The work “Machine-learning-based models in particle-in-cell codes for advanced physics modules” by EPP members Chiara B, Pablo J. B., Fábio C, and Luis OS is now out on Journal of Plasma Physics. This paper proposes a new method to use ML-based models in state-of-the-art plasma codes. ML models can be used in these codes to speed up or replace costly sub-algorithms typically used in advanced physics modules.

To showcase the method, the calculation of the probability of interaction between particles via collisions has been replaced, leading to an increase in computational performance without significant physics deterioration (see the plot shared here) from the classical algorithm.