Fast Mutation in Crossover-Based Algorithms

The heavy-tailed mutation operator proposed in Doerr et al. (GECCO 2017), called fast to agree with the previously used language, so far was proven be advantageous only mutation-based algorithms. There, it can relieve algorithm designer from finding optimal rate and nevertheless obtain a performance close one that gives. In this first runtime analysis of crossover-based using choice rate, we show an even stronger impact. For $$(1+(\lambda ,\lambda ))$$ genetic optimizing OneMax benchmark function, linear achieved. This is asymptotically faster than what obtained any static equivalent self-adjusting version parameters algorithm. result complemented by empirical study which shows effectiveness also on random satisfiable MAX-3SAT instances.