5
$\begingroup$

Let $\mathcal{K}$ be a $2$-category. Keeping in mind the Cat-intuition on $\mathcal{K}$ I say that:

Def : A $1$-cell $A \stackrel{f}{\to} B$ is a left split subobject if $f$ is a right adjoint and the counit is invertible.

In Cat this condition is equivalent to be fully faithful (and a right adjoint) for $f$.

In Lemma 2.3 of his 2-categories companion Steve Lack points out that:

In a 2-category, when the counit is invertible then the right adjoint is representably fully faithful.

Thus one implication remains true. Is it possible to have sort of a converse for this statement? Even adding somme additional structure on $\mathcal{K}$.

I was hoping to have something like the following:

A $1$-cell is a left split subobject if and only if $f$ is a right adjoint and a monomorphism in $\mathcal{K}$,

which - I understand - is a vain expectation. I was hoping to use the left cancellation in the triangle equality somehow. Unfortunately it is not working as I expected.

$\endgroup$

1 Answer 1

5
$\begingroup$

This is true if by "monomorphism" you mean "representably fully faithful" as in the result of Lack: just argue representably. If $f:A\to B$ has a left adjoint $\ell:B\to A$ in $\mathcal K$, then $\mathcal K(X,f): \mathcal K(X,A) \to \mathcal K(X,B)$ has a left adjoint $\mathcal K(X,\ell)$ in $\mathrm{Cat}$ for any $X\in \mathcal K$. If $f$ is representably fully faithful, then by definition $\mathcal K(X,f)$ is fully faithful. Thus, the counit of the adjunction $\mathcal K(X,\ell)\dashv \mathcal K(X,f)$ is an isomorphism in $\mathrm{Cat}$. This counit has components that are 2-cells $\ell f g \to g$ in $\mathcal K$ for $g:X\to A$, obtained by whiskering with the counit $\epsilon : \ell f \to 1_A$. So just take $X=A$ and $g = 1_A$ to conclude that $\epsilon$ is an isomorphism.

$\endgroup$
0

Your Answer

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

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