1: % \title{Conformally invariant scaling limits\\{\large an overview and a collection of problems}}

2: % \author{Oded Schramm\\Microsoft Research}

3: \date{}

4: \documentclass{icmart}

5: \usepackage{psfig,amsmath,amssymb,xspace,url,picinpar,graphicx}

6: \title[Conformally invariant scaling limits]{Conformally invariant scaling limits\\{\large an overview and a collection of problems}}

7: \author[Oded Schramm]{Oded Schramm}

8: \contact{Microsoft Corporation, Redmond, Washington, USA}

9: %\usepackage{psfig,amsmath,amssymb,amsthm,xspace,url,picinpar,graphicx}

10: %\usepackage[small,compact]{titlesec}

11: \newif\ifhyper\IfFileExists{hyperref.sty}{\hypertrue}{\hyperfalse}

12: \hyperfalse

13: \hypertrue

14: \ifhyper\usepackage{hyperref}\fi

15: \newcommand{\old}[1]{}

16: \def\note#1{{\bf[ #1 ]}}

17: \renewcommand{\th}{\ensuremath{^{\text{th}}}\xspace}

18: \newcommand{\nd}{\ensuremath{^{\text{nd}}}\xspace}

19: \newcommand{\st}{\ensuremath{^{\text{st}}}\xspace}

20: \newtheorem{thm}{Theorem}[section]

21: \newtheorem{theorem}{Theorem}[section]

22: \newtheorem{lem}[theorem]{Lemma}

23: \newtheorem{lemma}[theorem]{Lemma}

24: \newtheorem{cor}[theorem]{Corollary}

25: \newtheorem{prop}[theorem]{Proposition}

26: \newtheorem{conj}{Conjecture}

27: \newtheorem{problem}[theorem]{Problem}

28: \newtheorem{definition}[theorem]{Definition}

29: \newcommand{\remark}{\medskip\noindent\textbf{Remark:\ }}

30: \newcommand{\remarkend}{\medskip}

31: \newcommand{\R}{{\mathbb R}}

32: \newcommand{\N}{{\mathbb N}}

33: \newcommand{\G}{{G}}

34: \newcommand{\E}{{\mathbb E}}

35: \newcommand{\Eng}{{\cal E}}

36: \newcommand{\C}{{\mathbb C}}

37: \def\H{\mathbb H}

38: \def\U{\mathbb U}

39: \newcommand{\Z}{{\mathbb Z}}

40: \newcommand{\M}{{\cal M}}

41: \newcommand{\T}{{\mathbb T}}

42: \newcommand{\supp}{{\operatorname{supp}}}

43: \newcommand{\sign}{{\operatorname{sign}}}

44: \newcommand{\Arg}{{\operatorname{Arg}}}

45: \newcommand{\Log}{{\operatorname{Log}}}

46: \newcommand{\Lip}{{\operatorname{Lip}}}

47: \newcommand{\diam}{{\operatorname{diam}}}

48: \newcommand{\length}{{\operatorname{length}}}

49: \newcommand{\A}{{\mathbb A}}

50: \newcommand{\I}{{\rm I}\xspace}

51: \newcommand{\II}{{\rm II}\xspace}

52: \newcommand{\eps}{\varepsilon}

53: \def\eps{\epsilon}

54: \def\SLE/{SLE}

55: \def\SLEk#1/{$\mathrm{SLE}_{#1}$}

56: \def\pa{\partial}

57: \def\closure#1{\overline{#1}}

58: \newcommand{\Cov}{{\rm Cov}}

59: \newcommand{\Var}{{\bf Var}}

60: \renewcommand{\Re}{{\rm{Re}}}

61: \renewcommand{\Im}{{\rm{Im}}}

62: \def\TG{\mathrm{TG}}

63: \def\P{\mathbf{P}}

64: \def\ev{\mathcal}

65: \def\eref#1{(\ref{#1})}

66: \def\Bb#1#2{{\def\md{\bigm| }#1\bigl[#2\bigr]}}

67: \def\BB#1#2{{\def\md{\Bigm| }#1\Bigl[#2\Bigr]}}

68: \def\Bs#1#2{{\def\md{\mid}#1[#2]}}

69: \def\Pb{\Bb\P}

70: \def\Eb{\Bb\E}

71: \def\PB{\BB\P}

72: \def\EB{\BB\E}

73: \def\Ps{\Bs\P}

74: \def\Es{\Bs\E}

75: \def\bl{\bigl}\def\br{\bigr}\def\Bl{\Bigl}\def\Br{\Bigr}

76: \def\dhrule{\bigskip\hrule\smallskip

77: \addcontentsline{toc}{section}{--Revise--}

78: \hrule\bigskip

79: }

80: 

81: %BIB

82: \def\Hryniv{MR1602044}

83: \def\HaraSladeSAW{MR1171762}

84: \def\DLAdef{MR704464}

85: \def\DLAsurvey{MR1205417}

86: \def\MakarovCarleson{MR1815718}

87: \def\KestenDLA{MR1077203}

88: \def\LanglandsEtAl{MR1230963}

89: \def\HPSdynamical{MR1465800}

90: \def\TsirelsonFourierWalsh{math.PR/9903068}

91: \def\TsirelsonScalingLimit{math.PR/9903121}

92: \def\BousquetMelouSchaeffer{math.CO/0211070}

93: \def\MarckertMokkadem{math.PR/0403398}

94: \def\ChassaingSchaeffer{MR2031225}

95: \def\AngleGrowthUIPT{MR2024412}

96: \def\MadrasSlade{MR1197356}

97: \def\KennedyVariation{math.PR/0510604}

98: \def\FourierSurveys{KalaiSafra}

99: \def\BKSnoise{MR2001m:60016}

100: \def\KPZ{MR947880}

101: % V. G. Knizhnik, A. M. Polyakov and A. B. Zamolodchikov. Fractal structure of 2D-quantum gravity. Mod. Phys. Lett. A 3 (1988), no.  8, 819–826.

102: \def\Ziff{MR1745857}

103: \def\BollobasRiordanVoronoi{math.PR/0410336}

104: \def\CFNnearcrit{cond-mat/0510740}

105: \def\LSWuptoconst{MR1901950}

106: \def\RSWboolean{MR1398054}

107: \def\RSWvacant{MR1071809}

108: \def\WernerLoops{math.PR/0511605}

109: \def\BenjaminiSchrammVoronoi{MR1646475}

110: \def\SheffieldGFFsurvey{math.PR/0312099}

111: \def\KozmaLERWspace{math.PR/0508344}

112: \def\KozmaLERWplane{math.PR/0212338}

113: \def\Watts{MR98a:82059}

114: \def\DubedatWatts{math.PR/0405074}

115: \def\DubedatDuality{MR2118865}

116: \def\TaylorSurvey{MR857718}

117: % 1986 Math Proc (Measure Theory random ...)

118: \def\NienhuisOn{NienhuisOn}

119: %B. Nienhuis. Exact critical point and exponents of the O(n) model in

120: %two dimensions. Phys. Rev. L., 49:1062--1065, 1982.

121: \def\SSexceptional{math.PR/0504586}

122: \def\FortuinKasteleyn{MR50:12107}

123: \def\ItzyksonDrouffe{MR1175176}

124: \def\BeffaraSidoravicius{math.PR/0507220}

125: \def\BeffaraSLEdim{math.PR/0211322}

126: \def\KenyonConformalDomino{MR1782431}

127: \def\BurtonPemantle{MR94m:60019}

128: \def\MeesterRoy{MR1409145}

129: \def\GrimmettPercolationEdtwo{MR1707339}

130: \def\CamiaNewman{math.PR/0504036}

131: \def\KestenICM{MR1989192}

132: \def\LevyConfInvBM{MR0029120}

133: \def\KasteleynManhattan{MR28:2859}

134: \def\DuplantierManhattan{MR89b:82025}

135: \def\FukaiUchiyama{MR97m:60098}

136: \def\CM{MR1815718}

137: \def\MakFine{MR2000g:30018}

138: \def\JonMak{MR96k:30027}

139: \def\DurStoch{MR97k:60148}

140: \def\SmirnovPerc{MR1851632}

141: \def\Dup{MR2001c:82040}

142: \def\DupSal{MR88d:82073}

143: \def\WernerSurvey{math.PR/0007042}

144: \def\Dubedat{DubedatREF}

145: \def\SLEsurveys{listofslesurveypapers}

146: \def\higherFunc{MR15:419i}

147: \def\CardyPercolationSurvey{math-ph/0103018}

148: \def\CardySLESurvey{MR2148644}

149: \def\LWuniversality{MR2002g:60123}

150: \def\LSWsaw{MR2112127}

151: \def\LSWi{MR2002m:60159a}

152: \def\LSWii{MR2002m:60159b}

153: \def\LSWiii{MR1899232}

154: \def\LSWan{MR1992830}

155: \def\LSWlesl{math.PR/0112234}

156: \def\LSWlesl{MR2044671}

157: \def\LSWrestriction{MR1992830}

158: \def\LSWonearm{MR2002k:60204}

159: \def\SWpercexpo{math.PR/0109120}

160: \def\KenyonDD{MR99m:52026}

161: \def\SchSLE{MR1776084}

162: \def\SchTEL{MR96k:30009}

163: \def\FHM{MR90a:30017}

164: \def\Ahlfors{MR50:10211}

165: \def\KR{MR98k:28004}

166: \def\CardyFormula{MR92m:82048}

167: \def\RHAdd{MR98e:82018}

168: \def\AizenmanBurchard{MR2000i:60012}

169: \def\ABNW{MR2001c:60151}

170: \def\Dudley{MR91g:60001}

171: \def\DurrettProb{MR98m:60001}

172: \def\RSsle{MR2153402}

173: \def\Lbook{MR92f:60122}

174: \def\LV{MR49:9202}

175: \def\Llerw{MR2000k:60092}

176: \def\LawlerLERWdef{MR81j:60081}

177: \def\LawlerWernerUniversality{MR2002g:60123}

178: \def\LawlerWernerBMexponents{MR2000k:60165}

179: \def\LWi{\LawlerWernerBMexponents}

180: \def\LWii{\LawlerWernerUniversality}

181: \def\LawlerSLbook{MR2129588}

182: %\def\LSWoneArm{math.PR/0108211}

183: \def\HaggstromUSTlim{MR97b:60170}

184: \def\PemantleUST{MR92g:60014}

185: \def\SchPercForm{math.PR/0107096}

186: \def\WilsonAlg{MR1427525}

187: \def\LyonsUSFsurvey{MR99e:60029}

188: \def\KenyonLERW{MR1819995}

189: \def\KenyonUST{MR1757962}

190: \def\PommeBDRY{MR95b:30008}

191: \def\Dubins{MR38:2837}

192: \def\Strassen{MR35:4969}

193: \def\RevuzYor{MR92d:60053}

194: \def\BSsqharm{MR98d:60134}

195: \def\BSST{MR2:153d}

196: \def\KaratsasShreve{MR89c:60096}

197: \def\Collatz{MR22:322}

198: \def\LawlerSLEintro{LawlerSLEintro}

199: \def\LawlerStrict{MR1645225}

200: \def\LawlerFrontier{MR97g:60110}

201: \def\SchrammSheffieldHE{MR2184093}

202: \def\SchrammSheffieldDGFF{DGFF}

203: \def\SchrammWilsonCoord{math.PR/0505368}

204: \def\multiplyConnectedSLE{\DZhan,MR2111355,math.PR/0412060,math.PR/0503178}

205: \def\DZhan{MR2128237}

206: %\def\BauerFriedrichSLEmulti{MR2111355,math.PR/0412060,math.PR/0503178}

207: \def\WernerICM{WernerICM}

208: \def\LWsoup{MR2045953}

209: \def\WernerBMclusters{MR2023758}%{math.PR/0308164}

210: \def\SladeKestenFest{MR1703123}

211: \def\SladeSAWsurvey{MR1251665}

212: \def\BrydgesImbrie{MR2031859}

213: \def\KenyonDominoGFF{MR1872739}

214: \def\KestenPc{MR82c:60179}

215: \def\HarrisPc{MR0115221}

216: \def\WiermanPc{MR612205}

217: \def\RussoPercolation{MR0488383}

218: \def\ConiglioPotts{1989PhRvL..62.3054C}

219: \def\DupSalWinding{1988PhRvL..60.2343D}

220: \def\DupSalDenseSAW{1987NuPhB.290..291D}

221: \def\DupSalPercolationhull{1987PhRvL..58.2325S}

222: \def\Kadanoff{1978JPhA...11.1399K}

223: \def\WernerRestrictionSurvey{MR2178043}

224: \def\WernerStFlour{MR2079672}

225: \def\KondevHenleyninetyfive{1995PhRvL..74.4580K}

226: \def\KondevHenleytwothousand{2000PhRvE..61..104K}

227: \def\HuberKondevGeometryofloops{2001APS..DCM.Q2008H}

228: \def\HuberDuKondevGeometryofloops{2002APS..MAR.U4003K}

229: \def\Nienhuiseightytwo{1982PhRvL..49.1062N}

230: \def\SeymourWelsh{MR58:13410}

231: \def\KostThouTwo{1972JPhC....5L.124K}

232: \def\KostThouThree{1973JPhC....6.1181K}

233: \def\KostRoughening{1977JPhC...10.3753K}

234: \def\Foltin{2001JPhA...34.5327F}

235: \def\FroSpenRotator{1981PhRvL..46.1006F}

236: \def\FroSpenCoulombGasStats{MR610687}

237: \def\FroSpenAbelianSpin{MR634447}

238: \def\NaddafSpencer{MR1461951}

239: \def\KenyonDominoGFF{MR1872739}

240: \def\Mandelbrot{1967Sci...156..636M}

241: \def\IsichenkoSurvey{1992RvMP...64..961I}

242: \def\SpencerSurvey{MR1460292}

243: \def\BriMelFroSurvey{MR833220}

244: \def\DuplantierKwon{DuplantierKwon}

245: \def\GiacominSurvey{MR1919512}

246: \def\DuplantierSurvey{1989PhR...184..229D}

247: \def\KagerNienhuisSurvey{MR2065722}

248: \def\NienhuisSurvey{MR751711}

249: \def\CardySummerSchool{cardysummer}

250: \def\GFFSurvey{math.PR/0312099}

251: \def\Janson{MR1474726}

252: \def\GlimmJaffe{MR887102}

253: \def\GawcedzkiCFTStrings{MR1701610}

254: \def\diFrancescoetalCFT{MR1424041}

255: \def\Pommerenke{MR?}

256: 

257: \def\figdir{./}

258: \def\noopsort#1{}

259: 

260: 

261: \newcommand{\vs}{\vspace{.15 in}}

262: 

263: %\newenvironment{proof}{

264: %    \noindent\textbf{Proof:} \hspace*{1em}}{

265: %    \hspace*{\fill} $\square$\medskip }

266: 

267: \newcommand{\PSbox}[3]{\mbox{\rule{0in}{#3}\hspace{#2}\special{psfile=#1}}}

268: % \pagestyle{myheadings} \markright{Conformally invariant scaling limits

269:   % \hfil \rcsDate

270:   % \hfil Oded Schramm}

271: \begin{document}

272: %\maketitle

273: 

274: \begin{abstract}

275: Many mathematical models of statistical physics in two dimensions are

276: either known or conjectured to exhibit conformal invariance.

277: Over the years, physicists proposed predictions of various

278: exponents describing the behavior of these models.

279: Only recently have some of these predictions become accessible to

280: mathematical proof. One of the new developments is the discovery

281: of a one-parameter family of random curves

282: called Stochastic Loewner evolution

283: or \SLE/. The \SLE/ curves appear as limits of interfaces or paths

284: occurring in a variety of statistical physics models as the

285: mesh of the grid on which the model is defined tends to zero.

286: 

287: The main purpose of this article is to list a collection of

288: open problems.  Some of the open problems indicate aspects of the

289: physics knowledge that have not yet been understood mathematically.

290: Other problems are questions about the nature of the \SLE/ curves

291: themselves. 

292: Before we present the open problems, the definition of \SLE/

293: will be motivated and explained, and a brief sketch

294: of recent results will be presented.

295: \end{abstract}

296: 

297: \begin{classification}

298: Primary 60K35; Secondary 82B20, 82B43, 30C35.

299: \end{classification}

300: 

301: \begin{keywords}

302: Statistical physics, conformal invariance, stochastic Loewner evolutions, percolation.

303: \end{keywords}

304: 

305: % \tableofcontents

306: \maketitle

307: 

308: \section{Introduction}

309: In the past several years, many predictions from physics regarding the large-scale

310: behavior of random systems defined on a lattice in two dimensions have become

311: accessible to mathematical study and proof. Of central importance is the 

312: asymptotic conformal invariance of these systems. 

313: It turns out that paths associated with these random configurations

314: often fall into a one-parameter family of conformally invariant random curves called

315: stochastic Loewner evolutions, or \SLE/.

316: We start by motivating \SLE/ through a simple mathematical model of percolation.

317: After giving the definition of \SLE/, we present a narrative of recent developments.

318: However, since there are good surveys on the subject

319: in the literature~\cite{\WernerStFlour,\KagerNienhuisSurvey,\CardySLESurvey,%

320: \LawlerSLbook,\WernerRestrictionSurvey}, this introductory

321: part of the paper will be short and cursory.

322: The rest of the paper will consist of an annotated list of open problems in the subject.

323: 

324: \subsection{Motivation and definition of SLE}

325: 

326: To motivate SLE, we now discuss percolation.

327: More specifically, we define one particular model of percolation

328: in two dimensions.

329: Fix a number $p\in[0,1]$. Let $\omega$ be a random subset

330: of the set of vertices in the triangular grid $\TG$, where 

331: for vertices $v\in V(\TG)$ the events $v\in\omega$ are independent

332: and have probability $p$.

333: In percolation theory one studies the connected

334: components (a.k.a.\ clusters) of the random subgraph of $\TG$ whose vertex

335: set is $\omega$ and whose edges are the edges in $\TG$

336: connecting two elements of $\omega$. Equivalently,

337: one may study the connected components of the set of white

338: hexagons in Figure~\ref{f.perc}, where each hexagon

339: in the hexagonal grid dual to $\TG$ represents

340: a vertex of $\TG$ and hexagons corresponding to vertices

341: in $\omega$ are colored white. The reasons for considering this

342: dual representation are that the figures come out nicer and

343: that it makes some important definitions more concise.

344: 

345: \begin{figure}

346: \centerline{\psfig{file=\figdir/hexPerc.ps,height=3in}}

347: \caption{\label{f.perc}Site percolation on the triangular grid 

348: as represented by colored hexagons.}

349: \end{figure}

350: 

351: The above percolation model is site (or vertex) percolation on the

352: triangular grid. Likewise, there is a bond (or edge) model,

353: where one considers a random subgraph of a grid whose

354: vertex set is the set of all vertices of the grid, but

355: where each edge of the grid is in the percolation subgraph

356: with probability $p$, independently.

357: Additionally, there are various percolation models which

358: are not based on a lattice. Some of these will be discussed

359: in later sections.

360: 

361: There is an important value $p_c$ of the parameter $p$, which

362: is the threshold for the existence of an unbounded cluster

363: and is called the critical value of $p$.

364: The actual value of $p_c$ varies depending on the particular

365: percolation model. For site percolation on the triangular

366: grid, as well as for bond percolation on the square grid,

367: we have $p_c=1/2$. This is a theorem of Kesten~\cite{\KestenPc},

368: based on earlier work by Harris~\cite{\HarrisPc}, Russo~\cite{\RussoPercolation}

369: and Seymour and Welsh~\cite{\SeymourWelsh}. The underlying reason for this

370: nice value of $p_c$ is a duality which these two models have,

371: though the precise form of the duality they exhibit is different.

372: For bond percolation on the triangular grid $p_c=2\,\sin(\pi/18)$~\cite{\WiermanPc}, while

373: for site percolation on the square grid there is not even a prediction

374: for the value of $p_c$, though rigorous and experimental estimates exist.

375: As $p$ increases beyond $p_c$, the large scale behavior of 

376: percolation undergoes a rapid change.

377: This is perhaps the mathematically simplest model of a phase transition.

378: From now on, we will focus our attention on

379: critical percolation, that is, percolation with $p=p_c$, which is in

380: many ways the most interesting value of $p$.

381: 

382: We now define and discuss the percolation interface curve indicated in

383: Figure~\ref{f.domperc}.

384: Consider a bounded domain $D$ in the plane $\R^2=\C$ whose boundary is a 

385: simple closed curve. Let $\pa_+\subset\pa D$ be a proper arc on the boundary of $D$.

386: Given $\eps>0$ we may consider the collection of hexagons in a hexagonal grid of

387: mesh $\eps$ which intersect $\closure D$.

388: Each of these hexagons which meets $\pa_+$ we color white,

389: each of the hexagons which meet $\pa D$ but not $\pa_+$ we color black,

390: and each of the hexagons contained in $D$ we color white or black with

391: probability $1/2$, independently.

392: In addition to white clusters (connected components of white hexagons)

393: sometimes, black clusters are also  considered.

394: Percolation theory is the study of connected components of random sets,

395: such as these clusters.

396: 

397: \begin{figure}

398: \centerline{\psfig{file=\figdir/percdomain.eps,height=3in}}

399: \caption{\label{f.domperc}The interface associated with percolation.}

400: \end{figure}

401: 

402: For simplicity, we assume that $\pa D$ is sufficiently smooth and

403: $\eps$ is sufficiently small so that the union of hexagons

404: intersecting $\pa D$ but not $\pa_+$ is connected.

405: There is a unique (random) path $\beta$,

406: which is the common boundary of the white cluster 

407: meeting $\pa_+$ and the black cluster meeting 

408: $\pa D$.

409: 

410: The law $\mu_{D,\pa_+,\eps}$ of $\beta$ is a probability measure on the

411: space of closed subsets of $\closure D$ with the Hausdorff metric.

412: Smirnov~\cite{\SmirnovPerc} proved that as $\eps\searrow0$ the 

413: measure $\mu_{D,\pa_+,\eps}$ converges weakly to a measure

414: $\mu_{D,\pa_+}$, and that $\mu$ is conformally invariant,

415: in the following sense.

416: If $f:\closure D\to \closure{D'}\subset\R^2$ is a homeomorphism

417: that is analytic in $D$, then the push forward of $\mu_{D,\pa_+}$ under

418: $f$ is $\mu_{f(D),f(\pa_+)}$.

419: In other words, if $\eps$ is small, then $f(\beta)$ is a good approximation

420: for the corresponding path defined using a hexagonal grid of mesh $\eps$

421: in $f(D)=D'$.

422: This type of conformal invariance was believed to hold for many

423: \lq\lq critical\rq\rq\ random systems in two dimensions.

424: However, the only previous result establishing conformal invariance

425: for a random scaling limit is L\'evy's theorem~\cite{\LevyConfInvBM}

426: stating that for two-dimensional Brownian motion,

427: the scaling limit of simple random walk on $\Z^2$, is conformally invariant

428: up to a time-change.

429: 

430: Smirnov's proof is very beautiful, and the result is

431: important, but describing the proof will throw

432: us too far off course (since for this paper percolation is

433: just an example model, not the primary topic).

434: The interested reader is encouraged to consult~\cite{\BeffaraSidoravicius,%

435: \GrimmettPercolationEdtwo,\MeesterRoy,\KestenICM}

436: for background in percolation and highlights of percolation theory.

437: An elegant simplification of parts of Smirnov's proof has been

438: discovered by Vincent Beffara~\cite{BeffaraSmirnovThm}.

439: A more detailed version of other parts of Smirnov's argument appears

440: in~\cite{\CamiaNewman}.

441: 

442: Though this was not the original inspiration, 

443: we will now use Smirnov's result to motivate the definition of SLE.

444: By conformal invariance, we may venture to understand $\beta$

445: in the domain of our choice. The simplest situation turns out

446: to be when $D=\H$ is the upper half plane and $\pa_+$ is the

447: positive real ray $\R_+$, as in Figure~\ref{f.pcurve}.

448: (Though this domain $D$ is unbounded, that does not cause any problems.)

449: We may consider the discrete path

450: $\beta$ as a simple path $\beta:[0,T)\to\closure \H$ starting near $0$ and

451: satisfying $\lim_{t\to T}|\beta(t)|=\infty$ (where $T$ is finite or infinite).

452: 

453: \begin{figure}

454: \centerline{\psfig{file=\figdir/pcurve.ps,height=3in}}

455: \caption{\label{f.pcurve}The percolation interface in the upper half plane.}

456: \end{figure}

457: 

458: We would like to learn about $\beta$ by understanding the 

459: one-parameter family of conformal maps $g_t$ mapping

460: $\H\setminus \beta([0,t])$ onto $\H$.

461: To facilitate this, we must first recall a few basic facts and

462: discuss Loewner's theorem.

463: At this point, assume only that $\beta$ is a simple path

464: in $\closure{\H}$ with $\beta(0)\in\R$

465: and $\beta(t)\notin\R$ for $t>0$.

466: The existence of the conformal maps $g_t:\H\setminus\beta([0,t])\to\H$

467: is guaranteed by Riemann's mapping theorem.

468: However, $g_t$ is not unique.  In order to choose a specific

469: $g_t$ for every $t$, we first require that $g_t(\infty)=\infty$.

470: Schwarz reflection in the real axis implies that $g_t$ is analytic

471: in a neighborhood of $\infty$, and therefore admits a power series

472: representation in $1/z$, 

473: $$

474: g_t(z)=a_1\,z+a_0+a_{-1}\,z^{-1}+a_{-2}\,z^{-2}+\cdots,

475: $$

476: valid for all $z$ sufficiently large. Since $g_t$ maps the real line

477: near $\infty$ into the real line, it follows that

478: $a_j\in\R$ for all $j$, and because $g_t:\H\setminus\beta([0,t])\to\H$,

479: we find that $a_1>0$. We now pick a specific $g_t$ by imposing the

480: so-called hydrodynamic normalization at $\infty$, namely $a_1=1$ and

481: $a_0=0$. This can clearly be achieved  by post-composing

482: with a map of the form $z\mapsto a\,z+b$, $a>0$, $b\in\R$.

483: 

484: The coefficients $a_j$ of the series expansion of $g_t$ are now functions of $t$.

485: It is not hard to verify that $a_{-1}(t)$ is a continuous, strictly increasing function

486: of $t$. Clearly, $g_0(z)=z$ and hence $a_{-1}(0)=0$.

487: We may therefore reparametrize $\beta$ so as to have

488: $a_{-1}(t)=2\,t$ for all $t>0$. 

489: This is called the half-plane capacity parametrization of $\beta$.

490: With this parametrization, a variant of Loewner's theorem~\cite{Loewner} states

491: that the maps $g_t$ satisfy the differential equation

492: \begin{equation}

493: \label{e.chordal}

494: \frac{dg_t(z)}{dt}= \frac{2}{g_t(z)-W(t)}\,,

495: \end{equation}

496: where $W(t):=g_t\bl(\beta(t)\br)$

497: is called the Loewner driving term. A few comments are in order.

498: \begin{enumerate}

499: \item

500: Although $g_t$ is defined in $\H\setminus \beta([0,t])$,

501: it does extend continuously to $\beta(t)$, and therefore

502: $W(t)$ is well defined. 

503: \item

504: If $z=\beta(s)$ for some $s$,

505: then~\eref{e.chordal} makes sense only as long as $t<s$.

506: That is to be expected. The domain of definition of $g_t$ is shrinking as $t$ increases.

507: A point $z$ falls out of the domain of $g_t$ at the first time $\tau=\tau_z$ such that

508: $\liminf_{t\nearrow\tau} g_t(z)-W(t)=0$.

509: \item The main point here is that information about the path $\beta$ is encoded

510:  in $W(t)$, which is a path in $\R$.

511: \item The proof of~\eref{e.chordal} is not too hard. In~\cite[Theorem 2.6]{\LSWi} 

512: a proof (of a generalization) may be found.

513: \end{enumerate}

514:   

515: We now return to the situation where $\beta$ is the percolation interface chosen according

516: to $\mu_{\H,\R_+,\eps}$, parametrized by half-plane capacity.

517: It is easy to see that in this case $T=\infty$.

518: Fix some $s>0$. Suppose that we examine the colors of only those hexagons that

519: are necessary to determine $\beta([0,s])$. This can be done by sequentially

520: testing the hexagons adjacent to $\beta$ starting from $\beta(0)$

521: as follows.  Each time the already determined arc of $\beta$ meets a hexagon whose color

522: has not yet been examined, we test the color (which permits us to extend the

523: determined initial arc of $\beta$ by at least one segment), until

524: $\beta([0,s])$ has been determined. See Figure~\ref{f.pcurveini}.

525: 

526: \begin{figure}

527: \centerline{\psfig{file=\figdir/pcurveini1.eps,width=4in}}

528: \caption{\label{f.pcurveini}Initial segment of interface.}

529: \end{figure}

530: 

531: Now comes the main point. Let $D_s$ be the unbounded component

532: of the collection of hexagons of undetermined color in $\H$, and let

533: $\pa_+^s$ be the subset of $\pa D_s$ lying on the boundary

534: of hexagons of determined color white.

535: Then the distribution of the continuation $\beta([s,\infty))$ of the interface

536: given $\beta([0,s])$ is $\mu_{D_s,\pa_+^s,\eps}$.

537: By Smirnov's theorem, if $G:D_s\to\H$ is the conformal map 

538: satisfying the hydrodynamic normalization, then the image under

539: $G$ of $\mu_{D_s,\pa_+^s,\eps}$ is close to $\mu_{\H,G(\pa_+^s),\eps}$.

540: (Actually, to justify this, one needs a slightly stronger \lq\lq uniform\rq\rq\

541: version of Smirnov's theorem. But here we want to convey the main ideas, and

542: do not bother about being entirely precise.)

543: Now, since $D_s$ approximates $\H\setminus \beta([0,s])$,

544: it follows that $G$ is very close to $g_s$ and $G(\pa_+^s)$ is close

545: to $[W(s),\infty)=[g_s(\beta(s)),\infty)$. Therefore, in the limit as $\eps\searrow 0$, we have

546: for $\beta$ sampled according to $\mu_{\H,\R_+}$ that given

547: $\beta([0,s])$ the distribution of $ g_s\circ \beta([s,\infty))$

548: (which is the conformal image of the continuation of the path)

549: is $\mu_{\H,\R_+}$ translated by $W(s)$.

550: 

551: The Loewner driving term of the path 

552: $t\mapsto g_s\circ\beta(s+t)$ is $W(s+t)$,

553: because $g_{s+t}\circ g_s^{-1}$ maps

554: $\H\setminus g_s\bl(\beta([s,s+t])\br)$ onto $\H$.

555: The conclusion of the previous paragraph therefore implies that given $(W(t):t\in[0,s])$

556: the distribution of the continuation of $W$ is identical to the

557: original distribution of $W$ translated to start at $W(s)$.

558: This is a very strong property. 

559: Indeed, for every $n\in\N$ and $t>0$ we

560: may write $W(t)=\sum_{j=1}^n \bl(W(j\,t/n)-W((j-1)t/n)\br)$,

561: which by the above is a sum of $n$ independent identically distributed

562: random variables.

563: If we assume that the variance of $W(t)$ is finite,

564: then it is also the sum of the variances of the summands.

565: By the central limit theorem, $W(t)$ is therefore a Gaussian

566: random variable.

567: By symmetry, $W(t)$ has the same distribution as $-W(t)$,

568: and so $W(t)$ is a centered Gaussian. It now easily follows

569: that there is some constant $\kappa\ge 0$ such that

570: $W(t)$ has the same distribution as $B(\kappa\,t)$,

571: where $B$ is one-dimensional Brownian motion starting at $B(0)=0$.

572: Using results from the theory of stochastic processes

573: (e.g., the characterization of continuous martingales as time-changed

574: Brownian motion), the same conclusion can be reached while replacing

575: the assumption that $W(t)$ has finite variance with

576: the continuity of $W(t)$ in $t$.

577: 

578: \bigskip

579: 

580: We have just seen that Smirnov's theorem implies that

581: the Loewner driving term of a sample from $\mu_{\H,\R_+}$

582: is $B(\kappa\,t)$ for some $\kappa\ge 0$. This should serve

583: as adequate motivation for the following definition from~\cite{\SchSLE}.

584: 

585: \begin{definition}

586: Fix some $\kappa\ge 0$, and let $g_t$ be the solution of

587: Loewner's equation~\eref{e.chordal}

588: satisfying $g_0(z)=z$ with $W(t)=B(\kappa\,t)$, where $B$

589: is standard one-dimensional Brownian motion starting at $B(0)=0$.

590: Then $(g_t:t\ge 0)$ is called {\bf chordal Stochastic Loewner evolution}

591: with parameter $\kappa$ or \SLEk\kappa/.

592: \end{definition}

593: 

594: Of course, \SLEk\kappa/ is a random one-parameter family of maps;

595: the randomness is entirely due to the Brownian motion.

596: 

597: It has been proven~\cite{\RSsle,\LSWlesl} that with probability $1$

598: there is a (unique) random continuous path $\gamma(t)$ such that

599: for each $t\ge 0$ the domain of definition of $g_t$ is

600: the unbounded component of $\H\setminus\gamma([0,t])$.

601: The path is given by $\gamma(t)=g_t^{-1}(W(t))$, but

602: proving that $g_t^{-1}(W(t))$ is well defined is not easy.

603: It is also known~\cite{\RSsle} that a.s.\ $\gamma(t)$ is a simple

604: path if and only if $\kappa\le 4$ and is space-filling if and only

605: if $\kappa\ge 8$.

606: Sometimes the path $\gamma$ itself is called \SLEk\kappa/.

607: This is not too inconsistent, because $g_t$ can be reconstructed

608: from $\gamma([0,t])$ and vice versa.

609: 

610: If $D$ is a simply connected domain in the plane and $a,b\in\pa D$ are

611: two distinct points (or rather prime ends), then chordal \SLE/ from

612: $a$ to $b$ in $D$ is defined as the image of $\gamma$ under a conformal

613: map from $\H$ to $D$ taking $0$ to $a$ and $\infty$ to $b$.

614: Though the map is not unique, the choice of the map does not

615: effect the law of the \SLE/ in $D$. This follows from the easily

616: verified fact that up to a

617: rescaling of time, the law of the \SLE/ path is invariant

618: under scaling by a positive real constant, as is the case for Brownian

619: motion.

620: 

621: The reason for calling the \SLE/ \lq\lq chordal\rq\rq\ is

622: that it connects two boundary points of a domain $D$.

623: There is another version of \SLE/, which connects

624: a boundary point to an interior point, called {\bf radial}

625: \SLE/. Actually, there are a few other variations, but they

626: all have similar definitions and analogous properties.

627: 

628: \subsection{A historical narrative}

629: 

630: In this subsection we list some works and discoveries related to \SLE/

631: and random scaling limits in two dimensions.

632: The following account is not comprehensive. Some of the topics not covered

633: here are discussed in Wendelin Werner's~\cite{\WernerICM}

634: contribution to this ICM proceedings.

635: We start by very briefly discussing the historical background.

636: 

637: 

638: \bigskip

639: 

640: In the survey paper~\cite{\LanglandsEtAl}, Langlands, Pouliot and Saint-Aubin

641: present a collection of intriguing predictions from statistical physics.

642: They have discussed these predictions and some simulation

643: data with Aizenman, which prompted him

644: to conjecture that the critical percolation

645: crossing probabilities are asymptotically conformally invariant (see \cite{\LanglandsEtAl}).

646: This means that the probability $Q_\eps(D,\pa_1,\pa_2)$ that there

647: exists a critical percolation cluster in a domain $D\subset\C$ connecting two boundary arcs

648: $\pa_1$ and $\pa_2$ on a lattice with mesh $\eps$ has a limit

649: $Q(D,\pa_1,\pa_2)$ as $\eps\searrow 0$ and that the limit

650: is conformally invariant, namely,

651: $Q(D,\pa_1,\pa_2)=Q\bl(f(D),f(\pa_1),f(\pa_2)\br)$

652: if $f$ is a homeomorphism from $\closure D$ to $f(\closure {D})$

653: that is conformal in $D$.

654: This led John Cardy~\cite{\CardyFormula} to propose his formula (involving hypergeometric

655: functions) for the asymptotic crossing probability in a rectangle between two opposite edges.

656: The survey~\cite{\LanglandsEtAl} highlighted these predictions

657: and the role of the conjectured conformal invariance

658: in critical percolation, as well as several other statistical

659: physics models in two dimensions.

660: 

661: Prior to \SLE/ there were attempts to use compositions

662: of conformal slit mappings and even Loewner's equation

663: in the study of diffusion limited aggregation 

664: (DLA).  DLA is a random growth process, which

665: produces a random fractal and is notoriously

666: hard to analyse mathematically.

667: (See~\cite{\DLAdef,\DLAsurvey} for a definition and discussion of DLA.)

668: Makarov and Carleson~\cite{\MakarovCarleson} used Loewner's equation

669: to study a much simplified deterministic variant of DLA, which is

670: not fractal, and Hastings and Levitov~\cite{HastingsLevitov}

671: have used conformal mapping techniques for a non-rigorous

672: study of more realistic versions of DLA.

673: Given that the fractals produced by DLA are not conformally

674: invariant, it is not too surprising that

675: it is hard to faithfully model DLA using conformal maps.

676: Harry Kesten~\cite{\KestenDLA} proved that 

677: the diameter of the planar DLA cluster after $n$ steps grows asymptotically

678: no faster than $n^{2/3}$,

679: and this appears to be essentially the only theorem concerning

680: two-dimensional DLA, 

681: though several very simplified variants of DLA have been successfully analysed.

682: 

683: \bigskip

684: 

685: The original motivation for \SLE/ actually came from

686: investigating the Loop-erased random walk (a.k.a.\ LERW),

687: which is a random curve introduced by Greg Lawler~\cite{\LawlerLERWdef}.

688: Consider some bounded simply-connected domain $D$ in the plane.

689: Let $G=G(D,\eps)$ be the subgraph of 

690: a square grid of mesh $\eps$ that falls inside $D$ and let $V_\pa$ be the

691: set of vertices of $G$ that have fewer than $4$ neighbors in $D$.

692: Suppose that $0\in D$, and let $o$ be some vertex of $G$ closest to $0$.

693: Start a simple random walk on $G$ from $o$ (at each step the walk jumps to

694: any neighbor of the current position with equal probability). We keep

695: track of the trajectory of the walk at each step, except that every time

696: a loop is created, it is erased from the trajectory. The walk terminates

697: when it first reaches $V_\pa$, and the loop-erased random walk from $o$

698: to $V_\pa$ is the final trajectory.

699: See Figure~\ref{f.lerw}, where $D$ is a disk.

700: 

701: \begin{figure}

702: \centerline{\hskip3.5in\psfig{file=\figdir/lerw.ps,height=2.5in}}

703: \caption{\label{f.lerw}The LERW in a disk.}

704: \end{figure}

705: 

706: The LERW is intimately related to the uniform spanning tree.

707: In particular, if we collapse $V_\pa$ to a single vertex $v_\pa$

708: and take a random spanning tree of the resulting projection of $G$,

709: where each possible spanning tree is chosen with equal probability,

710: then the unique path in the tree joining $o$ to $v_\pa$ (as a set

711: of edges) has precisely the same law as the LERW

712: from $o$ to $V_\pa$~\cite{\PemantleUST}.

713:  This is not a particular property of the square

714: grid, the corresponding analog holds in an arbitrary finite graph.

715: In the other direction, there is a marvelous algorithm discovered

716: by David Wilson~\cite{\WilsonAlg} which builds the uniform spanning

717: tree by successively adding loop-erased random walks.

718: The survey~\cite{\LyonsUSFsurvey} is a good window into the

719: beautiful theory of uniform spanning trees and forests.

720: 

721: Using sophisticated determinant calculations

722: and Temperley's bijection between the collection of spanning trees and

723: a certain collection of dimer tilings (which is special to the

724: planar setting),

725: Richard Kenyon~\cite{\KenyonLERW,\KenyonUST,\KenyonConformalDomino}

726: was able to calculate several properties of the LERW.

727: For example, it was shown that the variance of the winding number

728: of the above LERW in $G(D,\eps)$ is $(2+o(1))\,\log(1/\eps)$

729: as $\eps\searrow0$, and that the growth exponent for the number

730: of edges in a LERW is $5/4$.

731: 

732: \medskip

733: In~\cite{\SchSLE} it was shown that {\it if\/} the limit of the law of

734: the LERW as $\eps\searrow0$ exists and is conformally invariant,

735: then it is a radial \SLEk2/ path, when parametrized

736: by capacity.

737: (See subsection~\ref{s.notions} for a description of two alternative

738: topologies on spaces of probability measures on curves,

739: for which this convergence may be stated.)

740: In broad strokes, the reason why it should be

741: an \SLE/ path is basically the same as the

742: argument presented above for the percolation interface.

743: Two important properties of the percolation interface

744: scaling limit were crucial in the above argument:

745: conformal invariance and the following Markovian property.

746: If we condition on an initial segment of the path,

747: the remainder is an instance of the path in the 

748: domain slitted by the initial segment starting from

749: the endpoint  of the initial segment.

750: Conformal invariance was believed to hold for the LERW

751: scaling limit, while the Markovian property

752: does hold for the reversal of the LERW.

753: 

754: The identification of the

755: correct value of the parameter $\kappa$ as $2$ is based on 

756: Kenyon's calculated LERW winding variance growth rate and

757: a calculation of the variance of the winding number

758: of the radial \SLEk\kappa/ path truncated at distance

759: $\eps$ from the interior target point.

760: The latter grows like $(\kappa+o(1))\,\log(1/\eps)$.

761: 

762: It was also conjectured in~\cite{\SchSLE} that the percolation

763: interface discussed above converges to \SLEk6/.

764: The identification of the parameter $\kappa$ as $6$ in this case

765: was based on Cardy's formula~\cite{\CardyFormula}

766: and the verification that the corresponding

767: formula holds for \SLEk\kappa/ if and only if $\kappa=6$.

768: 

769: The percolation interface satisfies the following locality property.

770: The evolution of the path (given its past) does not depend on the

771: shape of the domain away from the current location of the endpoint

772: of the path. Though this is essentially obvious, it should be noted

773: that other interesting paths (such as the LERW scaling limit) do

774: not satisfy locality. 

775: 

776: Another process that clearly satisfies locality is Brownian motion.

777: Greg Lawler and Wendelin Werner~\cite{\LWi,\LWii} studied the intersection

778: exponents of planar Brownian motion and the relations between them.

779: An example of an intersection exponent is the unique number

780: $\xi(1,1)$ such that the probability that the paths of two independent

781: Brownian motions started at distance $1$ apart within the unit

782: disk $\U$ and stopped when they first hit the circle $R\,\pa\U$

783: do not intersect one another is $R^{-\xi(1,1)+o(1)}$ as $R\to\infty$.

784: At the time, there were conjectures~\cite{\DuplantierKwon} for the

785: values of many of these exponents, which were rational numbers,

786: but only two of these could be proved rigorously (not accidentally,

787: those had values $1$ and $2$).

788: These exponents encode many fundamental properties of Brownian motion.

789: For example, Lawler~\cite{\LawlerFrontier} showed that the dimension of the outer

790: boundary of planar Brownian motion stopped at time $t=1$, say,

791: is $2\,(1-\alpha)$ for a certain intersection exponent $\alpha$.

792: Lawler and Werner~\cite{\LWi,\LWii} have proved certain relations

793: between the intersection exponents, and have shown that any process

794: which like Brownian motion satisfies conformal invariance and

795: a certain version of the locality property necessarily has

796: intersection exponents that are very simply related to the 

797: Brownian exponents.

798: 

799: Since \SLEk6/ was believed to be the scaling limit

800: of the percolation interface, it should satisfy locality.

801: It is also conformally invariant by definition.

802: Thus, the Brownian exponents should apply to \SLEk6/.

803: Indeed, in a series of papers~\cite{\LSWi,\LSWii,\LSWiii} Lawler,

804: Werner and the present author proved the conjectured values of

805: the Brownian exponents by calculating the corresponding exponents

806: for \SLEk6/ (and using the previous work by Lawler and Werner).

807: Very roughly, one can say that the reason why the exponents

808: of \SLE/ are easier to calculate than the Brownian exponents

809: is that the \SLE/ path, though it may hit itself,

810: does not cross itself. Thus, the outer boundary of the

811: \SLE/ path is drawn essentially in chronological order.

812: 

813: Later~\cite{\LSWrestriction} it became clear that the relation between 

814: \SLEk6/ and Brownian motion is even closer than previously

815: apparent: the outer boundary of Brownian motion started from

816: $0$ and stopped on hitting the unit circle $\pa\U$ has the

817: same distribution as the outer boundary of a variant of

818: \SLEk6/.

819: 

820: Lennart Carleson observed that, assuming conformal invariance, Cardy's formula 

821: is equivalent to the statement that $Q(D,\pa_1,\pa_2)=\length(\pa_2)$ when $D$ is an

822: equilateral triangle of sidelength $1$, $\pa_1$ is its base, and

823: $\pa_2\subset\pa D$ is a line segment having the vertex

824: opposite to $\pa_1$ as one of its endpoints.

825: Smirnov~\cite{\SmirnovPerc} proved Carleson's form of

826: Cardy's formula for

827: critical site percolation on the triangular lattice

828: (that is, the same percolation model we have described

829: above) and showed that crossing probabilities between 

830: two arcs on the boundary of a simply connected domain are

831: asymptotically conformally invariant. 

832: As a corollary, Smirnov concluded

833: that the scaling limit of the percolation

834: interface exists and is equal to the \SLEk6/ path.

835: This connection enabled proving many conjectures about

836: this percolation model. For example, the prediction~\cite{DN,NienhuisCoulomb}

837: that the probability that

838: the cluster of the origin has diameter larger than $R$ decays like

839: \begin{equation}\label{e.onearm}

840: \Pb{\text{the origin is in a cluster of diameter}\ge R}

841: = R^{-5/48+o(1)}

842: \end{equation}

843: as $R\to\infty$ was proved~\cite{\LSWonearm}.

844: This value $5/48$ is an example of what is commonly referred to

845: as a critical exponent. Building on earlier work by Kesten

846: and others, as well as on Smirnov's theorem and \SLE/,

847: Smirnov and Werner~\cite{\SWpercexpo} were able to determine many useful

848: percolation exponents.

849: Julien Dub\'edat~\cite{\DubedatWatts} has used \SLE/ to prove Watts'~\cite{\Watts} formula

850: for the asymptotic probability that in a given rectangle there are both a

851: white horizontal and vertical crossing (for the above percolation model

852: at $p=p_c=1/2$). 

853: 

854: \medskip

855: 

856: The next process for which conformal invariance and

857: convergence to \SLE/ was established is the LERW~\cite{\LSWlesl}.

858: Contrary to Smirnov's proof for percolation, where convergence

859: to \SLE/ was a consequence of conformal invariance, in the case of the

860: LERW the proof establishes conformal invariance as a consequence

861: of the convergence to \SLEk2/. More specifically, the argument in~\cite{\LSWlesl}

862: proceeds by considering the Loewner driving term of the discrete

863: LERW (before passing to the limit) and proving that the driving term

864: converges to an appropriately time-scaled Brownian motion.

865: The same paper also shows that the uniform spanning tree scaling limit is

866: conformally invariant, and the Peano curve associated with it 

867: (essentially, the boundary of a thickened uniform spanning tree)

868: converges to \SLEk8/.

869: Another difference between the results of~\cite{\SmirnovPerc} and~\cite{\LSWlesl}

870: is that while the former is restricted to site percolation on the triangular lattice,

871: the results in~\cite{\LSWlesl} are essentially lattice independent.

872: Figure~\ref{f.lerw} above shows a fine LERW, which gives an idea of what an

873: \SLEk2/ looks like. Likewise, Figure~\ref{f.peano} shows a sample

874: of an initial segment of the uniform spanning tree Peano curve in a rectangular domain.

875: Note that the curve is space filling, as is \SLEk8/.

876: 

877: \begin{figure}

878: \centerline{\hfill\psfig{file=\figdir/peano.ps,width=\hsize}}

879: \caption{\label{f.peano}An initial segment of the uniform spanning tree Peano path.}

880: \end{figure}

881: 

882: 

883: Meanwhile, Gady Kozma~\cite{\KozmaLERWplane} came up with a different proof that the

884: LERW scaling limit exists.

885: Although Kozma's proof does not identify the limit, it has the advantage

886: of generalizing to three dimensions~\cite{\KozmaLERWspace}.

887: 

888: 

889: \medskip

890: 

891: There are two discrete models for which convergence to \SLEk4/ has

892: been established by Scott Sheffield and the present author.

893: These models are the harmonic explorer~\cite{\SchrammSheffieldHE}

894: and the interface of the discrete Gaussian free field~\cite{\SchrammSheffieldDGFF}.

895: The discrete and continuous Gaussian free fields (a.k.a.\ the harmonic crystal)

896: play an important role in the heuristic

897: physics analysis of various statistical physics models.

898: The discrete Gaussian free field is a probability measure

899: on real valued functions defined on a graph, often a piece of a lattice.

900: If, for example, the graph is a triangulation of a domain in the plane,

901: an interface is a curve in the dual graph separating vertices where the function is positive

902: from vertices where the function is negative.

903: See~\cite{\SheffieldGFFsurvey} or~\cite{\SchrammSheffieldDGFF} for further details,

904: and see Figure~\ref{f.hepath} for a simulation of the harmonic explorer, and therefore

905: an approximation of \SLEk4/.

906: 

907: \begin{figure}

908: \centerline{\quad\psfig{file=\figdir/hepathsimulation.ps,height=2.5in}}

909: \caption{\label{f.hepath}The harmonic explorer path.}

910: \end{figure}

911: 

912: Sheffield also announced work in progress connecting the Gaussian free field

913: with \SLEk\kappa/ for other values of $\kappa$. The basic idea is

914: that while \SLEk4/ may be thought of as a curve solving the equation

915: $h=0$, where $h$ is the Gaussian free field, for other $\kappa$, the

916: \SLEk\kappa/ curve may be considered as a solution of 

917: \begin{equation}\label{e.shef}

918: c\,\mathrm{winding}(\gamma[0,t])=h(\gamma(t))\,,

919: \end{equation}

920: where $c$ is a constant depending on $\kappa$. When $\kappa=4$, the

921: corresponding constant $c$ is zero, which reduces to the setting

922: of~\cite{\SchrammSheffieldDGFF}.

923: Alternatively,~\eref{e.shef} can be heuristically written

924: as $c\,\gamma'(s)=\exp(i\,h(\gamma(t))$,

925: where $s$ is the length parameter of $\gamma$. However,

926: we stress that it is hard to make sense of these equations,

927: for the Gaussian free field is not a smooth function (in fact,

928: it is not even a function but rather a distribution).

929: Likewise, the \SLE/ path is not rectifiable and its winding

930: at most points is infinite.

931: 

932: \medskip

933: 

934: As mentioned above, there are several different variants of \SLE/ in simply

935: connected domains:

936: chordal, radial, as well as a few others, which we have not mentioned.

937: These variants are rather closely related to one another~\cite{\SchrammWilsonCoord}.

938: There are also variants defined in the multiply-connected setting~\cite{\multiplyConnectedSLE}.

939: One motivation for this study comes from statistical physics models, which are easy

940: to define on multiply connected domains. Since one can easily vary the boundary

941: conditions on different boundary components of the domain, it is clear that there is

942: often more than one reasonable choice for the definition of the \SLE/ path.

943: 

944: \medskip

945: 

946: Finally, we mention an intriguing connection between Brownian motion and

947: \SLEk\kappa/ for $\kappa\in(8/3,4]$. There is the notion of the Brownian

948: loop soup~\cite{\LWsoup}, which is a Poisson measure on the

949: space of Brownian motion loops.  

950: According to~\cite{\WernerBMclusters}, the boundaries of clusters of

951: a sample from the loop soup measure with intensity $c$ are \SLEk\kappa/-like

952: paths, where $\kappa=\kappa(c)\in(8/3,4]$.

953: The proof is to appear in a future joint work of Sheffield and Werner.

954: 

955: 

956: \bigskip

957: The above account describes some of the highlights of the developments

958: in the field in the past several years.

959: The rest of the paper will be devoted to a description of

960: some problems where we hope to see some future progress.

961: Some of these problems are obvious to anyone working in the field

962: (though the solution is not obvious), while

963: others are borrowed from several different sources. A few of the problems

964: appear here for the first time.

965: The paper~\cite{\RSsle} contains some additional problems.

966: 

967: \section{Random processes converging to \SLE/}\label{s.sleconv}

968: 

969: As we have seen, paths associated with several random processes have been proved to 

970: converge to various \SLE/ paths.

971: However, the list of processes where the convergence is expected but not proved

972: yet is longer. This section will present questions of this sort, most

973: of which have previously appeared in the literature.

974: 

975: \medskip

976: The strategy of the proofs of convergence to \SLE/ in the

977: papers~\cite{\LSWlesl,\SchrammSheffieldHE,\SchrammSheffieldDGFF} is very similar.

978: In these papers, a collection of martingales with respect to the filtration given by the

979: evolution of the curve is used to gain information about the Loewner driving

980: term of the discrete curve. Although such a proof is also possible for

981: the percolation interface (using Cardy's formula),

982: this technique was not available at the time,

983: and Smirnov used instead an argument which uses the independence properties

984: of percolation in an essential way and is therefore not likely to be applicable

985: to many other models. Thus, it seems that presently the most promising technique

986: is the martingale technique from~\cite{\LSWlesl}.

987: 

988: \subsection{Notions of convergence}\label{s.notions}

989: 

990: To be precise, we must describe the meaning of these scaling limits.

991: In fact, there are at least two distinct reasonable notions of convergence, which

992: we now describe.

993: Suppose that $\gamma_n$ are random paths in the closed upper half plane

994: $\closure\H$ starting from $0$. Consider the one-point compactification

995: $\hat\C=\C\cup\{\infty\}$ of $\C=\R^2$, which may be thought of as the sphere $S^2$.

996: The law of $\gamma_n$ may be thought of as a Borel probability measure

997: on the Hausdorff space of closed nonempty subsets of $\hat\C$.

998: Since that Hausdorff space is compact, the space of Borel probability measures on

999: it is compact with respect to weak convergence of measures~\cite{\Dudley}.

1000: We may say that $\gamma_n$ converges in the Hausdorff sense to a random set

1001: $\gamma\subset\closure\H\cup\{\infty\}$ if the law of $\gamma_n$ converges

1002: weakly to the law of $\gamma$. A similar definition applies to curves in the

1003: closure of a bounded domain $D\subset\C$.

1004: The above stated instances of convergence to \SLE/ hold with respect to this

1005: notion. However, in the case of convergence to \SLEk8/, this does not mean very much,

1006: for \SLEk8/ fills up the domain.

1007: 

1008: The second notion of convergence is stronger. Suppose that each $\gamma_n$ is a.s.\

1009: continuous with respect to the half-plane capacity parametrization from

1010: $\infty$. (This holds, in particular, if $\gamma_n$ is a.s.\ a [continuous]

1011: simple path.) 

1012: If $d$ is a metric on $\hat \C=\C\cup\{\infty\}$

1013: compatible with its topology, then we may consider

1014: the metric 

1015: $$

1016: d^*(\beta_1,\beta_2):= \sup_{t\in[0,\infty)}d\bl(\beta_1(t),\beta_2(t)\br)

1017: $$

1018: on the space of continuous paths defined on $[0,\infty)$.

1019: We may say that $\gamma_n$ converges to a random path $\gamma$

1020: weakly-uniformly if the law of $\gamma_n$ converges weakly to the law of

1021: $\gamma$ in the space of Borel measures with respect to the metric $d^*$.

1022: This implies Hausdorff convergence. Since $d^*$ is finer than the

1023: Hausdorff metric, there are more functions from the space of paths

1024: to $\R$ that are continuous with respect to $d^*$ than with respect to

1025: the Hausdorff metric. Consequently, weakly-uniform convergence

1026: is stronger than Hausdorff convergence.

1027: In all the results stated above saying that some random path converges to \SLE/,

1028: the convergence is weakly-uniform when the paths are parametrized by capacity

1029: or half-plane capacity (depending on whether the convergence is to radial or

1030: chordal \SLE/, respectively).

1031: 

1032: In the following, when we ask for convergence to \SLE/, we will mean weakly-uniform

1033: convergence. However, weaker nontrivial forms of convergence would also be very interesting.

1034: 

1035: 

1036: \subsection{Self avoiding walk}

1037: 

1038: Let $G$ be either the square, the hexagonal or the triangular

1039: grid in the plane, positioned so that $0$ is some vertex in $G$.

1040: For $n\in\N$ consider the uniform measures on all

1041: self avoiding $n$-step walks in $G$

1042: that start at $0$ and stay in the upper half plane.

1043: It has been shown in~\cite{\LSWsaw} that when 

1044: $G=\Z^2$ the limiting measure as $n\to\infty$ exists.

1045: (The same proof probably applies for the other alternatives for $G$,

1046: provided that $G$ is positioned so that horizontal lines

1047: through vertices in $G$ do not intersect the relative interior

1048: of edges of $G$ which they do not contain.)

1049: 

1050: \begin{problem}[\cite{\LSWsaw}]\label{p.saw}

1051: Let $\gamma$ be a sample from the $n\to\infty$ limit of the uniform

1052: measure on $n$-step self avoiding paths in the upper half plane described

1053: above. Prove that the limit as $s\searrow0$ of the law of $s\,\gamma$

1054: exists and that it is \SLEk{8/3}/.

1055: \end{problem}

1056: 

1057: The convergence may be considered with respect to either of the two

1058: topologies discussed in subsection~\ref{s.notions}.

1059: 

1060: In~\cite{\LSWsaw} some consequences of this convergence are indicated,

1061: as well as support for the conjecture.

1062: 

1063: There are some indications that the

1064: setting of the hexagonal lattice is easier: the rate of growth

1065: of the number of self avoiding paths on the hexagonal grid is

1066: predicted~\cite{\NienhuisOn} to be $\bl(2+\sqrt 2+o(1)\br)^{n/2}$;

1067: no such prediction exists

1068: for the square grid or triangular lattice.

1069: 

1070: In dimensions $d>4$ Takashi Hara and Gordon Slade~\cite{\HaraSladeSAW} proved

1071: that the scaling limit of self-avoiding random

1072: walk is Brownian motion. This is also believed to be the case

1073: for $d=4$. See~\cite{\MadrasSlade} for references and further background

1074: on the self-avoiding walk.

1075: 

1076: \subsection{Height models}\label{ss.heights}

1077: 

1078: 

1079: There is a vast collection of height model

1080: interfaces that should converge to \SLEk4/. 

1081: The one theorem in this regard is the convergence of the interface of the

1082: Gaussian free field~\cite{\SchrammSheffieldDGFF}.

1083: This was motivated by Kenyon's theorem stating that

1084: the domino tiling height function converges to the

1085: Gaussian free field~\cite{\KenyonDominoGFF} and by Kenyon's conjecture

1086: that the double domino interface converges to \SLEk4/ (see~\cite{\RSsle} for a statement

1087: of this problem).

1088: 

1089: The domino height function is a function on $\Z^2$ associated

1090: with a domino tiling (see~\cite{\KenyonDominoGFF}).

1091: Its distribution is roughly (ignoring boundary issues)

1092: the uniform measure on functions $h:\Z^2\to \Z$

1093: such that $h(0,0)=0$,

1094: $$

1095: h(x,y)\mod 4 =\begin{cases}

1096: 0 &\text{$x,y$ even}\\

1097: 1 &\text{$x$ odd $y$ even}\\

1098: 2 &\text{$x,y$ odd}\\

1099: 3 &\text{$x$ even $y$ odd}

1100: \end{cases}

1101: $$

1102: and $|h(z)-h(z')|\in\{1,3\}$ if $|z-z'|=1$, $z,z'\in\Z^2$.

1103: 

1104: Let $D$ be a bounded domain in the plane whose boundary

1105: is a simple path in the triangular lattice, say.

1106: Let $\pa_+$ and $\pa_-$ be complementary arcs

1107: in $\pa D$ such that  the two common endpoints of these arcs

1108: are midpoints of edges. Consider the uniform measure on functions

1109: $h$ taking odd integer values on vertices in $\closure D$

1110: such that $h=1$ on $\pa_+$, $h=-1$ on $\pa_-$, and

1111: $|h(v)-h(u)|\in\{0,2\}$ for neighbors $v,u$.

1112: We may extend such a function $h$ to $\closure D$ by affine

1113: interpolation within each triangle, and this interpolation

1114: is consistent along the edges. There is then a unique

1115: connected path $\gamma$ that is the connected component

1116: of $h^{-1}(0)$ that contains the two endpoints of each

1117: of the two arcs

1118: $\pa_\pm$.

1119: 

1120: \begin{problem}\label{p.h1}

1121: Is it true that the path $\gamma$ tends to \SLEk4/?

1122: Does the law of $h$ converge to the Gaussian free field?

1123: \end{problem}

1124: 

1125: The convergence we expect for $h$ is in the same sense

1126: as in~\cite{\KenyonDominoGFF}.

1127: 

1128: 

1129: Note that if we restrict in the above the image of

1130: $h$ to be $\{1,-1\}$, we obtain critical site

1131: percolation on the triangular grid, and the limit

1132: of the corresponding interface is in this case \SLEk6/.

1133: 

1134: Now suppose that $(D,\pa_+,\pa_-)$ is as above. Let $\lambda\in(0,1/2]$ be

1135: some constant and 

1136: consider the uniform measure on functions

1137: $h$ taking real values on vertices in $\closure D$

1138: such that $h=\lambda$ on $\pa_+$, $h=-\lambda$ on $\pa_-$

1139: and $|h(v)-h(u)|\le 1$ for every edge $[v,u]$.

1140: 

1141: \begin{problem}

1142: Is it true that for some value of $\lambda$ the corresponding

1143: interface converges to \SLEk4/?

1144: Does the law of $h$ converge in some sense to the

1145: Gaussian free field?

1146: \end{problem}

1147: 

1148: In the case of the corresponding questions for the

1149: discrete Gaussian free field, there is

1150: just one constant $\lambda$ such that the interface

1151: converges to \SLEk4/. For other choices

1152: of $\lambda$ the interface converges to a

1153: well-known variant of \SLEk4/~\cite{\SchrammSheffieldDGFF}.

1154: 

1155: There are some restricted classes of height models for which convergence

1156: to the Gaussian free field is known~\cite{\NaddafSpencer}.

1157: It may still be very hard to prove that the corresponding interface

1158: converges to \SLEk4/. One interesting problem of this sort 

1159: is the following.

1160: 

1161: 

1162: \begin{problem}[\cite{\SchrammSheffieldDGFF}]

1163: If we project the Gaussian free field 

1164: onto the subspace spanned by the eigenfunctions of the Dirichlet Laplacian

1165: with eigenvalues in $[-r,r]$ and add the harmonic function with boundary

1166: values $\pm\lambda$ on $\pa_\pm$, does the corresponding interface converge to

1167: \SLEk4/ as $r\to\infty$ when $\lambda$ is chosen appropriately?

1168: \end{problem}

1169: 

1170: The problem is natural, because the Gaussian free field is related to the

1171: Dirichlet Laplacian. In particular, the projections of the field onto the spaces spanned

1172: by eigenfunctions with eigenvalues in two disjoint intervals are independent.

1173: 

1174: 

1175: \subsection{The Ising, FK, and $O(n)$ loop models}

1176: 

1177: The Ising model is a fundamental physics model for magnetism.

1178: Consider again a domain $D$ adapted to the triangular lattice

1179: and a partition $\pa D=\pa_+\cup\pa_-$ as in subsection~\ref{ss.heights}.

1180: Now consider a function $h$ that take the values $\pm1$

1181: on vertices in $\closure D$ such that $h$ is $1$ on $\pa_+$ and $-1$ on $\pa_-$.

1182: On the collection of all such functions we put a probability measure

1183: such that the probability for a given $h$ is

1184: proportional to $e^{-2\beta k}$, where $\beta$ is a parameter and $k$ is the number of edges

1185: $[v,u]$ such that $h(v)\ne h(u)$.

1186: This is known as the Ising model and the value associate to a vertex is

1187: often called a spin.

1188: It is known that the critical value $\beta_c$ (which we do not define here in the context of the Ising model)

1189: for $\beta$ satisfies $e^{2\beta}=\sqrt 3$ (see~\cite{McCoyWu,\ItzyksonDrouffe}).  

1190: Again, the interface at the critical $\beta=\beta_c$ is believed to converge to an

1191: \SLE/ path, this time \SLEk3/. 

1192: For $\beta\in[0,\beta_c)$ fixed, the interface should converge

1193: to \SLEk6/. Note that when $\beta=0$, the model is again identical

1194: to critical site percolation on the triangular grid,

1195: and the interface does converge to \SLEk6/.

1196: 

1197: \begin{problem}

1198: Prove that when $\beta=\beta_c$, the interface converges to \SLEk3/

1199: and when $\beta\in(0,\beta_c)$ to \SLEk6/.

1200: \end{problem}

1201: 

1202: When $\beta>\beta_c$ we do not expect convergence to \SLE/,

1203: and do not expect conformal invariance. 

1204: The interface scaling limit is in this case a straight line segment

1205: if the domain is convex~\cite{MR1704276} (see also~\cite{MR2165255}).

1206: 

1207: 

1208: \bigskip

1209: 

1210: The Fortuin-Kasteleyn~\cite{\FortuinKasteleyn} (FK) model

1211: (a.k.a.\ the random cluster model)

1212: is a probability measure on the collection of all subsets of the

1213: set of edges $E$ of a finite graph $G=(V,E)$.

1214: In the FK model, the measure of each $\omega\subset E$ is proportional to

1215: $\bl(p/(1-p)\br)^{|\omega|}\,q^c$, where $q>0$ and $p\in(0,1)$ are parameters,

1216: $|\omega|$ is the

1217: cardinality of $\omega$, and $c$ is the number of connected components

1218: of the subgraph $(V,\omega)$. The FK model is very closely related

1219: to the well known Potts model~\cite{BaxterKellandWu},

1220: which is a generalization of the Ising model.

1221: Many questions about  the Potts model 

1222: can be translated to questions about the FK model and vice versa.

1223: 

1224: On the grid $\Z^2$, when

1225: $p=\sqrt q/(1+\sqrt q)$ the FK model satisfies a form of self-duality.

1226: 

1227: \begin{problem}[\cite{\RSsle}]

1228: Prove that

1229: when $q\in(0,4)$ and $p=\sqrt q/(1+\sqrt q)$,

1230: the interface of the FK model on $\Z^2$ with appropriate boundary conditions

1231: converges to \SLEk \kappa/, where

1232: $\kappa=4\,\pi/\cos^{-1}\bl(-\sqrt q/2\br)$.

1233: (See~\cite{\RSsle} for further details.)

1234: \end{problem}

1235: 

1236: 

1237: \bigskip

1238: 

1239: The $O(n)$ loop model on a finite graph $G=(V,E)$ is a measure on the collection

1240: of subgraphs of $G$ where the degree of every vertex in the subgraph is $2$.

1241: (The subgraph does not need to contain all the vertices.)

1242: The probability of each such subgraph is proportional to $x^e\,n^c$,

1243: where $c$ is the number of connected components, $e$ is the number of edges

1244: in the subgraph, and $x,n>0$ are parameters.

1245: When $n$ is a positive integer, the $O(n)$ loop model is derived from the $O(n)$ spin model,

1246: which is a measure on the set of functions which associate to every

1247: vertex a unit vector in $\R^n$. 

1248: 

1249: In order to pin down a specific long path in the $O(n)$ loop model,

1250: we pick two points on the boundary

1251: of the domain and require that in the random subgraph the degrees of two boundary vertices

1252: near these two points be $1$, while setting the degrees of all other boundary vertices to $0$, say.

1253: Then the measure is supported on configurations with one simple path and a collection of loops.

1254: 

1255: Now we specialize to the hexagonal lattice.

1256: Set $x_c(n):=\bl(2+\sqrt{2-n}\br)^{-1/2}$,

1257: which is the conjectured critical parameter~\cite{\NienhuisOn}.

1258: 

1259: 

1260: \begin{problem}[\cite{\KagerNienhuisSurvey}]

1261: Prove that when $n\in[0,2]$ and $x=x_c(n)$

1262: [respectively, $x>x_c(n)$]

1263: the scaling limit of the path containing the two special boundary vertices

1264: is chordal \SLEk\kappa/, where

1265: $\kappa\in[8/3,4]$ [respectively, $\kappa\in [4,8]$] and $n=-2\,\cos(4\,\pi/\kappa)$.

1266: \end{problem}

1267: 

1268: When $x<x_c(n)$, we expect the scaling limit to be a straight line segment

1269: (if the domain is convex).

1270: The fact that at $n=1$ we get the same limits as for the Ising model is

1271: no accident. It is not hard to see that the $O(1)$ loop measure coincides with

1272: the law of boundaries of Ising clusters.

1273: See~\cite{\KagerNienhuisSurvey} for further details.

1274: 

1275: Similar conjectures should hold in other lattices. However, the values of the

1276: critical parameters are expected to be different.

1277: 

1278: 

1279: \subsection{Lattice trees}\label{ss.lattree}

1280: 

1281: We now present an example of a discrete model where we

1282: suspect that perhaps conformal invariance might hold.

1283: However, we do not presently have a candidate for the

1284: scaling limit. 

1285: 

1286: 

1287: Fix $n\in\N_+$, and consider the collection 

1288: of all trees contained in the grid $G$ that contain

1289: the origin and have $n$ vertices. Select a tree $T$ from

1290: this measure, uniformly at random. 

1291: 

1292: \begin{problem}

1293: What is the growth rate of the expected diameter

1294: of such a tree?

1295: If we rescale the tree so that the expected (or median)

1296: diameter is $1$, is there a limit for the law of the

1297: tree as $n\to\infty$? What are its geometric and topological

1298: properties? Can the limit be determined?

1299: \end{problem}

1300: 

1301: It would be good to be able to produce some pictures.

1302: However, we presently do not know how to sample from

1303: this measure.

1304: 

1305: \begin{problem}

1306: Produce an efficient algorithm which samples lattice

1307: trees approximately uniformly, or prove that such an algorithm

1308: does not exist.

1309: \end{problem}

1310: 

1311: See~\cite{\SladeKestenFest} for background on lattice trees

1312: and for results in high dimensions and~\cite{\BrydgesImbrie}

1313: for an analysis of a related continuum model.

1314: 

1315: 

1316: 

1317: \subsection{Percolation interface}

1318: 

1319: It is natural to try to extend the understanding of percolation

1320: at $p_c$ to percolation at a parameter $p$ tending to $p_c$.

1321: One possible framework is as follows.

1322: Fix a parameter $q\in(0,1)$.

1323: Suppose that in Figure~\ref{f.pcurve} with small mesh

1324: $\epsilon>0$ we choose $p(\epsilon)$

1325: so that the probability to have a left to right

1326: crossing of white hexagons in some fixed $1\times 1$

1327: square in the upper half plane is $q$ at percolation parameter $p=p(\epsilon)$.

1328: The corresponding interface will still be an unbounded path starting

1329: at $0$, but its distribution will be different

1330: from the interface at $p=1/2$ if $q\ne 1/2$.

1331: Thus, it is natural to ask

1332: 

1333: \begin{problem}[Lincoln Chayes (personal communication)]

1334: What is the scaling limit of the interface as $\eps \searrow0$

1335: and $p=p(\eps)$ if $q\ne 1/2$ is fixed?

1336: \end{problem}

1337: This problem is also very closely related to a problem

1338: formulated by F.~Camia, L. Fontes and C.~Newman~\cite{\CFNnearcrit}.

1339: 

1340: Site percolation on the triangular lattice is only one of

1341: several different models for percolation in the plane.

1342: Among discrete models, widely studied is bond percolation

1343: on the square grid. As for site percolation on the triangular

1344: lattice, the critical probability is again $p_c=1/2$.

1345: 

1346: At present, Smirnov's proof does not work for bond percolation

1347: on the square grid. The proof uses the invariance of the

1348: model under rotation by $2\,\pi/3$.

1349:  Thus, the following

1350: problem presents itself.

1351: 

1352: \begin{problem}

1353: Prove Smirnov's theorem for critical bond percolation on $\Z^2$.

1354: \end{problem}

1355: 

1356: Some progress on this problem has been reported by Vincent Beffara~\cite{BeffaraTalk}.

1357: \medskip

1358: 

1359: 

1360: There are other natural percolation models which have been

1361: studied. Among them we mention Voronoi percolation and

1362: the boolean model. In Voronoi percolation one has two

1363: independent Poisson point processes in the plane, $W$ and $B$,

1364: with intensities $p$ and $1-p$, respectively.

1365: Let $\hat W$ be the closure of the set of points in $\R^2$ closer to 

1366: $W$ than to $B$, and let

1367: $\hat B$ be the closure of the set of points closer to $B$.

1368: The set $\hat W$ is a sample from Voronoi percolation

1369: at parameter $p$.

1370: Some form of conformal invariance

1371: was proved for Voronoi percolation~\cite{\BenjaminiSchrammVoronoi},

1372: but the version proved does not imply convergence

1373: to \SLE/. It is neither stronger nor weaker than

1374: the conformal invariance proved by Smirnov.

1375: Notable recent progress has been made for

1376: Voronoi percolation by Bollob\'as and Riordan~\cite{\BollobasRiordanVoronoi},

1377: who established the very useful Russo-Seymour-Welsh

1378: theorem, as well as $p_c=1/2$ for Voronoi percolation.

1379: 

1380: \begin{problem}

1381: Prove Smirnov's theorem for Voronoi percolation.

1382: \end{problem}

1383: 

1384: The boolean percolation model (a.k.a.\ continuum percolation)

1385: can be defined by taking a Poisson set

1386: of points $W\subset\R^2$ of intensity $1$ and letting

1387: $\hat W$ be the set of points in the

1388: plane at distance at most $r$ from $W$.

1389: Here, $r$ is the parameter of the model.

1390: (Alternatively, one may fix $r=1$, say, and

1391: let the intensity of the Poisson process be the parameter,

1392: but this is essentially the same, by scaling.)

1393: The Russo-Seymour-Welsh theorem is known for this model~\cite{\RSWboolean,\RSWvacant},

1394: but the critical value of the parameter has not been identified.

1395: A nice feature which the model shares with Voronoi percolation

1396: is invariance under rotations.

1397: 

1398: \begin{problem}

1399: Prove Smirnov's theorem for boolean percolation.

1400: \end{problem}

1401: 

1402: 

1403: \section{Critical exponents}

1404: 

1405: The determination of critical exponents has been one motivation

1406: to prove conformal invariance for discrete models.

1407: For example, den Nijs and Nienhuis predicted~\cite{DN,NienhuisCoulomb} that the

1408: probability that the critical percolation cluster of the origin 

1409: has diameter larger than $R$ is $R^{-5/48+o(1)}$ as

1410: $R\to\infty$.

1411: Likewise, the probability that a given 

1412: site in the square $[-R,R]^2$ is pivotal

1413: for a left-right crossing of the square $[-2\,R,2\,R]^2$

1414: was predicted to be $R^{-5/4+o(1)}$.

1415: (Here, pivotal means that the occurence or non-occurence of a crossing

1416: would be modified by flipping the status of the site.)

1417: These and other exponents were proved for site percolation on the

1418: triangular grid using Smirnov's theorem and

1419: \SLE/~\cite{\LSWonearm,\SWpercexpo}.

1420: The determination of the exponents is very useful for the

1421: study of percolation.

1422: 

1423: Richard Kenyon~\cite{\KenyonLERW} calculated by 

1424: enumeration techniques involving determinants

1425: the asymptotics of the probability that an edge

1426: belongs to a loop-erased random walk.

1427: The probability decays like $R^{-3/4+o(1)}$

1428: when the distance from the edge to the endpoints of the

1429: walk is $R$. However, Kenyon's estimate is much more precise;

1430: he shows that, in a specific domain, $R^{3/4}$ times the probability is bounded

1431: away from zero and infinity, and in fact estimates the

1432: probability as $f\,R^{-3/4}\,\bl(1+o(1)\br)$ as $R\to\infty$,

1433: where $f$ is an explicit function of the positiong of the edge.

1434: 

1435: Thus, it is natural to ask for such precise estimates for

1436: the important percolation events as well. Namely,

1437: 

1438: \begin{problem}

1439: Improve the estimates $R^{-5/48+o(1)}$ and

1440: $R^{-5/4+o(1)}$ mentioned above (as well as other similar estimates)

1441: to more precise formulas. It would be especially nice to obtain

1442: estimates that are sharp up to multiplicative constants.

1443: \end{problem}

1444: 

1445: In addition to the case of the loop-erased random walk mentioned

1446: above, estimates up to constants are known for events involving

1447: Brownian motions~\cite{\LSWuptoconst}.

1448: \medskip

1449: 

1450: The difficulty in getting more precise estimates is not in the

1451: analysis of \SLE/. Rather, it is due to the passage between the

1452: discrete and continuous setting. Consequently, the above

1453: problem seems to be related to the following.

1454: 

1455: \begin{problem}

1456: Obtain reasonable estimates for the speed of convergence of

1457: the discrete processes which are known to converge to \SLE/.

1458: \end{problem}

1459: 

1460: There are still critical exponents which do not seem accessible

1461: via an \SLE/ analysis. For example, we may ask

1462: 

1463: \begin{problem}

1464: Calculate the number $\alpha$ such that on the

1465: event that there is a left-right crossing

1466: in critical percolation in the square $[0,R]^2$,

1467: the expected length of the shortest crossing is

1468: $R^{\alpha+o(1)}$.

1469: \end{problem}

1470: 

1471: Ziff~\cite{\Ziff} predicts an exponent which is related

1472: to this $\alpha$, but it seems that there is currently

1473: no prediction for the exact value of $\alpha$.

1474: 

1475: 

1476: \section{Quantum gravity}

1477: 

1478: Consider the uniform measure $\mu_n$ on equivalence classes

1479: of $n$-vertex triangulations of the sphere, where

1480: two triangulations are considered equivalent

1481: if there is a homeomorphism of the sphere taking one to the

1482: other. One may view a sample from this measure with

1483: the graph metric as a random geometry on the sphere.

1484: Such models go under the name \lq\lq quantum gravity\rq\rq\ in

1485: physics circles.

1486: One may also impose statistical physics models on such

1487: random triangulations. For example, the sample space

1488: may include such a triangulation (or rather, equivalence

1489: class of triangulations) together with a map $h$ from the

1490: vertices to $\{-1,1\}$. The measure of such a pair

1491: may be taken proportional to $a^k$, where $k$ is the

1492: number of edges $[v,u]$ in the triangulation for

1493: which $h(v)\ne h(u)$ and $a>0$ is a parameter.

1494: Thus, we are in effect considering a triangulation weighted

1495: by the Ising model partition function.

1496: (The partition function is in this case the sum of all the weights

1497: of such functions $h$ on the given triangulation.)

1498: Likewise, one may weight the triangulation by other kinds

1499: of partition functions.

1500: 

1501: In some cases it is easier to make a heuristic analysis

1502: of such statistical physics models in the quantum gravity

1503: world than in the plane. The enigmatic KPZ formula

1504: of Knizhnik, Polyakov and Zamolodchikov~\cite{\KPZ}

1505: was used in physics to predict properties

1506: of statistical physics in the plane from the corresponding

1507: properties in quantum gravity. Basically, the KPZ formula

1508: is a formula relating exponents in quantum gravity to the

1509: corresponding exponents in plane geometry.

1510: 

1511: To date, there has been progress in the mathematical

1512: (as well as physical) understanding of the statistical

1513: physics in the plane as well as in quantum gravity~\cite{\AngleGrowthUIPT,math.PR/0501006,\BousquetMelouSchaeffer}.

1514: However, there is still no mathematical understanding of

1515: the KPZ formula. In fact, the author's understanding of KPZ

1516: is too weak to even state a concrete problem.

1517: 

1518: However, we may ask about the scaling limit of $\mu_n$.

1519: There has been significant progress lately describing some

1520: aspects of the geometry of samples from $\mu_n$~\cite{\AngleGrowthUIPT,\ChassaingSchaeffer}.

1521: In particular, it has been shown by Chassaing and Schaeffer~\cite{\ChassaingSchaeffer}

1522: that if $D_n$ is the graph-metric

1523: diameter of a sample from $\mu_n$, then $D_n/{n^{-1/4}}$

1524: converges in law to some random variable in $(0,\infty)$.

1525: However, the scaling limit of samples from $\mu_n$ is not known.

1526: On the collection of compact metric spaces, we may consider the

1527: Gromov-Hausdorff distance $d_{GH}(X,Y)$, which is the infimum

1528: of the Hausdorff distance between subsets $X^*$ and $Y^*$

1529: in a metric space $Z^*$ over all possible

1530: triples $(X^*,Y^*,Z^*)$

1531: such that $Z^*$ is a metric space,

1532: $X^*,Y^*\subset Z^*$, $X$ is isometric with $X^*$

1533: and $Y$ is isometric with $Y^*$.

1534: Let $X_n$ be a sample from $\mu_n$, considered as a metric space

1535: with the graph metric scaled by $n^{-1/4}$, and let $\mu^*_n$ denote

1536: the law of $X_n$.

1537: 

1538: \begin{problem}\label{p.UTL}

1539: Show that the weak limit $\lim_{n\to\infty}\mu^*_n$

1540: with respect to the Gromov-Hausdorff metric exists.

1541: Determine the properties of the limit.

1542: \end{problem}

1543: 

1544: Note that~\cite{\MarckertMokkadem} proves convergence of

1545: samples from $\mu_n$, but the metric used there on

1546: the samples from $\mu_n$ is very different and consequently a

1547: solution of Problem~\ref{p.UTL} does not seem to follow.

1548: 

1549: 

1550: 

1551: \section{Noise sensitivity, Fourier spectrum, and dynamical percolation}

1552: 

1553: The indicator function of the event of having a percolation crossing

1554: in a domain between two arcs on the boundary is a boolean function of

1555: boolean variables.

1556: Some fundamental results concerning percolation are based on general 

1557: theorems about boolean functions.

1558: (One can mention here the BK inequality,

1559: the Harris-FKG inequality and the Russo formula. See, e.g.~\cite{\GrimmettPercolationEdtwo}.)

1560: Central to the theory of boolean functions is the

1561: Fourier expansion. Basically, if $f:\{-1,1\}^n\to\R$

1562: is any function of $n$ bits, the Fourier-Walsh expansion of $f$ is

1563: $$

1564: f(x)=\sum_{S\subset[n]} \hat f(S)\,\chi_S(x)\,,

1565: $$

1566: where $\chi_S(x)=\prod_{j\in S} x_j$ for $S\subset [n]=\{1,\dots,n\}$.

1567: When we consider $\{-1,1\}^n$ with the uniform probability measure,

1568: the collection $\{\chi_S:S\subset[n]\}$ forms an orthonormal

1569: basis for $L^2(\{-1,1\}^n)$. (Often other measures are also considered.)

1570: If we suppose that $\|f\|_2=1$ (in particular, this holds if

1571: $f:\{-1,1\}^2\to\{-1,1\}$),

1572: then the Parseval identity gives

1573: $\sum_{S\subset[n]} \hat f(S)^2=1$. Thus, we get a probability measure $\mu_f$

1574: on $2^{[n]}=\{S:S\subset[n]\}$ for which $\mu_f(\{S\})=\hat f(S)^2$.

1575: The map $S\mapsto |S|$, assigning to each $S\subset[n]$ its cardinality

1576: pushes forward the measure $\mu_f$ to a measure $\tilde\mu_f$

1577: on $\{0,1,\dots,n\}$. This measure $\tilde \mu_f$ may be called

1578: the Fourier spectrum of $f$. The Fourier spectrum encodes important information

1579: about $f$, and quite a bit of research on the subject

1580: exists~\cite{\FourierSurveys}. For instance, one can read off from $\tilde\mu_f$ the sensitivity

1581: of $f$ to noise (see~\cite{\BKSnoise}).

1582: 

1583: When $f$ is a percolation crossing function (i.e.,\ $1$ if there is a crossing,

1584: $-1$ otherwise), the corresponding index set $[n]$ is identified

1585: with the collection of relevant sites or bonds, depending if it is a site or bond

1586: model. Though there is some partial understanding of the Fourier spectrum

1587: of percolation~\cite{\SSexceptional}, the complete picture is unclear.

1588: For example, if the domain is approximately an $\ell \times \ell$

1589: square in the triangular lattice, then for the indicator function $f_{\ell}$

1590: of  a crossing in critical

1591: site percolation it is known~\cite{\SSexceptional} that

1592: for every $\alpha<1/8$

1593: \begin{equation}\label{e.percnoise}

1594: \tilde \mu_{f_\ell}([1,\ell^\alpha])\to 0

1595: \end{equation}

1596: as $\ell\to\infty$.

1597: This is proved using the critical exponents for site percolation

1598: as well as 

1599: an estimate for the Fourier coefficients

1600: of general functions (based on the existence of an algorithm

1601: computing the function which is unlikely to examine any specific

1602: input variable).

1603: On the other hand, using the percolation exponents one can show that~\eref{e.percnoise}

1604: fails if $\alpha>3/4$. This is based on calculating the expected number of sites pivotal

1605: for a crossing (i.e., a change of the value of the corresponding input variable would change

1606: the value of the function) as well as showing that the second moment is bounded

1607: by a constant times the square of the expectation. The expected number of pivotals

1608: is known~\cite{\SWpercexpo} to be $\ell^{3/4+o(1)}$ as $\ell\to\infty$.

1609: It is reasonable to conjecture that~\eref{e.percnoise}

1610: holds for every $\alpha<3/4$.

1611: 

1612: \begin{problem}\label{p.muell}

1613: Is it true that $\lim_{\ell\to\infty}

1614: \tilde\mu_{f_\ell}([\ell^{\alpha_1},\ell^{\alpha_2}])=1$

1615: if $\alpha_1<3/4<\alpha_2$?

1616: Determine the asymptotic behavior of

1617: $\tilde\mu_{f_\ell}([\ell^{\alpha_1},\ell^{\alpha_2}])$ as $\ell\to\infty$

1618: for arbitrary $0\le \alpha_1<\alpha_2\le 2$.

1619: \end{problem}

1620: 

1621: Estimates on the Fourier coefficients of percolation crossings (in an annulus)

1622: play a central part in the proof~\cite{\SSexceptional} that dynamical percolation

1623: has exceptional times. Dynamical percolation (introduced in~\cite{\HPSdynamical})

1624: is a model in which at each fixed time one sees an ordinary percolation configuration,

1625: but the random bits determining whether or not a site (or bond) is open undergo

1626: random independent flips at a uniform rate, according to

1627: independent Poisson processes. The main result of~\cite{\SSexceptional}

1628: is that dynamical critical site percolation on the triangular lattice has

1629: exceptional times at which there is an infinite percolation component. These set of times

1630: are necessarily of zero Lebesgue measure. 

1631: A better understanding of the Fourier coefficients may lead to sharper

1632: results about dynamical percolation, such as the determination

1633: of the dimension of exceptional times. Some upper and lower bounds for

1634: the dimension are known~\cite{\SSexceptional}.

1635: 

1636: One would hope to understand the measure $\mu_f$ geometrically.

1637: Gil Kalai (personal communication) has suggested the

1638: problem of determining the scaling limit of $\mu_f$.

1639: More specifically, the Fourier index set $S$ for critical

1640: percolation crossing of a square

1641: is naturally identified with a subset

1642: of the plane. If we rescale the square to have edge length $1$

1643: while refining the mesh, then $\mu_f$ may be thought of as

1644: a probability measure on the Hausdorff space $\mathcal H$ of closed subsets

1645: of the square. It is reasonable to expect that $\mu_f$

1646: converges weakly to some probability measure $\mu$

1647: on $\mathcal H$.

1648: We really do not know what samples from $\mu$ look like.

1649: Could it be that $\mu$ is supported on singletons?

1650: Alternatively, is it possible that $\mu[S=\text{entire square}]=1$?

1651: Is $S$ a Cantor set $\mu$-a.s.?

1652: 

1653: \begin{problem}[Gil Kalai, personal communication]

1654: Prove that the limiting measure $\mu$ exists and determine properties of samples

1655: from $\mu$.

1656: \end{problem}

1657: 

1658: Kalai suspects (personal communication) that

1659: the set $S$ is similar to the set of pivotal sites

1660: (which is a.s.\ a Cantor set in the scaling limit).

1661: This is supported by the easily verified fact that

1662: $\Pb{i\in S}=\Pb{i\text{ pivotal}}$

1663: and $\Pb{i,j\in S}=\Pb{i,j\text{ pivotal}}$

1664: hold for arbitrary boolean functions.

1665: Examples of functions where the scaling limit of $S$

1666: has been determined are provided by Tsirelson~\cite{\TsirelsonFourierWalsh,\TsirelsonScalingLimit}.

1667: 

1668: \bigskip

1669: 

1670: One may try to study a scaling limit of dynamical percolation.

1671: Consider dynamical critical site percolation on a triangular lattice of mesh $\eps$, where the

1672: rate at which the sites flip is $\lambda>0$.

1673: We choose $\lambda=\lambda(\eps)$ so that

1674: the correlation between having a left-right crossing of a fixed square

1675: at time $0$ and at time $1$ is $1/2$, say.

1676: Noise sensitivity of percolation~\cite{\BKSnoise} shows that

1677: $\lim_{\eps\searrow 0}\lambda(\eps)=0$ and the results of~\cite{\SSexceptional}

1678: imply that $\lambda(\eps)= \eps^{O(1)}$ and $\eps=\lambda(\eps)^{O(1)}$

1679: for $\eps\in (0,1]$.

1680: It is not hard to invent (several different) notions in which

1681: to take the limit  of dynamical percolation as $\eps\searrow 0$.

1682: 

1683: \begin{problem}\label{p.dynexists}

1684: Prove that the scaling limit of dynamical critical percolation exists.

1685: Prove that correlations between crossing events at different times $t_1<t_2$

1686: decay to zero as $t_2-t_1\to\infty$ and that a change in a crossing event becomes

1687: unlikely if $t_2-t_1\to 0$.

1688: \end{problem}

1689: 

1690: Since the correlation between events occuring at different times can be expressed in

1691: terms of the Fourier coefficients~\cite{\BKSnoise,\SSexceptional},

1692: it follows that the second

1693: statement in Problem~\ref{p.dynexists} is very much related to strong concentration

1694: of the measure $\tilde\mu_{f_\ell}$, in the spirit of Problem~\ref{p.muell}.

1695: 

1696: \medskip

1697: 

1698: Because of the dependence of $\lambda$ on $\eps$, it is not reasonable to

1699: expect the dynamical percolation scaling limit to be invariant under 

1700: maps of the form $f\times \text{identity}$, where $f:D\to D'$

1701: is conformal and the identity map is applied to the time coordinate.

1702: In particular, in the case where $f(z)=a\,z$, $a>0$,

1703: one should expect dynamical percolation to be invariant under the map

1704: $f\times (t\mapsto a^\beta\,t)$,

1705: where $\beta:= -\lim_{\eps\searrow 0}\log \lambda(\eps)/\log\eps$

1706: (and this limit is expected to exist).

1707: It is not too hard to see that $\beta=3/4$ if the answer to the first question

1708: in Problem~\ref{p.muell} is yes.

1709: 

1710: This suggests a modified form of conformal invariance for dynamical percolation.

1711: Suppose that $F(z,t)$ has the form $F(z,t)=\bl(f(z),g(z,t)\br)$,

1712: where $f:D\to D'$ is conformal and $g$ satisfies $\pa_t g(z,t)=|f'(z)|^\beta$,

1713: with the above value of $\beta$.

1714: Is dynamical percolation invariant under such maps?

1715: If such invariance is to hold, it would be in a \lq\lq relativistic\rq\rq\ framework,

1716: in which one does not consider crossings occuring at a specific time slice, but rather

1717: inside a space-time set. It is not clear if one can make good sense of that.

1718: 

1719: 

1720: 

1721: 

1722: 

1723: 

1724: \section{LERW and UST}

1725: 

1726: The loop-erased random walk and the uniform spanning

1727: tree are models where very detailed knowledge

1728: exists. They may be studied using random walks

1729: and electrical network techniques,

1730: and in the two-dimensional setting also by

1731: \SLE/ as well as domino tiling methods.

1732: (See~\cite{\LyonsUSFsurvey} and the references

1733: cited there.)

1734: However, some open problems still remain.

1735: 

1736: \medskip

1737: 

1738: One may consider the random walk in a

1739: fine mesh lattice in the unit disk, which

1740: is stopped when it hits the boundary of the disk.

1741: The random walk converges to Brownian motion

1742: while its loop-erasure converges to \SLEk2/.

1743: It is therefore reasonable to expect that the law of the

1744: pair $(\text{random walk},\text{its loop-erasure})$

1745: converges to a coupling of Brownian motion and \SLEk2/.

1746: (If not, a subsequential limit will converge.)

1747: 

1748: \begin{problem}\label{p.bmsle}

1749: In this coupling, is the \SLEk2/ determined by the

1750: Brownian motion?

1751: \end{problem}

1752: 

1753: It seems that this question occurred to several researchers

1754: independently, including Wendelin Werner (personal communication).

1755: 

1756: Of course, one cannot naively loop-erase the Brownian motion

1757: path, because there is no first loop to erase and there are cases

1758: where the erasure of one loop eliminates some of the other

1759: loops.

1760: 

1761: \bigskip

1762: 

1763: It is also interesting to try to extend some of the

1764: understanding of probabilistic statistical physics

1765: models beyond the planar setting to higher genus.

1766: The following problem in this direction was

1767: proposed by Russell Lyons (personal communication).

1768: 

1769: Consider the uniform spanning tree on a fine square grid approximation

1770: of a torus. There is a random graph dual to the tree, which

1771: consists of the dual edges perpendicular to primal edges not in the

1772: tree. It is not hard to see that this random dual of the tree 

1773: contains precisely three edge-simple closed paths

1774: (i.e., no repeating edges), and that these

1775: paths are not null-homotopic.

1776: 

1777: \begin{problem}\label{p.ustgenus}

1778: Determine the distribution of the triple of homotopy classes

1779: containing these three closed paths.

1780: \end{problem}

1781: 

1782: The problem would already be interesting for a square torus, but

1783: one could hope to get the answer as a function of the geometry of the torus.

1784: 

1785: 

1786: 

1787: 

1788: \section{Non-discrete problems}

1789: 

1790: In this section we mention some problems about the behavior of

1791: \SLE/ itself, which may be stated without relation to

1792: any particular discrete model.

1793: 

1794: The parametrization of the \SLE/ path by capacity is

1795: very convenient for many calculations. However, in some

1796: situations, for example when you consider the reversal

1797: of the path, this parametrization is not so useful.

1798: It would be great if we had an understanding

1799: of a parametrization by a kind of Hausdorff measure.

1800: Thus we are led to

1801: 

1802: \begin{problem}\label{p.Hau}

1803: Define a Hausdorff measure on the \SLE/ path

1804: which is $\sigma$-finite.

1805: \end{problem}

1806: 

1807: We would expect the measure to be a.s.\ finite on

1808: compact subsets of the plane.

1809: 

1810: That the Hausdorff dimension of the \SLE/ path

1811: is $\min\{2,1+\kappa/8\}$ has been established

1812:  by Vincent Beffara~\cite{\BeffaraSLEdim}.

1813: When $\kappa<8$ (in which case the path has zero area),

1814: we expect the $\sigma$-finite Hausdorff measure to

1815: be the Hausdorff measure with respect to the gauge

1816: function $\phi(r)=r^d\,\log\log(1/r)$, where $d=1+\kappa/8$

1817: is the Hausdorff dimension.

1818: This is based on past experience with similar

1819: random paths~\cite{\TaylorSurvey}.

1820: However, in order to prove that this Hausdorff measure

1821: is $\sigma$-finite, one should probably find alternative

1822: constructions of the measure.

1823:  In the case $\kappa\le 4$,

1824: where the \SLE/ path is a simple path a.s.,

1825: one could try to use conformal maps

1826: from the unit disk to the two components

1827: in the complement of the curve in the upper half plane.

1828: If $f$ is such a map, it might be possible to show

1829: that the limit of the length measure of the

1830: image of the circle $r\,\pa\U$, rescaled appropriately,

1831: has a limit as $r\nearrow 1$. 

1832: Another approach, which was discussed by Tom Kennedy~\cite{\KennedyVariation},

1833: would be to study the $\alpha$-variation of the \SLE/ path,

1834: though this seems hard to handle.

1835: 

1836: One may also consider other measures of growth for the

1837: \SLE/ path. For example, when $\kappa>4$, we may study the

1838: area of the \SLE/ hull. It would be interesting to study

1839: the various relations between different measures of growth.

1840: 

1841: \bigskip

1842: It is also natural to ask what kind of sets are visited by

1843: the \SLE/ path. More precisely:

1844: 

1845: \begin{problem}\label{p.hit}

1846: Fix $\kappa<8$.  Find necessary or sufficient conditions

1847: on a deterministic compact set $K\subset\closure\H$

1848: to satisfy $\Pb{K\cap\gamma\ne\emptyset}>0$,

1849: where $\gamma$ is the \SLEk\kappa/ path.

1850: \end{problem}

1851: 

1852: The case $K\subset\R$ is of particular interest.

1853: \medskip

1854: 

1855: When $\kappa=8/3$ and $\H\setminus K$ is

1856: simply connected, there is a simple explicit

1857: formula~\cite{\LSWrestriction} for

1858: $\Pb{\gamma\cap K\ne\emptyset}$.

1859: It is not clear if such formulas are also available

1860: for other values of $\kappa$.

1861: Wendelin Werner~\cite{\WernerLoops} proved the

1862: existence of a random collection of \SLEk{8/3}/-like

1863: loops with some wonderful properties.

1864: In particular, the expected number of loops which

1865: separate two boundary components of an annulus

1866: is conformally invariant, and therefore a function

1867: of the conformal modulus of the annulus.

1868: However, this function is not known explicitly.

1869: 

1870: \bigskip

1871: 

1872: Many of the random interfaces which are known or believed

1873: to converge to \SLE/ are reversible, in the sense that the

1874: reversed path has the same law as the original path

1875: (with respect to a slightly modified setup).

1876: This motivates the following problem from~\cite{\RSsle}.

1877: 

1878: \begin{problem}\label{p.reverse}

1879: Let $\gamma$ be the chordal \SLEk\kappa/ path, where

1880: $\kappa\le 8$.

1881: Prove that up to reparametrization, the image of $\gamma$

1882: under inversion in the unit circle

1883: (that is, the map $z\mapsto 1/\bar z$) has the same law as $\gamma$

1884: itself.

1885: \end{problem}

1886: 

1887: The reason that we restrict to the case $\kappa\le 8$ is that

1888: this is known to be false as stated when $\kappa>8$~\cite{\RSsle}.

1889: Indeed, there are no known models from physics that are believed

1890: to be related to \SLEk\kappa/ when $\kappa>8$.

1891: Sheffield (personal communication) expects that at least

1892: in the case $\kappa<4$ Problem~\ref{p.reverse} can be 

1893: answered by studying the relationship between the

1894: Gaussian free field and \SLE/.

1895: 

1896: %% \medskip

1897: %% Bertrand Duplantier indicated that

1898: %% when $\kappa>4$ the outer boundary of \SLEk\kappa/

1899: %% stopped at capacity $1$, say, should locally look like

1900: %% \SLEk{16/\kappa}/. This property is referred to

1901: %% as {\bf duality}. A more precise formulation of

1902: %% duality was presented by Julien Dub\'edat~\cite{\DubedatDuality}.

1903: %% 

1904: %% \begin{problem}

1905: %% Prove \SLE/ duality, for example, as described in~\cite{\DubedatDuality}.

1906: %% \end{problem}

1907: 

1908: 

1909: \bigskip\noindent{\bf Acknowledgments:}

1910: Greg Lawler, Wendelin Werner and Steffen Rohde have collaborated

1911: with me during the early stages of the development of SLE. 

1912: Without them the subject would not be what it is today.

1913: I wish to thank Itai Benjamini, Gil Kalai, Richard Kenyon,

1914: Scott Sheffield, Jeff Steif

1915: and David Wilson for numerous inspiring conversations.

1916: Thanks are also due to Yuval Peres for useful advice, especially concerning

1917: Problem~\ref{p.Hau}.

1918: 

1919: 

1920: 

1921: \bibliography{mr,prep,notmr}

1922: \bibliographystyle{a}

1923: 

1924: \end{document}

1925: