Coupling of phase transition, anharmonicity, and thermal transport in CaSnF$_6$

The project is a specialized research repository accompanying a scientific paper. Its defensibility is low because it is a specific material study rather than a general-purpose tool; the value lies in the data and the specific ML model (NEP) parameters for CaSnF6, rather than a reusable software moat. While it has 5 forks within 21 days (suggesting academic peer interest or internal group use), it lacks the 'stars' or community traction that would indicate a broad-impact software framework. Frontier labs are unlikely to compete in this specific niche of fluoride perovskite thermal transport, but the methods (ML-based potentials for MD) are being standardized by projects like DeepMD, MACE, and CHGNet. The displacement risk is moderate as more accurate or universal potentials (like those from Microsoft's MatterSim or Google's GNoME) could supersede material-specific models. This is a classic example of an 'incremental' scientific contribution where the methodology (NEP + MD) is established, but the application to this specific material system is the new contribution.