Gaussian Process Regression and Bayesian Optimization for a 40-90 MeV Laser-Plasma Injector for the cSTART Storage Ring

Laser-plasma accelerators (LPAs) generate ultrashort high intensity electron bunches from a compact source size. At the Karlsruhe Institute of Technology (KIT), we will use an LPA as one of the injectors for the compact, high-momentum acceptance, non-equilibrium storage ring cSTART. The LPA injector with a length of only a few millimeters will be optimized to match the cSTART operation beam energy of 40– 90 MeV. It will be based on an ionization trapping scheme in combination with a tailored plasma den- sity profile to produce an electron beam with small energy spread that maximizes the spectral charge density at our tar- get energy, which is (for LPAs) comparably low. Moreover, the LPA injector must produce controlled electron beams with high shot-to-shot stability and avoid high-energy tails. These goals can be achieved largely by the detailed design of the plasma density profile and the laser pulse parameters. In an LPA, small changes across the high-dimensional pa- rameter space can have a disproportional influence on overall performance. To find parameters for stable high-quality LPA beams, we perform particle-in-cell (PIC) simulations and implement a machine-learning driven approach by using Bayesian Optimization (BO) based on Gaussian Process Re- gression (GPR). This procedure allows us to both optimize our gas target design and characterize the effects of the inter- action parameters, giving us a functional LPA with a simple tuning mechanism.