Shin-ichiro Sugimoto, Hachinohe Institute of Technology Amane Takei, University of Miyazaki Qinghe Yao, Sun Yat-sen University Masao Ogino, Daido University
To perform a numerical analysis of the finite element method on parallel computers such as modern supercomputers, which are consisted by multi-core processors, many-core co-processors, and/or graphical processing units, an efficient parallel algorithm is indispensable. A domain decomposition method (DDM) is well-known method as a parallel algorithm for a distributed-memory parallel computer, and there has been many researches for a numerical method to solve the boundary value problem of the partial different equation (PDE). However, the architecture of modern parallel computers are increasingly complex, making it very difficult to optimize the use of computer performance. The purpose of this minisymposium is to discuss wide-ranging topics of parallel computing and large-scale computation. The DDM is not only numerical methods for PDE but a broader sense of parallel computing.
