math0202144/ams.tex
1: 
2: \documentclass[twoside,final,reqno]{amsart}
3: 
4: \usepackage{amsfonts}
5: \usepackage{amssymb}
6: \usepackage{latexsym}
7: \usepackage{verbatim}
8: \usepackage{epsfig}
9: 
10: \renewcommand{\subjclassname}{\textup{2000} Mathematics Subject Classification}
11: 
12: % THEOREM Environments ---------------------------------------------------
13: \newtheorem{theorem}{Theorem}[section]
14: \newtheorem{lemma}[theorem]{Lemma}
15: \newtheorem{corollary}[theorem]{Corollary}
16: \newtheorem{proposition}[theorem]{Proposition}
17: \newtheorem{claim}[theorem]{Claim}
18: \newtheorem{conjecture}[theorem]{Conjecture}
19: \theoremstyle{definition}
20: \newtheorem{remark}[theorem]{Remark}
21: \newtheorem{remarks}[theorem]{Remarks}
22: \newtheorem{example}[theorem]{Example}
23: \newtheorem{definition}[theorem]{Definition}
24: \newtheorem{problem}[theorem]{Problem}
25: \newtheorem{note}[theorem]{Note}
26: 
27: \newenvironment{Proof}{\removelastskip\par\medskip
28: \noindent{\em Proof.} \rm}{\penalty-20\null\hfill$\square$\par\medbreak}
29: 
30: 
31: 
32: %--- A number of `math-words'
33: 
34: \newcommand{\alg}{{\rm{alg}}}
35: \newcommand{\var}{\mathop{{\rm{var}}}}
36: \newcommand{\iso}{{\rm{iso}}}
37: \newcommand{\reg}{{\rm{reg}}}
38: \renewcommand{\r}{{\rm{rhd\hspace{2pt}}}}
39: \newcommand{\rHd}{{\rm{rHd\hspace{2pt}}}}
40: \newcommand{\h}{{\rm{ht}}}
41: \newcommand{\homeo}{{\rm{homeo}}}
42: \newcommand{\diffeo}{{\rm{diffeo}}}
43: \newcommand{\diff}{{\rm{diff}}}
44: \newcommand{\inter}{{\rm{int}}}
45: \newcommand{\gen}{{\rm{gen}}}
46: \newcommand{\pol}{{\rm{pol}}}
47: \newcommand{\Diff}{\mathop{{\rm{Diff}}}\nolimits}
48: \newcommand{\Aut}{\mathop{{\rm{Aut}}}\nolimits}
49: \newcommand{\sing}{{\rm{sing}}}
50: \newcommand{\Sing}{{\rm{Sing\hspace{2pt}}}}
51: \newcommand{\Crit}{{\rm{Crit\hspace{2pt}}}}
52: \newcommand{\Disc}{{\rm{Disc\hspace{2pt}}}}
53: \newcommand{\Pol}{{\rm{Pol\hspace{1pt}}}}
54: \renewcommand{\th}{{\rm{th}}}
55: \newcommand{\id}{{\rm{id}}}
56: \newcommand{\Jac}{{\rm{Jac}}}
57: \newcommand{\Id}{{\rm{Id}}}
58: \newcommand{\im}{{\rm{Im}}}
59: \newcommand{\mult}{{\rm{mult}}}
60: \newcommand{\crt}{{\rm{Crt}}}
61: \newcommand{\Iso}{{\rm{Iso}}}
62: \newcommand{\cl}{{\rm{closure}}}
63: \newcommand{\grad}{\mathop{\rm{grad}}\nolimits}
64: \newcommand{\ity}{{\infty}}
65: \newcommand{\car}{{\rm{car}}}
66: \renewcommand{\d}{{\rm{d}}}
67: \newcommand{\e}{\varepsilon}
68: \newcommand{\fin}{\hspace*{\fill}$\square$\vspace*{2mm}}
69: 
70: 
71: % quite a number of \mathcal's
72: 
73: \newcommand{\cA}{{\mathcal A}}
74: \newcommand{\cB}{{\mathcal B}}
75: \newcommand{\cC}{{\mathcal C}}
76: \newcommand{\cD}{{\mathcal D}}
77: \newcommand{\cE}{{\mathcal E}}
78: \newcommand{\cF}{{\mathcal F}}
79: \newcommand{\cG}{{\mathcal G}}
80: \newcommand{\cH}{{\mathcal H}}
81: \newcommand{\cI}{{\mathcal I}}
82: \newcommand{\cJ}{{\mathcal J}}
83: \newcommand{\cK}{{\mathcal K}}
84: \newcommand{\cO}{{\mathcal O}}
85: \newcommand{\cP}{{\mathcal P}}
86: \newcommand{\cR}{{\mathcal R}}
87: \newcommand{\cS}{{\mathcal S}}
88: \newcommand{\cST}{{\mathcal S \mathcal T}}
89: \newcommand{\cT}{{\mathcal T}}
90: \newcommand{\cV}{{\mathcal V}}
91: \newcommand{\cW}{{\mathcal W}}
92: \newcommand{\cX}{{\mathcal X}}
93: \newcommand{\cN}{{\mathcal N}}
94: \newcommand{\cY}{{\mathcal Y}}
95: \newcommand{\cZ}{{\mathcal Z}}
96: 
97: % special
98: 
99: \renewcommand{\k}{{\mathbb K}}
100: \newcommand{\bR}{{\mathbb R}}
101: \newcommand{\bC}{{\mathbb C}}
102: \newcommand{\btC}{\mbox{\tiny\mathbb C}}
103: \newcommand{\bP}{{\mathbb P}}
104: \newcommand{\bN}{{\mathbb N}}
105: \newcommand{\bZ}{{\mathbb Z}}
106: \newcommand{\bX}{{\mathbb X}}
107: \newcommand{\bQ}{{\mathbb Q}}
108: \newcommand{\bG}{{\mathbb G}}
109: \newcommand{\bH}{{\mathbb H}}
110: \newcommand{\bY}{{\mathbb Y}}
111: 
112: 
113: %%% ----------------------------------------------------------------------
114: \begin{document}
115: 
116: \title[Singularities and Topology of Meromorphic Functions]
117:  {Singularities and Topology 
118:  of Meromorphic \\ Functions}
119: 
120: \author[Mihai Tib\u ar]{Mihai Tib\u ar}
121: 
122: \address{Math\' ematiques, UMR 8524 CNRS, 
123: Universit\'e des Sciences et Tech. de Lille,  
124:  59655 Villeneuve d'Ascq, France.}
125: 
126: \email{tibar@agat.univ-lille1.fr}
127: 
128: \thanks{Partially supported by the Newton Institute at Cambridge and by the European Commission and the Institute of Mathematics of the Romanian Academy under the EURROMMAT contract ICA1-CT-2000-70022.
129: }
130: 
131: \subjclass{Primary 32S50; Secondary 32A20, 32S30}
132: 
133: \keywords{vanishing cycles, singularities 
134: along the indeterminacy locus, topology of 
135: meromorphic functions.}
136: 
137: 
138: \begin{abstract}
139: We present several aspects of the ``topology of meromorphic functions'', which we conceive as a general theory which includes the topology of holomorphic functions, the topology of pencils on quasi-projective spaces and the topology of polynomial functions. 
140: \end{abstract}
141: 
142: 
143: \maketitle
144: 
145: 
146: \tableofcontents
147: \setcounter{section}{0}
148: \section{Introduction}
149: 
150: 
151: Milnor \cite{Mi} defined the basic ingredients for studying the topology of holomorphic germs of functions $f \colon (\bC^n,0)\to \bC$. 
152: One of the main motivations for Milnor's book concerns the link $K$ of an isolated singularity and its complement $S^{2n-1} \setminus K$: links which are exotic speres have been discovered by Hirzebruch \cite{Hi} and Brieskorn \cite{Bri}; in case $n=2$, the components of $K$ are iterated toric knots. The study of isolated singularities ever since revealed striking phenomena and established bridges between several branches of mathematics.
153: 
154: In another stream of research, there has been an increasing interest in the last decade
155:  for the study of the global topology of polynomial functions, especially
156:  in connection with the behaviour at infinity. This topic is closely related to the affine geometry and to dynamical systems on non-compact spaces.
157: 
158: This paper reports on how to extend the study from holomorphic germs to the class of {\em meromorphic functions}, local or global.  In the same time, polynomial functions $\bC^n \to \bC$ can be viewed as a special case of global meromorphic functions, as we explain futher on. A global meromorphic function defines a pencil of hypersurfaces, therefore our approach moreover yields a generalization of the theory of {\em Lefschetz pencils}. 
159:  
160:  
161:  Another motivation for the study of meromorphic functions is Arnold's approach to the classification of simple germs of 
162: meromorphic functions under certain equivalence relations \cite{Ar}. 
163: 
164: 
165: \begin{figure}[hbtp]\label{f:spec}
166: \begin{center}
167: \epsfxsize=7cm
168: \leavevmode
169: \epsffile{sfig.eps}
170: \end{center}
171: \caption{{\em 
172:  Specialization of topics}}
173: \label{f:1}   
174: \end{figure}
175: 
176: 
177:  Let us introduce the first definitions.
178: A meromorphic function, or pencil of hypersurfaces, on a compact complex analytic space $Y$, is a function $F \colon Y \dashrightarrow \bP^1$ defined as the ratio of two sections $P$ 
179: and $Q$ of a holomorphic line bundle over $Y$.  Then $F = P/Q$ is 
180:  a holomorphic function on $Y\setminus A$, where $A := \{ P=Q=0\}$ is the {\em  base locus} of the pencil (also called {\em  axis}, or {\em indeterminacy locus}).
181:  A germ of meromorphic function on a space germ is just the ratio of two holomorphic germs $f = p/q  \colon (\cY, y) \dashrightarrow \bP^1$. By definition, $f$ is equal to $f'=p'/q'$, as germs at $y$, if and only if there exists a holomorphic germ $u$ such that $u(y)\not= 0$ and that $p=up'$, $q=uq'$. Then $f$ is holomorphic on the germ at $y$ of the complement $\cY \setminus \cA$ of the axis $\cA = \{ p=q=0\}$.
182: 
183:  
184:   Meromorphic functions give rise to a new type of singularities, those occuring along the indeterminacy locus. To define them, we need to introduce some more objects attached to a meromorphic function.
185:   
186: \begin{definition}
187: Let $G := \{ (x, \tau)\in (Y\setminus A) \times \bP^1 \mid F(x) =\tau\}$
188: and let $\bY$ denote the analytic closure of $G$ in  $Y\times \bP^1$, namely:
189: \[ \bY =  \{ (x,[s:t])\in Y\times \bP^1 \mid tP(x) - sQ(x) =0\}.\]
190:   In case of a germ of meromorphic function at $(\cY,y)$, one similarly  defines space germs, which we denote by $(G,(y,\tau))$, resp. $(\bY,(y,\tau))$. See also Definition \ref{d:local}.
191: \end{definition}
192:   
193:  First note that $G$ is the graph of the restriction $F_{| Y\setminus A}$. Therefore 
194:  $Y\setminus A \simeq G$ embeds into $\bY$ and the projection
195:  $\pi : \bY \to \bP^1$ is an extension of the function $F_{| Y\setminus A}$.
196:   One may also say that $\bY \stackrel{\sigma}{\rightarrow} Y$ is a blow-up of $Y$ along the axis $A$, such that the meromorphic function $F \colon Y \dashrightarrow \bP^1$ pulls back to a well defined holomorphic function $\pi : \bY \to \bP^1$.
197:    
198: \begin{equation}\label{eq:blow}
199:  \begin{array}{ccc}
200: \bY & \  & \  \\
201:  \mbox{\tiny $\sigma$} \downarrow \ \ &  \ \  \ \  \searrow \mbox{\tiny $\pi$}& \\
202:  Y &  \stackrel{\mbox{\tiny $F$}}{\dashrightarrow} & \bP^1
203: \end{array}  
204: \end{equation}  
205: 
206:   We shall also consider the restriction of  $F$ (or of a germ $f$), to $X:= Y\setminus V$, where $V$ is some compact analytic subspace of $Y$. The case $V= \{ Q=0\}$ is of particular interest for the following reason.
207:   Let $P \colon \bC^n \to \bC$ be a polynomial of degree $d$, let $\tilde P$ be the homogenized of $P$ by the new variable $x_0$ and let $H^\ity =\{ x_0 =0\}$ be the hyperplane at infinity. Then 
208:   $\tilde P/x_0^d \colon \bP^n \dashrightarrow \bP^1$ is a meromorphic function on $Y := \bP^n$ which coincides with $P$ over $\bP^n\setminus H^\ity$. We shall briefly outline in \S \ref{s:poly} some results and literature on polynomials.
209: 
210:  Our meromorphic function (as a global one or as a germ) defines a family (=pencil) of hypersurfaces
211:   on each of the spaces defined above:  $\bY$, $Y$, $Y\setminus A$,  or $X=Y\setminus V$. The map to $\bP^1$, whenever defined, yields the pencil, as the family of its fibres. In other cases, we take the closures of fibres in the considered space. For instance, in case of $Y$, we take the closure
212:   of each hypersurface $F_{|Y\setminus A}^{-1}(\tau)$ within $Y$, for $\tau\in \bP^1$;
213:  each such closure contains $A$ and we actually have  $\overline{F_{|Y\setminus A}^{-1}(\tau)}= \pi^{-1}(\tau)$.
214:  The role of the ``completed space" $\bY$ is that it contains all these  pencils: we just restrict the fibres of $\pi$ to the particular  
215:  subspace of $Y$ and we get the pencil we are looking for.
216:  
217:  With this approach, one covers a large field. For instance, the class of holomorphic functions (or germs) represents the case when $A=\emptyset$. 
218: 
219: 
220:  Now, for any pencil on $Y$ or on $X$, there is only a finite number of atypical values, or atypical fibres. This finiteness result comes from ideas of Thom \cite{Th} and is based on the fact that we can stratify the space $Y$ (such that $V$ is union of strata, in case $V\not= \emptyset$), restrict to $Y\setminus A \simeq \bY\setminus (A\times \bP^1)$ and extend this to some Whitney stratification of $\bY$. 
221: 
222: %%%%%%%%%%%%%%%%%%%%
223:  
224:  In case of a germ at $(\cY,y)$ of a meromorphic function, one considers the germ of such a Whitney stratification at $\{y\}\times \bP^1 \subset \bY$. Local finiteness of the strata implies that non-transversality of the projection $\pi$ happens at discrete values only.
225: %%%%%%%%%%%%%%%%%%%
226:  
227: \begin{proposition}\label{p:1} 
228: There exists a finite set $\Lambda\subset\bP^1$ such that the map
229: $\pi : \bY \setminus \pi^{-1}(\Lambda) \to 
230: \bP^1\setminus \Lambda$ is a stratified
231: locally trivial {\rm C}$^0$ fibration.
232: 
233: In particular, the restrictions $\pi_| \colon \bY \setminus ((V\times \bP^1) \cup \pi^{-1}(\Lambda)) \to \bP^1 \setminus \Lambda$ and $F_|\colon
234: Y \setminus (V\cap A\cap F^{-1}(\Lambda)) \to \bP^1\setminus \Lambda$  
235:  are  stratified
236: locally trivial fibrations. 
237: \fin 
238: \end{proposition} 
239: %%%%%%%%%%%%%%%%%%%%%%
240: 
241: In our approach, the singularities of the meromorphic function $F$ along the indeterminacy locus $A$  
242: are the stratified singularities of $\pi$ at $(A\times \bP^1)\cap \bY$. Usually, singularities of functions on singular spaces
243: are defined with respect to some Whitney stratification. Here we use
244:  instead a {\em partial Thom stratification}, denoted $\cG$, as we already used in particular cases (cf. \cite{Ti-t,Ti-lef}, \cite{ST-x}). This is a more general type of stratification
245: since Whitney (b) condition is not required. Nevertheless, it allows one to study topological aspects, including homotopy type, at least for isolated singularities, in both local or global context.
246: 
247: Instead of endowing $\bY$ with a stratified structure, another strategy for studying the topology of the meromorphic function $F$
248: would be to further blow up $\bY$ in diagram (\ref{eq:blow}), such that the pull-back of $\{ P=0\} \cup \{ Q=0\}$ becomes a divisor with normal crossings. One may then use the data provided by this divisor in order to 
249: get informations. In this spirit, some results were found in the polynomial case, in two variables, by Fourrier \cite{Fo} and L\^e-Weber \cite{LW};  computation of the zeta functions of the monodromy has been done for polynomials and particular meromorphic germs (namely for $Y$ nonsingular and $X = Y\setminus A$) by Gusein-Zade, Melle and Luengo \cite{GLM, GLM-3, GLM-4}.
250: 
251: This paper revisits the techniques and results of \cite{Ti-t,Ti-lef}, \cite{ST-x} and extends them to the more general context introduced above. The main scope is to show how to study  
252: vanishing cycles of meromorphic functions in both 
253: local and global context. 
254: 
255: %%%%%%%%%%%%%%%%%%%%%%
256: 
257: %%%%%%%%%%%%%%%%%%%%%
258: %%%%%%%%%%%%%%%%%%%%%%%%
259: 
260: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
261: %%%%%%%%%%%%%%%%%%%%%%%%%
262: \section{Vanishing homology and singularities}\label{s:van}
263: %%%%%%%%%%%%%%%%%%%%%
264: %%%%%%%%%%%%%%%%%%%%%%%%
265: 
266: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
267: %%%%%%%%%%%%%%%%%%%%%%%%%
268: 
269: Let us first define vanishing homology attached to a global meromorphic function and relate it to the singularities along the indeterminacy locus.
270: The vanishing homology is important in detecting and controlling (whenever possible) the change of topology of the fibres.
271:   
272: We shall use the following notations. For any subset $W\subset \bP^1$,  $\bY_W := \pi^{-1}(W)$, 
273: $Y_W :=  A \cup F^{-1}(W)$, $X_W := X\cap Y_W$.
274:  The special case $X= Y\setminus A$ and this is the object of study, explicitly, in \cite{ST-x}, and implicitly, in \cite{Ti-t}. Our presentation follows the one of \cite{ST-x}, adapting it to our more general situation; in particular our notations are different 
275:  from those in \cite{ST-x}.
276:  
277:  Let $a_i\in \Lambda$ be an atypical value of $\pi$ and take a small enough disc $D_i$ at $a_i$ such that $D_i\cap \Lambda = \{ a_i \}$. 
278:  Let's fix some point $s_i\in\partial D_i$. Let $s \in \bP^1\setminus \Lambda$ be a general value, situated on the boundary of some  big closed disc $D \subset \bP^1$, such that $D \supset D_i$, $\forall a_i\in D \cap \Lambda$, and that $D \cap D_i =\emptyset$, $\forall a_i\in D \setminus \Lambda$. 
279:  
280:  The vanishing homology of meromorphic functions should be a natural
281: extension of the vanishing homology of local holomorphic functions. In the latter case, the total space of the Milnor fibration \cite{Mi} is contractible, by the local conical structure of analytic sets \cite{BV}. For global meromorphic functions, the total space one may take cannot be contractible anymore and the general fibre $X_s$ inherits 
282: part of its homology. 
283: 
284: %%%%%%%%%%%%%%%%%%%
285: \begin{definition}\label{d:vanhom} 
286: The  {\em vanishing homology of $F_{|X}$ at $a_i$} is 
287: the relative homology \\
288: $H_*(X_{D_i}, X_{s_i})$.
289: \end{definition}
290: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 
291:  In the case $X = Y\setminus A$, this corresponds to the definition 
292:  used by Siersma and the author in \cite{ST-x}.
293:  
294: We identify $X_s$ to $X_{s_i}$, in the following explicit manner.  
295: For each $i$, take a path $\gamma_i \subset D$ from $s$ 
296: to $s_i$, with the usual conditions: the 
297: path $\gamma_i$ has no self intersections and does not intersect any 
298: other path $\gamma_j$, except at the point $s$.
299:  Then Proposition \ref{p:1} allows identifying $X_s$ to $X_{s_i}$, 
300:  by parallel transport along $\gamma_i$.
301: 
302: 
303:  A general result tells that vanishing homologies can be ``patched" together.
304:  This type of result was observed before in different particular situations, see e.g. \cite[\S 5]{Br}, \cite{Si}. More precisely, we have the following result, extending the context of \cite[Proposition 2.1]{ST-x} to any $X= Y\setminus V$:
305: %%%%%%%%%%%%%%%%%%%%%%%%
306:  %%%%%%%%%%%%%%%%%%%%%%%
307: \begin{proposition}\label{p:basic} \
308: \begin{enumerate}
309: \item  $ H_*(X_{D}, X_s) = \oplus_{a_i\in\Lambda} H_*(X_{D_i}, X_{s_i}).$
310: \item The long exact sequence of the triple $(X_{D}, X_{D_i}, X_s)$
311: decomposes into short exact sequences which split:
312: \begin{equation}\label{eq:exact} 
313: 0\to H_*(X_{D_i}, X_{s_i}) \to H_*(X_{D},X_s) \to 
314: H_*(X_{D}, X_{D_i}) \to 0 .
315: \end{equation}
316:  \item There is a natural identification $H_*(X_{D},X_{D_i}) = 
317: \oplus_{a_j\in\Lambda, j\not= i} H_*(X_{D_j}, X_s)$. 
318: \end{enumerate}
319: \end{proposition}
320: \begin{Proof}
321:   By Proposition \ref{p:1}, 
322: the fibration $F_|: \bY \setminus ((v\times \bP^1) \cup \pi^{-1}(\Lambda) \to \bP^1 \setminus 
323: \Lambda$ is locally trivial. Its fibre over some $b$ is, by definition, $X_b$. We then get a sequence of excisions:
324: \[\oplus_{a_i\in\Lambda} H_*(X_{D_i}, X_{s_i})  \stackrel{\simeq}{\longrightarrow}
325: H_*( \pi^{-1}(\cup_{a_i\in \Lambda} D_i\cup\gamma_i), X_s) 
326: \stackrel{\simeq}{\longrightarrow} H_*(X_{D}, X_s).
327: \]
328: This also shows that each inclusion $(X_{D_i}, X_{s_i}) \subset 
329: (X_{D},X_{s_i})$ 
330: induces an injection in homology $H_*(X_{D_i}, X_{s_i}) \hookrightarrow H_*(X_{D}, X_s)$.
331:  The points (a), (b), (c) all follow from this. 
332: \end{Proof}
333: 
334: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
335: Vanishing homology of $F$ has its local counterpart and is closely related to singularities of $F$. One would like to say that vanishing homology is supported at the singular points of $F$. The typical problem for a meromorphic function is that it has singularities also outside the ground space, $Y$ or $X$. Therefore we need a larger space, such as $\bY$, to define singularities. Then the support of vanishing cycles is included into the singular locus of $\pi$ on $\bY$. (In cohomology, the sheaf of vanishing cycles of a function $h$ on a nonsingular space is indeed supported by the singular locus of $h$, see \cite{De}.) 
336: 
337: Let now give the precise definition of what we consider as singularities of $F$ (resp. of $f$). We relax the stratification conditions at $A\times \bP^1 \subset \bY$ and use only the Thom condition. Let us first recall the latter, following \cite{Ti-t}.
338: %%%%%%%%%%%%%
339: 
340: Let $\cG = \{ \cG_\alpha\}_{\alpha \in S}$ be a locally finite 
341: stratification such that $\bY \setminus X$ is union of strata.
342: Let $\xi :=(y,a)$ be a point on a stratum $\cG_\alpha$.
343: We assume, without loss of generality, that $a\not= [1:0]$. 
344: Let $f =p/q$ be our meromorphic germ on $(\cY, y)$ or a local representative of the germ of $F$ at $(Y, y)$. Then $q=0$ is a local equation for $A\times \bP^1$ at $\xi$. The {\em 
345: Thom regularity condition} (a$_q$) at  $\xi \in \cG_\alpha$
346: is satisfied (see e.g. 
347: \cite[ch. I]{GWPL}) if for any stratum $\cG_\beta$ such that $\cG_\alpha \subset \bar \cG_\beta$, the relative conormal space (see \cite{Te-2}, \cite{HMS} for a 
348: definition) of $q$ on $\bar \cG_\beta$
349: is included into the conormal of $\cG_\alpha$, locally at $\xi$, i.e.,  $(T_{\cG_\alpha}^*)_\xi \supset (T^*_{q| \overline{ \cG_\beta}})_\xi$.
350: This condition is known to be independent on $q$, up to 
351: multiplication by 
352: a unit \cite[Prop. 3.2]{Ti-t}. 
353: 
354: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
355: 
356: \begin{definition}\label{d:thom} 
357: Let $\cG$ be a stratification on $\bY$ as above such that it restricts to a Whitney stratification on $Y\setminus A$, where $V\setminus A$ is union of strata. We say that $\cG$ is a {\em partial 
358: Thom stratification} ($\partial \tau$-stratification) if Thom's condition (a$_q$) is satisfied at any point $\xi\in A\times \bP^1 \subset \bY$.
359: \end{definition}
360: 
361: %%%%%%%%%%%%
362: One may extend a Whitney stratification on $Y\setminus A$ to a locally finite $\partial \tau$-stratification of $\bY$, by usual stratification theory arguments (see e.g. 
363: \cite{GWPL}). For instance, the Whitney stratification $\cW$ of $\bX$ that we have considered before
364: is an example of $\partial 
365: \tau$-stratification. This follows from  \cite[Th\'eor\`eme 
366: 4.2.1]{BMM} or \cite[Theorem 3.9]{Ti-t}.
367: One can also construct a canonical (minimal) $\partial \tau$-stratification;
368: we send to \cite{Ti-t} for further details.
369: \hyphenation{stra-ti-fi-cation}
370:   
371: %%%%%%%%%%%%%%%%%%
372: 
373: \begin{definition}\label{d:sing}
374: Let $\cG$ be a $\partial \tau$-stratification on $\bY$. We say that the following closed subset of $\bY$:
375: \[ \Sing_{\cG}F := \cup_{
376:  \cG_\alpha \in \cG }
377:  \cl(\Sing \pi_{|\cG_\alpha} )\]
378:  is the singular locus of $F$ with respect to $\cG$.
379: We say that $F$ has {\em isolated singularities} with respect to $\cG$ 
380: if $\dim \Sing_\cG F \le 0$. 
381: \end{definition}
382: 
383: %%%%%%%%%%%%%%%%%%%%%%%%%%%
384: For the singular locus of a germ $f$, one modifies this definition accordingly.
385:   The singularities of the new type are those along the indeterminacy locus, namely on  $A\times \bP^1$. We shall further investigate the relation between singularities and 
386:   vanishing homology, in case of isolated singularities.
387:  %%%%%%%%%%%%%%
388: Let us define the local fibration of a meromorphic germ, already used in particular cases in \cite{ST}, \cite{Ti-t}, \cite{ST-x}), and which relates to the fibration of a holomorphic germ on a singular space defined by L\^e D.T. \cite{Le-oslo}. 
389: 
390: %%%%%%%%%%%%%%%%%%%%%%
391: 
392: \begin{definition}\label{d:local}
393: Let $f : ( \cY, y) \dashrightarrow \bP^1$ be a germ of a meromorphic 
394: function. For every $a\in \bP^1$, one associates the germ $\pi_{(y,a)} : (\bY , (y, a)) \to \bP^1$. Abusing language, by restricting this map to $X\subset \bY$, we have a germ  $f: (X , (y, 
395: a)) \to \bC$, where the point 
396: $(y, a)$ might be in the closure of the set $X$.
397: \end{definition}
398: 
399: When the point $y$ does not belong to the axis $\cA$, then we have the classical situation of a holomorphic germ; the point $(y,a)$ is uniquely determined by $y$. However, when $x\in \cA$, then for each $a\in \bP^1$ we get a different germ. Proposition \ref{p:1} together with Thom's Second Isotopy Lemma show that, in this family, all germs are isomorphic except of finitely many of them.
400: 
401: %%%%%%%%%%%%%%%%%%%%
402: There is a well defined local fibration at $(y,a)$, as follows.
403: Let $\cW$ be a Whitney stratification  of $\bY$ such that $\bY\setminus X$
404: is union of strata. For all small enough radii $\e$, the sphere $S_\e = 
405: \partial \bar B_\e(y,a)$ centered at $(y,a)$ intersects transversally all the (finitely many strata) in some neighbourhood of $(y,a)$. By \cite{Le-oslo}, the projection $\pi : \bY_{D} \cap B_\e(y,a)  \to D$ is stratified locally trivial over $D^*$, if the radius of $D$ is small enough.
406: It follows that the restriction:
407: \begin{equation}\label{eq:2}
408: \pi : X_{D^*} \cap B_\e(y,a)  \to D^*.
409: \end{equation} 
410: 
411: is also locally trivial. If $y$ is fixed, this fibration varies with the parametre $a$; the radius $\e$ of the ball depends also on the point $a$.
412: From Proposition \ref{p:1} and Thom's Isotopy Lemma it follows that, since $\pi$ is 
413: stratified-transversal to $\bY$ over $\bP^1 \setminus \Lambda$, the 
414: fibration  
415: $\pi : X_D \cap B_\e (y,a) \to D$ is trivial, for all but a finite number of values of $a\in \bP^1$.
416: %%%%%%%%%%%%
417: 
418: \begin{definition}
419: We call the locally trivial fibration (\ref{eq:2})  the 
420: {\em Milnor-L\^e fibration}  of the meromorphic function germ $f$ {\em at the 
421: point $(y, a) \in \bX$}.
422: \end{definition}
423: 
424: 
425: %%%%%%%%%%%%%%%%%%%%
426: %%%%%%%%%%%%%%%%%%%%%%%%%
427: 
428: \section{Isolated singularities and their vanishing cycles}\label{count} 
429: %%%%%%%%%%%%%%%%
430: %%%%%%%%%%%%%%%%
431: %%%%%%%%%%%%%%%%%%%%
432: %%%%%%%%%%%%%%%%%%%%%%%%%
433: 
434: We show that, if the singularities along the indeterminacy locus
435: are isolated, then
436: one can localize the variation of topology of fibres.
437: The same type of phenomenon exists in the previously known cases:
438:  holomorphic germs \cite{Mi} and of polynomial functions (e.g. \cite[4.3]{Ti-t}).
439: This has consequences on the problem of detecting variation of topology, especially when the underlying space $Y$ has maximal {\em rectified homotopical depth}\footnote{This notion was introduced by Hamm and L\^e \cite{HL}
440: in order to realize Grothendieck's predictions that homotopical depth was the cornerstone for the Lefschetz type theorems on singular spaces \cite{Gr}. We shall come back to Lefschetz theory in \S \ref{s:lef}.}. Before stating the localization result, let us give the definition here, for further use:  Let $Z$ be a complex space endowed with some Whitney stratification $\cW$; denote by $\cW_i$ the union of strata of dimension $\le i$. After \cite{HL}, one says that $\r_\cW Z \ge m$ if for any $i$ and any point $x\in\cW_i\setminus \cW_{i-1}$, the homotopy groups of $(U_\alpha, U_\alpha \setminus \cW_i)$ are trivial up to the order $m-1-i$, where  $\{U_\alpha\}$ is some fundamental system of neighbourhoods of $x$. It is shown in {\em loc.cit.} that this doest not depend on the chosen Whitney stratification. A similar definition holds in homology instead of homotopy, giving rise to the rectified homological depth, denoted $\rHd$. Let us mention  that $\r Y \ge n$ when $Y$ is  locally a complete intersection of dimension $n$ at all its points (see \cite{LM}).
441:  
442: %%%%%%%%%%%%%%%%%%%%%%%%%%%
443: \begin{proposition}\label{t:loc} 
444: Let $F$ have isolated singularities with respect to some $\partial \tau$-stratifi- \\
445: cation $\cG$ at $a\in 
446: \bP^1$ (i.e. $\dim \bY_a \cap \Sing_\cG F \le 0$).
447: Then the variation of topology of the fibres of $F$ at $X_a$ is
448: localizable at those points.
449: \fin
450: \end{proposition}
451: %%%%%%%%%%
452: What we mean by  ``localizable" is that there exist small enough balls in $\bY$ centered at the isolated singularities such that, outside these balls, the projection $\pi$ is trivial over a small enough disc centered at $a\in \bP^1$. The proof in the general case of meromorphic functions has the same structure as the proof presented  in \cite[Theorem 4.3]{Ti-t}, inspite the fact that in {\em loc.cit.} we consider a particular situation; we may safely skip it. 
453: %%%%%%%%%%%%%%%%%%%%%%%%%%%
454: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
455: The localization result implies, as in the particular cases \cite{Ti-t}, \cite{ST-x}, that the vanishing cycles are 
456: concentrated at the isolated singularities.
457: 
458: \begin{corollary}\label{c:conc}
459: Let $F$ have isolated singularities with respect to $\cG$ at $a\in \bP^1$ and let $\bY_a \cap \Sing_\cG F =\{ p_1, \ldots, p_k\}$. For any small enough balls $B_i\subset \bY$ centered 
460: at $p_i$, and for small enough closed disc centered at $a$, $s\in 
461: \partial 
462: D$,  we have:
463: \begin{equation}\label{eq:local}
464:  H_*(X_D, X_s)\simeq \oplus_{i=1}^k 
465: H_*(X_D\cap B_i, X_s\cap B_i) .
466: \end{equation}
467: \fin
468: \end{corollary}
469: %%%%%%%%%%%%%%%%%%%%%%%%%%
470:  One show, following \cite{ST-x}, that an isolated $\cG$-singularity at a point of 
471: $A\times \bP^1\subset \bY$ is detectable by the presence of a certain local polar locus. If the space $Y$ is ``nice enough", then the local vanishing homology (second term of the isomorphism (\ref{eq:local})) is concentrated in one dimension only. Then the polar locus defines a numerical invariant which measures the number of ``vanishing cycles at this point". 
472: %%%%%%%%%%%%%%%
473: \begin{definition}\label{d:polar}
474:  Let $\xi= (y,a)\in A\times \bP^1$ and let $f = p/q$ a local
475:  representation of $F$ at $y$. Then
476: the {\em polar locus} $\Gamma_\xi(\pi,q)$ is the germ at $\xi$ of the 
477: space: 
478:  \[ \cl \{ \Sing_\cG(\pi, q) \setminus (\Sing_\cG \pi \cup A\times \bP^1)\} \subset 
479: \bY .\]
480:  \end{definition}
481: 
482: As in \cite[\S 4]{ST-x}, we have the 
483: isomorphisms:
484: \[ \Gamma_\xi(\pi,q) \simeq \Gamma_\xi(f,q) \simeq \Gamma_\xi(p,q).\]
485:  
486:  The multiplication by a unit $u$ may change the polar locus:
487: $\Gamma_\xi(\pi,qu)$ is in general different from $\Gamma_\xi(\pi,q)$. 
488: Nevertheless, 
489: we have the following.
490: %%%%%%%%%%%%%%%%%%%% 
491: \begin{proposition}\label{p:inv}
492: Let $\xi= (y,a)\in A\times \bP^1$ be an isolated $\cG$-singularity
493: of $F$ and let $f_y = p/q$ a local representation of $F$. 
494: Then, for any multiplicative unit $u$, the polar locus 
495: $\Gamma_\xi(\pi,qu)$ is either void or $\dim \Gamma_\xi(\pi,qu) =1$.
496: If moreover $Y$ is of pure dimension $m$ and $\r (Y \setminus \{ q=0\}) \ge m$ in the neighbourhood of $\xi$, then  the intersection multiplicity $\inter_\xi 
497: (\Gamma_\xi(\pi,qu), \bY_a )$ is independent on the  
498: unit $u$. We call this multiplicity the {\em polar number} at $\xi$. 
499: \end{proposition}
500: \begin{Proof}
501: We follow essentially the proof of \cite[4.2]{ST-x}. The key argument to use is the independence of $\bP T^*_{qu}$ from the multiplicative unit 
502: $u$, proven in \cite[Prop. 3.2]{Ti-t}, where $\bP T^*_{q}$ denotes the 
503: projectivized relative conormal of $q$.
504: 
505: Since $\dim \bP T^*_q = m+1$, it follows that $\Gamma_\xi(\pi, q)$ is either void or of dimension at least $1$. On the other hand, since $\xi$ is a point belonging to $\Gamma_\xi(\pi, q)$, it follows that 
506: $\Gamma_\xi(\pi, q)$ has dimension at most $1$. The same argumentation holds for $qu$ instead of $q$. This proves the first claim.
507: 
508: To prove the second statement, let's suppose that $\Gamma_\xi(\pi, q)$ has dimension $1$. Consider the Milnor-L\^e fibration 
509:  of the function $\pi$ at $\xi$:
510: \begin{equation}\label{eq:le}
511: \pi_| :  \bY_{D^*}\cap B_\e (\xi)\to D^* ,
512: \end{equation} 
513: as explained at the end of \S \ref{s:van}.
514: We compute the homology $H_*(\bY_s \cap B)$ of the Milnor fibre of 
515: this fibration. Inside $B$, the restriction  of the 
516: function $q$ to $\bY_s\cap B$ has a finite number of stratified isolated singularities, which are precisely the points of intersection $\bY_s \cap B \cap \Gamma(\pi,q)$.  By the result due to Goresky-MacPherson that cylindrical neighbourhoods are conical \cite[pag. 165]{GM},
517: it follows that $\bY_s \cap B$ 
518: is homotopy equivalent to $\bY_s \cap B\cap q^{-1}(\delta)$, where $\delta$ is a small disc at $0\in \bC$ such that $\bY_s \cap B \cap \Gamma(\pi, q) = q^{-1}(\delta)\cap 
519: \bY_s  \cap B \cap \Gamma(\pi, q)$.
520: %%%%%%%%%%%%%%%%%%%%
521: 
522:  Let now take a small enough disc $\hat\delta$ centered at $0\in \bC$ such that $q^{-1}(\hat \delta)\cap 
523: \bY_s  \cap B  \cap \Gamma(\pi, q) =\emptyset$. By using the properties of the partial Thom stratification $\cG$ and by retraction, it follows that  $q^{-1}(\hat \delta) \cap 
524: \bY_s  \cap B$ is homotopy equivalent to the central fibre  $q^{-1}(0) \cap \bY_s \cap B$.  
525: 
526: Since the subspace $q^{-1}(0)$ at $\xi$ is the product $A\times \bP^1$,
527: the central fibre   
528: $q^{-1}(0) \cap \bY_s \cap B$ is contractible; hence $q^{-1}(\hat \delta) \cap \bY_s  \cap B$ is contractible too.
529: 
530: 
531: The total space $\bY_s \cap B \stackrel{\h}{\simeq} q^{-1}(\delta)\cap\bY_s \cap B$ is built by attaching 
532: to the space $q^{-1}(\hat \delta)\cap 
533: \bY_s  \cap B$  finitely many cells, which come from the isolated singularities of the function $q$ on $q^{-1} 
534: (\delta \setminus \hat\delta)\cap \bY_s \cap B$. Since $\r (Y_s \setminus \{ q=0\}) \ge m-1$ in some neighbourhood of $\xi$ (by our assumption and Hamm-L\^e's \cite[Theorem 3.2.1]{HL}), it follows that each singularity contributes with cells of dimension exactly $m-1$ and the number of cells is equal to the corresponding local Milnor number of 
535: the function $q$ (see \cite{Le-s} and \cite{Ti-b} for more details).
536:  The sum of these numbers is, by definition, the intersection multiplicity $\inter_\xi(\Gamma(\pi, q) , \bY_a)$.
537: 
538: By this we have proven that 
539:  $\dim H_{m-1}(\bY_s \cap B) = \inter ( \Gamma_\xi(\pi , q), \bY_a)$ and that $\tilde H_i(\bY_s \cap B) = 0$ for $i\not= n-1$. 
540:  Replacing $q$ by $qu$ in our proof yields the same equalities; this proves our statement. 
541: \end{Proof} 
542: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
543: \begin{remark}\label{r:spheres}
544: The last part of the above proof shows in fact more, that the fibre $\bY_s \cap B$ of the local fibration (\ref{eq:le})  is homotopy 
545: equivalent to a ball to 
546: which one attaches a certain number of $(m-1)$-cells. Therefore $\bY_s \cap B$ is homotopy equivalent to a bouquet of spheres $\bigvee S^{m-1}$ of dimension $m-1$.
547: \end{remark}
548: %%%%%%%%%%%%%%%%%%%%%
549: 
550: One can get more precise results when lowering the generality. The situations we consider in the following are more general that ``polynomial functions'', which we consider in \S \ref{s:poly}. Let then assume:
551: \vspace*{2mm}
552: 
553: \noindent $(*)$  $X := Y\setminus A$ has at most isolated singularities.
554: 
555: \vspace*{2mm}
556: %%%%%%%%%%%%%%%%%%%%%
557: In this case we have $\Sing_\cG F \cap A\times \bP^1 \subset \Sing \bY \cap A\times \bP^1$. In the notations of Cor. \ref{c:conc}, 
558: the following duality result holds (integral coefficients):
559: \[ 
560: H_*(X_D\cap B_i, X_s\cap B_i) \simeq H^{2m -*}(\bY_D \cap B_i , \bY_s \cap B_i),\]
561: where $m= \dim_{p_i}\bY$. This follows from Lefschetz duality for polyhedra, see Dold \cite[p. 296]{Do}.
562: 
563: Since $\bY_D \cap B_i$ is contractible, we get:
564: \begin{equation}\label{eq:dual}
565:  H_*(X_D\cap B_i, X_s\cap B_i) \simeq \tilde H^{2m -1 -*}(\bY_s \cap B_i) .
566: \end{equation}
567: When comparing this to Remark \ref{r:spheres} and to Definition \ref{d:vanhom}, Corollary \ref{c:conc} and Proposition  \ref{p:inv}, the following statement comes out:
568: %%%%%%%%%%%%%%%%%%%%
569: \begin{theorem}\cite{ST-x}
570: Let $F$ have isolated singularities with respect to $\cG$ at $a\in\bP^1$ and let $\xi \in A\times \{ a\} \cap \Sing_\cG F$. Then $F_{|Y\setminus A}$ has vanishing cycles at $\xi$ if and only if $\inter (\Gamma_\xi(\pi, q), \bY_a) \not= 0$.
571: The number of vanishing cycles at $\xi$ is $\lambda_\xi := \dim H_{m-1} ( \bY_s \cap B) = \inter (\Gamma_\xi(\pi, q), \bY_a)
572: $, where $m = \dim_\xi \bY_a$.
573: \fin
574: \end{theorem}
575: %%%%%%%%%%%%%%%%%%%%
576: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 
577: 
578:  Let us assume now that $\bY_a$ has an isolated singularity at $\xi \in A\times \{ a\}$. This happens for instance when $Y$ has at most isolated singularities on $A$. Since $(\bY_a, \xi)$ is a hypersurface germ in $Y$, it has a well defined Milnor fibration, and in particular a Milnor number, which we denote by  $\mu(a)$.
579:  
580:  It also follows that $\dim_\xi \Sing_\cG F \le 0$ and that $\bY$ has 
581: singularities of dimension at most 1 at $\xi$. If $\Sing \bY$ is a curve at $\xi$, this curve intersects $\bY_s$, for $s$ close enough to $a$, at some points $\xi_i(s)$, $1\le i\le k$. There is a well defined Milnor $\mu_i(s)$ at each hypersurface germ 
582: $(\bY_s, \xi_i(s))$. (In case $\Sing \bY$ is just the point $\xi$, we consider that $\mu_i(s)=0$, $\forall i$.)
583: We have the following computation of the number of vanishing cycles at $\xi$:
584: %%%%%%%%%%%
585: 
586: \begin{theorem}\label{t:numbers}\cite{ST-x}
587: Let $\dim_\xi \Sing \bY_a =0$. Then:
588:  \[ \lambda_\xi = \mu(a) - \sum_{i=1}^{k} \mu_i(s) .\]
589: \end{theorem}
590: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
591: \begin{Proof}
592:  We only sketch the proof and send to \cite{ST-x} for details. Consider the function:
593:  \[ G = p-sq : (Y\times \bC, \xi_i(s)) \to (\bC,0).\]
594:  For fixed $s$, this function is a smoothing of the germ $(\bY_s,\xi_i(s))$. Since $\xi$ is an isolated singularity, the polar locus at $\xi$ of the map $(G,\pi) : Y\times \bC \to \bC^2$, defined as $\Gamma_\xi(G,\pi) = \cl \{ \Sing (G,\pi) \setminus \Sing G\}$,
595: is a curve. By using polydisc neighbourhoods\footnote{ polydisc neighbourhoods were first used by L\^e D.T.\cite{Le-oslo}.} $(P_\alpha \times D_\alpha)$ at $\xi$ in $Y\times \bC$, one may show that $(G,\pi)^{-1}(\eta,s) \cap (P_\alpha \times D_\alpha)$ is homotopy 
596: equivalent to the Milnor fibre of the germ $(\bY_a, \xi)$. To obtain from this the space $\pi^{-1}(s)\cap (P_\alpha \times D_\alpha)$, one has to attach a number of $m$ cells, where $m$ denotes $\dim_\xi Y$. Part of these cells come from the singular points $\xi_i(s) \in \Sing G \cap \pi^{-1}(s)$: by definition, their total number is $\sum_{i=1}^{k} \mu_i(s)$. The other part of the cells come from the intersection with $\Gamma_\xi(G,\pi)$ and their number is $r= \inter(\Gamma_\xi(G,\pi), \pi{-1}(0))$.
597: The key observation is that $r$ turns out to be equal to $\dim H_{m-1}(\bY_s \cap B)$, which is the number of vanishing cycles at $\xi$.
598: Finally, since $\pi^{-1}(s)\cap (P_\alpha \times D_\alpha)$ is contractible (since being the Milnor fibre 
599: of a linear function $t$ on a smooth space), we have the following equality:
600: 
601:  \[  \mu(a) = r + \sum_{i=1}^k \mu_i(s).\]
602: Notice that, in case $\dim_\xi \Sing \bY =0$, we get just $\lambda_\xi = \mu(a)$.
603: \end{Proof}
604: We send to Corollary \ref{c:num} for the counting of the total number of vanishing cycles. 
605: %%%%%%%%%%%%%%%%%%%%%%%%% 
606: Let us give two examples of meromorphic functions, one on $\bP^2(\bC)$ and another on a nonsingular quadratic surface 
607: in  $\bP^3(\bC)$. 
608: 
609: \begin{example}\label{ex:1}(\cite{ST-x})
610: Let $F \colon Y=\bP^2 \dashrightarrow \bP^1$,   $\displaystyle F= \frac{x(z^{a+b} + 
611: x^ay^b)}{y^pz^q}$, where $a+b+1 = p+q$ and $a,b,p,q \ge 1$. 
612:  For some $s\in \bC = \bP^1\setminus [1:0]$, the space $\bY_s$ is given by:
613:  \begin{equation}\label{eq:1}
614:  x(z^{a+b} + x^ay^b) = s y^pz^q
615:  \end{equation}
616:  $\Sing \bY \cap (Y\times \bC)$ consists of three lines: $\{ [1:0:0], [0:1:0], [0:0:1]\} 
617: \times \bC$. We are under the assumptions of Theorem \ref{t:numbers} and we inspect each of these 3 families of germs with isolated singularity to see where the Milnor number jumps.
618:  
619:  Along $[1:0:0]\times \bC$, in chart $x=1$, there are no jumps, since the germs have uniform Brieskorn type $(b, a+b)$.
620:  Along $[0:0:1]$, in chart $z=1$, there are no jumps, since the type is constant $A_0$, for all $s$.
621:  Along $[0:1:0]$, in chart $y=1$: For $s\not= 0$, the Brieskorn type is $(a+1, 
622: q)$, with $\mu(s) = a(q-1)$. If $s=0$, then we have $x^{a+1} + xz^{a+b}=0$ 
623: with $\mu(0) = a^2 + ab + b$.
624: 
625:  There is only one jump, at $\xi = ([0:1:0], 0)$; according to the preceding theorem,
626: $\lambda_\xi = a^2 + ab + b - a(q-1) = b +ap$.
627:  
628: \end{example}
629: 
630: %%%%%%%%%%%%%%%%%%%%%%%%%
631: 
632: 
633: 
634: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
635: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
636: %%%%%%%%%%%%%%%%%%%%%
637: %%%%%%%%%%%%%%%%%
638: %%%%%%%%%%%%%%%%
639: 
640: \section{Homotopy type of fibres}
641: 
642: %%%%%%%%%%%%%%%%%%%%%
643: %%%%%%%%%%%%%%%%%
644: %%%%%%%%%%%%%%%%
645: 
646: 
647: In case of 
648: isolated singularities, we have the following result on the relative homotopy type.
649: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
650: \begin{theorem}\label{t:attach} 
651:  Let  $\{ a_i\}_{i=1}^p$ be the set of atypical values of $F$ within some  
652:  open disc $D\subset \bP^1$. Let $s\in D$ be some typical value of $F$.
653:  For all $1\le i\le p$, let  $F$ have an isolated $\cG$-singularity at $a_i$,  $Y$ be of pure dimension $m$ at $a_i$ and $\rHd (Y\setminus (V\cup A)) \ge m$ in some neighbourhood of $a_i$. If either of the two following conditons
654:  is fulfilled:
655: \begin{enumerate}
656:  \item $X$ is a Stein space,
657:  \item $X= Y\setminus A$ and $X$ has at most isolated singularities,
658: \end{enumerate} 
659:   then $X_D$
660:  is obtained from 
661: $X_s$ by attaching cells of real dimension $m$.
662: In particular, the topological space $X_D/X_s$ is homotopy equivalent to a bouquet of speres $\vee S^m$.
663: \end{theorem}
664: \begin{Proof} 
665: We prove that the reduced integral homology of $X_D/X_s$ is concentrated in dimension $m$. 
666: By Proposition \ref{t:loc}, the 
667: variation of topology of the fibres of $F_{| X_D}$ is
668: localizable at the points $\Sing_\cG F \cap \bY_D$. 
669: We have to take into account all the possible positions of such a singular point $\xi= (y,a)$, namely: on $X$, on $V\setminus A$ or on $A\times \bP^1 \subset \bY$. 
670: 
671: In all the cases, it turns out that the pair $(X_{D_a}\cap B_\xi, X_s \cap B_\xi)$ is ($m -1$)-connected, where $B_\xi \subset \bY$ is a small enough ball at $\xi$, $D_a$ is a small enough closed disc at $a$ and $s\in \partial D_a$.
672: 
673: For a point $\xi$ in the first case, this is just Milnor's classical 
674: result for holomorphic functions with isolated singularity 
675: \cite{Mi}.
676: In the two remaining cases, this follows by a result due to
677:  Hamm and L\^e \cite[Corollary 4.2.2]{HL}, in a slightly improved version for partial Thom 
678: stratifications (see \cite[2.7]{Ti-t}). This result needs the condition on the rectified homological depth.
679: 
680:   By the above proven connectivity  
681:  of the pair $(X_{D_a}\cap B_\xi, X_s \cap B_\xi)$ and the splitting of 
682:  vanishing homology into local contributions Corollary \ref{c:conc}, we get that the homology of $(X_D,X_s)$ is zero below dimension $m$.
683:  Above dimension $m$, we also have the annulation, due to the following reasons. In case (a) the space $X_D$, respectively $X_s$, is Stein of dimension $m$, resp. $m-1$. In case (b),  we may apply the duality (\ref{eq:dual}) and we have that the cohomology 
684:  $\tilde H^*(\bY_s\cap B_\xi)$ is concentrated in dimension $m-1$, by Remark \ref{r:spheres}.
685:  
686:   Then one can map a bouquet of $m$ spheres into $X_D/X_s$
687: such that this map is an isomorphism in homology. This implies, by Whitehead's theorem (see \cite[7.5.9]{Sp}), that the map induces an isomorphism of homotopy groups. (Remark that $X_D/X_s$ is simply connected whenever $m\ge 2$). Since we work with analytic objects, therefore triangulable, the space $X_D/X_s$ is a CW-complex. We may now conclude our proof, since for CW-complexes, weak homotopy equivalence coincides with homotopy equivalence. 
688: \end{Proof}
689: %%%%%%%%%%%%%%%%%%
690: 
691: In case (b), this result has been proved by Siersma and the author \cite{ST}. Let us point out that, in this case (b), we also have the local bouquet result: $(X_{D_a}\cap B_\xi) /  (X_s \cap B_\xi) \stackrel{\h}{\simeq} \bigvee S^m$. 
692: 
693: When assuming high connectivity of the space $X$, we get the following immediate consequence (proved in lower generality in \cite{ST-x}).
694: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
695: \begin{corollary}
696: Under the hypotheses of Theorem \ref{t:attach},  if in addition the space $X$ is Stein and $(m-1)$-connected, 
697: then $X_s \stackrel{\h}{\simeq} \bigvee S^{m-1}$.
698: \fin
699: \end{corollary}
700: %%%%%%%%%%%%%%%%%%%%%%%%%
701:   Particular cases of this corollary appeared previously in several circumstances: Milnor's bouquet result \cite{Mi} on holomorphic germs with isolated singularity;
702: bouquet results for generic fibres of polynomial maps with isolated singularities in the affine \cite{Br-1,Br} and with isolated singularities at infinity \cite{ST}, \cite{Ti-t}.
703: 
704:  As another consequence, we shall draw a formula for the total number of
705:  vanishing cycles in case of isolated 
706: $\cG$-singularities. Let us denote by $\lambda_a$ the sum of the polar  numbers at the singularities on $(A\times \bP) \cap \bY_a$ and by 
707: $\mu_a$ the sum of the Milnor numbers of the singularities on $\bY_a \setminus (A \times \{a\})$. One needs to note that the Milnor fiber (in our case $\bY_s \cap B$) of a holomorphic function with isolated singularity on a Whitney stratified space is homotopically a bouquet of spheres of dimension $=\dim \bY_s \cap B$, see \cite{Ti-b}.
708: %%%%%%%%%%%%%%%%%%%%%%%%
709: \begin{corollary}\label{c:num}
710:  Under the hypotheses of Theorem \ref{t:attach}, we have:
711: \[  \dim H_{m-1}(X_{D_a}, X_s) =  \mu_a + \lambda_a \ , 
712: \]
713: \[ \dim H_{m-1}(X_D, X_s) = \sum_{a\in D}\mu_a + \sum_{a\in D}\lambda_a . \]
714: \fin
715: \end{corollary}
716: %%%%%%%%%%%%%%%%%%%%%%
717: \begin{remark}\label{r:ex} 
718: In Example \ref{ex:1}, let us consider $X=\bP^2\setminus \{ yz=0\}$. Let $s\in \bC\subset \bP$.
719:  It is easy to see that the fiber $X_0$ is a disjoint union of $c+1$ 
720: disjoint copies of $\bC^*$, where $c= \gcd (a,b)$, therefore $\chi(X_0) =0$. For $s\not= 0$, by a branched covering 
721: argument, one shows $\chi(X_s) =- (b+ap)$. The vanishing homology is concentrated in dimension 2, by Theorem \ref{t:attach}. 
722:  When taking $D=\bC$, we get the Betti number
723:   $b_2(X,X_s) = \chi(X,X_s) = \chi(X) - \chi(X_s) = (b+ap)$.
724:   We have seen at \ref{ex:1} that the sum $\sum_{a\in \bC}\lambda_a$ 
725:   consists of a single term $\lambda_\xi = b+ap$. One can also see easily that there is no other singularity, in particular that $\sum_{a\in \bC}\mu_a =0$. Hence the second equality in Corollary 
726: \ref{c:num} is verified.
727:   
728: \end{remark}
729: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
730: %%%%%%%%%%%%%
731: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
732: %%%%%%%%%%%%%%%
733: 
734: \section{Monodromy}
735:   
736: %%%%%%%%%%%%%% 
737:  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
738: %%%%%%%%%%%%%%% 
739: 
740: 
741: Let $F\colon Y \dashrightarrow \bP^1$ be a meromorphic function and let $\Lambda\subset \bP^1$ denote the set of atypical values of the associated map $\pi \colon \bY \to \bP^1$. 
742: There is a well defined monodromy $h_i$ around an atypical value $a_i \in 
743: \Lambda$.  This is induced by a counterclockwise loop around the small circle $\partial D_i$. Let $D$ denote some large disc, like in \S \ref{s:van}, such that $D\cap \Lambda \not= \emptyset$. We have a geometric monodromy representation:
744: \[ \rho_i : \pi_1 (\partial \bar D_i, s_i) \to \Iso (X_D, X_{\partial \bar D_i},X_{s_i}), \]
745: where $\Iso (.,.,.)$ denotes the group of relative isotopy classes of stratified homeomorphisms (which are C$^\ity$ along each stratum).
746: Note that the retriction of this action to $X_D$, $X_{D_i}$ or to $X_{\partial D_i}$,
747: is trivial. Let $T_i$ denote the action induced  by $\rho_i$ in homology
748: (with integral coefficients).
749: 
750: Let us identify $H_*(X_D,X_s)$ to 
751: $\oplus_{a_i\in D\cap \Lambda} H_*(X_{D_i}, X_{s_i})$ as in Proposition \ref{p:basic}. This identification depends on the chosen system of paths $\gamma_i \subset D$ from $s\in \partial D$ to $s_i\in \partial D_i$, as explained in \S \ref{s:van}. 
752: There is the following general result, showing that the 
753: action of the monodromy $T_i$ on a vanishing cycle $\omega \in H_*(X_D,X_s)$ 
754: changes $\omega$ by adding to it only contributions from the homology 
755: vanishing at $a_i$. 
756: %%%%%%%%%%%%%%%%%%%%%
757: 
758: \begin{proposition}\label{p:picard}
759: For every $\omega\in H_*(X_D,X_s)$, there is $\psi_i(\omega) \in H_*(X_{D_i}, X_{s_i})$ such that 
760: $T_i (\omega) = \omega + \psi_i(\omega)$.
761: \end{proposition}
762: \begin{Proof}
763: The proof goes exactly as in the more particular that we consider in \cite[Prop. 6.1]{ST-x}. 
764: One may identify the map:
765: $T_i - \id : H_{q+1}(X_D,X_s) \to H_{q+1}(X_D,X_s)$
766: to the composed map:
767: \begin{equation}\label{eq:wang} 
768: H_{q+1}(X_D,X_s) \stackrel{\partial}{\to} H_q(X_s) \stackrel{w}{\to}
769: H_{q+1}(X_{\partial D_i},X_s) \stackrel{i_*}{\to} H_{q+1}(X_D,X_s),
770: \end{equation}
771: where $w$ denotes the Wang map, which $w$ is an isomorphism, by K\"unneth 
772: formula. The last morphism in (\ref{eq:wang}) factors as follows:
773: \[ \begin{array}{rcl}
774:   H_{q+1}(X_{\partial D_i},X_{s_i}) & \stackrel{i_*}{\longrightarrow}  & 
775: H_{q+1}(X_D,X_s) \\
776:   \searrow & \  & \nearrow  \\
777:  \ &  H_{q+1}(X_{D_i}, X_s) & \ 
778: \end{array} \] 
779: where all three arrows are induced by inclusion. It follows that 
780: the submodule of ``anti-invariant cycles'' $\im (T_i - \id : 
781: H_* (X_D,X_s)\to H_* (X_D,X_s))$ is 
782: contained in the direct summand $H_*(X_{D_i},X_{s_i})$ of $H_*(X_D,X_s))$.
783: \end{Proof}
784: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
785: One has the following easy consequence, in full generality. Assume that the paths in $D$, say $\gamma_1, \ldots 
786: \gamma_l$ are counterclockwise ordered. The chosen paths define a  decomposition of $H_*(X_D,X_s)$ into the direct sum $\oplus_{a_i\in D\cap \Lambda} H_*(X_{D_i}, X_{s_i})$. Denote by $T_{\partial D}$ the monodromy around the circle $\partial D$. One has the following imediate consequence,  remarked in \cite{DN2} for polynomial functions and in \cite{ST-x} for the particular case $Y$ nonsingular and $X=Y\setminus A$.
787: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
788: \begin{corollary}\label{c:pis}
789:  Assume that the direct sum decomposition of $H_*(X_D,X_s)$ is fixed. Then $T_{\partial D}$ determines $T_i$, $\forall 
790: i\in\{ 1, \ldots , l\}$.
791: \fin
792: \end{corollary}
793: 
794: \begin{note}\label{n:piclef}
795: One may say that Proposition \ref{p:picard} is a Picard type formula, since Picard showed it at the end of the XIX-th century, for algebraic functions of two variables with simple singularities. Lefschetz proved later the well known formula for a loop around a quadratic singularity, in which case $\psi_i(\omega)$ is, up to sign, equal to $c\Delta$, where $\Delta$
796: is the quadratic vanishing cycle and $c$ is the intersection number $(\omega, \Delta)$. This became the basis of what one calls now Picard-Lefschetz theory (which is the conterpart of the Morse theory, in case of complex spaces), see e.g. \cite{AGV}, \cite{Eb}, \cite{Va}. In case of polynomial functions, the Picard formula was singled out in \cite{DN1}, \cite{NN2}.
797: \end{note}   
798: %%%%%%%%%%%%%%%%%
799: \begin{remarks}\label{r:piclef}
800:  The statement and proof of Proposition \ref{p:picard} dualize easily from homology to cohomology. One 
801: obtains in this way statements about invariant cocycles $\ker (T^i -\id 
802: \colon H^*(X,F) \to H^*(X,F))$ instead of anti-invariant cycles.
803: 
804:   A special case is that of a polynomial function 
805: $F : \bC^n \to \bC$, for which $X = \bC^n$. 
806: Results on invariant cocycles were obtained in 
807: \cite{NN}; they can be proved also in our more general setting. 
808: 
809: \end{remarks}
810: 
811: In the rest of this section we review some results on the zeta function
812: of the monodromy. We shall only discuss  global meromorphic functions $F$; following the general remark in the Introduction, all results translate easily in case of meromorphic germs.
813: For the particular case of polynomial functions, we send the reader
814: to \S \ref{s:poly}, where we present more specific results. 
815: 
816: \begin{definition}
817:  Let $T_{\partial D}$ be the monodromy around some disk $D$ as above. One calls zeta function of $T_{\partial D}$ the following rational function in variable $t$:
818: \[ \zeta_{(X_D, X_s)} (t) = \prod_{i\ge 0} \det [\id - t T_{\partial D} : H_i(X_{D},X_s)\to H_i(X_{D},X_s)
819: ]^{(-1)^{i+1}}.\]
820: \end{definition}
821: 
822:  We are interested here in the zeta function  of the monodromy around a value
823:  $a\in \Lambda$. Let us first assume that $F$ has isolated
824:  $\cG$-singularities. By the direct sum splitting (Corollary \ref{c:conc})
825:  and since the monodromy acts on each local Milnor fibration, we get:
826:  \[ \zeta_{(X_{D_a}, X_s)} (t) = \prod_{i=1}^k \zeta_{(X_{D_a} \cap B_i, X_s \cap B_i)} (t),\]
827:  where $\{ p_1, \ldots, p_k\} = \bY_a\cap \Sing_\cG F$ and $B_i$ is a small Milnor ball centered at $p_i$.
828:  
829:  For the zeta function  of the monodromy $T_{\partial D_a}$ acting on the homology of the general fibre $X_s$ we also get:
830: \[ \zeta_{X_s} (t)= (1-t)^{-\chi(X_a)} \prod_{i=1}^k \zeta^{-1}_{(X_{D_a} \cap B_i, X_s \cap B_i)} (t),\]
831:  since the monodromy acts on $X_{D_a}$ as the identity and since $\chi(X_{D_a}) = \chi(X_a)$.
832:  
833: 
834: 
835:  Let us now suppose that $Y$ is nonsingular and consider $X= Y\setminus A$,
836:  but not assume anything about the singularities of the meromorphic function $F$.
837:  One may follow the method of A'Campo \cite{A'C-zeta} to produce a formula for  the zeta function, as follows. There exists a proper holomorphic modification
838: $\phi \colon \tilde Y \to Y$,  which is bi-holomorphic over $X\setminus \cup_{a\in \lambda} X_a$.
839:  The pull-back $\tilde F = F\circ \phi$ has a general fibre $\tilde X_s$ which is isomorphic to $F$. The action of the monodromy is also the same, therefore $\zeta_{\tilde X_s} (t) = \zeta_{X_s} (t)$. Then one can write down a formula for the zeta function around the value $a\in \bP^1$ in terms of the exceptional divisor and the axis $A$. By expressing the result as an integral with respect to the Euler characteristic (see e.g. Viro's paper \cite{Vi} for this technique), one can get rid of the resolution.
840: %%%%%%%%%%%%%%%%%%%%%%%%%%%% 
841: \begin{proposition}\label{p:glm} \cite{GLM-1} 
842:  Let $X= Y\setminus A$ and let $Y$ be nonsingular. Then: 
843:  \[ \zeta_{X_s}(t) = \int_{A\times \{ a\}\cup X_a} \zeta_p (t) d\chi,\]
844:  where $\zeta_p$ denotes the local zeta function at the point $p\in \bY$.
845: \fin 
846: \end{proposition}
847:  Further formulae for the zeta function and some consequences can be found in the papers by Gusein-Zade, Luengo and  Melle \cite{GLM-1, GLM, GLM-3}.
848:  
849:   From $\zeta_{X_s}$ one easily gets $\zeta_{(X_D, X_s)}$, since $\zeta_{X_s}= \zeta_{X_D}\cdot \zeta_{(X_D, X_s)}^{-1}$ and $\zeta_{X_D} = (1-t)^{-\chi(X_a)}$.
850: 
851: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
852: 
853: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
854: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
855: %%%%%%%%%%%%%
856: 
857: \section{Nongeneric pencils and Zariski-Lefschetz type results}\label{s:lef}
858: 
859: Exploring a space $Y$ by pencils of complex hyperplanes is an old idea 
860: in mathematics. It appeared in Lefschetz's work \cite{Lef}, which became the fundation of the so-called Lefschetz Theory. Almost in the same time the Morse Theory was born \cite{Mo}.
861: Each of the two theories provide a method for studying the topology of the space; both use scanning with levels of a function.\footnote{for hystorical notes and new developments until about 1987, see Goresky and MacPherson's  book \cite{GM}} 
862: The analogous of Morse function for the  Lefschetz Theory is ``Lefschetz pencil''.
863: 
864: One usually means by Lefschetz pencil a pencil having
865: singularities of simplest type (i.e.  A$_1$) and transversal axis. In the usual projective space, such pencils are generic, but on certain spaces they might not even  exist. A more general point of view is to allow pencils with isolated singularities, alias meromorphic functions $F \colon Y \dashrightarrow \bP^1$ with isolated singularities in the sense of this paper. We call them ``nongeneric pencils" and point out that they can have singularities also within the axis $A$ of the pencil. The case of isolated singularities outside the axis has been considered before by Hamm and L\^e (e.g. \cite{HL}) and by Goresky and MacPherson (see \cite{GM}).
866: 
867: 
868:  The following connectivity result of Zariski-Lefschetz type holds. 
869: %%%%%%%%%%%%%%%%%%%%
870: \begin{theorem}\label{t:lef} \cite{Ti-lef}
871: Let the pencil $F= P/Q \colon Y \dashrightarrow \bP^1$ have isolated $\cG$-singularities (Definition \ref{d:sing}).
872: Assume that 
873: $A\not\subset V$ and let $X_s$ denote a generic member of the pencil.
874: 
875: If $\r X \ge m$, $m\ge 2$, and if the pair $(X_s , A\cap X_s)$ is 
876: $(m-2)$-connected 
877: then the morphism induced by inclusion:
878: \[ \pi_i (X_s) \to \pi_i (X) \]
879: is an isomorphism for $i\le m-2$ and an epimorphism for $i=m-1$.
880:  \fin
881: \end{theorem} 
882: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
883: 
884: This represents a far-reaching extension of the classical Lefschetz 
885: theorem on hyperplane sections. The latter says that, if $X$ is a projective variety and $H$ is a hyperplane such that $X\setminus H$ is nonsingular of dimension $m$, then $\pi_i(X,X\cap H)= 0$, for all $i\le m-1$. This can be viewed as a statement about pencils with transversal axis (i.e. there are no singularities on $A$), since, even if $H$ is not a generic hyperplane, it is a member of some generic pencil in the projective space. Indeed, one may define 
886: such a pencil by  
887: choosing a generic axis inside the hyperplane $H$. Then our claim follows by the conjunction of the following 2 observations: 1). Theorem \ref{t:lef} is true when replacing $X_s$ by the tube $X_{D_a}$, where one  allows singularities of any type on $Y_a$; 2). $X_{D_a}$ is contractible to $X_a$ when $X$ is compact (i.e. $V=\emptyset$).
888: 
889:  One may draw the following consequence on the homotopy type of the pair space-section, which actually represents an extension of Theorem \ref{t:attach}:
890: 
891: \begin{corollary}\label{c:lef} \cite{Ti-lef}
892:  Under the hypotheses of Theorem \ref{t:lef}, up to homotopy type, 
893:  the space $X$ is built 
894: from $X_s$ by attaching cells of dimension $\ge m$. 
895: If $X$ 
896: is in addition a Stein space of dimension $m$, then the attaching 
897: cells are of dimension precisely $m$.
898: \fin
899: \end{corollary}
900: 
901: %%%%%%%%%%%%%%%%%%%%%
902: \begin{remark}\label{r:lef}
903: What happens when $A\subset V$? 
904: We prove in \cite{Ti-lef} that if $\{ Q=0\}\subset V$ (which is a special case of $A\subset V$, since $A\subset\{ Q=0\}$), then the conclusion of Theorem \ref{t:lef} holds, with the single assumption $\r X \ge m$.
905: 
906: This result concerns in particular 
907: the polynomial functions $P : \bC^n \to \bC$. Such a function defines a nongeneric pencil on $X= \bC^n$, since it can be regarded as a meromorphic 
908: function $\tilde P/ Q$ on the (weighted) projective space $Y = \bP_w^n$, where  $\{ Q=0\}$ is the hyperplane at infinity. Notice that in this case the condition $\r X \ge n$ holds since $X$ is nonsingular. We refer to \S \ref{s:poly} for some consequences.
909: \end{remark}
910: %%%%%%%%%%%%%%%%%%%%%
911: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
912: %%%%%%%%%%%%%%%%%
913: 
914: \section{Equisingularity at the indeterminacy locus}\label{s:equi}
915: %%%%%%%%%%%%%%%%%%%%
916: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
917: %%%%%%%%%%%%%%%%%
918: 
919: Equisingularity conditions are considered beginning with  Zariski's work
920: on families of algebraic (hyper)surfaces, see \cite{Za}. There are more recent contributions to local equisingularity theory, especially by Teissier \cite{Te-1, Te-2} and Gaffney (\cite{Ga} and several other papers of the same author).
921: 
922: The case of families on non-compact spaces, like our family $\{ X_a\}_{a\in \bP^1}$,  where $X = Y\setminus V$ and $A\cap V \not= \emptyset$,
923: is special. We have seen 
924:  that the singularities at the indeterminacy locus play an important role.
925:  The problem would be to find the weakest equisingularity condition at $A$ such that to imply topological triviality in the neighbourhood of $A$.
926:   If we stratify everything by Whitney conditions, then we may invoke Whitney equisingularity, which implies topological triviality; but Whitney equisingularity is too strong.  The search for a weaker alternative 
927:  has itself some history behind; maybe the first result in this sense is L\^e-Ramanujam's theorem for families of holomorphic germs with isolated singularity: ``$\mu$ constancy implies topological triviality'', see \cite{LR}.
928:   It had been found that $\mu$ constancy is really weaker than Whitney equisingularity \cite{BS} (i.e. that $\mu$ constancy is weaker than $\mu^*$ constancy\footnote{see Teissier's paper \cite{Te-2} for $\mu^*$ constancy}).
929:  
930:  In the same spirit, the problem was formulated (and solved) in case of a family of affine hypersurfaces in \cite{Ti-e}, where the equisingularity at infinity, respectively C$^\infty$-triviality at infinity, comes into the picture (see next section for details). In {\em loc.cit.}, these two notions are related to the partial Thom stratification $\cG$.
931:  For a family defined by a meromorphic function $F \colon Y \dashrightarrow \bP^1$, one may follow the ideas of \cite{Ti-e} up to some extent, as we pointed out in \cite{ST-x}. Let us give the main lines.  
932:  
933: \begin{definition}
934:  We say that $F_{|X}$ is topologically trivial at $\xi\in A\times \bP^1 \subset \bY$, resp. at $a\in \bP^1$,  if there is a neighbourhood
935:   $\cN$ of $\xi$, resp. of $\bY_a \cap (A\times \bP^1)$, and a small enough 
936: disc $D$ at $a$ such that the map $\pi_| : \cN \cap X_D \to D$ 
937: is a trivial fibration. 
938: \end{definition}
939: 
940: The points $\xi\in A\times \bP^1$ which pose problems are those in
941: $\Sing_\cG F$, since for the others we have the topological triviality.
942:  This claim follows by attentive re-reading of  
943:  \cite[Theorems 2.7, 4.6, 1.2]{Ti-e}, \cite[Theorem 7.2]{ST-x}; we actually get the following:
944: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
945: 
946: \begin{theorem}\label{t:equi}
947: Let $F = P/Q$ have isolated $\cG$-singularities at $\xi$, resp. at $a\in \bP^1$. Assume that $Y$ is of pure dimension $m$, that $X= Y\setminus \{ Q=0\}$ and that $\r X \ge m$.
948:  
949: Then $F_{|X}$ is  topologically trivial at $\xi$, resp. at 
950: $a\in \bP^1$ if and only if $\lambda_\xi = 0$, resp. $\lambda_a = 0$.
951: 
952: In particular, if $X$ has isolated singularities and $F$ has isolated $\cG$-singularities at $a\in \bP^1$, then 
953: $X_a$ is a general fibre of $F_{|X}$ if and only if $X_a$ is nonsingular and $\lambda_a = 0$.
954:  \fin
955: \end{theorem} 
956: 
957: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
958: Combining Theorem \ref{t:equi} with Corollary \ref{c:num}, we get 
959: the following consequence: 
960: 
961: \begin{corollary}\label{c:jump}
962: Under the hypotheses of Theorem \ref{t:attach}, a fibre $X_a$ of $F_{|X}$ is  general if and only if it has the same Euler characteristic of a general fibre.
963: \fin
964:  \end{corollary}
965:  
966:  Both results above have been stated, in
967: slightly lower generality, in \cite{ST-x}.
968: Corollary \ref{c:jump} extends the criteria for atypical 
969: fibres in case of polynomial functions in 2 variables \cite{HaLe}, and in 
970: $n$ variables \cite{ST}, \cite{Pa}. See also Proposition \ref{p:cond}. 
971: 
972: %%%%%%%%%%%%%%%%%%%%%%
973: %%%%%%%%%%%%%%%%%%
974: %%%%%%%%%%%%%%%%%%%%%
975: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
976: 
977: \section{More on polynomial functions}\label{s:poly}
978: 
979: In the last years there has been developed a flourishing activity in the topology of
980: polynomial maps, partly due to the links with affine geometry (see e.g. Kraft's Bourbaki talk \cite{Kr}). We give here a brief overview, throughout some of the multitude of the contributions.
981: 
982: %%%%%%%%%%%%%%%%%%%%
983: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
984: We have explained in the Introduction that a complex polynomial function $P: \bC^n \to \bC$, $\deg P =d$, can be extended to a meromorphic function $\frac{\tilde P}{x_0^d} : \bP^n \dashrightarrow \bP$. Here $X=\bC^n$ and $Y = \bP^n$, are nonsingular spaces. This is not the only way of extending $P$ and the space $\bC^n$; one may consider\footnote{see \cite{Ti-t} for a general treatment} for instance an embedding of $\bC^n$ into some toric variety, such as a weighted projective space $\bP^n_w$.
985: 
986: Maybe the first author who studied the topology of polynomial functions was Broughton \cite{Br-1}. In the same time Pham \cite{Ph} was interested in regularity conditions under which a polynomial has good behaviour at infinity. 
987:  Some of the challenging problems that have been under research ever since are:
988: \vspace*{2mm}
989: 
990: \noindent
991:  1. Determine the smallest set $\Lambda$ of atypical values of $P$.\\
992:  2. Describe the topology of the general fibre and of the atypical fibres.\\
993:  3. Describe the variation of topology in the family of fibres; monodromy.\\
994:  
995:  
996:  In problem 1., there are only partial answers.  One has to decide which are the atypical values among a finite set of values singled out by Proposition \ref{p:1}. For instance, our general result Corollary \ref{c:jump} applies here (see also the comment following it). It is easy to show that singular values of $P$ are atypical. Then, fixing a nonsingular fibre $X_a$ of $P$, one may try to prove topological triviality at infinity by constructing a controlled vector field and ``pushing'' $X_a$ along it. This is an idea due to Thom \cite{Th}.
997:   There are mainly two strategies: to work in the Euclidean space or to 
998:   compactify $\bC^n$ into some $Y$ and use the stratification $\cG$.
999:   
1000:   The first one leads to regularity conditions, in more and more generality: tameness \cite{Br}, quasi-tameness \cite{Ne}, M-tameness \cite{NZ}, $\rho$-regularity \cite{Ti-r}.
1001:   There are also the Malgrange condition (see \cite[2.1]{Ph})---which is a condition on the \L ojasiewicz number at infinity---and its generalization by Parusi\'nski \cite{Pa}. 
1002:   
1003:   The second strategy leads to the t-regularity \cite{ST}, or more generally, equisingularity at infinity (which has been discussed for meromorphic functions in \S \ref{s:equi}).
1004:     
1005:  There are of course relations between all these conditions; one may consult \cite{Ti-r} and its references.
1006:   Under certain circumstances, several of these conditions are equivalent.
1007:   We may quote the following result:
1008:   %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1009: \begin{proposition}\label{p:cond} \rm (after \it \cite[5.8]{ST} \rm and  \cite[2.15]{Ti-r})\it \\ 
1010: Let $P \colon \bC^n \to \bC$ have isolated $\cG$-singularities at infinity, at $a\in \bC$.
1011: Then the following are equivalent:
1012: \begin{enumerate}
1013: \item $P$ is M-tame at $X_a$.
1014: \item $P$ satisfies Malgrange condition at infinity at $X_a$.
1015: \item $P$ is t-regular at infinity at $X_a$.
1016: \item $\lambda_a =0$.
1017: \end{enumerate}
1018: \end{proposition}
1019:  It follows that, for a polynomial with isolated $\cG$-singularities at infinity, a fibre $X_a$ is general if and only if $\mu_a =0$ and one of the above conditions are satisfied.
1020:  
1021:    In case of 2 variables, the hypothesis of the above statement \ref{p:cond} is fulfilled and therefore the conclusions are valid. Indeed, 
1022: in 2 variables, any reduced fibre $X_a$ has at most isolated $\cG$-singularities at infinity. Moreover, there are several other criteria expressing non singularity at infinity, equivalent to the ones above; we send the reader to \cite{Du}, \cite{Ti-r}.
1023: 
1024:  Still in 2 variables, one may derive the following equivalent formulation of the well known Jacobian Conjecture, in terms of singularities at infinity \cite{LW,ST}: {\em Let $P\colon \bC^2 \to \bC$ be a polynomial without critical point.
1025:  If there exists a value $a\in \bC$ such that $\lambda_a \not= 0$, then for any other polynomial $g$, the zero locus of their jacobian ideal $\Jac(P,g)$ is not void.}
1026:  
1027:  
1028:  
1029:  Revisiting equisingularity at \S \ref{s:equi}, one gets more specific results in case of polynomials. By taking hyperplane sections with respect to $P$ in $\bC^n$, one may define global polar curves \cite{Ti-e} and use them in order to define an intersection number with some fibre $X_a$. By restricting $P$ to a general hyperplane and repeating generical cutting, we get a sequence of intersection numbers\footnote{they are invariant under linear change of coordinates but not under affine automorphisms.} $\gamma_a^{n-1}, \ldots , \gamma_a^{0}$. We show in \cite[Theorem 1.1]{Ti-e} that the constancy of $\gamma_s^*$, for $s$ in some neighbourhood of $a$, is equivalent to the equisingularity at infinity of $p$ at $X_a$.
1030:  
1031:  Secondly, the slicing processus just described gives a model of a fibre  $X_a$ as CW-complex. Let $\lambda_a^i := \gamma_s^i - \gamma_a^i$, where $s$ is a typical value of $P$.
1032:  %%%%%%%%%%%%%
1033: \begin{theorem}\label{t:cw}\cite{Ti-e}
1034:  Let $P \colon \bC^n \to \bC$ be a polynomial function. 
1035:  Suppose that the fibre $X_a = P^{-1}(a)$ has at most 
1036: isolated singularities. 
1037: Then $X_a$ is homotopy equivalent to a generic hyperplane 
1038: section $X_a \cap H$ to which one attaches $\gamma_a^{n-1} - \mu(X_a)$ cells of dimension $n-1$.
1039: 
1040: Moreover, $X_a$ is homotopy equivalent to the CW-complex obtained 
1041: by successively attaching 
1042: to $\gamma_a^{0} =\deg P$ points a number of $\gamma_a^1$ cells of dimension 1, then 
1043: $\gamma_a^2$ cells of dimension 2, $\ldots$, $\gamma_a^{n-2}$ cells of 
1044: dimension $n-2$ and finally $\gamma_a^{n-1} - \mu(X_a)$ cells of dimension 
1045: $n-1$.
1046:  In particular,
1047:  $\chi(X_a) = (-1)^n \mu(X_a) + \sum_{i=0}^{n-1} (-1)^i\gamma_a^i$
1048: and\\
1049:  $\chi(X_s) - \chi(X_a) = (-1)^{n-1} \mu(X_a) + \sum_{i=0}^{n-1} 
1050: (-1)^i\lambda_a^i$.
1051:   \fin
1052: \end{theorem} 
1053:  %%%%%%%%%%%%% 
1054:  One may compare this result to Corollary \ref{c:num} and notice that the
1055:  sequence $\lambda_a^*$ is a refinement of the number of vanishing cycles at infinity $\lambda_a$, in case 
1056:  of isolated $\cG$-singularities. Nevertheless, the numbers $\lambda_a^i$
1057:  are defined without any hypothesis on singularities at infinity.
1058:  
1059:  The vanishing cycles at infinity were described for the first time in \cite{ST}. It was shown in {\em loc.cit.} that, for $P$ with isolated $\cG$-singularities at infinity, the vanishing homology is concentrated in dimension $n$; this implies that the general fibre is homotopically a bouquet of $(n-1)$-spheres.  Further progress in describing the general fibre, the special fibres and the vanishing cycles was made by Neumann-Norbury \cite{NN, NN2}, Dimca-N\'emethi \cite{DN1}, the author \cite{Ti-lef}. The cohomology of fibres is investigated by Hamm \cite{Ha-m}.
1060: 
1061: The above construction of the model for the general fibre can be 
1062: pushed further; one may construct a global geometric monodromy group, acting on this model. This yields localization results and formulae for the zeta-function \cite{ST-m}. The more geometric point of view
1063: on monodromy at a singularity at infinity gives two types of singularities
1064: with local $\lambda$ equal to 1, see \cite[\S 6]{ST-m}. This may be contrasted with $\mu =1$ in case
1065: of holomorphic germs, when the singularity can only be of one type, A$_1$. 
1066: 
1067:   One of the monodromies in case of a polynomial $P$ is the one around a big disc containing all the atypical values, denoted $T_\ity$. In two variables, 
1068:  Dimca \cite{Di} shows that $T_\ity$ acting on the cohomology of the fibre is the identity if and only if the monodromy group of $P$ is trivial; the eigenvalue 1 occurs only in size one Jordan blocks. We send the reader to {\em loc.cit.} for further results and their discution in contrast to the holomorphic germs case. 
1069:  
1070:  Further aspects, such as mixed Hodge structure on fibres and algebraic Gauss-Manin systems, have been studied by several authors: Garc\'{\i}a-L\'opez and N\'emethi \cite{GN1, GN2}, respectively
1071:  Dimca-Saito \cite{DS}, Sabbah \cite{Sa1, Sa2}.
1072:  
1073: 
1074: \begin{thebibliography}{99}
1075: 
1076: \itemsep=\smallskipamount
1077: 
1078: \bibitem{A'C} N. A'Campo, {\em Le nombre de Lefschetz d'une monodromie}, Indag. 
1079: Math., {\bf 35} (1973), 113--118.
1080: 
1081: \bibitem{A'C-zeta}
1082: N. A'Campo, {\em La fonction zêta d'une monodromie}, Comment. Math. Helv., {\bf 50} 
1083: (1975), 233--248.
1084: 
1085: \bibitem{Ar}
1086: V.I. Arnol'd, {\em Singularities of fractions and behaviour of 
1087: polynomials at infinity},  Tr. Mat. Inst. Steklova, {\bf 221} (1998), 48--68. 
1088: 
1089: \bibitem{AGV}
1090: V. I. Arnol'd,   S. M. Guse\u\i n-Zade, A. N. Varchenko, {\em  Singularities of 
1091: differentiable maps. Vol. II. Monodromy and asymptotics
1092: of integrals}, Monographs in Mathematics, {\bf 83}.
1093: Birkh\" auser Boston, Inc., Boston, MA, 1988.
1094: 
1095: \bibitem{BS}
1096: J. Brian\c con, J.-P. Speder, {\em La trivialit\'e topologique 
1097: n'implique pas les conditions de Whitney}, C.R. Acad. Sci. Paris, s\'er. A, {\bf 280} 
1098: (1975), 365-367.
1099: 
1100: \bibitem{BMM}
1101: J. Brian\c{c}on, Ph. Maisonobe, M. Merle,  {\em Localisation de 
1102: syst\`emes diff\'erentiels, stratifications de Whitney et condition de Thom},   
1103: Inventiones Math.,
1104:  {\bf 117}, 3 (1994),  531--550.
1105:  
1106: \bibitem{Bri}
1107: E. Brieskorn, {\em Beispiele zur Differentialtopologie von Singularit\" aten}, 
1108: Invent. Math., {\bf 2} (1966), 1--14.   
1109: 
1110:  \bibitem{Br-1}
1111:   S.A. Broughton, {\em On the topology of polynomial
1112: hypersurfaces},
1113: Proceedings A.M.S. Symp. in Pure. Math., vol. {\bf 40}, I (1983),
1114: 165-178.
1115: 
1116: \bibitem{Br}
1117:   S.A. Broughton, {\em Milnor numbers and the topology 
1118:   of polynomial hypersurfaces}, Invent. Math.,  {\bf 92}, 2 (1988), 217--241. 
1119:   
1120: \bibitem{BV}
1121: D. Burghelea, A. Verona, 
1122:  {\em Local homological properties of analytic sets} 
1123: Manuscripta Math.,   {\bf 7} (1972), 55--66.
1124: 
1125: \bibitem{De}
1126: P. Deligne, {\em Groupes de monodromie en g\'eom\'etrie alg\'ebrique. II.}, 
1127: SGA 7 II, dirig\'e par P. Deligne et N. Katz. Lecture Notes in Mathematics, Vol. 
1128: {\bf 340}. Springer-Verlag, 1973. 
1129: 
1130: \bibitem{Di}
1131: A. Dimca, {\em Monodromy at infinity for polynomials in two variables}, J. Algebraic Geom., {\bf 7}, no. 4 (1998), 771--779. 
1132: 
1133: \bibitem{DN1}
1134: A. Dimca, A. N\'emethi, {\em Thom-Sebastiani construction and monodromy of 
1135: polynomials},  Universit\'e de Bordeaux, preprint no. 98/1999.
1136: 
1137: \bibitem{DN2}
1138: A. Dimca, A. N\'emethi, {\em On monodromy of complex polynomials},
1139:  Duke Math. J., 108, 2 (2001), 199--209.
1140: 
1141: \bibitem{DS}
1142: A. Dimca, M. Saito, {\em Algebraic Gauss-Manin systems}, Preprint no. 37 (1999), Universit\'e Bordeaux 1.
1143: 
1144: \bibitem{Do}
1145: A. Dold, {\em  Lectures on algebraic topology},
1146:  Grundlehren der mathematischen Wissenschaften, Band {\bf 200}. 
1147: Springer-Verlag, New York-Berlin, 1972.
1148: 
1149: \bibitem{Du}
1150: A.H. Durfee, {\em   Five definitions of critical point at infinity}, Singularities (Oberwolfach, 1996), 345--360, Progr. Math., {\bf 162},
1151: Birkh\" auser, Basel, 1998.
1152: 
1153: \bibitem{Eb}
1154: W. Ebeling, {\em  The monodromy groups of isolated singularities of complete 
1155: intersections}, Lecture Notes in Mathematics, {\bf 1293}.
1156: Springer-Verlag, Berlin, 1987. 
1157: 
1158: \bibitem{Fo}
1159:    L. Fourrier, {\em Topologie d'un polyn\^ome de deux variables complexes au
1160: voisinage de l'infini}, Ann. Inst. Fourier, Grenoble, {\bf 46}, 3 (1996),
1161:  645--687.
1162:  
1163: \bibitem{Ga} 
1164: T. Gaffney, {\em  Integral closure of modules and Whitney equisingularity}, Invent. Math., {\bf 107}, no. 2 (1992), 301--322.
1165:  
1166: \bibitem{GN1}
1167: R. Garc\'{\i}a L\' opez, A. N\'emethi, {\em On the monodromy at infinity of a polynomial map. II}, Compositio Math., {\bf 115}, no. (1999), 1--20. 
1168: 
1169: 
1170: \bibitem{GN2}
1171: R. Garc\'{\i}a L\' opez, A. N\'emethi, {\em  Hodge numbers attached to a polynomial map}, Ann. Inst. Fourier (Grenoble), {\bf 49}, no. 5 (1999), 1547--1579.
1172: 
1173:  \bibitem{GM}
1174:  M. Goresky, R. MacPherson, {\em Stratified Morse Theory},
1175: Springer-Verlag Berlin
1176: Heidelberg New-York, 1987.
1177: 
1178: \bibitem{GWPL}
1179:    C.G. Gibson, K. Wirthm\" uller, A.A. du Plessis, E.J.N. Looijenga, 
1180: {\em Topological
1181: Stability of Smooth Mappings}, Lect. Notes in Math., {\bf 552}, Springer 
1182: Verlag 1976. 
1183: 
1184: \bibitem{Gr}
1185: A. Grothendieck, 
1186: {\em Cohomologie locale des faisceaux coh\' erents et th\' eor\` emes de 
1187: Lefschetz locaux et globaux},  S\'eminaire de g\'eom\'etrie algebrique du 
1188: Bois-Marie 1962 (SGA 2),
1189: Advanced Studies in Pure Mathematics, {\bf 2}. Amsterdam: North-Holland 
1190: Publishing 
1191: Company; Paris: Masson \& Cie, 1968.
1192: 
1193:  \bibitem{GLM-1}
1194:  S. Gusein-Zade, I. Luengo, A. Melle, {\em Partial resolutions and the 
1195: zeta-function of a singularity}, Comment. Math. Helv., {\bf 72}, no. 2
1196: (1997), 244--256. 
1197:  
1198:  \bibitem{GLM}
1199:  S. Gusein-Zade, I. Luengo, A. Melle, {\em Zeta functions for germs of 
1200: meromorphic functions and Newton diagrams}, Funct. Anal. Appl.,  {\bf 32}, 2 (1998), 93--99. 
1201: 
1202: \bibitem{GLM-3}
1203:  S. Gusein-Zade, I. Luengo, A. Melle, {\em On atypical values and local 
1204: monodromies of meromorphic functions},
1205: Dedicated to S.P. Novikov on the occasion of his 60th birthday,  Tr. 
1206: Mat. Inst. Steklova, {\bf 225} (1999), 168--176.
1207:  
1208: \bibitem{GLM-4}
1209:  S. Gusein-Zade, I. Luengo, A. Melle, {\em  On the topology of germs of meromorphic functions and applications}, preprint, math.AG/9905127, 7 pages.
1210: 
1211: \bibitem{HaLe}
1212:  H\`a H.V.,  L\^e D.T., {\em Sur la topologie des polyn\^omes 
1213: complexes}, Acta Math.
1214: Vietnamica, {\bf 9} (1984), pp. 21--32.
1215: 
1216: \bibitem{Ha-m}
1217:  H. Hamm, {\em On the cohomology of fibres of polynomial maps}, this volume.
1218: 
1219: \bibitem{HL}
1220:  H. Hamm, L\^e D.T., {\em Rectified homotopical depth and Grothendieck 
1221: conjectures},
1222: in: P. Cartier et all. (eds) Grothendieck Festschrift II, pp. 311--351, 
1223: Birkh\" auser
1224: 1991.
1225: 
1226: \bibitem{HMS} 
1227: J.P. Henry, M. Merle, C. Sabbah, {\em Sur la condition de Thom stricte 
1228: pour un
1229: morphisme analytique complexe}, Ann. Scient. Ec. Norm. Sup.  4$^e$ 
1230: s\' erie, t. {\bf 17}
1231: (1984), 227--268.
1232: 
1233: \bibitem{Hi} 
1234: F. Hirzebruch, {\em  
1235: Singularities and exotic spheres}, 
1236: S\'eminaire Bourbaki, Vol. {\bf 10}, Exp. No. 314 (1966), 13--32. 
1237: Soc. Math. France, Paris, 1995.
1238: 
1239: \bibitem{Kr}
1240: H. Kraft, {\em Challenging problems on affine $n$-space}, S\'eminaire Bourbaki, Vol. 1994/95. Ast\'erisque, {\bf 237} (1996), Exp. No. 802, 5,
1241: 295--317.
1242: 
1243: \bibitem{Le-oslo}
1244:  L\^e D.T.,   {\em Some remarks on the relative monodromy}, in: Real and 
1245: Complex Singularities, Oslo 1976, Sijhoff en Norhoff, Alphen a.d. Rijn 
1246: 1977, p. 397--403.
1247: 
1248: \bibitem{Le-s}
1249:  L\^e D.T.,   {\em Complex analytic functions with isolated singularities}, J. 
1250: Algebraic Geom.  {\bf 1}, 1 (1992), 83--99. 
1251: 
1252: \bibitem{LR}
1253:  L\^{e} D.T.,  C. P. Ramanujam, {\em The invariance of Milnor's number 
1254:  implies the invariance of the topological type},
1255:   Amer. J. Math.,  {\bf 98}, no.1 (1976), 67--78. 
1256:   
1257: \bibitem{LM}  
1258: L\^{e} D.T., Z. Mebkhout, {\em Vari\'et\'es caract\'eristiques et vari\'et\'es polaires},
1259: C. R. Acad. Sci. Paris S\'er. I Math. {\bf 296}, 2 (1983), 129--132.
1260:  
1261: \bibitem{LW}
1262: L\^e D.T.,  C. Weber, {\em Polyn\^omes \`a fibres rationelles et conjecture 
1263: Jacobienne
1264: \`a 2 variables}, CRAS Paris, t. {\bf 320}, s\' erie I (1995), pp. 581--584. 
1265: 
1266: \bibitem{Lef} 
1267: S. Lefschetz,
1268: {\em L'analysis situs et la g\'eom\'etrie alg\'ebrique}, Gauthier-Villars, Paris 
1269: 1924, nouveau tirage 1950.
1270: 
1271: \bibitem{Mi}
1272: J. Milnor,  {\em Singular points of complex hypersurfaces}, Ann. of 
1273: Math. Studies, {\bf 61}, Princeton 1968.
1274: 
1275: \bibitem{Mo}
1276:  M. Morse,  {\em  Relations between the critical points of a real function of $n$ independent variables}, Trans. Amer. Math. Soc., {\bf 27}, no. 3 (1925), 345--396.
1277: 
1278: \bibitem{Ne}
1279: A. N\' emethi, {\em  Th\'eorie de Lefschetz pour les vari\'et\'es alg\'ebriques affines}, C.R. Acad. Sci. Paris S\'er. I Math., {\bf 303}, no. 12 (1986), 567--570.
1280: 
1281:  \bibitem{NZ}
1282: A. N\' emethi, A. Zaharia,  {\em  On the bifurcation set of a polynomial and Newton boundary}, Publ. Res. Inst. Math. Sci.,  {\bf 26} (1990),  681--689.
1283: 
1284: \bibitem{NN}
1285:  W. D. Neumann, P. Norbury, 
1286:  {\em Vanishing cycles and monodromy of complex polynomials}, Duke Math. J.  {\bf 101}, no. 3 (2000), 487--497.
1287:  
1288:  \bibitem{NN2}
1289:  W. D. Neumann, P. Norbury, {\em Unfolding polynomial maps at infinity}, 
1290: Math. Ann.,  {\bf 318}, no. 1 (2000), 149--180. 
1291:  
1292: \bibitem{Pa}
1293: A. Parusi\'nski,  {\em On the bifurcation set of a complex polynomial with 
1294: isolated singularities at infinity}, Compositio Math.,  {\bf 97}, 3 (1995), 
1295: 369--384. 
1296: 
1297: \bibitem{Ph}
1298:  F. Pham, {\em Vanishing homologies and the $n$ variable saddlepoint method},
1299: Arcata Proc. of Symp. in Pure Math., vol. {\bf 40}, II  (1983), 319--333. 
1300: 
1301: \bibitem{Sa1}
1302: C. Sabbah, {\em Monodromy at infinity and Fourier transform}, 
1303: Publ. Res. Inst. Math. Sci., {\bf 33}, no. 4 (1997), 643--685.
1304: 
1305: \bibitem{Sa2}
1306: C. Sabbah, {\em Hypergeometric period for a tame polynomial}, C. R. Acad. Sci. Paris S\'er. I Math., {\bf 328}, no. 7 (1999), 603--608.
1307: 
1308:  \bibitem{Si}
1309:  D. Siersma, {\em Vanishing cycles and special fibres},
1310:  in: Singularity theory and its applications, Part I (Coventry, 1988/1989),
1311:   292-301,
1312: Lecture Notes in Math., {\bf 1462}, Springer, Berlin, 1991. 
1313: 
1314: \bibitem{ST}
1315:  D. Siersma, M. Tib\u ar, {\em Singularities at infinity
1316: and their
1317: vanishing cycles},  Duke Math. Journal, {\bf  80}:3 (1995), 771-783.
1318: 
1319: \bibitem{ST-m}
1320:  D. Siersma, M. Tib\u ar, {\em Singularities at infinity
1321: and their
1322: vanishing cycles, II. Monodromy}, Publ. Res. Inst. Math. Sci.,  {\bf 36} (2000), 659-679.
1323: 
1324: \bibitem{ST-x}
1325:  D. Siersma, M. Tib\u ar, {\em Vanishing cycles and singularities of 
1326: meromorphic functions}, Utrecht University preprint no. 1105, May 1999,  
1327: math.AG/9905108.
1328: 
1329: \bibitem{Sp}
1330: E.H. Spanier,   {\em Algebraic topology}, 
1331: McGraw-Hill Book Co., New York-Toronto, Ont.-London 1966.
1332: 
1333: 
1334: \bibitem{Sw}
1335: R. Switzer, {\em Algebraic Topology - Homotopy and Homology}, Springer 
1336: Verlag,
1337: Berlin-Heidelberg-New York 1975.
1338: 
1339: \bibitem{Te-1}
1340:  B. Teissier, {\em Cycles \' evanescents, sections planes
1341: et conditions de Whitney}, Singularit\' es \`a Cargesse,
1342: Ast\'erisque, {\bf 7-8} (1973).
1343: 
1344: \bibitem{Te-2}
1345:  B. Teissier, {\em Variet\' es polaires 2: Multiplicit\' es polaires, 
1346: sections planes
1347: et conditions de Whitney}, G\' eom\' etrie Alg\`ebrique \`a la Rabida, 
1348: Springer Lecture Notes in Math., {\bf 961} (1981),  pp. 314--491.
1349:   
1350: 
1351: \bibitem{Th}
1352:   R. Thom, {\em Ensembles et morphismes stratifi\' es},
1353: Bull. Amer. Math.
1354: Soc.,  {\bf 75} (1969), 249-312.
1355: 
1356:  \bibitem{Ti-b}
1357:  M. Tib\u ar, {\em Bouquet decomposition of the Milnor fibre}, Topology, {\bf 35}, no. 
1358: 1 (1996),  227--241.
1359: 
1360:   \bibitem{Ti-t}
1361:  M. Tib\u ar, {\em Topology at infinity of polynomial maps and Thom
1362: regularity condition},  Compositio Math., {\bf 111}, 1 (1998),
1363: 89-109.
1364: 
1365:  \bibitem{Ti-r}
1366:  M. Tib\u ar, {\em  Regularity at infinity of real and complex polynomial functions}, Singularity theory (Liverpool, 1996), xx, 249--264,
1367: London Math. Soc. Lecture Note Ser., {\bf 263}, Cambridge Univ. Press, Cambridge, 1999. 
1368: 
1369: \bibitem{Ti-e}
1370:  M. Tib\u ar, {\em Asymptotic Equisingularity and Topology of Complex 
1371: Hypersurfaces},  
1372:  Int. Math. Research Notices,  {\bf 18} (1998), 979-990.
1373:  
1374: \bibitem{Ti-lef}
1375: M. Tib\u ar, {\em Connectivity via nongeneric pencils}, 
1376: preprint NI01016-SGT, Newton Institute, Cambridge; to appear in Internat. J. of Math.
1377: 
1378: 
1379:  \bibitem{Va}
1380: V. A. Vassiliev,  {\em Ramified integrals, singularities and lacunas}, 
1381: Mathematics and its Applications, {\bf 315}. Kluwer Academic Publishers
1382: Group, Dordrecht, 1995. 
1383: 
1384:   \bibitem{Ve}
1385:   J.-L. Verdier, {\em Stratifications de Whitney et th\'
1386: eor\`eme de
1387: Bertini-Sard}, Inventiones Math., {\bf 36} (1976), 295--312.
1388: 
1389: \bibitem{Vi}
1390: O.Ya. Viro,  {\em Some integral calculus based on Euler characteristic}, 
1391: Topology and geometry---Rohlin Seminar, 127--138, Lecture Notes
1392: in Math., {\bf 1346}, Springer, Berlin, 1988.
1393: 
1394: \bibitem{Za}
1395: O. Zariski, {\em  Collected papers. Vol. IV. Equisingularity on algebraic varieties},  Edited and with an introduction by J. Lipman and B.
1396: Teissier. Mathematicians of Our Time, {\bf 16}. MIT Press, Cambridge, Mass.-London, 1979.
1397: 
1398: 
1399: \end{thebibliography}
1400: 
1401: 
1402: \end{document}
1403: 
1404: 
1405: 
1406: 
1407: 
1408: 
1409: 
1410: 
1411: 
1412: 
1413: 
1414: 
1415: 
1416: 
1417: 
1418: