Machine Learning of Space-Fractional Differential Equations

DOI10.1137/18M1204991MaRDI QIDQ5230657zbMATH OpenFDO

Authors Mamikon Gulian, Maziar Raissi, Paris Perdikaris, George Em Karniadakis

Publication date 28 August 2019

Published in SIAM Journal on Scientific Computing (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1808.00931

zbMATH Keywords

anomalous diffusionfractional diffusionGaussian processesmatérn kernelstable process

Mathematics Subject Classification ID

Fractional partial differential equations (35R11) Gaussian processes (60G15) Stable stochastic processes (60G52) Bayesian inference (62F15) Time series, auto-correlation, regression, etc. in statistics (GARCH) (62M10) Numerical methods for inverse problems for boundary value problems involving PDEs (65N21)

Abstract: Data-driven discovery of "hidden physics" -- i.e., machine learning of differential equation models underlying observed data -- has recently been approached by embedding the discovery problem into a Gaussian Process regression of spatial data, treating and discovering unknown equation parameters as hyperparameters of a modified "physics informed" Gaussian Process kernel. This kernel includes the parametrized differential operators applied to a prior covariance kernel. We extend this framework to linear space-fractional differential equations. The methodology is compatible with a wide variety of fractional operators in $mathbb{R}^d$ and stationary covariance kernels, including the Matern class, and can optimize the Matern parameter during training. We provide a user-friendly and feasible way to perform fractional derivatives of kernels, via a unified set of d-dimensional Fourier integral formulas amenable to generalized Gauss-Laguerre quadrature. The implementation of fractional derivatives has several benefits. First, it allows for discovering fractional-order PDEs for systems characterized by heavy tails or anomalous diffusion, bypassing the analytical difficulty of fractional calculus. Data sets exhibiting such features are of increasing prevalence in physical and financial domains. Second, a single fractional-order archetype allows for a derivative of arbitrary order to be learned, with the order itself being a parameter in the regression. This is advantageous even when used for discovering integer-order equations; the user is not required to assume a "dictionary" of derivatives of various orders, and directly controls the parsimony of the models being discovered. We illustrate on several examples, including fractional-order interpolation of advection-diffusion and modeling relative stock performance in the S&P 500 with alpha-stable motion via a fractional diffusion equation.

Recommendations

• scientific article; zbMATH DE number 7604708
• Machine learning for semi linear PDEs
• A novel numerical scheme for fractional differential equations using extreme learning machine
• Integral equations and machine learning
• Machine learning approximation algorithms for high-dimensional fully nonlinear partial differential equations and second-order backward stochastic differential equations
• Machine learning for fast and reliable solution of time-dependent differential equations
• Machine learning of linear differential equations using Gaussian processes
• An exploratory study on machine learning to couple numerical solutions of partial differential equations
• Learning parameters of a system of variable order fractional differential equations
• Adaptive Learning Neural Network Method for Solving Time–Fractional Diffusion Equations

Cites work

• Asymptotic Statistics
• Automated reverse engineering of nonlinear dynamical systems
• Deep hidden physics models: deep learning of nonlinear partial differential equations
• Discovering governing equations from data by sparse identification of nonlinear dynamical systems
• Discovering variable fractional orders of advection-dispersion equations from field data using multi-fidelity Bayesian optimization
• Fast Numerical Solution of Parabolic Integrodifferential Equations with Applications in Finance
• Financial Modelling with Jump Processes
• Gaussian processes for machine learning.
• Hidden physics models: machine learning of nonlinear partial differential equations
• scientific article; zbMATH DE number 1639862 (Why is no real title available?)
• Inferring solutions of differential equations using noisy multi-fidelity data
• Interpolation of spatial data. Some theory for kriging
• Inverse problems: a Bayesian perspective
• Machine learning of linear differential equations using Gaussian processes
• Numerical Gaussian processes for time-dependent and nonlinear partial differential equations
• Option pricing beyond Black-Scholes based on double-fractional diffusion
• Paul Wilmott introduces quantitative finance. With CD-ROM
• Pricing Bermudan options in Lévy process models
• Stochastic models for fractional calculus
• The Bayesian formulation and well-posedness of fractional elliptic inverse problems
• The random walk's guide to anomalous diffusion: A fractional dynamics approach

Cited in

(25)

• Machine learning of linear differential equations using Gaussian processes
• Integral equations and machine learning
• An ergodic approach to Laplace transforms on time scales
• Towards a unified theory of fractional and nonlocal vector calculus
• Symplectic neural networks in Taylor series form for Hamiltonian systems
• An optimization-based approach to parameter learning for fractional type nonlocal models
• Error estimates for deep learning methods in fluid dynamics
• Discovery of subdiffusion problem with noisy data via deep learning
• Efficient optimization-based quadrature for variational discretization of nonlocal problems
• Isogeometric collocation method for the fractional Laplacian in the 2D bounded domain
• Topology optimization based on deep representation learning (DRL) for compliance and stress-constrained design
• Machine learning applied to pattern characterization in spatially extended dynamical systems
• scientific article; zbMATH DE number 7604708 (Why is no real title available?)
• APIK: Active Physics-Informed Kriging Model with Partial Differential Equations
• A cookbook for approximating Euclidean balls and for quadrature rules in finite element methods for nonlocal problems
• fPINNs: Fractional Physics-Informed Neural Networks
• Analysis of Anisotropic Nonlocal Diffusion Models: Well-Posedness of Fractional Problems for Anomalous Transport
• Efficient Monte Carlo Method for Integral Fractional Laplacian in Multiple Dimensions
• A general framework for substructuring‐based domain decomposition methods for models having nonlocal interactions
• Efficient quadrature rules for finite element discretizations of nonlocal equations
• Discovery of PDEs driven by data with sharp gradient or discontinuity
• Bounds on the approximation error for deep neural networks applied to dispersive models: nonlinear waves
• An implicit-explicit Monte Carlo method for semi-linear PDEs driven by -stable Lévy process and its error estimates
• Learning epidemic trajectories through kernel operator learning: from modelling to optimal control
• On the post-hoc Evaluation of PDE Discovery: A Multifaceted Challenge of Scientific Advancement

Describes a project that uses

Uses Software

• PDE-Net

This page was built for publication: Machine Learning of Space-Fractional Differential Equations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5230657)