arXiv subjects
Publications and source records attributed to Andrzej Weber.
We show that for a complete complex algebraic variety the pure component of homology coincides with the image of intersection homology. Therefore pure homology is topologically invariant. To obtain slightly more general results we introduce "image homology" for noncomplete varieties.
We present several approaches to equivariant intersection cohomology. We show that for a complete algebraic variety acted by a connected algebraic group $G$ it is a free module over $H^*(BG)$. The result follows from the decomposition theorem or from a weight argument.
Let $g$ be a reductive Lie algebra over a field of characteristic zero. Suppose $g$ acts on a complex of vector spaces $M$ by $i_λ$ and $L_λ$, which satisfy the identities as contraction and Lie derivative do for smooth differential forms. Out of this data one defines cohomology of the invariants and equivariant cohomology of $M$. We establish Koszul duality between each other.
We present classical and generalized Riemann-Hilbert problem in several contexts arising from $K$-theory and bordism theory. The language of Fredholm pairs turns out to be useful and unavoidable. We propose an abstract formulation of a notion of bordism in the context of Hilbert spaces equipped with splittings.
We consider a possibility of the existence of intersection homology morphism, which would be associated to a map of analytic varieties. We assume that the map is an inclusion of codimension one. Then the existence of a morphism follows from Saito's decomposition theorem. For varieties with conical singularities we show, that the existence of intersection homology morphism is exactly equivalent to the validity of Hard Lefschetz Theorem for links. For varieties with arbitrary analytic singularities we extract a remarkable property, which we call Local Hard Lefschetz.
Let $f:X-->Y$ be a map of algebraic varieties. Barthel, Brasselet, Fieseler, Gabber and Kaup have shown that there exists a homomorphism of intersection homology groups $f^*:IH^*(Y)-->IH^*(X)$ compatible with the induced homomorphism on cohomology. The crucial point in the argument is reduction to the finite characteristic. We give an alternative and short proof of the existence of a homomorphism $f^*$. Our construction is an easy application of the Decomposition Theorem.
There are several ways to generalize characteristic classes for singular algebraic varieties. The simplest ones to describe are Chern-Mather classes obtained by Nash blow up. They serve as an ingredient to construct Chern-MacPherson-Schwartz classes. Unfortunately, they all are defined in homology. There are examples showing, that they do not lie in the image of Poincaré morphism. On the other hand they are represented by an algebraic cycles. Barthel, Brasselet, Fiesler, Kaup and Gabber have shown that, any algebraic cycle can be lifted to intersection homology. Nevertheless, a lift is not unique. The Chern-Mather classes are represented by polar varieties. We show how to define a canonical lift of Chern-Mather classes to intersection homology. Instead of the polar variety alone, we consider it as a term in the whole sequence of inclusions of polar varieties. The inclusions are of codimension one. In this case the lifts are unique.
Suppose $M^{n+1}$ is a complex manifold and K is a hypersurface with isolated singularities. Let $ω$ be a holomorphic form on $M\setminus K$ with the first order pole on K. The Leray residue of such form gives an element in the n-th homology of K which is the Alexander dual to $[ω]\in H^{n+1}(M\setminus K)$. It always lifts to the intersection homology if 0 does not belong to the spectra of a singular points. We assume that singularities are described by the quasihomogeneous equations in certain coordinate systems. Suppose that oscillation indicators of the singular points are greater then -1. Then we find a metric on $K\setminusΣ$ in which the residue form is square integrable (and even L_p-integrable for p>2). Applying the isomorphism of L_p-cohomology and intersection homology we obtain a particular lift of the residue class in homology to intersection homology.
Suppose M is a complex manifold of dimension $n+1$ and K is a hypersurface in M. By Poincaré duality we define a residue morphism $res:H^{k+1}(M\setminus K)\longrightarrow H_{2n-k}(K)$ which generalizes the classical Leray residue morphism to cohomology for smooth K. We assume that K has isolated quasihomogeneous singularities. Suppose $ω$ is a holomorphic form of the type $(n+1,0)$ with the first order pole on K. The purpose of this note is to give a short, self contained proof of a criterion which tells us when the residue of $ω$ lifts to the intersection homology of K.