Analysis of solution searching algorithms on the model of functional relations

Here are the results of the study of the size of direct and inverted inference algorithm computations for solving the tasks of diagnostics, optimal designing, planning and complex systems control. The model of the subject area is set as a directed net without loops, where the attributes are the vertexes and functional dependencies between attributes are arches. The dependencies between attributes of the simulated subject area can be specified in different ways - as production rules, analytic dependences or function procedures. Formulae for defining the size of the subject area model calculus by direct and inverted inference algorithms are given. It is shown with the increasing number of sources (and the corresponding decreasing number of non-source attributes) that the size of computations of direct inference algorithm is increasing and the size of computations of inverted inference algorithm fails. At limited number of sources the task of optimization is rather solved by the method of direct inference, and the method of inverted inference is more efficient at large amount of sources.