My Community

Dear all,
I have met the next problem with relaxation: when I relax structure with magnetic impurity (Mn) using PAW potentials, I did not get self-consistence at all, but when I use US-PP, the structure relaxed properly. What can be the reason of such behaviour and should I include any special options, when I use PAW instead of US-PP?

Usually, PAW PPs are even better than the US-ones...
maybe it is necessary to decrease the (magnetic) mixing parameters
until a certain pre-convergence is reached?
have a look at the rms(c) column of CONTCAR to check