vasp machine learning force field run

Queries about input and output files, running specific calculations, etc.


Moderators: Global Moderator, Moderator

Post Reply
Message
Author
jpg
Newbie
Newbie
Posts: 10
Joined: Sat Sep 21, 2013 2:03 am
Contact:
Contact jpg

vasp machine learning force field run

#1 Post by jpg » Thu Feb 13, 2025 10:29 pm

Hi

On our expanse cluster ( https://www.sdsc.edu/systems/expanse/user_guide.html ) we have a user who was running the attached MD job using machine learning force fields with vasp 6.5.0 on 64 processors on a single node.

I was testing it for him and found that after 4 hours , not even a single SCF iteration was completed.

Is he doing something wrong?

Attached is a tgz file containing the input, standard output and OUTCAR files.

Thanks,

Jerry

You do not have the required permissions to view the files attached to this post.
Top
manuel_engel1
Global Moderator
Global Moderator
Posts: 188
Joined: Mon May 08, 2023 4:08 pm

Re: vasp machine learning force field run

#2 Post by manuel_engel1 » Fri Feb 14, 2025 11:42 am

Hi Jerry,

Thanks for reaching out. I cannot see any obvious fault with the calculation. In fact, I have run it on three different compute nodes with different hardware and the SCF cycles are run relatively quickly.

Could it maybe be an issue with buffering, so that the SCF cycles are computed but not yet written to the output? You could check with a smaller calculation (fewer atoms in the supercell) and see if the calculation finishes. Do regular DFT calculations run without issue?

Manuel
VASP developer
Top
Post Reply

Return to “Using VASP”