Minisymposium

Standard derivations of Gyrokinetic theory are not formally compatible with global MHD theory, even in the collisional and small-Larmor radius limit. On the other hand, simpler kinetic theories, like drift-kinetics, can be straightforwardly shown to be consistent with MHD in appropriate limits. By leveraging the known relationships between these theories, I explain how to formulate and solve such kinetic models in a way that allows system-scale MHD motion to be consistently treated. This provides some insight into which system-scale effects are absent in a conventional global gyro-kinetic approach, and how they might be calculated in a kinetic-MHD type framework. As an illustration, we show how certain equilibrium currents are 'missing' when we attempt to reconstruct them via standard gyrokinetics.

Magnetohydrodynamics (MHD) is widely used to study the stability of a given magnetic configuration with respect to potentially problematic machine-scale instabilities. The basic mechanism of these macroscopic modes are well described by this theory. Kinetic effects, through wave-particle interactions, can however significantly affect their stability. Some current kinetic-MHD studies estimating these effects rely on a number of assumptions, typically solving a drift-kinetic equation semi-analytically, which implies strong limitations on the type of orbits that can be reproduced by the model. Other codes integrate guiding-center orbits numerically in the framework of a Lagrangian approach, which requires less assumptions but strongly increases the computational requirements. We present a new spectral linear kinetic-MHD code which solves the MHD momentum equation along with an Eulerian discretization of the drift-kinetic equation. Following the Van Kampen approach, the problem is expressed as a standard linear generalized eigenvalue problem for the MHD displacement as well as the kinetic correction to the perturbed distribution function. The equations are discretized on an effective five-dimensional phase space and result in sparse matrices of very large dimension. The challenges associated with solving this eigenvalue problem on a parallel platform are presented as well as first benchmarks in simplified cylindrical geometry.

The fully kinetic Vlasov-Maxwell coupled to an appropriate collision operator contains all the physics for describing the evolution of a magnetic fusion plasma in a Tokamak or a Stellarator. However, the Vlasov equation is posed in a 6D phase space, so that it requires huge computational resources and very efficient codes on modern supercomputing architectures in particular on GPUs. Moreover, very long time computations are required for most relevant physics problems. For this reasons, keeping at the discrete level the hamiltonian structure of the continuous problem is a huge asset. This can be obtained by a so-called geometric Particle in Cell (PIC) discretization, which discretizes the Poisson bracket and the Hamiltonian leading to the Vlasov-Maxwell equations rather than directly the equations themselves.For an efficient and performance portable implementation we rely on the AMReX framework which offers performance portable data structures for grid based as well as for particle quantities. We will present the latest developments made in the code and the different models that have been implemented.

The ion-temperature-gradient driven instability (ITG) is a prominent challenge in contemporary magnetic fusion experiments. Turbulence driven by ITGs and other instabilities is most commonly simulated using gyrokinetic codes, which exploit the strong magnetization of such plasmas, but likely face limitations in regions with extreme pressure gradients. With the semi-Lagrangian code ssV, we aim to enable fully kinetic simulations for such scenarios, potentially serving as benchmarks. This presentation will offer a comprehensive overview of the numerical challenges encountered, the strategies employed to overcome them, and preliminary findings.