The integral term of the information entropy unadapted for being a measuring index
10.3969/j.issn.1005-202X.2001.01.026
- VernacularTitle:信息熵的积分项不宜用作为描述信息源的量度指标
- Author:
Hang YU
;
Yang HONG
;
Zhi QU
;
Li LIU
- Publication Type:Journal Article
- From:
Chinese Journal of Medical Physics
2001;18(1):57-59
- CountryChina
- Language:Chinese
-
Abstract:
The information entropy, H = - Pk log Pk, has been proved in this paper to be always positive and invariant for a linear coordinate transformation, so that it can be used as a measuring index. On the contrary, the integral term of the information entropy, H(x) = -f(x) log f(x) dx, is not always positive and invariant for a linear coordinate transformation. Therefore, H(x) can not be used as a measuring index. An example of a probability density function for response of nerve to electric current with different intensities is given to confirm above conclusions.
