Convergence and optimization of the parallel method of simultaneous directions for the solution of elliptic problems

For the solution of elliptic problems, fractional step methods and in particular alternating directions (ADI) methods are iterative methods where fractional steps are sequential. Therefore, they only accept parallelization at low level. In [T. Lu, P. Neittaanmaki, X.C. Tai, A parallel splitting-up method for partial differential equations and its applications to Navier-Stokes equations, RAIRO Model. Math. Anal. Numer. 26 (6) (1992) 673-708], Lu et al. proposed a method where the fractional steps can be performed in parallel. We can thus speak of parallel fractional step (PFS) methods and, in particular, simultaneous directions (SDI) methods. In this paper, we perform a detailed analysis of the convergence and optimization of PFS and SDI methods, complementing what was done in [T. Lu, P. Neittaanmaki, X.C. Tai, A parallel splitting-up method for partial differential equations and its applications to Navier-Stokes equations, RAIRO Model. Math. Anal. Numer. 26 (6) (1992) 673-708]. We describe the behavior of the method and we specify the good choice of the parameters. We also study the efficiency of the parallelization. Some 2D, 3D and high-dimensional tests confirm our results.

[1]  A. Iserles.  A First Course in the Numerical Analysis of Differential Equations: Stiff equations , 2008 .

[2]  Jim Douglas,et al.  Alternating Direction Methods for Parabolic Systems in m Space Variables , 1962, JACM.

[3]  H. H. Rachford,et al.  On the numerical solution of heat conduction problems in two and three space variables , 1956 .

[4]  J. Gillis,et al.  Matrix Iterative Analysis , 1961 .

[5]  Xue-Cheng Tai,et al.  Parallel finite element splitting‐up method for parabolic problems , 1991 .

[6]  Max A. Viergever,et al.  Efficient and reliable schemes for nonlinear diffusion filtering , 1998, IEEE Trans. Image Process..

[7]  H. H. Rachford,et al.  The Numerical Solution of Parabolic and Elliptic Differential Equations , 1955 .

[8]  Enrique Fernández-Cara,et al.  Simultaneous directions parallel methods for elliptic and parabolic systems , 2004 .

[9]  J. Weickert,et al.  Fast Methods for Implicit Active Contour Models , 2003 .

[10]  Enrique Fernández-Cara,et al.  Stability and convergence of a parallel fractional step method forthe solution of linear parabolic problems , 2005 .

[11]  Enrique Fernández-Cara,et al.  Convergence analysis and error estimates for a parallel algorithm for solving the Navier-Stokes equations , 2002, Numerische Mathematik.

[12]  Enrique Fernández Cara,et al.  Time and space parallelization of the navier-stokes equations , 2005 .

[13]  Enrique Fernández-Cara,et al.  A simultaneous directions parallel algorithm for the Navier–Stokes equations , 2004 .

[14]  Louis A. Hageman,et al.  Iterative Solution of Large Linear Systems. , 1971 .

[15]  S. Lennart Johnsson,et al.  Alternating direction methods on multiprocessors , 1987 .

[16]  William L. Briggs,et al.  A multigrid tutorial , 1987 .

[17]  Danny Barash,et al.  Nonlinear diffusion filtering on extended neighborhood , 2005 .

[18]  Xuecheng Tai,et al.  A parallel splitting-up method for partial differential equations and its applications to Navier-Stokes equations , 1992 .

[19]  William L. Briggs,et al.  A multigrid tutorial, Second Edition , 2000 .

[20]  Enrique Fernández-Cara,et al.  The smoothing effect of a simultaneous directions parallel method as applied to Poisson problems , 2006 .

[21]  S. Osher,et al.  Geometric Level Set Methods in Imaging, Vision, and Graphics , 2011, Springer New York.

[22]  R. Pavanello,et al.  Numerical methods for the dynamics of unbounded domains , 2005 .

[23]  A. R. Mitchell,et al.  The Finite Difference Method in Partial Differential Equations , 1980 .

[24]  G. Marchuk.  Splitting and alternating direction methods , 1990 .

[25]  G. Marchuk.  Methods of Numerical Mathematics , 1982 .

[26]  Richard S. Varga,et al.  Matrix Iterative Analysis , 2000, The Mathematical Gazette.