1
$\begingroup$

Suppose in some nice enough category, say abelian groups, we have a filtered colimit of compact objects, might as well say a colimit indexed by $\mathbb{N}$. If we are given an element $x$ in the colimit, must it have some representative at a finite stage? If so does this follow from the colimit being filtered, from the objects being compact, or both?

Thanks! Jon

$\endgroup$
1
$\begingroup$

Compactness and filtered are two different notions -though related. But the point is not the colimit being of compact objects.

The condition of being the colimit filtered already implies that any element $x \in \mathrm{colim}_i X_i$ has a representative in some $X_i$. No compact objects needed here.

The compactness condition for an object $K$ means that, for a filtered colimit, the universal morphism

$$ \mathrm{colim}_i \mathrm{Hom} (K, X_i) \longrightarrow \mathrm{Hom} (K, \mathrm{colim}_i X_i ) $$

is a bijection. In particular, any morphism $K \longrightarrow \mathrm{colim}_i X_i$ factors through some $X_i$.

$\endgroup$
  • $\begingroup$ Ah yes, thankyou! That second statement was one I had seen before, but could not remember the exact phrasing. $\endgroup$ – Jonathan Beardsley Oct 13 '11 at 18:06

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.