An Integrated Method of Bayesian Optimization and D-Optimal Design for Chemical Experiment Optimization

The smart chemical laboratory has recently emerged as a promising trend for future research, where experiment optimization is of vital importance. traditional Bayesian (BO) algorithm focuses on exploring the dependent variable space while overlooking independent space. Consequently, BO suffers from becoming stuck at local optima, which severely deteriorates performance, especially with bad-quality initial points. Herein, we propose novel stochastic framework D-optimal design (BODO) by integrating design. BODO can balance exploitation in and exploration We highlight excellent performance even poor points benchmark alpine2 function. Meanwhile, demonstrates better average objective function value than Summit SnAr process, showing its advantage potential application experiments.