Database of Graphs

[Gordon Royle]

> their intersection is an interesting class that is still looking for 
> publishable results. A database could provide an excellent initial 
> direction in the investigation. I know I don't want to draw up all these 
> graphs up to 10 vertices and look for common properties...
>   If anyone else out there sees a use in this and is interested in maybe 
> starting to build something useful, let me know. Maybe the integer 
> sequence crowd is not the right one to be asking about this, but I think 
> OEIS could eventually incorporate such functionality.

Although Brendan is correct when he says that re-generating graphs is
usually much better than downloading them from a database, I can see
a use for a fairly comprehensive catalogue.

Even if each property individually is relatively easy to compute, it
still may be tedious to download programs for a large number of 
properties, and if one included a number of more difficult to compute
properties (such as induced subgraphs etc), then it would be even 
harder.

Many of the graph classes mentioned by Jim Nastos are actually very
small (split graphs, ATF graphs etc) and so even a modest combination
of properties would rapidly reduce the files to manageable size. 

A database approach limited to the 12 million+ graphs of order up to 10
with a number of pre-computed properties would permit a server-side 
computation based on combinations of desired/forbidden properties and
could be implemented fairly easily, modulo the choice of properties and
developing the appropriate programs.  

I would be quite interested in adding such a facility to my existing
Website of combinatorial data (http://www.csse.uwa.edu.au/~gordon/data.html),
which for those interested is "more or less" complementary to Brendan's
in that we both use his "graph6" format and a basically similar set
of tools, but have chosen slightly different collections of data to 
reflect our interests. We have been planning to integrate the data 
portions of our sites for about 5 years now, without any action.
Again, like Brendan, momentum for site development often comes from
user requests, which I am always happy to consider.

Cheers

gordon


-- 
Dr. Gordon F Royle, http://www.cs.uwa.edu.au/~gordon, gordon at cs.uwa.edu.au
--