Dear all For this simple case of extended in the XY plane system the procedure of 2d basis reduction controlled by parameters ewind and epsproj should not be very efficient, in fact it is better if you use the full 2d basis, which is in this case n2d=271... You can do that specifying very big value for ewind, say ewind=400. I would also reduce delgep parameter to say delgep=5.d-10. It needs to be set to bigger values if the code crashes... good luck, Alex
2014-07-18 16:39 GMT+02:00 Sclauzero Gabriele < gabriele.sclauzero at mat.ethz.ch>: > Dear Vladislav, > > I mostly agree with what Nicki said here below. I also believe that in > your case the problem is the ewind value, which is too small and can lead > to some weirdnesses in the 2D-basis set reduction. You should try at least > 4.0 and check convergence as suggested by Nicki. > In my experience, the choice of epsproj depends a lot on the system. Even > values around 1.d-5 could assure a decent convergence, unless the basis set > reduction went bananas (e.g., because of a too small ewind). > Also, if you push epsproj very low, but you keep ewind very small, then > you could also get into trouble. > In normal calculations, I never found the need to increase the nz1 from > its default. > > Finally, a word about the crystal structure. Are you sure that?s what you > want? The neighbors of the Co atoms seem to have different bond distances. > Maybe you need to choose another value for celldm(3)? > > HTH > > > GS > > > > > > Il giorno 18/lug/2014, alle ore 12:53, nicvok <nicvok at freenet.de> ha > scritto: > > > The parameters epsproj and ewind control the amount of basis functions > > used in the underlying complex > > bandstructure (CBS) calculation in your case 121; cf. Choi&Ihm ( > http://journals.aps.org/prb/pdf/10.1103/PhysRevB.59.2267 > > ) and > > DalCorso et al. ( > http://journals.aps.org/prb/abstract/10.1103/PhysRevB.74.045429 > > ). So one should take a > > look at the CBS before and check for reasonableness and convergence. > > Then adapt step by step epsproj and ewind (even more nz1) and check > > for convergence of the CBS. > > In your case ewind=2 might be okay, nevertheless I would slightly > > increase this (3-5). Did you ever tried > > some meaningful epsproj< 1.d-7, which should be a good value? The > > number of integral points with 22 seems to > > be okay. That it worked for other systems could be by accident. Again, > > as always, one finds the optimal > > value of the parameters, by checking for convergence of the underlying > > physics. > > > > bests Nicki > > > > ------------------------------------------------------------- > > Nicki Frank Hinsche, Dr. rer. nat. > > Institute of physics - Theoretical physics, > > Martin-Luther-University Halle-Wittenberg, > > Von-Seckendorff-Platz 1, Room 1.07 > > D-06120 Halle/Saale, Germany > > Tel.: ++49 345 5525460 > > ------------------------------------------------------------- > > _______________________________________________ > > Pw_forum mailing list > > Pw_forum at pwscf.org > > http://pwscf.org/mailman/listinfo/pw_forum > > > Dr. Gabriele Sclauzero > Materials Theory - ETHZ > ETH Zurich, HIT G 43.2 > Wolfgang-Pauli-Str. 27 > 8093 Z?rich, Switzerland > > Phone +41 44 633 94 10 > Fax +41 44 633 14 59 > gabriele.sclauzero at mat.ethz.ch > www.theory.mat.ethz.ch > > > _______________________________________________ > 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/20140718/8a1bfa77/attachment.html
