Usually negative charges tend to disappear, not to increase, at large cutoff, but it is hard to say more.
P. On Nov 8, 2012, at 15:37 , Lat?vi Max LAWSON DAKU wrote: > Dear QE users and developers, > > > I am working on a large system, which I would like to characterize > in two spin states. The unit cell is made of 324 atoms and there are > 1220 electrons. The system is made of transition metal complexes > in a polymeric framework; and I am using the PBE GGA and USSPs > with nonlinear core corrections, a few of them which I have generated > and tested as much as possible. > > To proceed, I have taken in the USPPs the suggested largest cutoffs, > which I have raised to WFNc=70Ry and RHOc=750Ry, for the kinetic > energy and charge density, respectively. > I ran optimization calculations (vc-relax) with the total > magnetization > fixed to the desired values and the Gamma point only. The results > obtained for the structures in both spin states seem fine and > consistent > with available experimental data for similar and isostructural > systems. > To be sure about the choice made for the k-point sampling, I have > restarted the optimizations using a shifted 4x4x4 MP grid (actually > only 4 k-points): this did not change the results; this seems fine. > So I > felt quite confident. > > As a further test, and before moving to more demanding phonon > calculations, I increased the WFNc to 80Ry and RHOc to 950 Ry: > the cell parameters slightly increases by less than 1%, and the energy > difference between the two phases changes by less than > 2% (total energies - quite large values - vary by less than 0.01%). > However, for these cutoff values, a negative charge (rho_dwn) > appears in one case, with a final value of the order of 1e-4: > > Total force = 0.006379 Total SCF correction = > 0.000024 > > entering subroutine stress ... > > negative rho (up, down): 0.000E+00 0.302E-03 > > which, if I correctly understand the FAQ entry 6.12, is not harmful > at all. > I am however worried by the fact that this charge was not produced > in the calculations with the lower cutoff values. Did one of you > already > observe this? Should this be taken as a symptom of something wrong > in the calculation setups (USPPs, or WFNc and RHOc values)? > > Thanks for your time, answers and comments. > > Best regards, > Max > > > -- > Latevi Max LAWSON DAKU > Universite de Geneve - Sciences II > Quai Ernest-Ansermet 30 Tel: (41) 22/379 > 6548 > CH-1211 Geneve 4 Fax: (41) 22/379 > 6103 > > > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://pwscf.org/mailman/listinfo/pw_forum --- Paolo Giannozzi, Dept of Chemistry&Physics&Environment, Univ. Udine, via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
