This repository contains the dataset generation and network training code for the stability margin network detailed in the work of Orsolino. et al. 2021.
This guide assumes that $PROJECT_ROOT refers to the absolute path to this repository.
To setup a virtual environment inside the project directory:
sudo apt install -y python-venv
python -m venv $PROJECT_ROOT/venv
source $PROJECT_ROOT/venv/bin/activate
Now, you are ready to use the virtual environment. To install the dependencies related to this project, make sure you activate the environment and then run:
cd $PROJECT_ROOT/jet_leg_common
pip install -e .
And then
cd $PROJECT_ROOT
pip install -e .
If everything went well, you should be able to run the data generation script
data_generation/stability_margin.py, assuming you have activated the virtual environment,
like so:
cd $PROJECT_ROOT/scripts
python data_generation/stability_margin.py
If you want to generate a large amount of data, run
cd $PROJECT_ROOT/scripts/data_generation
./repeat_generation.sh
The script will repeatedly run the data generating script data_generation/stability_margin.py.
This avoids running the data generating script once with a large amount of data, filling up the RAM in the process.
The generated dataset will be stored in the $PROJECT_ROOT/data directory.
After generating the training dataset, you can start the training using the script
training/stability_margin.py:
cd $PROJECT_ROOT/scripts
python training/stability_margin.py
Use the script
utility/model_to_txt_converter.py to save the trained model parameters to a txt file for
use with the custom C++ MLP implementation:
cd $PROJECT_ROOT/scripts
python utility/model_to_txt_converter.py
- Siddhant Gangapurwala siddhant@gangapurwala.com
- Abdelrahman Abdalla abdulrahman.h.abdallah@gmail.com
You might need to set the PYTHONPATH to your Conda environment's site-packages
export PYTHONPATH=/path/to/miniconda3/envs/your-env/lib/pythonX.Y/site-packages