The tame infinitesimal groups of odd characteristic (Q2503364)

Let \(k\) be an algebraically closed field of characteristic \(p\geq 3.\) Let \( \mathcal{G}\) be an infinitesimal group over \(k\), and let \(H\left( \mathcal{G} \right) \) be its algebra of distributions (that is, the dual of its coordinate algebra). Then \(H\left( \mathcal{G}\right) \) decomposes into blocks -- the block belonging to the trivial \(H\left( \mathcal{G}\right) \) module is called the principal block and is denoted \(\mathcal{B} _{0}\left( \mathcal{G}\right) .\) This paper is concerned with the case where \(\mathcal{B}_{0}\left( \mathcal{G}\right) \) is tame, which is to say that for each \(d>0\) there exist a finite collection of \(\left( H\left( \mathcal{G} \right) ,k\left[ X\right] \right) \)-bimodules \(M_{i}\) which are free right \(k \left[ X\right] \)-modules such that all but finitely many isomorphism classes of \(d\)-dimensional indecomposable \(H\left( \mathcal{G}\right) \) -modules are isomorphism classes of the form \(M_{i}\otimes _{k\left[ X\right] }S\) for some simple \(k\left[ X\right] \)-module \(S.\) Given a group \(\mathcal{G},\) let \(\mathcal{G}_{1}\) denote the (first) Frobenius kernel of \(\mathcal{G}.\) The first result is that \(\mathcal{B} _{0}\left( \mathcal{G}\right) \) is tame if and only if \(\mathcal{B} _{0}\left( \mathcal{G}_{1}\right) \) is tame and \(\mathcal{G}/\mathcal{G}_{1}\) is multiplicative. Tameness is thus reduced to a question on Frobenius kernels, which can be studied via the equivalence of \(\mathcal{G}_{1}\) and Lie\(\left( \mathcal{G}\right) \) using previous work by the authors. Furthermore, if \(X\left( \mathcal{G}/\mathcal{G}_{1}\right) \) is the character group of \(\mathcal{G}/\mathcal{G}_{1}\) then \(\mathcal{B}_{0}\left( \mathcal{G}_{1}\right) \) is the set of elements of \(\mathcal{B}_{0}\left( \mathcal{G}\right) \) fixed by the canonical action of \(X\left( \mathcal{G}/ \mathcal{G}_{1}\right) \) on \(H\left( \mathcal{G}\right) .\) The other result consists of explicit computations on the Morita equivalence classes of groups with tame principal blocks. Suppose \(\mathcal{G\;}\) infinitesimal and of height \(r\), that is, \(r\) is the smallest number such that \(\mathcal{G}_{r}=\mathcal{G}\), where \(\mathcal{G}_{r}\) is the \(r^{\text{ th}}\) Frobenius kernel. Suppose further that \(\mathcal{G}\) has unipotent center of length \(n\). Letting \(\mathcal{Q}_{\left[ n,r\right] }=\left( \text{SL}\left( 2\right) _{1}^{n}\rtimes T_{r}\right) /\mathcal{M}\left( r,n\right) \), where \(T_{r}\) is the \(r^{\text{th}}\) Frobenius kernel of the maximal torus in \(\text{SL}\left( 2\right) \) and \(\mathcal{M}\left( r,n\right) \) is the multiplicative center, it is proved that if \(n\leq 1\) or \(r\geq 2\) then \(\mathcal{B}_{0}\left( \mathcal{G}\right) \) is tame if and only if \(\mathcal{G}\cong \mathcal{Q}_{\left[ r,n\right] }.\) If \(r=1\) and \( n\geq 2\) and \(\mathcal{B}_{0}\left( \mathcal{G}\right) \) is tame then there are infinitely many isomorphism classes of infinitesimal groups which give rise to a principal block of the same type as \(\mathcal{B}_{0}\left( \mathcal{G}\right) \). Finally, if \(\mathcal{B}_{0}\left( \mathcal{G}\right) \) then each block of \(H\left( \mathcal{G}\right) \) is either Morita equivalent to \(k\left[ X\right] /\left( X^{p^{n}}\right) \) or to the double Nakayama algebra \(\mathcal{N}^{2}\left( r,n\right) .\) In all, there are \( p^{r-1}+\left( p-1\right) /2\) blocks, \(p^{r-1}\) equivalent to \(k\left[ X \right] /\left( X^{p^{n}}\right) .\) It should be pointed out (and the authors do as this is article is very well written) that if \(\mathcal{G}\) does not have a unipotent center the situation is much more difficult. In general an infinitesimal group can have a tame principal block and other blocks which are wild.