Achieving high-quality proton beams for accelerators hinges on effective beam tuning. However, the conventional "Monkey Jump" method, widely used for tuning, proves labor-intensive and inefficient. Through harnessing Reinforcement Learning (RL), a novel beam tuning strategy can swiftly emerge, making informed decisions based on the prevailing system status and control demands, offering a promising alternative for accelerator systems. We explore novel techniques RL-based beam tuning and applying it to the beam tuning process of the CiADS Front End accelerator currently, with the aim of significantly enhancing the efficiency of the tuning process. To achieve this, we will first establish an RL-compatible environment based on dynamic simulation software. Subsequently, the policy is trained under different initial conditions. Finally, the strategy successfully trained in the simulation environment will be tested on real accelerator to verify its effectiveness.

Obtaining the complete distribution of a beam in high-dimensional phase space is crucial for predicting and controlling beam evolution. Previous studies on tomographic phase space reconstruction often required linear beam optics in the relevant transport section. In this paper, we show that the method of maximum entropy tomography can be generalized to incorporate nonlinear transformations, thereby widening its scope to the case of nonlinear beam transport. The improved method is verified using simulation results and potential applications are discussed.

Accurately assessing the difference between two beam distributions in high-dimensional phase space is crucial for interpreting experimental or simulation results. In this paper, we compare the common method of RMS moments and mismatch factors, and the method of statistical divergences that give the total contribution of differences at all points. We first show that, in the case of commonly used initial distributions, there is a one-to-one correspondence between mismatch factors and statistical divergences. This enables us to show how the values of several popular divergences vary with the mismatch factors, independent of the orientation of the phase space ellipsoid. We utilize these results to propose evaluation standards for these popular divergences, which will help interpret their values in the context of beam phase space distributions.