257}, 725 (2005).

3880: 

3881: \bibitem{HikLarNag80}

3882: S.~Hikami, A.~I. Larkin and Y.~Nagaoka, Spin-Orbit Interaction and

3883:   Magnetoresistance in the Two Dimensional Random System, Prog. Theor. Phys.

3884:   {\bf 63}(2), 707 (1980).

3885: 

3886: \bibitem{IglMon05}

3887: F.~Igloi and C.~Monthus, Strong disorder RG approach of random systems, Phys.

3888:   Rep. {\bf 412}, 277 (2005).

3889: 

3890: \bibitem{Iha66}

3891: Y.~Ihara, On discrete subgroup of the two by two projective linear group over

3892:   $p$-adic field, J. Math. Soc. Japan {\bf 18}(3), 219 (1966).

3893: 

3894: \bibitem{ItzDro89}

3895: C.~Itzykson and J.-M. Drouffe, {\em Th\'eorie statistique des champs\/},

3896:   Inter\'editions--{\sc Cnrs}, Paris (1989), Tomes 1 et 2.

3897: 

3898: \bibitem{JacPi90}

3899: R.~Jackiw and S.~Y. Pi, Classical and quantal nonrelativistic Chern-Simons

3900:   theory, Phys. Rev.~D {\bf 42}, 3500 (1990).

3901: 

3902: \bibitem{JayVijKum89}

3903: A.~M. Jayannavar, G.~V. Vijayagovindan and N.~Kumar, Energy dispersive

3904:   backscattering of electrons from surface resonances of a disordered medium

3905:   and $1/f$ noise, Z. Phys. B - Condens. Matter {\bf 75}, 77 (1989).

3906: 

3907: \bibitem{JeaPitYor97}

3908: M.~Jeanblanc, J.~Pitman and M.~Yor, The Feynman-Kac formula and decomposition

3909:   of Brownian paths, Comput. Appl. Math. {\bf 16}(1), 27 (1997).

3910: 

3911: \bibitem{Jon83}

3912: G.~Jona-Lasinio, Qualitative theory of stochastic differential equations and

3913:   quantum mechanics of disordered systems, Helv. Phys. Act. {\bf 56}, 61

3914:   (1983).

3915: 

3916: \bibitem{Kac46}

3917: M.~Kac, On the average of a certain Wiener functional and a related limit

3918:   theorem in calculus of probability, Trans. Am. Math. Soc. {\bf 59}, 401

3919:   (1946).

3920: 

3921: \bibitem{KeaMarWin03}

3922: J.~P. Keating, J.~Marklof and B.~Winn, Value distribution of the eigenfunctions

3923:   and spectral determinants of quantum star graphs, Commun. Math. Phys. {\bf

3924:   241}, 421 (2003).

3925: 

3926: \bibitem{Kel64}

3927: L.~V. Keldysh, Deep levels in semiconductors, Sov. Phys. JETP {\bf 18}(1), 253

3928:   (1964).

3929: 

3930: \bibitem{Kes73}

3931: H.~Kesten, Acta Math. {\bf 131}, 208 (1973).

3932: 

3933: \bibitem{KosSch99}

3934: V.~Kostrykin and R.~Schrader, Kirchhoff's rule for quantum wires, J.~Phys.~A:

3935:   Math. Gen. {\bf 32}, 595 (1999).

3936: 

3937: \bibitem{KotSmi97}

3938: T.~Kottos and U.~Smilansky, Quantum Chaos on Graphs, Phys. Rev. Lett. {\bf

3939:   79}(24), 4794 (1997).

3940: 

3941: \bibitem{KotSmi99}

3942: T.~Kottos and U.~Smilansky, Periodic Orbit Theory and Spectral Statistics for

3943:   Quantum Graphs, Ann. Phys. (N.Y.) {\bf 274}(1), 76 (1999).

3944: 

3945: \bibitem{KotSmi00}

3946: T.~Kottos and U.~Smilansky, Chaotic Scattering on Graphs, Phys. Rev. Lett. {\bf

3947:   85}(5), 968 (2000).

3948: 

3949: \bibitem{KotSmi03}

3950: T.~Kottos and U.~Smilansky, Quantum graphs~: A simple model for chaotic

3951:   scattering, J.~Phys.~A: Math. Gen. {\bf 36}, 3501 (2003).

3952: 

3953: \bibitem{Kre53}

3954: M.~G. Krein, Trace formulas in perturbation theory, Matem. Sbornik {\bf 33},

3955:   597 (1953).

3956: 

3957: \bibitem{KucSad98}

3958: E.~Z. Kuchinskii and M.~V. Sadovskii, Combinatorics of Feynman diagrams for the

3959:   problems with Gaussian random field, Sov. Phys. JETP {\bf 86}, 367 (1998).

3960: 

3961: \bibitem{Kuc04}

3962: P.~Kuchment, Quantum Graphs: I. Some basic structures, Waves Random Media {\bf

3963:   14}, S107 (2004).

3964: 

3965: \bibitem{Kuc05}

3966: P.~Kuchment, Quantum Graphs: II. Some spectral properties of quantum and

3967:   combinatorial graphs, J.~Phys.~A: Math. Gen. {\bf 38}, 4887 (2005).

3968: 

3969: \bibitem{LanLif66e}

3970: L.~D. Landau and E.~Lifchitz, {\em Physique statistique\/}, Mir (1966), tome 5.

3971: 

3972: \bibitem{LanMar94}

3973: R.~Landauer and T.~Martin, Barrier interaction time in tunneling, Rev. Mod.

3974:   Phys. {\bf 66}(1), 217 (1994).

3975: 

3976: \bibitem{LeDMonFis99}

3977: P.~{Le~Doussal}, C.~Monthus and D.~S. Fisher, Random walkers in one-dimensional

3978:   random environments: Exact renormalization group analysis, Phys. Rev.~E {\bf

3979:   59}(5), 4795 (1999).

3980: 

3981: \bibitem{LeG92}

3982: J.-F. {Le~Gall}, {\em Some properties of planar Brownian motion\/}, Lectures

3983:   Notes in Maths, Springer, Berlin (1992), p.~1527.

3984: 

3985: \bibitem{LeeRam85}

3986: P.~A. Lee and T.~V. Ramakrishnan, Disordered electronic systems, Rev. Mod.

3987:   Phys. {\bf 57}, 287 (1985).

3988: 

3989: \bibitem{LesWar04}

3990: H.~Leschke and S.~Warzel, Quantum-Classical Transitions in Lifshitz Tails with

3991:   Magnetic Fields, Phys. Rev. Lett. {\bf 92}, 086402 (2004).

3992: 

3993: \bibitem{Leu00}

3994: C.~Leuridan, Th\'eor\`eme de Ray-Knight dans un arbre~: une approche

3995:   alg\'ebrique, Pr\'epublication de l'institut Fourier n$^{\rm o}509$  (2000),

3996:   {\sf http://www-fourier.ujf-grenoble.fr/prepublications.html}.

3997: 

3998: \bibitem{Lev48}

3999: P.~{L\'evy}, {\em Processus stochastiques et mouvement brownien\/}, \'Editions

4000:   Jacques Gabay, Paris (1948).

4001: 

4002: \bibitem{LicTho75}

4003: D.~C. Licciardello and D.~J. Thouless, Constancy of Minimum Metallic

4004:   Conductivity in Two Dimensions, Phys. Rev. Lett. {\bf 35}(21), 1475 (1975).

4005: 

4006: \bibitem{LifGrePas88}

4007: I.~M. Lifshits, S.~A. Gredeskul and L.~A. Pastur, {\em Introduction to the

4008:   theory of disordered systems\/}, John Wiley \& Sons (1988).

4009: 

4010: \bibitem{Luc92}

4011: J.-M. Luck, {\em Syst\`emes d\'esordonn\'es unidimensionnels\/}, CEA,

4012:   collection Al\'ea Saclay, Saclay (1992).

4013: 

4014: \bibitem{LudMir04}

4015: T.~Ludwig and A.~D. Mirlin, Interaction-induced dephasing of Aharonov-Bohm

4016:   oscillations, Phys. Rev.~B {\bf 69}, 193306 (2004).

4017: 

4018: \bibitem{MajCom04}

4019: S.~N. Majumdar and A.~Comtet, Exact maximal height distribution of fluctuating

4020:   interfaces, Phys. Rev. Lett. {\bf 92}(22), 225501 (2004).

4021: 

4022: \bibitem{MajCom05}

4023: S.~N. Majumdar and A.~Comtet, Airy Distribution Function: From the Area Under a

4024:   Brownian Excursion to the Maximal Height of Fluctuating Interfaces, J. Stat.

4025:   Phys. {\bf 119}, 777 (2005).

4026: 

4027: \bibitem{MajKra03}

4028: S.~N. Majumdar and P.~L. Krapivsky, Extreme Value Statistics and Traveling

4029:   Fronts: Various Applications, Physica A {\bf 318}, 161 (2003).

4030: 

4031: \bibitem{McCOuv91}

4032: J.~McCabe and S.~Ouvry, Perturbative three-body spectrum and the third virial

4033:   coefficient in the anyon model, Phys. Lett.~B {\bf 260}, 113 (1991).

4034: 

4035: \bibitem{McKTar95}

4036: A.~J. McKane and M.~B. Tarlie, Regularization of functional determinants using

4037:   boundary perturbations, J.~Phys.~A: Math. Gen. {\bf 28}, 6931 (1995).

4038: 

4039: \bibitem{McK94}

4040: H.~P. McKean, A Limit Law for the Ground State of Hill's Equation, J.~Stat.

4041:   Phys. {\bf 74}(5/6), 1227 (1994).

4042: 

4043: \bibitem{MelPerKum88}

4044: P.~A. Mello, P.~Pereyra and N.~Kumar, Macroscopic approach to multichannel

4045:   disordered conductors, Ann. Phys. (N.Y.) {\bf 181}, 290 (1988).

4046: 

4047: \bibitem{Mol81}

4048: S.~A. Mol\v{c}anov, The Local Structure of the Spectrum of the One-Dimensional

4049:   Schr\"odinger Operator, Commun. Math. Phys. {\bf 78}, 429 (1981).

4050: 

4051: \bibitem{MonAkk05}

4052: G.~Montambaux and E.~Akkermans, Non exponential quasiparticle decay and phase

4053:   relaxation in low dimensional conductors, Phys. Rev. Lett. {\bf 95}, 016403

4054:   (2005).

4055: 

4056: \bibitem{Mon95}

4057: C.~Monthus, {\em \'Etude de quelques fonctionnelles du mouvement brownien et de

4058:   certaines propri\'et\'es de la diffusion unidimensionnelle en milieu

4059:   al\'eatoire\/}, Ph.D. thesis, Universit\'e Paris 6 (1995), Ann. Phys.

4060:   (France) {\bf20}, 341 (1995).

4061: 

4062: \bibitem{MonCom94}

4063: C.~Monthus and A.~Comtet, On the flux distribution in a one-dimensional

4064:   disordered system, J.~Phys.~I (France) {\bf 4}, 635--653 (1994).

4065: 

4066: \bibitem{MonOshComBur96}

4067: C.~Monthus, G.~Oshanin, A.~Comtet and S.~Burlatsky, Sample-size dependence of

4068:   the ground-state energy in a one-dimensional localization problem, Phys.

4069:   Rev.~E {\bf 54}(1), 231 (1996).

4070: 

4071: \bibitem{NerTsvWen94}

4072: A.~A. Nersesyan, A.~M. Tsvelik and F.~Wenger, Disorder effects in

4073:   two-dimensional $d$-wave superconductors, Phys. Rev. Lett. {\bf 72}(16), 2628

4074:   (1994).

4075: 

4076: \bibitem{OshMogMor93}

4077: G.~Oshanin, A.~Mogutov and M.~Moreau, Steady flux in a continuous-space Sinai

4078:   chain, J.~Stat. Phys. {\bf 73}, 379 (1993).

4079: 

4080: \bibitem{OssFyo05}

4081: A.~Ossipov and Y.~V. Fyodorov, Statistics of delay times in mesoscopic systems

4082:   as a manifestation of eigenfunction fluctuations, Phys. Rev.~B {\bf 71},

4083:   125133 (2005).

4084: 

4085: \bibitem{Ouv94}

4086: S.~Ouvry, $\delta$ perturbative interactions in Aharonov-Bohm and anyons

4087:   models, Phys. Rev.~D {\bf 50}, 5296 (1994).

4088: 

4089: \bibitem{OvcEri77}

4090: A.~A. Ovchinnikov and N.~S. Erikmann, Density of states in a one-dimensional

4091:   random potential, Sov. Phys. JETP {\bf 46}, 340 (1977).

4092: 

4093: \bibitem{Pas98}

4094: M.~Pascaud, {\em Magn\'etisme orbital de conducteurs m\'esoscopiques

4095:   d\'esordonn\'es et propri\'et\'es spectrales de fermions en interaction\/},

4096:   Ph.D. thesis, Universit\'e Paris 11 (1998).

4097: 

4098: \bibitem{PasMon99}

4099: M.~Pascaud and G.~Montambaux, Persistent currents on networks, Phys. Rev. Lett.

4100:   {\bf 82}, 4512 (1999).

4101: 

4102: \bibitem{PerWel96}

4103: M.~Perman and J.~A. Wellner, On the distribution of Brownian areas, Ann. Appl.

4104:   Probab. {\bf 6}(4), 1091 (1996).

4105: 

4106: \bibitem{PieGouAntPotEstBir03}

4107: F.~Pierre, A.~B. Gougam, A.~Anthore, H.~Pothier, D.~Esteve and N.~O. Birge,

4108:   Dephasing of electrons in mesoscopic metal wires, Phys. Rev.~B {\bf 68},

4109:   085413 (2003).

4110: 

4111: \bibitem{PitYor86}

4112: J.~Pitman and M.~Yor, Asymptotic laws of planar Brownian motion, Ann. Probab.

4113:   {\bf 14}(3), 733 (1986).

4114: 

4115: \bibitem{RamShe87}

4116: J.~Rammer and A.~L. Shelankov, Weak localization in inhomogeneous magnetic

4117:   fields, Phys. Rev.~B {\bf 36}(6), 3135 (1987).

4118: 

4119: \bibitem{Ric82}

4120: S.~O. Rice, The integral of the absolute value of the pinned Wiener process --

4121:   Calculation of its probability density by numerical integration, Ann. Probab.

4122:   {\bf 10}(1), 240 (1982).

4123: 

4124: \bibitem{Ric03}

4125: C.~Richard, Area distribution of the planar random loop boundary, J.~Phys.~A:

4126:   Math. Gen. {\bf 37}, 4493 (2003).

4127: 

4128: \bibitem{Rot83a}

4129: J.-P. Roth, Le spectre du Laplacien sur un graphe, in {\em Colloque de

4130:   Th\'eorie du potentiel - Jacques Deny\/}, p. 521, Orsay (1983).

4131: 

4132: \bibitem{Rot83}

4133: J.-P. Roth, Spectre du Laplacien sur un graphe, C. R. Acad. Sc. Paris {\bf

4134:   296}, 793 (1983).

4135: 

4136: \bibitem{RudSch53}

4137: K.~Rudenberg and C.~Scherr, J. Chem. Phys. {\bf 21}, 1565 (1953).

4138: 

4139: \bibitem{Sch57}

4140: H.~Schmidt, Disordered one-dimensional crystals, Phys. Rev. {\bf 105}(2), 425

4141:   (1957).

4142: 

4143: \bibitem{SchTit03}

4144: H.~Schomerus and M.~Titov, Band-center anomaly of the conductance distribution

4145:   in one-dimensional Anderson localization, Phys. Rev.~B {\bf 67}, 100201

4146:   (2003).

4147: 

4148: \bibitem{Sel56}

4149: A.~Selberg, Harmonic analysis and discontinuous groups in weakly symmetric

4150:   Riemannian spaces, with applications to Dirichlet series, J. Ind. Math. Soc.

4151:   {\bf 20}, 47 (1956).

4152: 

4153: \bibitem{Sha82}

4154: B.~Shapiro, Renormalization-Group transformation for the Anderson transition,

4155:   Phys. Rev. Lett. {\bf 48}(12), 823 (1982).

4156: 

4157: \bibitem{Sha83}

4158: B.~Shapiro, Quantum conduction on a Cayley tree, Phys. Rev. Lett. {\bf 50}(10),

4159:   747 (1983).

4160: 

4161: \bibitem{ShaSha82}

4162: D.~{\relax{Yu}}. Sharvin and {\relax{Yu}}.~V. Sharvin, Magnetic-flux

4163:   quantization in a cylindrical film of a normal metal, JETP Lett. {\bf 34}(5),

4164:   272 (1982).

4165: 

4166: \bibitem{SheTsv98}

4167: D.~G. Shelton and A.~M. Tsvelik, Effective theory for midgap states in doped

4168:   spin-ladder and spin Peierls systems: Liouville quantum mechanics, Phys.

4169:   Rev.~B {\bf 57}(22), 14242 (1998).

4170: 

4171: \bibitem{She82}

4172: L.~A. Shepp, On the integral of the absolute value of the pinned Wiener

4173:   process, Ann. Probab. {\bf 10}(1), 234 (1982), [Acknowledgement of priority:

4174:   Ann. Probab. {\bf 19}, 1397 (1991)].

4175: 

4176: \bibitem{Smi60}

4177: F.~T. Smith, Lifetime matrix in collision theory, Phys. Rev. {\bf 118}, 349

4178:   (1960).

4179: 

4180: \bibitem{StaTer96}

4181: H.~M. Stark and A.~A. Terras, Zeta functions of finite graphs and coverings,

4182:   Adv. Math. {\bf 121}, 124 (1996).

4183: 

4184: \bibitem{SteCheFabGog99}

4185: M.~Steiner, Y.~Chen, M.~Fabrizio and A.~O. Gogolin, Statistical properties of

4186:   localization-delocalization transition in one dimension, Phys. Rev.~B {\bf

4187:   59}(23), 14848 (1999).

4188: 

4189: \bibitem{SteFabGog98}

4190: M.~Steiner, M.~Fabrizio and A.~O. Gogolin, Random mass Dirac fermions in doped

4191:   spin-Peierls and spin-ladder systems: one-particle properties and boundary

4192:   effects, Phys. Rev.~B {\bf 57}(14), 8290 (1998).

4193: 

4194: \bibitem{TakLinMak80}

4195: H.~Takayama, Y.~R. {Lin-Liu} and K.~Maki, Continuum model for solitons in

4196:   polyacetylene, Phys. Rev.~B {\bf 21}(6), 2388 (1980).

4197: 

4198: \bibitem{Tan01b}

4199: G.~Tanner, Unitary-stochastic matrix ensembles and spectral statistics,

4200:   J.~Phys.~A: Math. Gen. {\bf 34}, 8485 (2001).

4201: 

4202: \bibitem{Tan05}

4203: G.~Tanner, From quantum graphs to quantum random walks, unpublished  (2005),

4204:   preprint quant-ph/0504224.

4205: 

4206: \bibitem{Tex99}

4207: C.~Texier, {\em Quelques aspects du transport quantique dans les syst\`emes

4208:   d\'esordonn\'es de basse dimension\/}, Ph.D. thesis, Universit\'e Paris 6

4209:   (1999), available at {\sf

4210:   http://ipnweb.in2p3.fr/$\sim$lptms/membres/texier/research.html}.

4211: 

4212: \bibitem{Tex00}

4213: C.~Texier, Individual energy level distributions for one-dimensional diagonal

4214:   and off-diagonal disorder, J.~Phys.~A: Math. Gen. {\bf 33}, 6095--6128

4215:   (2000).

4216: 

4217: \bibitem{Tex02}

4218: C.~Texier, Scattering theory on graphs (2): the Friedel sum rule, J.~Phys.~A:

4219:   Math. Gen. {\bf 35}, 3389--3407 (2002).

4220: 

4221: \bibitem{TexBut03}

4222: C.~Texier and M.~B{\"u}ttiker, Local Friedel sum rule in graphs, Phys. Rev.~B

4223:   {\bf 67}, 245410 (2003).

4224: 

4225: \bibitem{TexCom99}

4226: C.~Texier and A.~Comtet, Universality of the Wigner time delay distribution for

4227:   one-dimensional random potentials, Phys. Rev. Lett. {\bf 82}(21), 4220--4223

4228:   (1999).

4229: 

4230: \bibitem{TexDeg03}

4231: C.~Texier and P.~Degiovanni, Charge and current distribution in graphs,

4232:   J.~Phys.~A: Math. Gen. {\bf 36}, 12425--12452 (2003).

4233: 

4234: \bibitem{TexMon01}

4235: C.~Texier and G.~Montambaux, Scattering theory on graphs, J.~Phys.~A: Math.

4236:   Gen. {\bf 34}, 10307--10326 (2001).

4237: 

4238: \bibitem{TexMon04}

4239: C.~Texier and G.~Montambaux, Weak localization in multiterminal networks of

4240:   diffusive wires, Phys. Rev. Lett. {\bf 92}, 186801 (2004).

4241: 

4242: \bibitem{TexMon05b}

4243: C.~Texier and G.~Montambaux, Dephasing due to electron-electron interaction in

4244:   a diffusive ring, unpublished  (2005), preprint cond-mat/0505199, to appear

4245:   in Phys. Rev. B {\bf72}.

4246: 

4247: \bibitem{TexMon05}

4248: C.~Texier and G.~Montambaux, Quantum oscillations in mesoscopic rings and

4249:   anomalous diffusion, J.~Phys.~A: Math. Gen. {\bf 38}, 3455--3471 (2005).

4250: 

4251: \bibitem{ThoPepAhmAndDav86}

4252: T.~J. Thornton, M.~Pepper, H.~Ahmed, D.~Andrews and G.~J. Davies,

4253:   One-Dimensional Conduction in the 2D Electron Gas of a GaAs-AlGaAs

4254:   Heterojunction, Phys. Rev. Lett. {\bf 56}(11), 1198 (1986).

4255: 

4256: \bibitem{TraWid93}

4257: C.~A. Tracy and H.~Widom, Level-spacing distribution and the Airy kernel, Phys.

4258:   Lett. B {\bf 305}, 115 (1993).

4259: 

4260: \bibitem{TraWid02}

4261: C.~A. Tracy and H.~Widom, Distribution functions for largest eigenvalues and

4262:   their applications, in {\em Proceedings of the ICM\/}, volume~1, p. 587,

4263:   Beijing (2002).

4264: 

4265: \bibitem{Ver94}

4266: J.~Verbaarschot, The spectrum of the Dirac operator near zero virtuality for

4267:   $N_c=2$ and chiral random matrix theory, Nucl. Phys.~B [FS] {\bf 426}, 559

4268:   (1994).

4269: 

4270: \bibitem{Ver79}

4271: W.~Verwaat, On a stochastic difference equation and a representation of non

4272:   negative infinitely divisible random variables, Adv. Appl. Probab. {\bf 111},

4273:   750 (1979).

4274: 

4275: \bibitem{VidMonDou00}

4276: J.~Vidal, G.~Montambaux and B.~Dou\c{c}ot, Transmission through quantum

4277:   networks, Phys. Rev.~B {\bf 62}, R16294 (2000).

4278: 

4279: \bibitem{Weg79}

4280: F.~Wegner, The mobility edge problem: continuous symmetry and a conjecture, Z.

4281:   Phys. B {\bf 35}(3), 207 (1979).

4282: 

4283: \bibitem{Weg83}

4284: F.~Wegner, Exact density of states for Lowest Landau Level in white noise

4285:   potential. Superfield representation for interacting systems, Z. Phys. B -

4286:   Condens. Matter {\bf 51}, 279 (1983).

4287: 

4288: \bibitem{Weg76}

4289: F.~J. Wegner, Electrons in disordered systems. Scaling near the mobility edge,

4290:   Z. Phys. B {\bf 25}(4), 327 (1976).

4291: 

4292: \bibitem{WeiHav86}

4293: G.~H. Weiss and S.~Havlin, Some properties of a random walk on a comb

4294:   structure, Physica A {\bf 134}, 474 (1986).

4295: 

4296: \bibitem{Wer94}

4297: W.~Werner, Sur les points autour desquels le mouvement brownien plan tourne

4298:   beaucoup, Probab. Theory Rel. {\bf 99}, 111 (1994).

4299: 

4300: \bibitem{WinRooChaPro86}

4301: S.~Wind, M.~J. Rooks, V.~Chandrasekhar and D.~E. Prober, One-Dimensional

4302:   Electron-Electron Scattering with Small Energy Transfers, Phys. Rev. Lett.

4303:   {\bf 57}(5), 633 (1986).

4304: 

4305: \bibitem{Win03}

4306: B.~Winn, {\em The Laplacian on a graph and quantum chaology\/}, Ph.D. thesis,

4307:   University of Bristol (2003), available at {\sf

4308:   http://www.math.tamu.edu/$\sim$bwinn/thesis.htm}.

4309: 

4310: \bibitem{WuKun99}

4311: F.~Y. Wu and H.~Kunz, Restricted random walks on a graph, Ann. Combin. {\bf 3},

4312:   475 (1999).

4313: 

4314: \bibitem{Yor80}

4315: M.~Yor, Loi de l'indice du lacet brownien et distribution de Hartman-Watson,

4316:   Z.~Wahrscheinlichkeit. {\bf 53}, 71 (1980).

4317: 

4318: \bibitem{Yor00}

4319: M.~Yor, {\em Exponential functionals of Brownian motion and related

4320:   processes\/}, Springer (2000).

4321: 

4322: \bibitem{Zir96}

4323: M.~R. Zirnbauer, Riemannian symmetric superspaces and their origin in

4324:   random-matrix theory, J. Math. Phys. {\bf 37}, 4986 (1996).

4325: 

4326: \end{thebibliography}

4327: 

4328: 

4329: \end{document}

4330: