I think this is kind of true, since $\square\otimes C$ is a functor, so it preserves the isomorphism.

But what if we consider the example, $\mathbb{Z}\otimes\mathbb{Z}_2$ is not isomorphic to $2\mathbb{Z}\otimes\mathbb{Z}_2$. Because the latter is trivial while the first is not.

We can also consider the exact sequences: $$0\to\mathbb{Z}\to\mathbb{Z}\to\mathbb{Z}_2\to0,$$ where the first map is simply multiple by 2, and $$0\to2\mathbb{Z}\to\mathbb{Z}\to\mathbb{Z}_2\to0,$$ where the first map is the inclusion. If we tensor by $\mathbb{Z}_2$ we will get $$\mathbb{Z}_2\to\mathbb{Z}_2\to\mathbb{Z}_2\to0$$ and

$$(0\to)X\to\mathbb{Z}_2\to\mathbb{Z}_2\to0$$ respectively.

Apparently, the first one is the famous counterexample of non-left-exactness of tensor product. But the second one we just inject a submodule (ideal) to the whole module (ring). If $r\otimes m$ is 0 in $\mathbb{Z}\otimes\mathbb{Z}_2$, then apparently it is 0 in $2\mathbb{Z}\otimes\mathbb{Z}_2$. So we have the left exactness of the tensor product. This is again weird, since the two exact sequences are isomorphic (by wiki), why do I have different result?

  • 1
    $\begingroup$ @5xum but tensoring by $C$ is a functor in the category of $R$-modules, so it must preserve the isomorphism. $\endgroup$ – Upc Nov 22 '18 at 14:53
  • $\begingroup$ Are ${\Bbb Z}$ and ${\Bbb Z}_2$ considered as $\Bbb Z$ modules? $\endgroup$ – Wuestenfux Nov 22 '18 at 14:54
  • $\begingroup$ @Wuestenfux Yes. So, you mean $2\mathbb{Z}\otimes_\mathbb{Z}\mathbb{Z}_2$ is also $\mathbb{Z}_2$? $\endgroup$ – Upc Nov 22 '18 at 14:56
  • 3
    $\begingroup$ Why is the second trivial? Shouldn’t $2 \otimes 1$ be a nonzero element of $2 \mathbb{Z} \otimes \mathbb{Z}_2$? $\endgroup$ – user328442 Nov 22 '18 at 15:01
  • $\begingroup$ You've implicitly assumed that $(-) \otimes \mathbb{Z}_2$ preserves inclusions, but it doesn't; in general tensoring with a module preserves monomorphisms iff it's exact iff the module is flat. $\endgroup$ – Qiaochu Yuan Nov 22 '18 at 21:31

$2\mathbb{Z}\otimes\mathbb{Z}_2$ is not trivial. It is trivial inside of $\mathbb Z \otimes \mathbb{Z}_2$, a subtle but important difference. The "proof" you might have in mind that it is trivial would be

$$ 2n \otimes 0 = 0 \text{ and } 2n \otimes 1 = n \otimes 2 = 0 $$

but $n \otimes 2$ is only an element of $2\mathbb{Z}\otimes\mathbb{Z}_2$ if $n$ is even. Otherwise, you need the ambient module $\mathbb Z \otimes \mathbb{Z}_2$ to make sense of it.

Secondly, for your exact sequences. Do keep in mind that the isomorphism $2\mathbb{Z} \to \mathbb{Z}$ is a different map than the inclusion $2\mathbb{Z} \to \mathbb{Z}$.

So if we take the exact sequence

$$ 2\mathbb{Z} \to \mathbb{Z}\to\mathbb{Z}_2\to 0 $$

where the leftmost map is inclusion and tensor with $\mathbb{Z}_2$, we get the following commutative diagram $\require{AMScd}$ \begin{CD} 2\mathbb{Z} \otimes \mathbb{Z}_2 @>\text{inclusion}>> \mathbb{Z} \otimes \mathbb{Z}_2 @>>> \mathbb{Z}_2\otimes \mathbb{Z}_2 @>>> 0 \\ @VVV @VVV @VVV @VVV \\ \mathbb{Z}_2 @>\times 2>> \mathbb{Z}_2 @>>> \mathbb{Z}_2 @>>>0 \end{CD} where the rows are exact and the vertical arrows are isomorphisms.

That map $\mathbb{Z}_2 \xrightarrow{\times 2} \mathbb{Z}_2$ on the bottom is the zero map. To see that this must be the case, let us note that the isomorphisms $\to \mathbb{Z}_2$ are

$$ \begin{array}{ccc} 2\mathbb{Z} \otimes \mathbb{Z}_2 & \longrightarrow & \mathbb{Z}_2 \\ 2n \otimes 0 &\longmapsto & 0 \\ 2n \otimes 1 & \longmapsto & n \bmod 2 \end{array} \qquad \text{and} \qquad \begin{array}{ccc} \mathbb{Z} \otimes \mathbb{Z}_2 & \longrightarrow & \mathbb{Z}_2 \\ n \otimes 0 &\longmapsto & 0 \\ n \otimes 1 & \longmapsto & n \bmod 2 \end{array} $$

So the map on the bottom we can compute by looking at the composition $\mathbb{Z}_2 \to 2\mathbb{Z} \otimes \mathbb{Z}_2 \to \mathbb{Z} \otimes \mathbb{Z}_2 \to \mathbb{Z}_2$ and this composition takes $1 \mapsto 2 \otimes 1 \mapsto 2 \otimes 1 \mapsto 0$.

| cite | improve this answer | |

$2\mathbb{Z}$ and $\mathbb{Z}_2$ are not isomorphic as $\mathbb{Z}$-modules. Take a generator of each module and multiply it by 2.

| cite | improve this answer | |
  • $\begingroup$ Yes, this is clear. $\endgroup$ – Upc Nov 22 '18 at 17:16

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.