Abstract Details
Abstracts
Author: Maxwell H Rosen
Requested Type: Poster
Submitted: 2026-03-20 16:34:10
Co-authors: J. Juno, M. Francisquez, A. Hakim, G.W. Hammett
Contact Info:
Princeton University
12315 Winterberry Way
Princeton, NJ 08540
United States
Abstract Text:
Collisionless magnetized plasmas with a magnetic well, such as in tokamak, stellarators and mirrors, contain a region of phase space where collisions scatter particles between a trapped and passing region, and determine the particle loss rates or confinement times. Numerical kinetic studies of these systems often employ explicit time-stepping schemes with a time step (dt) set by the Courant condition, limiting dt to the fastest timescale of the system.
In order to reach confinement times, though, an extraordinary number of small dt steps must be taken, since one has to span collisional timescales that are long in such collisionless plasmas. These multiscale problems can span many orders of magnitude in time. Implicit integrators can step over the collisionless time scale; however, they can be expensive and challenging to implement. Analytic approaches reconcile this tension through the technique of bounce-averaging, providing a physics-based way to step over the advective timescale. However, the bounce-average approximation sets the distribution function to zero in the passing region, which is important for characterizing the dynamics of passing particles. This work presents an explicit multiscale time-integration scheme inspired by the bounce-averaged symmetry in these problems. The scheme has been implemented in the gyrokinetic solver in Gkeyll, and results will be presented. Results from single-field-line studies in a high-field magnetic mirror demonstrate equilibria consistent with theory. Specifically, our results agree with theoretical predictions of the dependence of the ion confinement time on the mirror ratio, and the amplitude of the ambipolar potential is consistent with prior theoretical work. Preliminary axisymmetric (2x2v) simulations will also be presented, showcasing Gkeyll’s ability to study new problems.
Characterization: 1.0
Comments:
Greg Hammett, Ammar Hakim, Manaurer Francisquez, James Juno, Antoine Hoffmann, Kolter Bradshaw