[CPMD-list] No convergence in wave functions optimization.

[Ari P Seitsonen]

Dear 'kdong_bj', (Sorry, I don't know your name)

   There are several possibilities:

  - Something wrong in the pseudo potentials; I cannot much about this, as 
you didn't provide any input

  - Something wrong in the geometry

  - Wrong multiplicity or charge; nearly-degenerate ground state

   A problem in the geometry is the usual cause; therefore I would first

  1) run the job with the default algorithm for the electrons (ie. 
commenting out 'PCG MINIMIZE' and see how far the convergence goes

  2) Set a value just slightly larger than this for 'CONVERGENCE ELECTRONS' 
and run a couple of steps of MD, with 'TEMPCONTROL IONS', keeping them at 
reasonable temperatures

   If the value that you reach by the point 1) is larger than 1e-4 I guess 
that there is a problem in the geometry (or there is a physical 
degeneracy, ie. not accidental), and that would show up in point 2) as 
very large forces on the ions, thus the temperature wanting to increase a 
lot (to thousands of degrees).

   A practical way to optimise geometries for such big systems is to start 
like above, but use first 'TEMPCONTROL IONS' and 'TEMPCONTROL ELECTRONS' 
with quite low target/average values and relatively small threshould for 
rescaling, and when this is stable, using simulated annealing ('ANNEALING 
IONS'). Or, if you are planning to run molecular dynamics, stopping at 
some point the rescaling(s) and switching to real/physical dynamics.

   Just my thoughts.

     Greetings,

        apsi

On Mon, 21 Jan 2008, kdong_bj at sina.com wrote:

> Dear CPMD Everyone,
>  
> now I am carrying out wavefunctions optimization, the maxstep was set up to 400, the convergence citerion is 1.0-10(-6), but the
>  
> GEMAX in output still oscillates in 1.0-10(-3) and 5.0-10(-4).
>  
> I have modified the parameters (the cutoff, the emass, the timestep ) for many times, but the convergence is very difficult
>  
> to complete.
>  
> Now I do not the reasons, my system includes 208 atoms.
>  
> my input is as follows:
>  
> &CPMD
OPTIMIZE WAVEFUNCTION
PCG MINIMIZE
TIMESTEP
10
CONVERGENCE ORBITALS
1.0d-6
RHOOUT
ELECTROSTATIC POTENTIAL
ELF
EMASS
600
MAXSTEP
400
&END
 
&DFT
NEWCODE
FUNCTIONAL BLYP
&END
 
&SYSTEM
ANGSTROM
SYMMETRY
1
CELL
13.350 1.0 1.0 0 0 0
CUTOFF
60.0
&END
> .....................................................................
>  
> The atomic species include C, H, N, F, P.
>  

-------------------------------------------------------------------
ÖÇÄÜÖ÷¶¯·ÀÓù£¬Ê·Éϳ¬ÓÅ»Ýɱ¶¾Èí¼þ( http://d1.sina.com.cn/sina/limeng3/mail_zhuiyu/2008/mail_zhuiyu_20080121.html )

===================================================================
»¹Ã»Âòµ½Æ±Âð£¿ÐÂÀËƱÎñ»¥Öúƽ̨£¬°ïÄã¸ã¶¨£¡£¨http://bbs.bj.sina.com.cn/tableforum/App/index.php?bbsid=143&subid=0&ismain=1£©

-- 
-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
   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