Beating the curse of dimension with accurate statistics for the Fokker–Planck equation in complex turbulent systems

Abstract

Solving the Fokker–Planck equation for high-dimensional complex dynamical systems is an important issue. Effective strategies are developed and incorporated into efficient statistically accurate algorithms for solving the Fokker–Planck equations associated with a rich class of high-dimensional nonlinear turbulent dynamical systems with strong non-Gaussian features. These effective strategies exploit a judicious block decomposition of high-dimensional conditional covariance matrices and statistical symmetry to facilitate an extremely efficient parallel computation and a significant reduction of sample numbers. The resulting algorithms can efficiently solve the Fokker–Planck equation in much higher dimensions even with orders in the millions and thus beat the curse of dimension. Skillful behavior of the algorithms is illustrated for highly non-Gaussian systems in excitable media and geophysical turbulence.

Document Details

Document Type
Pub Defense Publication
Publication Date
Nov 20, 2017
Source ID
10.1073/pnas.1717017114

Entities

People

  • Andrew J. Majda
  • Nan Chen

Organizations

  • New York University
  • Office of Naval Research

Tags

Readers

  • Distributed Systems and Data Platform Development
  • Finite Element Method (FEM) for solving Partial Differential Equations (PDEs)
  • Statistical inference.