Given that a Quiver (also known as directed multigraphs) is so important in category theory, being that every Quiver produces a free Category, and every Category has an underlying Quiver, I was wondering what a limit of a Quiver is. (See also the nice overview of Quivers and Diagrams in the answer on finding the limits of a diagram)

So if I take a category with two objects E and V and two functions s,t: E → V which given the start and end edges of vertices, then I should be able to have a functor from that to sets, which determines a quiver. Then all I need is to add a constant functor to some object of set say $L$ - the putative limit - and then find the real one. So for each vertex of the diagram, $v$, we have an arrow $p_v:L\to D(v)$. In this case it would be have the equations $p_V = s \circ p_E$, $p_V = t \circ p_E$. Which leads to $s \circ p_E = t \circ p_E$.

What would the co-version be? I suppose we turn the arrows around $p_v: D(v) \to L$ giving us $p_E = p_V \circ s$ and $p_E = p_V \circ t$, and so $p_V \circ s = p_V \circ t$

(I'd like to also ask what Monad one could build with the Free and Underlying Functors, but I guess that should be a different question).

  • $\begingroup$ To be clear, you're just asking for the (co)limit of a functor from your two-object category to $\mathtt{Set}$? $\endgroup$ – Eric Wofsey Aug 26 '18 at 15:42
  • $\begingroup$ Two object and two arrows. Two objects alone would give us (co)products. $\endgroup$ – Henry Story Aug 26 '18 at 15:44

(Co)limits of this form are called (co)equalizers. In sets, an equalizer of a functor $D$ from your category is just given by the set $L=\{x\in D(E): D(s)(x)=D(t)(x)\}$ with its inclusion map to $D(E)$. (So, in terms of quivers, it is just the set of loops.) Coequalizers are a bit more complicated, but a coequalizer can be constructed as the quotient of $D(V)$ by the smallest equivalence relation $\sim$ such that $D(s)(x)\sim D(t)(x)$ for all $x\in D(E)$. (In terms of quivers, this quotient is the set of connected components of the quiver.)

Note that colloquially, one typically just speaks of the (co)equalizer of a pair of morphisms $f,g:A\to B$ in some category, without talking about a functor explicitly. This refers to the (co)limit of the functor from your category which sends $E$ to $A$, $V$ to $B$, and $s$ and $t$ to $f$ and $g$.

| cite | improve this answer | |
  • $\begingroup$ Yes, that is quite neat. Having written down my question as above I was thinking this looked very much like (co)equalisers :-) Thinking about them in terms of quivers will I think make them more intuitive for me. $\endgroup$ – Henry Story Aug 26 '18 at 16:25
  • $\begingroup$ Interestingly this means that what corresponds to products here are looping arrows which are I suppose the most independent form of arrows, a bit like products bring two things together but in the most independent way (orthogonally). And co-products that are disjoint unions here correspond to groups of nodes that are somehow linked up, each group constituting such a disjoint set. $\endgroup$ – Henry Story Aug 27 '18 at 11:30

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.