Open jasongrout opened 16 years ago
[11:28am] rlm: given two graphs, and a surjection from one to the other (not necessarily a graph hom.), we can construct a bundle
[11:29am] rlm: further, all bundles arise in this way
[11:29am] rlm: so instead of a partition, give both graphs and a map
[11:29am] rlm: that allows for any bundle to be constructed, right?
[11:34am] jason: rlm: the first graph is the petersen graph, for example, and the second is K_2, right?
[11:34am] jason: Yes, I think you're right that every bundle can be described that way.
[11:35am] rlm: so at least we have one way of constructing, so that nothing can't be constructed
[11:55am] rlm: ok, G1 is a 5-cycle
[11:55am] rlm: G2 is K(3,3)
[11:55am] rlm: G3 is empty on 3 vertices
[11:55am] rlm: there's an obvious bundle here, and it should be brainless to construct
[11:56am] rlm: like GraphBundle(base=G1, edge_lifts=G2, vertex_lifts=G3)
[11:56am] rlm: and it just works
[11:56am] rlm: no matter how you glue, you'll get the same thing
[11:57am] rlm: maybe instead of, if the vertices match up, just match them according to their orderings
[11:57am] jason: That's what Chris was saying when he said "(There is an implicit orientation here)"
Also, there are several functions of generic graphs that should be overridden, especially things like add_vertex etc.
See #18028
1306 laid some of the groundwork for finishing this request, but it's not finished yet, so I'm opening another ticket with the original request.
Component: graph theory
Issue created by migration from https://trac.sagemath.org/ticket/1928