Revision history [back]

Hy, I've test several options:

Implement the pbitMCE algorithme : FAILED (I'm not enought skilled...)

Subdivise the primary graph in as many ego_graph as nodes (without previous nodes to prevent duplicate cliques), and solve MCE for each one simultanatly ; WORKS but isn't a good parallel way (some subgraph have many cliques and some none, so firts calls use all cores and lasts use only two or one.

As the second way : Subdivise the primary graph in as many ego_graph as nodes, test if they can contain a max_clique ; subdivise good one in as many ego_graph as nodes they contain, test if they can contain a max_clique ; subdivise each good one in as many ego_graph as nodes, test if they can contain a max_clique ; and finally solve MCE for each "good case"; WORKS in theory but i've issues... I think the node ordering is not the same each time i solve ego_graph, result : I've some duplicates cases and some accendently erased... so it's a FAIL...

So I'm still looking for a good way to enumarate max clique on multiple cores... Someone is skilled enought for the first way?