Implementation of the parallel level set method in a multi-core computer

Abstract

Parallel computing is a powerful tool to solve a large scale problem derived from finite element method or finite difference method related to the subjects such as fluid dynamics, weather forecasting, aerodynamics, and image problem. Currently a multi-core computer is available at a low price and the parallel computing with a small number of parallel tasks is possible at home. In the subjects listed above, propagating interface occurs in a wide variety of phenomena, especially in ocean waves, combustion and flame propagation, crystal growth, image restoration, image reconstruction, and image segmentation. To handle the propagating interface the level set method has been widely used. In this thesis, we study the parallel level set method and implement it in the multi-core computing environment. We use the Godunov scheme for numerical implementation of the level set method and divide the computational domain to keep the scalability. In other words, we develop a methodology for the parallel level set method to keep the scalability. In addition, we show the numerical results keeping the scalability.