Let $p>q$ be primes and $G$ be a group of order $pq$. When $p\neq 1\pmod q$, $G$ must be cyclic by a couple applications of Sylow's theorems. However, when $p=1\pmod q$ is it always possible to construct a non-cyclic $G$?

So far I tried letting $P,Q$ be Sylow $p,q$-subgroups of $G$. Then both are cyclic, $P$ is normal and $n_q=p$.


It's always possible to construct such a noncyclic group when $p \equiv 1 \pmod{q}$. Note that this means that $q$ divides $p-1$, which is the order of the automorphism group $\operatorname{Aut}(C_p)$ of a cyclic group of order $p$, so by Cauchy's theorem, $\operatorname{Aut}(C_p)$ contains an element of order $q$. Sending the generator of $C_q$ to that element in $\operatorname{Aut}(C_p)$ gives us a non-trivial homomorphism $\varphi: C_q \to \operatorname{Aut}(C_p)$. Now taking the semidirect product $C_p \rtimes_\varphi C_q$ gives us a noncyclic group of order $pq$.

It's also possible to construct this group without semidirect products.
Consider the field with $p$ elements $\Bbb F_p$. It's well known that the multiplicative group $\Bbb F_p^\times$ is cyclic of order $p-1$. So if $q \mid (p-1)$, we can find an element $a$ in $\Bbb{F}_p^\times$ of order $q$ by Cauchy's theorem. Now consider the following subgroup of $\operatorname{GL}_2(\Bbb{F_p})$:

$$\left \{\begin{pmatrix}a^n & x \\ 0 & 1 \end{pmatrix}, n \in \Bbb{F_q}, x \in \Bbb{F}_p\right \}$$ (Note that $a^n$ is well-defined because $a$ has order $q$ in $\Bbb{F}_p^\times$). It's evident that this has order $pq$ (using that $a$ has multiplicative order $q$). Checking that this is in fact a non-abelian subgroup of $\operatorname{GL}_2(\Bbb{F_p})$ requires only easy matrix computations. This construction turns out to give exactly the same group as above, but this form of presentation (which is not meant as a technical term here) is probably more accessible to those who are not familiar with semidirect products.

  • $\begingroup$ Would it be possible to describe this more concretely, for example with permutations? $\endgroup$ – Akababa Feb 17 '18 at 2:13
  • 2
    $\begingroup$ @Akababa I hope matrices are fine, too $\endgroup$ – MatheinBoulomenos Feb 17 '18 at 2:17
  • $\begingroup$ By Cauchy's theorem????? You just need a primitive root of $p$ $\endgroup$ – Kenny Lau Jun 13 '18 at 8:40

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.