If $A$ is a cocomplete category and $C$ is small, we can say that for all functor $C \stackrel{l}{\to} A$ there exists the Kan Extension Lan$_{y_C}(l): \hat{C} \to A$. Moreover, the Kan extension is pointwise.

Is it known if the other implication is true? Namely:

Conj: If for all functor $C \stackrel{l}{\to} A$ there exists the Kan Extension Lan$_{y_C}(l): \hat{C} \to A$ then $A$ is cocomplete.


Then answer is yes if you assume the extensions are pointwise.

Let $\ast$ be the terminal presheaf on $\hat C$. Because the extension is poinwise, we have $Lan_{y_C}(l)(\ast) = \varinjlim_{c \in C} l(c)$. This can be seen, for example, by calculating the colimit using the category of elements of $\ast$, which is just $C$ itself.

  • $\begingroup$ Since I am not assuming that A has colimits, how can I say that the extension is pointwise? $\endgroup$ – Ivan Di Liberti Mar 31 '18 at 15:36
  • 1
    $\begingroup$ There are many ways to say that an extension is poinwise. One way is to say that the Kan extension is preserved by representable functors. But now I see I did not read carefully enough -- although you mention that the extension is pointwise in the cocomplete case, you didn't assume it in your question statement. $\endgroup$ – Tim Campion Mar 31 '18 at 15:39
  • $\begingroup$ This is because I didn't know how to say that there are pointwise without colimits. Thanks for your answer. $\endgroup$ – Ivan Di Liberti Mar 31 '18 at 15:40
  • 1
    $\begingroup$ Actually, it's natural to guess that the answer is no if you don't assume pointwiseness. This might be a candidate example statement for the point of pointwise Kan extensions... $\endgroup$ – Tim Campion Mar 31 '18 at 15:41

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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