Suppose $F:C\rightarrow D$ and that $F\dashv U$ is an adjunction and $C^{T}$ the Eilenberg–Moore category for the monad $T=U◦F$, with the corresponding functors $F^{T}:C\rightarrow C^{T}$ and $U^{T}:C^{T}\rightarrow C$.

I have been able to prove that there is a comparison functor $Φ : D →C^{T}$ which satisfies

(1) $U^{T}◦Φ= U$


(2) $Φ◦F = F^{T}$

I am having trouble with uniqueness.

Here is what I have so far: Suppose $Φ'$ satisfies (1) and (2).

Let $U\in D$. Then using (1) with $Φ'(D)=(C',\alpha )$, it follows that $U◦Φ'(D)=U(C',\alpha )=C'$ whereas $Φ(D)=(UD,U\varepsilon_{D})$ and so $U◦Φ(D)=U(UD,\varepsilon_{D})=UD$ which says $C'=UD$

Now I need to show that $\alpha=U\varepsilon_{D}$. This is where I'm stuck.

edit: Using the hint below, the fact that the adjunctions have the same unit imply, after using (1) and (2) that

\begin{matrix} \operatorname{Hom}(FC, D) & \xrightarrow{{\phi}} & \operatorname{Hom}(C, UD) \\ \left\downarrow\vphantom{\int}\right. & & \left\downarrow\vphantom{\int}\right.\\ \operatorname{Hom}(F^{T}C, Φ'(D))& \xrightarrow{\phi^{T}} & \operatorname{Hom}(C, U^{T}Φ'(D)) \end{matrix}


($\phi$ and $\phi^{T}$ are the isomorphisms giving the adjunctions; the left downward arrow is the map $f\longmapstoΦ'(f)$ and the right downward arrow is the identity on the Hom$(C, UD)$.)

Then, setting $C=UD$ and following $id_{UD}$, you get that $Φ'\epsilon=\epsilon^T Φ'$, which is the hint. The rest follows easily.

| cite | improve this question | | | | |
  • $\begingroup$ You can find a proof in Mac Lane's CWM. $\endgroup$ – Martin Brandenburg Jun 4 '14 at 8:34
  • $\begingroup$ Right. I know these results are well-known, but before looking up the worked-out proof, I want to try it for myself, with a hint or two. $\endgroup$ – Matematleta Jun 4 '14 at 13:24
  • $\begingroup$ what does it mean $U\in D$ ? $U$ is a functor and $D$ is a category $\endgroup$ – magma Jun 4 '14 at 14:25

Here's a version of the proof that bypasses the $\Phi\epsilon=\epsilon^T\Phi$ lemma and proves uniqueness directly.

  • $U^T\Phi=U$ tells us that $\Phi d$ is a $T$-algebra with structure map $\gamma d: TUd\to Ud$
  • $\Phi$ sends $D$-arrows to $T$-homomorphisms, so $\gamma$ is a natural transformation $TU\to U$
  • $\Phi F=F^T$ tells us that $\gamma F=\mu =U\epsilon F$

Since $\gamma$ is natural we have $U\epsilon\circ\mu U=U\epsilon\circ\gamma FU=\gamma\circ TU\epsilon$

Precomposing with $T\eta U$ gives us $U\epsilon\circ\mu U\circ T\eta U=\gamma\circ TU\epsilon\circ T\eta U$

This rearranges as $U\epsilon\circ(\mu\circ T\eta)U=\gamma\circ T(U\epsilon\circ\eta U)$, which simplifies to $U\epsilon=\gamma$

| cite | improve this answer | | | | |

Just a hint....

First you should realize that

$$Φd=(Ud, h)$$

that is a T-algebra with underlying $Ud$, for any d in $D$. in order to obtain the structure h, observe that the 2 adjunctions have the same unit $\eta$, and deduce that $Φ\epsilon=\epsilon^T Φ$.

Then deduce that $\epsilon^TΦd=\epsilon^T(Ud, h)=h$ and so $Φ\epsilon=\epsilon^T Φ$ implies $U\epsilon_d=h$ so h is determined and $Φ$ is unique.

| cite | improve this answer | | | | |
  • $\begingroup$ Once you pointed put that I needed to prove that $Φ\epsilon=\epsilon^T Φ$, I was able to do it. It was a diagram chase, following identities and using (1) and (2) above. This was very helpful thanks. Where I live I literally have no one to talk math with, thanks for taking the time. $\endgroup$ – Matematleta Jun 4 '14 at 23:41
  • $\begingroup$ @Chilango you are welcome, my pleasure :-) $\endgroup$ – magma Jun 5 '14 at 5:44

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.