File that brings together all morphological evolution related function like the general evolution process and all related functions for mutation, crossover and selection. More...

Functions
List[Arm]	morphevo.evolution.run_evolution (Optional[List[Arm]] children=None)
	Evolve arms with following scheme: More...

List[Arm]	morphevo.evolution.selection (Callable selection_function, List[Arm] population)
	Do a selection on a list of parents given a selection function. More...

List[Arm]	morphevo.evolution.mutate (Callable mutation_function, List[Arm] parents)
	Do a mutation on a list of parents given a mutation function. More...

List[Arm]	morphevo.evolution.selection_fitness (List[Arm] population)
	Do selection on the fitness of the arm. More...

List[Arm]	morphevo.evolution.selection_success_rate (List[Arm] population)
	Do selection on the success_rate of the arm. More...

List[Arm]	morphevo.evolution.selection_fitness_diversity (List[Arm] population)
	Do selection in a fitness-diversity way. More...

Arm	morphevo.evolution.select_next_parent (List[Arm] population, List[Arm] parents)
	Select the next parent that will be added to the parents. More...

float	morphevo.evolution.calculate_fitness (Arm arm)
	Calculate the fitness of a genome by calculating its coverage. More...

float	morphevo.evolution.calculate_diversity (Arm arm, List[Arm] others)
	calculate how diverse an arm is compared to others. More...

List[float]	morphevo.evolution.calculate_selection_scores (List[float] population_fitnesses, List[float] population_diversities)
	Calculate a score on which selection can be done for fitness-diversity selection. More...

List[Arm]	morphevo.evolution.mutate_with_crossover (List[Arm] parents)
	Do mutation with morphevo configuration parameters. More...

List[Arm]	morphevo.evolution.mutate_with_crossover_coevolution (List[Arm] parents)
	Do mutation with coevolution configuration parameters. More...

def	morphevo.evolution.create_crossover_children (List[Arm] parents, int amount)
	Function to create crossover children, pick 2 random parents, check if they are the same. More...

def	morphevo.evolution.save_best_genome (Arm arm, int generation)
	Save the genome of an arm to a file. More...

Detailed Description

File that brings together all morphological evolution related function like the general evolution process and all related functions for mutation, crossover and selection.

Function Documentation

◆ calculate_diversity()

float morphevo.evolution.calculate_diversity	(	Arm	arm,
		List[Arm]	others
	)

calculate how diverse an arm is compared to others.

(others is probably the already selected parents in the fitness-diversity selection).

Parameters

arm	The arm you want to compare to others to calculate diversity score.
others	The other arms you want to compare to.

Returns: Average diversity compared to others.

◆ calculate_fitness()

float morphevo.evolution.calculate_fitness ( Arm arm )

Calculate the fitness of a genome by calculating its coverage.

Parameters

arm	The arm of which you want to calculate fitness.

◆ calculate_selection_scores()

List[float] morphevo.evolution.calculate_selection_scores	(	List[float]	population_fitnesses,
		List[float]	population_diversities
	)

Calculate a score on which selection can be done for fitness-diversity selection.

First normalize fitness and diversity scores. Then calculate the distance to (1,1). If fitness and diversity are 1,1 this means 100% fit and 100% diverse from all previously selected parents.

Parameters

population_fitnesses	The fitnesses of a population.
population_diversities	The diversities of a population.

Returns: List of scores that combined fitness and diversity.

◆ create_crossover_children()

def morphevo.evolution.create_crossover_children	(	List[Arm]	parents,
		int	amount
	)

Function to create crossover children, pick 2 random parents, check if they are the same.

If not do the crossover else pick again.

Parameters

parents	A list of parents to choose from.
amount	The amount of crossover children you want.

Returns: List of children resulting from crossover.

◆ mutate()

List[Arm] morphevo.evolution.mutate	(	Callable	mutation_function,
		List[Arm]	parents
	)

Do a mutation on a list of parents given a mutation function.

Parameters

mutation_function	The function used for mutation.
parents	The parents that will be mutated.

Returns: Mutated parents.

◆ mutate_with_crossover()

List[Arm] morphevo.evolution.mutate_with_crossover ( List[Arm] parents )

Do mutation with morphevo configuration parameters.

First make normal children by mutating parents, after that crossover children are added if specified in config.

Parameters

parents The parents of which you want to make children.

Returns: List of mutated children.

◆ mutate_with_crossover_coevolution()

List[Arm] morphevo.evolution.mutate_with_crossover_coevolution ( List[Arm] parents )

Do mutation with coevolution configuration parameters.

First make normal children by mutating parents, after that crossover children are added if specified in config.

Parameters

parents The parents of which you want to make children.

Returns: List of mutated children.

◆ run_evolution()

List[Arm] morphevo.evolution.run_evolution ( Optional[List[Arm]] children = None )

Evolve arms with following scheme:

evaluate arms
select best arms (selection includes best arms previous generation)
mutate best arms
restart loop with mutated arms
Parameters

children Optional list of children to start from. Else children will be randomly generated.

Returns
List of evolved arms.

◆ save_best_genome()

def morphevo.evolution.save_best_genome	(	Arm	arm,
		int	generation
	)

Save the genome of an arm to a file.

Parameters

arm	The arm of which you want to save the genome.
generation	The generation in which the genome was evaluated.

◆ select_next_parent()

Arm morphevo.evolution.select_next_parent	(	List[Arm]	population,
		List[Arm]	parents
	)

Select the next parent that will be added to the parents.

Do this by making a consideration between fitness and diversity.

Parameters

population	Current population.
parents	Current list of parents.

Returns: The parent that scores the highest on fitness-diversity compared to already selected parents.

◆ selection()

List[Arm] morphevo.evolution.selection	(	Callable	selection_function,
		List[Arm]	population
	)

Do a selection on a list of parents given a selection function.

Parameters

selection_function	The function used for selection.
population	The parents that will be selected from.

Returns: Selected parents.

◆ selection_fitness()

List[Arm] morphevo.evolution.selection_fitness ( List[Arm] population )

Do selection on the fitness of the arm.

Parameters

population The population on which you want to do selection.

Returns: List of selected parents.

◆ selection_fitness_diversity()

List[Arm] morphevo.evolution.selection_fitness_diversity ( List[Arm] population )

Do selection in a fitness-diversity way.

Parameters

population The population on which you want to do selection.

Returns: List of selected parents.

◆ selection_success_rate()

List[Arm] morphevo.evolution.selection_success_rate ( List[Arm] population )

Do selection on the success_rate of the arm.

This is for the coevolution loop, success_rate is calculated in the rl step.

Parameters

population The population on which you want to do selection.

Returns: List of selected parents.

Functions

Detailed Description

Function Documentation

◆ calculate_diversity()

◆ calculate_fitness()

◆ calculate_selection_scores()

◆ create_crossover_children()

◆ mutate()

◆ mutate_with_crossover()

◆ mutate_with_crossover_coevolution()

◆ run_evolution()

◆ save_best_genome()

◆ select_next_parent()

◆ selection()

◆ selection_fitness()

◆ selection_fitness_diversity()

◆ selection_success_rate()