Continuous-Scale Bridge Processes for Multiscale Generation
Please login to view abstract download link
Multiscale mechanics typically yields discrete observations across scales, consisting of microstructure ensembles and their coarsened representations. In this work, we take a first step toward characterizing the set of fine-scale realizations consistent with a given coarse description between observed scales. More specifically, we introduce a data-driven continuous-scale bridge whose marginals match observed distributions and whose paths respect coupled trajectories induced by physical upscaling. We present applications relevant to random heterogeneous materials and discuss performance from both physical and mathematical standpoints.
