Youhei Akimoto
ABSTRACT
We investigate evolution strategies with weighted recombination on general convex quadratic functions. We derive the asymptotic quality gain in the limit of the dimension to infinity, and derive the optimal recombination weights and the optimal step-size. This work is an extension of previous works where the asymptotic quality gain of evolution strategies with weighted recombination was derived on the infinite dimensional sphere function. Moreover, for a finite dimensional search space, we derive rigorous bounds for the quality gain on a general quadratic function. They reveal the dependency of the quality gain both in the eigenvalue distribution of the Hessian matrix and on the recombination weights. Taking the search space dimension to infinity, it turns out that the optimal recombination weights are independent of the Hessian matrix, i.e., the recombination weights optimal for the sphere function are optimal for convex quadratic functions.
- Y. Akimoto, A. Auger, and N. Hansen. Convergence of the continuous time trajectories of isotropic evolution strategies on monotonic C2-composite functions. In Problem Solving from Nature - PPSN XII, pages 42--51, 2012.Google Scholar
- Y. Akimoto and N. Hansen. Online model selection for restricted covariance matrix adaptation. In Parallel Problem Solving from Nature - PPSN XIV, pages 3--13, 2016. Google ScholarCross Ref
- Y. Akimoto and N. Hansen. Projection-based restricted covariance matrix adaptation for high dimension. In Genetic and Evolutionary Computation Conference, GECCO, pages 197--204, 2016. Google ScholarDigital Library
- D. V. Arnold. Optimal weighted recombination. In Foundations of Genetic Algorithms, pages 215--237, 2005. Google ScholarDigital Library
- D. V. Arnold. On the use of evolution strategies for optimising certain positive definite quadratic forms. In Genetic and Evolutionary Computation Conference, GECCO, pages 634--641, 2007. Google ScholarDigital Library
- A. Auger. Analysis of Comparison-based Stochastic Continuous Black-Box Optimization Algorithms. Habilitation, Universitè Paris-Sud, 2015.Google Scholar
- A. Auger, D. Brockhoff, and N. Hansen. Mirrored sampling in evolution strategies with weighted recombination. In Genetic and Evolutionary Computation Conference, GECCO, pages 861--868, 2011. Google ScholarDigital Library
- A. Auger and N. Hansen. Reconsidering the progress rate theory for evolution strategies in finite dimensions. In Genetic and Evolutionary Computation Conference, GECCO, pages 445--452, 2006. Google ScholarDigital Library
- A. Auger and N. Hansen. Linear convergence of comparison-based step-size adaptive randomized search via stability of markov chains. SIAM Journal on Optimization, 26(3):1589--1624, 2016. Google ScholarCross Ref
- H.-G. Beyer. Towards a theory of 'evolution strategies': Results for (1+, ł)-strategies on (nearly) arbitrary fitness functions. In Parallel Problem Solving from Nature - PPSN III, pages 58--67, 1994.Google Scholar
- H.-G. Beyer. The Theory of Evolution Strategies. Natural Computing Series. Springer-Verlag, 2001. Google ScholarCross Ref
- H.-G. Beyer and M. Hellwig. The dynamics of cumulative step size adaptation on the ellipsoid model. Evolutionary Computation, 24(1):25--57, 2016. Google ScholarDigital Library
- H.-G. Beyer and A. Melkozerov. The dynamics of self-adaptive multirecombinant evolution strategies on the general ellipsoid model. IEEE Transactions on Evolutionary Computation, 18(5):764--778, 2014. Google ScholarCross Ref
- A. DasGupta. Asymptotic theory of statistics and probability. Springer Science & Business Media, 2008.Google Scholar
- L. Devroye. Non-Uniform Random Variate Generation. Springer New York, 1986. Google ScholarCross Ref
- S. Finck and H.-G. Beyer. Weighted recombination evolution strategy on a class of pdqf's. In Foundations of Genetic Algorithms, FOGA, pages 1--12, 2009. Google ScholarDigital Library
- N. Hansen and A. Auger. Principled design of continuous stochastic search: From theory to practice. In Y. Borenstein and A. Moraglio, editors, Theory and Principled Methods for the Design of Metaheuristics. Springer, 2014. Google ScholarCross Ref
- N. Hansen and S. Kern. Evaluating the cma evolution strategy on multimodal test functions. In Parallel Problem Solving from Nature - PPSN VIII, pages 282--291, 2004. Google ScholarCross Ref
- J. Jägersküpper. How the (1 + 1)) es using isotropic mutations minimizes positive definite quadratic forms. Theoretical Computer Science, 361(1):38--56, 2006. Google ScholarDigital Library
- M. Jebalia and A. Auger. Log-linear convergence of the scale-invariant (μ/μω,λ)-es and optimal μ for intermediate recombination for large population sizes. In Parallel Problem Solving from Nature - PPSN XI, pages 52--62, 2010.Google Scholar
- M. Jebalia and A. Auger. Log-linear convergence of the scale-invariant (μ/μω,λ)-ES and optimal mu for intermediate recombination for large population sizes. Research Report RR-7275, Jun 2010.Google Scholar
- M. Jebalia, A. Auger, and P. Liardet. Log-linear convergence and optimal bounds for the (1 + 1)-es. In Evolution Artificielle (EA '07), pages 207--218, 2008.Google ScholarCross Ref
- I. Loshchilov. A computationally efficient limited memory cma-es for large scale optimization. In Genetic and Evolutionary Computation Conference, GECCO, pages 397--404, 2014. Google ScholarDigital Library
- I. Rechenberg. Evolutionsstrategie '94. Frommann-Holzboog, Stuttgart-Bad Cannstatt, 1994.Google Scholar
- R. Ros and N. Hansen. A simple modification in cma-es achieving linear time and space complexity. In Parallel Problem Solving from Nature - PPSN X, pages 296--305, 2008. Google ScholarDigital Library
- O. Teytaud and S. Gelly. General lower bounds for evolutionary algorithms. In Parallel Problem Solving from Nature - PPSN IX, pages 21--31, 2006. Google ScholarDigital Library
Index Terms
- Quality Gain Analysis of the Weighted Recombination Evolution Strategy on General Convex Quadratic Functions
Recommendations
Mirrored sampling in evolution strategies with weighted recombination
GECCO '11: Proceedings of the 13th annual conference on Genetic and evolutionary computationThis paper introduces mirrored sampling into evolution strategies (ESs) with weighted multi-recombination. Two further heuristics are introduced: pairwise selection selects at most one of two mirrored vectors in order to avoid a bias due to ...
Analysis of evolution strategies with the optimal weighted recombination
GECCO '18: Proceedings of the Genetic and Evolutionary Computation ConferenceThis paper studies the performance for evolution strategies with the optimal weighed recombination on spherical problems in finite dimensions. We first discuss the different forms of functions that are used to derive the optimal recombination weights ...
Empirical investigation of multiparent recombination operators in evolution strategies
An extension of evolution strategies to multiparent recombination involving a variable number ρ of parents to create an offspring individual is proposed. The extension is experimentally evaluated on a test suite of functions differing in their modality ...
Comments