I would like to improve my proof of the following result:

If $H$ is a finite, elementary abelian $p$-group, then $\Phi(H) = 1$.

Here, $\Phi(H)$ is the Frattini subgroup, defined as the intersection of all maximal subgroups of $H$. An elementary abelian $p$-group is an abelian group with the property that $x^p = 1$ for all $x \in H$.

I proved this by choosing an arbitrary nonidentity element $x \in H$ and showing that there exists a maximal subgroup $M < H$ which does not contain $x$.

My proof explicitly constructs $M$. I wonder if there is a cleaner, more elegant way to establish the existence of $M$ without resorting to a construction.

My proof: Let $x \in H$ be a nonidentity element. Define $N_1 = \langle x \rangle$. Then $|N_1| = p$. If $N_1 < H$, choose $x_2 \in H \setminus N_1$, and define $N_2 = \langle x_2\rangle$. Then $N_1 N_2$ is a subgroup, with $N_1 \cap N_2 = 1$, so $N_1 N_2$ is a direct product with order $p^2$.

If $N_1 N_2 < H$, we continue in the same way. By step $k$ we start with a direct product $N_1 N_2 \cdots N_{k-1}$ with order $p^{k-1}$. If this subgroup is proper, choose $x_k \in H \setminus N_1 N_2 \cdots N_{k-1}$ and define $N_k = \langle x_k \rangle$. Then $N_1 N_2 \cdots N_k$ is a direct product with order $p^k$.

Since $H$ is finite, this process eventually terminates and we end up with $H = N_1 N_2 \cdots N_m$, where $N_1 = \langle x \rangle$. Then $M = N_2 \cdots N_m$ is a maximal subgroup of $H$ which does not contain $x$:

  • Maximality is clear because the index of $N_2\cdots N_m$ is $p$.
  • If $x \in N_2 \cdots N_m$ then we can write $x$ in two different ways in $N_1 N_2 \cdots N_m$, namely $(x)(1)\cdots(1) = (1)(n_2)\cdots(n_m)$. The uniqueness of representation in a direct product then forces $x = 1$.

Context: This question addresses a remark made by Isaacs in Finite Group Theory, following problem 1.D.8:

If $G$ is a finite $p$-group, then $\Phi(G)$ is the unique normal subgroup of $G$ minimal with the property that the factor group is elementary abelian.

I proved this by establishing the following results:

  1. If $H$ is a finite, elementary abelian $p$-group, then $\Phi(H) = 1$.
  2. If $\theta : G \to H$ is a surjective homomorphism, then $\theta(\Phi(G)) \leq \Phi(H)$.
  3. Therefore, if $K \lhd G$ and $G/K$ is elementary abelian, then $\Phi(G) \leq K$.

Let $H$ be an elementary abelian $p$ group. Then we know that $H\cong Z_p\times Z_p...\times Z_p$.

Set $$M_1=1\times Z_p\times ... Z_p$$ $$M_2=Z_p\times 1\times ..\times Z_p$$ $$M_n=Z_p\times ...\ Z_p\times 1$$

Notice that $M_i$ are maximal in $G$ and $$1=\cap_{i=1}^n M_i\geq \Phi(H)$$.

Note: Notice that $A\times B\cap C\times D=(A\cap B)\times (C\cap D)$.

  • $\begingroup$ Thanks, this is a much nicer way! $\endgroup$ – Bungo Dec 15 '15 at 18:23
  • $\begingroup$ you are welcome. $\endgroup$ – mesel Dec 15 '15 at 18:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.