[gpaw-users] Unable to view(atoms) with mpirun gpaw-python

[Alex Eftimiades]

That is nice to know. I was however trying to view the atoms to figure  
out why I got an MPI_Bcast error after running a get_potential_energy  
calculation on some atoms. I may send another message out in a few  
days if I cannot figure it out.

Thanks,
Alex

PS: I understand your reasoning for that feature, but I cannot help  
but wonder whether it might be better to let the user decide whether  
he/she wants to use the "view" function to debug an parallel  
calculation rather than for deliberately preventing it.

On Dec 31, 2011, at 9:17 AM, Ask Hjorth Larsen wrote:

> Hi
>
> On Fri, 30 Dec 2011, Alex Eftimiades wrote:
>
>> I have ase, gpaw, mpi (openmpi), and python installed via aptitude on
>> Ubuntu 11.10.
>> I did a simple set of tests:
>>
>> I have two files:
>>
>> File 1:
>> from ase import view
>> from ase.structure import nanotube
>> from gpaw.mpi import *
>>
>> tube=nanotube(4,4)*(1,1,3)
>>
>> view(tube)
>> ___________________________________________
>>
>> File 2:
>> from ase import view
>> from ase.structure import nanotube
>> from gpaw.mpi import *
>>
>> tube=nanotube(4,4)*(1,1,3)
>>
>> if world.rank==0:
>>    view(tube)
>> world.barrier()
>> ____________________________________________
>>
>> I do not see any reason for putting world.barrier() at the end of  
>> File
>> 2, but considering how little I know about how gpaw-python works with
>> mpi, I thought I would be safe and leave it there.
>> I executed these two files with each of the following commands:
>>
>> python
>> gpaw-python
>> mpirun -np 4 gpaw-python
>> _____________________________________________
>>
>> I received the following results:
>>
>> 		python			gpaw-python		mpirun -np 4 gpaw-python
>> File 1:	correct result		correct result		nothing
>> File 2:	correct result		correct result		nothing
>>
>> So the question remains--how can I get a view of my atoms with mpirun
>> gpaw-python? If you are still skeptical of the functionality of my  
>> mpi
>> or gpaw-python, I can assure you I have been able to calculate the
>> potential energy of a nanotube(4,4) using mpirun -np 4 gpaw-python
>>
>> Thanks,
>> Alex
>
> In the ase.visualize.view function there's an if statement which  
> causes the view command to not do anything when run in parallel. The  
> idea appears to be that most people wouldn't want to open viewers  
> when running a 'serious' calculation.
>
> Regards
> Ask