On the Cohomology of GL ₂ and SL ₂ over Imaginary Quadratic Fields (Q6925282)

The article reports on extensive cohomology computations for the groups \(\mathrm{GL}_2(\mathcal{O})\) and \(\mathrm{SL}_2(\mathcal{O})\), where \(\mathcal{O}\) is the ring of integers of an imaginary quadratic field of discriminant \(D\) with \(|D| \leq 2099\) (resp. \(|D|\leq 1247\) in the \(\mathrm{SL}_2\) case). The computations are based on the Voronoi complex, which allows to calculate the integral cohomology away from the primes \(2\) and \(3\); this means that \(2\)- and \(3\)-torsion in the calculations might differ from the integral group cohomology.\N\NThe authors make a number of interesting observations.\N\begin{itemize}\N\item An exponential growth of torsion homology seems to occur for \(\mathrm{GL}_2\) as \(|D| \to \infty\) but is not apparent from the data for \(\mathrm{SL}_2\).\N\item The torsion away from the primes \(2\) and \(3\) is a square.\N\item A lower bound of Rohlfs for the cuspidal cohomology of \(\mathrm{SL}_2(\mathcal{O})\) appears to be sharp in many cases if \(|D| \not\equiv 0 \bmod 4\). For \(|D| \equiv 0 \bmod 4\) the gap between the bound and the actual size of cuspidal cohomology seems to grow with \(|D|\).\N\end{itemize}\NDetailed tables summarizing the calculated data are included in the article.