nLab
directed n-graph

Idea

A directed n-graph, or n-digraph, is a higher dimensional generalization of a directed graph with r-dimensional edges spanning (r1)-dimensional vertices.

A directed n-graph is like an n-category with units and composition forgotten. Indeed, an n-category is a directed n-graph with extra structure. To formalize this idea, we say there is a forgetful functor

U:nCatnDiGraphU : n\Cat \to n\DiGraph

where nDiGraph is the category of directed n-graphs and nCat is the category of small n-categories. Moreover, this forgetful functor has a left adjoint

F:nDiGraphnCatF : n\DiGraph \to n\Cat

sending each directed n-graph to the free n-category on that n-graph. A free n-category on an n-graph is called an n-quiver.

Definition

An abstract directed n-graph X is a category with

  • objects X r of r-dimensional edges (or r-edges) for 0rn;

  • morphisms s r,t r:X r+1X r, called sources and targets for 0r<n;

  • together with identity morphisms Id r:X rX r for 0rn.

A directed n-graph is a functor G:X Set.

A directed n-graph in a category C is a functor G:XC.

An r-edge eG r is called an identity (or r-loop) if

s(e)=t(e).s(e) = t(e).

A morphism i r:X rX r+1 sending an r-edge to its respective (r+1)-loop, when defined, is called an identity assigning map. Identity assigning maps satisfy

s ri r=Id rs_r\circ i_r = \mathrm{Id}_r
t ri r=Id r.t_r\circ i_r = \mathrm{Id}_r.

When identity assignment maps are defined for all 0r<n, the directed n-graph is referred to as a directed n-graph with identities.

Remarks

  • A directed n-graph is not required to contain identities.

Let G r=G(X r), s r=G(s r), t r=G(t r).

  • Any two consecutive objects G r,G r+1 together with morphism s r,t r,Id r,Id r+1 constitute a directed graph. In particular, a directed 1-graph is a directed graph.

  • One can define a (globular) directed ω-graph (or directed -graph) in a coalgebraic fashion as follows: A directed ω-graph G consists of a collection G 0 of vertices and, for each pair of vertices x and y, a directed ω-graph Hom(x,y). (This is a definition by structural coinduction?.) Can this be continued to a corecursive definition of -category?

See also


Mike Shulman: I had not seen this page before it was linked from finite category, but I don’t really like ”n-graph” as a name for these. When I see ”n-graph” I automatically think ”n-globular set”, and I expect the same will be true for many category theorists; I’ve seen the phrase used that way in print as well.

Eric: I really like the generality of this definition. Globular sets seem too limited to me, e.g. I like to have square 2-cells, but that is probably due to a less than perfect understanding. You can see some discussion and references on Revision 8 of this page. If we really decide to limit n-graph to globular sets we can certainly change the name.

Mike Shulman: But you don’t have square 2-cells here either! A square 2-cell would have to have the composite of two 1-cells as its source and target, but here your 2-cells only have a single 1-cell as their source and target. It sounds like maybe what you really want is an n-computad.

Eric: Thanks Mike! Todd mentioned computads in the discussion at finite category, but I didn’t realize we had a nice page: computad. Maybe we can think about how to clean things up. Based on the opening paragraph at computad, they do seem to be what I was after. I need to do some homework now. Thanks again.