porcupy.utils.population

Classes

PopulationCycle	Class for managing cyclic population reduction in the Crested Porcupine Optimizer.
SelectionStrategies	Class implementing various selection strategies for population reduction.

Module Contents

class porcupy.utils.population.PopulationCycle(initial_pop_size: int, min_pop_size: int, max_iter: int, cycles: int = 2, reduction_strategy: str = 'linear')

Class for managing cyclic population reduction in the Crested Porcupine Optimizer.

This class implements the cyclic population reduction strategy, which is a key feature of the CPO algorithm. The population size varies cyclically throughout the optimization process to balance exploration and exploitation.

Parameters:

• initial_pop_size (int) – Initial population size.

• min_pop_size (int) – Minimum population size during reduction.

• max_iter (int) – Maximum number of iterations.

• cycles (int) – Number of cycles for population reduction.

• reduction_strategy (str) – Strategy for population reduction (‘linear’, ‘cosine’, or ‘exponential’).

initial_pop_size

Initial population size.

Type:

int

min_pop_size

Minimum population size during reduction.

Type:

int

max_iter

Maximum number of iterations.

Type:

int

cycles

Number of cycles for population reduction.

Type:

int

reduction_strategy

Strategy for population reduction.

Type:

str

initial_pop_size

min_pop_size

max_iter

cycles = 2

reduction_strategy = 'linear'

calculate_pop_size(iteration: int) → int

Calculate the population size for the current iteration

Parameters:

iteration (int) – Current iteration number

Returns:

Population size for the current iteration

Return type:

int

class porcupy.utils.population.SelectionStrategies

Class implementing various selection strategies for population reduction.

This class provides methods for selecting which individuals to keep when reducing the population size.

static best_selection(positions: numpy.ndarray, fitness: numpy.ndarray, new_size: int) → Tuple[numpy.ndarray, numpy.ndarray]

Select the best individuals based on fitness

Parameters:

• positions (numpy.ndarray) – Current positions of all individuals

• fitness (numpy.ndarray) – Current fitness values of all individuals

• new_size (int) – New population size

Returns:

A tuple containing (selected_positions, selected_fitness)

Return type:

tuple

static tournament_selection(positions: numpy.ndarray, fitness: numpy.ndarray, new_size: int, tournament_size: int = 3) → Tuple[numpy.ndarray, numpy.ndarray]

Select individuals using tournament selection

Parameters:

• positions (numpy.ndarray) – Current positions of all individuals

• fitness (numpy.ndarray) – Current fitness values of all individuals

• new_size (int) – New population size

• tournament_size (int) – Size of each tournament

Returns:

A tuple containing (selected_positions, selected_fitness)

Return type:

tuple

static roulette_wheel_selection(positions: numpy.ndarray, fitness: numpy.ndarray, new_size: int) → Tuple[numpy.ndarray, numpy.ndarray]

Select individuals using roulette wheel selection

Parameters:

• positions (numpy.ndarray) – Current positions of all individuals

• fitness (numpy.ndarray) – Current fitness values of all individuals

• new_size (int) – New population size

Returns:

A tuple containing (selected_positions, selected_fitness)

Return type:

tuple

static diversity_selection(positions: numpy.ndarray, fitness: numpy.ndarray, new_size: int, elite_fraction: float = 0.2) → Tuple[numpy.ndarray, numpy.ndarray]

Select individuals based on fitness and diversity

Parameters:

• positions (numpy.ndarray) – Current positions of all individuals

• fitness (numpy.ndarray) – Current fitness values of all individuals

• new_size (int) – New population size

• elite_fraction (float) – Fraction of individuals to select based on fitness

Returns:

A tuple containing (selected_positions, selected_fitness)

Return type:

tuple