A global shape optimization method based on Bayesian optimization with exploitation of derivative information is studied. In order to improve the scalability of the approach for a large number of iterations N, an iterative inversion scheme is introduced that reduces the numerical effort from O(N^3) to O(N^2). The method is implemented in JCMsuite's analysis and optimization toolbox.
X. Garcia-Santiago, et al. Bayesian Optimization With Improved Scalability and Derivative Information for Efficient Design of Nanophotonic Structures. Journal of Lightwave Technology, 39(1), 167 (2020).