Development of semi-parametric models of Raleigh–Taylor instability by means of neuronet processing of the results of numerical experiment

Investigation of turbulent flows, and, in particular, the flows arising due to Raleigh–Taylor instability, is a fundamental hydrodynamic problem. Due to complexity of direct numerical calculation and scantiness of a parametric approach one has to look for new methods for the description of turbulent flows. A principally new approach to studying of the problem is proposed in this work, i.e. the development of a semi-parametric model by means of a neuronet analysis of the experimental (calculation) data. Below we describe a technique for coding the initial physical fields aimed at the creation of neuronet input vectors, present a stable description of Raleigh–Taylor mixing processes, show a possibility in principle to predict the development of turbulent flows.