[CPMD-list] error in MD calculations using Monkhorst Pack K-points

[Axel Kohlmeyer]

On Wed, 11 Oct 2006, Ari P Seitsonen wrote:

AS> Dear Fafa,
AS> 
AS>    Three ways to go around:

dear fafa,

actually, there is a fourth way (i may get excommunicated here 
  from the more fundamental believers in the 'the one true gospel
  of the CPMD', but since there is an increasing amount of people
  getting frustrated by 'undocumented features' like this, i think
  it is valid to state it): 

  don't force cpmd into doing something where it has not
  been used (and optimized) for a lot. ari's suggestion of
  implementing the features you need, is a solution. but
  there are other plane wave pseudopotential codes around,
  that are very well suited for problems like this. i would
  recommend trying out the quantum espresso package first,
  because it is a very good complement to the cpmd capabilities.
  e.g. it has very good support for k-points (_with_ ultra-soft
  pseudopotentials) including parallelization across k-points, 
  metals/semiconducters and commonly used optimization analysis
  methods (e.g. NEB, pDOS, bandstructure, phonons + dispersion
  to name a few). it is however lacking the sophistication 
  (thermostats, constraints, etc.) and extraordinary scaling 
  (on the right hardware ;-) ) for CP-dynamics that CPMD has. 

  in fact several people working on the QE package used to work
  on and with cpmd at times and the codes share a common ground.
  there are a bunch of other (commercial and free) packages
  as well that may or may not be more suitable for what you need.

i'll be putting on my asbestos underwear now, 
awaiting to get seriously flamed or worse. ;-)

best regards,
    axel.

AS> 
AS>     - One implements the Car-Parrinello scheme also for the case of k
AS>      points; would not be too difficult, however if you have almost
AS>      degenerate states and band structure effects one could get into
AS>      troubles with the occupation numbers; here I have to mention that
AS>      there are schemes for doing Car-Parrinello simulations with fractional
AS>      occupation numbers, yet they are not implemented in CPMD
AS> 
AS>     - You perform Born-Oppenheimer molecular dynamics instead of
AS>      Car-Parrinello dynamics
AS> 
AS>     - You take such a big unit cell that the Gamma point is sufficient;
AS>      might be, however, very heavy option CPU wise
AS> 
AS>      Greetings,
AS> 
AS>         apsi
AS> 
AS> PS I wanted to answer your previous mail concerning the physics of these
AS>    systems (I used to do defects in semi-conductors in my wild youth) where
AS>    Axel didn't already clarify the point, yet I haven't found time in the
AS>    last weeks, sorry!
AS> 
AS> -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
AS>    Ari P Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
AS>    CNRS & IMPMC, Université Pierre et Marie Curie
AS>    Tel: +33-1-4427 7542, Fax: +33-1-4427 3785, GSM: +33-6-6736 3820
AS> 
AS> On Wed, 11 Oct 2006, fafa chiker wrote:
AS> 
AS> >
AS> > Hi,
AS> >
AS> > Trying a MD calculations. I got this error:
AS> >  PROGRAM STOPS IN SUBROUTINE MDPT| K-POINTS NOT IMPLEMENTED [PROC=   8]. 
AS> > Does anyone know of a way of getting around this? 
AS> >
AS> > There are several considerations to take about the K-points in relaxing 
AS> > the system????
AS> >
AS> > Thanks in advance.
AS> >

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