The work considers the estimation of uncertainty of the regression model. The regression model is sought as a linear combination of basic functions. Coefficients in the linear combination are selected by minimizing the sum of the root-mean-square error of the approximation on the training set and the regularization term, which imposes restrictions on the set of solutions. Using a regularization term is one of the ways to combat the mathematical incorrectness of the problem. The regularization term consists of a penalty function and a regularization factor, which is an additional parameter of the regression model. The Bayesian approach allows to estimate the optimal value of the regularization factor directly from the data as being the most plausible. Constructing a regression model as a linear combination of basic functions (from a predefined set) allows us to reduce the high computational cost of the Bayesian approach by replacing the general iterative procedure by analytical expressions. A byproduct of estimating the regularization factor by Bayessian approach is the uncertainty of the regression model. The correctness of this assessment is the main subject of this research. The proposed approach of the uncertainty estimation is tested on the synthetic artificially noisy data. The proposed method estimated the noise magnitude close to the value used during data generation. The accuracy of the proposed method has outperformed the accuracy of the Gamma test, approach widely used to estimate uncertainty incorporated into the set of data.

Circuit simulation is an important part of the microelectronics design flow. This paper presents a design flow using Qucs-S circuit simulation tool with Ngspice or QucsatorRF backend. This solution may be targeted to both academic and industrial applications. Qucs-S combines a modern graphical user interface and a computation power of open-source circuit simulation kernels like Ngspice, QucsatorRF, and XYCE. The article provides an overview of Qucs-S software architecture and its application for integrated circuits design. The proposed workflow is illustrated by example of the semiconductor resistors parameter extraction and corner analysis using the mentioned open-source tools. 

The problem of schematic development (netlist generation), which arises in the development of analog integrated circuits, is formulated as an optimization problem for a differentiable smooth function using a combination of differentiable programming and machine learning methods. It is shown that this approach allows one to achieve the specification requirements and propose an optimal combination of circuit templates that make up an analog integrated circuit, without involving combinatorial optimization and reinforcement learning methods. It is shown that this approach provides significant speed advantages compared to traditional methods based on reinforcement learning. The possibility of fully automatic synthesis of an analog IC from specification to topology without expert participation using open-source software is investigated. The advantages and disadvantages of this approach are shown.

Hardware development is a time-consuming process that includes logic synthesis, placement and routing as its main steps. Despite that these steps are automated in modern CADs, their execution can take hours or even days. The application of machine learning methods can help predict synthesis results and thereby speed up the development process. This article describes the experience of creating and evaluating eight machine learning models for predicting area and delay of the synthesized ASIC using its netlist at the logic synthesis step. The results obtained show the benefits of this approach and indicate directions for further research.

Mandatory integrity control (MIC) is the security foundation of the Astra Linux operating system (OS) certified for the highest protection classes and trust levels, which, along with other mechanisms, including a closed software environment, ensures protection of privileged OS processes, integrity of executable and configuration system files and OS directories, as well as user data. The use of MIC is aimed for protecting against viruses (for example, ransomware), from the exploitation of many typical vulnerabilities in the software of the Linux family OS, including those leading to attacks by the adversaries with superuser’s root rights. The scientific basis for the implementation of MIC in the Astra Linux OS is the mandatory entity-role model of access and information flows security control in OS of Linux family (MROSL DP-model) that meets the criteria of GOST R 59453.1-2021. At the same time, the implementation of the MIC over the standard discretionary access control for the OS of Linux family presents significant difficulties and often requires the development of technologies and scenarios for the coordinated use of system and application software. In this regard, the authors conduct research on the design, development and effective use of MIC, a number of the results of which are devoted to this article. Firstly, there are modifications of the MROSL DP-model for the theoretical description of the MIC, including new features making for it. Secondly, adapting for MIC the container virtualization technology, when potentially "dangerous" software (for example, browsers) is launched at isolated intermediate integrity levels (in sessions of the system administrator with maximum integrity level) or negative integrity levels (in sessions of an unprivileged user with zero integrity level) in containers-sandboxes (for example, docker). Thirdly, technologies and scenarios for directly launching application software at intermediate or negative integrity levels with the configuration of the desktop menu of the system administrator or unprivileged user, respectively. Fourthly, the MIC configuration utility, which setting integrity levels or special flags for files and directories based on the rules of the AppArmor LSM module profiles.

The article presents a new tool, TSAR, designed for evaluating the effectiveness of static analyzers. TSAR includes three main components: a static analyzer assessment system, a test generator based on the Common Weakness Enumeration (CWE), and code transformation mechanisms (mutators) to challenge the analyzers. The assessment system identifies weaknesses in static analysis tools, while the test generator creates specific cases based on known vulnerabilities. Code transformations create complex structures that complicate analysis and intended to test the analyzers' ability in detecting real vulnerabilities. This tool provides researchers and developers with an opportunity for a deeper assessment of the quality of software static analyzers for their further improvement.

With the advancement of modern information technology, dynamic analysis is becoming an essential  part of software development. Fuzz testing is one of the most efficient and widely used techniques in this field. The core idea behind this approach is to input a large amount of random data into the program under the test. Mutation-based fuzzing tools generate test data by applying modifications (mutations) to successful variants that have already been identified, thus increasing the number of detected behaviors and code coverage. A common mutation strategy is to randomly select a mutation operator with a predefined probability.
This paper proposes a method to improve the effectiveness of mutation fuzzing through an adaptive mutation selection strategy. This approach was tested on commonly used Java packages and showed a statistically significant improvement in the number of errors detected and the diversity of program behaviors (execution traces).

This paper explores methods for enhancing the automated architecture search process for graph neural networks. We propose a novel approach that dynamically selects a priority direction within the search space, improving the efficiency and quality of the discovered architectures. Another proposed approach expands the search space by allowing combinations of different types of graph convolutional layers. The primary focus is on maximizing the quality of architectures within the expanded search space while maintaining a fixed search budget in terms of the number of models. Our experiments are conducted on datasets from citation networks, chemical molecules, and shopping graph domains. The experimental results show that the proposed approach enables the discovery of more effective and higher-quality models without increasing computational resources, demonstrating high potential for automating solutions to real-world graph data analysis tasks.

The work presents the construction of a hydraulic model of a heat exchanger (HE). The multi-section HE studied in this work includes a heat exchange matrix of 11 sections, each of which contains 6 small-diameter channels. The construction of a hydraulic mathematical model (HMM) is based on the results of experimental studies and the results of mathematical modeling. Numerical simulations were carried out in the OpenFOAM package using the simpleFoam solver. Based on mathematical modeling, full-size calculations of the TA were carried out in the operating range of Reynolds numbers, as well as additional calculations of flow in small-sized channels taking into account internal fins and the presence of heat transfer intensifiers. As a result of analyzing the distribution of hydraulic losses in a multi-section HE, a hydraulic mathematical model of the device’s operation was constructed. Based on a comparison of calculated data and the results of experimental studies, the parameters of the HMM heat exchanger were identified. The constructed model will be further generalized to a wide range of HE sizes of this type.

The paper presents the results of numerical investigation of the influence of the design parameter - the gap between the cooling plates of the convective heat exchanger for oil cooling on its efficiency. A single cooling section of an oil cooler consisting of cooling plates separated by a certain distance is considered. Each single plate of the oil cooler has 6 internal channels of complex geometry and row external fins, each row of which includes 11 elements of two standard sizes, and the number of rows is determined by the length of the plate. The efficiency of the device is evaluated on the basis of analyzing the change of heat transfer coefficients at the boundaries of working media from the increase of the distance between the plates. The problem of conjugate heat exchange between heated hydraulic oil, oil cooler plates and cold turbulent flow of air blown by a fan is considered. The mathematical model of fluid media motion is based on the Navier-Stokes equations. Modelling of heat transfer processes in the oil cooler plate is based on the heat conduction equation. To close the averaged system of conservation equations, the Menter SST turbulence model is applied. The numerical solution of the obtained system of equations is constructed by the method of control volumes using the chtMultiRegionFoam solver of the freely distributed OpenFOAM software. Numerical modelling of the working processes occurring in a single section of the oil cooler was performed using the establishment method. For discretization of non-viscous flows (in oil and air), a counter flow scheme of 2nd order of accuracy is applied, and for viscous flows, total variation minimization (TVD) schemes and the limitedLinear method were applied. The gradients were approximated based on the linear Gaussian method. The conjugate gradient method was used to accelerate convergence. As a result of numerical modelling, the fields of physical quantities, air and oil flow structure in the corresponding channels of the device were obtained. The influence of the plate gap size on the internal and external aerodynamics of the unit section of the oil cooler is identified and shown.  Non-uniform heating of the cooling section body with localization of the temperature maximum in the area of internal central channels has been revealed, described and substantiated. The analysis of the obtained thermophysical characteristics allowed to reveal the optimum distance between the plates of the oil cooler of 22 mm.

This study investigates the complex processes involved in the propagation of turbulent flames over solid combustible materials. Employing the Large Eddy Simulation (LES) technique within the Fire Dynamics Simulator (FDS), a detailed numerical investigation of flame structure and propagation were conducted. Obtained results reveal significant influences of turbulence on the flame's behavior, including fluctuations in temperature, velocity, and species concentrations. A notable finding of this study is the presence of a distinct laminar-like sublayer adjacent to the burning surface. This region exhibits significantly reduced turbulence intensity and is characterized by more stable temperature and species profiles compared to the fully turbulent regions of the flame. The coexistence of laminar and turbulent regimes within the flame has important implications for understanding flame spread rates and heat transfer mechanisms. Furthermore, our simulations highlight the role of buoyancy-driven flow in shaping the overall flame structure and propagation. The interaction between buoyancy forces and turbulent fluctuations leads to complex flow patterns and enhances mixing within the flame. Comparing our numerical results with experimental data, was demonstrated the ability of the LES model to accurately capture the essential features of turbulent flame spread. The findings of this study provide valuable insights into the underlying physics of turbulent flame spread. The detailed understanding of flame structure and propagation mechanisms gained from this work can be leveraged to develop more accurate and predictive models for fire safety engineering. Future research can focus on exploring the effects of different material properties, ambient conditions, and flame geometries on turbulent flame spread.

In this paper, a mathematical model for solving the problem of developed turbulent flow in a channel is proposed. The equations describing the fluid flow are the Reynolds equations and the equations of the k-omega turbulence model reduced to a quasi-hydrodynamic form. For the numerical solution of the equations of the mathematical statement, a combined approach of the control volume method and the finite element method on triangular adaptive grids was used. To verify the proposed mathematical model, the problem of turbulent flow in a rectangular channel was solved. The results obtained showed a good agreement between the results of the proposed model and the results of direct numerical simulation in the turbulent sub-layer region. For further verification of the model, a number of problems of the turbulent flow past fixed sand dunes with different lee-slope angles were calculated. A comparative analysis of the calculated flow characteristics with experimental data was performed, which showed their qualitative and quantitative agreement, with the exception of the values of the turbulent kinetic energy in the case of flowing past low-angle dunes. Good agreement of the values of the Reynolds shear stress averaged over one dune and the total shear stress obtained using the proposed model with the experimental data allows us to use the proposed model to calculate the characteristics of a hydrodynamic flow passing over time-varying bed forms.

The paper considers the problem of the impact of a turbulent jet on an eroding bottom. A mathematical model of the problem is proposed, including the Reynolds equations, kinetic energy transfer equations, turbulence dissipation, suspended particle concentrations and the equation of channel deformations. To describe changes in the bottom surface, an original equation of bottom deformations is used, constructed on the basis of an analytical model of the movement of traction sediments. An algorithm for solving the problem using the control volume method is proposed. Numerical modeling of the problem showed that when the bottom is eroded under the influence of a turbulent jet, a characteristic bottom wave arises, the parameters of which in the erosion region are consistent with known experimental data.

This paper presents the results of parametric studies of the interaction features of supersonic under-expanded argon jets, flowing from coaxial opposite micro-nozzles. A mathematical model of viscous compressible gas based on the classical Navier-Stokes equations supplemented by the equations of state of a perfect gas was used for numerical modelling of the gas dynamics of supersonic jets. The problem was solved in a two-dimensional axisymmetric formulation within the framework of the hypothesis of symmetric flow interaction. The problem was solved by the control volume method The Godunov method was used to discretize the convective terms, and linear interpolation of the values on the edges of the control cells was used to discretize the dissipative terms. Time integration was performed by the third-order Runge-Kutta method. Parametric analysis was aimed at evaluating the influence of the nozzle spacing on the size of the jet interaction region, as well as the density level in this zone. The obtained distributions of the main gasdynamic quantities allowed us to describe the flow structure and estimate the shape and size of local flow zones, as well as the density level in the contact region of two supersonic microjets. Analysis showed that a denser zone can be obtained by bringing the nozzles closer together, but in this case its longitudinal size decreases, which should be taken into account when planning experimental studies.

From the application point of view, the propagation of an admixture in a cylindrical volume filled with low-density air is of interest. It is associated with the evaporation of a substance from a small "glass" in which convective currents associated with the heating of its bottom take place. The propagation of an admixture is considered taking into account both diffusion and convective transfer due to thermal processes inside the "glass". The distribution of velocities in the main volume is sought by solving the Navier-Stokes equation, and the transfer equation with a diffusion term is solved for the admixture. A finite-difference numerical scheme implemented using our own program code is used. Solutions are obtained in cases corresponding to different heights of the "glass" walls, different ratios between the coefficients describing the processes of convective transfer and diffusion. It is shown that high walls significantly impede the process of admixture propagation into the main volume, and the substance is mainly concentrated inside the "glass" without moving beyond its limits. These results are similar to the data on the transfer of one of the components of the vector potential in the problem of amplification of the frozen magnetic field due to convection in the problem solved earlier. Just as there, structures are formed that repeat the features of the flow, and the maximum value is reached on the axis of symmetry. The issue of applying these results in practice and their experimental verification in laboratory conditions is discussed. It is noted that, in general, the propagation of the impurity corresponds to the data obtained in the course of experimental studies conducted earlier.

As part of the work, a specialized dictionary has been created to search for key terms in the texts of medical instructions, using data from VigiAccess, ICD-10 and rlsnet.ru. The text corpus was previously cleaned and brought to a single format to improve the quality of model training. In the future, it is planned to use the source grls.rosminzdrav.ru, as more authoritative and complete, for information about registered medicines. To automate data annotation, an algorithm has been developed that searches and marks terms from the dictionary in BIO (Begin, Inside, Outside) format, providing structured markup for model training. The model based on deep neural networks has demonstrated high efficiency in recognizing named entities by taking into account contextual dependencies. The semantic graph of medicines was constructed using algorithms for finding connections between named entities. However, automatic identification of deeper connections between graph nodes is difficult and requires additional data markup to account for complex grammatical structures, which will improve the analysis of interactions in the texts of medical instructions.

In this study the object of analysis is Vasyugan Khanty. Its status raises controversial opinions in Khanty studies. To clarify the status of the Vasyugan idiom as a separate dialect or as an accent of the Vakh-Vasyugan dialect, we employed modern methods of language data analysis. We used corpus data of the two varieties of the Khanty language, namely, Vakh Khanty and Vasyugan Khanty, available at the LingvoDoc platform to calculate their morphological proximity by means of the online virtual laboratory tool. The analysis results point to the fact that the morphological systems of the Vakh and Vasyugan Khanty vernaculars coincide by 98%, which confirms their morphological unity and affiliation with one and the same dialectal continuum. The machine analysis of the morphological dictionaries, cognate groups and transcriptions identified only three autonomous affixes in each idiom. Due to the fact that the volumes of the corpus data of the two varieties are unbalanced, the unique autonomous morphological affixes in each idiom can be regarded as a tentative argument subject to correction.

The article considers the names of jewelry in the Turkic and Finno-Ugric languages of the Ural-Volga region, reveals common and specific features in their nomination. The research was conducted taking into account the data of linguistic, ethnographic, archaeological works. The areas of distribution of names were constructed and an attempt was made to date the appearance of lexemes. The search for etymologies and mapping were carried out on the linguistic platform Lingvodok. The following features of the names of jewelry in the Turkic and Finno-Ugric languages of the Ural-Volga region were revealed: 1) lexemes denoting a ring, earrings and beads in the Turkic languages are more ancient and go back to the proto-forms; in the Finno-Ugric languages, they are either borrowings or derivatives of other words. This conclusion confirms the research of archaeologists who attribute the jewelry of the Turks to the Hunno-Sarmatian era; 2) the names confirm ethnographic data, for example, the Arabic loanword for coral beads is found in the languages in which this type of beads is found; 3) the names of the headbands correspond to the types and functions of the headbands. Thus, on the basis of linguistic data, the data of archaeological and ethnographic studies were confirmed, in a sense sometimes even dating the origin of the names.

The present article is devoted to determining the place of the Komi-Yaz'va idiom in the Komi dialect continuum. To this end, 8 dictionaries were analyzed using the programs of the linguistic platform LingvoDoc (search for cognates of languages / dialects, analysis of cognates of languages / dialects, glottochronology of languages / dialects, degree of morphological closeness between dialects / languages, and composite distance between languages). The dictionaries analyzed were Komi-Yaz'va, Upper Kama, Upper Sysola, Mysovsky and Kudymkarsky dialects. The comparative analysis of these dictionaries revealed that, in terms of phonetic, lexical, and morphological features, the Komi-Yaz'va idiom is significantly divergent from other Komi dialects. The degree of similarity with these dialects ranges from 86 to 88%, indicating that the Komi-Yaz'va is now considered a distinct language that has retained Proto-Komi language’s features.

Internet data serves as the foundation for a wide range of tasks, from information retrieval to analytical processing. With the rapid growth of data volumes, efficient metadata extraction from dynamic web resources has become critically important. Traditional information collection and extraction methods based on static templates are largely ineffective when processing interactive content. This paper presents the architecture of an adaptive information collection and extraction system that integrates standard data extraction techniques with machine learning technologies. The system has a modular structure comprising the following subsystems: task management, monitoring and logging, crawling, link management, and metadata extraction. The crawling subsystem processes both static and dynamic content through browser emulation. A hybrid approach combining structured rules and machine learning is used for metadata extraction. Experimental results demonstrated successful metadata extraction from various web resources, including pages with dynamic content and complex structures. The system exhibited high accuracy and resilience to changes in data formats while strictly adhering to ethical data collection standards, such as compliance with robots.txt directives and applying reasonable request intervals. Thus, the proposed solution represents a significant step toward the development of universal data collection and extraction systems for modern information environments. The developed software tools have been utilized in populating the index databases of the Neopoisk system.

Atrial fibrillation is the most common arrhythmia with a major impact on public health. This paper presents a model for automatic detection of atrial fibrillation episodes in ECG, using information compression and numerical differentiation for classification of beat-to-beat interval sequences. The core of the model is normalized compression distance based on the theory of universal similarity metrics. To enable class discrimination by compression we consider finite-difference representation of interval sequences with subsequent quantization procedure. In particular, we introduce a simple Δ5RR-interval representation which improves the sensitivity of the model to heart rhythm fluctuations. Our model achieves 96.37% sensitivity, 97.74% specificity and 0.935 MCC in 8x5-fold cross-validation on the MIT-BIH AFDB dataset using a segment window of 128 R-peaks. The particular advantage of the model is the classification quality in a few-shot learning setting, i.e., a training set with a small number of sequence observations can be used for classification of sufficiently large test sets.