[Davis Rens (davrens101)]

Summary

In mattersim.utils.supercell_utils.auto_grid_detection, calculating the number of supercell replicas along lattice directions was written as:

number_of_replicas = int(np.round(max_atoms / (len(atom))) ** (1 / 3))

Because the closing parenthesis of np.round was placed before ** (1 / 3), np.round was called on max_atoms / len(atom) instead of the cube root result.

In Python floating-point arithmetic, taking the cube root of perfect cubes often produces values slightly below integers (for example, 64 ** (1 / 3) == 3.9999999999999996 and 216 ** (1 / 3) == 5.999999999999999). Casting this with int(...) truncated the value, resulting in:

• max_atoms = 64 (with 1-atom unit cell) yielding (3, 3, 3) (27 atoms) instead of (4, 4, 4) (64 atoms).
• max_atoms = 216 yielding (5, 5, 5) (125 atoms) instead of (6, 6, 6) (216 atoms).

Similarly, in Case 2 (asymmetric lengths beyond tolerance), asymmetric_replica was computed with int(...) without rounding the cube root.

Changes

• Fixed parentheses so np.round operates on the cube root: int(np.round((max_atoms / len(atom)) ** (1 / 3))).
• Applied np.round to asymmetric_replica calculation as well.
• Added unit tests in tests/utils/test_supercell_utils.py verifying detection on cubic systems and within ratio_tolerance.

[Davis Rens (davrens101)]

@microsoft-github-policy-service agree [company=]

[microsoft-github-policy-service]

Davis Rens (@davrens101) the command you issued was incorrect. Please try again.

Examples are:

@microsoft-github-policy-service agree

and

@microsoft-github-policy-service agree company="your company"

[Davis Rens (davrens101)]

@microsoft-github-policy-service agree