See here for the Zenodo entry for this Artifact.

Programs: Either Docker or Python 3.11

Machine used for testing:

• CPU: AMD Ryzen Threadripper 1950X 16-Core Processor 3.7MHz
• RAM: 94GB total, 300G swap
• GPU: NVIDIA TITAN Xp 12GB x4
• OS: Ubuntu 18.04.5 LTS

1. Clone the repo:

   git clone https://github.com/jptalusan/ICCPS2024_AE_Stationing.git

2. Setup the repo:

   cd ICCPS2024_AE_Stationing
   wget --no-check-certificate "https://zenodo.org/records/10594255/files/ARTIFACT_FILES.tar.gz?download=1" -O ARTIFACT_FILES.tar.gz
   tar -xzvf ARTIFACT_FILES.tar.gz

3. Adjusting the experiment if desired:

• "early_end": Can be set to false, to run through the entire day.
• "method": "mcts" or "baseline"
  • If MCTS, "iter_limit" count can be adjusted while "pool_thread_count" should be at least 1.
• "noise_level": "", 1, 5, 10
• "overload_start_depots": Dictionary of where the substitute buses will start from.

4. Running the experiment:

   Run experiment using Python:

   cd code_root/experiments/TEST
   python run_mcts_no_inject.py -c configs/test

   Running using Docker:

   docker build -t iccps2024_stationing .
   docker run -v $PWD/code_root:/usr/src/app/code_root iccps2024_stationing

• Note: If running using Docker Desktop on a Mac, you might need to allow file sharing on the current git repo directory.

1. Clone the repo:

   git clone https://github.com/jptalusan/ICCPS2024_AE_Stationing.git

2. Setup the repo:

   cd ICCPS2024_AE_Stationing
   curl "https://zenodo.org/records/10594255/files/ARTIFACT_FILES.tar.gz?download=1" -o ARTIFACT_FILES.tar.gz
   tar -xzvf ARTIFACT_FILES.tar.gz

3. Adjusting the experiment if desired:

• "early_end": Can be set to false, to run through the entire day.
• "method": "mcts" or "baseline"
  • If MCTS, "iter_limit" count can be adjusted while "pool_thread_count" should be at least 1.
• "noise_level": "", 1, 5, 10
• "overload_start_depots": Dictionary of where the substitute buses will start from.

4. Running the experiment:

   Run experiment using Python:

   cd code_root\experiments\TEST
   python run_mcts_no_inject.py -c configs/test

   Running using Docker:

   docker build -t iccps2024_stationing .
   docker run -v "%cd%"/code_root:/usr/src/app/code_root iccps2024_stationing
   

Plots will appear in code_root/experiments/TEST/plots they will be labeled by figure number and correspond to the figures in the Appendix.

less -S code_root/experiments/TEST/logs/20210211_test/stream.log
less -S code_root/experiments/TEST/results/20210211_test/results.csv
less -S code_root/experiments/TEST/results/20210211_test/stops_results.csv
less -S code_root/experiments/TEST/results/20210211_test/buses_results.csv

Note the datetime in the results are in UTC time. The first one contains the raw logs detailing the bus movement and passenger pickups and dropoffs. The second one is a summary containing 3 distinct CSVs and a summary of the results at the bottom.

ARTIFACT_FILES.tar.gz contains the processed data and input files. They are extracted to the following folders:

• Config file -> code_root/experiments/TEST/configs
• prepped_disruptions.parquet -> code_root/experiments/TEST/
• Input data -> code_root/scenarios/REAL_WORLD
• Input data -> code_root/scenarios/TEST_WORLD
  • These are the inputs from the transit agency (test plan and bus plans).
  • Includes the occupancy predictions.
• code_root/scenarios/common/sampled_travel_times_dict.pkl
• code_root/scenarios/common/stops_tt_dd_node_dict.pkl
• code_root/scenarios/common/stops_node_matching_dict.pkl

To get all of this from the base code for use in the evaluation:

1. Navigate to the root mta_simulator_redo
2. Run:

    tar -cvzf ARTIFACT_FILES.tar.gz -C . \
    code_root/scenarios/common/sampled_travel_times_dict.pkl \
    code_root/scenarios/common/stops_node_matching_dict.pkl \
    code_root/scenarios/common/stops_tt_dd_node_dict.pkl \
    code_root/scenarios/REAL_WORLD \
    code_root/scenarios/TEST_WORLD \
    data_analysis/data/best_solutions.parquet \
    code_root/experiments/TEST/prepped_disruptions.parquet