Year: 2026
Type: article
Abstract: Developing models which are both accurate and computationally efficient is along-standing goal in plasma physics. For systems where small-scale kineticphenomena significantly influence global dynamics, this can be challenging, asglobal kinetic simulation – though highly accurate – is often computationallyintractable. Fluid models often provide a viable alternative, but their accuracydepends on how well the utilised closure captures the neglected kinetic physics.For collisional plasmas, closures can be constructed rigorously throughperturbative expansion around thermal equilibrium. However, no similarlygeneral framework exists in the collisionless case, prompting the developmentof a variety of collisionless closures with limited ranges of validity. Examplesinclude the Chew-Goldberger-Low closure, valid in the strongly magnetisedlimit, and Hammett-Perkins-like closures for linear Landau damping. Manyproblems of interest lie outside these regimes, however, often necessitatingad-hoc or theoretically poorly justified closures.This motivates the exploration of alternative approaches to closure construc-tion. In this thesis, we use a data-driven approach, employing machine learningmethods to systematically construct closures from first-principles kinetic simu-lation data. Initially restricting ourselves to one-dimensional setups, we developa two-step sparse regression procedure based on the SINDy framework anduse it to discover a new six-term closure for electrostatic phenomena. Theclosure accurately captures both linear and nonlinear regimes of the electrontwo-stream instability, as well as Landau damping of Langmuir waves. Addi-tionally, the closure generalises naturally to multi-species modelling, relevantfor streaming instabilities.Having identified the closure form in the first step of the procedure, wefurther show how its free parameters can be dynamically estimated fromfluid quantities. For this, we illustrate how both neural networks and moreinterpretable methods can be leveraged – in the latter case through a newlydeveloped framework for nonlinear sparse regression.
Author: Raaholt Ingelsten, Emil
Language: English
Subfield: Nuclear and High Energy Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7205615782