Efficient global crystal structure prediction using polynomial machine learning potential in the binary Al–Cu alloy system

Machine learning potentials (MLPs) are attracting much attention as powerful tools to accurately and efficiently perform atomistic simulations and crystal structure predictions. In this study, we develop a polynomial MLP for the Al–Cu system applicable to the robust global structure search and metastable structure enumeration. We then apply a combination of a global optimization method and the polynomial MLP to the Al–Cu alloy system. As a result of approximately 1010 times energy computations, the globally-stable and metastable structures are enumerated in the Al–Cu system.