All issues

Gradient-free optimization methods or zeroth-order methods are widely used in training neural networks, reinforcement learning, as well as in industrial tasks where only the values of a function at a point are available (working with non-analytical functions). In particular, the method of error back propagation in PyTorch works exactly on this principle. There is a well-known fact that computer calculations use heuristics of floating-point numbers, and because of this, the problem of finiteness of the mantissa arises.

In this paper, firstly, we reviewed the most popular methods of gradient approximation: Finite forward/central difference (FFD/FCD), Forward/Central wise component (FWC/CWC), Forward/Central randomization on $l_2$ sphere (FSSG2/CFFG2); secondly, we described current theoretical representations of the noise introduced by the inaccuracy of calculating the function at a point: adversarial noise, random noise; thirdly, we conducted a series of experiments on frequently encountered classes of problems, such as quadratic problem, logistic regression, SVM, to try to determine whether the real nature of machine noise corresponds to the existing theory. It turned out that in reality (at least for those classes of problems that were considered in this paper), machine noise turned out to be something between adversarial noise and random, and therefore the current theory about the influence of the mantissa limb on the search for the optimum in gradient-free optimization problems requires some adjustment.

A nonlinear oscillatory system described by ordinary differential equations with variable coefficients is considered, in which terms that are linearly dependent on coordinates, velocities and accelerations are explicitly distinguished; nonlinear terms are written as implicit functions of these variables. For the numerical solution of the initial problem described by such a system of differential equations, the one-step Galerkin method is used. At the integration step, unknown functions are represented as a sum of linear functions satisfying the initial conditions and several given correction functions in the form of polynomials of the second and higher degrees with unknown coefficients. The differential equations at the step are satisfied approximately by the Galerkin method on a system of corrective functions. Algebraic equations with nonlinear terms are obtained, which are solved by iteration at each step. From the solution at the end of each step, the initial conditions for the next step are determined.

The corrective functions are taken the same for all steps. In general, 4 or 5 correction functions are used for calculations over long time intervals: in the first set — basic power functions from the 2nd to the 4th or 5th degrees; in the second set — orthogonal power polynomials formed from basic functions; in the third set — special linear-independent polynomials with finite conditions that simplify the “docking” of solutions in the following steps.

Using two examples of calculating nonlinear oscillations of systems with one and two degrees of freedom, numerical studies of the accuracy of the numerical solution of initial problems at various time intervals using the Galerkin method using the specified sets of power-law correction functions are performed. The results obtained by the Galerkin method and the Adams and Runge –Kutta methods of the fourth order are compared. It is shown that the Galerkin method can obtain reliable results at significantly longer time intervals than the Adams and Runge – Kutta methods.

The paper deals with a heat wave motion problem for a degenerate second-order nonlinear parabolic equation with power nonlinearity. The considered boundary condition specifies in a plane the motion equation of the circular zero front of the heat wave. A new numerical-analytical algorithm for solving the problem is proposed. A solution is constructed stepby- step in time using difference time discretization. At each time step, a boundary value problem for the Poisson equation corresponding to the original equation at a fixed time is considered. This problem is, in fact, an inverse Cauchy problem in the domain whose initial boundary is free of boundary conditions and two boundary conditions (Neumann and Dirichlet) are specified on a current boundary (heat wave). A solution of this problem is constructed as the sum of a particular solution to the nonhomogeneous Poisson equation and a solution to the corresponding Laplace equation satisfying the boundary conditions. Since the inhomogeneity depends on the desired function and its derivatives, an iterative solution procedure is used. The particular solution is sought by the collocation method using inhomogeneity expansion in radial basis functions. The inverse Cauchy problem for the Laplace equation is solved by the null field method as applied to a circular domain with a circular hole. This method is used for the first time to solve such problem. The calculation algorithm is optimized by parallelizing the computations. The parallelization of the computations allows us to realize effectively the algorithm on high performance computing servers. The algorithm is implemented as a program, which is parallelized by using the OpenMP standard for the C++ language, suitable for calculations with parallel cycles. The effectiveness of the algorithm and the robustness of the program are tested by the comparison of the calculation results with the known exact solution as well as with the numerical solution obtained earlier by the authors with the use of the boundary element method. The implemented computational experiment shows good convergence of the iteration processes and higher calculation accuracy of the proposed new algorithm than of the previously developed one. The solution analysis allows us to select the radial basis functions which are most suitable for the proposed algorithm.

This article is dedicated to use of S-spline theory for solving equations in partial derivatives. For example, we consider solution of the Poisson equation. S-spline — is a piecewise-polynomial function. Its coefficients are defined by two states. The first part of coefficients are defined by smoothness of the spline. The second coefficients are determined by least-squares method. According to order of considered polynomial and number of conditions of first and second type we get S-splines with different properties. At this moment we have investigated order 3 S-splines of class C¹ and order 5 S-splines of class C² (they meet conditions of smoothness of order 1 and 2 respectively). We will consider how the order 3 S-splines of class C¹ can be applied for solving equation of Poisson on circle and other areas.

Newton and quasi-Newton methods of absolute optimization based on Cholesky factorization with adaptive step and finite difference approximation of the first and the second derivatives. In order to raise effectiveness of the quasi-Newton methods a modified version of Cholesky decomposition of quasi-Newton matrix is suggested. It solves the problem of step scaling while descending, allows approximation by non-quadratic functions, and integration with confidential neighborhood method. An approach to raise Newton methods effectiveness with finite difference approximation of the first and second derivatives is offered. The results of numerical research of algorithm effectiveness are shown.

WENO schemes (weighted, essentially non oscillating) are currently having a wide range of applications as approximate high order schemes for discontinuous solutions of partial differential equations. These schemes are used for direct numerical simulation (DNS) and large eddy simmulation in the gas dynamic problems, problems for DNS in MHD and even neutron kinetics. This work is dedicated to clarify some characteristics of WENO schemes and numerical simulation of specific tasks. Results of the simulations can be used to clarify the field of application of these schemes. The first part of the work contained proofs of the approximation properties, stability and convergence of WENO5, WENO7, WENO9, WENO11 and WENO13 schemes. In the second part of the work the modified wave number analysis is conducted that allows to conclude the dispersion and dissipative properties of schemes. Further, a numerical simulation of a number of specific problems for hyperbolic equations is conducted, namely for advection equations (one-dimensional and two-dimensional), Hopf equation, Burgers equation (with low dissipation) and equations of non viscous gas dynamics (onedimensional and two-dimensional). For each problem that is implying a smooth solution, the practical calculation of the order of approximation via Runge method is performed. The influence of a time step on nonlinear properties of the schemes is analyzed experimentally in all problems and cross checked with the first part of the paper. In particular, the advection equations of a discontinuous function and Hopf equations show that the failure of the recommendations from the first part of the paper leads first to an increase in total variation of the solution and then the approximation is decreased by the non-linear dissipative mechanics of the schemes. Dissipation of randomly distributed initial conditions in a periodic domain for one-dimensional Burgers equation is conducted and a comparison with the spectral method is performed. It is concluded that the WENO7–WENO13 schemes are suitable for direct numerical simulation of turbulence. At the end we demonstrate the possibility of the schemes to be used in solution of initial-boundary value problems for equations of non viscous gas dynamics: Rayleigh–Taylor instability and the reflection of the shock wave from a wedge with the formation a complex configuration of shock waves and discontinuities.

Modern methods of complex planning the execution of task packages in multistage systems are characterized by the presence of restrictions on the dimension of the problem being solved, the impossibility of guaranteed obtaining effective solutions for various values of its input parameters, as well as the impossibility of registration the conditions for the formation of sets from the result and the restriction on the interval duration of time of the system operating. The decomposition of the generalized function of the system into a set of hierarchically interconnected subfunctions is implemented to solve the problem of scheduling the execution of task packages with generating sets of results and the restriction on the interval duration of time for the functioning of the system. The use of decomposition made it possible to employ the hierarchical approach for planning the execution of task packages in multistage systems, which provides the determination of decisions by the composition of task groups at the first level of the hierarchy decisions by the composition of task packages groups executed during time intervals of limited duration at the second level and schedules for executing packages at the third level the hierarchy. In order to evaluate decisions on the composition of packages, the results of their execution, obtained during the specified time intervals, are distributed among the packages. The apparatus of the theory of hierarchical games is used to determine complex solutions. A model of a hierarchical game for making decisions by the compositions of packages, groups of packages and schedules of executing packages is built, which is a system of hierarchically interconnected criteria for optimizing decisions. The model registers the condition for the formation of sets from the results of the execution of task packages and restriction on duration of time intervals of its operating. The problem of determining the compositions of task packages and groups of task packages is NP-hard; therefore, its solution requires the use of approximate optimization methods. In order to optimize groups of task packages, the construction of a method for formulating initial solutions by their compositions has been implemented, which are further optimized. Moreover, a algorithm for distributing the results of executing task packages obtained during time intervals of limited duration by sets is formulated. The method of local solutions optimization by composition of packages groups, in accordance with which packages are excluded from groups, the results of which are not included in sets, and packages, that aren’t included in any group, is proposed. The software implementation of the considered method of complex optimization of the compositions of task packages, groups of task packages, and schedules for executing task packages from groups (including the implementation of the method for optimizing the compositions of groups of task packages) has been performed. With its use, studies of the features of the considered planning task are carried out. Conclusion are formulated concerning the dependence of the efficiency of scheduling the execution of task packages in multistage system under the introduced conditions from the input parameters of the problem. The use of the method of local optimization of the compositions of groups of task packages allows to increase the number of formed sets from the results of task execution in packages from groups by 60% in comparison with fixed groups (which do not imply optimization).

The paper is devoted to convex-concave saddle point problems where the objective is a sum of a large number of functions. Such problems attract considerable attention of the mathematical community due to the variety of applications in machine learning, including adversarial learning, adversarial attacks and robust reinforcement learning, to name a few. The individual functions in the sum usually represent losses related to examples from a data set. Additionally, the formulation admits a possibly nonsmooth composite term. Such terms often reflect regularization in machine learning problems. We assume that the dimension of one of the variable groups is relatively small (about a hundred or less), and the other one is large. This case arises, for example, when one considers the dual formulation for a minimization problem with a moderate number of constraints. The proposed approach is based on using Vaidya’s cutting plane method to minimize with respect to the outer block of variables. This optimization algorithm is especially effective when the dimension of the problem is not very large. An inexact oracle for Vaidya’s method is calculated via an approximate solution of the inner maximization problem, which is solved by the accelerated variance reduced algorithm Katyusha. Thus, we leverage the structure of the problem to achieve fast convergence. Separate complexity bounds for gradients of different components with respect to different variables are obtained in the study. The proposed approach is imposing very mild assumptions about the objective. In particular, neither strong convexity nor smoothness is required with respect to the low-dimensional variable group. The number of steps of the proposed algorithm as well as the arithmetic complexity of each step explicitly depend on the dimensionality of the outer variable, hence the assumption that it is relatively small.

A model of combustion of premixed mixture of gases with one global chemical reaction is considered, the model includes equations of the second order for temperature of mixture and concentrations of fuel and oxidizer, and the right-hand sides of these equations contain the reaction rate function. This function depends on five unknown parameters of the global reaction and serves as approximation to multistep reaction mechanism. The model is reduced, after replacement of variables, to one equation of the second order for temperature of mixture that transforms to a first-order equation for temperature derivative depending on temperature that contains a parameter of flame propagation velocity. Thus, for computing the parameter of burning velocity, one has to solve Dirichlet problem for first-order equation, and after that a model dependence of burning velocity on mixture equivalence ratio at specified reaction rate parameters will be obtained. Given the experimental data of dependence of burning velocity on mixture equivalence ratio, the problem of optimal selection of reaction rate parameters is stated, based on minimization of the mean square deviation of model values of burning velocity on experimental ones. The aim of our study is analysis of uniqueness of this problem solution. To this end, we apply computational experiment during which the problem of global search of optima is solved using multistart of gradient descent. The computational experiment clarifies that the inverse problem in this statement is underdetermined, and every time, when running gradient descent from a selected starting point, it converges to a new limit point. The structure of the set of limit points in the five-dimensional space is analyzed, and it is shown that this set can be described with three linear equations. Therefore, it might be incorrect to tabulate all five parameters of reaction rate based on just one match criterion between model and experimental data of flame propagation velocity. The conclusion of our study is that in order to tabulate reaction rate parameters correctly, it is necessary to specify the values of two of them, based on additional optimality criteria.

We study the computational properties of a parametric class of finite-volume schemes with customizable dissipative properties with splitting by physical processes into Lagrangian, Eulerian, and the final stages (the hybrid large-particle method). The method has a second-order approximation in space and time on smooth solutions. The regularization of a numerical solution at the Lagrangian stage is performed by nonlinear correction of artificial viscosity. Regardless of the grid resolution, the artificial viscosity value tends to zero outside the zone of discontinuities and extremes in the solution. At Eulerian and final stages, primitive variables (density, velocity, and total energy) are first reconstructed by an additive combination of upwind and central approximations weighted by a flux limiter. Then numerical divergent fluxes are formed from them. In this case, discrete analogs of conservation laws are performed.

The analysis of dissipative properties of the method using known viscosity and flow limiters, as well as their linear combination, is performed. The resolution of the scheme and the quality of numerical solutions are demonstrated by examples of two-dimensional benchmarks: a gas flow around the step with Mach numbers 3, 10 and 20, the double Mach reflection of a strong shock wave, and the implosion problem. The influence of the scheme viscosity of the method on the numerical reproduction of a gases interface instability is studied. It is found that a decrease of the dissipation level in the implosion problem leads to the symmetric solution destruction and formation of a chaotic instability on the contact surface.

Numerical solutions are compared with the results of other authors obtained using higher-order approximation schemes: CABARET, HLLC (Harten Lax van Leer Contact), CFLFh (CFLF hybrid scheme), JT (centered scheme with limiter by Jiang and Tadmor), PPM (Piecewise Parabolic Method), WENO5 (weighted essentially non-oscillatory scheme), RKGD (Runge –Kutta Discontinuous Galerkin), hybrid weighted nonlinear schemes CCSSR-HW4 and CCSSR-HW6. The advantages of the hybrid large-particle method include extended possibilities for solving hyperbolic and mixed types of problems, a good ratio of dissipative and dispersive properties, a combination of algorithmic simplicity and high resolution in problems with complex shock-wave structure, both instability and vortex formation at interfaces.