5
$\begingroup$

A torsionless abelian group $A$ is one for which any element $a\neq 0$ can be sent to a nonzero element of $Z$ by some homomorphism $A\rightarrow Z$ (integers). Equivalently, $A$ can be embedded as a subgroup of a Cartesian power of $Z$, $Z^I$.

A separable abelian group $A$ is one for which any element $a$ such that $\langle a \rangle$ is pure can be sent to $1\in Z$ by some homomorphism. Equivalently $A$ can be embedded as a pure subgroup of $Z^I$, equivalently any finitely generated pure subgroup is a direct summand.

There are torsionless abelian groups which are not separable, for example, take $A = Z^{(\omega)} + 2Z^\omega$. The element $\sum 2e_n$ generates a pure subgroup but any homomorphism to $Z$ must take it to a mulitple of $2$, since any such homomorphism has finite support in $\omega$, by a theorem of Specker.

I don't know of many other examples however, which leads to my problem:

Suppose $B$ is a countable subgroup of separable group $A$ so that $A/B$ is torsionless. Must $A/B$ be also separable?

$\endgroup$
1
  • 1
    $\begingroup$ The group $\mathbf{Q}$ is separable with the given definition. Probably it is assumed that a separable group is by definition torsionless. $\endgroup$
    – YCor
    May 29, 2017 at 19:57

2 Answers 2

2
$\begingroup$

The answer is no.

Consider your (renamed) example $V = \mathbf{Z}^{(\omega)} + 2\mathbf{Z}^\omega$. It is quotient of the direct sum $$A=\mathbf{Z}^{(\omega)} \oplus 2\mathbf{Z}^\omega\simeq \mathbf{Z}^{(\omega)} \oplus \mathbf{Z}^\omega$$ by a subgroup $B$ isomorphic to the intersection, namely the countable subgroup $2\mathbf{Z}^{(\omega)}$. As you say, $V\simeq A/B$ is torsionless and not separable, while $A$ is separable.

$\endgroup$
-1
$\begingroup$

All of these groups are torsion-free of integer type. A is separable in the sense of Fuchs, "Abelian Groups", Springer 2015, pp 501-508. B is free. A/B is quasi-separable in the sense of 39 Pacific J Math No. 3 (1971), pp 603-612. Is it possible that the question posed is undecidable in the sense of the Whitehead Problem, Fuchs, pp 519-524?

$\endgroup$

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.