On vanishing criteria of $L^2$-Betti numbers of groups
Pablo Sánchez-Peralta
TL;DR
The paper advances the theory of $L^2$-Betti numbers of groups by proving higher-degree vanishing criteria that transfer information from infinite-index subgroups, via purely algebraic and groupoid methods. It constructs and analyzes the crossed product $S=\,\\mathcal{U}(N)*G/N$ to translate finite-dimensional behavior of $S$-modules into vanishing of $L^2$-invariants for $G$, and it extends Sauer–Thom and Gaboriau-type results to a broader algebraic setting using translation groupoids and groupoid algebras. A central achievement is a fully algebraic proof of Gaboriau’s first Betti number vanishing, together with a higher-degree analogue, and the paper provides new vanishing criteria that connect subnormal and Hillman-type questions to groupoid techniques. Collectively, these results furnish powerful tools to control $L^2$-Betti numbers from substructures, with potential applications to 3-manifold groups and related areas in geometric group theory. The work highlights the utility of groupoid algebras and dimension theory (via SMRF/BSMRF and von Neumann dimensions) as a robust alternative to measured equivalence relations in studying $L^2$-invariants.
Abstract
Let $G$ be a countable group and $k$ a positive integer, we show that the $L^2$-Betti numbers of the group $G$ vanish up to degree $k$ provided that there is some infinite index subgroup $H$ with finite $k$th $L^2$-Betti number containing a normal subgroup of $G$ whose $L^2$-Betti numbers are all zero below degree $k$. This generalizes previous criteria of both Sauer and Thom, and Peterson and Thom. In addition, we exhibit a purely algebraic proof of a well-known theorem of Gaboriau concerning the first $L^2$-Betti number which was requested by Bourdon, Martin and Valette. Finally, we provide evidence of a positive answer for a question posted by Hillman that wonders whether the above statement holds for $k = 1$ and $H$ containing a subnormal subgroup instead.
