Bass' \(NK\) groups and \(cdh\)-fibrant Hochschild homology (Q982194)

The \(K\)-theory of a polynomial ring \(R[t]\) contains the \(K\)-theory of \(R\) as a direct summand. The obstruction for the map \(K_n(R[t]) \rightarrow K_n(R)\) given by evaluation at \(t=0\) to being an isomorphism is often related to the singularities of \(\operatorname{Spec} R\). Following a theorem of \textit{C. Traverso} [Ann. Sc. Norm. Super. Pisa, Sci. Fis. Mat., III. Ser. 24, 585--595 (1970; Zbl 0205.50501)] which asserts that \(\mathrm{Pic}(R[t_1]) = \mathrm{Pic}(R)\) implies \(\mathrm{Pic}(R[t_1,t_2]) = \mathrm{Pic}(R)\), Bass asked if \(K(R[t_1]) = K(R)\) implies \(K(R[t_1,t_2]) = K(R)\). He defined the following groups: \(NK_n(R) = \mathrm{Ker} (K_n(R[t]) \rightarrow K_n(R))\) and \(N^pK_n(R) = N(N^{p-1}K_n(R))\). With these notations, his question becomes: does \(NK_n(R)=0\) imply that \(N^2K_n(R)=0\)? After having recalled the fact that \(NK_n(R)\) has the structure of a module over the ring of big Witt vectors \(W(R)\), the authors show that there is an isomorphism of \(W(R)\)-modules \[ N^2K_n(R) \simeq (NK_n(R) \otimes t\mathbb{Q}[t]) \oplus (NK_{n-1}(R) \otimes \Omega^1_{\mathbb{Q}[t]}). \] Thus, the question of Bass can be reformulated as: does \(NK_n(R)=0\) imply that \(NK_{n-1}(R)=0\)? The proof relies on methods developed by the first three authors [Math. Ann. 344, No. 4, 891--922 (2009; Zbl 1189.19002)] and by \textit{J. F. Davis} [``Some remarks on Nil groups in algebraic K-theory'' (2007; \url{arXiv:0803.1641})] which allow to compute \(NK_n(R)\) in terms of the Hochschild homology of \(R\) and of the \(cdh\)-cohomology of the higher Kähler differentials \(\Omega^p\). These computations are used to prove that the answer to Bass' question is positive when \(R\) is essentially of finite type over a field \(F\) of infinite transcendence degree over \(\mathbb{Q}\), but that the answer is negative in general. In [Proc. Am. Math. Soc. 139, No. 4, 1187--1200 (2011; Zbl 1255.19001)], the authors provide a counterexample to Bass' question, namely \(R = F[x,y,z]/(z^2+y^3+x^{10}+x^7y)\) where \(F\) is any algebraic extension of \(\mathbb{Q}\).