Premium
Concurrent processing: The key to rapid probabilistic potential methods
Author(s) -
Fusco V. F.
Publication year - 1990
Publication title -
international journal of numerical modelling: electronic networks, devices and fields
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.249
H-Index - 30
eISSN - 1099-1204
pISSN - 0894-3370
DOI - 10.1002/jnm.1660030102
Subject(s) - computer science , monte carlo method , probabilistic logic , key (lock) , computation , concurrency , algorithm , computational science , mathematical optimization , topology (electrical circuits) , distributed computing , mathematics , artificial intelligence , statistics , computer security , combinatorics
This paper shows how probabilistic potential methods based on the Monte Carlo approach can be efficiently mapped onto an array of concurrent processors and how the resulting algorithm can provide rapid solutions of harmonic function problems. The method described here exploits the natural concurrency of the Monte Carlo approach. Computation is achieved by use of a low‐cost array of transputers linked to a PC‐AT. The resulting methodology is powerful and the ideas suggested should have application to many other modelling techniques used for field modelling. A microstrip geometry is used as the discussion vehicle throughout this paper. This geometry was selected since it exhibits a number of the important topological features required in a general problem, e.g. Dirichlet boundary conditions and mixed dielectric. The design of the parallel algorithm used is discussed in detail, and typical results are given.