ABSTRACT
In this chapter we focus on iterative procedures for the construction of optimal continuous (normalized) designs. In Section 3.1 we introduce first-order iterative procedures for the D-criterion. Then in Section 3.2 we describe first-order procedures for the general case of convex criteria. The major developments of the numerical procedures were undertaken by Fedorov (1969, 1972) [125], [127]; Wynn (1970, 1972) [416], [417]; Atwood (1973) [25]; Tsay (1976) [385]; Wu and Wynn (1978) [409]; Wu (1978) [408]. The case of finite sample size is discussed in Section 3.3. A short survey of publications on other classes of numerical procedures for the construction of optimal designs is provided in Section 3.4.
