This repository contains code, data sets and models corresponding to the following publication:
Neural functional theory for inhomogeneous fluids: Fundamentals and applications
Florian Sammüller, Sophie Hermann, Daniel de las Heras, and Matthias Schmidt, Proc. Natl. Acad. Sci. 120, e2312484120 (2023); arXiv:2307.04539.
Working in a virtual environment is recommended.
Set one up with python -m venv .venv, activate it with source .venv/bin/activate and install the required packages with pip install -r requirements.txt.
To use a GPU with Tensorflow/Keras, refer to the corresponding section in the installation guide at https://www.tensorflow.org/install/pip.
Simulation data can be found in data and trained models are located in models.
A sample script for training a model from scratch is given in learn.py.
The usage of a trained model, e.g. for the self-consistent calculation of density profiles, is illustrated in neuraldft.py.
Some useful utilities are provided in utils.py, such as tools for functional calculus as well as data generators and callbacks for training.
The reference data has been generated with grand canonical Monte Carlo simulations using MBD. The analytic DFT calculations with fundamental measure theory have been performed with the Julia library ClassicalDFT.jl.