[Pw_forum] vc-relax with BFGS
sclauzer at sissa.it
Mon Feb 14 15:18:15 CET 2011
Il giorno 14/feb/2011, alle ore 14.18, Srijan Kumar Saha ha scritto:
> > Where is the proof that the final symmetry is different from the initial one? <
> Here goes more direct proof:
> Initial symmetry : 48 Sym.Ops. (with inversion).
> Final symmetry : 6 Sym.Ops. (no inversion).
Where did you get this information from? Was it printed by the code? I never saw a printout of the "Final symmetry", but maybe I am wrong and the code computes and prints it. Which version of the code are you using?
> > I think that you should not pay attention to differences of the order of 10^-9, <
> If we compare the final & initial positions given in my previous email, then we can see that
> the differences are quite large.
From your previous email it looks to me that the atoms are still in the original arrangement with cubic symmetry. I can guess that the volume of the cell has increased a bit (less than 1%) and so the atomic positions, which are always expressed in terms of the INITIAL celldm(1), will scale accordingly on the final output. The largest deviations I can spot out are below 10^-8.
Moreover, I don't understand why you want to perform such kind of expensive calculation on a supercell. Have you tried with the fundamental unit cell first?
> These differences go away if we use
> Wentzcovitch Damped Dynamics, instead of BFGS. However, Damped Dynamics stops
> without convergence (50 iterations completed, stopping). Since this dynamics converge very
> slowly, I would like to know whether VC-RELAX with BFGS has the above capability of respecting the
> Thanks & regards,
> Srijan K.
> Indian Institute of Science
> India 560012
> Pw_forum mailing list
> Pw_forum at pwscf.org
§ Gabriele Sclauzero, EPFL SB ITP CSEA
PH H2 462, Station 3, CH-1015 Lausanne
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the users