Distributed architecture for implementing fast parallel algorithms for two combinatorial optimization problems
Using the Parallel Substitution Algorithm, as a formal model of parallel computations, a distributed architecture is designed for implementation of the following two fast parallel algorithms: one for the maximal independent set problem and the other for the minimum weighted vertex cover problem. The former is a randomizing algorithm based...
Discrete cellular neural networks for image processing
A discrete time version of Cellular–Neural Network (DCNN) is a paradigm of neural networks that has the realistic perspective to be implemented in VLSI. An investigation of DCNN capabilities for image processing is attempted. It is done through detailed study of two typical concrete problems: 1) associative storing, retrieving and...
Object-oriented approach in approximation of the boundary value problems
Several classes of objects for the object-oriented approach in programming the approximation stage of solving the boundary value problems are proposed. They are the classes for a computational cell, for boundary conditions and coefficients of the linear system of equations.
3D mixed boundary value problems: numerical algorithms, data structures and technologies of implementation
Technological aspects of the numerical solution to multi-dimensional boundary value problems (BVP) for partial differential equations (PDEs) using the finite difference, the finite element or the finite volume methods are described. Among them, the mathematical formulation of the problem, definitions of objects under consideration, algorithms of approximation with different accuracy...
Cellular algorithm architecture for long integers multiplication in arrays of restricted size
A new cellular algorithm architecture for multiplication of two long binary integers in arrays of restricted size is presented. The new algorithm is based on “divide and conquer” technique and performed in terms of a model of fine-grained parallelism – Parallel Substitution Algorithm. Time complexity of the new algorithm for...
TOPAS: Java-based system for investigation of parallel algorithms mapping
TOPAS (Test and Optimization of Parallel Algorithms and Structures), a programming tool for visualization, animation and investigation of sequential and parallel algorithms for mapping of parallel program graphs into graph structure of parallel computer systems is presented. The tool is implemented in Java and is accessed on http://rav.sscc.ru/~monakhov/topas.html.
Genetic and heuristic algorithms of synthesis of optimal multidimensional circulant networks
A solution to combinatorial optimization problem of constructing optimal circulant networks with the minimum diameter under given degree and number of graph nodes is considered. The circulant networks and their different applications are the object of intense investigations. Under degree of a graph δ > 4 and large number of nodes...
An associative version of Edmond's algorithm for finding optimum branchings
In this paper we propose a representation of Edmonds' algorithm for finding optimum branchings on an abstract model of the SIMD type with vertical data processing (the STAR-machine). To this end for a directed graph given as a list of triples (edge vertices and the weight) we construct associative algorithms...
The open architecture of WinALT
The necessity of the open architecture for fine-grained parallel model simulating system (WinALT) is discussed. The description of WinALT open architecture and external module interfaces is given. WinALT consists of the language and graphical user's interface subsystems and the kernel. The extensibility of WinALT is implemented in the kernel by...
Limiting capability of cellular-neural associative memory
The problem of achieving Cellular-Neural Associative Memory (CNAM) limiting capability is considered. At first a CNAM learning method based on the idea of Perceptron Learning Rule which provides maximal ability to restore distorted patterns is suggested. Next, expressions for determining self-connection weight values which increase attractivity and decrease the number...
Some results of autowave modeling by cellular neural network
Two-layer cellular network as a simulating model of wave processes is considered. The wave speed dependence from an intralayer connection weight is presented.