If $F: 𝒞 ⟶ 𝒟$ and $K: 𝒞 ⟶ ℰ$ are functors, where $𝒞$ is small and both $𝒟$ and $ℰ$ are cocomplete. How do the left Kan extension of $F$ along $K$ (${\rm Lan}_K(F)$) and the left Kan extension of $K$ along $F$ (${\rm Lan}_F(K)$) relate with one another? (Is it safe to say that they are adjoints for example?)

EDIT: Actually, I asked the question in general but what I had in mind was the special case where $K := y: 𝒞 ⟶ \widehat{𝒞}$ is the Yoneda embedding, so any tip to show that ${\rm Lan}_y(F) ⊣ {\rm Lan}_F(y)$ (or that ${\rm Lan}_F(y) \cong N_F := Hom(F(-), -)$) would be helpful as well.

Any insight on this would be very much appreciated!


If $y : A \to[A°,Set]$ is the Yoneda embedding it is a general fact that, given a functor $f : A \to B$, there is the adjunction $$ Lan_yf \dashv Lan_fy $$ this is called nerve-realization paradigm.

As you can see e.g. here, asking that the two functors $F,K$ are related in this way is a rather strong request (in fact, the tentative proof in the answer contains a flaw, and to this day no convincing example, different than Yoneda extensions, has been given)

  • $\begingroup$ Thank you! That was the special case I had in mind when I wrote the question actually (I have edited it): do you know any nice proof of $Lan_yf \dashv Lan_fy$ (avoiding if possible too much co/end fu (as in the answer you pointed out))? $\endgroup$ – Cooke4 Jul 30 '18 at 9:40
  • $\begingroup$ Or even better: a proof of ${\rm Lan}_F(y) \cong N_F := Hom(F(-), -)$ $\endgroup$ – Cooke4 Jul 30 '18 at 9:48
  • 1
    $\begingroup$ @Cooke4 The latter is easy: the category of elements of $\operatorname{Hom}(F(-),D)$ is precisely the comma category $(F\downarrow D)$ for any $D \in \mathcal D$. Hence the formula computing the pointwise left Kan extensions together with the Yoneda lemma (or mor precisely the density of the Yoneda embedding, which is a consequence) gives you the result. $\endgroup$ – Pece Jul 30 '18 at 10:20
  • 2
    $\begingroup$ You have to prove that $Nat(\hom(F,1),G)\cong Nat(y, GF)$; this can even be done directly! :-) $\endgroup$ – Fosco Loregian Jul 30 '18 at 15:10
  • $\begingroup$ @FoscoLoregian The answer of Pece leads to a pretty straightforward result, but I can't figure out how to show that $Nat(\hom(F,1),G)\cong Nat(y, GF)$, since it is supposed to be natural in $G$ if I'm not wrong: any hint? :-) $\endgroup$ – Cooke4 Aug 3 '18 at 11:04

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.