Efficient Simulation Budget Allocation With Bound Information

This paper proposes a bound-based simulation budget allocation (BSBA) procedure for solving ranking and selection (R&S) problems in simulation optimization. For many practical applications, strict bounds on system performances can be obtained through empirical and theoretical approaches. These b...

全面介紹

Saved in:
書目詳細資料
Main Authors: Li, Haidong, Xu, Xiaoyun, Zhao, Yaping
格式: text
出版: Archīum Ateneo 2019
主題:
在線閱讀:https://archium.ateneo.edu/gsb-pubs/3
https://ieeexplore.ieee.org/abstract/document/8675337?casa_token=1B5V8SdZvwYAAAAA:azZh4w_1RqMdeCy4F7opiSxV0r8aSZiBhs_ZWdYuugSbDTe96OasEUlRSG2tqld5SQVTEkSbZbw
標簽: 添加標簽
沒有標簽, 成為第一個標記此記錄!
實物特徵
總結:This paper proposes a bound-based simulation budget allocation (BSBA) procedure for solving ranking and selection (R&S) problems in simulation optimization. For many practical applications, strict bounds on system performances can be obtained through empirical and theoretical approaches. These bounds provide additional information which may help solve R&S problems. In this paper, a new method of objective function estimation is proposed using both bound information and simulation outputs. This new estimation method is demonstrated to be particularly effective. To solve R&S problems, several asymptotic optimal allocation rules are also derived. Using these allocation rules, a BSBA procedure is proposed to achieve high efficiency in identifying the best design. Numerical experiments are provided to examine the performance of the proposed BSBA procedure. The computational results show that BSBA outperforms three compared allocation procedures, especially when bounds are tight or the simulation budget is small.