Computing Residual Diffusivity by Adaptive Basis Learning via Super-Resolution Deep Neural Networks

It is expensive to compute residual diffusivity in chaotic in-compressible flows by solving advection-diffusion equation due to the formation of sharp internal layers in the advection dominated regime. Proper orthogonal decomposition (POD) is a classical method to construct a small number of adaptive orthogonal basis vectors for low cost computation based on snapshots of fully resolved solutions at a particular molecular diffusivity $D_{0}^{*}$. The quality of POD basis deteriorates if it is applied to $D_0\ll D_{0}^{*}$. To improve POD, we adapt a super-resolution generative adversarial deep neural network (SRGAN) to train a nonlinear mapping based on snapshot data at two values of $D_{0}^{*}$. The mapping models the sharpening effect on internal layers as $D_0$ becomes smaller. We show through numerical experiments that after applying such a mapping to snapshots, the prediction accuracy of residual diffusivity improves considerably that of the standard POD.

[1]  A. Majda,et al.  SIMPLIFIED MODELS FOR TURBULENT DIFFUSION : THEORY, NUMERICAL MODELLING, AND PHYSICAL PHENOMENA , 1999 .

[2]  L. Sirovich Turbulence and the dynamics of coherent structures. I. Coherent structures , 1987 .

[3]  Jiancheng Lyu,et al.  Computing Residual Diffusivity by Adaptive Basis Learning via Spectral Method , 2017 .

[4]  J. Xin,et al.  Front Quenching in the G-equation Model Induced by Straining of Cellular Flow , 2014 .

[5]  Camassa,et al.  Chaotic advection in a Rayleigh-Bénard flow. , 1991, Physical review. A, Atomic, molecular, and optical physics.

[6]  S. Heinze Diffusion-Advection in Cellular Flows with Large Peclet Numbers , 2003 .

[7]  Jack Xin,et al.  A computational study of residual KPP front speeds in time-periodic cellular flows in the small diffusion limit , 2014, 1406.2268.

[8]  Soumith Chintala,et al.  Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks , 2015, ICLR.

[9]  J. Xin,et al.  Sharp asymptotic growth laws of turbulent flame speeds in cellular flows by inviscid Hamilton–Jacobi models , 2013 .

[10]  A. Bensoussan,et al.  Asymptotic analysis for periodic structures , 1979 .

[11]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[12]  George Papanicolaou,et al.  Convection Enhanced Diffusion for Periodic Flows , 1994, SIAM J. Appl. Math..

[13]  S. Childress,et al.  Stretch, Twist, Fold , 1995 .

[14]  A. Crisanti,et al.  Eddy diffusivities in scalar transport , 1995 .

[15]  Geoffrey Ingram Taylor,et al.  Diffusion by Continuous Movements , 1922 .

[16]  Gianluigi Rozza,et al.  Reduced Order Methods for Modeling and Computational Reduction , 2013 .

[17]  Christian Ledig,et al.  Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[18]  L. Ryzhik,et al.  Boundary Layers and KPP Fronts in a Cellular Flow , 2007 .

[19]  Jack Xin,et al.  Computing Effective Diffusivity of Chaotic and Stochastic Flows Using Structure-Preserving Schemes , 2017, SIAM J. Numer. Anal..