Understanding the Difficulty of Training Physics-Informed Neural Networks on Dynamical Systems

03/25/2022
by   Franz M. Rohrhofer, et al.
0

Physics-informed neural networks (PINNs) seamlessly integrate data and physical constraints into the solving of problems governed by differential equations. In settings with little labeled training data, their optimization relies on the complexity of the embedded physics loss function. Two fundamental questions arise in any discussion of frequently reported convergence issues in PINNs: Why does the optimization often converge to solutions that lack physical behavior? And why do reduced domain methods improve convergence behavior in PINNs? We answer these questions by studying the physics loss function in the vicinity of fixed points of dynamical systems. Experiments on a simple dynamical system demonstrate that physics loss residuals are trivially minimized in the vicinity of fixed points. As a result we observe that solutions corresponding to nonphysical system dynamics can be dominant in the physics loss landscape and optimization. We find that reducing the computational domain lowers the optimization complexity and chance of getting trapped with nonphysical solutions.

READ FULL TEXT

page 2

page 6

page 7

page 8

research
10/28/2022

Physics-Informed Convolutional Neural Networks for Corruption Removal on Dynamical Systems

Measurements on dynamical systems, experimental or otherwise, are often ...
research
07/08/2022

Adaptive Self-supervision Algorithms for Physics-informed Neural Networks

Physics-informed neural networks (PINNs) incorporate physical knowledge ...
research
03/14/2022

Respecting causality is all you need for training physics-informed neural networks

While the popularity of physics-informed neural networks (PINNs) is stea...
research
03/19/2018

Comparing Dynamics: Deep Neural Networks versus Glassy Systems

We analyze numerically the training dynamics of deep neural networks (DN...
research
12/07/2022

Phase2vec: Dynamical systems embedding with a physics-informed convolutional network

Dynamical systems are found in innumerable forms across the physical and...
research
03/24/2023

Physics-informed neural networks in the recreation of hydrodynamic simulations from dark matter

Physics-informed neural networks have emerged as a coherent framework fo...
research
10/12/2018

Grand challenges in social physics: In pursuit of moral behavior

Methods of statistical physics have proven valuable for studying the evo...

Please sign up or login with your details

Forgot password? Click here to reset