Let $\mathcal{C}$ be a $k$-linear abelian category. Are the following two definitions of enough projectives equivalent?

(i) For every object $A$ there exists a projective object $P$ s.t. $P\twoheadrightarrow A$

(ii) Every simple object $U_i$ has a projective cover $p_i: P_i\twoheadrightarrow U_i$

The first one is taken from wikipedia, the second from the book Tensor Categories by Etingof et al.

I think this cannot be the same, for the superficial reason that nothing in (ii) makes a reference to arbitrary (generally indecomposable) objects of $\mathcal{C}$. However, I might be very wrong since I cannot seem to prove it.

Any pointers?

  • $\begingroup$ What's your definition of simple object? $\endgroup$ – Randall Apr 11 '18 at 15:31
  • $\begingroup$ One whose only subobjects are $0$ and itself. $\endgroup$ – Jo Mo Apr 11 '18 at 15:32

They do not, in general. I bet Etingof is assuming we're in a semisimple category, as is usually the case, if not always, in that book. If every object is a sum of simples, then it should be clear that these conditions are equivalent, since a sum of projectives is projective and a sum of epis is epi. The conditions are not equivalent in general, since abelian categories need not have any simple objects. See Jeremy Rickard's example here.

| cite | improve this answer | |
  • $\begingroup$ To be fair, I lied a bit (by omission). We are actually assuming local finiteness (Hom-finite and every object has finite length) and finitely many iso classes of simple objects. As you might have noticed, this is precisely the definition of a finite abelian cat over $k$, which is generally not semisimple. $\endgroup$ – Jo Mo Apr 11 '18 at 15:50

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.