Application of Parallel Substitution Algorithm for spatial dynamics simulation
Parallel Substitution Algorithm (PSA) is a formal model of fine-grained parallel computation. It has been developed and used for design and investigation of highly parallel algorithms and digital systems architecture. Here it is shown, that its expressive capabilities allow also to use PSA formalisms for representing a wide range of...
Use of parallel computation for estimation of coefficients of heat equation by Monte Carlo method
The inverse problem of determination of coefficients of multidimensional heat equation is considered. The method of statistical modeling of trajectories of corresponding systems of stochastic differential equations is applied to the solution of the direct problem and sensitivity analysis. The application of parallel computers allows a significant increase of the...
The domain decomposition parallel algorithm for multi-dimensional parabolic equations
In the present paper, one of the methods of parallel solution to the multidimensional parabolic equations on the multiprocessor computing systems of the MIMD type is proposed. Much attention is given to the domain decomposition method and to distribution of subdomains among computers. As a computing system, on which the...
The Akers problem for variable operation processing times
The well-known “ob-shop” problem in the wider formulation (many machines of the same sort and many workers per one operation) is discussed in the paper. The graphic approach to the problem with two jobs and variable processing times of operations is proposed. It is shown that the more general problem...
On existing optimal three-dimensional circulant networks
The optimal circulant graphs have minimum diameter for the given order N and degree ν and, respectively, optimal features with respect to communication delays, reliability and connectivity under implementation as interconnection networks in multimodule supercomputer systems. The questions of existence of optimal (nearly optimal) circulants with the degree ν =...
Program system environment for visual developing and mapping of parallel programs
A model of parallel program called program skeleton and a program system for manipulating and translating the program skeletons into parallel programs on various high-performance computing systems are considered. The main goal of this system is to make it easier to create and execute parallel programs on various parallel computing...
Finding single-source shortest paths using associative parallel processors
In this paper, we employ Dijkstra's algorithm for finding single-source shortest paths in directed graphs. We propose an efficient implementation of this algorithm on a model of associative parallel processors of the SIMD type with bit-serial (or vertical) processing (the STAR-machine). Moreover, we show how to extend this implementation for...
Expulsive tree data structures for fast data search by a key
We intend to draw a comparison between the different most known data indexing structures so as to outline the bottlenecks and faults from the point of view of their utilization in the fine grained algorithms simulating system WinALT. Then we would like to propose some ways to eliminate these drawbacks...
Formation of basic type autowave processes by a cellular neural network
In this work, a formal background for the choice of parameters of Cellular Neural Network generating basic types of autowaves, namely, a round traveling front, a round traveling pulse and a spiral wave, is presented. This background is based on investigation of phase plane properties of a CNN cell. The...
Numerical experiments with one ILP algorithm
In this paper, a branch and bound algorithm with branching in depth for problems of the integer linear programming (ILP) is considered. The results of solution to large problems of integer and mixed-integer linear programming are presented. A brief information is given on the software for solution to large ILP...