Hi Aaron,
Siesta does not stop if the SCF hasn't converged, it stops when
MaxSCFIterations has been reached (or if has converged, of course).
The problem you describe is known, but I can't recommend anything
except modifying the scf parameters until it converges.
Regards,
Eduardo
On 24/07/2007, at 0:58, P E LANDERVILLE III wrote:
Hi Everyone!
I think I'm experiencing some kind of glitch with siesta. Please
correct me if I'm wrong, but I believe that siesta is supposed to
stop the run when the self consistency criteria has not been met in
the allotted number of iscf steps. I've been having self
consistency problems for some time; I've been adjusting the pulay
number and mixing weight to fix this problem. However, if I have
MaxISCFIterations = 100 and convergence is not met, siesta
continues with the next ionic step. Then everything gets really
messy. Also, the ionic step at which scf fails is never the same
from run to run. This was a pain in that I was about to present my
results until I by chance discovered this problem. Anyone else
encounter such a thing? Is there some other option I don't know
about? I've done nothing to the code except change a setting in
the iomd.h file to enable the creation of the MDX data file.
Thanks
Aaron
