[Octopus-users] Parallel in k points

[Xavier Andrade]

Hi Abram,

Unfortunately we haven't implemented parallelization in k points yet but 
we will probably do it soon. Parallelization in domains is now implemented 
for periodic systems in the development version. What kind of information 
you would like to obtain for these systems?

Cheers,

Xavier

On Fri, 1 Aug 2008, avanderg at mines.edu wrote:

> Hi all,
>
> I apologize if this got sent twice my email client is acting up.
>
> I am trying to run a calculation for a single walled carbon nanotube
> that is periodic in one direction (I have tried it periodic in 1 and 3
> directions).  But when I  try to run the ground state with
> ParallelizationStrategy = par_kpoints I get an error message that says:
>
> * More than one node is available, but the this run mode can not run
> in parallel
> * Please select a ParallelizationStrategy compatible with
> *   - par_domains
>
> But par_domians isn't set up to work with periodic systems.  So I
> looked online and there is a to do list of things to go into version
> 3.0.0 (I am running 3.0.0) and parallelization over k points was on
> that list, but I couldn't find anything that confirmed that it had
> been included.    I would appreciate any help I can get.  Oh and this
> system runs in serial but takes a long time so I would like to be able
> to parallelize it for future systems.
>
> Thanks
>
> Abram Van Der Gest
>
> _______________________________________________
> Octopus-users mailing list
> Octopus-users at tddft.org
> http://www.tddft.org/mailman/listinfo/octopus-users
>