Simulation Data and Code for Deadline-Sensitive Online Routing in Stochastic Time-Varying Damaged Networks

Description

This dataset contains simulation results, ablation-study outputs, supplementary numerical files, and implementation code associated with the manuscript “Deadline-Sensitive Online Routing in Stochastic Time-Varying Damaged Networks.” The data were generated through numerical simulation for stochastic time-varying damaged-network routing experiments.

Steps to reproduce

1. Download and extract the dataset archive. 2. Install Python 3.10 or later and install the required packages listed in requirements.txt. The main dependencies are numpy, pandas, networkx, matplotlib, and torch. 3. To reproduce the full numerical experiment, run the implementation script in scripts/opstvn.py. The default configuration generates a 40-node, 120-arc directed network, constructs stochastic time-varying damaged-network scenarios, trains the travel-time predictor, trains the DQL critic, trains the diffusion-style candidate generator, and evaluates the routing policies. 4. A representative full run can be executed with: python scripts/opstvn.py --scenario ALL --run-ablation --output-dir results/reproduced 5. The default command uses the built-in random seed, training-sample size, training epochs, DQL episodes, evaluation-task count, and deadline multipliers specified in the script. The main command-line options include --tt-samples, --tt-epochs, --dql-episodes, --diff-epochs, --eval-tasks, --seed, --scenario, and --output-dir. 6. The script writes aggregate performance metrics, per-task results, runtime summaries, ablation results, hyperparameter records, generated figures, and LaTeX-ready result tables to the specified output directory. 7. The processed results reported in the manuscript are provided in the results/ directory. Minor numerical differences may occur across hardware platforms and PyTorch/CUDA versions because the experiments include stochastic simulation and neural-network training.

Institutions

• Shandong University

  Shandong, Jinan

Categories

Licence