ENH: VASP Frequency Results Handling

Wanted to get the results of an IBRION=5 calculation, i.e. a phonon/frequency run. I propose two new functions for the VASP calculator:

• read_vib_modes to return tuples of eigenvalues and eigenvectors (can also be complex eigenvalues)
• get_vibrations to get a representation of the vibrations by displacing coordinates along the normal modes.

While writing the test I stumbled upon a remaining open file (the sorting file), so I changed that. Don't see why this should remain open.

This collides with !1274, so if depending on what is merged first a bit of work is needed. Happy to do that though. If helpful, I can also try to incorporate the changes of !1274 and write some tests.

Happy to get feedback on this.

added 46 commits

• 6905a9c8...ad110ab2 - 45 commits from branch ase:master
• 587c9f41 - Merge branch 'master' into vasp-freq

Compare with previous version

Just found the new VibrationsData Class. It seems more reasonable to replace get_vibrations with a version that returns a VibrationsData object than to re-implement this here... Switching back to draft.

marked this merge request as draft

added 2 commits

• 9228d1bc - use VibrationsData and adjust test
• 0b0a11f4 - prettify

Compare with previous version

added 2 commits

• 56246027 - restore deleted function
• c219d1ef - fix missing lines

Compare with previous version

added 2 commits

• cc04e673 - test vasp freq functions against each other
• d4edd3a9 - Vasp vibrations doc, relating VibrationsData doc

Compare with previous version

Ok, I now added get_vibrations to the VASP calculator to return a VibrationsData object that is built using the vasprun.xml. It supports some constraints (VibrationsData only supports fully fixed atoms, so that is a limitation compared to VASP, if the VASP calculation was started manually).

I added a test, that tests the existing read_vib_freq (from the OUTCAR) against the new function.

Since VASP6 changes the content of the vasprun.xml I had some trouble, but think I figured it out in the end.

Added documentation and enabled documentation of the VibrationsData, since this was missing in my opinion anyhow.

The clash with !1274 remains, but I guess it is out of date and needs work. If I find the time I might do this later.

Happy to hear your opinion.

marked this merge request as ready

added 1 commit

• a0d1d2ec - make lint happy?!

Compare with previous version

Mmmh, no idea how I could fix the lint problem, I tried asserting that it is not None...

ase/calculators/vasp/vasp.py:1208: error: Unsupported target for indexed assignment ("Optional[Any]")
ase/calculators/vasp/vasp.py:1217: error: Unsupported operand types for * ("None" and "float")
ase/calculators/vasp/vasp.py:1217: note: Left operand is of type "Optional[Any]"

added 71 commits

• a0d1d2ec...51aaa4ac - 69 commits from branch ase:master
• 852f9ce2 - Merge branch 'master' into vasp-freq
• a5476a4b - insert assert

Compare with previous version

added 1 commit

• ef43a1bf - make mypy happy... hopefully

Compare with previous version

added 1 commit

• ccb537c3 - prettify test

Compare with previous version

Quick ping if I am missing something that remains to be done or if this can be merged?

Cheers for bump, I'm at a conference this week but happy to give it a test with Vasp 5.

assigned to @ajjackson