Dear Nicolas:
As you say, for a meaningful comparison
you should keep everything constant
(geometry, mesh cutoff, number of k-points,
pseudopotentials, etc.) during your variational
minimization of the energy or enthalpy
as a function of the basis set parameters.
Usually, what I have done is to choose a sensible
reference geometry with respect the problem
you are dealing with.
If the size of the basis is large enough,
then it would have flexibility to adapt to different
environments, and the transferability
(property of the basis to keep their performance
when transferred to different systems than the ones
they were optimized for) should be sufficiently good.
Hope this helps,
Javier
Hello,
I'm trying to use the simplex algorithm to optimize a basis for my
systems elements. I'm wondering whether I should do this on a fixed
geometrical structure for every tested node of the basis set or let
the system relax for each basis set tested. To do a meaningful
comparison between two basises, I would argue for the former case.
However, the latter may be better as the total energy of the system
(which is minimized by the simplex algorithm) may vary strongly if the
system is not relaxed.
Any thoughts on this?
Thanks in advance,
Nicolas Leconte
--
-------------------------------------------------------------------------
Javier Junquera
Ciencias de la Tierra y Fisica de la Materia Condensada (CITIMAC)
Avenida de Los Castros s/n
Facultad de Ciencias
Universidad de Cantabria
E-39005 Santander, Cantabria
Spain
web-page: http://personales.unican.es/junqueraj
e-mail: [email protected]
Tel: (+34) 942 20 15 16
Fax: (+34) 942 20 14 02
