A new meta-heuristic approach called the Marathon Optimization Algorithm has been developed to address complex engineering optimization problems. The proposed framework directly simulates the leading, following, and eliminating dynamics observed among competitors in marathon running events. By modeling these collective athletic behaviors, the algorithm establishes a structured optimization strategy designed to guide candidate solutions through complex search landscapes.
Under this computational strategy, all participating athletes within the population are categorized into three distinct operational groups based on their fitness ranking. These established categories consist of the leading runner, abbreviated as LR, the following runner, abbreviated as FR, and the eliminated runner, abbreviated as ER. This ranking-based structure determines the behavior and functional responsibility of each individual throughout the race.
Throughout the successive iterations of the algorithm, the leading runner actively directs the following runners to guide their forward movement. This guiding interaction enhances the overall movement efficiency of the trailing athletes while ensuring an extensive exploration across the solution space. As a result, the leading individual helps the broader group navigate the problem terrain more effectively.
Simultaneously, the following runners observe the leading runner and adjust their individual movement behaviors in response to achieve better optimization results. This observational adjustment mechanism endows the algorithm with substantial exploration capability built directly upon the foundation of exploitation capability, allowing the trailing runners to refine their trajectories during the search process.
The framework also accounts for competitors who are unable to complete the endurance race during the competitive event. Athletes who fail to finish are formally eliminated from the competition and replaced by entirely new athletes. This replacement mechanism effectively preserves and maintains the overall diversity of the algorithm population as iterations proceed.
Across the entire movement process, the leading, following, and eliminating strategies of marathoners are fully integrated within the computational model. This integrated design allows the marathoners to adjust their roles and positions in real time as fitness values change, which works to improve the overall convergence accuracy of the optimization system.
To evaluate the operational performance of the Marathon Optimization Algorithm, researchers tested the method against the CEC-2017 and CEC-2022 benchmark suites. The evaluation also examined performance across seven constrained engineering optimization problems and one path optimization problem. Under the adopted experimental settings, the algorithm demonstrated competitive performance compared with several recent metaheuristic algorithms.
Recorded convergence curves, statistical results, and engineering problem applications suggest that the strategy effectively balances exploration and exploitation while reducing premature stagnation during optimization. In addition, the source code implementing the Marathon Optimization Algorithm has been made publicly available through the MATLAB Central online repository for further verification.
Subject of Research: Meta-heuristic Marathon Optimization Algorithm for engineering optimization problems
Article Title: Marathon optimization algorithm: a novel approach for solving engineering optimization problems
Article References: Fu, Z., Guo, W., Li, C., Wang, Y., & Cao, D. (2027). Marathon optimization algorithm: a novel approach for solving engineering optimization problems. Cluster Computing, 30(1), Article 53. https://doi.org/10.1007/s10586-026-06831-1
Image Credits: AI Generated
DOI: 10.1007/s10586-026-06831-1
Keywords: Marathon Optimization Algorithm, meta-heuristic algorithm, engineering optimization, CEC-2017 benchmark, CEC-2022 benchmark, leading runner, following runner, eliminated runner, exploration and exploitation, premature stagnation, convergence accuracy, population diversity, constrained optimization, path optimization, fitness ranking, MATLAB Central, movement efficiency, solution space
News Source: Denise Maddox. (November 21, 2026). Marathon Optimization Algorithm Balances Exploration and Exploitation in Engineering Benchmarks. Scienmag.



