[Pw_forum] relax question

Paolo Giannozzi giannozz at nest.sns.it
Thu Dec 20 12:42:27 CET 2007

On Dec 20, 2007, at 9:57 , Miguel Marti nez wrote:

> Although 60 Ry looks reasonable, I don't really know.
> However, as ecutwfc is more-or-less system independent,
> you can take a simple Mn system (let's say fcc Mn with
> celldm(1)~14 bohr, no need to use the full I-43m structure)
> and check convergence.

convergence should be perfect with ecutwfc=55Ry, quite
reasonable with ecutwfc=35Ry and ecutrho=300Ry.
Coming back to the original problem:
- there are two paragraphs in the user guide, in section
  "troubleshooting", for self-consistency problems and
   for the "cdiaghg error" problem
- it is very unlikely that the latter is a bug of the code
- mathematical libraries are sometimes a source of
   trouble: please try to replace mkl with internal
   blas/lapack or with atlas
- the cvs version uses a different (and hoperfully better)
   distributed-matrix parallelization algorithm
By the way: I seriously doubt that LDA is the good choice
for Mn-doped GaAs

Paolo Giannozzi, Dept of Physics, University of Udine
via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222

More information about the users mailing list