MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Is every balanced pre-abelian category abelian? That is, given an additive category $\mathcal{A}$ in which cokernels and kernels exists, such that every morphism, which is a mono- and an epimorphism, is an isomorphism; does it follow that $\mathcal{A}$ is abelian? Note that it would suffice to prove that the canonical morphism $coim(f) \to im(f)$, where $f$ is an arbitrary morphism, is a mono- and an epimorphism.

Note that the usual examples for non-abelian categories somehow suggest this (filtered modules, topological abelian groups). See also this related question.

After a google search I have found the following theorem (in "Basic homological algebra" by M. Scott Osborne, Cor. 7.18): If $\mathcal{A}$ is a balanced pre-abelian category with a separating class of projectives and a coseparating class of injectives, then $\mathcal{A}$ is abelian. Ok then this does not seem to be true in general. Does anybody know an example?

share|cite|improve this question
The reference: "Rump: Almost abelian categories" might interest you. It is proven in particular that if $\mathcal{A}$ is preabelian, then all canonical morphisms $coim(f) \to im(f)$ are mono and epi if and only if the category is semi-abelian (that is kernels and stable under composition and so are cokernels). – Yann Palu Feb 22 '11 at 15:43
up vote 7 down vote accepted

Let $B$ be the abelian category of 3-term sequences of vector spaces and linear maps $V^{(1)}\to V^{(2)}\to V^{(3)}$ (the composition can be nonzero). There are 6 indecomposable objects in this category; denote them by $E_1$, $E_2$, $E_3$, $E_{12}$, $E_{23}$, and $E_{123}$. Here $dim E^{(i)}_J=1$ for $i\in J$ and $0$ otherwise. Let $A\subset B$ be the full additive subcategory whose objects are the direct sums of all the indecomposables except $E_{12}$ and $f$ be the morphism $E_3\to E_{123}$.

Then one has $Coker_B f=E_{12}$, $Coker_A f=E_1$ and $Im_A f=E_{23}$, while $Coim_A f=E_3$ and therefore $Coker_A(Coim_A f\to Im_A f)=E_2\ne0$. It is easy to see that any morphism in $A$ has a kernel and cokernel. Besides, any morphism in $A$ with zero kernel and cokernel has also zero kernel and cokernel in $B$, hence is an isomorphism.

Further discussion can be found in my paper , Example A.5(7) (pages 59-60 in version 2).

share|cite|improve this answer
Is there also a more natural example which comes up in "daily mathematics"? – Martin Brandenburg Oct 11 '10 at 14:28
I don't know of such an example. Its nonexistence may be the reason why quite a few people seem to believe that any balanced pre-abelian category is abelian. This mistake can be found in Gelfand-Manin "Methods of homological algebra", II.5.11b. and probably in some other sources, too. – Leonid Positselski Oct 11 '10 at 15:46

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.