Session: K9-09: NANOSCALE THERMAL TRANSPORT MODELING AND MACHINE LEARNING I

Paper Number: 137523

137523 - Quantum Annealing Facilitated Lattice Optimization for High Entropy Alloys 

Abstract: 

High Entropy Alloys (HEAs) have received more and more attention due to their exceptional mechanical and thermal properties. Many studies have focused on unraveling the relationship between the configuration of HEAs and their properties. However, the extremely large search space on the configuration has posed challenges for traditional optimization algorithms to find the group state of the HEAs. This work proposes a quantum annealing facilitated active learning algorithm for HEA lattice optimization. Utilizing data from Density Functional Theory (DFT) calculations, a Field-aware Factorization Machine (FFM) is trained to encapsulate the optimization problem into a Quadratic Unconstrained Binary Optimization (QUBO) format, and the active learning iterations are employed with integrating training and refinement of a machine learning potential within the pipeline to perform material property calculations without computational expensive DFT calculations. When applied to NbMoTaW alloys, our algorithm can take advantage of the parallelism of quantum annealing algorithms to quickly search for alloy structures with significantly low configurational energy. At the same time, the flexible application scenarios of FFM also allow us to implement a variety of different thermal and mechanical property optimizations. Our results highlight the potential of quantum computing in materials design and discovery, laying a solid foundation for further analysis of the relationship between the structures of alloys and their properties.

Presenting Author: Zhihao Xu University of Notre Dame

Presenting Author Biography: Zhihao Xu is a 4th year Ph.D. student at the University of Notre Dame. He is specializing in the domain of material science, thermal science and computational science. His recent work is mainly on the material design and discovery by machine learning algorithms.

Authors:

Zhihao Xu University of Notre Dame
Eungkyu Lee Kyung Hee University
Tengfei Luo University of Notre Dame