arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.16430math.PRmath.FA
收藏

uniform distribution在立方体上的投影--大偏差透视

Projections of the uniform distribution on the cube -- a large deviation perspective

Samuel G. G. Johnston, Zakhar Kabluchko, Joscha Prochno

摘要:

设$Θ^{(n)}$是均匀分布在$\mathbb R^n$中单位球面$\mathbb S^{n-1}$上的随机向量。考虑立方体$[-1,1]^n$上的均匀分布到$Θ^{(n)}$所跨越的直线上的投影。投影分布是$\mathbb{R}$的Borel子集合$A$的\[ μ_{Θ^{(n)}}(A) := \frac 1 {2^n} \int_{[-1,1]^n} \mathbb 1\{\langle u, Θ^{(n)} \rangle \in A\} du, \]给出的$\mathbb R$上的随机概率测度$μ_{Θ^{(n)}}$。众所周知,对于概率$1$,随机概率测度序列$μ_{Θ^{(n)}}$弱收敛到具有方差$1/3$的中心高斯分布。利用$n$证明了序列$μ_{Θ^{(n)}}$在概率测度空间上的一个大偏差原理。当$ν(α)$是\begin{align*} \sqrt{1 - \|α\|_2^2 } \frac{Z}{\sqrt 3} + \sum_{ k = 1}^\infty α_k U_k, \end{align*}形式的随机变量的规律时,$I(ν(α)) := - \frac{1}{2} \log ( 1 - \|α\|_2^2)$明确给出(好的)速率函数,其中$Z$是独立于$U_1,U_2,\ldots$的标准高斯,$U_1,U_2,\ldots$是独立同分布的。$\text{Unif}[-1,1]$和$α_1 \geq α_2 \geq \ldots $是一个非递增序列的非负实数与$\|α\|_2<1$。我们得到一个类似的结果为离散立方体$\{-1,+1\}^n$上的均匀分布的随机投影。

英文摘要:

Let $Θ^{(n)}$ be a random vector uniformly distributed on the unit sphere $\mathbb S^{n-1}$ in $\mathbb R^n$. Consider the projection of the uniform distribution on the cube $[-1,1]^n$ to the line spanned by $Θ^{(n)}$. The projected distribution is the random probability measure $μ_{Θ^{(n)}}$ on $\mathbb R$ given by \[ μ_{Θ^{(n)}}(A) := \frac 1 {2^n} \int_{[-1,1]^n} \mathbb 1\{\langle u, Θ^{(n)} \rangle \in A\} du, \] for Borel subets $A$ of $\mathbb{R}$. It is well known that, with probability $1$, the sequence of random probability measures $μ_{Θ^{(n)}}$ converges weakly to the centered Gaussian distribution with variance $1/3$. We prove a large deviation principle for the sequence $μ_{Θ^{(n)}}$ on the space of probability measures on $\mathbb R$ with speed $n$. The (good) rate function is explicitly given by $I(ν(α)) := - \frac{1}{2} \log ( 1 - \|α\|_2^2)$ whenever $ν(α)$ is the law of a random variable of the form \begin{align*} \sqrt{1 - \|α\|_2^2 } \frac{Z}{\sqrt 3} + \sum_{ k = 1}^\infty α_k U_k, \end{align*} where $Z$ is standard Gaussian independent of $U_1,U_2,\ldots$ which are i.i.d. $\text{Unif}[-1,1]$, and $α_1 \geq α_2 \geq \ldots $ is a non-increasing sequence of non-negative reals with $\|α\|_2<1$. We obtain a similar result for random projections of the uniform distribution on the discrete cube $\{-1,+1\}^n$.

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。

↑