Module 10: Differential Evolution - Simple, Powerful, and Underrated

DE creates new candidates by adding weighted differences between existing solutions. It's embarrassingly simple, incredibly effective, and sometimes beats CMA-ES. No covariance matrices, no gradient estimates - just vector arithmetic.

Learning Objectives

1. Implement DE/rand/1/bin from scratch
2. Understand mutation strategies (DE/best/1, DE/current-to-best/1)
3. Code binomial vs exponential crossover
4. Implement adaptive DE (JADE and SHADE)
5. Know when DE beats CMA-ES and vice versa

Concept Explanation

Coming soon.

Code Examples

Coming soon.

Exercises

Coming soon.

Milestone Checklist

• Built DE/rand/1/bin from scratch
• Understand the F and CR parameters
• Implemented JADE or SHADE
• Can compare DE vs CMA-ES on different problem types