The runner-root algorithm

摘要: A new optimization algorithm inspired by the plants propagated through runners is proposed.Global search with random large steps performed at all iterations (exploration).Local small (exploitation) only if global fails.Local roots and root hairs.It does not necessarily apply a same number of function evaluations iterations. This paper proposes metaheuristic, runner-root (RRA), some in nature. The which are look for water resources minerals developing (as well as hairs). first tool helps plant around big while second one appropriate steps. Moreover, placed very good location chance spreads larger area its longer roots. Similarly, proposed equipped two tools exploration: jumps steps, model nature, re-initialization strategy case trapping local optima, redistributes computational agents randomly domain problem models propagation being located position. Exploitation RRA so-called hairs respectively changes to variables best agent separately (in stagnation). Performance examined applying it standard CEC' 2005 benchmark problems then comparing results 9 state-of-the-art algorithms using nonparametric methods.