New paper published in Scripta Materialia

Excited to share our first open-access article of 2025, just published in Scripta Materialia. 

In this work with my talented students and postdocs Md Rajib Khan Musa, Yichen Qian, and Jie Peng, we propose a novel computational approach that integrates Monte Carlo (MC) sampling, first-principles DFT calculations, and machine learning techniques to accelerate the discovery of Minimum Energy Co­figurations (MECs), a critical step towards understanding the local chemical environment in complex alloying systems. Our framework, named a-MCDFT, is grounded in the principles of the Cluster Expansion method, and it also incorporates a Local Outlier Factor model.

We validate our methodology through tungsten-based quaternary and equiatomic high-entropy alloys,  but its applicability can be extended to other MPEAs. The results highlight the accuracy of our a-MCDFT method in identifying MECs and calculating their energies, achieving relative errors of just ∼0.022% when compared to DFT energy evaluations. Furthermore, it is capable of finding such MECs at a fraction of the computational cost of other traditional frameworks, enabling a much larger number of MC swap trials per atom. 

Huge THANKS to John Echols from the Office of Fusion Energy Sciences and the U.S. Department of Energy Office of Science for their support.

You can find it (it is open access!) at: https://www.sciencedirect.com/science/article/pii/S1359646224005682