Several improvements:

• Termination Checker can now also be a list. In that case, early termination is performed if any of the termination conditions apply.
• Generations now range from 1 to max_generation, instead of 0 to max_generation - 1. Individual now contains informative fields - id, gen, cloned_from, selected_by and applied_operators.
• Vector and its subtypes can now receive a list of the genome (vector cells) as constructor argument.

ProcessPoolExecutor Previously, fitness evaluation tasks could only be submitted to a ThreadPoolExecutor. From now on, the tasks can also be submitted to a ProcessPoolExecutor (enhanced performance in some cases).

When initializing an instance of SimpleEvolution, use the parameter executor=process for a ProcessPoolExecutor, or executor=thread for a ThreadPoolExecutor (the default is value is thread).

Multi-Objective Evolution

MOE - multi-objective evolution - from now on you could use EC-KitY to run an evolutionary algorithm with multiple objectives, to receive the Pareto front and find the best solution(s) for you.

IndividualEvaluator.evaluate signature change

IndividualEvaluator.evaluate now also receives the entire sub-population individuals, and not only the current individual. Useful for non-simple cases in which the individual's fitness also depends on the rest of the individuals in its sub-population.

bug fixes for float vector and tests

Bugfixes to GA IntVector, FloatVector and their appropriate Creator classes

EC-KitY - version 0.2.0

Support for Genetic Algorithms

This version supports:

• Genetic Algorithms (GA) Vector representation
• Bit Vectors, Integer Vectors and Float Vectors
• Crossover operators: K Point Crossover
• Mutation operators: One Point and N Point Mutation
• Two GA examples: One Max, Knapsack

EC-KitY - Version 0.1.1

In addition to the features described in version 0.1.0, EC-KitY now also includes Pypi support (i.e. pip-installable).

EC-KitY - version 0.1.0

Initial release of EC-KitY: Evolutionary Computation tool kit in Python

This version supports:

• Genetic Programming (GP) tree representation
• Two fundamental modes: Basic mode and sklearn mode
• Creating GP trees using either Grow, Full. or Ramped-Half-and-Half
• Selection methods: Tournament Selection
• Elitism
• Crossover operators: Subtree Crossover
• Mutation operators: Subtree Mutation, ERC Mutation
• Concurrent fitness evaluation
• Statistics
• Two basic-mode (non-sklearn) GP examples: Symbolic Regression, Multiplexer
• Two sklearn-mode GP examples: Symbolic Regression, Breast Cancer
• sklearn compatibility showcased through use of Pipeline and Grid search
• Adding user-defined problems and fitness functions