[ase-users] Inconsistent space group determination for ICSD_CollCode262928.cif

[Hongyi Zhao]

On Mon, Jul 29, 2024 at 10:30 AM Hongyi Zhao <hongyi.zhao at gmail.com> wrote:
>
> Dear ASE community,
>
> I've encountered an interesting case of space group determination
> inconsistency using ASE for the structure ICSD_CollCode262928.cif.
> Here are the details:
>
> The ase version:
>
> (datasci) werner at x13dai-t:~$ pip list |grep ^ase
> ase                           3.23.0
>
> The test is as follows:
>
> ```python
> In [15]: import numpy as np
>     ...: from ase.io import read
>     ...: from ase.spacegroup import get_spacegroup
>     ...:
>     ...: def analyze_space_group_ase(filename):
>     ...:     """Analyze space group using ASE with different tolerances."""
>     ...:     # Read the CIF file
>     ...:     atoms = read(filename)
>     ...:
>     ...:     # Define tolerances to test
>     ...:     tolerances = [1e-1, 1e-2, 1e-3, 1e-4, 1e-5, 1e-6, 1e-7]
>     ...:
>     ...:     print(f"Analysis for file: {filename}")
>     ...:     print(f"Cell parameters:\n{atoms.cell}")
>     ...:     print(f"Number of atoms: {len(atoms)}")
>     ...:     print(f"Atom types: {set(atoms.get_chemical_symbols())}")
>     ...:
>     ...:     for tol in tolerances:
>     ...:         spacegroup = get_spacegroup(atoms, symprec=tol)
>     ...:         print(f"Tolerance {tol:.1e}: {spacegroup.symbol}
> ({spacegroup.no})")
>     ...:
>     ...: if __name__ == "__main__":
>     ...:     cif_file = "ICSD_CollCode262928.cif"
>     ...:     analyze_space_group_ase(cif_file)
>     ...:
> Analysis for file: ICSD_CollCode262928.cif
> Cell parameters:
> Cell([[3.3717, 0.0, 0.0], [-1.6858499999999994, 2.9199778539399923,
> 0.0], [0.0, 0.0, 5.3855]])
> Number of atoms: 4
> Atom types: {'Mn', 'O'}
> Tolerance 1.0e-01: P 63 m c (186)
> Tolerance 1.0e-02: P 63 m c (186)
> Tolerance 1.0e-03: P 63 m c (186)
> Tolerance 1.0e-04: P 63 m c (186)
> Tolerance 1.0e-05: C m c 21 (36)
> Tolerance 1.0e-06: C m c 21 (36)
> Tolerance 1.0e-07: C m c 21 (36)
> ```
>
> As you can see:
>
> 1. Structure Information:
>    - File: ICSD_CollCode262928.cif
>    - Cell parameters:
>      [[3.3717, 0.0, 0.0],
>       [-1.6858499999999994, 2.9199778539399923, 0.0],
>       [0.0, 0.0, 5.3855]]
>    - Number of atoms: 4
>    - Atom types: Mn, O
>
> 2. Space group determination results using different tolerances:
>    - 1.0e-01 to 1.0e-04: P 63 m c (186)
>    - 1.0e-05 to 1.0e-07: C m c 21 (36)
>
> This structure shows a transition in the determined space group when
> the tolerance is changed from 1.0e-04 to 1.0e-05.
>
> Questions:
> 1. Is this behavior expected?
> 2. Which space group should be considered as the "correct" one for
> this structure?
> 3. Are there any recommendations for handling such cases where the
> space group determination is sensitive to the tolerance value?
>
> I would appreciate any insights or explanations about this behavior,
> as well as best practices for reporting space groups in such
> borderline cases.

I got it. The core reason for this issue can be succinctly summarized
as follows:

1. Input precision limit: The CIF file provides cell parameters
precise to 1e-4 Å:

celllength_a 3.3717
celllength_b 3.3717
celllength_c 5.3855(7)


2. Tolerance impact:
   - Tolerance ≥ 1e-4: Identified as P 63 m c (186)
   - Tolerance < 1e-5: Identified as C m c 21 (36)

3. Correct determination: P 63 m c (186) should be adopted because:
   - It matches the input data precision
   - It's consistent with the hexagonal system (Wurtzite) mentioned in the file

4. Best practices:
   - Use a tolerance comparable to the input data precision (1e-4 in this case)
   - Avoid using tolerances exceeding data precision to prevent
over-interpretation

So, the lower symmetry result (C m c 21) stems from over-interpreting
minute differences beyond the actual precision of the data.
This case emphasizes the importance of considering input data
precision and carefully selecting tolerance parameters in
crystallographic analysis.

Regards,
Zhao

> Thank you for your time and expertise.
>
> Best regards,
> Zhao
> --
> Assoc. Prof. Hongsheng Zhao <hongyi.zhao at gmail.com>
> Theory and Simulation of Materials
> Hebei Vocational University of Technology and Engineering
> No. 473, Quannan West Street, Xindu District, Xingtai, Hebei province