Hands-On Genetic Algorithms with Python

Differential evolution

Differential evolution (DE) is a specialized variant of genetic algorithms that's used for the optimization of real-valued functions. DE differs from genetic algorithms in the following aspects:

• The DE population is always represented as a collection of real-valued vectors.
• Instead of replacing the entire current generation with a new generation, DE keeps iterating over the population, modifying one individual at a time, or keeping the original individual if it's better than its modified version.
• The traditional crossover and mutation operators are replaced by specialized ones, thereby modifying the value of the current individual using the values of three other individuals that are chosen at random.