All issues

Authors describe a two-stage traffic assignment model. It contains of two blocks. The first block consists of a model for calculating a correspondence (demand) matrix, whereas the second block is a traffic assignment model. The first model calculates a matrix of correspondences using a matrix of transport costs (it characterizes the required volumes of movement from one area to another, it is time in this case). To solve this problem, authors propose to use one of the most popular methods of calculating the correspondence matrix in urban studies — the entropy model. The second model describes exactly how the needs for displacement specified by the correspondence matrix are distributed along the possible paths. Knowing the ways of the flows distribution along the paths, it is possible to calculate the cost matrix. Equilibrium in a two-stage model is a fixed point in the sequence of these two models. In practice the problem of finding a fixed point can be solved by the fixed-point iteration method. Unfortunately, at the moment the issue of convergence and estimations of the convergence rate for this method has not been studied quite thoroughly. In addition, the numerical implementation of the algorithm results in many problems. In particular, if the starting point is incorrect, situations may arise where the algorithm requires extremely large numbers to be computed and exceeds the available memory even on the most modern computers. Therefore the article proposes a method for reducing the problem of finding the equilibrium to the problem of the convex non-smooth optimization. Also a numerical method for solving the obtained optimization problem is proposed. Numerical experiments were carried out for both methods of solving the problem. The authors used data for Vladivostok (for this city information from various sources was processed and collected in a new dataset) and two smaller cities in the USA. It was not possible to achieve convergence by the method of fixed-point iteration, whereas the second model for the same dataset demonstrated convergence rate $k^{-1.67}$.

The quasistatic deformation problem for shape memory alloys is reviewed within the phenomenological mechanics of solids without microphysics analysis. The phenomenological approach is based on comparison of two material deformation diagrams. The first diagram corresponds to the active proportional loading when the alloy behaves as an ideal elastoplastic material; the residual strain is observed after unloading. The second diagram is relevant to the case when the deformed sample is heated to a certain temperature for each alloy. The initial shape is restored: the reverse distortion matches deformations on the first diagram, except for the sign. Because the first step of distortion can be described with the variational principle, for which the existence of the generalized solutions is proved under arbitrary loading, it becomes clear how to explain the reverse distortion within the slightly modified theory of plasticity. The simply connected surface of loading needs to be replaced with the doubly connected one, and the variational principle needs to be updated with two laws of thermodynamics and the principle of orthogonality for thermodynamic forces and streams. In this case it is not difficult to prove the existence of solutions either. The successful application of the theory of plasticity under the constant temperature causes the need to obtain a similar result for a more general case of variable external forces and temperatures. The paper studies the ideal elastoplastic von Mises model at linear strain rates. Taking into account hardening and arbitrary loading surface does not cause any additional difficulties.

The extended variational principle of the Reissner type is defined. Together with the laws of thermal plasticity it enables to prove the existence of the generalized solutions for three-dimensional bodies made of shape memory materials. The main issue to resolve is a challenge to choose a functional space for the rates and deformations of the continuum points. The space of bounded deformation, which is the main instrument of the mathematical theory of plasticity, serves this purpose in the paper. The proving process shows that the choice of the functional spaces used in the paper is not the only one. The study of other possible problem settings for the extended variational principle and search for regularity of generalized solutions seem an interesting challenge for future research.

This paper studies solids with internal degrees of freedom using the method of Cartan moving hedron. Strain compatibility conditions are derived in the form of structure equations for manifolds. Constitutive relations are reviewed and ultimate load theorems are proved for rigid plastic solids with internal degrees of freedom. It is demonstrated how the above theorems can be applied in behavior analysis of rigid plastic continual shells of shape memory materials. The ultimate loads are estimated for rotating shells under external forces and in case of shape recovery from heating.

The main statements and inferences of the Dynamic Theory Information (DTI) are considered. It is shown that DTI provides the possibility two reveal two essentially important types of information: objective (unconventional) and subjective (conventional) informtion. There are two ways of obtaining information: reception (perception of an already existing one) and generation (production of new) information. It is shown that the processes of generation and perception of information should proceed in two different subsystems of the same cognitive system. The main points of the Natural-Constructivist Approach to modeling the cognitive process are discussed. It is shown that any neuromorphic approach faces the problem of Explanatory Gap between the “Brain” and the “Mind”, i. e. the gap between objectively measurable information about the ensemble of neurons (“Brain”) and subjective information about the human consciousness (“Mind”). The Natural-Constructive Cognitive Architecture developed within the framework of this approach is discussed. It is a complex block-hierarchical combination of several neuroprocessors. The main constructive feature of this architecture is splitting the whole system into two linked subsystems, by analogy with the hemispheres of the human brain. One of the subsystems is processing the new information, learning, and creativity, i.e. for the generation of information. Another subsystem is responsible for processing already existing information, i.e. reception of information. It is shown that the lowest (zero) level of the hierarchy is represented by processors that should record images of real objects (distributed memory) as a response to sensory signals, which is objective information (and refers to the “Brain”). The next hierarchy levels are represented by processors containing symbols of the recorded images. It is shown that symbols represent subjective (conventional) information created by the system itself and providing its individuality. The highest hierarchy levels containing the symbols of abstract concepts provide the possibility to interpret the concepts of “consciousness”, “sub-consciousness”, “intuition”, referring to the field of “Mind”, in terms of the ensemble of neurons. Thus, DTI provides an opportunity to build a model that allows us to trace how the “Mind” could emerge basing on the “Brain”.

In this work we have developed a new efficient program for the numerical simulation of 3D global chemical transport on an adaptive finite-difference grid which allows us to concentrate grid points in the regions where flow variables sharply change and coarsen the grid in the regions of their smooth behavior, which significantly minimizes the grid size. We represent the adaptive grid with a combination of several dynamic (tree, linked list) and static (array) data structures. The dynamic data structures are used for a grid reconstruction, and the calculations of the flow variables are based on the static data structures. The introduction of the static data structures allows us to speed up the program by a factor of 2 in comparison with the conventional approach to the grid representation with only dynamic data structures.

We wrote and tested our program on a computer with 6 CPU cores. Using the computer microarchitecture simulator gem5, we estimated the scalability property of the program on a significantly greater number of cores (up to 32), using several models of a computer system with the design “computational cores – cache – main memory”. It has been shown that the microarchitecture of a computer system has a significant impact on the scalability property, i.e. the same program demonstrates different efficiency on different computer microarchitectures. For example, we have a speedup of 14.2 on a processor with 32 cores and 2 cache levels, but we have a speedup of 22.2 on a processor with 32 cores and 3 cache levels. The execution time of a program on a computer model in gem5 is 104–105 times greater than the execution time of the same program on a real computer and equals 1.5 hours for the most complex model.

Also in this work we describe how to configure gem5 and how to perform simulations with gem5 in the most optimal way.

The paper considers integro-differential equations of fractional order moisture transfer with the Bessel operator. The studied equations contain the Bessel operator, two Gerasimov – Caputo fractional differentiation operators with different orders $\alpha$ and $\beta$. Two types of integro-differential equations are considered: in the first case, the equation contains a non-local source, i.e. the integral of the unknown function over the integration variable $x$, and in the second case, the integral over the time variable τ, denoting the memory effect. Similar problems arise in the study of processes with prehistory. To solve differential problems for different ratios of $\alpha$ and $\beta$, a priori estimates in differential form are obtained, from which the uniqueness and stability of the solution with respect to the right-hand side and initial data follow. For the approximate solution of the problems posed, difference schemes are constructed with the order of approximation $O(h^2+\tau^2)$ for $\alpha=\beta$ and $O(h^2+\tau^{2-\max\{\alpha,\beta\}})$ for $\alpha\neq\beta$. The study of the uniqueness, stability and convergence of the solution is carried out using the method of energy inequalities. A priori estimates for solutions of difference problems are obtained for different ratios of $\alpha$ and $\beta$, from which the uniqueness and stability follow, as well as the convergence of the solution of the difference scheme to the solution of the original differential problem at a rate equal to the order of approximation of the difference scheme.

The work is devoted to the description of the author’s formal statements of the clustering problem for a given number of clusters, algorithms for their solution, as well as the results of using this toolkit in medicine.

The solution of the formulated problems by exact algorithms of implementations of even relatively low dimensions before proving optimality is impossible in a finite time due to their belonging to the NP class.

In this regard, we have proposed a hybrid algorithm that combines the advantages of precise methods based on clustering in paired distances at the initial stage with the speed of methods for solving simplified problems of splitting by cluster centers at the final stage. In the development of this direction, a sequential hybrid clustering algorithm using random search in the paradigm of swarm intelligence has been developed. The article describes it and presents the results of calculations of applied clustering problems.

To determine the effectiveness of the developed tools for optimal clustering of multidimensional objects according to a variety of heterogeneous indicators, a number of computational experiments were performed using data sets including socio-demographic, clinical anamnestic, electroencephalographic and psychometric data on the cognitive status of patients of the cardiology clinic. An experimental proof of the effectiveness of using local search algorithms in the paradigm of swarm intelligence within the framework of a hybrid algorithm for solving optimal clustering problems has been obtained.

The results of the calculations indicate the actual resolution of the main problem of using the discrete optimization apparatus — limiting the available dimensions of task implementations. We have shown that this problem is eliminated while maintaining an acceptable proximity of the clustering results to the optimal ones. The applied significance of the obtained clustering results is also due to the fact that the developed optimal clustering toolkit is supplemented by an assessment of the stability of the formed clusters, which allows for known factors (the presence of stenosis or older age) to additionally identify those patients whose cognitive resources are insufficient to overcome the influence of surgical anesthesia, as a result of which there is a unidirectional effect of postoperative deterioration of complex visual-motor reaction, attention and memory. This effect indicates the possibility of differentiating the classification of patients using the proposed tools.

In today’s technology-driven world, lite devices like Internet of Things (IoT) devices and microcontrollers (MCUs) are becoming increasingly common. These devices are more energyefficient and affordable, often with reduced features compared to the standard versions such as very limited memory and processing power for typical machine learning models. However, modern machine learning models can have millions of parameters, resulting in a large memory footprint. This complexity not only makes it difficult to deploy these large models on resource constrained devices but also increases the risk of latency and inefficiency in processing, which is crucial in some cases where real-time responses are required such as autonomous driving and medical diagnostics. In recent years, neural networks have seen significant advancements in model optimization techniques that help deployment and inference on these small devices. This narrative review offers a thorough examination of the progression and latest developments in neural network optimization, focusing on key areas such as quantization, pruning, knowledge distillation, and neural architecture search. It examines how these algorithmic solutions have progressed and how new approaches have improved upon the existing techniques making neural networks more efficient. This review is designed for machine learning researchers, practitioners, and engineers who may be unfamiliar with these methods but wish to explore the available techniques. It highlights ongoing research in optimizing networks for achieving better performance, lowering energy consumption, and enabling faster training times, all of which play an important role in the continued scalability of neural networks. Additionally, it identifies gaps in current research and provides a foundation for future studies, aiming to enhance the applicability and effectiveness of existing optimization strategies.

This paper investigates the effectiveness of Retrieval-Augmented Generation (RAG) combined with various Large Language Models (LLMs) for document retrieval and information access in corporate information systems. We survey typical use-cases of LLMs in enterprise environments, outline the RAG architecture, and discuss the major challenges that arise when integrating LLMs into a RAG pipeline. A system architecture is proposed that couples a text-vector encoder with an LLM. The encoder builds a vector database that indexes a library of corporate documents. For every user query, relevant contextual fragments are retrieved from this library via the FAISS engine and appended to the prompt given to the LLM. The LLM then generates an answer grounded in the supplied context. The overall structure and workflow of the proposed RAG solution are described in detail. To justify the choice of the generative component, we benchmark a set of widely used LLMs — ChatGPT, GigaChat, YandexGPT, Llama, Mistral, Qwen, and others — when employed as the answer-generation module. Using an expert-annotated test set of queries, we evaluate the accuracy, completeness, linguistic quality, and conciseness of the responses. Model-specific characteristics and average response latencies are analysed; the study highlights the significant influence of available GPU memory on the throughput of local LLM deployments. An overall ranking of the models is derived from an aggregated quality metric. The results confirm that the proposed RAG architecture provides efficient document retrieval and information delivery in corporate environments. Future research directions include richer context augmentation techniques and a transition toward agent-based LLM architectures. The paper concludes with practical recommendations on selecting an optimal RAG–LLM configuration to ensure fast and precise access to enterprise knowledge assets.

Langevin Dynamics, Monte Carlo, and all-atom Molecular Dynamics simulations in implicit solvent require a reliable source of pseudorandom numbers generated at each step of calculation. We present the two main approaches for implementation of pseudorandom number generators on a GPU. In the first approach, inherent in CPU-based calculations, one PRNG produces a stream of pseudorandom numbers in each thread of execution, whereas the second approach builds on the ability of different threads to communicate, thus, sharing random seeds across the entire device. We exemplify the use of these approaches through the development of Ran2, Hybrid Taus, and Lagged Fibonacci algorithms. As an application-based test of randomness, we carry out LD simulations of N independent harmonic oscillators coupled to a stochastic thermostat. This model allows us to assess statistical quality of pseudorandom numbers. We also profile performance of these generators in terms of the computational time, memory usage, and the speedup factor (CPU/GPU time).