Some iterative procedures for generating nonsingular optimal designs
- 1 January 1978
- journal article
- research article
- Published by Taylor & Francis in Communications in Statistics - Theory and Methods
- Vol. 7 (14) , 1399-1412
- https://doi.org/10.1080/03610927808827721
Abstract
The problem of optimal design algorithms is treated from the general viewpoint of optimization theory and algorithms. For designs of finite support, several iterative schemes are suggested, including vertex direction, gradient projection, normalized gradient projection and conjugate gradient projection. For an arbitrary design region, a general optimal design algorithm is proposed which incorporates the above algorithms for the finite support case ar.d methods of changing design support. The numerical behavior of these algorithms is investigated in detail on six examples and some experiences with the algorithms are summarized in the last section.Keywords
This publication has 9 references indexed in Scilit:
- Convergent Design Sequences, for Sufficiently Regular Optimality CriteriaThe Annals of Statistics, 1976
- Optimal design: Variation in structure and performance under change of criterionBiometrika, 1975
- D-Optimality for Regression Designs: A ReviewTechnometrics, 1975
- Optimal input signals for parameter estimation in dynamic systems--Survey and new resultsIEEE Transactions on Automatic Control, 1974
- General Equivalence Theory for Optimum Designs (Approximate Theory)The Annals of Statistics, 1974
- Sequences Converging to $D$-Optimal Designs of ExperimentsThe Annals of Statistics, 1973
- Results in the Theory and Construction of D-Optimum Experimental DesignsJournal of the Royal Statistical Society Series B: Statistical Methodology, 1972
- The Sequential Generation of $D$-Optimum Experimental DesignsThe Annals of Mathematical Statistics, 1970
- An algorithm for quadratic programmingNaval Research Logistics Quarterly, 1956