A comparison of efficient approximations for a weighted sum of chi-squared random variables

Abstract

In many applications, the cumulative distribution function (cdf) F Q N of a positively weighted sum of N i.i.d. chi-squared random variables Q N is required. Although there is no known closed-form solution for F Q N , there are many good approximations. When computational efficiency is not an issue, Imhof's method provides a good solution. However, when both the accuracy of the approximation and the speed of its computation are a concern, there is no clear preferred choice. Previous comparisons between approximate methods could be considered insufficient. Furthermore, in streaming data applications where the computation needs to be both sequential and efficient, only a few of the available methods may be suitable. Streaming data problems are becoming ubiquitous and provide the motivation for this paper. We develop a framework to enable a much more extensive comparison between approximate methods for computing the cdf of weighted sums of an arbitrary random variable. Utilising this framework, a new and comprehensive analysis of four efficient approximate methods for computing F Q N is performed. This analysis procedure is much more thorough and statistically valid than previous approaches described in the literature. A surprising result of this analysis is that the accuracy of these approximate methods increases with N . Electronic supplementary materialThe online version of this article (

Remaining useful life prediction for multivariable stochastic degradation systems with non‐Markovian diffusion processes

“…Lemma creates a bond between the weighted sum of variables and the Gaussian variable , which has been established by Bodenham and Adams . In this way, the original diffusion part in HI can be converted to a single time‐varied Gaussian process.…”

Subset testing and analysis of multiple phenotypes

“…Remark For the linear summary statistic , and are approximated well by normal densities. When the number of SNPs is large, is also approximately normally distributed with mean and variance (Bodenham & Adams, ; Chuang & Shih, ). However, as the normal approximation may be less precise for when is small, or there is high LD, we compute p ‐values for the LRT without relying on that assumption, as further described in the next Section.…”

Reliable Multi-Point Transmissions Over Directional MISO TWDP Fading Channels

“…j∈Sg\{u} I u,j does generally not follow a weighted chi-square distribution, only in the special case that all variances σ 2 d,u,j are equal, which is mostly not satisfied due to pathloss differences. 1 Nevertheless, the distribution of the sum interference, as a positively weighted sum of independent and identically distributed (iid) chi-square random variables, has been thoroughly studied in the literature and many efficient and accurate approximations of its cdf are available [55]. Additionally, as we will see below, we only need to evaluate this cdf at few points to obtain virtually optimal performance.…”

“…Additionally, the values p i of Eq. (35) must be calculated, by evaluating the cdf expression of [55]; again, in practice this could be realized by a table look-up. Thus, within each loop iteration, at most work of complexity O (N T ) has to be performed, if the look-ups in the tables of size T are realized by (inefficient) brute force search.…”