부산대학교 도서관

Applications of laser-plasma accelerators (LPA) require independent control of electron beam parameters. However, due to the complex coupling of the many variables governing the laser-plasma interaction, precisely tuning these parameters based on simple scalings is often impossible or at least suboptimal. Here, we apply multiobjective Bayesian optimization to derive optimal tuning curves for LPAs, both in simulations and experiments. For electron energies between 150 and 250 MeV, we demonstrate tuning of the charge over a range of nearly 100 pC, while preserving optimal beam loading conditions with energy spreads below 5%. The derived tuning curves can explain the sometimes counterintuitive interplay between laser and plasma control variables that is necessary to find the best trade-off between competing beam properties.