On Fri, Jul 3, 2009 at 8:58 PM, Tamas Nepusz
<address@hidden> wrote:
Hi Richard,
It doesn't look like igraph has support for the graph>RAM problem, though, but I thought I'd ask around to see how others deal with the problem (buying more RAM won't work for this machine).
We have a branch in igraph's Launchpad repository (maintained by Emmanuel Navarro) that replaces igraph's lowest in-memory graph storage layer with another one that stores the graph in an SQL database. Actually, there are two branches, one is based on the development tree, the other one is on the stable 0.5 tree:
https://code.launchpad.net/~enavarro222/igraph/0.5.1-db
https://code.launchpad.net/~enavarro222/igraph/0.6-db
I have never tried these branches myself, and I remember that Manu told that although it works, a caching layer on top the database layer would be desirable to make the thing efficient. There's another: some igraph functions start by converting whatever graph representation they get into an in-memory adjacency list, so at this point the graph must fit into the memory again (but of course one can rewrite these functions to use the original representation).
The relevant threads in the igraph-help mailing list archive are as follows (I think):
http://lists.gnu.org/archive/html/igraph-help/2009-03/msg00183.html
http://lists.gnu.org/archive/html/igraph-help/2008-12/msg00009.html
http://lists.gnu.org/archive/html/igraph-help/2008-12/msg00001.html
--
Tamas
_______________________________________________
igraph-help mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/igraph-help