In mathematics, specifically in category theory, a semi-abelian category is a pre-abelian category in which the induced morphism is a bimorphism, i.e., a monomorphism and an epimorphism, for every morphism .
The two properties used in the definition can be characterized by several equivalent conditions.[1]
Every semi-abelian category has a maximal exact structure.
If a semi-abelian category is not quasi-abelian, then the class of all kernel-cokernel pairs does not form an exact structure.
Every quasiabelian category is semiabelian. In particular, every abelian category is semi-abelian. Non-quasiabelian examples are the following.
The concept of a semiabelian category was developed in the 1960s. Raikov conjectured that the notion of a quasi-abelian category is equivalent to that of a semiabelian category. Around 2005 it turned out that the conjecture is false.[6]
By dividing the two conditions on the induced map in the definition, one can define left semi-abelian categories by requiring that is a monomorphism for each morphism . Accordingly, right semi-abelian categories are pre-abelian categories such that is an epimorphism for each morphism .[7]
If a category is left semi-abelian and right quasi-abelian, then it is already quasi-abelian. The same holds, if the category is right semi-abelian and left quasi-abelian.[8]