Unsupervised Discovery of Inertial-Fusion Plasma Physics using Differentiable Kinetic Simulations and a Maximum Entropy Loss Function

TL;DR

This paper introduces a differentiable kinetic plasma simulation framework combined with a maximum entropy loss to optimize neural network parameters, revealing a previously unknown physical effect in inertial fusion plasma behavior.

Contribution

The authors develop a novel differentiable 3D plasma kinetic solver and a domain-specific optimization method to uncover new physical phenomena in inertial fusion plasma dynamics.

Findings

Discovery of a new physical effect in inertial fusion plasma.

Effective neural network optimization of plasma forcing functions.

Framework applicable to complex plasma physics problems.

Abstract

Plasma supports collective modes and particle-wave interactions that leads to complex behavior in inertial fusion energy applications. While plasma can sometimes be modeled as a charged fluid, a kinetic description is useful towards the study of nonlinear effects in the higher dimensional momentum-position phase-space that describes the full complexity of plasma dynamics. We create a differentiable solver for the plasma kinetics 3D partial-differential-equation and introduce a domain-specific objective function. Using this framework, we perform gradient-based optimization of neural networks that provide forcing function parameters to the differentiable solver given a set of initial conditions. We apply this to an inertial-fusion relevant configuration and find that the optimization process exploits a novel physical effect that has previously remained undiscovered.