最近自学了一下粒子群算法的基本流程，写了一套可计算n维函数极值的python代码，对于边界条件和移动速度以及可视化还有待改进。仅供参考

import random
import math
# n维粒子群算法
class point(object):def __init__(self, n, position, p_best, p_best_position):self.speed = [0] * n                    # n维初始速度为0self.position = position                # 位置self.p_best = p_best                    # 个体最佳值self.p_best_position = p_best_position  # 最佳位置class main(object):def __init__(self, n, domain, peak, num, c1, c2, w, k):  # 几元函数，定义域，求最大或最小，粒子个数，学习因子c1和c2，惯性因子，迭代k次self.k = kself.w = wself.c1 = c1self.c2 = c2self.num = numself.domain = domainself.n = nif peak == "max":self.peak = Trueelse:self.peak = Falseself.point_list = []        # 定义包含所有粒子类的列表self.g_best = None          # 函数最佳值self.best_position = None   # 位置最佳值self.init_operation()       # 初始化位置和速度self.update()print("最佳位置：", self.best_position, "最{}值维：".format(peak), self.g_best)def init_operation(self):for i in range(self.num):  # 产生num个粒子position = []for j in range(self.n):position.append(random.uniform(self.domain[j][0], self.domain[j][1]))   # 生成一个定义域内的随机位置value = self.func(position)                                                 # 函数值if not self.best_position:                                                  # 是否为第一个self.g_best = valueself.best_position = positionelse:                                                                       # 判断大小,更新共享的最大值if self.peak:if self.g_best < value:self.g_best = valueself.best_position = positionelse:passelse:if self.g_best > value:self.g_best = valueself.best_position = positionelse:passself.point_list.append(point(self.n, position, value, position))                                    # 向粒子列表内添加初始化的粒子def update(self):for n in range(self.k):for i in range(self.num):                                                                           # 更新速度，第i个粒子for j in range(self.n):                                                                         # 遍历每个坐标，第j个坐标self.point_list[i].speed[j] += self.c1 * random.uniform(0, 1)\* (self.point_list[i].p_best_position[j] - self.point_list[i].position[j])\+ self.c2 * random.uniform(0, 1) * (self.best_position[j] - self.point_list[i].position[j])     # 计算速度self.point_list[i].position[j] += self.point_list[i].speed[j]if self.point_list[i].position[j] > self.domain[j][1]:self.point_list[i].position[j] = self.domain[j][1]if self.point_list[i].position[j] < self.domain[j][0]:self.point_list[i].position[j] = self.domain[j][0]if self.point_list[i].speed[j] > 10:                                                           # 限制速度最大为10self.point_list[i].speed[j] = 10value = self.func(self.point_list[i].position)if self.peak:if value > self.g_best:self.g_best = valueself.best_position = self.point_list[i].position.copy()                             # .copy非常重要else:passif value > self.point_list[i].p_best:self.point_list[i].p_best = valueself.point_list[i].p_best_position = self.point_list[i].position.copy()else:passelse:if value < self.g_best:self.g_best = valueself.best_position = self.point_list[i].position.copy()                             # .copy非常重要else:passif value < self.point_list[i].p_best:self.point_list[i].p_best = valueself.point_list[i].p_best_position = self.point_list[i].position.copy()else:passdef func(self, x):return 5.6 * math.sqrt(x[0]**2+(x[1]-5)**2)+6.0*math.sqrt((x[0]-15)**2+(x[2]-x[1])**2)+7.5*x[1]+21.5*math.sqrt(25+(8-x[2])**2)if __name__ == "__main__":main(3, [[-10, 12], [-100, 15], [-29, 20]], "min", 1000, 2, 2, 1, 1000)