Machine learning surrogate models for Landau fluid closure

Abstract

The first result of applying the machine/deep learning technique to the fluid closure problem is presented in this letter. As a start, three different types of neural networks (multilayer perceptron (MLP), convolutional neural network (CNN) and two-layer discrete Fourier transform (DFT) network) were constructed and trained to learn the well-known Hammett-Perkins Landau fluid closure in configuration space. We found that in order to train a well-preformed network, a minimum size of training data set is needed; MLP also requires a minimum number of neurons in the hidden layers equals to the degrees of freedom in Fourier space despite training data is fed in configuration space. Out of three models DFT performs the best for the clean data most likely due to the existence of nice Fourier expression for Hammett-Perkins closure but it is least robust with respect to input noise. Overall, with appropriate tuning and optimization, all three neural networks are able to accurately predict Hammett-Perkins closure and reproduce the inherit nonlocal feature, suggesting a promising path to calculate more sophisticated closures with the machine/deep learning technique.

VCNN-e: A vector-cloud neural network with equivariance for emulating Reynolds stress transport equations

“…In a different vein, many recent works [see, e.g. [23][24][25][26][27][28][29][30] used machine learning models to approximate global operators or surrogate models defined by the PDEs, which also hold the promise to nonlocal constitutive modeling. However, the objectivity of these modeling approaches, such as frame-independence and permutational invariance mentioned above, has rarely been discussed.…”

Model order reduction approach to the one-dimensional collisionless closure problem

“…See Goumiri et al 47 for an application to plasma physics. Other methods based on neural networks have also shown promising results 48 .…”

Machine learning moment closure models for the radiative transfer equation III: enforcing hyperbolicity and physical characteristic speeds

“…The traditional trade off in introducing a closure relation and solving a moment model instead of a kinetic equation is generic accuracy verses practical computability. However, thanks to the rapid development of machine learning (ML) and data-driven modeling [6,42,18], a new approach to solve the moment closure problem has emerged based on ML [19,44,25,5,35,48,36,23,24,41,43]. This approach offers a path for multi-scale problems that is relatively unique, promising to capture kinetic effects in a moment model with only a handful of moments.…”