This archive is distributed in association with the INFORMS Journal on Computing under the MIT License.
The repository contains the software, data, and computational results used in the paper Revenue Maximization for Seaport Terminal Operations Considering Multimodal Cargo Flows by Mohammad Sadegh Kazemi and Xinchang Wang.
To cite the contents of this repository, please cite both the paper and this repo, using their respective DOIs.
https://doi.org/10.1287/ijoc.2025.1238
https://doi.org/10.1287/ijoc.2025.1238.cd
Below is the BibTeX for citing this snapshot of the repository.
@misc{KazemiWang2026Code,
author = {Mohammad Sadegh Kazemi and Xinchang Wang},
publisher = {INFORMS Journal on Computing},
title = {Revenue Maximization for Seaport Terminal Operations
Considering Multimodal Cargo Flows},
year = {2026},
doi = {10.1287/ijoc.2025.1238.cd},
note = {Available for download at
https://github.com/INFORMSJoC/2025.1238},
}The repository contains five independent experiments. Each experiment directory includes its own README with detailed execution and output instructions.
| Experiment | Paper output | Purpose |
|---|---|---|
| Experiment 1 | Table 2 and Figure 2 | Calculates the mean revenue loss of NLP solvers relative to S1-multi-point and plots median solver runtime against mean revenue loss. |
| Experiment 2 | Table 3 | Generates the profit summary across distribution-correlation settings. |
| Experiment 3 | Figures EC.1a--EC.1c | Calculates bounds across 500 random seeds and produces the three EC.1 figures. |
| Experiment 4 | Table EC.1a | Summarizes mean losses and standard deviations when the true cargo-supplier choice model is a nested logit model with congestion. |
| Experiment 5 | Table EC.1b | Summarizes mean losses and standard deviations when the true cargo-supplier choice model is a multinomial logit model with congestion. |
.
├── experiment-1-table-2-figure-2/
├── experiment-2-table-3/
├── experiment-3-figure_EC1/
├── experiment-4-table-EC1a/
├── experiment-5-table-EC1b/
├── requirements.txt
├── AUTHORS
├── LICENSE
└── README.md
The experiments are independent and should be run from within their respective directories. See each experiment's README for its exact commands, inputs, outputs, logging, and resumability behavior. The full computational grids are expensive sequential runs; smaller grids can be configured using the constants near the top of each experiment's main.py.
The computational experiments were developed using Python. Install the common dependencies using:
python -m pip install -r requirements.txtEach experiment includes the processed data underlying its reported table or figure in its local results/ directory. Experiments that generate figures save both the visual artifact and the corresponding processed CSV data.
For support in using this software, submit an issue.