Luzar, Vesna; Olkin, Ingram Comparison of simulation methods in the estimation of the ordered characteristic roots of a random covariance matrix. (English) Zbl 0733.65116 Statistical multiple integration, Proc. AMS-IMS-SIAM Jt. Summer Res. Conf., Arcata/CA (USA) 1989, Contemp. Math. 115, 189-202 (1991). Summary: [For the entire collection see Zbl 0729.00014.] Several simulation methods are used to evaluate the expected values and correlations of the ordered characteristic roots of a sample covariance matrix arising from a normal distribution with zero mean and identity covariance matrix. The methods used are importance sampling coupled with regression and ratio estimates. The comparisons of the efficiency of these methods is determined, where efficiency is defined in terms of variability measures. Cited in 1 Review MSC: 65C99 Probabilistic methods, stochastic differential equations 65C05 Monte Carlo methods 62H10 Multivariate distribution of statistics Keywords:Wishart distribution; Cholesky decomposition; regression simulation; ratio simulation; mixture sampling; correlations; sample covariance matrix; normal distribution; importance sampling Citations:Zbl 0729.00014 PDFBibTeX XML