ML and DFT reveal local magnetic environments in magnetocaloric high-entropy alloys

bravo_abad · x · 2026-08-26

High-entropy alloys present a computational paradox due to their vast compositional freedom. Zhe Cui et al. combine DFT and machine learning to demonstrate that magnetostructural behavior can be understood through local atomic environments rather than long-range configurations. This enables a minimal-supercell strategy: using smaller first-principles calculations to sample relevant local environments, then letting ML map the larger design space, significantly reducing computational time.

Original post →

More from Research

Research channel →