Bayesian Optimization of a Laser-Plasma Accelerator

Generating high-quality laser-plasma accelerated electron beams requires carefully balancing a plethora of physical effects and is therefore challenging---both conceptually in experiments. Here, we use Bayesian optimization key laser plasma parameters to flatten the longitudinal phase space an ionization-injected bunch via optimal beam loading. We first study concept with particle-in-cell simulations then demonstrate it Starting from arbitrary set point, accelerator autonomously tunes energy spread subpercent level at 254 MeV $4.7\text{ }\text{ }\mathrm{pC}/\mathrm{MeV}$ spectral density. Finally, robust regime, which improves stability delivers sub-five-percent rms for 90% all shots.