Skip to content

CompPhysVienna/paper_NF_for_rare_events

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Conditioning Normalizing Flows for Rare Event Sampling

Code for the paper on conditioning normalizing flows for the generation of configurations in a parallel path sampling scheme. The publication can be found here:

Falkner, S., Coretti, A., Romano, S., Geissler, P., & Dellago, C. (2022). Conditioning Normalizing Flows for Rare Event Sampling (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2207.14530

Further Reading:

  • Noé, F., Olsson, S., Köhler, J., & Wu, H. (2019). Boltzmann generators: Sampling equilibrium states of many-body systems with deep learning. Science, 365(6457), eaaw1147. https://doi.org/10.1126/science.aaw1147

  • Ardizzone, L., Lüth, C., Kruse, J., Rother, C., & Köthe, U. (2019). Guided Image Generation with Conditional Invertible Neural Networks. ArXiv. http://arxiv.org/abs/1907.02392

  • Dinh, L., Sohl-Dickstein, J., & Bengio, S. (2016). Density estimation using Real NVP. 5th International Conference on Learning Representations, ICLR 2017 - Conference Track Proceedings. http://arxiv.org/abs/1605.08803

Python Distribution

Installation is recommended using a scientific Python distribution such as anaconda (https://www.anaconda.com).

Prerequisites

  1. Python >= 3.6 https://www.python.org

  2. PyTorch https://pytorch.org/

  3. Numba http://numba.pydata.org/

  4. NumPy https://www.numpy.org/

  5. SciPy https://www.scipy.org/

  6. PyYAML https://pyyaml.org/

  7. matplotlib https://matplotlib.org/

For the examples you will also need:

  1. Jupyter https://jupyter.org/

For free energy calculations PyEMMA is required: 8) PyEMMA http://emma-project.org/latest/

Installation

Head to the root folder of the project and type pip install .. The installation script checks automatically for the packages mentioned above except for pytorch and the example dependencies.

PyTorch needs to be installed manually. Instructions can be found at https://pytorch.org/.