1: \documentclass[10pt]{amsart}
2: \usepackage{epic,eepic,latexsym}
3:
4: % Picture stuff
5:
6: \newlength{\baseunit} % the basic unit length
7: \newlength{\templengthhoriz} % width of the picture
8: \newlength{\templengthvert} % depth of the picture
9: \newlength{\temprule} % with between left margin and picture
10: \newcount{\numlines} % depth of picture (in number of lines)
11: \setlength{\baseunit}{0.05ex}
12:
13: \newcommand{\getfig}[2] {
14: \setlength{\unitlength}{#2\baseunit}
15: \input #1.tex }
16:
17: % A macro to input the picture. Use
18: % \getfig{name}{scale}
19:
20:
21: \newcommand{\point}{\vspace{3mm}\par\refstepcounter{subsection}{\bf \thesubsection.} }
22: \newcommand{\tpoint}[1]{\vspace{3mm}\par\refstepcounter{subsection}{\bf \thesubsection.}
23: {\em #1. ---} }
24: \newcommand{\epoint}[1]{\vspace{3mm}\par\refstepcounter{subsection}{\bf \thesubsection.}
25: {\em #1.} }
26: \newcommand{\bpoint}[1]{\vspace{3mm}\par\refstepcounter{subsection}{\bf \thesubsection.}
27: {\bf #1.} }
28: % End template
29:
30: %\newlength{\baseunit} % the basic unit length
31: %\setlength{\baseunit}{0.05ex}
32:
33: \newcommand{\bpf}{\noindent {\em Proof. }}
34: \newcommand{\epf}{\qed \vspace{+10pt}}
35:
36: % I haven't really used this.
37: \newcommand{\theorem}{\noindent {\bf \addtocounter{subsection}{1} \arabic{section}.\arabic{subsection} Theorem.} }
38:
39: % I tend not to use these, and instead use tpoint above.
40: \newtheorem{tm}{Theorem}[subsection]
41: \newtheorem{pr}[tm]{Proposition}
42: \newtheorem{lm}[tm]{Lemma}
43: \newtheorem{co}[tm]{Corollary}
44: \newtheorem{df}[tm]{Definition}
45:
46:
47: \newcommand{\zed}{\mathbb{Z}}
48: \newcommand{\E}{\mathbb{E}}
49: \newcommand{\Q}{\mathbb{Q}}
50: \newcommand{\com}{\mathbb{C}}
51: \newcommand{\eff}{\mathbb{F}}
52: \newcommand{\ka}{\kappa}
53: \newcommand{\proj}{\mathbb P}
54: \newcommand{\oh}{{\mathcal{O}}}
55: \newcommand{\cC}{{\mathcal{C}}}
56: \newcommand{\cL}{{\mathcal{L}}}
57: \newcommand{\cm}{{\mathcal{M}}}
58: \newcommand{\cmbar}{{\overline{\cm}}}
59: \newcommand{\ch}{{\mathcal{H}}}
60: \newcommand{\al}{\alpha}
61: \newcommand{\om}{\omega}
62: \newcommand{\be}{\beta}
63: \newcommand{\ga}{\gamma}
64: \newcommand{\Ga}{\Gamma}
65: \newcommand{\De}{\Delta}
66: \newcommand{\si}{\sigma}
67: \newcommand{\la}{\lambda}
68: \newcommand{\vir}{\operatorname{vir}}
69: \newcommand{\rbr}{\operatorname{br}}
70: \newcommand{\bbr}{\operatorname{Br}}
71: \newcommand{\Pic}{\operatorname{Pic}}
72: \newcommand{\Spec}{\operatorname{Spec}}
73: \newcommand{\Bl}{\operatorname{Bl}}
74: \newcommand{\Sym}{\operatorname{Sym}}
75: \newcommand{\Aut}{\operatorname{Aut}}
76: \newcommand{\mult}{\operatorname{mult}}
77: \newcommand{\grass}{\mathbb{G}}
78: \newcommand{\etale}{\'{e}tale }
79:
80: \newcommand{\cited}{}
81: \newcommand{\photocopy}{}
82: \newcommand{\remind}[1]{{}}
83: % \newcommand{\remind}[1]{{}}
84:
85: \begin{document}
86: \pagestyle{plain}
87: \title{Hodge integrals and hurwitz numbers via virtual localization
88: \footnote{1991 Mathematics Subject Classification: Primary 14H10, Secondary 14H30, 58D29}}
89: \author{Tom Graber}
90: \address{Dept. of Mathematics, Harvard University, Cambridge MA~02138}
91: \email{graber@math.harvard.edu}
92: \author{Ravi Vakil}
93: \address{Dept. of Mathematics, MIT, Cambridge MA~02139}
94: \email{vakil@math.mit.edu}
95: \thanks{The second author is partially supported by NSF Grant DMS--9970101}
96: \date{February 29, 2000.}
97: \begin{abstract}
98: Ekedahl, Lando, Shapiro, and Vainshtein announced a
99: remarkable formula (\cite{elsv}) expressing Hurwitz numbers (counting covers of
100: $\proj^1$ with specified simple branch points, and specified branching
101: over one other point) in terms of Hodge integrals. We give a proof of
102: this formula using virtual localization on the moduli space of stable
103: maps, and describe how the proof could be simplified by the proper
104: algebro-geometric definition of a ``relative space''.
105: \end{abstract}
106: \maketitle
107: \tableofcontents
108:
109: {\parskip=12pt % closing bracket is just before the bibliography
110:
111:
112: \section{Introduction}
113:
114: Hurwitz numbers count certain covers of the projective line (or,
115: equivalently, factorizations of permutations into transposition). They
116: have been studied extensively since the time of Hurwitz, and have
117: recently been the subject of renewed interest in physics (\cite{ct}),
118: combinatorics (\cite{d}, \cite{a}, and the series starting with
119: \cite{gj0}), algebraic geometry (recursions from Gromov-Witten
120: theory, often conjectural), and symplectic geometry (e.g. \cite{lzz}).
121:
122: Ekedahl, Lando, Shapiro and Vainshtein have announced a remarkable
123: formula~(\cite{elsv}~Theorem~1.1; Theorem \ref{biggie} below) linking
124: Hurwitz numbers to Hodge integrals in a particularly elegant way.
125:
126: We prove Theorem \ref{biggie} using virtual localization on the moduli
127: space of stable maps, developed in \cite{gp}. In the simplest case, no
128: complications arise, and Theorem \ref{biggie} comes out immediately;
129: Fantechi and Pandharipande proved this case independently
130: (\cite{fp}~Theorem 2), and their approach inspired ours.
131:
132: We have chosen to present this proof because the formula of Ekedahl et al
133: is very powerful (see Sections \ref{mushroom} and \ref{celery} for
134: applications), and the program they propose seems potentially very
135: difficult to complete (e.g. \cite{elsv} Prop. 2.2, where they require
136: a compactification of the space of branched covers, with specified
137: branching at infinity, which is a bundle over $\cmbar_{g,n}$, such
138: that the branch map extends to the compactification).
139:
140: In Section \ref{relative}, we show that the proof would be much
141: simpler if there were a moduli space for ``relative maps'' in the
142: algebraic category (with a good two-term obstruction theory, virtual
143: fundamental class, and hence virtual localization formula). A space
144: with some of these qualities already exists in the symplectic category
145: (see \cite{lr} Section 7 and \cite{ip} for discussion). In the algebraic case,
146: not much is known, although Gathmann has obtained striking results in
147: genus 0 (\cite{g}).
148:
149:
150: \epoint{Acknowledgements} We are
151: grateful to Rahul Pandharipande, David M. Jackson, and Michael Shapiro
152: for helpful conversations.
153:
154: \section{Definitions and statement}
155:
156: \point \label{avocado} \remind{avocado}
157: Throughout, we work over $\com$, and we use the following notation.
158: Fix a genus $g$, a degree $d$, and a partition $(\al_1,\ldots,\al_m)$
159: of $d$ with $m$ parts. Let $b=2d+2g-2$, the ``expected number of
160: branch points of a degree $d$ genus $g$ cover of $\proj^1$'' by
161: the Riemann-Hurwitz formula.
162: We will identify $\Sym^b \proj^1$ with $\proj^b$ throughout. Let
163: $r = d+m+2(g-1)$,
164: so a branched cover of $\proj^1$, with monodromy above
165: $\infty$ given by $\al$, and $r$ other specified simple branch points
166: (and no other branching) has genus $g$.
167: Let $k = \sum_i (\al_i-1)$, so $r=b-k$.
168: Let $H^g_{\al}$ be the number of such branched covers that are connected.
169: (We do not take the points over $\infty$ to be labelled.)
170:
171:
172: \tpoint{Theorem (Ekedahl-Lando-Shapiro-Vainshtein, \cite{elsv}~Theorem 1.1)}
173: \label{biggie}
174: \remind{biggie} {\em
175: Suppose $g$, $m$ are integers ($g \geq 0$, $m \geq 1$) such that $2g-2+m>0$ (i.e.
176: the functor $\cmbar_{g,m}$ is represented by a Deligne-Mumford stack). Then
177: \begin{equation*}
178: H^g_\al = \frac {r!} { \# \Aut(\al)}
179: \prod_{i=1}^m \frac {{\al_i}^{\al_i}} {\al_i!}
180: \int_{\cmbar_{g,m}} \frac { 1-\la_1 + \dots \pm \la_g} {\prod (1-\al_i \psi_i)}
181: \end{equation*}
182: where $\la_i=c_i(\E)$ ($\E$ is the Hodge bundle).}
183:
184: Fantechi and Pandharipande's argument applies in the case where there is no
185: ramification above $\infty$, i.e. $\al = (1^d)$.
186:
187: The reader may check that a variation of our method also shows that
188: $$
189: H^0_{\al_1} = r! \frac {d^{d-2}} {d!}, \; \; \; H^0_{\al_1,\al_2} = \frac {r!}
190: {\# \Aut(\al_1, \al_2)} \cdot \frac {\al_1^{\al_1}} {\al_1!} \cdot
191: \frac {\al_2^{\al_2}} {\al_2!} \cdot d^{d-1}.
192: $$
193: As these formulas are known by other means (\cite{d} for the first, \cite{a} for the second,
194: \cite{gj0} for both), we omit the proof.
195:
196:
197: \epoint{Application: Hurwitz numbers to Hodge integrals}
198: \label{mushroom} \remind{mushroom}
199: (i) Theorem \ref{biggie} provides a way of computing all Hodge integrals
200: as follows. Define $$\langle \al_1, \dots, \al_m \rangle := \int_{\cmbar_{g,m}}
201: \frac { 1-\la_1 + \dots \pm \la_g} {\prod (1-\al_i \psi_i)},$$ a
202: symmetric polynomial in the $\al_i$ of degree $3g-3+m$ whose
203: coefficients are of the form $\int_{\cmbar_{g,m}} \psi_1^{d_1} \dots
204: \psi_m^{d_m} \la_k$. It is straightforward to recover the
205: coefficients of a symmetric polynomial in $m$ variables of known
206: degree from a finite number of values, and $\langle \al_1, \dots, \al_m \rangle$ can
207: easily be computed (as Hurwitz numbers are combinatorial objects that
208: are easily computable, see Section \ref{pea}). Once these integrals are
209: known, all remaining Hodge integrals (i.e. with more $\la$-classes) can be
210: computed in the usual way (\cite{m}). The only other methods known to us are
211: Kontsevich's theorem, formerly Witten's conjecture,
212: \cite{ko}, which has no known algebraic proof,
213: and methods of Faber and Pandharipande (making clever use of
214: virtual localization, \cite{p}). These methods of computation
215: are in keeping with an extension of Mumford's philosophy, which is that
216: much of the cohomology of $\cmbar_{g,n}$ is essentially combinatorial.
217:
218: (ii) Combinatorially straightforward relations among Hurwitz numbers
219: (e.g. ``cut-and-join'', see \cite{gj0} Section 2) yield nontrivial new
220: identities among Hodge integrals.
221:
222: \epoint{Application: Hodge integrals to Hurwitz numbers} \label{celery}
223: \remind{celery}
224: There has been much work on the structure of the Hurwitz numbers,
225: including various predictions from physics. Theorem \ref{biggie} is
226: the key step in a machine to verify these structures and predictions,
227: see \cite{gjv}.
228:
229: \section{Background: Maps of curves to curves}
230: \point \label{chickpea} \remind{chickpea}
231: Following \cite{g01pn} Section 4.2, define a {\em special locus} of a
232: map $f: X \rightarrow \proj^1$ (where $X$ is a nodal curve) as
233: a connected component of the locus in $X$ where $f$ is not \etale.
234: (Remark: No result in this section requires the target to be $\proj^1$.)
235: Then a special locus is of one of the following forms: (i) a nonsingular
236: point of $X$ that is an $m$-fold branch point (i.e. analytically
237: locally the map looks like $x \rightarrow x^m$, $m>1$), (ii) a node of
238: $X$, where the two branches of the node are branch points of order
239: $m_1$, $m_2$, or (iii) one-dimensional, of arithmetic genus $g$,
240: attached to $s$ branches of the remainder of the curve that are
241: $c_j$-fold branch points ($1 \leq j \leq s$). The form of the locus,
242: along with the numerical data, will be called the {\em type}. (For
243: convenience, we will consider a point {\em not} in a special locus to
244: be of type (i) with $m=1$.) We will use the fact that special loci
245: of type (ii) are smoothable (\cite{p1} Section 2.2).
246:
247: \epoint{Ramification number}
248: \label{broccoli} \remind{broccoli}
249: To each special locus, associate a {\em ramification number} as
250: follows: (i) $m-1$, (ii) $m_1 +m_2$, (iii) $2g-2+ 2s + \sum_{j=1}^s(c_j-1)$.
251: (Warning: in case (i), this is one less than what is normally called
252: the ramification index; we apologize for any possible confusion.) The
253: {\em total ramification} above a point of $\proj^1$ is the sum of the
254: ramification numbers of the special loci mapping to that point.
255: We will use the following two immediate facts: if the map is stable, then
256: the ramification number of each ``special locus'' is a positive
257: integer, and each special locus of type (iii) has ramification number
258: at least 2.
259:
260: \epoint{Extended Riemann-Hurwitz formula}
261: \label{pumpkin} \remind{pumpkin}
262: There is an easy generalization of the Riemann-Hurwitz formula:
263: $$
264: 2 p_a(X) - 2 = -2d + \sum r_i
265: $$
266: where $\sum r_i$ is the sum of the ramification numbers.
267: (The proof is straightforward. For example, consider the
268: complex $f^* \omega^1_{\proj^1} \rightarrow \omega^1_X$ as
269: in \cite{fp} Section 2.3,
270: and observe that its degree can be decomposed into contributions from
271: each special locus. Alternatively, it follows from the usual
272: Riemann-Hurwitz formula and induction on the number of nodes.)
273:
274: \epoint{Behavior of ramification number and type in families} \label{yam} \remind{yam} Ramification
275: number is preserved under deformations. Specifically, consider a
276: pointed one-parameter family of maps (of nodal cures). Suppose one map
277: in the family has a special locus $S$ with ramification number $r$.
278: Then the sum of the ramification numbers of the special loci in a
279: general map that specialize to $S$ is also $r$. (This can be shown by
280: either considering the complex $f^* \omega^1_{\proj^1} \rightarrow
281: \omega^1_X$ in the family or by deformation theory.)
282:
283: Next, suppose
284: $$\begin{array} {cc}
285: C & \rightarrow \proj^1 \\
286: \downarrow & \\
287: B &
288: \end{array}
289: $$
290: is a family of {\em stable} maps parametrized by a nonsingular curve $B$.
291:
292: \tpoint{Lemma} \label{radish} \remind{radish}
293: {\em Suppose there is a point $\infty$ of $\proj^1$ where the total
294: ramification number of special loci mapping to $\infty$ is a constant
295: $k$ for all closed points of $B$. Then the type of ramification above
296: $\infty$ is constant, i.e. the number of preimages of $\infty$ and their types are constant.}
297:
298: For example, if the general fiber is nonsingular, i.e. only has
299: special loci of type (i), then that is true for all fibers.
300:
301: {\em Proof.} Let $0$ be any point of $B$, and let $f: X \rightarrow
302: \proj^1$ be the map corresponding to $0$. We will show that the type
303: of ramification above $\infty$ for $f$ is the same as for the general
304: point of $B$.
305:
306: First reduce to the case where the general map has no contracted
307: components. (If the general map has a contracted component $E$, then consider
308: the complement of the closure of $E$ in the total general family. Prove the result
309: there, and then show that the statement of Lemma \ref{radish} behaves well
310: with respect to gluing a contracted component.)
311:
312: Similarly, next reduce to the case where general map is nonsingular.
313: (First show the result where the nodes that are in the closure of the
314: nodes in the generic curve are normalized, and then show that the
315: statement behaves well with respect to gluing a 2-section of
316: the family to form a node.)
317:
318: Pull back to an \etale neighborhood of 0 to separate special loci of
319: general fiber (i.e. so they are preserved under monodromy), and also
320: the fibers over $\infty$ for the general map.
321:
322: For convenience of notation, restrict attention to one special locus
323: $E$ of $f$. Assume first that $E$ is of type (iii), so $\dim E = 1$.
324: Let $g_E$ be the arithmetic genus of $E$. Suppose that $r$ preimages
325: of $\infty$ of the general fiber (of type (i) by reductions) meet $E$
326: in the limit, and that these have ramification numbers $b_1$, \dots,
327: $b_r$. Let $s$ be the number of other branches of $X$ meeting $E$, and $c_1$,
328: \dots, $c_s$ the ramification numbers of the branches (as in
329: Section \ref{chickpea}).
330:
331: The ramification number of $E$ is $(2 g_E-2) + 2s + \sum_{j=1}^s (c_j-1)$.
332: The total ramification number of the special loci specializing to $E$ is
333: $\sum_{i=1}^r (b_i-1)$.
334: Also,
335: $$\sum_{i=1}^r b_i = \sum_{j=1}^s c_j.$$
336: Hence by conservation
337: of ramification number,
338: $$
339: (2 g_E-2+s) + r = 0.
340: $$
341: But $r>0$,
342: and by the stability condition for $f$, $2g_E-2+s>0$, so we
343: have a contradiction.
344:
345: If $\dim E= 0$ is 0 (i.e. $E$ is of type (i) or (ii)), then
346: essentially the same algebra works (with the substitution ``$g_E=0$'',
347: resulting in $r+s-2=0$, from which $r=s=1$, from which the type is
348: constant).
349: \epf
350:
351:
352: A similar argument shows:
353: \tpoint{Lemma} \label{cucumber} \remind{cucumber}
354: {\em Suppose $E$ is a special locus in a specific fiber, and only one special locus $E'$ in the general fiber meets it.
355: Then the types of $E$ and $E'$ are the same.}
356:
357: \epoint{The Fantechi-Pandharipande branch morphism}
358: For any map $f$ from a nodal curve to a nonsingular curve, the ramification number defines a
359: divisor on the target: $\sum_L r_L f(L)$, where $L$ runs through the
360: special loci, and $r_L$ is the ramification number. This induces a
361: set-theoretic map $\bbr: \cmbar_g(\proj^1,d)
362: \rightarrow \Sym^b \proj^1 \cong \proj^b$.
363: In \cite{fp}, this was shown to be a morphism.
364:
365: Let $p$ be the point of $\Sym^b \proj^1 \cong \proj^b$
366: corresponding to $k(\infty) + (b-k)(0)$, let $L_\infty \subset
367: \proj^b$ be the linear
368: space corresponding to points of the form $k (\infty)+D$ (where $D$ is a divisor of degree $r=b-k$), and let
369: $\iota: L_\infty \rightarrow \proj^b$ be the inclusion.
370:
371: Define $M$ as the stack-theoretic pullback $\bbr^{-1}L_\infty$. It carries a virtual fundamental class
372: $[M]^{\vir} = \iota^! [ \cmbar_g(\proj^1,d) ]^{\vir}$ of dimension $r=b-k$
373: (i.e. simply intersect the class $[\cmbar_g(\proj^1,d)]^{\vir}$
374: with the
375: codimension $k$ operational Chow class $\bbr^*[L_\infty]$; the result
376: is supported on $\bbr^{-1}L_\infty$).
377: Denote the {\em restricted branch map} by $\rbr: M \rightarrow L_\infty$.
378: By abuse of notation,
379: we denote the top horizontal arrow in the following diagram by $\iota$ as well.
380: $$
381: \begin{array}{ccc}
382: M & \rightarrow & \cmbar_g(\proj^1,d) \\
383: \rbr \downarrow & & \downarrow \bbr \\
384: L_\infty & \stackrel \iota {\rightarrow} & \proj^b
385: \end{array}
386: $$
387: By the projection formula,
388: \begin{equation}
389: \label{leek}
390: \iota_* ( \rbr^*[p] \cap [M]^{\vir}) = \bbr^*[p] \cap [ \cmbar_g(\proj^1,d) ]^{\vir}.\end{equation}
391: \remind{leek}
392:
393: Define $M^\al$ as the union of irreducible components of $M$ whose
394: general members correspond to maps from irreducible curves, with
395: ramification above $\infty$ corresponding to $\al$ with the reduced substack
396: structure. (It is not hard
397: to show that $M^\al$ is irreducible, by the same group-theoretic
398: methods as the classical proof that the Hurwitz scheme is irreducible. None of
399: our arguments use this fact, so we will not give the details of the
400: proof. Still, for convenience, we will assume irreducibility in our
401: language.)
402:
403: \point \label{kidneybean} \remind{kidneybean}
404: Note that $M=\bbr^{-1} L_\infty$ contains $M^\al$ with some multiplicity $m_\al$,
405: as $M^\al$ is of the expected dimension $r$.
406: The Hurwitz number $H^g_\al$ is given by
407: $$\int_{M^\al} \rbr^* [p].$$
408: (The proof of \cite{fp} Proposition 2 carries over without change in this case, as does
409: the the argument of \cite{g1} Section 3.)
410: This is $1/ m_\al$ times the cap product of $\rbr^* [p]$ with
411: the part of the
412: class of $[M]^{\vir}$ supported on $M^\al$.
413:
414: \tpoint{Lemma} \label{lettuce} \remind{lettuce}
415: {\em $m_\al = k! \prod \left( \frac {\al_i^{\al_i-1}} {\al_i!} \right).$}
416:
417: \point \label{pea} \remind{pea}
418: In the proof, we will use the combinatorial interpretation of Hurwitz
419: numbers: $H^g_\al$ is $1/d!$ times the number of ordered $r$-tuples
420: $(\tau_1, \dots, \tau_r)$ of transpositions generating $S_d$, whose
421: product has cycle structure $\al$.
422:
423: {\em Proof.} Fix $r$ general points $p_1$, \dots, $p_r$ of $\proj^1$.
424: Let $L \subset \proj^b$ be the linear space corresponding to divisors
425: of the form $p_1 + \dots + p_r + D$ (where $\deg D = k$). By the
426: Kleiman-Bertini theorem, $(\bbr|_{M_\al})^{-1} L$ consists of $H^g_\al$
427: reduced points.
428:
429: Now $L_\infty \subset \Sym \proj^b$ can be interpreted as a ({\em
430: real} one-parameter) degeneration of the linear space
431: corresponding to divisors of the form $D'+ \sum_{i=1}^k
432: q_i$, where $q_1$, \dots $q_k$ are fixed generally chosen points of
433: $\proj^1$ and $D'$ is any degree $r$ divisor on $\proj^1$.
434:
435: Choose branch cuts to the points $p_1$, \dots, $p_r$, $q_1$, \dots,
436: $q_k$, $\infty$ from some other point of $\proj^1$. Choose a real
437: one-parameter path connecting $q_1$, \dots, $q_k$, $\infty$ (in that
438: order), not meeting the branch cuts (see the dashed line in Figure
439: \ref{degenfig}). Degenerate the points $q_i$ to $\infty$ along this
440: path one at a time (so the family parametrizing this degeneration is
441: reducible). If $\si_1$, \dots, $\si_k$, $\si_\infty$ are the
442: monodromies around the points $q_1$, \dots, $q_k$,
443: $\infty$ for a certain cover, then the monodromy around $\infty$ after
444: the branch points $q_i$, \dots, $q_k$ have been degenerated to
445: $\infty$ (along the path) is $\si_i \dots \si_k \si_\infty$.
446:
447: At a general point of the family parametrizing this real degeneration (before any of the
448: points $q_i$ have specialized, i.e. the $q_i$ are fixed general
449: points), $\bbr^{-1} (L \cap L_\infty)$ is a finite number of reduced
450: points. This number is the Hurwitz number $H^g_{(1^d)}$ (\cite{fp}
451: Prop. 2), i.e. $1/d!$ times the number of choices of $b=r+k$
452: transpositions $\tau_1$, \dots, $\tau_r$, $\si_1$,
453: \dots, $\si_k$ in $S_d$ such that $\tau_1 \dots \tau_r \si_1 \dots
454: \si_k$ is the identity and $\tau_1$, \dots, $\tau_r$, $\si_1$, \dots,
455: $\si_k$ generate $S_d$.
456:
457: \begin{figure}
458: \begin{center}
459: \getfig{degenfig}{.1}
460: \end{center}
461: \caption{Degenerating the points $q_1$, \dots, $q_k$ to $\infty$ one by one, along a real path \remind{degenfig}}
462: \label{degenfig}
463: \end{figure}
464:
465: As we specialize the $k$ branch points $q_1$, \dots, $q_k$
466: to $\infty$ one at a time, some
467: of these points tend to points of $M^\al$; these are the points for
468: which $\tau_1$, \dots, $\tau_r$ generate $S_d$, and their product has
469: cycle structure $\al$. The multiplicity $m_\al$ is the number of
470: these points that go to each point of $M^\al$. This is the number of
471: choices of $k$ transpositions $\si_1$, \dots, $\si_k$ whose product is
472: a {\em given} permutation $\xi$ with cycle structure $\al$. (Note that this
473: number is independent of the choice of $\xi$; hence the multiplicity is
474: independent of choice of component of $M^\al$.)
475:
476: If $k=\sum(\al_i-1)$ transpositions $\si_1$, \dots, $\si_k$ multiply
477: to a permutation $\xi=(a_{1,1} \dots a_{1,\al_1}) \dots (a_{m,1} \dots
478: a_{m, \al_m})$ (where $\{ a_{1,1}, \dots, a_{m,\al_m} \} = \{ 1,
479: \dots, d \}$), then for $1 \leq i \leq m$, $\al_i-1$ of the
480: transpositions must be of the form $(a_{i,j} a_{i,k})$. (Reason:
481: A choice of $k+1$ points $q_1$, \dots, $q_k$, $\infty$, of $\proj^1$
482: and the
483: data $\si_1$, \dots, $\si_k$, $\xi$ defines a degree $d$ branched cover
484: of $\proj^1$, simply branched above $q_j$ and with ramification type
485: $\al$ above $\infty$. By the Riemann-Hurwitz formula,
486: the arithmetic genus of
487: this cover is $1-m$; as the pre-image of $\infty$ contains $m$ smooth
488: points, the cover has at most $m$ components. Hence the cover has
489: precisely $m$ components, each of genus 0. The $i$th component is
490: simply branched at $\al_i-1$ of the points $\{ q_1, \dots, q_k \}$
491: away from $\infty$.)
492:
493: The number of ways of factoring an $\al_i$-cycle into $\al_i-1$
494: transpositions is $\al_i^{\al_i-2}$ (straightforward; or see \cite{d}
495: or \cite{gj0} Theorem 1.1). Hence $m_\al$ is the number of ways of partitioning
496: the $k$ points $q_1$, \dots, $q_k$ into subsets of size $\al_1-1$, \dots, $\al_m-1$,
497: times the number of ways of factoring the $\al_i$-cycles:
498: $$
499: m_\al = \binom {k} {\al_1 -1, \dots, \al_m-1} \prod \al_i^{\al_i-2} =
500: k! \prod \left( \frac {\al_i^{\al_i-1}} {\al_i!} \right).
501: $$
502: \epf
503:
504: \section{Virtual localization}
505: \bpoint{Virtual localization preliminaries}
506: We evaluate the integral using virtual localization (\cite{gp}). The
507: standard action of $\com^*$
508: on $\proj^1$ (so that the action on the tangent space at $\infty$ has weight 1) induces a natural $\com^*$-action
509: on $\cmbar_g(\proj^1,d)$, and the branch morphism $\bbr$ is
510: equivariant with respect to the induced torus action on $\Sym^b \proj^1
511: \cong \proj^b$. As a result, we can regard $\rbr^*[p]$ as an
512: equivariant Chow cohomology class in $A^r_{\com^*} M$. Let $\{ F_l
513: \}_{l \in L}$ be the set of components of the fixed locus of
514: the torus action on $\cmbar_g(\proj^1,d)$, where $L$ is some index set.
515: (Note that the connected components of the fixed locus are also irreducible.)
516:
517: Define $F_0$ to be the component of the fixed locus whose general point
518: parametrizes a
519: stable map with a single genus $g$ component contracted over 0, and
520: $m$ rational tails mapping with degree $\al_i$ ($1 \leq i \leq m$) to
521: $\proj^1$, totally ramified above 0 and $\infty$. $F_{0}$ is naturally
522: isomorphic to a quotient of $\cmbar _{g,m}$ by a finite group. See
523: \cite{k} or \cite{gp} for a discussion of the structure of the
524: fixed locus of the $\com ^{*}$ action on $\cmbar _{g}(\proj ^{1},d).$
525:
526: By the virtual localization formula, we can explicitly write down classes $\mu_l \in A_*^{\com^*}(F_l)_{(1/t)}$
527: such that $$
528: \sum_l i_*(\mu_l) = [M]^{\vir}$$ in $A_*^{\com^*}(M)$.
529: Here, and elsewhere, $i$ is the natural inclusion. It is important
530: to note that the $\mu_l$ are uniquely determined by this equation.
531: This follows from the Localization Theorem 1 of \cite{eg} (extended
532: to Deligne-Mumford stacks by \cite{kresch}), which says that pushforward gives
533: an isomorphism between the localized Chow group of the fixed locus and
534: that of the whole space.
535:
536: In order to pick out the contribution to this integral from a
537: single component $F_0$, we introduce more refined classes. We
538: denote the irreducible components of $M$ by $M_n$, and arbitrarily
539: choose a representation
540: $$
541: [M]^{\vir} = \sum_n i_* \Ga_n
542: $$
543: where $\Ga_n \in A_*^{\com^*}( M_n)$.
544: For a general
545: component, we can say little about these classes, but for our distinguished irreducible component $M^\al$
546: the corresponding $\Ga_\al$ is necessarily $m_\al [M^\al]$. (Note that
547: $M^\al$ has the expected dimension, so the Chow group in that
548: dimension is generated by the fundamental class).
549:
550: Next, we localize each of the $\Ga_n$. Define $\eta_{l,n}$ in $A_*^{\com^*}(F_l)_{(1/t)}$ by
551: \begin{equation}
552: \label{rutabaga}
553: \sum_l i_* \eta_{l,n}= \Ga_n
554: \end{equation}
555: \remind{rutabaga}
556: Once again (by \cite{eg}, \cite{kresch}), the $\eta_{l,n}$ are uniquely defined; this will be used in Lemma \ref{potato}.
557: Also, $\sum_n \eta_{l,n} = \mu_l$ (as the $\mu_l$ are uniquely determined).
558:
559: \tpoint{Lemma} \label{squash} \remind{squash} {\em
560: The equivariant class $\rbr^*[p]$ restricts to 0 on any component
561: of the fixed locus whose general map has total ramification number greater
562: than $k$ above $\infty$.}
563:
564: {\em Proof.}
565: Restricting the branch morphism to such a component, we see that it
566: gives a constant morphism to a point in $\proj ^{b}$ other than $p$.
567: Consequently, the pull-back of the class $p$ must vanish.
568: \epf
569:
570: \tpoint{Lemma} \label{melanzana} \remind{melanzana} {\em
571: $\int_{\Ga_n} \rbr^*[p] = 0$ for any irreducible component $M_n$ whose
572: general point corresponds to a map which has a contracted
573: component away from $\infty$.}
574:
575: {\em Proof.} A general cycle $\gamma \in L_\infty$ representing $p$ is
576: the sum of $r$ distinct points plus the point $\infty$ exactly $k$
577: times. However, a contracted
578: component always gives a multiple component of the branch divisor,
579: (Section \ref{broccoli}) so the image of $M_{n}$ cannot meet a general
580: point.
581: \epf
582:
583:
584: \tpoint{Lemma} {\em $\eta_{l,n} = 0$ if $F_l \cap M_n = \emptyset$.}
585: \label{potato} \remind{potato}
586:
587: {\em Proof.}
588: Since $\Ga_n$ is an element of $A_*^{\com^*}(M_n)$, there exist
589: classes $\tilde{\eta}_{l,n}$ in the localized equivariant Chow groups
590: of the fixed loci of $M_n$ satisfying equation (\ref{rutabaga}).
591: Pushing these forward to the fixed loci of $M$ gives classes in the Chow groups of the $F_l$
592: satisfying the same equation. By uniqueness, these must be the $\eta_{l,n}$. By this
593: construction, it follows that they can only be non-zero if $F_l$ meets $M_n$. \epf
594:
595: \tpoint{Lemma} {\em No irreducible component of $M$ can meet two distinct components of the
596: fixed locus with total ramification number exactly $k$ above $\infty$.}
597: \label{cabbage}
598: \remind{cabbage}
599:
600: {\em Proof.} To each map $f: X \rightarrow \proj^1$ with total ramification
601: number exactly $k$ above $\infty$, associate a graph as follows. The
602: connected components of the preimage of $\infty$ correspond to red
603: vertices; they are labelled with their type. The connected components of $Y=
604: \overline{X \setminus f^{-1}( \infty )}$ (where the closure is taken in
605: $X$) correspond to green vertices; they are labelled with their
606: arithmetic genus. Points of $Y \cap f^{-1}( \infty)$ correspond to
607: edges connecting the corresponding red and green points; they are
608: labelled with the ramification number of $Y \rightarrow \proj^1$ at
609: that point. Observe that this associated graph is constant in
610: connected families where the total ramification over $\infty$
611: is constant, essentially by Lemma \ref{radish}.
612:
613: If an irreducible component $M'$ of $M$ meets a component of the fixed
614: locus with total ramification number exactly $k$ above $\infty$, then the
615: general map in $M'$ has total ramification $k$ above $\infty$.
616: (Reason: the total ramification is at most $k$ as it specializes to a
617: map with total ramification exactly $k$; and the total ramification is
618: at least $k$ as it is a component of $M$.) There is only one
619: component of the fixed locus that has the same associated graph as the
620: general point in $M'$, proving the result. \epf
621:
622: \tpoint{Lemma} \label{arugula} \remind{arugula} {\em
623: The map parametrized by a general point of any
624: irreducible component of $M$ other
625: than $M^\al$ which meets $F_0$ must have a contracted component not mapping
626: to $\infty$.}
627:
628: {\em Proof.} Let $M'$ be an irreducible component of $M$ other than
629: $M^\al$. As in the proof of Lemma \ref{cabbage}, a general map $f: X \rightarrow \proj^1$ of
630: $M$ has total ramification exactly $k$ above $\infty$. By Lemma
631: \ref{radish}, we know the type of the special loci above $\infty$:
632: they are nonsingular points of the source curve, and the ramification
633: numbers are given by $\al_1, \dots, \al_m$.
634:
635: As $M' \neq M$, $X$ is singular. If $f$ has a special locus of type
636: (iii), then we are done. Otherwise, $f$ has only special loci of type
637: (ii), and none of these map to $\infty$. But then these type (ii)
638: special loci can be smoothed while staying in $M$ (Section
639: \ref{broccoli}), contradicting the assumption that $f$ is a general
640: map in a component of $M$.
641: \epf
642:
643: \tpoint{Proposition}
644: \label{onion} \remind{onion} $$
645: m_\al \int_{M^{\al}} \rbr^*[p] = \int_{F_0} \rbr^* [p] \cap \mu_0.$$
646:
647: It is the class $\mu_0$ that the Virtual Localization Theorem
648: of \cite{gp}
649: allows us to calculate explicitly. Thus this proposition is the main ingredient in giving us
650: an explicit formula for the integral we want to compute.
651:
652: {\em Proof.}
653: Now $\Ga_\al = m_\al [M^\al]$, so by definition of $\eta_{l,\al}$,
654: $$
655: m_\al [M^\al] = \sum_l i_* \eta_{l,\al}.$$
656:
657: By Lemma \ref{melanzana}, $M^\al$ meets only one component of the fixed locus which
658: has total ramification number $k$, $F_0$. Along with Lemmas \ref{squash}
659: and \ref{potato},
660: this implies that
661: $$
662: m_\al \int_{M^\al} \rbr^*[p] = \int_{F_0} \rbr^*[p] \cap \eta_{0,\al}.
663: $$
664: In other words, the only component of the fixed locus which
665: contributes to this integral is $F_0$. Since $\mu_0 = \sum_n
666: \eta_{0,n}$, the proposition will follow if we can show that
667: $$\int_{F_0} \rbr^* [p] \cap \eta_{0,n} = 0$$ for $n \neq \al$, i.e. that
668: no other irreducible component of $M$ contributes to
669: the localization term coming from $F_0$.
670:
671: If $F_0 \cap M_n = \emptyset$, this is true by Lemma \ref{potato}.
672: Otherwise, by Lemma \ref{arugula}, the general map in $M_n$
673: has a contracted component, so by Lemma \ref{melanzana}
674: $\int_{\Gamma_n} \rbr^*[p] = 0$. By equation (\ref{rutabaga}),
675: $$
676: \sum_l \int_{F_l} \rbr^*[p] \cap \eta_{l,n} = 0.$$
677: If $F_l$ generically corresponds to maps that have total ramification
678: number greater than $k$ above $\infty$, then $\rbr^*[p] \cap \eta_{l,n} = 0$
679: by Lemma \ref{melanzana} . If $l \neq 0$ and $F_l$ generically corresponds
680: to maps that have total ramification number $k$ above $\infty$,
681: then $\rbr^*[p] \cap \eta_{l,n} = 0$ by Lemma \ref{cabbage}, as $M_n$ meets
682: $F_0$. Hence $\int_{F_0} \rbr^*[p] \cap \eta_{0,n} = 0$ as desired.
683: \epf
684:
685: \bpoint{Proof of Theorem \ref{biggie}}
686: \label{proof} \remind{proof}
687: All that is left is to explicitly write down the right hand side of
688: Proposition \ref{onion}. By equation (\ref{leek}), this integral
689: can be interpreted as the contribution of $F_0$ to the integral of
690: $\bbr^*[p]$ against the virtual fundamental class of $\cmbar_g(\proj^1,d)$, divided by
691: $m_\al$. Since this means we are trying to compute an equivariant
692: integral over the entire space of maps to $\proj^1$, we are in exactly
693: the situation discussed in \cite{gp}. Let $\gamma$ be the natural
694: morphism from $\cmbar_{g,m}$ to $F_0$. The degree of $\gamma$ is $\#
695: \Aut(\alpha)\prod \alpha_i$. The pullback under $\gamma$ of the
696: inverse euler class of the virtual normal bundle is computed to be
697: $$c(E^\vee) \left( \prod \frac{1}{1-\alpha_i\psi_i} \cdot
698: \frac{(-1)^{\alpha_i}\alpha_i^{2\alpha_i}}{(\alpha_i!)^2} \right).$$
699: The class $\rbr^*[p]$ is easy to evaluate. Since $\rbr$ is constant when
700: restricted to $F_0$, this class is pure weight, and is given by the
701: product of the weights of the $\com^*$ action on $T_p \proj^b$. These
702: weights are given by the non-zero integers from $-(b-k)$ to $k$ inclusive.
703: The integral over $F_0$ is just the integral over $\cmbar_{g,m}$ divided
704: by the degree of $\gamma$. We conclude that
705: $$m_\al \int_{[M^\alpha]} br^*[p] =
706: \frac{k!(b-k)!}{\# \Aut(\alpha) \prod \alpha_i} \cdot \prod
707: \frac{\alpha_i^{2\alpha_i}}{(\alpha_i!)^2} \cdot \int_{\cmbar_{g,m}}
708: \frac{c(\E^\vee)}{\prod (1-\alpha_i \psi_i)}.
709: $$
710: Dividing by $m_\alpha$ (calculated
711: in Lemma \ref{lettuce}) yields the desired formula.
712: \epf
713:
714: \section{A case for an algebraic definition of a
715: space of ``relative stable maps''}
716: \label{relative}
717: \remind{relative}
718:
719: A space of ``relative stable maps'' has been defined in the symplectic category
720: (see \cite{lr} and \cite{ip}), but hasn't yet been properly defined in the algebraic
721: category (with the exception of Gathmann's work in genus 0, \cite{g}).
722:
723: The proof of Theorem \ref{biggie} would become quite short were such a
724: space $\cm$ to exist with expected
725: properties, namely the following. Fix $d$, $g$, $\al$, $m$, $k$, $r$ as
726: before (see Section \ref{avocado}).
727: \begin{enumerate}
728: \item $\cm$ is a proper Deligne-Mumford stack, which contains as an open substack $U$ the
729: locally closed substack of $\cmbar_g(\proj^1,d)$ corresponding to maps to $\proj^1$ where
730: the pre-image of $\infty$ consists of $m$ smooth points appearing with multiplicity
731: $\al_1$, \dots, $\al_m$.
732: \item There is a Fantechi-Pandharipande branch map $\bbr: \cm \rightarrow
733: \Sym^b \proj^1$. The image will be contained in $L_\infty$, so we may consider
734: the induced map $\rbr$ to $L_\infty \cong \Sym^r \proj^1$. Under this
735: map, the set-theoretic fiber of $k(\infty) +r(0)$ is precisely $F_{0}$.
736: \item There is a $\com^{*}$-equivariant perfect obstruction theory on
737: $\cm $ which when
738: restricted to $U$ is given (relatively over ${\mathfrak{M}}_g$) by
739: $R\pi _{*}(f^{*}(T\proj ^{1} \otimes \oh (- \infty)))$, where
740: $\pi$ is the structure morphism from the universal curve to $\cm$.
741: \end{enumerate}
742: With these axioms, the proof would require only Section \ref{proof}.
743:
744: All of these requirements are reasonable. However, as a warning, note that
745: the proof of Proposition \ref{onion} used special properties of the class
746: $\rbr^* [p]$ (Lemmas \ref{squash}--\ref{arugula}).
747:
748: One might expect this space to be a combination of Kontsevich's space
749: $\cmbar_g(\proj^1,d)$ and the space of twisted maps introduced by
750: Abramovich and Vistoli (see \cite{av} Section 3).
751:
752:
753: } % end of parskip; it started just before the introduction
754:
755: \begin{thebibliography}{[ELSV]}
756: \bibitem[AV]{av} {\sc D. Abramovich and A. Vistoli}, {\em Complete
757: moduli for families over semistable curves}, preprint 1998, math.AG/9811059.
758: \bibitem[A]{a} {\sc V. I. Arnol'd},
759: {\em Topological classification of trigonometric polynomials and
760: combinatorics of graphs with an equal number of vertices and edges},
761: Functional Analysis and its Applications {\bf 30} no. 1 (1996), 1--17. \cited
762: \bibitem[CT]{ct}{\sc M. Crescimanno and W. Taylor},
763: {\em Large N phases of chiral $QCD_2$},
764: Nuclear Phys. {\bf B~437} (1995), 3--24.
765: \bibitem[D]{d} {\sc J. D\'{e}nes},
766: {\em The representation of a permutation as the product of a minimal
767: number of transpositions and its connection with the theory of graphs},
768: Publ. Math. Ins. Hungar. Acad. Sci. {\bf 4} (1959), 63--70.
769: \bibitem[EG]{eg} {\sc D. Edidin and W. Graham}, {\em Localization
770: in equivariant intersection theory and the Bott residue formula},
771: Amer. J. Math. {\bf 120} (1998), no. 3, 619--636.
772: \bibitem[ELSV]{elsv} {\sc T. Ekedahl, S. Lando, M. Shapiro, and
773: A. Vainshtein},
774: {\em On Hurwitz numbers and Hodge integrals}, C. R. Acad. Sci. Paris, t. 328, S\'{e}rie I,
775: p. 1171--1180, 1999.
776: \bibitem[FP]{fp} {\sc B. Fantechi and R. Pandharipande},
777: {\em Stable maps and branch divisors}, preprint
778: 1999, math.AG/9905104. \cited
779: \bibitem[G]{g} {\sc A. Gathmann},
780: {\em Absolute and relative Gromov-Witten invariants of
781: very ample hypersurfaces},
782: preprint 1999, math.AG/9908054. \cited
783: \bibitem[GJ]{gj0} {\sc I. P. Goulden and D.M. Jackson},
784: {\em Transitive factorisations into transpositions
785: and holomorphic mappings on the sphere},
786: Proc. A.M.S.,
787: {\bf125} (1997), 51--60. \cited
788: \bibitem[GJV]{gjv} {\sc I. P. Goulden, D. M. Jackson, and R. Vakil}, {\em
789: The Gromov-Witten potential of a point, Hurwitz numbers,
790: and Hodge integrals}, preprint 1999, math.AG/9910004. \cited
791: \bibitem[GP]{gp} {\sc T. Graber and R. Pandharipande},
792: {\em Localization of virtual classes},
793: Invent. Math. {\bf 135} (1999), no. 2, 487--518. \cited
794: \bibitem[IP]{ip} {\sc E.-N. Ionel and T. Parker}, {\em Relative
795: Gromov-Witten invariants}, preprint 1999, math.SG/9907155.
796: \bibitem[K1]{ko} {\sc M. Kontsevich}, {\em Intersection theory on the moduli
797: space of curves and the matrix Airy function}, Comm. Math. Phys. {\bf 147} (1992), no. 1, 1--23.
798: \bibitem[K2]{k}{\sc M. Kontsevich}, {\em Enumeration of rational
799: curves via torus actions}, in {\em The moduli space of curves}, R. Dijkgraaf,
800: C. Faber, and G. van der Geer, eds., Birkhauser, 1995, 335-368.
801: \bibitem[Kr]{kresch} {\sc A. Kresch}, {\em Cycle groups for Artin stacks},
802: Invent. Math. {\bf 138} (1999), no. 3, 495--536.
803: \bibitem[LR]{lr} {\sc A.-M. Li and Y. Ruan}, {\em Symplectic
804: surgery and Gromov-Witten invariants of Calabi-Yau 3-folds I},
805: preprint 1998, math.AG/9803036.
806: \bibitem[LZZ]{lzz} {\sc A.-M. Li, G. Zhao, and Q. Zheng},
807: {\em The number of ramified covering of a Riemann surface
808: by Riemann surface}, preprint 1999, math.AG/9906053 v2.
809: \bibitem[P]{p} {\sc R. Pandharipande}, personal communication.
810: \bibitem[M]{m} {\sc D. Mumford}, {\em Towards an enumerative
811: geometry of the moduli space of curves}, in {\em Arithmetic
812: and Geometry} (M. Artin and J. Tate, eds.), Part II, Birkh\"{a}user,
813: 1983, 271--328.
814: \bibitem[V1]{g01pn} {\sc R. Vakil}, {\em The enumerative geometry
815: of rational and elliptic curves in projective space}, submitted for publication,
816: available at http://www-math.mit.edu/\~{}vakil/preprints.html, rewritten version of
817: much of math.AG/9709007. \cited
818: \bibitem[V2]{g1} {\sc R. Vakil}, {\em Recursions for characteristic
819: numbers of genus one plane curves}, available at
820: http://www-math.mit.edu/\~{}vakil/preprints.html, to appear in Arkiv f\"{o}r Matematik.
821: \cited
822: \bibitem[V3]{p1} {\sc R. Vakil}, {\em Recursions, formulas, and graph-theoretic
823: interpretations of ramified coverings of the sphere by surfaces of genus 0 and 1},
824: submitted for publication, math.CO/9812105.
825: \end{thebibliography}
826: \end{document}
827:
828:
829: