Interpreting neural network models of residual scalar flux

Abstract

We show that in addition to providing effective and competitive closures, when analysed in terms of dynamics and physically-relevant diagnostics, artificial neural networks (ANNs) can be both interpretable and provide useful insights in the on-going task of developing and improving turbulence closures. In the context of large-eddy simulations (LES) of a passive scalar in homogeneous isotropic turbulence, exact subfilter fluxes obtained by filtering direct numerical simulations (DNS) are used both to train deep ANN models as a function of filtered variables, and to optimise the coefficients of a turbulent Prandtl number LES closure. \textit{A-priori} analysis of the subfilter scalar variance transfer rate demonstrates that learnt ANN models out-perform optimised turbulent Prandtl number closures and Clark-type gradient models. Next, \textit{a-posteriori} solutions are obtained with each model over several integral timescales. These experiments reveal, with single- and multi-point diagnostics, that ANN models temporally track exact resolved scalar variance with greater accuracy compared to other subfilter flux models for a given filter length scale. Finally, we interpret the artificial neural networks statistically with differential sensitivity analysis to show that the ANN models feature dynamics reminiscent of so-called "mixed models", where mixed models are understood as comprising both a structural and functional component. Besides enabling enhanced-accuracy LES of passive scalars henceforth, we anticipate this work to contribute to utilising neural network models as a tool in interpretability, robustness and model discovery.

References in corpus (6)

• Perspectives on Machine Learning-augmented Reynolds-averaged and Large Eddy Simulation Models of Turbulence
• A neural network approach for the blind deconvolution of turbulent flows
• Deconvolutional artificial neural network models for large eddy simulation of turbulence
• A priori analysis on deep learning of subgrid-scale parameterizations for Kraichnan turbulence
• Asymptotic dynamics of high dynamic range stratified turbulence
• Neural network-based modelling of unresolved stresses in a turbulent reacting flow with mean shear

Cited by in corpus (16)

• Stable a posteriori LES of 2D turbulence using convolutional neural networks: Backscattering analysis and generalization to higher Re via transfer learning
• On the role of vorticity stretching and strain self-amplification in the turbulence energy cascade
• Attention-Enhanced Neural Network Models for Turbulence Simulation
• A posteriori learning for quasi-geostrophic turbulence parametrization
• Learning physics-constrained subgrid-scale closures in the small-data regime for stable and accurate LES
• Deep reinforcement learning for large-eddy simulation modeling in wall-bounded turbulence
• Physical invariance in neural networks for subgrid-scale scalar flux modeling
• Tempered Fractional LES Modeling
• A physics-inspired alternative to spatial filtering for large-eddy simulations of turbulent flows
• Validation and parameterization of a novel physics-constrained neural dynamics model applied to turbulent fluid flow
• A Parallel Integrated Computational-Statistical Platform for Turbulent Transport Phenomena
• Dynamic Nonlocal Passive Scalar Subgrid-Scale Turbulence Modeling
• Network Compression for Machine-Learnt Fluid Simulations
• Multigrid Solver With Super-Resolved Interpolation
• Probabilistic neural networks for predicting energy dissipation rates in geophysical turbulent flows
• A comparison of data-driven approaches to build low-dimensional ocean models