Accelerated materials design using batch Bayesian optimization: A case study for solving the inverse problem from materials microstructure to process specification

Microstructure-based process design is one of the main ingredients for materials design, under integrated computational engineering paradigm, which relies on inverting process-structure–property linkages. The specific inverse problem connecting microstructure to processing conditions exceedingly difficult solve, even in a setting. difficulty arises from challenges associated with properly representing space as well cost simulations used connect evolution. In this work, we attempt invert process-microstructure by implementing and deploying search scheme based multi-scale batch Bayesian optimization. We employ framework efficiently navigate manifold two examples involving phase-field simulations. these examples, volume fraction characteristic length scale phases resulting spinodal decompositions are considered different objective functions find synthetic target microstructures. show how optimization can be uncover connections through optimal parallel querying space, providing new pathway solving problems design.