Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Parallel algorithms for the finite-element method

Thesis/Dissertation ·
OSTI ID:5294853

The finite-element method is perhaps the most important numerical tool for the solution of problems in engineering today. Existing computer algorithms for implementing the finite-element method have been developed for execution on serial or single-processor computers. For the powerful new generation of parallel or multiprocessor computers coming into use, new parallel algorithms must be created and exploited. This work develops and implements a new class of parallel algorithms for the finite-element method. These algorithms are derived from a decomposition principle and utilize a mix of direct and iterative techniques that can be programmed in a wholly asynchronous mode. They have been implemented on a shared-memory multiprocessor and applied to some large test problems. The algorithms have been found to produce significant speedup and are very efficient.

Research Organization:
Columbia Univ., New York, NY (USA)
OSTI ID:
5294853
Country of Publication:
United States
Language:
English