TV距离不等式
TV homogenization inequalities
AI总结:
本文研究了在非均匀伯努利乘积测度上求和和均质化两种映射对总变差距离的影响,证明均质化映射在通用常数范围内减少总变差距离。
摘要:
我们研究了在非均匀伯努利乘积测度上两种信息消除映射下的总变差距离:求和和均质化。虽然求和是一个马尔可夫核,因此满足通常的数据处理不等式,但均质化——将每个伯努利参数映射到累积均值——并不满足。然而,我们证明均质化映射也减少总变差距离,最多一个通用常数。论证基于通过参数插值和二阶矩提取引理获得的总变差距离的双侧控制。
英文摘要:
We study the total variation distance under two information-erasing maps on inhomogeneous Bernoulli product measures: summation and homogenization. While summation is a Markov kernel and hence satisfies the usual data processing inequality, homogenization -- which maps each Bernoulli parameter to the cumulative mean -- is not. Nevertheless, we prove that the homogenization map also reduces the TV distance, up to a universal constant. The argument is based on an explicit two-sided control of the TV distance between Poisson binomials, obtained via a parameter interpolation and a second-moment extraction lemma.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。