1: \documentclass{article}

2: \usepackage{graphicx}

3: \usepackage{psfrag}

4: \usepackage{amssymb}

5: \usepackage{amsmath}

6: \usepackage{amsthm}

7: \usepackage{amscd}

8: 

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

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

11: \newtheorem{prop}[thm]{Proposition}

12: \newtheorem{cor}[thm]{Corollary}

13: \newtheorem{lem}[thm]{Lemma}

14: \newtheorem{qstn}[thm]{Question}

15: \newtheorem{rem}[thm]{Remark}

16: \newtheorem*{ec}{Ending Lamination Conjecture}

17: \newtheorem*{tc}{Tameness Conjecture}

18: \newtheorem*{thm;filling}{Theorem \ref{thm;filling}}

19: \newtheorem*{thm;surgery}{Theorem \ref{thm;surgery}}

20: \newtheorem*{twistthm}{Corollary \ref{cor;twist}}

21: \newtheorem*{SGL}{Theorem \ref{thm;SGL}}

22: \newtheorem*{thm;klarreich}{Theorem \ref{thm;klarreich}}

23: \newtheorem*{mainthm}{Main Theorem}

24: 

25: \theoremstyle{definition}

26: \newtheorem{defn}[thm]{Definition}

27: \newtheorem{exmp}[thm]{Example}

28: \newtheorem{excs}[thm]{Exercise}

29: 

30: % \newenvironment{proof}{\medskip \noindent

31: % {\bf Proof.}}{\hfill \rule{.5em}{1em} \\}

32: 

33: \newcommand{\A}{\mathcal{A}}

34: \newcommand{\T}{\mathcal{T}}

35: \newcommand{\TC}{\overline{\mathcal{T}}}

36: \newcommand{\ML}{\mathcal{ML}}

37: \newcommand{\MF}{\mathcal{MF}}

38: \newcommand{\GL}{\mathcal{GL}}

39: \newcommand{\PML}{\mathcal{PML}}

40: \newcommand{\PMF}{\mathcal{PMF}}

41: \newcommand{\UML}{\mathcal{UML}}

42: \newcommand{\EL}{\mathcal{EL}}

43: \newcommand{\C}{\mathcal{C}}

44: \newcommand{\MCG}{\mathcal{MCG}}

45: \newcommand{\el}{\ell}

46: \newcommand{\BC}{\partial_\infty{\mathcal{C}}}

47: \newcommand\N{{\mathbb N}}

48: \newcommand\Z{{\mathbb Z}}

49: \newcommand\R{{\mathbb R}}

50: \newcommand{\s}{\sigma}

51: \newcommand{\lam}{\lambda}

52: \newcommand{\mb}{\mathbf}

53: \newcommand{\mc}{\mathcal}

54: 

55: 

56: \title{The Thurston Boundary of Teichm\"uller Space and Complex of Curves}

57: \author{Young Deuk Kim \\ Department of Mathematics\\ Yale University\\ 

58: P.O. Box 208283\\New Haven, CT 06520-8283, USA 

59: \\(youngdeuk.kim@yale.edu)}

60: \date{\today}

61: 

62: \begin{document}

63: \maketitle

64: 

65: \begin{abstract}

66: Let $S$ be a closed orientable surface with genus $g\geq 2$. 

67: For a sequence $\s_i$ in the Teichm\"uller space of $S$, which converges to

68: a projective measured lamination $[\lam]$ in the Thurston boundary,

69: we obtain a relation between $\lam$ and the geometric limit of pants 

70: decompositions whose lengths are uniformly bounded by a Bers constant $L$. 

71: We also show that this bounded pants decomposition is related to the Gromov 

72: boundary of complex of curves. 

73: 

74: \vspace{.3cm}

75: \noindent

76: 2000 Mathematics Subject Classification ; 30F60, 32G15, 57M50, 57N05. 

77: \end{abstract}

78: 

79: 

80: %%%%%%%%%%%%%%%%%%%%%%

81: %%%%%%%%%%%%%%%%%%%%%%

82: \section{Introduction} 

83: %%%%%%%%%%%%%%%%%%%%%%

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

85: 

86: 

87: Let $S$ denote a closed orientable surface with genus $g\geq 2$, and $\T(S)$ 

88: the Teichm\"uller space of $S$. Although there exist some similarities between 

89: $\T(S)$ and a complete negatively curved space 

90: (see \cite{bers,FLP,krushkal,masur1}), Masur (\cite{masur}) showed that $\T(S)$

91:  is not negatively curved. In \cite{gromov}, Gromov introduced Gromov 

92: hyperbolic spaces which include complete negatively curved spaces and metric 

93: trees, but Masur and Wolf (\cite{MW}) showed  that $\T(S)$ is not Gromov 

94: hyperbolic, either (see \cite{MP1}, for another proof by McCarthy and 

95: Papadopoulos). 

96:  

97: \indent

98: In \cite{luo}, Luo classified surface theories into geometric theory, i.e. 

99: $\T(S)$,  algebraic theory, i.e. the mapping class group $\MCG(S)$, and 

100: topological theory, i.e. the complex of curves $\C(S)$. 

101: The complex of curves, which was introduced by Harvey in \cite{harvey}, is a 

102: finite dimensional simplicial complex  whose vertices are non-trivial homotopy 

103: classes of simple closed curves which are not boundary-parallel, and 

104: $k$-simplices are $k+1$ distinct vertices with disjoint representatives. 

105: Let $\C_0(S)$ denote the set of vertices and $\C_1(S)$ its 1-skeleton.

106: In \cite{MM}, Masur and Minsky defined a metric on $\C(S)$ by making each 

107: simplex regular Euclidean with side length 1 and taking shortest-path metric, 

108: and showed that $\C(S)$ is a non-proper Gromov hyperbolic space (see 

109: \cite{bowditch}, for a shorter proof by Bowditch). 

110: Every Gromov hyperbolic space has a natural boundary which is called Gromov

111: boundary (see \cite{BS,gromov,short}). We write $\BC(S)$ to denote the Gromov 

112: boundary of $\C(S)$.

113: 

114: \indent

115: Since $\T(S)$ is not Gromov hyperbolic, we can not define its Gromov boundary. 

116: For example, Kerckhoff(\cite{kerckhoff}) showed that the Teichm\"uller

117: boundary, which is the set of endpoints of geodesic rays, depends on the 

118: choice of base point. In \cite{thurston2}, Thurston introduced a 

119: compactification of $\T(S)$ with the boundary equal to the space of projective 

120: measured laminations $\PML(S)$, on which the action of $\MCG(S)$ extends 

121: continuously. Throughout this thesis, we will write $\TC(S)$ to denote this 

122: compactification of Thurston. See \cite{bers1,brock2} for a similar but 

123: different compactification by Bers. 

124: 

125: \indent

126: Thurston boundary $\PML(S)$ is the space of projective classes of measured 

127: laminations. We write $\ML(S)$ to denote the space of measured laminations, 

128: and $[\lam]$ the projective class of $\lam\in\ML(S)$. 

129: A measured lamination consists of a geodesic lamination and a transverse 

130: measure with full support on it. 

131: The topology on the space of geodesic laminations $\GL(S)$ is the Hausdorff 

132: metric topology on closed subsets. On $\ML(S)$, Thurston gave the weak-topology

133: induced by the measures on transverse arcs. Note that $\PML(S)$ has the 

134: natural quotient topology. Let $\UML(S)$ be the quotient space of $\PML(S)$ by 

135: forgetting measure. Although $\UML(S)$ is a subset of $\GL(S)$,

136: the quotient topology on $\UML(S)$ is not equal to the subspace topology.

137:    

138: \indent

139: In \cite{thurston2} \S 5, Thurston wrote {\em  ``... Intuitively, the 

140: interpretation is that a sequence of hyperbolic structures on $S$ can go to 

141: infinity by ``pinching'' a certain geodesic lamination $\lam$; then it 

142: converges to $\lam$. As a lamination is pinched toward $0$, lengths of paths 

143: crossing it are forced toward infinity. The ratios of these lengths determine 

144: the transverse invariant measure  ...'' }. 

145: 

146: \indent

147: More clearly, a sequence $\s_i\in\T(S)$ converges to $[\lam]\in\PML(S)$ in 

148: $\TC(S)$ if and only if for all simple closed curves $\alpha$, $\beta$ on $S$,

149: $${{\el_{\s_i}(\alpha)}\over {\el_{\s_i}(\beta)}}\quad\mbox{converges to}

150: \quad {{i(\alpha,\lam)}\over {i(\beta,\lam)}}\ ,$$ 

151: where $\el_{\s_i}(\alpha)$ is the length of closed $\s_i$-geodesic which is 

152: homotopic to $\alpha$, and $i(\alpha,\lam)$ is the intersection number of

153: $\alpha$ and $\lam$ which is a generalization of the geometric intersection 

154: number of simple closed curves (see \cite{bon,MO,papa}).  

155:      

156: \indent

157: The geometry of $\T(S)$ and complex of curves are well described by Minsky in 

158: \cite{m-geom}. They are related by the collar lemma (see \cite{buser,keen}). 

159: The collar lemma implies that there is a universal constant $\epsilon>0$ such 

160: that for any distinct $\alpha$, $\beta\in\C_0(S)$ and $\s\in\T(S)$, 

161: if $\el_\s(\alpha)<\epsilon$ and $\el_\s(\beta)<\epsilon$ then the two geodesic

162: representatives of $\alpha$ and $\beta$ are disjoint, i.e. $\alpha$ and $\beta$

163: are on a same simplex in the complex of curves. Therefore $\C_1(S)$ could be 

164: considered as the nerve of the family of regions 

165: $$T(\alpha)=\{\s\in\T(S)\mid\el_\s(\alpha)<\epsilon\},\quad \alpha\in\C_0(S).$$

166: Although $\T(S)$ is not Gromov hyperbolic, Masur and Minsky showed that $\T(S)$

167: is Gromov hyperbolic modulo this family of regions (see 

168: \cite{MM,m-geom,m-extremal}).

169: 

170: \indent

171: A lamination $\mu\in\ML(S)$ is called {\em a filling lamination} if 

172: $i(\mu,\mu')=0$ then

173: $$\mbox{support}(\mu)=\mbox{support}(\mu'),$$

174: for any $\mu'\in\ML(S)$.  

175: Minsky wrote $\EL(S)$ to denote the image of filling laminations in $\UML(S)$. 

176: In the celebrated proof of Thurston's ending lamination conjecture, by Brock,

177: Canary and Minsky, the laminations in $\EL(S)$ are appeared as ending 

178: laminations of Kleinian surface groups without accidental parabolics 

179: (see \cite{m-ending,m-ending1}). 

180: \begin{ec}

181: A hyperbolic 3-manifold with finitely generated fundamental group is uniquely 

182: determined by its topological type and its end invariants.

183: \end{ec}

184: \noindent

185: The proof of the ending lamination conjecture and the recent proof of Marden's 

186: tameness conjecture by Agol(\cite{agol}) give us rough picture of hyperbolic 

187: 3-manifolds. See \cite{marden} for the original conjecture, and see \cite{CG} 

188: for another proof  by Calegari-Gabai. 

189: \begin{tc} 

190: A hyperbolic 3-manifold with finitely generated fundamental group is 

191: homeomorphic to the interior of a compact manifold with boundary.

192: \end{tc}

193:   

194: \indent

195: The Gromov boundary of $\C(S)$ is homeomorphic to the Gromov boundary of its 

196: 1-skeleton $\C_1(S)$ because $\C(S)$ is quasi-isometric to $C_1(S)$. 

197: In \cite{MM}, Masur and Minsky showed that the relative hyperbolic space, which

198:  is roughly $\T(S)$ modulo the regions $T(\alpha)$, is quasi-isometric to 

199: $\C_1(S)$. 

200: Therefore we can expect some relation between Thurston boundary of $\T(S)$ and 

201: Gromov boundary of $\C(S)$ (see \cite{m-geom} for the first question of Minsky 

202: on this relation). In fact, Klarreich showed that $\BC(S)$ is homeomorphic to 

203: $\EL(S)$ (see \cite{hamenstadt} for a new proof by Hamenst\"adt). 

204: 

205: \begin{thm}[Klarreich \cite{klarreich}]\label{thm;klarreich}

206: There is a homeomorphism $$k:\BC(S)\to\EL(S)$$ such that for any sequence 

207: $\alpha_i$ in $\C_0(S)$, $\alpha_i$ converges to $\alpha\in\BC(S)$ if and only 

208: if $\alpha_i$, considered as a subset of $\UML(S)$, converges to $k(\alpha)$.

209: \end{thm} 

210: 

211: \indent

212: In the following theorem of Bers, $\T(\Sigma)$ stands for the Teichm\"uller 

213: space of $\Sigma$ with geodesic boundaries (see \cite{bers2, bers3,buser}).   

214: \begin{thm}[Bers \cite{bers3}]\label{thm;bers}

215: Let $\Sigma$ be a compact Riemann surface with genus $g\geq 2$ from which

216: $n$ points and $m$ disks have been removed. For any $\sigma\in\T(\Sigma)$, 

217: there exist $3g-3+n+m$ disjoint geodesics, which are not boundary parallel, 

218: whose lengths are bounded by a constant $L$ which depends only on $g,n,m$, and 

219: the largest length of the geodesics homotopic to the boundaries of $\Sigma$. 

220: \end{thm}

221: 

222: \indent

223: The constant $L$ is called {\em  a Bers constant}. Notice that for the closed 

224: surface $S$, for any $\s\in\T(S)$, $S$ has a pants decomposition with total 

225: length bounded by a constant $L$ which depends only on the genus $g$. 

226: In fact, Buser and Sepp\"al\"a(\cite{BS}) showed that we can choose 

227: $L=21g(3g-3)$. Throughout this thesis, we write $L$ to denote a fixed Bers 

228: constant.   

229: 

230: \indent

231: Motivated by Theorem \ref{thm;bers}, we define a function $\Phi$ on $\T(S)$ as 

232: follows. For $\s\in\T(S)$, let 

233: $$\Phi(\s)=\mbox{ a pants decomposition whose total length is bounded by L},$$ 

234: where all pants curves are geodesics in $\s$. Let 

235: $u:\PML(S)\to\UML(S)$ be the quotient map by forgetting measure. Suppose that

236: \begin{equation}\label{eq;basic}

237: \s_i\in\T(S)\mbox{ converges to }[\lam]\in\PML(S)\mbox{ in }\TC(S)

238: \end{equation}

239: and $\alpha_i$ is a pants curve in $\Phi(\s_i)$.

240: From Theorem \ref{thm;klarreich}, the following question is immediate.

241: \begin{qstn}\label{qstn1}

242: Suppose that $\lam$ is a filling lamination. Is the geodesic representative of 

243: $\alpha_i$ converging to $u([\lam])$ in $\UML(S)$? 

244: \end{qstn}

245: \noindent

246: This question is the motivation of all the work in this thesis. In Chapter

247: \ref{chap;ct}, we will solve this question positively.

248: Notice that if we identify $\BC(S)$ with $\EL(S)$ via the homeomorphism in 

249: Theorem \ref{thm;klarreich}, then we have 

250: \begin{thm}\label{thm;filling}

251: If $\lam$ is a filling lamination, then $\alpha_i$ converges to $u([\lam])$ in 

252: $\C(S)\cup\BC(S)$.

253: \end{thm}

254: 

255: \indent

256: Suppose that $\lam$ is not necessarily a filling lamination in eq. 

257: (\ref{eq;basic}). Is there any relation between the limit point of $\alpha_i$ 

258: and $u([\lam])$? 

259: The following simple example shows that a limit of $\alpha_i$ and $u([\lam])$

260: could be disjoint.

261: \begin{exmp}\label{exmp;disjoint}

262: Consider a fixed $\gamma\in\C_0(S)$. Suppose that $\s_i\in\T(S)$ converges to 

263: $[\gamma]\in\PML(S)$ and $\el_{\s_i}(\gamma)\to 0$, where 

264: we consider $\gamma$ as an element of $\ML(S)$ via the counting measure. 

265: Since $\el_{\s_i}(\gamma)\to 0$, there exists a pants decomposition 

266: $\Phi(\s_i)$ whose total length is bounded by L and

267: $$\gamma\in\Phi(\s_i)\quad\mbox{for all large enough }i.$$ 

268: Then we can find a pants curve $\alpha_i$ in $\Phi(\s_i)$ such that a limit 

269: point of $\alpha_i$ is disjoint from $\gamma$ in $\UML(S)$.  

270: \end{exmp} 

271: 

272: \indent

273: A nonempty geodesic lamination $\mu\in\GL(S)$ is called {\em  minimal} if no 

274: proper subset of $\mu$ is a geodesic lamination. For example, any closed 

275: geodesic is a minimal lamination. 

276: The following theorem on the structure of a geodesic lamination on $S$ works 

277: for any hyperbolic surface of finite type (see \cite{CEG} \S4.2, \cite{CB}

278: \S4 or \cite{thurston1} \S8).

279: \begin{thm}[Structure of Geodesic Lamination]\label{thm;SGL} 

280: A geodesic lamination on $S$ is the union of finitely many minimal 

281: sublaminations and of finitely many infinite isolated leaves whose ends spiral

282: along the minimal sublaminations. 

283: \end{thm}

284: \noindent

285: Furthermore, if $[\lam]\in\PML(S)$ then we can decompose $u([\lam])$ as a 

286: finite disjoint union of minimal laminations,

287: \begin{equation}\label{eq;decomposition}

288: u([\lam])=\lam_1\cup\lam_2\cup\cdots\cup\lam_m. 

289: \end{equation}

290: 

291: \indent

292: An essential subsurface $F$ of $S$ is a subsurface of $S$ whose boundaries are 

293: all homotopically non-trivial geodesics. Throughout this thesis, we assume that

294: all essential subsurfaces of $S$ are open, i.e. the boundaries are not 

295: included (see Figure \ref{fig;intro1}). 

296: Note that two distinct boundaries of $\overline{F}$ could be a same curve in

297: $S$, where $\overline{F}$ is the completion of $F$ with the path-metric in $F$.

298: 

299: \begin{figure}[ht]

300: \begin{center}

301: \psfrag{F}{$F$}

302: \includegraphics[width=2in,height=1.5in]{intro1.eps}

303: \end{center}

304: \caption{An essential subsurface}

305: \label{fig;intro1}

306: \end{figure}

307: 

308: \noindent

309: Suppose that $\mu\in\UML(S)$. An essential subsurface $F$ is called 

310: {\em filled by $\mu$}, if for any simple closed curve $\alpha$ in $F$ 

311: which is not parallel to a boundary of $\overline{F}$, $\alpha$ intersects 

312: $\mu$. 

313: 

314: \indent

315: Suppose that $\s_i\in\T(S)$ converges to $[\lam]\in\PML(S)$, and let 

316: $u([\lam])=\lam_1\cup\lam_2\cup\cdots\cup\lam_m$ be the decomposition into 

317: minimal sublaminations. The main theorem of this thesis is  

318: 

319: \begin{thm}[Main Theorem]\label{mainthm}

320: If $\Phi(\s_i)$ converges to a geodesic lamination $\nu$ in Hausdorff metric 

321: topology, then $u([\lam])\subset\nu$.

322: \end{thm}

323: 

324: \indent

325: There is no minimal lamination which fills a pair of pants (see \cite{penner}

326: \S 2.6), therefore in eq. (\ref{eq;decomposition}), if $\lam_j$ is not a simple

327: closed curve then the essential subsurface filled by $\lam_j$ is at least a 

328: 1-holed torus or 4-holed sphere. 

329: For a closed annulus $Y$, Minsky defined the arc complex $\A(Y)$. 

330: In this complex vertices are essential homotopy classes, rel endpoints, of 

331: properly embedded arcs, and simplices are sets of vertices with representatives

332: with disjoint interiors. Note that the endpoints are {\em not} allowed to move 

333: in the boundary. Let $\A_0(Y)$ be the set of vertices and $\A_1(Y)$ the 

334: 1-skeleton, and give shortest-path metrics to $\A(Y)$ and $\A_1(Y)$ as in 

335: $\C(S)$. For $a,b\in\A_0(Y)$, we write $a\cdot b$ to denote the {\em algebraic}

336: intersection number. 

337: 

338: \indent

339: Suppose that in eq. (\ref{eq;decomposition}), $\lam_1$ is a simple closed 

340: curve. Consider an annular covering $Y$ of $S$ in which a neighborhood of 

341: $\lam_1$ lifts homeomorphically. The following corollary follows from the

342: main theorem. 

343: 

344: \begin{cor}\label{cor;twist}

345: Suppose that $\alpha_i$ meets $\lam_1$ for all $i$. Then $|a_1\cdot a_i|$ 

346: approaches to $\infty$, where $a_1\in\A_0(Y)$ is a lift of $\alpha_1$ in $Y$ 

347: and so is $a_i$.

348: \end{cor}

349: 

350: \indent

351: Suppose that in eq. (\ref{eq;decomposition}), $\lam_2$ is not a simple closed 

352: curve. Suppose that $F$ is an essential subsurface of $S$ which is filled by 

353: $\lam_2$. Notice that there exists a pants curve $\alpha_i$ in $\Phi(\s_i)$ 

354: such that $\alpha_i\cap F\neq\emptyset$ for all $i$. Let $\beta_i$ be a 

355: component of $\alpha_i\cap F$. We can define a simple closed curve 

356: $\Tilde{\beta}_i$ from $\beta_i$ canonically (see Chapter \ref{chap;ct}). 

357: We will prove the following theorem which is a generalization of Theorem 

358: \ref{thm;filling}.

359: 

360: \begin{thm}\label{thm;surgery}

361: The geodesic representative of $\Tilde{\beta}_i$ converges to $\lam_2$ in 

362: $\UML(F)$.

363: \end{thm}

364: 

365: \indent

366: In Chapter \ref{chap;tbdry} and Chapter \ref{chap;gbdry}, we study some 

367: preliminaries. We prove the main theorem in Chapter \ref{chap;pfmainthm},

368: and we prove Corollary \ref{cor;twist} and Theorem \ref{thm;surgery} in 

369: Chapter \ref{chap;ct}. 

370: 

371: \vspace{.5cm}

372: \textbf{\large{Acknowledgements.}}

373: This paper is a part of Ph.D thesis of the author (May 2005, Stony Brook 

374: university). 

375: The author would like to express his deepest gratitude to his advisor 

376: Yair Minsky for suggesting this thesis topic, and for his patient guidance 

377: and continuous encouragement.  

378: 

379: 

380: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

381: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

382: \section{The Thurston Boundary of Teichm\"uller Space}\label{chap;tbdry} 

383: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

384: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

385: 

386: Let $S$ be a closed oriented surface of genus $g\geq 2$. In this chapter we 

387: study the Teichm\"uller space $\T(S)$ and its Thurston boundary $\PML(S)$.

388: See \cite{IT} and \cite{thurston5} as references. 

389: 

390: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

391: \subsection{Teichm\"uller space and its boundaries}\label{sec;teich}

392: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

393: 

394: A conformal structure $\s$ on $S$ is determined by an atlas of coordinate 

395: neighborhoods $(U_\alpha,z_\alpha)$, where $\{U_\alpha\}$ is an open cover of 

396: $M$, and $z_\alpha:U_\alpha\to\mathbb{C}$ has the property that 

397: $z_\alpha\circ z_\beta^{-1}$ is analytic whenever defined.

398: Teichm\"uller space $\T(S)$ is the space of conformal structures

399: on $S$, where two structures are considered to be equivalent if there

400: is a conformal map between them isotopic to the identity. 

401: Recall that two diffeomorphisms $f$ and $g$ on $S$ are called {\em isotopic}

402: if there exists a diffeomorphism $H(x,t)=(h_t(x),t):S\times[0,1]\to S

403: \times[0,1]$ such that $h_0(x)=f(x)$ and $h_1(x)=g(x)$ for all $x\in S$. 

404: 

405: \indent

406: By the uniformization theorem, $\T(S)$ can be considered as the space of 

407: complete hyperbolic metrics on $S$ with finite area. The area of 

408: hyperbolic surface $(S,\s)$ is equal to $-2\pi\chi(S)$, where $\chi(S)=2(1-g)$ 

409: is the Euler characteristic of $S$ (see \cite{BP} for details).

410: 

411: \indent

412: As usual, two closed curves $\alpha,\beta:[0,1]\to S$ are called 

413: {\em free homotopic} if there exists a continuous mapping 

414: $F:[0,1]\times[0,1]\to S$ such that 

415: $$F(t,0)=\alpha(t),\ F(t,1)=\beta(t)\quad\mbox{and}\quad F(0,s)=F(1,s)$$

416: for all $t,s\in [0,1]$. The following well-known lemma will be used frequently 

417: in this thesis (see \cite{BP} \S B.4 for a proof).

418: \begin{lem}\label{lem;gloop}

419: Suppose that $(S,\s)$ is a hyperbolic surface. Then each free homotopy class of

420: closed curves contains a unique geodesic representative.  

421: \end{lem}

422: \noindent

423: Two closed curves $\alpha,\beta:[0,1]\to S$ are called {\em isotopic} if there 

424: exists a diffeomorphism $G:S\times[0,1]\to S\times[0,1]$ such that for all 

425: $x\in S$ and $t,s\in [0,1]$,

426: $$G(x,t)=(g_t(x),t),\quad

427: g_0(x)=x\quad\mbox{and}\quad g_1(\alpha(s))=\beta(s).$$

428: Free homotopic simple curves on a connected surface are isotopic, too.

429: 

430: \indent

431: There is well-known classification of $\mathit{Isom}{\,\mb{H}}^n$ into 

432: elliptic, parabolic and hyperbolic isometries. 

433: We write $\overline{\mb{H}}^2$ to denote the compactification of $\mb{H}^2$ 

434: by the circle at infinity. The following two lemmas will be useful in Section

435: \ref{sec;tnumber} (see \cite{BP} \S B.4 for a proof).

436: \begin{lem}\label{lem;hisometry}

437: Every non-trivial elements of $\pi_1(S)$ are hyperbolic isometry.

438: \end{lem}   

439: \begin{lem}\label{lem;lift}

440: Suppose that $\alpha$ is a non-trivial simple closed geodesic in $S$. Then

441: any two lifts of $\alpha$ into $\mb{H}^2$ can not meet in the whole

442: $\overline{\mb{H}}^2$.  

443: \end{lem}   

444: \noindent

445: The following lemma will be useful in Chapter \ref{chap;pfmainthm} 

446: (see \cite{BP} \S B.4 for a proof).

447: \begin{lem}\label{lem;disjoint}

448: Suppose that $\alpha$ and $\beta$ are non-intersecting, non-isotopic and 

449: non-trivial simple closed curves in a hyperbolic surface $(S,\s)$. Then the 

450: geodesic representatives of $\alpha$ and $\beta$ are non-intersecting.  

451: In particular, if $\alpha$ is in a subsurface $F$ of $S$, then the geodesic

452: representative of $\alpha$ is in $F$, too.

453: \end{lem}   

454: 

455: \indent

456: The standard boundary $S^{n-1}_\infty$ of $\mb{H}^n$ is defined by the 

457: equivalence classes of geodesic rays (see \cite{BP} \S A.5). 

458: The Teichm\"uller boundary of $\T(S)$ is defined in the same way. 

459: But there are distinct geodesic rays from a point in $\T(S)$ that always remain

460: within a bounded distance of each other (see \cite{masur1}). 

461: 

462: \indent

463: The mapping class group $\MCG(S)$ is the group of orientation preserving 

464: diffeomorphisms of $S$ modulo those which are isotopic to identity, i.e.

465: $$\MCG(S)=\mbox{Diff}^+(S)/\mbox{Diff}^0(S).$$

466: The mapping class group acts by isometries on $\T(S)$ and its quotient space is

467: called {\em moduli space} of $S$. See \cite{maskit3} for a picture of 

468: moduli space. Kerckhoff proved in his thesis that the action of $\MCG(S)$ does 

469: not in general extend to Teichm\"uller boundary.

470:  

471: \indent

472: Let $\mathcal{V}(S)$ be the set of representations of $\pi_1(S)$ into 

473: $PSL(2,\mathbb{C})$ up to conjugacy with compact-open topology. The product

474: $\T(S)\times\T(S)$ can be identified with an open subset of $\mathcal{V}(S)$, 

475: consisting of faithful representations whose images are quasi-Fuchsian groups, 

476: by Bers simultaneous uniformization. If we fix the first factor, then we get

477: a holomorphic embedding of $\T(S)$ into $\mathcal{V}(S)$ which is called

478: {\em a Bers slices}. Although this embedding depends on the fixed first factor,

479: there is a biholomorphic mapping between any two slices. 

480: The closure of this slice is compact, and is called {\em a Bers 

481: compactification} (see \cite{bers1,mcmullen}). 

482: But Kerckhoff and Thurston showed that (see Theorem 1 and Theorem 2 of 

483: \cite{KT})

484: \begin{enumerate}

485: \item

486: For each genus $g\geq 2$, there are Bers slices for which the canonical 

487: homeomorphisms do not extend to homeomorphisms on their compactifications. 

488: \item

489: For $g=2$, there is a Bers slice for which the action of the mapping class 

490: group does not extend continuously to its compactification.

491: \end{enumerate}  

492: 

493: \indent

494: In the famous 1976 preprint, which is published in \cite{thurston5} later, 

495: Thurston introduced the space of projective measured laminations on $S$, which 

496: will be denoted by $\PML(S)$, and a compactification of $\T(S)$ whose boundary 

497: is equal to $\PML(S)$. 

498: Thurston boundary $\PML(S)$ is a natural boundary of $\T(S)$, in the sense

499: that the action of mapping class group extends continuously to the Thurston 

500: compactification $\TC(S)=\T(S)\cup\PML(S)$. Masur(\cite{masur1}) showed that 

501: Teichm\"uller boundary and Thurston boundary are same almost everywhere, but 

502: not everywhere.

503: 

504: \indent

505: With this compactifcation, Thurston classified surface diffeomorphisms as 

506: periodic, reducible or pseudo-Anosov, which is a generalization of the 

507: well-known classification of elements of $SL(2,\Z)$. 

508: Thurston boundary was also used by Kerckhoff to solve the Nielsen realization 

509: problem, i.e. {\em every finite subgroup of $\MCG(S)$ can be realized as a 

510: group of isometries of some hyperbolic structure on $S$} 

511: (see \cite{kerckhoff2} for the proof).  

512: 

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

514: \subsection{Measured laminations}

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

516: 

517: In this section we study measured laminations. See \cite{bon4,CEG,CB,hatcher,

518: levitt} and \cite{thurston1} as references. 

519: Consider a fixed hyperbolic structure $\s$ on $S$. A {\em geodesic lamination} 

520: $\mu$ is a closed subset of $S$, which is a disjoint union of simple geodesics 

521: which are called leaves of $\mu$. The leaves of a geodesic 

522: lamination are complete, i.e. each leaf is either closed or has infinite 

523: length in both of its ends, and a geodesic lamination is determined by its 

524: support, i.e. a geodesic lamination is a union of geodesics in just one way.  

525: Using $S^1_\infty$, a geodesic lamination on $(S,\s)$ can be naturally related 

526: to a geodesic lamination on $(S,\s')$ for any $\s'\in\T(S)$. 

527: We write $\GL(S)$ to denote the space of geodesic laminations on $S$, which is

528: equipped with the Hausdorff metric on closed subsets.

529: Note that $\GL(S)$ is compact and therefore, in particular, every infinite 

530: sequence of nontrivial simple closed geodesics has a convergent subsequence.

531: 

532: \indent

533: For an arbitrary topological space $X$, the Chabauty topology on the set of 

534: closed subsets of $X$ has the following sub-bases.

535: \begin{enumerate}

536: \item[(i)]

537: $O_1(K)=\{A\mid A\cap K=\emptyset\}$ where $K$ is compact.

538: \item[(ii)]

539: $O_2(U)=\{A\mid A\cap U\neq\emptyset\}$ where $U$ is open.

540: \end{enumerate}

541: 

542: \noindent

543: If $X$ is compact and metrizable, in particular for $S$, the Chabauty topology 

544: agrees with the topology induced by the Hausdorff metric. The following lemma 

545: will turn out to be useful (see \cite{CEG} \S 3.1). 

546: \begin{lem}[Geometric Convergence]\label{lem;gc}

547: Suppose that $X$ is a locally compact metric space. A sequence $A_n$ of closed 

548: subsets of $X$ converges to a closed subset $A$ in Chabauty topology if and

549: only if 

550: \begin{enumerate}

551: \item[(i)]

552: If $x_{n_k}\in A_{n_k}$ converges to $x\in X$ then $x\in A$.

553: \item[(ii)]

554: If $x\in A$, then there exists a sequence $x_n\in A_n$ which converges to $x$. 

555: \end{enumerate}

556: \end{lem}

557: 

558: \indent

559: In $\mb{H}^2$, a geodesic is determined by an element of the open M\"obius band

560: $$M=(S^1_\infty\times S^1_\infty-\Delta)/\mb{Z}_2,$$

561: where $\Delta=\{(x,x)\}$ is the diagonal and $\mb{Z}_2$ acts by interchanging 

562: coordinates. A geodesic in $\mb{H}^2$ projects to a simple geodesic on $S$ if 

563: and only if the covering translates of its pairs of end points never strictly 

564: separate each other. 

565: Notice that a geodesic lamination could be considered as a closed subset of 

566: $M$. The Chabauty topology on $\GL(S)$ as closed subsets of $M$ is equivalent 

567: to the Chabauty topology on $\GL(S)$ as closed subsets of $\mb{H}^2$.

568: Therefore

569: \begin{lem}\label{lem;HCtopology}

570: If $\mu_i$ converges to $\mu$ in $\GL(S)$ with Hausdorff metric topology, then

571: for any geodesic $\el\subset\mu$, there exist geodesics $\el_i\subset\mu_i$ 

572: which converge to $\el$.

573: \end{lem}

574: 

575: \indent

576: A geodesic lamination is called {\em maximal} if each complementary 

577: region is isometric to an ideal triangle. A nonempty geodesic lamination is 

578: called {\em minimal} if no proper subset is a geodesic lamination. 

579: For example, any simple closed geodesic is a minimal lamination.

580: The following lemma is about the structure of minimal laminations 

581: (see \cite{CEG} \S 4.2 for a proof).   

582: \begin{lem}[Structure of minimal lamination]

583: If $\mu$ is a minimal lamination then either $\mu$ is a single geodesic or

584: consists of uncountable leaves. 

585: \end{lem}

586: 

587: \noindent

588: The following theorem is about the structure of geodesic laminations.   

589: \begin{SGL}[Structure of Geodesic Lamination] 

590: A geodesic lamination on $S$ is the union of finitely many minimal 

591: sublaminations and of finitely many infinite isolated leaves whose ends spiral

592: along the minimal sublaminations. 

593: \end{SGL}

594: 

595: \begin{figure}[ht]

596: \begin{center}

597: \psfrag{a}{$\alpha$}

598: \psfrag{b}{$\beta$}

599: \includegraphics[width=3in,height=1.5in]{tbdry1.eps}

600: \end{center}

601: \caption{$\alpha$ and $\beta$ have same measure}

602: \label{fig;tb1}

603: \end{figure}

604: 

605: \indent

606: A transverse measure on a geodesic lamination $\mu$ is a rule, which assigns to

607: each transverse arc $\alpha$ a measure that is supported on $\mu\cap\alpha$, 

608: which is invariant under a map from $\alpha$ to another arc $\beta$ if it 

609: takes each point of intersection of $\alpha$ with a leaf of $\mu$ to a point 

610: of intersection $\beta$ with the same leaf (see Figure \ref{fig;tb1}).

611: A measured lamination on $S$ is a geodesic lamination $\mu$ with a transverse 

612: measure of full support, i.e. if $\alpha\cap\mu\neq\emptyset$ then $\alpha$ 

613: has nonzero measure for any transverse arc $\alpha$. For example, a simple 

614: closed geodesic equipped with counting measure is a measured lamination. 

615: We write $\ML(S)$ to denote the space of measured laminations on $S$.

616: There is a natural action of $\R^+$ on $\ML(S)$. Suppose that $r>0$. 

617: The measured lamination $r\mu$ has the same geodesic lamination as $\mu$ with 

618: the transverse measure scaled by $r$. 

619: We write $\PML(S)$ to denote the set of equivalence classes of projective 

620: measured laminations.

621: 

622: \indent

623: The support of a measured lamination has no infinite isolated leaves,  

624: therefore by Theorem \ref{thm;SGL}, it is a finite disjoint union of minimal 

625: sublaminations. 

626: 

627: \indent

628: For a transverse arc $\alpha$ to a measured lamination $\mu$, let $\theta$ be 

629: the angle between leaves of $\mu$ and $\alpha$, measured counterclockwise from

630: $\alpha$ to $\mu$. The {\em total angle} of $\alpha$ is defined by

631: $$\theta(\alpha,\mu)=\int_\alpha\theta\, d\mu.$$

632: Thurston gave $\ML(S)$ a topology with following basis, 

633: \begin{eqnarray*}

634: &&\mathcal{B}(\mu,\alpha_1,\cdots,\alpha_n,\epsilon)=\{\nu\in\ML(S):\\

635: &&\qquad\qquad\qquad |\left(\mu(\alpha_k),\theta(\alpha_k,\mu)\right)-

636: \left(\nu(\alpha_k),\theta(\alpha_k,\nu)\right)|<\epsilon,\ k=1,\cdots, n\},

637: \end{eqnarray*} 

638: where $\{\alpha_k\}$ is a finite set of transverse arcs to $\mu$, and 

639: $\epsilon>0$. The following theorem was proved by Thurston.  

640: \begin{thm}[Thurston]

641: \begin{itemize}

642: \item[(1)]

643: $\ML(S)$ is homeomorphic to the open ball $B^{6g-6}$ and $\PML(S)$ is 

644: homeomorphic to the sphere $S^{6g-7}$.

645: \item[(2)]

646: $\R\times\C_0(S)$ is dense in $\ML(S)$, and $C_0(S)$ is dense in $\PML(S)$. 

647: \end{itemize}

648: \end{thm}   

649: 

650: \indent

651: Suppose that $\alpha,\beta\in\C_0(S)$. The {\em geometric intersection number} 

652: $i(\alpha,\beta)$ is the minimal number of intersections of any two their 

653: representatives.

654: 

655: \indent

656: For a transverse arc $\alpha$ to $\mu\in\ML(S)$, we write $\int_\alpha\, d\mu$ 

657: to denote integration of the transverse measure over $\alpha$. 

658: For a simple closed curve $\gamma$, let 

659: $$i(\mu,\gamma)=\inf_{\gamma'}\int_{\gamma'}d\mu,$$ 

660: where the infimum is taken over all the simple closed curves $\gamma'$ which is

661: homotopic to $\gamma$. For a general transverse arc $\alpha$, we define 

662: $$i(\mu,\alpha)=\inf_{\alpha'}\int_{\alpha'}d\mu,$$   

663: where the infimum is taken over all the arcs $\alpha'$ which is homotopic to 

664: $\alpha$ with endpoints fixed. Note that, in both cases, the infimum is 

665: realized by the unique geodesic in the corresponding homotopy class. 

666: 

667: \indent

668: Suppose that $\mu\in\ML(S)$, $\gamma\in\C_0(S)$ and $r\in\R^+$, let

669: $i(\mu,r\gamma)=ri(\mu,\gamma)$. 

670: The following theorem of Thurston is useful in Chapter \ref{chap;ct} 

671: (see \cite{thurston4} for a proof).   

672: 

673: \begin{thm}[Continuity of the intersection number]

674: The intersection number $i$ extends to a continuous symmetric function on 

675: $\ML(S)\times\ML(S)$.   

676: \end{thm}   

677: 

678: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

679: \subsection{Topology of $\TC(S)$}

680: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

681: 

682: In this section we study the topology of the Thurston compactification 

683: $\TC(S)$. References are \cite {FLP,papa} and \cite{thurston4}. The topology 

684: on $\TC(S)=\T(S)\cup\PML(S)$ is determined by the following two properties.

685: \begin{enumerate}

686: \item[\textbf{(P1)}]

687: $\T(S)$ is open in $\TC(S)$.

688: \item[\textbf{(P2)}]

689: $\s_i\in\T(S)$ converge to $[\lam]\in\PML(S)$ if and only if, for all simple 

690: closed curves $\alpha$, $\beta$ on $S$ with $i(\beta,\lam)\neq 0$, 

691: $${{\el_{\s_i}(\alpha)}\over {\el_{\s_i}(\beta)}}\quad\mbox{converges to}

692: \quad {{i(\alpha,\lam)}\over {i(\beta,\lam)}}\ ,$$ 

693: where $\el_{\s_i}(\alpha)$ is the length of closed $\s_i$-geodesic which is 

694: homotopic to $\alpha$.     

695: \end{enumerate}

696: 

697: \indent

698: The following trivial lemma will turn out to be useful (see \cite{MP} for a

699: proof). 

700: \begin{lem}

701: For any infinite sequence of distinct simple closed curves in $S$, there is a

702: subsequence $\alpha_i$ and $c_i>0$ such that 

703: $$c_i\to 0\quad\mbox{and}\quad c_i\alpha_i\to\mu\quad\mbox{for some }

704: \mu\in\ML(S)-\{0\}.$$     

705: \end{lem}

706: 

707: \indent

708: Suppose that $\mu_i\in\ML(S)$. We write $\mu_i\to\infty$ to denote that there 

709: exists $\alpha\in\C_0(S)$ such that $i(\alpha,\mu_i)$ converges to $\infty$.

710: The following theorem is the most useful theorem in this thesis.

711: 

712: \begin{thm}[Theorem 2.2 of \cite{thurston4}]\label{thm;useful}

713: A sequence $\s_i\in\T(S)$ converges to $[\lam]\in\PML(S)$ if and only if 

714: there is a sequence $\mu_i\in\ML(S)$ converging projectively to $\lam$ such 

715: that $\mu_i\to\infty$ and $\el_{\s_1}(\mu_i)\to\infty$ but $\el_{\s_i}(\mu_i)$ 

716: remains bounded, and for all $\nu\in\ML(S)$, there exists a constant $C>0$ 

717: such that 

718: $$i(\nu,\mu_i)\leq \el_{\s_i}(\nu)\leq i(\nu,\mu_i)+C \el_{\s_1}(\nu).$$

719: \end{thm}

720: 

721: \noindent

722: In particular, for each $\gamma\in\C_0(S)$, there exists a constant $\Gamma>0$,

723: which does not depend on $i$ such that 

724: \begin{equation}\label{eq;papa}

725: i(\gamma,\mu_i)\leq \el_{\s_i}(\gamma)\leq i(\gamma,\mu_i)+\Gamma.

726: \end{equation}

727: 

728: \indent

729: The measured laminations $\mu_i$ were constructed in \cite{FLP}, \cite{papa} 

730: and \cite{thurston3} \S 9. In these constructions, the associated measured 

731: foliations are constructed first, and the measured laminations $\mu_i$ are 

732: induced later. 

733: In the following paragraphs, we study measured foliations and the construction 

734: of $\mu_i$ following Papadopoulos (see \cite{papa} for details). 

735: Another good French reference for this construction is $\cite{FLP}$. 

736: 

737: \indent

738: A measured foliation $F$ on $S$ is a foliation with finite number of 

739: singularities equipped with a invariant transverse measure, i.e. $F$ is 

740: determined by a finite number of points $p_k\in S$ and an atlas of coordinate 

741: neighborhoods $$(x_i,y_i):U_i\to\R^2$$ on the complement of $\{p_k\}$ such that

742: $x_j=f_{ij}(x_i,y_i)$ and $y_j=\pm y_i+C$ for any overlapping coordinate 

743: neighborhoods $(x_j,y_j)$, where $C$ is a constant and the transverse measure

744: is $dy$. The singularities have $p$-pronged saddles with $p\geq 3$.   

745: Suppose that $F$ is a measured foliation on $S$ and $\alpha\in\C_0(S)$. Let

746: $$i(F,\alpha)=\inf_{\alpha'}\int_{\alpha'}|dy|,$$

747: where the infimum is taken over all the representatives $\alpha'$ in the class 

748: $\alpha$. Two measured foliations $F$ and $G$ are called {\em equivalent} if 

749: $i(F,\alpha)=i(G,\alpha)$ for all $\alpha\in\C_0(S)$. We write $\MF(S)$ to 

750: denote the set of equivalence classes of measured foliations.

751: Measured foliations and measured laminations are related by the following 

752: theorem (see \cite{levitt} for the proof).

753: 

754: \begin{thm}\label{thm;FL}

755: There is a homeomorphism $h:\MF(S)\to\ML(S)$ which is identity on 

756: $\R\times\C_0(S)$ and preserves the intersection number.

757: \end{thm} 

758: 

759: \indent

760: Suppose that $\mu$ is a maximal geodesic lamination on $S$ and $\s\in\T(S)$. 

761: Thurston constructed a measured foliation $F_\mu(\s)$, which is called 

762: {\em horocyclic foliation} as follows.

763: Since $\mu$ is maximal, the complementary components of $\mu$ are all isometric

764: to ideal triangles. In each of these components, define a partial foliation,

765: i.e. a foliation whose support is subsurface, whose leaves are intersections of

766: the triangle and horocycle centered at vertices of the triangle

767: (see Figure \ref{fig;tb2}).

768: 

769: \begin{figure}[ht]

770: \begin{center}

771: \includegraphics[width=2in,height=2in]{tbdry2.eps}

772: \end{center}

773: \caption{The partial foliation}

774: \label{fig;tb2}

775: \end{figure}

776:     

777: \noindent

778: Notice that the horocycle meets the triangle with right angles, and the 

779: non-foliated region is equal to a little triangle whose edges are subarcs of

780: horocycles which meet tangentially at their endpoints. These partial foliations

781: in the ideal triangles fit together on the surface and define a partial 

782: foliation on $S$. The transverse measure on this partial foliation is uniquely 

783: determined by the fact that on the leaves of $\mu$ this transverse measure is 

784: equal to the hyperbolic distance.  

785: 

786: \indent

787: Suppose that $\mu$ is a maximal geodesic lamination on $S$. Let $\MF(\mu)$ be

788: the subset of $\MF(S)$ consisting of equivalence classes which has 

789: representative transverse to $\mu$. In \cite{thurston3} \S 9, Thurston showed 

790: \begin{thm}

791: The map $\phi_\mu:\T(S)\to\MF(\mu)$ such that $\phi_\mu(\s)=F_\mu(\s)$

792: is a homeomorphism.

793: \end{thm}

794: 

795: \noindent  

796: Suppose that $\s_i\in\T(S)$ converges to $[\lam]\in\PML(S)$ in $\TC(S)$. 

797: Then the measured lamination $\mu_i$, obtained from $F_\mu(\s_i)$ by 

798: Theorem \ref{thm;FL}, is the lamination in Theorem \ref{thm;useful}. 

799: 

800: 

801: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

802: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

803: \section{Complex of Curves}\label{chap;gbdry} 

804: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

805: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

806: 

807: 

808: In this chapter we study the complex of curves and its Gromov boundary.

809: All theories in this chapter work for any orientable surface of finite type, 

810: i.e. surface with genus $g$ and $n$ punctures. Throughout this chapter, 

811: we write $\Sigma=\Sigma_{g,n}$ to denote an orientable surface with genus $g$ 

812: and $n$ punctures. As before, we write $\MCG(\Sigma)$ to denote the mapping 

813: class group of $\Sigma$, and $\T(\Sigma)$ the Teichm\"uller space of $\Sigma$. 

814:  

815: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

816: \subsection{Complex of curves}\label{sec;complex of curves}

817: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

818: 

819: In \cite{harvey}, Harvey introduced the complex of curve $\C(\Sigma)$ to study 

820: the action of $\MCG(\Sigma)$ at the infinity of $\T(\Sigma)$. This complex 

821: encodes the asymptotic geometry of Teichm\"uller space, similarly as the 

822: Tits buildings for symmetric spaces. Let $\mc{S}(\Sigma)$ be the set of isotopy

823: classes of essential, unoriented, non-boundary parallel simple closed curves in

824: $\Sigma$. 

825: The vertices of $\C(\Sigma)$ are elements of $\mc{S}(\Sigma)$, i.e.

826: $\C_0(\Sigma)=\mc{S}(\Sigma)$, and the $k$-simplices of $\C(\Sigma)$ are  

827: subsets $\{\alpha_1,\cdots,\alpha_{k+1}\}$ of $\mc{S}(\Sigma)$ with mutually 

828: disjoint representatives. Notice that $\C(\Sigma)$ is empty if $g=0$ and 

829: $n\leq 3$. The maximal dimension of simplices is called the {\em dimension} of 

830: $\C(\Sigma)$, and it is equal to $3g+n-4$.

831: Masur and Minsky defined a metric on $\C(\Sigma)$ by making each simplex 

832: regular Euclidean with side length 1 and taking shortest-path metric. 

833: Let $d_\C$ denote this metric on $\C(\Sigma)$. 

834: 

835: \indent

836: The mapping class group acts on $\C(\Sigma)$ and Ivanov proved that, if 

837: $g\geq 2$ then all automorphisms of  $\C(\Sigma)$ are given by elements of

838: $\MCG(\Sigma)$ (see \cite{ivanov} for the proof, and \cite{korkmaz} 

839: for the related work of Korkmaz). Luo(\cite{luo3}) generalized this result 

840: by showing that, if $3g+n-4\geq 1$ and $(g,n)\neq (1,2)$, then all 

841: automorphisms of $\C(\Sigma)$  are given by elements of $\MCG(\Sigma)$. 

842: In \cite{harer}, Harer showed that $\C(\Sigma_{g,n})$ is homotopic to a 

843: wedge of spheres of dimension $r$, where 

844: \begin{eqnarray*}

845: r=\left\{

846: \begin{array}{lll}

847: 2g+n-3 &\mbox{ if }& g>0\mbox{ and }n>0\\ 

848: 2g-2 &\mbox{ if }& n=0\\ 

849: n-4 &\mbox{ if }& g=0.  

850: \end{array}

851: \right.

852: \end{eqnarray*}

853:             

854: \indent

855: If $\Sigma$ is a torus, once-punctured torus or 4-times punctured sphere, then 

856: any two essential simple closed curves intersects, i.e. there is no edge in 

857: $\C(\Sigma)$.

858: Masur and Minsky introduced a new definition for these cases so that it has 

859: edges. In this definition, $\{\alpha,\beta\}$ is an edge if $\alpha\neq\beta$, 

860: and $\alpha$ and $\beta$ have the lowest possible intersection number. 

861: For the tori this is 1, and for 4-holed sphere this is 2 (see \cite{HT} for 

862: Hatcher-Thurston complex which is related to this definition).

863: 

864: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

865: \subsection{Relative twist number in annulus complex}\label{sec;tnumber}

866: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

867: 

868: In this section we study the relative twist number in annulus complex, which 

869: was introduced by Minsky in \cite{FLM}, of two simple closed curves around a 

870: fixed simple closed curve. See \cite{FLM} \S2.1, \cite{MM2} \S2.4 and 

871: \cite{m-ending} \S4 as references. 

872: An annular domain in $\Sigma$ is an annulus with incompressible boundary.

873: A complex, which is called an {\em annulus complex}, is defined for such annuli

874: to keep track of Dehn twisting around their cores.

875: 

876: \indent

877: Consider an oriented annulus $Y=S^1\times [0,1]$. We write $\A_0(Y)$ to denote 

878: the set of arcs joining $S^1\times\{0\}$ to $S^1\times\{1\}$, up to homotopy 

879: with endpoints fixed. As in $\C(\Sigma)$, we put an edge between any 

880: two elements of $\A_0(Y)$ which have representatives with disjoint interiors, 

881: and define the annular complex $\A(Y)$ as for the complex of curves. 

882: We also make $\A(Y)$ a metric space with edge length 1 as in the curve complex.

883: Let $d_Y$ denote the path-metric.

884: 

885: \vspace{.5cm}  

886: \begin{figure}[ht]

887: \begin{center}

888: \psfrag{a}{$a$}

889: \psfrag{b}{$b$}

890: \includegraphics[width=2.5in,height=1in]{ccvs1.eps}

891: \end{center}

892: \caption{$|a\cdot b|=2$}

893: \label{fig;cc1}

894: \end{figure}

895: 

896: \indent

897: Suppose that $a,b\in\A_0(Y)$ and they do not share any endpoints. Notice that 

898: $a$ and $b$ inherit orientations from the orientation of $[0,1]$. Therefore we 

899: can define the {\em algebraic intersection number} $a\cdot b$ (for example, 

900: see Figure \ref{fig;cc1}). Let $a\cdot a=0$.

901: Consider a lift of $a\in\A_0(Y)$ to the covering space 

902: $\widetilde{Y}=\R\times[0,1]$ which has endpoints $(a_0,0)$ and $(a_1,1)$. 

903: Notice that these endpoints are determined by $a$, up to $\Z$, and 

904: $$a\cdot b=\lfloor b_1-a_1\rfloor-\lfloor b_0-a_0\rfloor, $$

905: where $\lfloor x\rfloor$ denotes the largest integer less than or equal to $x$.

906: It follows that 

907: \begin{equation}\label{eq;dot}

908: a\cdot c=a\cdot b+b\cdot c+\Delta\quad\mbox{with }\Delta\in\{0,1,-1\}

909: \end{equation}

910: for all $a,\ b,\ c\in\A_0(Y)$ such that the intersection numbers are defined. 

911: With an inductive argument, we can also check  

912: \begin{equation}\label{eq;dotdist}

913: d_Y(a,b)=1+|a\cdot b|\quad\mbox{for all distinct }a,b\in\A_0(Y). 

914: \end{equation}

915: Fix $a\in\A_0(Y)$. From eq. (\ref{eq;dot}) and eq. (\ref{eq;dotdist}), we can 

916: show that the map $f:\A_0(Y)\to\Z$ with $f(b)=a\cdot b$ is a 

917: quasi-isometry. Thus $\A(Y)$ is quasi-isometric to $\Z$. 

918: 

919: \indent

920: For a fixed finite generating set of $\MCG(\Sigma)$, let

921: $||\cdot ||$ be the minimal word length with respect to these generators. 

922: Masur and Minsky introduced the relative twist number, and Farb, Lubotzky and 

923: Minsky proved that every Dehn twist has linear growth in $\MCG(\Sigma)$.

924:     

925: \begin{thm}[Theorem 1.1 of \cite{FLM}]\label{thm;FLM}

926: For all Dehn twist $t$, there exists a constant $c>0$ such that 

927: $||t^m||\geq c|m|$ for all $m$.

928: \end{thm}

929: 

930: \begin{figure}[ht]

931: \begin{center}

932: \psfrag{a}{$a$}

933: \psfrag{b}{$b$}

934: \psfrag{Y}{$Y_\alpha$}

935: \psfrag{H}{$\overline{\mb{H}}^2$}

936: \includegraphics[width=2in,height=2.5in]{ccvs2.eps}

937: \end{center}

938: \caption{$Y_\alpha$}

939: \label{fig;cc2}

940: \end{figure}

941: 

942: \indent 

943: The relative twist number is defined as follows. For a fixed essential simple 

944: closed curve $\alpha$ in $\Sigma$, let $g_\alpha$ be an isometry of $\mb{H}^2$ 

945: representing the conjugacy class of $\alpha$. Let

946: $$Y=Y_\alpha=\left(\overline{\mb{H}}^2\setminus\mbox{Fix}(g_\alpha)\right)/

947: <g_\alpha>,$$

948: where $\overline{\mb{H}}^2$ is the closed-disk compactification of  

949: hyperbolic plane $\mb{H}^2$. See Figure \ref{fig;cc2}, where 

950: $\{a,b\}=\mbox{Fix}(g_\alpha)$. Notice that $Y$ is a 

951: closed annulus and a neighborhood of $\alpha$ lifts homeomorphically into $Y$.

952: Suppose that $\beta$ is an essential simple closed curve in $\Sigma$ such that 

953: $i(\alpha,\beta)\neq 0$. Notice that any lift of $\alpha$ into $Y$ does not 

954: share endpoints with a lift of $\beta$. Therefore any lift of $\beta$

955: extends to a properly embedded arc in $Y$. We write $\mb{lift}_\alpha(\beta)$

956: to denote the set of lifts of $\beta$ into $Y$ which connect the two boundaries

957: of $Y$ as elements in $\A_0(Y)$. Let $\gamma$ be another essential simple 

958: closed curve in $\Sigma$ with $i(\alpha,\gamma)\neq 0$. If $\beta$ and $\gamma$

959: are different, then $b$ and $c$ do not share endpoints for all 

960: $b\in\mb{lift}_\alpha(\beta)$ and $c\in\mb{lift}_\alpha(\gamma)$. 

961: The relative twist number is defined by  

962: $$\tau_\alpha(\beta,\gamma)=\left\{b\cdot c\mid b\in\mb{lift}_\alpha(\beta)\

963: \mbox{and }c\in\mb{lift}_\alpha(\gamma)\right\}.$$

964: 

965: \indent

966: From eq. (\ref{eq;dot}), we have $\mbox{diam}(\tau_\alpha(\beta,\gamma))

967: \leq 2$. In \cite{FLM}, Farb, Lubotzky and Minsky used the following equations 

968: to prove Theorem \ref{thm;FLM}.

969: \begin{enumerate}

970: \item[\textbf{(i)}]

971: If $t=T_\alpha$ is the leftward Dehn twist on $\alpha$, then 

972: $\tau_\alpha(\beta,t^n(\beta))\subset\{n,n+1\}$.

973: \item[\textbf{(ii)}]

974: If $\beta$ and $\gamma$ intersect $\alpha$, then their geometric intersection 

975: number bounds their relative twisting, i.e. 

976: $$\max|\tau_\alpha(\beta,\gamma)|\leq i(\beta,\gamma)+1.$$

977: \item[\textbf{(iii)}]

978: If $\beta$, $\gamma$ and $\delta$ intersect $\alpha$, then  

979: \begin{eqnarray*}

980: \max\tau_\alpha(\beta,\delta)\leq \max\tau_\alpha(\beta,\gamma)

981: +\max\tau_\alpha(\gamma,\delta)+2 \\

982: \min\tau_\alpha(\beta,\delta)\geq \min\tau_\alpha(\beta,\gamma)

983: +\min\tau_\alpha(\gamma,\delta)-2. 

984: \end{eqnarray*}

985: \end{enumerate}

986: 

987: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

988: \subsection{The theorem of Masur and Minsky, and of Klarreich}

989: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

990: 

991: In this section we study a theorem of Masur and Minsky, and of Klarreich. 

992: References are \cite{bowditch,hamenstadt,klarreich,MM} and \cite{m-geom}.

993: If $\Sigma$ is a torus, once-punctured torus or 4-times punctured sphere, then 

994: the complex of curves is the Farey graph (see Figure \ref{fig;cc3}). 

995: Minsky showed that $(\C(\Sigma),d_\C)$ is $3\over 2$-hyperbolic space for these

996: cases (see \cite{m-geom} \S 3 for the proof). For $\xi(\Sigma)=3g+n>4$, the 

997: following lemma gives an upper bound of $d_\C$. 

998: 

999: \begin{figure}[ht]

1000: \begin{center}

1001: \psfrag{0}{\tiny{0}}

1002: \psfrag{1/3}{\tiny{1/3}}

1003: \psfrag{1/2}{\tiny{1/2}}

1004: \psfrag{2/3}{\tiny{2/3}}

1005: \psfrag{1}{\tiny{1}}

1006: \psfrag{3/2}{\tiny{3/2}}

1007: \psfrag{2}{\tiny{2}}

1008: \psfrag{3}{\tiny{3}}

1009: \psfrag{1/0}{\tiny{1/0}}

1010: \psfrag{-1/3}{\tiny{-1/3}}

1011: \psfrag{-1/2}{\tiny{-1/2}}

1012: \psfrag{-2/3}{\tiny{-2/3}}

1013: \psfrag{-1}{\tiny{-1}}

1014: \psfrag{-3/2}{\tiny{-3/2}}

1015: \psfrag{-2}{\tiny{-2}}

1016: \psfrag{-3}{\tiny{-3}}

1017: \includegraphics[width=2.5in,height=2.5in]{ccvs3.eps}

1018: \end{center}

1019: \caption{Complex of curves of torus}

1020: \label{fig;cc3}

1021: \end{figure}

1022: 

1023: \begin{lem}[Lemma 1.1 of \cite{bowditch}]

1024: If $\xi(\Sigma)>4$ then $d_\C(\alpha,\beta)\leq i(\alpha,\beta)+1$.

1025: \end{lem} 

1026: \noindent

1027: Recall the classifications of elements of the mapping class group into 

1028: periodic, reducible and pseudo Anosov elements. An element $h\in\MCG(\Sigma)$ 

1029: is called {\em pseudo Anosov} if there exist $r>1$ and a pair of measured 

1030: foliations $\mc{F}^s$ and $\mc{F}^u$ such that

1031: $$h(\mc{F}^s)={1\over r}\mc{F}^s\quad\mbox{and}\quad h(\mc{F}^u)=r\mc{F}^u.$$ 

1032: If $(\C(\Sigma),d_\C)$ is a bounded metric space with upper bound $K$, then it 

1033: is trivially a $K$-hyperbolic space. 

1034: But from the following proposition, it is clear that 

1035: $\mbox{diam}(\C(\Sigma))=\infty$.

1036: \begin{prop}[Proposition 4.6 of \cite{MM1}]

1037: If $\xi(\Sigma)>4$ then there exists $c>0$ such that, for any pseudo-Anosov

1038: $h\in\MCG(\Sigma)$, $\gamma\in\mc{S}(\Sigma)$ and $n\in\Z$, we have

1039: $$d_\C(h^n(\gamma),\gamma)\geq c|n|.$$

1040: \end{prop}

1041: 

1042: \indent

1043: In \cite{MM1}, Masur and Minsky proved that $(\C(\Sigma),d_\C)$ is 

1044: $\delta$-hyperbolic for the case $\xi(\Sigma)=3g+n>4$, too. 

1045: \begin{thm}[Theorem 1.1 of \cite{MM}]\label{thm;MM}

1046: $\C(\Sigma_{g,n})$ is a $\delta$-hyperbolic space, where $\delta$ depends only 

1047: on $g$ and $n$.

1048: \end{thm} 

1049: \noindent

1050: Masur and Minsky used Teichm\"uller theory in the proof of Theorem 

1051: \ref{thm;MM}. In their proof, the constant $\delta$ is not constructive because

1052: it contains a compactness arguments on the spaces of quadratic differentials.

1053: In 2002, Bowditch proved Theorem \ref{thm;MM} in a more combinatorial way,

1054: and showed that the number $\delta$ is bounded by a logarithmic function

1055: of $3g+n-4$.    

1056: 

1057: \indent

1058: Since $\C(\Sigma)$ is $\delta$-hyperbolic, we can consider its Gromov boundary.

1059: For the case of torus, the boundary can be identified with the set of 

1060: irrational numbers. For general cases, it is clear that a sequence of curves 

1061: $\alpha_n$, whose distance from a fixed curve is going to infinity, must 

1062: converge to a maximal lamination. In fact, Klarreich showed that the Gromov 

1063: boundary of complex of curves is homeomorphic to the space of topological

1064: equivalence classes of filling lamination.  

1065: \begin{thm;klarreich}[Klarreich \cite{klarreich}]

1066: There is a homeomorphism $$k:\BC(\Sigma)\to\EL(\Sigma)$$ such that for any 

1067: sequence $\alpha_n$ in $\mc{S}(\Sigma)$, $\alpha_n$ converges to 

1068: $\alpha\in\BC(\Sigma)$ if and only if $\alpha_n$, considered as a subset of 

1069: $\UML(\Sigma)$, converges to $k(\alpha)$.

1070: \end{thm;klarreich}

1071: 

1072: \indent

1073: Klarreich used Teichm\"uller theory and the results of Masur and Minsky in

1074: \cite{MM}, to prove Theorem \ref{thm;klarreich}. It is clear that 

1075: $\BC(\Sigma)$ is homeomorphic to the Gromov boundary of its 1-skeleton 

1076: $\C_1(\Sigma)$ because they are quasi-isometric. 

1077: Suppose that $\epsilon>0$ satisfies the collar lemma. 

1078: For each $\alpha\in\C_0(\Sigma)$, let

1079: $$T(\alpha)=\{\s\in\T(\Sigma)\mid\ell_\s(\alpha)<\epsilon\}.$$

1080: Then a collection of sets $\T(\alpha_1),\cdots,\T(\alpha_n)$ has nonempty 

1081: intersection if and only if $\alpha_1,\cdots,\alpha_n$ form a simplex in 

1082: $\C(S)$. The set $\T_{el}(\Sigma)$ is defined from  $\T(\Sigma)$, by adding a 

1083: new point $P_\alpha$ for each set $T(\alpha)$ and an interval of length 

1084: $1\over 2$ from $P_\alpha$ to each point in $T(\alpha)$. 

1085: $\T_{el}(\Sigma)$ equipped with the minimal path-metric is called the 

1086: {\em relative Teichm\"uller space} following the terminology of 

1087: Farb(\cite{farb}).

1088: In \cite{MM}, Masur and Minsky showed that $\T_{el}(\Sigma)$ is quasi-isometric

1089: to $\C_1(\Sigma)$, and Klarreich showed that the Gromov boundary of

1090: $\T_{el}(\Sigma)$ is homeomorphic to the space of topological equivalence 

1091: classes of minimal singular foliations on $\Sigma$, which is homeomorphic to 

1092: $\EL(\Sigma)$.

1093: 

1094: 

1095: 

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

1097: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1098: \section{Proof of Main Theorem (Theorem \ref{mainthm})}

1099: \label{chap;pfmainthm}

1100: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1101: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1102: 

1103: Suppose that $\s_i\in\T(S)$ converges to $[\lam]\in\PML(S)$ in $\TC(S)$. 

1104: Recall the map 

1105: $$\Phi(\s_i)=\mbox{a pants decomposition whose total length is bounded by }L,$$

1106: where $L$ is a fixed Bers constant and all pants curves are geodesics in 

1107: $\s_i$. Recall also the quotient map $u:\PML(S)\to\UML(S)$ by forgetting 

1108: measure. Notice that $\Phi(\s_i)$ can be considered as a sequence in $\GL(S)$. 

1109: Since $\GL(S)$ is compact, it has a convergent subsequence. 

1110: In this chapter we prove our main theorem.  

1111: 

1112: \begin{mainthm}

1113: If $\Phi(\s_i)$ converge to $\nu\in\GL(S)$ in Hausdorff metric topology, then 

1114: $u([\lam])\subset\nu$. 

1115: \end{mainthm}

1116: 

1117: \noindent

1118: Consider the decomposition $u([\lam])=\lam_1\cup\lam_2\cup\cdots\lam_m$ as a 

1119: finite disjoint union of minimal laminations. To prove $\lam_j\subset\nu$ 

1120: for all $1\leq j\leq m$, it is enough to show that $\lam_1\subset\nu$ and 

1121: $\lam_2\subset\nu$, assuming that $\lam_1$ is a simple closed curve and 

1122: $\lam_2$ is not a simple closed curve.

1123: 

1124: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1125: \subsection{Proof of $\lam_1\subset\nu$}\label{sec;lam1}

1126: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1127: 

1128: Recall that $\lam_1$ is a simple closed curve. If $\lam_1\subset\Phi(\s_i)$ for

1129: infinitely many $i$, then it is clear that $\lam_1\subset\nu$. 

1130: If $\lam_1\subset\Phi(\s_i)$ for only finitely many $i$, then there exists 

1131: $N_1>0$ such that $\lam_1\not\subset\Phi(\s_i)$ for all $i>N_1$.

1132: Since $\Phi(\s_i)$ is a pants decomposition, there exists a pants curve 

1133: $\alpha_i$ in $\Phi(\s_i)$ such that $\alpha_i\cap\lam_1\neq\emptyset$ for all

1134: $i>N_1$. Choose $x_i\in\alpha_i\cap\lam_1$ and a limit point $x$ of $x_i$.

1135: By Lemma \ref{lem;gc}, there exists a leaf $\el$ of $\nu$ such that 

1136: $x\in\el$. If $\el=\lam_1$, we are done. 

1137: 

1138: To get a contradiction, suppose that $\el\neq\lam_1$. Choose an open 

1139: neighborhood $U$ of $x$ which is isometric to an open subset of $\mb{H}^2$ 

1140: (see Figure \ref{fig;pfm1}).

1141: 

1142: \begin{figure}[ht]

1143: \begin{center}

1144: \psfrag{r}{$\lam_1$}

1145: \psfrag{l}{$\el$}

1146: \psfrag{x}{$x$}

1147: \includegraphics[width=2in,height=2in]{pfm1.eps}

1148: \end{center}

1149: \caption{The open neighborhood $U$}

1150: \label{fig;pfm1}

1151: \end{figure}

1152: 

1153: \noindent

1154: Since $S$ is compact, by Lemma \ref{lem;HCtopology}, there exists a pants curve

1155: $\alpha_i$ in $\Phi(\s_i)$ such that an arc $\beta_i\subset\alpha_i$ 

1156: approaches to $\el\cap U$. Therefore there exists $N_2>0$ such that 

1157: $$i(\alpha_i,\lam)\geq \int_{\beta_i}d\lam_1=r>0\quad\mbox{for all }i>N_2,$$

1158: where $r>0$ is the transverse measure on $\lam_1$. 

1159: 

1160: \indent

1161: Choose $\mu_j\in\ML(S)$ which converges to $\lam$ in $\PML(S)$ as in 

1162: Theorem \ref{thm;useful}. Since $\mu_j\to\infty$, there exists a sequence 

1163: $c_j>0$ such that $c_j\mu_j$ converges to $\lam$ in $\ML(S)$ with 

1164: $\lim_{j\to\infty}c_j=0$. Notice that there exists $N_3>0$ which does not 

1165: depend on $i$ such that

1166: $$i(\alpha_i,c_j\mu_j)\geq {r\over 2}\quad \mbox{for all }i,j>N_3.$$

1167: Therefore $i(\alpha_i,\mu_j)$ approaches to $\infty$ as $i,j\to\infty$.

1168: But from eq. (\ref{eq;papa}), we have 

1169: $i(\alpha_i,\mu_i)\leq\el_{\s_i}(\alpha_i)\leq L$ for all $i$.

1170: This is a contradiction.

1171: 

1172: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1173: \subsection{Proof of $\lam_2\subset\nu$}

1174: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1175: 

1176: Recall that $\lam_2$ is not a simple closed curve. Let $F$ be the essential 

1177: subsurface of $S$ which is filled by $\lam_2$. Since $\Phi(\s_i)$ is a pants

1178: decomposition for all $i$, we have $\nu\cap F\neq\emptyset$. 

1179: To get contradictions, suppose that $\lam_2\not\subset\nu$ in the next two

1180: paragraphs. 

1181: 

1182: \indent

1183: Suppose that $\nu\cap\lam_2=\emptyset$. 

1184: Notice that $\nu\cup\lam_2$ is a geodesic lamination, too. 

1185: By Theorem \ref{thm;SGL}, we can decompose $\nu\cup\lam_2$ as a finite 

1186: disjoint union of minimal lamination, including $\lam_2$, and finite number of 

1187: infinite isolated leaves. Since $\lam_2$ is a filling lamination in $F$, 

1188: any isolated infinite leaf can not intersect $F$. Thus $\nu\cap F=\emptyset$. 

1189: This is a contradiction. 

1190: 

1191: \begin{figure}[ht]

1192: \begin{center}

1193: \psfrag{r}{$\lam_2$}

1194: \psfrag{l}{$\el$}

1195: \includegraphics[width=2in,height=2in]{pfm2.eps}

1196: \end{center}

1197: \caption{The open neighborhood $V$}

1198: \label{fig;pfm2}

1199: \end{figure}

1200: 

1201: \indent

1202: Suppose that $\nu\cap\lam_2\neq\emptyset$. There exists a leaf $\ell$ of

1203: $\nu$ which intersects $\lam_2$ transversely. Choose an open neighborhood $V$ 

1204: which is isometric to an open subset of $\mb{H}^2$, and in which $\el$ 

1205: intersects $\lam_2$ transversely (see Figure \ref{fig;pfm2}).

1206: As in Section \ref{sec;lam1}, there exists a pants curve 

1207: $\alpha_i$ in $\Phi(\s_i)$ and arc $\beta_i\subset\alpha_i$ such that

1208: $\beta_i$ converges to $\el\cap V$. Therefore there exist $r>0$ and $N>0$ 

1209: such that 

1210: $$i(\alpha_i,\lam)\geq\int_{\beta_i}d\lam_2=r>0\quad\mbox{for all }i>N.$$

1211: As in Section \ref{sec;lam1}, using Theorem \ref{thm;useful}, we can show 

1212: that this is a contradiction.

1213: 

1214: 

1215: 

1216: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1217: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1218: \section{Proof of Corollary \ref{cor;twist} and Theorem \ref{thm;surgery}}

1219: \label{chap;ct}

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

1221: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1222: 

1223: 

1224: In this chapter we prove Corollary \ref{cor;twist} and Theorem 

1225: \ref{thm;surgery}, and show that Theorem \ref{thm;filling} comes from

1226: Theorem \ref{thm;surgery}.

1227:  

1228: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1229: \subsection {Proof of Corollary \ref{cor;twist}}\label{sec;twist} 

1230: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1231: 

1232: Suppose that $\s_i\in\T(S)$ converges to $[\lam]\in\PML(S)$ in $\TC(S)$, and 

1233: let $u([\lam])= \lam_1\cup\lam_2\cup\cdots\cup\lam_m$ be the decomposition as a

1234: finite disjoint union of minimal laminations. 

1235: 

1236: \begin{figure}[ht]

1237: \begin{center}

1238: \psfrag{U}{$U$}

1239: \psfrag{r}{$\lam_1$}

1240: \psfrag{Y}{$Y$}

1241: \psfrag{S}{$S$}

1242: \psfrag{p}{$p$}

1243: \includegraphics[width=2.5in,height=2in]{ct1.eps}

1244: \end{center}

1245: \caption{The annular cover $Y$}

1246: \label{fig;ct1}

1247: \end{figure}

1248: 

1249: \noindent

1250: Suppose that $\lam_1$ is a simple

1251: closed curve. Suppose also that $\alpha_i$ is a pants curve in $\Phi(\s_i)$ and

1252: $\alpha_i\cap\lam_1\neq\emptyset$ for all $i$.  

1253: Construct an annular covering $Y$ of $S$ in which a neighborhood $U$ of 

1254: $\lam_1$ lifts homeomorphically (see Figure \ref{fig;ct1}). 

1255: We may assume that $U$ is a closed collar around $\lam_1$ and $U$ 

1256: does not intersect $\lam_j$ for all $j\neq 1$. 

1257: Let $a_i\in\mb{lift}(\alpha_i)$. In this section we prove Corollary 

1258: \ref{cor;twist}.

1259: 

1260: \begin{twistthm}

1261: $|a_1\cdot a_i|$ approaches to infinity as $i$ increases.

1262: \end{twistthm} 

1263: 

1264: \begin{proof}

1265: To get a contradiction, suppose that $|a_1\cdot a_i|$ does not approach to 

1266: $\infty$. Then there exists a subsequence of $a_i$, which we will call $a_i$ 

1267: again for the sake of simplicity, such that $|a_1\cdot a_i|=k$ for all $i$ 

1268: for some $k\in\N\cup\{0\}$. We may assume that $a_1\cdot a_i=k$ without loss

1269: of generality. 

1270: 

1271: \begin{figure}[ht]

1272: \begin{center}

1273: \psfrag{H}{${\overline{\mb{H}}}^2$}

1274: \psfrag{p}{$p$}

1275: \psfrag{U}{$U$}

1276: \psfrag{K}{$K$}

1277: \psfrag{b}{$\beta_i$}

1278: \includegraphics[width=1.8in,height=3in]{ct2.eps}

1279: \end{center}

1280: \caption{A lift of $\beta_i$ in $K$}

1281: \label{fig;ct2}

1282: \end{figure}

1283: 

1284: \noindent

1285: Suppose that $i\geq 3$. From eq. (\ref{eq;dot}), we have

1286: $a_1\cdot a_i=a_1\cdot a_2+a_2\cdot a_i+\Delta$, where $\Delta\in\{-1,0,1\}$. 

1287: Therefore 

1288: $$ |a_2\cdot a_i|\leq 1\quad\mbox{for all }i\geq 3.$$

1289: This equation implies that $\alpha_i=p(a_i)$ does not change so much in $U$  

1290: for $i\geq 2$. Let $\beta_i$ be a component of $\alpha_i\cap U$ (see Figure 

1291: \ref{fig;ct2}). 

1292: We can find a compact set $K\subset {\mb{H}}^2$ such that there 

1293: exists a lift of $\beta_i$ in $K$ for all $i$. A geodesic arc in $\mb{H}^2$ is 

1294: determined by its two endpoints. Therefore there exists a subsequence 

1295: of $\beta_i$, which we will call $\beta_i$ again, which converges to 

1296: a geodesic arc $\beta$ in Hausdorff metric topology with 

1297: $i(\beta,\lam_1)\neq 0$. For this subsequence $$\beta_i\subset\alpha_i

1298: \subset\Phi(\s_i),$$  

1299: $\Phi(\s_i)$ still converges to $\nu$ in Hausdorff metric topology.

1300: Notice that $\lam_1\not\subset\nu$. This is a contradiction to Theorem 

1301: \ref{mainthm}.

1302: \end{proof}

1303: 

1304: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1305: \subsection{Proof of Theorem \ref{thm;surgery}}\label{sec;surgery}

1306: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1307: 

1308: In this section we prove Theorem \ref{thm;surgery}.

1309: Suppose that $\lam_2$ is not a simple closed curve and let $F$ be the 

1310: subsurface of $S$ which is filled by $\lam_2$ (see Figure \ref{fig;ct3}). 

1311: 

1312: \begin{figure}[ht]

1313: \begin{center}

1314: \psfrag{b}{$\beta_i$}

1315: \psfrag{F}{$F$}

1316: \includegraphics[width=3in,height=2in]{ct3.eps}

1317: \end{center}

1318: \caption{$\beta_i$}

1319: \label{fig;ct3}

1320: \end{figure}

1321: 

1322: \noindent

1323: Notice that there exists a pants curve $\alpha_i$ in $\Phi(\s_i)$ with 

1324: $\alpha_i\cap F\neq\emptyset$ for all $i$. 

1325: Let $\beta_i$ be a component of $\alpha_i\cap F$. 

1326: As in Lemma 2.2 of \cite{MM2}, let

1327: \begin{eqnarray*}

1328: &&\Tilde{\beta_i}\mbox{ be a non-peripheral essential component of boundary}\\

1329: &&\quad\quad\mbox{of regular neighborhood of }\beta_i\cup\partial\overline{F},

1330: \end{eqnarray*}

1331: 

1332: \noindent

1333: where $\overline{F}$ is the completion of $F$ with path-metric. Note that two 

1334: different components of $\partial\overline{F}$ could be a same curve in $S$.

1335: 

1336: \indent

1337: Notice that if $\beta_i$ is a closed curve, then $\Tilde{\beta_i}$ is homotopic

1338: to $\beta_i$. Notice also that if $\beta_i$ is an arc, then there are two cases

1339: as in Figure \ref{fig;ct4}.

1340: 

1341: \begin{figure}[ht]

1342: \begin{center}

1343: \psfrag{b}{$\beta_i$}

1344: \psfrag{1}{\textbf{Case I}}

1345: \psfrag{2}{\textbf{Case II}}

1346: \psfrag{B}{$\partial\overline{F}$}

1347: \includegraphics[width=4.5in,height=2in]{ct4.eps}

1348: \end{center}

1349: \caption{Two Cases}

1350: \label{fig;ct4}

1351: \end{figure}

1352: 

1353: \noindent

1354: Recall that $\lam_2$ is not a closed curve. Since $F$ is filled by $\lam_2$, 

1355: it can not be a disk, an annulus or a pants. Therefore for both cases, the 

1356: regular neighborhood of $\beta_i\cup\partial\overline{F}$ has a boundary 

1357: component which is non-peripheral and essential in $F$.   

1358: Suppose that $\partial\overline{F}=\{\gamma_1,\cdots,\gamma_k\}$ and let

1359: $$\el_{\s_i}(\partial F)=\el_{\s_i}(\gamma_1)+\cdots\el_{\s_i}(\gamma_k).$$

1360: Let $\el_{\s_i}(\Tilde{\beta}_i)$ be the length of geodesic representative of 

1361: $\Tilde{\beta}_i$ in $\s_i$. From the definition of $\Tilde{\beta}_i$, we have 

1362: \begin{equation}\label{eq;simple}

1363: \el_{\s_i}(\Tilde{\beta}_i)<2\el_{\s_i}(\alpha_i)+\el_{\s_i}(\partial F).

1364: \end{equation} 

1365: 

1366: \indent

1367: We now prove Theorem \ref{thm;surgery}. 

1368: 

1369: \begin{thm;surgery}

1370: The geodesic representative of $\Tilde{\beta_i}$ converges to $\lam_2$ in 

1371: $\UML(F)$.   

1372: \end{thm;surgery} 

1373: 

1374: \noindent

1375: To prove Theorem \ref{thm;surgery}, it is enough to show that the geodesic

1376: representative of $\Tilde{\beta_i}$ converges to $\lam_2$ in $\UML(S)$. 

1377: For any limit point $[\beta]$ of $[\Tilde{\beta_i}]$ in $\PML(S)$, we will show

1378: that $i(\beta,\lam_2)=0$. Then Theorem \ref{thm;surgery} follows from the fact

1379: that $\lam_2$ fills $F$.

1380: 

1381: \begin{lem}

1382: If $[\beta]$ is a limit point of $[\Tilde{\beta_i}]$ in $\PML(S)$, then 

1383: $i(\beta,\lam_2)=0$.

1384: \end{lem}

1385: 

1386: \begin{proof}

1387: Suppose that $[\beta]$ is a limit point of $[\Tilde{\beta_i}]$. After possibly

1388: restricting to a subsequence, we may assume that $[\Tilde{\beta_i}]$ converges 

1389: to $[\beta]$ in $\PML(S)$. There exist a constant $K>0$ and a sequence $b_i>0$ 

1390: such that $b_i\Tilde{\beta_i}$ converges to $\beta$ in $\ML(S)$ with 

1391: $b_i\leq K$ for all $i$. 

1392: 

1393: \indent

1394: Choose $\mu_i\in\ML(S)$ as in Theorem \ref{thm;useful}. Since $\mu_i$ converges

1395: to $[\lam]\in\PML(S)$ with $\mu_i\to\infty$, there exists a sequence $c_i>0$ 

1396: such that $c_i\mu_i$ converges to $\lam$ in $\ML(S)$ with $c_i\to 0$. 

1397: By eq. (\ref{eq;papa}), there exists $\Gamma>0$ which does not depend on $i$ 

1398: such that 

1399: $i(\partial\overline{F},\mu_i)\leq \el_{\s_i}(\partial\overline{F})\leq 

1400: i(\partial\overline{F},\mu_i)+\Gamma$.

1401: Therefore

1402: $$i(\partial\overline{F},c_i\mu_i)\leq c_i\el_{\s_i}(\partial\overline{F})\leq 

1403: i(\partial\overline{F},c_i\mu_i)+c_i\Gamma.$$

1404: Since $i(\partial\overline{F},\lam)=0$, from the continuity of the intersection

1405: number, we have

1406: \begin{equation}\label{eq;crucial}

1407: \lim_{i\to\infty}c_i\el_{\s_i}(\partial\overline{F})=0.

1408: \end{equation}

1409: From eq. (\ref{eq;papa}) and eq. (\ref{eq;simple}), we have

1410: $$i(\Tilde{\beta_i},\mu_i)\leq \el_{\s_i}(\Tilde{\beta}_i)\leq 

1411: 2\el_{\s_i}(\alpha_i)+\el_{\s_i}(\partial\overline{F})$$

1412: Therefore 

1413: $$i(b_i\Tilde{\beta_i},c_i\mu_i)\leq K\left(2c_i L+c_i\el_{\s_i}(

1414: \partial\overline{F})\right).$$ Hence from eq. (\ref{eq;crucial}) and the 

1415: continuity of the intersection number, we have $i(\beta,\lam)=0$. 

1416: Thus $i(\beta,\lam_2)=0$.

1417: \end{proof}

1418: 

1419: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1420: \subsection{Theorem \ref{thm;filling} follows from Theorem \ref{thm;surgery}}

1421: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1422: 

1423: Suppose that $\s_i\in\T(S)$ converges to $[\lam]\in\PML(S)$ in $\TC(S)$, and 

1424: $\alpha_i$ is a pants curve in $\Phi(\s_i)$. If $\lam$ is a filling lamination,

1425: from Theorem \ref{thm;surgery}, then $\alpha_i$ converges to $u([\lam])$ in 

1426: $\UML(S)$. Therefore if we identify $\BC(S)$ with $\EL(S)$ via the 

1427: homeomorphism $k$ in Theorem \ref{thm;klarreich}, we have

1428: \begin{thm;filling} 

1429: If $\lam$ is a filling lamination, then $\alpha_i$ converges to $u([\lam])$ in 

1430: $\C(S)\cup\BC(S)$.

1431: \end{thm;filling}

1432: 

1433: %%%%%%%%%%%%%%%%%%%%%%%%%%%

1434: \begin{thebibliography}{00}

1435: %%%%%%%%%%%%%%%%%%%%%%%%%%%

1436: 

1437: \bibitem {abikoff} W. Abikoff, \textit{Kleinian groups-geometrically finite 

1438: and geometrically perverse}, Geometry of group representation (Boulder, CO, 

1439: 1987), 1-50, Contemp. Math. 74, Amer. Math. Soc., 1988.

1440: 

1441: \bibitem {agol} I. Agol, \textit{Tameness of hyperbolic 3-manifolds}, preprint,

1442: arXiv:math. GT/0405568.

1443: 

1444: \bibitem {ahlfors} L.V. Ahlfors, \textit{On quasiconformal mappings}, J. 

1445: Analyse Math. {\bf 3}(1954), 1-58.

1446: 

1447: \bibitem {BP} R. Benedetti and C. Petronio, \textit{Lectures on hyperbolic 

1448: geometry}, Springer-Verlag, 1992.

1449: 

1450: \bibitem{beardon} A.F. Beardon, \textit{The geometry of discrete group}, 

1451: Springer-Verlag, New York, 1983.

1452: 

1453: \bibitem {bers} L. Bers, \textit{Spaces of Riemann surfaces as bounded 

1454: domains}, Bull. Amer. Math. Soc. {\bf 66}(1960), 98-103.

1455: 

1456: \bibitem {bers1} L. Bers, \textit{On boundary of Teichm\"uller spaces

1457: and on Kleinian groups I}, Ann. of Math. {\bf 91}(1970), 570-600.

1458: 

1459: \bibitem {bers2} L. Bers, \textit{Spaces of degenerating Riemann surfaces}, 

1460: Discontinuous groups and Riemann surfaces, Ann. of Math. Stud 79, Princeton 

1461: Univ. Press, 1974, pp. 43-59.

1462: 

1463: \bibitem {bers3} L. Bers, \textit{An inequality for Riemann surfaces}, 

1464: Differential geometry and complex analysis, 87-93, Springer-Verlag, 1985.

1465: 

1466: \bibitem {bonahon} F. Bonahon, \textit{Bouts des vari\'et\'es hyperboliques 

1467: de dimension 3}, Ann. of Math. {\bf 124}(1986), 71-158.

1468:  

1469: \bibitem {bon} F. Bonahon, \textit{The geometry of Teichm\"uller space via 

1470: geodesic currents}, Invent. Math. {\bf 92}(1988), 139-162.

1471:  

1472: \bibitem {bon2} F. Bonahon, \textit{Transverse H\"oder distributions for 

1473: geodesic laminations}, Topology {\bf 36}(1997), no.1, 103-122.

1474:  

1475: \bibitem {bon3} F. Bonahon, \textit{Geodesic laminations with transverse 

1476: H\"oder distributions}, Ann. Sci. Ecole Norm. Sup.(4) {\bf 30}(1997), 

1477: 205-240.

1478: 

1479: \bibitem {bon4} F. Bonahon, \textit{Geodesic laminations on surfaces}, 

1480: Laminations and foliations in dynamics, geometry and topology (Stony Brook, NY,

1481: 1998), 1-37, Contemp. Math. 269, Amer. Math. Soc., 2001.

1482:  

1483: \bibitem {BoS} M. Bonk and O. Schramm, \textit{Embeddings of Gromov hyperbolic 

1484: spaces}, Geom. Funct. Anal. {\bf 10}(2000), no.2, 266-306.

1485: 

1486: \bibitem {bowditch} B.H. Bowditch, \textit{Intersection numbers and the 

1487: hyperbolicity of the curve complex}, preprint, 2002.

1488: 

1489: \bibitem {bowers} P.L. Bowers, \textit{Negatively curved graph and planar 

1490: metrics with applications to type}, Michigan Math. J. {\bf 45}(1989), no.1,

1491: 31-53.

1492: 

1493: \bibitem {bridson} M. Bridson, \textit{Geodesics and curvature in metric 

1494: simplicial complexes}, Group theory from a geometrical viewpoint (Triests, 

1495: 1990), 373-463, World Sci. Publishing, 1991.

1496: 

1497: \bibitem {BH} M. Bridson and A. Haefliger, \textit{Metric spaces of 

1498: non-positive curvature}, Springer-Verlag, 1999.

1499: 

1500: \bibitem {brock} J.F. Brock, \textit{Continuity of Thurston's length function},

1501: Geom. Funct. Anal. {\bf 10}(2000), no.4, 741-797. 

1502: 

1503: \bibitem {brock2} J.F. Brock, \textit{Boundaries of Teichm\"uller space and 

1504: end-invariants for hyperbolic 3-manifolds}, Duke Math. J. {\bf 106}(2001), 

1505: no.3, 527-552. 

1506: 

1507: \bibitem {buser} P. Buser, \textit{Geometry and spectra of compact Riemann 

1508: surfaces}, Birkh\"{a} user, 1992.

1509: 

1510: \bibitem {BS} P. Buser and M. Sepp\"al\"a, \textit{Symmetric pants 

1511: decompositions of Riemann surfaces}, Duke. Math. J. {\bf 67}(1992), no.1, 

1512: 39-55.

1513: 

1514: \bibitem {CG} D. Calegari and D. Gabai, \textit{Shrinkwrapping and the taming 

1515: of hyperbolic 3-manifolds}, preprint, arXiv:math.GT/0407161. 

1516: 

1517: \bibitem {CEG} R.D. Canary, D.B.A. Epstein and P. Green, \textit{Notes on notes

1518:  of Thurston}, Analytical and geometrical aspects of Hyperbolic spaces (

1519: D.B.A. Epstein, ed.), London Math. Lecture Note Ser. 111, Cambridge Univ. 

1520: Press, 1987, pp. 3-92.

1521: 

1522: \bibitem {canary} R.D. Canary, \textit{Ends of hyperbolic 3-manifolds}, 

1523: J. Amer. Math. Soc. {\bf 6}(1993), no.1, 1-35.

1524: 

1525: \bibitem {cannon} J. Cannon, \textit{The theory of negatively curved spaces and

1526: groups}, Ergodic theory, symbolic dynamics, and hyperbolic spaces 

1527: (Trieste, 1989), Oxford Univ. Press, 1991, 315-369.

1528: 

1529: \bibitem {CB} A.J. Casson and S.A. Bleiler, \textit{Automorphisms of surfaces 

1530: after Nielsen and Thurston}, Cambridge University Press, 1988.

1531: 

1532: \bibitem {farb} B. Farb, \textit{Relative hyperbolic and automatic groups with 

1533: applications to negatively curved manifolds}, Ph.D. thesis, Princeton Univ., 

1534: 1994. 

1535: 

1536: \bibitem {FLM} B. Farb, A. Lubotzky and Y. Minsky, \textit{Rank-1 phenomena for

1537: mapping class group},  Duke. Math. J. {\bf 106}(2001), no.3, 581-597.

1538: 

1539: \bibitem {FLP} A. Fathi, F. Laudenbach and V. Poenaru, \textit{Travaux de 

1540: Thurston sur les surfaces}, vol. 66-67, Asterisque, 1979.

1541: 

1542: \bibitem {gabai} D. Gabai, \textit{3 lectures on foliations and laminations on 

1543: 3-manifolds}, Laminations and foliations in dynamics, geometry and topology

1544: (Stony Brook, NY, 1998), 87-109, Contemp. Math. 269, Amer. Math. Soc., 2001.

1545: 

1546: \bibitem {gardiner} F.P. Gardiner, \textit{A correspondence between laminations

1547:  and quadratic differentials}, Complex Variables {\bf 6}(1986), 363-375. 

1548: 

1549: \bibitem {GM} F.P. Gardiner and H. Masur, \textit{Extremal length geometry 

1550: of Teichm\"uller space}, Complex Variables, {\bf 16}(2000), 209-237. 

1551: 

1552: \bibitem {gersten} S.M. Gersten, \textit{Introduction to hyperbolic and 

1553: automatic group}, Summer school in group theory in Banff, 1996, 45-70, CRM 

1554: Proc. Lecture Notes 17, Amer. Math. Soc., 1999. 

1555: 

1556: \bibitem {gromov} H. Gromov, \textit{Hyperbolic groups}, Essays in Group Theory

1557: (S.M. Gersten, editor), MSRI Publications no.8, Springer-Verlag, 1978.

1558: 

1559: \bibitem {hamenstadt} U. Hamenst\"{a}dt, \textit{Train tracks and the Gromov 

1560: boundary of the complex of curves}, preprint, arXiv:math.GT/0409611. 

1561: 

1562: \bibitem {harer} J.L. Harer, \textit{The virtual cohomological dimension of

1563: the mapping class group of an orientable surface}, Invent. Math. 

1564: {\bf 84}(1986), 157-176. 

1565: 

1566: \bibitem {harvey} W.J. Harvey, \textit{Boundary structure of the modular 

1567: group}, Riemann surfaces and related topics, 245-251, Ann. of Math. Stud. 97, 

1568: Princeton Univ. Press, 1981.

1569: 

1570: \bibitem {hatcher} A.E. Hatcher, \textit{Measured lamination spaces for 

1571: surfaces, from the topological viewpoint}, Topology Appl. {\bf 30}(1988),

1572: 63-88. 

1573: 

1574: \bibitem {HT} A. Hatcher and W. Thurston, \textit{A presentation for the 

1575: mapping class group of a closed orientable surface}, Topology {\bf 19}(1980),

1576: 221-237. 

1577: 

1578: \bibitem {HM} J. Hubbard and H. Masur, \textit{Quadratic differentials and 

1579: foliations}, Acta Math. {\bf 142}(1979), no.3-4, 221-274.

1580: 

1581: \bibitem {IT} Y. Imayoshi and M. Taniguchi, \textit{An introduction to 

1582: Teichm\"uller spaces}, Springer-Verlag, 1992.

1583: 

1584: \bibitem {ivanov} N.V. Ivanov, \textit{Automorphisms of complexes of curves 

1585: and of Teichm\"uller spaces}, Internat. Math. Res. Notices {\bf 1997}, no.14,

1586: 651-666.

1587: 

1588: \bibitem {KM} V.A. Kaimanovich and H. Masur, \textit{The Poisson boundary of 

1589: the mapping class group}, Invent. Math. {\bf 125}(1996), no.2, 221-264.

1590: 

1591: \bibitem {keen} L. Keen, \textit{Collars on Riemann surfaces}, Discontinuous

1592: groups and Riemann surfaces, Ann. of Math. Studies, No.79, Princeton Univ. 

1593: Press, 1974. 

1594: 

1595: \bibitem {kerckhoff} S.P. Kerckhoff, \textit{The asymptotic geometry of 

1596: Teichm\"uller space}, Topology {\bf 19}(1980), 23-41.

1597: 

1598: \bibitem {kerckhoff2} S.P. Kerckhoff, \textit{The Nielsen realization problem},

1599: Ann. Math. {\bf 117}(1983), 235-265.

1600: 

1601: \bibitem {KT} S.P. Kerckhoff and W.P. Thurston, \textit{Non-continuity of the 

1602: action of the modular group at Bers' boundary of Teichm\"uller space},

1603: Invent. Math. {\bf 100}(1990), 25-47.

1604: 

1605: \bibitem {kim} Y.D. Kim, \textit{A theorem on discrete, torsion free subgroups 

1606: of $Isom\, H^n$}, Geom. Dedicata {\bf 109}(2004), 51-57.

1607: 

1608: \bibitem {klarreich} E. Klarreich, \textit{The boundary at infinity of the 

1609: curve complex and the relative Teichm\"uller space}, preprint, 1998.

1610: 

1611: \bibitem {korkmaz} M. Korkmaz, \textit{Automorphisms of complexex of curves on 

1612: punctured spheres and on punctured tori}, Topology and its Applications 

1613: {\bf 95}(1999), 85-111.

1614: 

1615: \bibitem {kra} I. Kra, \textit{Quadratic differentials}, Rev. Roumaine Math. 

1616: Pures Appl. {\bf 39}(1994), no.8, 751-787.

1617: 

1618: \bibitem {krushkal} S.L. Krushkal, \textit{Teichm\"uller spaces are not 

1619: starlike}, Ann. Acad. Sci. Fenn. Ser.A/Math. {\bf 20}(1995), no.1, 167-173.

1620: 

1621: \bibitem {KS} R.S. Kulkarni and P.B. Shallen, \textit{On Ahlfors' finiteness

1622: theorem}, Adv. Math. {\bf 76}(1989), no.2, 155-169.

1623: 

1624: \bibitem {levitt} G. Levitt, \textit{Foliations and laminations on hyperbolic 

1625: surfaces}, Topology {\bf 22}(1983), no.2, 119-135. 

1626: 

1627: \bibitem {luo} F. Luo, \textit{Grothendieck's reconstruction principle and

1628: 2-dimensional topology and geometry}, Commun. Contemp. Math. {\bf 1}(1999), 

1629: no.2, 125-153.

1630: 

1631: \bibitem {luo2} F. Luo, \textit{Geodesic length functions and Teichm\"uller 

1632: space}, Electron. Res. Announc. Amer. Math. Soc. {\bf 2}(1996), no.1, 34-41.

1633: 

1634: \bibitem {luo3} F. Luo, \textit{Automorphisms of the complex of curves}, 

1635: Topology {\bf 39}(2000), 283-298.

1636: 

1637: \bibitem {marden} A. Marden, \textit{The geometry of finitely generated

1638: Kleinian groups}, Ann. of Math. {\bf 99}(1974), 383-462. 

1639: 

1640: \bibitem {maskit} B. Maskit, \textit{Comparison of hyperbolic and extremal 

1641: lengths}, Ann. Acad. Sci. Fenn. Series A. I. Mathematica

1642: {\bf 10}(1985), 381-386.

1643: 

1644: \bibitem {maskit2} B. Maskit, \textit{On boundary of Teichm\"uller spaces

1645: and on Kleinian groups II}, Ann. of Math. {\bf 91}(1970), 607-639.

1646: 

1647: \bibitem {maskit3} B. Maskit, \textit{A picture of moduli space}, Invent. Math.

1648:  {\bf 126}(1996), 341-390.

1649: 

1650: \bibitem {maskit4} B. Maskit, \textit{Kleinian groups}, Springer-Verlag, 1988.

1651: 

1652: \bibitem {masur} H. Masur, \textit{On a class of geodesics in Teichm\"uller 

1653: space}, Ann. of Math. {\bf 102}(1975), 205--221. 

1654: 

1655: \bibitem {masur1} H. Masur, \textit{Two boundaries of Teichm\"uller space}, 

1656: Duke Math. J. {\bf 49}(1982), no.1, 183-190. 

1657: 

1658: \bibitem {MM} H. Masur and Y. Minsky, \textit{Geometry of the complex of

1659: curves I, hyperbolicity}, Invent. Math. {\bf 138}(1999), 103-149.

1660: 

1661: \bibitem {MM1} H. Masur and Y. Minsky, \textit{Unstable quasi-geodesics

1662: in Teichm\"uller space}, Contemp. Math. 256, Amer. Math. Soc., 2000.

1663: 

1664: \bibitem {MM2} H. Masur and Y. Minsky, \textit{Geometry of the complex of 

1665: curves II, hierarchical structure}, Geom. Funct. Anal. {\bf 10}(2000), 902-974.

1666: 

1667: \bibitem {MW} H. Masur and M. Wolf, \textit{Teichm\"uller space is not Gromov 

1668: hyperbolic}, Ann. Acad. Sci. Fenn. Ser.A/Math. {\bf 20}(1995), no.2, 259-267.

1669: 

1670: \bibitem {MP} J. McCarthy and A. Papadopoulos, \textit{Dynamics on Thurston's 

1671: sphere of projective measured foliations}, Comment. Math. Helvetici 

1672: {\bf 64}(1989), 133-166.

1673: 

1674: \bibitem {MP1} J. McCarthy and A. Papadopoulos, \textit{The visual sphere of 

1675:  Teichm\"uller space and a theorem of Masur-Wolf}, Ann. Acad. Sci. Fenn. Math.

1676:  {\bf 24}(1999), 147-154.

1677: 

1678: \bibitem {mccullough} D. McCullough, \textit{Compact submanifolds of 

1679: 3-manifolds with boundary}, Quart. J. Math. Oxford {\bf 37}(1986), 299-306.

1680: 

1681: \bibitem {mcmullen} C. McMullen, \textit{Cusps are dense}, Ann. of Math. 

1682: {\bf 133}(1991), 217-247.

1683: 

1684: \bibitem {mosher} L. Mosher, \textit{Stable Teichm\"uller quasigeodesics 

1685: and ending laminations}, Geom. Topol. {\bf 7}(2003), 33-90.

1686: 

1687: \bibitem {m-thesis} Y. Minsky, \textit{Harmonic maps, length, and energy in

1688: Teichm\"uller space}, J. Differential Geom. {\bf 35}(1992), 151-217.

1689: 

1690: \bibitem {m-teich} Y. Minsky, \textit{Teichm\"uller geodesics and ends

1691: of hyperbolic 3-manifolds}, Topology {\bf 32}(1993), 625-647.

1692: 

1693: \bibitem {m-geom} Y. Minsky, \textit{A geometric approach to the complex of 

1694: curves on a surface}, Topology and Teichm\"uller space (Katinkulta, 1995), 

1695: 149-158, World Sci. Publishing, 1996.

1696: 

1697: \bibitem {m-extremal} Y. Minsky, \textit{Extremal length estimate and product 

1698: regions in Teichm\"uller space}, Duke. Math. J. {\bf 83}(1996), 249-286.

1699: 

1700: \bibitem {m-quasi} Y. Minsky, \textit{Quasi-projections in Teichm\"uller 

1701: space}, J. Reine Angew. Math. {\bf 473}(1996), 121-136.

1702: 

1703: \bibitem {m-torus} Y. Minsky, \textit{The classification of punctured-torus

1704: group}, Ann. of Math. {\bf 149}(1999), 559-626.

1705: 

1706: \bibitem {m-klein} Y. Minsky, \textit{Kleinian groups and the complex of 

1707: curves}, Geometry and Topology {\bf 4}(2000), 117-148.

1708: 

1709: \bibitem {m-combi} Y. Minsky, \textit{Combinatorial and geometrical aspects of

1710: hyperbolic 3-manifolds}, London Math. Soc. Lecture Note Ser. 299, Cambridge 

1711: Univ. Press, 2003.

1712: 

1713: \bibitem {m-bounded} Y. Minsky, \textit{Bounded geometry for Kleinian groups}, 

1714: Invent. Math. {\bf 146}(2001), no.1, 143-192.

1715: 

1716: \bibitem {m-cdm} Y. Minsky, \textit{End invariants and the classification of 

1717: hyperbolic 3-manifolds}, Current developments in mathematics, 2002, 181-217,

1718: Int. Press, 2003. 

1719: 

1720: \bibitem {m-ending} Y. Minsky, \textit{The classification of Kleinian surface

1721: groups I: Models and bounds}, preprint, arXiv:math.GT/0302208.

1722: 

1723: \bibitem {m-qconvex} Y. Minsky, \textit{Quasiconvexity in the curve complex}, 

1724: preprint, arXiv:math. GT/0307083.

1725: 

1726: \bibitem {m-ending1} Y. Minsky, \textit{The classification of Kleinian surface

1727: groups II: The ending lamination conjecture}, preprint, arXiv:math. GT/0412006.

1728: 

1729: \bibitem {MO} J.W. Morgan and J. Otal, \textit{Relative growth rates of closed 

1730: geodesics on a surface under varying hyperbolic structures}, 

1731: Comment. Math. Helv. {\bf 68}(1993), no.2, 171-208. 

1732: 

1733: \bibitem {papa} A. Papadopoulos, \textit{On Thurston's boundary of

1734: Teichm\"uller space and the extention of earthquakes}, Topology Appl. 

1735: {\bf 41}(1991), no.3, 147-177.

1736: 

1737: \bibitem {penner} R.C. Penner, \textit{An introduction to train tracks}, 

1738: Low-dimensional topology and Kleinian groups, 77-90, London Math. Soc. Lecture 

1739: Note Ser. 112, Cambridge Univ. Press, 1986. 

1740: 

1741: \bibitem {PH} R.C. Penner and J.L. Harer, \textit{Combinatorics of train 

1742: tracks}, Annals of Mathematics Studies 125, Princeton University Press, 

1743: Princeton, 1992.

1744: 

1745: \bibitem {PM} A. Portolano and I. Maniscalco, \textit{Curves as measured 

1746: foliation on noncompact surface}, Rend. Circ. Mat. Palermo(2) {\bf 42}(1993), 

1747: no.2, 161-180. 

1748: 

1749: \bibitem {ratcliffe} J.G. Ratcliffe, \textit{ Foundations of hyperbolic 

1750: manifolds}, Springer-Verlag, New York, 1994. 

1751: 

1752: \bibitem {schmutz} P. Schmutz, \textit{Riemann surfaces with shortest geodesic 

1753: of maximal length}, Geom. Funct. Anal. {\bf 3}(1993), no.6, 564-631.

1754: 

1755: \bibitem {schmutz2} P. Schmutz, \textit{Systoles on Riemann surfaces}, 

1756: Manuscripta Math. {\bf 85}(1994), no.3-4, 429-447. 

1757: 

1758: \bibitem {scott} G.P. Scott, \textit{Compact submanifolds of 3-manifolds}, 

1759: J. London Math. Soc. {\bf 7}(1973), 246-250. 

1760: 

1761: \bibitem {short} Edited by H. Short, \textit{Notes on word hyperbolic groups}, 

1762: Group theory from geometric viewpoint (Trieste, 1990), 3-63, World Sci. 

1763: Publishing, 1991.

1764: 

1765: \bibitem {sullivan} D. Sullivan, \textit{On the ergodic theory at infinity of

1766: an arbitrary discrete group of hyperbolic motions}, Riemann surfaces 

1767: and related topics: Proceedings of the 1978 Stony Brook Conference, 

1768: Ann. of Math. Stud. {\bf 97}, Princeton, 1981.

1769: 

1770: \bibitem {thurston} W. Thurston, \textit{3-dimensional geometry and topology}, 

1771: Princeton University Press, 1997, (S. Levy, ed.).

1772: 

1773: \bibitem {thurston1} W. Thurston, \textit{The geometry and topology of 

1774: 3-manifolds}, Princeton University Lecture Notes, 1982.

1775:   

1776: \bibitem {thurston2} W. Thurston, \textit{Three dimensional manifolds, Kleinian

1777: groups and hyperbolic geometry}, Bull. Amer. Math. Soc. {\bf 6}(1982), 357-381.

1778: 

1779: \bibitem {thurston3} W. Thurston, \textit{Minimal stretch maps between 

1780: hyperbolic surfaces}, preprint, arXiv:math.GT/9801039. 

1781: 

1782: \bibitem {thurston4} W. Thurston, \textit{Hyperbolic structures on 3-manifolds 

1783: II: Surface groups and 3-manifolds which fiber over the circle}, preprint, 

1784: arXiv:math. GT/9801045. 

1785: 

1786: \bibitem {thurston5} W. Thurston, \textit{On the geometry and dynamics of 

1787: diffeomorphisms of surfaces}, Bull. Amer. Math. Soc. {\bf 19}(1988), no.2, 

1788: 417-431.

1789: 

1790: \bibitem {weiss} H. Weiss, \textit{The geometry of measured geodesic 

1791: laminations and measured train tracks}, Ergodic Theory Dynam. Systems 

1792: {\bf 9}(1989), no.3, 587-604.

1793: 

1794: \end{thebibliography}

1795: \end{document}

1796: 

1797: 

1798: 

1799: 

1800: 

1801: 

1802: 

1803: 

1804: 

1805: 

1806: 

1807: 

1808: 

1809: 

1810: 

1811: 

1812: 

1813: 

1814: 

1815: 

1816: 

1817: 

1818: 

1819: 

1820: 

1821: 

1822: 

1823: 

1824: 

1825: 

1826: 

1827: 

1828: 

1829: 

1830: 

1831: 

1832: 

1833: 

1834: 

1835: 

1836: 

1837: 

1838: 

1839: 

1840: 

1841: 

1842: 

1843: 

1844: 

1845: 

1846: 

1847: 

1848: 

1849: 

1850: 

1851: 

1852: 

1853: 

1854: 

1855: 

1856: 

1857: 

1858: 

1859: 

1860: 

1861: 

1862: 

1863: 

1864: 

1865: 

1866: 

1867: 

1868: 

1869: 

1870: 

1871: 

1872: 

1873: 

1874: