Integrality of L^2-Betti numbers

From MaRDI portal
Publication:1581826



Abstract: The Atiyah conjecture predicts that the L2-Betti numbers of a finite CW-complex with torsion-free fundamental group are integers. We show that the Atiyah conjecture holds (with an additional technical condition) for direct and inverse limits of directed systems of groups for which it is true. As a corollary it holds for residually torsion-free solvable groups, e.g. for pure braid groups or for positive 1-relator groups with torsion free abelianization. Putting everything together we establish a new class of groups for which the Atiyah conjecture holds, which contains all free groups and in particular is closed under taking subgroups, direct sums, free products, extensions with elementary amenable quotient, and under direct and inverse limits of directed systems. This is a corrected version of an older paper with the same title. The proof of Proposition 2.1 of the earlier version contains a gap, as was pointed out to me by Pere Ara. This gap could not be fixed. Consequently, in this new version everything based on this result had to be removed. Please take the errata to "L2-determinant class and approximation of L2-Betti numbers" into account, which are added, rectifying some unproved statements about "amenable extension". As a consequence, throughout, amenable extensions should be extensions with normal subgroups.


The Atiyah conjecture asserts that if \(G\) is a finitely presentable torsion free group and \(X\) is a finite complex with \(\pi_1(X)\cong G\) the \(L^2\)-Betti numbers \(\beta_i^{(2)}(X)\) are integers. Generalizations allow (limited) torsion in \(G\) and predict the \(L^2\)-Betti numbers are rational, with bounded denominators. The main result of this paper is that the set of groups satisfying the Atiyah conjecture with denominator \(d\) is closed under certain HNN extensions. (In forthcoming work with W. Dicks the author has used the main technical result of this paper to establish closure under arbitrary HNN extensions). Editor's remark: In the meantime, we have been informed by the author that he discovered a gap in one of the proofs in the paper. An erratum is in preparation.











This page was built for publication: Integrality of \(L^2\)-Betti numbers

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1581826)