1: \begin{abstract}

2: High-dimensional optimization is a critical challenge for operating large-scale scientific facilities. % that push the limits of technical and physical understanding. 

3: We apply a physics-informed Gaussian process (GP) optimizer to tune a complex system by conducting efficient global search. Typical GP models learn from past observations to make predictions, but this reduces their applicability to new systems where archive data is not available. 

4: Instead, here we use a fast approximate model from physics simulations to design the GP model. The GP is then employed to make inferences from sequential online observations in order to optimize the system.

5: % {\color{black}In addition, we present an alternative procedure to learn the GP

6: % hyperparameters by using the physics model's basis functions 

7: % kernel by using approximations to the physics model as basis functions rather than by empirical kernel selection. 

8: % which maximizes the marginal likelihood.

9: % This method reduces the requirement for prior data, and removes the need for empirical prior GP model selection.

10: % is superior to the data or physics-informed ML-II in the absence of data.}

11: Simulation and experimental studies were carried out to demonstrate the method for online control of a storage ring. We show that the physics-informed GP outperforms current routinely used online optimizers in terms of convergence speed, and robustness on this task.

12: The ability to inform the machine-learning model with physics may have wide applications in science. 

13: \end{abstract}

14: