Data-enabled computational multiscale method in materials science and engineering

In the community of computational materials science, one of the challenges in hierarchical multiscale modeling is information-passing from one scale to another, especially from the molecular model to the continuum model. A machine-learning-enhanced approach, proposed in this paper, provides an alternative solution. In the developed hierarchical multiscale method, molecular dynamics simulations in the molecular model are conducted first to generate datasets, which represents physical phenomena at the nanoscale. The datasets are then used to train neural networks for failure classification and stress regressions. Finally, the well-trained learning machines are implemented in the continuum model to study the mechanical behaviors of materials at the macroscale. Randomized neural networks are employed due to their computational efficiency.