Supercomputing support and internet-oriented technologies

Тhis is a review of the 16-th IFIP World Computer Congress held in Beijing, China, in August, 2000 under the title "Information Processing Beyond Year 2000". The IFIP Computer congress is the most important world scientific forum on information technologies where the main problems and the most significant results in modern informatics are considered.

The main problem of program parallelizing is to find out data depedences. This article presents a precise test for data dependence analysis based on analysis of solvability of a system of linear inequalities in two variables on the basis of Shostak algorithm. The model of a source program, as well as the Shostak algorithm and its modification, is given and implementation of the project is described.

This paper is about some features of compilation and code generation for NUMA-architectures. The definition of a NUMA-architecture is given and the difference between multiprocessors and multicomputers is shown. Computers with NUMA-architectures and the peculiarities of software for them are described. A review of the papers related to cycle restructurization and data transformation is also given.

The paper describes the Sisal 3.0 language which is the initial version of the source language of the functional programming system (FPS). Semantic and syntactical features of this language have been briefly described and a review of the papers related to the Sisal 3.0 problems have also been done.

The parallel algorithm for solving a problem of interaction between low-density plasma flows is studied. The most elementary parallel architecture is considered. In the context of some natural assumptions, the run-time estimates have been made for the parallel and sequential cases; the acceleration coefficient has also been evaluated.

The problems connected with parallelizing the "MEDUZA" technique applied to a number of mathematical physics problems are considered.

The algorithms which are a base of "a test desk" program intended to develop applied software for image processing are briefly described.

Graph models and various methods of their processing are often and widely used when constructing software systems of different levels of complexity. Graphs (being a very convenient tool for description of data structure and information flows) are actively used in a number of mathematical problems. The GRAMAL system here described is aimed at 4 points:

• to provide a tool for graph model description;
• to represent this model graphically;
• to provide testing and debugging tools for graph processing;
• to support code generation and its further application.

A data input interface for the information retrieval system Transform is described. A source of data for this system is bibliography on program transformations. The aim of this work is to create a program complex which provides a remote user connected to IS Transform WWW-server with a convenient interface for registration in the system, input of data to the database, information retrieval, and output of results in a convenient form.

This article describes a number of tools which support application of graphs to programming; they are under development in the Laboratory for program construction and optimization of the A.P. Ershov Institute of Informatics Systems. A review of the papers on "encyclopedia" of graph algorithms for programmers is given. Methods and tools for graph (and graph model) visual processing are described. The dictionary on graph theory in programming and its electronic version are considered.

This paper considers equivalence of local computation models. There exist several models of local computations on graphs, such as the systems of graph rewriting with priority, the systems of graph rewriting with supressed context, Zhuravlev's local algorithms, and the finite automata nets. The paper discusses the question of comparison between classes of problems solvable in polynomial time invarious models of local computations. Some propositions and their proofs are given.

A cube-like graph was firstly introduced by Lovas. This is a graph with a special relation of vertex incidence. The article is devoted to a study of some cube-like graph invariants, in particular, chromatic ones. The value of an edge prescribed chromatic number is found, and the method for calculation of the cliquemathic number and the clique number of a cube-like graph is also described.

The mechanisms of graphic output in graph visualization systems are considered. The main problems of constructing such systems are presented and classified and the general solution methods are given for them. A particular implementation of these approaches in the Higres system, which is a visualizer and editor for hierarchical graph models, is described.

The user interface of the Higres system, a visualizer and editor for hierarchical graph models, are described. The methods for visualization of the hierarchical structure are presented which provide one with easy-to-use intuitive graph editor interface. The algorithms for automatic drawing refinement and some additional features implemented in the system are considered.

The Vegras system described in this article is an editor tool for making graphbased illustrations. The motivation to develop such software is the lack of easy-to-use graph editor tools which allow users to make graph-based illustrations for scientific literature. With the help of our system it is possible to create nice looking graph-based illustration in a few minutes. In addition, our system provides a possibility to add different special mathematical symbols to graph's labels, as well as subscripts and superscripts.

An extension of Lisp with a pattern-search procedure based on similar ones of Refal and Haskell languages is described. Some methods for extension of Lisp with pattern-search tools are considered and convenience of using them for structured data processing is estimated.

Today WEB is the most popular information storage. This article discusses the use of project patterns when solving different problems of constructing hypermediaapplications, namely, the process of using patterns aimed to improve constructing the architecture of WEB-Information Systems (WIS).

The decision-making methods in the TRIZ concept are investigated for the process of puzzle generation and solution.