I am trying to understand additive categories but I have trouble with basic properties.
First, want to show that in an additive category $A$ the product $M \times N$ is isomorphic to the direct sum $M \oplus N$. For construct a map $M \times N \to M \oplus N$, one can take $i_M \circ \pi_M + i_N \circ \pi_N$. For the other direction we can take the sum $(id_M,0)\oplus(0,id_N)$ gives us a map : $M \oplus N \to M \times N$. I don't see why these maps should be inverse.
Secondly, I saw that a functor $F : A \to A'$ between two additive categories is "$\mathbb Z$ linear on the morphisms" (i.e $F(f+g) = F(f) + F(g)$) if and only if the natural map $F(M) \oplus F(L) \to F(M \oplus L)$ is an isomorphism. I also don't see why.
Many thanks in advance !
$\renewcommand{\id}{\text{id}} $Given $\varphi \colon M \oplus N \to M \times N$ and $\psi \colon M \times N \to M \oplus N$, you get a map $\psi\circ\varphi \colon M \oplus N \to M \oplus N$. You can show that $$\psi \circ \phi \circ i_M = i_M = \id_{M \oplus N} \circ i_M \colon M \to M \oplus N$$ and $$\psi \circ \phi \circ i_N = i_N = \id_{M \oplus N} \circ i_N \colon N \to M \oplus N.$$ By the uniqueness property of the coproduct, this implies that $\psi \circ \varphi = \id_{M \oplus N}$. You can do a similar thing to show that $\varphi \circ \psi = \id_{M\times N}$.
For the second part, $F$ gives you two natural maps $F(M) \oplus F(L) \to F(M \oplus L)$ and $F(M \times L) \to F(M) \times F(L)$. By the argument above, you can use that $F$ is $\mathbb{Z}$-linear to show that these are inverses (when viewed through the isomorphism between product and coproduct). For the other direction, if $F \colon A \to A'$ preserves coproducts, then given $f,g \colon M \to N$ in $A$ you can consider the maps $$ F(M) \xrightarrow{F\begin{pmatrix}1\\1\end{pmatrix}} F(M \oplus M) \xrightarrow{F(f \; g)} F(N)$$ and $$ F(M) \xrightarrow{\begin{pmatrix}1\\1\end{pmatrix}} F(M) \oplus F(M) \xrightarrow{(F(f) \; F(g))} F(N)$$ and show that the first is $F(f+g)$, the second is $F(f)+F(g)$, and that they are equal.
(Note: I'm using a kind of matrix notation which is pretty handy; given maps $a_{ji} \colon M_i \to N_j$, I can write $(\begin{smallmatrix}a_{11}&a_{12}\\a_{21}&a_{22}\end{smallmatrix}) \colon M_1 \oplus M_2 \to N_1 \oplus N_2$, and composition of such morphisms corresponds to the usual matrix multiplication.)