Guys, I need your help and opinions.
Recently I invented an algorithm for the subject ( the C++ code at sf.net: sourceforge.net/projects/griso/).
But I can't find a counterexample for it. I tested it against a lot of just regular and strongly regular graphs. No avail.
If I test SRGs (of/with the same parameters, of course), say, 5000 items in bucket then I test all possible pairs, i.e., 5000 * (5000 - 1) / 2 pairs.
See what does its input/output look like here: funkybee.narod.ru/graphs.htm/ I need badly to construct a counterexample, but I can't.
If some of you have hard pairs of graphs for GI testing, plz, send them to me: funkybee@yandex.ru
PS
I need only in pairs of NON-isomorphic general indirected graphs, but which are very difficult distinguished, one from another.
With number of vertices < 1000. My algo is poly-time, ~ O(n^3), and so far I don't much elaborate its technical side.
Cheers!