Hyperbolicity of semigroup algebras. (Q934041)

Let \(A\) be a finite dimensional (unital) \(\mathbb{Q}\)-algebra (not necessarily semisimple). Recall that a \(\mathbb{Z}\)-order \(\Gamma\) in \(A\) is a subring of \(A\) (with the same identity) that is finitely generated as a \(\mathbb{Z}\)-module and contains a \(\mathbb{Q}\)-basis of \(A\). In the paper under review, the authors say that \(A\) satisfies the hyperbolic property if the unit group \(\mathcal U(\Gamma)\) of a (and thus every) \(\mathbb{Z}\)-order \(\Gamma\) in \(A\) does not contain a free Abelian subgroup \(\mathbb{Z}^2\) of rank \(2\). This property is equivalent with the unit group of \(\mathbb{Z}\)-orders being hyperbolic groups in the sense of \textit{M. Gromov} [Essays in group theory, Publ., Math. Sci. Res. Inst. 8, 75-263 (1987; Zbl 0634.20015)]. \textit{S. O. Juriaans}, \textit{I. B. S. Passi} and \textit{D. Prasad} [Proc. Am. Math. Soc. 133, No. 2, 415-423 (2005; Zbl 1067.16056)] described when a rational group algebra of a finite group has the hyperbolic property (they also dealt with infinite groups). A natural question is to extend these results to the much wider class of (unital) rational semigroup algebras of finite semigroups. In order to do so the authors first prove the following structure theorem. By \(J(A)\) we denote the Jacobson radical of an algebra \(A\). A finite dimensional \(\mathbb{Q}\)-algebra \(A\) has the hyperbolic property if and only if one of the following holds: (1) \(J(A)=\{0\}\) and \(A=\bigoplus_i D_i \oplus B\), (2) \(\dim_\mathbb{Q} J(A)=1\) and \(A/J(A)=\bigoplus_iD_i\); where each \(D_i\) is either \(\mathbb{Q}\), a quadratic imaginary extension of \(\mathbb{Q}\) or is a totally definite quaternion algebra over the rationals and where either \(B\in\{\{0\},M_2(\mathbb{Q})\}\) or \(B\) has a \(\mathbb{Z}\)-order whose unit group is non-torsion and hyperbolic. In case \(A\) has the hyperbolic property with \(J(A)\neq\{0\}\) then either \(J(A)\) is central in \(A\) or \(A\) is a direct product of division algebras and \(T_2(\mathbb{Q})\), the \(2\times 2\)-upper triangular matrices over \(\mathbb{Q}\). As an application, the authors classify the finite semigroups \(S\) such that \(\mathbb{Q} S\) has the hyperbolic property, this provided that either \(\mathbb{Q} S\) is semi-simple or \(S\) is not a semi-simple semigroup. The general case remains unsolved. It turns out that in the former case \(\mathbb{Q} S\) satisfies the hyperbolic property if and only \(S\) has a principal series with all factors, except possibly one, isomorphic to a Higman group (i.e. an Abelian group of exponent dividing \(4\) or \(6\) or a Hamiltonian \(2\)-group), and the exceptional factor is either a cyclic group of order 5, 8, or 12, or a finite group \(G\) so that the unit group of the integral group ring has a non-Abelian free subgroup of finite index (there are \(4\) such groups, [see \textit{E. Jespers}, Proc. Am. Math. Soc. 122, No. 1, 59-66 (1994; Zbl 0833.16032)]), or a completely \(0\)-simple semigroup of the form \(\mathcal M^0(\{1\},2,2;P)\) with \(P\in\left\{\left(\begin{smallmatrix} 1&0\\ 0&1\end{smallmatrix}\right),\left(\begin{smallmatrix} 1&1\\ 0&1\end{smallmatrix}\right)\right\}\). If \(S\) is non-semisimple then \(\mathbb{Q} S\) is hyperbolic if and only if \(S\) has a principal series with one factor a null semigroup (with one non-zero element) and all other factors are Higman groups. The authors also classify all finite semigroups \(S\) so that \(\mathbb{Q}(\sqrt{-d})S\) is hyperbolic, where \(d\) is a non-zero integer different from \(1\). The description also is in terms of a principal series with special factors.