Testing Homogeneity in Gamma Mixture Models (Q4455948)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 2059383
Language Label Description Also known as
default for all languages
No label defined
    English
    Testing Homogeneity in Gamma Mixture Models
    scientific article; zbMATH DE number 2059383

      Statements

      Testing Homogeneity in Gamma Mixture Models (English)
      0 references
      0 references
      0 references
      0 references
      16 March 2004
      0 references
      empirical process
      0 references
      likelihood ratio test
      0 references
      identifiability
      0 references
      Let \(X=(X_1,\dots, X_n)\) be an i.i.d. sample from the PDF \(f\), and \(g_\lambda(x)= (\Gamma(\kappa))^{-1}\lambda^\kappa x^{\kappa-1} e^{-x\lambda}\), (\(\kappa\) is fixed and known). The hypothesis \(H_0: f=g_{\lambda_0}\) is tested against \(H_1: f=pg_\lambda+(1-p)g_{\lambda_0}\), where \(\lambda_0\) is known, and \(\lambda\) and \(p\) are unknown. The likelihood ratio test (LRT) with statistics \(\Lambda_n\) and the Gaussian process test with statistics \(M_n\) are considered. It is shown that under \(H_0\) \(2\Lambda_n=M_n^2+o_p(1)\) and \[ \lim_{n\to\infty}\Pr\{M_n^2-\log\log n+\log(16\pi^2/\kappa)\leq x\}=\exp(-e^{-x^2/2}). \] Simulation of the \(M_n\) distribution for small and moderate sample sizes is considered.
      0 references

      Identifiers