My Community

Dear all,
I have installed VASP in a computer cluster and, firstly all the calculations ran without any problem but when I started with adsorptions of metal atoms on oxides I detected that some calculations do not convert properly. Thus, I ran these calculations in other machine to compare the results and it was obtained that the SCF diverges after the 3-4 step leading to erroneous energies.
anyone knows how to fit this problem?
I suppose that the problem is related with the compilation of the libraries.

thanks

Luis

Is your self consistent field diverging after the non-self consistent steps at the beginning of the convergence?

I saw with some of my systems I got fast convergence in the non-self consistent steps (determined by the NELMDL tag), but once it got into self consistent relaxation it immediately diverged and was slow to reconverge.

It may be something entirely separate, but I found that mine was just a slow converging system, and I upped NELM to a large number and got convergence after ~40 electronic steps.