A Markov Chain Multi-elimination Preconditioner for Elliptic PDE Problems on GPU

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/61181

Title:	A Markov Chain Multi-elimination Preconditioner for Elliptic PDE Problems on GPU
Authors:	簡宇則;Chien,Yu-tse
Contributors:	數學系
Keywords:	隨機漫步
Date:	2013-07-26
Issue Date:	2013-08-22 12:14:11 (UTC+8)
Publisher:	國立中央大學
Abstract:	本文討論解 Poisson's equation 的數值解在有界的定義域上，透過隨機漫步的方法，造出一種新的預處理 (precondition) 並且使用圖形處理器來加速此方程的數值解，而疊代法使用 Conjugate Gradient 方法並與現有的預處理方法做比較的有 ILU0，multigrid 在 CPU 上，其它還有對此新的預理討論收斂速度，以及給出一些改良策略。 This paper discuss to solve numerical solution of Poisson's equation on finite domain and create a new precondition with random walk method and use graphic process unit to accelerate numerical solution of this equation. At iteration method is used by Conjugate Gradient method and compare with ILU0 and Multigrid method on CPU, other discuss this new precondition has convergence rate and give some improvement strategy.
Appears in Collections:	[Graduate Institute of Mathematics] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	572	View/Open

社群 sharing