Bivariant theories of constructible functions and Grothendieck transformations (Q697613): Difference between revisions

The paper under review deals with a class of bivariant theories in the sense of \textit{W. Fulton} and \textit{R. MacPherson} [Categorical framework for the study of singular spaces, Mem. Am. Math. Soc. 243 (1981; Zbl 0467.55005)]. One of them, the {operational bivariant theory}, can always be constructed from any covariant functor. Let \(C\) be a category with a final object \(pt\), \(T_*\) a covariant theory with a distinguished element \(1\) in \(T_* (pt)\) and \(\mathbb T^{op}\) the associated operational bivariant theory, \(ev:\mathbb T^{op}(X\to pt) \to T_* (X)\) the {evaluation map}. In the above-mentioned article, Fulton and MacPherson state that the operational bivariant theory is the coarsest bivariant theory one can associate to \(T_*\): If \(\mathbb B\) is any bivariant theory on \(C\) and there are homomorphisms \(\Phi (X): B_*(X)= \mathbb B(X\to pt)\) to \(T_* (X)\), covariant for confinded morphisms, and taking \(1\in B^*(pt)\) to \(1\in T_*(pt)\), then there is a unique Grothendieck transformation \(\mathbb B \to \mathbb T^{op} \) of the bivariant theories such that the associated map \(B_*(X)\to \mathbb T^{op} (X\to pt)\) followed by the evaluation map \(ev: \mathbb T^{op}(X\to pt) \to T_*(X)\) is the given map \( \Phi (X):B_*(X)\to T_*(X)\). From the author's introduction: ``In fact, we will see that in general there does not exist such a Grothendieck transformation \(\mathbb B \to \mathbb T^{op}\).\dots So, a reasonable problem, motivated by the above statement due to Fulton and MacPherson, is how to modify or correct it \dots'' Theorem A. Suppose that in the above situation the product \(\times\) is defined on \(B_*\) and \(T_*\) and \(\Phi :B_*\to T_*\) preserves this operation. Then there is a bivariant subgroup \(\widetilde {\mathbb B} \subset \mathbb B \) and a Grothendieck transformation \(\widetilde {\mathbb B} \to \mathbb T^{op}\) such that the associated map \(B_*(X)\to \mathbb T^{op} (X\to pt)\) followed by the evaluation map \(ev: \mathbb T^{op}(X\to pt) \to T_*(X)\) is the given map \(B_*(X)\to T_*(X)\). The theorem is motivated by Theorem B. Let \(\mathbb H^{op}\) be the operational bivariant homology theory. Then there is a bivariant theory \(\widetilde {\mathbb F}\) of constructible functions and there is a Grothendieck transformation \(\gamma : \widetilde {\mathbb F} \to \mathbb H^{op}\) such that the associated map \(F(X) = \widetilde {\mathbb F} (X\to pt)\to \mathbb H^{op} (X\to pt)\) followed by the evaluation map \(ev: \mathbb H^{op} (X\to pt) \to H_*(X)\) is the Chern-Schwartz-MacPherson class \(c_*:F(X)\to H_*(X)\).