Vol 64, No 3 (2024)

Abstract. The paper devoted to the solution of nonlinear system of equations F(x) = 0, where the mapping F is quadratic, acting from ${\mathrm{ℝ}}^n\to {\mathrm{ℝ}}^n$. We consider the case, when the derivative F' is degenerate at the solution point. Based on the constructions of the p-regularity theory was proposed a 2-factor method for solving singular system of equations, which converges at a quadratic rate. Moreover, an exact formula is obtained for solving this quadratic system of equations in the 2-regular case of the mapping F(x).

Unique solvability of systems of linear algebraic equations is studied to which many in-verse problems of geophysics are reduced as a result of discretization after applying the method of integral equations or integral representations. Examples of singular and nonsingular systems of vari-ous dimensions that arise when processing magnetometric and gravimetric data from experimental observations are discussed. Conclusions are drawn about methods for constructing an optimal net-work of experimental observation points.

The problem of constructing eigenfunctions of a one-dimensional thermoelastic operator in Cartesian, cylindrical, and spherical coordinate systems is considered. The corresponding Sturm–Liouville problem is formulated using Fourier’s separation of variables applied to a coupled system of thermoelasticity equations, assuming that the heat transfer rate is finite. It is shown that the eigenfunctions of the one-dimensional thermoelastic operator are expressed in terms of well-known trigonometric, cylinder, and spherical functions. However, coupled thermoelasticity problems are solved analytically only under certain boundary conditions, whose form is determined by the properties of the eigenfunctions.

We study the multidimensional analogue of the Lavrent’ev integral equation to which an inverse problem of acoustic sounding is reduced. Conditions under which the studied equation has a unique solution are established. Results of numerical experiments concerning the solution of the inverse acoustic problem with variously located sets of sources and detectors are presented.

In this work, a system of kinetic equations is constructed to study the processes in a three-component plasma. It is an analogue of the Krook model, which is widely used in the dy-namics of rarefied gases. The model is supposed to be used to study the processes in the channels of rocket electric thrusters.

The paper considers algorithms for solving inverse scattering problems based on the discretization of the Gelfand–Levitan–Marchenko integral equations, associated with the system of nonlinear Schrödinger equations of the Manakov model. The numerical algorithm of the first order approximation for solving the scattering problem is reduced to the inversion of a series of nested block Toeplitz matrices using the Levinson-type bordering method. Increasing the approximation accuracy violates the Toeplitz structure of block matrices. Two algorithms are described that solve this problem for second order accuracy. One algorithm uses a block version of the Levinson bordering algorithm, which recovers the Toeplitz structure of the matrix by moving some terms of the systems of equations to the right-hand side. Another algorithm is based on the Toeplitz decomposition of an almost block-Toeplitz matrix and the Tyrtyshnikov bordering algorithm. The speed and accuracy of calculations using the presented algorithms are compared on an exact solution (the Manakov vector soliton).

The Cauchy problem for the heat equation with zero right-hand side is considered. The initial function is assumed to belong to the space of tempered distributions. The problem of determining the support of the initial function from solution values at some fixed time T > 0 is studied. Necessary and sufficient conditions for the support to lie in a given convex compact set are obtained. These conditions are formulated in terms of the solution’s decay rate at infinity. A sharp constant in the exponential for the Landis–Oleinik conjecture on the nonexistence of fast decaying solutions.

A direct variational method for solving the problem of reflection of ocean waves ray paths from the coastline with given positions of the source and the point of registration is considered. It is shown that the original boundary value problem can be reduced to the direct search of stationary points of the functional. Information about the objective function in the area of solutions of the ray problem allows us to construct a systematic procedure for finding minima and saddle points. A feature of the proposed approach is the optimization of the ray reflection point along a given coastline.

For a no Krasovskii Institute of Mathematics and Mechanics, Ural Branch, Russian Academy of Sciences, 620108, Yekaterinburg, Russianlinear controlled system, a fixed-time approach problem is considered in which the target point location becomes known only at the start of motion. According to the proposed solution method, node resolving program controls corresponding to a finite collection of target points from the set of their admissible locations are computed in advance and a refined control for the target point given at the start of motion is determined via linear interpolation of the node controls. The procedure for designing such a resolving control is formulated in the form of two algorithms, one of which is run before the start of the motion, and the other is executed in real time while the system is moving. The error in the transfer of the system’s state to the target point by applying these algorithms is estimated. As an example, we consider the approach problem for a modified Dubins car model and a target point about which only a compact set of its admissible locations is known before the start of motion.