On finite \(p\)-groups whose automorphisms are all central. (Q1760355)

This is an interesting paper, for it solves two natural problems that were around for some years. If \(G\) is a finite group, if \(I\) denotes its group of inner automorphisms, and if \(A\) denotes the full automorphism group \(\Aut(G)\), the centralizer in \(A\) of \(I\) is called the ``central automorphism group''. Let's call this centralizer, right here, for reasons of simplicity, \(C\). Two questions arise, namely: 1) Is it true that if \(A\) is Abelian, then \(G\) (must be then nilpotent) must have its Sylow \(p\)-subgroups special? The authors provide a family of counter-examples (all of them \(p\)-groups) and they give a \textit{negative} answer. 2) Can it be that \(A=C\) and \(A\) itself is \textit{nonabelian}? Again, the authors give a family of examples showing that this is possible. The examples, of course, are \(p\)-groups. The proofs are intricate, using (as one might expect) a lot of commutator calculations and, moreover, helped by an intensive use of GAP. The paper also contains a very large reference list on the mentioned problems, a list which, no doubt, will help those interested a lot in tracking back the history. To this reviewer's mind, this paper proves that almost \textit{nothing} suggested by ``intuition'' really ``works'' when it comes to finite nonabelian \(p\)-groups.