1: %%%%%%%%%%%%%% Geometry and Topology Monographs: m4-6.tex %%%%%%%%
2: %%%%
3: %%%% Quantum invariants of Seifert $3$--manifolds and their ...
4: %%%%
5: %%%% Soren Kold Hansen, Toshie Takata
6: %%%%
7: %%%% Published in Volume 4(2002) 69-87
8: %%%%
9: %%%% Publication date 19 September 2002
10: %%%%
11: %%%% This is a LaTeX file
12: %%%%
13: %%%%
14: %%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%
15: \documentclass{gtart}
16:
17: \input gtmonout
18: \volumenumber{4}
19: \volumename{Invariants of knots and 3-manifolds (Kyoto 2001)}
20: \volumeyear{2002}
21: \papernumber{6}
22: \pagenumbers{69}{87}
23: \received{3 December 2001}
24: %\revised{}
25: \accepted{22 July 2002}
26: \published{19 September 2002}
27:
28: \usepackage{amsmath,amssymb,texdraw}
29:
30:
31:
32: \newcommand{\Z}{\mathbb{Z}}
33: \newcommand{\Q}{\mathbb{Q}}
34: \newcommand{\R}{\mathbb{R}}
35: \newcommand{\C}{\mathbb{C}}
36: \newcommand{\Io}{\mathbb I}
37:
38:
39: \newcommand{\ep}{\epsilon}
40: \newcommand{\vep}{\varepsilon}
41: \newcommand{\I}{\sqrt{-1}}
42: \newcommand{\npr}{|\triangle_+|}
43: \newcommand{\wl}{\Lambda^W}
44: \newcommand{\rl}{\Lambda^R}
45: \newcommand{\la}{\langle}
46: \newcommand{\ra}{\rangle}
47: \newcommand{\ria}{\rightarrow}
48:
49: \newcommand{\vol}{\operatorname{vol}}
50: \newcommand{\sign}{\operatorname{sign}}
51: \newcommand{\inte}{\operatorname{int}}
52:
53:
54: \newcommand{\id}{\text{\rm id}}
55: \newcommand{\Hom}{\text{\rm Hom}}
56: \newcommand{\ns}{\text{\rm n}}
57: \newcommand{\os}{\text{\rm o}}
58: \newcommand{\s}{\text{\rm s}}
59: \newcommand{\tO}{\text{\rm O}}
60: \newcommand{\tr}{\text{\rm tr}}
61:
62:
63: \newcommand{\frg}{\mathfrak g}
64: \newcommand{\frh}{\mathfrak h}
65: \newcommand{\frhR}{{\mathfrak h}_{\mathbb{R}}}
66: \newcommand{\frsl}{\mathfrak sl}
67:
68:
69: \newcommand{\mA}{\mathcal A}
70: \newcommand{\mC}{\mathcal C}
71: \newcommand{\mD}{\mathcal D}
72: \newcommand{\mR}{\mathcal R}
73: \newcommand{\mT}{\mathcal T}
74: \newcommand{\mV}{\mathcal V}
75:
76:
77: \newtheorem{thm}{Theorem}[section]
78: \newtheorem{conj}[thm]{Conjecture}
79: \newtheorem{cor}[thm]{Corollary}
80: \newtheorem{lem}[thm]{Lemma}
81: \newtheorem{prop}[thm]{Proposition}
82:
83: \theoremstyle{definition}
84: \newtheorem{rem}[thm]{Remark}
85:
86: \newcommand{\refthm}[1]{Theorem~\ref{#1}}
87: \newcommand{\refconj}[1]{Conjecture~\ref{#1}}
88: \newcommand{\refcor}[1]{Corollary~\ref{#1}}
89: \newcommand{\reflem}[1]{Lemma~\ref{#1}}
90: \newcommand{\refprop}[1]{Proposition~\ref{#1}}
91: \newcommand{\refrem}[1]{Remark~\ref{#1}}
92:
93:
94: \let\HS\qed
95:
96:
97: \begin{document}
98:
99: \title{Quantum invariants of Seifert $3$--manifolds\\and their
100: asymptotic expansions}
101:
102: \asciititle{Quantum invariants of Seifert 3-manifolds and their
103: asymptotic expansions}
104:
105: \shorttitle{Quantum invariants of Seifert $3$--manifolds}
106:
107: \author{S\o ren Kold Hansen\\Toshie Takata}
108: \coverauthors{S\noexpand\o ren Kold Hansen\\Toshie Takata}
109: \asciiauthors{Soren Kold Hansen\\Toshie Takata}
110:
111:
112: \address{School of Mathematics, University of Edinburgh,
113: JCMB\\King's Buildings, Edinburgh EH9 3JZ, UK}
114: \email{hansen@maths.ed.ac.uk, takata@math.sc.niigata-u.ac.jp}
115: \secondaddress{Department of Mathematics, Faculty of Science\\Niigata University, Niigata 950-2181, Japan}
116:
117: \asciiaddress{Dept of Maths and Stats, University of Edinburgh,
118: JCMB\\King's Buildings, Edinburgh EH9 3JZ, UK\\Department of
119: Mathematics, Faculty of Science\\Niigata University, Niigata 950-2181,
120: Japan}
121:
122: \begin{abstract}
123: We report on recent results of the authors concerning calculations of
124: quantum invariants of Seifert $3$--manifolds. These results include a
125: derivation of the Reshetikhin--Turaev invariants of all oriented
126: Seifert manifolds associated with an arbitrary complex finite
127: dimensional simple Lie algebra, and a determination of the asymptotic
128: expansions of these invariants for lens spaces. Our results are in
129: agreement with the asymptotic expansion conjecture due to J.\ E.\
130: Andersen \cite{Andersen1}, \cite{Andersen2}.
131: \end{abstract}
132:
133: \asciiabstract{We report on recent results of the authors concerning
134: calculations of quantum invariants of Seifert 3-manifolds. These
135: results include a derivation of the Reshetikhin-Turaev invariants of
136: all oriented Seifert manifolds associated with an arbitrary complex
137: finite dimensional simple Lie algebra, and a determination of the
138: asymptotic expansions of these invariants for lens spaces. Our results
139: are in agreement with the asymptotic expansion conjecture due to JE
140: Andersen [The Witten invariant of finite order mapping tori I, to
141: appear in J. Reine Angew. Math.] and [The asymptotic expansion
142: conjecture, from `Problems on invariants of knots and $3$--manifolds',
143: edited by T. Ohtsuki, http://www.ms.u-tokyo.ac.jp/\char'176tomotada/proj01/].}
144:
145: \primaryclass{57M27}
146:
147: \secondaryclass{17B37, 18D10, 41A60}
148:
149: \keywords{Quantum invariants, Seifert manifolds, modular categories, quantum groups, asymptotic expansions}
150:
151: %\maketitlepage
152: \makeshorttitle
153:
154: \def\undersmile#1{\lower5.7pt\hbox{$\smallsmile$}\kern-0.6em #1}
155:
156:
157: \section{Introduction}\label{sec-Introduction}
158:
159:
160: In 1988 E.\ Witten \cite{Witten} proposed new invariants
161: $Z_{k}^{G}(X,L) \in \C$ of an
162: arbitrary closed oriented $3$--manifold $X$ with
163: an embedded colored link $L$ by
164: quantizing the Chern--Simons field theory associated to a
165: simple and simply connected compact Lie group $G$,
166: $k$ being an arbitrary positive integer called the (quantum) level.
167: The invariant $Z_{k}^{G}(X,L)$ is given by a Feynman path integral
168: over the (infinite dimensional) space of gauge equivalence
169: classes of connections in a $G$ bundle over $X$.
170: This integral should be understood in a formal way since,
171: at the moment of
172: writing, it seems that no mathematically rigorous definition is known,
173: cf.\ \cite[Sect.~20.2.A]{JohnsonLapidus}.
174: The invariants $Z_{k}^{G}$ are called the quantum
175: $G$--invariants or Witten's invariants associated to
176: $G$.
177:
178: Shortly afterwards, N.\ Reshetikhin and V.\ G.\ Turaev
179: \cite{ReshetikhinTuraev} defined in a mathematically rigorous
180: way invariants $\tau_{r}^{\frsl_{2}(\C)}(X,L) \in \C$
181: of the pair $(X,L)$
182: by combinatorial means using irreducible
183: representations of the quantum deformations of $\frsl_{2}(\C)$
184: at certain roots of unity, $r$ being an integer $\geq 2$ associated
185: to the order of the root of unity.
186: Later quantum invariants $\tau_{r}^{\frg}(X,L) \in \C$ associated to
187: other complex simple Lie algebras $\frg$ were constructed using
188: representations of the quantum deformations of $\frg$ at
189: `nice' roots of unity, see \cite{TuraevWenzl}.
190: We call $\tau_{r}^{\frg}$ for the quantum $\frg$--invariants or
191: the RT--invariants associated to $\frg$.
192:
193: Both in Witten's approach and in the
194: approach of Reshetikhin and Turaev the invariants are part of a
195: topological quantum field theory (TQFT) (or more correctly
196: a family of TQFT's). This implies that the
197: invariants are defined for compact oriented $3$--dimensional cobordisms
198: (perhaps with some extra structure on the boundary),
199: and satisfy certain cut-and-paste axioms, see \cite{Atiyah}, \cite{Blanchetetal},
200: \cite{Quinn}, \cite{Turaev}. The TQFT's of Reshetikhin and Turaev can from
201: an algebraic point of view be given a more general formulation by using
202: so-called modular (tensor) categories \cite{Turaev}. The representation theory
203: of the quantum deformations of $\frg$ at certain roots of unity,
204: $\frg$ an arbitrary finite dimensional
205: complex simple Lie algebra, induces such modular categories,
206: see e.g.\ \cite{Kirillov}, \cite{BakalovKirillov}, \cite{Le}.
207:
208: For an invariant to be powerful one should be able to calculate it.
209: A problem with the quantum invariants of knots and $3$--manifolds
210: is that they are rather hard to calculate. In fact people have only been
211: able to calculate these invariants for certain (families of)
212: knots and $3$--manifolds.
213: The lens spaces and more generally the Seifert $3$--manifolds constitute such a
214: family, and there is a wealth of literature about different
215: calculations of quantum invariants
216: of these spaces, see \cite[Introduction]{Hansen1} for some references.
217: In \cite{Hansen1} the RT--invariant associated to an
218: arbitrary modular category is calculated for any Seifert manifold,
219: cf.\ \cite[Theorem 4.1]{Hansen1}.
220: (Here and in the rest of this paper a $3$--manifold means a closed oriented
221: $3$--manifold. In particular, a Seifert manifold is an oriented Seifert manifold.)
222:
223: A solution to the above problem and to the general
224: problem of understanding the
225: topological `meaning' of
226: the quantum invariants could be to
227: determine relationships between the
228: quantum invariants and classical (well understood and calculable) invariants.
229: However, this seems to be a rather hard task.
230: This leads us into one of the themes in this article,
231: namely asymptotic expansions of the invariants.
232: By using stationary phase approximation techniques together with path
233: integral arguments Witten was able \cite{Witten}
234: to express the leading asymptotics
235: of $Z_{k}^{G}(X)$ in the limit $k \ria \infty$
236: as a sum over the set
237: of stationary points for the Chern--Simons functional.
238: The terms in this sum are expressed by such
239: topological/geometric invariants as Chern--Simons invariants, Reidemeister
240: torsions and spectral flows, so here we see a way to extract topological
241: information from the invariants.
242: A full asymptotic expansion of Witten's invariant
243: is expected on the basis of a full perturbative
244: analysis of the Feynman path integral, see \cite{AxelrodSinger1},
245: \cite{AxelrodSinger2}, \cite{Axelrod}.
246: It is generally believed that the family of TQFT's of Reshetikhin and Turaev is
247: a mathematical realization of Witten's family of TQFT's. This belief has together
248: with known results concerning asymptotics of the RT--invariants lead
249: to a conjecture, the asymptotic expansion conjecture (AEC),
250: which specifies in a rather precise way the asymptotic behaviour of the
251: RT--invariants.
252: The AEC was proposed by Andersen in \cite{Andersen1},
253: where he proved it for mapping
254: tori of finite order diffeomorphisms of orientable surfaces of
255: genus at least two
256: using the gauge theoretic definition of the quantum invariants.
257:
258: In this paper we explain recent results of the authors concerning the
259: RT--invariants of Seifert manifolds. Explicitly we state formulas
260: for the invariants $\tau_{r}^{\frg}$ of all Seifert manifolds in terms
261: of the Seifert invariants and standard data for $\frg$, $\frg$ being
262: an arbitrary complex finite dimensional simple Lie algebra,
263: cf.\ \refthm{Lie-Seifert}. Moreover,
264: we analyse more carefully the invariants $\tau_{r}^{\frg}(X)$ for $X$
265: any lens space, thereby determining a formula for
266: the large $r$ asymptotics of these invariants, cf.\ \refthm{asymp-lens}
267: and the remark following this theorem.
268: This formula is in agreement with the AEC.
269:
270: A part of the paper is concerned with studying
271: a certain family of
272: finite dimensional complex representations $\mR=\mR_{r}^{\frg}$
273: of $SL(2,\Z)$. These representations
274: are known from the study of theta functions and modular forms in
275: connection with the study of affine Lie algebras, cf.\ \cite{KacPeterson},
276: \cite[Sect.~13]{Kac}.
277: They also play a fundamental role in
278: conformal field theory and (therefore) in the Chern--Simons TQFT's of Witten,
279: see e.g.\ \cite{GepnerWitten}, \cite{Verlinde}, \cite{Witten}.
280: In case $\frg=\frsl_{2}(\C)$, Jeffrey \cite{Jeffrey1}, \cite{Jeffrey2} has
281: determined a nice formula for $\mR_{r}^{\frg}(U)$ in terms of the entries in
282: $U \in SL(2,\Z)$. \refthm{rep} is a direct
283: extension of Jeffrey's result to arbitrary $\frg$.
284: The representations $\mR_{r}^{\frg}$ are of interest when calculating
285: the RT--invariants of the Seifert manifolds since certain matrices,
286: which can be expressed through $\mR_{r}^{\frg}$, enter into the formulas of the invariants.
287:
288: The paper is organized as follows. In Sect.~\ref{sec-Seifert-manifolds}
289: we introduce notation for the Seifert manifolds and recall surgery
290: presentations for these manifolds. In
291: Sect.~\ref{sec-Quantum-invariants}
292: we explain our calculation of the $\frg$--invariants of the Seifert manifolds.
293: In Sect.~\ref{sec-The-asymptotic}
294: we state the asymptotic
295: expansion conjecture and determine the asymptotic expansions
296: of the $\frg$--invariants of the lens spaces.
297: In the appendix we sketch the proof of the formula for the entries
298: in $\mR_{r}^{\frg}(U)$, $U \in SL(2,\Z)$, \refthm{rep}.
299: The paper is to some extend expository.
300: Details and most technicalities, in particular in connection to
301: the proof of \refthm{rep}, will be given in \cite{HansenTakata}.
302:
303:
304: \rk{Acknowledgements} This work were done while the first author was
305: supported by a Marie Curie Fellowship of the European Commission (CEE
306: $\text{N}^{o}$ HPMF--CT--1999--00231).
307: He acknowledge hospitality of
308: l'Institut de Recherche Math\'{e}matique Avanc\'{e}e (IRMA), Universit\'{e}
309: Louis Pasteur and C.N.R.S., Strasbourg, while being a Marie Curie Fellow.
310: A part of this work was done while the second author visited IRMA.
311: She thanks this department for hospitality during here stay.
312: Another part was done while both authors visited the Research Institute
313: for Mathematical Sciences (RIMS), Kyoto University.
314: We would like to thank RIMS for hospitality during the special month on
315: {\it Invariants of knots and $3$--manifolds}, September 2001.
316: We also thank the organisers of this program for letting
317: us present this work at the workshop of the special month.
318: Finally the first author thanks J.\ E.\ Andersen for helpful
319: conversations about quantum
320: invariants in general and about asymptotics of these invariants in particular.
321:
322:
323: \section{Seifert manifolds}\label{sec-Seifert-manifolds}
324:
325:
326: For Seifert manifolds we will use the notation introduced by Seifert in his
327: classification results for these manifolds, see \cite{Seifert1},
328: \cite{Seifert2}, \cite[Sect.~2]{Hansen1}. That is,
329: $(\ep;g \, | \, b;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n}))$ is
330: the Seifert manifold with orientable base of genus $g \geq 0$ if
331: $\ep=\os$ and non-orientable base of genus $g>0$ if $\ep=\ns$
332: (where the genus of the non-orientable connected sum $\# ^{k} \R \text{P}^{2}$
333: is $k$). (In \cite{Seifert1}, \cite{Seifert2}
334: $(\ep;g \, | \, b;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n}))$ is
335: denoted $(\tO,\ep;g \, | \, b;\alpha_{1},\beta_{1};\ldots;\alpha_{n},\beta_{n})$,
336: but we leave out the $\tO$, since we are only dealing with
337: oriented Seifert manifolds.)
338: The pair $(\alpha_{j},\beta_{j})$ of coprime integers is the (oriented)
339: Seifert invariant of the $j$'th exceptional (or singular) fiber.
340: We have $0< \beta_{j}<\alpha_{j}$. The integer $-b$ is equal to the
341: Euler number of the Seifert fibration $(\ep;g \, | \, b)$
342: (which is a locally trivial $S^{1}$--bundle).
343: More generally, the Seifert Euler number of
344: $(\ep;g \, | \, b;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n}))$
345: is $E=-\left(b+\sum_{j=1}^{n} \beta_{j}/\alpha_{j} \right)$.
346: We note that lens spaces are Seifert manifolds with base $S^{2}$
347: and zero, one or two exceptional fibers.
348: According to \cite[Fig.~12 p.~146]{Montesinos}, the manifold
349: $( \ep ;g \, | \, b;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n}))$
350: has a surgery
351: presentation as shown in Fig.~\ref{fig-A1} if $\ep=\os$ and as
352: shown in Fig.~\ref{fig-A2} if $\ep=\ns$.
353: The $\undersmile{g}$ indicate $g$ repetitions.
354:
355:
356: \begin{figure}[ht!]
357:
358: \begin{center}
359: \begin{texdraw}
360: \drawdim{cm}
361:
362: \setunitscale 0.6
363:
364: \linewd 0.02 \setgray 0
365:
366: \move(0 4)
367:
368: \move(0 0) \lellip rx:3 ry:1.6
369:
370: \linewd 0.2 \setgray 1
371:
372: \move(4 0) \larc r:1.5 sd:218 ed:200
373: \move(4 0) \larc r:3 sd:213 ed:204
374: \move(-4 0) \larc r:1.5 sd:-20 ed:150
375: \move(-4 0) \larc r:2 sd:-23 ed:150
376: \move(-4 0) \larc r:3.5 sd:-20 ed:155
377: \move(-4 0) \larc r:3 sd:-23 ed:155
378:
379: \linewd 0.02 \setgray 0
380:
381: \move(-4 0) \larc r:1.5 sd:-20 ed:150
382: \move(-4 0) \larc r:2 sd:-23 ed:150
383: \move(-4 0) \larc r:3.5 sd:-20 ed:155
384: \move(-4 0) \larc r:3 sd:-23 ed:155
385:
386: \move(-5.88 0.67) \clvec(-5.85 0.9)(-5.6 0.95)(-5.45 0.8)
387: \move(-5.88 -0.67) \clvec(-5.85 -0.9)(-5.6 -0.95)(-5.45 -0.8)
388:
389: \move(-5.3 0.75) \clvec(-5.35 0.55)(-5.6 0.5)(-5.75 0.7)
390: \move(-5.3 -0.75) \clvec(-5.35 -0.55)(-5.6 -0.5)(-5.75 -0.7)
391:
392: \move(-7.3 1.17) \clvec(-7.25 1.4)(-7 1.45)(-6.85 1.3)
393: \move(-7.3 -1.17) \clvec(-7.25 -1.4)(-7 -1.45)(-6.85 -1.3)
394:
395: \move(-6.7 1.29) \clvec(-6.75 1.09)(-7 1.04)(-7.15 1.24)
396: \move(-6.7 -1.29) \clvec(-6.75 -1.09)(-7 -1.04)(-7.15 -1.24)
397:
398: \move(-4 0) \larc r:1.5 sd:210 ed:320
399: \move(-4 0) \larc r:2 sd:210 ed:318
400: \move(-4 0) \larc r:3.5 sd:205 ed:330
401: \move(-4 0) \larc r:3 sd:205 ed:328
402: \move(-4 0) \larc r:1.5 sd:160 ed:200
403: \move(-4 0) \larc r:2 sd:160 ed:200
404: \move(-4 0) \larc r:3.5 sd:160 ed:200
405: \move(-4 0) \larc r:3 sd:160 ed:200
406:
407: \move(4 0) \larc r:1.5 sd:218 ed:200
408: \move(4 0) \larc r:3 sd:213 ed:204
409:
410: \move(-6.8 -0.1) \htext{$\cdots$}
411: \move(1.5 -0.1) \htext{$\cdots$}
412: \move(-6.85 -0.8) \htext{$\undersmile{g}$}
413: \move(-6.65 2.8) \htext{$0$}
414: \move(-5.75 1.6) \htext{$0$}
415: \move(-7.95 0.3) \htext{$0$}
416: \move(-6.45 0.15) \htext{$0$}
417: \move(-3.95 -0.3) \htext{$-b$}
418: \move(5.1 1) \htext{$\frac{\alpha_{1}}{\beta_{1}}$}
419: \move(6.2 2.2) \htext{$\frac{\alpha_{n}}{\beta_{n}}$}
420:
421: \move(0 -4)
422:
423:
424: \end{texdraw}
425: \end{center}
426:
427: \caption{Surgery presentation of $(\os ;g \, | \, b;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n}))$}\label{fig-A1}
428: \end{figure}
429:
430:
431: \begin{figure}[ht!]
432:
433: \begin{center}
434: \begin{texdraw}
435: \drawdim{cm}
436:
437: \setunitscale 0.6
438:
439: \linewd 0.02 \setgray 0
440:
441: \move(0 4)
442:
443: \move(0 0) \lellip rx:3 ry:1.6
444: \move(-7.5 0) \lellip rx:0.5 ry:0.3
445: \move(-5.5 0) \lellip rx:0.5 ry:0.3
446:
447: \linewd 0.2 \setgray 1
448: \move(-4 0) \larc r:1.5 sd:-20 ed:180
449: \move(-4 0) \larc r:3.5 sd:-20 ed:180
450: \move(4 0) \larc r:1.5 sd:215 ed:200
451: \move(4 0) \larc r:3 sd:212 ed:205
452:
453:
454: \linewd 0.02 \setgray 0
455:
456: \move(-4 0) \larc r:1.5 sd:-20 ed:180
457: \move(-4 0) \larc r:3.5 sd:-20 ed:180
458:
459: \move(-4 0) \larc r:1.5 sd:200 ed:320
460: \move(-4 0) \larc r:3.5 sd:188 ed:330
461:
462: \move(4 0) \larc r:1.5 sd:215 ed:200
463: \move(4 0) \larc r:3 sd:212 ed:205
464:
465: \move(-6.8 -0.1) \htext{$\cdots$}
466: \move(1.5 -0.1) \htext{$\cdots$}
467: \move(-6.8 -0.9) \htext{$\undersmile{g}$}
468: \move(-8.3 0.25) \htext{$2$}
469: \move(-6.3 0.25) \htext{$2$}
470: \move(-7.4 2.2) \htext{$\frac{1}{2}$}
471: \move(-5.75 1) \htext{$\frac{1}{2}$}
472: \move(-3.95 -0.3) \htext{$-b$}
473: \move(5.1 1) \htext{$\frac{\alpha_{1}}{\beta_{1}}$}
474: \move(6.2 2.2) \htext{$\frac{\alpha_{n}}{\beta_{n}}$}
475:
476: \move(0 -4)
477:
478: \end{texdraw}
479: \end{center}
480:
481: \caption{Surgery presentation of $(\ns ; g \, | \, b;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n}))$}\label{fig-A2}
482: \end{figure}
483:
484:
485: For completeness we will also state
486: the results in terms of the non-normalized
487: Seifert invariants due to W.\ D.\ Neumann, see \cite{JankinsNeumann}.
488: For a Seifert manifold $X$ with non-normalized Seifert invariants
489: $\{\ep;g;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n})\}$
490: the invariants $\ep$ and $g$ are as above. The
491: $(\alpha_{j},\beta_{j})$ are here pairs of coprime integers with
492: $\alpha_{j} >0$ but not necessarily with $0< \beta_{j}<\alpha_{j}$.
493: These pairs are not invariants of $X$, but can be varied
494: according to certain rules.
495: In fact, $X$ has a surgery presentation as shown in
496: Fig.~\ref{fig-A1} with $b=0$ if $\ep=\os$ and
497: as shown in Fig.~\ref{fig-A2} with $b=0$ if $\ep=\ns$.
498: The Seifert Euler number of $X$ is $-\sum_{j=1}^{n} \beta_{j}/\alpha_{j}$
499: (which is an invariant of the Seifert fibration $X$).
500: For more details, see
501: \cite[Sect.~I.1]{JankinsNeumann}.
502:
503:
504: \section{Quantum invariants of Seifert manifolds}\label{sec-Quantum-invariants}
505:
506:
507: In this section we explain our calculation of
508: the $\frg$--invariants of the Seifert manifolds, $\frg$ being
509: an arbitrary complex finite dimensional simple Lie algebra.
510: Our starting point is a formula
511: for the RT--invariants associated to an arbitrary modular category of
512: the Seifert manifolds, derived in \cite{Hansen1}.
513:
514: \rk{The RT--invariants of the Seifert manifolds for modular categories}
515: Let us first give some preliminary remarks on modular categories. We use
516: notation as in \cite{Turaev}.
517: Let $\left( \mV, \{ V_{i} \}_{i \in I } \right)$
518: be an arbitrary modular category with braiding $c$ and
519: twist $\theta$. The ground ring is $K=\Hom_{\mV}(\Io,\Io)$,
520: where $\Io$ is the unit object.
521: Let $i \mapsto i^{*}$ be the involution in $I$
522: determined by the condition that $V_{i^{*}}$ is isomorphic to
523: the dual of $V_{i}$.
524: An element $i \in I$ is called self-dual if $i=i^{*}$. For such an element we
525: have a $K$--module isomorphism $\Hom_{\mV}(V \otimes V, \Io ) \cong K$, $V=V_{i}$.
526: The map $x \mapsto x(\id_{V} \otimes \theta_{V})c_{V,V}$ is a $K$--module
527: endomorphism of $\Hom_{\mV}(V \otimes V, \Io )$, so is a multiplication by a certain
528: $\vep_{i} \in K$. By the definition of the braiding and twist we have
529: $(\vep_{i})^{2} =1$. In particular $\vep_{i} \in \{ \pm 1 \}$ if $K$ is a field.
530: There is a distinguished element in $I$ denoted $0$, such that $V_{0}=\Io$.
531:
532: The $S$-- and $T$--matrices of $\mV$ are the matrices $S=(S_{ij})_{i,j \in I}$,
533: $T=(T_{ij})_{i,j \in I}$ given by
534: $S_{ij}=\tr(c_{V_{j},V_{i}} \circ c_{V_{i},V_{j}})$ and
535: $T_{ij}=\delta_{ij}v_{i}$, where $\tr$ is the categorical trace of $\mV$,
536: $\delta_{ij}$ is the Kronecher delta equal to $1$ if $i=j$ and zero elsewhere,
537: and $v_{i} \in K$ such that $\theta_{V_{i}}=v_{i}\id_{V_{i}}$.
538:
539: Assume that $\mV$ has a rank $\mD$, i.e.\ an element of $K$ satisfying
540: $$
541: \mD^{2}=\sum_{i \in I} \dim(i)^{2},
542: $$
543: where $\dim(i)=\dim(V_{i})=\tr(\id_{V_{i}})$. We let
544: $$
545: \Delta = \sum_{i \in I} v_{i}^{-1} \dim(i)^{2}.
546: $$
547: Moreover, let $\tau=\tau_{(\mV,\mD)}$ be the RT--invariant
548: associated to $\left( \mV, \{ V_{i} \}_{i \in I },\mD\right)$, cf.\ \cite[Sect.~II.2]{Turaev}.
549: For a tuple of integers $\mC =(m_{1},\ldots,m_{t})$, let
550: $$
551: G^{\mC} = T^{m_{t}}S\cdots T^{m_{1}}S.
552: $$
553: The Rademacher Phi function is
554: defined on $PSL(2,\Z)=SL(2,\Z)/\{\pm 1\}$ by
555: \begin{equation}\label{eq:Phi}
556: \Phi \left( \left[ \begin{array}{cc}
557: p & r \\
558: q & s
559: \end{array}
560: \right] \right) = \left\{ \begin{array}{ll}
561: \frac{p+s}{q} - 12(\sign (q))\s (s,|q|) & ,q \neq 0, \\
562: \frac{r}{s} & ,q=0,
563: \end{array}
564: \right.
565: \end{equation}
566: see \cite{RademacherGrosswald}.
567: Here, for $q \neq 0$, the Dedekind sum $\s (s,q)$ is given by
568: \begin{equation}\label{eq:dedekindsum}
569: \s (s,q)= \frac{1}{4|q|} \sum_{j=1}^{|q|-1} \cot\frac{\pi j}{q} \cot \frac{\pi s j}{q}
570: \end{equation}
571: for $|q|>1$ and $\s (s,\pm 1)=0$.
572: We put $a_{\os}=2$ and $a_{\ns}=1$. Moreover, let
573: $b_{j}^{(\os)}=1$ and
574: $b_{j}^{(\ns)}=\delta_{j,j^{*}}$, $j \in I$.
575: Given pairs $(\alpha_{j},\beta_{j})$ of coprime integers we
576: let $\mC_{j}=(a_{1}^{(j)},\ldots,a_{m_{j}}^{(j)})$ be a continued
577: fraction expansion of $\alpha_{j}/\beta_{j}$,
578: $j=1,2,\ldots,n$, i.e.\
579: $$
580: \frac{\alpha_{j}}{\beta_{j}}=a_{m_{j}}^{(j)}-\frac{1}{a_{m_{j}-1}^{(j)}-\dfrac{1}{\cdots -\dfrac{1}{a_{1}^{(j)}}}}.
581: $$
582:
583:
584: \begin{thm}\label{invariants}{\rm\cite{Hansen1}}\qua
585: Let
586: $M=(\ep;g\;|\;b;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n}))$, $\ep=\os,\ns$.
587: Then
588: \begin{eqnarray*}
589: &&\tau(M) = (\Delta\mD^{-1})^{\sigma_{\ep}} \mD^{a_{\ep}g-2-\sum_{j=1}^{n} m_{j}} \\
590: && \hspace{1.0in} \times \sum_{j \in I} \left(\vep_{j}\right)^{a_{\ep}g} b_{j}^{(\ep)} v_{j}^{-b} \dim(j)^{2-n-a_{\ep}g} \left( \prod_{i=1}^{n} (SG^{\mC_{i}})_{j,0} \right),
591: \end{eqnarray*}
592: where
593: $$
594: \sigma_{\ep}=(a_{\ep}-1)\sign(E) + \sum_{j=1}^{n} \sign(\alpha_{j}\beta_{j}) + \frac{1}{3} \sum_{j=1}^{n} \left( \sum_{k=1}^{m_{j}} a_{k}^{(j)} -\Phi(B^{\mC_{j}}) \right).
595: $$
596: Here $E=-\left( b+\sum_{j=1}^{n} \frac{\beta_{j}}{\alpha_{j}} \right)$ is the Seifert Euler
597: number.
598:
599: The RT--invariant $\tau(M)$ of the Seifert manifold $M$ with
600: non-normalized Seifert invariants
601: $\{ \ep;g;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n})\}$
602: is given by the same expression with the exceptions that the factor $v_{j}^{-b}$
603: has to be removed and $E=-\sum_{j=1}^{n} \frac{\beta_{j}}{\alpha_{j}}$.\HS
604: \end{thm}
605:
606:
607: The theorem is also valid in case $n=0$. In this case one just
608: has to put all sums $\sum_{j=1}^{n}$
609: equal to zero and all products $\prod_{i=1}^{n}$ equal to $1$.
610: Note that $\ep_{j}^{k}=1$ if $k$ is even and $\ep_{j}^{k}=\ep_{j}$
611: if $k$ is odd since $\ep_{j}^{2}=1$. In particular,
612: $\left(\vep_{j}\right)^{a_{\ep}g}=1$ if $\ep=\os$. The sum
613: $\sum_{j=1}^{n} \sign(\alpha_{j}\beta_{j})$ is of course
614: equal to $n$ for normalized Seifert invariants.
615:
616: Let us next consider the lens spaces. For $p,q$ a pair of
617: coprime integers, recall that $L(p,q)$ is given by surgery
618: on $S^{3}$ along the unknot with surgery coefficient $-p/q$.
619: In the following corollary we include the possibilities
620: $L(0,1)=S^{1} \times S^{2}$ and $L(1,q)=S^{3}$, $q \in \Z$.
621:
622:
623: \begin{cor}\label{cor-lens-spaces}{\rm\cite{Hansen1}}\qua
624: Let $p,q$ be a pair of coprime integers. If $q \neq 0$ we let
625: $(a_{1},\ldots,a_{m-1})$ be
626: a continued fraction expansion of $-p/q$.
627: If $q=0$, put $m=3$ and $a_{1}=a_{2}=0$.
628: Then
629: $$
630: \tau(L(p,q))=(\Delta\mD^{-1})^{\sigma} \mD^{-m} G^{\mC}_{0,0},
631: $$
632: where $\mC=(a_{1},\ldots,a_{m-1},0)$ and
633: $\sigma=\frac{1}{3} \left( \sum_{l=1}^{m-1} a_{l} - \Phi(B^{\mC}) \right)$.\HS
634: \end{cor}
635:
636:
637: \rk{The RT--invariants of the Seifert manifolds for the classical Lie algebras}
638: It is a well-known fact that quantum deformations of the classical Lie
639: algebras at roots of unity induce modular categories, see \cite{Kirillov},
640: \cite{BakalovKirillov}, \cite{Le}. Let us provide the details needed.
641: For simplicity we will only consider simply laced Lie algebras in this paper,
642: except in \refrem{non-simply} where we give a few remarks
643: with respect to what have to be adjusted
644: to include the general case. (See also \cite{HansenTakata} for the general case.)
645: Therefore, let in the following
646: $\frg$ be a fixed complex finite dimensional simple and simply laced Lie algebra.
647:
648: First let us fix some notation for $\frg$. Let $\frh$ be a Cartan subalgebra of
649: $\frg$, and let $\alpha_{1},\ldots,\alpha_{l}$ be a set of simple (basis)
650: roots in the dual space of $\frh$.
651: We denote by $\frhR^{*}$ the $\R$--vector space spanned
652: by $\alpha_1,\ldots,\alpha_l$ and let
653: $\la \;,\; \ra$ be the inner product on $\frhR^{*}$ defined by
654: $\la \alpha_{i},\alpha_{j} \ra=a_{ij}$, $\left( a_{ij} \right)_{1 \leq i,j \leq l}$
655: being the Cartan matrix for $\frg$. In particular, all roots have length $\sqrt{2}$.
656: The root lattice $\rl$ is the $\Z$--lattice generated by
657: $\alpha_1,\ldots,\alpha_l$, and the weight lattice $\wl$ is the
658: $\Z$--lattice generated by the fundamental weights
659: $\lambda_{1},\ldots,\lambda_{l}$,
660: i.e.\ $\lambda_{i} \in \frhR^{*}$ such that
661: $\la \lambda_{i},\alpha_{j} \ra=\delta_{ij}$ for all $i,j \in \{1,2,\ldots,l\}$.
662: The (open) fundamental Weyl chamber is the set
663: $$
664: C=\{ x \in \frhR^{*} \; | \; \langle x,\alpha_{i} \rangle > 0, i=1,\ldots,l\}.
665: $$
666: For a positive integer $k$, the $k$--alcove is the (closed) set
667: $$
668: C_{k}=\{ x \in \bar{C} \; | \; \langle x,\alpha_{0} \rangle \leq k\},
669: $$
670: where $\bar{C}$ is the topological closure of $C$ and
671: $\alpha_{0}$ is the highest root of $\frg$, i.e.\ $\alpha_{0}$ is the
672: unique root in $C$. The Weyl group is denoted $W$.
673:
674: Let $q=e^{\pi \I/r}$, where $r$ is an integer $\geq h^{\vee}$.
675: Here $h^{\vee}$ is the dual Coxeter number of $\frg$ (equal to the Coxeter
676: number $h$ of $\frg$, since $\frg$ is simply laced).
677: By $U_{q}(\frg)$ we denote the quantum group associated to these data
678: as defined by Lusztig, see \cite[Part V]{Lusztig}. We follow
679: \cite[Sect.~1.3 and 3.3]{BakalovKirillov} here but will mostly
680: use notation from \cite{Turaev} for modular categories as above.
681: (Note that what we denote
682: $U_{q}(\frg)$ here is denoted $U_{q}(\frg)|_{q=e^{\pi\I/r}}$
683: in \cite{BakalovKirillov}.)
684: Let $\left(\mV_{r}^{\frg}, \{ V_{i} \}_{i \in I} \right)$ be the modular category
685: induced by the representation theory of $U_{q}(\frg)$, cf.\ \cite[Theorem 3.3.20]{BakalovKirillov}.
686: In particular, the index set for the simple objects
687: is $I=\inte(C_{r}) \cap \wl$.
688: We use here the shifted indexes (shifted by $\rho$) (contrary to \cite{BakalovKirillov}).
689: (Normally the irreducible modules
690: of $U_{q}(\frg)$ (of type $1$), $q$ a formal variable, are indexed by
691: the cone of dominant integer weights $\wl_{+}$. Here we
692: denote the irreducible module associated to $\mu \in \wl_{+}$
693: by $V_{\mu + \rho}$.) For $q$ a root of unity as above, $V_{\lambda}$
694: is an irreducible module of $U_{q}(\frg)$ of non-zero dimension if
695: $\lambda \in I$.
696: The involution $I \rightarrow I$, $\lambda \mapsto \lambda^{*}$,
697: is given by
698: $\lambda^{*} = -w_{0}(\lambda-\rho)+\rho$, where $w_{0}$
699: is the longest element in $W$ and $\rho$ is
700: half the sum of positive roots. The distinguished element $0 \in I$
701: is equal to $\rho$. According to \cite[Theorem 3.3.20]{BakalovKirillov}
702: we can use
703: \begin{equation}\label{eq:rank}
704: \mD=r^{l/2} \left| \frac{\vol (\rl) }{\vol (\wl) }\right|^{1/2}
705: \left( \prod_{\alpha \in \Delta_{+}} 2\sin \left( \frac{\pi\la \alpha,\rho \ra }{r} \right) \right)^{-1}
706: \end{equation}
707: as a rank of $\mV_{r}^{\frg}$.
708: Here $\Delta_{+}$
709: is the set of positive roots.
710: According to the same theorem we have
711: \begin{equation}\label{eq:anomaly}
712: \Delta \mD^{-1} = \omega^{-3},
713: \end{equation}
714: where
715: \begin{equation}\label{eq:omega}
716: \omega = e^{\frac{2\pi\I c}{24}} = \exp\left( \frac{\pi\I}{h} |\rho|^{2} \right) \exp\left( -\frac{\pi\I}{r} |\rho|^{2} \right),
717: \end{equation}
718: where $c=\frac{r-h}{r}\dim(\frg)$ is the central
719: charge. The last equality in (\ref{eq:omega}) follows from Freudenthal's strange formula
720: $|\rho|^{2}/h=\dim \frg/12$.
721:
722: The matrices $S$ and $T$ for $\mV_{r}^{\frg}$ are tightly related to a certain
723: unitary representation $\mR=\mR_{r}^{\frg}$ of $SL(2,\Z)$. On the standard
724: generators
725: \begin{equation}\label{eq:generators}
726: \Xi = \left( \begin{array}{cc}
727: 0 & -1 \\
728: 1 & 0
729: \end{array}
730: \right),\hspace{.2in}
731: \Theta = \left( \begin{array}{cc}
732: 1 & 1 \\
733: 0 & 1
734: \end{array}
735: \right)
736: \end{equation}
737: of $SL(2,\Z)$ we have
738: \begin{eqnarray}\label{eq:mR}
739: &&\mR(\Xi)_{\lambda \mu} = \frac{\I^{\npr}}{r^{l/2}}
740: \left| \frac{\vol (\wl) }{\vol (\rl) }\right|^{1/2} \sum_{w \in W} \det(w)
741: \exp \left( -\frac{2\pi\I}{r} \la w(\lambda), \mu\ra \right), \nonumber \\
742: &&\mR(\Theta)_{\lambda \mu} = \delta_{\lambda \mu}
743: \exp \left( \frac{\pi\I}{r} \la \lambda,\lambda \ra
744: -\frac{\pi\I}{h} \la \rho,\rho \ra \right)
745: \end{eqnarray}
746: for $\lambda, \mu \in I$.
747: In the following we also write $\tilde{U}$ for $\mR(U)$.
748: By using the results in \cite[Sect.~3.3]{BakalovKirillov},
749: in particular \cite[Theorem 3.3.20]{BakalovKirillov}, we find
750: \begin{equation}\label{eq:ST}
751: S_{\lambda \mu} = \mD \tilde{\Xi}_{\lambda \mu}, \hspace{.2in}T_{\lambda \mu} = \omega \tilde{\Theta}_{\lambda \mu}
752: \end{equation}
753: for $\lambda,\mu \in I$.
754: Let $\mC=(a_{1},\ldots,a_{n}) \in \Z^{n}$ and let $m \in \{0,1\}$.
755: By (\ref{eq:ST}) we immediately get
756: \begin{equation}\label{eq:sumformula}
757: (S^{m}G^{\mC})_{\lambda \rho} = \mD^{m+n}\omega^{\sum_{j=1}^{n} a_{j}} \left( \tilde{\Xi}^{m} \tilde{\Theta}^{a_{n}}\tilde{\Xi}\tilde{\Theta}^{a_{n-1}}\cdots\tilde{\Theta}^{a_{1}}\tilde{\Xi}\right)_{\lambda \rho}
758: \end{equation}
759: for $\lambda \in I$. Finally we have for any $\lambda \in I$ that
760: $$
761: \dim(\lambda)=S_{\lambda \rho}=\mD \tilde{\Xi}_{\lambda \rho}=\mD r^{-l/2}
762: \left| \frac{\vol (\wl) }{\vol (\rl) }\right|^{1/2}
763: \prod_{\alpha \in \Delta_{+}} 2\sin \left( \frac{\pi\la \alpha,\lambda \ra }{r} \right),
764: $$
765: see also \cite[Formulas (3.3.2) and (3.3.5)]{BakalovKirillov}.
766: All the above data can now be put into the expression in \refthm{invariants}
767: to give a formula for $\tau_{r}^{\frg}(X)$,
768: $X$ an arbitrary Seifert manifold, where $\tau_{r}^{\frg}$ is the RT--invariant associated
769: to $\mV_{r}^{\frg}$. However, the formula to emerge is not detailed enough
770: to be of any use, at least not when it comes to a determination of asymptotics
771: of the invariants. The reason is, that the formula will contain matrix
772: products as in the right-hand side of (\ref{eq:sumformula}).
773:
774: A way out of this problem is to determine nice formulas for the entries of
775: $\mR(U)$ in terms of the entries of $U$. This has in fact been done for
776: $\frg=\frsl_{2}(\C)$ by Jeffrey \cite{Jeffrey1}, \cite{Jeffrey2}.
777: Here method is to
778: write the matrix $U \in SL(2,\Z)$ as a product in the generators
779: $\Xi$ and $\Theta$ and make a certain induction argument. A main ingredient
780: is a reciprocity formula for Gaussian sums. We use
781: a similar argument to extend Jeffrey's results to an arbitrary
782: complex finite dimensional simple Lie algebra. The
783: following theorem generalizes
784: \cite[Propositions 2.7 and 2.8]{Jeffrey2}.
785:
786:
787: \begin{thm}\label{rep}
788: Let $U=\left( \begin{array}{cc}
789: a & b \\
790: c & d
791: \end{array}
792: \right) \in SL(2,\Z)$ with $c \neq 0$.
793: Then there exists an $\ep \in \{ \pm 1 \}$ such that
794: \begin{eqnarray*}
795: &&\mR(\ep U)_{\lambda \mu} = \frac{ \I^{\npr} \sign(\ep c)^{\npr}}{(r|c|)^{l/2} \vol (\rl)}
796: \exp \left( -\frac{\pi \I}{h} |\rho|^{2} \Phi(U) \right)\\
797: && \hspace{1.0in} \times \exp \left( \frac{\pi \I}{r} \frac{d}{c} |\mu|^{2} \right) \sum_{\nu \in \rl/c \rl}
798: \exp \left( \frac{\pi \I} {r} \frac{a}{c} |\lambda+r\nu|^{2} \right) \\
799: && \hspace{1.0in} \times \sum_{w\in W} \det (w)
800: \exp\left( -\frac{2\pi\I}{r\ep c} \la \lambda + r\nu,w(\mu) \ra \right).
801: \end{eqnarray*}
802: \end{thm}
803:
804:
805: The function $\Phi$ is given in (\ref{eq:Phi}).
806: If $c=0$, then $U= \ep \Theta^{b}$ for some $b \in \Z$ and $\ep \in \{ \pm 1 \}$ and
807: the expression for $\mR(\ep U)_{\lambda \mu}$ follows immediately from (\ref{eq:mR}).
808: At first sight the above theorem looks a little strange because of the
809: undetermined sign $\ep$. This sign has to do with the fact
810: that $\mR$ is a representation of $SL(2,\Z)$ and not
811: of $PSL(2,\Z)$ (except for $\frg=\frsl_{2}(\C)$, where $\mR$
812: is in fact a representation of $PSL(2,\Z)$).
813: However, as we shall see now, we will get rid of this sign
814: in the cases we need.
815: In fact, according to \refthm{invariants} and (\ref{eq:sumformula}), we only need the expression for
816: $\mR(U)_{\lambda \mu}$ in case $\lambda$
817: or $\mu$ is equal to $\rho$
818: in the calculation of the invariants of the Seifert manifolds.
819: Since $\rho^{*}=\rho$, $\Xi^{2}=-1$, and
820: $\mR(\Xi^{2})_{\lambda \mu}=\delta_{\lambda \mu^{*}}$,
821: we have
822: $$
823: \mR(-U)_{\lambda \rho} = \mR(U)_{\lambda \rho},\hspace{.2in} \mR(-U)_{\rho \lambda} = \mR(U)_{\rho \lambda}
824: $$
825: for all $\lambda \in I$. By using this fact
826: and the Weyl denominator formula one can show the following corollary
827: to \refthm{rep}.
828: (To show the first formula in \refcor{cor:rho} one also has to use unitarity of
829: $\mR$.)
830:
831:
832: \begin{cor}\label{cor:rho}
833: Let $U=\left( \begin{array}{cc}
834: a & b \\
835: c & d
836: \end{array}
837: \right) \in SL(2,\Z)$ with $c \neq 0$. Then
838: \begin{eqnarray*}
839: &&\mR(U)_{\lambda \rho} = \frac{ \I^{\npr} \sign(c)^{\npr}}{(r|c|)^{l/2} \vol (\rl)}
840: \exp \left( -\frac{\pi \I}{h} |\rho|^{2} \Phi(U) \right) \\
841: && \hspace{0.8in} \times \exp \left( \frac{\pi \I}{r} \frac{a}{c} |\lambda|^{2} \right) \sum_{\nu \in \rl/c \rl}
842: \exp \left( \frac{\pi \I}{r} \frac{d}{c} |\rho+r\nu|^{2} \right) \\
843: && \hspace{0.8in} \times \sum_{w\in W} \det (w)
844: \exp\left( -\frac{2\pi\I}{rc}
845: \la \rho +r\nu,w(\lambda) \ra \right).
846: \end{eqnarray*}
847: If $a \neq 0$ we also have
848: \begin{eqnarray*}
849: &&\mR(U)_{\lambda \rho} = \frac{ \I^{\npr} \sign(c)^{\npr}}{(r|c|)^{l/2} \vol (\rl)} \exp \left( \frac{\pi \I}{r} \frac{b}{a} |\rho|^{2} \right) \\
850: && \hspace{0.8in} \times\exp \left( -\frac{\pi \I}{h} |\rho|^{2} \Phi(U) \right) \\
851: && \hspace{0.8in} \times \sum_{w\in W} \det (w)
852: \sum_{\nu \in \rl/c \rl}
853: \exp\left( \frac{\pi\I}{r} \frac{a}{c} |\lambda +r\nu-\frac {w(\rho)}{a}|^{2} \right).
854: \end{eqnarray*}\HS
855: \end{cor}
856:
857:
858: Because of the length and technical
859: nature of the proof of \refthm{rep}, we defer the argument to the Appendix,
860: and will only give the main ideas there. Detailed arguments will appear in \cite{HansenTakata}.
861:
862: Given a pair of coprime integers $(\alpha,\beta)$,
863: $\alpha>0$, we let $\beta^{*}$ be the inverse of $\beta$ in the multiplicative
864: group of units in $\Z/\alpha\Z$. The following theorem is a generalization
865: of \cite[Theorem 8.4]{Hansen1} (which concerns the case $\frg=\frsl_{2}(\C)$).
866: The proof follows closely the proof of \cite[Teorem 8.4]{Hansen1} and is
867: therefore left out here. (One has to use the first formula in \refcor{cor:rho}.)
868:
869:
870: \begin{thm}\label{Lie-Seifert}
871: Let $M=(\ep;g\;|\;b;(\alpha_{1},\beta_{1}),\ldots,(\alpha_{n},\beta_{n}))$,
872: $\ep \in \{ \os, \ns\}$. Then
873: \begin{eqnarray*}
874: \tau_{r}^{\frg}(M) &=& \exp \left( \frac{\pi\I}{r}|\rho|^{2} \left[ 3(a_{\ep}-1)\sign(E) -E - 12\sum_{j=1}^{n} \s (\beta_{j},\alpha_{j}) \right] \right) \\
875: &&\hspace{.1in} \times \frac{\I^{n\npr} r^{l(a_{\ep}g/2-1)}}{2^{\npr(n+a_{\ep}g-2)}\vol(\rl)^{2-a_{\ep}g}}
876: \frac{1}{\mA^{l/2}} e^{\frac{3\pi\I}{h}|\rho|^{2}(1-a_{\ep})\sign(E)} Z_{\ep}^{\frg}(M;r),
877: \end{eqnarray*}
878: where $\s (\beta_{j},\alpha_{j})$ is given by {\em (\ref{eq:dedekindsum})}, $\mA=\prod_{j=1}^{n} \alpha_{j}$, and
879: \begin{eqnarray*}
880: Z_{\ep}^{\frg}(M;r) &=& \sum_{\lambda \in I} b_{\lambda}^{(\ep)} \vep_{\lambda}^{a_{\ep}g}
881: \left( \prod_{\alpha \in \Delta_{+}}
882: \sin ^{2-n-a_{\ep}g} \left( \frac{\pi\la \lambda,\alpha \ra}{r} \right)\right)
883: \exp \left( \frac{\pi\I}{r} E |\lambda|^{2} \right) \\
884: & & \hspace{.2in} \times \sum_{w_{1},\ldots,w_{n}\in W}
885: \sum_{\nu_{1} \in \rl/\alpha_{1} \rl} \ldots \sum_{\nu_{n} \in \rl/\alpha_{n} \rl}
886: \left(\prod_{j=1}^{n} \det(w_{j})\right) \\
887: & & \hspace{0.6in} \times
888: \exp \left( -\pi \I \sum_{j=1}^{n} \frac{\beta_{j}^{*}}{\alpha_{j}} \left( r|\nu_{j}|^{2}
889: + 2\la w_{j}(\rho),\nu_{j} \ra \right) \right) \\
890: & & \hspace{0.6in} \times
891: \exp \left( - \frac{2 \pi\I}{r} \la \lambda, \sum_{j=1}^{n} \frac{ r\nu_{j} + w_{j}(\rho)}{\alpha_{j}} \ra \right).
892: \end{eqnarray*}
893: The RT--invariant $\tau_{r}^{\frg}(M)$ of the Seifert manifold $M$
894: with non-normalized Seifert invariants
895: $\{\ep;g;(\alpha_{1},\beta_{1}),\ldots,$ $(\alpha_{n},\beta_{n})\}$
896: is given by the same expression.\HS
897: \end{thm}
898:
899:
900: The theorem is also valid in case $n=0$. In this case one
901: just has to put the sum
902: $\sum_{w_{1},\ldots,w_{n}\in W}
903: \sum_{\nu_{1} \in \rl/\alpha_{1} \rl} \ldots \sum_{\nu_{n} \in \rl/\alpha_{n} \rl}$
904: in $Z_{\ep}(M;r)$ equal to $1$,$\ep=\os,\ns$,
905: and put $\mA=1$ and
906: $\sum_{j=1}^{n}\s (\beta_{j},\alpha_{j})=0$.
907:
908: Let us finally consider the lens space $L(p,q)$.
909: Let $b,d$ be any integers such that
910: $U=\left( \begin{array}{cc}
911: q & b \\
912: p & d
913: \end{array}
914: \right) \in SL(2,\Z)$. Assume $q \neq 0$, let $V=-\Xi U =\left( \begin{array}{cc}
915: p & d \\
916: -q & -b
917: \end{array}
918: \right)$, and let $C'=(a_{1},a_{2},\ldots,a_{m-1}) \in \Z^{m-1}$ such that
919: $B^{\mC'}=V$. Then $\mC'$ is a continued fraction expansion of $-p/q$ and
920: $U=\Xi V=B^{\mC}$, where $\mC=(a_{1},a_{2},\ldots,a_{m-1},0)$.
921: By \refcor{cor-lens-spaces}, (\ref{eq:anomaly}) and (\ref{eq:sumformula})
922: we therefore get
923: \begin{equation}\label{eq:RTlens}
924: \tau_{r}^{\frg}(L(p,q))= \omega ^{\Phi(U)} \tilde{U}_{\rho\rho},
925: \end{equation}
926: where $\omega$ is given by (\ref{eq:omega}).
927: If $q=0$ we have $p=1$ and $L(p,q)=S^{3}$. In this case we have
928: $\tau_{r}^{\frg}(L(p,q))=\mD^{-1}$. We also have
929: $U=\Xi \Theta^{d}$, so by using
930: (\ref{eq:mR}), (\ref{eq:rank}) and (\ref{eq:omega})
931: we find that the right-hand side of (\ref{eq:RTlens})
932: is also equal to $\mD^{-1}$.
933: The identity (\ref{eq:RTlens}) coincides with \cite[Formula (3.7)]{Jeffrey2}
934: for $\frg=\frsl_{2}(\C)$, see also \cite[Formula (49)]{Hansen1}.
935:
936:
937: \begin{rem}\label{non-simply}
938: Let us briefly mention the adjustments to be done for including the non-simply
939: laced Lie algebras. In the general case the root of unity $q=e^{\pi \I/(dr)}$,
940: where $d=1$ for $\frg$ simply laced, $d=2$ if $\frg$ belongs to the series
941: $BCF$ and $d=3$ if $\frg$ is of type $G_{2}$. Moreover, $\rl$ is in general
942: the coroot lattice, which is dual to the weight lattice.
943: The inner product $\la \; , \; \ra$ in $\frhR^{*}$ is induced by an invariant
944: bilinear form on $\frg$, and is normalized such that
945: a long root has length $\sqrt{2}$. We stress that $\alpha_{0}$ is
946: the long highest root of $\frg$.
947: \end{rem}
948:
949:
950: \section{The asymptotic expansion conjecture and Seifert manifolds}\label{sec-The-asymptotic}
951:
952:
953: \noindent For $X$ a fixed closed oriented $3$--manifold
954: we consider $r \mapsto \tau_{r}^{\frg}(X)$
955: as a complex valued function on $\{h^{\vee},h^{\vee}+1,h^{\vee}+2,\ldots\}$. We
956: are interested in the behaviour of this function in the limit of large $r$, i.e.
957: $r \rightarrow \infty$.
958:
959: It is believed that Witten's TQFT's associated to $G$
960: coincides with the TQFT's of Reshetikhin--Turaev associated to $\frg$, where
961: $G$ is a simply connected compact simple Lie group with complexified Lie
962: algebra $\frg$.
963: In particular it is conjectured that Witten's leading large $k$ asymptotics for
964: $Z_{k}^{G}(X)$
965: should be valid for the function $r \mapsto \tau_{r}^{\frg}(X)$ in the
966: limit $r \ria \infty$, and furthermore, that this function should have a full
967: asymptotic expansion.
968: The precise formulation of this is stated in the following conjecture,
969: called the asymptotic expansion conjecture (AEC).
970:
971:
972: \begin{conj}[J.\ E.\ Andersen \cite{Andersen1}, \cite{Andersen2}]\label{AEC}
973: Let $\{\alpha_{1},\ldots,\alpha_{M} \}$
974: be the set
975: of values of the Chern--Simons functional of flat $G$ connections on
976: a closed oriented $3$--manifold $X$. Then there exist $d_{j} \in \Q$,
977: $\tilde{I}_{j} \in \Q / \Z$, $b_{j} \in \R_{+}$ and $c_{m}^{j} \in \C$
978: for $j=1,\ldots,M$ and $m=1,2,3,\ldots$ such that
979: \begin{equation}\label{eq:AEC}
980: \tau_{r}^{\frg}(X) \sim_{r \ria \infty} \sum_{j=1}^{M} b_{j}e^{2\pi \sqrt{-1} r \alpha_{j}} r^{d_{j}} e^{\pi \sqrt{-1} \tilde{I}_{j}/4} \left(1+\sum_{m=1}^{\infty} c_{m}^{j} r^{-m} \right),
981: \end{equation}
982: that is, for all $N=0,1,2,\ldots$
983: $$
984: \tau_{r}^{\frg}(X) = \sum_{j=1}^{M} b_{j}e^{2\pi \sqrt{-1} r \alpha_{j}} r^{d_{j}} e^{\pi \sqrt{-1} \tilde{I}_{j}/4} \left(1+\sum_{m=1}^{N} c_{m}^{j} r^{-m} \right) + o(r^{d-N})
985: $$
986: in the limit $r \ria \infty$, where $d=\max\{d_{1},\ldots,d_{M} \}$.
987: \end{conj}
988:
989:
990: As noticed by Andersen \cite{Andersen1}, \cite{Andersen2}, a
991: complex function defined on the positive integers has
992: at most one asymptotic expansion on the form (\ref{eq:AEC})
993: if the $\alpha_{j}$'s are rational (and mutually different),
994: see also \cite{Hansen2}.
995: This means that if the AEC is true and if we futhermore
996: have that the Chern--Simons invariants are rational
997: (as conjectured by e.g.\ Auckly \cite{Auckly}), then all the quantities
998: $c_{m}^{j}$, $\alpha_{j}$, $d_{j}$,... are
999: topological invariants.
1000: The AEC was first proved for the mapping tori of finite order
1001: diffeomorphisms of orientable surfaces of genus at least $2$ and for any $\frg$
1002: by Andersen \cite{Andersen1} using the gauge theory definition of the invariants.
1003: Note that these mapping tori are Seifert manifolds with orientable base
1004: and Seifert Euler number equal to zero.
1005: Later on, the AEC was proved for all Seifert manifolds with
1006: orientable base or non-orientable base with even genus
1007: in case $\frg=\frsl_{2}(\C)$, cf.\ \cite{Hansen2}.
1008: The proof of this result is partly based on calculations of
1009: Rozansky \cite{Rozansky}.
1010: For more details about the AEC and conjectures about the topological
1011: interpretation of the different parts of the asymptotic formula (\ref{eq:AEC})
1012: we refer to \cite{Andersen1}, \cite{Andersen2},
1013: \cite{Hansen2}. One can also find a review about the status of the AEC in
1014: these references.
1015:
1016: By elaborating on the expression (\ref{eq:RTlens})
1017: along the same lines as in \cite[Sect.~3]{Jeffrey2}
1018: (using the last formula in \refcor{cor:rho}) we find
1019: the following generalization of \cite[Theorem 3.4]{Jeffrey2}
1020: (valid in case $p \neq 0$).
1021:
1022:
1023: \begin{thm}\label{asymp-lens}
1024: The RT--invariant associated to $\frg$ of the lens space $L(p,q)$
1025: is given by
1026: \begin{gather}
1027: \tau_{r}^{\frg}(L(p,q)) = \frac{\sign(p)^{\npr}\I^{\npr}}{(r|p|)^{l/2} \vol(\rl)}
1028: \exp \left( \frac{\pi \I}{r} 12 \sign(p)\s (q,|p|) |\rho|^{2} \right)\notag \\
1029: \hspace{1.0in} \times \sum_{w \in W} \det(w)
1030: \exp \left( -\frac{2\pi \I}{pr} \la \rho, w(\rho) \ra \right)\notag \\
1031: \hspace{0.6in} \times \sum_{\nu \in \rl/p\rl}
1032: \exp \left( \pi \I \frac{q}{p} r |\nu|^{2} \right)
1033: \exp \left( 2\pi \I \frac{1}{p} \la \nu, q\rho - w(\rho) \ra \right).\tag*{\qed}
1034: \end{gather}
1035: \end{thm}
1036:
1037:
1038: From this theorem
1039: it is obvious, that the large $r$ asymptotics
1040: of $\tau_{r}^{\frg}(L(p,q))$ is on the same form as in (\ref{eq:AEC})
1041: (expand the factor
1042: $\exp \left( \frac{\pi \I}{r} 12 \sign(p)\s (q,|p|) |\rho|^{2} \right)$ $\exp \left( -\frac{2\pi \I}{pr} \la \rho, w(\rho) \ra \right)$
1043: as a power series in $r^{-1}$).
1044: Proving the following conjecture will therefore finalize the proof of the AEC
1045: for the invariants $\tau_{r}^{\frg}(L(p,q))$.
1046:
1047:
1048: \begin{conj}
1049: The set of values of the Chern--Simons functional of flat $G$ connections on
1050: $L(p,q)$ is given by
1051: $$
1052: \left\{ \left. \; \frac{q}{2p} |\nu|^{2} \pmod{\Z} \;\; \right| \;\; \nu \in \rl/p\rl \; \right\}.
1053: $$
1054: \end{conj}
1055:
1056:
1057: For $\frg = \frsl_{n}(\C)$ ($G=SU(n)$) this conjecture should follow from results in
1058: \cite{Nishi}.
1059:
1060:
1061: \section{Appendix. The proof of \refthm{rep}}\label{sec-Appendix}
1062:
1063:
1064: In this section we will explain the ideas behind the
1065: proof of \refthm{rep}. The underlying Lie algebra $\frg$ is still assumed
1066: to be simply laced for simplicity. However, with some minor adjustments
1067: the arguments given are also
1068: true for the non-simply laced Lie algebras, see
1069: \refrem{non-simply}.
1070:
1071: The proof of \refthm{rep} builds mainly on the
1072: key-lemma, \reflem{lem:main}. Let us introduce some notation.
1073: For a tuple of integers $\mC=(m_{1},\ldots,m_{t})$, let
1074: \begin{equation}\label{eq:Bmatrix}
1075: B_{k}^{\mC} = \left( \begin{array}{cc}
1076: a_{k}^{\mC} & b_{k}^{\mC} \\
1077: c_{k}^{\mC} & d_{k}^{\mC}
1078: \end{array}
1079: \right) = \Theta^{m_{k}}\Xi\Theta^{m_{k-1}}\Xi \ldots \Theta^{m_{1}}\Xi
1080: \end{equation}
1081: for $k=1,2,\ldots,t$,
1082: and let $B^{\mC}=B_{t}^{\mC}$, where $\Xi$ and $\Theta$ are given by
1083: (\ref{eq:generators}). Moreover, we put
1084: $$
1085: a_{0}^{\mC}=d_{0}^{\mC}=1, \qquad b_{0}^{\mC}=c_{0}^{\mC}=0.
1086: $$
1087: We say that $\mC$ has length $|\mC|=t$.
1088: If it is clear from the context what $\mC$ is
1089: we write $a_{k}$ for $a_{k}^{\mC}$ etc.
1090: From \cite[Proposition 2.5]{Jeffrey2}, the elements $a_{i},b_{i},c_{i},d_{i}$
1091: satisfy the recurrence relations
1092: \begin{align}\label{eq:abcd}
1093: a_{k}=m_{k} a_{k-1}-c_{k-1},& \quad c_{k}=a_{k-1}, \\
1094: b_{k}=m_{k} b_{k-1}-d_{k-1},& \quad d_{k}=b_{k-1} \nonumber
1095: \end{align}
1096: for $k=1,2,\ldots,t$.
1097: One should note that the expressions (\ref{eq:mR}) for the entries of $\mR(\Xi)$
1098: and $\mR(\Theta)$ are well-defined for all $\lambda, \mu \in \wl$.
1099: Note also that if $\lambda$ or $\mu$ are elements of $\wl$
1100: belonging to the boundary of $C_{r}$ then $\mR(\Xi)_{\lambda \mu}=0$.
1101: This observation allows us to shift between $I=\inte(C_{r}) \cap \wl$
1102: and $C_{r} \cap \wl$ as summation index set in formulas below.
1103: This shift is important in the proof of \reflem{lem:main}.
1104: Following Jeffrey \cite[Sect.~2]{Jeffrey1}, \cite[Sect.~2]{Jeffrey2} we consider
1105: $$
1106: \mT^{\mC}_{\lambda_{0},\lambda_{t+1}} =
1107: \sum_{\lambda_{1},\ldots,\lambda_{t} \in C_{r} \cap \wl}
1108: \tilde{\Xi}_{\lambda_{t+1} \lambda_{t}}
1109: \tilde{\Theta}_{\lambda_t}^{m_t} \tilde{\Xi}_{\lambda_{t} \lambda_{t-1}}
1110: \tilde{\Theta}_{\lambda_{t-1}}^{m_{t-1}} \tilde{\Xi}_{\lambda_{t-1} \lambda_{t-2}}
1111: \cdots
1112: \tilde{\Theta}_{\lambda_{1}}^{m_{1}} \tilde{\Xi}_{\lambda_{1} \lambda_{0}}
1113: $$
1114: for $\lambda_{0},\lambda_{t+1} \in C_{r} \cap \wl$, where we write
1115: $\tilde{\Theta}_{\lambda}$ for $\tilde{\Theta}_{\lambda \lambda}$.
1116: Then we have the following generalization of \cite[Lemma 2.6]{Jeffrey2}:
1117:
1118:
1119: \begin{lem}\label{lem:main}
1120: Assume that $\mC=(m_{1},\ldots,m_{t})$ is a sequence of integers such that
1121: $a_{k}$ is nonzero for $k=1,\ldots,t$. Then
1122: $$
1123: \mT^{\mC}_{\lambda_{0},\lambda_{t+1}} = K^{\mC}_{\lambda_{0}} \sum_{w\in W} \det(w)
1124: \sum_{\mu \in \rl/a_{t} \rl}
1125: \exp\left( -\frac {\pi\I c_{t}}{a_{t} r}
1126: \left|\lambda_{t+1}+r\mu+\frac {w(\lambda_{0})}{c_{t}}\right|^{2} \right),
1127: $$
1128: where
1129: \begin{eqnarray*}
1130: K^{\mC}_{\lambda_{0}} &=& \frac{{\I}^{(t+1) \npr}}{(r|a_{t}|)^{l/2}\vol (\rl)}\; \zeta^{l \; D_{t}}
1131: \exp \left( -\frac{\pi\I}{h} (\sum_{i=1}^{t} m_{i}) |\rho |^{2} \right) \\
1132: & & \hspace{.3in} \times \exp \left( -\frac{\pi\I}{r}
1133: (\sum_{i=1}^{t-1} \frac{1}{a_{i-1}a_{i}}) |\lambda_{0} |^{2} \right).
1134: \end{eqnarray*}
1135: Here $\zeta=\exp \frac{\pi \I}{4}$
1136: and $D_{t}=\sign(a_{0}a_{1})+\cdots +\sign(a_{t-1}a_{t})$.\qed
1137: \end{lem}
1138:
1139:
1140: We will not give the proof of this lemma here, since it is long
1141: and technical.
1142: The lemma is proved by induction on the length of $\mC$. The
1143: reciprocity formula for Gaussian sums, \refprop{prop:gauss}, plays a
1144: prominent role in the proof. A proof of this reciprocity formula can be
1145: found in \cite[Sect.~2]{Jeffrey1}.
1146:
1147: Let $V$ be a real vector space of dimension $l$ with inner product
1148: $\langle \;, \;\rangle$, $\Lambda$ a lattice in $V$ and $\Lambda^{*}$
1149: the dual lattice.
1150: For an integer $r$, a self-adjoint automorphism
1151: $B \co V \to V$, and an element $\psi \in V$,
1152: we assume
1153: \begin{eqnarray*}
1154: &&\frac{1}{2} \la \lambda,B r \lambda \ra, \;\;
1155: \la \lambda,B \eta \ra, \;\;
1156: r \la \lambda, \psi \ra \in \Z, \hspace{.2in}\forall \lambda, \eta \in \Lambda,\\
1157: &&\frac{1}{2} \la \mu,B r \mu \ra, \;\;
1158: \la \mu,r \xi \ra, \;\;
1159: r \la \mu, \psi \ra \in \Z, \hspace{.2in}\forall \mu, \xi \in \Lambda^*\\
1160: \end{eqnarray*}
1161: and $B\Lambda^{*} \subseteq \Lambda^{*}$. Then we have the following:
1162:
1163: \begin{prop}[Reciprocity formula for Gauss sums]\label{prop:gauss}
1164: \begin{gather}
1165: \kern -1.4in \vol (\Lambda^{*}) \sum_{\lambda \in \Lambda / r\Lambda}
1166: \exp \left( \frac{\pi\I}{r} \langle \lambda,B \lambda\rangle\right)
1167: \exp \left(2\pi\I \la \lambda,\psi \ra \right)\notag \\
1168: \hspace{.2in} =\left( \det \frac B \I \right)^{-1/2} r^{l/2}
1169: \sum_{\mu \in \Lambda^{*} / B\Lambda^{*}}
1170: \exp \left( -\pi r\I \la \mu+\psi, B^{-1}(\mu+\psi) \ra \right).\tag*{\qed}
1171: \end{gather}
1172: \end{prop}
1173:
1174:
1175: In the proof of \reflem{lem:main} we use Proposition \ref{prop:gauss}
1176: with $\Lambda =\wl$, the dual lattice being the
1177: root lattice $\rl$. Basically we use the reciprocity formula
1178: ($t$ times, recursively)
1179: to change the sums in the expression for $\mT^{\mC}$
1180: to a sum with a range which does not depend on
1181: $r$. Another main ingredient in the proof of \reflem{lem:main}
1182: is symmetry considerations along the same lines as the discussion in
1183: \cite[pp.~584--586]{Jeffrey2}, see in particular
1184: \cite[Proposition 4.4]{Jeffrey2}.
1185: In the proof of \reflem{lem:main} and \refthm{rep} we
1186: use several results in
1187: \cite[Sect.~2]{Jeffrey2}, in particularly \cite[Proposition 2.5]{Jeffrey2}.
1188:
1189: \reflem{lem:main} nearly proves \refthm{rep}. There is, however, a small
1190: hurdle to overcome because of the assumption on the $a_{k}$'s in the lemma.
1191: The following small result does the job.
1192:
1193:
1194: \begin{lem}\label{lem:matrixdecomposition}
1195: Let $U=\begin{bmatrix}
1196: a & b \\
1197: c & d
1198: \end{bmatrix} \in PSL(2,\Z)$ with $c \neq 0$. Then we can write
1199: $U=V\Theta^{n}$,
1200: where $n \in \Z$ and where $V$ is given in the following way:
1201: If $a=0$, then $V =\Xi$; if $a\neq 0$, then
1202: there exists a sequence of integers $\mC$ such that
1203: $V=B^{\mC}$, see {\em (\ref{eq:Bmatrix})}, and such that $a^{\mC}_{k} \neq 0$,
1204: $k=1,2,\ldots,|\mC|$.\HS
1205: \end{lem}
1206:
1207:
1208: From this lemma we see the origin of the undetermined sign $\ep$.
1209: Let us use the above lemmas to sketch the proof of \refthm{rep}.
1210:
1211:
1212: \begin{proof}[Proof of \refthm{rep}]
1213: According to the previous lemma there exists an integer $n$, a sign $\ep \in \{ \pm 1 \}$,
1214: and a $V \in SL(2,\Z)$ as in the \reflem{lem:matrixdecomposition}
1215: such that $U=\ep V\Theta^{n}$.
1216: First assume $a \neq 0$ and that $n=0$, i.e.\ assume that $\ep U=B^{\mC}$, where $\mC=(m_{1},\ldots,m_{t})$
1217: and $a_{k} \neq 0$, $k=1,2,\ldots,t$. Let $\mC'=(m_{1},\ldots,m_{t-1})$.
1218: Then, by \reflem{lem:main}, (\ref{eq:abcd}), and a small calculation, we get
1219: \begin{eqnarray*}
1220: &&\mR(\ep U)_{\lambda \mu} = \tilde{\Theta}_{\lambda \lambda}^{m_t} \mT^{\mC'}_{\mu,\lambda} \\
1221: && \hspace{.5in} = \sum_{w\in W} \det (w)
1222: \sum_{\nu \in \rl/c \rl}
1223: \exp\left( \frac{\pi\I}{r} \frac{a}{c}
1224: \left|\lambda+r\nu-\frac {w(\mu)}{\ep a}\right|^{2} \right),
1225: \end{eqnarray*}
1226: where
1227: $$
1228: K = K^{\mC'}_{\mu} \exp \left( -\frac{\pi \I}{h} m_{t} |\rho|^{2} \right) \exp \left( -\frac{\pi \I}{a_{t-1}a_{t-2}r} |\mu|^{2} \right) \exp \left( -\frac{\pi \I}{a_{t}a_{t-1}r} |\mu|^{2} \right).
1229: $$
1230: The formula for the entries of $\mR(\ep U)$
1231: now follows by the fact $b/a+1/(ac)=d/c$
1232: (except for the factor $K$ which we will not rewrite here).
1233: Next assume that $\ep U = B^{\mC}\Theta^{n}$ with $n \neq 0$,
1234: where $\mC$ is as above. Then
1235: $$
1236: B^{\mC} = \ep U \Theta^{-n} = \ep \left( \begin{array}{cc}
1237: a & b \\
1238: c & d
1239: \end{array}
1240: \right)\left( \begin{array}{cc}
1241: 1 & -n \\
1242: 0 & 1
1243: \end{array}
1244: \right)=\ep \left( \begin{array}{cc}
1245: a & -na+b \\
1246: c & -nc+d
1247: \end{array}
1248: \right),
1249: $$
1250: and since \refthm{rep} is valid for $U=B^{\mC}$ (with $\ep =1$) we get
1251: the result after a small calculation.
1252: Finally one has to consider the case where $a=0$, in which case
1253: $\ep U=\Xi \Theta^{n}=
1254: \left( \begin{array}{cc}
1255: 0 & -1 \\
1256: 1 & n
1257: \end{array}
1258: \right)$. Here the result follows by inserting the expressions
1259: in (\ref{eq:mR}) into
1260: $\mR(\ep U)_{\lambda \mu}=\mR(\Xi)_{\lambda \mu}\mR(\Theta)^{n}_{\mu \mu}$.
1261: \end{proof}
1262:
1263:
1264: \begin{thebibliography}
1265:
1266: \itemsep2.5pt plus3pt minus2pt
1267:
1268: \bibitem{Andersen1} {\bf J\,E Andersen}, {\it The Witten invariant of finite order mapping tori I}, to appear in J. Reine Angew. Math.
1269:
1270: \bibitem{Andersen2} {\bf J\,E Andersen}, {\it The asymptotic expansion conjecture}, in {\it Problems on invariants of knots and $3$--manifolds}, edited by T. Ohtsuki, draft version available at {\tt http://www.ms.u-tokyo.ac.jp/\char'176tomotada/proj01/}
1271:
1272: \bibitem{Atiyah} {\bf M\,F Atiyah}, {\it Topological quantum field theories}, Inst. Hautes \'{E}tudes Sci. Publ. Math. 68 (1989) 175--186
1273:
1274: \bibitem{Auckly} {\bf D Auckly}, {\it Topological methods to compute Chern--Simons invariants}, Math. Proc. Camb. Phil. Soc. 115 (1994) 229--251
1275:
1276: \bibitem{Axelrod} {\bf S Axelrod}, {\it Overview and warmup example for perturbation theory with instantons}, Geometry and physics (Aarhus, 1995), Lecture Notes in Pure and Appl. Math. 184, Dekker, New York (1997) 321--338
1277:
1278: \bibitem{AxelrodSinger1} {\bf S Axelrod}, {\bf I\,M Singer}, {\it Chern--Simons perturbation theory}, Proceedings of the XXth International Conference on Differential Geometric Methods in Theoretical Physics, Vol. 1, 2, World Sci. Publishing (1992) 3--45
1279:
1280: \bibitem{AxelrodSinger2} {\bf S Axelrod}, {\bf I\,M Singer}, {\it Chern--Simons perturbation theory. II}, J. Differential Geom. 39 (1994) 173--213
1281:
1282: \bibitem{BakalovKirillov} {\bf B Bakalov}, {\bf A Kirillov Jr}, {\it Lectures on tensor categories and modular functors}, University Lecture Ser. 21, Amer. Math. Soc. (2001)
1283:
1284: \bibitem{Blanchetetal} {\bf C Blanchet}, {\bf N Habegger}, {\bf G Masbaum}, {\bf P Vogel}, {\it Topological quantum field theories derived from the Kauffman bracket}, Topology 34 (1995) 883--927
1285:
1286: \bibitem{GepnerWitten} {\bf D Gepner}, {\bf E Witten}, {\it String theory on group manifolds}, Nuclear Phys. B 278 (1986) 493--549
1287:
1288: \bibitem{Hansen1} {\bf S\,K Hansen}, {\it Reshetikhin--Turaev invariants of Seifert $3$--manifolds and a rational surgery formula}, Algebr. Geom. Topol. 1 (2001) 627--686
1289:
1290: \bibitem{Hansen2} {\bf S\,K Hansen}, {\it Analytic asymptotic expansions of the Reshetikhin--Turaev invariants of Seifert $3$--manifolds for $\frsl_{2}(\C)$}, in preparation
1291:
1292: \bibitem{HansenTakata} {\bf S\,K Hansen}, {\bf T Takata}, {\it Reshetikhin--Turaev invariants of Seifert $3$--manifolds for classical simple Lie algebras, and their asymptotic expansions}, in preparation
1293:
1294: \bibitem{JankinsNeumann} {\bf M Jankins}, {\bf W\,D Neumann}, {\it Lectures on Seifert manifolds}, Brandeis Lecture Notes 2, Brandeis University (1981)
1295:
1296: \bibitem{Jeffrey1} {\bf L\,C Jeffrey}, {\it On some aspects of Chern--Simons gauge theory}, D. Phil. thesis, University of Oxford (1991)
1297:
1298: \bibitem{Jeffrey2} {\bf L\,C Jeffrey}, {\it Chern--Simons--Witten invariants of lens spaces and torus bundles, and the semiclassical approximation}, Comm. Math. Phys. 147 (1992) 563--604
1299:
1300: \bibitem{JohnsonLapidus} {\bf G\,W Johnson}, {\bf M\,L Lapidus}, {\it The Feynman integral and Feynman's operational calculus}, Oxford mathematical monographs, Clarendon Press, Oxford University Press (2000)
1301:
1302: \bibitem{Kac} {\bf V\,G Kac}, {\it Infinite dimensional Lie algebras}, 3rd ed., Cambridge University Press (1990)
1303:
1304: \bibitem{KacPeterson} {\bf V\,G Kac}, {\bf D\,H Peterson}, {\it Infinite-dimensional Lie algebras, theta functions and modular forms}, Adv. in Math. 53 (1984) 125--264
1305:
1306: \bibitem{Kirillov} {\bf A\,A Kirillov}, {\it On an inner product in modular tensor categories}, J. Amer. Math. Soc. 9 (1996) 1135--1169
1307:
1308: \bibitem{Le} {\bf T\,T\,Q Le}, {\it Quantum invariants of $3$--manifolds: Integrality, splitting, and perturbative expansions}, {\tt arXiv:math.QA/0004099}
1309:
1310: \bibitem{Lusztig} {\bf G Lusztig}, {\it Introduction to quantum groups}, Birkh\"{a}user (1993)
1311:
1312: \bibitem{Montesinos} {\bf J\,M Montesinos}, {\it Classical tessellations and three-manifolds}, Universitext, Springer-Verlag (1987)
1313:
1314: \bibitem{Nishi} {\bf H Nishi}, {\it $SU(n)$--Chern--Simons invariants of Seifert fibered $3$--manifolds}, Internat. J. Math. 9 (1998) 295--330
1315:
1316: \bibitem{Quinn} {\bf F Quinn}, {\it Lectures on axiomatic topological quantum field theory}, (Parc City, UT, 1991), IAS/Park City Math. Ser. 1, Amer. Math. Soc. (1995) 323--453
1317:
1318: \bibitem{RademacherGrosswald} {\bf H Rademacher}, {\bf E Grosswald}, {\it Dedekind sums}, The Carus Mathematical Monographs 16, The Mathematical Association of America (1972)
1319:
1320: \bibitem{ReshetikhinTuraev} {\bf N Reshetikhin}, {\bf V\,G Tureav}, {\it Invariants of $3$--manifolds via link polynomials and quantum groups}, Invent. Math. 103 (1991) 547--597
1321:
1322: \bibitem{Rozansky} {\bf L Rozansky}, {\it Residue formulas for the large $k$ asymptotics of Witten's invariants of Seifert manifolds. The case of $SU(2)$}, Comm. Math. Phys. 178 (1996) 27--60
1323:
1324: \bibitem{Seifert1} {\bf H Seifert}, {\it Topologie dreidimensionaler gefaserter r\"{a}ume}, Acta Math. 60 (1933) 147--238
1325:
1326: \bibitem{Seifert2} {\bf H Seifert}, {\it Topology of $3$--dimensional fibered spaces} (english translation by {\bf W Heil} of: {\it Topologie dreidimensionaler gefaserter r\"{a}ume}, Acta Math. 60 (1933) 147--238). In H. Seifert, W. Threlfall, {\em A textbook of topology}, Pure Appl. Math. 89, Academic Press (1980) 359--422
1327:
1328: \bibitem{Turaev} {\bf V\,G Turaev}, {\it Quantum invariants of knots and $3$--manifolds}, de Gruyter Stud. Math. 18, Walter de Gruyter (1994)
1329:
1330: \bibitem{TuraevWenzl} {\bf V\,G Turaev}, {\bf H Wenzl}, {\it Quantum invariants of $3$--manifolds associated with classical simple Lie algebras}, Internat. J. Math. 4 (1993) 323--358
1331:
1332: \bibitem{Verlinde} {\bf E Verlinde}, {\it Fusion rules and modular transformations in 2D conformal field theory}, Nuclear Phys. B 300 (1988) 360--376
1333:
1334: \bibitem{Witten} {\bf E Witten}, {\it Quantum field theory and the Jones polynomial}, Comm. Math. Phys. 121 (1989) 351--399
1335:
1336: \end{thebibliography}
1337:
1338: \Addresses\recd
1339:
1340: \end{document}
1341:
1342: