Parallelization of Algorithms for Linear Discrete Optimization Using ParaPhrase

In industry optimization of processes, production planing, or resource usage is important to reduce costs and increase profit. Mathematical models for optimization can contribute to achieve this, but they also pose some challenges. Not only expertise in mathematics is needed to apply these optimization models, but furthermore expertise in programming is needed for implementation and integration into the software landscape of the company. Additionally most optimization algorithms are computationally very expensive and finding a solution takes a long time. Parallelization reduces the time and can lead to better results, but makes implementation even more challenging. How the high-level pattern-based approach of ParaPhrase [5] and its provided tools reduces this challenges will be described in this paper using a real-world example from industry.