[CPMD-list] experienced problem

[Axel Kohlmeyer]

On 12/17/07, liuzhim <liuzhim at mail.jl.cn> wrote:

>       By the way, I am interested in what we need every day-- the phase of water/ice. Say: The best way to look into a package is to run the examples. The cpmd_test provides a cubic WAT32 example for STANDARD MD simulations.

properties of water/ice have been studied to death with classical
and first principles MD calculations. so my first recommendation is
to look into the existing literature, _PARTICULARLY_ into the papers
discussing the accuracy/benefits/shortcomings of DFT in general
and the supported functionals in CPMD, then also the impact of
the CP method and the choice of fictitious mass on the dynamics
_and_ structure and finally the impact of k-space sampling and
finite size effects.

>      During the trial, I found the PP for H and O are using inconsistent correlation functionals. and the kinetic energy for electrons, Ekin converges to about 0.02.

which pseudopotentials _exactly_?, which functional?

>     My work starts from water /ice phase X with 16 molecules and lattice parameter varies from 5.4 angstrom to 6.2 angstrom to phase VII, at 300K(ambient temperature). When a0 >= 5.7  and =< 6.1, the Ekin oscillates around 1.0 a. u., exhibits convergency. Otherwise it increases to 0.01~0.02 with no oscillate curvatures observed(although it seems to be converged).
>      I wonder if what above mentioned is physically reasonable. Additionally, I have tried to re-optimize wavefunctions every 500 MD steps. Surprisingly, at the second round of 500-Step MD, it increases to ~1.0 and oscillates, but delines to ~0.02 in the next round.

please try to rationalize what happens and how the CP method works.
this is something that _has_ to happen. you start from a converged wavefunction,
with velocities set to zero, i assume you restart the atomic velocities, so
now your atoms move and the electrons cannot follow right away. there are
ways to avoid this peak, but they are very time consuming (they are
_required_ for (near) metallic systems to avoid divergence of the wavefunction).

>     I am looking forward to your advices. the main part of my CPMD script is as what follows.
>                                 Thanks a lot
>                                        Z. M. Liu
>                                         liuzhim at mail.jl.cn
>                                            Dec. 17, 2007
> &CPMD
>       OPTIMIZE WAVEFUNCTION/MOLECULAR DYNAMICS cp
>       TIMESTEP
>       4
>       MAXSTEP
>       500
>      EMASS
>      1100.

you are missing quite a few important input parameters, so it is
hard to judge the adequacy of your flags this far.

as noted in previous discussions. this is a huge electron mass
for a system containing hydrogens. please see above.

>      TEMPCONTROL IONS
>      300.0 70.0
>      TRAJECTORY XYZ
>      TEMPERATURE ELECTRONS
>      1000.0


this has _no_ effect unless you use the free energy functional.
why do you put it here? are you using FEMD?

cheers,
    axel.

>      &END
>
>
>
> ----- Original Message -----
> From: "Ari P Seitsonen" <ari.p.seitsonen at iki.fi>
> To: "liuzhim" <liuzhim at mail.jl.cn>
> Cc: <cpmd-list at cpmd.org>
> Sent: Sunday, December 16, 2007 10:17 PM
> Subject: Re: [CPMD-list] HELP: Explaination to manual 9.5.2
>
>
> >
> >
> > Dear Mme/Mlle/M Liu,
> >
> >   That sentence ONLY applies to calculations in the electronically excited
> > state (employing TDDFT, that chapter is meant to be read only when using
> > TDDFT). However, it is always preferable to perform an optimisation of the
> > wave functions before a molecular dynamics run when running the
> > Car-Parrinello MD, it is a common source of error to try start directly
> > with the dynamics.
> >
> >     Greetings from Strasbourg,
> >
> >        apsi
> >
> > PS Which version of the code are you using?
> >
> > On Sun, 16 Dec 2007, liuzhim wrote:
> >
> >> Hi, every body:
> >>   I have a question to the manual for CPMD, and hope somebody may give help.
> >>    I want to handle some finete temperature simulations via CPMD.  All
> >> the examles were tried. Recently, when I read the manual carefully, the
> >> sentence  "geometry optimizations and molecular dynamics simulations can
> >> only be performed after an electronic spectra calculation" in section
> >> 9.5.2 makes me much confused. Does it mean that I should calculate the
> >> electronic spectra even though I just want to perform a dynamical
> >> simulation  to investigate the mechanical or thermal properties for ions
> >> while the electronic properties are neglected? If I do not do the
> >> spectra work, what will occur to the results?
> >>                                                        Z. M. Liu from , Changchun, P.R.C
> >>                                                         Dec. 16, 2007
> >>                                                        liuzhim at mail.jl.cn
> >>                                                                 .
> >> _______________________________________________
> >> CPMD-list mailing list
> >> CPMD-list at cpmd.org
> >> http://cpmd.org/mailman/listinfo/cpmd-list
> >>
> >
> > --
> > -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
> >   Ari P Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
> >   IMPMC, CNRS & Université Pierre et Marie Curie
> >   Tel: +33-1-4427 7542, Fax: +33-1-4427 3785, GSM: +33-6-6736 3820
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://cpmd.org/mailman/listinfo/cpmd-list
>
>


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