PBE Sm + LAPACK: Routine ZPOTRF failed! 52

[Danny]

Hi,

I'm doing some calculations on substitutions, and when I'm trying to use Sm (PBE functional) then the run crashes even before the first electronic iteration with the error message:

LAPACK: Routine ZPOTRF failed! 52

Up to now more than a dozen systems worked out perfectly without any such error. What has gone wrong here, and how can it be fixed?

I know there is an Sm3 functional...but the basic idea of its existence opposes any usefullness for calculations, so I believe one should not use these type of potentials except in some very specific cases.

I am using VASP 5.2.2 and PBE functionals.

Danny

[admin]

whether or not one has to use PPs that localize the f-electrons localized in the core or not depends on the electronic structure of the compound (to be checked against experiments, for example PES)
if you are sure that your Sm-substitutional atom has the f-electrons in valence, the most probable reason for the LAPACK ZPOTRF error is that the charge density is unreasonable because e.g. the inter-atomic distances are too small (please check the NN-distances of the atoms as written in OUTCAR)

[Danny]

NN are all 2.37A which is perfectly OK.

I do not entirely agree with your argument. One could state that it is necessary to do both calculations (and hope to have agreement with experiment if it exists...if there is no agreement then it's an interesting case to learn and understand what causes it to fail), and decide on the basis of their results(energies) which one to use. This however means that at least both calculations should start...I will even accept that they crash because the energies blow up during the calculation...but not that it simply doesn't even start.

Danny