Diseños experimentales para modelos no lineales de las ciencias experimentales y de la ingeniería
The main objective of this project is to provide referential designs in different fields of application, where there is some collaboration of the members of the three teams (http://areaestadistica.uclm.es/oed/). This will require the development of new theoretical tools in order to produce algorithms to compute designs for some families of nonlinear models, which are most frequent in real applications. In particular, -Exponential models and some extensions. -Characterizing and computating optimal designs for multidimentional and heteroscedastic models. -Standardized A-optimal and maximin desgins for nonlinear multifactorial models. -Theoretical approach for t regression models. - Measurments censored by a threshold in multi exponential models. -MV -optimal designs for binary response models in the whole design space. -Characterizating the information matrices for multiresponse models. -Optimal design theory for non-standard information matrices. -Semi-Bayesian approach in the computation of designs for discriminating between models. -Models with correlated observations in pharmacokinetics, chemical-physics, radiology and astronomy. -Designs with support points following a specified pattern within in a interval (Filling Designs). -Optimal designs for MTE models in Bayesian networks for biometric indices in spatial data. -Rational models with industrial interest for different optimality criteria. There is special interest in obtaining efficient algorithms for obtaining optimal designs in the situations mentioned above. Moreover, there is much interest in implementating these algorithms in a code easy to use by the practitioners. KEYWORDS: Algorithms, optimal experimental design, random effects, information matrix, non-linear models, survival models, correlation structures
Ministerio de Economía y competitividad (2011-2013)
Principal investigator
01/01/2011 hasta: 31/12/2013