A finiteness property of infinite resolutions (Q2570677): Difference between revisions

Given a noetherian local ring \((R,{\mathfrak m})\) or a standard graded ring over a field with \({\mathfrak m} = R_+\) and a finitely generated (graded) \(R\)-module \(M\), the minimal (graded) free resolution \[ {\mathbf F}: \cdots\to F_n @>f_n>> F_{n-1}@>f_{n-1}>> \cdots \to F_0 \] of \(M\) is in general infinite. But for instance in the graded case one may ask the question whether there are bases for the free modules \(F_n\) such that the matrices of the maps \(f_n\) with respect to these bases have entries of bounded degree. In this paper, the authors partially answer the following related question: If \(R\) is local and \(I\subset R\) and ideal, is there an integer \(q\) such that \(I^N F_{n-1}\cap \text{Im}\,f_n\subset I^{N-q}\text{Im}\,f_n\) for all \(N\geq q\) and \(n\)? If \(R\) is graded and \(I ={\mathfrak m}\), this uniform Artin-Rees property for free resolutions would follow from a positive answer to the first question. The main result of the paper is that the second question has a positive answer if \(M\) is locally of finite projective dimension of constant rank on the punctured spectrum of \(R\). The proof rests on the existence of an \(\mathfrak m\)-primary ideal generated by elements \(d\) with the property that multiplication by \(c_i\) is homotopic to zero on \({\mathbf F}_{\geq 1}\). In particular, this implies the following uniform annihilation of Tor: There exists an integer \(q\) such that for \(n\gg 0\) and all \(N \geq 0\) the equality \(I^q\text{Tor}^R_n(M,R/I^N)=0\) holds. In fact, the authors prove that the uniform annihilation of Tor, together with a uniform bound connected with the modules of boundaries, is equivalent to the uniform Artin-Rees property for free resolutions. Finally, the authors show that a positive answer to the second question implies that there is a number \(q\) such that any complex which is a perturbation of \(\mathbf F\) to order \((q,q,\dots)\) (i.e. for which \(f_n'\) is given by \(f_n' = f_n +g_n\) with \(\text{Im\,}g_n\subset{\mathfrak m}^q F_{n-1})\) is exact, too. Furthermore, they verify the last property for modules over Cohen-Macaulay rings and ask whether it holds in general.