My Community

Hello! I am attempting to relax a perovskite system with VASP 6.4.1 on Perlmutter, a cluster on NERSC. They have a special compilation of 6.4.1 to utilize the GPU nodes they provide so that may be relevant to the issue I am encountering. I would appreciate advice as to what I am doing incorrectly! Thank you!

Hi,

Thanks for reaching out. In order to assist you with your issue, I will have to ask you to provide more information. First, please provide relevant input and output files of the failed run (including the standard output that shows the error) according to our forum posting guidelines. Next, I would also need more information of what the "special compilation" entails. Could you attach the makefile.include from this compilation process? Are there modifications of the VASP source code on top of our 6.4.1 release?

Any information you can provide in this regard will be helpful. Thank you.

Absolutely, I think there may be some sort of bug on the forum because I was not allowed to attach my zip file containing all the relevant information even though it was only 1.1MB when I created the post originally (it said the file size quota was reached for some reason).

The requested files should be attached here.

Thank you!

Also here is the OUTCAR file, I forgot to include it in the zip. Regarding what modifications were made to the VASP version available on Perlmutter, I am not familiar or sure if any modifications were made. I did include the makefile.include in my previous reply.

Thanks for sharing the files. We are aware of an issue regarding file upload and are working on a permanent solution.

I examined the input and output files and noticed that some of the atoms specified in the POSCAR file are in very close proximity to each other. There is a warning about this in the standard output. Usually, this warning hints at severe problems with the atomic structure of the system that can cause all kinds of errors and unphysical behavior down the line. I suspect that the internal error you are encountering is also a symptom of the atoms being too close together. Therefore, I suggest that you carefully inspect and modify or recreate your POSCAR file so that these conflicting atomic sites are resolved.

If the problem still persists, feel free to reach out again.