Fix lattice velocities formatting #3433
Conversation
pymatgen/io/vasp/inputs.py
Outdated
| @@ -545,7 +547,8 @@ def get_str(self, direct: bool = True, vasp4_compatible: bool = False, significa | |||
| lines.append("Lattice velocities and vectors") | |||
| lines.append(" 1") | |||
| for v in self.lattice_velocities: | |||
| lines.append(" ".join(format_str.format(i) for i in v)) | |||
| # vasp is strict about the format when reading this quantity | |||
| lines.append(" ".join(f" {i: .7E}" for i in v)) | |||
There was a problem hiding this comment.
If VASP is strict about this, could you add a test that checks for the right float precision?
janosh
added
io
|
Hi @janosh, I added a check on the format of the produced POSCAR. It should be noted that in this case it just checks that the format is equivalent to the one written from the |
janosh
removed
the
needs testing
Summary
This PR fixes a problem in #3428 that I encountered while performing some more tests with NpT MD simulations: in the case of lattice velocities VASP seems to be strict in the format when reading the POSCAR. After some tests it seems that this should be the best format to allow the correct parsing of this quantity in VASP.
Sorry for the error in the previous PR.