Pattern-Based Control for Dynamic Stochastic Multi-Knapsack Problems with Multiple Item Types

This repository provides the complete dataset and source code for reproducing all numerical experiments reported in the corresponding manuscript submitted to Computers & Operations Research, in compliance with the journal’s data availability policy.
The numerical experiments are conducted to verify the effectiveness, efficiency, and robustness of the proposed optimization method. Tests are performed on standard benchmark instances and randomly generated cases of different scales, aiming to compare the proposed approach with state-of-the-art methods from related literature, analyze parameter sensitivity, and demonstrate computational performance.
The dataset includes benchmark test instances widely used in the operations research field and synthetic instances with various problem sizes. All data are preprocessed consistently to ensure fair comparison. Experimental results such as objective values, computational time, and performance indicators are also included for verification.
The source code implements the proposed algorithm, benchmark models, parameter settings, and result output modules. It is structured to be executable and reproducible, with fixed random seeds and clear running instructions. Dependencies and environment requirements are briefly documented to help users replicate the results efficiently.
All materials are shared for academic research and reproducible verification. Users may access and reuse the data and code with proper citation of the related manuscript and this repository.