Pseudo-Hamiltonian neural networks

This repository contains the package phlearn for modelling pseudo-Hamiltonian ODE and PDE systems with neural networks, and example scripts for training and simulation.

See (Eidnes et al., 2022) for a description of PHNN for ODEs. See (Eidnes and Lye, 2023) for a description of PHNN for PDEs.

Documentation available here

Installation

The phlearn package is available via PyPi:

pip install phlearn

Alternatively, to get the latest updates not yet available on PyPi, you can clone this repository and install directly:

git clone https://github.com/SINTEF/pseudo-Hamiltonian-neural-networks.git
cd pseudo-Hamiltonian-neural-networks
pip install -e phlearn

Authors

Camilla Sterud: c.sterud@icloud.com Sølve Eidnes: solve.eidnes@sintef.no Eivind Bøhn: eivind.bohn@sintef.no Signe Riemer-Sørensen: signe.riemer-sorensen@sintef.no Alexander Stasik: alexander.stasik@sintef.no Kjetil Olsen Lye: kjetil.olsen.lye@sintef.no Ben Tapley: ben.tapley@sintef.no