Re: [igraph] bug in igraph_shortest_paths_dijkstra?

[Tamas Nepusz]

Hi Davide,

> I'm using igraph from bzr because with igraph-0.5.2 the "giant" graph
> is not strongly connected (maybe a problem with DiGraphs?)
It seems to be strongly connected for me in igraph 0.5.2:

>>> from igraph import *
>>> g = load("graphCurrent.net")
>>> giant = g.clusters().giant()
>>> print giant
Directed graph (|V| = 19970, |E| = 237535)
>>> giant.is_connected(STRONG)
True

> With the bzr version (0.6 branch) the giant graph has ~20000 nodes
> (like networkx says), but the shortest path function doesn't work.
This is my output in igraph 0.6:

>>> from igraph import *
>>> g = load("graphCurrent.net")
>>> giant = g.clusters().giant()
>>> print giant
Directed graph (|V| = 19970, |E| = 237535)
>>> giant.is_connected(STRONG)
True
>>> sps = giant.shortest_paths(0, weights="weight")[0]
>>> len(sps)
19970
>>> sps[17]
5.0

Note that you don't have to use shortest_paths_dijkstra, igraph will
make that choice automatically when you supply weights (and it even
switches to Bellman-Ford if it detects negative weights).

So I'm quite lost here, as I can't reproduce your problem. Can you
please copy the exact commands you use to load the graph, construct the
giant component and call shortest_paths?

Thanks,
-- 
Tamas

pgpzHzsjxOvJ2.pgp
Description: PGP signature