混合导数测度的性质及其应用

The Properties of the Mixed Derivative Measures of Interaction and Its Application

  • 摘要: 本文讨论了三维Bernoulli分布与正态分布在混合导数测度下的可压缩性的问题,导出了这两个分布在混合导数测度下可压缩的充分必要条件,同时利用混合导数测度,我们在适当条件下解决了1提到的无向图的Gibbs分解问题。

     

    Abstract: This paper concerns whether some random variable can be collapsed under the mixed derivative measures of interaction, we will give the necessary and sufficient conditions about both the three dimensional Bernoulli distribution and the Normal distribution. By the mixed derivative measures interaction, we then proved the guess of Gibbs factorization which was described by Frydenberg (1990) with a moderate condition.

     

/

返回文章
返回