1
$\begingroup$

Let given torsion free abelian group $A$ of finite rank. Let for prime number $p$, given that $\cap_i p^iA =\{0\}$. Is it true that for any $p$- torsion abelian group $B$, $\text{Hom}_{\mathbb{Z}}(A, B)$ is torsion $\mathbb{Z}$ module.

$\endgroup$
3
  • 1
    $\begingroup$ Would you define what you mean by "rank"? There are many possible non-equivalent meanings. $\endgroup$
    – YCor
    Nov 18, 2015 at 17:19
  • $\begingroup$ en.m.wikipedia.org/wiki/Rank_of_an_abelian_group $\endgroup$
    – solver6
    Nov 18, 2015 at 18:12
  • $\begingroup$ OK thanks. To avoid ambiguity it's easy to just say $\mathbb{Q}$-rank or torsion-free rank. (Well there's indeed less ambiguity as soon as the group is assumed torsion-free.) $\endgroup$
    – YCor
    Nov 18, 2015 at 18:18

2 Answers 2

3
$\begingroup$

Let $p$ be any prime. There exists a subgroup $A$ of $\mathbf{Z}[1/p]^2$ containing $\mathbf{Z}^2$ such that $\bigcap_n p^nA=\{0\}$ and $A/\mathbf{Z}^2$ is infinite (isomorphic to the quasi-cyclic group $P_p=\mathbf{Z}[1/p]/\mathbf{Z}$). Then $\mathrm{Hom}(A,P_p)$ contains $\mathrm{Hom}(P_p,P_p)\simeq\mathbf{Z}_p$ (the $p$-adics), so is not torsion.

$\endgroup$
0
$\begingroup$

Yes, i think it is true. Take $A\otimes_{\mathbb{Z}}\mathbf{Z}_p$ (tensor product with $p$-adic ring). If $e_1, ..., e_n\in A$ is basis of $A/pA$, then $A\otimes_{\mathbb{Z}}\mathbf{Z}_p = e_1\otimes_{\mathbb{Z}}\mathbf{Z}_p +... + e_n\otimes_{\mathbb{Z}}\mathbf{Z}_p$. Note that $\text{Hom}(A, B)\hookrightarrow\text{Hom}_{\mathbf{Z}_p}(A\otimes_{\mathbb{Z}}\mathbf{Z}_p, B\otimes_{\mathbb{Z}}\mathbf{Z}_p)$. For any $\phi\in\text{Hom}_{\mathbf{Z}_p}(A\otimes_{\mathbb{Z}}\mathbf{Z}_p, B\otimes_{\mathbb{Z}}\mathbf{Z}_p)$ and some $k$, $k\phi(e_i) = 0$, so $k\phi = 0$. So $\text{Hom}_{\mathbf{Z}_p}(A\otimes_{\mathbb{Z}}\mathbf{Z}_p, B\otimes_{\mathbb{Z}}\mathbf{Z}_p)$-torsion and so $\text{Hom}(A, B)$ - torsion. $\Box$

$\endgroup$
4
  • $\begingroup$ The problem in this argument is that you lose the condition $\bigcap p^nA=0$ when you tensor with $\mathbf{Z}_p$ (this is not very visible since you don't explicitly use it... anyway if $A$ is in my example, $A\otimes\mathbf{Z}_p\simeq\mathbf{Q}_p^2$ and $\mathrm{Hom}(\mathbf{Q}_p^2,\mathbf{Q}_p/\mathbf{Z}_p)$ is not torsion. Or you seem to assume that $A\otimes\mathbf{Z}_p$ is finitely generated as $\mathbf{Z}_p$-module, which is not the case. $\endgroup$
    – YCor
    Nov 18, 2015 at 23:25
  • $\begingroup$ But $A\otimes\mathbf{Z}_p = e_1\otimes\mathbf{Z}_p+...+e_n\otimes\mathbf{Z}_p$ can't be isomorfic to $\mathbf{Q}_p^2$? $\endgroup$
    – Ring king
    Nov 19, 2015 at 6:29
  • $\begingroup$ I used that $A\otimes\mathbf{Z}_p\cong\varprojlim_i p^iA = e_1\otimes\mathbf{Z}_p+...+e_n\otimes\mathbf{Z}_p$ $\endgroup$
    – Ring king
    Nov 19, 2015 at 6:39
  • $\begingroup$ Your claim that lifts of generators mod $p$ generate at the $p$-adic level is false. $\endgroup$
    – YCor
    Nov 19, 2015 at 12:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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