The method of choosing the technology for creating parallel special software with time parameterization

A new method of choosing the technology for creating a parallel program is proposed, which is based on the fact that a parallel program code written once can be executed by different parallel programming technologies, depending on the tasks, cycle parameters and the available time statistics of previous launches. At the same time, one of the parallelization methods can be used during the operation of a ready-made parallel program. Using Bayes' theorem, we move from a priori distributions of unknown magnitude to a posteriori distributions. We have a complete group of incompatible events, if we do not know their probabilities before the experiment, they are equally probable. As a result of the experiment, a certain technology number selection event appears and conditional probabilities of classification selection are known for this event. Then a statistical drawing is made to determine the number of the selected technology.