[CPMD-list] Re: Re:questions on PIMD

[qfzhang]

Hi,
  Thanks for the reply and the advice!I still have some questions in what you ha
ve suggested.
  What do you mean by 'study the Trotter convergence'? Does that mean check the 
convergence with Trotter dimension? Which properties should I check?
  I want to calculate the barrier of the proton trasfer in the oxide using both 
constrained PIMD and constrained CPMD, Must I use centroid PIMD? 
  How can I provide the adiabaticity between centroid and other beads?
How large should I specify facstage with respect to the Trotter dimension? Must 
the EMASS be very small to provide the adiabaticity between other beads and elec
trons? And how smaller should the time step be? Can I use BO PIMD? In that case,
How smaller should the time step be?
  I am sorry I am a beginner at it and maybe ask some foolish questions.
Can you advise some good paper and reference?

  Best wishes
                                                      Zhang Qianfan
                                                       2007-03-09 

Serge Ivanov д:

>     Hi !
> 
>  There are obviosly problems with what you are trying to do.
> 1) The temperature of 600K is not the interesting one for applying PI
>    formalism, since the system behaves classically being heated that much.
> 2) Don't use massive thermostat for electrons. 
> 3) Before applying centroid dynamics, one should study the Trotter
>    convergence for given system using 'regular' PI dynamics.
> 4) Centroid dynamics will require a much smaller time step, than the
>    classical one. One also has to provide two adiabaticity levels, i.e.
>    between centroid and other beads and between other beads and electrons.
>    You can read this and other connected issues in more detail, e.g. in 
>    Marx,Hutter book here:
> 
>    http://www.theochem.ruhr-uni-bochum.de/research/marx/cprev.en.html
> 
> 5) You should always check how stable is your Verlet scheme, at least,
>    by looking at the conserved quantity (9th column in the ENERGIES file
>    for PI). I'm pretty sure that your current runs are very unstable.
> 
> WBR, Serge   
> 
> On Fri, 9 Mar 2007, qfzhang wrote:
> 
>> Date: Fri, 09 Mar 2007 10:13:18 GMT
>> From: qfzhang <qfzhang at aphy.iphy.ac.cn>
>> To: cpmd-list at cpmd.org
>> Subject: [CPMD-list] questions on PIMD
>> 
>> Hi,dear all!
>>   I want to use centroid PIMD in periodic oxide system. and the energy of the


>> w
>> hole system is about 1000 Hatree.But the EKINC/P term in the output file, i.e


>> th
>> e electron kenetic energy, is always very high, about 0.2 Hatree. Even after
>> lon
>> g-time equilibrium, it still will rise rapidly when beginning the PIMD
>> simulatio
>> n.Is there any trick to solve the problem? the input file is as below.
>> 
>> &CPMD
>> PATH INTEGRALS
>> MOLECULAR DYNAMICS CP RESTART WAVEFUNCTION COORDINATES VELOCITIES LATEST
>> TEMPERATURE
>> 600.d0
>> NOSE IONS MASSIVE
>> 600.d0   5000.d0
>> NOSE ELECTRONS MASSIVE
>> 0.012  18000.d0
>> 
>> CONVERGENCE ORBITALS
>> 1.0E-04
>> EMASS
>> 150
>> MAXSTEP
>> 7500 TIMESTEP
>>  2.0
>> TRAJECTORY XYZ SAMPLE
>> 1
>> 
>> &END
>> 
>> &DFT
>> NEWCODE
>> FUNCTIONAL BLYP 
>> &END
>> 
>> &SYSTEM
>> ANGSTROM
>> SYMMETRY
>> 1 CELL
>> 8.52 1.0 1.0  0  0  0
>> CUTOFF
>> 100.0
>> CHARGE
>> 1.0
>> &END
>> 
>> 
>> &ATOMS
>> ........
>> ........
>> CONSTRAINTS
>> FIX STRUCTURE 1
>> DIFFER 23 41 24  -0.7287
>> END CONSTRAINTS             
>> &END
>> 
>> &PIMD
>> CENTROID DYNAMICS
>> TROTTER DIMENSION
>>  8 NORMAL MODES
>>  50.0
>> FACMASS
>>  1.0
>> PRINT LEVEL
>>  1
>> PROCESSOR GROUPS
>> 1
>> &END
>> 
>> 
>> _______________________________________________
>> CPMD-list mailing list
>> CPMD-list at cpmd.org
>> http://cpmd.org/mailman/listinfo/cpmd-list
>> 
>>