pymatgen 2025.10.7

v2025.10.7

on Python PyPI

• Fixes a bug related to indices when skipping repeated k-points while reading PROCAR (for example, the double-counting of symmetry points when using a k-path for plotting bandstructure).
  Previously, projections, eigenvalues, occupations and weights would be attributed to the wrong k-points. Also, the k-point following the skipped k-point would have complex projections of NaN, as well as an occupation, weight, eigenvalue and squared projections of 0.

• replace print with logging, I assume pymatgen as a lib code shouldn't use print...
• remove some print from tests

• fix #4495
• Check ArrayWithUnit behaviour
• Fix unpickle ArrayWithUnit loses unit_type
• Add some tests for core.units, https://app.codecov.io/gh/materialsproject/pymatgen/blob/master/pymatgen%2Fcore%2Funits.py

• Replace linear_assignment with scipy linear_sum_assignment
• Speed benchmark

Benchmark

n | avg_time (s) |  std_dev (s)
------------------------------------
100 |     0.000104 |     0.000018
500 |     0.004302 |     0.002333
1000 |     0.018374 |     0.006702
5000 |     1.049615 |     0.327349

n | avg_time (s) |  std_dev (s)
------------------------------------
100 |     0.000121 |     0.000030
500 |     0.005082 |     0.002957
1000 |     0.018487 |     0.006484
5000 |     1.062805 |     0.332552

• feature 1: Parsing and storage of vibrational modes within a JDFTXOutfileSlice
  -- Stored as a list[dict], where each dict contains
  --- "Type": str
  ---- either "Imaginary", "Zero", or "Real"
  --- "Type index": int
  ---- (Base 1) index of the mode among other modes of the same frequency type
  --- "Degeneracy": int
  ---- Degeneracy of the mode (ie if mode["Degeneracy"] == 2, than there is one other mode in the list with the same frequency)
  --- "Frequency": float | complex
  ---- Frequency of the vibration in eV
  --- "IR intensity": float
  ---- IR intensity of the vibrational mode in e^2/amu as derived from dipole of the vibration
  --- "Displacements": NDArray[float]
  ---- Displacement vector representing the vibrational mode (converted from Bohr to Angstrom as magnitude of vector is a input variable for vibrational analysis in JDFTx)
  -- Testing for vibrational mode parsing
• feature 2: Parsing and storage of vibration energy components within a JDFTXOutfileSlice
  -- Stored as a dict[str, float], containing
  --- "T" : Temperature at which free energy was evaluated (in K)
  --- Dumped free energy components in eV
  -- Testing for vibrational energy components parsing

• Google format doc strings added. Check with ruff.
• Type annotations included. Check with mypy.
• Tests added for new features/fixes.
• If applicable, new classes/functions/modules have duecredit @due.dcite decorators to reference relevant papers by DOI (example)
  Tip: Install pre-commit hooks to auto-check types and linting before every commit:

pip install -U pre-commit
pre-commit install

Summary

• Loosen requests upper pin, current pin seems too strict and might cause unresolvable dependency
• Fix a few deprecation warnings
• Fix some new ruff errors from pre-commit

• Incorrect label (and value)
• glob now would use exact match first (['path_1_Li_sv2004_test/00/POSCAR.xyz', 'path_1_Li_sv2004_test/00/POSCAR', 'path_1_Li_sv2004_test/00/POSCAR.vasp'])

  pymatgen/src/pymatgen/analysis/transition_state.py

  Line 261 in f01009c

  poscar = glob(f"{neb_dir}/POSCAR*")

• Deprecate spline_options dict with boolean flag zero_slope_saddle as there's only one boolean config available:

  pymatgen/src/pymatgen/analysis/transition_state.py

  Line 70 in f01009c

  if self.spline_options.get("saddle_point", "") == "zero_slope":

• moved FHI-aims specific content into a separate package (pyfhiaims) for ease of maintenance (stop multiple libraries from implementing the same basic interface)

Todos

• Workout how to best test the aims io interface/dependency tracker
• check with community this does not break their setups

• Patches ReDOS vulnerability mentioned in #2755

• Remove LOBSTER output file trailing line sensitivity, fix #4215
  str.splitlines(keepends=False):

Unlike split() when a delimiter string sep is given, this method returns an empty list for the empty string, and a terminal line break does not result in an extra line:
cc @JaGeo @naik-aakash

Summary

• Remove unit in docstring for Element properties, to close #4453
• Check other units in docstring

Summary

• Handle numpy array for selective dynamics in Structure, to fix #4460
• Double check consistent behaviour from json and orjson, also check round trip?
• Unit test

• feature 0:
• fix 1: Missing a letter "s" for site.species in pymatgen.symmetry.analyser.SpacegroupAnalyzer.get_primitive_standard_structure()

• Better comment handling for CP2K input file, to fix #4442