[Pw_forum] Phonon computation related problem

Lorenzo Paulatto paulatz at gmail.com
Wed Feb 14 09:32:29 CET 2018

On 02/13/2018 07:10 PM, Anindya Bose wrote:
> I am performing a phonon calculation of a Graphene 5*5 
> supercell(consisting defects) with 32 processors but it is taking too 
> much time and the predicted time of computation can be endless.Is there 
> any other way to do such phonon calculation where all the phonon modes 
> can be computed parallely so that the computation time will be 
> minimized.The supercell having 150 phonon modes.Is there any way 
> out.Please help.

This can be done in several ways, using phonon with images 
parallelization (automatically, as a big job with 32xn_images cpus¹) or 
using the grid parallelization scheme (by hand, as many small jobs of 32 
CPUs each²).

However, before wasting a lot of CPU time, I would share the input file 
to verify that you are setting the input parameters optimally.

Finally, for a large supercell doing a finite differences calculations 
can be more practical than DFPT at Gamma. There are several packages 
that promise to do it user-friendly (none does, in my opinion). The 
already mentioned phonopy, PHON by Dario Alfé, or the FD package in 
PHonon/FD (check the example)


1) see section 5 of 

2) see PHonon/examples/GRID_example

> Thanks and regards,
> Anindya Bose,
> Research Fellow,
> IIIT Allahabad
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum

Lorenzo Paulatto - Paris

More information about the users mailing list