Locally D-optimal design for weighted exponential model and its computation

Abstract

Abstract Weighted exponential model is used as a growth curve. We introduce locally D-optimal design for weighted exponential model with three parameters. The assumption of this model ishomoscedastic error. D-optimal criterion is a criterion which is the value of standardized variance less than or same the number of parameters. The design in this paper, number of supported designis three and have same weight 1 3 . Determination of the supported designs by maximizing the determinant of the information matrix. We present this formula, this is a nonlinear model, so to maximizing it we use numerically aproach.