Data-driven modelling of collisional kinetic plasma dynamics

Computational plasma physics has seen significant advances in the ability to model nonlinear plasma dynamics from first-principles. However, capturing the complex multi-scale dynamics of plasmas, especially in non-equilibrium and strongly coupled regimes, remains challenging. This talk will highlight recent work that explores the potential of combining machine learning algorithms, differentiable simulators, and traditional plasma kinetic simulation frameworks to address these challenges. In particular, how electromagnetic Particle in Cell (PIC) simulations that resolve the inter particle fields that mediate collisional interactions can guide the development of data-driven collision operators. The first part of the talk will highlight how to efficiently learn collision operators from self-consistent PIC data. As a test case, we will focus on extracting an interpretable collision operator for an electromagnetic PIC code and compare the retrieved operator against existing theory. In the second part, we will motivate the usage of Graph Neural Networks (GNNs) as surrogate models for kinetic plasma simulators. More specifically, we wiill show that GNNs can accurately replicate the dynamics of a one-dimensional plasma and learn a computationally efficient collisional operator.