Die DPG-Frühjahrstagung in Hannover musste abgesagt werden! Lesen Sie mehr ...

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

P: Fachverband Plasmaphysik

P 5: Poster Session 1

P 5.29: Poster

Montag, 9. März 2020, 16:30–18:30, Empore Lichthof

Viability of NN-based Predictor-Corrector Schemes for Plasma Simulations — •Robin Greif¹, Frank Jenko¹, and Nils Thuerey² — ¹Max-Planck-Institute für Plasmaphysik, Boltzmannstr. 2, 85748 Garching, Germany — ²TU Munich, Boltzmannstr. 3, 85748 Garching, Germany

We investigate the viability of using neural network driven simulation methods based on novel predictor-corrector schemes developed for fluid and smoke simulations for turbulence in plasma. The approach builds on top of successful pioneering work on numerical schemes from Mantaflow and its successor, Phi-Flow, a soon to be published data-driven first framework for fluid and smoke simulations. In this project, we extend Phi-Flow to solve the Hasegawa-Wakatani equations as a proof-of-concept of the viability of modern neural-network based numerical simulation techniques for simple plasma models. The use of deep-learning based numerical integration schemes explored here has been shown to provide superior accuracy at coarser grids than classical methods in fluid simulations and is a promising candidate to reduce the computational cost for the next generation of plasma simulations.