An algorithm for the computation of optimum designs under a given covariance structure (Q1966363): Difference between revisions

The authors present a first order iterative design optimization algorithm based on the notion of extended information matrices which is applicable when the observations are correlated, with a given covariance structure, and there are restrictions on the number of observations. To avoid nonoptimal solutions a certain heuristic is introduced; its usefulness is demonstrated by some typical examples from the literature.