For the testing purposes of another program that uses the output of wannier90 I'm trying to run a Gamma centered Monkhorst-Pack k-sampled calculation and then run the equivalent Gamma sampled supercell calculation. My first test system was a loosely bound H2 chain, which behaved as expected. The scf/nscf/wannier90 (using QEv6.5) all matched exactly for the k-sampled and explicit supercell calculations in Bloch orbital energy spectrum and Wannier function spreads and centers. Then I attempted to do the same for silicon, and have been unable to get the same results. The scf/nscf Bloch orbital energy spectrums agree as expected, but the supercell calculation does not localize correctly. During the descent phase the SPRD jumps around wildly and does not nicely descend like it does in the k-sampled case. The initial guesses I have tried are use_bloch_phases=.true., and SCDM isolated since I'm only using the occupied orbitals. I have also tried using the gamma branch algorithms for the supercell and surprisingly saw that the SPRD does not always decrease, which makes no sense to me from my understanding of how the 2x2 rotation algorithm works. I don't know if I'm doing something wrong or if this is expected behavior? Any help would be appreciated. Attached are the input files I used for the silicon calculation with an mp_grid = 4 4 4 and the equivalent unfolded supercell. Cheers, -Aaron Mahler Duke University
PS: I'm using these pseudos: https://github.com/pipidog/ONCVPSP
<<attachment: si-inputs-k4.zip>>
<<attachment: si-inputs-sc4.zip>>
_______________________________________________ Wannier mailing list Wannier@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/wannier
