ADEPT is an Automatic Differentation Enabled Plasma Transport code.
- Install
conda(we recommendmamba) mamba env create -f env.yamlormamba env create -f env_gpu.yamlmamba activate adept
python3 -m venv venvsource venv/bin/activatepip3 install -r requirements.txt
https://adept.readthedocs.io/en/latest/
python3 run.py --cfg {config_path} without the .yaml extension
This runs the simulation defined in the config and stores the output to an mlflow server.
Unless you have separately deployed an mlflow server somewhere, it simply writes files using the mlflow specification to the current working directory.
To access and visualize the results, it is easiest to use the UI from the browser by typing mlflow ui in the command line from the same directory.
The contributing guide is in development but for now, just make an issue / pull request and we can go from there :)
[1] A. S. Joglekar and A. G. R. Thomas, “Machine learning of hidden variables in multiscale fluid simulation,” Mach. Learn.: Sci. Technol., vol. 4, no. 3, p. 035049, Sep. 2023, doi: 10.1088/2632-2153/acf81a.