2
$\begingroup$

Let $C$ be category, let $PSh(C) = [C^{op},$ *Set*$]$ be the category of presheaves on $C$, the Karoubi envelope of $C$, denoted $\overline{C}$, is defined as the full subcategory of $PSh(C)$ which are retracts of objects in $C$ under the Yoneda embedding.

I'm reading this paper on Karoubi envelopes ("Cauchy completion in category theory" by Borceux and Dejean), on page 3 (theorem 1) they say:

"(4) The category $PSh(\overline{C})$ of presheaves on $\overline{C}$ is equivalent to the category $PSh(C)$ of presheaves on $C$"

I'm interested to see how they proved this equivalence so I've been trying to follow the whole proof to get to where they prove it but frankly I'm stuck and a lot of what they're saying is going over my head, like for example this:

"If every idempotent of $C$ splits, every retract of a representable functor $C(-,c)$ induces an idempotent on $C(-,c)$, thus an idempotent $e$ on $C$, $e$ splits in $C$, which produces a retraction of $c$ and thus a retraction of $C(-,c)$, which..."

That's where I first got lost, every retract of a representable functor induces an idempotent on itself thus an idempotent $e$ on $C$ that splits?? Don't follow at all, I tried to skip to the next paragraph (where they prove the equivalence), it says:

"To prove (4), it suffices to show that every presheaf $F$ on $C$ can be uniquely (up to an isomorphism) extended in a presheaf $\overline{F}$ on $\overline{C}$."

("extended IN a presheaf $\overline{F}$"? I guess they mean "extended TO a presheaf $\overline{F}$" right?), then they write:

"From the uniqueness of the splitting in $\overline{C}$ of an idempotent $e \in C$ and the Cauchy completeness of the category of sets (Proposition 1), $\overline{F}$ has to map the splitting of $e$ on the splitting of $F e$."

(Don't they mean $\overline{F}$ has to map the splitting of $e$ TO the splitting of $F(e)$? What do they mean by ON?)

The proof is longer but I just about stopped there, couldn't follow the rest of the proof, can anyone provide some clarification on this?

Please be nice to me :)

$\endgroup$
3
  • 4
    $\begingroup$ You shouldn't have skipped ahead. The bit about idempotents is very important! The key point is that the Yoneda embedding is fully faithful, so any idempotent on $\mathcal{C}(-, c)$ comes from a unique idempotent on $C$, and that the splitting of an idempotent is preserved by any functor whatsoever. $\endgroup$
    – Zhen Lin
    Nov 14, 2013 at 0:47
  • 2
    $\begingroup$ Try reading this paper projecteuclid.org/DPubS/Repository/1.0/…, which is one of the original references on the subject. The terminology is a bit outdated but the paper is very readable. $\endgroup$ Nov 14, 2013 at 16:22
  • $\begingroup$ what paper is it ? link seems broken $\endgroup$
    – nicolas
    Feb 24, 2023 at 14:23

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.