GPT-PINN: Generative Pre-Trained Physics-Informed Neural Networks toward non-intrusive Meta-learning of parametric PDEs

Physics-Informed Neural Network (PINN) has proven itself a powerful tool to obtain the numerical solutions of nonlinear partial differential equations (PDEs) leveraging the expressivity of deep neural networks and the computing power of modern heterogeneous hardware. However, its training is still t...

Full description

Saved in:
Bibliographic Details
Published in:Finite elements in analysis and design Vol. 228; p. 104047
Main Authors: Chen, Yanlai, Koohy, Shawn
Format: Journal Article
Language:English
Published: Elsevier B.V 01.01.2024
Subjects:
Meta-learning
model order reduction
Network of networks
Non-intrusive learning
Parametric PDEs
Physics-Informed Neural Networks
Physics-Informed Neural Networks
Parametric PDEs
Non-intrusive learning
model order reduction
Meta-learning
Network of networks
ISSN:0168-874X, 1872-6925
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Physics-Informed Neural Network (PINN) has proven itself a powerful tool to obtain the numerical solutions of nonlinear partial differential equations (PDEs) leveraging the expressivity of deep neural networks and the computing power of modern heterogeneous hardware. However, its training is still time-consuming, especially in the multi-query and real-time simulation settings, and its parameterization often overly excessive. In this paper, we propose the Generative Pre-Trained PINN (GPT-PINN) to mitigate both challenges in the setting of parametric PDEs. GPT-PINN represents a brand-new meta-learning paradigm for parametric systems. As a network of networks, its outer-/meta-network is hyper-reduced with only one hidden layer having significantly reduced number of neurons. Moreover, its activation function at each hidden neuron is a (full) PINN pre-trained at a judiciously selected system configuration. The meta-network adaptively “learns” the parametric dependence of the system and “grows” this hidden layer one neuron at a time. In the end, by encompassing a very small number of networks trained at this set of adaptively-selected parameter values, the meta-network is capable of generating surrogate solutions for the parametric system across the entire parameter domain accurately and efficiently. •The use of whole pre-trained networks as the activation functions of another network•A network of networks that is adaptively generated•The adoption of the training loss of the meta-network as an error indicator•The adaptation of the classical greedy algorithm to the neural network setting
ISSN:0168-874X
1872-6925
DOI:10.1016/j.finel.2023.104047