The runner-root algorithm: A metaheuristic for solving unimodal and multimodal optimization problems inspired by runners and roots of plants in nature

Abstract

This paper proposes a new metaheuristic, the runner-root algorithm (RRA), inspired by the function of runners and roots of some plants in nature. The plants which are propagated through runners look for water resources and minerals by developing runners and roots (as well as root hairs). The first tool helps the plant for search around with random big steps while the second one is appropriate for search around with small steps. Moreover, the plant which is placed at a very good location by chance spreads in a larger area through its longer runners and roots. Similarly, the proposed algorithm is equipped with two tools for exploration: random jumps with big steps, which model the function of runners in nature, and a re-initialization strategy in case of trapping in local optima, which redistributes the computational agents randomly in the domain of problem and models the propagation of plant in a larger area in case of being located in a good position. Exploitation in RRA is performed by the so-called roots and root hairs which respectively apply random large and small changes to the variables of the best computational agent separately (in case of stagnation). Performance of the proposed algorithm is examined by applying it to the standard CEC’ 2005 benchmark problems and then comparing the results with 9 state-of-the-art algorithms using nonparametric methods.