Juliane Mueller is a Research Scientist in the Computational Research Division at Berkley Lab. Her research focuses on algorithm development for computationally expensive black-box optimization problems. These problems arise, for example, whenever the parameters of a time-consuming simulation model must be adjusted such that its outputs best agree with observational data. Mueller approximates the computational expensive simulations with fast-to-compute surrogate models. These surrogate models enable the adaptive exploration of the parameter space and thus the optimal solution will be found within very few expensive simulation evaluations. Mueller employed these optimization methods in various science applications, including combustion simulations, modeling in earth sciences, particle physics, design optimization in engineering, and materials sciences. Recently, Mueller extended the application of her methods to optimizing the hyperparameters of machine learning models that are used to predict future groundwater levels in California. Read more about her optimization work.