| Abstract Scope |
Predicting atomic ordering in HEAs remains a significant challenge due to the vast configurational phase space inherent to these systems. While ordering is known to critically influence a wide range of material properties, conventional methods like MC-DFT are often computationally prohibitive. To solve this, we developed MECx: the Minimum Energy Configuration Explorer. Our proposed framework integrates high‑fidelity energetic evaluations, materials‑informed ML models, and Monte‑Carlo sampling to efficiently discover deep, low‑energy configurations in multicomponent alloys. Rather than aiming for absolute energy precision relative to DFT, our approach focuses on accurately capturing energy trends to identify stable configurations. This strategy drastically reduces computational costs and is capable of predicting the minimum energy configuration by training on as few as 100 DFT data points. Ultimately, this methodology offers an efficient pathway for predicting the structural evolution of existing and next-generation HEAs, facilitating a deeper understanding of how atomic ordering dictates overall material behavior. |