The file for the diffusion map code is src/diffusion_map.py. This uses scikitlearn extensively and is based largely on the "pydiffmap" library.
The examples available in rough .ipynb form are:
- a 2d toy system with position-dependent diffusion, see 2D_example.ipynb.
- a 2d molecular dynamics example, alanine dipeptide with two dihedral angles, see aladip_2var.ipynb
- a 4d molecular dynamics example, alanine dipeptide with four dihedral angles, see aladip_4var.ipynb
There are python scripts available for the `Ksum test' on the datasets, Ksumtest_datasetname.py
- Step-by-step process for alanine dipeptide in vacuum data (at current state these aren't runnable by themselves, they currently serve as a reference):
- Gromacs simulation files for alanine dipeptide in vacuum (see Some_Gromacs_Files), used from Plumed Masterclass 21.6
- Simulation with a static metadynamics bias (see Some_Plumed_Files)
- Free energy computation on
$\Phi, \Psi$ grid with metadynamics (see Some_Plumed_Files) - Restrained Simulations for diffusion matrices on grid (for finite element method validation) and on trajectory (see multirun files)
- Note: The plumed used for this was a custom plumed recompiled particularly with an adjusted DUMPPROJECTIONS (plumed documentation) in plumed/src/core/value.cpp so that it computes with the mass matrix as in equation (6) of our tm-mmap paper.
- Cleaning up the .ipynb for readability