The higher \(K\)-theory of complex varieties (Q5935572)

The first main result of the paper reads as follows: Let \(F\) be a field of finite transcendence degree over \(\mathbb C\), and let \(m\) be a power of two. Then the Bott element \(\beta\in K_2(F;\mathbb Z/m)\) is a nonzero divisor in \(K_*(F;\mathbb Z/m)\), and there are isomorphisms of graded rings \[ K_*(F,\mathbb Z/m)\cong (K^M_*(F)/m)[\beta]\cong H^*_{\text{ét}}(F;\mathbb Z/m)[\beta]. \] In order to achieve this, the authors first consider the Adams operations to show that the Bloch-Lichtenbaum and Atiyah-Hirzebruch spectral sequences both degenerate at the \(E_2\)-term and determine the additive structure of K-theory and étale K-theory (as the paper was written before the announcement in [\textit{G. M. Friedlander} and \textit{A. Suslin}, ``The spectral sequence relating algebraic K-theory to motivic cohomology'', preliminary draft (2000) \url{http://www.math.uiuc.edu/K-theory/0432}] it does not rely on the multiplicativity of the Bloch-Lichtenbaum spectral sequence). Comparing these two results the authors establish the ring structure using an induction on the transcendence degree. Furthermore, the integral \(K\)-groups \(K_n(F)\) are \(2\)-divisible for all \(n\) greater than the transcendence degree of \(F\) over \(\mathbb C\). Using the description of \(K_*(F,\mathbb Z/m)\) and \textit{A. Suslin}'s argument for curves and surfaces [in: Proc. 1994 ICM Zürich, Birkhäuser, Basel, 342-351 (1995; Zbl 0841.19003)], the authors prove the Quillen-Lichtenbaum conjecture for smooth complex varieties \(X\): the homomorphism \[ K_n(X;\mathbb Z/m)\to K_n^{\text{ét}}(X;\mathbb Z/m)\cong KU^{-n}(X;\mathbb Z/m) \] is an isomorphism for all \(n\geq\dim(X)-1\), and an injection for \(n=dim(X)-2\), where \(KU^*(X;\mathbb Z/m)\) is the topological K-theory of the associated topological space \(X(\mathbb C)\) with coefficients in \(\mathbb Z/m\). In consequence, if \(K^{\text{rel}}_*(X;\mathbb Z/m)\) is the relative term in the natural long exact sequence \[ \dots K_{n+1}(X)\to KU^{-n-1}(X)\to K^{\text{rel}}_n(X)\to K_{n}(X)\to KU^{-n}(X)\to \dots \] then \(K^{\text{rel}}_n(X)\) is uniquely \(2\)-divisible if \(n\geq dim(X)-2\), and without \(2\)-torsion if \(n=dim(X)-3\). The last of the main results of the paper gives a noncanonical description of the integral \(K\)-groups \(K_n(X)\) for \(n\) above roughly the \(dim(X)\) in terms of direct summands that are either \(2\)-divisible or finite (corresponding to the torsion subgroup of \(KU^{-n}(X)\)). This uses a comparison with Deligne cohomology and the compatibility of Deligne and analytic Chern classes. These results all depend on \textit{V. Voevodsky}'s [``The Milnor conjecture'', preprint (1997)], and extend from \(2\) to any other prime \(\ell\) where one might be able to prove that the norm residue homomorphism \[ K^M_*(F)/\ell^\nu\to H^*_{\text{ét}}(F,Z/\ell^\nu) \] is an isomorphism (called the norm residue conjecture in this paper; the reason for the constant coefficients on the étale cohomology is that all roots of unity are present in \(F\)). The paper is clean and well-written, and contains interesting examples and background.