[Pw_forum] optimization of any structure

[Stefano de Gironcoli]

Dar Sathya,
 Starting from version 4.0 there is an additional option for the 
vc-relaxation that may be worth trying.
Just specify cell_dymanics='bfgs' in the &CELL namelist (and 
ion_dymanics='bfgs' in &IONS) and the optimization will be done with a 
quasi-newton bfgs algorithm.
It has been recently implemented, report to the list in case of problems
best regards,
   stefano
PS: as already noticed by others  etot_conv_thr = 1.0E-10  and  
forc_conv_thr = 1.0D-10 are extremely tight thresholds ...
Start using the default values and then modify them to get your own 
feeling on what is a good threshold for your system...

Prasenjit Ghosh wrote:
> Dear Sathya,
>
> I have no experience in using 'vc-relax'.......so I can't help you out 
> on this. Either you wait for other people's comments or you can search 
> the pw forum mailing list. There were discussions regarding this 
> earlier in this forum. I said its tricky because you need to choose 
> different parameters correctly to reach the minima quickly and 
> correctly. What I would suggest is to play around with different 
> parameters for some small system and understand how they affects the 
> calculation. Later you can go for production run.
>
> Prasenjit.
>
> 2008/7/30 sathya subramanyan <sathya_sheela1985001 at rediffmail.com 
> <mailto:sathya_sheela1985001 at rediffmail.com>>
>
>      
>     Dear Dr. Prasenjit,
>                         Thank you so much for your reply and the link.
>                       I'm doing the vc-relax calculation but it has
>     been running for more than 36 hours.Is this why you have mentioned
>     that vc-relax is tricky ? I have attached the I/P file for
>     reference and if you could please take a look at it.
>                       I'm also trying the manual relaxation and will
>     let you know the status of my results.
>
>     INPUT FILE FOR VC-RELAX
>
>     &CONTROL
>       calculation =  "vc-relax"  ,
>       restart_mode = 'from_scratch' ,
>       outdir='/home/413107002/tmp/' ,
>       pseudo_dir = '/home/413107002/espresso-4.0.1/pseudo' ,
>       nstep =  1000  ,
>                   etot_conv_thr = 1.0E-10  ,
>                   forc_conv_thr = 1.0D-10 ,
>                               dt = 150 ,
>     /
>     &SYSTEM
>       ibrav = 6 ,
>       celldm(1)=7.373 ,
>       celldm(3)=1.065,
>         nat  = 5 ,
>         ntyp  =  3 ,
>         ecutwfc  = 30.0,
>         ecutrho  =  300.0 ,
>         occupations  =  'fixed' ,
>         degauss  =  0.00 ,
>     /
>     &ELECTRONS
>                         conv_thr  =  1.0d-10  ,
>                 diagonalization  =  'david'  , 
>     /
>     &IONS
>       ion_dynamics = 'damp'
>     /
>     &CELL
>       cell_dynamics = 'damp-w' ,
>
>     /
>     ATOMIC_SPECIES
>       Pb  207.2    Pb.vdb.UPF
>       Ti  47.867  Ti.vdb.UPF
>       O    15.9994  O.vdb.UPF
>     ATOMIC_POSITIONS
>     Pb    0.000    0.000    0.000
>     Ti    0.500    0.500    0.538
>     O    0.500    0.000    0.612
>     O    0.000    0.500    0.612
>     O    0.500    0.500    0.117
>     K_POINTS {automatic}
>       4 4 4  1 1 1
>
>     Thanking you,
>
>     SATHYA SHEELA.S
>     Grad. Student
>     Department of Physics
>     National Institute of Technology
>     Tiruchirapalli - 620015
>     India
>
>     HP
>     <http://adworks.rediff.com/cgi-bin/AdWorks/click.cgi/www.rediff.com/signature-home.htm/1050715198@Middle5/2208526_2200903/2203540/1?PARTNER=3&OAS_QUERY=null>
>
>
>
>     _______________________________________________
>     Pw_forum mailing list
>     Pw_forum at pwscf.org <mailto:Pw_forum at pwscf.org>
>     http://www.democritos.it/mailman/listinfo/pw_forum
>
>
>
>
> -- 
> PRASENJIT GHOSH,
> POST-DOC,
> ROOM NO: 265, MAIN BUILDING,
> CM SECTION, ICTP,
> STRADA COSTERIA 11,
> TRIESTE, 34104,
> ITALY
> PHONE: +39 040 2240 369 (O)
> +39 3807528672 (M)
> ------------------------------------------------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>