geneticengine.problems

Submodules

Package Contents

Classes

Fitness

Problem

An Abstract class that SingleObjectiveProblem and MultiObjectiveProblem

SingleObjectiveProblem

SingleObjectiveProblem is a class that extends the Problem class.

MultiObjectiveProblem

MultiObjectiveProblem is a class that extends the Problem class.

Functions

wrap_depth_minimization(p)

This wrapper takes a SingleObjectiveProblem and adds a penalty for

Attributes

P

class geneticengine.problems.Fitness

Bases: NamedTuple

maximizing_aggregate: float
fitness_components: list[float]
__str__()

Return str(self).

geneticengine.problems.P
class geneticengine.problems.Problem

Bases: abc.ABC

An Abstract class that SingleObjectiveProblem and MultiObjectiveProblem extends to.

Parameters:

  • minimize (bool | list[bool]) – When switch on, the fitness function is reversed, so that a higher result from the fitness function corresponds to a less fit solution.

  • fitness_function (Callable[[P], float] | Callable[[P], list[float]]) – The fitness function. Should take in any valid individual and return a float or a list of floats, depending if its a single objetive problem or a multi objective problem.

minimize: list[bool] | bool
epsilon: float
abstract evaluate(phenotype)
Parameters:

phenotype (P)

Return type:

Fitness

key_function(a)

Returns the (maximizing) fitness of the individual as a single float.

Parameters:

a (Fitness)

Return type:

float

is_better(a, b)

Returns whether the first fitness is better than the second.

Parameters:
Return type:

bool

abstract number_of_objectives()
Return type:

int

class geneticengine.problems.SingleObjectiveProblem(fitness_function, minimize=False)

Bases: Problem

SingleObjectiveProblem is a class that extends the Problem class.

Parameters:

  • minimize (bool) – When switch on, the fitness function is reversed, so that a higher result from the fitness function corresponds to a less fit solution.

  • fitness_function (Callable[[P], float]) – The fitness function. Should take in any valid individual and return a float.

evaluate(phenotype)
Parameters:

phenotype (P)

Return type:

Fitness

number_of_objectives()
Return type:

int

class geneticengine.problems.MultiObjectiveProblem(minimize, fitness_function, best_individual_criteria_function=None, aggregate_fitness=None)

Bases: Problem

MultiObjectiveProblem is a class that extends the Problem class.

Parameters:

  • minimize (list[bool] | bool) – When switch on, the fitness function is reversed, so that a higher result from the fitness function corresponds to a less fit solution. When a list is passed, each element of the list corresponds to a fitness component. When a bool is passed all the fitness component of the problem are minimized or maximized. when giving a bool, the selection algorithm must have that into consideration and create a list of bools with the same size as the number of the fitness components of an Individual.

  • fitness_function (Callable[[P], list[bool]]) – The fitness function. Should take in any valid individual and return a list of float.

  • best_individual_criteria_function (Optional(Callable[[P], float]) – This function allow the user to choose how to find the best individual in a generation (default = None , this means that the individual with the best fitness is the one considered as the best in that generation)

  • aggregate_fitness (Callable[[list[float]], float] | None)

minimize: list[bool] | bool
ff: dict[str, Any]
evaluate(phenotype)
Parameters:

phenotype (P)

Return type:

Fitness

number_of_objectives()
Return type:

int

geneticengine.problems.wrap_depth_minimization(p)

This wrapper takes a SingleObjectiveProblem and adds a penalty for bigger trees.

Parameters:

p (SingleObjectiveProblem)

Return type:

SingleObjectiveProblem