requirements.txt

# =============================================================================
# requirements.txt — CityLearn HVAC RL Project
# =============================================================================
#
# DO NOT run:  pip install -r requirements.txt
#
# Use the install script instead — it handles platform-specific issues:
#   Windows (Anaconda Prompt):  install.bat
#   macOS / Linux:              bash install.sh
#
# -----------------------------------------------------------------
# Why the install script exists:
#
#   1. CityLearn requires openstudio<=3.3.0, which is unavailable on
#      Windows / Python 3.11 via PyPI. We install with --no-deps and
#      skip openstudio (only needed for EnergyPlus generation, which
#      this project does not use).
#
#   2. CityLearn requires gymnasium<=0.28.1. Stable Baselines3 >= 2.3
#      automatically upgrades gymnasium to 1.x, breaking CityLearn.
#      We pin SB3 to 2.2.1 (last version compatible with gymnasium 0.28).
#
#   3. pip-installed PyTorch often fails with DLL errors in conda
#      environments on Windows. Use conda to install PyTorch instead.
# =============================================================================

# --- PyTorch (install via conda, NOT pip; see install.bat) ---
# conda install pytorch cpuonly -c pytorch -y

# --- CityLearn (install with --no-deps; see install.bat) ---
# citylearn==2.5.0

# --- CityLearn minimal runtime deps ---
gymnasium==0.28.1
scikit-learn==1.2.2
simplejson
pyyaml
platformdirs          # CityLearn uses this for cache directory resolution

# --- RL library (pinned — last version compatible with gymnasium 0.28) ---
stable-baselines3==2.2.1

# --- Utilities ---
# numpy capped at <2.0.0: CityLearn 2.5.0 is not compatible with numpy 2.x
matplotlib>=3.7.0
pandas>=2.0.0
numpy>=1.24.0,<2.0.0
cloudpickle