Dear Sai, your figures show that the variations are actually very small and possibly within the numerical errors these quantities are affected by. I would certainly try to use larger ranges for the perturbation (e.g., from -0.1 to 0.1).
best, Matteo On Wed, May 7, 2014 at 5:07 PM, Sai Kumar Ramadugu <sramadugu at gmail.com>wrote: > Dear Pietro: > > My calculations use a unit cell of Fe2O3. Currently I am trying to > optimize 2x1 and 2x2 supercells. Once I have the optimized structure, I > want redo the linear response calculations. > > Dear Matteo: > > The deviations are smaller. For the case of intersection at 0.0, I have > very small range from 0.00001 to 0.00025 for the difference between the > occupations at alpha=0.0 for bare and scf converged responses. > > About linearity, I have attached two figures for the response of atom 5 > (Fe) with respect to the perturbed atom (Fe) at the origin and atom 6 (Fe) > with respect to the perturbed atom (Fe) at the origin. > The alpha values for these two graphs are -0.01, 0.0 and +0.01. > The response of atom 5 with perturbed atom does intersect (not at 0.0) but > for the response of atom 6, the two lines do not intersect. > > Would you suggest to use larger values of alpha (close to 0.1) or use a > super-cell for the calculations? > > Thanks very much for the time and help. > Sai > > > On Wed, May 7, 2014 at 7:59 AM, Matteo Cococcioni <matteo at umn.edu> wrote: > >> Dear Sai Kumar Ramadug, >> >> just another idea: how wide are the deviations from linearity? on-site >> occupations are difficult to converge beyond the 3rd digit. Sometimes, when >> they are minimally affected by the perturbation, their variation can be >> within numerical noise. >> >> regards, >> >> Matteo >> >> >> >> >> On Wed, May 7, 2014 at 2:44 PM, Pietro Bonfa' <pietro.bonfa at >> fis.unipr.it>wrote: >> >>> Dear Sai Kumar Ramadug, >>> >>> I can comment only on the sentence below: >>> >>> On 05/06/2014 05:47 PM, Sai Kumar Ramadugu wrote: >>> > The problem I am having as of now is that some of the responses dn0/da >>> > and dn/da are not linear and do not intersect at 0. >>> >>> It happened to me as well. In my case I think that it was due to the >>> size of my supercell (i.e. to the spurious perturbation introduced by >>> the PBC). With a bigger supercell I obtained reasonable results. >>> >>> Kind regards, >>> Pietro Bonfa' >>> >>> -- >>> Pietro Bonfa' - PhD student >>> Dipartimento di Fisica e Scienze della Terra "Macedonio Melloni" >>> Viale delle Scienze 7A >>> 43124 Parma - Italy >>> _______________________________________________ >>> Pw_forum mailing list >>> Pw_forum at pwscf.org >>> http://pwscf.org/mailman/listinfo/pw_forum >>> >> >> >> _______________________________________________ >> Pw_forum mailing list >> Pw_forum at pwscf.org >> http://pwscf.org/mailman/listinfo/pw_forum >> > > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://pwscf.org/mailman/listinfo/pw_forum > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140508/7c627b30/attachment.html
