Error: supplied exchange-correlation table is too small, maxmimal index: 16672

[phillip_rotzell]

Hello,

I've been trying to optimize TiO2 anatase 010 for some time now, but I repeatedly receive this error. I've previously tried setting ISIF=3 but it yielded a structure that shrunk the vacuum too much. Attached is the INCAR and POSCAR I've been using. Any help is appreciated.

Screenshot 2025-09-03 112508.png

[manuel_engel1]

Hello,

Welcome to the VASP forum and thanks for reaching out. Could you please upload all the relevant input and output files according to our forum posting guidelines. Otherwise, it will be very difficult to say what the issue in this calculation is.

[phillip_rotzell]

Apologies, here's the zip file you requested.

[manuel_engel1]

No worries. Unfortunately, it seems that you have not uploaded any files. Could you please try again?

[phillip_rotzell]

Apologies, does this work?

[manuel_engel1]

Thanks. There are a few things potentially wrong with your calculation. I believe that the error you see is just a symptom that is somehow related to these things. Could you please try to address these issues and then see if you still run into the same problem?

• Your NCORE value does not make sense. VASP writes a warning that it will reset it to 52, which is all the available CPU cores on this node/job. This is usually not very inefficient. I suggest that you change it to 4 or even 1 to see if it resolves the issue.

• Do not set NPAR and NCORE at the same time.

• For spin-polarized calculations, you should set MAGMOM manually in the INCAR file.

• You specify IALGO twice with different values. In your case, VASP chooses pure RMM-DIIS (IALGO=48). This might be the main reason you see your instability. It's usually best to use blocked Davidson first and then switch to RMM-DIIS later. I suggest to switch to ALGO=Normal or ALGO=Fast instead.

Let me know if that helps.