Re: [igraph] Community Detection using leading eigenvector

[Gábor Csárdi]

Thanks for the suggestion. Yes, this is in theory a relatively easy change.

In practice it is not, however. The reason for that is that the
non-symmetric ARPACK eigensolver that is used to find the leading
eigenvector is not very well behaved. Many times it gives you the
wrong eigenvector, or no useful result at all.

Best,
Gabor

On Mon, Dec 19, 2011 at 6:19 AM, Claudio Martella
<address@hidden> wrote:
> I too second your feature request, it would be highly appreciated.
>
> On Mon, Dec 19, 2011 at 3:40 AM, Andrei Boutyline
> <address@hidden> wrote:
>> I would like to enthusiastically second this feature request.  It would be
>> really wonderful if the eigenvector community detection method was
>> generalized to weighted graphs.  As far as I can tell, the change just
>> consists of using the sum of incoming edge weights instead of vertix degree,
>> and of normalizing the modularity by the total sum of all edge weights
>> instead of by the total number of ties.  I would be very grateful if someone
>> on the igraph team could make this change!
>>
>> Thanks a lot!
>> Andrei
>>
>> On Sat, Dec 17, 2011 at 3:21 AM, Denzil Correa <address@hidden> wrote:
>>>
>>> Hi,
>>>
>>> I would like Prof. Newman's "Leading Eigenvector for Community Detection"
>>> algorithm on a weighted directed graph. I currently see an implementation
>>> for just the undirected (and unweighted) version. Am I missing something OR
>>> has this been left out?
>>>
>>> Thanks.
>>>
>>> --
>>> Regards,
>>> Denzil
>>>
>>>
>>> _______________________________________________
>>> igraph-help mailing list
>>> address@hidden
>>> https://lists.nongnu.org/mailman/listinfo/igraph-help
>>>
>>
>>
>>
>> --
>> Andrei Boutyline
>> University of California, Berkeley
>> PhD Student, Sociology
>> www.ocf.berkeley.edu/~andrei
>>
>> _______________________________________________
>> igraph-help mailing list
>> address@hidden
>> https://lists.nongnu.org/mailman/listinfo/igraph-help
>>
>
>
>
> --
>    Claudio Martella
>   address@hidden
>
> _______________________________________________
> igraph-help mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/igraph-help



-- 
Gabor Csardi <address@hidden>     MTA KFKI RMKI