D-optimal designs for polynomial regression models through origin

Abstract

In this article we consider D-optimal designs for polynomial regression models with low-degree terms being missed, by applying the theory of canonical moments. It turns out that the optimal design places equal weight on each of the zeros of some Jacobi polynomial when the number of unknown parameters in the model is even. The procedure and examples of finding the optimal supports and weights are given when the number of unknown parameters in the model is odd.