SSD Main Publications

NOTES.

Programmirovanie is considered as "number one" Russian Journal in theoretical and application programming.
Cybernetica is concedered as a main Jornal in theoretical computer sciences.

BOOKS

Ach90
Authors: S.M.Achasova, O.L.Bandman
Title: Correctness of Distributed Computational Processes
In: Novosibirsk: Nauka, 1990. - 250pp. (in Russian)
Abstract: Theory of parallel processes behavior and methods for providing their validity are studied from the point of view of algorithmical-oriented architectural design of special purpose processors. The concept of validity is defined in terms of behavioral properties of parallel processes, such as determinism, termination, safeness, absence of deadlocks. Validity conditions are formulated using the languages of parallel microprogramming and parallel flowcharts, Petri Nets and computational graphs being used as formal computation models. Validity conditions for both synchronous and asynchronous modes of parallel computation are defined. Validity checking procedures are reduced to those for behavioral properties of Petri Nets. The most attention is paid to the analysis of general and free-choice Petri Nets as abstract models of parallel flowcharts.

Ban94.1
Authors: S.Achasova, O.Bandman, V.Markova, S.Piskunov
Title: Parallel Substitution Algorithm. Theory and Application
In: World Scientific: Singapore, 1994. - 220pp.
Abstract: A Parallel Substitution Algorithm (PSA) is a new model of distributed (cellular) computations. It provides a concise mapping of distributed computation processes into cellular arrays. A PSA is specified by a set of parallel substitutions operating over a cellular array. Two concepts make PSA a powerful tool for modelling cellular computations: 1) naming functions which allow to specify any type of interactions in the computation space, 2) a context which serves to represent control of a computation process in time. The foundation of PSA theory comprises validity conditions of computations in the synchronous and asynchronous modes, space-time, space-space (2D 6 3D) and global-local equivalent transformations of PSAs, composition and decomposition of PSAs, interpretation of PSAs with automata nets. On the basis of PSA theory a variety of tools and techniques is developed for designing algorithmic-oriented cellular VLSI and optical architecture.

Val88
Authors: V.A.Valkovskii, V.E.Malyshkin.
Title: Synthesis of Parallel Programs and Systems on the Basis of Computational Models.
In: Nauka, Novosibirsk, 1988. (In Russian).
Abstract: The book is devoted to the description of the pragmatic method of parallel programs synthesis on the bases of special formalization of the object domain, so called Computational Model. The method enable us to synthesize a high performance parallel program for MIMD multicomputers.

Val90
Authors: V.A.Valkovskii,V.E.Malyshkin
Title: The Elements of Modern Programming and Supercomputers.
In: Nauka, Novosibirsk, 1990. (In Russian).
Abstract: This is a popular book devoted to supercompers and their software. It is written on the materials of the classes given the authors to the participants of annual Summer schools of young prigrammers (Novosibirsk).

PAPERS:

Ach96
Authors: S.M.Achasova.
Title: Cellular-Neural Combinatorial Optimization Algorithms.
In: Proceedings of the Conference "New Information Technologies in Discrete Structures Study". Ekaterinburg, 1996. - P. 156 - 161.
Abstract: The neural approach to combinatorial optimization problems is outlined on the basis of a generalized neural model. Cellular-neural combinatorial optimization algorithms are presented which are based on two models of distributed computations, namely, Parallel Substitution Algorithm (generalization of Cellular Automaton) and Artificial Neuron Networks. The potentials of the model of cellular-neural computations for organizing different modes of execution of the algorithms are discucced.

Ach91
Authors: S.M.Achasova.
Title: Correctness of Interpretations of Parallel Substitution Systems.
In: J. New Gener. Comput. Syst. 4 (1991) 1, P. 19 - 27.
Abstract: Parallel substitution systems are intended to process large data arrays. Systems of stationary parallel substitutions are parallel microprograms. Synchronous and asynchronous modes of execution of parallel microprograms are considered. The necessary and sufficient conditions of determinacy and termination of parallel substitution systems are stated. Theorems are formulated on the relationship of correctness conditions for parallel microprograms in synchronous and asynchronous modes.

Ach93
Authors: S.M.Achasova.
Title: Correctness of Synchronized Cellular Computations.
In: Proceeding of the International Conference "Parallel Computing Technologies", Obninsk, Russia, Aug 30 - Sept 4, 1993, P. 543-556.
Abstract: Operation of the parallel microprograms with microinstructions of various execution time over cellular arrays is studied. Correctness conditions for such microprograms is stated. A method of recognition of incorrectness is presented.

Ach95
Authors: S.M.Achasova.
Title: Synchronous-Asynchronous Cellular Computations.
In: Lect. Notes in Comput. Sci., 964, 1995, P. 1-6.
Abstract: Operation of parallel substitutions over cellular arrays in synchronous - asyncronous mode is studied. Correctness conditions for parallel substitution system in this mode of execution are stated.

Ani91
Authors: V.A.Anisimov, V.E.Malyshkin.
Title: Assemble Parallel Programming System INIA.
In: In Proceedings of PaCT-91 International Conference (Novosibirsk, USSR, 1991), 316- 324. World Scientific, Singapore.
Abstract: Procedural language and system INIA supporting assembly technology are presented. This sytem was implemented for multicomputer system SIBERIA assembled out mainfraim ES-1068 as a host and 8 ES-2706 as processor elements. At the bases of this system the method of linearized computation lies.

Ban95
Authors: O.L.Bandman
Title: Cellular-Neural Computations. Formal Model and Possible Applications
In: Lecture Notes in Computer Science, 964 (1995). Ed. Malyshkin. - P.21-35
Abstract: A formal model for fine-grain parallel computations is presented, which combines the connectionist approach of Artificial Neuron Networks with the cellular-like communication structure. The model is based on the concepts and formalisms of Parallel Substitution Algorithm, which is considered be the most theoretically advanced generalization of Cellular Automaton. Principles of cellular-neural algorithms synthesis are discussed and three simple examples (optimization on graphs, pattern retrieval, image processing) are given.

Ban93.2
Authors: O.L.Bandman
Title: Electrooptical implementation of cellular arrays for data processing
In: Zarubezhnaya Radioelectronica, N2, 1993. - P.64-75
Abstract: A method for logical structure synthesis of electrooptical cellular special purpose processors is proposed, the implemented algorithm being given by a parallel microprogram (cellular automaton description language). The target of such a proposition is twofold: to show the possibilities of electrooptical circuits, and to demonstrate the effectiveness of parallel microprogramming techniques for design and investigation of electrooptical devices of cellular architecture. A method for implementation complexity and expected characteristics assessment is given.The proposed methods are illustrated by an example of cellular multiplier.

Ban93.1
Authors: O.L.Bandman
Title: Organization of massive computations in Optical Computers (a survey)
In: Zarubezhnaya Radioelectronica, N2, 1993. - P.64-75.
Abstract: A survey of research in the field of optical computing and optical computer architecture is presented. Most attention is paid on the investigations directed to explore fine-grained parallelism, such that shadow casting principle, simbolic substitution method and neuron networks. Both algorithmic aspect and implementation techniques are considered in details. Some projects of optical computers under construction are also described.

Ban94
Authors: O.L.Bandman
Title: Synchronous-Asynchronous Transformation of cellular Algorithms
In: Bulletin of the Novosibirsk Computing Center. Series: Computer Science, issue 2 (1994). - P.25-44
Abstract: Asynchronous versus synchronous cellular computations are inmvestigated in terms of Parallel Substitution Algorithm ( a formal model of fine-grained parallel data processing). A general method for constructing asynchronous cellular computation equivalent to a given cellular algorithm is presented. It is shown that more than a twofold increase in cell-complexity should be paid for the abandonment of synchronizing clock in cellular VLSI systems.

Ban93.3
Authors: O.L.Bandman, V.P.Markova, S.N.Piskunov.
Title: 2D->3D Transformation of Cellular Algorithms
In: Parallel Computing Technologies. Proceedings of the International Conference on Parallel Computing Technologies, ed. V. Malyshkin.
Abstract: The most efficient solution of connection problem in VLSI technology lies in using multiplanar architecture, in which connection between layers are implemented by light signals. Since the representation of algorithms distributed in the 3D-space is a difficult task, the transformation 3D one is developed. Three methods of 2D-algorithm stratification are presented. All of them in the framework of the abstract model of cellular computations, named Parallel Substitution Algorithm. The methods differ in the degree of resulting parallelism.

Ban91
Authors: O. Bandman, S. Piskunov.
Title: Parallel Substitution Algorithm as a Model for Distributed Computations
In: Journal of New Generation Computer Systems.-1991.-Vol. 4, N 1.-P. 3.
Abstract: A Parallel Substitution Algorithm (PSA) is an abstract model of distributed computations which provides formal ways of creating facilities for the computer-aided design of algorithmically oriented VLSI and optical architectures. A PSA is specified by a set of parallel substitutions acting over a cellular array. The paper presents main theoretical results in PSA investigation. It starts with an informal description of PSA features. Then definitions of main concepts are given in a strict mathematical form. Methods of composition and interpretation of PSA automata nets are developed. In the conclusion PSA aplications and future investigations are discussed.

Ban93
Authors: O.L.Bandman, V.P.Markova
Title: Spectral Analysis and Synthesis of Logical functions for special-purpose VLSI implementation
In: Bulletin of the Novosibirsk Computing Center. Series: Computer Science, issue 1(1993). - P.5-23
Abstract: Some results of the investigation of spectral properties of k-valued (k>2) logical functions are presented. The main goal is to show the power of spectral methods in implicant extraction and recognition of some useful logical function properties. The relation between the sum-of-product form of a logical function and its spectrum is established by means of a group representation of implicants. It allowed to use some results of abstract harmonic analysis of discrete-valued functions. A number of theorems which state the necessary and sufficient conditions for the existence or absence of an implicant and for recognition of monotonicity and symmetry are obtained. The application of the results is shown by presenting an algorithm for deriving an irredundant sum-of-product form of a logical function.

Kol96
Authors: Yu.I.Kolosova
Title: General scheme of algorithms for preservation of causal ordering events and its applications
In: Avtometria, N3, Novosibirsk, 1996 (to appear)
Abstract: The existing algorithms to control the ordering of events of the reception of messages to separate processes according to the ordering of events of the emission of the same messages from other processes of distributed programs have specially been developed to reduce nondeterminism, which is due to the asynchronous nature of the communication channels. The general scheme of these algorithms is considered, and a new algorithm, which is built on its basis, is proposed. It is shown, that this algorithm could be better suited in a certain class of applications. An example of the use of the control scheme is demonstrated for the development of a concrete task.

Kol95.1
Authors: Yu.I.Kolosova
Title: Implementation and analysis of causal ordering of events for distributed computations
In: Novosibirsk, 1995. - 46 pp. Preprint, Siberian Division of RAS, Computing Center; no.1046
Abstract: We consider the existing protocols to implement the causal ordering of events in message-passing systems and propose an off-line algorithm to analyze the underlying ordering. The algorithm uses logical time and detects all departures of the actual execution order of events of an indicated process from an expected one. The proof of correctness of the algorithm is also given and its usefulness for the correct evaluation of global predicates is shown.

Kol91
Authors: Yu.I.Kolosova
Title: On Studying Information of Dynamic Analysis of Parallel Programs
In: Upravlyajushie Systems i Machine (USiM), N4, Kiev, 1991. - P.53-61
Abstract: A problem of debugging the interactions of processes of parallel programs is considered. We propose an approach where accumulation, storage and mapping of information of such interactions are based on a relational model. A logic analyzer in the PROLOG language is presented. It allows an inference of extensive information on parallel processes run in terms of the source programming language. The promising directions of further development are discussed.

Kol93
Authors: Yu.I.Kolosova
Title: Parallel Histories - a Basis for Revealing Race-Competing Processes
In: Kibernetika i System Analysis, N4, Kiev, 1993. - P.144-154
Abstract: An approach permitting the detection of race-competing processes in program execution on computers of MIMD-architecture is considered. The approach is based on the post-program analysis of the information intended for execution replay, which ensures the reexecution of a parallel program to be equivalent to the initial execution. Processes communicate via a message passing on the basis of blocking and non-blocking primitives.

Kol95
Authors: Yu.I.Kolosova
Title: The method for preservation of causal ordering events
In: Proceedings of the 5th International Workshop on Distributed Data Processing,
Abstract: The existing algorithms to implement causal ordering of events for distributed programs, which are built with message-passing communications only, are mainly introduced to reduce nondeterminism, that is due to the asynchronous nature of the communication channels. A new algorithm proposed in this paper has some similarities with these implementations. The algorithm is much more involved for the actual causality than for the potential causality and can prove to be sufficient in some applications. The proof of correctness of the algorithm is also given.

Mal94
Authors: V.E.Malyshkin.
Title: Assembly Environment for Development of Application Parallel Programs.
In: Proceedings of HPCN (High Performance Computing and Networking) International Conference (Munich, Germany, 1994), Vol.II: Networking and Tools, Springer Verlag, LNCS No.797, p. 302-307
Abstract

Mal93
Authors: V.E.Malyshkin.
Title: Assembly Parallel Programming: some examples.
In: -In Proceedings of PaCT-93 International Conference (Obninsk, Russia, 1993), 191-198.
Abstract: Examples of assembly technology applications are presented.

Mal95
Authors: V.Malyshkin.
Title: Functionality in ASSY System and Language of Functional Programming.
In: Proc. "The First Aizu International Symposium on Parallel Algorithms/Architecture Synthesis", 1995, Aizu-Wakamatsu, Fukushima, Japan. IEEE Comp. Soc. Press, Los Alamitos, California. P. 92-97.
Abstract: The notion of functions in ASSY system, their representation and implementation (synthesis) are discussed. Main features of Assembly System of parallel programming are presented.

Mal91
Authors: V.E.Malyshkin.
Title: Linearized Mass Computation.
In: - In Proceedings of PaCT-91 International Conference (Novosibirsk, USSR,1991), pp. 339 - 353, World Scientific, Singapore.
Abstract: An approach to organization of parallel computations on large-block high-performance hiearchical multicomputer systems (MCS) is presented which is named the linearized mass computations. This approach includes the design of hierarchical MCS for large-scale computation, representation of the algorithm as a system of linearly ordered processes with linear interactions and algorithms of MCS resources allocation. This approach has been used in Novosibirsk Computing Center for the development of the MCS Siberia. On the basis of this MCS a user-friendly parallel programming system Inya has been developed. The Inya system supports the development of applied parallel programs which tune to available MCS resources in the course of their execution.
Text: HTML file (128Kb)

Authors
M.Kraeva, V.Malyshkin.
Title
Implementation of PIC Method on MIMD Multicomputers with Assembly Technology.

In
Proceed. of HPCN'Europe 1997 (High Performance Computing and
Networking) International Conference, LNCS, Vol.1255, Springer Verlag, 1997. pp. 541-549.

Abstract
Linearization of the Particle-In-Cell method and its implementation for multicomputers is discussed. The parallel program is dynamically tuneable to the available resources of multicomputer and particles distribution. Actually, the program is able to be tuned to the behavours of physical phenomenon!

Mal92
Authors: V.E.Malyshkin.
Title: Organization of Parallel Computations on Large-Block Multicomputer Systems.
In: Programmirovanie, No.4, 1992,.(In Russian).
Abstract: The paper describs the method of linearized mass computation which forms the mathematical basis for the INIA programming language and system. This system supports the procedural assembly technology of parallel programming oriented to realization of numerical algorithms first.

Mar95.2
Authors: V.P.Markova
Title: 3D Cellular Pipelined Algorithm for Many Number Pairs Multiplication
In: To appear in Proc. PARCELLA'95
Abstract: 3D Cellular Pipelined Algorithm for Many Number Pairs Multiplication has a minimal period (4 steps) and is well adapted to electro-optical implementation. A small period is accomplished first by using the binary signed-digit number system, and second by pipelining at both the initial data and the computation process levels. Pipelining at the computation process level is achieved by the stratification of a 2D pairwise summation over a 3D array. Here the stratification is based on the replacement of the neighborhood in the rows in the 2D array by the neighborhood in the layers in the 3D one. The use of the third dimension allows not only to speed up the 2D algorithm, but to simplify the topology of each layer of the 3D structure, interpreting the 3D algorithm, and to transfer the greater part of interconnections of the 2D cellular structure into interlayer connections of the 3D cellular one as well.

Mar93
Authors: V.P.Markova
Title: Electrooptical implementation of cellular multiplier
In: Avtometriya, N3, 1993. - P.14
Abstract: By means of the STAR-machine with bit-serial processing for graphs represented as a list of triples (edge vertices and their weights) we have compared the implementations of the associative version of the Prim-Dijkstra algorithm and the Baase algorithm.

Mar95
Authors: V.P.Markova
Title: Multilayer Cellular Algorithm for Complex Number Multiplication
In: Proceedings of ASAP-95, Strasburg, France, July 16-20, 1995. - IEEE Computer Society Press: Los-Alamos
Abstract: Presented 3D cellular algorithm for complex number multiplication based on Knuth number system allows to multiply two complex numbers with the speed close to the multiplication speed of conventional binary numbers of the same length in differ from algorithm [5]. Such a speed is achieved by means of the processing the staff and the unstaff digits of the intermediate results in parallel in different layers of a 3D array, i.e., the summation of the intermediate results are performed without waiting for them to be reduced.

Mar94
Authors: V.P.Markova
Title: The Cellular Knuth Algorithm for Complex Number Multiplication
In: Proceedings of the VI International Workshop on Parallel Processing by Cellular Automata and Arrays,
Abstract: 2D cellular algorithm for complex number multiplication based on Knuth number system (NS) is proposed. It is known that this NS possesses the interesting property: the possibility of representing a complex number as a single vector. However it has a disadvantage which absent in all positional NS. It is the reduction of the intermediate results (the modification unstaff digits, i.e., digits non belonging to the base) which precedes by their summation. Here the influence of the reduction on the time complexity is decreased due to following features algorithm: 1) the generation of intermediate results is performed concurrently with their reduction, 2) the reduction of the bad digit to the base is done in parallel and 3) the pairwise summation of numbers is starting without waiting all result to be re- duced.

Mar96
Authors: V.P.Markova, U.Pogudin, S.N.Piskunov
Title: Formal Methods, Language Tools and Techniques for Synthesis of Cellular Algorithm and Architecture
In: To appear in Programmirovanie, N4, 1996.
Abstract: By means of the STAR-machine with bit-serial processing for graphs represented as a list of triples (edge vertices and their weights) we have compared the implementations of the associative version of the Prim-Dijkstra algorithm and the Baase algorithm.

Nep96.1
Authors: A.S.Nepomniaschaya
Title: An Associative Version of the Prim-Dijkstra Algorithm and its Application to Some Graph Problems
In: LNCS, Proceedings of the Second Intern. Memorial Conf. "Perspectives of System Informatics", 1996
Abstract: By means of the STAR-machine for undirected weighted graphs represented as a list of triples (edge vertices and their weights) we have analyzed associative algorithms for finding a connected component, a minimal spanning forest, transitive closure of a graph, for verifying an articulation point and verifying a bridge. These algorithms are based on an associative version of the Prim-Dijkstra algorithm. They are written as the corresponding STAR procedures whose correctness is proved and time complexity is evaluated.

Nep95
Authors: A.S.Nepomniaschaya
Title: Comparison of two MST Algorithms for Associative Parallel Processors
In: Proc. of the 3-d Intern. Conf. "Parallel Computing Technologies", PaCT-95,
Abstract: By means of the STAR-machine with bit-serial processing for graphs represented as a list of triples (edge vertices and their weights) we have compared the implementations of the associative version of the Prim-Dijkstra algorithm and the Baase algorithm.

Nep93.1
Authors: A.S.Nepomniaschaya
Title: High level language STAR for associative parallel processors and its application to relational algebra
In: Bulletin of Novosibirsk Comput. Center, Series: Computer Science, 1993, 1
Abstract: By means of the language STAR the realization algorithms of relational algebra operations have been constructed and investigated. The constructed procedures form a hierarchy in which the operation Selection is the simplest one and operations Division and Join are the most compound procedures.

Nep93
Authors: A.S.Nepomniaschaya
Title: Investigation of Associative Search Algorithms in Vertical Processing Systems
In: Proc. of the Intern. Conf. "Parallel Computing Technologies",
Abstract: Search algorithms based on different specifications including equality, maximum, minimum, greater (less) than, nearest to and between limits have been analysed. For the constructed procedures we prove theorems justifying their correctness. For some procedures we have found conditions allowing one to decrease their access number to the parallel memory. In addition, we have improved the well-known Falkoff's tree algorithm for defining the nearest to a given number.

Nep91
Authors: A.S.Nepomniaschaya
Title: Language STAR for Associative and Parallel Computation with Vertical Data Processing
In: Proc. of the Intern. Conf. "Parallel Computing Technologies"
Abstract: A high-level language STAR has been worked out. Its operational semantics is described by means of an abstract model (the STAR-machine). The language STAR is related to Pascal, but has some special data types and the corresponding operations for them providing associative and parallel processing. Algorithms are written as the corresponding STAR procedures whose complexity is evaluated by the access number to the matrix memory during its execution.

Nep96.2
Authors: A.S.Nepomniaschaya
Title: Representation of the Gabow algorithm for finding smallest spanning trees with a degree constraint on associative parallel processors
In: LNCS, Proceedings of the EuroPar'96
Abstract: Using the STAR-machine we have described an associative version of the Gabow algorithm for finding smallest spanning trees with a degree constraint. We have obtained that on an associative parallel processor this algorithm takes the same time as a minimal spanning tree algorithm, that is, 0(n\log n) time, where n is the number of vertices in a given graph.

Nep95.1
Authors: A.S.Nepomniaschaya, Y.I.Fet
Title: Investigation of Some Hardware Accelerators for Relational Algebra Operations
In: IEEE, Proceedings of the First Aizu Intern. Symp. on Parallel Algorithms/Architectures Synthesis,
Abstract: The speed up of implementation of a relational algebra operation group at the expense of two specialized co-procesors application, introduced by Y.I.Fet, is investigated. To this end we have itilized an extended version of the language STAR and the modified abstract model for simulating both the STAR-machine run and the run of the co-processors. For each operation from the group we have compared complexity of corresponding procedures written both for the STAR-machine and for modified abstract model.

Nep95.2
Authors: A.S.Nepomniaschaya, V.J.Tskhay,
Title: Modelling of an Associative Parallel Processor Run (Description of a Language and its Translator)
In: Teoriya i Sistemy Upravleniya, Izvestia Akademii Nauk, 1995, N3 (in Russian)
Abstract: We have described a translator implemented in Turbo-Pascal of version 6.0. For convenient debugging the STAR-system includes an auxiliary module where for any STAR construction there is its equivalent in Pascal. At first, the translator initializes the vocabulary of reserved words. Then, if necessary, it parses the STAR program using the auxiliary module. Moreover, the STAR-system includes a shell allowing one to edit the STAR program, to create libraries and to execute the program immediately after its translation.

Pis95

Authors

S.Piskunov.

Title

A Multilayered Universal Cellular Structure and Its Electrooptical Implementation

Optoelectronics, Instrumentation and Data Processing (Avtometriya).-1995.-N 3.

Abstract

A 3-D universal cellular structure is suggested. It was developed using a computer system for simulating cellular computations (Parallel Substitution Algorithm). The structure contains four lauers. All the layers, exept one, are actually a memory with optical inputs and outputs of cells. The only layer other than memory is rather simple: it consists of an iteratively repeated match circuit. It is shown that the structure may have a high manufacturability and speed, allows a compact replacement of the simulated digital device.

Pis93
Authors: S.Piskunov.
Title: Construction of Maximally Stratified Electrooptical Cell Devices
In: Optoelectronics, Instrumentation and Data Processing (Avtometriya).-1993.-N. 3.-P. 3.
Abstract: Conversion of a two-dimensional cell structure to a three- dimensional (multilayer) one is suggested. The conversion is based on the distributed calculation model refered to as a Parallel Substitution Algorithm. A characteristic feature of the conversion is that it provides a possibility for designing an electrooptical device of maximal stratification in implementing a three-dimensional structure in circuitry. This device has a layer consisting of memory cells with optical inputs and outputs; the other lauers consist of repeating ordinary electrooptical cells; all interlauyer connections are optical.

Pis92
Authors: S.Piskunov.
Title: Multilayer Electrooptical Structure Design
In: Optoelectronics, Instrumentation and Data Processing (Avtomrtriya).-1992.-N. 4.-P. 48.
Abstract: A method for transition from a two-dimensional computational structure to a multilayer (three-dimensional) structure, which, in essence, implements the same algorithm for date processing, has been developed. The method is based on the distributed-computation model which is called a Parallel Substitution Algorithm. It is shown that when the three-dimensional structure is implemented within the framework of the electrooptical circuitry, the method makes it possible to choose the complexity of an individual cell and the layout of electric connections between the cells which are in one and the same layer and convert a considerable part of electrical connections between the two-dimensional-structure cells to optical interlayer connections in the three-dimensional structure as well. The presentation of the method is followed by the example of designing a four-layer parallel many-number summator.

Roy95
Authors: M.Royak,E.Shurina,Y.Soloveichik, V.Malyshkin.
Title: Parallelization of Computer Code MASTAC Three-Dimensional Finite Elements Method Implementing.
In: LNCS No.964, pp.304-313, Proceedings of PaCT-95, St.Petersburg, Russia, 12-15, September 1995.
Abstract: Program package implementing 3D FEM and its parallelization on the bases of the method of linearized mass computation are presented.

Val81
Authors: V.A.Valkovskii,V.E.Malyshkin.
Title: To More Detailed Definition of the Programming Languages Nonprocedurality.
In: Kibernetica, No.3, 1981.
Abstract: Comparetive nonprocedurality of programming languages is defined

Vsh95
Authors: V.Vshivkov, M.Kraeva, V.Malyshkin.
Title: On PIC method implementation on MIMD multicomputers.
In: preprint N 1052, Computing Centre RAS, Novosibirsk, 1995, 37 pages.
Abstract: We consider the Particle-in-cell (PIC) method in its application to solving the problem of interaction of rarefied plasma flows as well as discretization of this problem and the difficulties of its solving parallelization for MIMD computers. A brief review of publications on parallel realization of PIC method is given. The assembly method of parallel program construction in order to support the numerical solving large-size problems on MIMD computers is discussed.

Vsh96
Authors: V.Vshivkov, M.Kraeva, V.Malyshkin.
Title: Parallel Implementations of PIC Method.
In: Programmirovanie, Vol.2, 1997, pp.39-51
Abstract: The most interesting implementations of PIC method on multicomputers are described. The application of assembly technology to PIC method implementation is discussed.

Vsh96.2
Authors: V.Vshivkov, G.I.Dudnikova, M.Kraeva, V.Malyshkin.
Title: Implementation of particle-in-cell method on distributed memory multicomputers.
In: Voprosy atomnoi nauki i teckniki, to be published in 1996.
Abstract: Assembly technology of PIC method implementation is described.

Best viewed with Netscape Navigator. The page last updated on October 2, 2000. Maintained by Elvire Kuksheva.