A machine learning-based Bayesian optimization solution to nonlinear responses in dusty plasmas

Ding, ZhiyueMatthews, LorinHyde, Truell2022-03-212022-03-212021-06Machine Learning Science and Technology, 2, 035017, 2021https://hdl.handle.net/2104/11776Nonlinear frequency response analysis is a widely used method for determining system dynamics in the presence of nonlinearities. In dusty plasmas, the plasma–grain interaction (e.g. grain charging fluctuations) can be characterized by a single-particle non-linear response analysis, while grain–grain non-linear interactions can be determined by a multi-particle non-linear response analysis. Here a machine learning-based method to determine the equation of motion in the non-linear response analysis for dust particles in plasmas is presented. Searching the parameter space in a Bayesian manner allows an efficient optimization of the parameters needed to match simulated non-linear response curves to experimentally measured non-linear response curves.enA machine learning-based Bayesian optimization solution to nonlinear responses in dusty plasmasArticle10.1088/2632-2153/abe7b7