A parallel approach to estimation of the approximate optimal control

In this paper the method for computing a priori estimates of the approximate optimal control based on the Krotov sufficient conditions for optimality is considered. These estimates provide us with information about the quality of the approximate optimal solution obtained by applying the improvement control procedure. The method is implemented in the form of a parallel algorithm and may be used at the stage of finding out initial control. This algorithm is an essential part of the developed software package intended for optimization of controllable dynamical systems with piecewise constant and piecewise linear control. We also consider the scalability of the parallel algorithm in the OpenTS parallel programming system for bifunctional catalyst blend optimization problem and production of secreted protein in a fed-batch reactor problem.