如何在 DEAP 中将参数传递给"toolbox.population"

Question

我尝试在DEAP中自定义"个人"。原因是个体是由几个解释变量组成的。每个解释变量都有下限和上限。此外，它可能具有步长或精度。

这是我正在处理的代码：

import random
from deap import base, creator, tools
from typing import Optional, List, Any

creator.create("FitnessMin", base.Fitness, weights=(-1.0,))   
creator.create("Individual", list, fitness=creator.FitnessMin)   # individual data-type : List
toolbox = base.Toolbox()

'''
Functions
'''
def random_pick(lower: float, upper: float, step: Optional[int] = None) -> float:
# Pick random numbers within specified bounds 
if step is None:
return random.uniform(lower, upper)
else:
candidates_num = round((upper - lower) / step)
return lower + random.randint(0, candidates_num) * step

def make_individual(explanatories) -> List[Any]:
individual = []
for explanatory in explanatories:
individual.append(
random_pick(
lower=explanatories[explanatory][0],
upper=explanatories[explanatory][1],
step=explanatories[explanatory][2] if len(explanatories[explanatory]) == 3 else None,
)
)
return creator.Individual(individual)

'''
Main
'''
pop_size = 10
explanatories={
"a": (30, 100, 1),   # (low, high, step)
"b": (80, 200, 1),
"c": (15, 80, 1),
"d": (1.5, 5, 0.1),
"e": (15, 200, 1),
"f": (5, 1699, 1),
"g": (0.5, 0.5),     # (low, high)
}

toolbox.register("individual", make_individual)
# Check value of individual
ind = toolbox.individual(explanatories)   # list

# Population (following is the Bag type of population which is common)
toolbox.register("population", tools.initRepeat, list, toolbox.individual)
# Check population
pop = toolbox.population(pop_size)   # ERROR  

函数random_pick通过考虑步长随机选取下限和上限内所有解释的值。而职能make_individual负责使个性化。

上述代码的问题在于我试图使人口。当我运行以下行时：

pop = toolbox.population(pop_size)    

我收到错误：

TypeError: make_individual() missing 1 required positional argument: 'explanatories'

我的问题是，在制作人口时，我如何将解释作为输入参数传递给"工具箱.population"或"make_individual()"。

谢谢你的帮助。

Answer 1

您的问题似乎源于tools.initRepeat只接受三个参数：container、func和n(文档在这里)。

因此，它不能将explanatories参数"传递"给func(在本例中为toolbox.individual)。

您可以通过在注册toolbox.individual时将explanatories指定为make_individual参数来解决此问题。

也就是说，这有效：

toolbox.register(
"individual", 
make_individual,
explanatories=explanatories  # now toolbox.individual uses default explanatories
)
ind = toolbox.individual()  # no longer provide explanatories arg
toolbox.register("population", tools.initRepeat, list, toolbox.individual)
pop = toolbox.population(pop_size)
print(int)
print(pop)

它打印的东西类似于

<class 'int'>
[[71, 122, 50, 1.9, 62, 679, 0.5], [88, 192, 24, 3.4000000000000004, 40, 1377, 0.5], [78, 182, 59, 4.1, 65, 1297, 0.5], [39, 174, 49, 3.0, 104, 883, 0.5], [89, 153, 36, 5.0, 102, 718, 0.5], [57, 114, 23, 2.7, 54, 1628, 0.5], [78, 146, 32, 2.1, 145, 961, 0.5], [96, 174, 40, 5.0, 19, 974, 0.5], [39, 150, 80, 4.7, 140, 801, 0.5], [31, 150, 37, 2.8, 109, 1230, 0.5]]