Page 1 of 1

q points weight in epw input files

Posted: Mon Sep 04, 2017 12:57 pm
by nadeemnatt
Hi
Please tell me how to give weights to the q points that we have to mention at the end of input file for epw. As I think these points are same as in dynamical matrices calculations on coarse grid. since ph.x output file does show those points without their weights. Thanks

Best Regards
Nadeem
Sungkyunkwan university, Korea

Re: q points weight in epw input files

Posted: Tue Sep 05, 2017 8:50 am
by sponce
Dear Nadeem,

The list of coarse q-point that you need to provide at the end of an epw calculation indeed need weights.

However those weights are not used at the moment so you can put whatever you want (like 1.0).

I have changed EPW so that it is not needed anymore (will be in the next release).


Best,
Samuel

Re: q points weight in epw input files

Posted: Wed Sep 06, 2017 7:52 am
by nadeemnatt
Thank you Samuel
I have another question. I am running a 70 atom calculation with following details
coarse grid:
2x2x2 qpoints (8 qpoints)
8x8x8 kpoints (512 points) in nscf calculation
fine grid:
4x4x4 qpoints
16x16x16
nbndsub=136
nbndskip=0

I am running on 40 cpu in 5 nodes and memory is 64G. How much time will it take? How can I estimate the time needed? would you please recommend me how can I shorten the time? Thank you

Best Regards
Nadeem

Re: q points weight in epw input files

Posted: Wed Sep 06, 2017 10:34 am
by sponce
Dear Nadeem,

Computing el-ph properties in a 70 atoms system is very challenging.

The largest system I personally did had 48 atoms and it is very expansive.

If you want to estimate the time that it will take, you first need to do a phonon bandstructure calculation. Based on the time taken for that, you can estimate that EPW will
take the same order of magnitude amount of time.

Best,
Samuel