Skip to main content

Module 06: Population-Based ES - From One to Many

Real evolution needs populations. The (mu,lambda)-ES and (mu+lambda)-ES introduce recombination, self-adaptation, and the idea of learning the mutation distribution from the population. This is where ES starts to get interesting.

Learning Objectives​

  1. Implement (mu,lambda)-ES and (mu+lambda)-ES
  2. Understand weighted recombination (intermediate and discrete)
  3. Code self-adaptation of strategy parameters
  4. Understand derandomized step-size control
  5. Compare comma vs plus selection strategies

Concept Explanation​

Coming soon.

Code Examples​

Coming soon.

Exercises​

Coming soon.

Milestone Checklist​

  • Built (mu,lambda) and (mu+lambda) ES
  • Self-adaptation of sigma working
  • Can explain comma vs plus selection trade-offs
  • Understand why self-adaptation is better than 1/5th rule

Was this page helpful?