For more details check our publication.
Linear Multi-fidelity model based on Gaussian Process regression can be used to combine qualitative descriptioons known from simulations and quantitative accuracy from a few simulations to obtain something thats better than both on their own.
Shear viscosities from a few molecular dynamics simulations as well as a few experimental shear viscosities enable a high-quality prediction of this transport property over a large range of thermodynamic state points.
The model relies on the univariate relationship between the reduced shear viscosity and the reduced residual entropy allowing the molecular simulation to extrapolate to regions not covered by the experimental training data.
builds on:
Core of this repository. The entire approach can be reproduced using this notebook.
Creates a reference based on all available data. Intended for comparison with the actual low-data approach.
Contains .py files with main code base of the project
Implementation of the core functions of our entropy scaling approach.
Useful functions that build on entropy scaling and multi-fidelity approach.
Contains all simulation results in a excel sheet
Contains pc-saft parameter files for FeOs
Contains result plots and experimental data for the respective species
If you find this repo useful for your own scientific studies, consider citing our publication accompanying this library.
@article{fleck2024multifidelity,
title={Multifidelity Gaussian Processes for Predicting Shear Viscosity over Wide Ranges of Liquid State Points Based on Molecular Dynamics Simulations},
author={Fleck, Maximilian and Gross, Joachim and Hansen, Niels},
journal={Industrial \& Engineering Chemistry Research},
year={2024},
publisher={ACS Publications}
}