Fix structure field shadowing sort_structure in VaspInputGenerator

[ThomasWarford]

Summary

• VaspInputGenerator redeclared structure as a plain dataclass field, overriding the structure property (with its sort_structure/reduce_structure/validate_magmom setter) inherited from pymatgen's VaspInputSet. As a result, sort_structure silently never ran for any atomate2 VASP input-set generator.
• This surfaces destructively in MDSetGenerator's npt ensemble: LANGEVIN_GAMMA is sized to the number of distinct elements, while VASP counts POSCAR "types" as contiguous same-element runs. With an unsorted structure the two counts diverge and VASP aborts with Error reading item LANGEVIN_GAMMA from file INCAR.
• Fix: remove the redundant field declaration so the inherited property is no longer shadowed.
• Added regression tests (tests/vasp/test_sets.py) and a standalone repro (example.py).

Alternatively, we modify MDSetGenerator something like:

class MDSetGenerator(VaspInputGenerator):
    def get_input_set(self, structure=None, prev_dir=None, potcar_spec=False):
        if structure is not None and self.sort_structure:
            structure = structure.get_sorted_structure()
        return super().get_input_set(structure, prev_dir=prev_dir, potcar_spec=potcar_spec)

for a more targeted fix.

Additional dependencies introduced (if any)

• None.

TODO (if any)

• None.

Checklist

Work-in-progress pull requests are encouraged, but please put [WIP] in the pull request
title.

Before a pull request can be merged, the following items must be checked:

• Code is in the standard Python style.
  The easiest way to handle this is to run the following in the correct sequence on
  your local machine. Start with running ruff and ruff format on your new code. This will
  automatically reformat your code to PEP8 conventions and fix many linting issues.
• Doc strings have been added in the Numpy docstring format.
  Run ruff on your code.
• Type annotations are highly encouraged. Run mypy to
  type check your code.
• Tests have been added for any new functionality or bug fixes.
• All linting and tests pass.

Note that the CI system will run all the above checks. But it will be much more
efficient if you already fix most errors prior to submitting the PR. It is highly
recommended that you use the pre-commit hook provided in the repository. Simply run
pre-commit install and a check will be run prior to allowing commits.

Issue

Closes #1529. See this issue for bug reproduction instructions.

[JaGeo]

@esoteric-ephemera do you have time to look into this? I don't have much experience with the MD workflow

[esoteric-ephemera]

Thanks @ThomasWarford. I'm not sure the analysis is quite right, since VaspInputSet itself defines structure as an init field, and then later defines it as a property with a setter. My guess is that there's some unhealthy handling/collision of fields in the upstream model that need to be changed

That being said, it is redundant to redeclare the field here so I'm fine with this change

Guessing example.py was an AI-generated MWE?

[ThomasWarford]

example.py was AI generated yes. It's quite verbose sorry. Here's a simpler demonstration which I've read and cut down more closely.

It looks like redefining structure as a field in VaspInputGenerator means the setter function defined in VaspInputSet is no longer used.

from pathlib import Path
from atomate2.vasp.sets.core import MDSetGenerator

defining_class = next(
    cls for cls in MDSetGenerator.__mro__ if "structure" in cls.__dict__
)
structure_attr = defining_class.__dict__["structure"]

print(f"`structure` is defined on: {defining_class.__module__}.{defining_class.__qualname__}")
print(f"`structure` is a property: {isinstance(structure_attr, property)}")
print(f"`structure` has a setter: {isinstance(structure_attr, property) and structure_attr.fset is not None}")

---

Before removing structure field definition from VaspInputGenerator:

(dft) twarford:~/dev/atomate2> python setter_example.py
`structure` is defined on: pymatgen.io.vasp.sets.VaspInputSet
`structure` is a property: True
`structure` has a setter: True

After removing structure field from VaspInputGenerator:

(dft) twarford:~/dev/atomate2> python setter_example.py
`structure` is defined on: atomate2.vasp.sets.base.VaspInputGenerator
`structure` is a property: False
`structure` has a setter: False

Without the setter function sorting based on elements, we get the bug in issue #1530.

[esoteric-ephemera]

Gotcha, thanks for the analysis and fix @ThomasWarford! Merging asap