The ABINIT Users Mailing List ( CLOSED )

[paul.boulanger@umontreal.ca]

Hi Eric,

            I am testing the PAW implementation in Abinit 4.6.4 (and before 
that 4.6.3) and I have observed the same error for all the calculations, not 
just the relaxations. If you make a convergence test with a lot of datasets, 
the calculation will stop, giving a max memory error after a few datasets. It 
seems like there is an internal memory that always grows. 

           I have not located the source of the error. Maybe Marc and 
Francois would know what to do. But this is not the only bizarre behavior I 
have found. It seems the molecular dynamics are a little off, because when I 
relax a single molecule (cobaltocene) in a big box the force does not 
converge. When I visualize the movements of the atoms at each Broyden Step, 
the whole molecule moves in the box. The motion is oscillatory, some 
iteration pushes the molecule out of its initial position and others bring it 
back. I do not understand, because I though that the center of mass was 
strictly conserved in the molecular dynamic part of Abinit. 

        Finally, the last puzzling behavior is in the SCF convergence. The 
energy seems to converge in the wrong direction: the first SCF begins with 
energy lower then the converged value. But the converged value is suppose to 
be the fundamental of our system, hence the minimum. Maybe it is the 
compensation charge that is non-physical for the first iterations or that the 
negative correction of the pseupotential energy inside the augmentation 
region that is to big in the not converged state?? This behavior is apparent 
in the tutorials used in the summer school, but the difference is less 
apparent then in my cobaltocene system. Which explains why I didn’t notice it 
before. Maybe Marc and Francois could comment on that.

  Cheers,

paul