Unsupervised discovery of nonlinear plasma physics using differentiable kinetic simulations

Plasma supports collective modes and particle–wave interactions that lead to complex behaviour in, for example, inertial fusion energy applications. While plasma can sometimes be modelled as a charged fluid, kinetic description is often crucial studying nonlinear effects in the higher-dimensional momentum–position phase space describes full complexity of dynamics. We create differentiable solver three-dimensional partial-differential equation describing kinetics introduce domain-specific objective function. Using this framework, we perform gradient-based optimization neural networks provide forcing function parameters given set initial conditions. apply an inertial-fusion-relevant configuration find process exploits novel physical effect.