


Vol 64, No 5 (2024)
General numerical methods
NUMERICAL-ANALYTICAL METHOD OF DECOMPOSITION-AUTOCOMPENSATION FOR SOLVING THE SIGNAL RECOGNITION PROBLEM BASED ON INACCURATE OBSERVATIONS
Abstract
A numerical-analytical method is developed to solve the problem of optimal recognition of a set of possible signals observed as an additive mixture containing not only a fluctuation observation error (with an unknown statistical distribution) but also a singular interference (with parametric uncertainty). The method allows for both the detection of signals present in the mixture and the estimation of their parameters within a specified quality criterion and associated constraints. The proposed method, based on the idea of generalized invariant-unbiased estimation of linear functional values, enables decomposition of the computational procedure and autocompensation of singular interference without resorting to the traditional expansion of the state space. Linear spectral decompositions in specified functional bases are used for the parametric finite-dimensional representation of signals and interference, while knowledge of the correlation matrix of the observation error is sufficient for error description. Random and systematic errors are analyzed, and an illustrative example is provided.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):699-712






SPECTRAL METHODS FOR SOLVING DIFFERENTIAL AND FUNCTIONAL EQUATIONS
Abstract
The operator approach previously developed for the spectral method using Legendre polynomials is generalized here to any systems of basis functions (not necessarily orthogonal) that satisfy two conditions: the result of the operation of multiplication by x or differentiation with respect to x is expressed in the same functions. All systems of classical orthogonal polynomials meet these conditions. In particular, a spectral method utilizing Chebyshev polynomials is constructed, which is most efficient for numerical calculations. This method is applied for the numerical solution of linear functional equations that arise in generalized series summation problems, aswell as in problems of analytic continuation of discrete mappings. It is also shown how these methods solve nonstandard and nonlinear boundary value problems for which conventional algorithms are not applicable.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):713-728



Optimal control
ON THE EXISTENCE OF OPTIMAL CONTROL FOR A SEMILINEAR EVOLUTION EQUATION WITH AN UNBOUNDED OPERATOR
Abstract
The paper studies the problem of optimal control for an abstract firstorder semilinear differential equation in a Hilbert space, with an unbounded operator and a control linearly entering the righthand side. The objective functional is assumed to be additively separable with respect to the state and control, with a fairly general dependence on the state. A theorem on the existence of an optimal control is proved for this problem, and properties of the set of optimal controls are established. Due to the nonlinearity of the equation under study, the author further develops previous results on total preservation of unique global solvability and solution estimates for similar equations. This estimate proves essential for the investigation. As examples, a nonlinear heat conduction equation and a nonlinear wave equation are considered.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):745-765



ASYMPTOTICS OF THE SOLUTION TO A BISINGULAR PROBLEM OF OPTIMAL DISTRIBUTED CONTROL IN A CONVEX DOMAIN WITH A SMALL PARAMETER IN ONE OF THE HIGHER DERIVATIVES
Abstract
The paper considers the problem of optimal distributed control in a strictly convex planar domain with a smooth boundary and a small parameter in one of the higher derivatives of the elliptic operator. In this problem, a zero Dirichlet boundary condition is imposed, and the control enters additively into the inhomogeneity. The set of admissible controls is the unit ball in the corresponding space of square-integrable functions. The solutions to the resulting boundary value problems are treated in the generalized sense as elements of a certain Hilbert space. The optimality criterion is the sum of the square of the norm of the state deviation from a given state and the square of the norm of the control, with a weighting coefficient. This structure of the optimality criterion allows either the first or the second term to be emphasized, depending on the need. In the first case, achieving the desired state is prioritized, while in the second case, minimizing resource costs becomes more important. The asymptotics of the problem are studied in detail, arising from a second-order differential operator with a small coefficient in one of the higher derivatives, to which a zero-order differential operator is added.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):732-744



Ordinary differential equations
CONVERGENCE RATE OF ALGORITHM FOR SOLVING LINEAR EQUATIONS BY QUANTUM ANNEALING
Abstract
Various iterative algorithms for solving the linear equation ax = b using a quantum computer operating on the principle of quantum annealing are studied. Assuming that the result produced by the computer is described by the Boltzmann distribution, conditions under which these algorithms converge are obtained and an estimate of their convergence rate is provided. Application of this approach for algorithms that use an infinite number of qubits and a small number of qubits is considered.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):766-779



Partial Differential Equations
IDENTITIES FOR MEASURES OF DEVIATIONS FROM SOLUTIONS OF PARABOLO-HYPERBOLIC EQUATIONS
Abstract
The article presents integral identities that hold for the difference between the exact solution of the initial-boundary value problem for a parabolo-hyperbolic equation and any function from the corresponding energy class. These identities allow for the derivation of two-sided a posteriori estimates for approximate solutions to the corresponding Cauchy problem. The left side of the estimate provides a natural measure of deviation from the solution, while the right side depends only on the problem data and the approximate solution itself, making it computable. The obtained estimates are utilized to compare solutions of Cauchy problems for both the parabolic equation and the parabolo-hyperbolic equation with a small parameter in the second time derivative. Additionally, the estimates enable a quantitative assessment of the effects arising from inaccuracies in initial data and coefficients of the equation.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):819-834



ON A NUMERICAL METHOD FOR SOLVING THE CAUCHY PROBLEM FOR SINGULARLY PERTURBED DIFFERENTIAL EQUATIONS
Abstract
The paper proposes a new method for numerically solving nonlinear stiff problems based on the numerical implementation of the holomorphic regularization method of the Cauchy problem for singularly perturbed nonlinear differential equations.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):804-818



GREEN’S FUNCTION FOR THE RIEMANN–NEUMANN PROBLEM FOR A POLYHARMONIC EQUATION IN THE UNIT SPHERE
Abstract
The Green’s function for the Riemann–Neumann problem for a polyharmonic equation in the unit sphere is constructed, and an integral representation of the solutions to the Riemann–Neumann problem is provided. Two examples are presented.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):791-803



ON THE STRUCTURE OF HELICAL AXISYMMETRIC SOLUTIONS OF THE NAVIER-STOKES SYSTEM FOR INCOMPRESSIBLE FLUIDS
Abstract
A class of exact solutions to the Navier-Stokes equations for axisymmetric vortex flows of incompressible fluids is obtained. Invariant manifolds of flows with rotational symmetry relative to a given axis in three-dimensional coordinate space are identified, and the structure of the solutions is described. It is established that typical invariant regions of such flows are rotational figures homeomorphic to a torus, forming a structure of topological fibration, such as in a sphere, cylinder, and more complex configurations composed of such figures. The results are extended to similar solutions of the magnetohydrodynamics (MHD) system and Maxwell’s electrodynamics equations, which possess ℝ3 analogous properties. Examples of axisymmetric vortex vector fields and the topological fibrations they generate on manifolds invariant ℝ3 under the dynamical systems defined by these fields are provided.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):780-790



Mathematical physics
ON THE SIMULTANEOUS DETERMINATION OF THE DENSITY DISTRIBUTION OF EQUIVALENT SOURCES UNDER AN EXTERNAL FIELD AND THE SPECTRUM OF USEFUL SIGNAL
Abstract
The article investigates the possibility of simultaneously recovering equivalent sources under an external field and the spectral characteristics of a useful signal. Examples of variational formulations for different versions of the method of linear integral representations are presented, and the problem of finding the density distribution of gravitating or magnetic masses on several horizontal planes is formulated. Additionally, the Fourier transform of the anomalous field element based on the known signal values at certain observation points, complicated by noise, is discussed.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):867-880



APPLICATION OF CABARET AND WENO SCHEMES FOR SOLVING THE NONLINEAR TRANSPORT EQUATION IN THE MODELING OF SOUND WAVE PROPAGATION IN THE ATMOSPHERE
Abstract
The most convenient model for describing the phenomenon of shock wave propagation in the atmosphere is the extended Burgers equation. This work investigates the influence of the numerical scheme on the results of solving the equation, which accounts for the nonlinear nature of shock wave propagation in the atmosphere. This equation is a key component of the extended Burgers equation and defines the transformation of the perturbed pressure profile during its propagation. Two numerical schemes were applied for the solution: CABARET and WENO, which are quasi-monotonic finite difference schemes that allow for solutions without significant numerical oscillations. An analysis of the applicability of these schemes for solving the considered problem was conducted.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):852-866



INVESTIGATION AND OPTIMIZATION OF THE N-PARTIAL NUMERICAL STATISTICAL ALGORITHM FOR SOLVING THE BOLTZMANN EQUATION
Abstract
The primary goal of the study is to test the hypothesis that the known N-partial statistical algorithm provides an estimate of the solution to the nonlinear Boltzmann equation with an error of order O(1/N). To achieve this, practically important optimal relationships between the value of N and the number n of sample estimates are determined. Numerical results for a problem with a known solution confirm the adequacy of the formulated estimates and conclusions.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):842-851



ON THE STABILITY OF A STABILIZING CORRECTION SCHEME WITH CENTRAL DIFFERENCES FOR SPATIAL VARIABLES IN THE 3D TRANSPORT EQUATION
Abstract
It is generally accepted that the stabilizing correction scheme with central differences for spatial variables in the 3D transport equation is conditionally stable. The work shows that, strictly speaking, this scheme is absolutely unstable. However, the region of unstable harmonics in the wave vector space and the magnitude of their increments rapidly approach zero as the Courant parameter tends to zero, allowing successful use of this scheme. Therefore, it is more accurate to refer to this scheme as practically conditionally stable.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):835-841



CALCULATION OF PLASMA HEATING BY CHARGED PRODUCTS OF THERMONUCLEAR REACTIONS BASED ON A SIMPLIFIED FOKKER–PLANCK EQUATION
Abstract
A two-time-layer scheme has been developed for solving the simplified kinetic Fokker–Planck equation related to the transport of charged products of thermonuclear reactions, which includes an interpolation procedure in four-dimensional grid space. Instabilities in the scheme were detected at low particle velocities and for a specific choice of particle deceleration in the ion field, which enters the kinetic equation as a parameter. It was shown that the thermalization condition, which prohibits solving the kinetic equation for particles with energy lower than the average ion energy, significantly limits the number of thermonuclear reactions where instability can manifest. The scheme was tested on the problem of relaxation to a stationary state and on a problem with a prescribed time-dependent thermonuclear reaction rate, for which an exact solution to the kinetic equation can be found.
Žurnal vyčislitelʹnoj matematiki i matematičeskoj fiziki. 2024;64(5):881-892


