PyBias: PLUMED plug-in with embedded Python¶
This project is still under construction!
PLUMED is a library for free energy calculations with molecular dynamics (MD). The calculations can be performed as a function of various collective variables (CV), using state-of-the-art enhanced sampling methods: metadynamics, parallel-bias metadynamics, etc. PLUMED works with most MD engines.
Python environment has a variety of high-quality numerical algorithm (NumPy, SciPy, SimPy, etc.) and machine learning (scikit-learn, Theano, TensorFlow, etc.) libraries, which can be used to compute bias.
- Can use a system or custom Python interpreter
- All installed Python modules can be used
- A built-in Python module to access data from PLUMED
- No PLUMED re-compilation required
If your are lucky (i.e. your system has all dependencies), it is just a 5-step process:
git clone https://github.com/raimis/plumed-pybias.git cd plumed-pybias cmake . -DCMAKE_INSTALL_PREFIX=~/opt/pybias make make install
This is not yet functional!
# plumed.dat # Load PyBias plug-in LOAD FILE=libpybias.so # Create a collective variable dist: DISTANCE ATOMS=1,2 # Creat a PyBias action, which executes "bias.py" PYBIAS ARG=dist ENDPLUMED
# bias.py def bias(coord, force, extra): force[:] = 0. return 0.