mealpy.utils package

• mealpy.utils.visualize
  • mealpy.utils.visualize.linechart

mealpy.utils.history

class mealpy.utils.history.History

Bases: object

A History class is responsible for saving each iteration’s output.

Notes

• Access to variables in this class:

  • list_global_best: List of global best SOLUTION found so far in all previous generations

  • list_current_best: List of current best SOLUTION in each previous generations

  • list_epoch_time: List of runtime for each generation

  • list_global_best_fit: List of global best FITNESS found so far in all previous generations

  • list_current_best_fit: List of current best FITNESS in each previous generations

  • list_diversity: List of DIVERSITY of swarm in all generations

  • list_exploitation: List of EXPLOITATION percentages for all generations

  • list_exploration: List of EXPLORATION percentages for all generations

  • list_population: List of POPULATION in each generations

  • Warning, the last variable ‘list_population’ can cause the error related to ‘memory’ when using pickle to save model. Better to delete that variable or assign to empty list [] to reduce the ‘memory’.

• There are 8 methods to draw available in this class:

  • save_global_best_fitness_chart()

  • save_local_best_fitness_chart()

  • save_global_objectives_chart()

  • save_local_objectives_chart()

  • save_exploration_exploitation_chart()

  • save_diversity_chart()

  • save_runtime_chart()

  • save_trajectory_chart()

Examples

>>> import numpy as np
>>> from mealpy.swarm_based.PSO import BasePSO
>>>
>>> def fitness_function(solution):
>>>     return np.sum(solution**2)
>>>
>>> problem_dict = {
>>>     "fit_func": fitness_function,
>>>     "lb": [-10, -15, -4, -2, -8],
>>>     "ub": [10, 15, 12, 8, 20],
>>>     "minmax": "min",
>>>     "verbose": True,
>>> }
>>> model = BasePSO(problem_dict, epoch=1000, pop_size=50)
>>>
>>> model.history.save_global_objectives_chart(filename="hello/goc")
>>> model.history.save_local_objectives_chart(filename="hello/loc")
>>> model.history.save_global_best_fitness_chart(filename="hello/gbfc")
>>> model.history.save_local_best_fitness_chart(filename="hello/lbfc")
>>> model.history.save_runtime_chart(filename="hello/rtc")
>>> model.history.save_exploration_exploitation_chart(filename="hello/eec")
>>> model.history.save_diversity_chart(filename="hello/dc")
>>> model.history.save_trajectory_chart(list_agent_idx=[3, 5], list_dimensions=[3], filename="hello/tc")
>>>
>>> ## Get list of population after all generations
>>> print(model.history.list_population)

get_global_repeated_times(id_fitness, id_target, epsilon)

save_diversity_chart(title='Diversity Measurement Chart', algorithm_name='GA', filename='diversity-chart', verbose=True)

save_exploration_exploitation_chart(title='Exploration vs Exploitation Percentages', list_colors=('blue', 'orange'), filename='exploration-exploitation-chart', verbose=True)

save_global_best_fitness_chart(title='Global Best Fitness', color='b', x_label='#Iteration', y_label='Function Value', filename='global-best-fitness-chart', verbose=True)

save_global_objectives_chart(title='Global Objectives Chart', x_label='#Iteration', y_label='Function Value', filename='global-objectives-chart', verbose=True)

save_initial_best(best_agent)

save_local_best_fitness_chart(title='Local Best Fitness', color='b', x_label='#Iteration', y_label='Function Value', filename='local-best-fitness-chart', verbose=True)

save_local_objectives_chart(title='Local Objectives Chart', x_label='#Iteration', y_label='Objective Function Value', filename='local-objectives-chart', verbose=True)

save_runtime_chart(title='Runtime chart', color='b', x_label='#Iteration', y_label='Second', filename='runtime-chart', verbose=True)

save_trajectory_chart(title='Trajectory of some first agents after generations', list_agent_idx=(1, 2, 3), list_dimensions=(1, 2), filename='trajectory-chart', verbose=True)

mealpy.utils.problem

class mealpy.utils.problem.Problem(**kwargs)

Bases: object

Define the mathematical form of optimization problem

Notes

• fit_func (callable): your fitness function

• lb (list, int, float): lower bound, should be list of values

• ub (list, int, float): upper bound, should be list of values

• minmax (str): “min” or “max” problem (Optional, default = “min”)

• verbose (bool): print out the training process or not (Optional, default = True)

• n_dims (int): number of dimensions / problem size (Optional)

• obj_weight: list weights for all your objectives (Optional, default = [1, 1, …1])

• problem (dict): dictionary of the problem (contains at least the parameter 1, 2, 3) (Optional)

• amend_position(callable): Depend on your problem, may need to design an amend_position function (Optional for continuous domain, Required for discrete domain)

Examples

>>> ## 1st way:
>>> import numpy as np
>>> from mealpy.swarm_based.PSO import BasePSO
>>>
>>> def fitness_function(solution):
>>>     return np.sum(solution**2)
>>>
>>> problem_dict = {
>>>     "fit_func": fitness_function,
>>>     "lb": [-10, -15, -4, -2, -8],
>>>     "ub": [10, 15, 12, 8, 20],
>>>     "minmax": "min",
>>>     "verbose": True,
>>> }
>>> model1 = BasePSO(problem_dict, epoch=1000, pop_size=50)
>>>
>>> ## 2nd way:
>>> from mealpy.utils.problem import Problem
>>>
>>> problem_obj2 = Problem(fit_func=fitness_function, lb=[-10, -15, -4, -2, -8], ub=[10, 15, 12, 8, 20], minmax="min", verbose=True)
>>> model3 = BasePSO(problem_obj2, epoch=1000, pop_size=50)

DEFAULT_LB = -1

DEFAULT_UB = 1

ID_MAX_PROB = -1

ID_MIN_PROB = 0

amend_position(position=None)

Depend on what kind of problem are we trying to solve, there will be an different amend_position function to rebound the position of agent into the valid range.

Parameters

position – vector position (location) of the solution.

Returns

Amended position (make the position is in bound)

mealpy.utils.termination

class mealpy.utils.termination.Termination(**kwargs)

Bases: object

Define the Stopping Condition (Termination) for the Optimizer

Notes

• By default, the stopping condition is maximum generations (epochs/iterations) in Optimizer class.

• By using this class, the default termination will be overridden

• In general, there are 4 termination cases: FE, MG, ES, TB

  • FE: Number of Function Evaluation

  • MG: Maximum Generations / Epochs - This is default in all algorithms

  • ES: Early Stopping - Same idea in training neural network (If the global best solution not better an epsilon after K epochs then stop the program)

  • TB: Time Bound - You just want your algorithm run in K seconds. Especially when comparing different algorithms.

• Parameters for Termination class

  • mode (str): FE, MG, ES or TB

  • quantity (int): value for termination type

  • problem (dict): dictionary of the termination (contains at least the parameter ‘mode’ and ‘quantity’) (Optional)

Examples

>>> ## 1st way to define and use termination object
>>> import numpy as np
>>> from mealpy.swarm_based.PSO import BasePSO
>>>
>>> def fitness_function(solution):
>>>     return np.sum(solution**2)
>>>
>>> problem_dict = {
>>>     "fit_func": fitness_function,
>>>     "lb": [-10, -15, -4, -2, -8],
>>>     "ub": [10, 15, 12, 8, 20],
>>> }
>>>
>>> term_dict = {
>>>     "mode": "FE",
>>>     "quantity": 100000  # 100000 number of function evaluation
>>> }
>>> model1 = BasePSO(problem_dict, epoch=1000, pop_size=50, termination=term_dict)
>>>
>>> ## 2nd and 3rd ways:
>>> from mealpy.utils.termination import Termination
>>>
>>> term_obj2 = Termination(termination = term_dict)
>>> model2 = BasePSO(problem_dict, epoch=1000, pop_size=50, termination=term_obj2)
>>>
>>> term_obj3 = Termination(termination = term_dict)
>>> model3 = BasePSO(problem_dict, epoch=1000, pop_size=50, termination=term_obj3)

DEFAULT_MAX_ES = 20

DEFAULT_MAX_FE = 100000

DEFAULT_MAX_MG = 1000

DEFAULT_MAX_TB = 20

logging(verbose=True)