On Previdi's delooping conjecture for \(K\)-theory (Q2018347): Difference between revisions

Let \(\mathcal{A}\) be an exact category and let \(\mathbb{K}(\mathcal{A})\) be its nonconnective algebraic \(K\)-theory spectrum, so that the homotopy groups \(\pi_i(\mathbb{K}(\mathcal{A}))\) are isomorphic to the algebraic \(K\)-theory groups \(K_i(\mathcal{A})\), for all \(i \in \mathbb{Z}\), provided that \(\mathcal{A}\) is idempotent-complete. On the other hand, we have the category \(\underleftrightarrow{\lim}\mathcal{A}\) introduced by \textit{A. A. Beilinson} [Lect. Notes Math. 1289, 42--51 (1987; Zbl 0651.14009)]. The main result of the paper under review states that \(\mathbb{K}(\mathcal{A})\) is delooped by \(\mathbb{K}(\underleftrightarrow{\lim}\mathcal{A})\), i.e. there is a homotopy equivalence of spectra \(\mathbb{K}(\mathcal{A})\cong \Omega \mathbb{K}(\underleftrightarrow{\lim}\mathcal{A})\). This proves a modified version of a conjecture of Previdi, stated as Conjecture 1.1 in the paper. As an immediate consequence there are isomorphisms \(K_i(\mathcal{A})\cong K_{i+1}(\underleftrightarrow{\lim}\mathcal{A})\) for every \(\mathcal{A}\) and every \(i\geq 0\), provided that \(\mathcal{A}\) is idempotent complete. Also, we can recover the lower algebraic \(K\)-theory groups \(K_{-i}(\mathcal{A})\) as the 0th \(K\)-group of the idempotent completion of the \(i\)-times iterated Beilinson category \(\underleftrightarrow{\lim}^i\mathcal{A}\). The autor also obtains, as an application of the main result, a purely formal way to associate to a \(1\)-Tate \(R\)-module \(M\) a \(\mathcal{K}\)-torsor with a canonical action of the sheaf.