Exploring plasma turbulence with a gyrokinetic moment-based approach: from the core to the edge, from reduced fluid to full gyrokinetic modelling

We present the first detailed analysis of local delta-f nonlinear gyrokinetic (GK) simulations based on a gyromoment (GM) approach [1], which exploits the projection of the distribution functions onto a Hermite-Laguerre velocity-space space basis to solve the GK equation [2,3]. We first demonstrate that, in contrast to gyrofluid models, the GM approach reproduces the Dimits shift, i.e. the nonlinear upshift in the gradient required for the onset of significant turbulent transport with respect to the linear growth rate [4]. Notably, we report that the width of the shift can be retrieved with a coarser velocity space resolution than state-of-the-art continuum GK codes, such as GENE. The convergence properties of the GM moment further improves with the gradient strength and collisionality. In addition, we reveal that the choice of collision operator model (Dougherty, Sugama, Lorentz and Landau [5]) significantly impacts the level of turbulent transport, in particular through zonal flow damping, questioning the use of simplified collision operators [6]. The full potential of the GM approach is then shown by discussing tokamak relevant edge turbulence simulations [7]. Finally, our results demonstrate, for the first time, the unique multi-fidelity property of the GM approach, which bridges the gap between fully GK and reduced fluid models [8], retrieving fluid results at low velocity space resolution and GK results at high velocity space resolution.