Problems of intellectualization and quality of informatics systems

One of the important problems in loop parallelization is a revelation of data dependences. Tests for dependence should find whether there exist integer solutions of a system of linear Diophantine equations obtained at revealing the dependences that satisfy the constraints on the loop bounds. In this work a new modified variant of the λ-test is presented, in which and λ-test is integrated with the exact IR-test. This allowed us to obtain more exact results in the analysis of dependences in multidimensional arrays. A practical result of this work is a test, which can be used in the block of the data dependences analysis in the system of compiler fast prototyping which is now under development.

The paper briefly considers the possibility of application of modified constructions of the language of symbolic transformations REFAL to formation of a tree-like representation of a sentence in a natural language and a "question-answer" scheme. The algorithm of using them in search systems is described.

The paper considers polyhedral graphs (graphs of polyhedron or planar 3-connected graphs). A face α of size k of a polyhedral graph is of type (a1, a2, ..., ak) if the vertices incident with α in a cyclic order have degrees a1, a2, ..., ak and this sequence is lexicographically minimal. A polyhedral graph is oblique if it has no two faces of the same type. The number of oblique polyhedral graphs with up to 12 vertices is found. Some additional properties of such graphs are also examined.

The paper presents the algorithms and software support for research in the field of physiology and for the process of biotraining (biological feedback) for medical and other purposes.

The article is devoted to the development of a research system for analysis of texts in a natural language which is intended to carry out linguistic research and to test various algorithms of text processing.

With the advent of the digital age and the Web, museums and cultural heritage institutions are rethinking their roles. More and more museums make a decision to maintain their websites (digital museums) in order to provide useful information and attract new visitors.

The paper describes some new possibilities related to presentation of museums in the Web. The main properties of the museum sites and virtual museums are also described. Some approaches to access unification and integration of information resources are presented. The methods of adaptive hypermedia are considered. The possibilities of open virtual museums are studied. A short description of an open virtual adaptive museum of informatics history in Siberia is given.

The paper presents a project of an adaptive system WAPE aimed at support for distant education in programming.

This work presents analysis of algorithms constructed for the development of an application for a mobile device which could detect its position with respect to the previous one by analyzing images made by its video camera.

The paper considers a process of video camera calibration. Some calibration techniques are described.

The number of vertices in the longest simple path of a graph G is denoted by τ(G). A subset S of the vertex set V(G) is called Pn+1-free set in the graph G, if τ(G[S]) ≤ n. Pn+1-free set of maximum cardinality in the graph G is called maximal Pn+1-free set of the graph G.

There is a well known hypothesis that for any graph G and n < τ(G)/2 there exists a Pn+1-free set M such that τ(G - M) ≤ τ(G) - n.

In this work we had proved this hypothesis for n ≤ 8.

The article describes a block of high-level reducing optimizations for Sisal 3.0 compiler. The reducing optimizing transformations are transformations which do not decrease any qualities of a program. The block performs the following high-level optimizations: Common Subexpression Elimination, Constant Folding, Constant Propagation, IF-transformations, Dead Code Elimination, etc.

The paper presents an algorithm that rewrites a loop nest under any nonsingular loop transformation. This algorithm, first, calculates the loop bounds using the Fourier-Motzkin elimination method and, next, adjusts these loop bounds and calculates the loop strides using a method based on the theory of Hermite normal form. The algorithm works nicely with unimodular transformations being treated as a special case.

This paper describes the process of deciding a problem related to determination of blood parameters using the spectrum reflected from the body surface. The decision is found in four stages: description of physical processes taking place in light reflection and creation of a physical model, analysis of spectrums, data preprocessing, the choice of the most appropriate network architecture and of the algorithm of its learning. Some specific features of the Significo environment are also considered.

This paper consists of two parts: the place of the Significo environment among other modern neural packages is shown in the first part, and the second part presents its description. The first part generally describes the applications of neural network modeling and gives the benchmark analysis of a variety of neural packages like NeuroSolutions, NeuralWorks, Process Advisor, NeuroShell 2, etc. The second part presents the base architecture of the Significo environment and the interaction scheme of its elements, as well as its user groups.

The paper considers the general defining qualities of dataflow programming languages and briefly describes the specific features of some widespread and experimental dataflow languages, such as Lucid, Id, Val, Post, Sisal and Pythagoras. All versions of the Sisal language, including those being under development, and their basic distinctions are presented.

In biomedical research, gene expression is studied with the help of DNA-microchips and, to analyze the increasing data amount obtained by this technology, clusterization is used. The paper considers the methods of clusterization, which are divided to hierarchical and iteration ones, and is aimed at the development of a new clusterization algorithm that can find a solution of the clusterization problem close to optimal.

The paper presents a review of investigations related to the consensus problems: variants of the problem statements and results related to the quality of solutions: time-, bit- and message-complexities.

The subject of research are web site content management systems also known as CMS. Aim of work is development of CMS, that is optimized for laboratorial web site management. Several existing CMS were examined with principles of their work, demands, made to them, and methods of their evaluation during the research. These demands were examined and expanded with demands for laboratorial web sites. Distinctive for these sites is the management of information that is atypical for average statistical sites, for which the existing CMS were made. The next stage of work was the development of new CMS, satisfying all demands and with number of advantages in comparison with other CMS. On the basis of the new CMS the Program Construction and Optimization Laboratory site was created. This site was used for examination of CMS functioning.