logo

Graduate student Changho Hong won the Best Poster Award at 2021 Virtual Spring Meeting. MRS meeting has been a prestigious and esteemed conference in the field of material science and engineering. This year, the best posters award was bestowed upon only 23 people among thousands of posters. Changho Hong presented “Developing Machine-Learning Potentials from Disordered Structures for Crystal Structure Prediction”, which introduced a new method of generating training sets for machine-learning potential to accelerate crystal structure prediction. It was observed that machine-learning potentials trained from disordered structures can distinguish stable crystalline phases out of metastable phases. With this finding, authors also investigated machine-learning potentials can replace first-principles calculation during crystal structure prediction, which accelerates the prediction process with 1~2 order of magnitude. Authors expect this method paves a new way to identifying unexplored ternary or higher (multinary) phases efficiently.