1: %

2: % This is a LaTeX file

3: %

4: 

5: \documentclass[12pt]{article}

6: \usepackage{latexsym,color,amsmath,amsthm,amssymb,amscd,amsfonts,epsfig}

7: 

8: \setlength{\textwidth}{6.0in} \setlength{\evensidemargin}{0.25in}

9: \setlength{\oddsidemargin}{0.25in} \setlength{\textheight}{9.0in}

10: \setlength{\topmargin}{-0.5in} \setlength{\parskip}{2mm}

11: \setlength{\baselineskip}{1.7\baselineskip}

12: \newtheorem{lemma}{Lemma}[section]

13: \newtheorem{proposition}[lemma]{Proposition}

14: \newtheorem{remark}[lemma]{Remark}

15: \newtheorem{theorem}[lemma]{Theorem}

16: \newtheorem{definition}[lemma]{Definition}

17: \newtheorem{corollary}[lemma]{Corollary}

18: \newtheorem{thrm}{Theorem}

19: \newtheorem{lem}[thrm]{Lemma}

20: \newtheorem{cor}[thrm]{Corollary}

21: \newtheorem{defn}[thrm]{Definition}

22: \newtheorem{rem}[thrm]{Remark}

23: \newtheorem{example}[lemma]{Example}

24: \newtheorem{claim}[lemma]{Claim}

25: \newtheorem{conj}[lemma]{Conjecture}

26: \newtheorem{ques}[lemma]{Question}

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

28: \newtheorem*{remark*}{Remark}

29: \newtheorem*{remnonum}{Remark}

30: \newtheorem*{rems}{Remarks}

31: \newtheorem*{corr}{Corollaries}

32: \def\HH{{\cal H}}

33: 

34: \newcommand{\Id}{{{\mathchoice {\rm 1\mskip-4mu l} {\rm 1\mskip-4mu l}

35:       {\rm 1\mskip-4.5mu l} {\rm 1\mskip-5mu l}}}}

36: 

37: 

38: \begin{document}

39: \title {  The geometry of partial order on contact transformations of prequantization manifolds    }\author{Gabi Ben Simon

40: \footnote{This paper is a part of the author's Ph.D. thesis, being

41: carried out under the supervision of Professor Leonid Polterovich,

42: at Tel-Aviv University.} \footnote{Partially supported by Israel

43: Science Foundation grant 11/03.}

44: %\author{Gabi Ben Simon

45: %\thanks{This paper is a part of the author's Ph.D thesis, being carried out under the supervision of Professor

46: %Leonid Polterovich, at Tel-Aviv  University.}

47: \\

48: School of Mathematical Sciences \\

49: Tel Aviv University

50: 69978 Tel Aviv, Israel \\

51:  gabiben@post.tau.ac.il} \maketitle

52: \begin{abstract} \noindent

53: In this paper we find  connection between the Hofer's metric of

54: the group of Hamiltonian diffeomorphisms of a

55:  closed symplectic manifold, with an integral symplectic form, and the

56:  geometry, defined in~\cite{El-P}, of the quantomorphisms group

57:  of its prequantization manifold. This gives two main results: First, we calculate, partly, the geometry

58:  of the quantomorphisms groups of a prequantization manifolds

59:  of an integral symplectic manifold which admits certain Lagrangian

60:  foliation. Second, for every prequantization manifold we give a formula

61:  for the distance between a point and a distinguished curve in the metric

62:  space associated to its group of quantomorphisms.

63:   Moreover, our first result is a

64:  full computation of the geometry related to the symplectic linear group which can be considered

65:  as a subgroup of the contactomorphisms group of suitable prequantization manifolds of the

66:  complex projective space.

67:  In the course of the proof we use in an essential way the Maslov quasimorphism.

68: 

69: %Let $(M,\omega)$ be one of the symplectic manifolds: the standard

70: %symplectic $n-1$ dimentional complex projective space or the

71: %standard 2 dimentional symplectic torus. In this paper we study

72: %the geometry, defined in, for certain subgroups of the

73: %contact transformations group of $(P,\xi)$ where $(P,\xi)$ is a

74: %prequantization space of $(M,\omega)$.

75: 

76: 

77: 

78: \end{abstract}

79: 

80: 

81: \section{Introduction and results}

82:  The motivating background of this paper can be described  as

83: follows. Let $(M,\omega)$ be a closed symplectic manifold. Denote

84: by $Ham(M,\omega)$ the group of all Hamiltonian symplectomorphisms

85: of $(M,\omega).$ This important group carries a natural

86: bi-invariant metric called the Hofer metric (a detailed

87: description of the

88:  investigation of this metric structure can be found in~\cite{P}).

89: An analogue of $Ham(M,\omega)$ in the case of contact geometry

90:  is the group of Hamiltonian

91: contactomorphisms $Cont(P,\xi)$ where $(P,\xi)$ is a contact

92: manifold (see~\cite{A} appendix-4,~\cite{AG} and~\cite{AM} for

93: basic information on contact manifolds). The groups $Ham$ and

94: $Cont$ are closely related. However in contrast to the case of

95: $Ham$, no interesting bi-invariant metrics on the group $Cont$ are

96: known. In~\cite{El-P} Eliashberg and Polterovich noticed that for

97: certain contact manifold the universal cover of $Cont$ carries a

98: bi-invariant partial order from which we get a natural metric

99: space $Z$ associated to $Cont$. The definition of $Z$ is somewhat

100: indirect, and its geometry is far from being understood even in

101: the simplest examples.

102: 

103: \noindent In this paper we study the geometry  of $Z$ for certain

104: subgroups of $Cont(P,\xi)$ where $(P,\xi)$ is a prequanization

105: space of a closed symplectic manifold $(M,\omega)$. As we remarked

106: above an important tool is the establishment of a connection

107: between the geometry of $Z$ and the geometry of the universal

108: cover of $Ham(M,\omega)$ endowed with the Hofer metric.

109: 

110: \noindent Finally let us mention that further developments on this

111: subject appear in~\cite{E-K-P}, where in this work we refer mostly

112: to~\cite{El-P}.

113: 

114: 

115: \subsection{Preliminaries on partially ordered groups} \label{basics on p.o.g}

116: A basic idea in this work is the implementation of basic notions

117: from the theory of partially ordered groups, to the universal

118: cover of the groups of contact transformations, of the relevant

119: contact manifolds. So we start with the following basic

120: definitions and constructions.

121: 

122: \begin{definition}{\rm Let $\mathcal{D}$ be a group. A subset $\mathcal{C} \subset \mathcal{D}$ is called a\textit{ normal

123: cone} if

124: 

125: $ \textbf{a.}\text{ } f,g\in \mathcal{C}\Rightarrow fg \in

126: \mathcal{C}$

127: 

128: $ \textbf{b.}\text{ }f \in \mathcal{C}, h \in \mathcal{D}

129: \Rightarrow hfh^{-1} \in \mathcal{C}$

130: 

131: $ \textbf{c.}\text{ } \textbf{1}_{\mathcal{D}} \in \mathcal{C}$}

132: 

133: 

134: 

135: \end{definition}

136: 

137: We define for $f,g \in \mathcal{D}$ that $f \geq g$ if $fg^{-1}\in

138: \mathcal{C}$. It is not hard to check that this relation is

139: reflexive and transitive.

140: 

141: \begin{definition}{\rm If the above relation is also anti-symmetric,

142: then we call it a \textit{ bi-invariant partial order} induced by

143: $\mathcal{C}$. In this situation, an element $f\in

144: \mathcal{C}\setminus \mathbf{1}$ is called a \textit{dominant} if

145: for every $g \in \mathcal{D}$ there exists $n \in \mathbb{N}$ such

146: that $f^{n}\geq g$.}

147: \end{definition}

148: 

149: \noindent $\mathbf{Remark.}$ Notice that the normality of the cone

150: $\mathcal{C}$ implies that for every $f,g,d,e \in \mathcal{D}$

151: \begin{equation}\text{ if  } f\geq g \text{

152:  and  } d \geq e \text{  then  } fd \geq ge.

153: \end{equation}

154: \begin{definition}{\rm Let $f$ be a dominant and $g\in \mathcal{D}$.

155: Then the \textit{relative growth} of $f$ with respect to $g$ is

156: $$\gamma(f,g)=\lim \limits_{n\rightarrow \infty}

157: \frac{\gamma_{n}(f,g)}{n}$$ where $$\gamma_{n}(f,g)=inf \{p\in

158: \mathbb{Z}|\text{ } f^{p}\geq g^{n}\}.$$}

159: \end{definition} The above limit exists as the reader can check by

160: himself (see also~\cite{El-P} section 1).

161: 

162: Now we want to relate a geometrical structure to the function

163: $\gamma$ defined above. Denote by $\mathcal{C}^{+}\in\mathcal{C}$

164: the set of all dominants. We define the metric space $(Z,d)$ in

165: the following way. First note that $$f,g,h\in

166: \mathcal{C}^{+}\Rightarrow \gamma(f,h)\leq \gamma(f,g)

167: \cdot\gamma(g,h).$$ We define the function

168: $$K:\mathcal{C}^{+}\times \mathcal{C}^{+}\rightarrow [0,\infty)$$

169: by $$K(f,g)= \max\{\log\gamma(f,g),\log\gamma(g,f)\}.$$ It is

170: straightforward to check that $K$ is non negative, symmetric,

171: vanishes on the diagonal and satisfies the triangle inequality.

172: Thus $K$ is a pseudo-distance.

173: 

174: \noindent Define an equivalence relation on $\mathcal{C}^{+}$ by

175: setting $f\sim g$ provided $K(f,g)=0$. Put

176: $$Z=\mathcal{C}^{+}/\sim.$$

177: 

178: The function $K$ on $\mathcal{C}^{+}$ projects in a natural way to

179: a genuine metric $d$ on $Z$. Thus we get a metric space naturally

180: associated to a partially ordered groups (see also~\cite{El-P} for

181: more information).

182: \subsection{Geometry of the symplectic linear group}\label{linear

183: geometry}

184: 

185: 

186: Let $Sp(2n,{\mathbb{R}})$ be the symplectic linear group. We

187: denote by $\mathcal{S}$ its universal cover with base point $\Id$,

188: the identity matrix. We can think of $\mathcal{S}$ as the space of

189: paths starting at $\Id$ up to a homotopy relation between paths

190: with the same end point. Throughout this subsection we consider

191: the group $\mathcal{D}$ to be $\mathcal{S}$.

192: 

193: \noindent Next, consider the equation:\begin{equation}\label{real

194: positive} \dot{X}(t){X^{-1}(t)}=JH(x,t)\end{equation} Here, $X(t)

195: \in {\mathcal{S}}$, $H$ is a time dependant symmetric matrix on

196: $\mathbb{R}^{2n}$ and $J$ is the matrix

197: $\left( \! \! \! \begin{array}{rr} 0 & -{\Id} \\

198:  {\Id} &  0 \end{array} \! \! \! \right) $ on $\mathbb{R}^{2n}$.

199: The quadratic form represented by the matrix $H$, will be called

200: the {\it Hamiltonian} generating $X(t)$.

201:  It is easy to verify the following two facts:\\

202: 

203: \noindent {\bf A}). The set of elements $X(t)$ in $\mathcal{S}$,

204: generated by $H(x,t)$ which are non negative as a quadratic form

205: for each $t$, establishes a normal cone.

206: 

207: \noindent {\bf B}). In particular, those elements which are

208: generated by a {\bf strictly} positive $H(x,t)$ are dominants in

209: $\mathcal{S}$.

210: 

211: \noindent G.I. Olshanskii has proved the following theorem

212: (see~\cite{Ol} as a reference and~\cite{CSM} as a background

213: source).

214: 

215: \begin{theorem}\label{linear n.t.p.o}

216: In the above setting the cone establishes a non trivial partial

217: order on $\mathcal{S}$.

218: \end{theorem}In view of theorem~\ref{linear n.t.p.o} we can explore the metric

219: space $Z$. And indeed we have the following result.

220: 

221: 

222: \begin{theorem}\label{main linear theorem} The metric space $(Z,d)$,

223: derived from the partial order above, is isometric to

224: $\mathbb{R}$ with the standard metric.

225: \end{theorem}

226: 

227: 

228: \textbf{Remark.}  One can prove theorem~\ref{linear n.t.p.o} using

229: a criterion developed in~\cite{El-P}. Briefly this criterion says

230: that in order to establish a non trivial partial order it is

231: enough to prove that there is no contractible loop in $Sp$ which

232: is generated by a strictly positive Hamiltonian. We should remark

233: that the criterion appears in a larger context in~\cite{ El-P}. In

234: our case the criterion follows from the fact that the Maslov index

235: of any positive loop is positive, and hence the loop can not be

236: contractible (see~\cite{M-S} and~\cite{R-S} for more information

237: on the Maslov index).

238: 

239: 

240: 

241: \subsection{The geometry of quantomorphisms and Hofer's

242: metric}\label{pre. hofer}We start with some preliminary

243: definitions and constructions before presenting the main results

244: of this section (see~\cite{E-K-P} and~\cite{El-P}).

245: 

246: Let $(P,\xi)$ be a prequantization space of a symplectic manifold

247: $(M,\omega)$ where $[\omega] \in H^{2}(M,\mathbb{Z})$.

248: Topologically, $P$ is a principal $S^1$-bundle over $M$. It

249: carries a distinguished $S^1$-invariant contact form $\alpha$

250: whose differential coincides with the lift of $\omega$. The

251: subgroup $Q \subset Cont(P,\xi)$ consisting of all

252: contactomorphisms which preserve $\alpha$ is called {\it the group

253: of quantomorphisms} of the prequantization space, (we will give

254: some more basic information at the beginning of Section~\ref{quant

255: section}).

256: 

257: We Denote by $\widetilde{Q}$ the universal cover of $Q$, and by

258: $\widetilde{Ham}(M,\omega)$ the universal cover of $Ham(M,\omega)$

259: relative to the identity. We have the following important lemma.

260: 

261: \begin{lemma}\label{cw} There is an inclusion \begin{equation}\widetilde{Ham}(M,\omega)\hookrightarrow \widetilde{Q}\end{equation}

262: of $\widetilde{Ham}(M,\omega)$ into $\widetilde{Q}.$ \end{lemma}

263: 

264: \begin{proof}

265: 

266: Let $\alpha$ be the contact form of the prequantization. The field

267: of hyperplans $\xi:=\{ker \alpha \}$ is called the contact

268: structure of $P$. Every smooth function $H:P \times S^{1}

269: \rightarrow \mathbb{R}$ gives rise to a smooth isotopy of

270: diffeomorphisms, which preserve the contact structure, as we now

271: explain. First, define the Reeb vector field of $\alpha$, $Y$, to

272: be the unique vector field which satisfies the following equations

273: $$\iota(Y)d \alpha=0 \text{   }, \alpha(Y)=1 .$$

274: 

275: 

276: Now given a smooth function $H$ as above define $X_{H}$ to be the

277: unique vector field which satisfies the following two equations

278: \begin{equation}\label{good}(1)\text{  }\iota(X_{H}) \alpha =H \text{   }, (2) \text{  }\iota(X_{H})d

279: \alpha=-dH+(\iota(Y)dH) \alpha \end{equation}

280: 

281: It is well known that the elements of the flow generated by

282: $X_{H}$ preserve the contact structure. Another important fact is

283: that the vector field which correspond to the constant Hamiltonian

284: $H\equiv 1$ is the Reeb vector field, and the flow is the obvious

285: $S^{1}$ action on $P$.

286: 

287: Now, let $F:M{\times}S^{1}{\rightarrow}\mathbb{R}$ be any

288: Hamiltonian on $M$ and let $X_{F}$ be its Hamiltonian vector

289: field. Let $p^{*}F$ be its lift to $P$, by definition $p^{*}F$ is

290: constant along the fibers of $P$.

291: 

292: We claim that

293: $$p_{*}X_{p^{*}F}=X_{F}.$$

294: 

295: To see this we substitute $X_{p^{*}F}$ in equation 2

296: of~\eqref{good}. We get $$\iota(X_{p^{*}F})d

297: \alpha=-dp^{*}F+(\iota(Y)dp^{*}F) \alpha$$ remembering that $d

298: \alpha=p^{*}\omega$ we have the equation

299: 

300: $$\iota(X_{p^{*}F})p^{*}\omega=-dp^{*}F+(\iota(Y)dp^{*}F) \alpha$$

301: 

302: $\Longleftrightarrow$

303: 

304: For every tangent vector to $P$, $v$, we have

305: 

306: 

307: $$\iota(p_{*}X_{p^{*}F})\omega(p_{*}v)=-dp^{*}F(v)+(\iota(Y)dp^{*}F) \alpha(v)$$

308: 

309: 

310: $\Longleftrightarrow$

311: 

312: \begin{equation}\label{almost}\iota(p_{*}X_{p^{*}F})\omega(p_{*}v)=-p^{*}dF(v)+(\iota(Y)dp^{*}F)

313: \alpha(v).\end{equation}

314: 

315: Now note that $\iota(Y)dp^{*}F\equiv 0.$ This is since $Y$ is

316: tangent to the fibers of $P$, and since $p^{*}F$ is constant along

317: the fibers, its differential has no vertical component with

318: respect to any local system of coordinates.

319: 

320: Thus, we get from equation~\eqref{almost} the equation

321: 

322: $$\iota(p_{*}X_{p^{*}F})\omega(p_{*}v)=-dF(p_{*}v)$$

323: for every tangent vector $v$. Now since the linear map $p_{*}$ is

324: surjective on every point of $P$, we get the equation

325: $$\iota(p_{*}X_{p^{*}F})\omega=-dF.$$

326: The fact that $\omega$ is non degenerate gives the desired

327: equation $p_{*}X_{p^{*}F}=X_{F}.$

328: 

329: 

330: 

331: 

332: 

333: 

334: 

335:  Denote by $\mathcal{F}$ the set of all time-1-periodic functions

336: 

337: 

338:  \begin{equation}F:M{\times}S^{1}{\rightarrow}\mathbb{R}\text{    such that   }

339:  \int_{M}{F(x,t)\omega^{n}}=0 \text{ for every  } t \in

340:  [0,1].\end{equation}

341:  %Let $F \in \mathcal{F}$ be

342: %a Hamiltonian function generating an element of

343: %$\widetilde{Ham}(M,\omega)$. Let $p:P \rightarrow M$ be the fiber

344: %map of $P$ over $M$. Then using $p$ we can lift $F$.

345: 

346: 

347: 

348: 

349: 

350: 

351: 

352: 

353: 

354: 

355: 

356: 

357: 

358: 

359: 

360: 

361: 

362: 

363: 

364: 

365: 

366: 

367:  It is not hard to check that for every representative,

368: $\{f_{t}\}_{t\in[0,1]},$ of an element of

369: $\widetilde{Ham}(M,\omega)$ there is a unique Hamiltonian

370: $F\in\mathcal{F}$ which generates it.

371: 

372: 

373:  Let $F$ be any Hamiltonian. The lift of $F$ is the contact Hamiltonian

374: $$\widetilde{F}:P{\times}S^{1}{\rightarrow}\mathbb{R}$$ defined by $\widetilde{F}:=p^{*}F$

375: where $p:P \rightarrow M$ is the projection from $P$ to $M$  (Note

376: that the function $\widetilde{F}$ is constant along the fibers of

377: $P$).  Now, $\widetilde{F}$ is a contact Hamiltonian function of

378: the contact manifold $P$ which generates an element of

379: $\widetilde{Q}$. It is well known that every representative of an

380: element of $\widetilde{Q}$, is generated by a unique contact

381: Hamiltonian which is a lift of an Hamiltonian defined on $M.$ We

382: further remind the reader that $Q$ is a central extension of

383: $Ham$. That is we have the short exact sequence $$\Id \rightarrow

384: S^{1} \rightarrow Q \rightarrow \text{  } Ham \text{  }

385: \rightarrow  \Id .$$

386: 

387: Which is from the Lie algebra point of view, the Poisson bracket

388: extension of the algebra of Hamiltonian vector fields. That is

389: 

390: $$0 \rightarrow \mathbb{R} \rightarrow  C^{\infty}(M) \rightarrow

391: \text{  }\mathcal{F} \text{  } \rightarrow 0 .$$

392: 

393: 

394: We first explain how to map the identity element of

395: $\widetilde{Ham}$, $[\Id]$, into $\widetilde{Q}$. Let $f_{t}$ be

396: any contractible loop representing $[\Id]$, and let $F_{t}$ be the

397: unique normalized Hamiltonian which generate it. The path

398: $\hat{f}_{t}$, in $Q$, which is generated by the lift of $F_{t}$,

399: $p^{*}F_{t}$, is a lift of $f_{t}$. Nevertheless, it is not

400: necessarily closed (actually it is closed, as we show below). It

401: easy to check that there is a constant, $c$, such that

402: $p^{*}F_{t}+c$ generates a lift, $\hat{h}(t)$, of the loop $f_{t}$

403: and this lift is a loop based at the identity of $Q.$ We claim

404: that

405: 

406: \textbf{a.} The loop $\hat{h}(t)$ is contractible.

407: 

408: \textbf{b.} The constant $c$ equals zero.

409: 

410: We first prove $a$. Since the loop $f_{t}$ is contractible, there

411: is a homotopy $$K_{s}(t):[0,1] \times [0,1] \rightarrow Ham$$ such

412: that for every $s$, $K_{s}(t)$ is a contractible loop. We denote

413: by $F_{s}(x,t)$ the normalized Hamilonian which generate the loop

414: $K_{s}$. As we explained, for every $s$ there is a constant $c(s)$

415: such that $p^{*}F_{s}+c(s)$ generates a loop, $\hat{h}_{s}(t)$, in

416: $Q$ which is a lift of $K_{s}.$

417: 

418: Now since  $\hat{h}_{s}(t)$ is a lift of the homotopy $K_{s}(t)$,

419: (which is a homotopy between  $f_{t}$ and $\Id$) it is a homotopy

420: between $\hat{h}(t)$ and the identity element of $Q.$ Thus

421: $\hat{h}(t)$ is contractible.

422: 

423: We now prove $b$. First we remind the reader the Calabi-Weinstein

424: invariant, which is well defined on the group $\pi_{1}(Q)$. It was

425: proposed by Weinstein in~\cite{W1}. For a loop $\gamma \in Q$

426: define the Calabi Weinstein invariant by

427: $$cw(\gamma)=\int_{0}^{1}dt \int \limits_{M}F_{t} \omega^{n}$$

428: 

429: where $p^{*}F_{t}$ is the unique Hamiltonian generating $\gamma.$

430: 

431: Now it is clear that $$cw([\Id])=0.$$

432: 

433: We conclude that for every $s$ $$\int_{0}^{1}dt \int

434: \limits_{M}(F_{s}(t,x)+c(s)) \omega^{n}=0$$

435: 

436: $\Longleftrightarrow$

437: $$c(s)vol(M)=-\int_{0}^{1}dt \int

438: \limits_{M}F_{s}(t,x) \omega^{n}.$$

439: 

440: Now since $F_{s}$ is normalized for every $s$ we get that $c(s)=0$

441: as desired.

442: 

443: The map of an arbitrary element of $\widetilde{Ham}$ is defined in

444: the same manner. Let $f_{t}$ be any representative of an element,

445: $[f]$, of $\widetilde{Ham}.$ Let $F_{t}$ be the unique normalized

446: Hamiltonian generating $f_{t}$. Then the image of $[f]$ is the

447: homotopy class of the path, $\hat{f}_{t}$, generated by the lift

448: of the Hamiltonian $F_{t}$, namely $p^{*}F_{t}$. By using the same

449: type of argument as above it can be shown that this map is well

450: defined. Finally, this map is clearly injective.

451: 

452: 

453: 

454: 

455: \end{proof}

456: 

457: 

458: The universal cover of $Q$ carries a natural normal cone $C$

459: consisting of all elements generated by non-negative contact

460: Hamiltonians. The set of dominants consists of those elements of

461: $\widetilde{Q}$ which are generated by strictly positive

462: Hamiltonian. The normal cone always gives rise to a genuine

463: partial order on $\widetilde{Q}$. Thus one can define the

464: corresponding metric space $(Z,d)$ as it was explained in

465: subsection \ref{basics on p.o.g}. We state this as a theorem.

466: 

467: \begin{theorem} Let $(M,\omega)$ be a symplectic form such that

468: $\omega \in H^{2}(M,\mathbb{Z})$. Let $(P,\xi)$ be a

469: prequantization of $(M,\omega).$ Then $\widetilde{Q}$, the

470: universal cover of the group of quantomorphisms of $Q$, is

471: orderable.

472: 

473: \end{theorem}

474: 

475: \begin{proof} According to ~\cite{El-P} all we need to check is

476: that there are no contractible loops in $Q$ which are generated by

477: a strictly positive contact Hamiltonian. Such contractible loops

478: can not exists since the Calabi-Weinstein invariant of any

479: representative of a contractible loop must be zero, and thus can

480: not be generated by a strictly positive Hamiltonian.

481: 

482: \end{proof}

483: 

484: In light of theorem~\ref{linear n.t.p.o} we should remark the

485: following.

486: 

487: 

488:  \noindent \textbf{Remark.}  Let $(M,\omega)$ be the

489:  complex projective space $\mathbb{C}P^{n-1}$ endowed with

490:  the Fubini-Study symplectic form normalized to be integral. Let $(P,\xi)$ be the sphere

491:  $S^{2n-1}$ endowed with the standard contact structure, and let $(\mathbb{R}P^{2n-1},\beta)$

492:  be the standard contact real projective space. It is well known that these two contact manifolds are

493:  prequantizations

494:  of $(M,\omega)$ (see~\cite{Ki},~\cite{W} and~\cite{Wo} for preliminaries on

495:  prequantization). Let

496:    $Cont(P,\xi)$ and $Cont(\mathbb{R}P^{2n-1},\beta)$ be the groups of all Hamiltonian contact transformations of

497:   these manifolds.

498: 

499: 

500: \noindent The relations of the symplectic linear group to these

501: groups is as follows. The group $ Sp(2n,\mathbb{R})$ is  subgroup

502: of all elements of $Cont(P,\xi)$ which commutes with $-\Id$, and $

503: Sp(2n,\mathbb{R})/\pm \Id$ is a subgroup of

504: $Cont(\mathbb{R}P^{2n-1},\beta).$

505:   Now, it is known that $Cont(\mathbb{R}P^{2n-1},\beta)$ admits a nontrivial partial order

506:   (while $Cont(P,\xi)$

507:   does not). The non orderabilty of  $Cont(P,\xi)$

508:   follows from~\cite{E-K-P}, this is done by

509:   using the criterion mentioned above. The orderabilty of $Cont(\mathbb{R}P^{2n-1},\beta)$ follows from the theory of the

510:   nonlinear Maslov index introduce by Givental see~\cite{G}.

511: 

512: 

513: \bigskip In what follows we establish a connection between the

514: partial order on $\widetilde{Q}$ (and its metric space $Z$) and

515: the Hofer's metric. So first let us recall some basic definitions

516: related to the Hofer's metric. The \textit{Hofer's distance}

517: between  $f \in \widetilde{Ham}(M,\omega)$ and $\Id$ is defined by

518: $$\rho(f,\Id)=\inf_{G} \int_0^1 \left ( \max_{x{\in}M}G(x,t)-\min_{x{\in}M}G(x,t) \right )dt, $$

519: where the infimum is taken over all time $1$-periodic Hamiltonian

520: functions $G$ generating paths in $Ham$ which belongs to the

521: homotopy class represented by $f$ with fixed end points

522: %of $\widetilde{Ham}$$f$ where

523: %the paths generated by these Hamiltonians are whiten the same

524: %homotopy class of the original path which we denote here by

525: %$fh^{-1}$

526:  (see~\cite{H-Z},~\cite{M-S} and~\cite{P}, for further

527: information). Furthermore, we define the\textit{ positive and the

528: negative part of the Hofer's distance} as

529: $$\rho_{+}(f,\Id) = \inf { \int_0^1 \max_{x \in M} {G(x,t)}}dt, \ \ \ \ \rho_{-}(f,\Id)=

530: \inf{\int_0^1-\min_{x{\in}M}{G(x,t)}dt},$$ % respectively.

531: where the infimum is taken as above.

532: 

533:  \noindent Now, for each $f \in \widetilde{Ham}(M,\omega)$, generated by some Hamiltonian, set

534:  \begin{equation} \| f \|_{+} = \inf \left \{ \max F(x,t) \right \} , \ \ \ \| f \|_{-} =

535:  \inf \left \{ - \min F(x,t) \right \}\end{equation}

536:  where the infimum is taken over all Hamiltonian functions $F \in

537:  \mathcal{F}$ generating paths in $Ham$ which belongs to the

538: homotopy class represented by $f$.

539:  %generating the flow having

540:  %$f$ as its time-1-map

541:   %and are within the same homotopy class of this path.

542:    The following result is due to Polterovich

543: see~\cite{P-1}.

544: %\begin{theorem}\label{p theorem}

545: 

546: 

547:  \begin{lemma} \textit{For every $f \in

548: \widetilde{Ham}(M,\omega)$ we have

549: \begin{equation}\label{+ equation}\! \! \! \rho_{+}(f)=\|f\|_{+}\end{equation}

550: \begin{equation}\rho_{-}(f)=\|f\|_{-}\ . \end{equation}}

551: \end{lemma}

552: 

553: 

554: We further define the positive and negative \textit{asymptotic}

555: parts of the Hofer's metric:

556: $$\|f\|_{+,\infty}:=\lim\limits_{n\rightarrow

557: \infty}\frac{\|f^{n}\|_{+}}{n}\text{ and }

558: \|f\|_{-,\infty}:=\lim\limits_{n\rightarrow

559: \infty}\frac{\|f^{n}\|_{-}}{n}.$$

560: 

561: 

562: 

563: Before stating our first main result we fix the following

564: notation.  Given $(P,\xi)$  (a prequantization of a symplectic

565: manifold $(M,\omega)$) we denote by $e^{is}$ the diffeomorphism of

566: $P$ obtained by rotating the fibers in total angle $s.$

567: %rotation in an angle $s$ of the fibers in $(P,\xi)$.

568:  Note that the set $\{e^{is}\}_{s\in \mathbb{R}}$ is a one parameter

569: family of contact transformations in $Q.$

570: 

571: 

572: \begin{theorem}\label{main theorem 1}

573: 

574:  Let $f$ be the

575: time-1-map of a flow generated by  Hamiltonian $F \in

576: \mathcal{F}$. Let $\widetilde{F}$ be the lift of $F$ to the

577: prequantization space and let $\tilde{f}$ be its time-1- map. Take

578: any $s \geq 0$ such that $e^{is}\tilde{f}$ is a dominant. Then

579: \begin{equation}\label{log}dist(\{e^{it}\},e^{is}\tilde{f}):=\inf\limits_{t}K(e^{it},e^{is}\tilde{f})=

580: \frac{1}{2}\log\frac{s+\|f\|_{+,\infty}}{s-\|f\|_{-,\infty}}.\end{equation}

581: \end{theorem}

582: 

583: Before stating our second result we need the following

584: preliminaries.

585: 

586: \begin{definition}{\rm  Let $(M,\omega)$ be a closed symplectic manifold.

587: Let $L$ be a closed Lagrangian submanifold of $M$. We say that $L$

588: has the \textit{Lagrangian intersection property} if $L$

589: intersects its image under any exact Lagrangian isotopy.}

590: \end{definition}

591: 

592: \begin{definition}{\rm  We say that \textit{L has the stable Lagrangian

593: intersection property} if $L\times \{r=0\}$ has the Lagrangian

594: intersection property in $(M\times T^{\ast}S^{1},\omega \oplus

595: dr\wedge dt)$ where $(r,t)$ are the standard coordinates on the

596: symplectic manifold $T^{\ast}S^{1}$ and $M\times T^{\ast}S^{1}$ is

597: considered as a symplectic manifold with the symplectic form

598: $\omega \oplus dr\wedge dt$.}

599: \end{definition}

600: 

601: For more information on Lagrangian intersections see~\cite{P}

602: chapter 6.

603: 

604: Consider now the following situation. For $(M,\omega)$ a closed

605: symplectic manifold, $[\omega] \in H^{2}(M,\mathbb{Z})$, assume

606: that we have an open dense subset $M_{0}$ of $M$, such that

607: $M_{0}$ is foliated by a family of closed Lagrangian submanifolds

608: $\{L_{\alpha}\}_{\alpha\in \Lambda}$, and each Lagrangian in this

609: family has the stable Lagrangian intersection property.

610: 

611: 

612: We denote by \texttt{F} the family of autonomous functions, in

613: $\mathcal{F}$, defined on $M$ which are constant when restricted

614: to $L_{\alpha}$ for every $\alpha,$ note that this means that the

615: elements of \texttt{F} commutes relative to the Poisson brackets.

616: That is for every $F,G\in \texttt{F}$ we have $\{F,G\}=0.$

617: 

618:  We denote by $\{\texttt{F},\|\text{ }\|_{max}\}$ the metric space of the family

619: of functions \texttt{F} endowed with the $max$ norm where the

620: $max$ norm is $\|F\|_{max}:=\max\limits_{x\in M}|F(x)|.$

621: 

622: In the course of the proof we will use the following subspace: Let

623: $V$ be the subspace of $\widetilde{Q}$ constitutes of all elements

624: generated by contact Hamiltonians of the form $s+\widetilde{F}$

625: where $\widetilde{F}$ is a lift of an Hamiltonian $F\in

626: \texttt{F}$ and $s+\widetilde{F}> 0.$

627: 

628: 

629: We now state our second result.

630: 

631: 

632: 

633: 

634: 

635: 

636:  \begin{theorem}\label{main theorem}

637: 

638:    There is an isometric injection of the metric space

639: $(\texttt{F},\|\text{ }\|_{max})$ into the metric space $Z$.

640: 

641: 

642: 

643: 

644: 

645: \end{theorem}

646: 

647: 

648: 

649: 

650: Note that the existence of $s$ in theorem~\ref{main theorem} is

651: justified due to the fact that the function $F$, and thus

652: $\widetilde{F}$, attains a minimum.

653: 

654: As we will show in section~\ref{quant section} theorem~\ref{main

655: theorem} can be implemented to the case of the standard even

656: dimensional symplectic torus

657: $T^{2n}=\mathbb{R}^{2n}/\mathbb{Z}^{2n}$. Our second example will

658: be for oriented surfaces of genus greater equal 2. In this case we

659: foliate the surface (minus a finite family of curves) with closed

660: non contractible loops which are all closed lagrangians with the

661: stable Lagrangian intersection property.

662: 

663: 

664: 

665: 

666: \section{ Proof of theorem~\ref{main linear theorem}}

667: \subsection{The unitary case}\label{unitary case}

668: In this subsection we prove theorem~\ref{main linear theorem} for

669: the unitary group $U(n)$. This is a result by itself. Moreover

670: elements of the proof will be used to prove theorem~\ref{main

671: linear theorem} for the symplectic case. Here we use occasionally

672: basic properties of the exponential map on the Lie algebra of a

673: matrix Lie group (see~\cite{H} chapter 2).

674: 

675: We denote by $\mathcal{A}$ the universal cover of $U(n)$. We view

676: $\mathcal{A}$ as the set of all paths starting at $\Id$, the

677: identity element, up to homotopy relation between paths with the

678: same end point. We will denote an element $[u]\in \mathcal{A}$,

679: with a slight abuse of notation, by $u$ where $u$ is a path

680: representing $[u]$.

681: 

682: As in the symplectic case (see definition~\ref{maslov quasi}) we

683: can define for an element $u \in \mathcal{A}$ its Maslov index as

684: $\alpha(1)-\alpha(0)$ where $e^{i2 \pi \alpha(t)}$= det$(u(t))$.

685: We will denote it by $\mu(u)$. Using the same definitions

686: of~\ref{linear geometry} we have the following. An element $u \in

687: \mathcal{A}$ is called \textit{semi positive} if for some

688: representative, the hermitian matrix $h_{u}$ defined by the

689: equation

690: \begin{equation}\label{complex positive} \dot{u}u^{-1}=ih_{u} \end{equation}

691: is semi positive definite. Note that if we consider $U(n)$ as a

692: subgroup of $Sp(2n,\mathbb{R})$ via the

693: realization$$ A+iB\mapsto \left( \! \! \! \begin{array}{rr} A & {-B} \\

694:  {B} &  A \end{array} \right)  $$

695:  equation~\eqref{complex positive} is the same as equation~\eqref{real

696:  positive}. We proceed as in~\ref{linear geometry}, define a

697:  partial ordering on $\mathcal{A}$, $\geq$, by $f \geq g$ if and

698:  only if $fg^{-1}$ is semi positive. Let $\mathcal{C}$ denote the

699:  subset of all semi positive elements and denote by $\mathcal{C^{+}}$ the set of positive

700:  definite elements

701:   of $\mathcal{A}$. These are, respectively, the normal

702:  cone and the set of dominants  of $\mathcal{A}$ ( see~\ref{basics on p.o.g}).

703: 

704:  %An element $u\in \mathcal{A}$ is \textit{eventually positive} if

705:  %there exists $n_{0}$ such that $u^{n}$ is positive for every

706:  %$n\geq n_{0}$, we denote the set of eventually positive elements by ${\mathcal{C}}^{+}_{0}$ .

707:   We define the function $\gamma_{n}$, the relative growth, and the metric

708:  space $Z$ in the same way as in~\ref{basics on p.o.g}. %As we will see the full characterization of the

709:  %eventually positive elements will enable us to calculate the

710:  %metric space $Z$.

711: 

712: \subsubsection{Preliminary basic lemmas}

713: 

714:  We need

715: the following two elementary lemmas.

716: \begin{lemma}\label{homotopy}Let $u$ and $v$ be two paths in $U(n)$ with the same

717: endpoints. Then $[u]=[v]$ if and only if $[|u|]=[|v|]$ in $U(1)$.

718: 

719: \end{lemma}

720: 

721: \begin{proof} It is well known that the determinant map

722: $U(n)\rightarrow U(1)$ induces an isomorphism between the

723: fundamental groups of these spaces. Let $\hat{\gamma}$ be the

724: inverse path to $\gamma$ and let $\ast$ be the juxtaposition of

725: paths. Thus $[u]=[v]$ if and only if $u \ast \hat{v}$ is homotopic

726: to the identity, if and only if $$|u \ast \hat{v}| = |u| \ast

727: \widehat{|v|}$$ is homotopic to the identity, if and only if

728: $[|u|]=[|v|].$

729: \end{proof}

730: 

731: \begin{lemma}\label{tr lemma} Assume that $u$, $h_{u}$ satisfies

732: equation~\eqref{complex positive}. Then \begin{equation}

733: \mu(u)=\int_{0}^{1}\text{tr}h_{u}(t)dt .\end{equation}

734: 

735: \end{lemma}

736: 

737: \begin{proof} For a complex path $z(t)\neq0$ we have of course

738: $\frac{d}{dt}\log z(t)=\frac{\dot{z}(t)}{z(t)}$. Taking the

739: imaginary part we have:

740: $$\frac{d}{dt}argz(t)=Im\frac{\dot{z}(t)}{z(t)}.$$ Fix $t=t_{0}$

741: in $[0,1]$ and write $u(t)=u(t_{0})v(t)$ where $v(t_{0})=\Id$. Let

742: $D$ be the differentiation with respect to $t$ at $t=t_{0}$. Note

743: that $$\dot{v}v^{-1}=ih_{u}.$$ Then $D|v|=\text{tr}(\dot{v})$ at

744: $t_{0}$, which implies that $$D\text{ }

745: arg|u|=Im\frac{D|u|}{|u|}=Im\frac{D|v|}{|v|}$$ $$=Im\text{

746: }\text{tr}( \dot{v})=Im\text{ }\text{tr}(ih_{u}v)=\text{tr}

747: (h_{u}).$$

748: 

749: In the computation above remember that $v=1$ at $t=t_{0}$. Now the

750: above formula follows as $arg|u(0)|=0.$

751: 

752: \end{proof}

753: 

754: As a corollary of the lemma we have that if $u$ is semi positive

755: then $\mu(u)\geq0$ (since if $h_{u}$ is semi positive then

756: $\text{tr}h_{u} \geq 0$). %Notice that the same thing is true if

757: %$u$ is only eventually positive, since $\mu(u^{n})=n\mu(u)$.

758: The converse of the corollary is not true (we will not give an

759: example). Nevertheless we have:

760: 

761: \begin{lemma}\label{main linear lemma} If $\mu(v) \geq 2\pi n$ then $v \geq \Id$. %In

762: %particular, every path with positive Maslov index number is

763: %eventually positive.

764: In this case, we can have a representative

765: $v(t)=e^{itA}$ where $\text{tr}(A)=\mu(v)$, $A$ is hermitian and

766: positive semidefinite, and if $\lambda_{1},...,\lambda_{n}$ are

767: the eigenvalues of $A$, then

768: $\max_{i,j}|\lambda_{i}-\lambda_{j}|\leq2 \pi n.$

769: 

770: \end{lemma}

771: 

772: \begin{proof} Suppose that $\mu(v) \geq 2\pi n$. Then according to

773: lemmas~\ref{homotopy} and~\ref{tr lemma} we can write

774: $v(t)=e^{itA}$ with $A$ hermitian. The choice of $A$ is not

775: unique, however $\text{tr}(A)=\mu(v)$ up to a multiple of $2\pi$.

776: Still, we can modify $A$ so that $\text{tr}(A)=\mu(v)$. To see

777: this, diagonalize $A$ by a unitary matrix and then modify the

778: eigenvalues by multiples of $2 \pi$ (here we are using the fact

779: that if $C$ is an invertible matrix then

780: $e^{CXC^{-1}}=Ce^{X}C^{-1}$ for any arbitrary matrix $X$).

781: 

782: In fact we can do better: We may still modify without changing

783: $\text{tr}A$ such that now $A$ will be positive semidefinite and

784: $\max_{i,j}|\lambda_{i}-\lambda_{j}|\leq2 \pi n.$ To see this,

785: after $A$ has been diagonalize, let $\lambda_{1},...,\lambda_{n}$

786: be the eigenvalues of $A$ (in an increasing order). Then we have

787: $$\sum \limits_{i}\lambda_{i} = \text{tr}A= \mu(v) \geq 2\pi

788: n.$$

789: 

790: \noindent Now let $0< \bar{\lambda}_{i}\leq 2 \pi$ be the unique

791: number that is congruent to $\lambda_{i}$ modulo $2 \pi

792: \mathbb{Z}.$  Then we have

793: 

794: $$\sum \limits_{i}\bar{\lambda}_{i}=\text{tr}A-2\pi k,\text{

795: }k>0.$$ Note that

796: $\max_{i,j}|\bar{\lambda}_{i}-\bar{\lambda}_{j}|\leq2 \pi .$ We

797: first assume that $k<n$ (the case $n=1$ is trivial so we assume $n

798: \geq 2$). In this case we change the $\bar{\lambda}_{i}\text{'s}$

799: by distributing the $k$ $2 \pi \text{'s}$ to the first $k$

800: $\bar{\lambda}_{i}\text{'s}.$ It is clear that the new maximal

801: $\bar{\lambda}_{i}$ will come from the first, new, $k$

802: $\bar{\lambda}_{i}\text{'s}$ and that the condition

803: $\max_{i,j}|\bar{\lambda}_{i}-\bar{\lambda}_{j}|\leq2 \pi n$ is

804: kept.

805: 

806: Now if $k \geq n$ then write $k=l n+m$ where $l,m \in \mathbb{N}$

807: and $m<n$. In this case we add $2 \pi$ to all the

808: $\bar{\lambda}_{i}\text{'s}$ $l$ times and then we add to the

809: first $m$ $\bar{\lambda}_{i}\text{'s}$ the remaining $m$ $2 \pi

810: \text{'s}.$ Now we are at the same position as in the first case

811: where $k < n.$ The resulting matrix $A$ (before diagonalization)

812: has the desired properties.

813: 

814: \end{proof}

815: 

816: \subsubsection{The Metric induced by $\mathcal{C}^{+}$}\label{c^+}

817: 

818: Let $f,g \in {\mathcal{C}}^{+}.$ We wish to compute $\gamma(f,g).$

819: We have the following theorem.

820: \begin{theorem}\label{hand theorem} For every non constant $f,g\in

821: {\mathcal{C}}^{+}$ we have: \begin{equation}

822: \gamma(f,g)=\frac{\mu(g)}{\mu(f)}.\end{equation}

823: 

824: \end{theorem}

825: \begin{proof} %First note that since the definition of $\gamma$ is

826: %eventually applied to dominants (see subsection~\ref{basics on

827: %p.o.g}) we can change the definition of $\gamma_{n}$ in the

828: %following away

829: %$$\gamma_{n}(f,g)=inf \{p\in \mathbb{N}|\text{ } f^{p}\geq

830: %g^{n}\}.$$ This is since if $f,g \in \mathcal{C}^{+}$ then if

831: %$f^{p}\geq g^{n}$ for $p$ negative integer then it will apply that

832: %$1 \geq f^{-p}g^{n}$ meaning that $1 \geq f^{-p}g^{n}$ and $1 \leq

833: %f^{-p}g^{n}$  meaning that $1=f^{-p}g^{n}$ meaning that one of the

834: %elements $f$ or $g$ can not be a dominant. We extend the

835: %definition of $\gamma_{n}$ in the obvious way to

836: %${\mathcal{C}}^{+}_{0}.$

837: 

838: We start with the following definition.

839: \begin{equation}\label{gamma*}\gamma^{*}(f,g)=\inf\{\frac{r}{s}|f^{r}\geq

840: g^{s}\text{ , }r \in \mathbb{Z}\text{ , }s \in

841: \mathbb{N}\}\end{equation} We claim that $\gamma^{*}=\gamma$.

842: Indeed, denote by $T$ the set of numbers which satisfies the

843: condition of the right hand side of~\eqref{gamma*}. Assume that $

844: \frac{r}{s}\in T $, then the equivalence of the definitions

845: follows from the inequality

846: $$r\geq \gamma_{s}(f,g)\geq s\gamma^{*}(f,g).$$ We claim that a

847: sufficient condition for $f^{r}\geq g^{s}$ is that $r

848: \mu(f)-s\mu(g) \geq 2 \pi n.$ Indeed, if so then

849: $\mu(f^{r}g^{-s})=r \mu(f)-s\mu(g) \geq 2 \pi n,$ so by

850: lemma~\ref{main linear lemma}, $$f^{r}g^{-s} \geq

851: \Id\Leftrightarrow f^{r}\geq g^{s}.$$ Now, given $\varepsilon >0$

852: choose $s> \frac{1}{\varepsilon}.$ Let $r \geq 0$ be the smallest

853: integer so that $$r \mu(f)-s\mu(g) \geq 2 \pi n.$$ Thus, $$r

854: \mu(f)-s\mu(g)<2 \pi n+\mu(f)\Leftrightarrow$$ $$\frac{r}{s}

855: \mu(f)-\mu(g) <(2 \pi n+\mu(f))\varepsilon \Leftrightarrow$$

856: $$\frac{r}{s}< \frac{\mu(g)}{\mu(f)}+\frac{(2 \pi

857: n+\mu(f))\varepsilon}{\mu(f)}.$$ Since $\varepsilon$ is arbitrary

858: then $$\gamma^{*}(f,g)=\inf\{\frac{r}{s}|f^{r}\geq g^{s}\} \leq

859: \frac{\mu(g)}{\mu(f)} .$$ Now $$\gamma(f,g) \gamma(g,f) \geq 1$$

860: gives the desired equality

861: 

862: 

863: $$\gamma(f,g)=\frac{\mu(g)}{\mu(f)}.$$

864: 

865: 

866: \end{proof}

867: 

868: Now we can finish the proof of theorem~\ref{main linear theorem}

869: in the unitary case. First, we remind the definition of the metric

870: $Z$. $Z=\mathcal{C}^{+}/\sim $ where $f\sim g$ provided $K(f,g)=0$

871: and $K(f,g)= \max\{\log\gamma(f,g),\log\gamma(g,f)\}.$ Define a

872: map $p:\mathcal{A}\rightarrow \mathbb{R}$ by $p(u)=\log(\mu(u)).$

873: We claim that the map $p$ induces an isomorphism of metric spaces

874: that is

875: $$\frac{\mathcal{C}^{+}}{\sim}=Z \cong \mathbb{R}.$$ Indeed, by

876: theorem~\ref{hand theorem} we get

877: $$|p(f)-p(g)|=\max \{\log \mu(g)/\mu(f), \log \mu(f)/\mu(g)\}=K(f,g).$$

878: Thus $p$ is an isometry.

879: 

880: \noindent \textbf{Remark.} Note also that $p$ preserve order as

881: well. Indeed, $f\geq g$ implies that $0\leq

882: \mu(fg^{-1})=\mu(f)-\mu(g)$ which implies that $p(f)\geq p(g).$

883: See~\cite{El-P} subsection 1.7 for more details on this phenomena,

884: in a larger context.

885: 

886: \subsection{The Maslov qusimorphism}\label{quasi maslov}

887: 

888: Before proving theorem~\ref{main linear theorem} in the symplectic

889: case we define an important property which we use in a crucial way

890: in the course of the proof.

891: 

892: \begin{definition} \label{quasi}

893: {\rm Let $G$ be a group. A \textit{quasimorphism} $r$ on $G$ is a

894: function $r:G\rightarrow\mathbb{R}$ which satisfies the

895: homomorphism equation up to a bounded error: there exists $R>0$

896: such that

897: $$|r(fg)-r(f)-r(g)|\leq R$$

898: for all $f,g\in G$}.

899: \end{definition} Roughly speaking a quasimorphism is a

900: homomorphism up to a bounded error. See ~\cite{Ba} for

901: preliminaries on quasimorphisms. A quasimorphism $r_{h}$ is called

902: \textit{homogeneous} if $r_{h}(g^{m})=mr_{h}(g)$ for all $g\in G$

903: and $m\in \mathbb{Z}$. Every quasimorphism $r$ gives rise to a

904: homogeneous one

905: $$r_{h}(g)=\lim \limits_{m\rightarrow\infty}\frac{r(g^{m})}{m}.$$

906: 

907: As we said a basic tool in the proof of Theorem \ref{main linear

908: theorem} is the {\it{Maslov quasimorphism}}  whose definition is a

909: generalization of the Maslov index from loops to paths in

910: $Sp(2n,{\mathbb{R}})$.

911: 

912: \begin{definition}\label{maslov quasi}{\rm  Let $\Psi(t)=U(t)P(t)$ be a representative of a point in $\widetilde{Sp}$

913: ,where $U(t)P(t)$ is the polar decomposition in $\widetilde{Sp}$

914: of $\Psi(t)$, $U(t)$ is unitary and $P(t)$ is symmetric and

915: positive definite. Choose $\alpha(t)$ such that $e^{i2\pi

916: \alpha(t)}$= det$(U(t))$. Define the Maslov quasimorphism $\mu$ by

917: $\mu([\Psi])=\alpha(1)-\alpha(0)$ }.

918: \end{definition}

919: \begin{theorem}\label{the quasi maslov} The Maslov quasimorphism is a

920: quasimorphism.

921: \end{theorem}

922: For the proof of theorem~\ref{the quasi maslov} see~\cite{BG}

923: and~\cite{D}. \noindent We denote by $\widetilde{\mu}$ the

924: homogeneous quasimorphism corresponding to $\mu$:

925: $${\widetilde{\mu}}(x)=\lim_{k{\rightarrow}\infty}{\frac{\mu(x^k)}{k}}. $$

926: We further remark that the restriction of the (homogeneous) Maslov

927: quasimorphism to  $\mathcal{A}$ (since there is an isomorphism

928: $\pi_{1}(Sp(2n,\mathbb{R}))\cong \pi_{1}(U(n))$, we may consider

929: $\mathcal{A}$ as a subset of $\widetilde{Sp}$) is the Maslov index

930: we have defined in subsection~\ref{unitary case} on $\mathcal{A}$.

931: 

932: \subsection{The Symplectic case}

933: 

934: We now turn to the symplectic case. Let $J$ be the standard

935: symplectic structure on $\mathbb{R}^{2n}$ (see~\ref{linear

936: geometry}). We recall that the symplectic linear group

937: $Sp(2n,\mathbb{R})$, is the group of all matrices $A$ which

938: satisfies $A^{T}JA=J.$ As we already remarked $U(n)$ is a subgroup

939: of $Sp(2n,\mathbb{R}).$ The elements of $U(n)$ are precisely those

940: which commute with $J$.

941: 

942:  \noindent We denote by $\mathcal{S}$ the universal

943: cover of this group having as base point the identity matrix

944: ${\Id}$. As before we use the same letter to denote an element in

945: $\mathcal{S}$ and a representing path. However we shall use

946: capital letters instead.

947: 

948: Now, assume that $X,Y \in \mathcal{S}$. Let $H_{X}, H_{Y}, H_{XY}$

949: be the Hamiltonians generating respectively $X,Y, XY$

950: (see~\ref{linear geometry}). For later use we need the following

951: formula. The formula is well known and it is easy to prove.

952: \begin{equation}\label{prod. formula} H_{XY}=

953: H_{X}+{X^{-1}}^{T}H_{Y}X^{-1}. \end{equation}

954: 

955: We denote by $\mathcal{S}^{+}$ the set of dominants of

956: $\mathcal{S}$. Note that we have the inclusion

957: $\mathcal{C}^{+}\hookrightarrow \mathcal{S}^{+}.$ Recall that

958: $\mathcal{S}^{+}$ induces a non trivial partial order on

959: $\mathcal{S}$ (see~\ref{linear geometry}), thus we can define for

960: every $f,g \in\mathcal{S}^{+}$ the function $\gamma(f,g)$. Here

961: one must be cautious as this magnitude may have two different

962: meanings for unitary $f,g$. According to the following theorem

963: this causes no difficulties (see the remark following

964: theorem~\ref{sp lemma}).

965: 

966: Recall that $\tilde{\mu}$ denote the homogenous Maslov

967: quasimorphism. The key theorem of this subsection is the

968: following.

969: \begin{theorem}\label{sp lemma} For all $X,Y \in \mathcal{S}^{+}$ we

970: have:\begin{equation}\label{main symp.

971: step}\gamma(X,Y)=\frac{\tilde{\mu}(Y)}{\tilde{\mu}(X)}.\end{equation}

972: 

973: \end{theorem}

974: 

975: 

976:  Theorem~\ref{main linear theorem}, now, follows from

977:  theorem~\ref{sp lemma}. Indeed define $p(X)=\log(\tilde{\mu}(X))$

978: for

979:  every $X \in \mathcal{S}^{+}.$ Now repeat verbatim the last part

980:  of~\ref{c^+}.

981: 

982: 

983: \textbf{Remark.} Notice that if $X$ and $Y$ are unitary, the

984: theorem shows that their symplectic relative growth is the same as

985: the unitary relative growth, since on unitary paths,

986: $\tilde{\mu}=\mu.$

987: 

988: We now prove theorem~\ref{sp lemma} in two steps.

989: \subsubsection{Proof of Theorem~\ref{sp lemma} - Step 1}

990: In step 1 we prove the following lemma.

991: \begin{lemma} For every positive definite symmetric symplectic

992: \textbf{matrix} $P$, there exists a positive path $X$ connecting

993: $\Id$ with $P$, such that \begin{equation}\mu(X)\leq 4 \pi

994: n.\end{equation}

995: \end{lemma}

996: \begin{proof} Without the loss of generality we may assume that

997: $P$ is diagonal. Otherwise, diagonalize $P$ by a unitary matrix

998: $U_{0}$ and notice that for every path $Y$, we have by a direct

999: calculation that $$H_{U_{0}YU_{0}^{-1}}=U_{0}H_{Y}U_{0}^{-1}$$

1000: where $H_{Y}$ is a Hamiltonian generating $Y$. Thus $X$ is

1001: positive if and only if $U_{0}XU_{0}^{-1}$ is.  Now, assume that

1002: the lemma has been proved for the case $n=1$. Using this

1003: assumption we prove the general case.

1004: 

1005: We first define $n$ different embeddings

1006: $$j_{i}:Sp(2,\mathbb{R})\hookrightarrow Sp(2n,\mathbb{R})\text{,        }1\leq i \leq n$$

1007: as follows. We copy a $2\times 2$ matrix $A$ to the $(i,n+i)$

1008: block of a $2n\times 2n$ matrix $B$. That is

1009: $$B_{ii}:=A_{11},\text{  }B_{ii+n}:=A_{12}$$ $$B_{i+ni}:=A_{21},\text{  }B_{i+ni+n}:=A_{22}.$$

1010:  The rest of the elements of $B$ are defined as follows (we assume

1011:  of course that $(k,l)\neq (i,i),(i,i+n),(i+n,i),(i+n,i+n)$).

1012:  $$B_{kl}:=\begin{cases}0 &k\neq l \\1&k=l\text{ .}\end{cases}$$

1013: 

1014:  These are at least homomorphisms to $GL(2n,\mathbb{R})$, but in

1015:  fact one can easily check that the image is symplectic. These

1016:  embeddings preserve transpose operator and thus symmetry and

1017:  orthogonality. As a result the embeddings respect

1018:  symmetric-unitary decomposition. In particular

1019:  \begin{equation}\label{mu}\mu(X)=\mu(j_{i}(X))\end{equation}

1020: 

1021:  \noindent for every $X\in Sp(2,\mathbb{R})$ and $1\leq i \leq n$.

1022: 

1023: 

1024: Note the following property of the embeddings $j_{i}$. For every

1025: $i\neq k$ and any $X,Y\in Sp(2,\mathbb{R})$ we have

1026: \begin{equation}\label{j commu.} j_{i}(X)j_{k}(Y)=j_{k}(Y)j_{i}(X)\end{equation}

1027: 

1028:  Now, by assumption $P$ is diagonal, let

1029:  $(\lambda_{1},...,\lambda_{n},1/\lambda_{1},...,1/\lambda_{n})$

1030:  be the diagonal of $P$. Define

1031:  $P_{i}=diag(\lambda_{i},1/\lambda_{i})$, and notice that

1032:  \begin{equation}\label{prod.1}P=\prod\limits_{i=1}^{n}j_{i}(P_{i})\end{equation}

1033: 

1034:  Having proved the theorem for $n=1$, we can find paths $X_{i}$ connecting

1035: $\Id$ (of $Sp(2,\mathbb{R})$) with $P_{i}$, such that

1036: $\mu(X_{i})\leq 4 \pi$. Define

1037: \begin{equation}\label{prod.2}X=\prod\limits_{i=1}^{n}j_{i}(X_{i})\end{equation}

1038: We write $j_{i}(X_{i})=P_{i}(X_{i})U_{i}(X_{i})$ for the polar

1039: decomposition of the matrix $j_{i}(X_{i})$ in $Sp(2n,\mathbb{R})$.

1040: Note that according to what we have remarked above, this

1041: representation of this polar decomposition is the image of the

1042: polar decomposition of $X_{i}$ with respect to the homomorphism

1043: $j_{i}$ for all $i$. Anyhow we get

1044: \begin{equation}\label{polar prod.}X=\prod\limits_{i=1}^{n}P_{i}(X_{i})U_{i}(X_{i})\end{equation}

1045: Note that according to~\eqref{j commu.} all the elements in the

1046: product of~\eqref{polar prod.} commutes. Thus we can write

1047: \begin{equation}\label{decom.polar.}X=\prod\limits_{i=1}^{n}P_{i}(X_{i})\prod\limits_{i=1}^{n}U_{i}(X_{i})\end{equation}

1048: We claim that the r.h.s. of~\eqref{decom.polar.} is the polar

1049: decomposition of $X$. Indeed, all the elements in the product

1050: $\prod\limits_{i=1}^{n}P_{i}(X_{i})$, according to~\eqref{j

1051: commu.}, commutes. Thus, this product is a symmetric (positive

1052: definite) matrix. The fact that the product

1053: $\prod\limits_{i=1}^{n}U_{i}(X_{i})$ is unitary prove our claim.

1054: We get that

1055: $$\mu(X)=\mu(\prod\limits_{i=1}^{n}U_{i}(X_{i}))=\sum\limits_{i=1}^{n}\mu(U_{i}(X_{i}))$$

1056: $$=\sum\limits_{i=1}^{n}\mu(X_{i}) \leq 4 \pi n.$$ Note that the

1057: last equality follows from~\eqref{mu}. Finally, $H_{X}$ is

1058: positive definite since it is a direct sum of positive definite

1059: Hamiltonians (see formula~\eqref{prod. formula}).

1060: 

1061: It remains to prove the case $n=1$. We produce here a concrete

1062: example. As argued above, we may assume that

1063: $P=diag(\lambda,1/\lambda)$. Consider the function

1064: $f(t)=\tan(\pi/4+at)$, where $|a|<\pi/4$ is chosen so that

1065: $f(1)=\lambda$. We then have $$|f'|=|a(f^{2}+1)|<f^{2}+1.$$ Now,

1066: let

1067: 

1068: 

1069: $$U(t):= \left( \! \! \! \begin{array}{rr} \cos(2\pi t) & -\sin(2 \pi t) \\

1070: \sin(2 \pi t)  &  \cos(2\pi t) \end{array} \right).  $$

1071: 

1072: \noindent Define the path $$X(t)=U(t)F(t)U(t)$$ where $F(t):=

1073: diag(f(t),1/f(t))$. We claim that $\mu(X)=4 \pi$. Indeed,

1074: $\mu(X)=\mu(U)+\mu(f)+\mu(U)=2 \pi +0+2 \pi =4 \pi$.

1075: 

1076: 

1077: 

1078: Finally we want to show that $X$ is positive.

1079: 

1080: \noindent Using~\eqref{prod. formula} we get:

1081: $$H_{X}=H_{U}+UH_{F}U^{-1}+UF^{-1}H_{U}F^{-1}U^{-1},$$

1082: 

1083: \noindent and we know that $$H_{U}=2 \pi I; \text{  and  }H_{F}= \left( \begin{matrix} 0  & f'(t)/f(t) \\

1084: f'(t)/f(t) & 0 \end{matrix} \right).$$ Thus $H_{X}$ is positive if

1085: and only if $$U^{-1}H_{X}U=H_{U}+H_{F}+F^{-1}H_{U}F^{-1}$$

1086: 

1087: $$=2\pi (I+F^{-2})+H_{F}=\left( \begin{matrix} 2

1088: \pi (1+f^{-2})  & f'(t)/f(t) \\

1089: f'(t)/f(t) & 2 \pi (1+f^{2}) \end{matrix} \right)$$

1090: 

1091: is positive. Since the diagonal entries of the latter matrix are

1092: positive, then the matrix is positive if and only if its

1093: determinantis positive. Indeed we have:$$4

1094: \pi^{2}(1+f^{2})(1+f^{-2})-(f'/f)^{2}=4

1095: \pi^{2}(f^{2}+1/f)^{2}-(f'/f)^{2}>0$$ where in the last inequality

1096: we have used the fact that $|f'|<1+f^{2}.$ This complete the proof

1097: of the lemma.

1098: 

1099: \end{proof}

1100: 

1101: 

1102: \subsubsection{Proof of Theorem~\ref{sp lemma} - Step 2}

1103: 

1104:  In what follows we denote by $C$ the constant which appears in

1105:  the definition of the Maslov quasimorphism. That is:$$|\mu(fg)-\mu(f)-\mu(g)|\leq

1106:  C$$for all $f,g \in \mathcal{S}$. It is not hard to prove the

1107:  following fact (the proof appears in the literature see~\cite{Ba}).

1108: \begin{lemma}\label{constants}

1109: Let $\tilde{\mu}$ be the homogeneous quasimorphism of $\mu$.

1110:  Then $$|\tilde{\mu}(X)-\mu(X)| \leq C\text{  for all X   }\in\mathcal{S}. $$

1111:  We then have $|\tilde{\mu}(X)-\tilde{\mu}(Y)| \leq 4C$.

1112:  \end{lemma} We will denote the constant $4C$ by $C_{1}$. The main

1113:  lemma of step-2 is the following.

1114:  \begin{lemma}\label{positive Y}Every element $Y \in \mathcal{S}$ for which

1115:  $\mu({Y})\geq 6 \pi n+C$ is positive.

1116:  \end{lemma}

1117:  \begin{proof} Let $Y(t)=P(t)V(t)$ be the polar decomposition of

1118:  $Y$. Let $X$ be the positive path, guaranteed by the main lemma of

1119:  step-1, ending at $P(1).$ Define $$Z(t)=X^{-1}(t)P(t).$$ Note

1120:  that $Z$ is a closed path, moreover we have \begin{equation}\label{eq.1}\mu(Z)\geq

1121:  \mu(P)-\mu(X)-C \geq -4 \pi n-C\end{equation} where in the first inequality

1122:  we have used the quasimorphism property, and in the second

1123:  inequality we have used the fact that for every $X   \in\mathcal{S}

1124:  $ we have $\mu(X)=-\mu(X^{-1}).$ From~\eqref{eq.1} we

1125:  get:$$\mu(X^{-1}Y)=\mu(X^{-1}PV)=\mu(ZV)$$

1126:  $$=\mu(Z)+\mu(V)=\mu(Z)+\mu(Y)>2 \pi n.$$ Now, since $X^{-1}Y(1)$

1127:  is unitary, then $X^{-1}Y$ is homotopic to its unitary

1128:  projection, we denote by $T$. So we have $\mu(T)=\mu(Y)>2 \pi n$

1129:  and from~\ref{main linear lemma} it follows that $T\geq\Id$ in the

1130:  unitary sense and thus in the symplectic sense. Consequently

1131:  $X^{-1}Y$is positive and thus $Y=XX^{-1}Y$ is positive.

1132: 

1133:  \end{proof}

1134: 

1135: 

1136:  We are now ready to prove the main theorem:

1137: 

1138:  We wish to compute $\gamma(X,Y)$. We claim that a sufficient

1139:  condition for $X^{r}\geq Y^{s}$ to hold is

1140:  \begin{equation}\label{condition}

1141:  r\tilde{\mu}(X)-s\tilde{\mu}(Y)\geq 6 \pi

1142:  n+2C+C_{1}.\end{equation}

1143: 

1144:  Indeed, if this condition holds, then $$\mu(X^{r}Y^{-s}) \geq

1145:  \tilde{\mu}(X^{r}Y^{-s})-C \geq \tilde{\mu}(X^{r}) +

1146:  \tilde{\mu}(Y^{-s})-C-C_{1}$$ $$=r

1147:  \tilde{\mu}(X)-s\tilde{\mu}(Y)-C-C_{1} \geq 6 \pi n+C.$$

1148: 

1149: 

1150:  Where we have used in the first and the second inequalities

1151:  lemma~\ref{constants}, and in the third inequality we have used

1152:  our assumption.

1153: 

1154:  Now, by lemma~\ref{positive Y}, we see that $X^{r}Y^{-s} \geq \Id$

1155:  or that  $X^{r} \geq Y^{s}.$ Since the r.h.s.

1156:  of~\eqref{condition} is independent of $r$ and $s$, then for

1157:  every $\varepsilon \geq 0$ we can find large $r$ and $s$, which satisfy~\eqref{condition},

1158:  such that

1159:  $$0<\frac{r}{s}(\tilde{\mu}(X))-\tilde{\mu}(Y)<\varepsilon.$$

1160:  Since $\gamma^{*}(X,Y) \leq r/s$, we deduce that

1161:  $$\gamma^{*}(X,Y)=\gamma(X,Y) \leq \frac{\tilde{\mu}(Y)}

1162: {\tilde{\mu}(X))}.$$ The fact that $\gamma(X,Y)\gamma(Y,X)\geq 1$

1163: finishes the proof of theorem~\ref{sp lemma} and thus of

1164: theorem~\ref{main linear theorem}.

1165: 

1166: 

1167: 

1168: 

1169: 

1170: 

1171: 

1172: 

1173: 

1174: 

1175: 

1176: \section{Proof of Theorems ~\ref{main theorem 1} and ~\ref{main theorem}}\label{quant

1177: section}

1178: 

1179: %First, let us recall some basic properties of prequantization

1180: %manifolds (we refer the reader to~\cite{Ki},~\cite{M-S},~\cite{W}

1181: %and~\cite{Wo} for further information).

1182: %First, we add the preliminary data of the construction of contact

1183: %vector fields generating elements of $Quant$.

1184: Before we start proving the main theorems

1185:  we add here one more

1186: basic construction. That is, the construction of vector fields

1187: generating the elements of the subgroup, of the contactomorphisms

1188: group, $Quant$ (recall that $Quant$ is the subgroup of all

1189: contactomorphisms which preserve the contact form, see

1190: subsection~\eqref{pre. hofer}). The purpose of this construction

1191: is to give an intuitive geometrical description of the group

1192: $quant$.

1193: 

1194: submitted

1195: 

1196: 

1197: Given  $(M,\omega)$ a symplectic manifold, such that $\omega\in

1198: H^{2}(M,\mathbb{Z})$, and  $(P,\xi)$ its prequantization (as we

1199: remarked in~\ref{pre. hofer} $P$ carries the structure of a

1200: principal $S^{1}$ bundle over $M$) recall that in such a case the

1201: manifold $P$ carries a 1-form, $\alpha$, globally defined on $P$,

1202: such that $\xi=\{v\in T_{a}P|\alpha(v)=0,\text{ }\forall a\in

1203: P\}.$ Moreover $\alpha$ is a connection form on $P$ such that

1204: $\omega$ is its curvature . That is $d\alpha=p^{*}\omega$ where

1205: $p:P\rightarrow M$ is the fiber bundle projection from $P$ to $M$

1206: ( See~\cite{Na} and~\cite{Sp} chapter 8 as a source on connections

1207: and curvature of principal bundles). This means that for every

1208: $a\in P$ we have the smooth decomposition

1209: \begin{equation}\label{conn.decom.}T_{a}P=\xi_{|a}\oplus V_{a}\end{equation} where $V_{a}$ is the vertical

1210: subspace of $T_{a}P$ defined canonically as the subspace tangent

1211: to the fiber over the point $p(a)$, at the point $a.$ Equivalently

1212: $V_{a}$ can be defined as the restriction to $a$ of all vector

1213: fields which are the image under the Lie algebras homomorphisms

1214: between the Lie algebra of $S^{1}$ into the Lie algebra of vector

1215: fields of $P$ under the $S^{1}$ action (recall that this Lie

1216: algebra homomorphism is induced by the action of the Lie group

1217: $S^{1}$ on

1218: $P$).% derivative of all paths passing through $a$ obtained by the

1219: %right action of the Lie group $S^{1}$ on $P.$

1220: 

1221: Now, it can be easily verified that for every $x \in M$ and $a\in

1222: p^{-1}(x)$ $p_{*}(\xi_{|a})$ is an isomorphism between $T_{x}M$

1223: and $\xi_{|a}$. Assume now that $p^{*}F$ is a contact Hamiltonian

1224: on $P$ which is a lift of an Hamiltonian $F$ on $M$ (see

1225: subsection~\ref{pre. hofer}).

1226: 

1227: Using decomposition~\eqref{conn.decom.} one can describe the

1228: contact vector field obtained from $p^{*}F$ as follows: Denote by

1229: $X_{F}(x)$ the Hamiltonian vector field obtained by $F$ at the

1230: point $x$. Let $a \in p^{-1}(x)$ be a point above $x$. Then the

1231: contact vector vector field $X_{p^{*}F}$ at the point $a$ is the

1232: sum \begin{equation}X_{p^{*}F}=p_{*}^{-1}(X_{F}(x))\oplus

1233: v\end{equation} where $p_{*}^{-1}(X_{F}(x))\in \xi_{|a}$ and $v$

1234: is the unique vector in $V_{a}$ determined by the condition

1235: $\alpha(v)=F(x).$ One can say that the horizontal component of the

1236: contact vector field is determined by the (symplectic) Hamiltonian

1237: vector field and its vertical componenet is the measure of

1238: transversality (determined by $F$) to the horizontal field $\xi$.

1239: 

1240: Take for example any constant time dependant Hamiltonian

1241: $c(t):M\rightarrow \mathbb{R}.$ Then in this case we have

1242: $p_{*}^{-1}(X_{F}(x))=0$ and the vertical component of the contact

1243: vector field is $v(t)=\alpha(c(t))$. It is easily verified that

1244: the dynamics on $P$ after time $t$ is $$a\mapsto e^{ic(t)}(a)$$

1245: where we have used the notation from subsection~\ref{pre. hofer}.

1246: 

1247: 

1248: 

1249: 

1250: 

1251: \subsection{Proof of Theorem~\ref{main theorem 1}}

1252: We start with the following lemma in which we establish a

1253: connection between the Hofer's metric and the partial order.

1254: 

1255: %In the second one we give a formula for the Hofer's metric for the

1256: %family of Hamiltonians $\texttt{F}.$

1257: 

1258: \begin{lemma} Let $f$ be the time-1-map of a flow generated by

1259: Hamiltonian $F \in \mathcal{F}$. Let $\widetilde{F}$ be its lift

1260: to the prequantization space and let $\tilde{f}$ be its

1261: time-1-map. Then we have the formulas:

1262: \begin{equation}\label{onee}\|f\|_{+}=\inf\{s|\text{ }e^{is}\geq\tilde{f}\}

1263: \end{equation}

1264: 

1265: \begin{equation}\label{two}\|f\|_{-}=\inf\{s|\text{ }\tilde{f}\geq e^{-is}\}

1266: \end{equation}

1267: \end{lemma}

1268: 

1269: 

1270: \begin{proof}

1271:  We prove here formula~\eqref{onee}, formula~\eqref{two} is

1272: proved along the same lines. Assume that

1273: $$e^{is}\geq \tilde{f}\Leftrightarrow e^{is}\tilde{f}^{-1} \geq

1274: \Id $$

1275: 

1276: This means that $\exists H_{1} \geq 0$ which generates the element

1277: $e^{is}\tilde{f}^{-1}$.

1278: 

1279: \noindent Moreover we have that the contact Hamiltonian

1280: $H=H_{1}-s$ connects $\Id$ to $\tilde{f}^{-1}$ which implies that

1281: $-H$ connects $\Id$ to $\tilde{f}$. Now the fact that $H_{1} \geq

1282: 0 $ gives us that $-H \leq s$ $\Leftrightarrow$

1283:  $\max(-H) \leq s$.

1284: 

1285:  \noindent Remembering that $-H$

1286: generates $\tilde{f}$ (note that $H \in \mathcal{F}$, this can be

1287: seen by using the Calabi-Weinstein invariant in a similar way to

1288: the way we have use it in lemma~\ref{cw}) we have

1289: $$\|f\|_{+} \leq \inf\{s|\text{

1290: }e^{is}\geq\tilde{f}\}.$$

1291: 

1292: \noindent On the other hand assume that $s \geq \|f\|_{+}.$ This

1293: means that $\exists F$ such that $F \in \mathcal{F}$ and $\max F

1294: \leq s$. Define $-H=F.$ So we have $H+s \geq 0$ which implies that

1295: $e^{is}\tilde{f}^{-1}\geq \Id$ which implies that $e^{is} \geq

1296: \tilde{f}.$

1297: 

1298: \noindent Thus we conclude: $$\|f\|_{+} \geq \inf\{s|\text{

1299: }e^{is}\geq\tilde{f}\}$$  which is the desired.

1300: 

1301: \end{proof}

1302: 

1303: As a result we have the following corollary.

1304: 

1305: 

1306: \begin{corollary}\label{polt.lemma}Let $F:M\rightarrow \mathbb{R}$ be an Hamiltonian.

1307:  Let $\{f_{t}\}$ be its flow and let $f$ be the time-1-map of the

1308: flow. Let $\widetilde{F}$ be the lift of $F$ and denote by

1309: $\tilde{f}$ its time-1-map. Then

1310: \begin{equation}\label{1 of main} e^{is} \widetilde{f} \geq \Id

1311: \Longleftrightarrow \|f\|_{-} \leq s

1312: \end{equation} and \begin{equation}\label{2 of main} e^{is}

1313: \widetilde{f} \leq \Id \Longleftrightarrow \|f\|_{+} \leq -s

1314: \text{

1315:   .}

1316: \end{equation}

1317: \end{corollary}

1318: \begin{proof}

1319:  We use formula~\eqref{onee} to derive formula~\eqref{2 of

1320:  main}. The derivation of formula~\eqref{1 of main} from

1321:  formula~\eqref{two} can be shown in the same way.

1322: 

1323:  For the first direction note that $$\|f\|_{+} \leq -k \Rightarrow

1324:  \inf\{s|\text{ }e^{is}\geq\tilde{f}\} \leq -k$$

1325: 

1326:  $$\Rightarrow \text{    } \Id \geq e^{-i(-k)}\tilde{f} \text{   }

1327:  \Rightarrow \Id \geq e^{ik} \tilde{f}.$$

1328: 

1329:  We now show the other direction. $$e^{ik}\tilde{f} \leq \Id \text{  }

1330:  \Rightarrow \text{   } -k \geq \inf\{s|\text{

1331:  }e^{is}\geq\tilde{f}\}$$

1332:  $$\Rightarrow \|f\|_{+} \leq -k \text{,  as required  .}$$

1333: 

1334:  \end{proof}

1335: 

1336: We now prove the following formulas,

1337:  ($t,s,\tilde{f}$ are

1338: as in the theorem).

1339: 

1340: 

1341: \begin{equation}\label{first}\gamma(e^{it},e^{is}\tilde{f})=\frac{s+\|f\|_{+,\infty}}{t}\end{equation}

1342: 

1343: \begin{equation}\label{second}\gamma(e^{is}\tilde{f},e^{it})=\frac{t}{s-\|f\|_{-,\infty}}.\end{equation}

1344: 

1345: We will prove formula~\eqref{first}. Formula~\eqref{second} is

1346: proved along the same lines. First, recall that the function

1347: $\gamma$ can be defined alternatively as in~\eqref{gamma*}, which

1348: is the definition we use here. So assume that

1349: $$c:=\frac{r}{p}\geq \gamma(e^{it},e^{is}\tilde{f})$$

1350: $\Rightarrow$ $$e^{itr} \geq e^{isp}\tilde{f}^{p}$$ (here we use

1351: the fact that all maps commutes, and the alternative definition of

1352: $\gamma$) $$\Rightarrow e^{i(tr-sp)}\geq \tilde{f}^{p} \Rightarrow

1353: tr-sp \geq \|f^{p}\|_{+}$$ (here we use formula~\eqref{onee})

1354: $$\Rightarrow t \frac{r}{p} \geq \frac{\|f^{p}\|_{+}}{p}+s

1355: \Rightarrow c \geq \frac{s+\frac{\|f^{p}\|_{+}}{p}}{t}$$

1356: 

1357: since $p$ can be chosen big as we want, we conclude that $$ c \geq

1358: \frac{s+{\|f\|_{+,\infty}}}{t}$$ $\Rightarrow$

1359: $$\gamma(e^{it},e^{is}\tilde{f}) \geq

1360: \frac{s+\frac{\|f^{p}\|_{+}}{p}}{t}.$$

1361: 

1362: On the other hand assume that $$ c \geq

1363: \frac{s+{\|f\|_{+,\infty}}}{t}.$$ Then there exists a sequence of

1364: positive real numbers, $\{k_{n}\}$, such that $k_{n}\rightarrow

1365: \infty$ and

1366: $$c \geq \frac{s+\frac{\|f^{k_{n}}\|_{+}}{k_{n}}}{t}.$$ We

1367: conclude that

1368: $$tk_{n}c-sk_{n} \geq \|f^{k_{n}}\|_{+} \Rightarrow e^{i(\frac{trk_{n}}{p}-sk_{n})} \geq

1369: \tilde{f}^{k_{n}}$$

1370: 

1371: \noindent $\Rightarrow$

1372: \begin{equation}\label{frac}{e^{it}}^{\frac{rk_{n}}{p}} \geq

1373: (e^{is}\tilde{f})^{k_{n}}\end{equation}

1374: 

1375: Now choose a sequence $\alpha_{n}$ such that $0 \leq \alpha_{n}

1376: \leq 1$ and $\frac{rk_{n}}{p}+\alpha_{n} \in \mathbb{N}.$

1377: 

1378: From inequality~\eqref{frac} and the choice of $\alpha_{n}$ we

1379: get:

1380: $$e^{it({\frac{rk_{n}}{p}+\alpha_{n}})} \geq

1381: (e^{is}\tilde{f})^{k_{n}}.$$

1382: 

1383: From the last inequality and the definition of $\gamma$ we use

1384: here we get,

1385: 

1386: $$\frac{ \frac{rk_{n}}{p}+\alpha_{n}}{k_{n}} \geq

1387: \gamma(e^{it},e^{is}\tilde{f})$$

1388: 

1389:  $\Rightarrow$

1390: 

1391: $$\frac{r}{p}+\frac{\alpha_{n}}{k_{n}} \geq

1392: \gamma(e^{it},e^{is}\tilde{f}).$$

1393: 

1394: 

1395: \noindent Now, since $\lim \limits_{n \rightarrow \infty}

1396: \frac{\alpha_{n}}{k_{n}}=0$ we get that

1397: 

1398: 

1399: 

1400: 

1401: $$c=\frac{r}{p} \geq \gamma(e^{it},e^{is}\tilde{f})$$ which is

1402: what we need.

1403: 

1404: At this point we remark that due to~\eqref{first} and~\eqref

1405: {second} and the fact that $\gamma(f,g)\gamma(g,f) \geq 1$ for

1406: every $f$ and $g$, we infer that the r.h.s of~\eqref{log} is

1407: defined.

1408: 

1409: Now we can actually calculate $K(e^{it},e^{is}\tilde{f})$. From

1410: this calculation we will derive formula~\eqref{log}.

1411: 

1412:   By the very definition of $K$ and formulas~\eqref{first}

1413: and~\eqref{second} we have

1414: $$K(e^{it},e^{is}\tilde{f})=\max\{\log (s+\|f\|_{+,\infty})-\log

1415: t\text{, }\log t- \log (s-\|f\|_{-,\infty})\}$$

1416: 

1417: 

1418: $$=\frac{\log

1419: (s+\|f\|_{+,\infty})- \log (s-\|f\|_{-,\infty})}{2}+\frac{|2\log t

1420: -(\log (s+\|f\|_{+,\infty})+ \log (s-\|f\|_{-,\infty}))|}{2}$$ Now

1421: clearly this expression attains its infimum when

1422: 

1423: 

1424: $$t=e^{\frac{-(\log (s+\|f\|_{+,\infty})+ \log

1425: (s-\|f\|_{-,\infty}))}{2}}.$$

1426: 

1427: Thus we conclude that the l.h.s of~\eqref{log} equals

1428: $$=\frac{\log (s+\|f\|_{+,\infty})- \log

1429: (s-\|f\|_{-,\infty})}{2}$$ or simply

1430: $$\frac{1}{2}\log\frac{s+\|f\|_{+,\infty}}{s-\|f\|_{-,\infty}}$$

1431: which is the desired.

1432: 

1433: 

1434: 

1435: 

1436: 

1437: 

1438: 

1439: 

1440: 

1441: \subsection{Proof of Theorem~\ref{main theorem}}

1442: 

1443: We begin with the following lemma.

1444: 

1445: \begin{lemma} Let $F \in \texttt{F}$ be an Hamiltonian. Then we

1446: have the formulas:\begin{equation}\label{first eq.}\max

1447: F=\|f\|_{+}\end{equation} and \begin{equation}\label{second

1448: eq.}-\min F=\|f\|_{-}.\end{equation}

1449: \end{lemma}

1450: 

1451: \begin{proof}

1452: 

1453: 

1454: 

1455: Our starting point is the following fact which is due to

1456: Polterovich which can be easily deduced from~\cite{P-1}.

1457: 

1458: 

1459: \textbf{Fact.} Let $(M,\omega)$ be a symplectic manifold. Let $L$

1460: be a closed Lagrangian in $M$ with the stable Lagrangian

1461: intersection property. Moreover, let $F$ be an autonomous

1462: Hamiltonian defined on $M$ such that $F \in \mathcal{F}$ (see

1463: subsection~\ref{pre. hofer}) and for some positive constant $C$ we

1464: have $F_{|L} \geq C$. Denote by $f$ the time-1-map defined by $F$.

1465: Then we have

1466: \begin{equation}\label{p}\|f\|_{+} \geq C.\end{equation}

1467: 

1468: 

1469: 

1470: 

1471: 

1472: Now let $L_{\alpha}$ be a Lagrangian of the family of Lagrangians

1473: which foliate $M_{0}$ as in theorem~\ref{main theorem}. Assume

1474: that $F_{|L_{\alpha}} \geq C_{\alpha}.$ Then using~\eqref{p} we

1475: have the following double inequality.

1476: $$\max F\geq \|f\|_{+} \geq C_{\alpha}\text{, }\forall \alpha$$ where in the first

1477: inequality we have used the very definition of the norm $\|\text{

1478: }\|_{+}$. Now due to the fact that the set of Lagrangians foliate

1479: an open dense set in $M$ we get that

1480: $$\max F=\max \limits_{\alpha} \max \limits_{L_{\alpha}}F.$$ From this

1481: we get we get formula~\eqref{first eq.}.

1482: 

1483: In the same manner we obtain formula \eqref{second eq.}.

1484: \end{proof}

1485: %Now we can prove part $a$ of theorem~\ref{main theorem}. Recall

1486: %that the path $e^{is}:=\{e^{it}\}_{0\leq t \leq s}$, considered as

1487: %generated by any time dependant constant contact Hamiltonian

1488: %$\dot{s}(t)$ such that $\int_{0}^{1}\dot{s}(t)dt=s.$ Assume that

1489: %$e^{is} \widetilde{f} \geq \Id.$ This means that we have a

1490:  %contact Hamiltonian $G$ generating $e^{is} \widetilde{f}$ such that $$\min G \geq 0.$$

1491: %Now define $F:=G-s$. Clearly we have $\min F=\min G -s$. So we

1492: %have that $$-\min F \leq s.$$ Finally note that since the constant

1493: %Hamiltonian $s$ generates $e^{is}$ we infer that $F$ generates

1494: %$f.$ So we have that

1495: %$$\|f\|_{-} \leq s.$$

1496: % we finish this direction of the proof.

1497:  %$\dot{s}+\widetilde{F}(x) \geq 0$ such that $s(0)=0$ and

1498: %$s(1)=s.$ So, we have

1499: %\begin{equation}\label{arrows}\min \limits_{x}F(x) \geq

1500: %-\dot{s}\Rightarrow-\int_{0}^{1}\min \limits_{x}F(x)dt

1501: %\leqint_{0}^{1}\dot{s}dt=s \end{equation}

1502: 

1503: %For the other direction we choose the constant contact Hamiltonian

1504: %$\dot{s}(t)=s.$

1505: %and all is left to do is to go the other direction

1506:  %in~\eqref{arrows}.

1507:  %So we have

1508: %$$0\leq \dot{s}+\widetilde{F}\Leftarrow\min \limits_{x}F(x)+\dot{s}

1509: %\geq0 \Leftarrow-\min \limits_{x}F(x) \leq \dot{s}=s $$ which

1510: %implies, by formula~\eqref{second eq.}, that $e^{is} \widetilde{f}

1511: %\geq \Id.$ This complete the proof of formula~\eqref{1 of main}.

1512: %Formula~\eqref{2 of main} is proved in the same way.

1513: %\end{proof}

1514: 

1515: 

1516:  \noindent We now calculate the relative growth on elements of $V$ (see~\ref{pre. hofer} for the definition of $V$).

1517: 

1518:  \noindent Let

1519: $F,G \in \texttt{F}$. Denote by $\varphi_{F} \text{  the time-1-

1520: map of }F$ and by $\varphi_{G} \text{  the time-1-map of }G.$ Let

1521: $\widetilde{\varphi}_{F}$, $\widetilde{\varphi}_{G}$ be their lift

1522: to the prequantization space. Now let $s,t$ be any real numbers

1523: such that

1524: $$e^{is}\widetilde{\varphi}_{F},e^{it}\widetilde{\varphi}_{G}\geq

1525: \Id.$$ Note that this means that

1526: $e^{is}\widetilde{\varphi}_{F},e^{it}\widetilde{\varphi}_{G}$ are

1527: in $V$ and of course all elements of $V$ can be characterize in

1528: this way. Now, by the very definition of the relative growth we

1529: have

1530: $$\gamma(e^{is}\widetilde{\varphi}_{F},e^{it}\widetilde{\varphi}_{G})

1531: =\lim\limits_{n\rightarrow\infty}\frac{\gamma_{n}(e^{is}\widetilde{\varphi}_{F},e^{it}\widetilde{\varphi}_{G})}{n}.$$

1532: 

1533: We calculate $\gamma$ by a direct calculation of the functions

1534: $\gamma_{n}$ on elements of $V$.

1535: 

1536: By the very definition of $\gamma_{n}$ we have

1537: 

1538: 

1539: $$\gamma_{n}(e^{is}\widetilde{\varphi}_{F},e^{it}\widetilde{\varphi}_{G})=

1540: \inf\{m|\text{ }e^{ims}{\widetilde{\varphi}}^{m}_{F}\geq

1541: e^{int}{\widetilde{\varphi}}^{n}_{G}\}$$

1542: 

1543: $$=\inf\{m|\text{

1544: }e^{i(ms-nt)}\widetilde{\varphi}_{mF-nG} \geq \Id\}=\inf\{m|\text{

1545: }\|\varphi_{mF-nG}\|_{-}\leq ms-nt\}$$

1546: 

1547: (where in the second equality we have used the fact

1548:   that the functions of the union $\texttt{F}\cup \{e^{is}\}_{s\in \mathbb{R}}$ are all Poisson

1549: commutes, in the third equality we have used

1550: corollary~\ref{polt.lemma})

1551: 

1552: $$=\inf\{m|\text{

1553: }-\min(mF-nG)\leq ms-nt\}.$$ this follows  from

1554: formula~\eqref{second eq.}.

1555: 

1556:  So we need $m$ that will satisfy

1557: $$\max(nG-mF) \leq ms-nt\Leftrightarrow  $$

1558: 

1559:  $$ nG-mF\leq ms-nt \Leftrightarrow n(G+t)-m(F+s) \leq 0$$

1560: $$\Leftrightarrow n(G+t) \leq m(F+s) $$

1561: 

1562: dividing both sides of the last inequality by the positive

1563: function $F+s$ (recall that $F+s$ generates the dominant

1564: $e^{is}\widetilde{\varphi}_{F}$) we get

1565:  $$\Leftrightarrow \max (\frac{G+t}{F+s})n \leq m.$$ Thus we have

1566: 

1567: 

1568: $$\gamma_{n}(e^{is}\widetilde{\varphi}_{F},e^{it}

1569:  \widetilde{\varphi}_{G})-1 \leq \max(\frac{G+t}{F+s})n \leq \gamma_{n}(e^{is}\widetilde{\varphi}_{F},e^{it}

1570:  \widetilde{\varphi}_{G})$$ thus

1571: 

1572: 

1573:  $$\frac{\gamma_{n}(e^{is}\widetilde{\varphi}_{F},e^{it}

1574:  \widetilde{\varphi}_{G})-1}{n} \leq \max(\frac{G+t}{F+s}) \leq \frac{\gamma_{n}(e^{is}\widetilde{\varphi}_{F},e^{it}

1575:  \widetilde{\varphi}_{G})}{n}.$$

1576: 

1577: 

1578:  So we have the formula

1579: 

1580:  \begin{equation}\label{gamma for.}\gamma(e^{is}\widetilde{\varphi}_{F},e^{it}\widetilde{\varphi}_{G})

1581:  =\max(\frac{G+t}{F+s}).

1582:  \end{equation}

1583: 

1584: 

1585:  %Now we can define the embedding of $(\texttt{F})$ into $Z$.

1586:  Now, let $\widetilde{f},\widetilde{g}\in V$ generated by the

1587:  Hamiltonians $\widetilde{F},\widetilde{G}$ respectively. Then

1588:  according to formula~\eqref{gamma for.} we have

1589:  \begin{equation}\label{K}

1590:  K(\widetilde{f},\widetilde{g})=\max|\log\widetilde{F}-\log\widetilde{G}|

1591: \end{equation}

1592: 

1593: 

1594: 

1595: Using formula~\eqref{K}, for $K$, we define the isometry of

1596: $\{\texttt{F},\|\text{ }\|_{max}\}$ to $Z.$

1597: %the space of smooth

1598: %function depending only on $\Lambda$ with the $\max$ norm.

1599: 

1600: Let $\widetilde{f} \in V $ generated by the Hamiltonian

1601: $\widetilde{F}$. Then the correspondence

1602: $\widetilde{f}\leftrightarrow \log\widetilde{F}$ clearly induces

1603: the required isometric imbedding of $\{\texttt{F},\|\text{

1604: }\|_{max}\}$ into $Z$. This conclude the proof of this part of the

1605: theorem.

1606: 

1607: 

1608: 

1609: \subsubsection{Examples}

1610: 

1611: \textbf{Example 1.} Consider the $2n$ dimensional standard

1612: symplectic torus. That is $T^{2n}=\mathbb{R}^{2n}/\mathbb{Z}^{2n}$

1613: with the symplectic structure $dP \wedge dQ=\sum

1614: \limits_{i=1}^{n}dp_{i} \wedge dq_{i}.$ We now foliate the tours

1615: by a family of Lagrangians which depends only on the $P$

1616: coordinate. This way we exhibit the tours as a Lagrangian

1617: fiberation parameterize by an $n$ tours (the $P$ coordinate which

1618: in the notations of theorem~\ref{main theorem} is the parameter

1619: $\Lambda$) and with Lagrangian fibers (the $Q$ coordinate). It is

1620: known that this family of Lagrangian has the stable Lagrangian

1621: intersection property. Now, define the family of autonomous

1622: functions on $T^{2n}$ which depends only on the $P$ coordinate.

1623: Thus all the conditions of theorem~\ref{main theorem} are

1624: satisfied (we remained the reader that all functions of this

1625: family are Poisson commuting). \noindent We conclude that the

1626: metric space $Z$ of a prequantization space of the standard $2n$

1627: dimensional symplectic torus contains an infinite dimensional

1628: metric space. Denote by $C^{\infty}_{nor}(T^{n})$ the space of

1629: smooth normalized functions on the $n$-dimensional torus. Then we

1630: have the isometric injection:$$(C^{\infty}_{nor}(T^{n}),\|\text{

1631: }\|_{\max})\hookrightarrow Z.$$

1632: 

1633: 

1634: 

1635: 

1636: \begin{figure}[h!t]

1637: \begin{minipage}{4.5cm}

1638: \begin{center}

1639: \epsfig{file=surface2.eps,width=1.9cm} \caption{The curves we

1640: remove from $\Sigma$. } \label{surface_2}

1641: \end{center}

1642: \end{minipage}

1643: \hfill

1644: \begin{minipage}{4.5cm}

1645: \begin{center}

1646: \epsfig{file=foliat.eps,width=1.9cm} \caption{The foliation of

1647: $\Sigma_{0}$. } \label{foliation}

1648: \end{center}

1649: \end{minipage}

1650: \end{figure}

1651: 

1652: \bigskip

1653: \textbf{Example 2.} Let $\Sigma$ be any surface of genus greater

1654:  or equal 2. It is known that non contractible loops on $\Sigma$

1655:  has the stable Lagrangian property (see for example~\cite{P-S}). For such surfaces one can

1656:  foliate a subset, $\Sigma_{0}$, of $\Sigma$ by a family of disjoint non

1657:  contractible closed loops such that $\Sigma_{0}$ is an open dense

1658:  subset of $\Sigma$ (actually $\Sigma\setminus \Sigma_{0}$

1659:  is a finite collection of closed arcs-see the figures 1 and 2).

1660:  Note that all the conditions of theorem~\eqref{main theorem} are satisfied. In the figures 1 and 2 we show an

1661:  example of such family of Lagrangians for surface of genus equal

1662:  2.

1663: 

1664: 

1665: 

1666: 

1667: 

1668: 

1669: 

1670: 

1671: 

1672: 

1673: 

1674: 

1675: 

1676: 

1677: 

1678: 

1679: \bigskip

1680: 

1681: 

1682: 

1683: 

1684: 

1685: 

1686: 

1687: 

1688: 

1689: 

1690: 

1691: 

1692: 

1693: 

1694: 

1695: 

1696: 

1697: 

1698: 

1699: 

1700: 

1701: 

1702: 

1703: 

1704: 

1705: 

1706: 

1707: 

1708: 

1709: 

1710: 

1711: 

1712: 

1713: 

1714: 

1715: 

1716: 

1717: 

1718: 

1719: 

1720: 

1721: 

1722: 

1723: 

1724: 

1725: 

1726: \textbf{Acknowledgements.} I want to thank deeply  Leonid

1727: Polterovich for reaching an helping hand each time it was needed,

1728:  in all aspects of academic research activity: support, ideas, and

1729: encouragement. I also want to thank Yasha Eliashberg for inviting

1730: me to Stanford university and for sharing with me his opinion on

1731: this work, and his kind hospitality. The "linear" part of this

1732: work has started at the beginning of my Ph.D studies and has

1733: benefit a lot from many discussions with Assaf Goldberger. I thank

1734: him very much for his help.

1735: 

1736: 

1737: 

1738: 

1739: 

1740: 

1741: 

1742: 

1743: 

1744: \begin{thebibliography}{}

1745: 

1746: 

1747: \bibitem{AM}Abraham,R.,Marsden,J. {\em Foundations of Mechanics}, second edition, 1994. Addison-Wesley Publishing Company.

1748: 

1749: 

1750: \bibitem{A} Arnold,V.I. {\it Mathematical Methods of Classiccal

1751: Mechanics}, second edition, Graduate Texts in Mathematics;60,

1752: 1989.

1753: 

1754: \bibitem{AG}Arnold,V.I.,Givental,A. Symplectic Geometry, in, {\it

1755: Encyclopaedia of Mathematical Sciences, Vol. 4, Dynamical systems

1756: IV,} V. Arnold, S. Novikov eds., Springer, 1990, pp.1-136.

1757: 

1758: 

1759: 

1760: \bibitem{B} Banyaga, A.,{\it Sur la structure du groupe des

1761: diff\'{e}omorphismes qui pr\'{e}servent une forme symplectique,}

1762: comm. Math. Helv. {\bf 53}:2 (1978), 174-227.

1763: 

1764: 

1765: \bibitem{BG} Barge,J.,Ghys,E. {\em Cocycle D'Euler et de Maslov} Math.Ann.{\bf{294}}:2(1992)\\

1766: 235-265.

1767: 

1768: 

1769: 

1770: \bibitem{BW} Bates,S.,Weinstein,A. {\em Lectures on the Geometry

1771: of Quantization.} AMS,Berkely Center for Pure and Applied

1772: Mathematics,1997.

1773: 

1774: 

1775: \bibitem{Ba} Bavard, C. {\it Longueur stable des commutateurs,}

1776: L'Enseign. Math. {\bf 37}:1-2 (1991), 109-150.

1777: %\bibitem[B-En-P] nBiran,P., Entov, M., Polterovich L. {\em Calabi quasimorphisms for the symplectic ball}, Forschungsinstitut f\"{u}r

1778: %Mathematik ETH Z\"{u}rich, 2003.

1779: 

1780: 

1781: \bibitem{BT} Bott,R.,Tu,L.W. {\em Differential Forms in algebraic

1782: Topology}, Springer,1982.

1783: 

1784: 

1785: \bibitem{CSM} Carter,R.,Segal,G.,MacDonald,I. {\em Lectures on Lie

1786: Groups and Lie Algebras.} London Mathematical Society Student

1787: Texts 32, Cambridge University Press.

1788: 

1789: \bibitem{D}  Dupont,J. {\em Bounds for the chercteristic numbers of flat bundles.} Notes Math.,vol 763

1790:   p.109-119 Berlin Heidelberg New York Springer 1979.

1791: 

1792: \bibitem{E-K-P} Eliashberg,Y., Kim,S.K., Polterovich,P. {\it

1793: Geometry of contact transformations and domains: orderability vs.

1794: squeezing,}  Preprint math.SG/0511658, 2005.

1795: 

1796: \bibitem{El-P} Eliashberg,Y.;Polterovich,L. {\em Partially ordered groups and geometry of contact

1797:   transformations.} Geom.Funct.Anal.{\bf{10}}(2000),no.6, 1448-1476.

1798: 

1799: 

1800: \bibitem{G} Givental A.B. {\it Nonlinear generalization of the

1801: Maslov index.} Theory of singularities and its applications, Adv.

1802: Soviet Math., vol,1,American Mathematical society, Rhode

1803: Island,1990,p.71-103.

1804: 

1805: 

1806: \bibitem{H} Hall,B.C. {\em Lie groups, Lie Algebras, and

1807: Representations: An Elementarey Introduction.}, Graduate Texts in

1808: Mathematics;222, Springer, 2003.

1809: 

1810: \bibitem{H-Z} Hofer,H.,Zehnder,E. {\em Symplectic Invariants and

1811: Hamiltonian Dynamics}, Birkh\"{a}user Advanced Texts,

1812: Birkh\"{a}user Verlag, 1994.

1813: 

1814: 

1815: \bibitem{Ki} Kirillov,A. Geometric quantization, in {\it

1816: Encyclopaedia of Mathematical Sciences, Vol. 4, Dynamical systems

1817: IV,} V. Arnold, S. Novikov eds., Springer, 1990, pp.137-172.

1818: 

1819: \bibitem {M-S} McDuff,D., Salamon, D., {\it Introduction to symplectic

1820: topology}, second edition, Oxford University Press, Oxford, 1998.

1821: 

1822: 

1823:   %\bibitem[En-P]{F-2} Entov,M.;Polterovich,L. {\em Calabi quasimorphism and quantum

1824:   %homology.}Int.math.Res.Not.{\bf{2003}} no.30,1635-1676.

1825: 

1826: \bibitem{Na} Nakahara,M. {\em Geometry, Topology and Physics},

1827: Institute Of Physics Publishing Bristol And Philadelphia,1990.

1828: 

1829:   %\bibitem[G]{Book-Spectral} Givental A.B. {\em Non linear generalization of the Maslov index.} Theory of

1830:   %singularities and its applications ,Adv.Soviet Math.,vol,1,American Mathematical society,Rhode

1831:   %Island,1990,p.71-103.

1832: 

1833: 

1834: \bibitem{Ol} Olshanskii,G.I. {\em Invariant orderings in simple

1835: Lie groups. The solution to E.B Vinberg's problem.} Functional

1836: Analysis Appl. 16 (1982), No. 4, 311-313.

1837: 

1838: \bibitem{P} Polterovich L.  {\em The geometry of the group of symplectic diffeomorphisms.} Lectures in Mathematics ETH

1839:   Z\"{u}rich, Birkh\"{a}user Verlag,Basel, 2001.

1840: 

1841: \bibitem{P-1} Polterovich L. {\em Hofer's diameter and Lagrangian intersections,.}

1842:    Int.Math.Res.Notices {\bf{4}}(1998), 217-223.

1843: 

1844: \bibitem{P-S} Polterovich,L.;Siburg,K.F.;{\em On the asymptotic geometry of

1845:    area preserving maps}

1846:      .Math.Res.Lett.{\bf{7.}} 233-243(2000).

1847: 

1848: \bibitem{R-S} Robbin, J.W. and Salamon, D.A. {\em The Maslov index

1849: for paths}, Topology,\textbf{32}, 827-44, 1993.

1850: 

1851: \bibitem{Sp} Spivak,M. {\em A Comprehensive Introduction to

1852: Differential Geometry, Volume 2}, Third edition, 1999, Publish or

1853: Perish,Inc.

1854: 

1855: \bibitem{W} Weinstein,A. {\em Lectures on  Symplectic manifolds}, Regional Conference Series in Mathematics

1856: \textbf{29}, Amer. Math. Soc., Providence, 1977.

1857: 

1858: 

1859: 

1860: \bibitem{W1} Weinstein,A. {\it Cohomology of Symplectomorphism Groups

1861: and Critical Values of Hamiltonians.} Math.Zeit., 201, 75-82,

1862: 1989.

1863: 

1864: \bibitem{Wo} Woodhouse,N.M.J. {\em Geometric Quantization}, second edition, Oxford

1865: university press, 1991.

1866: 

1867: 

1868: \end{thebibliography}

1869: 

1870: 

1871: 

1872: 

1873: 

1874: 

1875: \end{document}

1876: