Re: [igraph] Re: Calculating measures for a bunch of ego-networks

[Gábor Csárdi]

On Thu, May 6, 2010 at 4:36 PM, Magnus Torfason <address@hidden> wrote:
[...]
> Again, I find it very interesting that E(g, P=temp) is much much faster than
> E(g)[edge.ids]. It seems that there must be some problem with the subsetting
> operator [] for edge sequences.
[...]

Yes, there is. I think I have fixed this, so it should be much faster
in the 0.6 version.

> Now, unfortunately all of the above applies to version 0.5.2-2.
>
> I have downloaded the 0.6 version of igraph and am looking into installing
> it, but I am finding the prospect a little bit daunting. I'm on Windows, and
> have never installed R packages from source. Are there directions on the web
> somewhere specific to igraph, or is this a matter of installing R-tools and
> that whole environment as per the (somewhat lengthy) guidelines for
> developing your own packages on Windows?

There is an online service to build windows packages:
http://win-builder.r-project.org/

All you need is to extract the package, change the maintainer email
address in the DESCRIPTION file, recompress the package, and upload it
to the site.

So theoretically it is simply, practically it is not, at least not
right now, because we started to use the GLPK library and that is not
available on the win-builder machine.

I don't have time right now, but will solve this issue asap, probably
this evening.

Best,
Gabor

> Best regards,
> Magnus
>
>
> On 5/6/2010 8:34 AM, Gábor Csárdi wrote:
>>
>> Magnus,
>>
>> you can try the 0.6 development version, it has some improvements for
>> V() and E() and also for %--%. You can download it from
>> http://code.google.com/p/igraph/downloads/list
>>
>> If you want all edges that connect two vertex sets, and your graph is
>> not changing much, then you can query the edge list of the graph and
>> extract the edges from there:
>>
>> el<- get.edgelist(G)
>> which(el[,1] %in% myfrom &  el[,2] %in% myto)-1
>>
>> On a second thought, this is more or less what %--% is doing, so I am
>> not sure that it will be any faster.
>>
>> Please tell me if the 0.6 version solves your problem. Thanks,
>> G.
>>
>
>
> _______________________________________________
> igraph-help mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/igraph-help
>



-- 
Gabor Csardi <address@hidden>     UNIL DGM