1: 

2: %\documentclass[reqno]{amsart}

3: \documentclass{birkmult}

4: 

5: %\usepackage{amscd}

6: \usepackage[all]{xy}

7: %\usepackage{showkeys}

8: %\usepackage[notcite,notref]{showkeys}

9: %\usepackage{showkeys}

10: %\usepackage{hyperref}

11: 

12: \theoremstyle{plain}

13:   \newtheorem{theorem}{Theorem}

14:   \newtheorem{corollary}{Corollary}

15:   \newtheorem{conjecture}{Conjecture}

16:   \newtheorem{lemma}{Lemma}

17:   \newtheorem{proposition}{Proposition}

18:   \newtheorem{observation}{Observation}

19: \theoremstyle{definition}

20:   \newtheorem{definition}{Definition}

21: \theoremstyle{remark}

22:   \newtheorem{remark}{Remark}

23:   \newtheorem{example}{Example}

24: 

25: 

26: \newcommand{\sm}{\text{\rm sm}}

27: \newcommand{\la}{\text{\rm la}}

28: \newcommand{\an}{\text{\rm an}}

29: \newcommand{\sing}{\text{\rm sing}}

30: \newcommand\even{\text{\rm even}}

31: \newcommand\odd{\text{\rm odd}}

32: \newcommand\RS{\text{\rm RS}}

33: \newcommand\ac{\text{\rm ac}}

34: \newcommand\Hodge{\text{\rm Hodge}}

35: \newcommand\cc{\mathfrak c}

36: \newcommand\bb{\mathfrak b}

37: \newcommand\hh{\mathfrak h}

38: \newcommand\ff{\mathfrak f}

39: %\newcommand\resp{resp.\ }

40: \newcommand\id{\text{\rm id}}

41: \newcommand{\Z}{\mathbb Z}

42: \newcommand{\N}{\mathbb N}

43: \newcommand{\X}{\mathbb X}

44: \newcommand{\R}{\mathbb R}

45: \newcommand{\C}{\mathbb C}

46: \newcommand{\T}{\mathbb T}

47: \newcommand{\K}{\mathbb K}

48: \newcommand{\Eul}{\mathfrak{Eul}}

49: \newcommand\vf{\mathfrak X}

50: %\newcommand{\ie}{i.e.\ }

51: %\newcommand{\eg}{e.g.\ }

52: %\newcommand{\cf}{cf.\ }

53: \newcommand{\e}{\mathfrak e}

54: %\newcommand{\o}{\mathfrak o}

55: \newcommand\Or{\mathcal O}

56: \newcommand\OM{\mathcal O_M}

57: \newcommand\dEul{\mathfrak{Eul}^*}

58: \newcommand\comb{\text{\rm comb}}

59: 

60: \renewcommand\aa{\mathfrak a}

61: \renewcommand\ss{\mathfrak s}

62: 

63: \DeclareMathOperator{\rank}{rank}

64: \DeclareMathOperator{\llet}{let}

65: \DeclareMathOperator{\GL}{GL}

66: \DeclareMathOperator\Hom{Hom}

67: \DeclareMathOperator\eend{end}

68: \DeclareMathOperator\coker{coker}

69: \DeclareMathOperator\Tor{Tor}

70: \DeclareMathOperator\img{img}

71: \DeclareMathOperator\Emb{Emb}

72: \DeclareMathOperator\Gr{Gr}

73: \DeclareMathOperator\tr{tr}

74: \DeclareMathOperator\inter{int}

75: \DeclareMathOperator\pt{pt}

76: \DeclareMathOperator{\ind}{ind}

77: \DeclareMathOperator{\Det}{Det}

78: \DeclareMathOperator{\sgn}{sgn}

79: \DeclareMathOperator{\Lef}{Lef}

80: \DeclareMathOperator{\sign}{sign}

81: \DeclareMathOperator{\Id}{Id}

82: \DeclareMathOperator{\IND}{IND}

83: \DeclareMathOperator{\Vol}{vol}

84: \DeclareMathOperator{\Maps}{Maps}

85: \DeclareMathOperator{\Spect}{Spect}

86: \DeclareMathOperator{\dvol}{dvol}

87: \DeclareMathOperator{\Int}{Int}

88: \DeclareMathOperator{\grad}{grad}

89: \DeclareMathOperator{\Zero}{Zero}

90: \DeclareMathOperator{\Cr}{Cr}

91: \DeclareMathOperator{\Diff}{Diff}

92: \DeclareMathOperator{\cs}{cs}

93: \DeclareMathOperator{\ec}{e}

94: \DeclareMathOperator\PD{PD}

95: \DeclareMathOperator\Rep{Rep}

96: 

97: \newcommand\itemref[1]{(\ref{#1})}

98: \renewcommand{\theenumi}{\roman{enumi}}

99: \renewcommand{\labelenumi}{(\theenumi)}

100: 

101: 

102: 

103: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

104: 

105: 

106: 

107: \begin{document}

108: 

109: \title[Torsions,   as a functions on the  space of representations]

110:       {Torsion,   as a function on the  space of\\ representations}

111: 

112: \author{Dan Burghelea}

113: 

114: \address{%Dan Burghelea,

115:          Dept. of Mathematics, The Ohio State University,\\

116:          231 West 18th Avenue, Columbus, OH 43210, USA.}

117: 

118: \email{burghele@mps.ohio-state.edu}

119: 

120: \author{Stefan Haller}

121: 

122: \address{%Stefan Haller,

123:          Department of Mathematics, University of Vienna,\\

124:          Nordbergstrasse 15, A-1090, Vienna, Austria.}

125: 

126: \email{stefan.haller@univie.ac.at}

127: 

128: \thanks{Part of this work was done while both authors enjoyed the

129:         hospitality of the Max Planck Institute for Mathematics in Bonn.

130:         A previous version was written while the second author enjoyed the

131:         hospitality of the Ohio State University.

132:         The second author was partially supported by the \emph{Fonds zur F\"orderung der

133:         wissenschaftlichen Forschung} (Austrian Science Fund),

134:         project number {\tt P17108-N04}}

135: 

136: \keywords{Euler structure; coEuler structure; combinatorial torsion;

137:           analytic torsion; theorem of Bismut--Zhang; Chern--Simons theory;

138:           geometric regularization; mapping torus; rational function}

139: 

140: %\subjclass[2000]{57R20, 58J52}

141: \subjclass{57R20, 58J52}

142: 

143: 

144: %\date{\today}

145: 

146: 

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

148: 

149: 

150: \begin{abstract}

151: Riemannian Geometry, Topology and 

152: Dynamics permit to introduce partially defined holomorphic 

153: functions on the variety of representations of the fundamental group of a manifold.

154: The functions we consider are the complex valued Ray--Singer torsion, the Milnor--Turaev 

155: torsion, and the dynamical torsion. They are associated essentially to a 

156: closed smooth manifold equipped with a (co)Euler structure and a Riemannian metric 

157: in the first case, a smooth triangulation in the second case, and a smooth 

158: flow of type described in section~\ref{S:2} in the third case.

159: In this paper we define these functions, describe 

160: some of their properties and calculate them in some case. 

161: We conjecture that they are essentially equal and  

162: have analytic continuation to rational functions on the variety of representations. 

163: We discuss the case of one dimensional representations 

164: and other relevant situations when the conjecture is true.  

165: As particular cases of our torsions, we recognize familiar rational functions in 

166: topology such as the Lefschetz zeta function of a diffeomorphism, the dynamical 

167: zeta function of closed trajectories, and the Alexander polynomial of a knot. 

168: A numerical invariant derived from Ray--Singer torsion and associated to 

169: two homotopic acyclic representations is discussed in the last section.

170: \end{abstract}

171: 

172: \maketitle

173: 

174: \newpage

175: 

176: \setcounter{tocdepth}{1}

177: \tableofcontents

178: 

179: 

180: 

181: 

182: \section{Introduction}\label{S:intro}

183: 

184: 

185: For a finitely presented group $\Gamma$ denote by $\Rep(\Gamma;V)$ 

186: the algebraic set of all complex representations of $\Gamma$ on

187: the complex vector space $V$.

188: For a closed base pointed manifold $(M,x_0)$ with $\Gamma=\pi_1(M,x_0)$ 

189: denote by $\Rep^M(\Gamma;V)$ the algebraic closure of $\Rep^M_0(\Gamma;V)$,

190: the Zariski open set of representations $\rho\in\Rep(\Gamma;V)$ so that 

191: $H^*(M;\rho)=0$.

192: The manifold $M$ is called $V$-acyclic iff $\Rep^M(\Gamma;V)$, or

193: equivalently $\Rep^M_0(\Gamma;V)$, is non-empty. If $M$ is $V$-acyclic

194: then the Euler--Poincar\'e characteristic $\chi(M)$ vanishes. There are plenty of

195: $V$-acyclic manifolds. 

196: 

197: 

198: If $\dim V=1$ then $\Rep(\Gamma;V)=(\mathbb C\setminus 0)^k\times F$, where $k$ denotes

199: the first Betty number of $M$, and $F$ is a finite Abelian group. If in addition $M$ 

200: is $V$-acyclic and $H_1(M;\mathbb Z)$ is torsion free, then 

201: $\Rep^M(\Gamma;V)=(\mathbb C\setminus 0)^k$. 

202: There are plenty of $V$-acyclic ($\dim V=1$) manifolds $M$ with $H_1(M;\mathbb Z)$ 

203: torsion free.

204: 

205: 

206: In this paper, to a $V$-acyclic manifold and an Euler or coEuler structure

207: we associate three partially defined holomorphic functions on

208: $\Rep^M(\Gamma;V)$, the complex valued Ray--Singer torsion, the

209: Milnor--Turaev torsion, and the dynamical torsion, and describe some of

210: their properties.  They are defined with the help of a Riemannian metric,

211: resp.\ smooth triangulation resp.\ a vector field with the properties listed

212: in section~\ref{SS:2.3}, but are independent of these data.

213: 

214:  

215: We conjecture that they are essentially equal and have analytic continuation to rational

216: functions on $\Rep^M(\Gamma;V)$ and discuss the cases when we know that this is true.

217: If $\dim V=1$ they are genuine rational functions of $k$ variables.

218: 

219: 

220: We calculate them in some cases and recognize familiar rational functions in topology

221: (Lefschetz zeta function of a diffeomorphism, dynamical zeta function of some flows, 

222: Alexander polynomial of a knot) as particular cases of our torsions, cf.\ section~\ref{S:7}.

223: 

224: 

225: The results answer the question

226: \begin{itemize}

227: \item[(Q)]

228: \emph{Is the Ray--Singer torsion the absolute value of a  holomorphic 

229: function on the space of representations?}\footnote{A similar question was 

230: considered in \cite{Q85} and a positive answer provided.}

231: \end{itemize} 

232: (for a related result consult \cite{BK05})

233: and  establish the analytic continuation of the dynamical torsion.

234: Both issues are  subtle when formulated inside the field of spectral geometry  

235: or of dynamical systems and can hardly be decided using internal technologies 

236: in these fields. There are  interesting dynamical implications on the growth 

237: of the number of instantons  and of closed trajectories, some of them improving 

238: on a conjecture formulated by S.P.~Novikov about the gradients of closed Morse one 

239: form, cf.\ section~\ref{S:8}.

240: 

241:   

242: \vskip .1in

243: 

244: 

245: This paper surveys results from \cite{BH04}, \cite{BH03'}, \cite{BH05} and reports 

246: on additional work in progress on these lines. Its contents is the following.

247: 

248: 

249: In section~\ref{S:2}, for the reader's convenience, we recall some less 

250: familiar characteristic forms used in this paper and describe the class of vector 

251: fields we use to define the dynamical torsion. These vector fields  have finitely 

252: many rest points  but  infinitely many instantons and closed trajectories.  However, 

253: despite their infiniteness, they can be counted by appropriate counting functions 

254: which can be  related to  the topology and  the geometry  of the underlying manifold cf.\ \cite{BH03}. 

255: The dynamical torsion is derived from them.

256: 

257: 

258: All torsion functions referred to above involve some additional topological data; 

259: the Milnor--Turaev and dynamical torsion involve an Euler structure  while the 

260: complex Ray--Singer torsion a coEuler structure, a sort of Poincar\'e dual of the first.  

261: In section~\ref{S:3} we define Euler and coEuler structures and discuss some of their 

262: properties. Although they can be defined for arbitrary base pointed manifolds $(M,x_0)$ 

263: we present the theory only in the case $\chi(M)=0$ when the base point is irrelevant.

264: 

265: 

266: While the complex Ray--Singer torsion and dynamical torsion are new concepts the 

267: Milnor--Turaev torsion is not, however our presentation is somehow different from 

268: the traditional one. In section~\ref{S:4} we discuss the algebraic variety 

269: of cochain complexes of finite dimensional vector spaces and introduce  

270: the Milnor torsion as  a rational function on this variety.  

271: The Milnor--Turaev torsion is obtained as a pull back by a characteristic map  

272: of this rational function. 

273: 

274: 

275: Section~\ref{S:5} is about analytic torsions. In section~\ref{SS:5.1}, we recall  

276: the familiar Ray--Singer torsion slightly modified with the help of a coEuler 

277: structure. This is a positive real valued function defined on $\Rep^M_0(\Gamma;V)$, 

278: the variety of the acyclic representations. We show that this function is independent 

279: of the Riemannian metric, and that it is the absolute value of a rational function,

280: provided the coEuler structure is integral.

281: %%%%%%%%%%%%%%%%

282: %locally is the absolute value of holomorphic functions,  

283: %more precisely of a \emph{holomorphic closed  one m-form}.  

284: %but globally is not in general. However its square always is. 

285: %The same happens for the unmodified Ray--Singer torsion.

286: %%%%%%%%%%%%%%%%%%%%%

287: In section~\ref{SS:5.3} we introduce the complex valued Ray--Singer torsion,  

288: and show the relation to the first.

289: % its absolute value is conjecturally the square of the modified Ray--Singer torsion.

290: The complex Ray--Singer torsion, denoted $\mathcal S\mathcal T$, is a meromorphic 

291: function on a finite cover of the  space of representations and is defined analytically

292: using regularized determinants of elliptic operators but not self adjoint.

293: 

294: 

295: The  Milnor--Turaev torsion, defined in section~\ref{SS:6.1}, is associated with a smooth 

296: manifold, a given Euler structure and a homology orientation and is constructed  

297: using a smooth triangulation. Its square is  conjecturally  equal to  the 

298: complex Ray--Singer torsion as defined in section~\ref{SS:5.3},  

299: when the coEuler structure for Ray--Singer corresponds, by Poincar\'e

300: duality map, to the Euler structure for Milnor--Turaev. The conjecture  

301: is true in many relevant cases, in particular for $\dim V=1$.

302: 

303: 

304: Up to sign the dynamical torsion, introduced in section~\ref{SS:6.2}, is associated to 

305: a smooth manifold and a given Euler structure and is constructed using a smooth vector 

306: field in the class described in section~\ref{SS:2.3}. The sign can be  fixed with the help 

307: of an equivalence class of orderings of the rest points of $X$, cf.\ section~\ref{SS:6.2}. 

308: A priori the dynamical torsion is only a partially defined holomorphic function on 

309: $\Rep^M(\Gamma;V)$ and is defined using the instantons and the closed trajectories of $X$.

310: For a representation $\rho$ the dynamical torsion is expressed as  a series which might 

311: not be convergent for each $\rho$ but is certainly convergent for $\rho$ in a subset 

312: $U$ of $\Rep^M(\Gamma;V)$ with non-empty interior. At present this convergence was 

313: established only in the case of rank one representations. The existence of $U$ is 

314: guaranteed by the exponential growth property (EG) (cf.\ section~\ref{SS:2.3} 

315: for the definition) required from the vector field. 

316: 

317: 

318: The main results, Theorems~\ref{T:1}, \ref{T:2} and \ref{T:3}, establish the 

319: relationship between these torsion functions, at least in the case $\dim V=1$, 

320: and a few other relevant cases. The  same relationship is expected to hold for 

321: $V$ of arbitrary dimension.

322: 

323: 

324: One can calculate the Milnor--Turaev torsion when $M$ has a structure of mapping torus  

325: of a diffeomorphism $\phi$ as the ``twisted  Lefschetz zeta function''  of the 

326: diffeomorphism  $\phi$, cf.\ section~\ref{SS:7.1}.

327: The Alexander polynomial as well as the twisted Alexander polynomials of a knot 

328: can also be recovered from these torsions cf.\ section~\ref{SS:7.3}.

329: If the vector field has no rest points but admits a closed Lyapunov cohomology class,  

330: cf.\ section~\ref{SS:7.2}, the dynamical torsion can be expressed in terms of closed 

331: trajectories only, and the dynamical zeta function of the vector field (including all 

332: its twisted versions) can be recovered from the dynamical torsion described here.

333: 

334: 

335: In section~\ref{SS:8.1} we express the phase difference of the Milnor--Turaev torsion 

336: of two representations in the same connected component of

337: $\Rep^M_0(\Gamma;V)$ in terms of the Ray--Singer torsion. This invariant is analogous to 

338: the Atiyah--Patodi--Singer spectral flow but has not been investigated so far.

339: Section~\ref{S:8} discusses some progress towards a conjecture of Novikov which came 

340: out from the work on dynamical torsion.

341: 

342: 

343: 

344: 

345: \section{Characteristic forms and vector fields}\label{S:2}

346: 

347: 

348: 

349: 

350: \subsection{Euler, Chern--Simons, and Mathai--Quillen form}\label{SS:2.1}

351: 

352: 

353: 

354: Let $M$ be smooth closed manifold of dimension $n$. 

355: Let $\pi:TM\to M$ denote the

356: tangent bundle, and $\Or_M$ the orientation bundle, which is a flat real line

357: bundle over $M$. 

358: For a Riemannian metric $g$ denote by 

359: $$

360: \ec(g)\in\Omega^n(M;\Or_M)

361: $$

362: its Euler form, and for two Riemannian metrics $g_1$ and $g_2$ by 

363: $$

364: \cs(g_1,g_2)\in

365: \Omega^{n-1}(M;\Or_M)/d(\Omega^{n-2}(M;\Or_M))

366: $$ 

367: their Chern--Simons class. 

368: %The definition of both quantities is

369: %implicit in the formulas \eqref{dpsie} and \eqref{pg1pg2cs} below.

370: The following properties follow from

371: \eqref{dpsie} and \eqref{pg1pg2cs} below.

372: \begin{eqnarray}

373: d \cs(g_1,g_2) &=& \ec(g_2)-\ec(g_1)

374: \label{E:csg:i}

375: \\

376: \cs(g_2,g_1) &=& -\cs(g_1,g_2)

377: \label{E:csg:ii}

378: \\

379: \cs(g_1,g_3) &=& \cs(g_1,g_2) + \cs(g_2,g_3)

380: \label{E:csg:iii}

381: \end{eqnarray}

382: If the dimension of $M$ is odd both $\ec(g)$ and $\cs(g_1, g_2)$ vanish.

383: 

384: 

385: Denote by $\xi$ the Euler vector field on $TM$ which

386: assigns to a point $x\in TM$ the vertical vector $-x\in T_xTM$.

387: A Riemannian metric $g$ determines the Levi--Civita connection in the bundle

388: $\pi:TM\to M$. There is a canonic $n$-form $\Vol(g)\in\Omega^n(TM;\pi^*\Or_M)$,

389: which assigns to an $n$-tuple of vertical vectors \emph{their volume times their

390: orientation} and vanishes when contracted with horizontal vectors and a global 

391: angular form, see for instance \cite{BT82}, is the differential form

392: $$

393: A(g)

394: :=\frac{\Gamma(n/2)}{(2\pi)^{n/2}|\xi|^n}

395: i_\xi\Vol(g)\in\Omega^{n-1}(TM\setminus M;\pi^*\Or_M).

396: $$

397: In \cite{MQ86} Mathai and Quillen have introduced a differential form 

398: $$

399: \Psi(g)\in \Omega^{n-1}(TM\setminus M;\pi^*\Or_M).

400: $$ 

401: When pulled back to the fibers of $TM\setminus M\to M$ the form $\Psi(g)$

402: coincides with $A(g)$. If $U\subseteq M$ is an open subset on

403: which the curvature of $g$ vanishes, then $\Psi(g)$ coincides with $A(g)$

404: on $TU\setminus U$.

405: In general we have the equalities

406: \begin{eqnarray}

407: d\Psi(g) &=& \pi^*\ec(g).

408: \label{dpsie}

409: \\

410: \Psi(g_2)-\Psi(g_1) &=& \pi^*\cs(g_1,g_2) \mod d\Omega^{n-2}(TM \setminus M;\pi^* \Or_M).

411: \label{pg1pg2cs}

412: \end{eqnarray}

413: 

414: 

415: \subsection{Euler and Chern--Simons chains}\label{SS:2.22}

416: 

417: 

418: 

419: For a vector field $X$ with non-degenerate rest points 

420: we have the singular $0$-chain $\ec(X)\in C_0(M;\mathbb Z)$ defined by 

421: $\ec(X):=\sum_{x\in\mathcal X}\IND(x)x$, with $\IND(x)$ the Hopf index.

422: 

423: 

424: For two  vector fields $X_1$ and $X_2$ with non-degenerate rest points 

425: we have the singular $1$-chain rel.\ boundaries 

426: $\cs(X_1,X_2)\in C_1(M;\mathbb Z)/\partial C_2(M;\mathbb Z)$ defined from the 

427: zero set of a homotopy from $X_1$ to $X_2$ 

428: cf.\ \cite{BH04}. They are related by the formulas, see \cite{BH04},

429: \begin{eqnarray}

430: \partial \cs(X_1,X_2) &=& \ec(X_2)-\ec(X_1)

431: \label{E:11}

432: \\

433: \cs(X_2,X_1) &=& -\cs(X_1,X_2)

434: \label{E:12i}

435: \\

436: \cs(X_1,X_3) &=& \cs(X_1,X_2) + \cs(X_2,X_3).

437: \label{E:13}

438: \end{eqnarray}

439: 

440: 

441: 

442: \subsection{Kamber--Tondeur one form}\label{SS:2.2}

443: 

444: 

445: 

446: Let $E$ be a real or complex vector bundle over $M$.

447: For a connection $\nabla$ and a Hermitian structure $\mu$ on $E$

448: define a real valued one form

449: $\omega(\nabla,\mu)\in\Omega^1(M;\R)$ by

450: \begin{equation}\label{E:6}

451: \omega(\nabla,\mu)(Y):=-\frac12\tr\bigl(\mu^{-1}\cdot (\nabla_Y\mu)\bigr),

452: \quad Y\in TM.

453: \end{equation}

454: Here we consider $\mu$ as an element in $\Omega^0(M;\hom(E,\bar E^*))$, where

455: $\bar E^*$ denotes the dual of the complex conjugate bundle. With respect to the 

456: induced connection on $\hom(E,\bar E^*)$ we have $\nabla_Y\mu\in\Omega^1(M;\hom(E,\bar E^*))$

457: and therefore $\mu^{-1}\cdot\nabla_Y\mu\in\Omega^1(M;\eend(E,E))$. Actually the latter

458: one form has values in the endomorphisms of $E$ which are symmetric with respect to $\mu$,

459: and thus the (complex) trace, see \eqref{E:6}, will indeed be real.

460: Since any two Hermitian structures $\mu_1$ and $\mu_2$ are homotopic,

461: the difference $\omega(\nabla,\mu_2)-\omega(\nabla,\mu_1)$ will be exact.

462: If $\nabla$ is flat then $\omega(\nabla,\mu)$ is 

463: closed and its cohomology class independent of $\mu$.

464: %Moreover, 

465: %$\omega(\nabla^{\det E},\mu^{\det E})=\omega(\nabla,\mu)$

466: %where $\nabla^{\det E}$ and $\mu^{\det E}$ denote the induced connection

467: %and Hermitian structure on the determinant line 

468: %$\det E:=\Lambda^{\rank E}E$.

469: 

470: 

471: %A similar form (closed for a flat connections) can be associated to a symmetric

472: %non-degenerate bilinear form $b$

473: %on $E$ instead of a Hermitian structure by a similar formula

474: Replacing the Hermitian structure by a non-degenerate symmetric bilinear form $b$,

475: we define a complex valued one form $\omega(\nabla,b)\in\Omega^1(M;\C)$ by

476: a similar formula

477: \begin{equation}\label {E:7}

478: \omega(\nabla,b)(Y):=-\frac12\tr\bigl(b^{-1}\cdot(\nabla_Yb)\bigr),\quad Y\in TM.

479: \end{equation}

480: Here we regard $b$ as an element in $\Omega^0(M;\hom(E,E^*))$.

481: If two non-degenerate symmetric bilinear forms $b_1$ and $b_2$ are homotopic, then 

482: $\omega(\nabla,b_2)-\omega(\nabla,b_1)$ is exact. If $\nabla$ is flat,

483: then $\omega(\nabla,b)$ is closed.

484: Note that $\omega(\nabla,b)\in\Omega^1(M;\mathbb C)$ depends holomorphically

485: on $\nabla$.

486: 

487: 

488: 

489: 

490: \subsection{Vector fields, instantons and closed trajectories}\label{SS:2.3}

491: 

492: 

493: 

494: Consider a vector field $X$ which satisfies the following properties:

495: 

496: 

497: \begin{itemize}

498: \item[(H)]

499: All rest points are of hyperbolic type.

500: \item[(EG)]

501: The vector field has exponential growth at all rest points.

502: \item[(L)]

503: The vector field is of Lyapunov type.

504: \item[(MS)]

505: The vector field satisfies Morse--Smale condition.

506: \item[(NCT)]

507: The vector field has all closed trajectories non-degenerate.

508: \end{itemize}

509: Precisely this means that:

510: \begin{itemize}

511: \item[(H)]

512: In the neighborhood of each rest point the differential of $X$ has all 

513: eigenvalues with non-trivial real part; the number of eigenvalues with 

514: negative real part is called the index and denoted by $\ind(x)$; 

515: as a consequence the stable

516: and unstable stable sets are images of one-to-one immersions 

517: $i_x^\pm:W^\pm_x\to M$ with $W^\pm_x$ diffeomorphic to $\mathbb R^{n-\ind(x)}$ 

518: resp.\ $\mathbb R^{\ind(x)}$.

519: \item[(EG)]

520: With respect to one and then any Riemannian metric $g$ on

521: $M$, the volume of the disk of radius $r$ in $W^-_x$ (w.r.\ to the induced Riemannian

522: metric) is $\leq e^{Cr}$, for some constant $C>0$.

523: \item[(L)]

524: There exists a real valued closed one form $\omega$ so that $\omega(X)_x<0$ for 

525: $x$ not a rest point.\footnote{This $\omega$ has nothing in common with 

526: $\omega(\nabla,b)$ notation used in the previous section.}

527: \item[(MS)]

528: For any two rest points $x$ and $y$ the mappings $i^-_x$ and $i^+_y$ are transversal 

529: and therefore the space of non-parameterized trajectories form $x$ to $y$, 

530: $\mathcal T(x,y)$, is a smooth manifold of dimension $\ind(x)-\ind(y)-1$. 

531: Instantons are exactly the elements of $\mathcal T(x,y)$ when this is a smooth 

532: manifold of dimension zero, i.e.\ $\ind(x)-\ind(y)-1=0$.

533: \item[(NCT)]

534: Any closed trajectory is non-degenerate, i.e.\ the differential of the return map

535: in normal direction at one and then any point of a closed trajectory does not have

536: non-zero fixed points.

537: \end{itemize}

538: 

539: Recall that a trajectory $\theta$ is an equivalence class of parameterized trajectories and 

540: two parameterized trajectories $\theta_1$ and $\theta_2$ are equivalent iff 

541:   $\theta_1(t+c)= \theta_2(t)$

542: for some real number $c.$  

543: Recall that a closed trajectory $\hat\theta$ is a pair consisting of a trajectory $\theta$ and a positive real number $T$ so that $\theta(t+T)= \theta(t).$ 

544: 

545: %an equivalence class of parameterized closed trajectory 

546: %consisting of ais a pair $(\theta, T)$ with $\theta$ a parameterized 

547: %trajectory  and $T$ a positive real number so that $\theta(t)= \theta(t+T).$  A closed trajectory $\hat %\theta$ is an equivalence class and parameterized closed trajectories;  $(\theta_1, T_1)\equiv %(\theta_2, T_2)$ iff $T_1= T_2$ and $\theta_1$ and $\theta_2$ are equivalent parameterized %trajectories. 

548: 

549: 

550:  

551: Property (L), (H), (MS)  imply that for any real number $R$ the set of instantons 

552: $\theta$ from $x$ to $y$ 

553: with $-\omega([\theta])\leq R$ 

554: is finite  and properties (L), (H), (MS), (NCT) imply that for any real number $R$ the set of 

555: the closed trajectory $\hat\theta$ with $-\omega([\hat \theta])\leq R$ is finite. 

556: Here 

557: $[\theta]$ resp $[\hat\theta]$ denote the homotopy class of instantons resp. closed trajectories

558: \footnote{ For a closed trajectory the map whose homotopy class is considered is $\hat \theta: \mathbb R/ T\mathbb Z\to M.$}.

559:   

560: 

561: Denote by $\mathcal P _{x,y}$ the set of homotopy classes of paths from 

562: $x$ to $y$ and by $\mathcal X_q$ the set of rest points of index $q$. 

563: Suppose a collection $\mathcal O=\{\mathcal O_x\mid x\in\mathcal X\}$

564: of orientations of the unstable manifolds is given and (MS) is satisfied.

565: Then any instanton $\theta$ has a sign $\epsilon(\theta)=\pm1$ and therefore,

566: if (L) is also satisfied, for any two rest points $x\in{\mathcal X}_{q+1}$

567: and $y\in \mathcal X_{q}$ we have the counting function of instantons

568: $\mathbb I^{X,\mathcal O}_{x,y}: \mathcal P_{x,y}\to\mathbb Z$

569: defined by 

570: \begin{equation}\label {E:8}

571: \mathbb I^{X,\mathcal O}_{x,y}(\alpha):=\sum_{\theta\in\alpha}\epsilon(\theta).

572: \end{equation}

573: 

574: 

575: Under the hypothesis (NCT) any closed trajectory $\hat\theta$ has a sign 

576: $\epsilon(\hat\theta)=\pm1$ and a period $p(\hat\theta)\in\{1,2,\dotsc\}$, cf.\ \cite{H02}.

577: If (H), (L), (MS), (NCT) are satisfied, as the set of closed trajectories in a fixed 

578: homotopy class $\gamma\in[S^1,M]$ is compact, we have the counting function of closed 

579: trajectories $\mathbb Z_X:[S^1,M]\to\mathbb Q$ defined by  

580: \begin{equation}\label{E:9}

581: \mathbb Z_X(\gamma):=\sum_{\hat\theta\in\gamma}\epsilon(\hat\theta)/p(\hat\theta).

582: \end{equation}

583: Here are a few properties about vector fields which satisfy (H) and (L). 

584: 

585: 

586: 

587: \begin{proposition}\label{P:1}

588: 1. Given a vector field $X$ which satisfies (H) and (L) arbitrary close in the

589: $C^r$-topology for any $r\geq0$ there exists a vector field $Y$ which agrees  

590: with $X$ on a neighborhood of the rest points and satisfies (H), (L), (MS) and (NCT).

591: 

592: 

593: 2. Given a vector field $X$ which satisfies (H) and (L) arbitrary close in the

594: $C^0$-topology there exists a vector field $Y$ which agrees with $X$ on a 

595: neighborhood of the rest points and satisfies (H), (EG), (L), (MS) and (NCT).

596:  

597: 

598: 3. If $X$ satisfies (H), (L) and (MS) and a collection $\mathcal O$ of orientations 

599: is given then  for any  $x\in\mathcal X_q$, $z\in\mathcal X_{q-2}$ and 

600: $ \gamma\in\mathcal P_{x,z}$ one has\footnote{It is understood that only finitely many 

601: terms from the left side of the equality are not zero. Here $*$ denotes juxtaposition.}

602: \begin{equation}\label{E:10}

603: %\sum_{ \begin{aligned} y\in{\mathcal X}_{q-1},   \alpha\in &\mathcal P_{x,y},  

604: %\beta\in \mathcal P_{y,z} \\  \alpha* \beta&= \gamma\end{aligned}}

605: \sum_{\begin{smallmatrix} y\in{\mathcal X}_{q-1},\alpha\in\mathcal P_{x,y},  

606: \beta\in \mathcal P_{y,z} \\  \alpha* \beta=\gamma\end{smallmatrix}}

607: \mathbb I^{X,\mathcal O}_{x,y}(\alpha)\cdot

608: \mathbb I^{X,\mathcal O}_{y,z}(\beta)=0.

609: \end{equation}

610: \end{proposition}

611: 

612: 

613: This proposition is a recollection of some of the main results in \cite{BH03}, see

614: Proposition~3, Theorem~1 and Theorem~5 in there.

615:  

616: 

617: 

618: \section{Euler and coEuler structures}\label{S:3}

619: 

620: 

621: 

622: Although not always necessary in this section as in fact always in this 

623: paper $M$ is supposed to be closed connected smooth manifold. 

624: 

625: 

626: 

627: \subsection{Euler structures}\label{SS:3.1}

628: 

629: 

630: 

631: Euler structures have been introduced by Turaev \cite{Tu90} for

632: manifolds $M$ with $\chi(M)=0$. If one

633: removes the hypothesis $\chi(M)=0$ the concept of Euler structure 

634: can still be considered for any connected base pointed manifold

635: $(M,x_0)$ cf.\ \cite{B99} and \cite{BH04}. Here we will consider only

636: the case $\chi(M)=0$. The set of Euler structures,

637: denoted by $\Eul(M;\Z)$, is equipped with a free and transitive action

638: $$

639: m:H_1(M;\Z)\times\Eul(M;\Z)\to\Eul(M;\Z) 

640: $$ 

641: which makes $\Eul(M;\Z)$ an affine version of $H_1(M;\Z)$.

642: If $\e_1,\e_2$ are two Euler structure we write $\e_2-\e_1$ for the 

643: unique element in $H_1(M;\Z)$ with $m(\e_2-\e_1,\e_1)=\e_2$.

644: 

645: 

646: To define the set $\Eul(M;\Z)$ we consider pairs $(X,c)$ with $X$ a vector field with

647: non-degenerate zeros and $c\in C_1(M;\mathbb Z)$ so that $\partial

648: c=\ec(X)$. We make $(X_1,c_1)$ and $(X_2,c_2)$ equivalent iff $c_2-c_1=\cs(X_1,X_2)$

649: and write $[X,c]$ for the equivalence class represented by $(X,c)$.

650: The action $m$ is defined by $m([c'],[X, c]):=[X,c'+c]$.

651: 

652: 

653: \begin{observation}\label{O:1}

654: Suppose $X$ is a vector field with non-degenerate zeros, and assume its zero set

655: $\mathcal X$ is non-empty. Moreover, let $\e\in\Eul(M;\mathbb Z)$ be an Euler structure

656: and $x_0\in M$. Then there exists a collection of

657: paths $\{\sigma_x\mid x\in\mathcal X\}$ with $\sigma_x(0)=x_0$,

658: $\sigma_x(1)=x$ and such that $\e=[X,c]$ where $c=\sum_{x\in\mathcal X}\IND(x)\sigma_x$.

659: \end{observation}

660: 

661: 

662: A remarkable source of Euler structures is the set of homotopy classes

663: of nowhere vanishing vector fields. Any nowhere vanishing vector field $X$

664: provides an Euler structure $[X,0]$ which only depends on the homotopy class

665: of $X$. Still assuming $\chi(M)=0$, every 

666: Euler structure can be obtained in this way provided $\dim(M)>2$.

667: Be aware, however, that different homotopy classes may give rise to the

668: same Euler structure. 

669: 

670: 

671: To construct such a homotopy class one can proceed as follows. Represent

672: the Euler structure $\e$ by a vector field $X$ and a collection of paths

673: $\{\sigma_x\mid x\in\mathcal X\}$ as in Observation~\ref{O:1}.

674: Since $\dim(M)>2$ we may assume that the interiors

675: of the paths are mutually disjoint. Then the set $\bigcup_{x\in\mathcal

676: X}\sigma_x$ is contractible. A smooth regular neighborhood

677: of it is the image by a smooth embedding $\varphi:(D^n,0)\to(M,x_0)$. 

678: Since $\chi(M)=0$, the restriction of the vector field $X$ to

679: $M\setminus\inter(D^n)$ can be extended to a non-vanishing vector field 

680: $\tilde X$ on $M$. It is readily checked that $[\tilde X,0]=\e$.

681: For details see \cite{BH04}.

682: 

683: 

684: If $M$  dimension larger than 

685: 2 an alternative description of $\Eul(M;\Z)$

686: %%%%%%%%%%%%%%

687: % and of the action $m$  

688: %%%%%%%%%%%

689: with respect to a base point $x_0$ is 

690: $\Eul(M;\Z)=\pi_0(\vf(M,x_0))$, where

691: $\vf(M,x_0)$ denotes the space of vector fields of class $C^r$, $r\geq 0$, which vanish 

692: at $x_0$ and are non-zero elsewhere. We equip this space with the $C^r$-topology and 

693: note that the result $\pi_0(\vf(M,x_0))$ is the same for all $r$, and

694: since $\chi(M)=0$, canonically identified for different base points.

695: 

696: 

697: Let $\tau$ be a smooth triangulation of $M$ and consider the function 

698: $f_\tau:M\to\R$ linear on any simplex of the first barycentric subdivision 

699: and taking the value $\dim(s)$ on the barycenter $x_s$ of

700: the simplex $s\in\tau$. A smooth vector field $X$ on $M$ 

701: with the barycenters as the only rest points all of them hyperbolic 

702: %as hyperbolic rest points

703: and $f_\tau$ strictly decreasing on

704: non-constant trajectories is called an Euler vector field of $\tau$.

705: By an argument of convexity two Euler vector fields are homotopic by a homotopy of

706: Euler vector fields.\footnote{Any Euler vector field $X$ satisfies (H), (EG), (L) 

707: and has no closed trajectory, hence also satisfies (NCT).

708: The counting functions of instantons are exactly the same as the incidence numbers  

709: of the triangulation hence take the values $1$, $-1$ or $0$.}

710: Therefore, a triangulation $\tau$, a base point $x_0$ and a collection of

711: paths $\{\sigma_s\mid s\in\tau\}$ with $\sigma_s(0)=x_0$ and

712: $\sigma_s(1)=x_s$ define an Euler structure $[X_\tau,c]$, where

713: $c:=\sum_{s\in\tau}(-1)^{n+\dim(s)}\sigma_s$, $X_\tau$ is any Euler vector

714: field for $\tau$, and this Euler structure does not depend on the

715: choice of $X_\tau$. Clearly, for fixed $\tau$ and $x_0$, every Euler

716: structure can be realized in this way by an appropriate choice of

717: $\{\sigma_s\mid s\in\tau\}$, cf.\ Observation~\ref{O:1}.

718: 

719: 

720: 

721: 

722: \subsection{Co-Euler structures}\label{SS:3.2}

723: 

724: 

725: 

726: Again, suppose $\chi(M)=0$.\footnote{The hypothesis is not necessary and the theory 

727: of coEuler structure can be pursued for an arbitrary base pointed smooth manifold

728: $(M,x_0)$, cf.\ \cite{BH04}.} Consider pairs

729: $(g,\alpha)$ where $g$ is a Riemannian metric on $M$ and

730: $\alpha\in\Omega^{n-1}(M;\mathcal O_M)$ with $d\alpha=\ec(g)$

731: where $\ec(g)\in\Omega^n(M;\mathcal O_M)$ denotes the Euler form of 

732: $g$, see section~\ref{SS:2.1}. We call two

733: pairs $(g_1,\alpha_1)$ and $(g_2,\alpha_2)$ equivalent if

734: $$

735: \cs(g_1,g_2)=\alpha_2-\alpha_1

736: \in\Omega^{n-1}(M;\mathcal O_M)/

737: d\Omega^{n-2}(M;\mathcal O_M).

738: $$

739: We will write $\Eul^*(M;\R)$ for the set of equivalence classes and

740: $[g,\alpha]$ for the equivalence class represented by the pair $(g,\alpha)$.

741: Elements of $\Eul^*(M;\R)$ are called \emph{coEuler structures.}

742: 

743: 

744: There is a natural action

745: $$

746: m^*:H^{n-1}(M;\Or_M)\times\Eul^*(M;\R)\to\Eul^*(M;\R)

747: $$

748: given by

749: $$

750: m^*([\beta],[g,\alpha]):=[g,\alpha-\beta]

751: $$

752: for $[\beta]\in H^{n-1}(M;\mathcal O_M)$. 

753: This action is obviously free and transitive. In this sense

754: $\Eul^*(M;\R)$ is an affine version of $H^{n-1}(M;\mathcal O_M)$.

755: If $\e^*_1$ and $\e^*_2$ are two coEuler structures we write $\e^*_2-\e^*_1$

756: for the unique element in $H^{n-1}(M;\Or_M)$ with 

757: $m^*(\e^*_2-\e^*_1,\e^*_1)=\e^*_2$.

758:   

759: 

760: \begin{observation}\label {O:2}

761: Given a Riemannian metric $g$ on $M$ any coEuler structure can be

762: represented as a pair $(g,\alpha)$ 

763: for some $\alpha\in\Omega^{n-1}(M;\Or_M)$ with $d\alpha=\ec(g)$.

764: \end{observation}

765: 

766: 

767: There is a natural map $\PD:\Eul(M;\Z)\to\Eul^*(M;\R)$

768: which combined with the Poincar\'e duality map

769: $D:H_1(M;\Z)\to H_1(M;\R)\to H^{n-1}(M;\Or_M)$, the composition of the coefficient

770: homomorphism for $\Z\to\R$ with the Poincar\'e duality isomorphism,\footnote{We 

771: will use the same notation D for the Poincar\'e duality isomorphism

772: $D:H_1(M;\R)\to H^{n-1}(M; \mathcal O_M)$.} makes the diagram below commutative:

773: $$

774: \xymatrix{

775: H_1(M;\Z)\times\Eul(M;\Z)

776: \ar[d]_{D\times\PD}\ar[r]^-{m}   

777: & 

778: \Eul(M;\Z)\ar[d]^{\PD}

779: \\

780: H^{n-1}(M;\Or_M)\times\Eul^*(M;\R) 

781: \ar[r]^-{m^*} 

782: &  

783: \Eul^*(M;\R)

784: }

785: $$

786: 

787: 

788: There are many ways to define the map $\PD$, cf.\ \cite{BH04}. For example, 

789: assuming $\chi(M)=0$ and $\dim M>2$ one can proceed as follows.

790: Represent the Euler structure by a nowhere vanishing vector field

791: $\e=[X,0]$. Choose a Riemannian metric $g$, regard $X$ as 

792: mapping $X:M\to TM\setminus M$, set $\alpha:=X^*\Psi(g)$, put 

793: $\PD(\e):=[g,\alpha]$ and check that this does indeed only depend on $\e$.

794: 

795: 

796: A coEuler structure $\e^*\in\Eul^*(M;\R)$ is called \emph{integral}

797: if it belongs to the image of $\PD$. Integral coEuler structures 

798: constitute a lattice in the affine space $\Eul^*(M;\R)$.

799: 

800: 

801: \begin{observation} If $\dim M$ is odd, then there is a canonical coEuler structure

802: $\e^*_0\in\Eul^*(M;\R)$; it is represented by the pair $[g,0]$, with any $g$ 

803: Riemannian metric. In general this coEuler structure is not integral. 

804: \end{observation}

805: 

806: 

807: 

808: \section{Complex representations and cochain complexes}\label{S:4}

809: 

810: 

811: 

812: 

813: 

814: \subsection{Complex representations}\label{SS:4.1}

815: 

816: 

817: 

818: Let $\Gamma$ be a finitely presented group with generators $g_1,\dotsc,g_r$ 

819: and relations 

820: $$

821: R_i(g_1,g_2,\dotsc,g_r)=e,\quad i=1,\dotsc,p,

822: $$ 

823: and $V$ be a

824: complex vector space of dimension $N$. Let $\Rep(\Gamma;V)$ be the set of 

825: linear representations of $\Gamma$ on $V$, i.e.\ group homomorphisms 

826: $\rho:\Gamma\to\GL_\C(V)$. By identifying $V$ to $\C^N$ this set is, in a

827: natural way, an algebraic set inside the space $\C^{rN^2+1}$ given

828: by $pN^2+1$ equations. Precisely if $A_1,\dotsc,A_r,z$ represent the 

829: coordinates in $\C^{rN^2+1}$ with $A:=(a^{ij})$, $a^{ij}\in\C$, so 

830: $A\in\C^{N^2}$ and $z\in\C$, then the equations defining $\Rep(\Gamma;V)$

831: are 

832: \begin{eqnarray*}

833: z\cdot\det(A_1)\cdot\det(A_2)\cdots\det(A_r)&=&1

834: \\

835: R_i(A_1,\dotsc,A_r)&=&\id,\qquad i=1,\dotsc,p

836: \end{eqnarray*}

837: with each of the equalities $R_i$ representing $N^2$ polynomial equations.

838: 

839: 

840: Suppose $\Gamma=\pi_1(M,x_0)$, $M$ a closed manifold.

841: Denote by $\Rep^M_0(\Gamma;V)$ the set of representations $\rho$ with 

842: $H^*(M;\rho)=0$ and notice that they form a Zariski

843: open set in $\Rep(\Gamma;V)$. Denote the closure of this set by  

844: $\Rep^M(\Gamma;V)$. This is an algebraic set 

845: which depends only on the homotopy type of $M$,

846: and is a union of irreducible components of $\Rep(\Gamma;V)$.

847: 

848: 

849: Recall that every representation $\rho\in\Rep(\Gamma;V)$ induces a

850: canonical vector bundle $F_\rho$ equipped with a canonical flat

851: connection $\nabla_\rho$. They are obtained from the

852: trivial bundle $\tilde M\times V\to\tilde M$ and the trivial connection by

853: passing to the $\Gamma$ quotient spaces. Here $\tilde M$ is the canonical

854: universal covering provided by the base point $x_0$. The $\Gamma$-action is

855: the diagonal action of deck transformations on $\tilde M$ and of the action 

856: $\rho$ on $V$. The fiber of $F_\rho$ over $x_0$ identifies canonically with

857: $V$. The holonomy representation determines a right $\Gamma$-action on the

858: fiber of $F_\rho$ over $x_0$, i.e.\ an anti homomorphism $\Gamma\to\GL(V)$.

859: When composed with the inversion in $\GL(V)$ we get back the representation

860: $\rho$. The pair $(F_\rho,\nabla_\rho)$ will be denoted by $\mathbb F_\rho$.

861: 

862: 

863: If $\rho_0$ is a representation in the connected component 

864: $\Rep_\alpha(\Gamma;V)$ one can identify $\Rep_\alpha(\Gamma;V)$ to the

865: connected component of $\nabla_{\rho_0}$ in the complex analytic space of

866: flat connections of the bundle $F_{\rho_0}$ modulo the group of bundle

867: isomorphisms of $F_{\rho_0}$ which fix the fiber above $x_0$.

868: 

869: 

870: \begin{remark}

871: An element $a\in H_1(M;\Z)$ defines a holomorphic function  

872: $$

873: {\det}_a:\Rep^M(\Gamma;V)\to\mathbb C_*.

874: $$

875: The complex number $\det_a(\rho)$ is

876: the evaluation on $a\in H_1(M;\Z)$ of $\det(\rho):\Gamma\to\C_*$

877: which factors through $H_1(M;\Z)$. Note that for $a,b\in H_1(M;\Z)$ we have 

878: $\det_{a+b}=\det_a\det_b$. 

879: If $a$ is a torsion element, then $\det_a$ is constant equal

880: to a root of unity of order, the order of $a$. 

881: \end{remark}

882: 

883: 

884: 

885: \subsection{The space of cochain complexes}\label{SS:4.3}

886: 

887: 

888: 

889: Let $k= (k_0,k_1,\dotsc,k_n)$ be a string of non-negative integers. The string is 

890: called admissible, and will write $k\geq0$ in this case, 

891: if the following requirements are satisfied

892: \begin{eqnarray}

893: k_0-k_1+k_2\mp\cdots+(-1)^nk_n&=&0

894: \label{E:14}

895: \\

896: k_i-k_{i-1}+ k_{i-2}\mp\cdots+(-1)^ik_0&\geq&0

897: \quad\text{for any $i\leq n-1$.}

898: \label{E:15}

899: \end{eqnarray}

900: 

901: 

902: Denote by $\mathbb D(k)=\mathbb D(k_0,\dotsc,k_n)$

903: the collection of cochain complexes of the form

904: $$

905: C=(C^*,d^*):

906: 0\to C^0 \xrightarrow{d^0} 

907: C^1\xrightarrow{d^1}

908: \cdots 

909: \xrightarrow{d^{n-2}}

910: C^{n-1}

911: \xrightarrow{d^{n-1}} C^n\to 0

912: $$

913: with $C^i:=\C^{k_i}$, and by $\mathbb D_\ac(k)\subseteq\mathbb D(k)$ 

914: the subset of acyclic complexes. Note that $\mathbb D_\ac(k)$ is non-empty

915: iff $k\geq0$. The cochain complex $C$ is determined by 

916: the collection $\{d^i\}$ of linear maps $d^i:\C^{k_i}\to\C^{k_{i+1}}$.

917: If regarded as the subset of those

918: $\{d^i\}\in\bigoplus_{i=0}^{n-1}L(\C^{k_i},\C^{k_{i+1}})$, 

919: with $L(V,W)$ the space of linear maps from $V$ to $W$,

920: which satisfy the quadratic equations $d^{i+1}\cdot d^i=0$,

921: the set $\mathbb D(k)$ is an affine algebraic 

922: set given by degree two homogeneous polynomials and 

923: $\mathbb D_\ac(k)$ is a Zariski open set.

924: The map $\pi_0:\mathbb D_\ac(k)\to\Emb(C^0,C^1)$

925: which associates to $C\in\mathbb D_\ac(k)$

926: the linear map $d^0$, is a bundle whose fiber is isomorphic to 

927: $\mathbb D_\ac(k_1-k_0,k_2,\dotsc,k_n)$. 

928: 

929: 

930: This can be easily generalized as follows.

931: %%%%%%%%%%%

932: %Consider two $n$-strings $k$ and $b=(b_0,\dotsc,b_n)$ and suppose that 

933: %$k-b=(k_0-b_0,\dotsc,k_n-b_n)$ is admissible. We write this as $k\geq b$.  

934: %%%%%%%%%%%

935: Consider a string $b=(b_0,\dotsc,b_n)$. We will write $k\geq b$ if

936: $k-b=(k_0-b_0,\dotsc,k_n-b_n)$ is admissible, i.e.\ $k-b\geq0$.

937: Denote by $\mathbb D_b(k)=\mathbb D_{(b_0,\dotsc,b_n)}(k_0,\dotsc,k_n)$ 

938: the subset of cochain complexes $C\in\mathbb D(k)$ with $\dim(H^i(C))=b_i$.

939: Note that $\mathbb D_b(k)$ is non-empty iff $k\geq b$.

940: The obvious map $\pi_0:\mathbb D_b(k)\to L(C^0,C^1;b_0)$, 

941: $L(C^0,C^1;b_0)$ the space of linear maps in $L(C^0,C^1)$ whose kernel 

942: has dimension $b_0$, 

943: %%%%%%%%%%%%%%%%%%%%%%%%%%%

944: %%$ 

945: %\pi_0:\mathbb D_\ac (k_0,\dotsc, k_n)\to   \Emb (\ker d^0, C^0)\times \Emb(C^0/ \ker d^0 ,C^1)

946: %$

947: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

948: is a bundle whose fiber is isomorphic to 

949: $\mathbb D_{b_1,\dotsc,b_n}(k_1-k_0+b_0,k_2,\dotsc,k_n)$. Note that

950: $L(C^0, C^1; b_0)$ is the total space of a bundle

951: $\Emb(\underline{\C}^{k_0}/L,\underline{\C}^{k_1})\to\Gr_{b_0}(k_0)$ with

952: $L\to\Gr_{b_0}(k_0)$ the tautological bundle over $\Gr_{b_0}(k_0)$ and 

953: $\underline{\C}^{k_0}$ resp.\ $\underline{\C}^{k_1}$ the trivial bundles over 

954: $\Gr_{b_0}(k_0)$ with fibers of dimension $k_0$ resp.\ $k_1$.

955: As a consequence we have

956: \begin{proposition}\label{P:2}

957: 1. $\mathbb D_\ac(k)$ and $\mathbb D_b(k)$ are connected smooth quasi affine  

958: algebraic sets whose dimension is 

959: $$

960: \dim\mathbb D_b(k)=\sum_j(k^j-b^j)\cdot\Bigl(

961: k^j-\sum_{i\leq j}(-1)^{i+j}(k^i-b^i)

962: \Bigr).

963: $$

964: 

965: 

966: 2. The closures $\hat{\mathbb D}_\ac(k)$ and $\hat{\mathbb D}_b(k)$ are irreducible 

967: algebraic sets, hence affine algebraic varieties, and

968: $\hat{\mathbb D}_b(k)=\bigsqcup_{k\geq b'\geq b}{\mathbb D}_{b'}(k)$.

969: \end{proposition}

970: 

971: 

972: For any cochain complex in $C\in\mathbb D_\ac(k)$ denote by 

973: $B^i:=\img(d^{i-1})\subseteq C^i= \C^{k_i}$

974: and consider the short exact sequence 

975: $0\to B^i \xrightarrow{\textrm{inc}} C^i \xrightarrow {d^i} B^{i+1}\to 0$. 

976: Choose a base $\bb_i$ for each $B_i$, and choose lifts $\overline{\bb}_{i+1}$ of

977: $\bb_{i+1}$ in $C^i$ using $d^i$, i.e.\ $d^i(\overline{\bb}_{i+1})=\bb_{i+1}$. 

978: Clearly $\{\bb_i,\overline\bb_{i+1}\}$ is a base of $C^i$.

979: Consider the base $\{\bb_i,\overline{\bb}_{i+1}\}$ as a collection of vectors in 

980: $C^i=\C^{k_i}$ and write them as columns of a matrix 

981: $[\bb_i,\overline{\bb}_{i+1}]$. Define the torsion of the acyclic complex $C$, by 

982: $$

983: \tau(C):=(-1)^{N+1}\prod^n_{i=0}\det[\bb_i,\overline{\bb}_{i+1}]^{(-1)^i}

984: $$

985: where $(-1)^N$ is Turaev's sign, see \cite{FT99}.

986: The result is independent of the choice of 

987: the bases $\bb_i$ and of the lifts $\overline{\bb}_i$ cf.\ \cite{M66}

988: \cite{FT99}, and leads to the function 

989: $$

990: \tau:\mathbb D_\ac(k)\to\C_*.

991: $$ 

992: Turaev provided a simple formula for this function, cf.\ \cite{Tu01}, which  

993: permits to recognize $\tau$ as the restriction of a rational function on  

994: $\hat{\mathbb D}_\ac(k)$.

995: 

996: 

997: For $C\in\hat{\mathbb D}_\ac(k)$ denote by $(d^i)^t:\C^{k_{i+1}}\to\C^{k_i}$ 

998: the transpose of $d^i:\C^{k_i}\to\C^{k_{i+1}}$, and define 

999: $P_i=d^{i-1}\cdot(d^{i-1})^t+(d^i)^t\cdot d^i$. Define $\Sigma(k)$ as the 

1000: subset of cochain complexes in $\hat{\mathbb D}_\ac(k)$ where $\ker P\neq0$, and consider

1001: $S\tau:\hat{\mathbb D}_\ac(k)\setminus\Sigma(k)\to\mathbb C_*$ defined by 

1002: $$

1003: S\tau(C)

1004: :=\Bigl(\prod_{i\ \even}(\det P_i)^{i}

1005: \big/\prod_{i\ \odd}(\det P_i)^{i}\Bigr)^{-1}.

1006: $$

1007: One can verify 

1008: 

1009: 

1010: \begin{proposition} 

1011: Suppose $k=(k_0,\dotsc,k_n)$ is admissible. 

1012: 

1013: 

1014: 1. $\Sigma(k)$ is a proper subvariety containing the singular set of 

1015: $\hat{\mathbb D}_\ac(k)$.

1016: 

1017: 

1018: 2. $S\tau=\tau^2$ and implicitly $S\tau$ has an analytic continuation 

1019: to ${\mathbb D}_\ac(k)$.

1020: \end{proposition}

1021: 

1022: 

1023: In particular $\tau$ defines a square root of $S\tau$.  

1024: We will not use explicitly $S\tau $ in this writing

1025: however it justifies the definition of complex Ray--Singer torsion.

1026:  

1027: 

1028: 

1029: 

1030: \section{Analytic torsion}\label{S:5}

1031: 

1032: 

1033: 

1034: 

1035: Let $M$  be a closed manifold, $g$ Riemannian metric and $(g,\alpha)$ 

1036: a representative of a coEuler structure $\e^*\in\Eul^*(M;\R)$.

1037: Suppose $E\to M$ is a complex vector bundle and denote by $\mathcal C(E)$ 

1038: the space of connections and by $\mathcal F(E)$ the subset of flat connections.

1039: $\mathcal C(E)$ is a complex affine (Fr\'echet) space while $\mathcal F(E)$  

1040: a closed complex analytic subset (Stein space) of $\mathcal C(E)$.

1041: Let $b$ be a non-degenerate symmetric bilinear form and $\mu$ a Hermitian 

1042: (fiber metric) structure on $E$.  While Hermitian structures always exist,  

1043: non-degenerate symmetric bilinear forms exist 

1044: iff the bundle is the complexification of some real vector bundle, and in

1045: this case $E\simeq E^*$.

1046: 

1047: 

1048: The connection $\nabla\in\mathcal C(E)$ can be interpreted as a first order 

1049: differential operator

1050: $d^\nabla:\Omega^*(M;E)\to\Omega^{*+1}(M;E)$ and $g$ and $b$ resp.\ $g$ and $\mu$

1051: can be used to define the formal $b$-adjoint resp.\ $\mu$-adjoint 

1052: $\delta^\nabla_{q;g,b}$ resp.\ $\delta^\nabla_{q;g,\mu}:\Omega^{q+1}(M;E)\to\Omega^q(M;E)$

1053: and therefore the Laplacians

1054: $$

1055: \Delta^\nabla_{q;g,b} 

1056: \ \text{resp.}\

1057: \Delta^\nabla_{q;g,\mu}:\Omega^q(M;E)\to\Omega^q(M;E).

1058: $$

1059: They are elliptic second order differential operators with principal symbol 

1060: $\sigma_\xi=|\xi|^2$. Therefore they have a

1061: unique well defined zeta regularized determinant (modified determinant) 

1062: $\det(\Delta^\nabla_{q;g,b})\in\C$ (${\det}'(\Delta^\nabla_{q;g,b})\in\C_*$)

1063: resp.\ $\det(\Delta^\nabla_{q;g,\mu})\in\R_{\geq 0}$  

1064: (${\det}'(\Delta^\nabla_{q;g,\mu})\in\R_{>0}$) calculated with respect to

1065: a non-zero Agmon angle avoiding the spectrum cf.\ \cite{BH05}.

1066: % Agmon angle $\pi$.

1067: %%%%%%%%%%%%

1068: %This determinant is zero if $\ker\Delta^\nabla_q\neq 0$.

1069: %%%%%%%%%%%%%%%

1070: Recall that the zeta regularized determinant (modified determinant) is the 

1071: zeta regularized product of all (non-zero) eigenvalues.

1072: %%%%%%%%%%%%%%

1073: %(sometimes called  the modified zeta regularized determinant.)

1074: %%%%%%%%%%%%%%%

1075: 

1076: 

1077: Denote by 

1078: \begin{align*}

1079: \Sigma(E,g,b)

1080: &:=\bigl\{\nabla\in\mathcal C(E)\bigm| 

1081: \ker(\Delta^\nabla_{*;g,b})\neq0\bigr\}

1082: \\

1083: \Sigma(E,g,\mu)

1084: &:=\bigl\{\nabla\in\mathcal C(E)\bigm| 

1085: \ker(\Delta^\nabla_{*;g,\mu})\neq0\bigr\}

1086: \end{align*}

1087: and by 

1088: $$

1089: \Sigma(E):=\bigl\{\nabla\in\mathcal F(E) 

1090: \bigm| H^*(\Omega^*(M;E),d^\nabla)\neq 0\bigr\}.

1091: $$

1092: Note that $\Sigma(E,g,\mu)\cap\mathcal F(E)=\Sigma(E)$ for any $\mu$,

1093: and $\Sigma(E,g,b)\cap\mathcal F(E)\supseteq\Sigma(E)$. Both,

1094: $\Sigma(E)$ and $\Sigma(E,g,b)\cap\mathcal F(E)$, are closed complex 

1095: analytic subsets of $\mathcal F(E)$, and

1096: $\det(\Delta^\nabla_{q;g,\cdots})

1097: ={\det}'(\Delta^\nabla_{q;g,\cdots})$ on $\mathcal F(E)\setminus \Sigma(E,g,\cdots)$.

1098:  

1099: 

1100: We consider the real analytic functions:

1101: $T^\even_{g,\mu}:\mathcal C(E)\to\mathbb R_{\geq 0}$, 

1102: $T^\odd_{g,\mu}:\mathcal C(E)\to\mathbb R_{\geq 0}$, 

1103: $R_{\alpha, \mu}:\mathcal C(E)\to\mathbb R_{>0}$

1104: and the holomorphic functions 

1105: $T^\even_{g,b}:\mathcal C(E)\to\mathbb C$,

1106: $T^\odd_{g,b}:\mathcal C(E)\to\mathbb C$,

1107: $R_{\alpha,b}:\mathcal C(E)\to \mathbb C_*$

1108: defined by: 

1109: \begin{equation}\label{E;16'}

1110: \begin{aligned}

1111: T^\even_{g,\cdots}(\nabla):=& \prod_{q\ \even}(\det\Delta^\nabla_{q;g,\cdots})^q,

1112: %T'^{\,\even}_{g,\cdots}(\nabla):=\prod_{q\ \even}({\det}'\Delta^\nabla_{q;g,\cdots})^q,

1113: \\

1114: T^\odd_{g,\cdots}(\nabla) :=& \prod_{q\ \odd }(\det\Delta^\nabla_{q;g,\cdots})^q ,

1115: %T'^{\,\odd}_{g,\cdots}(\nabla) := &\prod_{q\ \odd }({\det}'\Delta^\nabla_{q;g,\cdots})^q

1116: \\

1117: R_{\alpha,\cdots}(\nabla):= &e^{\int_M\omega(\cdots,\nabla)\wedge\alpha}.

1118: \end{aligned}

1119: \end{equation}

1120: We also write $T'^{\,\even}_{g,\cdots}$ resp.\ $T'^{\,\odd}_{g,\cdots}$ for the same 

1121: formulas with ${\det}'$ instead of $\det$. These functions are discontinuous on 

1122: $\Sigma(E,g,\cdots)$ and coincide with $T^\even_{g,\cdots}$ resp.\ 

1123: $T^\odd_{g,\cdots}$ on $\mathcal F(E)\setminus\Sigma(E,g,\cdots)$.

1124: Here $\cdots$ stands for either $b$ or $\mu$.

1125: For the  definition of real or complex analytic space/set, holomorphic/meromorphic function/map 

1126: in infinite dimension the reader can consult \cite {D} and \cite{KM}, although the definitions used here are rather straightforward.

1127: 

1128: 

1129: Let $E_r\to M$ be a smooth real vector bundle equipped with a non-degenerate symmetric

1130: positive definite bilinear form $b_r$.

1131: Let $\mathcal C(E_r)$ resp.\ $\mathcal F(E_r)$ the space of connections resp.\ 

1132: flat connections in $E_r$. Denote by $E\to M$ the complexification of $E_r$, 

1133: $E=E_r\otimes\mathbb C$, and by $b$ resp.\ $\mu$ the complexification of $b_r$ 

1134: resp.\ the Hermitian structure  extension of $b_r$. We continue to denote by   

1135: $\mathcal C(E_r)$ resp.\ $\mathcal F(E_r)$ the subspace of $\mathcal C(E)$ 

1136: resp.\ $\mathcal F(E)$ consisting of connections which are complexification of connections  

1137: resp.\ flat connections in $E_r$, and by $\nabla$ the complexification of 

1138: the connection $\nabla\in\mathcal C(E_r)$. If $\nabla\in\mathcal C(E_r)$, then

1139: $$

1140: \Spect\Delta^\nabla_{q;g,b}=\Spect\Delta^\nabla_{q;g,\mu}\subseteq\mathbb R_{\geq 0}

1141: $$

1142: and therefore  

1143: \begin{equation}\label{E:16}

1144: \begin{aligned}

1145: T^{\even/\odd}_{g,b}(\nabla)&=\bigl|T^{\even/\odd}_{g,b}(\nabla)\bigr|

1146: =T^{\even/\odd}_{g,\mu}(\nabla),

1147: \\

1148: T'^{\,\even/\odd}_{g,b}(\nabla)&=\bigl|T'^{\,\even/\odd}_{g,b}(\nabla)\bigr|

1149: =T'^{\,\even/\odd}_{g,\mu}(\nabla),

1150: \\

1151: R_{\alpha,b}(\nabla)&=\bigl|R_{\alpha,b}(\nabla)\bigr|=R_{\alpha,\mu}(\nabla).

1152: \end{aligned}

1153: \end{equation}

1154: 

1155: 

1156: Observe that $\Omega^*(M;E)(0)$ the (generalized) eigen space of

1157: $\Delta^\nabla_{*;g,b}$ corresponding to the eigen value zero is a finite 

1158: dimensional vector space of dimension the multiplicity of $0$. The 

1159: restriction of the symmetric bilinear form induced by $b$ remains 

1160: non-degenerate and defines for each component $\Omega^q(M;E)(0)$ an equivalence 

1161: class of bases. Since $d^\nabla$ commutes with $\Delta^\nabla_{*;g,b}$,  

1162: $\bigl(\Omega^*(M;E)(0),d^\nabla\bigr)$ is a finite dimensional complex. 

1163: When acyclic, i.e.\ $\nabla\in\mathcal F(E)\setminus\Sigma(E)$, denote by

1164: $$

1165: T_\an(\nabla,g,b)(0)\in\mathbb C_*

1166: $$ 

1167: the Milnor torsion associated to the equivalence class of bases induced by $b$.

1168: 

1169: 

1170: 

1171: \subsection {The modified Ray--Singer torsion}\label{SS:5.1}

1172: 

1173: 

1174: 

1175: Let $E\to M$ be a complex vector bundle, 

1176: and let $\e^*\in\Eul^*(M;\mathbb R)$ be a coEuler structure.

1177: Choose a Hermitian structure (fiber metric) $\mu$ on $E$,

1178: a Riemannian metric $g$ on $M$ and $\alpha\in\Omega^{n-1}(M;\mathcal O_M)$

1179: so that $[g,\alpha]=\e^*$, see section~\ref{SS:3.2}.

1180: For $\nabla\in\mathcal F(E)\setminus\Sigma(E)$ consider the quantity

1181: $$

1182: T_\an(\nabla,\mu,g,\alpha):=\bigl(T^{\even}_{g,\mu}(\nabla)

1183: /T^{\odd}_{g,\mu}(\nabla)\bigr)^{-1/2}

1184: \cdot R_{\alpha,\mu}(\nabla)\in\mathbb R_{>0}

1185: $$ 

1186: referred to as the \emph{modified Ray--Singer torsion.} 

1187: The following proposition is a reformulation of one of the main 

1188: theorems in \cite{BZ92}, cf.\ also \cite{BFK01} and \cite{BH04}.

1189: 

1190: 

1191: \begin{proposition}\label{P:3}

1192: If $\nabla\in\mathcal F(E)\setminus\Sigma(E)$, then $T_\an(\nabla,\mu,g,\alpha)$

1193: is gauge invariant and independent of $\mu,g,\alpha$.

1194: \end{proposition}

1195: 

1196: 

1197: When applied to $\mathbb F_\rho$ the number 

1198: $T_\an^{\e^*}(\rho):=T_\an(\nabla_\rho,\mu,g,\alpha)$

1199: defines a real analytic function

1200: $T_\an^{\e^*}:\Rep^M_0(\Gamma;V)\to\R_{>0}$.

1201: It is natural to ask if $T_\an^{\e^*}$ is the absolute value of a holomorphic function.

1202: 

1203: 

1204: The answer is no as one can see on the simplest possible example $M= S^1$ 

1205: equipped with the the canonical coEuler structure $\e^*_0$. 

1206: In this case $\Rep^M(\Gamma;\mathbb C)=\mathbb C\setminus 0$,

1207: and $T_\an^{\e^*_0}(z)=|\frac{(1-z)}{z^{1/2}}|$, cf.\ \cite {BH05}.

1208: However, Theorem~\ref{T:2} in section~\ref{SS:6.1} below provides

1209: the following answer to the question (Q) from the introduction.

1210: 

1211: 

1212: \begin{observation}  

1213: If $\e^*$ is an integral coEuler structure, then 

1214: $T_\an^{\e^*}$ is the absolute value of a holomorphic function on 

1215: $\Rep^M_0(\Gamma;V)$ which is the restriction of a rational function on 

1216: $\Rep^M(\Gamma;V)$. For a general coEuler structure $T^{\e^*}_\an$

1217: still locally is the absolute value of a holomorphic function.

1218: \end{observation}

1219: 

1220: 

1221: 

1222: \subsection{Complex Ray--Singer torsion}\label{SS:5.3}

1223: 

1224: 

1225: 

1226: Let $E$ be a complex vector bundle equipped with a non-degenerate symmetric 

1227: bilinear form $b$. Suppose $(g,\alpha)$ is a pair consisting of a Riemannian 

1228: metric $g$ and a differential form $\alpha\in\Omega^{n-1}(M;\mathcal O_M)$ 

1229: with $d\alpha=\ec(g)$. For any $\nabla\in\mathcal F(E)\setminus\Sigma(E)$ 

1230: consider the complex number 

1231: \begin{equation}\label{E:00}

1232: \mathcal{ST}_\an(\nabla,b,g,\alpha)

1233: :=\bigl(T'^{\,\even}_{g,b}(\nabla)/T'^{\,\odd}_{g,b}(\nabla)\bigr)^{-1}

1234: \cdot R_{\alpha,b}(\nabla)^2

1235: \cdot T_\an(\nabla,g,b)(0)^2\in\mathbb C_*

1236: \end{equation}

1237: referred to as the \emph{complex valued Ray--Singer torsion.}\footnote{The idea 

1238: of considering $b$-Laplacians for torsion was brought to the

1239: attention of the  first author by W.~M\"uller \cite{M}.

1240: The second author came to it independently.}

1241: 

1242: 

1243: It is possible to provide an alternative definition of 

1244: $\mathcal{ST}_\an(\nabla,b,g,\alpha)$.

1245: Suppose $R>0$ is a positive real number so that the Laplacians $\Delta^\nabla_{q;g,b}$ 

1246: have no eigen values of absolute value $R$.

1247: In this case denote by $\det^R\Delta^\nabla_{q;g,b}$ the regularized product of all eigen

1248: values larger than $R$ w.r.\ to a non-zero Agmon angle disjoint from the spectrum  

1249: $T^{R,\even}_{g,b}$ resp.\ $T^{R,\even}_{g,b}$ the quantities defined by the 

1250: formulae~\eqref{E;16'} with $T^{R,\even/\odd}(\Delta)$ instead of 

1251: $T^{'\,\even/\odd}(\Delta)$.

1252: Consider $\Omega^*(M;E)(R)$ to be the sum of generalized  eigen spaces  

1253: of $\Delta^\nabla_{*;g,b}$  corresponding to eigen values smaller in 

1254: absolute value than $R$. $(\Omega^*(M;E)(R), d^\nabla)$ is a finite 

1255: dimensional complex. As before $b$ remains non-degenerate  and when 

1256: acyclic (and this is the case iff $(\Omega^*(M;E),d^\nabla)$ is acyclic) 

1257: denote by $T_\an(\nabla,g,b)(R)$ the Milnor torsion associated to the equivalence 

1258: class of bases induced by $b$.

1259: It is easy to check that 

1260: \begin{equation}\label{E:000}

1261: \mathcal{ST}_\an(\nabla,b,g,\alpha)

1262: =\bigl(T^{R,\even}_{g,b}(\nabla)/T^{R,\odd}_{g,b}(\nabla)\bigr)^{-1}

1263: \cdot R_{\alpha,b}(\nabla)^2

1264: \cdot T_\an(\nabla,g,b)(R)^2

1265: \end{equation}

1266: 

1267: 

1268: \begin{proposition}\label{P:6}

1269: 1. $\mathcal{ST}_\an(\nabla,b,g,\alpha)$ is a holomorphic function  

1270: on $\mathcal F(E)\setminus\Sigma(E)$ and the restriction of a meromorphic 

1271: function on $\mathcal F(E)$ with poles and zeros in $\Sigma(E)$.

1272: 

1273: 

1274: 2. If $b_1$ and $b_2$ are two non-degenerate symmetric bilinear forms which are 

1275: homotopic, then $\mathcal{ST}_\an(\nabla,b_1,g,\alpha)=\mathcal{ST}_\an(\nabla,b_2,g,\alpha)$.

1276: 

1277: 

1278: 3. If $(g_1,\alpha_1)$ and $(g_2,\alpha_2)$ are two pairs representing the same 

1279: coEuler structure, then 

1280: $\mathcal{ST}_\an(\nabla,b,g_1,\alpha_1)=\mathcal{ST}_\an(\nabla,b,g_2,\alpha_2)$.

1281: 

1282: 

1283: 4. We have 

1284: $\mathcal{ST}_\an(\gamma\nabla,\gamma b,g,\alpha)=\mathcal{ST}_\an(\nabla,b,g,\alpha)$

1285: for every gauge transformation $\gamma$ of $E$.

1286: 

1287: 5. $\mathcal{ST}_\an(\nabla_1\oplus\nabla_2,b_1\oplus b_2,g,\alpha)= 

1288: \mathcal{ST}_\an(\nabla_1,b_1,g,\alpha)\cdot\mathcal{ST}_\an(\nabla_2,b_2,g,\alpha)$.

1289: \end{proposition}

1290: 

1291: 

1292: To check the first part of this proposition, one shows that for $\nabla_0\in\mathcal F(E)$  

1293: one can find $R>0$ and an open neighborhood $U$ of $\nabla_0\in \mathcal F(E)$ such that  

1294: no eigen value of $\Delta^\nabla_{q;g,b}$, $\nabla\in U$, has absolute value $R$. 

1295: The function $\bigl(T^{R,\even}_{g,b}(\nabla)/T^{R,\odd}_{g,b}(\nabla)\bigr)^{-1}$ 

1296: is holomorphic in $\nabla\in U$. Moreover, on $U$ the function $T_\an(\nabla,g,b)(R)^2$ is 

1297: meromorphic in $\nabla$, and holomorphic when restricted to $U\setminus\Sigma(E)$.

1298: The statement thus follows from \eqref{E:000}.

1299: 

1300: 

1301: The second and third part of Proposition~\ref{P:6} are derived from formulas for 

1302: $d/dt(\mathcal{ST}_\an(\nabla,b(t),g,\alpha))$ 

1303: resp.\ $d/dt(\mathcal{ST}_\an(\nabla,b,g(t),\alpha)$

1304: which are similar to such formulas for Ray--Singer torsion in the case of a 

1305: Hermitian structure instead of a non-degenerate symmetric bilinear form, cf.\ \cite {BH05}. 

1306: The proof of 4) and 5) require a careful inspection of the definitions.

1307: The full arguments are contained in \cite{BH05}.

1308: 

1309: 

1310: As a consequence to each homotopy class of non-degenerate symmetric bilinear forms $[b]$

1311: %(equivalently isomorphisms between $E^*$ and $E$)

1312: and coEuler structure $\e^*$ we can

1313: associate a meromorphic function on $\mathcal F(E)$.  The reader unfamiliar with the basic concepts 

1314: of complex  analytic geometry on Banach/ Frechet manifolds  can  consult \cite{D} and \cite {KM}.   Changing the coEuler structure our 

1315: function changes by multiplication with a non-vanishing holomorphic function as one can 

1316: see from \eqref{E:00}. Changing the homotopy class $[b]$ is actually more subtle. We 

1317: expect however that $\mathcal{ST}$ remains unchanged when the coEuler structure is integral.

1318: 

1319: 

1320: Denote by $\Rep^{M,E}(\Gamma;V)$ the union of components of $\Rep^M(\Gamma;V)$

1321: which consists of representations equivalent to holonomy representations of flat 

1322: connections in the bundle $E$. Suppose $E$ admits non-degenerate symmetric bilinear 

1323: forms and let $[b]$ be a homotopy class of such forms. Let $x_0\in M$ be a base point

1324: and denote by $\mathcal G(E)_{x_0,[b]}$ 

1325: the group of gauge transformations which leave fixed $E_{x_0}$ and the class $[b]$.

1326: In view of Proposition~\ref{P:6}, $\mathcal{ST}_\an(\nabla, b, g,\alpha)$ defines

1327: a meromorphic function $\mathcal{ST}_\an^{\e^*,[b]}$ on 

1328: $\pi^{-1}(\Rep^{M,E}(\Gamma;V) \subseteq \mathcal F(E)/\mathcal G_{x_0,[b]}$. Note that

1329: $\pi:\mathcal F(E)/\mathcal G_{x_0,[b]}\to\Rep(\Gamma;V)$

1330: is an principal holomorphic covering of its image which contains $\Rep^{M,E}(\Gamma;V)$.

1331: We expect that the absolute value of this function is the square of 

1332: modified Ray--Singer torsion. The expectation is true when $(E,b)$ satisfies $P_r$ below.

1333: 

1334: 

1335: \begin{definition}

1336: The pair $(E,b)$ satisfies \emph{Property $P_r$} if it is the complexification of a 

1337: pair $(E_r,b_r)$ consisting of a real vector bundle $E_r$ and a 

1338: non-degenerate symmetric positive definite $\mathbb R$-bilinear form $b_r$

1339: and the space of flat connections $\mathcal F(E_r)$ 

1340: is a real form of the space $\mathcal F(E)$.

1341: \end{definition}

1342: 

1343: 

1344: We summarize this in the following Theorem.

1345: 

1346: 

1347: \begin{theorem}\label{T:1}

1348: With the hypotheses above we have.

1349: 

1350: 1.

1351: If $\e^*_1$ and $\e^*_2$ are two coEuler structures then

1352: $$

1353: \mathcal{ST}_\an^{\e^*_1,[b]}=\mathcal{ST}_\an^{\e^*_2,[b]}\cdot 

1354: e^{2([\omega(\nabla,b)],D^{-1}(\e^*_1-\e^*_2))}

1355: $$

1356: with $D:H_1(M;\R)\to H^{n-1}(M;\mathcal O_M)$ the Poincar\'e duality isomorphism.

1357: 

1358: Suppose that $(E,b)$ satisfies  property $(P_r)$. Then:

1359: 

1360: 2. 

1361: If $\e^* $ is integral then $\mathcal{ST}_\an^{\e^*,[b]}$ is independent of

1362: $[b]$ and descends to a rational function on 

1363: $\Rep^{M,E}(\Gamma;V)$ denoted $\mathcal{ST}_\an^{\e^*}$.

1364: 

1365: 

1366: 3.

1367: We have

1368: \begin{equation}

1369: \bigl|\mathcal{ST}_\an^{\e^*,[b]}\bigr|=(T_\an^{\e^*}\cdot\pi)^2.

1370: \end{equation}

1371: \end{theorem}

1372: 

1373: 

1374: We expect that both 2) and 3) remain true for an arbitrary pair $(E,b)$.

1375:  

1376: 

1377: \begin{observation}

1378: Property 5) in Proposition~\ref{P:6} shows that up to multiplication with a root of unity 

1379: the complex Ray--Singer torsion can be defined on all components of

1380: $\Rep^M(\Gamma;V)$, since $F=\oplus_kE$ is trivial for sufficiently large $k$.

1381: \end{observation}

1382: 

1383: 

1384: 

1385: \section{Milnor--Turaev and dynamical torsion}\label{S:6}

1386: 

1387: 

1388: 

1389: 

1390: 

1391: \subsection{Milnor--Turaev torsion}\label{SS:6.1}

1392: 

1393: 

1394: 

1395: Consider a smooth triangulation $\tau$ of $M$, and 

1396: choose a collection of orientations $\mathcal O$ of the simplices of

1397: $\tau$. Let $x_0\in M$ be a base point, and set $\Gamma:=\pi_1(M,x_0)$.

1398: Let $V$ be a finite dimensional complex vector space.

1399: For a representation $\rho\in\Rep(\Gamma;V)$, 

1400: consider the chain complex $(C^*_\tau(M;\rho),d^{\mathcal O}_\tau(\rho))$ associated with

1401: the triangulation $\tau$ which computes the cohomology $H^*(M;\rho)$.

1402: 

1403: 

1404: Denote the set of simplexes of dimension $q$ by $\mathcal X_q$, and

1405: set $k_i:=\sharp(\mathcal X_i)\cdot\dim(V)$.

1406: Choose a collection of paths $\sigma:=\{\sigma_s\mid s\in\tau\}$ from

1407: $x_0$ to the barycenters of $\tau$ as in section~\ref{SS:3.1}.

1408: Choose an ordering $o$ of the barycenters

1409: and a framing $\epsilon$ of $V$.

1410: Using $\sigma$, $o$ and $\epsilon$ one can identify $C^q_\tau(M;\rho)$ with

1411: $\C^{k_q}$. We obtain in this way a map 

1412: $$

1413: t_{\mathcal O,\sigma,o,\epsilon}:\Rep(\Gamma;V)\to\mathbb D(k_0,\dotsc,k_n)

1414: $$

1415: which sends $\Rep^M_0(\Gamma;V)$ to 

1416: $\mathbb D_\ac(k_0,\dotsc,k_n)$. A look at the explicit definition of

1417: $d^{\mathcal O}_\tau(\rho)$ implies that $t_{\mathcal O,\sigma,o,\epsilon}$ is actually a 

1418: regular map between two algebraic sets. Change of $\mathcal O,\sigma,o,\epsilon$

1419: changes the map $t_{\mathcal O,\sigma,o,\epsilon}$. 

1420: 

1421: 

1422: Recall that the triangulation $\tau$ determines Euler vector fields $X_\tau$ which 

1423: together with $\sigma$ determine an Euler structure $\e\in\Eul(M;\mathbb Z)$, see

1424: section~\ref{SS:3.1}. Note that the ordering $o$ induces

1425: a cohomology orientation $\mathfrak o$ in $H^*(M;\R)$. In view of the arguments of 

1426: \cite{M66} or \cite{Tu86} one can conclude (cf.\ \cite{BH04}):

1427: 

1428: 

1429: \begin{proposition}\label{P:7}

1430: If $\rho\in\Rep^M_0(\Gamma;V)$ different choices of $\tau, \mathcal O, \sigma, o, \epsilon$

1431: provide the same composition $\tau\cdot t_{\mathcal O, \sigma, o, \epsilon}(\rho)$

1432: provided they define the same Euler structure $\e$ and

1433: homology orientation $\mathfrak o$.

1434: \end{proposition}

1435: 

1436: 

1437: In view of Proposition~\ref{P:7}

1438: we obtain a well defined complex valued rational function

1439: on $\Rep^M(\Gamma;V)$ called the Milnor--Turaev torsion

1440: and denoted from now on by $\mathcal T_\comb^{\e,\mathfrak o}$.

1441: 

1442: 

1443: \begin{theorem}\label{T:2}

1444: 1.

1445: The poles and zeros of $\mathcal T_\comb^{\e,\mathfrak o}$ are contained 

1446: in $\Sigma(M)$, the subvariety of representations $\rho$ with $H^*(M;\rho)\neq 0$.

1447: 

1448: 

1449: 2.

1450: The absolute value of $\mathcal T_\comb^{\e,\mathfrak o}(\rho)$

1451: calculated on $\rho\in\Rep^M_0(\Gamma;V)$ is the modified Ray--Singer 

1452: torsion $T_{\an}^{\e^*}(\rho)$, where $\e^*=\PD(\e)$.

1453: 

1454: 

1455: 3.

1456: If $\e_1$ and $\e_2$ are two Euler structures then 

1457: $\mathcal T_\comb^{\e_2,\mathfrak o}

1458: =\mathcal T_\comb^{\e_1,\mathfrak o}

1459: \cdot\det_{\e_2-\e_1}$

1460: and $\mathcal T_\comb^{\e,-\mathfrak o}=

1461: (-1)^{\dim V}\cdot\mathcal T_\comb^{\e,\mathfrak o}$ 

1462: where $\det_{\e_2-\e_1}$ is the regular function on $\Rep^M(\Gamma;V)$ 

1463: defined in section~\ref{SS:4.1}.

1464: 

1465: 

1466: 4.

1467: When restricted to $\Rep^{M,E}(\Gamma;V)$, $E$ a complex vector bundle equipped with a

1468: non-degenerate symmetric bilinear form  $b$ so that $(E,b)$ satisfies Property $P_r$, 

1469: $(\mathcal T_\comb^{\e,\mathfrak o})^2=\mathcal{ST}_\an^{\e^*}$,

1470: where $\e^*=\PD(\e)$.

1471: \end{theorem}

1472: 

1473: 

1474: We expect that 4) remains true without any hypothesis.

1475: Parts 1) and 3) follow from the definition and the general properties of

1476: $\tau$, part 2) can be derived from the work of Bismut--Zhang \cite{BZ92} 

1477: cf.\ also \cite{BFK01}, and part 4) is discussed in \cite{BH05}, Remark 5.11.

1478:  

1479: 

1480: 

1481: \subsection{Dynamical torsion}\label{SS:6.2}

1482: 

1483: 

1484: 

1485: Let $X$ be a vector field on $M$ satisfying 

1486: (H), (EG), (L), (MS) and (NCT) from section~\ref{SS:2.3}. 

1487: Choose orientations $\mathcal O$ of the unstable manifolds.

1488: Let $x_0\in M$ be a base point and set $\Gamma:=\pi_1(M,x_0)$.

1489: Let $V$ be a finite dimensional complex vector space.

1490: For a representation $\rho\in\Rep(\Gamma;V)$ 

1491: consider the associated flat bundle $(F_\rho,\nabla_\rho)$, and set

1492: $C^q_X(M;\rho):=\Gamma(F_\rho|_{\mathcal X_q})$, where $\mathcal X_q$

1493: denotes the set of zeros of index $q$. Recall that for 

1494: $x\in\mathcal X$, $y\in\mathcal X$ and every 

1495: homotopy class $\hat\alpha\in\mathcal P_{x,y}$ parallel transport

1496: provides an isomorphism $(\pt^\rho_{\hat\alpha})^{-1}:(F_\rho)_y\to(F_\rho)_x$.

1497: For $x\in\mathcal X_q$ and $y\in\mathcal X_{q-1}$

1498: consider the expression:

1499: %\begin{equation}\label{E:18}

1500: %\delta_X^{\mathcal O}(\rho)_{x,y}:=\sum_{\theta\in\mathcal T(x,y)}

1501: %\epsilon(\theta)(\pt^\rho_\theta)^{-1}

1502: %\end{equation}

1503: \begin{equation}\label{E:18}

1504: \delta_X^{\mathcal O}(\rho)_{x,y}:=\sum_{\hat\alpha\in\mathcal P_{x,y}}

1505: \mathbb I_{x,y}^{X,\mathcal O}(\hat\alpha)(\pt^\rho_{\hat\alpha})^{-1}.

1506: \end{equation}

1507: If the right hand side of \eqref{E:18} is absolutely convergent for all $x$

1508: and $y$ they provide a linear mapping $\delta^{\mathcal O}_X(\rho):C^{q-1}_X(M;\rho)\to

1509: C^q_X(M;\rho)$ which, in view of Proposition~\ref{P:1}(3), makes 

1510: $\bigl(C^*_X(M;\rho),\delta_X^{\mathcal O}(\rho)\bigr)$ a cochain complex.

1511: There is an integration homomorphism

1512: $\Int_X^{\mathcal O}(\rho):\bigl(\Omega^*(M;F_\rho),d^{\nabla_\rho}\bigr)\to

1513: \bigl(C^*_X(M;\rho),\delta_X^{\mathcal O}(\rho)\bigr)$

1514: which does not always induce an isomorphism in cohomology.

1515: 

1516: 

1517: Recall that for every $\rho\in\Rep(\Gamma;V)$ 

1518: the composition $\tr\cdot\rho^{-1}:\Gamma\to\C$ factors through conjugacy classes

1519: to a function $\tr\cdot\rho^{-1}:[S^1,M]\to\C$.

1520: Let us also consider the expression

1521: %\begin{equation}\label{E:19}

1522: %P_X(\rho):=\sum_{\hat\theta}(\epsilon(\hat\theta)/p(\hat\theta))

1523: %\tr(\rho(\hat\theta)^{-1})

1524: %\end{equation}

1525: \begin{equation}\label{E:19}

1526: P_X(\rho):=\sum_{\gamma\in[S^1,M]}\mathbb Z_X(\gamma)(\tr\cdot\rho^{-1})(\gamma).

1527: \end{equation}

1528: Again, the right hand side of \eqref{E:19} will in general not converge.

1529: 

1530: 

1531: \begin{proposition}\label{P:8}

1532: 

1533: There exists an open set $U$ in $\Rep^M(\Gamma;V)$, intersecting every irreducible 

1534: component, s.t.\ for any representation $\rho\in U$ we have:

1535: 

1536: a) The differentials $\delta^{\mathcal O}_X(\rho)$ converge absolutely.

1537: 

1538: b) The integration $\Int_X^{\mathcal O}(\rho)$ converges absolutely.

1539: 

1540: c) The integration $\Int_X^{\mathcal O}(\rho)$ induces an isomorphism in cohomology.

1541:  

1542: d) If in addition $\dim V=1$, then 

1543: \begin{equation}\label{E:19'}

1544: \sum_{\sigma\in H_1(M;\Z)/\Tor(H_1(M;\Z))}\,\biggl|\sum_{[\gamma]\in\sigma}

1545: \mathbb Z_X(\gamma)(\tr\cdot\rho^{-1})(\gamma)\biggr|

1546: \end{equation}

1547: converges, cf.~\eqref{E:19}. Here the inner (finite) sum is over all

1548: $\gamma\in[S^1,M]$ which give rise to $\sigma\in H_1(M;\Z)/\Tor(H_1(M;\Z))$.

1549: \end{proposition}

1550: 

1551: %%%%%%%%%%%%%%%%%

1552: %a) The set $U$ of representations $\rho$ in $\Rep^M_0(\Gamma;V)$ for which the 

1553: %right side of the formula \eqref{E:18} is absolutely convergent 

1554: %has an open interior in any connected component of $\Rep^M(\Gamma;V)$.

1555: %The subset $\Sigma\subset U$ consisting of the representations $\rho\in U$  

1556: %where $\Int_X^{\mathcal O}(\rho)$ does not induce an isomorphism in cohomology 

1557: %is a closed proper complex analytic subset of $U$. 

1558: %b) If $\dim V=1$ the set $U$ of representations $\rho$ in $\Rep^M_0(\Gamma;V)$ for which the 

1559: %right side of the formula 

1560: % \eqref{E:19} 

1561: %is absolutely convergent 

1562: %has an open interior in any connected component of $\Rep^M(\Gamma;V)$. I

1563: %The subset $\Sigma\subset U$ consisting of the representations $\rho\in U$  

1564: %where $\Int_X^{\mathcal O}(\rho)$ does not induce an isomorphism in cohomology 

1565: %is a closed proper complex analytic subset of $U.$ 

1566: %\end{proposition}

1567: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1568: This  Proposition is a consequence of exponential growth property (EG) and requires 

1569: (for d)) Hutchings--Lee or Pajitnov results. 

1570: A proof in the case $\dim V=1$ is presented in \cite{BH03'}.

1571: The convergence of \eqref{E:19'} is derived 

1572: from the interpretation of this sum as the Laplace transform of a Dirichlet 

1573: series with a positive abscissa of convergence.

1574: 

1575: 

1576: We expect d) to remain true for $V$ of arbitrary dimension.\footnote{Even more, 

1577: we conjecture that \eqref{E:19} converges absolutely on

1578: an open set $U$ as in Proposition~\ref{P:8}.} In this case we make \eqref{E:19}

1579: precise by setting

1580: \begin{equation}\label{E:19''}

1581: P_X(\rho):=\sum_{\sigma\in H_1(M;\Z)/\Tor(H_1(M;\Z))}\,\sum_{[\gamma]\in\sigma}

1582: \mathbb Z_X(\gamma)(\tr\cdot\rho^{-1})(\gamma).

1583: \end{equation}

1584: 

1585: 

1586: \begin{observation}

1587: A Lyapunov closed one form $\omega$ for $X$ permits  to consider the family

1588: of regular functions $P_{X;R}$, $R\in\R$, on the variety $\Rep(\Gamma;V)$ 

1589: defined by: 

1590: $$

1591: P_{X;R}(\rho):=\sum_{\hat\theta,-\omega(\hat\theta)\leq R}

1592: (\epsilon(\hat\theta)/p(\hat\theta))\tr(\rho(\hat\theta)^{-1}).

1593: $$ 

1594: If \eqref{E:19'} converges then   

1595: $\lim_{R\to \infty}P_{X;R}$ exists for $\rho$ in an open set of 

1596: representations. We expect that by analytic continuation this can be defined for 

1597: all representations except ones in a proper algebraic subvariety. This is the case 

1598: when $\dim V=1$ or, for $V$ of arbitrary dimension,  when  the  vector field $X$ 

1599: has only finitely many simple closed trajectories.

1600: In this case $\lim_{R\to \infty}P_{X;R}$ has an analytic continuation to a  

1601: rational function on $\Rep(\Gamma;V)$, see section~\ref{S:8} below. 

1602: \end{observation}

1603: 

1604: As in section~\ref{SS:6.1}, we choose a collection of paths

1605: $\sigma:=\{\sigma_x\mid x\in\mathcal X\}$ from $x_0$ to the zeros of $X$,

1606: an ordering $o'$ of $\mathcal X$, and a framing $\epsilon$ of $V$.

1607: Using $\sigma,o,\epsilon$ we can identify $C^q_X(M;\rho)$ with $\mathbb

1608: C^{k_q}$, where $k_q:=\sharp(\mathcal X_q)\cdot\dim(V)$.

1609: As in the previous section we obtain in this way a holomorphic map

1610: $$

1611: t_{\mathcal O,\sigma,o',\epsilon}:

1612: U\to\hat{\mathbb D}_\ac(k_0,\dotsc,k_n).

1613: $$

1614: 

1615: 

1616: An ordering $o'$ of $\mathcal X$ is given by orderings $o'_q$ of 

1617: $\mathcal X_q$, $q=0,1,\dotsc,n$. Two orderings $o'_1$ and $o'_2$ 

1618: are equivalent if $o'_{1,q}$ is obtained from $o'_{2,q}$ by a permutation 

1619: $\pi_q$ so that $\prod_q\sgn(\pi_q)=1$. 

1620: We call an equivalence class of such orderings a \emph{rest point orientation.}

1621: Let us write $\mathfrak o'$ for the rest point orientation determined by $o'$.

1622: Moreover, let $\e$ denote the Euler structure represented by $X$ and $\sigma$, see

1623: Observation~\ref{O:1}.

1624: As in the previous section, the composition $\tau\cdot t_{\mathcal

1625: O,\sigma,o',\epsilon}:U\setminus\Sigma\to\mathbb C_*$ is a holomorphic map

1626: which only depends on $\e$ and $\mathfrak o'$, and will be denoted by

1627: $\tau_X^{\e,\mathfrak o'}$. Consider the holomorphic map 

1628: $P_X:U\to\mathbb C$ defined by formula~\eqref{E:19}. 

1629: The \emph{dynamical torsion} 

1630: is the partially defined holomorphic function

1631: $$

1632: \mathcal T^{\e,\mathfrak o'}_X:=\tau^{\e,\mathfrak o'}_X\cdot e^{P_X}:

1633: U\setminus\Sigma\to\mathbb C_*.

1634: $$

1635: 

1636: %%%%%%%%%%%%%%%%%%%%%%%%

1637: %%\begin{observation}

1638: %%The absolute convergence of  (\ref {E:19}) is derived in case $\dim V=1$ from the interpretation 

1639: %%of this sum as the Laplace transform of a Dirichlet series with a positive abscissa of convergence.

1640: %A Lyapunov closed one form $\omega$ for $X$ permits to consider the family

1641: %of regular functions on the  variety  on $\R(\Gamma; V),$  $P_{X,R}$  

1642: %$R\in \mathbb R $  defined by :%of regular functions on $\R(\Gamma; V)$,

1643: %%write down   $P_X:\equiv

1644: %%\lim _{R\to \infty} P_{X; R}$ with 

1645: %$$P_{X; R}(\rho):=\sum_{\hat\theta, \omega(\hat\theta)<R}(\epsilon(\hat\theta)/p(\hat\theta))

1646: %\tr(\rho(\hat\theta)^{-1}$$

1647: % We expect that  %as in the case $\dim V=1$ 

1648: %$\lim _{R\to \infty} P_{X; R}$ exists  at least  for $\rho$ in an open set of representations 

1649: %and  by analytic continuation be defined for  all representations except ones in a proper algebraic subvariety.

1650: %This is the case when $\dim V=1$ or,  for $V$ of arbitrary dimension,  when  the  vector field $X$ has only finitely many simple closed trajectories.

1651: %%when

1652: %In this case $\lim _{R\to \infty} P_{X; R}$ has an analytic continuation to a  a rational function 

1653: %on $\R(\Gamma;V), $

1654: %see section 8 below. 

1655: %\end{observation}

1656: %%%%%%%%%%%%%%%%

1657: 

1658: 

1659: 

1660: 

1661: The following result is based on a theorem of 

1662: Hutchings--Lee and Pajitnov \cite{H02} cf.\  \cite {BH03'}.

1663: 

1664: 

1665: \begin{theorem}\label{T:3}  If $\dim V=1$

1666: the partially defined holomorphic function $\mathcal T^{\e,\mathfrak o'}_X$

1667: has an analytic continuation to a rational function equal to 

1668: $\pm\mathcal T_\comb^{\e,\mathfrak o}$.

1669: \end{theorem}

1670: 

1671: 

1672: It is hoped  that a  generalization  of Hutchings--Lee and Pajitnov  results 

1673: which will be elaborated in subsequent 

1674: work \cite{BH05b} might led to the proof of the above result for $V$ of arbitrary 

1675: dimension. 

1676: 

1677: 

1678: 

1679: \section{Examples}\label{S:7}

1680: 

1681: 

1682: \subsection{Milnor--Turaev torsion for mapping tori and twisted Lefschetz zeta function}\label{SS:7.1}

1683: 

1684: 

1685: 

1686: Let $\Gamma_0$ be a group, $\alpha:\Gamma_0\to \Gamma_0$ an isomorphism

1687: and $V$ a complex vector space. Denote by $\Gamma:=\Gamma_0\times_{\alpha}\mathbb Z$ 

1688: the group whose underlying set is $\Gamma_0\times\mathbb Z$ and group operation 

1689: $(g',n) * (g'',m) :=(\alpha^m(g')\cdot g'', n+m)$. 

1690: A representation $\rho:\Gamma\to\GL(V)$ determines a representation 

1691: $\rho_0(\rho):\Gamma_0\to\GL(V)$ the restriction of $\rho$ to $\Gamma_0\times 0$ 

1692: and an isomorphism of $V$, $\theta(\rho)\in\GL(V)$.

1693: 

1694: 

1695: Let $(X,x_0)$ be a based  point compact space with $\pi_1(X,x_0)=\Gamma_0$ and 

1696: $f:(X,x_0) \to (X,x_0)$ a homotopy equivalence. For any integer $k$ the map $f$ 

1697: induces the linear isomorphism $f^k: H^k(X; V) \to H^k(X; V)$ and then the standard 

1698: Lefschetz zeta function 

1699: $$

1700: \zeta_f (z):=\frac{\prod_{k\ \even}\det(I-zf^k)}{\prod_{k\ \odd}\det(I-zf^k)}.

1701: $$

1702: 

1703: More general if $\rho$ is a representation of $\Gamma$ then $f$ and 

1704: $\rho=(\rho_0(\rho), \theta(\rho))$ induce the linear 

1705: isomorphisms $f^k_\rho: H^k(X;\rho_0(\rho)) \to H^k(X; \rho_0(\rho))$ 

1706: and then the $\rho$-twisted  Lefschetz zeta function 

1707: $$

1708: \zeta_f(\rho,z):=\frac{\prod_{k\ \even}\det(I-zf^k_\rho)}{\prod_{k\ \odd}\det(I-zf^k_\rho)}.

1709: $$     

1710: %%%%%%%%%%%%%%%%

1711: %When $\rho$ is the trivial representation the twisted zeta function is the standard zeta function.

1712: %%%%%%%%%%%%%%%%

1713: 

1714: 

1715: Let $N$ be a closed connected manifold and $\varphi:N\to N$ 

1716: a diffeomorphism. Without  loss of generality one can suppose 

1717: that $y_0\in N$ is a fixed point of $\varphi$. Define the mapping torus 

1718: $M=N_\varphi$, the manifold obtained from $N\times I$ identifying 

1719: $(x,1)$ with $(\varphi(x),0)$. Let $x_0=(y_0,0)\in M$ be a base point of $M$.

1720: Set $\Gamma_0:=\pi_1(N,n_0)$ and denote by $\alpha:\pi_1(N,y_0)\to \pi_1(N,y_0)$ 

1721: the isomorphism induced by $\varphi$.  We are in the situation considered above 

1722: with $\Gamma=\pi_1(M,x_0)$. The mapping torus structure on $M$ equips $M$ with a 

1723: canonical Euler structure $\e$ and canonical homology orientation $\mathfrak o$. 

1724: The Euler structure $\e$ is defined by any vector field $X$ with $\omega(X)<0$ 

1725: where $\omega:=p^*dt\in\Omega^1(M;\mathbb R)$; all are homotopic. 

1726: The Wang sequence 

1727: \begin{equation}\label{E:Wang}

1728: \cdots\to H^*(M;\mathbb F_\rho)\to H^*(N;i^* (\mathbb F_\rho))

1729: \xrightarrow{\varphi^*_{ \rho} -\id} 

1730: H^*(N;i^*(\mathbb F_\rho) )

1731: \to H^{*+1}(M;\mathbb F_\rho)\to\cdots

1732: \end{equation}

1733: implies $H^*(M;\mathbb F_\rho)=0$ iff $\det(I-\varphi^k_\rho)\neq0$ for all $k$.

1734: The cohomology orientation is derived from the

1735: Wang long exact sequence for the trivial one dimensional real

1736: representation. For details see \cite{BH04}.

1737: We have 

1738: 

1739: 

1740: \begin{proposition}\label{P:11}

1741: With these notations $\mathcal T_\comb^{\e,\mathfrak o}(\rho)=\zeta_\varphi(\rho,1)$.

1742: \end{proposition}

1743: 

1744: 

1745: This result is known cf.\ \cite{BJ96}. A proof can be also derived easily from \cite{BH04}.

1746: 

1747: 

1748: 

1749: \subsection{Vector fields without rest points and Lyapunov cohomology class}\label{SS:7.2}

1750: 

1751: 

1752: 

1753: Let $X$ be a vector field without rest points,

1754: and suppose $X$ satisfies (L) and (NCT).

1755: As in the previous section $X$ defines an Euler structure $\e$.

1756: Consider the expression 

1757: \eqref{E:19}.

1758: By Theorem~\ref{T:3} we have:  

1759: 

1760: 

1761: \begin{observation}\label{O:5}

1762: With the hypothesis above there exists an open set $U\subseteq\Rep^M(\Gamma;V)$ 

1763: so that \eqref{E:19''} converges,

1764: %$P_X(\rho):=\sum_{\hat\theta}(\epsilon(\hat\theta)/p(\hat\theta))\tr(\rho(\hat\theta)^{-1})$

1765: and $e^{P_X}$ is a well defined holomorphic function on $U$. The function $e^{P_X}$ has 

1766: an analytic continuation to a rational function on $\Rep^M(\Gamma;V)$ equal to 

1767: $\pm\mathcal T^{\e,\mathfrak o}_\comb$. The set $U$ intersects 

1768: non-trivially each connected component of $\Rep^M(\Gamma;V)$.

1769: \end{observation}

1770: 

1771: 

1772: 

1773: \subsection{The Alexander polynomial}\label{SS:7.3}

1774: 

1775: 

1776: 

1777: If $M$ is obtained by surgery on a framed knot, and $\dim V=1$, then

1778: %$\pi_1(M)/[\pi_1(M),\pi_1(M)]=\Z$, 

1779: $\Rep(\Gamma;V)=\C\setminus0$,

1780: and the function

1781: $(z-1)^2  \mathcal T_\comb^{\e,\mathfrak o}$ equals the Alexander polynomial of the knot,

1782: see \cite{Tu02}. Any twisted Alexander polynomial

1783: of the knot can be also recovered from $\mathcal T_\comb^{\e,\mathfrak o}$ for $V$ of 

1784: higher dimension.  One expects  that passing to higher dimensional

1785: representations $\mathcal T_\comb^{\e,\mathfrak o}$ captures even more subtle

1786: knot invariants.

1787: 

1788: 

1789: 

1790: \section{Applications}\label{S:8}

1791: 

1792: 

1793: 

1794: 

1795: 

1796: \subsection{The invariant $A^{\e^*}(\rho_1,\rho_2)$}\label{SS:8.1}

1797: 

1798: 

1799: 

1800: Let $M$ be a $V$-acyclic manifold and $\e^*$ a coEuler structure.

1801: Using the modified Ray--Singer torsion we define a $\R/\pi\Z$ valued

1802: invariant (which resembles the Atiyah--Patodi--Singer

1803: spectral flow) for two representations $\rho_1$, $\rho_2$ in the same 

1804: component of $\Rep^M_0(\Gamma;V)$.

1805: 

1806: 

1807: By a holomorphic path in $\Rep^M_0(\Gamma;V)$ we understand a holomorphic map

1808: $\tilde\rho:U\to\Rep^M_0(\Gamma;V)$ where $U$ is an open neighborhood 

1809: of the segment of real numbers $[1,2]\times\{0\}\subset\C$ in the 

1810: complex plane. For a coEuler structure $\e^*$

1811: and a holomorphic path $\tilde\rho$ in $\Rep_0^M(\Gamma;V)$ define

1812: \begin{equation}\label{E:01}

1813: \arg^{\e^*}(\tilde\rho)

1814: :=\Re\biggl(2/i\int_1^2\frac{{\partial}(T_{\an}^{\e^*}\circ\tilde\rho)}

1815: {T_{\an}^{\e^*}\circ\tilde\rho}\biggr) \mod\pi.

1816: \end{equation}

1817: Here, for a smooth function $\varphi$ of complex variable $z$, 

1818: $\partial\varphi$ denotes the complex valued $1$-form $(\partial\varphi/\partial z)dz$

1819: and the integration is along the path $[1,2] \times0\subset U$.

1820: Note that

1821: 

1822: 

1823: \begin{observation}\label {O:4}

1824: 1. 

1825: Suppose $E$ is a complex vector bundle with a non-degenerate bilinear form

1826: $b$, and suppose $\tilde\rho$

1827: is a holomorphic path in $\Rep^{M,E}_0(\Gamma;V)$. Then

1828: $$

1829: \arg^{\e^*}(\tilde\rho)=\arg\Bigl(\mathcal{ST}_\an^{\e^*,[b]}(\tilde\rho(2))

1830: \big/\mathcal{ST}_\an^{\e^*,[b]}(\tilde\rho(1))\Bigr) \mod\pi.

1831: $$

1832: As consequence

1833: 

1834: 

1835: 2. 

1836: If $\tilde\rho'$ and $\tilde\rho''$ are two holomorphic paths in

1837: $\Rep^M_0(\Gamma;V)$ with $\tilde\rho'(1)=\tilde\rho''(1)$

1838: and $\tilde\rho'(2)=\tilde\rho''(2)$ then

1839: $$

1840: \arg^{\e^*}(\tilde\rho')=\arg^{\e^*}(\tilde\rho'') \mod\pi.

1841: $$

1842: 

1843: 

1844: 3. 

1845: If $\tilde\rho'$, $\tilde\rho''$ and $\tilde\rho'''$ are three holomorphic

1846: paths in $\Rep^M_0(\Gamma;V)$ with $\tilde\rho'(1)=\tilde\rho'''(1)$,

1847: $\tilde\rho'(2)=\tilde\rho''(1)$ and $\tilde\rho''(2)=\tilde\rho'''(2)$ then

1848: $$

1849: \arg^{\e^*}(\tilde\rho''')=\arg^{\e^*}(\tilde\rho')+\arg^{\e^*}(\tilde\rho'')\mod\pi.

1850: $$

1851: \end{observation}

1852: 

1853: 

1854: Observation~\ref{O:4} permits to define a $\R/\pi\Z$ valued numerical

1855: invariant $A^{\e^*}(\rho_1,\rho_2)$ associated to a coEuler 

1856: structure $\e^*$ and two

1857: representations $\rho_1,\rho_2$ in the same connected component of

1858: $\Rep^M_0(\Gamma;V)$.

1859: If there exists a holomorphic path with $\tilde\rho(1)=\rho_1$

1860: and $\tilde\rho(2)=\rho_2$ we set

1861: $$

1862: A^{\e^*}(\rho_1,\rho_2):=\arg^{\e^*}(\tilde\rho)\mod\pi.

1863: $$

1864: Given any two representations $\rho_1$ and $\rho_2$ in the same component of 

1865: $\Rep^M_0(\Gamma;V)$ one can always find a finite collection of holomorphic paths 

1866: $\tilde\rho_i$, $1\leq i\leq k$, in $\Rep^M_0(\Gamma;V)$ so that

1867: $\tilde\rho_i(2)=\tilde\rho_{i+1}(1)$ for all $1\leq i<k$, and such that

1868: $\tilde\rho_1(1)=\rho_1$ and $\tilde\rho_k(2)=\rho_2$.

1869: Then take

1870: $$

1871: A^{\e^*}(\rho_1,\rho_2):=\sum_{i=1}^k\arg^{\e^*}(\tilde\rho_i)\mod\pi.

1872: $$

1873: 

1874: 

1875: In view of Observation~\ref{O:4} the invariant is well defined, and if

1876: $\e^*$ is integral it is actually well defined in $\R/2\pi\Z$.

1877: This invariant was first introduced  when the authors were not fully aware of

1878: ``the complex Ray--Singer torsion.''

1879: The formula \eqref{E:01} is a more or less obvious

1880: expression of the phase of a holomorphic function in terms of its absolute value,

1881: the Ray--Singer torsion, as positive real valued function.

1882: By Theorem~\ref{T:2} the invariant can  be computed

1883: with combinatorial topology and by section~\ref{S:7}

1884: quite explicitly in some cases. If the representations $\rho_1, \rho_2$

1885: are unimodular then the

1886: coEuler structure is irrelevant. It is interesting to compare this 

1887: invariant to the Atiyah--Patodi--Singer spectral flow; it is not the same but are related.

1888: 

1889: 

1890: \subsection{Novikov conjecture}

1891: 

1892: 

1893: Let $X$ be  a smooth  vector field which satisfies (H), (L), (MS), (NCT).  

1894: Suppose $\omega$ is a real valued closed one form so that $\omega(X)_x<0$, $x$ 

1895: not a rest point (Lyapunov form). Define the  functions  $I^X_{x,y}:\mathbb R\to\mathbb Z$ 

1896: and $Z^X:\mathbb R\to \mathbb Q$ by  

1897: \begin{equation}

1898: \begin{aligned}

1899: I^{X,\mathcal O}_{x,y}(R):= &\sum _{\hat \alpha,\  \omega (\hat \alpha)<R} 

1900: \mathbb I^{X,\mathcal O}_{x,y} (\hat \alpha)\\

1901: Z^X(R):= &\sum _{\hat \theta,\  \omega (\hat \theta)<R} \mathbb Z^X(\hat\theta) 

1902: \end{aligned}

1903: \end{equation}

1904: 

1905: Part (a) of the following conjecture was formulated by Novikov for $X=\grad_g\omega$,

1906: $\omega$ a Morse closed one  form  when this vector field satisfies the above properties.

1907: 

1908: 

1909: \begin{conjecture}

1910: a) The function $I^{X,\mathcal O}_{x,y} (R)$ has exponential growth.

1911: 

1912: b) The function $Z^X(R)$ has exponential growth.

1913: \end{conjecture}

1914: 

1915: 

1916: Recall that a function $f:\mathbb R \to \mathbb R$ is said to have exponential growth iff

1917: there exists constants $C_1, C_2$ so that  $| f(x) |  <C_1 e^{C_2}$.

1918: 

1919: 

1920: As a straight forward consequence of Proposition~\ref{P:8} we have  

1921: 

1922: 

1923: %%%%%%%%%%%%%%%%%%%%%%%

1924: %\begin{theorem}

1925: %If $M$ is $V$-acyclic for some $V$ then for any vector field $X$ which 

1926: %satisfies (H), (EG), (L), (MS), (NCT)  and $\omega$ a Lyapunov closed one 

1927: %form the above conjecture is true.

1928: %

1929: %The set of vector fields which satisfy (H), (EG), (L), (MS), (NCT)  

1930: %is $C^0$ dense in the space of vector fields which satisfy (H), (L), (MS), (NCT).

1931: %\end{theorem}

1932: %%%%%%%%%%%%%%%%%%%%

1933: 

1934: \begin{theorem}

1935: a) Suppose $X$ satisfies (H), (MS), (L) and (EG). Then part a) of the

1936: conjecture above holds.

1937: 

1938: b) Suppose $M$ is $V$-acyclic for some $V$ with $\dim(V)=1$. Moreover, assume

1939: $X$ satisfies (H), (MS), (L), (NTC) and (EG). Then part b) of the conjecture above holds.

1940: \end{theorem}

1941: 

1942: 

1943: This result is proved in \cite {BH05}; The $V$-acyclicity in part b) is not

1944: necessary if (EG) is replaced by an apparently stronger assumption (SEG).

1945: Prior to our work Pajitnov has  considered for vector fields which satisfy  (H), 

1946: (L), (MS), (NCT) an additional property, condition  $(\mathfrak C \mathcal Y)$, 

1947: and has verified  part (a) of this conjecture. He has also shown that the vector 

1948: fields which satisfy  (H), (L), (MS), (NCT)  and $(\mathfrak C \mathcal Y)$

1949: are actually $C^0$ dense in the space of vector fields which satisfy (H), 

1950: (L), (MS), (NCT). It is shown in 

1951: \cite{BH05} that Pajitnov vector fields satisfy (EG), and in fact (SEG).

1952: 

1953: 

1954: 

1955: \subsection{A question in dynamics}\label{SS:7.4}

1956: 

1957: 

1958: 

1959: Let $\Gamma$ be a finitely presented group, $V$ a complex vector space and  

1960: $\Rep(\Gamma;V)$ the variety of complex representations.

1961: Consider triples $\underline a:=\{a,\epsilon_-,\epsilon_+\}$  

1962: where $a$ is a conjugacy class of $\Gamma$ and $\epsilon_\pm\in\{\pm1\}$.

1963: Define the rational function $\llet_{\underline a}:\Rep(\Gamma;V)\to\mathbb C$

1964: by

1965: $$

1966: \llet_{\underline a}(\rho)

1967: :=\Bigl(\det\bigl(\id-(-1)^{\epsilon_-}\rho(\alpha)^{-1}\bigr)

1968: \Bigr)^{(-1)^{\epsilon_-+\epsilon_+}}

1969: $$

1970: where $\alpha \in \Gamma$ is a representative of $a.$

1971: 

1972: 

1973: Let $(M,x_0)$ be a $V$-acyclic manifold and $\Gamma=\pi_1(M,x_0)$.

1974: Note that $[S^1,M]$ identifies with the conjugacy classes of $\Gamma$.

1975: Suppose $X$ is a vector field satisfying (L) and (NCT).

1976: Every closed trajectory $\hat\theta$ 

1977: gives rise to a conjugacy class $[\hat\theta]\in[S^1,M]$

1978: and two signs $\epsilon_\pm(\hat\theta)$. These signs are obtained

1979: from the differential of the return map in normal direction;

1980: $\epsilon_-(\hat\theta)$ is the parity of the number of real eigenvalues 

1981: larger than $+1$ and $\epsilon_-(\hat\theta)$ is the parity of

1982: the number of real eigenvalues smaller than $-1$.

1983: For a simple closed trajectory, i.e.\ of period $p(\hat\theta)=1$, let

1984: us consider the triple $\underline{\hat\theta}

1985: :=\bigl([\hat\theta],\epsilon_-(\hat\theta),\epsilon_+(\hat\theta)\bigr)$.

1986: This gives a (at most countable) set of triples as in the previous

1987: paragraph. 

1988: 

1989: 

1990: Let $\xi\in H^1(M;\mathbb R)$ be a Lyapunov cohomology class for $X$.

1991: Recall that for every $R$ there are only finitely many closed

1992: trajectories $\hat\theta$ with $-\xi([\hat\theta])\leq R$.

1993: Hence, we get a rational function $\zeta_R^{X,\xi}:\Rep(\Gamma;V)\to\mathbb C$

1994: $$

1995: \zeta_R^{X,\xi}:=\prod_{-\xi([\hat\theta])\leq R}\llet_{\underline{\hat\theta}}

1996: $$

1997: where the product is over all triples $\underline{\hat\theta}$

1998: associated to simple closed trajectories with $-\xi([\hat\theta])\leq R$.

1999: It is easy to check that formally we have 

2000: $$

2001: \lim_{R\to\infty}\zeta_R^{X,\xi}=e^{P_X}.

2002: $$

2003: It would be interesting to understand in what sense (if any) this can be made precise.

2004: We conjecture that there exists an open set with non-empty interior in each 

2005: component of $\Rep(\Gamma;V)$ on which we have true convergence. 

2006: In fact there exist vector fields $X$ where the sets of triples are 

2007: finite in which case the conjecture is obviously true.  

2008: %%%%

2009: %The case of vector fields with no rest points when $\zeta_{R,\infty}$ is of particular interest.

2010: %%%

2011: 

2012: 

2013: 

2014: 

2015: 

2016: \begin{thebibliography}{XXXX}

2017: 

2018: 

2019: 

2020: 

2021: \bibitem[APS75]{APS75}

2022:   M.F. Atiyah, V.K. Patodi and I.M. Singer,

2023:   \emph{Spectral asymmetry and Riemannian geometry. II,}

2024:   Math. Proc. Cambridge Philos. Soc. \textbf{78}(1975), 405--432.

2025: 

2026: 

2027: \bibitem[BZ92]{BZ92}

2028:   J.M. Bismut and W. Zhang,

2029:   \emph{An extension of a theorem by Cheeger and M\"uller,}

2030:   Ast\'erisque \textbf{205},

2031:   Soci\'et\'e Math\'ematique de France, 1992.

2032: 

2033: 

2034: \bibitem[BT82]{BT82}

2035:   R. Bott and L.W. Tu,

2036:   \emph{Differential forms in algebraic topology.}

2037:   Graduate texts in Mathematics \textbf{82}.

2038:   Springer Verlag, New York--Berlin, 1982.

2039: 

2040: 

2041: \bibitem[BK05]{BK05}

2042:   M. Braverman and T. Kappeler,

2043:   \emph{Refined analytic torsion,}\\

2044:   preprint {\tt math.DG/0505537.}

2045:   

2046: 

2047: \bibitem[B99]{B99}

2048:   D. Burghelea, 

2049:   \emph{Removing Metric Anomalies from Ray--Singer Torsion,}

2050:   Lett. Math. Phys. \textbf{47}(1999), 149--158.

2051: 

2052: 

2053: \bibitem[BFK01]{BFK01}

2054:   D. Burghelea, L. Friedlander and T. Kappeler,  

2055:   \emph{Relative Torsion,}

2056:   Commun. Contemp. Math. \textbf{3}(2001), 15--85.

2057: 

2058: 

2059: \bibitem[BH03]{BH04}

2060:   D. Burghelea and S. Haller, 

2061:   \emph{Euler structures, the variety of representations and the Milnor--Turaev torsion,}

2062:   preprint {\tt math.DG/0310154.}

2063: 

2064: 

2065: \bibitem[BH04]{BH03}

2066:   D. Burghelea and S. Haller,

2067:   \emph{Laplace transform, dynamics and spectral geometry (version 1),}

2068:   preprint {\tt math.DG/0405037.}

2069:     

2070:  

2071: \bibitem[BH05]{BH03'}

2072:   D. Burghelea and S. Haller,

2073:   \emph{Dynamics, Laplace transform, and spectral geometry (version 2),}

2074:   preprint {\tt math.DG/0508216.}

2075:   

2076: 

2077: \bibitem[BH06]{BH05}

2078:   D. Burghelea and S. Haller, 

2079:   \emph{Complex valued Ray--Singer torsion.}

2080:   Preprint {\tt math.DG/0604484.}

2081: 

2082: 

2083: \bibitem[BH]{BH05b}

2084:   D. Burghelea and S. Haller, 

2085:   \emph{Dynamical torsion, a non-commutative version of a theorem of

2086:         Hutchings--Lee and Pajitnov,} 

2087:   in preparation.

2088: 

2089: 

2090: \bibitem[D81]{D}

2091:   S. Dineen, 

2092:   \emph{Complex Analysis in Locally Convex Spaces.}

2093:   Mathematics Studies \textbf{57}, North-Holland, 1981.

2094: 

2095: 

2096: \bibitem[F95]{F02}

2097:   M. Farber, 

2098:   \emph{Singularities of the analytic torsion,}

2099:   J. Differential Geom. \textbf{41}(1995), 528--572.

2100: 

2101:   

2102: \bibitem[FT00]{FT99}

2103:   M. Farber and V. Turaev, 

2104:   \emph{Poincar\'e--Reidemeister metric, Euler structures and torsion,}

2105:   J. Reine Angew. Math. \textbf{520}(2000), 195--225.

2106: 

2107: 

2108: \bibitem[H77]{Ha}

2109:   R. Hartshorne,

2110:   \emph{Algebraic Geometry.}

2111:   Graduate Texts in Mathematics \textbf{52}.

2112:   Sprin\-ger Verlag, New York--Heidelberg, 1977.

2113: 

2114: 

2115: \bibitem[H02]{H02}

2116:   M. Hutchings, 

2117:   \emph{Reidemeister torsion in generalized Morse Theory,}

2118:   Forum Math. \textbf{14}(2002), 209--244.

2119: 

2120: 

2121: \bibitem[J96]{BJ96}

2122:   B. Jiang,

2123:   \emph{Estimation of the number of periodic orbits,}

2124:   Pac. J. Math. \textbf{172}(1996), 151--185.

2125: 

2126: 

2127: \bibitem[KM97]{KM}

2128:   A. Kriegl and P.W. Michor, 

2129:   \emph{The convenient setting of global analysis.} 

2130:   Mathematical Surveys and Monographs \textbf{53}, 

2131:   American Mathematical Society, Providence, RI, 1997.

2132: 

2133: 

2134: \bibitem[M95]{M98}

2135:   J. Marcsik,

2136:   \emph{Analytic torsion and closed one forms.}

2137:   Thesis, Ohio State University, 1995.

2138: 

2139: 

2140: \bibitem[MQ86]{MQ86}

2141:   V. Mathai and D. Quillen,  

2142:   \emph{Superconnections, Thom Classes, and Equivariant differential forms,}

2143:   Topology \textbf{25}(1986), 85--110.

2144: 

2145: 

2146: \bibitem[M66]{M66}

2147:   J. Milnor,

2148:   \emph{Whitehead torsion,}

2149:   Bull. Amer. Math. Soc. \textbf{72}(1966), 358--426.    

2150: 

2151: 

2152: \bibitem[M]{M}

2153:   W. M\"uller,

2154:   private communication.

2155: 

2156: 

2157: \bibitem[N03]{N03}

2158:   L. Nicolaescu,

2159:   \emph{The Reidemeister Torsion of $3$--manifolds.}

2160:   De Gruyter Studies in Mathematics {\bf 30}, 

2161:   Walter de Gruyter \& Co., Berlin, 2003.

2162: 

2163: 

2164: \bibitem[Q85]{Q85}

2165:   D. Quillen,

2166:   \emph{Determinants of Cauchy--Riemann operators on Riemann surfaces,}

2167:   Functional Anal. Appl. \textbf{19}(1985), 31--34. Translation of

2168:   Funktsional. Anal. i Prilozhen. \textbf{19}(1985), 37--41.

2169: 

2170: 

2171: \bibitem[RS71]{RS71}

2172:   D.B. Ray and I.M. Singer,

2173:   \emph{R--torsion and the Laplacian on Riemannian manifolds,}

2174:   Adv. in Math. \textbf{7}(1971), 145--210.

2175: 

2176: 

2177: \bibitem[S99]{S99}

2178:   N. Steenrod,

2179:   \emph{The topology of fiber bundles.}

2180:   Reprint of the 1957 edition.

2181:   Princeton University Press, Princeton, NJ, 1999.

2182: 

2183: 

2184: \bibitem[T86]{Tu86}

2185:   V. Turaev,

2186:   \emph{Reidemeister torsion in Knot theory,} 

2187:   Uspekhi Mat. Nauk \textbf{41}(1986), 119--182.

2188: 

2189: 

2190: \bibitem[T90]{Tu90}

2191:   V. Turaev,

2192:   \emph{Euler structures, nonsingular vector fields, and Reidemeister-type torsions,}

2193:   Math. USSR--Izv. \textbf{34}(1990), 627--662.

2194: 

2195: 

2196: \bibitem[T01]{Tu01}

2197:   V. Turaev,

2198:   \emph{Introduction to combinatorial torsions.}

2199:   Notes taken by Felix Schlenk.

2200:   Lectures in Mathematics ETH Z\"urich.

2201:   Birkh\"auser Verlag, Basel, 2001.

2202: 

2203: 

2204: \bibitem[T02]{Tu02}

2205:   V. Turaev,

2206:   \emph{Torsion of 3-dimensional manifolds.}

2207:   Progress in Mathematics \textbf{208}. 

2208:   Birkh\"auser Verlag, Bassel--Boston--Berlin, 2002.

2209: 

2210: 

2211: \bibitem[W78]{W78}

2212:   G.W. Whitehead,

2213:   \emph{Elements of homotopy theory.}

2214:   Graduate Texts in Mathematics {\bf 61}.

2215:   Springer Verlag, New York--Berlin, 1978.

2216: 

2217: 

2218: \end{thebibliography}

2219: 

2220: 

2221: 

2222: 

2223: \end{document}

2224: 

2225: 

2226: 

2227: