Equivalent functors carrying representability

195 Views Asked by At

Emily Riehl's "Category Theory in Context, ${\rm Exercise}~2.1.{\rm iii}.$

Suppose $F:{\rm C}\to{\rm Set}$ is equivalent to $G:{\rm D}\to{\rm Set}$ in the sense that there is an equivalence of categories $H:{\rm C}\to{\rm D}$ so that $GH$ and $F$ are naturally isomorphic.

$~~~~~~(i)$ If $G$ is representable, then $F$ is representable?
$~~~~~(ii)$ If $F$ is representable, then $G$ is representable?

Proof of $(i)$:
Consider the following diagram $$\require{AMScd}\begin{CD} Fc @>{\alpha_c}>> GHc @>{\mu_c}>>{\rm Hom}(Hc^*,Hc) @>{\eta_c}>>{\rm Hom}(c^*,c)\\ @V{Ff}VV @V{GHf}VV @V{Hf\circ-}VV @V{f\circ-}VV\\ Fc' @>{\alpha_{c'}}>> GHc' @>{\mu_{c'}}>>{\rm Hom}(Hc^*,Hc') @>{\eta_{c'}}>>{\rm Hom}(c^*,c') \end{CD}$$ The first square commutes by the naturality of $\alpha:F\Rightarrow GH$, the second square commutes as $G$ is representable and thus $G\cong{\rm Hom}(d^*,-)$ for some $d^*\in{\rm D}$ (decoded by the maps $\mu_c,~\mu_{c'}$). As $H$ is an equivalence of categories it is in particular essentially surjective on objects and thus $d^*$ is isomorphic to the image of an element $c^*\in{\rm C}$, that is $d^*\cong Hc^*$. Furthermore, the equivalence $H$ asserts that there is a bijection $\eta$ between the ${\rm Hom}$-sets of ${\rm C}$ and ${\rm D}$ for any objects $c,~c'$ (such that $f:c\to c'$) guaranteeing the commutivity of the last square.
Since every intermediate square commutes, we have a commutative rectangle from which it follows that $F\cong{\rm Hom}(c^*,-)$, so $F$ is representable.

Proof of $(ii):$
Consider the following diagram $$\require{AMScd}\begin{CD} GHc @>{\alpha'_c}>> Fc @>{\mu'_c}>>{\rm Hom}(c^*,c) @>{\eta'_c}>>{\rm Hom}(Hc^*,Hc)\\ @V{GHf}VV @V{Ff}VV @V{f\circ-}VV @V{Hf\circ-}VV\\ GHc' @>{\alpha_{c'}}>> Fc' @>{\mu'_{c'}}>>{\rm Hom}(c^*,c') @>{\eta'_{c'}}>>{\rm Hom}(Hc^*,Hc') \end{CD}$$ In a similiar manner to $(i)$, the first square commutes by the naturality of $\alpha:GH\to F$ (using the opposite direction given due to the isomorphism), the second square commutes as $F$ is representable by some object $c^*\in{\rm C}$, and the last square commutes by the bijection of ${\rm Hom}$-sets induced by the equivalence $H$ (again, utilizing the opposite direction).
Since every intermediate square commutes, we have a commutative rectangle from which it follows that $G\cong{\rm Hom}(Hc^*,-)$, so $G$ is representable.$~~~\square$

Is my proof correct? If so, can it be improved; if not, where did I went wrong?

Thanks in advance!

1

There are 1 best solutions below

2
On BEST ANSWER

Your proof is essentially correct, you could improve (of course, whether or not it's an improvement depends on who you ask) it by making it into a series of reduction steps in the following way :

(a) If $F\cong K$ and $K$ is representable, so is $F$. Proof : trivial, as natural isomorphism is transitive.

(a) reduces the question to proving that $GH$ is representable. But also :

(b) If $G\cong K$ then $GH\cong KH$. Proof : easy, as if $\eta$ is a natural iso, so is $\eta H$.

(b) reduces the question to $\hom (d,-)\circ H$. At this point you have a smaller diagram to draw (the little square at the right of your diagram), so it's a bit easier.

And your (ii) can be improved even more (depending on your definition of equivalence - if you take it to mean "there is a quasi-inverse"). Indeed, if there is a quasi-inverse $L$ to $H$, then you can just apply your (i) to $G$ and $FL$, given that $FL \cong GHL \cong G$ (use (b) at some point)