科研成果详情

Parameter estimation in generalized linear mixed models (GLMM) is highly challenging due to the problem of analytically intractable integrated likelihood. In this paper we propose to use Quasi-Monte Carlo (QMC) approach to approximate the maximum likelihood estimates (MLE) in GLMM. Particulary, the sequential number-theoretic optimization (SNTO) approach is proposed to calculate the MLE. Numerical comparison to existing approaches is made through analyzing the infamous salamander mating data and conducting simulation studies.