Tutorial 4

[lionelalexandre]

Tutorial4 : post processing

• contains: charge, band densities, band structure, DOS, projected DOS and STM
• issue with band densities
• deviation of values for some of them wrt. version v1.4-44-gd8899c46 (cf. README files in directories)
• added also print forces in xsf file

[lionelalexandre]

TODO

• provide a systematic plotting environment using python (also xmgrace and gnuplot are welcome)
• clean the input files ; to help in identifying what are the important variables
• PostProcess seems also print atomic magnetic moments ; something which is handled as well with ASE and not implemented yet in the Conquest calculator

[davidbowler]

I am checking all directories where you have noted something in README. I'll add comments for each.

[davidbowler]

1. pp_band_density_C2H4_mol says "bug here" in README: I can't see any problems with the calculations. There is some issue with band densities and gfortran that I've seen on macOS but haven't managed to track down: is that what you meant?

[davidbowler]

2. pp_band_structu_SrTiO3_bulk gives me no differences at all (once I remembered that we had to run an SCF calculation followed by the band output calculation!).

[davidbowler]

3. pp_DOS_C2H4_mol has a different DOS broadening (we changed the default value of sigma, the smearing, in a recent release) but the overall DOS is essentially the same (peaks in the same place, just different width and height).

[davidbowler]

4. pp_DOS_Si_bulk I suspect that the reference value you have may be from a calculation with a much higher k-mesh (15x15x15 k points). I can reproduce the result in the tutorial.

[davidbowler]

5. pp_pDOS_C2H4_mol Again, the difference is broadening on DOS peaks (and the energy limits).

[lionelalexandre]

1. pp_band_density_C2H4_mol says "bug here" in README: I can't see any problems with the calculations. There is some issue with band densities and gfortran that I've seen on macOS but haven't managed to track down: is that what you meant?

yes, this is not a real bugg, sorry

[lionelalexandre]

Well, if you want to do any update of references, please feel free. Thank you!