Hermitian lattices and bounds in \(K\)-theory of algebraic integers (Q273853)

Let \(F\) be a number field of degree \(d\) with ring of integers \(\mathcal{O}_F\) and discriminant \(D_F\). Then \(K_n(\mathcal{O}_F)\), the \(n\)-th Quillen \(K\)-theory group of \(\mathcal{O}_F\), is a finitely-generated abelian group whose rank is given by a well-known formula of \textit{A. Borel} [Ann. Sc. Norm. Super. Pisa, Cl. Sci., IV. Ser. 4, 613--636 (1977; Zbl 0382.57027)]. The current article gives a bound on \(\mathrm{card}\left(K_n(\mathcal{O}_F)_{\mathrm{tors}}\right)\), up to small torsion, in terms of \(n\), \(d\) and \(D_F\): For a finite abelian group \(A\), let \(\mathrm{card}_{\ell}(A)\) denote \(\mathrm{card}(A/B)\), where \(B\) is the subgroup of \(A\) generated by elements of order \(\leq \ell\). The authors prove that there exist \(\alpha, \beta\) both polynomials in \(n\) and \(d\) such that \( \log \mathrm{card}_{\ell}K_n(\mathcal{O}_F)_{\mathrm{tors}}\leq (nd)^{\alpha}|D_F|^\beta \), where \(\ell=\max(d+1,2n+2)\) and \(n\geq 2\). Although these are very large bounds, they improve by an exponential factor on the bounds previously calculated by \textit{C. Soulé} [J. Reine Angew. Math. 517, 209--221 (1999; Zbl 1012.11094)]. The authors follow the general approach of Soulé. The Hurewicz homomorphism maps \(K_n(\mathcal{O}_F)\) to \(H_n(\mathrm{GL}(\mathcal{O}_F))\) and a result of \textit{D. Arlettaz} [J. Pure Appl. Algebra 71, No. 1, 1--12 (1991; Zbl 0728.19002)] shows that there is only small torsion in the kernel. Homology stability theorems in turn show that \(H_n(\mathrm{GL}(\mathcal{O}_F))=H_n(\mathrm{GL}_N(\mathcal{O}_F))\) for \(N\geq 2n+1\). The authors study the torsion in these latter homology groups by using the complex of cellular chains on \(\mathrm{GL}_N(\mathcal{O}_F)\backslash\tilde{W}\), where \(\tilde{W}\) is the \textit{well-rounded retract} of \textit{A. Ash} [Math. Ann. 225, no. 1 (1977), 69-76] together with a lemma of Gabber which gives a method for determining bounds on the torsion in the homology of a complex of free abelian groups. Estimating the number of cells of given dimension in \(\mathrm{GL}_N(\mathcal{O}_F)\backslash\tilde{W}\) involves techniques from the geometry of numbers. The improvements over Soulé's estimates in the article under review are based on better estimates on Hermitian lattices, building on previous work of the first author [J. Number Theory 121, No. 2, 305--323 (2006; Zbl 1130.11066)].