Implement flexible k-grid calculation

[RandomDefaultUser]

Currently, every unit cell regarding of shape gets a regular cubic k-grid assigned. Here, we should have an algorithm that reflects the unit cell geometry.

[elcorto]

What you want is a constant k-point density. I don't get why this isn't the default input format in any DFT code, instead of "k-grid" :)

• https://elcorto.github.io/pwtools/written/tutorial.html#find-monkhorst-pack-k-grid-sampling-for-a-given-unit-cell
• https://elcorto.github.io/pwtools/_modules/pwtools/crys.html#kgrid
• https://elcorto.github.io/pwtools/generated/api/pwtools.calculators.Pwscf.html#pwtools.calculators.Pwscf -> kpts keyword

ASE:

• https://wiki.fysik.dtu.dk/ase/ase/dft/kpoints.html
• ase.calculators.calculator.kpts2mp
• ase.dft.kpoints.monkhorst_pack

[RandomDefaultUser]

Ah yes, you're absolutely right! Thanks for all the links, I will have a look :)

[elcorto]

I bet there are also tools for that in https://pymatgen.org

[RandomDefaultUser]

That's also a good idea.

[elcorto]

Also, the folks over at https://www.aiida.net/ and https://materialsproject.github.io/fireworks/ have to solve the same problem, so everybody doing automated ("high throughput") DFT.

[RandomDefaultUser]

I'll make an issue for this one. I didn't really think about maybe using other existing workflows but i might be wiser in the long run to have this repository contain the custom code we actually need (analysis of MD trajectories, LDOS analysis, maybe MD performance analysis) and do the rest by using an established workflow.