Parse bad float values in vasprun.xml for MD runs and OSZICAR

[DaanHolleman]

Python version

3.9.18

Pymatgen version

2024.8.9 / 2025.10.7

Operating system version

Red Hat Enterprise Linux 9.4

Current behavior

Vaspruns or Oszicars with unconverged ionic steps have ******** to indicate a wrong value but this throws an exception when trying to convert to float.

The parser should still parse these files as it is still interesting to see how the energies behaved up until this point.

Expected Behavior

Convert the ***** string to nan and raise a warning. There already exists a function for this called _vasprun_float

Minimal example

Use the following for an OSZICAR file:
       N       E                     dE             d eps       ncg     rms          rms(c)
DAV:   1     0.831012218308E+03    0.83101E+03   -0.25939E+05  4672   0.894E+02
DAV:   2    -0.461632861706E+04   -0.54473E+04   -0.53016E+04  5716   0.208E+02
DAV:   3    -0.541060909256E+04   -0.79428E+03   -0.78895E+03  5280   0.860E+01
DAV:   4    -0.544402022598E+04   -0.33411E+02   -0.33288E+02  5632   0.205E+01
          1 T= ****** E= 0.64876404E+06 F= 0.49220367E+06 E0= 0.49220367E+06  EK= 0.15656E+06 SP= 0.00E+00 SK= 0.00E+00

Then try to load the file:

from pymatgen.io.vasp import Oszicar
Oszicar('OSZICAR')

Relevant files to reproduce this bug

The float conversion in these places could maybe be changed to _vasprun_float

This one for vasprun.xml files

pymatgen/src/pymatgen/io/vasp/outputs.py

Lines 602 to 605 in 770791a

	elif tag == "energy":
	d = {i.attrib["name"]: float(i.text) for i in elem.findall("i")} # type:ignore[arg-type]
	if "kinetic" in d:
	md_data[-1]["energy"] = {i.attrib["name"]: float(i.text) for i in elem.findall("i")} # type:ignore[arg-type]

This one for Oszicar files

pymatgen/src/pymatgen/io/vasp/outputs.py

Line 4545 in 770791a

ionic_steps.append({key: float(value) for key, value in matches})

There might be more places where the float needs to be converted