Block predictor – corrector method on Python-based distributed computing environment

• Main Authors: chih-chieh cheng, 鄭至傑
• Other Authors: Ming-Gong Lee
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/40855202325662510816

碩士 === 中華大學 === 應用統計學系碩士班 === 99 === In this thesis, we study how to apply Python to do parallel computation in a distributed computing structure. MPI-2 (Message Passing Interface) is implemented to be a basic standard on parallel processing to communicate between each CPU inside the structure. MPICH2 (Message Passing Interface Chameleon) is implemented to let individual CPU to do data transmission and reception between other CPUs. While controlling and manipulation of the process is proceed by applying the functions; send、recv、get_size、get_rank, etc, in the package of mpi4py by Python. Numerical experiments are carried on the block method to solve some stiff ordinary differential equations, and their speed up between 1 CPU and two CPU are explored to show positive achievement of the proposed distributed parallel structure.