Progress toward automatic identification of under-resolved gyrokinetic turbulence simulations

As the fusion industry approaches the goal of a fusion pilot plant, minimizing turbulent transport remains an important objective. Workflows using ensembles of turbulence simulations are now computationally feasible due to recent advancements in algorithms and hardware. As these workflows become increasingly automated, it is desirable to have the capability to automatically identify whether simulations are using sufficiently high enough resolution to accurately capture dynamics. As a step toward that goal, we evaluate the use of correlation length [1] and energy transfer [2] diagnostics as indicators of under-resolved simulations. We present the initial results of a survey of gyrokinetic turbulence simulations, implementing in situ diagnostics in the GX code [3] to study correlation lengths and energy transfers in a range of magnetic geometries.

[1] M. Barnes et al. Phys. Rev. Lett. 107, 115003 (2011).

[2] B. Teaca et al. New J. Phys. 19, 045001 (2017).

[3] N. Mandell et al. J. Plasma Phys. 90, 905900402 (2024).