Magnetic control of tokamak plasmas through deep reinforcement learning with privileged information

Dmitry Igorevich Sorokin, Alexander A. Granovskiy, Ivan Kharitonov, Maxim Stokolesov, Igor Prokofyev, Evgeny Adishchev, Georgy Subbotin, Maxim Nurgaliev

Keywords: magnetic plasma control, reinforcement learning, privileged information

TL;DR: We propose a novel approach to real-time plasma control using reinforcement learning with privileged information. We validate our approach on various plasma scenarios and successfully test it on a physical DIII-D tokamak.

Confirmation Of Submission Requirements: I submit an abstract. It uses the template provided on the submission page and is no longer than 2 pages.

Submission Number: 52