How do we prove the following without using Cauchy's theorem? (From Pinter p156 problem H3.) Let $G$ be a finite abelian group, and the order $|G|$ be a multiple of a prime number, $p$.
Let $a$ be an element of $G$ and $\text{ord}(a)$ not a multiple of $p$. It can be shown that the order of the quotient group $|G/\langle a \rangle|$ is a multiple of $p$. Prove $G/\langle a\rangle$ must have an element of order $p$.

  • $\begingroup$ I think the best thing to do here would be to prove cauchy $\endgroup$ – Jorge Fernández Hidalgo Sep 22 '14 at 23:43
  • $\begingroup$ Yes, but I think the proof of this without Cauchy must reveal something interesting about the order of quotient groups compared to the order of their elements, which does hold for non-quotient groups. $\endgroup$ – MPitts Sep 22 '14 at 23:55
  • 1
    $\begingroup$ This implies Cauchy's theorem, just taking $a$ to be the identity of $G$, so there's no real way to get around it. $\endgroup$ – Kevin Carlson Sep 23 '14 at 0:16

Looking at page 156, I see that this question is in the context of "induction on $|G|$: an example." The series of four exercises is walking you through a proof of Cauchy's theorem for the special case of finite abelian groups, using induction.

For the benefit of others reading this question, I'll summarize the context here.

Claim: if $G$ is a finite abelian group, and $p$ is any prime factor of $|G|$, then $G$ has an element of order $p$. The base case, $|G| = 1$, is trivially true since there are no prime factors of $|G|$. Assume that $|G| = k$ and the result is proven for orders smaller than $k$. Take any nonidentity element $a \in G$. If $\text{ord}(a) = p$, then we're done. Otherwise:

  1. If $\text{ord}(a) = tp$ for some positive integer $t$, what element of $G$ has order $p$?

  2. Suppose $\text{ord}(a)$ is not a multiple of $p$. Then $G/\langle a \rangle$ is a group having fewer than $k$ elements. (Explain why.) The order of $G/\langle a \rangle$ is a multiple of $p$. (Explain why.)

  3. Why must $G / \langle a \rangle$ have an element of order $p$?

  4. Conclude that $G$ has an element of order $p$.

I assume you have answered 2 and now want to know why 3 is true. But this is simply because $|G / \langle a \rangle| < k$ and $p$ divides $|G / \langle a \rangle|$ (by problem 2), and we are assuming that we have proved the result for finite abelian groups of order is less than $k$ and divisible by $p$. (Induction hypothesis.)

By the way, problem 2 doesn't mention it, but for completeness you need to show that $G/\langle a \rangle$ is abelian. Fortunately, this is easy.

Problem 4 is actually the key part of the proof. Do you see why it's true? Hint: by 3, there is an element of $G/\langle a \rangle$ with order $p$. The elements of this group are cosets of $\langle a \rangle$, so let's call the element $g\langle a\rangle$. If $g\langle a\rangle$ has order $p$, then what does that tell you about the order of $g$?

  • $\begingroup$ Sorry if I'm being dense here, but there are two aspects of the proof and explanation that I don't see: (1) where did we prove that the proposition is true for k if it is true for values less than k ? And (2) problem 2 concerns the order of the class G/<a>, whereas problem 3 concerns the order of an element (in this case a coset) in that class. Obviously, these can be two very different orders. $\endgroup$ – MPitts Sep 23 '14 at 4:17
  • $\begingroup$ @user54301: (1) We are doing precisely that in problems 1-4. We showed (trivially) that it's true for $k=1$, then, during all four problems, we assume that it's true for all groups with order less than $k$. The goal of the problems is to prove that this implies it's true for $k$. Since $G/\langle a\rangle$ has order less than $k$ and divisible by $p$ (not necessarily equal to $p$), the proposition is true for that group: namely, that group contains an element of order $p$. $\endgroup$ – Bungo Sep 23 '14 at 4:35
  • $\begingroup$ Ok, I now see that problems 1-4 as a whole are the proof by induction. But can you help me see why problem (3) is true? $\endgroup$ – MPitts Sep 23 '14 at 21:41
  • $\begingroup$ In the explanation of (3), you mention induction on |G/<a>|, but doesnt the proof instead concern an induction on k = |G| ? $\endgroup$ – MPitts Sep 23 '14 at 21:55
  • $\begingroup$ @user54301: The induction hypothesis is that the proposition is true for any finite group which has order less than $k$ and divisible by $p$. $\endgroup$ – Bungo Sep 23 '14 at 21:57

As explained by Kevin Carlson, this implies Cauchy's Theorem anyway. So let's prove Cauchy's Theorem for finite abelian groups.

Write $G = \bigoplus_q G_q$ with $G_q = \{a \in G : \exists n \geq 0 ( q^n a = 0)\}$ for primes $q$. Then $|G_q|$ is a power of $q$, because any element $a \in G_q$ has order a power of $q$ and we can do induction by considering $G_q / \langle a \rangle$. Now if a prime $p$ divides $|G|$, it divides some $|G_q|$. By the result before, we must have $q=p$. This means $G_p \neq 1$, i.e. there is some element $a$ of order $p^n$, $n \geq 1$. Then $p^{n-1} a$ has order $p$.


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.