no convergence blues with bulk sio2

Axel akoh... at
Sat Aug 4 00:26:43 UTC 2007

hi everybody!

thanks for the suggestions.

using CG, changing the preconditioner,
all helped incrementally, but changing
the basis set to the MOLOPT version was
the ticket. i'll be trying the 'confined'
basis later. at the moment, i'm testing whether
DIIS will work after the system is a bit more
'in balance' after a few MD steps and starting
from a reasonably converged wavefunction.


yep. and works nicely, too. about 4 scf steps.
now i only need to get cp2k working on the new
NCSA cluster without killing the infiniband... :-(

thanks and cheers,

More information about the CP2K-user mailing list