[Pw_forum] Broken degeneracy using B3LYP

[Lorenzo Paulatto]

On Wed, 15 Dec 2010 23:03:12 +0100, Jeremy McMinis <mcminis2 at illinois.edu>  
wrote:
> I am using the B3LYP functional and notice that degeneracies are being
> broken. Do I need to use a specific diagonalization method or is there
> something else I need to be doing to avoid this problem? I am running
> with sym and inv .false. because I want the k-points to generate
> orbitals. This doesn't effect the LDA calculation.
> [...]
>    nosym=.true.,
>    noinv=.true.,

Dear Jeremy,
I don't think the problem is b3lyp itself, but the just the gradient  
correction: it can be quite noisy and thus break the symmetry of the  
crystal. If you wish the symmetry to be actively enforced you have to  
lease nosym to its default value, namely .false.

best regards


-- 
Lorenzo Paulatt (IdR)
IMPMC - CNRS UMR 7590 & Université P&M Curie
T23-C13/23-5e27 - 4 place Jussieu - 75252 Paris Cedex5
phone: +33 (0)144 27 5211
www:   http://www-int.impmc.upmc.fr/~paulatto/