Examples

JCP paper experiments

Experiments on the modified Muller-Brown potential (3 states) reported in the mlcolvar JCP paper (https://doi.org/10.1063/5.0156343), describing different learning scenarios.

• Modified Muller-Brown
  • Unsupervised setting: state discovery
  • Supervised setting: CVs as classifiers
  • Time-lagged setting: improving CVs
  • Multitask learning: a semi-supervised application

Literature examples

Examples of usage of data-driven CVs or other tools from the mlcolvar library in the literature.

• DeepLDA: Alanine dipeptide and aldol reaction
• DeepTICA: Alanine dipeptide from multithermal simulation
• TPI-DeepTDA: Chignolin mini-protein
• Stateinterpreter (characterizing DeepTICA states)
• Learning the committor for Alanine with distances as inputs
• Learning the committor for Tropolone intramolecular proton-transfer with distances as inputs