Polyhedral groups, McKay quivers, and the finite algebraic groups with tame principal blocks. (Q2509015): Difference between revisions

Let \(k\) be an algebraically closed field, \(\text{char\,}k>3\). Let \(\mathcal G\) be a finite algebraic \(k\)-group. We denote by \(\mathcal G^0\) its connected component, \(\mathcal G_{\text{red}}\) its reduced component, \(\mathcal G_{\text{lr}}\) its unique largest linearly reductive subgroup, and \(H(\mathcal G)\) its algebra of measures. The unique block \(\mathcal B_0(\mathcal G)\) on which the counit does not vanish is called the principal block of \(H(\mathcal G)\). In this paper, the author classifies, up to isomorphism, the finite algebraic \(k\)-groups \(\mathcal G\) such that \(\mathcal B_0(\mathcal G)\) is tame (i.e. \(\mathcal B_0(\mathcal G)\) has an infinite number of isomorphism classes of indecomposable modules and each dimension contains a finite number of discrete and one-parameter families of isoclasses). The study of groups with tame principal block leads to a collection of groups referred to as amalgamated polyhedral groups. For \(T\) the standard maximal torus in \(\text{SL}(2)\) we let \(T_{(m)}(R)\) be the diagonal matrices in \(\text{SL}(2)(R)\) whose entries are \(m\)-th roots of unity in \(R\), and let \(N_{(m)}=T_{(m)}(h^{-1}T_{(4)}h)\), where \(h=\left(\begin{smallmatrix}\zeta_4 &1\\ 1&\zeta_4\end{smallmatrix}\right)\in\text{GL}(2)(k)\). Furthermore, let \(\widehat T\), \(\widehat O\), and \(\widehat I\) denote the binary tetrahedral, octahedral and icosahedral groups, respectively. (The reviewer believes the matrix definition for \(\widehat I\) in the paper to be incorrect.) For each of the five groups \(\mathcal J\) above we define \(\mathcal{SJ}\) to be the product of the first Frobenius kernel of \(\text{SL}(2)\) with \(\mathcal{HJ}\), moreover \(P\mathcal{SJ}:=\mathcal{SJ}/\mathcal Z(\mathcal{SJ})\), where \(\mathcal{Z(SJ)}\) is the center of \(\mathcal{SJ}\). The groups of the form \(P\mathcal{SJ}\) are called amalgamated polyhedral groups. Given a finite algebraic \(k\)-group \(\mathcal G\), let \(\mathcal M\) denote the multiplicative center of \(\mathcal G^0\), and let \(\mathcal N\) denote the centralizer of \(\mathcal G^0/\mathcal M\) in \(\mathcal G\). In the case where the centralizer of \(\mathcal G^0\) in \(\mathcal G\) is trivial it is shown that \(\mathcal B_0(\mathcal G)\) is tame if and only if \(\mathcal G\) is an amalgamated polyhedral group. More generally, if \(\mathcal G\) has tame principal block then \(\mathcal G/\mathcal N\) is an amalgamated polyhedral group, \(\mathcal N^0\cong\mathcal M\times(\mathcal W_\ell)_1\), and \(\mathcal N_{\text{red}}\) is linearly reductive, where \(\mathcal W_\ell\) is the Witt vectors of length \(\ell\) and \((-)_1\) represents its first Frobenius kernel. It terms of quivers, let \(\mathcal G\) be a finite algebraic \(k\)-group such that \(\mathcal B_0(\mathcal G)\) is tame. It is shown that there exists a linearly reductive group \(\widehat{\mathcal G}\subset\text{SL}(2)\) such that the Gabriel quiver of \(\mathcal B_0(\mathcal G)\) is isomorphic to the McKay quiver \(\Upsilon_{L(1)}(\widehat{\mathcal G})\) relative to its two-dimensional, self-dual module. (This is the opposite quiver to the one McKay originally used.) Furthermore, the underlying graph is an extended Dynkin diagram of type \(\widetilde A\), \(\widetilde D\), or \(\widetilde E\). The Gabriel quiver is determined by \(\mathcal G(k)/\mathcal C_{\mathcal G}(k)\), \(\mathcal C_{\mathcal G}\) the kernel of the action of conjugation on \(\mathcal G^0/\mathcal M\) by \(\mathcal G_{\text{red}}\) and is explicitly given in a table.