Machine Learning Molecular Dynamics (mlmd)

The machine learning molecular dynamics (mlmd) package (link to the source code) is a command line open source python program. It calculates the feature representation of a structure (molecule, cluster, crystal), using the Structural Information Filter Features (SIFF), in addition, mlmd is also able to perform molecular dynamics simulations with machine learning potentials (also known as machine learning force fields or machine learning empirical potentials).

The mlmd program generates machine learning potentials (mlp) processing the energies and forces of DFT calculations carried out over systems of particles, then uses the trained potential to do NVT molecular dynamics on a structure compatible with the mlp. So far mlmd can use calculations from VASP, Abinit, and Fireball as inputs.

• mlmd Requirements
• Installation
  • Installing mlmd
  • Uninstalling mlmd
• Creating (Training) a Machine Learning Potential
• Performing Molecular Dynamics With a Machine Learning Potential
• Input Files
  • Training File
  • Structure File
  • md File
• Dimension of the Feature Space
  • The Feature Space in the Structural Information Fiter Features
  • The fami_2b and fami_3b
• Tutorials
  • Creating (Training) A Machine Learning Potential With Abinit Data
  • Creating (Training) A Machine Learning Potential With Fireball Data
  • Creating (Training) A Machine Learning Potential With Vasp Data
  • Performing Molecuar Dyanmics With A Machine Leaning Potential
  • Calculate The Structural Information Fiter Features Only

Indices and tables

• Index
• Module Index
• Search Page