自适应拟蒙特卡罗多极值优化方法

摘要

Quasi-Monte Carlo random search is useful in nondifferentiable optimization. By borrowing the ideas of population from genetic algorithms, we introduce an adaptive random search in quasi-Monte Carlo method(AQMC) for global optimization. The adaptive search technique enables local search to head for local extuema quickly. The low discrepancy of quasirandom sequence ensures that the function field be searched evenly and various local extrema including global extremum be found.%拟蒙特卡罗搜索方法能用来有效地解决不可微优化问题.借用遗传算法中种群的概念,介绍了一种解全局优化的拟蒙特卡罗自适应搜索算法.由于应用了自适应搜索技术,局部搜索能够快速找到局部极值.同时,拟随机序列的低偏差性保证了函数定义域能够被均匀地搜索,为找到多个局部极值包括全局极值提供了保证.