Automatic control of the distribution of tasks under conditions of uncertainty with the use of adaptive approach

The subject of this article is the development of an adaptive approach for the optimal distribution of problems among solvers in conditions of uncertainty. Despite the large amount of research related to the construction of solutions for automatic control of task distribution, this issue remains relevant. As an alternative approach, a multi-level adaptive algorithm is proposed, which at each level filters incoming tasks according to solution methods, thereby significantly reducing the computational load. A distinctive feature of this algorithm is taking into account the time of task preprocessing, in particular, related to the current load of solvers and the distribution of tasks by solvers, in accordance with the maximum load.