Analysis of a meta-ES on a conically constrained problem
Michael Hellwig, Hans-Georg Beyer · Proceedings of the Genetic and Evolutionary Computation Conference · 2019
The paper presents the theoretical performance analysis of a hierarchical Evolution Strategy (meta-ES) variant for mutation strength control on a conically constrained problem. Infeasible offspring are repaired by projection onto the boundary of the feasibility region. Closed-form approximations are used for the one-generation progress of the lower-level evolution strategy. An interval that brackets the expected progress over a single isolation period of the meta-ES is derived. Approximate deterministic evolution equations are obtained that characterize the upper-level strategy dynamics. It is shown that the dynamical behavior of the meta-ES is determined by the choice of the mutation strength control parameter. The obtained theoretical results are compared to experiments for assessing the approximation quality.