Asif Hamid, Danish Rafiq, Shahkar A. Nahvi and M. A. Bazaz
This repository is to help the users reproduce the results presented in "Hierarchical deep learning based adaptive time-stepping of multiscale systems", Eng Appl Artif Intell, vol. 133 part D, 2024
- Clone the entire directory
git clone https://github.com/DanishRaf32/Adaptive-HiTS.git
conda create -n <ENV_NAME> python=3.7
conda activate <ENV_NAME>
conda install pytorch torchvision -c pytorch
pip install -r requirements.txt
To allow tqdm (the progress bar library) to run in a notebook, you also need:
conda install -c conda-forge ipywidgets
- Generate data for all benchmark systems by running the script "scripts/data_generation.ipynb" (for the KS system, the data is already available)
- Train models for all benchmark systems via script "scritps/model_training.ipynb" (I have already provided the models for quick run, but new models can also be trained)
- Finally, run the file "scripts/adaptive-HiTs.ipynb"
For any technical issues, feel free to contact danishrafiq32@gmail.com
[1] Yuying Liu, Nathan Kutz, and Steven Brunton, Hierarchical deep learning of multiscale differential equation time-steppers Phil. Trans. R. Soc. A., (380), 2022.