Controlled wild algebras (Q2766418)

Let \(K\) be an algebraically closed field. Given a \(K\)-algebra \(B\), we denote by \(\text{mod }B\) the category of finitely generated right \(B\)-modules, and by \(\text{modf }B\) the full subcategory of \(\text{mod }B\) consisting of finite dimensional \(B\)-modules. If \(B\) is of finite \(K\)-dimension, we have \(\text{modf }B=\text{mod }B\).NEWLINENEWLINENEWLINEWe recall that a finite dimensional \(K\)-algebra \(R\) is `wild' if there is a \(K\)-linear exact functor \(F\colon\text{modf }K\langle t_1,t_2\rangle\to\text{mod }R\) which carries indecomposable modules to indecomposable ones, and non-isomorphic modules to non-isomorphic ones. If the functor \(F\) is fully faithful, \(R\) is called `strictly wild'. The author defines the algebra \(R\) to be `controlled wild', if there exists a \(K\)-linear exact faithful functor \(F\colon\text{modf }K\langle t_1,t_2\rangle\to\text{mod }R\) and a full subcategory \(\mathcal C\) of \(\text{mod }R\) which is closed under direct sums and direct summands such that, for each pair of modules \(X\) and \(Y\) in \(\text{modf }K\langle t_1,t_2\rangle\), there is a decomposition NEWLINE\[NEWLINE\Hom_R(FX,FY)=F(\Hom(X,Y))\oplus\Hom_R(FX,FY)_{\mathcal C},NEWLINE\]NEWLINE where the \(K\)-subspace NEWLINE\[NEWLINE\Hom_R(FX,FY)_{\mathcal C}=\{f\in\Hom_R(FX,FY)\mid f\text{ factors through an object of }{\mathcal C}\}NEWLINE\]NEWLINE of \(\Hom_R(FX,FY)\) is contained in \(\text{rad}_R(FX,FY)\). It is clear that every strictly wild algebra is controlled wild. It is shown in the paper that if \(R\) is controlled wild then \(R\) is wild and for any finite dimensional \(K\)-algebra \(B\) there is a module \(M\) in \(\text{mod }R\) such that \(B\cong\text{End }M/{\mathfrak A}\), where \(\mathfrak A\) is a nilpotent ideal of \(\text{End }M\).NEWLINENEWLINENEWLINEOne of the main results of the paper is the controlled wild covering criterion asserting that if \((Q',I')\to(Q,I)\) is a Galois covering of a finite bound quiver \((Q,I)\), with a torsion-free Galois group, and the bound quiver \((Q',I')\) is strictly wild, then the \(K\)-algebra \(KQ/I\) is controlled wild. As a consequence, it is shown that any local wild \(K\)-algebra is controlled wild. Further, if \(A\) is an arbitrary finite dimensional \(K\)-algebra, the characteristic of \(K\) is a prime number \(p\) and \(G\) is a non-trivial finite \(p\)-group such that the group algebra \(AG\) is wild then \(AG\) is controlled wild. A new proof of the controlled wild covering criterion, by applying the cleaving functors, is presented in [\textit{P. Dräxler}, J. Pure Appl. Algebra 169, No. 1, 33-42 (2002; see the preceding review Zbl 1019.16006)]. For a discussion of the controlled wildness and endo-wildness the reader is referred to [the reviewer, J. Pure Appl. Algebra 172, No. 2-3, 293-303 (2002; Zbl 1017.16019)].