Dear Igor, I looked into this subtle problem and came to the following conclusions:
a) There is nothing wrong with the code. Calculations in i) I4/mmm (B-lattic, a a c), ii) Cccm (doubled unit cell, CXY, a*sqrt2,c,a*sqrt2) and the iii) Pma2 (primitive (again doubled) supercell, a+sqrt2,a*sqrt2,c)yield total energies of E, 2E and 4E within microRy precision, identical charges, ....
b) with respect to band structures there seems to be a subtle difficulty in interpreting the back-folding, which occurs in these cases.
I attach two plots for only a small (not over crowded) E-window to make the effects more clear. i) The first plot in the smallest cell (I4/mmm) shows the bands from Gamma-X, but shown actually from Gamma-AB(1/4,1/4,0) and X(labelled 2AB)(1/2,1/2,0)-AB(1/4,1/4,0), because this is what you will see when backfolding the BZ in the larger cell.
ii) The other plot in Cmmc (doubled cell, rotated) shows the bands from C (0,0,1/2) - Gamma - A(1/2,0,0) - 2A(1,0,0). Obviously, the results of C-Gamma agree with the backfolded bands of i) (after overlaying them and mirroring). On the other hand, out of the 4 bands C-Gamma, only the 1st and 4th of Gamma-A agree, while the other two disagree. However, if I continue in this direction up to 2A, I can find in A-2A the 2 "missing bands". So actually the C-Gamma direction is equivalent to Gamma-2A, but the latter also contains some other backfolded direction of the original I4/mmm cell (which I'm too lazy to identify ...)
So obviously, backfolding in these various centered cells is more difficult, but for sure consistently included.
Best regards Peter On 06/27/2017 08:31 PM, mazin wrote:
Hi, I am running a pretty standard calculation for the BaFe2As2 superconductor; have done zillions of those before. For testing purposes, I wanted to plot bands for the double-cell AF structure (.struct attached), but for the nonmagnetic case and, again, I do think I did the same before with no problems. Now I am using the 14.2, and it may be that I had used other versions before. Without magnetism, this structure is equivalent to I4/mmm (and WIEN2k does find this symmetry, if both Fe are made equivalent), so the resulting band must, up to numerical noise, possess full tetragonal symmetry. Yet, the bands along Gamm-X and Gamm-Y are noticeably different (note that in this setting X=(1,0,0), but Y=(0,0,1)). If I use the conventional cell (doubled, Pccm), the resulting bands have proper symmetry - even though the underlying unit cell is orthorhombic. So, the problem appears if the cell is base-centered. Thanks! _______________________________________________ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
-- P.Blaha -------------------------------------------------------------------------- Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-1-58801-165300 FAX: +43-1-58801-165982 Email: bl...@theochem.tuwien.ac.at WIEN2k: http://www.wien2k.at WWW: http://www.imc.tuwien.ac.at/TC_Blaha --------------------------------------------------------------------------
RUN1.spaghetti_ps.gz
Description: GNU Zip compressed data
tetra.spaghetti_ps.gz
Description: GNU Zip compressed data
_______________________________________________ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html