IFMIF DONES hosts pioneering research into the design and optimisation of fusion facilities, as published in the scientific journal MDPI. The results of this paper demonstrate the existing synergy between deep learning models and differentiable programming, offering a promising collaboration between physicists and computer scientists to improve the design of future fusion power plants.

As part of the DONES-FLUX project, this work has applied surrogate deep learning models to the design and optimisation of the IFMIF-DONES linear accelerator.

Galo Gallardo Romero, Guillermo Rodríguez-Llorente, Lucas Magariños Rodríguez, Rodrigo Morant Navascués, Nikita Khvatkin Petrovsky, Rubén Lorenzo Ortegay, and Roberto Gómez Espinosa Martín participated in the study.