« first day (40 days earlier)      last day (312 days later) » 

Thomas Klimpel
Thomas Klimpel
10:21 AM
klimpel@foxme:~/x/nauty/Fahad$ g++ iso_inter.cpp
klimpel@foxme:~/x/nauty/Fahad$ ./a.out graphA.txt graphB.perm.txt
0 0111111111111000000000000 | 13 0001100001111001011011010
1 1011111000000111111000000 | 23 0011111000000011001111001
2 1101111000000000000111111 | 10 0101001110100001011100110
3 1110000111000111000111000 | 14 1110101111110010000001000
4 1110000100110100110100110 | 16 1101000011000111100110010
5 1110000010101010101010101 | 6 0100001001100010110011111
6 1110000001011001011001011 | 18 0111010011001100111001000
@Jim OK, the example graph was completely labeled by individualizing the vertices 0, 1, and 3. So I will need another graph as counterexample. The formatting of the pasted results is also not great yet. (The subset of 2D Weisfeiler Leman that you algorithm uses is also not yet implemented.) But you can download the .cpp program from the link above, compile it, and play with it, if you like. (I have to leave now...)
 
Jim
Jim
10:58 AM
Received.
 
Jim
Jim
11:36 AM
Do you need to tell me anything?
 
Thomas Klimpel
Thomas Klimpel
@Jim No, I just tried whether I could download "iso_inter.cpp" myself from my computer at work.
 
Jim
Jim
11:58 AM
ok
@ThomasKlimpel , what is your statement ? I mean this is not proving anything wrong, is it? it looks fine to my "hypothesis"
 
Thomas Klimpel
Thomas Klimpel
@Jim When I have time, I will try it with some graphs from the link above. The initial goal would be to end in the "equipartition" case. When this doesn't work, then I will have to build the counterexample myself.
 
Jim
Jim
@ThomasKlimpel so you are looking for a counter example only?
 
Jim
Jim
12:21 PM
@ThomasKlimpel 1. I think algorithm can work without 2 dim W-L method 2.Would you please check the post math.stackexchange.com/questions/1610806/… 3.I need to relearn "Individualization", would you please give a source/ link for that?
4. You have spent a lot of time on a dimwitted undergrad, I am not sure whether i deserve it or not, thanks for your time. I think you should ask me making this kind example, also ask Questions/proof as much as you want, it does not matter how little\trivial they are.. I am keen to give elaborate answer, that way, you will need less time.
 
Jim
Jim
12:34 PM
Jan 9 at 16:43, by Thomas Klimpel
@Jim "It is assumed that regular subgraphs in Partition 1.1, 1.2, are not complete graph, cycle graph, trivial graph(these graphs’ isomrphism test is trivial)." So you already try to exclude the case where the bipartite graph would occur most naturally. But how can you claim that isomorphism testing for those graphs would be trivial?
5. "It is assumed that regular subgraphs in Partition 1.1, 1.2, are not complete graph, cycle graph, trivial graph(these graphs’ isomrphism test is trivial)." So you already try to exclude the case where the bipartite graph would occur most naturally." -- these conditions are lifted,you can consider complete graph, cycle graph, trivial graph.
 
Thomas Klimpel
Thomas Klimpel
12:50 PM
@Jim "Individualization" is really trivial. Whenever you do a partition (of the vertices), I just assign a "unique" color to the vertices of the partition. And when you select a vertex with the intention to create a partition into two submatrices, I just assign a "unique" color to that vertex. This assigning a "unique" color is called "Individualization". (The subsequent 1D Weisfeiler Leman automatically takes care to create the intended submatrices.)
@Jim I would hope that you try to compile "iso_iter.cpp" and try to play with it a bit.
 

« first day (40 days earlier)      last day (312 days later) »