ViscNet is a machine learning model that can predict the temperature-dependency of viscosity of oxide liquids and the fragility index and the glass transition temperature.
Python 3.6+ is required to run the code. The recommended procedure is to create a new virtual environment and install the necessary modules by running
pip install -r requirements.txt- models.py: class to build the models and function to train them.
- data.py: reads the data and splits it.
- train.py: train ViscNet, ViscNet-Huber, and ViscNet-VFT models.
- cross-validation.py: computes the cross-validation metrics.
- metrics.py: compute the metrics of the ViscNet, ViscNet-Huber, and ViscNet-VFT models.
- plots.py: generate the plots to check the performance of the models.
If you find bugs or have questions, please open an issue. PRs are most welcome.
Cassar, D.R. (2021). ViscNet: Neural network for predicting the fragility index and the temperature-dependency of viscosity. Acta Materialia 206, 116602.
Portions of the data from these databases are used and available in this repository:
- SciGlass Copyright (c) 2019 EPAM Systems
- matminer Copyright (c) 2015, The Regents of the University of California
- mendeleev Copyright (c) 2015 Lukasz Mentel
ViscNet, a machine learning model to predict viscosity. Copyright (C) 2020-2021 Daniel Roberto Cassar
This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.