Improving robustness in multiscale methods

An important constraint on our ability to simulate physical processes numerically is our ability to solve the resulting linear systems. Multiscale methods, such as multigrid, provide optimal or near-optimal order solution techniques for a wide range of problems. Current multigrid methods (both geometric and algebraic) rely on the use of effective coarse-scale operators to achieve this efficiency. Classical geometric multigrid methods construct these operators based on the geometry of a given problem and are, thus, most effective when this geometry is known and simple. Algebraic multigrid methods are free from most geometric constraints, instead relying on assumptions about the character of the matrix that limit the applicability of such methods to a relatively small class of problems. In this thesis, we first study the applicability of multigrid methods to the problem of flow through porous media, particularly motivated by the size of the linear systems resulting from discretization. For such problems, it is well known how to construct effective coarse-scale models for a correction-based multigrid method. For these robust multigrid solvers, we consider whether sufficient information is contained in the multigrid hierarchy to allow for efficient approximation of solutions of these linear systems. In particular, we examine important properties of the coarse-scale operators and the solutions to these coarse-scale equations and demonstrate their relevance to both the fine-scale and continuum-scale models. Additionally, we consider the question of improvements to multigrid algorithms in the form of adaptivity or self-correction. By utilizing the multigrid method itself to expose slow-to-converge components, we may adapt the multigrid hierarchy in order to improve performance, resulting in improved efficiency for a new class of problems. We present results for both the classical algebraic multigrid and smoothed aggregation frameworks illustrating typical multigrid efficiency with fewer assumptions on the given matrix systems. We consider theoretical issues of this adaptive procedure in the reduction-based algebraic multigrid setting.

[1]  L. Durlofsky Numerical calculation of equivalent grid block permeability tensors for heterogeneous porous media , 1991 .

[2]  P. King The use of renormalization for calculating effective permeability , 1989 .

[3]  J. Pasciak,et al.  Convergence estimates for product iterative methods with applications to domain decomposition , 1991 .

[4]  J. Hyman,et al.  The Black Box Multigrid Numerical Homogenization Algorithm , 1998 .

[5]  A. Brandt Multi-level adaptive technique (MLAT) for fast numerical solution to boundary value problems , 1973 .

[6]  A. Gilbert A Comparison of Multiresolution and Classical One-Dimensional Homogenization Schemes , 1998 .

[7]  Thomas A. Manteuffel,et al.  Algebraic Multigrid Based on Element Interpolation (AMGe) , 2000, SIAM J. Sci. Comput..

[8]  I. Gustafsson A class of first order factorization methods , 1978 .

[9]  L. Durlofsky Accuracy of mixed and control volume finite element approximations to Darcy velocity and related quantities , 1994 .

[10]  Thomas A. Manteuffel,et al.  Adaptive Smoothed Aggregation (αSA) , 2004, SIAM J. Sci. Comput..

[11]  J. Bourgat Numerical experiments of the homogenization method , 1979 .

[12]  S. V. Ahamed,et al.  Accelerated Convergence of Numerical Solution of Linear and Non-Linear Vector Field Problems , 1965, Comput. J..

[13]  J. Pasciak,et al.  New convergence estimates for multigrid algorithms , 1987 .

[14]  J. Dvo A Reliable Numerical Method for Computing Homogenized Coeecients , 1995 .

[15]  A. Brandt Multi-Level Adaptive Techniques (MLAT) for Partial Differential Equations: Ideas and Software , 1977 .

[16]  J. E. Dendy,et al.  Black box multigrid for systems , 1986 .

[17]  V. Zhikov,et al.  Homogenization of Differential Operators and Integral Functionals , 1994 .

[18]  G. Beylkin,et al.  A Multiresolution Strategy for Reduction of Elliptic PDEs and Eigenvalue Problems , 1998 .

[19]  Steve McCormick,et al.  Algebraic multigrid methods applied to problems in computational structural mechanics , 1989 .

[20]  J. Ruge,et al.  Efficient solution of finite difference and finite element equations by algebraic multigrid (AMG) , 1984 .

[21]  David Moulton,et al.  Multilevel Upscaling in Heterogeneous Porous Media , 2004 .

[22]  H. S. Price,et al.  Flow in Heterogeneous Porous Media , 1961 .

[23]  M. Hestenes,et al.  Methods of conjugate gradients for solving linear systems , 1952 .

[24]  D. Brandt,et al.  Multi-level adaptive solutions to boundary-value problems math comptr , 1977 .

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

[26]  J. E. Dendy Black box multigrid for periodic and singular problems , 1988 .

[27]  Thomas Y. Hou,et al.  A Multiscale Finite Element Method for Elliptic Problems in Composite Materials and Porous Media , 1997 .

[28]  G. TEMPLE,et al.  Relaxation Methods in Engineering Science , 1942, Nature.

[29]  Stephan Knapek,et al.  Matrix-Dependent Multigrid Homogenization for Diffusion Problems , 1998, SIAM J. Sci. Comput..

[30]  Clayton V. Deutsch,et al.  GSLIB: Geostatistical Software Library and User's Guide , 1993 .

[31]  Panayot S. Vassilevski,et al.  Spectral AMGe (ρAMGe) , 2003, SIAM J. Sci. Comput..

[32]  J. Gómez-Hernández,et al.  Upscaling hydraulic conductivities in heterogeneous media: An overview , 1996 .

[33]  Jacob Fish,et al.  Generalized Aggregation Multilevel solver , 1997 .

[34]  W. T. Cardwell,et al.  Average Permeabilities of Heterogeneous Oil Sands , 1945 .

[35]  James Demmel,et al.  Applied Numerical Linear Algebra , 1997 .

[36]  Van Emden Henson,et al.  Robustness and Scalability of Algebraic Multigrid , 1999, SIAM J. Sci. Comput..

[37]  Anne Greenbaum,et al.  Iterative methods for solving linear systems , 1997, Frontiers in applied mathematics.

[38]  Petr Vaněk,et al.  A modification of the two-level algorithm with overcorrection , 1992 .

[39]  Eduard Stiefel,et al.  Über einige Methoden der Relaxationsrechnung , 1952 .

[40]  D. Ron,et al.  Multigrid Solvers and Multilevel Optimization Strategies , 2003 .

[41]  Petr Vaněk Acceleration of convergence of a two-level algorithm by smoothing transfer operators , 1992 .

[42]  R. P. Fedorenko A relaxation method for solving elliptic difference equations , 1962 .

[43]  Daniel M. Tartakovsky,et al.  Effective Hydraulic Conductivity of Bounded, Strongly Heterogeneous Porous Media , 1996 .

[44]  Louis J. Durlofsky,et al.  Numerical Calculation of Equivalent Cell Permeability Tensors for General Quadrilateral Control Volumes , 2002 .

[45]  James T. Smith Iterative Solution of Systems of Equations , 1999 .

[46]  Robert D. Falgout,et al.  hypre: A Library of High Performance Preconditioners , 2002, International Conference on Computational Science.

[47]  Jinchao Xu,et al.  Convergence estimates for multigrid algorithms without regularity assumptions , 1991 .

[48]  N. Bakhvalov On the convergence of a relaxation method with natural constraints on the elliptic operator , 1966 .

[49]  S. F. McCormick,et al.  Multigrid Methods for Variational Problems , 1982 .

[50]  E. C. Childs Dynamics of fluids in Porous Media , 1973 .

[51]  P. Vassilevski,et al.  ON GENERALIZING THE AMG FRAMEWORK , 2003 .

[52]  A. Brandt Algebraic multigrid theory: The symmetric case , 1986 .

[53]  A. Bensoussan,et al.  Asymptotic analysis for periodic structures , 1979 .

[54]  Petr Vaněk,et al.  Acceleration of convergence of a two-level algebraic algorithm by aggregation in smoothing process , 1992 .

[55]  Marian Brezina,et al.  Two-grid Method for Linear Elasticity on Unstructured Meshes , 1999, SIAM J. Sci. Comput..

[56]  Thomas Y. Hou,et al.  Convergence of a multiscale finite element method for elliptic problems with rapidly oscillating coefficients , 1999, Math. Comput..

[57]  R. P. Fedorenko The speed of convergence of one iterative process , 1964 .

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

[59]  R D Falgout,et al.  A Note on the Relationship Between Adaptive AMG and PCG , 2004 .

[60]  J. Mandel,et al.  An Iterative Method with Convergence Rate Chosen a priori , 1999 .

[61]  Edmond Chow,et al.  An Aggregation Multilevel Method Using Smooth Error Vectors , 2005, SIAM J. Sci. Comput..

[62]  Marian Brezina,et al.  Convergence of algebraic multigrid based on smoothed aggregation , 1998, Numerische Mathematik.

[63]  Gregory Beylkin,et al.  LU Factorization of Non-standard Forms and Direct Multiresolution Solvers , 1998 .

[64]  V. E. Henson,et al.  BoomerAMG: a parallel algebraic multigrid solver and preconditioner , 2002 .

[65]  William P. Timlake,et al.  An Accelerated Relaxation Algorithm for Iterative Solution of Elliptic Equations , 1968 .

[66]  Marian Brezina,et al.  Algebraic multigrid by smoothed aggregation for second and fourth order elliptic problems , 2005, Computing.

[67]  J. E. Dendy Black box multigrid for nonsymmetric problems , 1983 .

[68]  A. Brandt,et al.  The Multi-Grid Method for the Diffusion Equation with Strongly Discontinuous Coefficients , 1981 .

[69]  O. E. Livne,et al.  Coarsening by compatible relaxation , 2004, Numer. Linear Algebra Appl..

[70]  Richard Vynne Southwell,et al.  Stress-calculation in frameworks by the method of "systematic relaxation of constraints"—I and II , 1935, Proceedings of the Royal Society of London. Series A - Mathematical and Physical Sciences.

[71]  J. Dendy Black box multigrid , 1982 .