Sizing the population with respect to the local progress in (1,/spl lambda/)-evolution strategies-a theoretical analysis

This paper presents an analysis of the local serial rate of progress with respect to the number of offspring λ for the (1,λ)-evolution strategy. It is shown that local serial progress is maximized when the expected progress of the second best offspring is zero. The theoretical results lead to a simple but efficient adaptation rule for λ, which needs no extra fitness function evaluations and only small computational expense. Simulations of the λ-adaptation on simple test functions are shown