论文标题
用随机正交矩阵模拟靶向kollo偏度
Targetting Kollo Skewness with Random Orthogonal Matrix Simulation
论文作者
论文摘要
建模多元系统对于工程和运营研究中的许多应用都很重要。审查下的多元分布通常没有分析或封闭形式。因此，它们的建模采用数值技术，通常是多元模拟，可以具有很高的维度。随机正交矩阵（ROM）模拟是一种由于没有某些模拟误差而获得了一些普及的方法。具体而言，它与每个模拟都完全匹配目标均值，协方差矩阵和某些更高的矩。本文扩展了Hanke等人提出的ROM模拟算法。 （2017年），以下称为HPSW，它与目标平均值，协方差矩阵和Kollo Skewness向量完全匹配。我们的第一个贡献是为HPSW算法建立必要和充分的条件。我们的第二个贡献是开发一种在HPSW中构建可允许价值的一般方法。我们的第三个理论贡献是分析多元样本串联对目标kollo偏度的影响。最后，我们说明了使用仿真研究在这里开发的扩展。
Modelling multivariate systems is important for many applications in engineering and operational research. The multivariate distributions under scrutiny usually have no analytic or closed form. Therefore their modelling employs a numerical technique, typically multivariate simulations, which can have very high dimensions. Random Orthogonal Matrix (ROM) simulation is a method that has gained some popularity because of the absence of certain simulation errors. Specifically, it exactly matches a target mean, covariance matrix and certain higher moments with every simulation. This paper extends the ROM simulation algorithm presented by Hanke et al. (2017), hereafter referred to as HPSW, which matches the target mean, covariance matrix and Kollo skewness vector exactly. Our first contribution is to establish necessary and sufficient conditions for the HPSW algorithm to work. Our second contribution is to develop a general approach for constructing admissible values in the HPSW. Our third theoretical contribution is to analyse the effect of multivariate sample concatenation on the target Kollo skewness. Finally, we illustrate the extensions we develop here using a simulation study.
