InfoMap is complex and all of igraph is all single threaded to my
knowledge. I think your best bet would be to export an edge list and run
something like the label propagation algorithm (which only needs direct
neighbors) on multiple cores. That's what I ended up doing for a graph with
about a million nodes. The code I wrote is at:
https://github.com/computermacgyver/network-label-propagation

Cheers,
Scott

---Original Message---
Date: Mon, 12 May 2014 13:42:38 +0200
From: Alexander Struck <[email protected]>
To: Help for igraph users <[email protected]>
Subject: Re: [igraph] infomap unresponsive? on large graph in R env
Message-ID: <[email protected]>
Content-Type: text/plain; charset="iso-8859-1"

Hi,

has anyone any info on a memory limit of igraph which I may have missed in
the documentation?
I'd really like to use igraph's implementation of InfoMap but run into the
problem described below.

Thanks,
Alexander


On May 6, 2014, at 11:21 AM, Alexander Struck <
[email protected]> wrote:

> Dear all,
>
> while InfoMap runs smoothly on small graphs it appears to become
unresponsive on larger ones. My .RObj has the size of 28 MB and igraph
reports:
>> G
> IGRAPH DN-- 1282336 2507034 --
> + attr: name (v/c)
>
> The R v3.0.2 process now runs 14+ hours on one core out of 16, Ubuntu
12.04 uses 16 of the available 72 GB RAM.  The latest igraph library
version is installed.
> Should I be more patient? What are my options here? I could make the
.RObj available on request.
>
> Many thanks and best regards,
>
> Alexander
>
>
> _______________________________________________
> igraph-help mailing list
> [email protected]
> https://lists.nongnu.org/mailman/listinfo/igraph-help
_______________________________________________
igraph-help mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/igraph-help

Reply via email to