Bi-relative algebraic \(K\)-theory and topological cyclic homology (Q2433887)

The theorem proved in this paper says that the \(p\)-primary part of the homotopy fiber of the cyclotomic trace \(K\to TC\) satisfies excision, making the way open for calculations of algebraic \(K\)-theory in many non-smooth examples through examination of the corresponding situation in topological cyclic homology. More precisely, let \[ \begin{tikzcd} A\rar["f"]\dar & B\dar\\ A/I\rar & B/J \end{tikzcd} \] be a commuting diagram of unital associative rings with \(I\subseteq A\) and \(J\subseteq B\) two-sided ideals such that \(f\) restricted to \(I\) gives an isomorphism from \(I\) to \(J\) (i.e., the square is cartesian). A central question in algebraic \(K\)-theory from the very start of the subject has been: to what extent does \(K\)-theory satisfies excision? I.e., how far from exact will the Mayer-Vietoris sequence in \(K\)-theory associated to the cartesian diagram above be? Let \(K_q(A,B,I)\) be the \(q\)th homotopy group of the iterated homotopy fiber of the algebraic \(K\)-theory applied to the square above. Then \(K_q(A,B,I)\) can be viewed as the correction term, measuring the failure of the Mayer-Vietoris sequence to be exact. The paper under review gives the answer to the \(p\)-primary part of this question. Theorem. Let \(p\) be a prime, \(\upsilon\) a natural number and \(q\) an integer. Then the cyclotomic trace map \[ K_q(A,B,I,\mathbb Z/p^\upsilon)\to \{\mathrm{TC}^n_q(A,B,I;p,\mathbb Z/p^\upsilon)\}_{n\geq 1} \] is an isomorphism of pro-abelian groups. The groups in question are the homotopy groups with coefficients in \(\mathbb Z/p^\upsilon\), and \(\mathrm{TC}^n_q(-;p,\mathbb Z/p^\upsilon)\) is a theory derived from \textit{M. Böksted, W. C. Hsiang} and \textit{I. Madsen}'s [Invent. Math. 111, 465--539 (1993; Zbl 0804.55004)] setup. The proof is roughly as follows: First of all it is enough to prove the statement that \(I\) can be embedded as a two-sided ideal in some free unital associative ring. Secondly, under this assumption on \(I\) both \(K\)-theory and \(\mathrm{TC}^n\)-theory with \(\mathbb Z/p^\upsilon\)-coefficients satisfy excision ``infinitesimally'', that is, the pro-abelian groups \(\{K_q(A,B,I^m,\mathbb Z/p^\upsilon)\}_{m\geq 1}\) and \(\{\mathrm{TC}^n_q(A,B,I^m,\mathbb Z/p^\upsilon)\}_{n,m\geq 1}\) are both zero (and hence isomorphic). But as soon as one knows this infinitesimal result one obtain the theorem by using \textit{R. McCarthy}'s theorem (which says that the fiber of the cyclotomic trace is ``locally constant'') [Acta Math. 179, 197--222 (1997; Zbl 0913.19001)]. The proofs of the ``infinitesimal'' results follow \textit{A. A. Suslin} and \textit{M. Wodzicki}'s [Ann. Math. (2) 136, 51--122 (1992; Zbl 0756.18008)], but require some care for the current situation, and some seemingly new constructions with occasional hands-on encounters. The result is very interesting since it opens up for calculations of algebraic \(K\)-theory in non-smooth situations, examples of which are displayed in the paper. Given the result of the paper, very many other calculations can be envisioned. The corresponding rational statement was conjectured by among others Loday and Weibel in the mid 80's, and was proved recently by \textit{G. Cortiñas} [Invent. Math. 164, 143--173 (2006; Zbl 1092.19001)]. The \(p\)-primary statement was conjectured soon after Bökstedt, Hsiang and Madsen invented topological cyclic homology. There is an extension of these results to connective ring spectra due to Kittang and the reviewer.