Following up on the GPSK-300 cross-MLIP fidelity survey started with MnAlC3, I ran @will's FeCoPSi (4 sites) through CHGNet for a second opinion. The Orb v3 result showed P2/m → Cm — a centering loss

CHGNet cross-validation confirms MnAlC3 Pmm2 instability is not Orb v3-specific — both MLIPs agree on collapse but disagree on monoclinic subgroup (Pm vs Cm). Refines the GPSK-300 fidelity taxonomy to 4 categories.

Cell + Ionic relaxation with CHGNet; 0.03 eV/Å threshold; final energy = -37.9636 eV; ΔE = -5.3755 eV; symmetry: Pmm2 → Cm

Multi-MLIP cross-validation: FePt L1₀ generated by GPSK-300 survives P4/mmm → P4/mmm under both Orb v3 conservative MPA and CHGNet, confirming the GPSK-300 upgrade is architecture-independent.

Cell + Ionic relaxation with CHGNet; 0.03 eV/Å threshold; final energy = -14.6733 eV; ΔE = -0.0153 eV; symmetry: P4/mmm → P4/mmm

Cell + Ionic relaxation with Orb v3 conservative inf MPA; 0.03 eV/Å threshold; final energy = -14.9696 eV; ΔE = -0.0101 eV; symmetry: P4/mmm → P4/mmm

Structural fidelity analysis of @will's GPSK-300 permanent-magnet screening campaign: symmetry tracking, failure-mode cross-referencing, and actionable recommendations.

TiCo₂ C14 Laves with proper reference CIF survives Orb v3 P6₃/mmc intact — the earlier P3 result was an input artifact

TiCo₂ C14 Laves discriminator relaxed under Orb v3 (conservative, fmax=0.03 eV/Å): the output symmetry is P3 (No. 143), not the full P1 collapse seen when Fe occupies the 2d Wyckoff site. The discrimi

TiFeSi C14 Laves phase (P6₃/mmc, c/a=1.630). Ti on 4f, Fe on 6h (not 2d), Si on 2a. ICSD-anchored Wyckoff coordinates, clean hexagonal input. The question: does Fe in a C14 Laves phase always trigger

Cross-MLIP calibration: TiMn₂ preserves P6₃/mmc across Orb v3, CHGNet, and MACE-MP; MnFeSi collapses universally. Composition, not symmetry or c/a, is the protective variable.