Let $F : I \to C$ be a diagram in $C$, and $N$ an object of $C$. A cone from $N$ to $F$ is a family of morphism $P_X : N \to F(X)$ such that for every morphism $f : i1 \to i2$ in $I$, $F(f) \circ P_X = P_Y$. And the limit of $F$ is the universal cone.

If $C$ is now a $Cat$-enriched category and $W : I \to Cat$ is given as a weight, is there a notion of $W$-weighted cone such that a $W$-weighted limit is a universal $W$-weighted cone? If yes, how do you define such a $W$-weighted cone? And what does it mean for it to be universal?


Recall that a cone from $N$ to $F$ is a natural transformation from the constant functor $\Delta N$ to $F$. A cone is universal if it represents the functor $[I,C](\Delta -,F) : C^\text{op} \to \textbf{Set}$, i.e. if it induces isomorphisms $$C(X,N) \cong [I,C](\Delta X,F)$$ for all objects $X$ of $C$. Note that there is a natural isomorphism $$[I,C](\Delta X,F) \cong [I,\textbf{Set}](\Delta 1, C(X,F-)),$$ so we can see ordinary (universal) cones as $\textbf{Set}$-enriched $\Delta 1$-weighted (universal) cones.

Now in the 2-categorical context, let $\hat{F}$ denote the 2-functor $C^\text{op} \to [I,\textbf{Cat}]$ with $\hat{F}X = C(X,F-)$. A $W$-weighted cone from $N$ to $F$ is then a 2-natural transformation $W \to \hat{F}N$, and such a cone is universal if it represents the 2-functor $[I,\textbf{Cat}](W,\hat{F}-) : C^\text{op} \to \textbf{Cat}$, i.e. if it induces isomorphisms of categories $$C(X,N) \cong [I,\textbf{Cat}](W,C(X,F-))$$ for all objects $X$ of $C$. This defines the universal property of the $W$-weighted limit of $F$.

Re your comment: A 2-natural transformation $\theta : F\to G : \mathcal{A} \to \mathcal{B}$ is given by a family of 1-morphisms $\theta_A : FA \to GA$ of $\mathcal{B}$, indexed by the objects A of $\mathcal{A}$, such that the diagram of functors

$$ \require{AMScd} \begin{CD} \mathcal{A}(A,B) @>F>> \mathcal{B}(FA,FB)\\ @VGVV @VV\mathcal{B}(1,\theta_B)V\\ \mathcal{B}(GA,GB) @>>\mathcal{B}(\theta_A,1)> \mathcal{B}(FA,GB) \end{CD}$$

commutes for all objects $A,B$ of $\mathcal{A}$. This amounts to the usual naturality condition, plus an extra condition that for every 2-cell $\alpha : f \to g : A \to B$ of $\mathcal{A}$, we have $$\theta_B \circ F\alpha = G\alpha \circ \theta_A,$$ where $\circ$ denotes horizontal composition (''whiskering'' in this case) in $\mathcal{B}$.

Weighted limits and enriched natural transformations are elementary definitions of enriched category theory, which may be found in the standard reference on the subject, Kelly's book Basic concepts of enriched category theory.

  • $\begingroup$ I see indeed that the commutative diagram $F(f) \circ P_X = P_Y$ is the naturality of the cone. Your answer would be complete if you could spell out the commutative diagram coming from the 2-naturality of the weighted cone. $\endgroup$
    – Bob
    May 20 '16 at 6:57
  • $\begingroup$ @Bob, I have spelled out the definition of 2-natural transformation. $\endgroup$ May 20 '16 at 7:58
  • $\begingroup$ My point was that I expect as an answer an elementary definition of a weighted cone similar to the elementary definition of a cone in my question. Although, I could find the definition of a 2-natural transformation in Borceux's book, I still have a hard time unfolding everything to arrive at an elementary definition of a weighted cone. $\endgroup$
    – Bob
    May 20 '16 at 20:50
  • $\begingroup$ @Bob What is the difficulty? A $W$-weighted cone from $N$ to $F$ is precisely a 2-natural transformation $W \to \hat{F}N : I \to \textbf{Cat}.$ Just substitute these specific 2-functors and 2-categories for the arbitrary ones in the definition of 2-natural transformation and you have your answer. $\endgroup$ May 21 '16 at 0:47

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.