[Rsiena-help] ML estimation for 2-mode networks

[Ruth M. Ripley]

Dear Marco,

I am afraid the ML estimation routine for bipartite networks is not 
correct, despite my hopeful note in the changelog. I will fix it very 
soon.

I will look into the error you got. It should be possible to use two cores 
for three waves.

Regards,

Ruth
--
Ruth M. Ripley,                         Email:ruth at stats.ox.ac.uk
Dept. of Statistics,                    http://www.stats.ox.ac.uk/~ruth/
University of Oxford,                   Tel:   01865 282851
1 South Parks Road, Oxford OX1 3TG, UK  Fax:   01865 272595


On Wed, 23 Feb 2011, marco tonellato wrote:

> Dear all,
>
> while trying out the new ML option with Revision 138 of RSiena on my
> two-mode data (3 panels, 135x719 actors, very sparse networks), I
> encountered this error when running it on multiple cores:
>
> Error in robmon(z, x, useCluster, nbrNodes, initC, clusterString,
> clusterIter,  :
>  Not enough observations to use the nodes
>
> Yet, when I switched to single core I was able to start the model, although
> it's taking forever: after 3 hours on my server it's still at Phase 2
> Subphase 1 Iteration 50 Progress: 4%
>
> My effects object included just density and 4cycles, and the standard
> initial values option was set to TRUE
>
> Do you have any clues on how to proceed?
>
> Thanks in advance,
>
> Marco
>