Speaker
Description
The Isotope Separation On-Line (ISOL) technique has enabled advances in many fields spanning in nuclear, atomic, molecular, solid-state and medical physics by producing radioisotopes at facilities like CERN ISOLDE and the emerging ISOL@MYRRHA. Tuning these facilities is a complex task that requires manual intervention by experienced operators, a process that is often time-consuming due to the many parameters involved. In recent years, optimization algorithms have emerged as effective tools to support this tuning process. Among the key tuning tasks, the adjustment of ion source parameters plays a crucial role in maximizing the yield of the extracted ion beam. Since modifications to the ion source parameters can affect the beam energy and emittance, automatic re-tuning of the transport beamline parameters is required to ensure that beam intensity and shape performance criteria are satisfied. In this study, a nested optimization approach is proposed, utilizing Gaussian processes and Bayesian optimization to maximize the beam intensity of a selected isotope or molecule. Developed for ISOL@MYRRHA at SCK CEN and implemented in its ISOL offline system, the method was experimentally validated at CERN’s ISOLDE Offline 2 facility by maximizing the intensity of various isotopes across different operation parameters.
| Email address | santiago.ramos.garces@sckcen.be |
|---|---|
| Supervisor's Name | João Pedro Ramos |
| Supervisor's email | joao.pedro.ramos@sckcen.be |
| Funding Agency | MYRRHA INPO and SCK CEN Academy |
| Classification | Machine Learning and AI |
