Nuclear fusion is going from strength to strength, with several technological breakthroughs in recent years, including the ability to build high-temperature superconducting (HTS) magnets that can generate high magnetic fields with almost no energy loss. This is important not only because it increases the obvious net energy production, but also because it may be possible to build more compact magnetic fusion devices such as tokamaks and stellarators. The dimensions of these machines are much smaller than those of ITER, and this has attracted the attention of industry in a variety of ways, from larger and more experienced companies focused on manufacturing, to a proliferation of start-ups focused on novel reactor designs and software modelling.
Part of the industry is developing artificial intelligence (AI) solutions to fusion problems. Two of these companies, digilab and Next Step Fusion, have recently visited the Fusion Group at the Barcelona Supercomputing Center headquarters.
During their visits, Dr Cyd Cowley, Fusion Solutions Engineer, and Mr Alexei Zurba, Chief Product Officer (CPO), explained the relevance of AI and its applications in nuclear fusion. We enjoyed both presentations, which ranged from optimising diagnostics to real-time control (RTC) of plasma shape, a clear demonstration of the potential and breadth of AI. Afterwards, we discussed opportunities for collaboration in the near future and the need for high performance computing (HPC) resources when it comes to managing large amounts of data. It should be noted that MareNostrum 5, which is mainly a GPU supercomputer, is particularly capable of accelerating AI tasks.