数学
可分性规则
乘法函数
半群
无限可除性
双射
度量(数据仓库)
概率测度
卷积(计算机科学)
离散数学
猜想
组合数学
数学分析
数据库
机器学习
计算机科学
人工神经网络
作者
Octavio Arizmendi,Tadashi Hasebe
出处
期刊:Studia Mathematica
[Institute of Mathematics, Polish Academy of Sciences]
日期:2013-01-01
卷期号:215 (2): 157-185
被引量:15
摘要
Belinschi and Nica introduced a composition semigroup on the set of probability measures. Using this semigroup, they introduced a free divisibility indicator, from which one can know whether a probability measure is freely infinitely divisible or not. In this paper we further investigate this indicator, introduce a multiplicative version of it and are able to show many properties. Specifically, on the first half of the paper, we calculate how the indicator changes with respect to free and Boolean powers; we prove that free and Boolean 1/2-stable laws have free divisibility indicators equal to infinity; we derive an upper bound of the indicator in terms of Jacobi parameters. This upper bound is achieved only by free Meixner distributions. We also prove Bozejko's conjecture which says the Boolean power of a probability measure mu by 0 < t < 1 is freely infinitely divisible if mu is so. In the other half of this paper, we introduce an analogous composition semigroup for multiplicative convolutions and define free divisibility indicators for these convolutions. Moreover, we prove that a probability measure on the unit circle is freely infinitely divisible concerning the multiplicative free convolution if and only if the indicator is not less than one. We also prove how the multiplicative divisibility indicator changes under free and Boolean powers and then the multiplicative analogue of Bozejko's conjecture. We include an appendix, where the Cauchy distributions and point measures are shown to be the only fixed points of the Boolean-to-free Bercovici-Pata bijection.
科研通智能强力驱动
Strongly Powered by AbleSci AI