An improved Thomas Algorithm for finite element matrix parallel computing

With the expansion of the scale of linear finite element analyzing data, efficiency of computation is a main technical bottleneck. For solving the bottleneck, parallel computation is playing an increasingly prominent role. At present, the researches on linear finite element parallel computation are mainly concentrated on the preprocessing phase, and goal of these researches is to reduce the communication overhead and improve the homogeneous degree. The researches on computation phase dealing with linear finite element analyzing data are rare. However, most of computation cost is in this phase. In this paper, we study and analyze stiffness matrix decomposition and after comparing the different matrix decomposition algorithms, an improved algorithm for parallel computation based on Thomas algorithm is proposed. Verification by a large amount of data proves that the improved algorithm greatly enhances the parallel performance of linear finite element computation.