I want to prove if the following assertion from Rotmans Advanced Algebra page 205 is true:

For any finite abelian group $G$, there is some integer $m$ with $G$ isomorphic to a subgroup of $U(\mathbb{Z}_{m})$, where $U(\mathbb{Z}_{m})$ are the units of Integers module m.

Is this sentence true or false? The book I'm studying says this is not true, but I cannot find a proper counterexample to understand the mentioned claim. Thanks

  • $\begingroup$ What book are you studying? $\endgroup$
    – Shaun
    Apr 24 '19 at 20:58
  • $\begingroup$ What is $U(\Bbb Z_m)$? $\endgroup$
    – Lukas
    Apr 24 '19 at 21:23
  • $\begingroup$ units in the ring of integers modulo $m$ ? $\endgroup$ Apr 24 '19 at 21:27
  • 2
    $\begingroup$ Already edited. I'm studying Rotmans Advanced Algebra and U(Zm) means units of Integers module m $\endgroup$
    – Cos
    Apr 24 '19 at 22:17
  • 1
    $\begingroup$ Probably follows from the Kronecker-Weber Theorem and this. $\endgroup$
    – Dzoooks
    Apr 24 '19 at 22:39

This is, in fact, a true statement.

There are two essential facts that I use here without proof.

  1. Any finite abelian group is a product of cyclic groups. Look up the structure theorem for modules over a PID for more on this.

  2. For any $ n \geq 2 $ there are infinitely many primes $p$ such that $ p \equiv 1 \text{ mod } n $. This is a special case of Dirchlet's theorem on primes in an arithmetic progression.

Now, take such a finite abelian group $ G $. We have $ G = \prod_{i=1}^n \mathbb Z/n_i \mathbb Z$, some integers $n_i$. Let $p_i$ be a prime such that $p_i \equiv 1 \text{ mod } n_i$. This exists by the second fact listed above. In fact, as there are infinitely many such primes we can take these $p_i$ to be distinct. Now, as $p_i \equiv 1 \text{ mod } n_i$, we have $ n | p_i - 1 $. Thus, we have $\mathbb Z / n_i \mathbb Z \subseteq \mathbb Z/(p_i - 1) \mathbb Z$. Strictly speaking, this means that there is an injective homomorphism between these groups, but identitying a group with its isomorphic image doesn't affect us in this case. As $p_i$ is prime, we know that $\mathbb Z / (p_i-1) \mathbb Z = (\mathbb Z/p_i \mathbb Z)^{\times}$. Hence, we have $\mathbb Z/ n_i \mathbb Z \subseteq (\mathbb Z/ p_i \mathbb Z)^{\times}$ Thus, we have $ G = \prod_{i=1}^n \mathbb Z/n_i \mathbb Z \subseteq \prod_{i=1}^n (\mathbb Z/p_i \mathbb Z)^{\times}$. One can show easily that the unit group of the product of rings is the product of their unit groups, so we have $\prod_{i=1}^n (\mathbb Z/p_i \mathbb Z)^{\times} = \left(\prod_{i=1}^n \mathbb Z/p_i \mathbb Z\right)^{\times}$. We took the $p_i$ to be distinct, hence they are pairwise relatively prime. By the Chinese Remainder Theorem, $\prod_{i=1}^n \mathbb Z/p_i \mathbb Z = \mathbb Z/ p_1 p_2 \dots p_n \mathbb Z$. Thus, $G \subseteq \left(\prod_{i=1}^n \mathbb Z/p_i \mathbb Z\right)^{\times} = (\mathbb Z/p_1 p_2 \dots p_n \mathbb Z)^{\times}$.

Field theory is listed as a tag for this, so I should mention that once you have that $\mathbb Q(\zeta_n)$ has Galois group $(\mathbb Z/n \mathbb Z)^{\times}$, this result proves the inverse Galois problem for finite abelian groups, i.e. all finite abelian groups are Galois groups over $\mathbb Q$.

  • $\begingroup$ And I'm wondering why Rotman would say otherwise. $\endgroup$ Apr 25 '19 at 1:18
  • 1
    $\begingroup$ You didn't explicitly answer the question "yes" or "no" (it is conventional to do so at the start of the answer). $\endgroup$ Apr 25 '19 at 3:01
  • $\begingroup$ @BillDubuque I've added this, thanks for catching that. $\endgroup$ Apr 25 '19 at 4:08
  • 2
    $\begingroup$ @QuangHoang In the copy of Rotman I have, the remark on page 205 says "We cannot conclude more from the proposition; given any finite abelian group G, there is some integer m with G isomorphic to a subgroup of U(Im)". It seems to me that Rotman is saying that the result is true, but it isn't a corollary of the result he just proved (that $\text{Gal}(\mathbb F(\zeta_n)/F)$ is a subgroup of $(\mathbb Z/m \mathbb Z)^{\times})$. The phrasing is kind of poor, so I see where the confusion arises. $\endgroup$ Apr 25 '19 at 4:17

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.