# Maximal subgroups of a certain finite 2-group

The following came up in a problem on reconstruction of digraphs. I determined enough about the answer to satisfy the application completely, but still I am curious to know what the complete solution is. My group theory is weak, so apologies if this is too simple.

Let $T$ be a full binary tree with depth $k$. Call its levels $L_0,\ldots,L_k$. Here is the case $k=4$:

The number of leaves is $n=2^k$. Let $A$ be the full automorphism group of $T$ and let $f$ be its (faithful) action on the leaves of the tree, i.e. on $L_k$. Obviously $f(A)$ is an iterated wreath product of $\mathbb{Z}_2$ with itself and has order $2^{n-1}$. It is, indeed, the Sylow 2-subgroup of $S_n$.

The problem is: what are the subgroups of $f(A)$ of index 2?

Here is what I think the answer is. Let $X$ be a union of levels of $T$, including at least one level other than $L_0$. Let $P$ be the set of all $f(\gamma)$ such that $\gamma\in A$ and the action of $\gamma$ on $X$ is an even permutation. Then $P$ is a subgroup of the desired index. I'm guessing there are no others...

-

You can easily count the number of maximal subgroups of $W(k)$, the $k$-fold interated wreath product of $\mathbb{Z}_{2},$ by calculating the index of the Frattini subgroup. You can inductively prove that the number of generators is $k,$ which is clear for $k =1,2.$ To proceed, note that $W(k) = W(k-1) \wr \mathbb{Z}_{2}.$ Factor out the Frattini subgroup of the base group, and by induction, you are left with $E(k-1) \wr \mathbb{Z}_{2}$, where $E(k-1)$ is elementary Abelian of order $2^{k-1}.$ If $x$ is an element of order $2$ outside the new base group, then $[E(k-1) \times E(k-1),x]$ has order $2^{k-1},$ so that the largest elementary Abelian factor group of the original wreath product does have order $2^{k},$ as claimed. Hence the group $W(k)$ has $2^{k}-1$ maximal subgroups, since there is a bijection between maximal subgroups of $W(k)$ and maximal subgroups of $W(k)/\Phi(W(k)).$
That's great. Since the subgroups I identified are all distinct (ignoring whether $L_0\subseteq X$) and there are exactly $2^k-1$ of them, your proof establishes that my wild guess was correct (for once). Thanks! –  Brendan McKay Sep 2 '12 at 11:41