A new method for obtaining sharp compound Poisson approximation error estimates for sums of locally dependent random variables
Authors:
Michael V. Boutsikas,
Eutichia Vaggelatou
Abstract:
Let $X_1,X_2,...,X_n$ be a sequence of independent or locally dependent random variables taking values in $\mathbb{Z}_+$. In this paper, we derive sharp bounds, via a new probabilistic method, for the total variation distance between the distribution of the sum $\sum_{i=1}^nX_i$ and an appropriate Poisson or compound Poisson distribution. These bounds include a factor which depends on the smoothne…
▽ More
Let $X_1,X_2,...,X_n$ be a sequence of independent or locally dependent random variables taking values in $\mathbb{Z}_+$. In this paper, we derive sharp bounds, via a new probabilistic method, for the total variation distance between the distribution of the sum $\sum_{i=1}^nX_i$ and an appropriate Poisson or compound Poisson distribution. These bounds include a factor which depends on the smoothness of the approximating Poisson or compound Poisson distribution. This "smoothness factor" is of order $\mathrm{O}(σ^{-2})$, according to a heuristic argument, where $σ^2$ denotes the variance of the approximating distribution. In this way, we offer sharp error estimates for a large range of values of the parameters. Finally, specific examples concerning appearances of rare runs in sequences of Bernoulli trials are presented by way of illustration.
△ Less
Submitted 8 October, 2010;
originally announced October 2010.
On the relation between the distributions of stopping time and stopped sum with applications
Authors:
M. V. Boutsikas,
A. C. Rakitzis,
D. L. Antzoulakos
Abstract:
Let $T\$ be a stopping time associated with a sequence of independent random variables $Z_{1},Z_{2},...$ . By applying a suitable change in the probability measure we present relations between the moment or probability generating functions of the stopping time $T$ and the stopped sum $%S_{T}=Z_{1}+Z_{2}+...+Z_{T}$. These relations imply that, when the distribution of $S_{T}$\ is known, then the di…
▽ More
Let $T\$ be a stopping time associated with a sequence of independent random variables $Z_{1},Z_{2},...$ . By applying a suitable change in the probability measure we present relations between the moment or probability generating functions of the stopping time $T$ and the stopped sum $%S_{T}=Z_{1}+Z_{2}+...+Z_{T}$. These relations imply that, when the distribution of $S_{T}$\ is known, then the distribution of $T$\ is also known and vice versa. Applications are offered in order to illustrate the applicability of the main results, which also have independent interest. In the first one we consider a random walk with exponentially distributed up and down steps and derive the distribution of its first exit time from an interval $(-a,b).$ In the second application we consider a series of samples from a manufacturing process and we let $Z_{i},i\geq 1$, denoting the number of non-conforming products in the $i$-th sample. We derive the joint distribution of the random vector $(T,S_{T})$, where $T$ is the waiting time until the sampling level of the inspection changes based on a $k$-run switching rule. Finally, we demonstrate how the joint distribution of $%(T,S_{T})$ can be used for the estimation of the probability $p$ of an item being defective, by employing an EM algorithm.
△ Less
Submitted 27 June, 2011; v1 submitted 31 July, 2010;
originally announced August 2010.
