D-Optimum Designs for Optimum Mixture in a Quadratic Log Contrast Model

作者：Pal Manisha^*; Mandal Nripes Kumar

来源：Communications in Statistics - Simulation and Computation, 2012, 41(2): 253-264.

DOI：10.1080/03610918.2011.585003

• D-optimal designs
• Equivalence theorem
• Quadratic log contrast model
• Symmetric subspace

Optimal designs for estimating the optimum mixing proportions in a quadratic mixture model was first investigated by Pal and Mandal (2006). In this article, similar investigation is carried out when mean response in a mixture experiment is described by a quadratic log contrast model. It is found that in a symmetric subspace of the finite dimensional simplex, there exists a D-optimal design that puts weights at the centroid of the sub-space and the vertices of the experimental domain. The optimality is checked by numerical computation using Equivalence Theorem.

• 出版日期2012