Cohomology of \(\mathrm{PSL}_2(q)\) (Q2068182)

Let \(G\) be a finite group and \(k\) be an algebraically closed field of prime characteristic \(r\). Of interest here is the cohomology \(H^n(G,V)\) for a \(kG\)-module \(V\) (or extensions of modules more generally), with two distinct goals: (i) development of a general cohomological relationship and (ii) computations of extensions between simple modules for a rank 1 finite group of Lie type in cross (or non-defining) characteristic. The general result relates cohomology to extensions (of lower degree) involving the permutation induced module \({\mathcal L} := \mathrm{Ind}_{H}^{G}k\) for a subgroup \(H\) of \(G\). This generalizes a result of \textit{R. M. Guralnick} and \textit{P. H. Tiep} [Ann. Math. (2) 174, No. 1, 543--559 (2011; Zbl 1237.20039)] that is used to gain information on \(H^1(G,V)\) for \(G\) being a finite group of Lie type in cross characteristic. For a group \(H\), let \(O_{r'}(H)\) denote the largest normal subgroup of \(H\) with order not divisible by \(r\) and \(O^r(H)\) denote the smallest normal subgroup of \(H\) such that \(H/O^r(H)\) is an \(r\)-group. The case of interest is when \(G\) admits a subgroup \(H\) with \(O_{r'}(H) = O^r(H)\). Suppose further that \(V\) is a \(kG\)-module with fixed points \(V^{H} = 0\). The assumptions on \(H\) and \(V\) allow one to conclude that \(H^n(G,V) = 0\) for all \(n\). From this, one obtains the main result that \(H^n(G,V) \cong\mathrm{Ext}_{G}^{n-1}(V^*,{\mathcal L}/k)\), where \(V^*\) denotes the linear dual of \(V\). In degree 1, one has \(H^1(G,V) \cong\mathrm{Hom}_G(V^*, {\mathcal L}/k)\). The author also discusses some further consequences, particularly for dimension computations. \par The majority of the paper focuses on \(G := \mathrm{PSL}_2(q)\) where \(q = p^n\) for a prime \(p \neq r\) (and related groups). The author determines \(\mathrm{Ext}_{G}^n(V,W)\) for all irreducible \(kG\)-modules \(V, W\) for all \(n\). The main tool used here are syzygy modules (or Heller shifts); specifically the well-known relationship \(\mathrm{Ext}_{G}^n(V,W) \cong\mathrm{Hom}_{G}(\Omega^nV,W)\). The task is thus reduced to studying the principal indecomposable modules for \(G\). For odd \(r\), the Sylow \(r\)-subgroups are cyclic and the Brauer tree associated to a block is used to understand the PIMs and obtain the results. For \(r = 2\), the 2-Sylow subgroups are no longer cyclic, necessitating more complex arguments and the use of Brauer graphs. \par The author then discusses how to use the \(\mathrm{PSL}_2(q)\) results to compute the extensions between irreducible modules over \(\mathrm{PGL}_2(q)\). In most cases, this is straightforward considering \(\mathrm{PSL}_2(q)\) as a normal subgroup of \(\mathrm{PGL}_2(q)\). The most challenging case is computing \(H^n(\mathrm{PGL}_2(q),k)\) when \(r = 2\), which requires embedding \(\mathrm{PGL}_2(q)\) as a maximal subgroup in \(\mathrm{PSL}_2(q^2)\). Lastly, the author considers the case \(G := \mathrm{SL}_2(q)\) (with \(q\) odd). Again, the \(r = 2\) case is more difficult requiring the use of (known) information on the structure of PIMs in the principal block.