Are FFT grids (NGX,NGY,NGZ) effective in a simulation with ML_MODE=run?

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


Moderators: Global Moderator, Moderator

Post Reply
Message
Author
xiancai
Newbie
Newbie
Posts: 11
Joined: Sun Nov 17, 2019 9:42 am

Are FFT grids (NGX,NGY,NGZ) effective in a simulation with ML_MODE=run?

#1 Post by xiancai » Fri Mar 21, 2025 2:28 pm

Dear VASP developers,

I am performing a simulation using a generated machine-learning force field (ML_FF) with ML_MODE=run. At the beginning of the output file, a warning appears stating: “Your FFT grids (NGX,NGY,NGZ) are not sufficient for an accurate calculation. Thus, the results might be wrong. Good settings for NGX, NGY and NGZ are 6, 6 and 6, respectively.” The full message is shown in the enclosed figure.

According to my understanding, the FFT grids (NGX,NGY,NGZ) are only relevant for first-principles calculations, and thus, this warning can be disregarded. I would like to confirm whether my understanding is correct. Thank you for your time!

Sincerely,
Tian Hua

You do not have the required permissions to view the files attached to this post.

martin.schlipf
Global Moderator
Global Moderator
Posts: 594
Joined: Fri Nov 08, 2019 7:18 am

Re: Are FFT grids (NGX,NGY,NGZ) effective in a simulation with ML_MODE=run?

#2 Post by martin.schlipf » Fri Mar 21, 2025 3:13 pm

Yes, your understanding is correct. In ML_MODE = run, VASP does not need to perform FFTs so it is not important that NGX, NGY, and NGZ are large enough.

Martin Schlipf
VASP developer


Post Reply