Re: [igraph] cohesive.blocks V0.5

[Gabor Csardi]

Uri,

On Tue, Feb 12, 2008 at 02:46:00PM -0500, Uri Shwed wrote:
> Hi Gabor and Peter,
> See what I get and some data below. 

Please next time attach a file with the network, then we don't need 
to make tricks to transform the "printed graph" into a graph object.

For me cohesive.blocks works fine on it. I'll try R 2.6.2, 
i don't think this is the problem, but we'll see.

> I'm using R 2.6.2 on Scientific 
> Linux 4.5. I've just installed R today, and I'm a newby in both R and unix.
> Regarding bigger networks - If I have sufficient computing power, you 
> don't think this could work? 

Personally i don't, but if your networks are very special (for 
example tree-like) then it might.

> Are you familiar with other packages that 
> could work better?

No. I don't know any open source implementation of the cohesive
blocking algorithm.

> I have access to a cluster of 9 machines, with a total of 82G memory on 
> them. 

Not very useful for this calculation, as you cannot make the R process 
use all this memory, can you?

I'll check R 2.6.2, stay tuned,

G.

> My largest network is 12k nodes and 70k edges, but maybe I can 
> omit it. the second largest is 3k edges and 5k nodes. (these are 
> scientific citation networks).
> Thanks so much for all your work,
> Uri
> 
> 
[...]

-- 
Csardi Gabor <address@hidden>    UNIL DGM