0

As the question says, I want to show that

If $Q$ is a non-trivial $\mathbb{Z}$-module, that is divisible, then $Q$ can not be a projective $\mathbb{Z}$-module.

Now, here is my reasoning:

Let $F$ be a free $\mathbb{Z}$-module. Then, for any basis $$A \subset F$$ of $F$ we have that $$F \cong \bigoplus_{a \in A} \mathbb{Z}.$$

Now, I think we can say that $$S = \bigcap_{n = 1}^{\infty} nF = 0$$ since $$S = \bigcap_{i = 1}^{\infty} \cong \bigcap_{n = 1}^{\infty} n \Big(\bigoplus_{a \in A}\mathbb{Z}\Big).$$

Now, it is clear this must be $0$, since, take any element $h$ supposedly contained in $S$;

Then $$h = (\ldots,z_{\alpha},\ldots)$$ with only finitely many nonzero-entries. Prime-factorize each $z_{\alpha}$. Since there are infinitely many primes, and the union of the primes in each entry will be a union of finite sets (just picking out the prime factors, disregarding the possible powers), hence finite, we can choose a prime $p$ such that $p$ is not in the prime-factorization of any entry of $h$, hence $$h \not \in p\bigoplus_{a \in A} \mathbb{Z} \implies h \not \in S \quad(\text{contradiction!}).$$

Now, I think there are two ways to show that if $Q$ was projective, we would get a contradiction. I'll just write what I think is the easier one; note that if $R$ (here $\mathbb{Z}$) is a P.I.D. then free is equivalent to projective, so that if $Q$ was a projective $\mathbb{Z}$-module, it would have to be free, hence we would get $$Q \cong \bigoplus_{b \in \beta} \mathbb{Z}$$ for some basis $\beta \subset Q$.

Now, since $Q$ is a divisible $\mathbb{Z}$-module, we have $$nQ = Q \quad(\forall n \in \mathbb{Z},n \neq 0)$$ it follows that $$\bigcap_{n = 1}^{\infty} nQ = Q$$ but $$\bigcap_{n = 1} nQ \cong \bigcap_{n = 1}^{\infty} n\Big(\bigoplus_{b \in \beta}\mathbb{Z}\Big) = 0 \quad(\text{contradiction!})$$

Is this proof correct?

Ben123
  • 1,847
  • 1
    The solution given at the duplicate is given in terms of a divisible module $Q$, not just the originally stated $\mathbb Q.$ – rschwieb Sep 29 '23 at 01:51
  • That is not the same question even? – Ben123 Sep 29 '23 at 07:39
  • That is, unless you can show that every divisible $\mathbb{Z}$-module is isomorphic to $\mathbb{Q}$, those are different questions. – Ben123 Sep 29 '23 at 09:03
  • Although, ofc very related. – Ben123 Sep 29 '23 at 09:04
  • 1
    To recapitulate, my comment means that the most upvoted solution there works for any divisible $\mathbb Z$ module, not just $\mathbb Q$. I have linked a second question which is essentially the same ("a nonzero module over a domain cannot be both projective and injective") because for $\mathbb Z$ divisible=injective. – rschwieb Sep 29 '23 at 11:32
  • I am interested in whether this particular way to do it works, not an answer in general.

    If this is of no interest to the community, I accept that. But that was the reason I asked it.

    Although, I also am somewhat interested in a general solution. The way the one you linked did the problem is not how D & F recommends one to do it, not at a first glance, since there is nothing about looking at the infinite interseciton of a free module $F$, multiplied by the index $n$, i.e. $$\bigcap_{n = 1}^{\infty} nF$$ there, but that is the hint given in D & F for the problem.

    – Ben123 Sep 29 '23 at 11:43
  • 1
    If you want you reader to know you're reading D&F and that you have to follow an exclusive path without saying so, you can try psychic.math.stackexchange. But also it points the way of how to revise this question for the non-psychic site! Revise to explain where the problem comes from and that it has this specific hint, and that you're trying to make sure you've used this hint. You can also include the links I've attached saying "i've looked at these but they do not follow the lines of the strategy given." If so, I'd be willing to reopen (provided I don't find a duplicate D&F post.) – rschwieb Sep 29 '23 at 13:45
  • 1
    Also, I'd like to draw your attention to this meta post, to explain why extra care needs to be taken when trying to write something like a solution verification question. The bar is set higher for such posts for reasons given there. i should have said, in the last comment, before I ran out of characters, that you also need to justify what exactly you're asking about (as opposed to "is everything correct?") – rschwieb Sep 29 '23 at 14:00

0 Answers0