In Steve Awodey's book Category Theory, he has some proof that seems wrong to me on page $60$, he mentioned a proposition that

Proposition $3.11.$In the category $Ab$ of abelian groups, there is a canonical isomorphism between the binary coproduct and product,

$$A+B \cong A \times B.$$

And his proof goes as follows,

Proof. To define an arrow $\theta:A+B \to A \times B,$ we need one $A \to A \times B$ (and one $B \to A \times B$), so we need arrows $A \to A$ and $A \to B$ (and $B \to A$ and $B \to B$). For these, we take $1_A : A \to A$ and the zero homomorphism $0_B : A \to B$ (and $0_A:B \to A$ and $1_B : B \to B$). Thus, all together, we get $$\theta = [\langle1_A,0_B\rangle,\langle0_A,1_B\rangle]:A+B \to A \times B.$$ Then given any $(a,b) \in A+B,$ we have $$\theta(a,b) = [\langle1_A,0_B\rangle,\langle0_A,1_B\rangle](a,b) \\ =\langle1_A,0_B\rangle(a)+\langle0_A,1_B\rangle (b) \\ = \langle1_A(a),0_B(a)\rangle+\langle0_A(b),1_B(b)\rangle \\ = (a,0_B) + (0_A,b) \\= (a+0_A,0_B+b)\\ = (a,b) $$

My question is that: where did he exactly use the assumption that the group $A$ and $B$ are abelian. He did not explicitly stated the fact, so I am a little bit suspicious about the proof.

And another thing, he leave the identity $[f,g](a,b) = f(a)+_X g(b)$ (page $60$) as an exercise. I did not get the way to prove it. If anyone has the proof, please share in the post. Thank you.

  • $\begingroup$ @RghtHndSd How so? $\endgroup$ – Zack Ni Feb 8 '19 at 2:53
  • 1
    $\begingroup$ I really don't like this proof either. When he says $(a,b) \in A + B$ he comes awfully close to just assuming that the coproduct is the product. $\endgroup$ – Chessanator Feb 8 '19 at 2:53
  • $\begingroup$ @Chessanator Yeah I have to agree on it. But is there any better proof, though? $\endgroup$ – Zack Ni Feb 8 '19 at 2:54
  • $\begingroup$ @RghtHndSd There's a canonical morphism from coproduct to product in any category with a zero, but it's not necessarily an isomorphism. In particular, the coproduct of non-abelian groups is not the product. $\endgroup$ – Chessanator Feb 8 '19 at 3:06

As I mentioned in the comment, I really don't like Awodey's proof.

If I was proving this, I'd construct the inverse $\varphi: A \times B \rightarrow A + B$ to the morphism $\theta$ that Awodey gave by using the projections $\pi_A: A \times B \rightarrow A$, $\pi_B: A \times B \rightarrow B$ and the inclusions $\iota_A:A \rightarrow A+B$, $\iota_B:B \rightarrow A+B$.

I would define $\varphi = \iota_A \pi_A + \iota_B \pi_B$. (This is where abelian-ness is used. You can't define the addition of group homomorphisms like this unless the groups are commutative.)

You can check for yourself that this is an inverse to $\theta$.

Edited: I've also just realised that this helps with your second question about the exercise, about $[f,g](a,b)$. $(a,b)$ is an element of $A \times B$, but $[f,g]$ is a morphism out of $A + B$. Awodey has implicitly used the fact that they are isomorphic again.

In other words, there's an implied instance of $\varphi$ in the expression: it should be $[f,g] \varphi (a,b)$.

| cite | improve this answer | |

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.