Half mirror algorithm: a metaheuristic that hybridizes swarm intelligence and evolution-based system

This paper promotes a new metaheuristic called the half mirror algorithm(HMA). As its name suggests, HMA offers a new kind of mirroring search.HMA is developed by hybridizing swarm intelligence and the evolutionsystem. Swarm intelligence is adopted by constructing several autonomousagents called swarms. On the other hand, the evolution system is adoptedusing arithmetic crossover based on a particular reference called a mirror.Four mirrors are used in HMA: the best swarm member, a randomly selectedswarm member, the central point of the space, and the corresponding swarmmember. During the confrontative assessment, HMA is confronted withaverage and subtraction-based optimization (ASBO), total interactionalgorithm (TIA), walrus optimization algorithm (WaOA), coati optimizationalgorithm (COA), and clouded leopard optimization (CLO). The resultshows that HMA is superior to ASBO, TIA, WaOA, COA, and CLO in 20,19, 19, 20, and 20 out of 23 functions, respectively. Moreover, HMA hasfound the global optimal of eight functions. It means the superiority of HMAoccurs in almost entire functions. In the future, the mirroring search can becombined with the guided and neighborhood search to construct a morepowerful metaheuristic.