資訊專欄INFORMATION COLUMN
摘要：是一個遺傳算法框架，這里是它的簡介。就是這個評價功能函數得自己寫，其實，也就是適應度函數了就是注意評價函數返回值必須是可迭代的。完整性的目的我們將開發完整的分代算法。對于同一個種子值的輸入，之后產生的隨機數序列也一樣。
DEAP-Overview
DEAP是一個python遺傳算法框架，這里是它的簡介。DEAP documentation
今天整理一下DEAP的概覽，大體了解一下它的流程。初學，不嚴謹，僅作為自己的備忘學習筆記。
The first thing to do is to think of the appropriate type for your problem.This is done with the creator module.
第一件事就是思考你問題的合適類型(是什么樣的優化問題，最小值？最大值？單目標？多目標？）。在creator模塊中可以實現。
For example, the following creates a FitnessMin class for a minimization problem and an Individual class that is derived from a list with a fitness attribute set to the just created fitness.
例如,下面為一個最小化問題創建一個FitnessMin類和個體類，它來自于剛剛創建的適應度帶有fitness屬性值集合的列表。
from deap import base, creator
creator.create("FitnessMin", base.Fitness, weights=(-1.0,))
creator.create("Individual", list, fitness=creator.FitnessMin)
That’s it. More on creating types can be found in the Creating Types tutorial.
更多關于創建類型的問題可以在Creating Types tutorial查看。
Once the types are created you need to fill them with sometimes random values, sometime guessed ones. Again, DEAP provides an easy mechanism to do just that.
一旦這些類型被創建，你就需要用一些隨機的值來填充它們，有時是猜測值。同樣的， DEAP極大的提供了這樣的一個簡單的機制來做這件事兒。
The Toolbox is a container for tools of all sorts including initializers that can do what is needed of them. The following takes on the last lines of code to create the initializers for individuals containing random floating point numbers and for a population that contains them.
toolbox是一個工具容器，可用來包含各種“初始化體”（可以做它們需要的事兒）。以下是以代碼來創建包含隨機浮點數的初始化和個人人口包含他們最后的線。下面最后幾行代碼用來創建“初始化體”，為包含那些隨機值的“個體”和包含它們的種群。
import random
from deap import tools
IND_SIZE = 10
toolbox = base.Toolbox()
toolbox.register("attribute", random.random)
toolbox.register("individual", tools.initRepeat, creator.Individual,
toolbox.attribute, n=IND_SIZE)
toolbox.register("population", tools.initRepeat, list, toolbox.individual) #是吧，創建的individual放入了population中
This creates functions to initialize populations from individuals that are themselves initialized with random float numbers. The functions are registered in the toolbox with their default arguments under the given name.
這將創建函數來初始化種群（來自那些用隨機浮點數初始化而成的個體們）。這些函數在給定名字的默認參數下被“注冊到”toolbox中，
register感覺這個單詞在這兒的理解可以看做是“創建”的意思，然后toolbox中就有了toolbox.individual。具體理解，還得看后面的詳細介紹。
toolbox貌似和function有關，里邊的第一個參數就是函數名。因為在Algorithm步驟里時，出現了這樣的用法toolbox.population()。
More initialization methods are found in the Creating Types tutorial and the various Examples.
更多初始化的方法可以在 Creating Types tutorial和Examples中查看。
toolbox.register("mate", tools.cxTwoPoint)
Operators都在toolbox模塊里，要給每個算子選擇合適算法。在Operator里邊可以給每個步驟的算子分別選擇合適的算法。
In addition you must create your evaluation function. This is how it is done in DEAP.
就是這個評價功能函數得自己寫，其實，也就是適應度函數了！
Note also that fitness values must be iterable, that is why we return a tuple in the evaluate function.
就是注意評價函數返回值必須是可迭代的。
def evaluate(individual):
return sum(individual),
toolbox.register("mate", tools.cxTwoPoint)
toolbox.register("mutate", tools.mutGaussian, mu=0, sigma=1, indpb=0.1)
toolbox.register("select", tools.selTournament, tournsize=3)
toolbox.register("evaluate", evaluate)
四. Algorithms
Now that everything is ready, we can start to write our own algorithm. It is usually done in a main function. For the purpose of completeness we will develop the complete generational algorithm.
現在一切具備，我們可以開始寫自己的算法了。通常在主函數里邊寫。完整性的目的,我們將開發完整的分代算法。
def main():
pop = toolbox.population(n=50)
CXPB, MUTPB, NGEN = 0.5, 0.2, 40
# Evaluate the entire population
fitnesses = map(toolbox.evaluate, pop)
for ind, fit in zip(pop, fitnesses):
ind.fitness.values = fit
for g in range(NGEN):
# Select the next generation individuals
offspring = toolbox.select(pop, len(pop))
# Clone the selected individuals
offspring = map(toolbox.clone, offspring)
# Apply crossover and mutation on the offspring
for child1, child2 in zip(offspring[::2], offspring[1::2]):
if random.random() < CXPB:
toolbox.mate(child1, child2)
del child1.fitness.values
del child2.fitness.values
for mutant in offspring:
if random.random() < MUTPB:
toolbox.mutate(mutant)
del mutant.fitness.values
# Evaluate the individuals with an invalid fitness
invalid_ind = [ind for ind in offspring if not ind.fitness.valid]
fitnesses = map(toolbox.evaluate, invalid_ind)
for ind, fit in zip(invalid_ind, fitnesses):
ind.fitness.values = fit
# The population is entirely replaced by the offspring
pop[:] = offspring
return pop
縱觀整個步驟，可以發現DEAP在實現GA過程中的思路一般是很清晰的。
確定Types--creator創建，選好解決問題的類型
初始化Initialization--toolbox注冊個體啊，種群啊等等函數。
Operator--算子選擇，交叉，變異，選擇，進化等等。每個算子都有不同的算法，可以選擇的！
Algorithms--算法就是具體將上面注冊的函數啊，等等應用結合起來，編寫流程。
例子
