The subject of the book is a functional theory of optimal designs elaborated by the author during the last two decades. This theory relates to points and weight of optimal designs considered as functions of some values. For linear models these values are metric characteristics of the set of admissible experimental conditions, for example, the bounds of a segment. For nonlinear models they are true values of the parameter to be estimated. Particularly locally D- optimal designs for exponential regression as an important example of nonlinear models and E-optimal designs for polynomial regression on arbitrary segments will be fully studied.