Boost logo

Boost :

From: Oliver Kullmann (O.Kullmann_at_[hidden])
Date: 2006-09-19 13:03:26

Yes, I'm very interested.

On Tue, Sep 19, 2006 at 09:29:21AM -0700, Brian Makin wrote:
> Pinging interest in a hypergraph library.
> (there seems to be interest as per mailing list and
> boost summer of code)
> Couple questions to start the topic.
> Extension to BGL or seperate library? (prefered,
> mandatory, depends)

The problem with BGL is in my opinion, that for the central part of
the library, the concept development, there was a good start, but
then development didn't continue.
Now hypergraphs are much more diverse beasts than graphs (among other
things, they generalise the different types of undirected(!) graphs), and
thus the concepts are even more critical here than for a graph library.
So a hypergraph library needs a fresh start, with a completely new
concept discussion, and, if done right, it will also be bigger than
the graph library, so it is a separate thing.

And, as I just mentioned, for BGL the directed case is the base case, while
for the hypergraph case the undirected case must be the base case (directed
hypergraphs are, at this time, only a sort of curiosity).

Thus a hypergraph library cannot extend BGL, but must
go its own ways.
> Previous or current efforts towards this end?

The whole field of SAT solving extends the hypergraph business,
and much can be learned.

I'm not aware of a hypergraph library.

> Anyone particularly interested in this who also wants
> to begin a side discussion as to design and
> architecture?

This is the most important thing in my opinion --- and not a
trivial one. I'll be happy to contribute (though the next 3 weeks
I'll be not available (most of the time), unfortunately.


Boost list run by bdawes at, gregod at, cpdaniel at, john at