[Pw_forum] CP simulation with ensemble-dft algorithm

Nicola Marzari nicola.marzari at materials.ox.ac.uk
Sun Jun 19 18:24:28 CEST 2011




Dear All,


just to clear up a few things - the ensemble-DFT algorithm coded in the
CP code is a direct minimization algorithm, to do Born-Oppenheimer
molecular dynamics in metallic systems, and not to do Car-Parrinello
molecular dynamics (this latter breaks down in the absence of a gap).

Hence, the ion dynamics parameters are typical of BOMD (large ionic
steps, of the order of 0.5 to 5 fs, i.e. 20 to 200 a.u.), while for the
electrons there is no fictitious electronic kinetic energy (ekinc).

Although the code has been used very little, it does seem to work
(actually very well - we just checked it last week for some tungsten
clusters, that were impossible to converge with another of the popular
codes, and it did get convergence in ~40 conjugate gradient steps).

Pascal, let us know if you encounter problems.

			nicola




On 6/19/11 9:21 AM, pascal boulet wrote:
>
> Dear Giovanni,
>
> Humm... hard to say because the data are not printed in the output file.
> Here is what I get, for instance:
>
>>>>>
>    nfi    ekinc  temph  tempp        etot      enthal       econs
> econt    vnhh   xnhh0    vnhp   xnhp0
>          nfi   tempp            E -T.S-mu.nbsp         +K_p #Iter
> Step    20       0   -502.49200   -502.62143   -502.62143     6
>   cg_sub: missed minimum, case 1, iteration           1
>   cg_sub: missed minimum, case 1, iteration           2
>   cg_sub: missed minimum, case 1, iteration           3
>   cg_sub: missed minimum, case 1, iteration           4
>   cg_sub: missed minimum, case 1, iteration           5
> NOTE: eigenvalues are not computed without ortho
> Occupations  :
> 2.000000 2.000000 2.000000 2.000000 2.000000 2.000000 1.999999 1.999999
> 1.999999 1.999998
>   1.999997 1.999995 1.999994 1.999989 1.999983 1.999977 1.999958 1.999933
> 1.999884 1.999793 ... 0.001473 0.000546 0.000502 0.000278 0.000134
> 0.000079 0.000033 0.000019 0.000000
>          nfi   tempp            E -T.S-mu.nbsp         +K_p #Iter
> Step    21       0   -502.49200   -502.62143   -502.62143     6
>
> <<<<<
> etc.
>
>
> So, there is nothing about ekinc, temph, tempp, ...
>
>
> However, the data "nfi    ekinc  temph  tempp        etot     enthal
>     econs       econt    vnhh   xnhh0    vnhp   xnhp0" look
> like those given in the file xxx.evp (except for one missing column).
> If this is right, then in this file ekinc = 0. during the whole
> simulation. That's strange!
> So, I guess there is something wrong somewhere in my input.
>
> Pascal
>
>
>> Date: Fri, 17 Jun 2011 10:27:09 +0200 (CEST)
>> From: Giovanni La Penna<glapenna at iccom.cnr.it>
>> Subject: Re: [Pw_forum] CP simulation with ensemble-dft algorithm
>> To: PWSCF Forum<pw_forum at pwscf.org>
>> Message-ID:<alpine.LNX.2.00.1106171015110.8707 at biurca.iccom.cnr.it>
>> Content-Type: TEXT/PLAIN; format=flowed; charset=US-ASCII
>>
>>
>> Dear Pascal,
>>
>> The input shows that you are trying to
>> carry the system to 627 K starting with
>> random velocities and a time-step of 20 au (about 0.5 fs).
>> I expect a huge Kel after a few time-steps, due
>> to huge current atomic velocities, not manageable
>> with the given time-step.
>> Please, tell me (us) if this won't happen.
>>
>>                Giovanni La Penna
>>
>>
>
>      *
>      * Anglais
>      * Français
>
>      * Anglais
>      * Français
>
>   <javascript:void(0);>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum


-- 
----------------------------------------------------------------------
Prof Nicola Marzari    Department of Materials    University of Oxford
Chair of Materials Modelling  Director, Materials Modelling Laboratory
nicola.marzari at materials.ox.ac.uk     http://mml.materials.ox.ac.uk/NM



More information about the users mailing list