A theorem of Jon F. Carlson on filtrations of modules. (Q856328)

Let \(G\) be a finite group and \(k\) be a field of characteristic \(p>0\). Carlson in 2000 proved that any \(kG\)-module is a direct summand of a module filtered by modules induced from elementary Abelian \(p\)-subgroups. After reducing to \(p\)-groups, the key step in proving the result is that, if \(G\) is a non-elementary Abelian \(p\)-group, then there is a sequence of maximal subgroups of \(G\) such that, up to projective modules, \(k\oplus\Omega^{1-n}(k)\) has a filtration \[ 0=L_0\subseteq L_1\subseteq\cdots\subseteq L_n=k\oplus\Omega^{1-n}(k), \] where \(L_i/L_{i-1}\cong\Omega^{1-i}(k)\!\!\uparrow^G_{H_i}\), for \(i=1,\dots,n\). The filtration comes from Serre's theorem on vanishing products in group cohomology and hypercohomology with coefficients in a chain complex. This paper gives an alternative proof of Carlson's theorem without hypercohomology calculation. For any \(\xi\in H^n(G,k)\), \(L_\xi\) is defined as the kernel, up to projective modules, of \(\widehat\xi\colon\Omega^n(k)\to k\). The argument is that, given \(\xi_1,\dots,\xi_n\in H^1(G,\mathbb{F}_p)\) with the Bockstein map \(\beta(\xi_1)\cdots\beta(\xi_n)=0\), \(L_{\beta(\xi_1)\cdots\beta(\xi_n)}=\Omega(k)\oplus\Omega^{2n}(k)\) has a filtration whose sections are isomorphic to Heller shifts of \(L_{\beta(\xi_i)}\)'s. Moreover, for any \(\xi\in H^1(G,\mathbb{F}_p)\) with kernel \(H\leq G\), there is a 2-step filtration for \(L_{\beta(\xi)}\) such that the sections are induced from \(H\). If \(\xi\) is represented by the extension \[ 0\to k\to M_{n-1}\to\cdots\to M_0\to k\to 0, \] the paper proves that \(L_\xi\) has a filtration \[ 0=L_0\subseteq L_1\subseteq\cdots\subseteq L_n=L_\xi,\quad\text{with}\quad L_i/L_{i-1}\cong\Omega^{n-i+1}(M_{i-1}). \] As corollary, if \(G\) is an elementary Abelian 2-group, and each \(M_i\) is a direct sum of modules induced from proper subgroups, then \(\xi\) is a non-zero scalar multiple of a product of one dimensional classes in \(H^1(G,\mathbb{F}_2)\). In particular, the extension is equivalent to an extension of a topological group acting on a sphere. For the varieties of modules, the paper proves that, for a finitely generated \(kG\)-module \(M\), \(V_G(M)=\bigcup_{H\leq G}\text{res}^*_{G,H}(V_H(M\!\!\downarrow^G_H))\) if and only if there is a finitely generated \(kG\)-module \(V\) such that \(M\oplus V\) is filtered by modules induced from subgroups of \(G\).