VASP error when using many atoms in large supercell

Message

vasp_user · #1 Post by **vasp_user** » Thu Oct 18, 2012 11:10 am

This is the error that I'm concerned about:

WARNING: CHECK: NIOND is too small

I'm using 450 atoms in my supercell.

I have two questions and right now I'm more concerned about the first one:
1) Can I trust the final "energy(sigma->0)" that I'm getting???
2) What is the solution? I saw a 2006 post that said to recompile VASP... I can do that if the answer to this question is the same... but please, I would like to know the answer to the first question because I didn't notice this error until a couple days ago and I have a lot of previous results with this large supercell.

Thanks!!!!

tlchan · #2 Post by **tlchan** » Tue Oct 23, 2012 4:25 am

The main program calls a subroutine called check. The subroutine uses the evaluated total energies, the forces, and the atomic movements to check if the force and the change in the total energy between atomic movements are consistent. For some reason, the variables within the subroutine are defined to have a fixed size. At most 256 atoms can be handled. I think the checking is for a diagnostic purpose, and it won't affect your result. If you want to have the checking done after each ionic movement, you can modify the definition of the variables and re-compile to get rid of the warning message.