Journal of Theoretical and Computational Studies
http://situs.jurnal.lipi.go.id/jtcs/
THE PARALLEL COMPUTATION FOR TRIDIAGONAL SYSTEM IN ONEDIMENSION DIFFUSION MODEL
M. Susmikanti, A. Adhiyaksa and T. Setiadipura
Pusat Pengembangan Informatika Nuklir BATAN, Kawasan Puspiptek Serpong Gedung 71 (lt. 1), Tangerang 15314, Indonesia
Abstract
In applications transport neutron of nuclear reactor has to solve the onedimension diffusion model. The model diffusion onedimension has the tridiagonal linear equation system. Many linear systems that arise from discritization of partial differential equations are sparse and specifically banded. The tridiagonal systems are the most difficult case and needs to be handled differently. Several algorithms have been developed for the parallel solution of the tridiagonal systems. Parallel computing can be help to solve a big system of linear equations for fast computation and efficiently shared memory of integrated computer. Parallel computing has been made breakthroughs that have decreased costs, increased memory space and computing power. The serial and parallel computing for the solution of linear equations problem as known are use the $ \it Gauss$ elimination method. These papers present the use of cyclic reduction method for solving tridiagonal system in diffusion model one dimension, which has been one of the most successful approaches for the tridiagonal system in parallel computing. The process of this algorithm is focused in elements tridiagonal by one dimension such that efficient memory and faster than the elimination $ \it Gauss $ method. Most of the operation in the cyclic reduction can be done in parallel. The parallel computing of tridiagonal for solving the onedimension diffusion model are implementation with integrated used program $ \it C $ and $\it Message$ $\it Passing$ $\it Interface$ $\it (MPI)$.
