Vestnik SibGUTI


Papers of Journal no. 2019.3

G. B. Abdikerimova, A. L. Bychkov, Wei Xinyu, F. A. Murzin, N. E. Russkikh, E. I. Ryabchikova, S. S. Khayrulin "Methods for detecting and highlighting areas in textural images"  // More...

The article deals with methods for analyzing textural images. Micrographs of plant materials obtained by the transmission electron microscopy are considered. This work was carried out for the Institute of Solid State Chemistry and Mechanochemistry SB RAS. The main goal of the research is the development and implementation of algorithms allowing us to detect and highlight areas of interest to chemists in the image. For example, areas, in which the initial structure of the material is disordered after applying various mechano-chemical methods of processing. To solve this problem, we use: analysis of textural features, clustering, R/S-analysis, orthogonal transformations, wavelet analysis. Much attention was paid to the development of software tools that allow us to select features describing textural differences in order to segment textural regions into subregions. So, the question of the applicability of sets of textural features and other parameters for the analysis of experimental data is being investigated in order to identify characteristic areas in microphotographs that can be associated in future with the porosity, chemical reactivity, etc.


A. I. Adamovich, A. V. Klimov "An approach to deterministic parallel programming system construction based on monotonic objects"  // More...

Due to the explosive growth of the complexity of programs for multi-core processors and supercomputers, the idea of parallel computation with determinism guaranteed by the programming language and system is becoming increasingly significant. This paper analyzes the problem of making parallel programming as deterministic as possible and some of the existing approaches to its solution. It describes the principles of constructing the object-oriented programming system developed by the authors, which allows expert programmers to write both deterministic and non-deterministic code with guarantees to application de-velopers that their programs are deterministic. The system and its input language have two levels: the higher level is intended for application developers; the lower level is for devel-opers of class libraries referred to as monotonic. The input language of the higher-level subsystem is like a functional language with the possibility of creating and using immutable and monotonic objects. The libraries of monotonic classes ensure that all programs in the higher-level sublanguage that use only these classes are deterministic and idempotent when they are parallelized by asynchronous calls of all functions. Some representative applications implemented on this system are described.


T. V. Batura, L. V. Efimova, A. S. Yerimbetova, A. B. Kasekeeva, F. A. Murzin "Temporal and spatial concepts in natural language texts and their investigation"  // More...

The aim of the work is to create a knowledge base containing information on temporal and spatial concepts found in natural language texts. The content of the base: the most important concepts related to time and space, from the S. I. Ozhegov explanatory dictionary; rephrased sentences; the results of the analysis of vocabulary articles (interpretations) and examples from a literature by means of Link Grammar Parser and DIALING software systems, etc. Results of the work can be used in intelligent information retrieval systems. The diagrams obtained at the output of Link Grammar Parser and DIALING systems are extremely interesting material for further research. It is advisable to investigate the possibility of using a number of constructions and concepts of mathematical logic in computer lin-guistics, such as: the construction of L. Henkin, realizability and omitting of types, model completeness, forcing, as well as a number of non-classical logics.


A. Bocharnikov, E. Ignatieva, O. Vishenvskiy "GPU-based algorithm for context analysis of the core promoter region of mouse genes differently expressed in hypothalamic energy-sensing neurons in response to weight-loss"  // More...

De novo motif discovery in the regulatory regions of eukaryotic genes poses a complex computational problem due to the large size of datasets and huge diversity of motifs. This article suggests a new algorithm for measuring the presence of degenerate oligonucleotide motifs written as a 15-letter IUPAC code in a DNA dataset. Its performance has increased 10 times compared with the previous one. There are three key ingredients of this method. The first one is the prefix trees. The second is the relation between motif prefixes and hash ranges in the analyzed nucleotide sequences. The third consists of applying CUDA framework to the massive parallelization allowing to use affordable graphic accelerators.
The context analysis of promoter regions of mouse genes differently expressed (DEG) in hypothalamic AGRP neurons after food deprivation was performed with the proposed method. When an animal is deprived of food, AGRP neurons produce molecules that in-crease appetite and stimulate weight gain. The understanding of how AGRP neurons re-spond to weight loss is important to confront the obesity. Nowadays, this hereditary dis-ease lacks methods of treatment and intervention strategies which would be both safe and efficient in the long term. The performed analysis revealed relevant oligonucleotide motifs that were associated with starvation.


M. Bulyonkov, T. Nesterenko "Research automation of the transport network development"  // More...

A research automation system of forecasting the development of a basic transport network in Russia is considered. A formal model of the transport network allows various types of transport, products, production and consumption, and the task is to minimize the total cost of transportation with respect to a set of linear restrictions. The system provides a user interface for setting and editing all pa-rameters of the transport network. Particular attention is paid to the visual and interactive presenta-tion of simulation results. A holistic perception of the results is achieved by displaying both the input data (mode of transport, throughput, transportation cost) and the modeling results (transported volume and functioning capacity) directly either on the map or on the network plan. The system al-lows us to display either a selected product or all products together. In practice, it is often required to solve the problem that is inverse to simulation, for example, to find out at what tariffs the volumes of transportation over a certain transportation shoulder will exceed a given value. To do this, it is proposed to use stochastic methods based on the mass solution of the transport problem for a variety of network parameters, such as the capacity of the transport shoulder, the tariff for transpor-tation of cargo or its processing in the transport hub. Clustering methods make it possible to distin-guish a relatively small number of “typical” solutions from the entire set of variants. This enables the expert to evaluate both the conditions and the probability of the predicted transport situation. It is also shown that this approach allows us to identify automatically emerging transport corridors and determine the dependence of the volume of transportation on a given shoulder from a specific variable parameter.


L. A. Golubeva, V. S. Gorshunov, V. P. Il’in "Management of an application software package for solving problems of mathematical physics via semantic network"  // More...

The numerical solution of mathematical physics problems using a computer can be divided into several stages: geometric model of the computational domain construction, grid model construction, function approximation, derivatives and integrals, as well as equations solving. There are many grid generators and algorithms for constructing two-dimensional and three-dimensional grids, programs for systems of equations solving, approximators and geometric modeling tools. While creating an application software package for solving problems of mathematical physics based on the concept of basic modeling system, each stage of the problem solving representing as a separate module. Each module, in turn, can be a set of algorithms and subroutines. Such a software package ensures the integrity of the solution to a computational problem thanks to a wide range of tools for going through any of the computational stages and allows you to vary the input parameters and choose the most suitable algorithms at different stages. Moreover, such a system allows the decomposition of the initial computational domain into subdomains when constructing the geometry and the generation of a quasistructured grid model. However, when new algorithms and programs are included into the computing complex, the complexity of its use increases inevitably. Thus, there is a need for the design of a supersystem allowing us to determine the best set of algorithms for subtasks solving at each stage in terms of some quality criteria defined in advance. The purpose of this work is to develop and describe such a model for managing this computing complex using a knowledge base presented in the form of semantic network.


T. Yesikova, S. Vakhrusheva "Agent environment simulation while developing multi-agent system with large-scale infrastructure projects serving as an example"  // More...

The article poses the problem of simulation processes associated with the construction of large-scale infrastructural projects in the context of the Bering Strait tunnel. The research is based on such a simulation method as a multi-agent modeling. The article describes the basics of building a multi-agent system environment in relation to this task. It contains description of the environment interactions and identified agents.


G. Zagorulko, L. Massel "Development of intelligent DSS for preventing energy security threats"  // More...

The paper considers an intelligent decision support system (IDSS) for preventing energy security threats. This system is created by means of a methodology that combines modern approaches and principles for developing systems of this class. IDSS provides access to systematized information on energy security threats and helps to choose preventive measures for solving two specific tasks: simulation of the cooling threat and accident sim-ulation caused by energy producer.


Е. Sidorova "The integrated approach to text lexical characteristics study"  // More...

The integrated approach and software environment for multi-aspect study of the text lexical characteristics are considered. This work is at the junction of corpus linguistics and lexicographical research. The basis of the research is the corpus of text and the problem-oriented dictionary. The proposed environment for supporting the researcher provides tools and interfaces for developing vocabularies and a system of domain features, terms markup, automatic generation of lexical content and accumulation of statistical information, etc. To extract terms the morphological analysis and the construction of phrases based on the rules of matching the grammatical characteristics of words are carried out. To study the contexts of the terms use, concordance construction tools are provided. Concordances allow the researcher to test his or her hypothesis about the functionality of a particular lexical unit. The considered environment allows to solve various text analysis tasks because it integrates various tools for conducting language research and supports customization of vocabularies to a problem area.


V. Shelekhov "Certified compiler design of predicate programs"  // More...

A compiler with deductive verification facilities should be thoroughly certified to obtain high level of assurance. The development of the compiler for a predicate program using a compiler model in accordance with a model-based design method makes the compiler to be of a higher level of assurance. The main part of a compiler model is the model of inner program representation for a predicate program. The model is designed in the framework of third release of the compiler. The model architecture for inner representation of predicate program is described. Type checking is specially analyzed as an important aspect in the compiler development. The verification of the model and compiler development on the base of the model should guarantee a high level of compiler reliability.