# Number of generators of the automorphism group of an abelian group

Let $G$ be a finite abelian $p$-group. What is known about the minimal number of generators of a $p$-sylow of $Aut(G)$? is it bounded in terms of $d(G)$ the minimal number of generators of $G$ (and perhaps $p$)?

-
Are there examples showing that the minimal number of generators of $Aut(G)$ can be $\ge 5$? – Mark Sapir Aug 6 '13 at 3:16
I'm sure that $Aut(G)$ can have arbitrary rank (that is $Aut(G)$ may contain subgroups with arbitrary minimal number of generators). I will be pleased if you can explain why exactely 5? – Yassine Guerboussa Aug 6 '13 at 20:13
I did not notice that you are asking about the Sylow $p$ subgroup, not about the group $Aut(G)$ itself. My comment is about the whole $Aut(G)$. – Mark Sapir Aug 6 '13 at 20:28
Yes, I see. I find the question in your comment interesting. – Yassine Guerboussa Aug 6 '13 at 20:35

For the special case that $G=(\mathbb{Z}/p\mathbb{Z})^n$, this is true. Then we have $d(G)=n$ and $Aut (G)=GL(n,p)$. In the article of A. Patterson, "The minimal number of generators for $p$-subgroups of $GL(n, p)$" of $1974$ it is shown that any $p$-subgroup of $\text{GL}(n,p)$, where $p$ is an odd prime, can be generated by ${\textstyle\frac 1{4}}n^2$ elements. So the bound is $\frac{1}{4}d(G)^2$ in this case.
Without resorting to a reference, the exact number of minimal generators in this case clearly is $d(G)-1$. – Guntram Aug 6 '13 at 17:02
@Guntram: no, not any $p$-subgroup of $GL(n,p)=Aut(G)$ can be generated by exactly $n-1=d(G)-1$ generators. – Dietrich Burde Aug 6 '13 at 18:08
Thank you so much dear Professors. Now I think the bound on the rank $5/4 d(G)^2$ works for arbitrary abelian $p$-groups. – Yassine Guerboussa Aug 6 '13 at 20:21