Evaluating Stationary Distribution of the Binary GA Markov Chain in Special Cases

Authors Boris S. Mitavskiy, Chris Cannings



PDF
Thumbnail PDF

File

DagSemProc.08051.4.pdf
  • Filesize: 305 kB
  • 24 pages

Document Identifiers

Author Details

Boris S. Mitavskiy
Chris Cannings

Cite As Get BibTex

Boris S. Mitavskiy and Chris Cannings. Evaluating Stationary Distribution of the Binary GA Markov Chain in Special Cases. In Theory of Evolutionary Algorithms. Dagstuhl Seminar Proceedings, Volume 8051, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2008) https://doi.org/10.4230/DagSemProc.08051.4

Abstract

The evolutionary algorithm stochastic process is well-known to be
Markovian. These have been under investigation in much of the
theoretical evolutionary computing research. When mutation rate is
positive, the Markov chain modeling an evolutionary algorithm is
irreducible and, therefore, has a unique stationary distribution,
yet, rather little is known about the stationary distribution. On the other
hand, knowing the stationary distribution may provide
some information about the expected times to hit optimum, assessment of the biases due to recombination and is of importance in population
genetics to assess what's called a ``genetic load" (see the
introduction for more details). In this talk I will show how the quotient
construction method can be exploited to derive rather explicit bounds on the ratios of the stationary distribution values of various subsets of
the state space. In fact, some of the bounds obtained in the current
work are expressed in terms of the parameters involved in all the
three main stages of an evolutionary algorithm: namely selection,
recombination and mutation. I will also discuss the newest developments which may allow for further improvements of the bounds

Subject Classification

Keywords
  • Genetic algorithms
  • Markov chains
  • stationary distribution
  • lumping quotient

Metrics

  • Access Statistics
  • Total Accesses (updated on a weekly basis)
    0
    PDF Downloads
Questions / Remarks / Feedback
X

Feedback for Dagstuhl Publishing


Thanks for your feedback!

Feedback submitted

Could not send message

Please try again later or send an E-mail