5

From this post I see that $\mathbb{Z}^{\mathbb{N}}$ is not a free abelian group. Someone can explain me why?

Thanks a lot!

YCor
  • 17,878
EQJ
  • 4,369
  • 1
    Did you click the link in your link? – Alex Youcis Jul 25 '14 at 07:12
  • I don't understand the concept of dual then I don't understand why the problem in the link implies what I am asking. – EQJ Jul 25 '14 at 07:16
  • There is not other way? Or, What I need to understand this prove? – EQJ Jul 25 '14 at 07:21
  • @Alex: The link doesn't prove $\mathbb{Z}^\mathbb{N}$ is free. And despite the phrasing of the answer in that link, the link in the answer in that link doesn't prove $\mathbb{Z}^\mathbb{N}$ is free either. Although the question is about a fact that I imagine could be used in a (nontrivial) proof it's not free. –  Jul 25 '14 at 07:21
  • Related: https://math.stackexchange.com/questions/320444 – Watson Feb 20 '18 at 16:52

1 Answers1

4

The answer https://mathoverflow.net/a/10249 to Is it true that, as $\mathbb{Z}$-modules, the polynomial ring and the power series ring over integers are dual to each other? on MathOverflow shows that $$ \operatorname{Hom}(\mathbb{Z}^{\mathbb{N}},\mathbb{Z}) $$ is isomorphic to $\mathbb{Z}^{(\mathbb{N})}$.

If $\mathbb{Z}^{\mathbb{N}}$ were free it would be isomorphic to $\mathbb{Z}^{(X)}$ for some infinite set $X$, so we'd have $$ \mathbb{Z}^{(\mathbb{N})}\cong \operatorname{Hom}(\mathbb{Z}^{\mathbb{N}},\mathbb{Z})\cong \operatorname{Hom}(\mathbb{Z}^{(X)},\mathbb{Z})\cong \mathbb{Z}^X $$ which is false because the first group is countable, whereas the final one isn't.


There are different proofs of the result above. An abelian group $L$ is called slender if, for every homomorphism $f\colon \mathbb{Z}^{\mathbb{N}}\to L$, $f(e_n)=0$ for all but a finite number of $n\in\mathbb{N}$ (where $e_n$ is the function which is $1$ on $n$ and $0$ elsewhere).

A theorem by Nunke (Acta Sci. Math. Szeged, 23 (1962)) characterizes slender groups:

An abelian group $L$ is slender if and only if

  • $L$ is reduced and torsionfree and
  • $L$ has no subgroup isomorphic to $\mathbb{Z}^{\mathbb{N}}$ or to the group of $p$-adic integers for any prime $p$.

(I'm pretty sure you can find a proof of Nunke's theorem in Fuchs's Infinite Abelian Groups.)

In particular $\mathbb{Z}^{(X)}$ is slender for any countable set $X$. A property of slender groups is the following.

If $L$ is a slender group, $(G_n)$ is a countable family of torsionfree abelian groups and $f\colon \prod_n G_n\to L$ is a homomorphism, then

  • $f(G_n)=0$ for all but a finite number of $n\in\mathbb{N}$ (the canonical embedding of $G_n$ in the product is considered)
  • $f$ is zero if and only if it is zero on $\bigoplus_n G_n$ (considered as a subgroup of the product).

A consequence of this is that for every homomorphism $f\colon\mathbb{Z}^{\mathbb{N}}\to L$ a slender group, the image of $f$ is finitely generated.

So, since $\mathbb{Z}^{(\mathbb{N})}$ is slender, no homomorphism $\mathbb{Z}^{\mathbb{N}}\to\mathbb{Z}^{(\mathbb{N})}$ is surjective, which proves $\mathbb{Z}^{\mathbb{N}}$ is not free.

egreg
  • 238,574