Numerical Methods

for Engineers and Scientists

An Introduction with
®

Applications using MATLAB

Third Edition

Amos Gilat
Vish- Subramaniam
Department of Mechanical Engineering
The Ohio State University

Wiley
Contents
Preface iii

Chapter 1 Introduction l

1.1 Background 1
1.2 Representation of Numbers on a Computer 4
1.3 Errors in Numerical Solutions 10
1.3.1 Round-OffErrors 10
1.3.2 Truncation Errors 13
1.3.3 Total Error 14

1.4 Computers and Programming 15

1.5 Problems 18

Chapter 2 Mathematical Background 23

2.1 Background 23
2.2 Concepts from Pre-Calculus and Calculus 24
2.3 Vectors 28
2.3.1 Operations with Vectors 30

2.4 Matrices and Linear Algebra 32

2.4.1 Operations with Matrices 33
2.4.2 Special Matrices '35
2.4.3 Inverse ofa Matrix 36

2.4.4 Properties ofMatrices 37

2.4.5 Determinant of a Matrix 37
2.4.6 Cramer's Rule and Solution of a System of Simultaneous Linear Equations 38
2.4.7 Norms 40
2.5 Ordinary Differential Equations (ODE) 41
2.6 Functions of Two or More Independent Variables 44
2.6.1 Definition ofthe Partial Derivative 44
2.6.2 Chain Rule 45
2.6.3 TheJacobian 46

2.7 Taylor Series Expansion of Functions 47

2.7.1 Taylor Series for a Function ofOne Variable 47
2.7.2 Taylor Series for a Function of Two Variables 49
2.8 Inner Product and Orthogonality 50
2.9 Problems 51

xi
 Contents

Chapter 3 Solving Nonlinear Equations 57

3.1 Background 57

3.2 Estimation of Errors in Numerical Solutions 59

3.3 Bisection Method 61

3.4 Regula Falsi Method 64

3.5 Newton's Method 66

3.6 Secant Method 71

3.7 Fixed-Point Iteration Method 74

Solving Nonlinear Equations 77

3.8 Use of MATLAB Built-in Functions for
3.8.1 The f zero Command 78

3.8.2 The roots Command 79

79
3.9 Equations with Multiple Solutions
3.10 Systems of Nonlinear Equations 81
3.10.1 Newton's Methodfor Solving a System ofNonlinear Equations 82
86
3.10.2 Fixed-Point Iteration Methodfor Solving a System ofNonlinear Equations
3.11 Problems 88

Chapter 4 Solving a System ofLinear Equations 99

4.1 Background 99
100
4.1.1 Overview ofNumerical Methods for Solving a System ofLinear Algebraic Equations
4.2 Gauss Elimination Method 102
110
4.2.1 Potential Difficulties When Applying the Gauss Elimination Method
4.3 Gauss Elimination with Pivoting 112

4.4 Gauss-Jordan Elimination Method 115

4.5 LU Decomposition Method 118

4.5.1 LU Decomposition Using the Gauss Elimination Procedure 120

4.5.2 LU Decomposition Using Crout's Method 121

4.5.3 LU Decomposition with Pivoting 128

4.6 Inverse of a Matrix 128
4.6.1 Calculating the Inverse with the LU Decomposition Method 129

the Inverse Using the Gauss-Jordan Method 131

4.6.2 Calculating
4.7 Iterative Methods 132
4.7.1 JacobiIterative Method 133
4.7.2 Gauss-Seidel Iterative Method 133

4.8 Use of MATLAB Built-in Functions for Solving a System of Linear Equations 136

4.8.1 Solving a System ofEquations Using MATLAB's Left and Right Division 136
4.8.2 Solving
a Using MATLAB"s Inverse Operation 137
System of Equations
4.8.3 MATLAB's Built-in Function for LUDecomposition 138

- 4.8.4 Additional MATLAB Built-in Functions 139

4.9 Tridiagonal Systems of Equations 141

 Contents xiii

4.10 Error, Residual, Norms, and Condition Number 146

4.10.1 Error and Residual 146
4.10.2 Norms and Condition Number 148
4.11 Ill-Conditioned Systems 153
4.12 Problems 155

Chapter 5 Eigenvalues and Eigenvectors 165

5.1 Background 165

5.2 The Characteristic Equation 167
5.3 The Basic Power Method 167
5.4 The Inverse Power Method 172
5.5 The Shifted Power Method 173
5.6 The QR Factorization and Iteration Method 174
5.7 Use of MATLAB Built-in Functions for Determining Eigenvalues and
Eigenvectors 184
5.8 Problems 186

Chapter 6 Curve Fitting and Interpolation 193

6.1 Background 193

6.2 Curve Fitting with a Linear Equation 195
6.2.1 Measuring How Good Is a Fit 195
6.2.2 Linear Least-Squares Regression 197
6.3 Curve Fitting with Nonlinear Equation by Writing the Equation in a Linear Form 201
6.4 Curve Fitting with Quadratic and Higher-Order Polynomials 205
6.5 Interpolation Using a Single Polynomial 210
6.5.1 Lagrange Interpolating Polynomials 212
6.5.2 Newton's Interpolating Polynomials 216
6.6 Piecewise (Spline) Interpolation 223
6.6.1 Linear Splines 223
6.6.2 Quadratic Splines 225
6.6.3 Cubic Splines 229
6.7 Use of MATLAB Built-in Functions for Curve Fitting and Interpolation 236
6.8 Curve Fitting with a Linear Combination of Nonlinear Functions 238
6.9 Problems 241

Chapter 7 Fourier Methods 251

7.1 Background 251

7.2 Approximating a Square Wave by a Series of Sine Functions 253
7.3 General (Infinite) Fourier Series 257
7.4 Complex Form of the Fourier Series 261
xiv Contents

7.5 The Discrete Fourier Series and Discrete Fourier Transform 263

7.6 Complex Discrete Fourier Transform 268

7.7 Power (Energy) Spectrum 271

7.8 Aliasing and Nyquist Frequency 272

7.9 Alternative Forms of the Discrete Fourier Transform 278

7.10 Use of MATLAB Built-in Functions for Calculating Discrete Fourier Transform 278

7.11 Leakage and Windowing 284

7.12 Bandwidth and Filters 286

7.13 The Fast Fourier Transform (FFT) 289

7.14 Problems 298

Chapter 8 Numerical Differentiation 303

8.1 Background 303

8.2 Finite Difference Approximation of the Derivative 305
8.3 Finite Difference Formulas Using Taylor Series Expansion 310
8.3.1 Finite Difference Formulas ofFirst Derivative 310
8.3.2 Finite Difference Formulasfor the Second Derivative 315

8.4 Summary of Finite Difference Formulas for Numerical Differentiation 317

8.5 Differentiation Formulas Using Lagrange Polynomials 319

8.6 Differentiation Using Curve Fitting 320

8.7 Use of MATLAB Built-in Functions for Numerical Differentiation 320

8.8 Richardson's Extrapolation 322

8.9 Error in Numerical Differentiation 325

8.10 Numerical Partial Differentiation 327
8.11 Problems 330

Chapter 9 Numerical Integration 341

9.1 Background 341

9.1.1 Overview ofApproaches in Numerical Integration 342

9.2 Rectangle and Midpoint Methods 344

9.3 Trapezoidal Method 346
9.3.1 Composite Trapezoidal Method 347
9.4 Simpson's Methods 350
9.4.1 Simpson's 1/3 Method 350
9.4.2 Simpson's3/8Method 353
9.5 Gauss Quadrature 355
9.6 Evaluation of Multiple Integrals 360
9.7 Use of MATLAB Built-in Functions for Integration 362
9.8 Estimation of Error in Numerical Integration 364
9.9 Richardson's Extrapolation 366
 Contents xv

9.10 Romberg Integration 369

9.11 Improper Integrals 372
9.11.1 Integrals with Singularities 372
9.11.2 Integrals with Unbounded Limits 373
9.12 Problems 374

Chapter 10 Ordinary Differential Equations: Initial- Value

Problems 385

10.1 Background 385

10.2 Euler's Methods 390
10.2.1 Euler 's Explicit Method 390
10.2.2 Analysis of Truncation Error in Euler's Explicit Method 394
10.2.3 Euler's Implicit Method 398

10.3 Modified Euler's Method 401

10.4 Midpoint Method 404
10.5 Runge-Kutta Methods 405
10.5.1 Second-Order Runge-Kutta Methods 406
10.5.2 Third-Order Runge-Kutta Methods 410
10.5.3 Fourth-Order Runge-Kutta Methods 411 -

10.6 Multistep Methods 417

10.6.1 Adams-BashforthMethod 418
10.6.2 Adams-Moulton Method 419
10.7 Predictor-Corrector Methods 420
10.8 System of First-Order Ordinary Differential Equations 422
10.8.1 Solving a System ofFirst-Order ODEs Using Euler's Explicit Method 424
10.8.2 Solving a System ofFirst-Order ODEs Using Second-Order Runge-Kutta Method
(Modified Euler Version) 424
10.8.3 Solving a System ofFirst-Order ODEs Using the Classical Fourth-Order Runge-Kutta
Method 431
10.9 Solving a Higher-Order Initial Value Problem 432
10.10 Use of MATLAB Built-in Functions for Solving Initial-Value Problems 437
10.10.1 Solving a Single First-Order ODE Using MATLAB 438
10.10.2 Solving a System ofFirst-Order ODEs Using MATLAB 444
10.11 Local Truncation Error in Second-Order Range-Kutta Method 447
10.12 Step Size for Desired
Accuracy 448
10.13 Stability 452
10.14 Stiff Ordinary Differential Equations 454
10.15 Problems 457

,
xvi Contents

Chapter 11 Ordinary Differential Equations: Boundary- Value

Problems 471

11.1 Background 471

11.2 The Shooting Method 474
11.3 Finite Difference Method 482

11.4 Use of MATLAB Built-in Functions for Solving Boundary Value Problems 492
11.5 Error and Stability in Numerical Solution of Boundary Value Problems 497
11.6 Problems 499

Appendix A Introductory MATLAB 509

A. 1 Background 509
A.2 Starting with MATLAB 509
A.3 Arrays 514
A.4 Mathematical Operations with Arrays 519

A.5 Script Files 524

A.6 Plotting 526
A.7 User-Defined Functions and Function Files 528
A. 8 Anonymous Functions 530
A.9 Function functions 532
A. 10 Subfunctions 535
A. 11 Programming in MATLAB 537
A. 11.1 Relational and Logical
Operators 537
A.11.2 Conditional Statements, if-else Structures 538
A. 11.3 Loops 541
A. 12 Problems 542

Appendix B MATLAB Programs 547

Appendix C Derivation of the Real Discrete Fourier Transform

(DFT) 551

C. 1 Orthogonality of Sines and Cosines for Discrete Points 551

C.2 Determination of the Real DFT 553

Index 555