Minisymposium

Understanding and predicting turbulence-driven transport across the edge and scrape-off layer (SOL) are crucial for the success of magnetic confinement fusion devices. However, the challenging plasma conditions in the edge and SOL require first-principles gyrokinetic (GK) simulations. To address these challenges, the GENE-X code has been developed to solve the full-f GK model in magnetic geometries with X-points. While GENE-X simulations provide an accurate description of edge and SOL turbulence, their computational cost remains prohibitively high, often requiring millions of CPU hours and spanning over weeks or months. Although GPUs and exascale architectures offer potential speed-ups, further advances in numerical algorithms are essential.We present a novel velocity-space spectral method, recently implemented in the GENE-X code for the first time. The numerical implementation is rigorously verified, and performance benchmarks are conducted. Through a detailed analysis of spectral simulations, we demonstrate excellent agreement with Eulerian (grid-based) GENE-X simulations, validated against experimental measurements. This allows us to identify the minimum spectral resolution required to achieve accurate results, comparable to those from grid-based simulations. We find that the spectral approach enables a significant computational speed-up, reducing the velocity-space resolution by a factor of 50. The first applications in reactor-relevant scenarios will be presented.

Gyselalib++ is a portable, GPU-accelerated C++ library designed for high-performance gyrokinetic semi-Lagrangian simulations. It uses Kokkos to ensure performance portability across diverse hardware architectures, including modern multi-core CPUs and GPUs, making it well-suited for exascale computing. Additionally, Gyselalib++ makes use of DDC (Discrete Domain Computation library), a library that provides a framework for strongly typing mathematical concepts. Gyselalib++ is the result of a rewriting of GYSELA, a Gyrokinetic Semi-Lagrangian code written in Fortran. While the original code was highly optimised to run petascale simulations, the lack of modularity makes it difficult to add non-trivial extensions, such as X-point geometries, to the code. It was also difficult to optimise for new GPU architectures. This talk will introduce the design and capabilities of Gyselalib++, including its approach to parallelism, memory management, and performance optimisation for large-scale gyrokinetic modelling. We will present benchmarking results on a 4D simulation and discuss ongoing work to extend the capabilities of the library. In particular, we will highlight early developments toward a patch-based approach for handling complex magnetic field geometries, such as those found near the X-point in fusion devices.

The construction of the ITER tokamak is motivating a whole field of research concerning the influence of the tungsten wall on the fusion plasma efficiency. Tungsten ions are sputtered from the plasma facing components and contaminate the plasma where they radiate a significant amount of its energy. This can lead to a radiative collapse, as observed in the WEST tokamak. Understanding this physics requires modeling the different ionization states and radiation levels of each species in the fusion device. Atomic interactions calculations, such as ionization and recombination, provides this information but the delta-f or total-f modeling of all the ionization states of the different impurity species is still out of computational capabilities, even on exascale supercomputers. Instead we introduced a bundling technique that regroups ions of similar charge together while accounting for their bundled atomic physics. This bundling technique has been implemented in a modular way to be integrated in different transport codes. Application to the WEST and ASDEX-U tokamak with the neoclassical transport code FACIT and with the 5D gyrokinetic code XGC will be presented. This includes the whole device simulation of ASDEX-U plasma with multiple low-Z impurities and bundles of tungsten ions.

In current fusion reactors the heat and particle loads from the scrape-off-layer (SOL) plasma to the plasma-facing components (PFCs) are one of the main limiting factors in the process of designing devices and operational scenarios for future fusion power plants. The transport into the SOL plasma is dominated by filamented structures that penetrate the last closed flux surface, the so-called blobs. Blobs are coherent structures of high density and high temperature plasma which travel through SOL in directions parallel and perpendicular to the magnetic field. The perpendicular transport can mostly be adequately described using fluid codes, but the parallel transport requires kinetic approach, since the parallel energy distribution function is often not Maxwellian. We have used a fully-kinetic 1d3v code BIT1 to simulate a flux tube in a tokamak SOL where the particle and energy source consists of a series of blobs injected at the outer midplane. We have parametrically studied the influence of source properties on the self-consistent development of SOL and target plasma. The simulations show significant effects of source temperature and blob size on the SOL particle and energy transport.