Dear Nathan,
I am unsure why the pseudopotential included in the example no longer works; maybe there was a change in the format that is acceptable by QE. Since this error has been reported before, I have created an issue https://github.com/Z2PackDev/Z2Pack/issues/108 to investigate it.
In any case, at least for the "Bi.pbe-dn-kjpaw_psl.1.0.0.UPF https://www.quantum-espresso.org/upf_files/Bi.pbe-dn-kjpaw_psl.1.0.0.UPF" pseudopotential the input files need changing because the "z_valence" of the pseudo is 15 (instead of 5 for the previous pseudopotential). That means that we will have 30 instead of 10 occupied states.
That means `nbnd` in `bi.nscf.in` needs to be changed to some number larger than 30, let's pick 36 for example (adding a few to make sure all occupied states are converged). In the .win file, we will need to change `num_wann` and `num_bands` to 30, and `exclude_bands` to `31 - 36`.
These values generally need to be changed when investigating different systems, such that the occupied subspace is considered.
From the plots, you can also see that the results are incorrect: The WCC should follow a smooth evolution. Abrupt changes in WCC position are generally a sign of trouble. They can happen (mostly when the direct band gap is very low) but it should eventually converge to a continuous evolution as more "lines" are added in the convergence routine. The evaluation of the topological invariants relies on this continuous evolution, meaning that the results are meaningless if the WCC evolution is discontinuous.
Hope this helps,
Best regards,
Dominik
On 11.01.21 14:36, Johnson, Nathan wrote:
Dear recipient,
I have Z2Pack 2.1.1 set up with first principles code as Quantum Espresso version 6.4.1 with interface to Wannier90 3.1.0. I have run calculations for Z2Pack's Bi example and would like to see how our output compares.
Calculations have been performed for two Bi systems with pseudopotentials Bi.pbe-dn-kjpaw_psl.1.0.0.UPF https://www.quantum-espresso.org/upf_files/Bi.pbe-dn-kjpaw_psl.1.0.0.UPF and _Bi.pbe-d-mt.UPF_ using the provided run.py script. These pseudopotentials were used instead of the supplied pseudopotential as the error,
"
Error in routine readpp (1):
file ../input/Bi_MT_PBE.UPF not readable
", was returned during an attempted SCF run using the pseudopotential. The values used for converge parameters were the same values as those corresponding to converge parameters in the _run.py file located in the 'surface' folder of the Z2Pack 2.1.1 distribution.
What ought the output for this example be?
I have found a Z2Pack WCC plot corresponding to 'half the k_z =0 plane of Bismuth' on page 82 of the Topological Matter School Lectures, however, the surfaces of the example seem to be half the k_x =0 and k_x =0.5 planes. With the working assumption that the allotrope of Bi corresponding to the Topological Matter School plot is the same allotrope of Bi as that for which the ionic configuration of the example corresponds to, I also performed calculations for both Bi systems using ‘surface=lambda s, t: [t, s / 2, 0]’. I am uncertain, however, that the vertical axis of the WCC plots generated by calculations on this surface would correspond to the x-axis of direct space.
I attach the output of aforementioned calculations - please find attached the .win file for both Bi systems and WCC plots.
Looking forward to hearing from you,
Best wishes,
Nathan Johnson, PhD student UK