×

Estimation and bimodality testing in the cusp model. (English) Zbl 1463.62039

The article deals with statistical inference methods under the (four-parametric) stochastic cusp model. First, a reduction technique is proposed to reduce the parameter space dimension in the context of maximum likelihood estimation of the cusp parameters. Second, several tests (beta-test, delta-test, dip test, likelihood ratio test) for bimodality (or unimodality) of the cusp density are proposed and compared in simulation studies. Except for the (generic) dip test, these tests rely on properties of Cardan’s discriminant.

MSC:

62F03 Parametric hypothesis testing
62F10 Point estimation

Software:

R; diptest
PDFBibTeX XMLCite
Full Text: DOI Link