Cross-entropy-based adaptive importance sampling using von Mises-Fisher mixture for high dimensional reliability analysis
详细信息 查看全文
- 作者：Ziqi Wanga ; Junho Songb ; junhosong@snu.ac.kr" class="auth_mail" title="E-mail the corresponding author
- 关键词：Adaptive sampling ; Cross entropy ; High dimensional space ; Importance sampling ; Monte Carlo simulations ; Structural reliability ; Von Mises-Fisher distribution
- 刊名：Structural Safety
- 出版年：2016
- 出版时间：March 2016
- 年：2016
- 卷：59
- 期：Complete
- 页码：42-52
- 全文大小：1713 K
文摘
- •
An adaptive importance sampling is developed using Kullback–Leibler cross entropy.
- •
The importance sampling density with minimum cross entropy is found by pre-samples.
- •
The von Mises-Fisher mixture model is utilized for applications to high dimensions.
- •
Simple formulas and rules are developed to facilitate minimizing the cross entropy.
- •
Superb performance is not affected by probability level, dimension or nonlinearity.