Details:

Title
Author(s)	Kurt Siegl
Text	Submitted to ISSAC '92.
Type	Technical Report, Misc
Abstract	This paper presents a new system for parallel symbolic computation called kMAPLEk (speak: parallel Maple), in which logic programming provides parallelism and imperative programming provides efficiency. The system is built as a combination between the parallel declarative programming language Strand and the sequential computer algebra system Maple. We describe a novel approach to the parallelization of Buchberger's Groebner bases algorithm using a medium grain pipe-line principle for the polynomial reduction. Additionally we give a reformulation of the "optimization criteria" suitable for a distributed algorithm. The approach is implemented in kMAPLEk on a 17 Processor transputer distributed memory system and tested on a 20 Processor Sequent shared memory machine where it shows a remarkable speed-up.
Keywords
Length	12
File
Language	English
Number	92-11
Address	Johannes Kepler University, Linz, Austria
Year	1992
Edition	0
Translation	No
Refereed	No
Organization	Johannes Kepler University Linz
Institution	RISC (Research Institute for Symbolic Computation)