An Efficient Simulation Budget Allocation Method Incorporating Regression for Partitioned Domains.

An Efficient Simulation Budget Allocation Method Incorporating Regression for Partitioned Domains.
复制标题

DOI:
10.1016/j.automatica.2014.03.011
复制
发表时间:
2014-05-01
期刊:
影响因子:
6.4
通讯作者:
Xu, Jie
Xu, Jie
中科院分区:
计算机科学2区
文献类型:
--
作者:
Brantley, Mark W.;Lee, Loo Hay;Chen, Chun-Hung;Xu, Jie

文献摘要

参考文献

被引文献

相似文献

模拟可以是一个非常强大的工具,以帮助决策在许多应用程序中,但探索多个过程的行动可能是耗时的。许多排名和选择(R&S)的程序已经开发,以提高找到最佳设计的仿真效率。为了进一步提高效率,一种方法是将来自整个域的信息合并到回归方程中。然而,回归元模型的使用也继承了大多数回归方法的一些典型假设,例如基础二次函数的假设,并且模拟噪声在感兴趣的域中是均匀的。为了扩展限制,同时保持效率的好处,我们建议划分感兴趣的域,使得在每个分区的基础函数的平均值是近似二次的。我们的新方法为分区之间和分区内提供了近似最优的规则,这些规则确定了分配给每个设计位置的样本数量。目标是最大化正确选择最佳设计的概率。数值实验表明,我们的新方法可以显着提高效率比现有的高效R&S方法。
Simulation can be a very powerful tool to help decision making in many applications but exploring multiple courses of actions can be time consuming. Numerous ranking & selection (R&S) procedures have been developed to enhance the simulation efficiency of finding the best design. To further improve efficiency, one approach is to incorporate information from across the domain into a regression equation. However, the use of a regression metamodel also inherits some typical assumptions from most regression approaches, such as the assumption of an underlying quadratic function and the simulation noise is homogeneous across the domain of interest. To extend the limitation while retaining the efficiency benefit, we propose to partition the domain of interest such that in each partition the mean of the underlying function is approximately quadratic. Our new method provides approximately optimal rules for between and within partitions that determine the number of samples allocated to each design location. The goal is to maximize the probability of correctly selecting the best design. Numerical experiments demonstrate that our new approach can dramatically enhance efficiency over existing efficient R&S methods.
DOI: 10.1080/03610927808827671
发表时间: 1978-01-01
期刊: COMMUNICATIONS IN STATISTICS PART A-THEORY AND METHODS
影响因子: --
作者:
RINOTT, Y
通讯作者: RINOTT, Y
DOI: 10.1023/a:1008349927281
发表时间: 2000-07-01
影响因子: 2
作者:
Chen, CH;Lin, JW;Chick, SE
通讯作者: Chick, SE
DOI: 10.1214/09-aos787
发表时间: 2010-08-01
影响因子: 4.5
作者:
Yang, Min
通讯作者: Yang, Min
DOI: 10.1214/aos/1176347963
发表时间: 1991-03-01
影响因子: 4.5
作者:
FRIEDMAN, JH
通讯作者: FRIEDMAN, JH
DOI: 10.1080/07408170903116360
发表时间: 2010-01-01
期刊: IIE TRANSACTIONS
影响因子: --
作者:
Chen, Chun-Hung;Yuecesan, Enver;Chen, Hsiao-Chang
通讯作者: Chen, Hsiao-Chang