[XtalOpt-devel] custom chemical elements

[David Lonie]

Hi Charles,

On Sun, Sep 16, 2012 at 4:24 PM, charles quarra
<charllsnotieneningunputocorreo at gmail.com> wrote:
> The documentation for input files for both XtalOpt and USPEX seems to rely
> on the chemical element name for the involved species. So i'm wondering if
> there is a way to provide custom element properties in some configuration
> files or something. Would the software assumptions fail for this case?
> otherwise, what would you need to do to add positrons and electrons as
> separate elements for the optimization algorithm?

This is possible, so long as the optimization code (e.g. GULP, VASP,
pwSCF, etc) can handle the positron/electron interactions. I've had
XtalOpt users that were analyzing systems made up of non-elemental
"atoms" using GULP and appropriate forcefield parameters. I forget the
exact details, but it can be done. The only assumption XtalOpt makes
about the atoms is the atomic radius, which is used to prevent atoms
from getting too close together, but this can be disabled in the user
interface (minimum interatomic distance settings).

The molecular crystal search implementation makes some extra
assumptions by using a standard forcefield to preoptimize/analyse the
candidates, but the atomistic search will only optionally use the
radius.

The easiest way to accomplish this would be to just pick an element
for each particle, and make sure that the composition in XtalOpt and
the potential parameters in the input file are using the same
element-particle mapping.

Hope this helps,

Dave