Abstract:Throughout many fields, practitioners often rely on differential equations to model systems. Yet, for many applications, the theoretical derivation of such equations and/or accurate resolution of their solutions may be intractable. Instead, recently developed methods, including those based on parameter estimation, operator subset selection, and neural networks, allow for the data-driven discovery of both ordinary and partial differential equations (PDEs), on a spectrum of interpretability. The success of these strategies is often contingent upon the correct identification of representative equations from noisy observations of state variables and, as importantly and intertwined with that, the mathematical strategies utilized to enforce those equations. Specifically, the latter has been commonly addressed via unconstrained optimization strategies. Representing the PDE as a neural network, we propose to discover the PDE by solving a constrained optimization problem and using an intermediate state representation similar to a Physics-Informed Neural Network (PINN). The objective function of this constrained optimization problem promotes matching the data, while the constraints require that the PDE is satisfied at several spatial collocation points. We present a penalty method and a widely used trust-region barrier method to solve this constrained optimization problem, and we compare these methods on numerical examples. Our results on the Burgers' and the Korteweg-De Vreis equations demonstrate that the latter constrained method outperforms the penalty method, particularly for higher noise levels or fewer collocation points. For both methods, we solve these discovered neural network PDEs with classical methods, such as finite difference methods, as opposed to PINNs-type methods relying on automatic differentiation. We briefly highlight other small, yet crucial, implementation details.

MORL4PDEs: Data-driven discovery of PDEs based on multi-objective optimization and reinforcement learning

DISCOVER: Deep Identification of Symbolically Concise Open-Form Partial Differential Equations Via Enhanced Reinforcement Learning

DISCOVER: Deep identification of symbolically concise open-form PDEs via enhanced reinforcement-learning

Physics-constrained robust learning of open-form partial differential equations from limited and noisy data

DL-PDE: Deep-learning based data-driven discovery of partial differential equations from discrete and noisy data

Monte Carlo Neural PDE Solver for Learning PDEs via Probabilistic Representation

Constrained or Unconstrained? Neural-Network-Based Equation Discovery from Data

Neural-PDE: a RNN Based Neural Network for Solving Time Dependent PDEs

PDE-READ: Human-readable partial differential equation discovery using deep learning

Physics-Guided Discovery of Highly Nonlinear Parametric Partial Differential Equations

Meta-Auto-Decoder for Solving Parametric Partial Differential Equations

Meta-PDE: Learning to Solve PDEs Quickly Without a Mesh

Symbolic Physics Learner: Discovering governing equations via Monte Carlo tree search

Deep-learning of parametric partial differential equations from sparse and noisy data

DLGA-PDE: Discovery of PDEs with incomplete candidate library via combination of deep learning and genetic algorithm

Discovering Physics-Informed Neural Networks Model for Solving Partial Differential Equations through Evolutionary Computation

Identification of Partial Differential Equations from Noisy Data with Integrated Knowledge Discovery and Embedding Using Evolutionary Neural Networks

M$^{2}$M: Learning controllable Multi of experts and multi-scale operators are the Partial Differential Equations need

Symbolic Genetic Algorithm for Discovering Open-Form Partial Differential Equations (SGA-PDE)

M^2M: Learning Controllable Multi of Experts and Multi-Scale Operators Are the Partial Differential Equations Need