| 一类由概率分布生成的新度量 |
| |
| 引用本文： | 卢国祥,沈世镒.一类由概率分布生成的新度量[J].系统科学与数学,2009,29(4):451-459. |
| |
| 作者姓名： | 卢国祥 沈世镒 |
| |
| 作者单位： | 南开大学数学科学学院,天津,300071 |
| |
| 基金项目： | 天津市科技发展计划项目 |
| |
| 摘 要： | 主要得到了一类由概率分布生成的新度量.以信息理论中的重要概念-相对熵为基础,对前人文章中的重要结论进行推广,通过利用改进的初等方法在离散的可测空间中得到了这类新度量,并且证明了得到的新度量成立的充要条件.由此再将这类新度量推广到连续的可测空间中,得到了同样的结果.最后讨论了新度量的最值问题.
|
| 关 键 词： | 相对熵 度量 三角不等式 |
| 收稿时间： | 2007-11-1 |
| 修稿时间： | 2008-5-26 |
A Class of New Metrics Generated by Probability Distributions |
| |
| LU Guoxiang,SHEN Shiyi.A Class of New Metrics Generated by Probability Distributions[J].Journal of Systems Science and Mathematical Sciences,2009,29(4):451-459. |
| |
| Authors: | LU Guoxiang SHEN Shiyi |
| |
| Institution: | School of Mathematical Science, Nankai University, Tianjin 300071 |
| |
| Abstract: | In this paper, a class of new metrics generated by probability distributions is introduced. The new metrics, based on the important concept of relative entropy, are the generalization of the existing conclusions. Using the improved elementary method, the new metrics are obtained in a discrete measurable space. The sufficient and necessary conditions for the new merics to hold are given. The new metrics is then generalized to a continuous measurable space. Finally, the maxima and minima of the new metrics are discussed. |
| |
| Keywords: | Relative entropy metric triangle inequality |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《系统科学与数学》浏览原始摘要信息 |
| 点击此处可从《系统科学与数学》下载免费的PDF全文 |
|
