Exponential integration of large systems of ODEs

Size: px
Start display at page:

Download "Exponential integration of large systems of ODEs"

Transcription

1 Exponential integration of large systems of ODEs Jitse Niesen (University of Leeds) in collaboration with Will Wright (Melbourne University) 23rd Biennial Conference on Numerical Analysis, June 2009

2 Plan Aim: To solve the ODE u = Lu + N(u) where L is a big matrix and the stiffness is concentrated in L. Motivation: Such equations arise when semi-discretizing semi-linear PDEs. Method: Some exponential integrator with Krylov methods.

3 Outline Exponential integrators Definition of the ϕ-functions Connection ϕ-functions and ODEs Exponential Euler method An exponential predictor-corrector method Computing ϕ-function with matrix argument Reducing to the exponential Krylov methods Timestepping Numerical examples Uni Florida sparse matrices Heston PDE from finance (linear) Allen Cahn PDE in 2d (nonlinear) Conclusions

4 The solution of inhomogeneous linear DEs The solution of u = au, u(0) = u 0 is u(t) = e at u 0 = ϕ 0 (at)u 0. The solution of u = au + b, u(0) = 0 is u(t) = t 0 e as b ds = 1 a eas b where ϕ 1 (x) = (e x 1)/x. t s=0 = eat 1 b = tϕ 1 (at)b, a The solution of u = au + ct, u(0) = 0 is u(t) = t 2 ϕ 2 (at)c, where ϕ 2 (x) = (e x 1 x)/x 2. These solutions are also valid for vector ODEs.

5 The solution of inhomogeneous linear DEs II The solution of u (t) = Lu(t) + v ! tv ! t2 v 2 +, u(0) = u 0, where u(t), u 0, v 0, v 1,... are vectors and L is a matrix, is u(t) = ϕ 0 (tl)u 0 + tϕ 1 (tl)v 0 + t 2 ϕ 2 (tl)v 1 + t 3 ϕ 3 (tl)v 2 + where the ϕ-functions are ϕ 0 (x) = e x = 1 + x x ! x 3 + ϕ 1 (x) = ex 1 x = x + 1 3! x ! x 3 + ϕ 2 (x) = ex 1 x x 2 = ! x + 1 4! x ! x 3 + ϕ 3 (x) = ex 1 x 1 2 x 2 x 3 = 1 3! + 1 4! x + 1 5! x ! x 3 +

6 Exponential Euler method Back to u = Lu + N(u) (L = linear, N = nonlinear) Replace the nonlinear term with the constant N(u(0)) N(u(t)) (for small t) and use the results from the previous slide: u(t) = e tl u(0) + tϕ 1 (tl) N(u(0)). This leads to the exponential Euler method u n+1 = e hl u n + hϕ 1 (hl)n(u n ). This method has order 1 (like the normal Euler method). If DN(u n ) = 0 then it has order 2. (Certaine 1960) Exponential integrators are exact for linear equations. Thus, they can solve stiff equations if the stiffness is in the linear part.

7 An exponential predictor corrector method Higher-order exponential integrators are built from higher-order normal integrators with the higher ϕ-functions. We like exponential integrators based on AB/AM predictor corrector methods. E.g., for order 2, base method: u n+1 = u n hf (u n) 1 2 hf (u n 1) u n+1 = u n h( f (u n+1) 2f (u n ) + f (u n 1 ) ) Exponential integrator: u n+1 = e hl u n + hϕ 1 (hl)n(u n ) + h 2 ϕ 2 (hl) ( N(u n ) N(u n 1 ) ) u n+1 = u n+1 + h 2 ϕ 2 (hl) ( N(u n+1) 2N(u n ) + N(u n 1 ) ) Other exponential methods also exist (without corrector step, based on Runge Kutta, using exact Jacobian,... ). All need to evaluate ϕ-function of matrix times vector.

8 Outline Exponential integrators Definition of the ϕ-functions Connection ϕ-functions and ODEs Exponential Euler method An exponential predictor-corrector method Computing ϕ-function with matrix argument Reducing to the exponential Krylov methods Timestepping Numerical examples Uni Florida sparse matrices Heston PDE from finance (linear) Allen Cahn PDE in 2d (nonlinear) Conclusions

9 Evaluating the ϕ-functions The biggest problem for EIs is the evaluation of (linear combinations of) ϕ-functions with a matrix argument. Sometimes (e.g., spectral methods) you can diagonalize the matrix easily and use L = V 1 DV = ϕ k (hl) = hv 1 ϕ k (D)V. General case: Reduce problem to evaluation of matrix exponential (this step enlarges the matrix slightly, by two rows and columns): A v 0 e A ϕ 1 (A)v ϕ 2 (A)v exp = (Saad 1992; Sidje 1998) Now use well-known method to compute matrix exponential (scaling-and-squaring with Padé).

10 Krylov methods Padé approximation with squaring and scaling is too costly for large matrices (N 10 3 ). Idea of Krylov-subspace methods is to project the action of A on the Krylov subspace with dimension m N: span{v, Av, A 2 v,..., A m 1 v}. The Lanczos (if A symmetric) or Arnoldi algorithm produces an orthogonal N-by-m matrix V m and an m-by-m matrix H m such that A V m H m V T m = ϕ k (A) V m ϕ k (H m )V T m = ϕ k (A)v v V m ϕ k (H m )e 1 H m is small so we can compute ϕ k (H m )e 1 via Padé. (Friesner, Tuckerman & Dornblaser 1989)

11 Dimension of Krylov subspace The error in the Krylov approximation may be estimated with error v h m+1,m e T mϕ k+1 (H m )e 1 v m+1. This can be computed very cheaply. (Saad 1992; Sidje 1998) We use this error estimate to choose the dimension m of the Krylov subspace adaptively. That is, we compute the error estimate and if it is too big, we increase m and try again. (Hochbruck, Lubich & Selhofer 1998) We also use the error estimate to correct our initial approximation.

12 Timestepping The error in the Krylov approximation is bounded by error 2 v (ρ(a))m (1 + o(1)) m! for m large, where ρ(a) is the spectral radius. (Saad 1992) This suggest that m has to be large if ρ(a) is large. An alternative is to use timestepping. Remember that the ϕ-functions solve ODEs: The solution of u = Au + tv, u(0) = 0 is u(t) = t 2 ϕ 2 (ta)v. To compute ϕ 2 (A)v, we can split the interval [0, 1] up in several sub-intervals and solve the ODE on each sub-interval by computing a ϕ-function. (Sidje 1998; Sofroniou & Spaletta 2007) The length of the sub-intervals is chosen adaptively with standard techniques for ODE solvers.

13 Outline Exponential integrators Definition of the ϕ-functions Connection ϕ-functions and ODEs Exponential Euler method An exponential predictor-corrector method Computing ϕ-function with matrix argument Reducing to the exponential Krylov methods Timestepping Numerical examples Uni Florida sparse matrices Heston PDE from finance (linear) Allen Cahn PDE in 2d (nonlinear) Conclusions

14 Uni Florida sparse matrices We compare our routine with expokit, using the same experiments as in Sidge (1998). Compute ϕ 1 (A)v for: orani678: N = 2529, nnz = 90158, TOL = bcspwr10: N = 5300, nnz = 21842, TOL = 10 5, symm. gr 30 30: N = 900, nnz = 7744, TOL = 10 14, symm. helm2d03: N 400k, nnz 2.7M, TOL = 10 8, symm. Estimate error by repeating computation with tight tolerance; for gr 30 30, by integrating forward and backward. expokit our code time error time error orani bcspwr gr helm2d

15 Heston PDE The Heston PDE is used in mathematical finance to prize European call options with stochastic volatility: u t = 1 2 xy 2 u yy + ρλxyu xy λ2 xu xx + ryu y + κ(η x)u x ru. Use standard second-order finite differences and incorporate boundary conditions to get ODE of the form u = Au + v 1, u(0) = v 0. The matrix A has size 5100 and non-zero elements. The ODE can be solved by evaluating ϕ-functions. In t Hout (2007) advocates the use of ADI. We compare his ADI schemes with the standard matlab routine ode15s and our code.

16 Error vs time maximum error Phipm Crank Nicolson Douglas Craig Hundsdorfer Ode15s cpu time

17 Krylov dimension m and step size τ 55 m x τ t

18 Error estimate (blue) and actual error (red) 9 x error t

19 Allen Cahn in two dimensions u t = α 2 u + u u 3, x, y [0, 1], t [0, 1] Finite differences, Neumann BCs, α = 0.1, grid D Allen Cahn α = 0.1 odeexp5 oderos4 ode15s radau5 exp4 global error cpu time Implicit: ode15s (Matlab), radau5 (Hairer & Wanner) Exponential: odeexp5 (us), oderos4 (Caliari & Ostermann), exp4 (Hochbruck, Lubich & Selhofer)

20 Conclusion EIs are meant for ODEs with stiffness concentrated in the linear part, e.g., semi-discretized semi-linear PDEs. EIs work very well with spectral discretization. Otherwise need to calculate ϕ(a)v challenging if A large. EIs with Krylov subspace methods are competitive with state-of-the-art for mildly stiff problems. Performance degrades for very stiff problems. Current work / future plans: How to combine with finite elements Mu = Lu + N(u)? Krylov subspace may not be best, so look at: Leja point interpolation (Caliari & Ostermann) Rational Krylov (Grimm & Hochbruck) RD-rational approximations (Moret & Novati) Using contour integrals (Schmelzer & Trefethen) Using inverse Laplace transform (López-Fernández)

Krylov methods for the computation of matrix functions

Krylov methods for the computation of matrix functions Krylov methods for the computation of matrix functions Jitse Niesen (University of Leeds) in collaboration with Will Wright (Melbourne University) Heriot-Watt University, March 2010 Outline Definition

More information

Exponential Almost Runge-Kutta Methods for Semilinear Problems

Exponential Almost Runge-Kutta Methods for Semilinear Problems Exponential Almost Runge-Kutta Methods for Semilinear Problems John Carroll Eóin O Callaghan 9 January 3 Abstract We present a new class of one-step, multi-value Exponential Integrator (EI) methods referred

More information

Exponential integrators and functions of the matrix exponential

Exponential integrators and functions of the matrix exponential Exponential integrators and functions of the matrix exponential Paul Matthews, Stephen Cox, Hala Ashi and Linda Cummings School of Mathematical Sciences, University of Nottingham, UK Introduction to exponential

More information

The ReLPM Exponential Integrator for FE Discretizations of Advection-Diffusion Equations

The ReLPM Exponential Integrator for FE Discretizations of Advection-Diffusion Equations The ReLPM Exponential Integrator for FE Discretizations of Advection-Diffusion Equations Luca Bergamaschi 1, Marco Caliari 2, and Marco Vianello 3 1 Dip.to di Met. e Mod. Mat. per le Scienze Appl., Università

More information

Implementation of exponential Rosenbrock-type integrators

Implementation of exponential Rosenbrock-type integrators Implementation of exponential Rosenbrock-type integrators Marco Caliari a,b, Alexander Ostermann b, a Department of Pure and Applied Mathematics, University of Padua, Via Trieste 63, I-35121 Padova, Italy

More information

Approximation of functions of large matrices. Part I. Computational aspects. V. Simoncini

Approximation of functions of large matrices. Part I. Computational aspects. V. Simoncini Approximation of functions of large matrices. Part I. Computational aspects V. Simoncini Dipartimento di Matematica, Università di Bologna and CIRSA, Ravenna, Italy valeria@dm.unibo.it 1 The problem Given

More information

Computing the Action of the Matrix Exponential

Computing the Action of the Matrix Exponential Computing the Action of the Matrix Exponential Nick Higham School of Mathematics The University of Manchester higham@ma.man.ac.uk http://www.ma.man.ac.uk/~higham/ Joint work with Awad H. Al-Mohy 16th ILAS

More information

MOX EXPONENTIAL INTEGRATORS FOR MULTIPLE TIME SCALE PROBLEMS OF ENVIRONMENTAL FLUID DYNAMICS. Innsbruck Workshop October

MOX EXPONENTIAL INTEGRATORS FOR MULTIPLE TIME SCALE PROBLEMS OF ENVIRONMENTAL FLUID DYNAMICS. Innsbruck Workshop October Innsbruck Workshop October 29 21 EXPONENTIAL INTEGRATORS FOR MULTIPLE TIME SCALE PROBLEMS OF ENVIRONMENTAL FLUID DYNAMICS Luca Bonaventura - Modellistica e Calcolo Scientifico Dipartimento di Matematica

More information

FDM for parabolic equations

FDM for parabolic equations FDM for parabolic equations Consider the heat equation where Well-posed problem Existence & Uniqueness Mass & Energy decreasing FDM for parabolic equations CNFD Crank-Nicolson + 2 nd order finite difference

More information

Comparing Leja and Krylov approximations of large scale matrix exponentials

Comparing Leja and Krylov approximations of large scale matrix exponentials Comparing Leja and Krylov approximations of large scale matrix exponentials L. Bergamaschi 1, M. Caliari 2, A. Martínez 2, and M. Vianello 2 1 Dept. of Math. Methods and Models, University of Padova, berga@dmsa.unipd.it

More information

The LEM exponential integrator for advection-diffusion-reaction equations

The LEM exponential integrator for advection-diffusion-reaction equations The LEM exponential integrator for advection-diffusion-reaction equations Marco Caliari a, Marco Vianello a,, Luca Bergamaschi b a Dept. of Pure and Applied Mathematics, University of Padova. b Dept. of

More information

Implicit-explicit exponential integrators

Implicit-explicit exponential integrators Implicit-explicit exponential integrators Bawfeh Kingsley Kometa joint work with Elena Celledoni MaGIC 2011 Finse, March 1-4 1 Introduction Motivation 2 semi-lagrangian Runge-Kutta exponential integrators

More information

Exponential multistep methods of Adams-type

Exponential multistep methods of Adams-type Exponential multistep methods of Adams-type Marlis Hochbruck and Alexander Ostermann KARLSRUHE INSTITUTE OF TECHNOLOGY (KIT) 0 KIT University of the State of Baden-Wuerttemberg and National Laboratory

More information

Exponential integrators for semilinear parabolic problems

Exponential integrators for semilinear parabolic problems Exponential integrators for semilinear parabolic problems Marlis Hochbruck Heinrich-Heine University Düsseldorf Germany Innsbruck, October 2004 p. Outline Exponential integrators general class of methods

More information

APPROXIMATION OF THE LINEAR THE BLOCK SHIFT-AND-INVERT KRYLOV SUBSPACE METHOD

APPROXIMATION OF THE LINEAR THE BLOCK SHIFT-AND-INVERT KRYLOV SUBSPACE METHOD Journal of Applied Analysis and Computation Volume 7, Number 4, November 2017, 1402 1416 Website:http://jaac-online.com/ DOI:10.11948/2017085 APPROXIMATION OF THE LINEAR COMBINATION OF ϕ-functions USING

More information

Multi-Factor Finite Differences

Multi-Factor Finite Differences February 17, 2017 Aims and outline Finite differences for more than one direction The θ-method, explicit, implicit, Crank-Nicolson Iterative solution of discretised equations Alternating directions implicit

More information

Co-sponsored School/Workshop on Integrable Systems and Scientific Computing June Exponential integrators for stiff systems

Co-sponsored School/Workshop on Integrable Systems and Scientific Computing June Exponential integrators for stiff systems 2042-2 Co-sponsored School/Workshop on Integrable Systems and Scientific Computing 15-20 June 2009 Exponential integrators for stiff systems Paul Matthews University of Nottingham U.K. e-mail: Paul.Matthews@nottingham.ac.uk

More information

Comparing Leja and Krylov Approximations of Large Scale Matrix Exponentials

Comparing Leja and Krylov Approximations of Large Scale Matrix Exponentials Comparing Leja and Krylov Approximations of Large Scale Matrix Exponentials L. Bergamaschi 1,M.Caliari 2,A.Martínez 2, and M. Vianello 2 1 Dept. of Math. Methods and Models, University of Padova berga@dmsa.unipd.it

More information

Accurate evaluation of divided differences for polynomial interpolation of exponential propagators

Accurate evaluation of divided differences for polynomial interpolation of exponential propagators Computing 80, 189 201 (2007) DOI 10.1007/s00607-007-0227-1 Printed in The Netherlands Accurate evaluation of divided differences for polynomial interpolation of exponential propagators M. Caliari, Padua

More information

Semi-implicit Krylov Deferred Correction Methods for Ordinary Differential Equations

Semi-implicit Krylov Deferred Correction Methods for Ordinary Differential Equations Semi-implicit Krylov Deferred Correction Methods for Ordinary Differential Equations Sunyoung Bu University of North Carolina Department of Mathematics CB # 325, Chapel Hill USA agatha@email.unc.edu Jingfang

More information

Chapter 9 Implicit Methods for Linear and Nonlinear Systems of ODEs

Chapter 9 Implicit Methods for Linear and Nonlinear Systems of ODEs Chapter 9 Implicit Methods for Linear and Nonlinear Systems of ODEs In the previous chapter, we investigated stiffness in ODEs. Recall that an ODE is stiff if it exhibits behavior on widelyvarying timescales.

More information

The family of Runge Kutta methods with two intermediate evaluations is defined by

The family of Runge Kutta methods with two intermediate evaluations is defined by AM 205: lecture 13 Last time: Numerical solution of ordinary differential equations Today: Additional ODE methods, boundary value problems Thursday s lecture will be given by Thomas Fai Assignment 3 will

More information

Chapter Two: Numerical Methods for Elliptic PDEs. 1 Finite Difference Methods for Elliptic PDEs

Chapter Two: Numerical Methods for Elliptic PDEs. 1 Finite Difference Methods for Elliptic PDEs Chapter Two: Numerical Methods for Elliptic PDEs Finite Difference Methods for Elliptic PDEs.. Finite difference scheme. We consider a simple example u := subject to Dirichlet boundary conditions ( ) u

More information

EXPONENTIAL ROSENBROCK-TYPE METHODS

EXPONENTIAL ROSENBROCK-TYPE METHODS EXPONENTIAL ROSENBROCK-TYPE METHODS MARLIS HOCHBRUCK, ALEXANDER OSTERMANN, AND JULIA SCHWEITZER Abstract. We introduce a new class of exponential integrators for the numerical integration of large-scale

More information

Chapter 5 Exercises. (a) Determine the best possible Lipschitz constant for this function over 2 u <. u (t) = log(u(t)), u(0) = 2.

Chapter 5 Exercises. (a) Determine the best possible Lipschitz constant for this function over 2 u <. u (t) = log(u(t)), u(0) = 2. Chapter 5 Exercises From: Finite Difference Methods for Ordinary and Partial Differential Equations by R. J. LeVeque, SIAM, 2007. http://www.amath.washington.edu/ rjl/fdmbook Exercise 5. (Uniqueness for

More information

Index. higher order methods, 52 nonlinear, 36 with variable coefficients, 34 Burgers equation, 234 BVP, see boundary value problems

Index. higher order methods, 52 nonlinear, 36 with variable coefficients, 34 Burgers equation, 234 BVP, see boundary value problems Index A-conjugate directions, 83 A-stability, 171 A( )-stability, 171 absolute error, 243 absolute stability, 149 for systems of equations, 154 absorbing boundary conditions, 228 Adams Bashforth methods,

More information

Linear algebra for exponential integrators

Linear algebra for exponential integrators Linear algebra for exponential integrators Antti Koskela KTH Royal Institute of Technology Beräkningsmatematikcirkus, 28 May 2014 The problem Considered: time integration of stiff semilinear initial value

More information

PARTIAL DIFFERENTIAL EQUATIONS

PARTIAL DIFFERENTIAL EQUATIONS MATHEMATICAL METHODS PARTIAL DIFFERENTIAL EQUATIONS I YEAR B.Tech By Mr. Y. Prabhaker Reddy Asst. Professor of Mathematics Guru Nanak Engineering College Ibrahimpatnam, Hyderabad. SYLLABUS OF MATHEMATICAL

More information

Lecture 9 Approximations of Laplace s Equation, Finite Element Method. Mathématiques appliquées (MATH0504-1) B. Dewals, C.

Lecture 9 Approximations of Laplace s Equation, Finite Element Method. Mathématiques appliquées (MATH0504-1) B. Dewals, C. Lecture 9 Approximations of Laplace s Equation, Finite Element Method Mathématiques appliquées (MATH54-1) B. Dewals, C. Geuzaine V1.2 23/11/218 1 Learning objectives of this lecture Apply the finite difference

More information

MATHEMATICAL METHODS INTERPOLATION

MATHEMATICAL METHODS INTERPOLATION MATHEMATICAL METHODS INTERPOLATION I YEAR BTech By Mr Y Prabhaker Reddy Asst Professor of Mathematics Guru Nanak Engineering College Ibrahimpatnam, Hyderabad SYLLABUS OF MATHEMATICAL METHODS (as per JNTU

More information

Butcher tableau Can summarize an s + 1 stage Runge Kutta method using a triangular grid of coefficients

Butcher tableau Can summarize an s + 1 stage Runge Kutta method using a triangular grid of coefficients AM 205: lecture 13 Last time: ODE convergence and stability, Runge Kutta methods Today: the Butcher tableau, multi-step methods, boundary value problems Butcher tableau Can summarize an s + 1 stage Runge

More information

Meshfree Exponential Integrators

Meshfree Exponential Integrators Meshfree joint work with A. Ostermann (Innsbruck) M. Caliari (Verona) Leopold Franzens Universität Innsbruck Innovative Integrators 3 October 2 Meshfree Problem class: Goal: Time-dependent PDE s with dominating

More information

Test you method and verify the results with an appropriate function, e.g. f(x) = cos 1 + sin

Test you method and verify the results with an appropriate function, e.g. f(x) = cos 1 + sin Advanced methods for ODEs - Lab exercises Verona, 20-30 May 2015 1. FFT and IFFT in MATLAB: Use the Fast Fourier Transform to compute compute the k-th derivative of an appropriate function. In MATLAB the

More information

Motivation: Sparse matrices and numerical PDE's

Motivation: Sparse matrices and numerical PDE's Lecture 20: Numerical Linear Algebra #4 Iterative methods and Eigenproblems Outline 1) Motivation: beyond LU for Ax=b A little PDE's and sparse matrices A) Temperature Equation B) Poisson Equation 2) Splitting

More information

Lecture 8: Calculus and Differential Equations

Lecture 8: Calculus and Differential Equations Lecture 8: Calculus and Differential Equations Dr. Mohammed Hawa Electrical Engineering Department University of Jordan EE201: Computer Applications. See Textbook Chapter 9. Numerical Methods MATLAB provides

More information

Lecture 8: Calculus and Differential Equations

Lecture 8: Calculus and Differential Equations Lecture 8: Calculus and Differential Equations Dr. Mohammed Hawa Electrical Engineering Department University of Jordan EE21: Computer Applications. See Textbook Chapter 9. Numerical Methods MATLAB provides

More information

Computing the Action of the Matrix Exponential, with an Application to Exponential Integrators. Al-Mohy, Awad H. and Higham, Nicholas J.

Computing the Action of the Matrix Exponential, with an Application to Exponential Integrators. Al-Mohy, Awad H. and Higham, Nicholas J. Computing the Action of the Matrix Exponential, with an Application to Exponential Integrators Al-Mohy, Awad H. and Higham, Nicholas J. 2010 MIMS EPrint: 2010.30 Manchester Institute for Mathematical Sciences

More information

Strong Stability-Preserving (SSP) High-Order Time Discretization Methods

Strong Stability-Preserving (SSP) High-Order Time Discretization Methods Strong Stability-Preserving (SSP) High-Order Time Discretization Methods Xinghui Zhong 12/09/ 2009 Outline 1 Introduction Why SSP methods Idea History/main reference 2 Explicit SSP Runge-Kutta Methods

More information

Selected HW Solutions

Selected HW Solutions Selected HW Solutions HW1 1 & See web page notes Derivative Approximations. For example: df f i+1 f i 1 = dx h i 1 f i + hf i + h h f i + h3 6 f i + f i + h 6 f i + 3 a realmax 17 1.7014 10 38 b realmin

More information

Time-adaptive methods for the incompressible Navier-Stokes equations

Time-adaptive methods for the incompressible Navier-Stokes equations Time-adaptive methods for the incompressible Navier-Stokes equations Joachim Rang, Thorsten Grahs, Justin Wiegmann, 29.09.2016 Contents Introduction Diagonally implicit Runge Kutta methods Results with

More information

Bindel, Fall 2011 Intro to Scientific Computing (CS 3220) Week 12: Monday, Apr 18. HW 7 is posted, and will be due in class on 4/25.

Bindel, Fall 2011 Intro to Scientific Computing (CS 3220) Week 12: Monday, Apr 18. HW 7 is posted, and will be due in class on 4/25. Logistics Week 12: Monday, Apr 18 HW 6 is due at 11:59 tonight. HW 7 is posted, and will be due in class on 4/25. The prelim is graded. An analysis and rubric are on CMS. Problem du jour For implicit methods

More information

Stability of Krylov Subspace Spectral Methods

Stability of Krylov Subspace Spectral Methods Stability of Krylov Subspace Spectral Methods James V. Lambers Department of Energy Resources Engineering Stanford University includes joint work with Patrick Guidotti and Knut Sølna, UC Irvine Margot

More information

Krylov Implicit Integration Factor Methods for Semilinear Fourth-Order Equations

Krylov Implicit Integration Factor Methods for Semilinear Fourth-Order Equations mathematics Article Krylov Implicit Integration Factor Methods for Semilinear Fourth-Order Equations Michael Machen and Yong-Tao Zhang * Department of Applied and Computational Mathematics and Statistics,

More information

Krylov single-step implicit integration factor WENO methods for advection-diffusion-reaction equations

Krylov single-step implicit integration factor WENO methods for advection-diffusion-reaction equations Accepted Manuscript Krylov single-step implicit integration factor WENO methods for advection diffusion reaction equations Tian Jiang, Yong-Tao Zhang PII: S0021-9991(16)00029-2 DOI: http://dx.doi.org/10.1016/j.jcp.2016.01.021

More information

Newton s Method and Efficient, Robust Variants

Newton s Method and Efficient, Robust Variants Newton s Method and Efficient, Robust Variants Philipp Birken University of Kassel (SFB/TRR 30) Soon: University of Lund October 7th 2013 Efficient solution of large systems of non-linear PDEs in science

More information

The Leja method revisited: backward error analysis for the matrix exponential

The Leja method revisited: backward error analysis for the matrix exponential The Leja method revisited: backward error analysis for the matrix exponential M. Caliari 1, P. Kandolf 2, A. Ostermann 2, and S. Rainer 2 1 Dipartimento di Informatica, Università di Verona, Italy 2 Institut

More information

An Algorithmic Framework of Large-Scale Circuit Simulation Using Exponential Integrators

An Algorithmic Framework of Large-Scale Circuit Simulation Using Exponential Integrators An Algorithmic Framework of Large-Scale Circuit Simulation Using Exponential Integrators Hao Zhuang 1, Wenjian Yu 2, Ilgweon Kang 1, Xinan Wang 1, and Chung-Kuan Cheng 1 1. University of California, San

More information

A Composite Runge Kutta Method for the Spectral Solution of Semilinear PDEs

A Composite Runge Kutta Method for the Spectral Solution of Semilinear PDEs Journal of Computational Physics 8, 57 67 (00) doi:0.006/jcph.00.77 A Composite Runge Kutta Method for the Spectral Solution of Semilinear PDEs Tobin A. Driscoll Department of Mathematical Sciences, University

More information

Krylov subspace exponential time domain solution of Maxwell s equations in photonic crystal modeling

Krylov subspace exponential time domain solution of Maxwell s equations in photonic crystal modeling Krylov subspace exponential time domain solution of Maxwell s equations in photonic crystal modeling Mike A. Botchev Department of Applied Mathematics and MESA+ Institute for Nanotechnology, University

More information

On the Diagonal Approximation of Full Matrices

On the Diagonal Approximation of Full Matrices On the Diagonal Approximation of Full Matrices Walter M. Lioen CWI P.O. Box 94079, 090 GB Amsterdam, The Netherlands ABSTRACT In this paper the construction of diagonal matrices, in some

More information

Finite Difference Methods Assignments

Finite Difference Methods Assignments Finite Difference Metods Assignments Anders Söberg and Aay Saxena, Micael Tuné, and Maria Westermarck Revised: Jarmo Rantakokko June 6, 1999 Teknisk databeandling Assignment 1: A one-dimensional eat equation

More information

Computing the Action of the Matrix Exponential, with an Application to Exponential Integrators. Al-Mohy, Awad H. and Higham, Nicholas J.

Computing the Action of the Matrix Exponential, with an Application to Exponential Integrators. Al-Mohy, Awad H. and Higham, Nicholas J. Computing the Action of the Matrix Exponential, with an Application to Exponential Integrators Al-Mohy, Awad H. and Higham, Nicholas J. 2010 MIMS EPrint: 2010.30 Manchester Institute for Mathematical Sciences

More information

Review for Exam 2 Ben Wang and Mark Styczynski

Review for Exam 2 Ben Wang and Mark Styczynski Review for Exam Ben Wang and Mark Styczynski This is a rough approximation of what we went over in the review session. This is actually more detailed in portions than what we went over. Also, please note

More information

Experiments on Temporal Variable Step BDF2 Algorithms

Experiments on Temporal Variable Step BDF2 Algorithms University of Wisconsin Milwaukee UWM Digital Commons Theses and Dissertations May 2014 Experiments on Temporal Variable Step BDF2 Algorithms Anja Katrin Denner University of Wisconsin-Milwaukee Follow

More information

Approximating the matrix exponential of an advection-diffusion operator using the incomplete orthogonalization method

Approximating the matrix exponential of an advection-diffusion operator using the incomplete orthogonalization method Approximating the matrix exponential of an advection-diffusion operator using the incomplete orthogonalization method Antti Koskela KTH Royal Institute of Technology, Lindstedtvägen 25, 10044 Stockholm,

More information

A parallel exponential integrator for large-scale discretizations of advection-diffusion models

A parallel exponential integrator for large-scale discretizations of advection-diffusion models A parallel exponential integrator for large-scale discretizations of advection-diffusion models L. Bergamaschi 1, M. Caliari 2, A. Martínez 3, and M. Vianello 3 1 Department of Mathematical Methods and

More information

Comparison of various methods for computing the action of the matrix exponential

Comparison of various methods for computing the action of the matrix exponential BIT manuscript No. (will be inserted by the editor) Comparison of various methods for computing the action of the matrix exponential Marco Caliari Peter Kandolf Alexander Ostermann Stefan Rainer Received:

More information

Eigenvalue Problems CHAPTER 1 : PRELIMINARIES

Eigenvalue Problems CHAPTER 1 : PRELIMINARIES Eigenvalue Problems CHAPTER 1 : PRELIMINARIES Heinrich Voss voss@tu-harburg.de Hamburg University of Technology Institute of Mathematics TUHH Heinrich Voss Preliminaries Eigenvalue problems 2012 1 / 14

More information

Applied Numerical Analysis

Applied Numerical Analysis Applied Numerical Analysis Using MATLAB Second Edition Laurene V. Fausett Texas A&M University-Commerce PEARSON Prentice Hall Upper Saddle River, NJ 07458 Contents Preface xi 1 Foundations 1 1.1 Introductory

More information

Partial Differential Equations

Partial Differential Equations Partial Differential Equations Introduction Deng Li Discretization Methods Chunfang Chen, Danny Thorne, Adam Zornes CS521 Feb.,7, 2006 What do You Stand For? A PDE is a Partial Differential Equation This

More information

Stability of the Parareal Algorithm

Stability of the Parareal Algorithm Stability of the Parareal Algorithm Gunnar Andreas Staff and Einar M. Rønquist Norwegian University of Science and Technology Department of Mathematical Sciences Summary. We discuss the stability of the

More information

Krylov integrators for Hamiltonian systems. Eirola, Timo

Krylov integrators for Hamiltonian systems. Eirola, Timo https://helda.helsinki.fi Krylov integrators for Hamiltonian systems Eirola, Timo 2018-10-10 Eirola, T & Koskela, A 2018, ' Krylov integrators for Hamiltonian systems ' BIT Numerical Mathematics. https://doi.org/10.1007/s10543-018-0732-y

More information

Study guide: Generalizations of exponential decay models

Study guide: Generalizations of exponential decay models Study guide: Generalizations of exponential decay models Hans Petter Langtangen 1,2 Center for Biomedical Computing, Simula Research Laboratory 1 Department of Informatics, University of Oslo 2 Sep 13,

More information

The Newton-ADI Method for Large-Scale Algebraic Riccati Equations. Peter Benner.

The Newton-ADI Method for Large-Scale Algebraic Riccati Equations. Peter Benner. The Newton-ADI Method for Large-Scale Algebraic Riccati Equations Mathematik in Industrie und Technik Fakultät für Mathematik Peter Benner benner@mathematik.tu-chemnitz.de Sonderforschungsbereich 393 S

More information

Chapter 6 - Ordinary Differential Equations

Chapter 6 - Ordinary Differential Equations Chapter 6 - Ordinary Differential Equations 7.1 Solving Initial-Value Problems In this chapter, we will be interested in the solution of ordinary differential equations. Ordinary differential equations

More information

Finite Difference Methods (FDMs) 2

Finite Difference Methods (FDMs) 2 Finite Difference Methods (FDMs) 2 Time- dependent PDEs A partial differential equation of the form (15.1) where A, B, and C are constants, is called quasilinear. There are three types of quasilinear equations:

More information

Numerical Solutions to Partial Differential Equations

Numerical Solutions to Partial Differential Equations Numerical Solutions to Partial Differential Equations Zhiping Li LMAM and School of Mathematical Sciences Peking University The Implicit Schemes for the Model Problem The Crank-Nicolson scheme and θ-scheme

More information

CS 450 Numerical Analysis. Chapter 9: Initial Value Problems for Ordinary Differential Equations

CS 450 Numerical Analysis. Chapter 9: Initial Value Problems for Ordinary Differential Equations Lecture slides based on the textbook Scientific Computing: An Introductory Survey by Michael T. Heath, copyright c 2018 by the Society for Industrial and Applied Mathematics. http://www.siam.org/books/cl80

More information

Exponential integrators

Exponential integrators Acta Numerica (2), pp. 29 286 c Cambridge University Press, 2 doi:.7/s962492948 Printed in the United Kingdom Exponential integrators Marlis Hochbruck Karlsruher Institut für Technologie, Institut für

More information

Kasetsart University Workshop. Multigrid methods: An introduction

Kasetsart University Workshop. Multigrid methods: An introduction Kasetsart University Workshop Multigrid methods: An introduction Dr. Anand Pardhanani Mathematics Department Earlham College Richmond, Indiana USA pardhan@earlham.edu A copy of these slides is available

More information

High-order ADI schemes for convection-diffusion equations with mixed derivative terms

High-order ADI schemes for convection-diffusion equations with mixed derivative terms High-order ADI schemes for convection-diffusion equations with mixed derivative terms B. Düring, M. Fournié and A. Rigal Abstract We consider new high-order Alternating Direction Implicit ADI) schemes

More information

Time-Parallel Algorithms for Weather Prediction and Climate Simulation

Time-Parallel Algorithms for Weather Prediction and Climate Simulation Time-Parallel Algorithms for Weather Prediction and Climate Simulation Jean Côté Adjunct Prof. ESCER Centre (Étude et la Simulation du Climat à l Échelle Régionale) Sciences de la Terre & de l Atmosphère

More information

Study guide: Generalizations of exponential decay models. Extension to a variable coecient; Crank-Nicolson

Study guide: Generalizations of exponential decay models. Extension to a variable coecient; Crank-Nicolson Extension to a variable coecient; Forward and Backward Euler Study guide: Generalizations of exponential decay models Hans Petter Langtangen 1,2 Center for Biomedical Computing, Simula Research Laboratory

More information

MTH 452/552 Homework 3

MTH 452/552 Homework 3 MTH 452/552 Homework 3 Do either 1 or 2. 1. (40 points) [Use of ode113 and ode45] This problem can be solved by a modifying the m-files odesample.m and odesampletest.m available from the author s webpage.

More information

An Overly Simplified and Brief Review of Differential Equation Solution Methods. 1. Some Common Exact Solution Methods for Differential Equations

An Overly Simplified and Brief Review of Differential Equation Solution Methods. 1. Some Common Exact Solution Methods for Differential Equations An Overly Simplified and Brief Review of Differential Equation Solution Methods We will be dealing with initial or boundary value problems. A typical initial value problem has the form y y 0 y(0) 1 A typical

More information

Numerical Mathematics

Numerical Mathematics Alfio Quarteroni Riccardo Sacco Fausto Saleri Numerical Mathematics Second Edition With 135 Figures and 45 Tables 421 Springer Contents Part I Getting Started 1 Foundations of Matrix Analysis 3 1.1 Vector

More information

Scientific Computing: An Introductory Survey

Scientific Computing: An Introductory Survey Scientific Computing: An Introductory Survey Chapter 11 Partial Differential Equations Prof. Michael T. Heath Department of Computer Science University of Illinois at Urbana-Champaign Copyright c 2002.

More information

Coordinate Update Algorithm Short Course Operator Splitting

Coordinate Update Algorithm Short Course Operator Splitting Coordinate Update Algorithm Short Course Operator Splitting Instructor: Wotao Yin (UCLA Math) Summer 2016 1 / 25 Operator splitting pipeline 1. Formulate a problem as 0 A(x) + B(x) with monotone operators

More information

MATH 590: Meshfree Methods

MATH 590: Meshfree Methods MATH 590: Meshfree Methods Chapter 43: RBF-PS Methods in MATLAB Greg Fasshauer Department of Applied Mathematics Illinois Institute of Technology Fall 2010 fasshauer@iit.edu MATH 590 Chapter 43 1 Outline

More information

Time-Parallel Algorithms for Weather Prediction and Climate Simulation

Time-Parallel Algorithms for Weather Prediction and Climate Simulation Time-Parallel Algorithms for Weather Prediction and Climate Simulation Jean Côté Adjunct Prof. ESCER Centre (Étude et la Simulation du Climat à l Échelle Régionale) Sciences de la Terre & de l Atmosphère

More information

Inexact shift-invert Arnoldi method for evolution equations

Inexact shift-invert Arnoldi method for evolution equations ANZIAM J. 58 (E) pp.e1 E27, 2016 E1 Inexact shift-invert Arnoldi method for evolution equations Yuka Hashimoto 1 Takashi Nodera 2 November 16, 2016 Abstract Linear and nonlinear evolution equations with

More information

1.6: 16, 20, 24, 27, 28

1.6: 16, 20, 24, 27, 28 .6: 6, 2, 24, 27, 28 6) If A is positive definite, then A is positive definite. The proof of the above statement can easily be shown for the following 2 2 matrix, a b A = b c If that matrix is positive

More information

Scientific Computing: An Introductory Survey

Scientific Computing: An Introductory Survey Scientific Computing: An Introductory Survey Chapter 9 Initial Value Problems for Ordinary Differential Equations Prof. Michael T. Heath Department of Computer Science University of Illinois at Urbana-Champaign

More information

Lecture 42 Determining Internal Node Values

Lecture 42 Determining Internal Node Values Lecture 42 Determining Internal Node Values As seen in the previous section, a finite element solution of a boundary value problem boils down to finding the best values of the constants {C j } n, which

More information

Efficient Solution of Parabolic Equations by Krylov Approximation Methods

Efficient Solution of Parabolic Equations by Krylov Approximation Methods Efficient Solution of Parabolic Equations by Krylov Approximation Methods E. Gallopoulos and Y. Saad Abstract In this paper we take a new look at numerical techniques for solving parabolic equations by

More information

Outline. 1 Partial Differential Equations. 2 Numerical Methods for PDEs. 3 Sparse Linear Systems

Outline. 1 Partial Differential Equations. 2 Numerical Methods for PDEs. 3 Sparse Linear Systems Outline Partial Differential Equations Numerical Methods for PDEs Sparse Linear Systems 1 Partial Differential Equations 2 Numerical Methods for PDEs 3 Sparse Linear Systems Michael T Heath Scientific

More information

Exponential Integrators

Exponential Integrators Exponential Integrators John C. Bowman (University of Alberta) May 22, 2007 www.math.ualberta.ca/ bowman/talks 1 Exponential Integrators Outline Exponential Euler History Generalizations Stationary Green

More information

Dynamics and Time Integration. Computer Graphics CMU /15-662, Fall 2016

Dynamics and Time Integration. Computer Graphics CMU /15-662, Fall 2016 Dynamics and Time Integration Computer Graphics CMU 15-462/15-662, Fall 2016 Last Quiz: IK (and demo) Last few classes Keyframing and motion capture - Q: what can one NOT do with these techniques? The

More information

Martin Luther Universität Halle Wittenberg Institut für Mathematik

Martin Luther Universität Halle Wittenberg Institut für Mathematik Martin Luther Universität Halle Wittenberg Institut für Mathematik Implicit peer methods for large stiff ODE systems St. Beck, R. Weiner, H. Podhaisky and B. A. Schmitt Report No. 07 (2010) Editors: Professors

More information

Richarson Extrapolation for Runge-Kutta Methods

Richarson Extrapolation for Runge-Kutta Methods Richarson Extrapolation for Runge-Kutta Methods Zahari Zlatevᵃ, Ivan Dimovᵇ and Krassimir Georgievᵇ ᵃ Department of Environmental Science, Aarhus University, Frederiksborgvej 399, P. O. 358, 4000 Roskilde,

More information

1 Finite difference example: 1D implicit heat equation

1 Finite difference example: 1D implicit heat equation 1 Finite difference example: 1D implicit heat equation 1.1 Boundary conditions Neumann and Dirichlet We solve the transient heat equation ρc p t = ( k ) (1) on the domain L/2 x L/2 subject to the following

More information

ANALYSIS OF SOME KRYLOV SUBSPACE APPROXIMATIONS TO THE MATRIX EXPONENTIAL OPERATOR

ANALYSIS OF SOME KRYLOV SUBSPACE APPROXIMATIONS TO THE MATRIX EXPONENTIAL OPERATOR ANALYSIS OF SOME KRYLOV SUBSPACE APPROXIMATIONS TO THE MATRIX EXPONENTIAL OPERATOR Y. SAAD Abstract. In this note we present a theoretical analysis of some Krylov subspace approximations to the matrix

More information

Applied Mathematics 205. Unit III: Numerical Calculus. Lecturer: Dr. David Knezevic

Applied Mathematics 205. Unit III: Numerical Calculus. Lecturer: Dr. David Knezevic Applied Mathematics 205 Unit III: Numerical Calculus Lecturer: Dr. David Knezevic Unit III: Numerical Calculus Chapter III.3: Boundary Value Problems and PDEs 2 / 96 ODE Boundary Value Problems 3 / 96

More information

A Fifth Order Flux Implicit WENO Method

A Fifth Order Flux Implicit WENO Method A Fifth Order Flux Implicit WENO Method Sigal Gottlieb and Julia S. Mullen and Steven J. Ruuth April 3, 25 Keywords: implicit, weighted essentially non-oscillatory, time-discretizations. Abstract The weighted

More information

Numerical solution of ODEs

Numerical solution of ODEs Numerical solution of ODEs Arne Morten Kvarving Department of Mathematical Sciences Norwegian University of Science and Technology November 5 2007 Problem and solution strategy We want to find an approximation

More information

Research Article Diagonally Implicit Block Backward Differentiation Formulas for Solving Ordinary Differential Equations

Research Article Diagonally Implicit Block Backward Differentiation Formulas for Solving Ordinary Differential Equations International Mathematics and Mathematical Sciences Volume 212, Article ID 767328, 8 pages doi:1.1155/212/767328 Research Article Diagonally Implicit Block Backward Differentiation Formulas for Solving

More information

Iterative Methods for Incompressible Flow

Iterative Methods for Incompressible Flow Iterative Methods for Incompressible Flow Melanie McKay Thesis submitted to the Faculty of Graduate and Postdoctoral Studies in partial fulfillment of the requirements for the degree of Master s of Science

More information

Exponentials of Symmetric Matrices through Tridiagonal Reductions

Exponentials of Symmetric Matrices through Tridiagonal Reductions Exponentials of Symmetric Matrices through Tridiagonal Reductions Ya Yan Lu Department of Mathematics City University of Hong Kong Kowloon, Hong Kong Abstract A simple and efficient numerical algorithm

More information

ME Computational Fluid Mechanics Lecture 5

ME Computational Fluid Mechanics Lecture 5 ME - 733 Computational Fluid Mechanics Lecture 5 Dr./ Ahmed Nagib Elmekawy Dec. 20, 2018 Elliptic PDEs: Finite Difference Formulation Using central difference formulation, the so called five-point formula

More information

Background. Background. C. T. Kelley NC State University tim C. T. Kelley Background NCSU, Spring / 58

Background. Background. C. T. Kelley NC State University tim C. T. Kelley Background NCSU, Spring / 58 Background C. T. Kelley NC State University tim kelley@ncsu.edu C. T. Kelley Background NCSU, Spring 2012 1 / 58 Notation vectors, matrices, norms l 1 : max col sum... spectral radius scaled integral norms

More information