Maximum entropy sampling and general. Equivalence theory

The relationship between maximum entropy sampling (MES) and the General Equivalence Theory of Kiefer and Wolfowitz is discussed. In MES a subsample with maximum entropy is selected from a population to minimise the expected posterior entropy for the unsampled units. Taking limits of MES as the prior variation on a regression parameter tends to infinity leads to constiained Doptimality. The general case benefits from a spectral resolution of the covariance function.