QMoDeM: Quantum-aided Models for Decision-Making

Installation

pip install -e .
# or with uv:
uv sync

Running the scripts

Scripts log to MLFlow, a machine learning experiment tracking software. This design choice allows to compare runs easily in the same UI.

Data generation

Run one of the generate_discharge_history.py scripts.

Setup of the environment file (.env)

1. Specify RAW_DATA_DIR, i.e. the location where to write the battery simulation data. For the multiple loading scenarios case, you can do the same with RAW_DATA_DIR_MULTI.
2. Copy-paste the MLFlow run ID to DATA_GEN_RUN_ID (DATA_GEN_RUN_ID_MULTI for the multiple scenarios case).
3. Set MLFLOW_USE_LAST_TRAINED=true if you plan to run the test script right after the corresponding training one. If MLFLOW_USE_LAST_TRAINED=true, the program will prompt you to provide the experiment name and run ID for MLFlow to retrieve the correct run.
4. Set a meaningful MLFLOW_EXPERIMENT_NAME for your numerical campaign.

Training/testing

1. Make sure the data genration run ID is the correct one.
2. Run one of the _train.py scripts to train the network.
3. Run the corresponding test script.

Development

See CONTRIBUTING.md for full guidelines. Quick reference:

uv run pre-commit install            # install hooks (once, before contributing)
uv run pytest                        # run all tests