[Nauty] geng changes

[Sterten at aol.com]

 >Ok, but don't call it geng.

I chose geng2.c,geng3.c. I'm not planning to make it public or give
it to others, unless maybe someone asks for it (unlikely)
I also have gengoh.c(oddhole),gengpf.c(perfect) etc.
and was wondering whether to put them all in a single big file, but there
are just too many possible interesting graph-types.

 >> I included a count[n]++; in my PRUNE  subroutine in geng,
 >> and display the count[] values before geng exists.
 >> Then not only all graphs of size n are counted but
 >> also all graphs of smaller size.
 >
 >There is no guarantee that all graphs smaller than the output
 >size are generated, so this is invalid.

in practice it seems to work, so I used it for my counts:
just copying the output to my counts-list.
I found no counterexample so far.

 >> I want to remove the -t, -f switches, instead include a list
 >> of forbidden subgraphs or read a list of forbidden subgraphs from file.
 >
 >However, -t, -f, -b are implemented much more efficiently than
 >is possible using PRUNE and should work correctly in conjunction
 >with PRUNE.  So removing them is a mistake.

efficiency is not so much a problem for me actually. And I still could use
the old geng.c for efficency.
Well, I'm not really removing them, just using lists of forbidden
subgraphs instead so the counts can be combined.
I wondered: what sorts of graphs are subgraphs of many(n) graphs
and which of few ?


Guenter
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.anu.edu.au/mailman/private/nauty/attachments/20021205/147d2e5d/attachment.html