Total Betti numbers of modules of finite projective dimension

From MaRDI portal



Abstract: The Buchsbaum-Eisenbud-Horrocks Conjecture predicts that if M is a non-zero module of finite length and finite projective dimension over a local ring R of dimension d, then the i-th Betti number of M is at least d choose i. This conjecture implies that the sum of all the Betti numbers of such a module must be at least 2^d. We prove the latter holds in a large number of cases.


The long standing Buchsbaum-Eisenbud-Horrocks conjecture on the Betti numbers of non zero finitely generated modules \(M\) with finite projective dimension over a commutative Noetherian ring \(R\), [\textit{D. Eisenbud} and \textit{D. Buchsbaum}, Am. J. Math. 99, 447--485 (1997; Zbl 0373.13006)], implies that the total rank \(\sum_i \beta_i(M)\) must be at least \(2^c\) where \(c=\text{height}_R(\mathrm{ann}_R(M))\). In the paper, the author proves that the above inequality holds in a larger number of cases. He supposes that \(R\) is a local Noetherian commutative ring and \(M\) has a finite length. He proves Theorem 2. Assume \((R, m, k)\) is a local (Noetherian, commutative) ring of Krull dimension \(d\) and \(M\) is a non-zero \(R\) module of finite length and finite projective dimension. If either (1) \(R\) is the quotient of a regular local ring by a regular sequence of elements and \(2\) is invertible in \(R\), or (2) \(R\) contains \(\mathbb Z /p\) as a subring for an odd prime \(p\), then \(\sum \beta_i(M) \geq 2^d\). Moreover, if the assumptions in (1) hold and \(\sum \beta_i(M)=2^d\), then \(M\) is isomporphic to the quotient of \(R\) by a regular sequence of \(d\) elements. As a consequence, the inequality holds when the ring \(R\) and the module \(M\) are as in the Buchsbaum-Eisenbud-Horrocks conjecture. He proves Theorem 1. Let \(R\) be a commutative Noetherian ring such that spec \((R)\) is connected, and let \(M\) be a nonzero, finitely generated \(R\)-module of finite projective dimension. Let \(P\) be a finite projective resolution of \(M\), and suppose that (1) \(R\) is locally a complete intersection and \(M\) is \(2\)-torsion free or (2) \(R\) contains \(\mathbb Z/p\) as a subring for an odd prime \(p\) then \(\sum_i \text{rank}_R(P_i)\) must be at least \(2^c\) where \(c=\text{height}_R(\mathrm{ann}_R(M))\).




Cited in
(35)








This page was built for publication: Total Betti numbers of modules of finite projective dimension

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2404048)