Bloch–Ogus properties for topological cycle theory (Q4951735)

This paper provides a unified construction of both a topological cycle cohomology and a topological cycle homology theory for quasi-projective complex varieties (or reduced schemes locally of finite type). It is proven that for smooth varieties, topological cycle cohomology coincides with the morphic cohomology earlier defined by the author and \textit{H. B. Lawson} [Ann. Math. (2) 136, 361-8 (1992; Zbl 0788.14014)] and that topological cycle homology is always equal to Lawson homology. The author constructs two families \({\mathcal M}(-,s)\), \({\mathcal L}(-,r)\) of complexes of Zariski presheaves whose ``cohomology groups'' are respectively the topological cycle cohomology \[ {\mathcal H}^j(X,s)\equiv \text{Hom}_{{\mathcal D}_X}(\mathbb{Z}_X,{\mathcal M}(-,s)[j]) \] and the topological cycle homology \[ {\mathcal H}_i(X,r)\equiv \text{Hom}_{{\mathcal D}_X}(\mathbb{Z}_X[i],{\mathcal L}(-,r)). \] The definition of \({\mathcal L}(-,r)\) is made in terms of the topological abelian monoid of effective cocycles and the definition of \({\mathcal M}(-,s)\) uses a topological abelian monoid of effective relative \(r\)-cycles on a variety \(X\) with values in a projective variety \(Y\), whose construction when \(X\) is not normal relies on Voevodsky's qfh-topology [see \textit{A. Suslin} and \textit{V. Voevodsky}, in: Cycles, transfers, and motivic homology theories, Ann. Math. Stud. 143, 10-86 (2000; Zbl 1019.14004)]. Thanks to this presentation, the author proves that topological cycle theory satisfies the axioms of \textit{S. Bloch} and \textit{A. Ogus} for a Poincaré duality with supports [Ann. Sci. Éc. Norm. Supér., IV. Sér. 7, 181-201 (1974; Zbl 0307.14008)] simplifying, clarifying and generalizing earlier duality theorems due to the author and Lawson [cf. \textit{E. M. Friedlander}, Compos. Math. 110, 127-162 (1998; Zbl 0915.14004) and Topology 36, No. 2, 533-565 (1997; Zbl 0889.14004)]. In the last paragraph, it is shown that topological cycle theory satisfies the axioms of \textit{H. Gillet} [Adv. Math. 40, 203-289 (1981; Zbl 0478.14010)] for the existence of a cohomology theory and one can then expect, as the author says, a theory of Chern classes valued in topological cycle cohomology and a Riemann-Roch formula. Topological cycle theory, as presented in the current paper, shares considerable technical difficulties with former approaches, but it has the major advantage of being conceptually clear, natural and elegant.