There’s such thing as an Abelian category. Was wondering if you can formally start adding and subtracting parallel arrows in such a way that composition always distributes over +. If not using the usual definition of Abelian category then is it true for turning any category into a preadditive category?
In the same way you can take the group ring $\Bbb{Z}[G]$ for any group, I’d like to do the same to a category $C$ . In other words abelianization does not generalize usual group abelianization, but instead it generalizes taking the group ring wrt the ring of integers.
The free abelian category on a category $C$ is the composition of the following constructions:
First, one needs to turn $C$ into a preadditive category. This can be done by applying the free abelian group functor (which is a strong monoidal functor) to all of the hom-sets of $C$ to get an $\mathbf{AbGrp}$-enriched category $C'$.
Then, one needs to turn $C'$ into an additive category. As usual, the objects of the resulting additive category $C''$ will be the finite tuples of objects of $C'$, while a morphism from $(X_i)_{1 \le i \le m}$ to $(Y_j)_{1 \le j \le n}$ in $C''$ will be a tuple of morphisms $(f_{i,j}:X_i \to Y_j)_{1 \le i \le m, 1 \le j \le n}$, and composition is given by the usual multiplication of matrices.
Finally, one needs to adjoin kernels and cokernels to $C''$ to get the free abelian category $C'''$ on $C$. It turns out that it does not matter whether one adjoins kernels first and then cokernels, or cokernels first and then kernels. Also, for an additive category $D$ with kernels, it turns out that the abelian category formed by adjoining cokernels to $D$ is equivalent to the ex/lex completion of $D$.
The free cocompletion $E$ of an additive category $D$ under cokernels is constructed as follows (the free completion under kernels is likewise constructed dually):