EcoSta 2021: Start Registration
View Submission - EcoSta2021
A0707
Title: A general approach for constrained optimal design Authors:  Saumen Mandal - University of Manitoba (Canada) [presenting]
Abstract: A brief introduction to optimal design theory and an overview of constrained optimal design using a Lagrangian approach will be given. The approach is quite appealing and has some potential applications. One such application is the problem of determining maximum likelihood estimates of cell probabilities under a hypothesis of marginal homogeneity for data in a square contingency table. This is an optimization problem with respect to variables that are nonnegative and satisfy several linear constraints. A second application is the problem of constructing optimal designs subject to achieving zero correlation among certain parameter estimators. To estimate certain parameters independently of the others, we require that the correlations between relevant parameter estimates be zero. The methodology can be applied to a wide class of estimation problems where constraints are imposed on the parameters. We will conclude with another potential application of the approach in model discrimination in which we optimize one objective subject to achieving a given efficiency of a parameter.