/Sophon.jl

Efficient, Accurate, and Streamlined Training of Physics-Informed Neural Networks

Primary LanguageJuliaMIT LicenseMIT

Sophon

Stable Dev Build Status Coverage DOI

Sophon.jl provides specialized neural networks and neural operators for Physics-informed machine learning.

Use the documentation to explore the features.

Please star this repository if you find it useful.

Installation

To install Sophon, please open Julia's interactive session (REPL) and press ] key in the REPL to use the package mode, then type the following command

pkg> add Sophon

Examples

More examples can be found in Sophon-Examples.

Gallery

Function Fitting Multi-scale Poisson Equation Convection Equation Helmholtz Equation
Allen-Cahn Equation Schrödinger Equation L-shaped Domain SOD Shock Tube

🏠 Stable & Mature Framework

The current version is highly stable and you can confidently use it for your projects. In the immediate future, we will not be introducing new features to the framework. Our focus is on maintaining its stability and robustness, ensuring that it stands resilient in various use-cases and scenarios.

We're here to help you with any issues or challenges you might encounter. Feel free to reach out!

Related Libraries

What's the difference between this package and NeuralPDE.jl?

The biggest difference is the explicit control over data sampling. Note that we have an example of an L-shape domain, and there is an example of a disk with a hole in this file.