Численная аппроксимация табличных функций на основе методов многомерной оптимизации при моделировании физико-технических задач

Abstract

The aim of the article is to present an approximation method based on multidimensional optimization methods which does not require preliminary information about the derivatives of the approximating function. It is proposed to use a combination of the grid method, coordinate descent method and the golden section method to solve the task of optimizing the objective function (standard deviation in the nodes of the tabular function). The considered example of the approximation of the flight trajectory of a body in the atmosphere illustrates the comparability of the results of the proposed method and the least squares method.