1991 Mathematics Subject Classification. 14B05, 58A10, 55S99.Partially supported by the DFG-programme ”Global methods in complex geometry” (Eb 102/4–3), grants RFBR–04–01–00762, NSh–1972.2003.1.
<ph f="cmbx">Chern obstructions for collections of 1-forms on singular varieties</ph>

### W. Ebeling

Dedicated to Jean-Paul Brasselet on the occasion of his 60th birthday
Universitat Hannover, Institut fur Mathematik, Postfach 6009, D-30060 Hannover, Germany. E-mail address : ebeling@math.uni-hannover.de Moscow State University, Faculty of Mechanics and Mathematics, Moscow, 119992, Russia. E-mail address : sabir@mccme.ru

• Abstract. We introduce a certain index of a collection of germs of 1-forms on a germ of a singular variety which is a generalization of the local Euler obstruction corresponding to Chern numbers different from the top one.

Introduction

The aim of this paper is to bring together some ideas of [3and [6.
A germ of a vector field or of a 1-form on the complex affine space ${\mathbb{C}}^{n}$  at the origin not vanishing in a punctured neighbourhood of it has a topological invariant — the Poincaré–Hopf index. The sum of the Poincaré–Hopf indices of the singular points of a vector field on a compact complex manifold is equal to the Euler characteristic of the manifold. There are several generalizations of this notion to vector fields and/or to 1-forms on complex analytic varieties with singularities (isolated or not) started by M.-H. Schwartz: [11, 4, 8, 1, 12, 5, 3,.... For the case of an isolated complete intersection singularity there is defined an index which is sometimes called the GSV index: [8, 12, 5. Another generalization which makes sense not only for isolated complete intersection singularities and also not only for varieties with isolated singularities is the so called local Euler obstruction: [2, 3(its analogue for 1-forms was considered in [7). One can say that in some sense all these invariants correspond to the Euler characteristic, which, for a compact complex analytic manifold ${M}^{n}$  , coincides with the top Chern number $〈{c}_{n}\left(M\right),\left[M\right]〉$  .
A generalization of the GSV-index corresponding to other Chern numbers (different from the top one) was introduced and studied in [6. It is defined for a collection of germs of 1-forms on an isolated complete intersection singularity. For a collection of 1-forms on a projective complex complete intersection with isolated singularities, the sum of these indices of the singular points is equal to plus-minus the corresponding Chern number of a smoothing of the variety.
Here we define and study an index of a collection of germs of 1-forms on a germ of a singular variety which is an analogue of the local Euler obstruction corresponding to a Chern number different from the top one.

1 Special points of 1-forms

Let $\left({X}^{n},0\right)\subset \left({\mathbb{C}}^{N},0\right)$  be the germ of a purely $n$  -dimensional reduced complex analytic variety at the origin (generally speaking with a non-isolated singularity). Let $\mathbf{k}=\left\{{k}_{i}\right\}$  , $i=1,\dots ,s$  , be a fixed partition of $n$  (i.e., ${k}_{i}$  are positive integers, ${\sum }_{i=1}^{s}{k}_{i}=n$  ). Let $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  ( $i=1,\dots ,s$  , $j=1,\dots ,n-{k}_{i}+1$  ) be a collection of germs of 1-forms on $\left({\mathbb{C}}^{N},0\right)$  (not necessarily complex analytic; it suffices that the forms ${\omega }_{j}^{\left(i\right)}$  are complex linear functions continuously depending on a point of ${\mathbb{C}}^{N}$  ).
Let $\varepsilon >0$  be small enough so that there is a representative $X$  of the germ $\left(X,0\right)$  and representatives ${\omega }_{j}^{\left(i\right)}$  of the germs of 1-forms inside the ball ${B}_{\varepsilon }\left(0\right)\subset {\mathbb{C}}^{N}$  .
Definition: A point $P\in X$  is called a special point of the collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  of 1-forms on the variety $X$  if there exists a sequence $\left\{{P}_{m}\right\}$  of points from the non-singular part ${X}_{reg}$  of the variety $X$  converging to $P$  such that the sequence ${T}_{{P}_{m}}{X}_{reg}$  of the tangent spaces at the points ${P}_{m}$  has a limit $L$  as $m\to \infty$  (in the Grassmann manifold of $n$  -dimensional vector subspaces of ${\mathbb{C}}^{N}$  ) and the restrictions of the 1-forms ${\omega }_{1}^{\left(i\right)}$  , . . . , ${\omega }_{n-{k}_{i}+1}^{\left(i\right)}$  to the subspace $L\subset {T}_{P}{\mathbb{C}}^{N}$  are linearly dependent for each $i=1,\dots ,s$  .
Definition: The collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  of 1-forms has an isolated special point on the germ $\left(X,0\right)$  if it has no special points on $X$  in a punctured neighbourhood of the origin.
Remarks. 1. If the 1-forms ${\omega }_{j}^{\left(i\right)}$  are complex analytic, the property to have an isolated special point is a condition on the classes of these 1-forms in the module ${\Omega }_{X,0}^{1}={\Omega }_{{\mathbb{C}}^{N},0}^{1}/\left\{f\cdot {\Omega }_{{\mathbb{C}}^{N},0}^{1}+df\cdot {\mathcal{O}}_{{\mathbb{C}}^{N},0}|f\in {\mathcal{J}}_{X}\right\}$  of germs of 1-forms on the variety $X$  ( ${\mathcal{J}}_{X}$  is the ideal of germs of holomorphic functions vanishing on $X$  ).
2. For the case $s=1$  (and therefore ${k}_{1}=n$  ), i.e. for one 1-form $\omega$  , there exists a notion of a singular point of the 1-form $\omega$  on $X$  (see, e.g., [7). It is defined in terms of a Whitney stratification of the variety $X$  . A point $x\in X$  is a singular point of the 1-form $\omega$  on the variety $X$  if the restriction of the 1-form $\omega$  to the stratum of $X$  containing $x$  is equal to zero at the point $x$  . (One should consider points of all zero-dimensional strata as singular ones.) One can easily see that a special point of the 1-form $\omega$  on the variety $X$  is singular, but not vice versa. (E.g. the origin is a singular point of the 1-form $dx$  on the cone $\left\{{x}^{2}+{y}^{2}+{z}^{2}=0\right\}$  , but not a special one.) On a smooth variety these two notions coincide.
The notion of a non-degenerate special (singular) point of a collection of germs of 1-forms on a smooth variety was introduced in [6. The index of a non-degenerate point of a collection of germs of holomorphic 1-forms is equal to 1.
Let ${\mathcal{ℒ}}^{\mathbf{k}}={\prod }_{i=1}^{s}{\prod }_{j=1}^{n-{k}_{i}+1}{\mathbb{C}}_{ij}^{N*}$  be the space of collections of linear functions on ${\mathbb{C}}^{N}$  (i.e. of 1-forms with constant coefficients).
Proposition 1. There exists an open and dense subset $U\subset {\mathcal{ℒ}}^{\mathbf{k}}$  such that each collection $\left\{{\ell }_{j}^{\left(i\right)}\right\}\in U$  has only isolated special points on $X$  and, moreover, all these points belong to the smooth part ${X}_{reg}$  of the variety $X$  and are non-degenerate.
• Proof. Let $Y\subset X×{\mathcal{ℒ}}^{\mathbf{k}}$  be the closure of the set of pairs $\left(x,\left\{{\ell }_{j}^{\left(i\right)}\right\}\right)$  where $x\in {X}_{reg}$  and the restrictions of the linear functions ${\ell }_{1}^{\left(i\right)}$  , . . . , ${\ell }_{n-{k}_{i}+1}^{\left(i\right)}$  to the tangent space ${T}_{x}{X}_{reg}$  are linearly dependent for each $i=1,\dots ,s$  .
Let $\pi :Y\to {\mathcal{ℒ}}^{\mathbf{k}}$  be the projection to the second factor. One has $codimY={\sum }_{i=1}^{s}{k}_{i}=n$  and therefore $dimY=dim{\mathcal{ℒ}}^{\mathbf{k}}$  . Moreover, $Y\\left({X}_{reg}×{\mathcal{ℒ}}^{\mathbf{k}}\right)$  is a proper subvariety of $Y$  and therefore its dimension is strictly smaller than $dim{\mathcal{ℒ}}^{\mathbf{k}}$  . A generic point of the space ${\mathcal{ℒ}}^{\mathbf{k}}$  is a regular value of the map $\pi$  which means that it has only finitely many preimages, all of them belong to ${X}_{reg}×{\mathcal{ℒ}}^{\mathbf{k}}$  and the map $\pi$  is non-degenerate at them. This implies the statement.
Corollary 1. Let $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  be a collection of 1-forms on $X$  with an isolated special point at the origin. Then there exists a deformation $\left\{{\stackrel{~}{\omega }}_{j}^{\left(i\right)}\right\}$  of the collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  whose special points lie in ${X}_{reg}$  and are non-degenerate. Moreover, as such a deformation one can use $\left\{{\omega }_{j}^{\left(i\right)}+\lambda {\ell }_{j}^{\left(i\right)}\right\}$  with a generic collection $\left\{{\ell }_{j}^{\left(i\right)}\right\}\in {\mathcal{ℒ}}^{\mathbf{k}}$  , $\lambda \ne 0$  small enough.
Corollary 2. The set of collections of holomorphic 1-forms with a non-isolated special point at the origin has infinite codimension in the space of all holomorphic collections.

2 Local Chern obstructions

Let $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  be a collection of germs of 1-forms on $\left(X,0\right)$  with an isolated special point at the origin. Let $\nu :\stackrel{^}{X}\to X$  be the Nash transformation of the variety $X\subset {B}_{\varepsilon }\left(0\right)$  defined as follows. Let $G\left(n,N\right)$  be the Grassmann manifold of $n$  -dimensional vector subspaces of ${\mathbb{C}}^{N}$  .
There is a natural map $\sigma :{X}_{reg}\to {B}_{\varepsilon }\left(0\right)×G\left(n,N\right)$  which sends a point $x\in {X}_{reg}$  to $\left(x,{T}_{x}{X}_{reg}\right)$  . The Nash transform $\stackrel{^}{X}$  of the variety $X$  is the closure of the image $Im\sigma$  of the map $\sigma$  in ${B}_{\varepsilon }\left(0\right)×G\left(n,N\right)$  , $\nu$  is the natural projection. The Nash bundle $\stackrel{^}{T}$  over $\stackrel{^}{X}$  is a vector bundle of rank $n$  which is the pullback of the tautological bundle on the Grassmann manifold $G\left(n,N\right)$  . There is a natural lifting of the Nash transformation to a bundle map from the Nash bundle $\stackrel{^}{T}$  to the restriction of the tangent bundle $T{\mathbb{C}}^{N}$  of ${\mathbb{C}}^{N}$  to $X$  . This is an isomorphism of $\stackrel{^}{T}$  and $T{X}_{reg}\subset T{\mathbb{C}}^{N}$  over the non-singular part ${X}_{reg}$  of $X$  .
The collection of 1-forms $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  gives rise to a section $\stackrel{^}{\omega }$  of the bundle $\stackrel{^}{\mathbb{T}}{=}^{s}{{\oplus }_{i=1}}^{n-{k}_{i}+1}{\oplus }_{j=1}{\stackrel{^}{T}}_{i,j}^{*}$  where ${\stackrel{^}{T}}_{i,j}^{*}$  are copies of the dual Nash bundle ${\stackrel{^}{T}}^{*}$  over the Nash transform $\stackrel{^}{X}$  numbered by indices $i$  and $j$  . Let $\stackrel{^}{\mathbb{D}}\subset \stackrel{^}{\mathbb{T}}$  be the set of pairs $\left(x,\left\{{\alpha }_{j}^{\left(i\right)}\right\}\right)$  where $x\in \stackrel{^}{X}$  and the collection $\left\{{\alpha }_{j}^{\left(i\right)}\right\}$  of elements of ${\stackrel{^}{T}}_{x}^{*}$  (i.e. of linear functions on ${\stackrel{^}{T}}_{x}$  ) is such that ${\alpha }_{1}^{\left(i\right)}$  , . . . , ${\alpha }_{n-{k}_{i}+1}^{\left(i\right)}$  are linearly dependent for each $i=1,...,s$  . The image of the section $\stackrel{^}{\omega }$  does not intersect $\stackrel{^}{\mathbb{D}}$  outside of the preimage ${\nu }^{-1}\left(0\right)\subset \stackrel{^}{X}$  of the origin.
The map $\stackrel{^}{\mathbb{T}}\\stackrel{^}{\mathbb{D}}\to \stackrel{^}{X}$  is a fibre bundle. The fibre ${W}_{x}={\stackrel{^}{\mathbb{T}}}_{x}\{\stackrel{^}{\mathbb{D}}}_{x}$  of it is $\left(2n-2\right)$  -connected, its homology group ${H}_{2n-1}\left({W}_{x};\mathbb{Z}\right)$  is isomorphic to $\mathbb{Z}$  and has a natural generator: see, e.g., [6. The latter fact implies that the fibre bundle $\stackrel{^}{\mathbb{T}}\\stackrel{^}{\mathbb{D}}\to \stackrel{^}{X}$  is homotopically simple in dimension $2n-1$  , i.e. the fundamental group ${\pi }_{1}\left(\stackrel{^}{X}\right)$  of the base acts trivially on the homotopy group ${\pi }_{2n-1}\left({W}_{x}\right)$  of the fibre, the last one being isomorphic to the homology group ${H}_{2n-1}\left({W}_{x};\mathbb{Z}\right)$  : see, e.g., [14.
Definition: The local Chern obstruction $C{h}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}$  of the collections of germs of 1-forms $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  on $\left(X,0\right)$  at the origin is the (primary, and in fact the only) obstruction to extend the section $\stackrel{^}{\omega }$  of the fibre bundle $\stackrel{^}{\mathbb{T}}\\stackrel{^}{\mathbb{D}}\to \stackrel{^}{X}$  from the preimage of a neighbourhood of the sphere ${S}_{\varepsilon }=\partial {B}_{\varepsilon }$  to $\stackrel{^}{X}$  , more precisely its value (as an element of ${H}^{2n}\left({\nu }^{-1}\left(X\cap {B}_{\varepsilon }\right),{\nu }^{-1}\left(X\cap {S}_{\varepsilon }\right);\mathbb{Z}\right)$  ) on the fundamental class of the pair $\left({\nu }^{-1}\left(X\cap {B}_{\varepsilon }\right),{\nu }^{-1}\left(X\cap {S}_{\varepsilon }\right)\right)$  .
The definition of the local Chern obstruction $C{h}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}$  can be reformulated in the following way. Let ${\mathcal{D}}_{X}^{\mathbf{k}}\subset {\mathbb{C}}^{N}×{\mathcal{ℒ}}^{\mathbf{k}}$  be the closure of the set of pairs $\left(x,\left\{{\ell }_{j}^{\left(i\right)}\right\}\right)$  such that $x\in {X}_{reg}$  and the restrictions of the linear functions ${\ell }_{1}^{\left(i\right)}$  , . . . , ${\ell }_{n-{k}_{i}+1}^{\left(i\right)}$  to ${T}_{x}{X}_{reg}\subset {\mathbb{C}}^{N}$  are linearly dependent for each $i=1,\dots ,s$  . (For $s=1$  , $\mathbf{k}=\left\{n\right\}$  , ${\mathcal{D}}_{X}^{\mathbf{k}}$  is the (non-projectivized) conormal space of $X$  [15.) The collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  of germs of 1-forms on $\left({\mathbb{C}}^{N},0\right)$  defines a section $\stackrel{ˇ}{\omega }$  of the (trivial) fibre bundle ${\mathbb{C}}^{N}×{\mathcal{ℒ}}^{\mathbf{k}}\to {\mathbb{C}}^{N}$  . Then $C{h}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}=\left(\stackrel{ˇ}{\omega }\left({\mathbb{C}}^{N}\right)\circ {\mathcal{D}}_{X}^{\mathbf{k}}{\right)}_{0}$  where $\left(\cdot \circ \cdot {\right)}_{0}$  is the intersection number at the origin in ${\mathbb{C}}^{N}×{\mathcal{ℒ}}^{\mathbf{k}}$  . This description can be considered as a generalization of an expression of the local Euler obstruction as a microlocal intersection number defined in [9, see also [10,Sections5.0.3and5.2.1.
Remarks. 1. On a smooth manifold $X$  the local Chern obstruction $C{h}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}$  coincides with the index $in{d}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}$  of the collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  defined in [6.
2. The local Euler obstruction is defined for vector fields as well as for 1-forms. One can see that vector fields are not well adapted to a definition of the local Chern obstruction. A more or less direct version of the definition above for vector fields demands to consider vector fields on a singular variety $X\subset {\mathbb{C}}^{N}$  to be sections $v=v\left(x\right)$  of $T{\mathbb{C}}^{N}\text{}|{\text{}}_{X}$  such that $v\left(x\right)\in {T}_{x}X\subset {T}_{x}{\mathbb{C}}^{N}$  ( $dim{T}_{x}X$  is not constant).
(Traditionally vector fields tangent to smooth strata of the variety $X$  are considered.) There exist only continuous (non-trivial, i.e. with $s>1$  ) collections of such vector fields ”on $X$  ” with isolated special points, but not holomorphic ones.
3. The definition of the local Chern obstruction $C{h}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}$  may also be formulated in terms of a collection $\left\{{\omega }^{\left(i\right)}\right\}$  of germs of 1-forms with values in vector spaces ${L}_{i}$  of dimensions $n-{k}_{i}+1$  . Therefore (via differentials) it is also defined for a collection $\left\{{f}^{\left(i\right)}\right\}$  of germs of maps ${f}^{\left(i\right)}:\left({\mathbb{C}}^{N},0\right)\to \left({\mathbb{C}}^{n-{k}_{i}+1},0\right)$  (just as the Euler obstruction is defined for a germ of a function).
Being a (primary) obstruction, the local Chern obstruction satisfies the law of conservation of number, i.e. if a collection of 1-forms $\left\{{\stackrel{~}{\omega }}_{j}^{\left(i\right)}\right\}$  is a deformation of the collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  and has isolated special points on $X$  , then $C{h}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}=\sum C{h}_{X,Q}\left\{{\stackrel{~}{\omega }}_{j}^{\left(i\right)}\right\}$  where the sum on the right hand side is over all special points $Q$  of the collection $\left\{{\stackrel{~}{\omega }}_{j}^{\left(i\right)}\right\}$  on $X$  in a neighbourhood of the origin. With Corollary  1 this implies the following statements.
Proposition 2. The local Chern obstruction $C{h}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}$  of a collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  of germs of holomorphic 1-forms is equal to the number of special points on $X$  of a generic (holomorphic) deformation of the collection.
This statement is an analogue of Proposition 2.3 in [13.
Proposition 3. If a collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  of 1-forms on a compact (say, projective) variety $X$  has only isolated special points, then the sum of the local Chern obstructions of the collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  at these points does not depend on the collection and therefore is an invariant of the variety.
It is reasonable to consider this sum as ( $\left(-1{\right)}^{n}$  times) the corresponding Chern number of the singular variety $X$  .
Let $\left(X,0\right)$  be an isolated complete intersection singularity. As it was mentioned above, a collection of germs of 1-forms $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  on $\left(X,0\right)$  with an isolated special point at the origin has an index $in{d}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}$  which is an analogue of the GSV–index of a vector field: [5. The fact that both the Chern obstruction and the index satisfy the law of conservation of number and they coincide on a smooth manifold yields the following statement.
Proposition 4. For a collection $\left\{{\omega }_{j}^{\left(i\right)}\right\}$  of germs of 1-forms on an isolated complete intersection singularity $\left(X,0\right)$  the difference $in{d}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}-C{h}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}$  does not depend on the collection and therefore is an invariant of the germ of the variety.
Since, by Proposition  1 , $C{h}_{X,0}\left\{{\ell }_{j}^{\left(i\right)}\right\}=0$  for a generic collection $\left\{{\ell }_{j}^{\left(i\right)}\right\}$  of linear functions on ${\mathbb{C}}^{N}$  , one has the following statement.
Corollary 3. One has $C{h}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}=in{d}_{X,0}\left\{{\omega }_{j}^{\left(i\right)}\right\}-in{d}_{X,0}\left\{{\ell }_{j}^{\left(i\right)}\right\}$  for a generic collection $\left\{{\ell }_{j}^{\left(i\right)}\right\}$  of linear functions on ${\mathbb{C}}^{N}$  .
References

1. Ch. Bonatti, X. Gómez-Mont: The index of holomorphic vector fields on singular varieties I. Astérisque 222, 9–35 (1994).
2. J.-P. Brasselet, Lê Du͂ng Tráng, J. Seade: Euler obstruction and indices of vector fields. Topology 39, 1193–1208 (2000).
3. J.-P. Brasselet, D. Massey, A. J. Parameswaran, J. Seade: Euler obstruction and defects of functions on singular varieties. J. London Math. Soc. (2) 70, 59–76 (2004).
4. J.-P. Brasselet, M.-H. Schwartz: Sur les classes de Chern d'un ensemble analytique complexe. In: Caractéristique d'Euler-Poincaré, Astérisque 82–83, 93–147 (1981).
5. W. Ebeling, S. M. Gusein-Zade: Indices of 1-forms on an isolated complete intersection singularity. Moscow Math. J. 3, 439–455 (2003).
6. W. Ebeling, S. M. Gusein-Zade: Indices of vector fields or 1-forms and characteristic numbers. math.AG/0303330. Bull. London Math. Soc. (to appear).
7. W. Ebeling, S. M. Gusein-Zade: Radial index and Euler obstruction of a 1-form on a singular variety. math.AG/0402388. Geom. Dedicata (to appear).
8. X. Gómez-Mont, J. Seade, A. Verjovsky: The index of a holomorphic flow with an isolated singularity. Math. Ann. 291, 737–751 (1991).
9. M. Kashiwara, P. Schapira: Sheaves on Manifolds. Springer-Verlag, 1990.
10. J. Schürmann: Topology of Singular Spaces and Constructible Sheaves. Birkhäuser, 2003.
11. M.-H. Schwartz: Classes caractéristiques définies par une stratification d'une variété analytique complexe. C. R. Acad. Sci. Paris Sér. I Math. 260, 3262–3264, 3535–3537 (1965).
12. J. A. Seade, T. Suwa: A residue formula for the index of a holomorphic flow. Math. Ann. 304, 621–634 (1996).
13. J. Seade, M. Tibǎr, A. Verjovsky: Milnor numbers and Euler obstruction. math.CV/0311109.
14. N. Steenrod: The Topology of Fibre Bundles. Princeton Math. Series, Vol. 14, Princeton University Press, Princeton, N. J., 1951.
15. B. Teissier: Variétés polaires. II. Multiplicités polaires, sections planes, et conditions de Whitney. In: Algebraic geometry (La Rábida, 1981), Lecture Notes in Math., Vol. 961, Springer, Berlin, 1982, pp. 314–491.

Universitat Hannover, Institut fur Mathematik, Postfach 6009, D-30060 Hannover, Germany. E-mail address : ebeling@math.uni-hannover.de Moscow State University, Faculty of Mechanics and Mathematics, Moscow, 119992, Russia. E-mail address : sabir@mccme.ru