--- configs: - config_name: CA-easy data_files: - split: test path: CA/easy/test* - split: train path: CA/easy/train* - split: validation path: CA/easy/val* - config_name: CA-medium data_files: - split: test path: CA/medium/test* - split: train path: CA/medium/train* - split: validation path: CA/medium/val* - config_name: CA-very-hard data_files: - split: test path: CA/very-hard/test* - split: train path: CA/very-hard/train* - split: validation path: CA/very-hard/val* - config_name: SC-easy data_files: - split: test path: SC/easy/test* - split: train path: SC/easy/train* - split: validation path: SC/easy/val* - config_name: SC-medium data_files: - split: test path: SC/medium/test* - split: train path: SC/medium/train* - split: validation path: SC/medium/val* - config_name: SC-hard data_files: - split: test path: SC/hard/test* - split: train path: SC/hard/train* - split: validation path: SC/hard/val* - config_name: SC-very-hard data_files: - split: test path: SC/very-hard/test* - split: train path: SC/very-hard/train* - split: validation path: SC/very-hard/val* - config_name: MIS-easy data_files: - split: test path: MIS/easy/test* - split: train path: MIS/easy/train* - split: validation path: MIS/easy/val* - config_name: MIS-medium data_files: - split: test path: MIS/medium/test* - split: train path: MIS/medium/train* - split: validation path: MIS/medium/val* - config_name: MIS-very-hard data_files: - split: test path: MIS/very-hard/test* - split: train path: MIS/very-hard/train* - split: validation path: MIS/very-hard/val* - config_name: MVC-easy data_files: - split: test path: MVC/easy/test* - split: train path: MVC/easy/train* - split: validation path: MVC/easy/val* - config_name: MVC-medium data_files: - split: test path: MVC/medium/test* - split: train path: MVC/medium/train* - split: validation path: MVC/medium/val* - config_name: MVC-hard data_files: - split: test path: MVC/hard/test* - split: train path: MVC/hard/train* - split: validation path: MVC/hard/val* - config_name: MVC-very-hard data_files: - split: test path: MVC/very-hard/test* - split: train path: MVC/very-hard/train* - split: validation path: MVC/very-hard/val* - config_name: LB-hard data_files: - split: test path: LB/hard/test* - split: train path: LB/hard/train* - split: validation path: LB/hard/val* - config_name: IP-very-hard data_files: - split: test path: IP/very-hard/test* - split: train path: IP/very-hard/train* - split: validation path: IP/very-hard/val* - config_name: GISP-easy data_files: - split: test path: GISP/easy/test* - split: train path: GISP/easy/train* - split: validation path: GISP/easy/val* - config_name: GISP-medium data_files: - split: test path: GISP/medium/test* - split: train path: GISP/medium/train* - split: validation path: GISP/medium/val* - config_name: GISP-hard data_files: - split: test path: GISP/hard/test* - split: train path: GISP/hard/train* - split: validation path: GISP/hard/val* - config_name: GISP-very-hard data_files: - split: test path: GISP/very-hard/test* - split: train path: GISP/very-hard/train* - split: validation path: GISP/very-hard/val* - config_name: GISP-ext-hard data_files: - split: test path: GISP/ext-hard/test* - split: train path: GISP/ext-hard/train* - split: validation path: GISP/ext-hard/val* - config_name: SRPN-easy data_files: - split: test path: SRPN/easy/test* - split: train path: SRPN/easy/train* - split: validation path: SRPN/easy/val* - config_name: SRPN-hard data_files: - split: test path: SRPN/hard/test* - split: train path: SRPN/hard/train* - split: validation path: SRPN/hard/val* - config_name: CFLP-easy data_files: - split: test path: CFLP/easy/test* - split: train path: CFLP/easy/train* - split: validation path: CFLP/easy/val* - config_name: CFLP-medium data_files: - split: test path: CFLP/medium/test* - split: train path: CFLP/medium/train* - split: validation path: CFLP/medium/val* - config_name: MIRP-medium data_files: - split: test path: MIRP/medium/test* - split: train path: MIRP/medium/train* - config_name: MMCN-medium-BC data_files: - split: test path: MMCN/medium-BC/test* - split: train path: MMCN/medium-BC/train* - split: validation path: MMCN/medium-BC/val* - config_name: MMCN-medium-BI data_files: - split: test path: MMCN/medium-BI/test* - split: train path: MMCN/medium-BI/train* - split: validation path: MMCN/medium-BI/val* - config_name: MMCN-hard-BI data_files: - split: test path: MMCN/hard-BI/test* - split: train path: MMCN/hard-BI/train* - split: validation path: MMCN/hard-BI/val* - config_name: MMCN-very-hard-BI data_files: - split: test path: MMCN/very-hard-BI/test* - split: train path: MMCN/very-hard-BI/train* - split: validation path: MMCN/very-hard-BI/val* - config_name: NNV-easy data_files: - split: test path: NNV/easy/test* - split: train path: NNV/easy/train* - split: validation path: NNV/easy/val* - config_name: OTS-easy data_files: - split: test path: OTS/easy/test* - split: train path: OTS/easy/train* - split: validation path: OTS/easy/val* - config_name: OTS-hard data_files: - split: test path: OTS/hard/test* - split: train path: OTS/hard/train* - split: validation path: OTS/hard/val* - config_name: OTS-medium data_files: - split: test path: OTS/medium/test* - split: train path: OTS/medium/train* - split: validation path: OTS/medium/val* --- # Dataset Card for Distributional MIPLIB (D-MIPLIB) 🌐 [website](https://sites.google.com/usc.edu/distributional-miplib) Introducing Distributional MIPLIB (D-MIPLIB), the first comprehensive standardized dataset for evaluating ML-guided Mixed Integer Linear Programming (MILP) solving. Details of D-MIPLIB can be found on our [website](https://sites.google.com/usc.edu/distributional-miplib). ### Composition Distributional MIPLIB contains MILP distributions from 13 application domains and of different hardness levels. A list of all the domains and all the distributions is provided at [website](https://sites.google.com/usc.edu/distributional-miplib). Each distribution contains a set of similar MILP optimization problem instances that correspond to the same model (a class of combinatorial optimization with specific parameters). For each distribution, there are 1000 instances in total, except for cases where more instances have been made publicly available previously or when limited instances are available. For Neural Network Verification, there are 3692 instances in total. Maritime Inventory Routing Problem contains 118 instances. Seismic-Resilient Pipe Network Planning domain contains 219 and 206 instances in the Easy and Hard distribution, respectively. ### Loading MILP instances MILP instances in this library are stored as text strings. You can download the MILP as *.lp or *.mps files by ```python from datasets import load_dataset # load the first test instance dataset = load_dataset("weiminhu/D-MIPLIB","CA-easy", split='test') test_id = 0 test_instance = dataset[test_id]['MILP'] # check file format file_format='lp' if 'format' in list(dataset[0].keys()): if dataset[0]['format']=='mps': file_format='mps' ins_save_path = "./test_instance_%s.%s" % (test_id,file_format) # write file to disk with open(ins_save_path, "w") as file: tmp=test_instance[2:-1] tmp=tmp.replace("\\n","\n") file.write(tmp) ``` To load and solve the MILP in Gurobi, run ```python import gurobipy as gp model=gp.read(ins_save_path) model.optimize() ``` ### License Instances data in different domains is licensed under different licenses, as we curated the data from different sources. | Domain | License | |--------|----------------------| | CA | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) | | SC | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) | | MIS | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) | | MVC | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) | | GISP | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) | | CFLP | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) | | LB | BSD 3-Clause License [(Source repository)](https://github.com/ds4dm/ml4co-competition/tree/main)[(Details and copyright notice)](https://github.com/ds4dm/ml4co-competition/blob/main/LICENSE) | | IP | BSD 3-Clause License [(Source repository)](https://github.com/ds4dm/ml4co-competition/tree/main)[(Details and copyright notice)](https://github.com/ds4dm/ml4co-competition/blob/main/LICENSE) | | MIRP | BSD 3-Clause License [(Source repository)](https://github.com/ds4dm/ml4co-competition/tree/main) [(Details and copyright notice)](https://github.com/ds4dm/ml4co-competition/blob/main/LICENSE) | | NNV | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) [(Source repository)](https://github.com/google-deepmind/deepmind-research/tree/master/neural_mip_solving) | | OTS | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) | | MMCN | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) | | SRPN | [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode) | ### Citation **BibTeX:** ```BibTeX @misc{huang2024distributional, title={Distributional MIPLIB: a Multi-Domain Library for Advancing ML-Guided MILP Methods}, author={Weimin Huang and Taoan Huang and Aaron M Ferber and Bistra Dilkina}, year={2024}, eprint={2406.06954}, archivePrefix={arXiv}, primaryClass={cs.LG} } ```