Optimal design of sequential experiments for error-in-variables models

The optimal design of sequential experiments is well-established for explicit models, i.e. for models in which variables subject to error are expressed as functions of parameters and of independent variables measured accurately. Optimal design means therefore the determination of convenient values of the independent variables for the next experiment(s). In error-in-variables models there is no difference between independent and dependent variables and it is no longer possible to determine values of variables at which the experiment is to be carried out, but rather optimum nominal settings, from which the actual values of the experiment will differ due to the presence of experimental inaccuracy. In this paper optimum conditions for error-in-variables models are derived and an important special case is examined for comparison.