摘要:

This paper reports on a comparative study of alternative algorithmic approaches for the decomposition and solution of large multicommodity network flow problems. Three specific algorithms are considered: (1) a smooth penalty function algorithm, (2) a coercion function algorithm, and (3) a row-action iterative algorithm. Qualitative comparisons are based on the applicability, ease of implementation and robustness of each algorithm. Quantitative comparisons are based on the performance of the algorithms when applied to a common set of test problems. Special attention is given to the suitability of the algorithms for parallel computations. Direct comparisons with an interior point code, OBI, are also included.

ISSN:1063-7192    

DOI:10.1080/10637199308915446

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

In this paper, we present a set of heterogeneous algorithms for computer vision tasks using the Image Understanding Architecture [IUA]. The full-scale IUA developed jointly by Hughes Research Labs and University of Massachusetts at Amherst is a multiple level heterogeneous architecture. Each lcvel is constructed to perform tasks most suitable to its mode of processing. The lowest level called CAAPP is an SIMD bit-serial mesh. The second level is an MIMD organization of numerically powerful digital signal processing chips. At the top level there are fewer number of MIMD general purpose processors. We propose a set of algorithms utilizing multiple levels of this organization, concurrently. The problems studied include Hough Transform-line detection, finding geometric properties of images, and high level image understanding tasks such as object matching.

ISSN:1063-7192    

DOI:10.1080/10637199308915447

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

In this paper we discuss two parallel data-driven models together with their implementations on multiprocessor systems. The parallel models use a dynamic scheduling strategy, and are for a rule-based expert system. All the models are domain independent. To support the use of these models, a “rulebase compiler” has been built to translate a rule base in text format into the data structure needed by the system. The results indicate satisfactory speed up performance for a small number of processors (< 10) and a reasonably large number of rules.

ISSN:1063-7192    

DOI:10.1080/10637199308915448

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

A special class non uniform recurrent equations produced data dependency graphs which involve non-constant data dependencies on the input dataf(i)is considered. The mathematical expectation of the execution time for the problems described by such recurrences under the assumption thatf(i)are uniformly distributed is found. In case whenf(i)are exponentially distributed it is proved that the mathematical expectation of time tends to the optimal one.

ISSN:1063-7192    

DOI:10.1080/10637199308915449

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor

摘要:

A concurrent object-oriented programming (COOP) model is presented for computer-aided engineering using a blackboard architecture. The basic components of the model are a hierarchy of tasks, global and local controllers, knowledge sources, global and local blackboards, controlling data slots, and numerical data slots. For interprocess communications, named pipes, message queues, and shared memory are used. The model is implemented inC + + on a simulated concurrent processing platform using the system functions provided by UNIX operating system. The COOP model is applied to a complex engineering problem, analysis of high-performance composite laminates and performance results are reported.

ISSN:1063-7192    

DOI:10.1080/10637199308915450

出版商:Taylor & Francis Group    

年代:1993

数据来源: Taylor