Re: [igraph] Issue with Bonacich centrality: "cs_lu(A) failed"

[Gábor Csárdi]

Sorry for the late reply. power_centrality (which is the same as
bonpow) is fragile with degenerate graphs, e.g. graphs that are
disconnected, have isolates, etc.

AFAIK we implement the method as it was published. Maybe UCInet does
some workaround. I wouldn't mind implementing one. I have opened an
issue for this:
https://github.com/igraph/igraph/issues/904

Gabor

On Tue, Sep 8, 2015 at 10:32 AM, Martin Goossen <address@hidden> wrote:
> Hi everyone,
>
> When computing the Bonacich centrality of a large, sparsely-connected
> network (1310 vertices with 6717 edges), i get the following error message:
>
> Error in .local(a, b, ...) :
>   cs_lu(A) failed: near-singular A (or out of memory)
>
> The graph is available via
> https://www.dropbox.com/s/9ba0l3i9ozm0axe/graph.txt?dl=1 (in graphml
> format).
>
> What i have tried so far:
> - using both igraphs' "power_centrality" function or the former  "bonpow"
> function provide the same error
> - computed eigenvector centrality (no errors), alpha centrality (same
> error), computed Bonacich centrality in UCInet (no errors).
> - increased memory to max (around 4GB)
> - checked earlier posts and found one relevant, but unresolved thread about
> six years ago:
> http://lists.nongnu.org/archive/html/igraph-help/2009-04/msg00040.html
>
> Any help is highly appreciated, thanks in advance!
>
> Best regards,
> martin
>
>
> _______________________________________________
> igraph-help mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/igraph-help
>