[CPMD-list] Is steepest-descent for electronic degrees of freedom coupled with those for ions?

[Eduard Schreiner]

Masroor Ahmad wrote:
> Hi List,
> I've this apparently curious behaviour:- Using (STEEPEST DESCENT IONS)
> unexpectedly speeds up the wave-function convergence, despite never
> specifying ELECTRONS on the same line. If I switch my geometry update
> method to some other (e.g RFO or default GDIIS), wave-function
> convergence is dramatically changed for worse.( altough without
> changing anything for minimizing the energy gradient).
> I'll be quite happy if STEEPEST DESCENT works for my geometry updates
> and simultaneously smoothens wave-function calculation. But I want to
> understand, if this option is really coupled for both ionic and
> electronic cases?  
> Thanks, with regards.

hey,
in the code there are indeed some special things if one is specifying SD 
for electrons AND ions. I have no idea how it works precisely, but the 
subsequent calls differ from the "standard" geometry optimization if the 
SD-Algorithm is defined for ions and electrons.


Eddi

-- 

--

=======================================================================
Eduard Schreiner                        e-mail: eduard.schreiner at rub.de
Lehrstuhl fuer Theoretische Chemie          Phone: ++49 (0)234/32-22121
Ruhr-Universitaet Bochum - NC 03/52         Fax:   ++49 (0)234/32-14045
D-44780 Bochum                               http://www.theochem.rub.de
=======================================================================