Abstract Details
Abstracts
Author: Federico D Halpern
Requested Type: Poster
Submitted: 2025-03-04 16:06:17
Co-authors: M.-G. Yoo, B.C. Lyons, J.D. Colmenares
Contact Info:
General Atomics
P.O. Box 85608
San Diego, 92186
United States
Abstract Text:
Diffusive transport processes in magnetized plasmas are highly anisotropic, with fast parallel transport along the magnetic field lines faster than perpendicular transport by several orders of magnitude. We describe and validate a new numerical approach for parallel diffusion in magnetized plasmas based on the anti-symmetry representation [Halpern and Waltz, Phys. Plasmas 25, 060703 (2018)]. In the anti-symmetry formalism, diffusion manifests as a flow operator involving the logarithmic derivative of the transported quantity. Qualitative plane wave analysis shows that the new operator naturally yields extended discrete spectral resolution compared to its conventional counterpart. We carry out 2D and 3D simulations to assess the numerical convergence of the new method, and comparing directly against existing finite difference methods. Numerical routines are constructed using Python-based symbolic mathematics to generate tedious finite difference stencil expressions consistently, without errors, and at any precision order. The resulting symbolic expressions are translated into GPU-accelerated Fortran modules and are imported into the ALMA numerical engine. Convergence tests using these routines show that the new anti-symmetry diffusion method produces significantly less error while also more precisely retaining the field-aligned property of the operator. Combining anti-symmetry with finite differences in diagonally staggered grids essentially eliminates the so-called "artificial numerical diffusion" that affects conventional finite difference and finite volume methods. --Acknowledgement: This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Fusion Energy Sciences, Theory Program, under Award DE-FG02-95ER54309.
Characterization: 4.0
Comments: