Turbulence model reduction by deep learning

Abstract

A central problem of turbulence theory is to produce a predictive model for turbulent fluxes. These have profound implications for virtually all aspects of the turbulence dynamics. In magnetic confinement devices, drift-wave turbulence produces anomalous fluxes via cross-correlations between fluctuations. In this work, we introduce a new, data-driven method for parameterizing these fluxes. The method uses deep supervised learning to infer a reduced mean-field model from a set of numerical simulations. We apply the method to a simple drift-wave turbulence system and find a significant new effect which couples the particle flux to the local \emph{gradient} of vorticity. Notably, here, this effect is much stronger than the oft-invoked shear suppression effect. We also recover the result via a simple calculation. The vorticity gradient effect tends to modulate the density profile. In addition, our method recovers a model for spontaneous zonal flow generation by negative viscosity, stabilized by nonlinear and hyperviscous terms. We highlight the important role of symmetry to implementation of the new method.

To be published in Phys. Rev. E Rap. Comm. 6 pages, 7 figures

References in corpus (5)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• A high-bias, low-variance introduction to Machine Learning for physicists
• BOUT++: a framework for parallel plasma fluid simulations
• Deep Learning is Robust to Massive Label Noise
• Bifurcation in electrostatic resistive drift wave turbulence

Cited by in corpus (8)

• 2022 Review of Data-Driven Plasma Science
• Wave-kinetic approach to zonal-flow dynamics: recent advances
• Scientific Machine Learning Based Reduced-Order Models for Plasma Turbulence Simulations
• Identification of high order closure terms from fully kinetic simulations using machine learning
• Data assimilation empowered neural network parameterizations for subgrid processes in geophysical flows
• Phase and amplitude evolution in the network of triadic interactions of the Hasegawa-Wakatani system
• Deep learning regression for inverse quantum scattering
• Flux-driven turbulent transport using penalisation in the Hasegawa-Wakatani system