Some remarks concerning mod-\(n\) \(K\)-theory. (Q5950699): Difference between revisions

The existence of a suitable spectral sequence relating motivic cohomology to algebraic \(K\)-theory, once predicted by A. Beilinson, has recently been established for smooth quasi-projective varieties over a field by the results of E. Friedlander, A. Suslin, M. Levine and others in the 1990s. Among other crucial properties, this spectral sequence has the expected multiplicative behavior with respect to the cup product in motivic cohomology and the usual product in algebraic \(K\)-theory) and moreover a good multiplicative ``mod-\(n\) reduction'' relating mod-\(n\) motivic cohomology to mod-\(n\) algebraic \(K\)-theory. In the paper under review, the authors continue the study of this spectral sequence, mainly with regard to its mod-\(n\) reduction property. Working with smooth quasi-projective varieties \(X\) over an algebraically closed field \(k\), they add three types of results in this direction. The first result compares the spectral sequence to its localization by the Bott element for \(n\in\mathbb{N}\), \(n\) being invertible in \(k\) and \(n\not\equiv 2\pmod 4\). This comparison leads to a natural map of strongly convergent spectral sequences, on the mod-\(n\) level, which then induces the spectral sequence from mod-\(n\) étale cohomology to mod-\(n\) étale \(K\)-theory of the variety \(X\). The second main result establishes the split surjectivity of the natural map \(K_m(X,\mathbb{Z}/n)\to K^{\text{ét}}_m(X,\mathbb{Z}/n)\) from mod-\(n\) algebraic \(K\)-theory to mod-\(n\) étale \(K\)-theory for \(m\geq 2\dim(X)\) and \(n\in\mathbb{N}\) as before. The proof is based on the first comparison result mentioned above, and the theorem as such refines previous surjectivity results by \textit{W. Dwyer}, \textit{E. Friedlander}, \textit{V. Snaith} and \textit{R. Thomason} [Invent. Math. 66, 481--491 (1982; Zbl 0501.14013)], on the one hand, and an argument by \textit{A. Suslin} [in: Proc. ICM Zürich 1994, 342--351 (1995; Zbl 0841.19003)] on the other. The third result of the present paper establishes the fact that, in mod-\(n\) algebraic \(K\)-theory, the multiplication by the Bott element \(\beta\in K_2(X,\mathbb{Z}/n)\) is generally not injective. It is shown that certain smooth projective threefolds constructed by \textit{S. Bloch} and \textit{H. Esnault} [Math. Ann. 304, No. 2, 303--314 (1996; Zbl 0868.14004)] have non-zero elements in \(K_0(X,\mathbb{Z}/n)\) which are annihilated by a sufficiently high power of the Bott element. As there have been attempts, in the past, to prove that multiplication by the Bott element is always injective, these counterexamples clarify the situation once for all.