Intersection homology: general perversities and topological invariance (Q2417019)

\textit{M. Goresky} and \textit{R. MacPherson} [Topology 19, 135--165 (1980; Zbl 0448.55004)] introduced intersection homology as a tool to study singular spaces (pseudomanifolds), having in mind the main application: singular complex algebraic varieties. The construction is purely topological and makes sense for topological objects -- pseudomanifolds. The intersection homology of a pseudomanifold \(X\) is the homology of a complex of \(p\)-allowable chains \(IC^p_*(X)\), where \(p\) is a perversity, i.e. a function which measures the defect of transversality of chains with respect to the strata of \(X\). The authors modify the definition of allowable chains and they introduce \textit{filtered singular simplices} and chains. The resulting homology remains the same. Moreover non-classical perversities are considered, also depending not only on the dimension of the strata, but on the strata themselves. The following question arises: how the resulting homology behaves when the stratification is refined. In particular, what is the relation between the corresponding invariants for the given stratification and the intristic minimal stratification. To that end the authors introduce the notion of \(K\)-perversity and show that for such perversities the resulting homology can be computed from the intristic minimal stratification. The results of the paper hold in a slightly wider generality compared with Goresky-MacPherson's approach, namely \(X\) is assumed to be a locally conical set. The authors avoid the language of sheaf theory, the Mayer-Vietoris sequence is the main tool to compare the considered homology theories.