Physics-informed Gaussian Process for Online Optimization of Particle Accelerators

TL;DR

This paper introduces a physics-informed Gaussian process optimizer that leverages physics simulations to efficiently and robustly optimize complex large-scale scientific systems like particle accelerators in real-time.

Contribution

It presents a novel approach combining physics simulations with Gaussian processes for online optimization, improving convergence speed and robustness over existing methods.

Findings

Outperforms current online optimizers in convergence speed.

Demonstrates robustness in simulation and experimental studies.

Effective for online control of a storage ring.

Abstract

High-dimensional optimization is a critical challenge for operating large-scale scientific facilities. 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. 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. 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. The ability to inform the machine-learning model with physics may have wide applications in science.