Converge problem with nanoparticle calculation

[WT S]

Dear all CP2K developers and users,

First of all, I am a totally new user of CP2K.

I am trying to use the CP2K to do some geometry and energy calculations on small cobalt nanoparticles, the first one I started is the smallest icosahedron with 13 atoms. However, I just cannot get my calculation converged, the best situation I got now is the energy change stuck around 1E-01. Please see my input files, I am not sure where is the problem.

Another problem is that my calculation uses too many memory, it usually requires more than 130g memory to run, not sure whether it is a problem or not.

Best

WT S

[Sam Broderick]

Hi WT S

While 13 atoms doesn't sound like a lot, many transition elements have a lot of valence electrons and higher order obrital shapes. As a result, your matrices are very large due to the number of basis functions needed.

To be quite honest, I am having trouble with my calculations, too. Your &MGRID CUTOFF seems low...

Kind Regards

Sam

[Sam Broderick]

Quick update. I have been using Davidson, which has helped a bit. Also consider &XC_GRID with XC_SMOOTH_RHO and XC_DERIV.

I am trying to figure out if &VDW_POTENTIAL would help, too.