All issues

This article reviews both historical achievements and modern results in the field of Markov Decision Process (MDP) and convex optimization. This review is the first attempt to cover the field of reinforcement learning in Russian in the context of convex optimization. The fundamental Bellman equation and the criteria of optimality of policy — strategies based on it, which make decisions based on the known state of the environment at the moment, are considered. The main iterative algorithms of policy optimization based on the solution of the Bellman equations are also considered. An important section of this article was the consideration of an alternative to the $Q$-learning approach — the method of direct maximization of the agent’s average reward for the chosen strategy from interaction with the environment. Thus, the solution of this convex optimization problem can be represented as a linear programming problem. The paper demonstrates how the convex optimization apparatus is used to solve the problem of Reinforcement Learning (RL). In particular, it is shown how the concept of strong duality allows us to naturally modify the formulation of the RL problem, showing the equivalence between maximizing the agent’s reward and finding his optimal strategy. The paper also discusses the complexity of MDP optimization with respect to the number of state–action–reward triples obtained as a result of interaction with the environment. The optimal limits of the MDP solution complexity are presented in the case of an ergodic process with an infinite horizon, as well as in the case of a non-stationary process with a finite horizon, which can be restarted several times in a row or immediately run in parallel in several threads. The review also reviews the latest results on reducing the gap between the lower and upper estimates of the complexity of MDP optimization with average remuneration (Averaged MDP, AMDP). In conclusion, the real-valued parametrization of agent policy and a class of gradient optimization methods through maximizing the $Q$-function of value are considered. In particular, a special class of MDPs with restrictions on the value of policy (Constrained Markov Decision Process, CMDP) is presented, for which a general direct-dual approach to optimization with strong duality is proposed.

This paper presents a methodology for modeling centrifugal pumps using the example of the NM 1250 260 main oil centrifugal pump. We use FlowVision CFD software as the numerical modeling instrument. Bench tests and numerical modeling use water as a working fluid. The geometrical model of the pump is fully three-dimensional and includes the pump housing to account for leakages. In order to reduce the required computational resources, the methodology specifies leakages using flow rate rather than directly modeling them. Surface roughness influences flow through the wall function model. The wall function model uses an equivalent sand roughness, and a formula for converting real roughness into equivalent sand roughness is applied in this work. FlowVision uses the sliding mesh method for simulation of the rotation of the impeller. This approach takes into account the nonstationary interaction between the rotor and diffuser of the pump, allowing for accurate resolution of recirculation vortices that occur at low flow rates.

The developed methodology has achieved high consistency between numerical simulations results and experiments at all pump operating conditions. The deviation in efficiency at nominal conditions is 0.42%, and in head is 1.9%. The deviation of calculated characteristics from experimental ones increases as the flow rate increases and reaches a maximum at the far-right point of the characteristic curve (up to 4.8% in head). This phenomenon occurs due to a slight mismatch between the geometric model of the impeller used in the calculation and the real pump model from the experiment. However, the average arithmetic relative deviation between numerical modeling and experiment for pump efficiency at 6 points is 0.39%, with an experimental efficiency measurement error of 0.72%. This meets the accuracy requirements for calculations. In the future, this methodology can be used for a series of optimization and strength calculations, as modeling does not require significant computational resources and takes into account the non-stationary nature of flow in the pump.

An approximate mathematical model of blood flow in an axisymmetric blood vessel is studied. Such a vessel is understood as an infinitely long circular cylinder, the walls of which consist of elastic rings. Blood is considered as an incompressible fluid flowing in this cylinder. Increased pressure causes radially symmetrical stretching of the elastic rings. Following J. Lamb, the rings are located close to each other so that liquid does not flow between them. To mentally realize this, it is enough to assume that the rings are covered with an impenetrable film that does not have elastic properties. Only rings have elasticity. The considered model of blood flow in a blood vessel consists of three equations: the continuity equation, the law of conservation of momentum and the equation of state. An approximate procedure for reducing the equations under consideration to the Korteweg – de Vries (KdV) equation is considered, which was not fully considered by J. Lamb, only to establish the dependence of the coefficients of the KdV equation on the physical parameters of the considered model of incompressible fluid flow in an axisymmetric vessel. From the KdV equation, by a standard transition to traveling waves, ODEs of the third, second and first orders are obtained, respectively. Depending on the different cases of arrangement of the three stationary solutions of the first-order ODE, a cnoidal wave and a soliton are standardly obtained. The main attention is paid to an unbounded periodic solution, which we call a degenerate cnoidal wave. Mathematically, cnoidal waves are described by elliptic integrals with parameters defining amplitudes and periods. Soliton and degenerate cnoidal wave are described by elementary functions. The hemodynamic meaning of these types of decisions is indicated. Due to the fact that the sets of solutions to first-, second- and third-order ODEs do not coincide, it has been established that the Cauchy problem for second- and third-order ODEs can be specified at all points, and for first-order ODEs only at points of growth or decrease. The Cauchy problem for a first-order ODE cannot be specified at extremum points due to the violation of the Lipschitz condition. The degeneration of the cnoidal wave into a degenerate cnoidal wave, which can lead to rupture of the vessel walls, is numerically illustrated. The table below describes two modes of approach of a cnoidal wave to a degenerate cnoidal wave.

Considered the application of physics-informed neural networks using multi layer perceptrons to solve Cauchy initial value problems in which the right-hand sides of the equation are continuous monotonically increasing, decreasing or oscillating functions. With the use of the computational experiments the influence of the construction of the approximate neural network solution, neural network structure, optimization algorithm and software implementation means on the learning process and the accuracy of the obtained solution is studied. The analysis of the efficiency of the most frequently used machine learning frameworks in software development with the programming languages Python and C# is carried out. It is shown that the use of C# language allows to reduce the time of neural networks training by 20–40%. The choice of different activation functions affects the learning process and the accuracy of the approximate solution. The most effective functions in the considered problems are sigmoid and hyperbolic tangent. The minimum of the loss function is achieved at the certain number of neurons of the hidden layer of a single-layer neural network for a fixed training time of the neural network model. It’s also mentioned that the complication of the network structure increasing the number of neurons does not improve the training results. At the same time, the size of the grid step between the points of the training sample, providing a minimum of the loss function, is almost the same for the considered Cauchy problems. Training single-layer neural networks, the Adam method and its modifications are the most effective to solve the optimization problems. Additionally, the application of twoand three-layer neural networks is considered. It is shown that in these cases it is reasonable to use the LBFGS algorithm, which, in comparison with the Adam method, in some cases requires much shorter training time achieving the same solution accuracy. The specificity of neural network training for Cauchy problems in which the solution is an oscillating function with monotonically decreasing amplitude is also investigated. For these problems, it is necessary to construct a neural network solution with variable weight coefficient rather than with constant one, which improves the solution in the grid cells located near by the end point of the solution interval.

The paper constructs and studies a generalized model describing two-component systems of reaction – diffusion type with power nonlinearity, considering the influence of external noise. A methodology has been developed for analyzing the generalized model, which includes linear stability analysis, nonlinear stability analysis, and numerical simulation of the system’s evolution. The linear analysis technique uses basic approaches, in which the characteristic equation is obtained using a linearization matrix. Nonlinear stability analysis realized up to third-order moments inclusively. For this, the functions describing the dynamics of the components are expanded in Taylor series up to third-order terms. Then, using the Novikov theorem, the averaging procedure is carried out. As a result, the obtained equations form an infinite hierarchically subordinate structure, which must be truncated at some point. To achieve this, contributions from terms higher than the third order are neglected in both the equations themselves and during the construction of the moment equations. The resulting equations form a set of linear equations, from which the stability matrix is constructed. This matrix has a rather complex structure, making it solvable only numerically. For the numerical study of the system’s evolution, the method of variable directions was chosen. Due to the presence of a stochastic component in the analyzed system, the method was modified such that random fields with a specified distribution and correlation function, responsible for the noise contribution to the overall nonlinearity, are generated across entire layers. The developed methodology was tested on the reaction – diffusion model proposed by Barrio et al., according to the results of the study, they showed the similarity of the obtained structures with the pigmentation of fish. This paper focuses on the system behavior analysis in the neighborhood of a non-zero stationary point. The dependence of the real part of the eigenvalues on the wavenumber has been examined. In the linear analysis, a range of wavenumber values is identified in which Turing instability occurs. Nonlinear analysis and numerical simulation of the system’s evolution are conducted for model parameters that, in contrast, lie outside the Turing instability region. Nonlinear analysis found noise intensities of additive noise for which, despite the absence of conditions for the emergence of diffusion instability, the system transitions to an unstable state. The results of the numerical simulation of the evolution of the tested model demonstrate the process of forming spatial structures of Turing type under the influence of additive noise.

Aortic stenosis and other forms of aortic valve dysfunction are associated with impaired intracardiac hemodynamics, left ventricular overload, and an increased risk of cardiovascular complications. Assessment of valve function requires not only integral clinical indicators but also local mechanical and hemodynamic characteristics, which, as a rule, cannot be measured directly in vivo. Therefore, mathematical modeling is regarded as one of the main tools for the quantitative analysis of the aortic valve. Despite the widespread use of various deformable-solid models and coupled fluid-structure interaction formulations, FSI, for describing leaflet dynamics, the limits of applicability of simplified formulations relative to the fully coupled problem remain insufficiently defined. In this study, an idealized model of the aortic root with the sinuses of Valsalva and a tricuspid valve was considered. The leaflets were described using an anisotropic hyperelastic material model. Five computational scenarios were compared, including a fully coupled FSI formulation that accounts for both solid and fluid dynamics, as well as a deformable-solid model with four loading variants replacing the effect of blood flow, differing in the way pressure was represented and in the direction of load application to the leaflets. The comparison criteria included deformation, displacement, von Mises stress, leaflet oscillatory dynamics, and the geometric opening area of the valve. It was shown that the FSI model provides the most consistent description of valve function, including asymmetric leaflet opening, smoother opening dynamics, and the absence of pronounced nonphysiological flutter. Structural formulations with loads applied along the local normal to the leaflet surface lead to overestimation of deformation and stress, as well as to more pronounced oscillatory regimes. Scenarios with restricted load direction produce a more moderate response, but they also fail to reproduce the spatial load structure and the temporal organization of leaflet opening. It was concluded that, in aortic valve modeling, not only the magnitude of the pressure difference but also the way it is applied to the leaflets in space and time is of decisive importance. Structural deformable-solid models may be used for the qualitative assessment of selected mechanical trends, but they cannot serve as a full substitute for the FSI formulation in the analysis of leaflet kinematics, oscillatory regimes, stress-strain state, and valve opening dynamics.

The paper presents an optimization approach to microstructure simulation. Porosity function was optimized by numerical method, grain-size model was optimized by complex method based on criteria of model quality. Methods have been validated on examples. Presented new regression model of model quality. Actual application of proposed method is 3D reconstruction of core sample microstructure. Presented results suggest to prolongation of investigations. 

This article presents an analysis of the impact of the transcatheter prosthesis frame design features on the results of its implantation in the aortic root model. In this paper we analyzed the various approaches to the design of such structures, as well as modifications in order to improve their functional characteristics during the implantation. As a general method for obtaining the results of interaction of the objects was used finite element method with nonlinear materials description and analysis of the main parameters: the stress-strain state, radial and friction forces.

Investigation of logical deterministic cellular automata models of population dynamics allows to reveal detailed individual-based mechanisms. The search for such mechanisms is important in connection with ecological problems caused by overexploitation of natural resources, environmental pollution and climate change. Classical models of population dynamics have the phenomenological nature, as they are “black boxes”. Phenomenological models fundamentally complicate research of detailed mechanisms of ecosystem functioning. We have investigated the role of fecundity and duration of resources regeneration in mechanisms of population growth using four models of ecosystem with one species. These models are logical deterministic cellular automata and are based on physical axiomatics of excitable medium with regeneration. We have modeled catastrophic death of population arising from increasing of resources regeneration duration. It has been shown that greater fecundity accelerates population extinction. The investigated mechanisms are important for understanding mechanisms of sustainability of ecosystems and biodiversity conservation. Prospects of the presented modeling approach as a method of transparent multilevel modeling of complex systems are discussed.

The short-term prediction of road traffic condition is one of the main tasks of transportation modelling. The main purpose of which are traffic control, reporting of accidents, avoiding traffic jams due to knowledge of traffic flow and subsequent transportation planning. A number of solutions exist — both model-driven and data driven had proven to be successful in capturing the dynamics of traffic flow. Nevertheless, most space-time models suffer from high mathematical complexity and low efficiency. Artificial Neural Networks, one of the prominent datadriven approaches, show promising performance in modelling the complexity of traffic flow. We present a neural network architecture for traffic flow prediction on a real-world road network graph. The model is based on the combination of a recurrent neural network and graph convolutional neural network. Where a recurrent neural network is used to model temporal dependencies, and a convolutional neural network is responsible for extracting spatial features from traffic. To make multiple few steps ahead predictions, the encoder-decoder architecture is used, which allows to reduce noise propagation due to inexact predictions. To model the complexity of traffic flow, we employ multilayered architecture. Deeper neural networks are more difficult to train. To speed up the training process, we use skip-connections between each layer, so that each layer teaches only the residual function with respect to the previous layer outputs. The resulting neural network was trained on raw data from traffic flow detectors from the US highway system with a resolution of 5 minutes. 3 metrics: mean absolute error, mean relative error, mean-square error were used to estimate the quality of the prediction. It was found that for all metrics the proposed model achieved lower prediction error than previously published models, such as Vector Auto Regression, LSTM and Graph Convolution GRU.