On Wednesday, 8. December 2010 16:08:42 Mattia Lambertini wrote:
I am not sure if this correct, but respecting the concept of PBGL, where the vertices are local to processors, it seems reasonable. After all, the edges are the same, except they are split up into different subgraphs, which again looks reasonable if one thinks of the subgraphs local to the processors (due to the local vertices).
I am sorry, i wasn't clear enough. I am talking about the 'label' attribute that should be the distance (number of edges) from vertex 0 to each other reachable vertex (the owner of the attribute). Maybe i misunderstood the distance computed by the PBGL algorithm.. is it not the global distance computed by bfs algorithm usually?
Ok, got you now on this one. Following your explanation, this seems strange to me too. Perhaps you could attach the code of a minimal working example? I have no way to test distributed graphs (therefore no example files available for me), nevertheless I would like to have a look at it. Thank you. Best, Cedric