Packages for general use

• MolANN
  • This package implements PyTorch artificial neural network classes for molecular applications.
  • It can be used to define neural network functions that are invariant under rotation and translation, or functions that take molecular features as inputs.
  • Documentation and installation on this page.
• TorchANN-Plumed  
  • This is a plugin for package PLUMED that implements the TorchFunc function class and the TorchColVar class.
  • This package is useful as it allows one to define functions and collective variables in PLUMED that are represented by artificial neural networks.
• Colvars Finder  
  • This Python package implements algorithms that allow to identify collective variables of dynamical systems using neural networks.
  • The collective variables are found by training autoencoders or computing eigenfunctions of the system's generator.
  • Documentation and installation on this page.

Packages from past research

• EigenPDE-NN
  • Python package for solving eigenvalue PDE problems by training neural networks.
• Constrained-HMC
  • Julia package that implements a hybrid Monte Carlo (HMC) sampler on submanifolds.

More on github page.