Bayesian optimization for design of high-repetition-rate laser-driven muon source

Abstract

With the increasing repetition rate of ultra-intense laser pulses, ion beams accelerated by these lasers show great potential for achieving high-repetition-rate, high-average-flux muon sources. Nonetheless, generating high-quality ion beams is a challenging feat as it demands a careful balance among numerous physical effects. In this study, we utilize Bayesian optimization to fine-tune laser and plasma parameters to produce high-charge energetic ion beams, consequently leading to a high-yield muon source via pitcher-catcher scheme. Beginning with initial points steered by Latin hypercube sampling, Bayesian optimization conducts an adaptive, multi-parameter exploration of input parameter space, significantly faster than univariate uniform scans, and results in a mm-scale ps-duration laser-ion-based muon source scheme providing 106π± and 104μ+ at a 10 Hz frequency, using only several tens of simulations.