随机游走轨迹的收缩原理与核加权和的 Cramer 定理
概率论
2021-04-05 v3 最优化与控制
摘要
2013 年 A.A. Borovkov 和 A.A. Mogulskii 证明了 中增量在零的邻域内拉普拉斯变换有限的随机游走轨迹的一个弱于标准的“度量”大偏差原理(LDP)。我们证明一般的度量 LDP 在一致连续映射下保持不变。这使我们能将 Borovkov 和 Mogulskii 的结果转化为标准 LDP。我们还给出了他们所发现速率函数的显式积分表示。作为一个应用,我们通过证明 中 i.i.d. 随机向量的核加权和的 LDP,推广了经典的 Cramér 定理。
引用
@article{arxiv.1909.00374,
title = {Contraction principle for trajectories of random walks and Cramer's theorem for kernel-weighted sums},
author = {Vladislav Vysotsky},
journal= {arXiv preprint arXiv:1909.00374},
year = {2021}
}
备注
This is a version to be published. Proposition 2.1 added. Section 3 had a number of new results added. An error was fixed -- the M_1 topology and the space BV_0[0,1] were replaced throughout by M_1' and BV[0,1], respectively. Minor changes were made throughout