All issues

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.

The gradual incorporation of automated vehicles into the global transport networks leads to the need to develop tools to assess the impact of this process on various aspects of traffic. This implies a more organized movement of automated vehicles which can form uniformly moving platoons. The influence of the formation and movement of these platoons on the dynamics of traffic flow is of great interest. The currently most developed traffic flow models are based on the cellular automaton approach. They are mainly developed in the direction of increasing accuracy. This inevitably leads to the complication of models, which in their modern form have significantly moved away from the original philosophy of cellular automata, which implies simplicity and schematicity of models at the level of evolution rules, leading, however, to a complex organized behavior of the system. In the present paper, a simulation model of connected automated vehicles platoon dynamics in a heterogeneous transport system is proposed, consisting of two types of agents (vehicles): human-driven and automated. The description of the temporal evolution of the system is based on modified rules 184 and 240 for elementary cellular automata. Human-driven vehicles move according to rule 184 with the addition of accidental braking, the probability of which depends on the distance to the vehicle in front. For automated vehicles, a combination of rules is used depending on the type of nearest neighbors, regardless of the distance to them, which brings non-local interaction to the model. At the same time, it is considered that a group of sequentially moving connected automated vehicles can form an organized platoon. The influence of the ratio of types of vehicles in the system on the characteristics of the traffic flow during free movement on a circular one-lane and two-lane roads, as well as in the presence of a traffic light, is studied. The simulation results show that the effect of platoon formation is significant for a freeway traffic flow; the presence of a traffic light reduces the positive effect by about half. The movement of platoons of connected automated vehicles on two-lane roads with the possibility of lane changing was also studied. It is shown that considering the types of neighboring vehicles (automated or human-driven) when changing lanes for automated vehicles has a positive effect on the characteristics of the traffic flow.

This paper presents the results of experimental validation of some structural issues concerning the practical use of methods to overcome catastrophic forgetting of neural networks. A comparison of current effective methods like EWC (Elastic Weight Consolidation) and WVA (Weight Velocity Attenuation) is made and their advantages and disadvantages are considered. It is shown that EWC is better for tasks where full retention of learned skills is required on all the tasks in the training queue, while WVA is more suitable for sequential tasks with very limited computational resources, or when reuse of representations and acceleration of learning from task to task is required rather than exact retention of the skills. The attenuation of the WVA method must be applied to the optimization step, i. e. to the increments of neural network weights, rather than to the loss function gradient itself, and this is true for any gradient optimization method except the simplest stochastic gradient descent (SGD). The choice of the optimal weights attenuation function between the hyperbolic function and the exponent is considered. It is shown that hyperbolic attenuation is preferable because, despite comparable quality at optimal values of the hyperparameter of the WVA method, it is more robust to hyperparameter deviations from the optimal value (this hyperparameter in the WVA method provides a balance between preservation of old skills and learning a new skill). Empirical observations are presented that support the hypothesis that the optimal value of this hyperparameter does not depend on the number of tasks in the sequential learning queue. And, consequently, this hyperparameter can be picked up on a small number of tasks and used on longer sequences.

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.

The presence of a contact boundary between water and transformer oil greatly reduces the electrical strength of the oil phase. The presence of an electric field leads to varying degrees of polarization at the interface and the appearance of a force acting on a liquid with a higher dielectric constant (water) in the direction of a liquid with a lower dielectric constant (oil). This leads to the contact surface instability development. Instability as a result of its development leads to a stream of water being drawn into oil volume and a violation of the insulating gap. In this work, we experimentally and numerically study electrohydrodynamic instability at the phase boundary between electrically weakly conductive water and transformer oil in a highly inhomogeneous electric field directed perpendicular to the contact boundary. The results of a full-scale and numerical experiment of studying of the electrohydrodynamic instability development in a strong electric field at the interface between water and transformer oil are presented. The system consists of a spherical electrode with a radius of 3.5 mm, placed in water with a conductivity of 5 $\mu S/cm$, and a thin blade electrode 0.1 mm thick, placed in transformer oil of the GK brand. The contact boundary passes at the same distance from the nearest points of the electrodes, equal to 3 mm. The work shows that at a certain electric field strength, the cone-shaped structure of water grows towards the electrode immersed in transformer oil. A numerical correspondence was obtained for both the shape of the resulting water structure (cone) during the entire growth time and the size measured from its top to the level of the initial contact boundary of phase separation. The dynamics of this structure growth has been studied. Both in numerical calculations and in experiment, it was found that the size of the resulting cone along the electrode connection line depends linearly on time.

The study of the development of π-periodic perturbations of a converging spherical shock wave leading to cumulation limitation is performed. The study is based on 3D hydrodynamic calculations with the Carnahan – Starling equation of state for hard sphere fluid. The method of solving the Euler equations on moving (compressing) grids allows one to trace the evolution of the converging shock wave front with high accuracy in a wide range of its radius. The compression rate of the computational grid is adapted to the motion of the shock wave front, while the motion of the boundaries of the computational domain satisfy the condition of its supersonic velocity relative to the medium. This leads to the fact that the solution is determined only by the initial data at the grid compression stage. The second order TVD scheme is used to reconstruct the vector of conservative variables at the boundaries of the computational cells in combination with the Rusanov scheme for calculating the numerical vector of flows. The choice is due to a strong tendency for the manifestation of carbuncle-type numerical instability in the calculations, which is known for other classes of flows. In the three-dimensional case of the observed force, the carbuncle effect was obtained for the first time, which is explained by the specific nature of the flow: the concavity of the shock wave front in the direction of motion, the unlimited (in the symmetric case) growth of the Mach number, and the stationarity of the front on the computational grid. The applied numerical method made it possible to study the detailed flow pattern on the scale of cumulation termination, which is impossible within the framework of the Whitham method of geometric shock wave dynamics, which was previously used to calculate converging shock waves. The study showed that the limitation of cumulation is associated with the transition from the Mach interaction of converging shock wave segments to a regular one due to the progressive increase in the ratio of the azimuthal velocity at the shock wave front to the radial velocity with a decrease in its radius. It was found that this ratio is represented as a product of a limited oscillating function of the radius and a power function of the radius with an exponent depending on the initial packing density in the hard sphere model. It is shown that increasing the packing density parameter in the hard sphere model leads to a significant increase in the pressures achieved in a shock wave with broken symmetry. For the first time in the calculation, it is shown that at the scale of cumulation termination, the flow is accompanied by the formation of high-energy vortices, which involve the substance that has undergone the greatest shock-wave compression. Influencing heat and mass transfer in the region of greatest compression, this circumstance is important for current practical applications of converging shock waves for the purpose of initiating reactions (detonation, phase transitions, controlled thermonuclear fusion).

In this paper, we consider the interaction of a kink of the $\varphi^4$ equation with two identical extended impurities. An extended impurity is described using a rectangular function. The case of an attractive impurity is analyzed. Using analytical methods, we consider the case of small amplitudes of localized waves, when it is possible to linearize the equations of motion. For the numerical solution, the method of lines for partial differential equations was used. To find the oscillation frequencies of waves localized on impurities, the discrete Fourier transform is used. The kink was launched in the direction of the impurities with different initial velocities. The distance between the two impurities was also varied. It is shown that when a kink interacts with impurities, long-lived localized breather-type waves are excited on them. Their structure and coupled dynamics are investigated. It is determined how, by changing the parameters of the impurities and the distance between them, it is possible to control the type and dynamic parameters of the coupled oscillations of the waves localized on the impurities. Possible solutions in the form of in-phase, antiphase oscillations, in the form of beats are found. The oscillations of localized waves occur with the emission of small-amplitude waves. The spectrum of these emissions consists of two frequencies. The first is approximately equal to $\sqrt{2}$, which corresponds to the frequency value for the wobbling breather tail of the $\varphi^4$ equation. The second is approximately equal to the doubled frequency of impurity mode oscillations. The presence of two possible frequencies for coupled localized oscillations is found both analytically and numerically. It is shown that the frequencies strongly depend on the distance between impurities. With increasing distance between impurities, the frequencies merge into one — frequency obtained for the case of a single impurity. The dependences of the frequencies on the distance between impurities found numerically and analytically coincide well for large distances, when the interaction between impurities is weak, and begin to differ noticeably at small distances, when the interaction between impurities is strong. The analytical value of the obtained frequencies is always greater than the numerical ones. It is shown that the dependence of the amplitude of localized waves on the initial kink velocity has several minima and maxima.

Modeling of gas-particle suspensions with large pressure and density gradients is of practical interest in the study of volcanic phenomena, explosions at different altitudes, as well as in technogenic problems related to the operation of space technology and the formation of space debris. This work presents numerical and analytical investigations of the expansion of gas suspensions with a high ratio (up to six orders of magnitude) of pressures and densities. For numerical modeling, a high-resolution hybrid large-particle method was employed. Under the conditions considered, the accuracy of the method was confirmed by comparison with asymptotically exact solutions. The study examined the wave and structural characteristics of concentrated gas suspension expansion depending on particle volume fraction, particle size, and initial pressure ratio. It was found that the polytropic index and sound speed in the gas suspension depend not only on temperature but also on pressure and particle concentration. With increasing pressure, both the polytropic index and sound speed rise, while with increasing particle volume fraction they decrease. In the case of an arbitrary discontinuity decay, an unusual effect is observed compared with “pure” gas dynamics: the relative velocity of the mixture in the uniform flow region decreases as the initial pressure increases. This is explained by the nonlinear dependence of the sound speed in a gas-dispersed mixture on pressure. With increasing particle size (Stokes number), the mixture flow splits into gaseous and dispersed components. At the initial moment, the contact discontinuity separating the mixture from the rarefied gas region splits into two contact boundaries: gaseous and dispersed. A practical conclusion is that when the particle size changes by two orders of magnitude, the gas-dynamic parameters of the mixture in the rarefaction wave region and up to the medium interface remain close to each other. During spatial expansion, the initial cylindrical shape of the dispersed medium successively transforms into a cross-section resembling a hexagon. At the next stage of expansion, the particles redistribute to form a bilateral conical structure. Eventually, a dispersed formation close to a spherical shape emerges.

Статья представляет собой пример компьютерного моделирования в области инженерной механики. Численным методом находятся поля скорости в гидроциклоне, которые недоступны прямому измерению. Рассматривается численное моделирование трехмерной гидродинамики на основе k-ε RNG модели турбулентности в гидроциклоне со встроенным инжектором, содержащим 5 тангенциально направленных сопла. Показано, что направление движения инжектируемой жидкости зависит от расхода жидкости через инжектор. Расчеты показывают в соответствии с экспериментами, что зависимость сплит-параметра от расхода инжектируемой жидкости имеет немонотонный характер, связанный с отношением мощности основного потока и инжектируемой жидкости.

The purpose of this paper is to generalize the macroscopic hydrodynamic vehicular traffic models by using the algorithm for constructing the adequate state equation — dependence the pressure from traffic density by taking into account the real experimental data (possibly using the parametric solutions for model equations). It is proved that this kind of state equation which closed model equations system and obtained from the experimentally observed form of the fundamental diagram — dependence the traffic intensity from its density, completely determines the all properties of the used phenomenological model.