Evolutionary%20Computation

1
Evolutionary Computation

• Module for HON207

2
Introduction

• This lecture is an introduction to the topic of
  Evolutionary Computation, but it is not a
  substitute for the assigned readings
• We will review key concepts, and we will
  introduce a few simple models in R

3
Topics

• What is Evolutionary Computation (EC)?
• Basic principles
• Fitness
• Simple models of asexual reproduction
• Description of the R programs needed for the
  assignments

4
What is Evolutionary Computation (EC)?

• EC is an emerging field that focuses on the
  design and application of computational models
  inspired in Darwinian-like formulations.

5
The goals of EC

• Developing and understanding better models of
  natural evolution
• Engineers try to use evolution principles to
  build artifacts (e.g., a stock trading tool)
• Artificial-life researchers experimenting with
  artificial evolutionary worlds

6
Basic Concepts

• Evolution in a Darwinian sense.
• Individuals in Population(s) compete for limited
  resources.
• Dynamically changing populations due to
  death/birth
• Offspring resembling parent but not identical

7
Genotype / Phenotype

• The genotype is the specific genetic makeup (the
  specific genome) of an individual, in the form of
  DNA.
• The phenotype of an individual organism is either
  its total physical appearance and constitution or
  a specific manifestation of a trait.
• For our purpose, we will assume a one-to-one
  correspondence between the two.

8
Gene/Pheno Space

• The geno/pheno space is a vector that specifies
  the genetic makeup of an individual.
• lt trait1, trait2, trait3, , traitN gt
• lthair color, eye color, height, etcgt
• In Genetics, an allele is any one of a number of
  viable DNA codings.
• In other words, the possible values for the
  elements in the vector (e.g., a number between 0
  and 5)

9
Fitness

• In biology fitness is, in general, an ex post
  facto measure based on the individuals ability
  to survive and reproduce. This could be related
  to the environment, the make up of the
  population, etc.
• In our simplified models, fitness will be a
  function solely of individuals geno/phenotype.
• e.g., f(x)50-x2

10
Mutation

• In biology mutations are changes to the genetic
  material
• We need to make some assumptions on how likely
  mutations would be in our model, and how would
  those mutations be implemented
• If there are L genes, we assume a probability of
  mutation 1/L
• The mutation would be the value in the gene
  inherited from the parent ??

11
First simulation

• 1.3 EV on a Simple Fitness Landscape (page 6)
• Fitness function f(x)50-x2
• 1 trait
• Mutation change by ? 1
• Population 10
• 100 generations

12
First simulation

• 1.3 EV on a Simple Fitness Landscape (page 6)
• Fitness function f(x)50-x2
• 1 trait
• Mutation change by ? 1
• Population 10
• 100 generations
• We have a function in R that can do this
• EC(M,L,generations,lb,ub,formula,mutrule,Delta,fix
  )