Bayesian Inference for a Change-Point in Nonlinear Modeling

Under the setting of a nonlinear model with normal additive errors, the nonlinear change-point problem is formulated and inferential procedures are developed adapting the Bayesian approach. Assuming an exponential type regres-sor, numerical methods are implemented for approximating the marginal posterior densities. Calculations are carried out for approximating these posterior densities by means of the Gibbs sampler. The outcome of this Bayesian analysis are the calculation and graphical display of posterior densities of the parameters of interest. Results of simulation studies are presented.