Machine learning atomic-scale stiffness in metallic glass

Due to lack of either translational or rotational symmetries at atomic-scale, predicting properties amorphous materials from static structure is a challenging task. To circumvent the dilemma, supervised machine-learning strategy via neural network proposed predict atomic stiffness metallic glass discretized radial distribution function. The predicted and its spatial nature are calibrated with molecular dynamics simulations. After which, origin constraint interpreted learning structural input. Inadequacy model discussed in terms incompleteness both configurational space descriptor.