Logical-combinatorial algorithm for experimental data fragmentation of in "Cut-Glue" approximation problems
Abstract
Logical-combinatorial algorithm for experimental data fragmentation of in "Cut-Glue" approximation problems
Incoming article date: 19.12.2018The construction of mathematical models of technical objects is most often associated with the mathematical processing of experimental data. Due to significant non-linearities, approximation of such dependencies using known methods (polynomial expansions, spline approximation, etc.) is, firstly, difficult, and, secondly, is associated with large errors. Studies have been conducted to create and substantiate an alternative to the existing “Cut-Glue” method of approximation (CGA).In the CGA problem, there are three subtasks: fragmentation, approximation, and multiplicatively additive processing of the results of the first two subtasks. The initial subtask is data fragmentation. It is a preparatory stage for the subsequent stages of the implementation of the Cut-Glue approximation method.The method of its solution is based on logical-combinatorial features and conditions for partitioning multidimensional data. It allows you to consider the maximum possible number of various solutions when searching for the best.The CGA method implements a fundamentally new approach and is designed to solve the problems of experimental description of nonlinear dependencies. To implement the CGA method, the software package ""Algorithmic structure and functionality of the"" CutGlueApproximation ""software complex was developed. The article describes the nature and capabilities of the functional implemented in software tool ST. ST can be used as part of the entire software package (SP) or independently to perform intermediate tasks in isolation from the entire complex. The connection between the parts is realized with the help of auxiliary libraries, which allow both to use input data and to present the resulting data in various forms, which makes the SP more flexible. A convenient user interface contains many variable parameters for flexible configuration and convenient structured information output at all stages. With the help of the developed ST, many demonstration experiments have been carried out.
Keywords: Approximation, optimization, mathematical model, combinatorics, heuristic algorithms, modeling, software complex, fragmentation