General Lower Bounds for the Running Time of Evolutionary Algorithms

Abstract. We present a new method for proving lower bounds in evolutionary computation based on fitness-level arguments and an additional condition on transition probabilities between fitness levels. The method yields exact or near-exact lower bounds for LO, OneMax, and all functions with a unique optimum. All lower bounds hold for every evolutionary algorithm that only uses standard mutation as variation operator.

• Type: Journal
• Year: 2010
• Where: PPSN
• Authors: Dirk Sudholt
• Keywords: Distributed And Parallel Computing, Evolutionary Computation, Lower Bounds, Near-exact Lower Bounds, PPSN 2010