PHYSBO

PHYSBO is a Python library for fast and scalable Bayesian optimization. It is useful for finding combinations of parameters (material composition, structure, process and simulation parameters, etc.) that improve the value of the objective function (material properties, etc.) with the fewest number of experiments and simulations. Users can set the parameters and the objective function and apply them to problems in a wide variety of fields. It is designed to be highly scalable, using Thompson sampling, random feature maps, 1-rank Cholesky updating, and automatic tuning of hyperparameters to allow the application of large amounts of training data.