[ase-users] ase-vasp calculation problem
Alexander Sougaard Tygesen
alexty at dtu.dk
Mon Jun 22 15:41:57 CEST 2020
Hi Liu,
You need to tell the calculator how to execute, so like you mentioned, you need to execute through MPI.
You achieve this in the VASP calulator by setting the "VASP_COMMAND" environment variable, so you do something like
export VASP_COMMAND=mpirun -np 24 vasp_std
The calculators are not aware of any parallelization, they only execute the above command, so that command needs to be exactly how you would execute VASP without the calculator.
I hope this helps.
Regards,
Alexander
________________________________
From: ase-users-bounces at listserv.fysik.dtu.dk <ase-users-bounces at listserv.fysik.dtu.dk> on behalf of Liu via ase-users <ase-users at listserv.fysik.dtu.dk>
Sent: Monday, June 22, 2020 1:56:07 PM
To: ase-users
Subject: [ase-users] ase-vasp calculation problem
Hi everyone,
I have some python scripts which are based on modules of ASE and use VASP to caculate continuously (adjust the number of electrons according to result obtain by last VASP caculation and then start next vasp caculation). The scripts worked well when use one core to caculate parallel. But when I use more than one cores to run the scripts parrall, it just could finish the first VASP calculation and then it caused errors like this.
[cid:56721995 at 5F406148.579CF05E]
The parameters of caculator settings are as follows:
calc = Vasp(setups='recommended',
gga = 'RP',
xc = 'PBE',
kpts = (3,6,1),
ispin = 1,
encut = 400,
prec = "accurate",
ismear = 2,
sigma = 0.1,
nelm = 100,
ibrion = 2,
isif = 0,
nsw = 0,
ediff = 1E-5,
algo = "All",
# Parameters for quicker Convergence
nbands = 432,
npar = 8,
lreal = "A",
lvtot = True,
lvhar = True,
# Solvation
lsol =True,
eb_k = 78.4,
lambda_d_k =3.00,
nc_k = .0025, # default
tau = 0)
Are there any parallel tags like 'command= ‘mpirun -np 24’' in ase-vasp calculator settings?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20200622/3c76bc4d/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 56721995 at 5F406148.579CF05E.jpg
Type: image/jpeg
Size: 22780 bytes
Desc: 56721995 at 5F406148.579CF05E.jpg
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20200622/3c76bc4d/attachment-0001.jpg>
More information about the ase-users
mailing list