求帝王蝶优化算法（MBOA）Python代码及结果报错解决指导

以下是帝王蝶优化算法的Python实现，以经典Sphere函数作为测试目标函数，你可根据需求替换为实际优化问题的目标函数：

import numpy as np

def sphere_function(x):
    """测试用Sphere目标函数，求最小值"""
    return np.sum(x ** 2)

def monarch_butterfly_optimization(obj_func, dim, bounds, max_iter, npop=50, p=0.7):
    """
    帝王蝶优化算法实现
    参数:
        obj_func: 目标函数，输入个体向量，输出适应度值
        dim: 变量维度
        bounds: 变量取值范围，格式为[(min1, max1), (min2, max2), ..., (dim, maxdim)]
        max_iter: 最大迭代次数
        npop: 种群规模，默认50
        p: 迁徙比例，默认0.7（p*Npop个个体参与迁徙操作）
    返回:
        best_fitness: 最优适应度值
        best_position: 最优位置向量
    """
    # 初始化种群
    lower_bound = np.array([b[0] for b in bounds])
    upper_bound = np.array([b[1] for b in bounds])
    population = np.random.uniform(low=lower_bound, high=upper_bound, size=(npop, dim))
    
    # 计算初始适应度
    fitness = np.array([obj_func(ind) for ind in population])
    best_idx = np.argmin(fitness)
    best_fitness = fitness[best_idx]
    best_position = population[best_idx].copy()
    
    for iter_num in range(max_iter):
        # 迁徙操作（Migration operator）
        num_migrate = int(np.round(p * npop))
        migrate_indices = np.random.choice(npop, num_migrate, replace=False)
        
        for idx in migrate_indices:
            source_idx = np.random.choice([i for i in range(npop) if i != idx])
            population[idx] = population[source_idx] + np.random.uniform(-1, 1, dim) * (best_position - population[source_idx])
            population[idx] = np.clip(population[idx], lower_bound, upper_bound)
        
        # 调整操作（Adjustment operator）
        adjust_indices = np.array([i for i in range(npop) if i not in migrate_indices])
        
        for idx in adjust_indices:
            r1, r2 = np.random.choice([i for i in range(npop) if i != idx], 2, replace=False)
            population[idx] = population[idx] + np.random.uniform(-1, 1, dim) * (population[r1] - population[r2])
            population[idx] = np.clip(population[idx], lower_bound, upper_bound)
        
        # 更新最优解
        fitness = np.array([obj_func(ind) for ind in population])
        current_best_idx = np.argmin(fitness)
        current_best_fitness = fitness[current_best_idx]
        if current_best_fitness < best_fitness:
            best_fitness = current_best_fitness
            best_position = population[current_best_idx].copy()
    
    return best_fitness, best_position

# 测试示例
if __name__ == "__main__":
    dim = 30
    bounds = [(-100, 100)] * dim
    max_iter = 100
    
    best_fit, best_pos = monarch_butterfly_optimization(sphere_function, dim, bounds, max_iter)
    
    print(f"最优适应度值: {best_fit:.6f}")
    print(f"最优位置向量: {best_pos}")

关键检查点 #

若你自己的代码结果错误，可对比上述实现排查：

• 迁徙、调整阶段的核心公式是否正确
• 个体更新后的边界处理是否到位
• 适应度计算是否匹配你的优化目标（求最小/最大值）
• 种群迭代过程中的索引操作是否无逻辑错误

内容的提问来源于stack exchange，提问作者ezabil