ISSN:1063-7192    

DOI:10.1080/10637199308915426

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

The two most popular models of sequential and parallel computation lead, once defined precisely, to a computational paradox. Specifically, we show that for a wide family of problems the cost of a PRAM solution is smaller than that of its RAM counterpart. This contradicts the currently established belief, and does not appear to be amenable to explanation using existing approaches. We use the termparallel synergyto refer to this phenomenon

ISSN:1063-7192    

DOI:10.1080/10637199308915427

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

Knowledge bases are currently being investigated by database researchers in order to extend the expressiveness of the relational model to include deductive capabilities in the form of rules. The motivation behind this paper is to present a formalism which extends the relational approach to homogeneous distributed databases to the knowledge base paradigm. The basis of our formalism of distributed knowledge bases is the S5 modal logic system. We define an extensional distributed database as a modal structure and an intentional distributed database as a DatalogKlogic program. We interpret a modal structure, (W,V), as a network of user sites, W, together with a valuation function, V, which assigns a database to each site w ∈ W DatalogKextends Datalog + negation by incorporating into it the modal knowledge operator, K. Knowing in our context can be viewed as data replication, and possibility can be viewed as data fragmentation. We define the model-theoretic semantics of DatalogKvia implication in a proper subset of first-order modal logic and its operational semantics in terms of two inflationary fixpoint operators, the local fixpoint and the global fixpoint operators. It is shown that the operational semantics of DatalogKcoincides with its model-theoretic semantics and that DatalogKqueries are computable in polynomial time. We also demonstrate that DatalogKcan be implemented in a parallel knowledge base environment over a shared-nothing architecture for a multiprocessor database.

ISSN:1063-7192    

DOI:10.1080/10637199308915428

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

One of the fundamental algorithmic problems in computer science involves selecting the κth smallest element in a collectionAofnelements. We propose a simple selection algorithm running in O(log2n) time on a reconfigurable mesh of size √n × -√n.

ISSN:1063-7192    

DOI:10.1080/10637199308915429

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

In this paper we present an integrated model for decomposition and mapping (D&M) of Locally Connected Layered Neural Networks (LCLNs) on message-passing multiprocessors. Within the framework of this model we analyze two previously proposed D&M strategies for a particular class of LCLNs. The model is compared with the performance of a neural network simulation environment, running on a transputer array. We find that both strategies may be applicable, depending on the network size, and we can determine the problem size at which one strategy is preferred over the other. Furthermore, we find that the regularity of the communication pattern between the processors is an unexpected factor in the D&M decision.

ISSN:1063-7192    

DOI:10.1080/10637199308915430

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

The design proposed here concerns bidirectional linear Very Large Scale Integration (VLSI) processor arrays. No technique has yet been developed to enable such a design where the data items enter the array in consecutive time moments. The problem to be resolved was how to organize the array and the cells to enable each data item to meet all the elements from the other data stream.

ISSN:1063-7192    

DOI:10.1080/10637199308915431

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

A partitioning algorithm is presented, which applies to composite grids for explicit difference methods. Results from four case studies are promising. In fact, close to optimal partitionings were generated by the algorithm. The new algorithm is intended for use in a set of software tools for parallelization of explicit difference methods on MIMD (multiple instruction, multiple data) computers with distributed memory.

ISSN:1063-7192    

DOI:10.1080/10637199308915432

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor