Introduction

DEAP-ER is a complete rework of the original DEAP library, which includes features such as:

• Genetic algorithms using any imaginable containers like:

  • List, Array, Set, Dictionary, Tree, Numpy Array, etc.

• Genetic programming using prefix trees

  • Loosely typed, Strongly typed

  • Automatically defined functions

• Evolution Strategies (Covariance Matrix Adaptation)

• Multi-objective optimisation (SPEA-II, NSGA-II, NSGA-III, MO-CMA)

• Co-evolution (cooperative and competitive) of multiple populations

• Parallelization of evolution processes using multiprocessing or with Ray

• Records to track the evolution and to collect the best individuals

• Checkpoints to persist the progress of evolutions to disk

• Benchmarks to test evolution algorithms against common test functions

• Genealogy of an evolution, that is also compatible with NetworkX

• Examples of alternative algorithms:

  • Symbolic Regression,

  • Particle Swarm Optimization,

  • Differential Evolution,

  • Estimation of Distribution Algorithm