Block Cyclic Reduction 기법에 의한 대형 Sparse Matrix 선형 2계편미분방정식의 효율적인 병렬 해 알고리즘

An efficient parallel solution algorithm on the linear second-order partial differential equations with large sparse matrix being based on the block cyclic reduction technique

선계2계 편미분 방정식의 일반식에 대한 계수 메트릭스를 (n-1)x(n-1) submatrices로 나누어서 block tridiagonal system으로 변환한 후 cyclic odd-even reduction 기법을 응용하여 large-grain data granularity로서 미지벡타를 구하는 block cyclic reduction 알고리즘을 작성했다. 그런데 이 block cyclic reduction 기법은 매 연산의 단계마다 병렬성이 변하여 병렬처리형 컴퓨터에는 적합하지 못하므로 이 기법을 변형해서 병렬성이 일정하며 실행시간이 보다 단축되는 block cyclic reduction 기법을 제안하고 이 기법에 의한 선형2계 편미분 방정식의 일반식의 解를 구하는 알고리즘을 작성하여 기존의 기법과 비교 고찰했다.

The co-efficient matrix of linear second-order partial differential equations in the general form is partitioned with (n-1)x(n-1) submartices and is transformed into the block tridiagonal system. Then the cyclic odd-even reduction technique is applied to this system with the large-grain data granularity and the block cyclic reduction algorithm to solve unknown vectors of this system is created. But this block cyclic reduction technique is not suitable for the parallel processing system because of its parallelism chanigng at every computing stages. So a new algorithm for solving linear second-order partical differential equations is presentes by the block cyclic reduction technique which is modified in order to keep its parallelism constant, and to reduce gteatly its execution time. Both of these algoriths are compared and studied.