KL-optimum designs: theoretical properties and practical computation

In this paper some new properties and computational tools for finding KL-optimum designs are provided. KL-optimality is a general criterion useful to select the best experimental conditions to discriminate between statistical models. A KL-optimum design is obtained from a minimax optimization problem, which is defined on a infinite-dimensional space. In particular, continuity of the KL-optimality criterion is proved under mild conditions; as a consequence, the first-order algorithm converges to the set of KL-optimum designs for a large class of models. It is also shown that KL-optimum designs are invariant to any scale-position transformation. Some examples are given and discussed, together with some practical implications for numerical computation purposes.

[1]  R. Kirk.  Convergence of Baire measures , 1973 .

[2]  Steven E. Rigdon,et al.  Model-Oriented Design of Experiments , 1997, Technometrics.

[3]  Elijah Polak,et al.  Optimization: Algorithms and Consistent Approximations , 1997 .

[4]  William J. Hill,et al.  Discrimination Among Mechanistic Models , 1967 .

[5]  J. López–Fidalgo,et al.  An optimal experimental design criterion for discriminating between non‐normal models , 2007 .

[6]  D. Wiens.  Robust discrimination designs , 2009 .

[7]  A. Atkinson,et al.  Optimal design : Experiments for discriminating between several models , 1975 .

[8]  George E. P. Box,et al.  Follow-up designs to resolve confounding in multifactor experiments , 1996 .

[9]  Chiara Tommasi,et al.  Optimal Designs for Discriminating among Several Non-Normal Models , 2007 .

[10]  Caterina May,et al.  MODEL SELECTION AND PARAMETER ESTIMATION IN NON-LINEAR NESTED MODELS: A SEQUENTIAL GENERALIZED DKL-OPTIMUM DESIGN , 2013 .

[11]  D. Cox,et al.  Planning Experiments for Discriminating between Models , 1974 .

[12]  Chiara Tommasi,et al.  Bayesian optimum designs for discriminating between models with any distribution , 2010, Comput. Stat. Data Anal..

[13]  J. Knowles.  Measures on Topological Spaces , 1967 .

[14]  V. Melas,et al.  Robust T-optimal discriminating designs , 2013, 1309.4652.

[15]  W. J. Studden,et al.  Theory Of Optimal Experiments , 1972 .

[16]  V. Melas,et al.  T-optimal designs for discrimination between two polynomial models , 2012, 1205.6283.

[17]  Holger Dette,et al.  Optimal Designs for Identifying the Degree of a Polynomial Regression , 1995 .

[18]  M. C. Spruill.  GOOD DESIGNS FOR TESTING THE DEGREE OF A POLYNOMIAL MEAN , 2016 .

[19]  A. Atkinson,et al.  The design of experiments for discriminating between two rival models , 1975 .

[20]  Giacomo Aletti,et al.  A Convergent Algorithm for Finding KL-Optimum Designs and Related Properties , 2013 .

[21]  Drew Seils,et al.  Optimal design , 2007 .

[22]  R. W. Wedderburn,et al.  On the existence and uniqueness of the maximum likelihood estimates for certain generalized linear models , 1976 .

[23]  Holger Dette.  Discrimination designs for polynomial regression on compact intervals , 1994 .

[24]  Byron Jones,et al.  Kullback–Leibler divergence for evaluating bioequivalence , 2003, Statistics in medicine.

[25]  D. Ucinski.  T-optimum design for discrimination between two mul-tiresponse dynamic models , 2003 .

[26]  Chiara Tommasi,et al.  Optimal designs for both model discrimination and parameter estimation , 2008 .

[27]  Solomon Kullback,et al.  Information Theory and Statistics , 1960 .

[28]  Adrian S. Lewis,et al.  Convex Analysis And Nonlinear Optimization , 2000 .

[29]  D. Ucinski,et al.  T‐optimum designs for discrimination between two multiresponse dynamic models , 2005 .

[30]  Peter D. H. Hill,et al.  A Review of Experimental Design Procedures for Regression Model Discrimination , 1978 .

[31]  Alison L Gibbs,et al.  On Choosing and Bounding Probability Metrics , 2002, math/0209021.