Where Elitists Start Limping Evolution Strategies at Ridge Functions
نویسندگان
چکیده
How well an optimization algorithm satisses short-term and long-term goals, can be veriied using appropriate test functions, respective convergence measures, theoretical analysis, and simulations. This paper analyses the convergence behavior of the evolution strategy (ES) at the parabolic ridge function using the standard (1 + ;)-ES. Some further results are given for the case of more general ridge functions. The results obtained are counter-intuitive and diierent from if not contrary to those obtained from the sphere model theory. Furthermore, using static analysis, we show that the progress rate and the quality gain possess entirely diierent characteristics.
منابع مشابه
Modelling Genetic Algorithms: From Markov Chains to Dependence with Complete Connections
A Timing Analysis of Convergence to Fitness Sharing Equilibrium p. 23 Where Elitists Start Limping: Evolution Strategies at Ridge Functions p. 34 A Bit-Wise Epistasis Measure for Binary Search Spaces p. 47 Inside GA Dynamics: Ground Basis for Comparison p. 57 The Effect of Spin-Flip Symmetry on the Performance of the Simple GA p. 67 Fitness Distance Correlation and Ridge Functions p. 77 Acceler...
متن کاملMultiswarm comprehensive learning particle swarm optimization for solving multiobjective optimization problems
Comprehensive learning particle swarm optimization (CLPSO) is a powerful state-of-the-art single-objective metaheuristic. Extending from CLPSO, this paper proposes multiswarm CLPSO (MSCLPSO) for multiobjective optimization. MSCLPSO involves multiple swarms, with each swarm associated with a separate original objective. Each particle's personal best position is determined just according to the c...
متن کاملCumulative Step Length Adaptation on Ridge Functions
The ridge function class is a parameterised family of test functions that is often used to evaluate the capabilities and limitations of optimisation strategies. Past research with the goal of analytically determining the performance of evolution strategies on the ridge has focused either on the parabolic case or on simple one-parent strategies without step length adaptation. This paper extends ...
متن کاملConvergence Behavior of the ( 1 + ; ) Evolution
The convergence behavior of (1 +; ){ES is investigated at parabolic ridge, sharp ridge, and at the general case of the ridge functions. The progress rate, the distance to the ridge axis, the success rate, and the success probability are used in the analysis. For the parabolic ridge, we observed in our simulations with large values using (1; ){ES a larger progress rate asymptote than the theoret...
متن کاملOn the Performance of (1; )-evolution Strategies at the Ridge Function Class
This paper presents the N-dependent analysis of the (1;) Evolution Strategy (ES) with isotropic mutations at the ridge functions including the special cases sharp and parabolic ridge. The new approach presented allows for the prediction of the dynamics in ridge direction as well as in radial direction. The central quantities are the corresponding progress rates which are determined in terms of ...
متن کامل