Re: [igraph] Re: community detection

[Gabor Csardi]

Matthieu,

(i'm posting this to the list as well in case anyone is interested.)

i've added walktrap to igraph. This algorithm is a very nice piece of
work, congratulations. The igraph version is considered to be a bit
'nasty' for some reasons, but that is not your fault of course. :) 
It looks like this:

int igraph_walktrap_community(const igraph_t *graph, 
                              const igraph_vector_t *weights,
                              int steps,
                              graph_matrix_t *merges,
                              igraph_vector_t *modularity);

So it returns the history of the merges and/or the modularity after
each merge. I might add a function which can 'cut' such a dendogram
at the requested place. Also, the memory management is not used 
right now, but that can be added in two minutes. So there
will be some minor improvements, but the main part is done.
Hopefully the code will work on MacOSX, Windows, etc. as well without
too much hassle.

Our version control system is here:
http://cneurocvs.rmki.kfki.hu/cgi-bin/archzoom.cgi/address@hidden/igraph--main--0.4
in case you want to take a look at it. The web-server is quite slow,
sorry for that.

Thanks for the triangle algorithms, i could have used this paper
very well about a month ago when i rewrote transitivity calculation 
in igraph. I'll run some benchmarks to see how much improvement 
it would be such replace the current implementation.

Best,
Gabor

On Thu, Mar 29, 2007 at 05:08:13PM +0200, Matthieu Latapy wrote:
> Gabor,
> 
> thanks for your message, and passing mine on the list. I am
> actually a member, but I changed my affectation recently and
> forgot to update; sorry for the inconvenience.
> 
> I would like very much to incorporate our code into igraph.
> Unfortunately, I have no time for this currently, and the
> PhD student that implemented it will leave soon :p
> 
> If you are interested in more C code for complex network
> studies, you may have a look at my implementations of
> algorithms to compute triangles:
>  http://www.liafa.jussieu.fr/~latapy/Triangles/
> 
> Thanks again,
> ML
> 

-- 
Csardi Gabor <address@hidden>    MTA RMKI, ELTE TTK