Solving multiscale dynamical systems by deep learning
Authors:
- Zhi-Qin John XuZXZhi-Qin John Xu
- Weinan EWEWeinan E
- Tianhan ZhangTZTianhan Zhang
Preprint
Abstract
Multiscale dynamical systems, modeled by high-dimensional stiff ordinary differential equations (ODEs) with wide-ranging characteristic timescales, arise across diverse fields of science and engineering, but their numerical solvers often encounter severe efficiency bottlenecks. This paper introduces a novel DeePODE method, which consists of a global multiscale sampling method and a fitting by deep neural networks to handle multiscale systems. DeePODE's primary contribution is to address the multiscale challenge of efficiently uncovering representative training sets by combining the Monte Carlo method and the ODE system's intrinsic evolution without suffering from the ``curse of dimensionality''. The DeePODE method is validated in multiscale systems from diverse areas, including a predator-prey model, a power system oscillation, a battery electrolyte auto-ignition, and turbulent flames. Our methods exhibit strong generalization capabilities to unseen conditions, highlighting the power of deep learning in modeling intricate multiscale dynamical processes across science and engineering domains.Keywords
All Related Versions
This publication has 33 references indexed in Scilit:
- A multi-scale sampling method for accurate and robust deep neural network to predict combustion chemical kineticsCombustion and Flame, 2022
- DLBFoam: An open-source dynamic load balancing model for fast reacting flow simulations in OpenFOAMComputer Physics Communications, 2021
- Machine learning tabulation of thermochemistry in turbulent combustion: An approach based on hybrid flamelet/random data and multiple multilayer perceptronsCombustion and Flame, 2021
- Mathematical Modeling and Analysis of Spatial Neuron Dynamics: Dendritic Integration and BeyondCommunications on Pure and Applied Mathematics, 2021
- On-the-fly artificial neural network for chemical kinetics in direct numerical simulations of premixed combustionCombustion and Flame, 2021
- Linear eddy mixing based tabulation and artificial neural networks for large eddy simulations of turbulent flamesCombustion and Flame, 2010
- Library-based numerical reduction of the Hodgkin–Huxley neuron for network simulationJournal of Computational Neuroscience, 2009
- Artificial neural network implementation of chemistry with pdf simulation of H2/CO2 flamesCombustion and Flame, 1996
- Neural‐network‐based approximations for solving partial differential equationsCommunications in Numerical Methods in Engineering, 1994
- Simplifying chemical kinetics: Intrinsic low-dimensional manifolds in composition spaceCombustion and Flame, 1992