[Pw_forum] Problem in Ionic relaxation and MD runs

[Axel Kohlmeyer]

On Wed, 28 May 2008, Holger Hesske wrote:

HH> Dear pwscf-users, -developers and other readers,

holger,


[...]

HH> Maybe with the answer of the question above the next part might be unimportant 
HH> or useless. But: Starting md's from ideal atomic positions (with and without 
HH> temp rescaling, dt ={2, 5, 10, 15, 20}) it always ended up finally with pairs 
HH> of BaO having a interatomic distance below 1.3 Angstrom! I guess it is the 
HH> same problem...  
HH> 
HH> Thanks for reading the complete text and in advance for every hint I get.

could be you have a crappy Ba potential. is that one using the 
original data from the vanderbilt site? that is one of the dacapo
parameter sets, right? you may need to "tune" the potential fit
parameters to get something reasonable. at least that is my experience
with most of those parameter sets. if you run the pseudopotential
generation code on them, they usually have large "negative densities"
and one has to correct for those by adding constraints to the 
potential fitting (or generate a better potential from scratch).

cheers,
    axel.

HH>    
HH> best regards
HH> 
HH> Holger Hesske 
HH>  
HH> PS. I could provide a lot of input and output files but i dont want to waste 
HH> the time of the readers, in case I made a simple (stupid) mistake in setting 
HH> up the calculations.
HH> 
HH> 

-- 
=======================================================================
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.