3
$\begingroup$

This question is inspired by Characterization of functors whose right adjoint is monadic?.

Let $F : \mathbf C \rightleftarrows \mathbf D : U$ be an adjunction, and suppose that we want to establish when the canonical comparison functor $\mathbf{Kl}(UF) \to \mathbf D$ is an equivalence. A necessary and sufficient condition is that $F$ be essentially surjective on objects.

Is it possible to characterise this condition in terms only of the right adjoint $U : \mathbf D \to \mathbf C$?

$\endgroup$
0

1 Answer 1

3
$\begingroup$

Try this. I confess I haven't written out the proofs yet, but will do so if you (and the community) think this is an appropriate answer. Please excuse my renaming your categories according to my personal convention.

First, $U:{\mathcal A}\to{\mathcal S}$ must be faithful and reflect invertibility, cf Beck's theorem.

We rewrite essential surjectivity of $F:{\mathcal S}\to{\mathcal A}$ in terms of univeral properties:

For every "algebra" $A$, ie object of $\mathcal A$,

there are a "set" $X$ and a "function" $e:X\to U A$ in $\mathcal S$ that is universal in the sense that

for every other "algebra" $B$ and "function" $f:X\to U B$ in $\mathcal S$ there is a unique "homomorphism" $h:A\to B$ in $\mathcal A$ that extends $f$ in the sense that $f=e;U h$.

How do we find the object $X$? This is going to invoke $F$ (if that's allowed by the question!)

It need not be unique (up to iso), but the construction of the replacement for $\mathcal S$ is essentially the one in this paper of mine and the ones by Hayo Thielecke and Peter Selinger that it cites.

If appropriate equalisers exist in $\mathcal S$ then the canonical $e:X\to U A$ (in fact the terminal one) is given by the equaliser of $U A\rightrightarrows U F U A$.

In terms of $X$, the two maps are $\eta U F X=\eta U A$ and $U F\eta X$. We can derive the second of these from the same problem starting with $F U A$ in place of $A$, in which case we get a split equaliser.

$\endgroup$
4
  • $\begingroup$ Thanks, I shall digest this and get back to you. $\endgroup$
    – varkor
    Commented Oct 4, 2021 at 18:17
  • $\begingroup$ Could you spell out a little more why this property is the same as essential surjectivity? (In particular, I do not want to assume equalisers.) I'm happy to assume $F$ is bijective-on-objects or even identity-on-objects if that makes things simpler. $\endgroup$
    – varkor
    Commented Oct 25, 2021 at 14:40
  • 1
    $\begingroup$ I wrote out the definition of when $A$ is free on $X$, ie $A\cong F X$, and said that every $A$ is free on something. $\endgroup$ Commented Oct 25, 2021 at 14:44
  • $\begingroup$ Ah, of course! Thank you, I am satisfied this answers my question. Sorry for the delay. $\endgroup$
    – varkor
    Commented Oct 25, 2021 at 22:25

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .