pyccea.optimizers package

class pyccea.optimizers.BinaryGeneticAlgorithm(subpop_size: int, n_features: int, conf: dict)

Bases: object

Binary genetic algorithm.

Attributes:

mutation_rate: float

Probability of a mutation occurring.

crossover_rate: float

Probability of a crossover occurring.

tournament_sample_size: int

Sample size of the subpopulation that will be used in Tournament Selection.

selection_method: str

Population update method employed in each generation, which can take one of two options: (i) generational, where the entire population is replaced in each generation (excluding a specified ‘elite_size’ individuals), or (ii) steady-state, where only the two least fit individuals are replaced.

elite_size: int, optional

Number of individuals that will be preserved from the current to the next generation. They are selected according to the fitness, i.e., if it is N, the N best individuals of the current generation will be preserved for the next generation. This parameter is only used if the selection_method is generational.

Methods

evolve(subpop, fitness)

Evolve a subpopulation for a single generation.

evolve(subpop: ndarray, fitness: list)

Evolve a subpopulation for a single generation.

Parameters:

subpop: np.ndarray

Individuals of the subpopulation, where each individual is an array of size equal to the number of features.

fitness: list

Evaluation of all individuals in the subpopulation.

Returns:

next_subpop: np.ndarray

Individuals in the subpopulation of the next generation.

selection_methods = ['generational', 'steady-state']

class pyccea.optimizers.DifferentialEvolution(subpop_size: int, n_features: int, conf: dict)

Bases: object

Differential Evolution (AMDE) algorithm (rand/1/exp).

Storn, Rainer, and Kenneth Price. “Differential Evolution - A Simple and Efficient Heuristic for Global Optimization over Continuous Spaces” Journal of global optimization 11 (1997): 341-359.

Attributes:

scaling_factor: float

The mutation constant. In the literature this is also known as differential weight, being denoted by F. It should be in the range [0, 2].

crossover_probability: float

The recombination constant, should be in the range [0, 1]. In the literature this is also known as the crossover probability. Increasing this value allows a larger number of mutants to progress into the next generation, but at the risk of population stability.

bounds: tuple[float, float]

Bounds for continuous variables (min, max).

Methods

evolve(subpop, fitness)

Evolve a subpopulation for a single generation.

evolve(subpop: ndarray, fitness: list)

Evolve a subpopulation for a single generation.

Parameters:

subpop: np.ndarray

Individuals of the subpopulation, where each individual is an array of size equal to the number of features.

fitness: list

Evaluation of all individuals in the subpopulation.

Returns:

next_subpop: np.ndarray

Individuals in the subpopulation of the next generation.

Submodules

pyccea.optimizers.differential_evolution module

class pyccea.optimizers.differential_evolution.DifferentialEvolution(subpop_size: int, n_features: int, conf: dict)

Bases: object

Differential Evolution (AMDE) algorithm (rand/1/exp).

Storn, Rainer, and Kenneth Price. “Differential Evolution - A Simple and Efficient Heuristic for Global Optimization over Continuous Spaces” Journal of global optimization 11 (1997): 341-359.

Attributes:

scaling_factor: float

The mutation constant. In the literature this is also known as differential weight, being denoted by F. It should be in the range [0, 2].

crossover_probability: float

The recombination constant, should be in the range [0, 1]. In the literature this is also known as the crossover probability. Increasing this value allows a larger number of mutants to progress into the next generation, but at the risk of population stability.

bounds: tuple[float, float]

Bounds for continuous variables (min, max).

Methods

evolve(subpop, fitness)

Evolve a subpopulation for a single generation.

evolve(subpop: ndarray, fitness: list)

Evolve a subpopulation for a single generation.

Parameters:

subpop: np.ndarray

Individuals of the subpopulation, where each individual is an array of size equal to the number of features.

fitness: list

Evaluation of all individuals in the subpopulation.

Returns:

next_subpop: np.ndarray

Individuals in the subpopulation of the next generation.

pyccea.optimizers.genetic_algorithm module

class pyccea.optimizers.genetic_algorithm.BinaryGeneticAlgorithm(subpop_size: int, n_features: int, conf: dict)

Bases: object

Binary genetic algorithm.

Attributes:

mutation_rate: float

Probability of a mutation occurring.

crossover_rate: float

Probability of a crossover occurring.

tournament_sample_size: int

Sample size of the subpopulation that will be used in Tournament Selection.

selection_method: str

Population update method employed in each generation, which can take one of two options: (i) generational, where the entire population is replaced in each generation (excluding a specified ‘elite_size’ individuals), or (ii) steady-state, where only the two least fit individuals are replaced.

elite_size: int, optional

Number of individuals that will be preserved from the current to the next generation. They are selected according to the fitness, i.e., if it is N, the N best individuals of the current generation will be preserved for the next generation. This parameter is only used if the selection_method is generational.

Methods

evolve(subpop, fitness)

Evolve a subpopulation for a single generation.

evolve(subpop: ndarray, fitness: list)

Evolve a subpopulation for a single generation.

Parameters:

subpop: np.ndarray

Individuals of the subpopulation, where each individual is an array of size equal to the number of features.

fitness: list

Evaluation of all individuals in the subpopulation.

Returns:

next_subpop: np.ndarray

Individuals in the subpopulation of the next generation.

selection_methods = ['generational', 'steady-state']