Data

Our platform provides corresponding data for different problems. There are two methods of data generation:

• Random data generator: Automatically creates randomized datasets based on specified size and parameters for quick experiments or benchmarking.

• Customized data loader: Loads user-provided datasets from a specified path and file type for custom or real-world data usage.

The class random_{problem}_generator and class customized_{problem}_loader are integrated into the function def {problem}Generator, allowing users to select their preferred data generation method.

Please see individual problem descriptions for details.

• Routing Problems: TSP, ATSP, PCTSP, CVRP, OP, SOP

• Scheduling Problems: FFSP, RCPSP, SMTWTP

• Packing Problems: KP, MKP, BPP

• Other Graph Problems: MIS

Random data generator

Bases: Dataset

Attributes:

• num_sample: Number of samples in the dataset.

• num_nodes: Scale of the problem instance.

• device: Device to store the data (CPU/GPU).

Methods:

• __getitem__: Returns the samples.

• __len__: Returns the total number of samples.

Customized data loader

Bases: Dataset

Attributes:

• num_sample: Number of samples in the dataset.

• num_nodes: Scale of the problem instance.

• device: Device to store the data (CPU/GPU).

• path: Path to the custom data file.

• file_type: File type of the custom data file.

Methods:

• __getitem__: Returns the samples.

• __len__: Returns the total number of samples.