Bayesian optimization for the local bump injection in the HLS-II storage ring

Abstract Currently, a conventional local bump injection system with four pulsed dipole kicker magnets is adopted in the Hefei Light Source II (HLS-II) storage ring to achieve top-off operation. Due to the multipole magnets located between the kickers in the injection section, the local bump injection presents technical challenges in forming a perfect closed bump, which causes oscillation to the stored beam. In order to reduce the injection disturbance on the stored beam, the Bayesian Optimization (BO) method is employed to determine the kick angles of the four bump kickers.