[CPMD-list] Energy components in PIMD

[Axel Kohlmeyer]

On Fri, 9 May 2008, ukasz Walewski wrote:

LW> 
LW> I did not write this but the rest of the plots (energy components) are
LW> in the same directory, i.e.
LW> http://bioexploratorium.pl/ljw/cpmd/nvt/test1/ 

yes, but i cannot see them. your index.html doesn't list them.

LW> > looks as if your choice of reference frequency for the
LW> > ionic nose-hoover chains does not couple well to the
LW> > prevalent modes in your system (how did you determine it?).
LW> 
LW> 181 cm^-1 is one of the vibrational eigenmodes calculated with
LW> VIBRATIONAL ANALYSIS option. It was proved to promote the proton
LW> transfer (by making the N-N distance shorter) so I want to have this
LW> mode excited.

that would be for production, but first you have to equilibrate
and there you are much better off by using some other settings.

[...]

LW> I see. Then I wonder what is the origin of that, i.e. is it the
LW> specificity of my system that makes USPPs unusable (and if yes, where
LW> does it come from) ? How is it connected with the implementation in
LW> CPMD ?

the USPP support in cpmd is based on a very old implementation of
the USPP code and has not been seriously maintained or updated
(other than to keep it running). it seems to do reasonably well for
bulk systems and geometry optimizations. the fact that you don't 
have a working dual grid support (or box grids for the augmentation
charges) is a severe limitation. also NLCC supports seems a bit 
shaky (doesn't apply to your system). stress tensor and higher
angular momentum is does not work at all.

i just made a benchmark run on a different system (with dual=8) and 
compared to the cp.x code from quantum espresso. the cpmd uspp 
implementation is basically a factor of two slower. OTOH, cp.x is
missing PI-MD support (and quite a few other features that CPMD has).

LW> Could you please post your input that conserves the fictitious kinetic
LW> energy ? On what number of processors did you run it ?

will do later, when i'm in the office and have checked a couple more
things. i've been running on varying number (12-40) pentium-3 
processors (it is a very old machine. the fast machines in our group
are all fully used...).

cheers,
    axel.

LW> 
LW> Regards,
LW> Lukasz
LW> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.