ALPI-PIAVE performance at INFN-LNL with advanced optimization algorithms

The ALPI-PIAVE heavy-ion accelerator complex at INFN-LNL requires precise tuning to achieve optimal performance, particularly given its complex beam dynamics and high-dimensional parameter space. Although the facility is designed for 60% transmission, actual transmission rates are significantly lower, around 15–20%, highlighting the need for improved optimization strategies. In this work, we compare advanced optimization techniques, specifically Bayesian Optimization (BO) and Trust Region Bayesian Optimization (TuRBO), to enhance the efficiency, reliability, and adaptability of ALPI-PIAVE operations. Since Bayesian optimization algorithms are sensitive to accelerator parameter drifts, we integrated Particle Swarm Optimization (PSO) to improve robustness. Additionally, we conducted controlled instability tests to evaluate performance under dynamic conditions. The results of these experiments and their implications for real-time accelerator tuning are presented in this paper.