DOCSLIB.ORG

Lecture Notes in Computational Science and Engineering 76

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Lecture Notes in Computational Science and Engineering 76 Lecture Notes in Computational Science and Engineering 76 Editors: Timothy J. Barth Michael Griebel David E. Keyes Risto M. Nieminen Dirk Roose Tamar Schlick For further volumes: http://www.springer.com/series/3527 • Jan S. Hesthaven • Einar M. Rønquist Editors Spectral and High Order Methods for Partial Differential Equations Selected papers from the ICOSAHOM ’09 conference, June 22-26, Trondheim, Norway 123 Editors Jan S. Hesthaven Einar M. Rønquist Brown University Norwegian University of Science Division of Applied Mathematics and Technology 182 George Street Department of Mathematical Sciences Providence, RI 02912 7491 Trondheim USA Norway [email protected] [email protected] ISBN 978-3-642-15336-5 e-ISBN 978-3-642-15337-2 DOI 10.1007/978-3-642-15337-2 Springer Heidelberg Dordrecht London New York Mathematics Subject Classification (2010): 65M70, 65N35, 65N30, 74S25, 76M10, 76M22, 78M10, 78M22 c Springer-Verlag Berlin Heidelberg 2011 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer.Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Cover illustration: The nudg++ team - Tim Warburton (Rice University), Nigel Nunn, Nico Gödel (Helmut-Schmidt-University, University of the Federal Armed Forces Hamburg) Cover design: deblik, Berlin Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com) Foreword This volume presents selected papers from the eigth ICOSAHOM (International Conference On Spectral and High Order Methods) conference which was held at the Norwegian University of Science and Technology (NTNU), Trondheim, Norway, during the week June 22–26, 2009. These papers were refereed by members of the scientific committee of ICOSAHOM as well as by other leading scientists. The first ICOSAHOM conference was held in Como, Italy, in 1989. At that point the importance of high order methods was recognized and it was deciced to organize a series of conferences to be held every 3 years (Montpelier, France, 1992; Houston, TX, USA, 1995; Tel Aviv, Israel, 1998; Uppsala, Sweden, 2001; Providence, RI, USA, 2004; and Beijing, China, 2007). At the conference in Beijing in 2007, it was decided to organize the conferences every other year. This decision was made partly due to the growing importance and the growing activity in this field. From the interest seen at ICOSAHOM 2009, this seems to have been an appropriate decision. The number of registered participants was over 200, while the total number of talks was 215, comprising nine invited talks, 153 talks in 19 different topic-specific minisymposia, and 53 talks in various contributed sessions. The ICOSAHOM conferences remain the main meeting place for researchers with interest in the theoretical, applied and computational aspects of high order methods for the numerical solution of partial differential equations. The content of the proceedings is organized as follows. First, contributions from the invited speakers are included, listed in alphabetical order according to the invited speaker. Next, contributions from the speakers at all the minisymposia are included, listed in alphabetical order according to the first author of each paper. Finally, con- tributions from the speakers at the various contributed sessions are included, also listed in alphabetical order according to the the first author. As part of the conference, a special minisymposium was organized in memory of David Gottlieb who passed away in December 2008, and who left an indelible mark on the field of applied mathematics in general and spectral methods in particular. The success of the meeting was ensured through the generous financial support by the Research Council of Norway, the National Science Foundation (NSF), the Norwegian University of Science and Technology (NTNU – through the Faculty of Information Technology, Mathematics and Electrical Engineering and through v vi Foreword the Program for Computational Science and Visualization), and Simula Research Laboratory (through Center for Biomedical Computing). Finally, the conference could not have happened without the invaluable support and assistance of conference coordinator Anne Kajander. Special thanks also go to Tormod Bjøntegaard for all his contributions. The assistance from the members of the Numerical Analysis Group and from the graduate students at the Department of Mathematical Sciences at NTNU is also gratefully acknowledged. Jan S. Hesthaven Einar M. Rønquist Contents hp-FEM for the Contact Problem with Tresca Friction in Linear Elasticity: The Primal Formulation .................................. 1 P. Dorsek ¨ and J.M. Melenk On Multivariate Chebyshev Polynomials and Spectral Approximations on Triangles ..................................................... 19 Brett N. Ryland and Hans Z. Munthe-Kaas Stochastic Spectral Galerkin and Collocation Methods for PDEs with Random Coefficients: A Numerical Comparison ............. 43 Joakim B¨ack, Fabio Nobile, Lorenzo Tamellini, and Raul Tempone Hybridizable Discontinuous Galerkin Methods ................................ 63 N.C. Nguyen, J. Peraire, and B. Cockburn Multivariate Modified Fourier Expansions ..................................... 85 Ben Adcock and Daan Huybrechs Constraint Oriented Spectral Element Method ................................ 93 E. Ahusborde, M. Aza¨ıez, and R. Gruber Convergence Rates of Sparse Tensor GPC FEM for Elliptic sPDEs ................................................................................101 Marcel Bieri, Roman Andreev, and Christoph Schwab A Conservative Spectral Element Method for Curvilinear Domains ..........................................................111 Mick Bouman, Artur Palha, Jasper Kreeft, and Marc Gerritsma An Efficient Control Variate Method for Parametrized Expectations ...................................................121 S´ebastien Boyaval vii viii Contents A Proof, Based on the Euler Sum Acceleration, of the Recovery of an Exponential (Geometric) Rate of Convergence for the Fourier Series of a Function with Gibbs Phenomenon ................131 John P. Boyd A Seamless Reduced Basis Element Method for 2D Maxwell’s Problem: An Introduction.........................................................141 Yanlai Chen, Jan S. Hesthaven, and Yvon Maday An hp-Nitsche’s Method for Interface Problems with Nonconforming Unstructured Finite Element Meshes.........................153 Alexey Chernov and Peter Hansbo Hybrid Explicit–Implicit Time Integration for Grid- Induced Stiffness in a DGTD Method for Time Domain Electromagnetics ...................................................................163 Victorita Dolean, Hassan Fahs, Loula Fezoui, and St´ephane Lanteri High-Order Quasi-Uniform Approximation on the Sphere Using Fourier-Finite-Elements ...................................................171 T. Dubos An hp Certified Reduced Basis Method for Parametrized Parabolic Partial Differential Equations ........................................179 Jens L. Eftang, Anthony T. Patera, and Einar M. Rønquist Highly Accurate Discretization of the Navier–Stokes Equations in Streamfunction Formulation...................................................189 D. Fishelov, M. Ben-Artzi, and J.-P. Croisille Edge Functions for Spectral Element Methods .................................199 Marc Gerritsma Modeling Effects of Electromagnetic Waves on Thin Wires with a High-Order Discontinuous Galerkin Method...........................209 N. G¨odel, T. Warburton, and M. Clemens A Hybrid Method for the Resolution of the Gibbs Phenomenon .............219 Jae-Hun Jung Numerical Simulation of Fluid–Structure Interaction in Human Phonation: Verification of Structure Part ..........................229 Martin Larsson and Bernhard M¨uller Contents ix A New Spectral Method on Triangles ............................................237 Youyun Li, Li-Lian Wang, Huiyuan Li, and Heping Ma The Reduced Basis Element Method: Offline-Online Decomposition in the Nonconforming, Nonaffine Case ........................247 A.E. Løvgren, Y. Maday, and E.M. Rønquist The Challenges of High Order Methods in Numerical Weather Prediction ...........................................................................255 Catherine Mavriplis GMRES for Oscillatory Matrix-Valued Differential Equations ..............267 Sheehan Olver Sensitivity Analysis of Heat Exchangers Using Perturbative Methods .............................................................................275 J.C. Pacio, C.A. Dorao, and M. Fernandino Spectral Element Approximation of the Hodge-? Operator in Curved Elements ................................................................283 Artur Palha and Marc Gerritsma Uncertainty Propagation for Systems of Conservation Laws, High Order Stochastic Spectral Methods .......................................293 G. Po¨ette, B. Despr´es, and
Load more

Recommended publications
  • Efficient Spectral-Galerkin Method I. Direct Solvers for the Second And
    SIAM J. SCI. COMPUT. °c 1994 Society for Industrial and Applied Mathematics Vol. 15, No. 6, pp. 1489{1505, November 1994 013 E±cient Spectral-Galerkin Method I. Direct Solvers for the Second and Fourth Order Equations Using Legendre Polynomials¤ Jie Sheny Abstract. We present some e±cient algorithms based on the Legendre-Galerkin approxima- tions for the direct solution of the second and fourth order elliptic equations. The key to the e±ciency of our algorithms is to construct appropriate base functions, which lead to systems with sparse matri- ces for the discrete variational formulations. The complexities of the algorithms are a small multiple of N d+1 operations for a d dimensional domain with (N ¡ 1)d unknowns, while the convergence rates of the algorithms are exponential for problems with smooth solutions. In addition, the algorithms can be e®ectively parallelized since the bottlenecks of the algorithms are matrix-matrix multiplications. Key words. spectral-Galerkin method, Legendre polynomial, Helmholtz equation, biharmonic equation, direct solver AMS subject classi¯cations. 65N35, 65N22, 65F05, 35J05 1. Introduction. This article is the ¯rst in a series for developing e±cient spectral Galerkin algorithms for elliptic problems. The spectral method employs global polynomials as the trial functions for the discretization of partial di®erential equations. It provides very accurate approximations with a relatively small number of unknowns. Consequently it has gained increasing popularity in the last two decades, especially in the ¯eld of computational fluid dynamics (see [11], [8] and the references therein). The use of di®erent test functions in a variational formulation leads to three most commonly used spectral schemes, namely, the Galerkin, tau and collocation versions.
    [Show full text]
  • Family Name Given Name Presentation Title Session Code
    Family Name Given Name Presentation Title Session Code Abdoulaev Gassan Solving Optical Tomography Problem Using PDE-Constrained Optimization Method Poster P Acebron Juan Domain Decomposition Solution of Elliptic Boundary Value Problems via Monte Carlo and Quasi-Monte Carlo Methods Formulations2 C10 Adams Mark Ultrascalable Algebraic Multigrid Methods with Applications to Whole Bone Micro-Mechanics Problems Multigrid C7 Aitbayev Rakhim Convergence Analysis and Multilevel Preconditioners for a Quadrature Galerkin Approximation of a Biharmonic Problem Fourth-order & ElasticityC8 Anthonissen Martijn Convergence Analysis of the Local Defect Correction Method for 2D Convection-diffusion Equations Flows C3 Bacuta Constantin Partition of Unity Method on Nonmatching Grids for the Stokes Equations Applications1 C9 Bal Guillaume Some Convergence Results for the Parareal Algorithm Space-Time ParallelM5 Bank Randolph A Domain Decomposition Solver for a Parallel Adaptive Meshing Paradigm Plenary I6 Barbateu Mikael Construction of the Balancing Domain Decomposition Preconditioner for Nonlinear Elastodynamic Problems Balancing & FETIC4 Bavestrello Henri On Two Extensions of the FETI-DP Method to Constrained Linear Problems FETI & Neumann-NeumannM7 Berninger Heiko On Nonlinear Domain Decomposition Methods for Jumping Nonlinearities Heterogeneities C2 Bertoluzza Silvia The Fully Discrete Fat Boundary Method: Optimal Error Estimates Formulations2 C10 Biros George A Survey of Multilevel and Domain Decomposition Preconditioners for Inverse Problems in Time-dependent
    [Show full text]
  • A Space-Time Discontinuous Galerkin Spectral Element Method for the Stefan Problem
    Noname manuscript No. (will be inserted by the editor) A Space-Time Discontinuous Galerkin Spectral Element Method for the Stefan problem Chaoxu Pei · Mark Sussman · M. Yousuff Hussaini Received: / Accepted: Abstract A space-time Discontinuous Galerkin (DG) spectral element method is presented to solve the Stefan problem in an Eulerian coordinate system. In this scheme, the level set method describes the time evolving interface. To deal with the prior unknown interface, two global transformations, a backward transformation and a forward transformation, are introduced in the space-time mesh. The idea of the transformations is to combine an Eulerian description, i.e. a fixed frame of reference, with a Lagrangian description, i.e. a moving frame of reference. The backward transformation maps the unknown time- varying interface in the fixed frame of reference to a known stationary inter- face in the moving frame of reference. In the moving frame of reference, the transformed governing equations, written in the space-time framework, are discretized by a DG spectral element method in each space-time slab. The forward transformation is used to update the level set function and then to project the solution in each phase onto the new corresponding time-dependent domain. Two options for calculating the interface velocity are presented, and both methods exhibit spectral accuracy. Benchmark tests in one spatial di- mension indicate that the method converges with spectral accuracy in both space and time for both the temperature distribution and the interface ve- locity. The interrelation between the interface position and the temperature makes the Stefan problem a nonlinear problem; a Picard iteration algorithm is introduced in order to solve the nonlinear algebraic system of equations and it is found that only a few iterations lead to convergence.
    [Show full text]
  • Spectral Element Simulations of Laminar and Turbulent Flows in Complex Geometries
    Applied Numerical Mathematics 6 (1989/90) 85-105 85 North-Holland SPECTRAL ELEMENT SIMULATIONS OF LAMINAR AND TURBULENT FLOWS IN COMPLEX GEOMETRIES George Em KARNIADAKIS* Center for Turbulence Research, Stanford University/NASA Ames Research Center, Stanford, CA 94305, USA Spectral element methods are high-order weighted residual techniques based on spectral expansions of variables and geometry for the Navier-Stokes and transport equations . Their success in the recent past in simulating flows of industrial complexity derives from the flexibility of the method to accurately represent nontrivial geometries while preserving the good resolution properties of the spectral methods . In this paper, we review some of the main ideas of the method with emphasis placed on implementation and data management . These issues need special attention in order to make the method efficient in practice, especially in view of the fact that high computing cost as well as strenuous storage requirements have been a major drawback of high-order methods in the past . Several unsteady, laminar complex flows are simulated, and a direct simulation of turbulent channel flow is presented, for the first time, using spectral element techniques . 1 . Introduction Spectral methods have proven in recent years a very powerful tool for analyzing fluid flows and have been used almost exclusively in direct simulations of transitional and turbulent flows [8]. To date, however, only simple geometry turbulent flows have been studied accurately [14], chiefly because of the poor performance of global spectral methods in representing more complex geometries. The development of the spectral element method [26] and its success to accurately simulate highly unsteady, two-dimensional laminar flows [6,13] suggests that a three-dimensional implementation may be the right approach to direct simulations of turbulent flows in truly complex geometries.
    [Show full text]
  • Hybrid Multigrid/Schwarz Algorithms for the Spectral Element Method
    Hybrid Multigrid/Schwarz Algorithms for the Spectral Element Method James W. Lottes¤ and Paul F. Fischery February 4, 2004 Abstract We study the performance of the multigrid method applied to spectral element (SE) discretizations of the Poisson and Helmholtz equations. Smoothers based on finite element (FE) discretizations, overlapping Schwarz methods, and point-Jacobi are con- sidered in conjunction with conjugate gradient and GMRES acceleration techniques. It is found that Schwarz methods based on restrictions of the originating SE matrices converge faster than FE-based methods and that weighting the Schwarz matrices by the inverse of the diagonal counting matrix is essential to effective Schwarz smoothing. Sev- eral of the methods considered achieve convergence rates comparable to those attained by classic multigrid on regular grids. 1 Introduction The availability of fast elliptic solvers is essential to many areas of scientific computing. For unstructured discretizations in three dimensions, iterative solvers are generally optimal from both work and storage standpoints. Ideally, one would like to have computational complexity that scales as O(n) for an n-point grid problem in lRd, implying that the it- eration count should be bounded as the mesh is refined. Modern iterative methods such as multigrid and Schwarz-based domain decomposition achieve bounded iteration counts through the introduction of multiple representations of the solution (or the residual) that allow efficient elimination of the error at each scale. The theory for these methods is well established for classical finite difference (FD) and finite element (FE) discretizations, and order-independent convergence rates are often attained in practice. For spectral element (SE) methods, there has been significant work on the development of Schwarz-based methods that employ a combination of local subdomain solves and sparse global solves to precondition conjugate gradient iteration.
    [Show full text]
  • SCHWARZ Rr'"YPE DOMAIN DECOMPOSITION ]\1ETHODS FOR
    SCHWARZ rr'"YPE DOMAIN DECOMPOSITION ]\1ETHODS FOR St<EC'l'RAL ELEMENT DISCRETIZATIONS Shannon S. Pahl University of the Witwatersrand December 1993 .() Degree awarded with distinction on 30 June 1994. , o A research report submitted to the Faculty of Science in partial fulfillment of the requirements 'L for the degree of Mastel' of Science at the University of the Witwatersrand, Johannesburg. () II " Abstract Most cf the theory for domain decomposition methods of Schwarz tyne has been set in the framework of the h ~',ndp~version finite element method. In this study, Schwarz methods are formulated for both the conforming and nonconforming spectral element discretizations applied to linear scalar self adjoint second order elliptic problems in two dimensions. An overlapping additive Schwarz method for the conforming spectral element discretization is formulated. However, unlike p~nnite element discretizations, a minimum overlap strategy com- mon to h-J11niteelement discret)zations can be accommodated. Computational results indicate if that the convergence rate of the overlapping Schwarz method is similar for the corresponding method defined for the h-finite element discretization. It is also shown that the minimum over- lap additive Schwarz method results in significantly fewer floating point operations than finite element preconditioning for the spectral element method. Iterative substructuring and Neumann-Neumann methods carealso formulated for the conforming spectral element discretization. These methods demonstrate greater robustness when increasing the degree p of the elements than the minimum overlap additive Schwarz method. Computa- tionally, convergencerates are also similar for the corresponding methods defined for the h-finite element discretization. In addition, an efficient interface preconditioner for iterative substruc- turing methods is employed which does not degrade the performance of the method for tY..e problems considered.
    [Show full text]
  • NONCONFORMING FORMULATIONS with SPECTRAL ELEMENT METHODS a Dissertation by CUNEYT SERT Submitted to the Office of Graduate Studi
    View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by Texas A&amp;M Repository NONCONFORMING FORMULATIONS WITH SPECTRAL ELEMENT METHODS A Dissertation by CUNEYT SERT Submitted to the Office of Graduate Studies of Texas A&M University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY August 2003 Major Subject: Mechanical Engineering NONCONFORMING FORMULATIONS WITH SPECTRAL ELEMENT METHODS A Dissertation by CUNEYT SERT Submitted to Texas A&M University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Approved as to style and content by: Ali Beskok (Chair of Committee) J. N. Reddy N. K. Anand (Member) (Member) Paul Cizmas Dennis O’Neal (Member) (Head of Department) August 2003 Major Subject: Mechanical Engineering iii ABSTRACT Nonconforming Formulations with Spectral Element Methods. (August 2003) Cuneyt Sert, B.S., Middle East Technical University; M.S., Middle East Technical University Chair of Advisory Committee: Dr. Ali Beskok A spectral element algorithm for solution of the incompressible Navier-Stokes and heat transfer equations is developed, with an emphasis on extending the classical conforming Galerkin formulations to nonconforming spectral elements. The new algorithm employs both the Constrained Approximation Method (CAM), and the Mortar Element Method (MEM) for p-and h-type nonconforming elements. Detailed descriptions, and formulation steps for both methods, as well as the performance com- parisons between CAM and MEM, are presented. This study fills an important gap in the literature by providing a detailed explanation for treatment of p-and h-type nonconforming interfaces. A comparative eigenvalue spectrum analysis of diffusion and convection operators is provided for CAM and MEM.
    [Show full text]
  • Fekete-Gauss Spectral Elements for Incompressible Navier-Stokes Flows: the Two-Dimensional Case 1 1, 1 Laura Lazar , Richard Pasquetti ∗ and Francesca Rapetti 1 Lab
    Commun. Comput. Phys. Vol. 13, No. 5, pp. 1309-1329 doi: 10.4208/cicp.180112.110612a May 2013 Fekete-Gauss Spectral Elements for Incompressible Navier-Stokes Flows: The Two-Dimensional Case 1 1, 1 Laura Lazar , Richard Pasquetti ∗ and Francesca Rapetti 1 Lab. J.A. Dieudonn´e, UMR 7351 CNRS UNS, Universit´ede Nice - Sophia Antipolis, 06108 Nice Cedex 02, France. Received 18 January 2012; Accepted (in revised version) 11 June 2012 Available online 8 October 2012 Abstract. Spectral element methods on simplicial meshes, say TSEM, show both the advantages of spectral and finite element methods, i.e., spectral accuracy and geomet- rical flexibility. We present a TSEM solver of the two-dimensional (2D) incompressible Navier-Stokes equations, with possible extension to the 3D case. It uses a projection method in time and piecewise polynomial basis functions of arbitrary degree in space. The so-called Fekete-Gauss TSEM is employed, i.e., Fekete (resp. Gauss) points of the triangle are used as interpolation (resp. quadrature) points. For the sake of consistency, isoparametric elements are used to approximate curved geometries. The resolution al- gorithm is based on an efficient Schur complement method, so that one only solves for the element boundary nodes. Moreover, the algebraic system is never assembled, therefore the number of degrees of freedom is not limiting. An accuracy study is car- ried out and results are provided for classical benchmarks: the driven cavity flow, the flow between eccentric cylinders and the flow past a cylinder. AMS subject classifications: 65M60, 65M70, 76D05 Key words: Spectral elements, simplicial meshes, Fekete-Gauss approximation, Navier-Stokes equations, projection methods, domain decomposition.
    [Show full text]
  • Additive Schwarz Preconditioners with Minimal Overlap For
    Commun. Comput. Phys. Vol. 13, No. 2, pp. 411-427 doi: 10.4208/cicp.080711.160212a February 2013 Additive Schwarz Preconditioners with Minimal Overlap for Triangular Spectral Elements Yuen-Yick Kwan∗ Center for Computational Science, Tulane University, New Orleans, LA 70118, USA. Received 8 July 2011; Accepted (in revised version) 16 February 2012 Communicated by Jie Shen Available online 28 June 2012 Abstract. The additive Schwarz preconditioner with minimal overlap is extended to triangular spectral elements (TSEM). The method is a generalization of the correspond- ing method in tensorial quadrilateral spectral elements (QSEM). The proposed precon- ditioners are based on partitioning the domain into overlapping subdomains, solving local problems on these subdomains and solving an additional coarse problem asso- ciated with the subdomain mesh. The results of numerical experiments show that the proposed preconditioner are robust with respect to the number of elements and are more efficient than the preconditioners with generous overlaps. AMS subject classifications: 65F08, 65N35 Key words: Spectral elements, additive Schwarz. 1 Introduction Spectral element method, which combines the flexibility of the low-order finite element methods and the high accuracy of the spectral methods, are popular in computational science and engineering. The efficiency of the spectral element method depends on the solution method employed to solve the resultant linear system. Since direct methods are infeasible when the number of elements get large, preconditioned iterative methods are usually used. Preconditioners can be based on an overlapping or non-overlapping domain decomposition [24, 28]. The former include the multiplicative Schwarz [18], ad- ditive Schwarz [9] and the restricted Schwarz preconditioners [6].
    [Show full text]
  • The Spectral Element Method (Sem): Formulation and Implementation of the Method for Engineering Seismology Problems
    th The 14 World Conference on Earthquake Engineering October 12-17, 2008, Beijing, China THE SPECTRAL ELEMENT METHOD (SEM): FORMULATION AND IMPLEMENTATION OF THE METHOD FOR ENGINEERING SEISMOLOGY PROBLEMS 1 2 Kristel C. Meza-Fajardo and Apostolos S. Papageorgiou 1Professor, Depto. de Ingenieria Civil,Universidad Nacional Autónoma de Honduras,Tegucigalpa, Honduras 2 Professor, Department of Civil Engineering, University of Patras, GR-26500 Patras, Greece Email: [email protected], [email protected] ABSTRACT Various numerical methods have been proposed and used to investigate wave propagation in realistic earth media. Recently an innovative numerical method, known as the Spectral Element Method (SEM), has been developed and used in connection with wave propagation problems in 3D elastic media (Komatitsch and Tromp, 1999). The SEM combines the flexibility of a finite element method with the accuracy of a spectral method and it is not cumbersome in dealing with non-flat free surface and spatially variable anelastic attenuation. The SEM is a highly accurate numerical method that has its origins in computational fluid dynamics. One uses a weak formulation of the equations of motion, which are solved on a mesh of hexahedral elements that is adapted to the free surface and to the main internal discontinuities of the model. The wavefield on the elements is discretized using high-degree Lagrange interpolants, and integration over an element is accomplished based upon the Gauss-Lobatto-Legendre integration rule. This combination of discretization and integration results in a diagonal mass matrix, which greatly simplifies the algorithm. The most important property of the SEM is that the mass matrix is exactly diagonal by construction, which drastically simplifies the implementation and reduces the computational cost because one can use an explicit time integration scheme without having to invert a linear system.
    [Show full text]
  • Fast Iterative Solver for Convection-Diffusion Systems with Spectral Elements P
    Fast Iterative Solver for Convection-Diffusion Systems with Spectral Elements P. Aaron Lott,1 Howard Elman2 1Mathematical and Computational Sciences Division, National Institute of Standards and Technology, Gaithersburg, Maryland 20899 2Department of Computer Science and Institute for Advanced Computer Studies, University of Maryland, College Park, Maryland 20742 Received 9 March 2009; accepted 12 July 2009 Published online 16 October 2009 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/num.20518 We introduce a solver and preconditioning technique based on Domain Decomposition and the Fast Diagonal- ization Method that can be applied to tensor product based discretizations of the steady convection–diffusion equation. The method is based on a Robin–Robin interface preconditioner coupled to a fast diagonalization solver which is used to efficiently eliminate the interior degrees of freedom and perform subsidiary subdo- main solves. Using a spectral element discretization, we first apply our technique to constant wind problems, and then propose a means for applying the technique as a preconditioner for variable wind problems. We demonstrate that iteration counts are mildly dependent on changes in mesh size and convection strength. © 2009 Wiley Periodicals, Inc.∗ Numer Methods Partial Differential Eq 27: 231–254, 2011 Keywords: convection-diffusion; domain decomposition; preconditioning; spectral element method I. INTRODUCTION Numerical simulation of fluid flow allows for improved prediction and design of natural and engi- neered systems such as those involving water, oil, or blood. The interplay between inertial and viscous forces in a fluid flow dictates the length scale where energy is transferred, thus determin- ing the resolution required to capture flow information accurately.
    [Show full text]
  • Hybrid Schwarz-Multigrid Methods for the Spectral Element Method: Extensions to Navier-Stokes
    Hybrid Schwarz-Multigrid Methods for the Spectral Element Method: Extensions to Navier-Stokes Paul F. Fischer1 and James W. Lottes2 1 Argonne National Laboratory, Mathematics and Computer Science Division (http://www.mcs.anl.gov/~fischer/) 2 University of Illinois, Dept. of Theoretical and Applied Mechanics Summary. We study the performance of the multigrid method applied to spectral element (SE) discretizations of the Poisson and Helmholtz equations. Smoothers based on finite element (FE) discretizations, overlapping Schwarz methods, and point-Jacobi are considered in conjunction with conjugate gradient and GMRES acceleration techniques. It is found that Schwarz methods based on restrictions of the originating SE matrices converge faster than FE-based methods and that weighting the Schwarz matrices by the inverse of the diagonal counting matrix is essential to effective Schwarz smoothing. Several of the methods considered achieve convergence rates comparable to those attained by classic multigrid on regular grids. 1 Introduction The spectral element method (SEM) is a high-order weighted residual tech- nique that combines the geometric flexibility of finite elements with the rapid convergence properties and tensor-product efficiencies of global spectral meth- ods. Globally, elements are coupled in an unstructured framework with in- terelement coupling enforced through standard matching of nodal interface values. Locally, functions are represented as tensor products of stable Nth- order Lagrangian interpolants based on Gauss-Lobatto (GL) or Gauss (G) quadrature points. For problems having smooth solutions, such as the incom- pressible Navier-Stokes equations, the SEM converges exponentially fast with the local approximation order N. Because of its minimal numerical dissipa- tion and dispersion, the SEM is particularly well suited for the simulation of flows at transitional Reynolds numbers, where physical dissipation is small and turbulence-model dissipation is absent.
    [Show full text]