Abstract
This thesis aims to parallelize an NP-Hard inventory management problem using the OpenMP and OpenACC parallel programming models. Over the course of the thesis, the significance of metaheuristic algorithms will be explained, and the Variable Neighborhood Search (VNS) algorithm will be analyzed in particular. A presentation of inventory control problems is also provided. After establishing the framework addressed by this work, an analysis is conducted on parallel metaheuristic algorithms and the technologies used to implement them—specifically OpenMP and OpenACC—along with examples of parallelizing the VNS algorithm, upon which we will base our work. By studying the algorithms and parallelization techniques, we determine the model that best fits our problem and implement it, detailing the technical aspects. Once the implementation is complete, the parallelized algorithm is tested using both OpenMP and OpenACC technologies, and the results are presented using relevant tables and diagrams. The thesis concludes with the research findings and its future directions.
Figure: Exploring parallelization of Metaheuristics for Inventory Control Problems
