io.pwscf.PWInput.from_file() incorrectly reads decimal numbers and the K_POINTS card in automatic mode

[jsukpark]

Python version

Python 3.12.4

Pymatgen version

2024.10.3

Operating system version

Ubuntu 22.04.5 LTS

Current behavior

The io.pwscf.PWInput.from_file() method fails to correctly parse:

• decimal numbers written in scientific notation with exponent part prefixed with e or E (note that as of QE v7.3, pw.x can still parse such cases correctly; i.e., exponent prefix need not be d), and
• the K_POINTS card with mode set to automatic.

Example input file scf.in:

&CONTROL
  calculation = 'scf',
/
&SYSTEM
  ecutwfc = 100,
  ibrav = 0,
  nat = 2,
  ntyp = 1,
/
&ELECTRONS
  conv_thr = 2e-12,    ! same behavior with 2E-12
/
ATOMIC_SPECIES
  C  12.0107 C.upf
ATOMIC_POSITIONS crystal
  C  0.2500000000  0.2500000000 0.2500000000
  C -0.0000000000 -0.0000000000 0.0000000000
K_POINTS automatic
  4 4 4 1 1 1
CELL_PARAMETERS angstrom
  -1.786119886   0.000000000   1.786119886
  -0.000000000   1.786119886   1.786119886
  -1.786119886   1.786119886   0.000000000

Current behavior:

>>> from pymatgen.io.pwscf import PWInput
>>> pwi = PWInput.from_file("scf.in")
>>> print(pwi.sections["electrons"]["conv_thr"])
2.0
>>> print(pwi.kpoints_grid, pwi.kpoints_shift)
(1, 1, 1) (0, 0, 0)

Expected Behavior

I expect both the conv_thr value and the K_POINTS card in the above example input file to be parsed correctly.

>>> from pymatgen.io.pwscf import PWInput
>>> pwi = PWInput.from_file("scf.in")
>>> print(pwi.sections["electrons"]["conv_thr"])
2e-12
>>> print(pwi.kpoints_grid, pwi.kpoints_shift)
(4, 4, 4) (1, 1, 1)

Minimal example

contents = """&CONTROL
  calculation = 'scf',
/
&SYSTEM
  ecutwfc = 100,
  ibrav = 0,
  nat = 2,
  ntyp = 1,
/
&ELECTRONS
  conv_thr = 2e-12,    ! same behavior with 2E-12
/
ATOMIC_SPECIES
  C  12.0107 C.upf
ATOMIC_POSITIONS crystal
  C  0.2500000000  0.2500000000 0.2500000000
  C -0.0000000000 -0.0000000000 0.0000000000
K_POINTS automatic
  4 4 4 1 1 1
CELL_PARAMETERS angstrom
  -1.786119886   0.000000000   1.786119886
  -0.000000000   1.786119886   1.786119886
  -1.786119886   1.786119886   0.000000000
"""

with open("scf.in", "w") as fp:
    fp.write(contents)

from pymatgen.io.pwscf import PWInput

pwi = PWInput.from_file("scf.in")
print(pwi.sections["electrons"]["conv_thr"])  # Prints "2.0", should print "2e-12"
print(pwi.kpoints_grid, pwi.kpoints_shift)  # Prints "(1, 1, 1) (0, 0, 0)", should print "(4, 4, 4) (1, 1, 1)"

Relevant files to reproduce this bug

No response

[jsukpark]

Seems there is one more bug in parsing K_POINTS card, adding that to this issue