Closed davidbowler closed 2 years ago
Let me ask you a question. I thought this branch was made from the latest "develop" branch. If so, I cannot find XC_LibXC_v5_module.f90. Is it okay?
This branch started some time ago, so is behind the latest develop, which explains the lack of LibXC5. However, when it is merged into develop it will only add to the code, and not remove the new changes.
Thank you for your reply. I am relieved.
A few comments:
I just wonder for 3 or 4, we should change them now or later.
I have found that Conquest will not produce Process0000001WF.dat etc. if I use the following Conquest_input.
If the lines
are activated, I could get the files.
Is it a correct behavior?
---- Conquest_input ------- IO.outputWF T IO.maxnoWF 6
%block WaveFunctionsOut 62 63 64 65 66 67 %endblock ---- Conquest_input -------
No, this is not correct behaviour - thank you for finding it! I can see where the problem is (in initial_read_module.f90
) and will fix it.
Thank you very much for the update. I have confirmed that the problems I pointed out above have been solved.
I have some questions, comments or suggestions for DOS.
I am sorry, but I have not checked the STM image. Others look perfect to me.
I have now updated DOS as discussed: improving default parameters (1001 bins, min/max eigenvalues, sigma); added integrated DOS (including importing smearing from Conquest for Fermi or Methfessel-Paxton).
I have also refactored the storage to account for bands output by Conquest and bands to be processed (which do not have to be the same set, though the second has to be a subset of the first).
Also the output of chden.nnn files from Conquest is now turned on by IO.DumpChargeDensity
and happens at SCF, and optionally at each SCF step if IO.Iprint_SC
is larger than 2.
Also the output of chden.nnn files from Conquest is now turned on by
IO.DumpChargeDensity
and happens at SCF, and optionally at each SCF step ifIO.Iprint_SC
is larger than 2.
This sounds very reasonable to me.
Thank you very much for the update of DOS procedure. It looks wonderful, but I have found a small problem and suggestion.
process_dos
, E_DOS_min
or E_DOS_max
are shifted by 2xsigma_DOS
. But, this should be done AFTER the setting of sigma_DOS
, I think.E_DOS_min
) to the integrated DOS. (It may be a little difficult to consider this when E_DOS_min
is within the dispersion of some band.)I will send an e-mail showing the corresponding part.
Thank you very much for the update of DOS related tools.
Let me ask you one more.
You have changed the keywords for DOS, from IO.min_DOS_E
etc. to Process.min_DOS_E
.
I agree that it should be Process.***
, but I have noticed that the explanation in manual has not been changed.
Please change them.
As far as I have checked, others are fine and let me approve this pull request. Thank you very much for your efforts.
I have made the changes to the documentation of the parameter names (Process.XXX
) and added an option to allow integrated DOS to be local (i.e. just within the energy window) or total (i.e. from lowest eigenvalue).
I've also realised that the band density and STM are not (yet) compatible with MSSF, so I've added a check to quit if post-processing is attempted on these areas with MSSF.
New post-processing utility for output: structure, charge density, band density, STM simulation, DOS.