This action might not be possible to undo. Are you sure you want to continue?
and Nonlinear Equations
C. T. Kelley
North Carolina State University
Society for Industrial and Applied Mathematics
Philadelphia 1995
Untitled1 9/20/2004, 2:59 PM 3
To Polly H. Thomas, 19061994, devoted mother and grandmother
1
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
2
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Contents
Preface xi
How to Get the Software xiii
CHAPTER 1. Basic Concepts and Stationary Iterative Methods 3
1.1 Review and notation . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 The Banach Lemma and approximate inverses . . . . . . . . . . 5
1.3 The spectral radius . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Matrix splittings and classical stationary iterative methods . . 7
1.5 Exercises on stationary iterative methods . . . . . . . . . . . . 10
CHAPTER 2. Conjugate Gradient Iteration 11
2.1 Krylov methods and the minimization property . . . . . . . . . 11
2.2 Consequences of the minimization property . . . . . . . . . . . 13
2.3 Termination of the iteration . . . . . . . . . . . . . . . . . . . . 15
2.4 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Preconditioning . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6 CGNR and CGNE . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.7 Examples for preconditioned conjugate iteration . . . . . . . . 26
2.8 Exercises on conjugate gradient . . . . . . . . . . . . . . . . . . 30
CHAPTER 3. GMRES Iteration 33
3.1 The minimization property and its consequences . . . . . . . . 33
3.2 Termination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3 Preconditioning . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 GMRES implementation: Basic ideas . . . . . . . . . . . . . . . 37
3.5 Implementation: Givens rotations . . . . . . . . . . . . . . . . . 43
3.6 Other methods for nonsymmetric systems . . . . . . . . . . . . 46
3.6.1 BiCG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.6.2 CGS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.6.3 BiCGSTAB. . . . . . . . . . . . . . . . . . . . . . . . . 50
vii
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
viii CONTENTS
3.6.4 TFQMR. . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.7 Examples for GMRES iteration . . . . . . . . . . . . . . . . . . 54
3.8 Examples for CGNR, BiCGSTAB, and TFQMR iteration . . . 55
3.9 Exercises on GMRES . . . . . . . . . . . . . . . . . . . . . . . . 60
CHAPTER 4. Basic Concepts and FixedPoint Iteration 65
4.1 Types of convergence . . . . . . . . . . . . . . . . . . . . . . . . 65
4.2 Fixedpoint iteration . . . . . . . . . . . . . . . . . . . . . . . . 66
4.3 The standard assumptions . . . . . . . . . . . . . . . . . . . . . 68
CHAPTER 5. Newton’s Method 71
5.1 Local convergence of Newton’s method . . . . . . . . . . . . . . 71
5.2 Termination of the iteration . . . . . . . . . . . . . . . . . . . . 72
5.3 Implementation of Newton’s method . . . . . . . . . . . . . . . 73
5.4 Errors in the function and derivative . . . . . . . . . . . . . . . 75
5.4.1 The chord method. . . . . . . . . . . . . . . . . . . . . . 76
5.4.2 Approximate inversion of F
. . . . . . . . . . . . . . . . 77
5.4.3 The Shamanskii method. . . . . . . . . . . . . . . . . . 78
5.4.4 Diﬀerence approximation to F
. . . . . . . . . . . . . . . 79
5.4.5 The secant method. . . . . . . . . . . . . . . . . . . . . 82
5.5 The Kantorovich Theorem . . . . . . . . . . . . . . . . . . . . . 83
5.6 Examples for Newton’s method . . . . . . . . . . . . . . . . . . 86
5.7 Exercises on Newton’s method . . . . . . . . . . . . . . . . . . 91
CHAPTER 6. Inexact Newton Methods 95
6.1 The basic estimates . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.1.1 Direct analysis. . . . . . . . . . . . . . . . . . . . . . . . 95
6.1.2 Weighted norm analysis. . . . . . . . . . . . . . . . . . . 97
6.1.3 Errors in the function. . . . . . . . . . . . . . . . . . . . 100
6.2 Newtoniterative methods . . . . . . . . . . . . . . . . . . . . . 100
6.2.1 Newton GMRES. . . . . . . . . . . . . . . . . . . . . . . 101
6.2.2 Other Newtoniterative methods. . . . . . . . . . . . . . 104
6.3 NewtonGMRES implementation . . . . . . . . . . . . . . . . . 104
6.4 Examples for NewtonGMRES . . . . . . . . . . . . . . . . . . 106
6.4.1 Chandrasekhar Hequation. . . . . . . . . . . . . . . . . 107
6.4.2 Convectiondiﬀusion equation. . . . . . . . . . . . . . . 108
6.5 Exercises on inexact Newton methods . . . . . . . . . . . . . . 110
CHAPTER 7. Broyden’s method 113
7.1 The Dennis–Mor´e condition . . . . . . . . . . . . . . . . . . . . 114
7.2 Convergence analysis . . . . . . . . . . . . . . . . . . . . . . . . 116
7.2.1 Linear problems. . . . . . . . . . . . . . . . . . . . . . . 118
7.2.2 Nonlinear problems. . . . . . . . . . . . . . . . . . . . . 120
7.3 Implementation of Broyden’s method . . . . . . . . . . . . . . . 123
7.4 Examples for Broyden’s method . . . . . . . . . . . . . . . . . . 127
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONTENTS ix
7.4.1 Linear problems. . . . . . . . . . . . . . . . . . . . . . . 127
7.4.2 Nonlinear problems. . . . . . . . . . . . . . . . . . . . . 128
7.5 Exercises on Broyden’s method . . . . . . . . . . . . . . . . . . 132
CHAPTER 8. Global Convergence 135
8.1 Single equations . . . . . . . . . . . . . . . . . . . . . . . . . . 135
8.2 Analysis of the Armijo rule . . . . . . . . . . . . . . . . . . . . 138
8.3 Implementation of the Armijo rule . . . . . . . . . . . . . . . . 141
8.3.1 Polynomial line searches. . . . . . . . . . . . . . . . . . 142
8.3.2 Broyden’s method. . . . . . . . . . . . . . . . . . . . . . 144
8.4 Examples for Newton–Armijo . . . . . . . . . . . . . . . . . . . 146
8.4.1 Inverse tangent function. . . . . . . . . . . . . . . . . . 146
8.4.2 Convectiondiﬀusion equation. . . . . . . . . . . . . . . 146
8.4.3 Broyden–Armijo. . . . . . . . . . . . . . . . . . . . . . . 148
8.5 Exercises on global convergence . . . . . . . . . . . . . . . . . . 151
Bibliography 153
Index 163
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
x CONTENTS
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Preface
This book on iterative methods for linear and nonlinear equations can be used
as a tutorial and a reference by anyone who needs to solve nonlinear systems
of equations or large linear systems. It may also be used as a textbook for
introductory courses in nonlinear equations or iterative methods or as source
material for an introductory course in numerical analysis at the graduate level.
We assume that the reader is familiar with elementary numerical analysis,
linear algebra, and the central ideas of direct methods for the numerical
solution of dense linear systems as described in standard texts such as [7],
[105], or [184].
Our approach is to focus on a small number of methods and treat them
in depth. Though this book is written in a ﬁnitedimensional setting, we
have selected for coverage mostly algorithms and methods of analysis which
extend directly to the inﬁnitedimensional case and whose convergence can be
thoroughly analyzed. For example, the matrixfree formulation and analysis for
GMRES and conjugate gradient is almost unchanged in an inﬁnitedimensional
setting. The analysis of Broyden’s method presented in Chapter 7 and
the implementations presented in Chapters 7 and 8 are diﬀerent from the
classical ones and also extend directly to an inﬁnitedimensional setting. The
computational examples and exercises focus on discretizations of inﬁnite
dimensional problems such as integral and diﬀerential equations.
We present a limited number of computational examples. These examples
are intended to provide results that can be used to validate the reader’s own
implementations and to give a sense of how the algorithms perform. The
examples are not designed to give a complete picture of performance or to be
a suite of test problems.
The computational examples in this book were done with MATLAB
(version 4.0a on various SUN SPARCstations and version 4.1 on an Apple
Macintosh Powerbook 180) and the MATLAB environment is an excellent one
for getting experience with the algorithms, for doing the exercises, and for
smalltomedium scale production work.
1
MATLAB codes for many of the
algorithms are available by anonymous ftp. A good introduction to the latest
1
MATLAB is a registered trademark of The MathWorks, Inc.
xi
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
xii PREFACE
version (version 4.2) of MATLAB is the MATLAB Primer [178]; [43] is also
a useful resource. If the reader has no access to MATLAB or will be solving
very large problems, the general algorithmic descriptions or even the MATLAB
codes can easily be translated to another language.
Parts of this book are based upon work supported by the National
Science Foundation and the Air Force Oﬃce of Scientiﬁc Research over
several years, most recently under National Science Foundation Grant Nos.
DMS9024622 and DMS9321938. Any opinions, ﬁndings, and conclusions or
recommendations expressed in this material are those of the author and do not
necessarily reﬂect the views of the National Science Foundation or of the Air
Force Oﬃce of Scientiﬁc Research.
Many of my students and colleagues discussed various aspects of this
project with me and provided important corrections, ideas, suggestions, and
pointers to the literature. I am especially indebted to Jim Banoczi, Jeﬀ Butera,
Steve Campbell, Tony Choi, Moody Chu, Howard Elman, Jim Epperson,
Andreas Griewank, Laura Helfrich, Ilse Ipsen, Lea Jenkins, Vickie Kearn,
Belinda King, Debbie Lockhart, Carl Meyer, Casey Miller, Ekkehard Sachs,
Jeﬀ Scroggs, Joseph Skudlarek, Mike Tocci, Gordon Wade, Homer Walker,
Steve Wright, Zhaqing Xue, Yue Zhang, and an anonymous reviewer for their
contributions and encouragement.
Most importantly, I thank ChungWei Ng and my parents for over one
hundred and ten years of patience and support.
C. T. Kelley
Raleigh, North Carolina
January, 1998
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
How to get the software
A collection of MATLAB codes has been written to accompany this book. The
MATLAB codes can be obtained by anonymous ftp from the MathWorks server
ftp.mathworks.com in the directory pub/books/kelley, from the MathWorks
World Wide Web site,
http://www.mathworks.com
or from SIAM’s World Wide Web site
http://www.siam.org/books/kelley/kelley.html
One can obtain MATLAB from
The MathWorks, Inc.
24 Prime Park Way
Natick, MA 01760,
Phone: (508) 6531415
Fax: (508) 6532997
Email: info@mathworks.com
WWW: http://www.mathworks.com
xiii
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Chapter 1
Basic Concepts and Stationary Iterative Methods
1.1. Review and notation
We begin by setting notation and reviewing some ideas from numerical linear
algebra that we expect the reader to be familiar with. An excellent reference
for the basic ideas of numerical linear algebra and direct methods for linear
equations is [184].
We will write linear equations as
Ax = b, (1.1)
where A is a nonsingular N N matrix, b ∈ R
N
is given, and
x
∗
= A
−1
b ∈ R
N
is to be found.
Throughout this chapter x will denote a potential solution and ¦x
k
¦
k≥0
the
sequence of iterates. We will denote the ith component of a vector x by (x)
i
(note the parentheses) and the ith component of x
k
by (x
k
)
i
. We will rarely
need to refer to individual components of vectors.
In this chapter   will denote a norm on R
N
as well as the induced matrix
norm.
Definition 1.1.1. Let   be a norm on R
N
. The induced matrix norm
of an N N matrix A is deﬁned by
A = max
x=1
Ax.
Induced norms have the important property that
Ax ≤ Ax.
Recall that the condition number of A relative to the norm   is
κ(A) = AA
−1
,
where κ(A) is understood to be inﬁnite if A is singular. If   is the l
p
norm
x
p
=
_
_
N
j=1
[(x)
i
[
p
_
_
1/p
3
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
4 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
we will write the condition number as κ
p
.
Most iterative methods terminate when the residual
r = b −Ax
is suﬃciently small. One termination criterion is
r
k

r
0

< τ, (1.2)
which can be related to the error
e = x −x
∗
in terms of the condition number.
Lemma 1.1.1. Let b, x, x
0
∈ R
N
. Let A be nonsingular and let x
∗
= A
−1
b.
e
e
0

≤ κ(A)
r
r
0

. (1.3)
Proof. Since
r = b −Ax = −Ae
we have
e = A
−1
Ae ≤ A
−1
Ae = A
−1
r
and
r
0
 = Ae
0
 ≤ Ae
0
.
Hence
e
e
0

≤
A
−1
r
A
−1
r
0

= κ(A)
r
r
0

,
as asserted.
The termination criterion (1.2) depends on the initial iterate and may result
in unnecessary work when the initial iterate is good and a poor result when the
initial iterate is far from the solution. For this reason we prefer to terminate
the iteration when
r
k

b
< τ. (1.4)
The two conditions (1.2) and (1.4) are the same when x
0
= 0, which is a
common choice, particularly when the linear iteration is being used as part of
a nonlinear solver.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BASIC CONCEPTS 5
1.2. The Banach Lemma and approximate inverses
The most straightforward approach to an iterative solution of a linear system
is to rewrite (1.1) as a linear ﬁxedpoint iteration. One way to do this is to
write Ax = b as
x = (I −A)x +b, (1.5)
and to deﬁne the Richardson iteration
x
k+1
= (I −A)x
k
+b. (1.6)
We will discuss more general methods in which ¦x
k
¦ is given by
x
k+1
= Mx
k
+c. (1.7)
In (1.7) M is an NN matrix called the iteration matrix. Iterative methods of
this form are called stationary iterative methods because the transition from x
k
to x
k+1
does not depend on the history of the iteration. The Krylov methods
discussed in Chapters 2 and 3 are not stationary iterative methods.
All our results are based on the following lemma.
Lemma 1.2.1. If M is an N N matrix with M < 1 then I − M is
nonsingular and
(I −M)
−1
 ≤
1
1 −M
. (1.8)
Proof. We will show that I − M is nonsingular and that (1.8) holds by
showing that the series
∞
l=0
M
l
= (I −M)
−1
.
The partial sums
S
k
=
k
l=0
M
l
form a Cauchy sequence in R
N×N
. To see this note that for all m > k
S
k
−S
m
 ≤
m
l=k+1
M
l
.
Now, M
l
 ≤ M
l
because   is a matrix norm that is induced by a vector
norm. Hence
S
k
−S
m
 ≤
m
l=k+1
M
l
= M
k+1
_
1 −M
m−k
1 −M
_
→ 0
as m, k → ∞. Hence the sequence S
k
converges, say to S. Since MS
k
+ I =
S
k+1
, we must have MS +I = S and hence (I −M)S = I. This proves that
I −M is nonsingular and that S = (I −M)
−1
.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
6 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Noting that
(I −M)
−1
 ≤
∞
l=0
M
l
= (1 −M)
−1
.
proves (1.8) and completes the proof.
The following corollary is a direct consequence of Lemma 1.2.1.
Corollary 1.2.1. If M < 1 then the iteration (1.7) converges to
x = (I −M)
−1
c for all initial iterates x
0
.
A consequence of Corollary 1.2.1 is that Richardson iteration (1.6) will
converge if I − A < 1. It is sometimes possible to precondition a linear
equation by multiplying both sides of (1.1) by a matrix B
BAx = Bb
so that convergence of iterative methods is improved. In the context of
Richardson iteration, the matrices B that allow us to apply the Banach lemma
and its corollary are called approximate inverses.
Definition 1.2.1. B is an approximate inverse of A if I −BA < 1.
The following theorem is often referred to as the Banach Lemma.
Theorem 1.2.1. If A and B are NN matrices and B is an approximate
inverse of A, then A and B are both nonsingular and
A
−1
 ≤
B
1 −I −BA
, B
−1
 ≤
A
1 −I −BA
, (1.9)
and
A
−1
−B ≤
BI −BA
1 −I −BA
, A−B
−1
 ≤
AI −BA
1 −I −BA
. (1.10)
Proof. Let M = I −BA. By Lemma 1.2.1 I −M = I −(I −BA) = BA is
nonsingular. Hence both A and B are nonsingular. By (1.8)
A
−1
B
−1
 = (I −M)
−1
 ≤
1
1 −M
=
1
1 −I −BA
. (1.11)
Since A
−1
= (I −M)
−1
B, inequality (1.11) implies the ﬁrst part of (1.9). The
second part follows in a similar way from B
−1
= A(I −M)
−1
.
To complete the proof note that
A
−1
−B = (I −BA)A
−1
, A−B
−1
= B
−1
(I −BA),
and use (1.9).
Richardson iteration, preconditioned with approximate inversion, has the
form
x
k+1
= (I −BA)x
k
+Bb. (1.12)
If the norm of I − BA is small, then not only will the iteration converge
rapidly, but, as Lemma 1.1.1 indicates, termination decisions based on the
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BASIC CONCEPTS 7
preconditioned residual Bb − BAx will better reﬂect the actual error. This
method is a very eﬀective technique for solving diﬀerential equations, integral
equations, and related problems [15], [6], [100], [117], [111]. Multigrid methods
[19], [99], [126], can also be interpreted in this light. We mention one other
approach, polynomial preconditioning, which tries to approximate A
−1
by a
polynomial in A [123], [179], [169].
1.3. The spectral radius
The analysis in ¸ 1.2 related convergence of the iteration (1.7) to the norm of
the matrix M. However the norm of M could be small in some norms and
quite large in others. Hence the performance of the iteration is not completely
described by M. The concept of spectral radius allows us to make a complete
description.
We let σ(A) denote the set of eigenvalues of A.
Definition 1.3.1. The spectral radius of an N N matrix A is
ρ(A) = max
λ∈σ(A)
[λ[ = lim
n→∞
A
n

1/n
. (1.13)
The term on the righthand side of the second equality in (1.13) is the limit
used by the radical test for convergence of the series
A
n
.
The spectral radius of M is independent of any particular matrix norm of
M. It is clear, in fact, that
ρ(A) ≤ A (1.14)
for any induced matrix norm. The inequality (1.14) has a partial converse that
allows us to completely describe the performance of iteration (1.7) in terms of
spectral radius. We state that converse as a theorem and refer to [105] for a
proof.
Theorem 1.3.1. Let A be an N N matrix. Then for any > 0 there is
a norm   on R
N
such that
ρ(A) > A −.
A consequence of Theorem 1.3.1, Lemma 1.2.1, and Exercise 1.5.1 is a
characterization of convergent stationary iterative methods. The proof is left
as an exercise.
Theorem 1.3.2. Let M be an NN matrix. The iteration (1.7) converges
for all c ∈ R
N
if and only if ρ(M) < 1.
1.4. Matrix splittings and classical stationary iterative methods
There are ways to convert Ax = b to a linear ﬁxedpoint iteration that are
diﬀerent from (1.5). Methods such as Jacobi, Gauss–Seidel, and sucessive
overrelaxation (SOR) iteration are based on splittings of A of the form
A = A
1
+A
2
,
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
8 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
where A
1
is a nonsingular matrix constructed so that equations with A
1
as
coeﬃcient matrix are easy to solve. Then Ax = b is converted to the ﬁxed
point problem
x = A
−1
1
(b −A
2
x).
The analysis of the method is based on an estimation of the spectral radius of
the iteration matrix M = −A
−1
1
A
2
.
For a detailed description of the classical stationary iterative methods the
reader may consult [89], [105], [144], [193], or [200]. These methods are usually
less eﬃcient than the Krylov methods discussed in Chapters 2 and 3 or the more
modern stationary methods based on multigrid ideas. However the classical
methods have a role as preconditioners. The limited description in this section
is intended as a review that will set some notation to be used later.
As a ﬁrst example we consider the Jacobi iteration that uses the splitting
A
1
= D, A
2
= L +U,
where D is the diagonal of A and L and U are the (strict) lower and upper
triangular parts. This leads to the iteration matrix
M
JAC
= −D
−1
(L +U).
Letting (x
k
)
i
denote the ith component of the kth iterate we can express Jacobi
iteration concretely as
(x
k+1
)
i
= a
−1
ii
_
_
b
i
−
j=i
a
ij
(x
k
)
j
_
_
. (1.15)
Note that A
1
is diagonal and hence trivial to invert.
We present only one convergence result for the classical stationary iterative
methods.
Theorem 1.4.1. Let A be an N N matrix and assume that for all
1 ≤ i ≤ N
0 <
j=i
[a
ij
[ < [a
ii
[. (1.16)
Then A is nonsingular and the Jacobi iteration (1.15) converges to x
∗
= A
−1
b
for all b.
Proof. Note that the ith row sum of M = M
JAC
satisﬁes
N
j=1
[m
ij
[ =
j=i
[a
ij
[
[a
ii
[
< 1.
Hence M
JAC

∞
< 1 and the iteration converges to the unique solution of
x = Mx + D
−1
b. Also I − M = D
−1
A is nonsingular and therefore A is
nonsingular.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BASIC CONCEPTS 9
Gauss–Seidel iteration overwrites the approximate solution with the new
value as soon as it is computed. This results in the iteration
(x
k+1
)
i
= a
−1
ii
_
_
b
i
−
j<i
a
ij
(x
k+1
)
j
−
j>i
a
ij
(x
k
)
j
_
_
,
the splitting
A
1
= D +L, A
2
= U,
and iteration matrix
M
GS
= −(D +L)
−1
U.
Note that A
1
is lower triangular, and hence A
−1
1
y is easy to compute for vectors
y. Note also that, unlike Jacobi iteration, the iteration depends on the ordering
of the unknowns. Backward Gauss–Seidel begins the update of x with the Nth
coordinate rather than the ﬁrst, resulting in the splitting
A
1
= D +U, A
2
= L,
and iteration matrix
M
BGS
= −(D +U)
−1
L.
A symmetric Gauss–Seidel iteration is a forward Gauss–Seidel iteration
followed by a backward Gauss–Seidel iteration. This leads to the iteration
matrix
M
SGS
= M
BGS
M
GS
= (D +U)
−1
L(D +L)
−1
U.
If A is symmetric then U = L
T
. In that event
M
SGS
= (D +U)
−1
L(D +L)
−1
U = (D +L
T
)
−1
L(D +L)
−1
L
T
.
From the point of view of preconditioning, one wants to write the stationary
method as a preconditioned Richardson iteration. That means that one wants
to ﬁnd B such that M = I − BA and then use B as an approximate inverse.
For the Jacobi iteration,
B
JAC
= D
−1
. (1.17)
For symmetric Gauss–Seidel
B
SGS
= (D +L
T
)
−1
D(D +L)
−1
. (1.18)
The successive overrelaxation iteration modiﬁes Gauss–Seidel by adding a
relaxation parameter ω to construct an iteration with iteration matrix
M
SOR
= (D +ωL)
−1
((1 −ω)D −ωU).
The performance can be dramatically improved with a good choice of ω but
still is not competitive with Krylov methods. A further disadvantage is that
the choice of ω is often diﬃcult to make. References [200], [89], [193], [8], and
the papers cited therein provide additional reading on this topic.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
10 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
1.5. Exercises on stationary iterative methods
1.5.1. Show that if ρ(M) ≥ 1 then there are x
0
and c such that the iteration
(1.7) fails to converge.
1.5.2. Prove Theorem 1.3.2.
1.5.3. Verify equality (1.18).
1.5.4. Show that if A is symmetric and positive deﬁnite (that is A
T
= A and
x
T
Ax > 0 for all x ,= 0) that B
SGS
is also symmetric and positive
deﬁnite.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Chapter 2
Conjugate Gradient Iteration
2.1. Krylov methods and the minimization property
In the following two chapters we describe some of the Krylov space methods
for linear equations. Unlike the stationary iterative methods, Krylov methods
do not have an iteration matrix. The two such methods that we’ll discuss in
depth, conjugate gradient and GMRES, minimize, at the kth iteration, some
measure of error over the aﬃne space
x
0
+/
k
,
where x
0
is the initial iterate and the kth Krylov subspace /
k
is
/
k
= span(r
0
, Ar
0
, . . . , A
k−1
r
0
)
for k ≥ 1.
The residual is
r = b −Ax.
So ¦r
k
¦
k≥0
will denote the sequence of residuals
r
k
= b −Ax
k
.
As in Chapter 1, we assume that A is a nonsingular N N matrix and let
x
∗
= A
−1
b.
There are other Krylov methods that are not as well understood as CG or
GMRES. Brief descriptions of several of these methods and their properties
are in ¸ 3.6, [12], and [78].
The conjugate gradient (CG) iteration was invented in the 1950s [103] as a
direct method. It has come into wide use over the last 15 years as an iterative
method and has generally superseded the Jacobi–Gauss–Seidel–SOR family of
methods.
CG is intended to solve symmetric positive deﬁnite (spd) systems. Recall
that A is symmetric if A = A
T
and positive deﬁnite if
x
T
Ax > 0 for all x ,= 0.
11
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
12 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
In this section we assume that A is spd. Since A is spd we may deﬁne a norm
(you should check that this is a norm) by
x
A
=
√
x
T
Ax. (2.1)
 
A
is called the Anorm. The development in these notes is diﬀerent from
the classical work and more like the analysis for GMRES and CGNR in [134].
In this section, and in the section on GMRES that follows, we begin with a
description of what the algorithm does and the consequences of the minimiza
tion property of the iterates. After that we describe termination criterion,
performance, preconditioning, and at the very end, the implementation.
The kth iterate x
k
of CG minimizes
φ(x) =
1
2
x
T
Ax −x
T
b (2.2)
over x
0
+/
k
.
Note that if φ(˜ x) is the minimal value (in R
N
) then
∇φ(˜ x) = A˜ x −b = 0
and hence ˜ x = x
∗
.
Minimizing φ over any subset of R
N
is the same as minimizing x −x
∗

A
over that subset. We state this as a lemma.
Lemma 2.1.1. Let o ⊂ R
N
. If x
k
minimizes φ over o then x
k
also
minimizes x
∗
−x
A
= r
A
−1 over o.
Proof. Note that
x −x
∗

2
A
= (x −x
∗
)
T
A(x −x
∗
) = x
T
Ax −x
T
Ax
∗
−(x
∗
)
T
Ax + (x
∗
)
T
Ax
∗
.
Since A is symmetric and Ax
∗
= b
−x
T
Ax
∗
−(x
∗
)
T
Ax = −2x
T
Ax
∗
= −2x
T
b.
Therefore
x −x
∗

2
A
= 2φ(x) + (x
∗
)
T
Ax
∗
.
Since (x
∗
)
T
Ax
∗
is independent of x, minimizing φ is equivalent to minimizing
x −x
∗

2
A
and hence to minimizing x −x
∗

A
.
If e = x −x
∗
then
e
2
A
= e
T
Ae = (A(x −x
∗
))
T
A
−1
(A(x −x
∗
)) = b −Ax
2
A
−1
and hence the Anorm of the error is also the A
−1
norm of the residual.
We will use this lemma in the particular case that o = x
0
+/
k
for some k.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 13
2.2. Consequences of the minimization property
Lemma 2.1.1 implies that since x
k
minimizes φ over x
0
+/
k
x
∗
−x
k

A
≤ x
∗
−w
A
(2.3)
for all w ∈ x
0
+/
k
. Since any w ∈ x
0
+/
k
can be written as
w =
k−1
j=0
γ
j
A
j
r
0
+x
0
for some coeﬃcients ¦γ
j
¦, we can express x
∗
−w as
x
∗
−w = x
∗
−x
0
−
k−1
j=0
γ
j
A
j
r
0
.
Since Ax
∗
= b we have
r
0
= b −Ax
0
= A(x
∗
−x
0
)
and therefore
x
∗
−w = x
∗
−x
0
−
k−1
j=0
γ
j
A
j+1
(x
∗
−x
0
) = p(A)(x
∗
−x
0
),
where the polynomial
p(z) = 1 −
k−1
j=0
γ
j
z
j+1
has degree k and satisﬁes p(0) = 1. Hence
x
∗
−x
k

A
= min
p∈P
k
,p(0)=1
p(A)(x
∗
−x
0
)
A
. (2.4)
In (2.4) P
k
denotes the set of polynomials of degree k.
The spectral theorem for spd matrices asserts that
A = UΛU
T
,
where U is an orthogonal matrix whose columns are the eigenvectors of A and
Λ is a diagonal matrix with the positive eigenvalues of A on the diagonal. Since
UU
T
= U
T
U = I by orthogonality of U, we have
A
j
= UΛ
j
U
T
.
Hence
p(A) = Up(Λ)U
T
.
Deﬁne A
1/2
= UΛ
1/2
U
T
and note that
x
2
A
= x
T
Ax = A
1/2
x
2
2
. (2.5)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
14 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Hence, for any x ∈ R
N
and
p(A)x
A
= A
1/2
p(A)x
2
≤ p(A)
2
A
1/2
x
2
≤ p(A)
2
x
A
.
This, together with (2.4) implies that
x
k
−x
∗

A
≤ x
0
−x
∗

A
min
p∈P
k
,p(0)=1
max
z∈σ(A)
[p(z)[. (2.6)
Here σ(A) is the set of all eigenvalues of A.
The following corollary is an important consequence of (2.6).
Corollary 2.2.1. Let A be spd and let ¦x
k
¦ be the CG iterates. Let k be
given and let ¦¯ p
k
¦ be any kth degree polynomial such that ¯ p
k
(0) = 1. Then
x
k
−x
∗

A
x
0
−x
∗

A
≤ max
z∈σ(A)
[ ¯ p
k
(z)[. (2.7)
We will refer to the polynomial ¯ p
k
as a residual polynomial [185].
Definition 2.2.1. The set of kth degree residual polynomials is
T
k
= ¦p [ p is a polynomial of degree k and p(0) = 1.¦ (2.8)
In speciﬁc contexts we try to construct sequences of residual polynomials,
based on information on σ(A), that make either the middle or the right term
in (2.7) easy to evaluate. This leads to an upper estimate for the number of
CG iterations required to reduce the Anorm of the error to a given tolerance.
One simple application of (2.7) is to show how the CG algorithm can be
viewed as a direct method.
Theorem 2.2.1. Let A be spd. Then the CG algorithm will ﬁnd the
solution within N iterations.
Proof. Let ¦λ
i
¦
N
i=1
be the eigenvalues of A. As a test polynomial, let
¯ p(z) =
N
i=1
(λ
i
−z)/λ
i
.
¯ p ∈ T
N
because ¯ p has degree N and ¯ p(0) = 1. Hence, by (2.7) and the fact
that ¯ p vanishes on σ(A),
x
N
−x
∗

A
≤ x
0
−x
∗

A
max
z∈σ(A)
[ ¯ p(z)[ = 0.
Note that our test polynomial had the eigenvalues of A as its roots. In
that way we showed (in the absence of all roundoﬀ error!) that CG terminated
in ﬁnitely many iterations with the exact solution. This is not as good as it
sounds, since in most applications the number of unknowns N is very large,
and one cannot aﬀord to perform N iterations. It is best to regard CG as an
iterative method. When doing that we seek to terminate the iteration when
some speciﬁed error tolerance is reached.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 15
In the two examples that follow we look at some other easy consequences
of (2.7).
Theorem 2.2.2. Let A be spd with eigenvectors ¦u
i
¦
N
i=1
. Let b be a linear
combination of k of the eigenvectors of A
b =
k
l=1
γ
l
u
i
l
.
Then the CG iteration for Ax = b with x
0
= 0 will terminate in at most k
iterations.
Proof. Let ¦λ
i
l
¦ be the eigenvalues of A associated with the eigenvectors
¦u
i
l
¦
k
l=1
. By the spectral theorem
x
∗
=
k
l=1
(γ
l
/λ
i
l
)u
i
l
.
We use the residual polynomial,
¯ p(z) =
k
l=1
(λ
i
l
−z)/λ
i
l
.
One can easily verify that ¯ p ∈ T
k
. Moreover, ¯ p(λ
i
l
) = 0 for 1 ≤ l ≤ k and
hence
¯ p(A)x
∗
=
k
l=1
¯ p(λ
i
l
)γ
l
/λ
i
l
u
i
l
= 0.
So, we have by (2.4) and the fact that x
0
= 0 that
x
k
−x
∗

A
≤ ¯ p(A)x
∗

A
= 0.
This completes the proof.
If the spectrum of A has fewer than N points, we can use a similar technique
to prove the following theorem.
Theorem 2.2.3. Let A be spd. Assume that there are exactly k ≤ N
distinct eigenvalues of A. Then the CG iteration terminates in at most k
iterations.
2.3. Termination of the iteration
In practice we do not run the CG iteration until an exact solution is found, but
rather terminate once some criterion has been satisﬁed. One typical criterion is
small (say ≤ η) relative residuals. This means that we terminate the iteration
after
b −Ax
k

2
≤ ηb
2
. (2.9)
The error estimates that come from the minimization property, however, are
based on (2.7) and therefore estimate the reduction in the relative Anorm of
the error.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
16 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Our next task is to relate the relative residual in the Euclidean norm to
the relative error in the Anorm. We will do this in the next two lemmas and
then illustrate the point with an example.
Lemma 2.3.1. Let A be spd with eigenvalues λ
1
≥ λ
2
≥ . . . λ
N
. Then for
all z ∈ R
N
,
A
1/2
z
2
= z
A
(2.10)
and
λ
1/2
N
z
A
≤ Az
2
≤ λ
1/2
1
z
A
. (2.11)
Proof. Clearly
z
2
A
= z
T
Az = (A
1/2
z)
T
(A
1/2
z) = A
1/2
z
2
2
which proves (2.10).
Let u
i
be a unit eigenvector corresponding to λ
i
. We may write A = UΛU
T
as
Az =
N
i=1
λ
i
(u
T
i
z)u
i
.
Hence
λ
N
A
1/2
z
2
2
= λ
N
N
i=1
λ
i
(u
T
i
z)
2
≤ Az
2
2
=
N
i=1
λ
2
i
(u
T
i
z)
2
≤ λ
1
N
i=1
λ
i
(u
T
i
z)
2
= λ
1
A
1/2
z
2
2
.
Taking square roots and using (2.10) complete the proof.
Lemma 2.3.2.
b
2
r
0

2
b −Ax
k

2
b
2
=
b −Ax
k

2
b −Ax
0

2
≤
_
κ
2
(A)
x
k
−x
∗

A
x
∗
−x
0

A
(2.12)
and
b −Ax
k

2
b
2
≤
_
κ
2
(A)r
0

2
b
2
x
k
−x
∗

A
x
∗
−x
0

A
. (2.13)
Proof. The equality on the left of (2.12) is clear and (2.13) follows directly
from (2.12). To obtain the inequality on the right of (2.12), ﬁrst recall that if
A = UΛU
T
is the spectral decomposition of A and we order the eigenvalues
such that λ
1
≥ λ
2
≥ . . . λ
N
> 0, then A
2
= λ
1
and A
−1

2
= 1/λ
N
. So
κ
2
(A) = λ
1
/λ
N
.
Therefore, using (2.10) and (2.11) twice,
b −Ax
k

2
b −Ax
0

2
=
A(x
∗
−x
k
)
2
A(x
∗
−x
0
)
2
≤
¸
λ
1
λ
N
x
∗
−x
k

A
x
∗
−x
0

A
as asserted.
So, to predict the performance of the CG iteration based on termination on
small relative residuals, we must not only use (2.7) to predict when the relative
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 17
Anorm error is small, but also use Lemma 2.3.2 to relate small Anorm errors
to small relative residuals.
We consider a very simple example. Assume that x
0
= 0 and that the
eigenvalues of A are contained in the interval (9, 11). If we let ¯ p
k
(z) =
(10 −z)
k
/10
k
, then ¯ p
k
∈ T
k
. This means that we may apply (2.7) to get
x
k
−x
∗

A
≤ x
∗

A
max
9≤z≤11
[ ¯ p
k
(z)[.
It is easy to see that
max
9≤z≤11
[ ¯ p
k
(z)[ = 10
−k
.
Hence, after k iterations
x
k
−x
∗

A
≤ x
∗

A
10
−k
. (2.14)
So, the size of the Anorm of the error will be reduced by a factor of 10
−3
when
10
−k
≤ 10
−3
,
that is, when
k ≥ 3.
To use Lemma 2.3.2 we simply note that κ
2
(A) ≤ 11/9. Hence, after k
iterations we have
Ax
k
−b
2
b
2
≤
√
11 10
−k
/3.
So, the size of the relative residual will be reduced by a factor of 10
−3
when
10
−k
≤ 3 10
−3
/
√
11,
that is, when
k ≥ 4.
One can obtain a more precise estimate by using a polynomial other than
p
k
in the upper estimate for the righthand side of (2.7). Note that it is always
the case that the spectrum of a spd matrix is contained in the interval [λ
N
, λ
1
]
and that κ
2
(A) = λ
1
/λ
N
. A result from [48] (see also [45]) that is, in one
sense, the sharpest possible, is
x
k
−x
∗

A
≤ 2x
0
−x
∗

A
_
_
κ
2
(A) −1
_
κ
2
(A) + 1
_
k
. (2.15)
In the case of the above example, we can estimate κ
2
(A) by κ
2
(A) ≤ 11/9.
Hence, since (
√
x −1)/(
√
x + 1) is an increasing function of x on the interval
(1, ∞).
_
κ
2
(A) −1
_
κ
2
(A) + 1
≤
√
11 −3
√
11 + 3
≈ .05.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
18 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Therefore (2.15) would predict a reduction in the size of the Anorm error by
a factor of 10
−3
when
2 .05
k
< 10
−3
or when
k > −log
10
(2000)/ log
10
(.05) ≈ 3.3/1.3 ≈ 2.6,
which also predicts termination within three iterations.
We may have more precise information than a single interval containing
σ(A). When we do, the estimate in (2.15) can be very pessimistic. If the
eigenvalues cluster in a small number of intervals, the condition number can
be quite large, but CG can perform very well. We will illustrate this with an
example. Exercise 2.8.5 also covers this point.
Assume that x
0
= 0 and the eigenvalues of A lie in the two intervals (1, 1.5)
and (399, 400). Based on this information the best estimate of the condition
number of A is κ
2
(A) ≤ 400, which, when inserted into (2.15) gives
x
k
−x
∗

A
x
∗

A
≤ 2 (19/21)
k
≈ 2 (.91)
k
.
This would indicate fairly slow convergence. However, if we use as a residual
polynomial ¯ p
3k
∈ T
3k
¯ p
3k
(z) =
(1.25 −z)
k
(400 −z)
2k
(1.25)
k
400
2k
.
It is easy to see that
max
z∈σ(A)
[ ¯ p
3k
(z)[ ≤ (.25/1.25)
k
= (.2)
k
,
which is a sharper estimate on convergence. In fact, (2.15) would predict that
x
k
−x
∗

A
≤ 10
−3
x
∗

A
,
when 2 (.91)
k
< 10
−3
or when
k > −log
10
(2000)/ log
10
(.91) ≈ 3.3/.04 = 82.5.
The estimate based on the clustering gives convergence in 3k iterations when
(.2)
k
≤ 10
−3
or when
k > −3/ log
10
(.2) = 4.3.
Hence (2.15) predicts 83 iterations and the clustering analysis 15 (the smallest
integer multiple of 3 larger than 3 4.3 = 12.9).
From the results above one can see that if the condition number of A is near
one, the CG iteration will converge very rapidly. Even if the condition number
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 19
is large, the iteration will perform well if the eigenvalues are clustered in a few
small intervals. The transformation of the problem into one with eigenvalues
clustered near one (i.e., easier to solve) is called preconditioning. We used
this term before in the context of Richardson iteration and accomplished the
goal by multiplying A by an approximate inverse. In the context of CG, such
a simple approach can destroy the symmetry of the coeﬃcient matrix and a
more subtle implementation is required. We discuss this in ¸ 2.5.
2.4. Implementation
The implementation of CG depends on the amazing fact that once x
k
has been
determined, either x
k
= x
∗
or a search direction p
k+1
,= 0 can be found very
cheaply so that x
k+1
= x
k
+ α
k+1
p
k+1
for some scalar α
k+1
. Once p
k+1
has
been found, α
k+1
is easy to compute from the minimization property of the
iteration. In fact
dφ(x
k
+αp
k+1
)
dα
= 0 (2.16)
for the correct choice of α = α
k+1
. Equation (2.16) can be written as
p
T
k+1
Ax
k
+αp
T
k+1
Ap
k+1
−p
T
k+1
b = 0
leading to
α
k+1
=
p
T
k+1
(b −Ax
k
)
p
T
k+1
Ap
k+1
=
p
T
k+1
r
k
p
T
k+1
Ap
k+1
. (2.17)
If x
k
= x
k+1
then the above analysis implies that α = 0. We show that
this only happens if x
k
is the solution.
Lemma 2.4.1. Let A be spd and let ¦x
k
¦ be the conjugate gradient iterates.
Then
r
T
k
r
l
= 0 for all 0 ≤ l < k. (2.18)
Proof. Since x
k
minimizes φ on x
0
+/
k
, we have, for any ξ ∈ /
k
dφ(x
k
+tξ)
dt
= ∇φ(x
k
+tξ)
T
ξ = 0
at t = 0. Recalling that
∇φ(x) = Ax −b = −r
we have
∇φ(x
k
)
T
ξ = −r
T
k
ξ = 0 for all ξ ∈ /
k
. (2.19)
Since r
l
∈ /
k
for all l < k (see Exercise 2.8.1), this proves (2.18).
Now, if x
k
= x
k+1
, then r
k
= r
k+1
. Lemma 2.4.1 then implies that
r
k

2
2
= r
T
k
r
k
= r
T
k
r
k+1
= 0 and hence x
k
= x
∗
.
The next lemma characterizes the search direction and, as a side eﬀect,
proves that (if we deﬁne p
0
= 0) p
T
l
r
k
= 0 for all 0 ≤ l < k ≤ n, unless the
iteration terminates prematurely.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
20 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Lemma 2.4.2. Let A be spd and let ¦x
k
¦ be the conjugate gradient iterates.
If x
k
,= x
∗
then x
k+1
= x
k
+ α
k+1
p
k+1
and p
k+1
is determined up to a scalar
multiple by the conditions
p
k+1
∈ /
k+1
, p
T
k+1
Aξ = 0 for all ξ ∈ /
k
. (2.20)
Proof. Since /
k
⊂ /
k+1
,
∇φ(x
k+1
)
T
ξ = (Ax
k
+α
k+1
Ap
k+1
−b)
T
ξ = 0 (2.21)
for all ξ ∈ /
k
. (2.19) and (2.21) then imply that for all ξ ∈ /
k
,
α
k+1
p
T
k+1
Aξ = −(Ax
k
−b)
T
ξ = −∇φ(x
k
)
T
ξ = 0. (2.22)
This uniquely speciﬁes the direction of p
k+1
as (2.22) implies that p
k+1
∈ /
k+1
is Aorthogonal (i.e., in the scalar product (x, y) = x
T
Ay) to /
k
, a subspace
of dimension one less than /
k+1
.
The condition p
T
k+1
Aξ = 0 is called Aconjugacy of p
k+1
to /
k
. Now, any
p
k+1
satisfying (2.20) can, up to a scalar multiple, be expressed as
p
k+1
= r
k
+w
k
with w
k
∈ /
k
. While one might think that w
k
would be hard to compute, it
is, in fact, trivial. We have the following theorem.
Theorem 2.4.1. Let A be spd and assume that r
k
,= 0. Deﬁne p
0
= 0.
Then
p
k+1
= r
k
+β
k+1
p
k
for some β
k+1
and k ≥ 0. (2.23)
Proof. By Lemma 2.4.2 and the fact that /
k
= span(r
0
, . . . , r
k−1
), we need
only verify that a β
k+1
can be found so that if p
k+1
is given by (2.23) then
p
T
k+1
Ar
l
= 0
for all 0 ≤ l ≤ k −1.
Let p
k+1
be given by (2.23). Then for any l ≤ k
p
T
k+1
Ar
l
= r
T
k
Ar
l
+β
k+1
p
T
k
Ar
l
.
If l ≤ k −2, then r
l
∈ /
l+1
⊂ /
k−1
. Lemma 2.4.2 then implies that
p
T
k+1
Ar
l
= 0 for 0 ≤ l ≤ k −2.
It only remains to solve for β
k+1
so that p
T
k+1
Ar
k−1
= 0. Trivially
β
k+1
= −r
T
k
Ar
k−1
/p
T
k
Ar
k−1
(2.24)
provided p
T
k
Ar
k−1
,= 0. Since
r
k
= r
k−1
−α
k
Ap
k
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 21
we have
r
T
k
r
k−1
= r
k−1

2
2
−α
k
p
T
k
Ar
k−1
.
Since r
T
k
r
k−1
= 0 by Lemma 2.4.1 we have
p
T
k
Ar
k−1
= r
k−1

2
2
/α
k
,= 0. (2.25)
This completes the proof.
The common implementation of conjugate gradient uses a diﬀerent form
for α
k
and β
k
than given in (2.17) and (2.24).
Lemma 2.4.3. Let A be spd and assume that r
k
,= 0. Then
α
k+1
=
r
k

2
2
p
T
k+1
Ap
k+1
(2.26)
and
β
k+1
=
r
k

2
2
r
k−1

2
2
. (2.27)
Proof. Note that for k ≥ 0
p
T
k+1
r
k+1
= r
T
k
r
k+1
+β
k+1
p
T
k
r
k+1
= 0 (2.28)
by Lemma 2.4.2. An immediate consequence of (2.28) is that p
T
k
r
k
= 0 and
hence
p
T
k+1
r
k
= (r
k
+β
k+1
p
k
)
T
r
k
= r
k

2
2
. (2.29)
Taking scalar products of both sides of
r
k+1
= r
k
−α
k+1
Ap
k+1
with p
k+1
and using (2.29) gives
0 = p
T
k+1
r
k
−α
k+1
p
T
k+1
Ap
k+1
= r
T
k

2
2
−α
k+1
p
T
k+1
Ap
k+1
,
which is equivalent to (2.26).
To get (2.27) note that p
T
k+1
Ap
k
= 0 and hence (2.23) implies that
β
k+1
=
−r
T
k
Ap
k
p
T
k
Ap
k
. (2.30)
Also note that
p
T
k
Ap
k
= p
T
k
A(r
k−1
+β
k
p
k−1
)
= p
T
k
Ar
k−1
+β
k
p
T
k
Ap
k−1
= p
T
k
Ar
k−1
.
(2.31)
Now combine (2.30), (2.31), and (2.25) to get
β
k+1
=
−r
T
k
Ap
k
α
k
r
k−1

2
2
.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
22 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Now take scalar products of both sides of
r
k
= r
k−1
−α
k
Ap
k
with r
k
and use Lemma 2.4.1 to get
r
k

2
2
= −α
k
r
T
k
Ap
k
.
Hence (2.27) holds.
The usual implementation reﬂects all of the above results. The goal is to
ﬁnd, for a given , a vector x so that b−Ax
2
≤ b
2
. The input is the initial
iterate x, which is overwritten with the solution, the right hand side b, and a
routine which computes the action of A on a vector. We limit the number of
iterations to kmax and return the solution, which overwrites the initial iterate
x and the residual norm.
Algorithm 2.4.1. cg(x, b, A, , kmax)
1. r = b −Ax, ρ
0
= r
2
2
, k = 1.
2. Do While
√
ρ
k−1
> b
2
and k < kmax
(a) if k = 1 then p = r
else
β = ρ
k−1
/ρ
k−2
and p = r +βp
(b) w = Ap
(c) α = ρ
k−1
/p
T
w
(d) x = x +αp
(e) r = r −αw
(f) ρ
k
= r
2
2
(g) k = k + 1
Note that the matrix A itself need not be formed or stored, only a routine
for matrixvector products is required. Krylov space methods are often called
matrixfree for that reason.
Now, consider the costs. We need store only the four vectors x, w, p, and r.
Each iteration requires a single matrixvector product (to compute w = Ap),
two scalar products (one for p
T
w and one to compute ρ
k
= r
2
2
), and three
operations of the form ax +y, where x and y are vectors and a is a scalar.
It is remarkable that the iteration can progress without storing a basis for
the entire Krylov subspace. As we will see in the section on GMRES, this is
not the case in general. The spd structure buys quite a lot.
2.5. Preconditioning
To reduce the condition number, and hence improve the performance of the
iteration, one might try to replace Ax = b by another spd system with the
same solution. If M is a spd matrix that is close to A
−1
, then the eigenvalues
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 23
of MA will be clustered near one. However MA is unlikely to be spd, and
hence CG cannot be applied to the system MAx = Mb.
In theory one avoids this diﬃculty by expressing the preconditioned
problem in terms of B, where B is spd, A = B
2
, and by using a twosided
preconditioner, S ≈ B
−1
(so M = S
2
). Then the matrix SAS is spd and its
eigenvalues are clustered near one. Moreover the preconditioned system
SASy = Sb
has y
∗
= S
−1
x
∗
as a solution, where Ax
∗
= b. Hence x
∗
can be recovered
from y
∗
by multiplication by S. One might think, therefore, that computing
S (or a subroutine for its action on a vector) would be necessary and that a
matrixvector multiply by SAS would incur a cost of one multiplication by A
and two by S. Fortunately, this is not the case.
If y
k
, ˆ r
k
, ˆ p
k
are the iterate, residual, and search direction for CG applied
to SAS and we let
x
k
= Sˆ y
k
, r
k
= S
−1
ˆ r
k
, p
k
= Sˆ p
k
, and z
k
= Sˆ r
k
,
then one can perform the iteration directly in terms of x
k
, A, and M. The
reader should verify that the following algorithm does exactly that. The input
is the same as that for Algorithm cg and the routine to compute the action of
the preconditioner on a vector. Aside from the preconditioner, the arguments
to pcg are the same as those to Algorithm cg.
Algorithm 2.5.1. pcg(x, b, A, M, , kmax)
1. r = b −Ax, ρ
0
= r
2
2
, k = 1
2. Do While
√
ρ
k−1
> b
2
and k < kmax
(a) z = Mr
(b) τ
k−1
= z
T
r
(c) if k = 1 then β = 0 and p = z
else
β = τ
k−1
/τ
k−2
, p = z +βp
(d) w = Ap
(e) α = τ
k−1
/p
T
w
(f) x = x +αp
(g) r = r −αw
(h) ρ
k
= r
T
r
(i) k = k + 1
Note that the cost is identical to CG with the addition of
• the application of the preconditioner M in step 2a and
• the additional inner product required to compute τ
k
in step 2b.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
24 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Of these costs, the application of the preconditioner is usually the larger. In
the remainder of this section we brieﬂy mention some classes of preconditioners.
A more complete and detailed discussion of preconditioners is in [8] and a
concise survey with many pointers to the literature is in [12].
Some eﬀective preconditioners are based on deep insight into the structure
of the problem. See [124] for an example in the context of partial diﬀerential
equations, where it is shown that certain discretized secondorder elliptic
problems on simple geometries can be very well preconditioned with fast
Poisson solvers [99], [188], and [187]. Similar performance can be obtained from
multigrid [99], domain decomposition, [38], [39], [40], and alternating direction
preconditioners [8], [149], [193], [194]. We use a Poisson solver preconditioner
in the examples in ¸ 2.7 and ¸ 3.7 as well as for nonlinear problems in ¸ 6.4.2
and ¸ 8.4.2.
One commonly used and easily implemented preconditioner is Jacobi
preconditioning, where M is the inverse of the diagonal part of A. One can also
use other preconditioners based on the classical stationary iterative methods,
such as the symmetric Gauss–Seidel preconditioner (1.18). For applications to
partial diﬀerential equations, these preconditioners may be somewhat useful,
but should not be expected to have dramatic eﬀects.
Another approach is to apply a sparse Cholesky factorization to the
matrix A (thereby giving up a fully matrixfree formulation) and discarding
small elements of the factors and/or allowing only a ﬁxed amount of storage
for the factors. Such preconditioners are called incomplete factorization
preconditioners. So if A = LL
T
+ E, where E is small, the preconditioner
is (LL
T
)
−1
and its action on a vector is done by two sparse triangular solves.
We refer the reader to [8], [127], and [44] for more detail.
One could also attempt to estimate the spectrum of A, ﬁnd a polynomial
p such that 1 −zp(z) is small on the approximate spectrum, and use p(A) as a
preconditioner. This is called polynomial preconditioning. The preconditioned
system is
p(A)Ax = p(A)b
and we would expect the spectrum of p(A)A to be more clustered near z = 1
than that of A. If an interval containing the spectrum can be found, the
residual polynomial q(z) = 1 − zp(z) of smallest L
∞
norm on that interval
can be expressed in terms of Chebyshev [161] polynomials. Alternatively
q can be selected to solve a least squares minimization problem [5], [163].
The preconditioning p can be directly recovered from q and convergence rate
estimates made. This technique is used to prove the estimate (2.15), for
example. The cost of such a preconditioner, if a polynomial of degree K is
used, is K matrixvector products for each application of the preconditioner
[5]. The performance gains can be very signiﬁcant and the implementation is
matrixfree.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 25
2.6. CGNR and CGNE
If A is nonsingular and nonsymmetric, one might consider solving Ax = b by
applying CG to the normal equations
A
T
Ax = A
T
b. (2.32)
This approach [103] is called CGNR [71], [78], [134]. The reason for this name
is that the minimization property of CG as applied to (2.32) asserts that
x
∗
−x
2
A
T
A
= (x
∗
−x)
T
A
T
A(x
∗
−x)
= (Ax
∗
−Ax)
T
(Ax
∗
−Ax) = (b −Ax)
T
(b −Ax) = r
2
is minimized over x
0
+/
k
at each iterate. Hence the name Conjugate Gradient
on the Normal equations to minimize the Residual.
Alternatively, one could solve
AA
T
y = b (2.33)
and then set x = A
T
y to solve Ax = b. This approach [46] is now called CGNE
[78], [134]. The reason for this name is that the minimization property of CG
as applied to (2.33) asserts that if y
∗
is the solution to (2.33) then
y
∗
−y
2
AA
T
= (y
∗
−y)
T
(AA
T
)(y
∗
−y) = (A
T
y
∗
−A
T
y)
T
(A
T
y
∗
−A
T
y)
= x
∗
−x
2
2
is minimized over y
0
+/
k
at each iterate. Conjugate Gradient on the Normal
equations to minimize the Error.
The advantages of this approach are that all the theory for CG carries over
and the simple implementation for both CG and PCG can be used. There
are three disadvantages that may or may not be serious. The ﬁrst is that the
condition number of the coeﬃcient matrix A
T
A is the square of that of A.
The second is that two matrixvector products are needed for each CG iterate
since w = A
T
Ap = A
T
(Ap) in CGNR and w = AA
T
p = A(A
T
p) in CGNE.
The third, more important, disadvantage is that one must compute the action
of A
T
on a vector as part of the matrixvector product involving A
T
A. As we
will see in the chapter on nonlinear problems, there are situations where this
is not possible.
The analysis with residual polynomials is similar to that for CG. We
consider the case for CGNR, the analysis for CGNE is essentially the same.
As above, when we consider the A
T
A norm of the error, we have
x
∗
−x
2
A
T
A
= (x
∗
−x)
T
A
T
A(x
∗
−x) = A(x
∗
−x)
2
2
= r
2
2
.
Hence, for any residual polynomial ¯ p
k
∈ T
k
,
r
k

2
≤ ¯ p
k
(A
T
A)r
0

2
≤ r
0

2
max
z∈σ(A
T
A)
[ ¯ p
k
(z)[. (2.34)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
26 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
There are two major diﬀerences between (2.34) and (2.7). The estimate is
in terms of the l
2
norm of the residual, which corresponds exactly to the
termination criterion, hence we need not prove a result like Lemma 2.3.2. Most
signiﬁcantly, the residual polynomial is to be maximized over the eigenvalues
of A
T
A, which is the set of the squares of the singular values of A. Hence the
performance of CGNR and CGNE is determined by the distribution of singular
values.
2.7. Examples for preconditioned conjugate iteration
In the collection of MATLAB codes we provide a code for preconditioned
conjugate gradient iteration. The inputs, described in the comment lines,
are the initial iterate, x
0
, the right hand side vector b, MATLAB functions for
the matrixvector product and (optionally) the preconditioner, and iteration
parameters to specify the maximum number of iterations and the termination
criterion. On return the code supplies the approximate solution x and the
history of the iteration as the vector of residual norms.
We consider the discretization of the partial diﬀerential equation
−∇
˙
(a(x, y)∇u) = f(x, y) (2.35)
on 0 < x, y < 1 subject to homogeneous Dirichlet boundary conditions
u(x, 0) = u(x, 1) = u(0, y) = u(1, y) = 0, 0 < x, y < 1.
One can verify [105] that the diﬀerential operator is positive deﬁnite in the
Hilbert space sense and that the ﬁvepoint discretization described below is
positive deﬁnite if a > 0 for all 0 ≤ x, y ≤ 1 (Exercise 2.8.10).
We discretize with a ﬁvepoint centered diﬀerence scheme with n
2
points
and mesh width h = 1/(n + 1). The unknowns are
u
ij
≈ u(x
i
, x
j
)
where x
i
= ih for 1 ≤ i ≤ n. We set
u
0j
= u
(n+1)j
= u
i0
= u
i(n+1)
= 0,
to reﬂect the boundary conditions, and deﬁne
α
ij
= −a(x
i
, x
j
)h
−2
/2.
We express the discrete matrixvector product as
(Au)
ij
= (α
ij
+α
(i+1)j
)(u
(i+1)j
−u
ij
)
−(α
(i−1)j
+α
ij
)(u
ij
−u
(i−1)j
) + (α
i(j+1)
+α
ij
)(u
i(j+1)
−u
ij
)
−(α
ij
+α
i(j−1)
)(u
ij
−u
i(j−1)
)
(2.36)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 27
for 1 ≤ i, j ≤ n.
For the MATLAB implementation we convert freely from the representa
tion of u as a twodimensional array (with the boundary conditions added),
which is useful for computing the action of A on u and applying fast solvers,
and the representation as a onedimensional array, which is what pcgsol ex
pects to see. See the routine fish2d in collection of MATLAB codes for an
example of how to do this in MATLAB.
For the computations reported in this section we took a(x, y) = cos(x) and
took the right hand side so that the exact solution was the discretization of
10xy(1 −x)(1 −y) exp(x
4.5
).
The initial iterate was u
0
= 0.
In the results reported here we took n = 31 resulting in a system with
N = n
2
= 961 unknowns. We expect secondorder accuracy from the method
and accordingly we set termination parameter = h
2
= 1/1024. We allowed
up to 100 CG iterations. The initial iterate was the zero vector. We will report
our results graphically, plotting r
k

2
/b
2
on a semilog scale.
In Fig. 2.1 the solid line is a plot of r
k
/b and the dashed line a plot of
u
∗
−u
k

A
/u
∗
−u
0

A
. Note that the reduction in r is not monotone. This is
consistent with the theory, which predicts decrease in e
A
but not necessarily
in r as the iteration progresses. Note that the unpreconditioned iteration is
slowly convergent. This can be explained by the fact that the eigenvalues are
not clustered and
κ(A) = O(1/h
2
) = O(n
2
) = O(N)
and hence (2.15) indicates that convergence will be slow. The reader is asked
to quantify this in terms of execution times in Exercise 2.8.9. This example
illustrates the importance of a good preconditioner. Even the unpreconditioned
iteration, however, is more eﬃcient that the classical stationary iterative
methods.
For a preconditioner we use a Poisson solver. By this we mean an operator
G such that v = Gw is the solution of the discrete form of
−v
xx
−v
yy
= w,
subject to homogeneous Dirichlet boundary conditions. The eﬀectiveness of
such a preconditioner has been analyzed in [124] and some of the many ways
to implement the solver eﬃciently are discussed in [99], [188], [186], and [187].
The properties of CG on the preconditioned problem in the continuous
case have been analyzed in [48]. For many types of domains and boundary
conditions, Poisson solvers can be designed to take advantage of vector and/or
parallel architectures or, in the case of the MATLAB environment used in
this book, designed to take advantage of fast MATLAB builtin functions.
Because of this their execution time is less than a simple count of ﬂoating
point operations would indicate. The fast Poisson solver in the collection of
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
28 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
0 10 20 30 40 50 60
10
4
10
3
10
2
10
1
10
0
10
1
Iterations
R
e
l
a
t
i
v
e
R
e
s
i
d
u
a
l
a
n
d
A

n
o
r
m
o
f
E
r
r
o
r
Fig. 2.1. CG for 2D elliptic equation.
0 10 20 30 40 50 60
10
4
10
3
10
2
10
1
10
0
10
1
Iterations
R
e
l
a
t
i
v
e
R
e
s
i
d
u
a
l
Fig. 2.2. PCG for 2D elliptic equation.
codes fish2d is based on the MATLAB fast Fourier transform, the builtin
function fft.
In Fig. 2.2 the solid line is the graph of r
k

2
/b
2
for the preconditioned
iteration and the dashed line for the unpreconditioned. The preconditioned
iteration required 5 iterations for convergence and the unpreconditioned
iteration 52. Not only does the preconditioned iteration converge more
rapidly, but the number of iterations required to reduce the relative residual
by a given amount is independent of the mesh spacing [124]. We caution
the reader that the preconditioned iteration is not as much faster than the
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 29
unpreconditioned one as the iteration count would suggest. The MATLAB
flops command indicates that the unpreconditioned iteration required roughly
1.2 million ﬂoatingpoint operations while the preconditioned iteration required
.87 million ﬂoatingpoint operations. Hence, the cost of the preconditioner is
considerable. In the MATLAB environment we used, the execution time of
the preconditioned iteration was about 60% of that of the unpreconditioned.
As we remarked above, this speed is a result of the eﬃciency of the MATLAB
fast Fourier transform. In Exercise 2.8.11 you are asked to compare execution
times for your own environment.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
30 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
2.8. Exercises on conjugate gradient
2.8.1. Let ¦x
k
¦ be the conjugate gradient iterates. Prove that r
l
∈ /
k
for all
l < k.
2.8.2. Let A be spd. Show that there is a spd B such that B
2
= A. Is B
unique?
2.8.3. Let Λ be a diagonal matrix with Λ
ii
= λ
i
and let p be a polynomial.
Prove that p(Λ) = max
i
[p(λ
i
)[ where   is any induced matrix norm.
2.8.4. Prove Theorem 2.2.3.
2.8.5. Assume that A is spd and that
σ(A) ⊂ (1, 1.1) ∪ (2, 2.2).
Give upper estimates based on (2.6) for the number of CG iterations
required to reduce the A norm of the error by a factor of 10
−3
and for
the number of CG iterations required to reduce the residual by a factor
of 10
−3
.
2.8.6. For the matrix A in problem 5, assume that the cost of a matrix vector
multiply is 4N ﬂoatingpoint multiplies. Estimate the number of ﬂoating
point operations reduce the A norm of the error by a factor of 10
−3
using
CG iteration.
2.8.7. Let A be a nonsingular matrix with all singular values in the interval
(1, 2). Estimate the number of CGNR/CGNE iterations required to
reduce the relative residual by a factor of 10
−4
.
2.8.8. Show that if A has constant diagonal then PCG with Jacobi precondi
tioning produces the same iterates as CG with no preconditioning.
2.8.9. Assume that A is N N, nonsingular, and spd. If κ(A) = O(N), give
a rough estimate of the number of CG iterates required to reduce the
relative residual to O(1/N).
2.8.10. Prove that the linear transformation given by (2.36) is symmetric and
positive deﬁnite on R
n
2
if a(x, y) > 0 for all 0 ≤ x, y ≤ 1.
2.8.11. Duplicate the results in ¸ 2.7 for example, in MATLAB by writing the
matrixvector product routines and using the MATLAB codes pcgsol
and fish2d. What happens as N is increased? How are the performance
and accuracy aﬀected by changes in a(x, y)? Try a(x, y) =
√
.1 +x and
examine the accuracy of the result. Explain your ﬁndings. Compare
the execution times on your computing environment (using the cputime
command in MATLAB, for instance).
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
CONJUGATE GRADIENT ITERATION 31
2.8.12. Use the Jacobi and symmetric Gauss–Seidel iterations from Chapter 1
to solve the elliptic boundary value problem in ¸ 2.7. How does the
performance compare to CG and PCG?
2.8.13. Implement Jacobi (1.17) and symmetric Gauss–Seidel (1.18) precondi
tioners for the elliptic boundary value problem in ¸ 2.7. Compare the
performance with respect to both computer time and number of itera
tions to preconditioning with the Poisson solver.
2.8.14. Modify pcgsol so that φ(x) is computed and stored at each iterate
and returned on output. Plot φ(x
n
) as a function of n for each of the
examples.
2.8.15. Apply CG and PCG to solve the ﬁvepoint discretization of
−u
xx
(x, y) −u
yy
(x, y) +e
x+y
u(x, y) = 1, 0 < x, y, < 1,
subject to the inhomogeneous Dirichlet boundary conditions
u(x, 0) = u(x, 1) = u(1, y) = 0, u(0, y) = 1, 0 < x, y < 1.
Experiment with diﬀerent mesh sizes and preconditioners (Fast Poisson
solver, Jacobi, and symmetric Gauss–Seidel).
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
32 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Chapter 3
GMRES Iteration
3.1. The minimization property and its consequences
The GMRES (Generalized Minimum RESidual) was proposed in 1986 in [167]
as a Krylov subspace method for nonsymmetric systems. Unlike CGNR,
GMRES does not require computation of the action of A
T
on a vector. This is
a signiﬁcant advantage in many cases. The use of residual polynomials is made
more complicated because we cannot use the spectral theorem to decompose
A. Moreover, one must store a basis for /
k
, and therefore storage requirements
increase as the iteration progresses.
The kth (k ≥ 1) iteration of GMRES is the solution to the least squares
problem
minimize
x∈x
0
+K
k
b −Ax
2
. (3.1)
The beginning of this section is much like the analysis for CG. Note that
if x ∈ x
0
+/
k
then
x = x
0
+
k−1
j=0
γ
j
A
j
r
0
and so
b −Ax = b −Ax
0
−
k−1
j=0
γ
j
A
j+1
r
0
= r
0
−
k
j=1
γ
j−1
A
j
r
0
.
Hence if x ∈ x
0
+/
k
then r = ¯ p(A)r
0
where ¯ p ∈ T
k
is a residual polynomial.
We have just proved the following result.
Theorem 3.1.1. Let A be nonsingular and let x
k
be the kth GMRES
iteration. Then for all ¯ p
k
∈ T
k
r
k

2
= min
p∈P
k
¯ p(A)r
0

2
≤ ¯ p
k
(A)r
0

2
. (3.2)
From this we have the following corollary.
Corollary 3.1.1. Let A be nonsingular and let x
k
be the kth GMRES
iteration. Then for all ¯ p
k
∈ T
k
r
k

2
r
0

2
≤ ¯ p
k
(A)
2
. (3.3)
33
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
34 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
We can apply the corollary to prove ﬁnite termination of the GMRES
iteration.
Theorem 3.1.2. Let A be nonsingular. Then the GMRES algorithm will
ﬁnd the solution within N iterations.
Proof. The characteristic polynomial of A is p(z) = det(A − zI). p has
degree N, p(0) = det(A) ,= 0 since A is nonsingular, and so
¯ p
N
(z) = p(z)/p(0) ∈ T
N
is a residual polynomial. It is well known [141] that p(A) = ¯ p
N
(A) = 0. By
(3.3), r
N
= b −Ax
N
= 0 and hence x
N
is the solution.
In Chapter 2 we applied the spectral theorem to obtain more precise infor
mation on convergence rates. This is not an option for general nonsymmetric
matrices. However, if A is diagonalizable we may use (3.2) to get information
from clustering of the spectrum just like we did for CG. We pay a price in that
we must use complex arithmetic for the only time in this book. Recall that
A is diagonalizable if there is a nonsingular (possibly complex!) matrix V such
that
A = V ΛV
−1
.
Here Λ is a (possibly complex!) diagonal matrix with the eigenvalues of A on
the diagonal. If A is a diagonalizable matrix and p is a polynomial then
p(A) = V p(Λ)V
−1
A is normal if the diagonalizing transformation V is orthogonal. In that case
the columns of V are the eigenvectors of A and V
−1
= V
H
. Here V
H
is the
complex conjugate transpose of V . In the remainder of this section we must
use complex arithmetic to analyze the convergence. Hence we will switch to
complex matrices and vectors. Recall that the scalar product in C
N
, the space
of complex Nvectors, is x
H
y. In particular, we will use the l
2
norm in C
N
.
Our use of complex arithmetic will be implicit for the most part and is needed
only so that we may admit the possibility of complex eigenvalues of A.
We can use the structure of a diagonalizable matrix to prove the following
result.
Theorem 3.1.3. Let A = V ΛV
−1
be a nonsingular diagonalizable matrix.
Let x
k
be the kth GMRES iterate. Then for all ¯ p
k
∈ T
k
r
k

2
r
0

2
≤ κ
2
(V ) max
z∈σ(A)
[ ¯ p
k
(z)[. (3.4)
Proof. Let ¯ p
k
∈ T
k
. We can easily estimate ¯ p
k
(A)
2
by
¯ p
k
(A)
2
≤ V 
2
V
−1

2
¯ p
k
(Λ)
2
≤ κ
2
(V ) max
z∈σ(A)
[ ¯ p
k
(z)[,
as asserted.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 35
It is not clear how one should estimate the condition number of the
diagonalizing transformation if it exists. If A is normal, of course, κ
2
(V ) = 1.
As we did for CG, we look at some easy consequences of (3.3) and (3.4).
Theorem 3.1.4. Let A be a nonsingular diagonalizable matrix. Assume
that A has only k distinct eigenvalues. Then GMRES will terminate in at most
k iterations.
Theorem 3.1.5. Let A be a nonsingular normal matrix. Let b be a linear
combination of k of the eigenvectors of A
b =
k
l=1
γ
l
u
i
l
.
Then the GMRES iteration, with x
0
= 0, for Ax = b will terminate in at most
k iterations.
3.2. Termination
As is the case with CG, GMRES is best thought of as an iterative method.
The convergence rate estimates for the diagonalizable case will involve κ
2
(V ),
but will otherwise resemble those for CG. If A is not diagonalizable, rate
estimates have been derived in [139], [134], [192], [33], and [34]. As the set of
nondiagonalizable matrices has measure zero in the space of N N matrices,
the chances are very high that a computed matrix will be diagonalizable. This
is particularly so for the ﬁnite diﬀerence Jacobian matrices we consider in
Chapters 6 and 8. Hence we conﬁne our attention to diagonalizable matrices.
As was the case with CG, we terminate the iteration when
r
k

2
≤ ηb
2
(3.5)
for the purposes of this example. We can use (3.3) and (3.4) directly to estimate
the ﬁrst k such that (3.5) holds without requiring a lemma like Lemma 2.3.2.
Again we look at examples. Assume that A = V ΛV
−1
is diagonalizable,
that the eigenvalues of A lie in the interval (9, 11), and that κ
2
(V ) = 100.
We assume that x
0
= 0 and hence r
0
= b. Using the residual polynomial
¯ p
k
(z) = (10 −z)
k
/10
k
we ﬁnd
r
k

2
r
0

2
≤ (100)10
−k
= 10
2−k
.
Hence (3.5) holds when 10
2−k
< η or when
k > 2 + log
10
(η).
Assume that I − A
2
≤ ρ < 1. Let ¯ p
k
(z) = (1 − z)
k
. It is a direct
consequence of (3.2) that
r
k

2
≤ ρ
k
r
0

2
. (3.6)
The estimate (3.6) illustrates the potential beneﬁts of a good approximate
inverse preconditioner.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
36 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
The convergence estimates for GMRES in the nonnormal case are much
less satisfying that those for CG, CGNR, CGNE, or GMRES in the normal
case. This is a very active area of research and we refer to [134], [33], [120],
[34], and [36] for discussions of and pointers to additional references to several
questions related to nonnormal matrices.
3.3. Preconditioning
Preconditioning for GMRES and other methods for nonsymmetric problems is
diﬀerent from that for CG. There is no concern that the preconditioned system
be spd and hence (3.6) essentially tells the whole story. However there are two
diﬀerent ways to view preconditioning. If one can ﬁnd M such that
I −MA
2
= ρ < 1,
then applying GMRES to MAx = Mb allows one to apply (3.6) to the
preconditioned system. Preconditioning done in this way is called left
preconditioning. If r = MAx − Mb is the residual for the preconditioned
system, we have, if the product MA can be formed without error,
e
k

2
e
0

2
≤ κ
2
(MA)
r
k

2
r
0

2
,
by Lemma 1.1.1. Hence, if MA has a smaller condition number than A, we
might expect the relative residual of the preconditioned system to be a better
indicator of the relative error than the relative residual of the original system.
If
I −AM
2
= ρ < 1,
one can solve the system AMy = b with GMRES and then set x = My. This is
called right preconditioning. The residual of the preconditioned problem is the
same as that of the unpreconditioned problem. Hence, the value of the relative
residuals as estimators of the relative error is unchanged. Right preconditioning
has been used as the basis for a method that changes the preconditioner as the
iteration progresses [166].
One important aspect of implementation is that, unlike PCG, one can
apply the algorithm directly to the system MAx = Mb (or AMy = b). Hence,
one can write a single matrixvector product routine for MA (or AM) that
includes both the application of A to a vector and that of the preconditioner.
Most of the preconditioning ideas mentioned in ¸ 2.5 are useful for GMRES
as well. In the examples in ¸ 3.7 we use the Poisson solver preconditioner for
nonsymmetric partial diﬀerential equations. Multigrid [99] and alternating
direction [8], [182] methods have similar performance and may be more
generally applicable. Incomplete factorization (LU in this case) preconditioners
can be used [165] as can polynomial preconditioners. Some hybrid algorithms
use the GMRES/Arnoldi process itself to construct polynomial preconditioners
for GMRES or for Richardson iteration [135], [72], [164], [183]. Again we
mention [8] and [12] as a good general references for preconditioning.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 37
3.4. GMRES implementation: Basic ideas
Recall that the least squares problem deﬁning the kth GMRES iterate is
minimize
x∈x
0
+K
k
b −Ax
2
.
Suppose one had an orthogonal projector V
k
onto /
k
. Then any z ∈ /
k
can
be written as
z =
k
l=1
y
l
v
k
l
,
where v
k
l
is the lth column of V
k
. Hence we can convert (3.1) to a least squares
problem in R
k
for the coeﬃcient vector y of z = x −x
0
. Since
x −x
0
= V
k
y
for some y ∈ R
k
, we must have x
k
= x
0
+V
k
y where y minimizes
b −A(x
0
+V
k
y)
2
= r
0
−AV
k
y
2
.
Hence, our least squares problem in R
k
is
minimize
y∈R
kr
0
−AV
k
y
2
. (3.7)
This is a standard linear least squares problem that could be solved by a QR
factorization, say. The problem with such a direct approach is that the matrix
vector product of A with the basis matrix V
k
must be taken at each iteration.
If one uses Gram–Schmidt orthogonalization, however, one can represent
(3.7) very eﬃciently and the resulting least squares problem requires no extra
multiplications of A with vectors. The Gram–Schmidt procedure for formation
of an orthonormal basis for /
k
is called the Arnoldi [4] process. The data are
vectors x
0
and b, a map that computes the action of A on a vector, and a
dimension k. The algorithm computes an orthonormal basis for /
k
and stores
it in the columns of V .
Algorithm 3.4.1. arnoldi(x
0
, b, A, k, V )
1. Deﬁne r
0
= b −Ax
0
and v
1
= r
0
/r
0

2
.
2. For i = 1, . . . , k −1
v
i+1
=
Av
i
−
i
j=1
((Av
i
)
T
v
j
)v
j
Av
i
−
i
j=1
((Av
i
)
T
v
j
)v
j

2
If there is never a division by zero in step 2 of Algorithm arnoldi, then
the columns of the matrix V
k
are an orthonormal basis for /
k
. A division by
zero is referred to as breakdown and happens only if the solution to Ax = b is
in x
0
+/
k−1
.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
38 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Lemma 3.4.1. Let A be nonsingular, let the vectors v
j
be generated by
Algorithm arnoldi, and let i be the smallest integer for which
Av
i
−
i
j=1
((Av
i
)
T
v
j
)v
j
= 0.
Then x = A
−1
b ∈ x
0
+/
i
.
Proof. By hypothesis Av
i
∈ /
i
and hence A/
i
⊂ /
i
. Since the columns of
V
i
are an orthonormal basis for /
i
, we have
AV
i
= V
i
H,
where H is an i i matrix. H is nonsingular since A is. Setting β = r
0

2
and e
1
= (1, 0, . . . , 0)
T
∈ R
i
we have
r
i

2
= b −Ax
i

2
= r
0
−A(x
i
−x
0
)
2
.
Since x
i
− x
0
∈ /
i
there is y ∈ R
i
such that x
i
− x
0
= V
i
y. Since r
0
= βV
i
e
1
and V
i
is an orthogonal matrix
r
i

2
= V
i
(βe
1
−Hy)
2
= βe
1
−Hy
R
i+1,
where  
R
k+1 denotes the Euclidean norm in R
k+1
.
Setting y = βH
−1
e
1
proves that r
i
= 0 by the minimization property.
The upper Hessenberg structure can be exploited to make the solution of
the least squares problems very eﬃcient [167].
If the Arnoldi process does not break down, we can use it to implement
GMRES in an eﬃcient way. Set h
ji
= (Av
j
)
T
v
i
. By the Gram–Schmidt
construction, the k+1k matrix H
k
whose entries are h
ij
is upper Hessenberg,
i.e., h
ij
= 0 if i > j +1. The Arnoldi process (unless it terminates prematurely
with a solution) produces matrices ¦V
k
¦ with orthonormal columns such that
AV
k
= V
k+1
H
k
. (3.8)
Hence, for some y
k
∈ R
k
,
r
k
= b −Ax
k
= r
0
−A(x
k
−x
0
) = V
k+1
(βe
1
−H
k
y
k
).
Hence x
k
= x
0
+V
k
y
k
, where y
k
minimizes βe
1
−H
k
y
2
over R
k
. Note that
when y
k
has been computed, the norm of r
k
can be found without explicitly
forming x
k
and computing r
k
= b −Ax
k
. We have, using the orthogonality of
V
k+1
,
r
k

2
= V
k+1
(βe
1
−H
k
y
k
)
2
= βe
1
−H
k
y
k

R
k+1. (3.9)
The goal of the iteration is to ﬁnd, for a given , a vector x so that
b −Ax
2
≤ b
2
.
The input is the initial iterate, x, the righthand side b, and a map that
computes the action of A on a vector. We limit the number of iterations
to kmax and return the solution, which overwrites the initial iterate x and the
residual norm.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 39
Algorithm 3.4.2. gmresa(x, b, A, , kmax, ρ)
1. r = b −Ax, v
1
= r/r
2
, ρ = r
2
, β = ρ, k = 0
2. While ρ > b
2
and k < kmax do
(a) k = k + 1
(b) for j = 1, . . . , k
h
jk
= (Av
k
)
T
v
j
(c) v
k+1
= Av
k
−
k
j=1
h
jk
v
j
(d) h
k+1,k
= v
k+1

2
(e) v
k+1
= v
k+1
/v
k+1

2
(f) e
1
= (1, 0, . . . , 0)
T
∈ R
k+1
Minimize βe
1
−H
k
y
k

R
k+1 over R
k
to obtain y
k
.
(g) ρ = βe
1
−H
k
y
k

R
k+1.
3. x
k
= x
0
+V
k
y
k
.
Note that x
k
is only computed upon termination and is not needed within
the iteration. It is an important property of GMRES that the basis for the
Krylov space must be stored as the iteration progress. This means that in
order to perform k GMRES iterations one must store k vectors of length N.
For very large problems this becomes prohibitive and the iteration is restarted
when the available room for basis vectors is exhausted. One way to implement
this is to set kmax to the maximum number m of vectors that one can store,
call GMRES and explicitly test the residual b−Ax
k
if k = m upon termination.
If the norm of the residual is larger than , call GMRES again with x
0
= x
k
,
the result from the previous call. This restarted version of the algorithm is
called GMRES(m) in [167]. There is no general convergence theorem for the
restarted algorithm and restarting will slow the convergence down. However,
when it works it can signiﬁcantly reduce the storage costs of the iteration. We
discuss implementation of GMRES(m) later in this section.
Algorithm gmresa can be implemented very straightforwardly in MAT
LAB. Step 2f can be done with a single MATLAB command, the backward
division operator, at a cost of O(k
3
) ﬂoatingpoint operations. There are more
eﬃcient ways to solve the least squares problem in step 2f, [167], [197], and we
use the method of [167] in the collection of MATLAB codes. The savings are
slight if k is small relative to N, which is often the case for large problems, and
the simple oneline MATLAB approach can be eﬃcient for such problems.
A more serious problem with the implementation proposed in Algo
rithm gmresa is that the vectors v
j
may become nonorthogonal as a result of
cancellation errors. If this happens, (3.9), which depends on this orthogonality,
will not hold and the residual and approximate solution could be inaccurate. A
partial remedy is to replace the classical Gram–Schmidt orthogonalization in
Algorithm gmresa with modiﬁed Gram–Schmidt orthogonalization. We replace
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
40 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
the loop in step 2c of Algorithm gmresa with
v
k+1
= Av
k
for j = 1, . . . k
v
k+1
= v
k+1
−(v
T
k+1
v
j
)v
j
.
While modiﬁed Gram–Schmidt and classical Gram–Schmidt are equivalent in
inﬁnite precision, the modiﬁed form is much more likely in practice to maintain
orthogonality of the basis.
We illustrate this point with a simple example from [128], doing the
computations in MATLAB. Let δ = 10
−7
and deﬁne
A =
_
_
_
1 1 1
δ δ 0
δ 0 δ
_
_
_.
We orthogonalize the columns of A with classical Gram–Schmidt to obtain
V =
_
_
_
1.0000e + 00 1.0436e −07 9.9715e −08
1.0000e −07 1.0456e −14 −9.9905e −01
1.0000e −07 −1.0000e + 00 4.3568e −02
_
_
_.
The columns of V
U
are not orthogonal at all. In fact v
T
2
v
3
≈ −.004. For
modiﬁed Gram–Schmidt
V =
_
_
_
1.0000e + 00 1.0436e −07 1.0436e −07
1.0000e −07 1.0456e −14 −1.0000e + 00
1.0000e −07 −1.0000e + 00 4.3565e −16
_
_
_.
Here [v
T
i
v
j
−δ
ij
[ ≤ 10
−8
for all i, j.
The versions we implement in the collection of MATLAB codes use modi
ﬁed Gram–Schmidt. The outline of our implementation is Algorithm gmresb.
This implementation solves the upper Hessenberg least squares problem using
the MATLAB backward division operator, and is not particularly eﬃcient. We
present a better implementation in Algorithm gmres. However, this version is
very simple and illustrates some important ideas. First, we see that x
k
need
only be computed after termination as the least squares residual ρ can be used
to approximate the norm of the residual (they are identical in exact arithmetic).
Second, there is an opportunity to compensate for a loss of orthogonality in
the basis vectors for the Krylov space. One can take a second pass through the
modiﬁed Gram–Schmidt process and restore lost orthogonality [147], [160].
Algorithm 3.4.3. gmresb(x, b, A, , kmax, ρ)
1. r = b −Ax, v
1
= r/r
2
, ρ = r
2
, β = ρ, k = 0
2. While ρ > b
2
and k < kmax do
(a) k = k + 1
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 41
(b) v
k+1
= Av
k
for j = 1, . . . k
i. h
jk
= v
T
k+1
v
j
ii. v
k+1
= v
k+1
−h
jk
v
j
(c) h
k+1,k
= v
k+1

2
(d) v
k+1
= v
k+1
/v
k+1

2
(e) e
1
= (1, 0, . . . , 0)
T
∈ R
k+1
Minimize βe
1
−H
k
y
k

R
k+1 to obtain y
k
∈ R
k
.
(f) ρ = βe
1
−H
k
y
k

R
k+1.
3. x
k
= x
0
+V
k
y
k
.
Even if modiﬁed Gram–Schmidt orthogonalization is used, one can still
lose orthogonality in the columns of V . One can test for loss of orthogonality
[22], [147], and reorthogonalize if needed or use a more stable means to create
the matrix V [195]. These more complex implementations are necessary if A is
ill conditioned or many iterations will be taken. For example, one can augment
the modiﬁed Gram–Schmidt process
• v
k+1
= Av
k
for j = 1, . . . k
h
jk
= v
T
k+1
v
j
v
k+1
= v
k+1
−h
jk
v
j
• h
k+1,k
= v
k+1

2
• v
k+1
= v
k+1
/v
k+1

2
with a second pass (reorthogonalization). One can reorthogonalize in every
iteration or only if a test [147] detects a loss of orthogonality. There is nothing
to be gained by reorthogonalizing more than once [147].
The modiﬁed Gram–Schmidt process with reorthogonalization looks like
• v
k+1
= Av
k
for j = 1, . . . , k
h
jk
= v
T
k+1
v
j
v
k+1
= v
k+1
−h
jk
v
j
• h
k+1,k
= v
k+1

2
• If loss of orthogonality is detected
For j = 1, . . . , k
h
tmp
= v
T
k+1
v
j
h
jk
= h
jk
+h
tmp
v
k+1
= v
k+1
−h
tmp
v
j
• h
k+1,k
= v
k+1

2
• v
k+1
= v
k+1
/v
k+1

2
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
42 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
One approach to reorthogonalization is to reorthogonalize in every step.
This doubles the cost of the computation of V and is usually unnecessary.
More eﬃcient and equally eﬀective approaches are based on other ideas. A
variation on a method from [147] is used in [22]. Reorthogonalization is done
after the Gram–Schmidt loop and before v
k+1
is normalized if
Av
k

2
+δv
k+1

2
= Av
k

2
(3.10)
to working precision. The idea is that if the new vector is very small relative
to Av
k
then information may have been lost and a second pass through
the modiﬁed Gram–Schmidt process is needed. We employ this test in the
MATLAB code gmres with δ = 10
−3
.
To illustrate the eﬀects of loss of orthogonality and those of reorthogonal
ization we apply GMRES to the diagonal system Ax = b where b = (1, 1, 1)
T
,
x
0
= (0, 0, 0)
T
, and
A =
_
_
_
.001 0 0
0 .0011 0
0 0 10
4
_
_
_. (3.11)
While in inﬁnite precision arithmetic only three iterations are needed to solve
the system exactly, we ﬁnd in the MATLAB environment that a solution to full
precision requires more than three iterations unless reorthogonalization is ap
plied after every iteration. In Table 3.1 we tabulate relative residuals as a func
tion of the iteration counter for classical Gram–Schmidt without reorthogonal
ization (CGM), modiﬁed Gram–Schmidt without reorthogonalization (MGM),
reorthogonalization based on the test (3.10) (MGMPO), and reorthogonaliza
tion in every iteration (MGMFO). While classical Gram–Schmidt fails, the
reorthogonalization strategy based on (3.10) is almost as eﬀective as the much
more expensive approach of reorthogonalizing in every step. The method based
on (3.10) is the default in the MATLAB code gmresa.
The kth GMRES iteration requires a matrixvector product, k scalar
products, and the solution of the Hessenberg least squares problem in step 2e.
The k scalar products require O(kN) ﬂoatingpoint operations and the cost
of a solution of the Hessenberg least squares problem, by QR factorization or
the MATLAB backward division operator, say, in step 2e of gmresb is O(k
3
)
ﬂoatingpoint operations. Hence the total cost of the m GMRES iterations is
m matrixvector products and O(m
4
+m
2
N) ﬂoatingpoint operations. When
k is not too large and the cost of matrixvector products is high, a brute
force solution to the least squares problem using the MATLAB backward
division operator is not terribly ineﬃcient. We provide an implementation
of Algorithm gmresb in the collection of MATLAB codes. This is an appealing
algorithm, especially when implemented in an environment like MATLAB,
because of its simplicity. For large k, however, the bruteforce method can be
very costly.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 43
Table 3.1
Eﬀects of reorthogonalization.
k CGM MGM MGMPO MGMFO
0 1.00e+00 1.00e+00 1.00e+00 1.00e+00
1 8.16e01 8.16e01 8.16e01 8.16e01
2 3.88e02 3.88e02 3.88e02 3.88e02
3 6.69e05 6.42e08 6.42e08 6.34e34
4 4.74e05 3.70e08 5.04e24
5 3.87e05 3.04e18
6 3.35e05
7 3.00e05
8 2.74e05
9 2.53e05
10 2.37e05
3.5. Implementation: Givens rotations
If k is large, implementations using Givens rotations [167], [22], Householder
reﬂections [195], or a shifted Arnoldi process [197] are much more eﬃcient
than the bruteforce approach in Algorithm gmresb. The implementation in
Algorithm gmres and in the MATLAB code collection is from [167]. This
implementation maintains the QR factorization of H
k
in a clever way so that
the cost for a single GMRES iteration is O(Nk) ﬂoatingpoint operations. The
O(k
2
) cost of the triangular solve and the O(kN) cost of the construction of
x
k
are incurred after termination.
A 2 2 Givens rotation is a matrix of the form
G =
_
c −s
s c
_
, (3.12)
where c = cos(θ), s = sin(θ) for θ ∈ [−π, π]. The orthogonal matrix G rotates
the vector (c, −s), which makes an angle of −θ with the xaxis through an
angle θ so that it overlaps the xaxis.
G
_
c
−s
_
=
_
1
0
_
.
An N N Givens rotation replaces a 2 2 block on the diagonal of the
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
44 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
N N identity matrix with a 2 2 Givens rotation.
G =
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
1 0 . . . 0
0
.
.
.
.
.
.
.
.
. c −s
.
.
. s c 0
.
.
.
0 1
.
.
.
.
.
.
.
.
. 0
0 . . . 0 1
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
. (3.13)
Our notation is that G
j
(c, s) is an N N givens rotation of the form (3.13)
with a 2 2 Givens rotation in rows and columns j and j + 1.
Givens rotations are used to annihilate single nonzero elements of matrices
in reduction to triangular form [89]. They are of particular value in reducing
Hessenberg matrices to triangular form and thereby solving Hessenberg least
squares problems such as the ones that arise in GMRES. This reduction can be
accomplished in O(N) ﬂoatingpoint operations and hence is far more eﬃcient
than a solution by a singular value decomposition or a reduction based on
Householder transformations. This method is also used in the QR algorithm
for computing eigenvalues [89], [184].
Let H be an N M (N ≥ M) upper Hessenberg matrix with rank M.
We reduce H to triangular form by ﬁrst multiplying the matrix by a Givens
rotation that annihilates h
21
(and, of course, changes h
11
and the subsequent
columns). We deﬁne G
1
= G
1
(c
1
, s
1
) by
c
1
= h
11
/
_
h
2
11
+h
2
21
and s
1
= −h
21
/
_
h
2
11
+h
2
21
. (3.14)
If we replace H by G
1
H, then the ﬁrst column of H now has only a single
nonzero element h
11
. Similarly, we can now apply G
2
(c
2
, s
2
) to H, where
c
2
= h
22
/
_
h
2
22
+h
2
32
and s
2
= −h
32
/
_
h
2
22
+h
2
32
. (3.15)
and annihilate h
32
. Note that G
2
does not aﬀect the ﬁrst column of H.
Continuing in this way and setting
Q = G
N
. . . G
1
we see that QH = R is upper triangular.
A straightforward application of these ideas to Algorithm gmres would
solve the least squares problem by computing the product of k Givens rotations
Q, setting g = βQe
1
, and noting that
βe
1
−H
k
y
k

R
k+1 = Q(βe
1
−H
k
y
k
)
R
k+1 = g −R
k
y
k

R
k+1,
where R
k
is the k + 1 k triangular factor of the QR factorization of H
k
.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 45
In the context of GMRES iteration, however, we can incrementally perform
the QR factorization of H as the GMRES iteration progresses [167]. To see
this, note that if R
k
= Q
k
H
k
and, after orthogonalization, we add the new
column h
k+2
to H
k
, we can update both Q
k
and R
k
by ﬁrst multiplying h
k+2
by Q
k
(that is applying the ﬁrst k Givens rotations to h
k+2
), then computing
the Givens rotation G
k+1
that annihilates the (k + 2)nd element of Q
k
h
k+2
,
and ﬁnally setting Q
k+1
= G
k+1
Q
k
and forming R
k+1
by augmenting R
k
with
G
k+1
Q
k
h
k+2
.
The MATLAB implementation of Algorithm gmres stores Q
k
by storing
the sequences ¦c
j
¦ and ¦s
j
¦ and then computing the action of Q
k
on a vector
x ∈ R
k+1
by applying G
j
(c
j
, s
j
) in turn to obtain
Q
k
x = G
k
(c
k
, s
k
) . . . G
1
(c
1
, s
1
)x.
We overwrite the upper triangular part of H
k
with the triangular part of the
QR factorization of H
k
in the MATLAB code. The MATLAB implementation
of Algorithm gmres uses (3.10) to test for loss of orthogonality.
Algorithm 3.5.1. gmres(x, b, A, , kmax, ρ)
1. r = b −Ax, v
1
= r/r
2
, ρ = r
2
, β = ρ,
k = 0; g = ρ(1, 0, . . . , 0)
T
∈ R
kmax+1
2. While ρ > b
2
and k < kmax do
(a) k = k + 1
(b) v
k+1
= Av
k
for j = 1, . . . k
i. h
jk
= v
T
k+1
v
j
ii. v
k+1
= v
k+1
−h
jk
v
j
(c) h
k+1,k
= v
k+1

2
(d) Test for loss of orthogonality and reorthogonalize if necessary.
(e) v
k+1
= v
k+1
/v
k+1

2
(f) i. If k > 1 apply Q
k−1
to the kth column of H.
ii. ν =
_
h
2
k,k
+h
2
k+1,k
.
iii. c
k
= h
k,k
/ν, s
k
= −h
k+1,k
/ν
h
k,k
= c
k
h
k,k
−s
k
h
k+1,k
, h
k+1,k
= 0
iv. g = G
k
(c
k
, s
k
)g.
(g) ρ = [(g)
k+1
[.
3. Set r
i,j
= h
i,j
for 1 ≤ i, j ≤ k.
Set (w)
i
= (g)
i
for 1 ≤ i ≤ k.
Solve the upper triangular system Ry
k
= w.
4. x
k
= x
0
+V
k
y
k
.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
46 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
We close with an example of an implementation of GMRES(m) . This
implementation does not test for success and may, therefore, fail to terminate.
You are asked to repair this in exercise 7. Aside from the parameter m, the
arguments to Algorithm gmresm are the same as those for Algorithm gmres.
Algorithm 3.5.2. gmresm(x, b, A, , kmax, m, ρ)
1. gmres(x, b, A, , m, ρ)
2. k = m
3. While ρ > b
2
and k < kmax do
(a) gmres(x, b, A, , m, ρ)
(b) k = k +m
The storage costs of m iterations of gmres or of gmresm are the m + 2
vectors b, x, and ¦v
k
¦
m
k=1
.
3.6. Other methods for nonsymmetric systems
The methods for nonsymmetric linear systems that receive most attention in
this book, GMRES, CGNR, and CGNE, share the properties that they are easy
to implement, can be analyzed by a common residual polynomial approach, and
only terminate if an acceptable approximate solution has been found. CGNR
and CGNE have the disadvantages that a transposevector product must be
computed for each iteration and that the coeﬃcient matrix A
T
A or AA
T
has
condition number the square of that of the original matrix. In ¸ 3.8 we give
an example of how this squaring of the condition number can lead to failure
of the iteration. GMRES needs only matrixvector products and uses A alone,
but, as we have seen, a basis for /
k
must be stored to compute x
k
. Hence,
the storage requirements increase as the iteration progresses. For large and
illconditioned problems, it may be impossible to store enough basis vectors
and the iteration may have to be restarted. Restarting can seriously degrade
performance.
An ideal method would, like CG, only need matrixvector products, be
based on some kind of minimization principle or conjugacy, have modest
storage requirements that do not depend on the number of iterations needed
for convergence, and converge in N iterations for all nonsingular A. However,
[74], methods based on shortterm recurrences such as CG that also satisfy
minimization or conjugacy conditions cannot be constructed for general
matrices. The methods we describe in this section fall short of the ideal, but
can still be quite useful. We discuss only a small subset of these methods and
refer the reader to [12] and [78] for pointers to more of the literature on this
subject. All the methods we present in this section require two matrixvector
products for each iteration.
Consistently with our implementation of GMRES, we take the view that
preconditioners will be applied externally to the iteration. However, as with
CG, these methods can also be implemented in a manner that incorporates
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 47
the preconditioner and uses the residual for the original system to control
termination.
3.6.1. BiCG. The earliest such method BiCG (Biconjugate gradient)
[122], [76], does not enforce a minimization principle; instead, the kth residual
must satisfy the biorthogonality condition
r
T
k
w = 0 for all w ∈ /
k
, (3.16)
where
/
k
= span(ˆ r
0
, A
T
ˆ r
0
, . . . , (A
T
)
k−1
ˆ r
0
)
is the Krylov space for A
T
and the vector ˆ r
0
. ˆ r
0
is a usersupplied vector and is
often set to r
0
. The algorithm gets its name because it produces sequences of
residuals ¦r
k
¦, ¦ˆ r
k
¦ and search directions ¦p
k
¦, ¦ˆ p
k
¦ such that biorthogonality
holds, i. e. ˆ r
T
k
r
l
= 0 if k ,= l and the search directions ¦p
k
¦ and ¦ˆ p
k
¦ satisfy
the biconjugacy property
ˆ p
T
k
Ap
l
= 0 if k ,= l.
In the symmetric positive deﬁnite case (3.16) is equivalent to the minimization
principle (2.2) for CG [89].
Using the notation of Chapter 2 and [191] we give an implementation of Bi
CG making the choice ˆ r
0
= r
0
. This algorithmic description is explained and
motivated in more detail in [122], [76], [78], and [191]. We do not recommend
use of BiCG and present this algorithm only as a basis for discussion of some
of the properties of this class of algorithms.
Algorithm 3.6.1. bicg(x, b, A, , kmax)
1. r = b −Ax, ˆ r = r, ρ
0
= 1, ˆ p = p = 0, k = 0
2. Do While r
2
> b
2
and k < kmax
(a) k = k + 1
(b) ρ
k
= ˆ r
T
r, β = ρ
k
/ρ
k−1
(c) p = r +βp, ˆ p = ˆ r +βˆ p
(d) v = Ap
(e) α = ρ
k
/(ˆ p
T
v)
(f) x = x +αp
(g) r = r −αv; ˆ r = ˆ r −αA
T
ˆ p
Note that if A = A
T
is spd, BiCG produces the same iterations as CG
(but computes everything except x twice). If, however, A is not spd, there is
no guarantee that ρ
k
in step 2b or ˆ p
T
Ap, the denominator in step 2e, will not
vanish. If either ρ
k−1
= 0 or ˆ p
T
Ap = 0 we say that a breakdown has taken
place. Even if these quantities are nonzero but very small, the algorithm can
become unstable or produce inaccurate results. While there are approaches
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
48 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
[80], [78], [148], [79], [81], to avoid some breakdowns in many variations of
BiCG, there are no methods that both limit storage and completely eliminate
the possibility of breakdown [74]. All of the methods we present in this section
can break down and should be implemented with that possibility in mind.
Once breakdown has occurred, one can restart the BiCG iteration or pass
the computation to another algorithm, such as GMRES. The possibility of
breakdown is small and certainly should not eliminate the algorithms discussed
below from consideration if there is not enough storage for GMRES to perform
well.
Breakdowns aside, there are other problems with BiCG. A transpose
vector product is needed, which at the least will require additional program
ming and may not be possible at all. The performance of the algorithm can
be erratic or even unstable with residuals increasing by several orders of mag
nitude from one iteration to the next. Moreover, the eﬀort in computing ˆ r
at each iteration is wasted in that ˆ r makes no contribution to x. However,
BiCG sometimes performs extremely well and the remaining algorithms in
this section represent attempts to capture this good performance and damp
the erratic behavior when it occurs.
We can compare the bestcase performance of BiCG with that of GMRES.
Note that there is ¯ p
k
∈ T
k
such that both
r
k
= ¯ p
k
(A)r
0
and ˆ r
k
= ¯ p
k
(A
T
)ˆ r
0
. (3.17)
Hence, by the minimization property for GMRES
r
GMRES
k

2
≤ r
Bi−CG
k

2
,
reminding us that GMRES, if suﬃcient storage is available, will always
reduce the residual more rapidly than BiCG (in terms of iterations, but not
necessarily in terms of computational work). One should also keep in mind that
a single BiCG iteration requires two matrixvector products and a GMRES
iterate only one, but that the cost of the GMRES iteration increases (in terms
of ﬂoatingpoint operations) as the iteration progresses.
3.6.2. CGS. A remedy for one of the problems with BiCG is the Conjugate
Gradient Squared (CGS) algorithm [180]. The algorithm takes advantage of
the fact that (3.17) implies that the scalar product ˆ r
T
r in BiCG (step 2b of
Algorithm bicg) can be represented without using A
T
as
r
T
k
ˆ r
k
= (¯ p
k
(A)r
0
)
T
(¯ p
k
(A
T
)ˆ r
0
) = (¯ p
k
(A)
2
r
0
)
T
ˆ r
0
.
The other references to A
T
can be eliminated in a similar fashion and an
iteration satisfying
r
k
= ¯ p
k
(A)
2
r
0
(3.18)
is produced, where ¯ p
k
is the same polynomial produced by BiCG and used in
(3.17). This explains the name, Conjugate Gradient Squared.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 49
The work used in BiCG to compute ˆ r is now used to update x. CGS
replaces the transposevector product with an additional matrixvector product
and applies the square of the BiCG polynomial to r
0
to produce r
k
. This may,
of course, change the convergence properties for the worse and either improve
good convergence or magnify erratic behavior [134], [191]. CGS has the same
potential for breakdown as BiCG. Since ¯ p
2
k
∈ T
2k
we have
r
GMRES
2k

2
≤ r
CGS
k

2
. (3.19)
We present an algorithmic description of CGS that we will refer to in our
discussion of TFQMR in ¸ 3.6.4. In our description of CGS we will index the
vectors and scalars that would be overwritten in an implementation so that we
can describe some of the ideas in TFQMR later on. This description is taken
from [77].
Algorithm 3.6.2. cgs(x, b, A, , kmax)
1. x
0
= x; p
0
= u
0
= r
0
= b −Ax
2. v
0
= Ap
0
; ˆ r
0
= r
0
3. ρ
0
= ˆ r
T
0
r
0
; k = 0
4. Do While r
2
> b
2
and k < kmax
(a) k = k + 1
(b) σ
k−1
= ˆ r
T
0
v
k−1
(c) α
k−1
= ρ
k−1
/σ
k−1
(d) q
k
= u
k−1
−α
k−1
v
k−1
(e) x
k
= x
k−1
+α
k−1
(u
k−1
+q
k
)
r
k
= r
k−1
−α
k−1
A(u
k−1
+q
k
)
(f) ρ
k
= ˆ r
T
0
r
k
; β
k
= ρ
k
/ρ
k−1
u
k
= r
k
+β
k
q
k
p
k
= u
k
+β
k
(q
k
+β
k
p
k−1
)
v
k
= Ap
k
Breakdowns take place when either ρ
k−1
or σ
k−1
vanish. If the algorithm
does not break down, then α
k−1
,= 0 for all k. One can show [180], [77], that
if ¯ p
k
is the residual polynomial from (3.17) and (3.18) then
¯ p
k
(z) = ¯ p
k−1
(z) −α
k−1
z¯ q
k−1
(z) (3.20)
where the auxiliary polynomials ¯ q
k
∈ T
k
are given by ¯ q
0
= 1 and for k ≥ 1 by
¯ q
k
(z) = ¯ p
k
(z) +β
k
¯ q
k−1
(z). (3.21)
We provide no MATLAB implementation of BiCG or CGS. BiCGSTAB
is the most eﬀective representative of this class of algorithms.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
50 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
3.6.3. BiCGSTAB. The Biconjugate gradient stabilized method (Bi
CGSTAB) [191] attempts to smooth the convergence of CGS by replacing
(3.18) with
r
k
= q
k
(A)p
k
(A)r
0
(3.22)
where
q
k
(z) =
k
i=1
(1 −ω
i
z).
The constant ω
i
is selected to minimize r
i
= q
i
(A)p
i
(A)r
0
as a function of ω
i
.
The examples in [191] indicate the eﬀectiveness of this approach, which can
be thought of [12] as a blend of BiCG and GMRES(1). The performance in
cases where the spectrum has a signiﬁcant imaginary part can be improved by
constructing q
k
to have complex conjugate pairs of roots, [98].
There is no convergence theory for BiCGSTAB and no estimate of the
residuals that is better than that for CGS, (3.19). In our description of the
implementation, which is described and motivated fully in [191], we use ˆ r
0
= r
0
and follow the notation of [191].
Algorithm 3.6.3. bicgstab(x, b, A, , kmax)
1. r = b −Ax, ˆ r
0
= ˆ r = r, ρ
0
= α = ω = 1, v = p = 0, k = 0, ρ
1
= ˆ r
T
0
r
2. Do While r
2
> b
2
and k < kmax
(a) k = k + 1
(b) β = (ρ
k
/ρ
k−1
)(α/ω)
(c) p = r +β(p −ωv)
(d) v = Ap
(e) α = ρ
k
/(ˆ r
T
0
v)
(f) s = r −αv, t = As
(g) ω = t
T
s/t
2
2
, ρ
k+1
= −ωˆ r
T
0
t
(h) x = x +αp +ωs
(i) r = s −ωt
Note that the iteration can break down in steps 2b and 2e. We provide an
implementation of Algorithm bicgstab in the collection of MATLAB codes.
The cost in storage and in ﬂoatingpoint operations per iteration re
mains bounded for the entire iteration. One must store seven vectors
(x, b, r, p, v, ˆ r
0
, t), letting s overwrite r when needed. A single iteration re
quires four scalar products. In a situation where many GMRES iterations are
needed and matrixvector product is fast, BiCGSTAB can have a much lower
average cost per iterate than GMRES. The reason for this is that the cost of
orthogonalization in GMRES can be much more than that of a matrixvector
product if the dimension of the Krylov space is large. We will present such a
case in ¸ 3.8.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 51
3.6.4. TFQMR. Now we consider the quasiminimal residual (QMR)
family of algorithms [80], [81], [77], [37], [202]. We will focus on the transpose
free algorithm TFQMR proposed in [77], to illustrate the quasiminimization
idea. All algorithms in this family minimize the norm of an easily computable
quantity q = f − Hz, a quasiresidual over z ∈ R
N
. The quasiresidual is
related to the true residual by a fullrank linear transformation r = Lq and
reduction of the residual can be approximately measured by reduction in q.
The speciﬁc formulation of q and L depends on the algorithm.
Returning to Algorithm cgs we deﬁne sequences
y
m
=
_
¸
_
¸
_
u
k−1
if m = 2k −1 is odd,
q
k
if m = 2k is even,
and
w
m
=
_
¸
_
¸
_
(¯ p
k
(A))
2
r
0
if m = 2k −1 is odd,
¯ p
k
(A)¯ p
k−1
(A)r
0
if m = 2k is even.
Here the sequences ¦¯ p
k
¦ and ¦¯ q
k
¦ are given by (3.20) and (3.21). Hence, the
CGS residual satisﬁes
r
CGS
k
= w
2k+1
.
We assume that CGS does not break down and, therefore, α
k
,= 0 for all
k ≥ 0. If we let ¸r be the nearest integer less than or equal to a real r we
have
Ay
m
= (w
m
−w
m+1
)/α
(m−1)/2
, (3.23)
where the denominator is nonzero by assumption. We express (3.23) in matrix
form as
AY
m
= W
m+1
B
m
, (3.24)
where Y
m
is the N m matrix with columns ¦y
j
¦
m
j=1
, W
m+1
the N (m+1)
matrix with columns ¦w
j
¦
m+1
j=1
, and B
m
is the (m+ 1) m matrix given by
B
m
=
_
_
_
_
_
_
_
_
_
_
1 0 . . . 0
−1 1
.
.
.
.
.
.
0
.
.
.
.
.
. 0
.
.
.
.
.
. −1 1
0 . . . 0 −1
_
_
_
_
_
_
_
_
_
_
diag(α
0
, α
0
, . . . , α
(m−1)/2
)
−1
.
Our assumptions that no breakdown takes place imply that /
m
is the span
of ¦y
j
¦
m
j=1
and hence
x
m
= x
0
+Y
m
z (3.25)
for some z ∈ R
m
. Therefore,
r
m
= r
0
−AY
m
z = W
m+1
(e
1
−B
m
z), (3.26)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
52 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
where, as in our discussion of GMRES, e
1
= (1, 0, . . . , 0)
T
∈ R
m
. In a sense,
the conditioning of the matrix W
m+1
can be improved by scaling. Let
Ω = diag(ω
1
, ω
2
, . . . , ω
m
) (3.27)
and rewrite (3.26) as
r
m
= r
0
−AY
m
z = W
m+1
Ω
−1
m+1
(f
m+1
−H
m
z) (3.28)
where f
m+1
= Ω
m+1
e
1
= ω
m+1
e
1
and H
m
= Ω
m+1
B
m
is bidiagonal and hence
upper Hessenberg. The diagonal entries in the matrix Ω
m+1
are called weights.
The weights make it possible to derive straightforward estimates of the residual
in terms of easily computed terms in the iteration.
If W
m+1
Ω
−1
m+1
were an orthogonal matrix, we could minimize r
m
by solving
the least squares problem
minimize
z∈R
mf
m+1
−H
m
z
2
. (3.29)
The quasiminimization idea is to solve (3.29) (thereby quasiminimizing the
residual) to obtain z
m
and then set
x
m
= x
0
+Y
m
z
m
. (3.30)
Note that if k corresponds to the iteration counter for CGS, each k produces
two approximate solutions (corresponding to m = 2k −1 and m = 2k).
The solution to the least squares problem (3.29) can be done by using
Givens rotations to update the factorization of the upper Hessenberg matrix
H
m
[77] and this is reﬂected in the implementation description below. As
one can see from that description, approximate residuals are not computed in
Algorithm tfqmr. Instead we have the quasiresidual norm
τ
m
= f
m+1
−H
m
z
m

2
.
Now if we pick the weights ω
i
= w
i

2
then the columns of W
m+1
Ω
−1
m+1
have
norm one. Hence
r
m

2
= W
m+1
Ω
−1
m+1
(f
m+1
−H
m
z)
2
≤ τ
m
√
m+ 1. (3.31)
We can, therefore, base our termination criterion on τ
m
, and we do this in
Algorithm tfqmr.
One can [80], [78], also use the quasiminimization condition to estimate
the TFQMR residual in terms of residual polynomials.
Theorem 3.6.1. Let A be an N N matrix and let x
0
, b ∈ R
N
be given.
Assume that the TFQMR iteration does not break down or terminate with
the exact solution for k iterations and let 1 ≤ m ≤ 2k. Let r
GMRES
m
be the
residual for the mth GMRES iteration. Let ξ
m
be the smallest singular value
of W
m+1
Ω
−1
m+1
. Then if ξ
m
> 0
τ
m
≤ r
GMRES
m

2
/ξ
m
. (3.32)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 53
Proof. We may write
r
GMRES
m
= r
0
+Y
m
z
GMRES
m
and obtain
r
GMRES
m

2
= W
m+1
Ω
−1
m+1
(f
m+1
−H
m
z
GMRES
m
)
2
≥ ξ
m
f
m+1
−H
m
z
GMRES
m

2
.
Hence, by the quasiminimization property
r
GMRES
m

2
≥ ξ
m
τ
m
as desired.
As a corollary we have a ﬁnite termination result and a convergence
estimate for diagonalizable matrices similar to Theorem 3.1.3.
Corollary 3.6.1. Let A be an N N matrix and let x
0
, b ∈ R
N
be given.
Then within (N +1)/2 iterations the TFQMR iteration will either break down
or terminate with the solution.
We apply Theorem 3.1.3 to (3.32) and use (3.31) to obtain the following
result.
Theorem 3.6.2. Let A = V ΛV
−1
be a nonsingular diagonalizable matrix.
Assume that TFQMR does not break down or terminate with the solution for k
iterations. For 1 ≤ m ≤ 2k let ξ
m
be the smallest singular value of W
m+1
Ω
−1
m+1
and let x
m
be given by (3.30). Then, if ξ
m
> 0,
r
m

2
r
0

2
≤
√
m+ 1ξ
−1
m
κ(V ) max
z∈σ(A)
[φ(z)[
for all φ ∈ T
m
.
The implementation below follows [77] with ˆ r
0
= r
0
used throughout.
Algorithm 3.6.4. tfqmr(x, b, A, , kmax)
1. k = 0, w
1
= y
1
= r
0
= b −Ax, u
1
= v = Ay
1
, d = 0
ρ
0
= r
T
0
r
0
, τ = r
2
, θ = 0, η = 0
2. Do While k < kmax
(a) k = k + 1
(b) σ
k−1
= r
T
0
v, α = ρ
k−1
/σ
k−1
, y
2
= y
1
−α
k−1
v, u
2
= Ay
2
(c) For j = 1, 2 (m = 2k −2 +j)
i. w = w −α
k−1
u
j
ii. d = y
j
+ (θ
2
η/α
k−1
)d
iii. θ = w
2
/τ, c = 1/
√
1 +θ
2
iv. τ = τθc, η = c
2
α
k−1
v. x = x +ηd
vi. If τ
√
m+ 1 ≤ b
2
terminate successfully
(d) ρ
k
= r
T
0
w, β = ρ
k
/ρ
k−1
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
54 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
(e) y
1
= w +βy
2
, u
1
= Ay
1
(f) v = u
1
+β(u
2
+βv)
Note that y
2
and u
2
= Ay
2
need only be computed if the loop in step 2c
does not terminate when j = 1. We take advantage of this in the MATLAB
code tfqmr to potentially reduce the matrixvector product cost by one.
3.7. Examples for GMRES iteration
These examples use the code gmres from the collection of MATLAB codes. The
inputs, described in the comment lines, are the initial iterate x
0
, the righthand
side vector b, a MATLAB function for the matrixvector product, and iteration
parameters to specify the maximum number of iterations and the termination
criterion. We do not pass the preconditioner to the GMRES code, rather
we pass the preconditioned problem. In all the GMRES examples, we limit
the number of GMRES iterations to 60. For methods like GMRES(m), Bi
CGSTAB, CGNR, and TFQMR, whose storage requirements do not increase
with the number of iterations, we allow for more iterations.
We consider the discretization of the partial diﬀerential equation
(Lu)(x, y) = −(u
xx
(x, y) +u
yy
(x, y)) +a
1
(x, y)u
x
(x, y)
+a
2
(x, y)u
y
(x, y) +a
3
(x, y)u(x, y) = f(x, y)
(3.33)
on 0 < x, y < 1 subject to homogeneous Dirichlet boundary conditions
u(x, 0) = u(x, 1) = u(0, y) = u(1, y) = 0, 0 < x, y < 1.
For general coeﬃcients ¦a
j
¦ the operator L is not selfadjoint and its discretiza
tion is not symmetric. As in ¸ 2.7 we discretize with a ﬁvepoint centered dif
ference scheme with n
2
points and mesh width h = 1/(n + 1). The unknowns
are
u
ij
≈ u(x
i
, x
j
)
where x
i
= ih for 1 ≤ i ≤ n. We compute Lu in a matrixfree manner
as we did in ¸ 2.7. We let n = 31 to create a system with 961 unknowns.
As before, we expect secondorder accuracy and terminate the iteration when
r
k

2
≤ h
2
b
2
≈ 9.8 10
−4
b
2
.
As a preconditioner we use the fast Poisson solver fish2d described in ¸ 2.7.
The motivation for this choice is similar to that for the conjugate gradient
computations and has been partially explained theoretically in [33], [34], [121],
and [139]. If we let Gu denote the action of the Poisson solver on u, the
preconditioned system is GLu = Gf.
For the computations reported in this section we took
a
1
(x, y) = 1, a
2
(x, y) = 20y, and a
3
(x, y) = 1.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 55
u
0
= 0 was the initial iterate. As in ¸ 2.7 we took the right hand side so that
the exact solution was the discretization of
10xy(1 −x)(1 −y) exp(x
4.5
).
In Figs. 3.1 and 3.2 we plot iteration histories corresponding to precondi
tioned or unpreconditioned GMRES. For the restarted iteration, we restarted
the algorithm after three iterations, GMRES(3) in the terminology of ¸ 3.4.
Note that the plots are semilog plots of r
2
/b
2
and that the righthand
sides in the preconditioned and unpreconditioned cases are not the same. In
both ﬁgures the solid line is the history of the preconditioned iteration and the
dashed line that of the unpreconditioned. Note the importance of precondi
tioning. The MATLAB flops command indicates that the unpreconditioned
iteration required 7.6 million ﬂoating point operations and converged in 56
iterations. The preconditioned iteration required 1.3 million ﬂoating point op
erations and terminated after 8 iterations. Restarting after 3 iterations is more
costly in terms of the iteration count than not restarting. However, even in
the unpreconditioned case, the restarted iteration terminated successfully. In
this case the MATLAB flops command indicates that the unpreconditioned
iteration terminated after 223 iterations and 9.7 million ﬂoating point opera
tions and the preconditioned iteration required 13 iterations and 2.6 million
ﬂoating point operations.
The execution times in our environment indicate that the preconditioned
iterations are even faster than the diﬀerence in operation counts indicates.
The reason for this is that the FFT routine is a MATLAB intrinsic and is not
interpreted MATLAB code. In other computing environments preconditioners
which vectorize or parallelize particularly well might perform in the same way.
3.8. Examples for CGNR, BiCGSTAB, and TFQMR iteration
When a transpose is available, CGNR (or CGNE) is an alternative to GMRES
that does not require storage of the iteration history. For elliptic problems like
(3.33), the transpose (adjoint) of L is given by
L
∗
u = −u
xx
−u
yy
−a
1
u
x
−a
2
u
y
+a
3
u
as one can derive by integration by parts. To apply CGNR to the precondi
tioned problem, we require the transpose of GL, which is given by
(GL)
∗
u = L
∗
G
∗
u = L
∗
Gu.
The cost of the application of the transpose of L or GL is the same as that
of the application of L or GL itself. Hence, in the case where the cost of the
matrixvector product dominates the computation, a single CGNR iteration
costs roughly the same as two GMRES iterations.
However, CGNR has the eﬀect of squaring the condition number, this eﬀect
has serious consequences as Fig. 3.3 shows. The dashed line corresponds to
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
56 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
0 10 20 30 40 50 60
10
4
10
3
10
2
10
1
10
0
Iterations
R
e
l
a
t
i
v
e
R
e
s
i
d
u
a
l
Fig. 3.1. GMRES for 2D elliptic equation.
0 50 100 150 200 250
10
4
10
3
10
2
10
1
10
0
Iterations
R
e
l
a
t
i
v
e
R
e
s
i
d
u
a
l
Fig. 3.2. GMRES(3) for 2D elliptic equation.
CGNR on the unpreconditioned problem, i.e., CG applied to L
∗
Lu = L
∗
f.
The matrix corresponding to the discretization of L
∗
L has a large condition
number. The solid line corresponds to the CGNR applied to the preconditioned
problem, i.e., CG applied to L
∗
G
2
Lu = L
∗
G
2
f. We limited the number of
iterations to 310 = 10m and the unpreconditioned formulation of CGNR had
made very little progress after 310 iterations. This is a result of squaring
the large condition number. The preconditioned formulation did quite well,
converging in eight iterations. The MATLAB flops command indicates that
the unpreconditioned iteration required 13.7 million ﬂoatingpoint operations
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 57
and the preconditioned iteration 2.4 million. However, as you are asked to
investigate in Exercise 3.9.8, the behavior of even the preconditioned iteration
can also be poor if the coeﬃcients of the ﬁrst derivative terms are too large.
CGNE has similar performance; see Exercise 3.9.9.
0 50 100 150 200 250 300 350
10
4
10
3
10
2
10
1
10
0
Iterations
R
e
l
a
t
i
v
e
R
e
s
i
d
u
a
l
Fig. 3.3. CGNR for 2D elliptic equation.
The examples for BiCGSTAB use the code bicgstab from the collection
of MATLAB codes. This code has the same input/output arguments as gmres.
We applied BiCGSTAB to both the preconditioned and unpreconditioned
forms of (3.33) with the same data and termination criterion as for the GMRES
example.
We plot the iteration histories in Fig. 3.4. As in the previous examples, the
solid line corresponds to the preconditioned iteration and the dashed line to
the unpreconditioned iteration. Neither convergence history is monotonic, with
the unpreconditioned iteration being very irregular, but not varying by many
orders of magnitude as a CGS or BiCG iteration might. The preconditioned
iteration terminated in six iterations (12 matrixvector products) and required
roughly 1.7 million ﬂoatingpoint operations. The unpreconditioned iteration
took 40 iterations and 2.2 million ﬂoatingpoint operations. We see a
considerable improvement in cost over CGNR and, since our unpreconditioned
matrixvector product is so inexpensive, a better cost/iterate than GMRES in
the unpreconditioned case.
We repeat the experiment with TFQMR as our linear solver. We use
the code tfqmr that is included in our collection. In Fig. 3.5 we plot the
convergence histories in terms of the approximate residual
√
2k + 1τ
2k
given
by (3.31). We plot only full iterations (m = 2k) except, possibly, for the
ﬁnal iterate. The approximate residual is given in terms of the quasiresidual
norm τ
m
rather than the true residual, and the graphs are monotonic. Our
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
58 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
0 5 10 15 20 25 30 35 40
10
4
10
3
10
2
10
1
10
0
10
1
Iterations
R
e
l
a
t
i
v
e
R
e
s
i
d
u
a
l
Fig. 3.4. BiCGSTAB for 2D elliptic equation.
0 10 20 30 40 50 60 70
10
5
10
4
10
3
10
2
10
1
10
0
Iterations
R
e
l
a
t
i
v
e
A
p
p
r
o
x
i
m
a
t
e
R
e
s
i
d
u
a
l
Fig. 3.5. TFQMR for 2D elliptic equation.
termination criterion is based on the quasiresidual and (3.31) as described in
¸ 3.6.4, stopping the iteration when τ
m
√
m+ 1/b
2
≤ h
−2
. This led to actual
relative residuals of 7 10
−5
for the unpreconditioned iteration and 2 10
−4
for the preconditioned iteration, indicating that it is reasonable to use the
quasiresidual estimate (3.31) to make termination decisions.
The unpreconditioned iteration required roughly 4.1 million ﬂoatingpoint
operations and 67 iterations for termination. The preconditioned iteration
was much more eﬃcient, taking 1.9 million ﬂoatingpoint operations and 7
iterations. The plateaus in the graph of the convergence history for the
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 59
unpreconditioned iteration are related (see [196]) to spikes in the corresponding
graph for a CGS iteration.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
60 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
3.9. Exercises on GMRES
3.9.1. Give an example of a matrix that is not diagonalizable.
3.9.2. Prove Theorem 3.1.4 and Theorem 3.1.5.
3.9.3. Prove that the Arnoldi process (unless it terminates prematurely with a
solution) produces a basis for /
k+1
that satisﬁes (3.8).
3.9.4. Duplicate Table 3.1 and add two more columns corresponding to classical
Gram–Schmidt orthogonalization with reorthogonalization at every step
and determine when the test in (3.10) detects loss of orthogonality. How
do your conclusions compare with those in [160]?
3.9.5. Let A = J
λ
be an elementary Jordan block of size N corresponding to
an eigenvalue λ ,= 0. This means that
J
λ
=
_
_
_
_
_
_
_
_
_
_
λ 1 0
0 λ 1 0
.
.
.
.
.
.
.
.
.
.
.
.
0 λ 1 0
0 λ 1
0 λ
_
_
_
_
_
_
_
_
_
_
.
Give an example of x
0
, b ∈ R
N
, and λ ∈ R such that the GMRES
residuals satisfy r
k

2
= r
0

2
for all k < N.
3.9.6. Consider the centered diﬀerence discretization of
−u
+u
+u = 1, u(0) = u(1) = 0.
Solve this problem with GMRES (without preconditioning) and then
apply as a preconditioner the map M deﬁned by
−(Mf)
= f, (Mf)(0) = (Mf)(1) = 0.
That is, precondition with a solver for the highorder term in the
diﬀerential equation using the correct boundary conditions. Try this
for meshes with 50, 100, 200, . . . points. How does the performance of
the iteration depend on the mesh? Try other preconditioners such as
Gauss–Seidel and Jacobi. How do other methods such as CGNR, CGNE,
BiCGSTAB, and TFQMR perform?
3.9.7. Modify the MATLAB GMRES code to allow for restarts. One way to do
this is to call the gmres code with another code which has an outer loop
to control the restarting as in Algorithm gmresm. See [167] for another
example. How would you test for failure in a restarted algorithm? What
kinds of failures can occur? Repeat Exercise 6 with a limit of three
GMRES iterations before a restart.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GMRES ITERATION 61
3.9.8. Duplicate the results in ¸ 3.7 and ¸ 3.8. Compare the execution times
in your computing environment. Experiment with diﬀerent values of m.
Why might GMRES(m) be faster than GMRES? Vary the dimension and
the coeﬃcient a
y
. Try a
y
= 5y. Why is the performance of the methods
sensitive to a
y
and a
x
?
3.9.9. Apply CGNE to the PDE in ¸ 3.7. How does the performance diﬀer from
CGNR and GMRES?
3.9.10. Consider preconditioning from the right. For the PDE in ¸ 3.7, apply
GMRES, CGNR, BiCGSTAB, TFQMR, and/or CGNE, to the equation
LGw = f and then recover the solution from u = Gw. Compare timings
and solution accuracy with those obtained by left preconditioning.
3.9.11. Are the solutions to (3.33) in ¸ 3.7 and ¸ 3.8 equally accurate? Compare
the ﬁnal results with the known solution.
3.9.12. Implement BiCG and CGS and try them on some of the examples in
this chapter.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
62 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Chapter 4
Basic Concepts and FixedPoint Iteration
This part of the book is about numerical methods for the solution of systems
of nonlinear equations. Throughout this part, we will let   denote a norm
on R
N
as well as the induced matrix norm.
We begin by setting the notation. We seek to solve
F(x) = 0. (4.1)
Here F : R
N
→ R
N
. We denote the ith component of F by f
i
. If the
components of F are diﬀerentiable at x ∈ R
N
we deﬁne the Jacobian matrix
F
(x) by
F
(x)
ij
=
∂f
i
∂x
j
(x).
The Jacobian matrix is the vector analog of the derivative. We may express
the fundamental theorem of calculus as follows.
Theorem 4.0.1. Let F be diﬀerentiable in an open set Ω ⊂ R
N
and let
x
∗
∈ Ω. Then for all x ∈ Ω suﬃciently near x
∗
F(x) −F(x
∗
) =
_
1
0
F
(x
∗
+t(x −x
∗
))(x −x
∗
) dt.
4.1. Types of convergence
Iterative methods can be classiﬁed by the rate of convergence.
Definition 4.1.1. Let ¦x
n
¦ ⊂ R
N
and x
∗
∈ R
N
. Then
• x
n
→ x
∗
qquadratically if x
n
→ x
∗
and there is K > 0 such that
x
n+1
−x
∗
 ≤ Kx
n
−x
∗

2
.
• x
n
→ x
∗
qsuperlinearly with qorder α > 1 if x
n
→ x
∗
and there is
K > 0 such that
x
n+1
−x
∗
 ≤ Kx
n
−x
∗

α
.
• x
n
→ x
∗
qsuperlinearly if
lim
n→∞
x
n+1
−x
∗

x
n
−x
∗

= 0.
65
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
66 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
• x
n
→ x
∗
qlinearly with qfactor σ ∈ (0, 1) if
x
n+1
−x
∗
 ≤ σx
n
−x
∗

for n suﬃciently large.
Definition 4.1.2. An iterative method for computing x
∗
is said to be
locally (qquadratically, qsuperlinearly, qlinearly, etc.) convergent if the
iterates converge to x
∗
(qquadratically, qsuperlinearly, qlinearly, etc.) given
that the initial data for the iteration is suﬃciently good.
Note that a qsuperlinearly convergent sequence is also qlinearly conver
gent with qfactor σ for any σ > 0. A qquadratically convergent sequence is
qsuperlinearly convergent with qorder 2.
In general, a qsuperlinearly convergent method is preferable to a qlinearly
convergent one if the cost of a single iterate is the same for both methods. We
will see that often a method that is more slowly convergent, in terms of the
convergence types deﬁned above, can have a cost/iterate so low that the slow
iteration is more eﬃcient.
Sometimes errors are introduced into the iteration that are independent of
the errors in the approximate solution. An example of this would be a problem
in which the nonlinear function F is the output of another algorithm that
provides error control, such as adjustment of the mesh size in an approximation
of a diﬀerential equation. One might only compute F to low accuracy in the
initial phases of the iteration to compute a solution and tighten the accuracy
as the iteration progresses. The rtype convergence classiﬁcation enables us to
describe that idea.
Definition 4.1.3. Let ¦x
n
¦ ⊂ R
N
and x
∗
∈ R
N
. Then ¦x
n
¦ converges
to x
∗
r(quadratically, superlinearly, linearly) if there is a sequence ¦ξ
n
¦ ⊂ R
converging q(quadratically, superlinearly, linearly) to zero such that
x
n
−x
∗
 ≤ ξ
n
.
We say that ¦x
n
¦ converges rsuperlinearly with rorder α > 1 if ξ
n
→ 0
qsuperlinearly with qorder α.
4.2. Fixedpoint iteration
Many nonlinear equations are naturally formulated as ﬁxedpoint problems
x = K(x) (4.2)
where K, the ﬁxedpoint map, may be nonlinear. A solution x
∗
of (4.2) is
called a ﬁxed point of the map K. Such problems are nonlinear analogs of the
linear ﬁxedpoint problems considered in Chapter 1. In this section we analyze
convergence of ﬁxedpoint iteration, which is given by
x
n+1
= K(x
n
). (4.3)
This iterative method is also called nonlinear Richardson iteration, Picard
iteration, or the method of successive substitution.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
FIXEDPOINT ITERATION 67
Before discussing convergence of ﬁxedpoint iteration we make two deﬁni
tions.
Definition 4.2.1. Let Ω ⊂ R
N
and let G : Ω → R
M
. G is Lipschitz
continuous on Ω with Lipschitz constant γ if
G(x) −G(y) ≤ γx −y
for all x, y ∈ Ω.
Definition 4.2.2. Let Ω ⊂ R
N
. K : Ω → R
N
is a contraction mapping
on Ω if K is Lipschitz continuous on Ω with Lipschitz constant γ < 1.
The standard result for ﬁxedpoint iteration is the Contraction Mapping
Theorem [9]. Compare the proof to that of Lemma 1.2.1 and Corollary 1.2.1.
Theorem 4.2.1. Let Ω be a closed subset of R
N
and let K be a contraction
mapping on Ω with Lipschitz constant γ < 1 such that K(x) ∈ Ω for all x ∈ Ω.
Then there is a unique ﬁxed point of K, x
∗
∈ Ω, and the iteration deﬁned by
(4.3) converges qlinearly to x
∗
with qfactor γ for all initial iterates x
0
∈ Ω.
Proof. Let x
0
∈ Ω. Note that ¦x
n
¦ ⊂ Ω because x
0
∈ Ω and K(x) ∈ Ω
whenever x ∈ Ω. The sequence ¦x
n
¦ remains bounded since for all i ≥ 1
x
i+1
−x
i
 = K(x
i
) −K(x
i−1
) ≤ γx
i
−x
i−1
 . . . ≤ γ
i
x
1
−x
0
,
and therefore
x
n
−x
0
 = 
n−1
i=0
x
i+1
−x
i

≤
n−1
i=0
x
i+1
−x
i
 ≤ x
1
−x
0

n−1
i=0
γ
i
≤ x
1
−x
0
/(1 −γ).
Now, for all n, k ≥ 0,
x
n+k
−x
n
 = K(x
n+k−1
) −K(x
n−1
)
≤ γx
n+k−1
−x
n−1

≤ γK(x
n+k−2
) −K(x
n−2
)
≤ γ
2
x
n+k−2
−x
n−2
 ≤ . . . ≤ γ
n
x
k
−x
0

≤ γ
n
x
1
−x
0
/(1 −γ).
Hence
lim
n,k→∞
x
n+k
−x
n
 = 0
and therefore the sequence ¦x
n
¦ is a Cauchy sequence and has a limit x
∗
[162].
If K has two ﬁxed points x
∗
and y
∗
in Ω, then
x
∗
−y
∗
 = K(x
∗
) −K(y
∗
) ≤ γx
∗
−y
∗
.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
68 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Since γ < 1 this can only hold if x
∗
−y
∗
 = 0, i. e. x
∗
= y
∗
. Hence the ﬁxed
point is unique.
Finally we note that
x
n+1
−x
∗
 = K(x
n
) −K(x
∗
) ≤ γx
n
−x
∗
,
which completes the proof.
One simple application of this theorem is to numerical integration of
ordinary diﬀerential equations by implicit methods. For example, consider
the initial value problem
y
= f(y), y(t
0
) = y
0
and its solution by the backward Euler method with time step h,
y
n+1
−y
n
= hf(y
n+1
). (4.4)
In (4.4) y
k
denotes the approximation of y(t
0
+ kh). Advancing in time from
y
n
to y
n+1
requires solution of the ﬁxedpoint problem
y = K(y) = y
n
+hf(y).
For simplicity we assume that f is bounded and Lipschitz continuous, say
[f(y)[ ≤ M and [f(x) −f(y)[ ≤ M[x −y[
for all x, y.
If hM < 1, we may apply the contraction mapping theorem with Ω = R
since for all x, y
K(x) −K(y) = h[f(x) −f(y)[ ≤ hM[x −y[.
From this we conclude that for h suﬃciently small we can integrate forward
in time and that the time step h can be chosen independently of y. This time
step may well be too small to be practical, and the methods based on Newton’s
method which we present in the subsequent chapters are used much more often
[83]. This type of analysis was used by Picard to prove existence of solutions
of initial value problems [152]. Nonlinear variations of the classical stationary
iterative methods such as Gauss–Seidel have also been used; see [145] for a
more complete discussion of these algorithms.
4.3. The standard assumptions
We will make the standard assumptions on F.
Assumption 4.3.1.
1. Equation 4.1 has a solution x
∗
.
2. F
: Ω → R
N×N
is Lipschitz continuous with Lipschitz constant γ.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
FIXEDPOINT ITERATION 69
3. F
(x
∗
) is nonsingular.
These assumptions can be weakened [108] without sacriﬁcing convergence
of the methods we consider here. However the classical result on quadratic
convergence of Newton’s method requires them and we make them throughout.
Exercise 5.7.1 illustrates one way to weaken the standard assumptions.
Throughout this part, we will always denote a root of F by x
∗
. We let
B(r) denote the ball of radius r about x
∗
B(r) = ¦x[ e < r¦,
where
e = x −x
∗
.
The notation introduced above will be used consistently. If x
n
is the nth iterate
of a sequence, e
n
= x
n
−x
∗
is the error in that iterate.
Lemma 4.3.1 is an important consequence of the standard assumptions.
Lemma 4.3.1. Assume that the standard assumptions hold. Then there is
δ > 0 so that for all x ∈ B(δ)
F
(x) ≤ 2F
(x
∗
), (4.5)
F
(x)
−1
 ≤ 2F
(x
∗
)
−1
, (4.6)
and
F
(x
∗
)
−1

−1
e/2 ≤ F(x) ≤ 2F
(x
∗
)e. (4.7)
Proof. Let δ be small enough so that B(δ) ⊂ Ω. For all x ∈ B(δ) we have
F
(x) ≤ F
(x
∗
) +γe.
Hence (4.5) holds if γδ < F
(x
∗
).
The next result (4.6) is a direct consequence of the Banach Lemma if
I −F
(x
∗
)
−1
F
(x) < 1/2.
This will follow from
δ <
F
(x
∗
)
−1

−1
2γ
since then
I −F
(x
∗
)
−1
F
(x) = F
(x
∗
)
−1
(F
(x
∗
) −F
(x))
≤ γF
(x
∗
)
−1
e ≤ γδF
(x
∗
)
−1
 < 1/2.
(4.8)
To prove the ﬁnal inequality (4.7), we note that if x ∈ B(δ) then x
∗
+te ∈
B(δ) for all 0 ≤ t ≤ 1. We use (4.5) and Theorem 4.0.1 to obtain
F(x) ≤
_
1
0
F
(x
∗
+te)e dt ≤ 2F
(x
∗
)e
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
70 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
which is the right inequality in (4.7).
To prove the left inequality in (4.7) note that
F
(x
∗
)
−1
F(x) = F
(x
∗
)
−1
_
1
0
F
(x
∗
+te)e dt
= e −
_
1
0
(I −F
(x
∗
)
−1
F
(x
∗
+te))e dt,
and hence, by (4.8)
F
(x
∗
)
−1
F(x) ≥ e(1 −
_
1
0
I −F
(x
∗
)
−1
F
(x
∗
+te)dt) ≥ e/2.
Therefore
e/2 ≤ F
(x
∗
)
−1
F(x) ≤ F
(x
∗
)
−1
F(x),
which completes the proof.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Chapter 5
Newton’s Method
5.1. Local convergence of Newton’s method
We will describe iterative methods for nonlinear equations in terms of the
transition from a current iterate x
c
to a new iterate x
+
. In this language,
Newton’s method is
x
+
= x
c
−F
(x
c
)
−1
F(x
c
). (5.1)
We may also view x
+
as the root of the twoterm Taylor expansion or linear
model of F about x
c
M
c
(x) = F(x
c
) +F
(x
c
)(x −x
c
).
In the context of single systems this method appeared in the 17th century
[140], [156], [13], [137].
The convergence result on Newton’s method follows from Lemma 4.3.1.
Theorem 5.1.1. Let the standard assumptions hold. Then there are K > 0
and δ > 0 such that if x
c
∈ B(δ) the Newton iterate from x
c
given by (5.1)
satisﬁes
e
+
 ≤ Ke
c

2
. (5.2)
Proof. Let δ be small enough so that the conclusions of Lemma 4.3.1 hold.
By Theorem 4.0.1
e
+
= e
c
−F
(x
c
)
−1
F(x
c
) = F
(x
c
)
−1
_
1
0
(F
(x
c
) −F
(x
∗
+te
c
))e
c
dt.
By Lemma 4.3.1 and the Lipschitz continuity of F
e
+
 ≤ (2F
(x
∗
)
−1
)γe
c

2
/2.
This completes the proof of (5.2) with K = γF
(x
∗
)
−1
.
The proof of convergence of the complete Newton iteration will be complete
if we reduce δ if needed so that Kδ < 1.
Theorem 5.1.2. Let the standard assumptions hold. Then there is δ such
that if x
0
∈ B(δ) the Newton iteration
x
n+1
= x
n
−F
(x
n
)
−1
F(x
n
)
converges qquadratically to x
∗
.
71
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
72 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Proof. Let δ be small enough so that the conclusions of Theorem 5.1.1
hold. Reduce δ if needed so that Kδ = η < 1. Then if n ≥ 0 and x
n
∈ B(δ),
Theorem 5.1.1 implies that
e
n+1
 ≤ Ke
n

2
≤ ηe
n
 < e
n
 (5.3)
and hence x
n+1
∈ B(ηδ) ⊂ B(δ). Therefore if x
n
∈ B(δ) we may continue the
iteration. Since x
0
∈ B(δ) by assumption, the entire sequence ¦x
n
¦ ⊂ B(δ).
(5.3) then implies that x
n
→ x
∗
qquadratically.
The assumption that the initial iterate be “suﬃciently near” the solution
(x
0
∈ B(δ)) may seem artiﬁcial at ﬁrst look. There are, however, many
situations in which the initial iterate is very near the root. Two examples are
implicit integration of ordinary diﬀerential equations and diﬀerential algebraic
equations, [105], [83], [16], where the initial iterate is derived from the solution
at the previous time step, and solution of discretizations of partial diﬀerential
equations, where the initial iterate is an interpolation of a solution from a
coarser computational mesh [99], [126]. Moreover, when the initial iterate is
far from the root and methods such as those discussed in Chapter 8 are needed,
local convergence results like those in this and the following two chapters
describe the terminal phase of the iteration.
5.2. Termination of the iteration
We can base our termination decision on Lemma 4.3.1. If x ∈ B(δ), where δ
is small enough so that the conclusions of Lemma 4.3.1 hold, then if F
(x
∗
) is
well conditioned, we may terminate the iteration when the relative nonlinear
residual F(x)/F(x
0
) is small. We have, as a direct consequence of applying
Lemma 4.3.1 twice, a nonlinear form of Lemma 1.1.1.
Lemma 5.2.1. Assume that the standard assumptions hold. Let δ > 0 be
small enough so that the conclusions of Lemma 4.3.1 hold for x ∈ B(δ). Then
for all x ∈ B(δ)
e
4e
0
κ(F
(x
∗
))
≤
F(x)
F(x
0
)
≤
4κ(F
(x
∗
))e
e
0

,
where κ(F
(x
∗
)) = F
(x
∗
)F
(x
∗
)
−1
 is the condition number of F
(x
∗
)
relative to the norm  .
From Lemma 5.2.1 we conclude that if F
(x
∗
) is well conditioned, the size
of the relative nonlinear residual is a good indicator of the size of the error.
However, if there is error in the evaluation of F or the initial iterate is near a
root, a termination decision based on the relative residual may be made too
late in the iteration or, as we will see in the next section, the iteration may
not terminate at all. We raised this issue in the context of linear equations
in Chapter 1 when we compared (1.2) and (1.4). In the nonlinear case, there
is no “righthand side” to use as a scaling factor and we must balance the
relative and absolute size of the nonlinear residuals in some other way. In all
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 73
of our MATLAB codes and algorithms for nonlinear equations our termination
criterion is to stop the iteration if
F(x) ≤ τ
r
F(x
0
) +τ
a
, (5.4)
where the relative error tolerance τ
r
and absolute error tolerance τ
a
are input
to the algorithm. Combinations of relative and absolute error tolerances are
commonly used in numerical methods for ordinary diﬀerential equations or
diﬀerential algebraic equations [16], [21], [23], [151].
Another way to decide whether to terminate is to look at the Newton step
s = −F
(x
c
)
−1
F(x
c
) = x
+
−x
c
,
and terminate the iteration when s is suﬃciently small. This criterion is
based on Theorem 5.1.1, which implies that
e
c
 = s +O(e
c

2
).
Hence, near the solution s and e
c
are essentially the same size.
For methods other than Newton’s method, the relation between the step
and the error is less clear. If the iteration is qlinearly convergent, say, then
e
+
 ≤ σe
c

implies that
(1 −σ)e
c
 ≤ s ≤ (1 +σ)e
c
.
Hence the step is a reliable indicator of the error provided σ is not too near
1. The diﬀerential equations codes discussed in [16], [21], [23], and [151], make
use of this in the special case of the chord method.
However, in order to estimate e
c
 this way, one must do most of the work
needed to compute x
+
, whereas by terminating on small relative residuals or
on a condition like (5.4) the decision to stop the iteration can be made before
computation and factorization of F
(x
c
). If computation and factorization of
the Jacobian are inexpensive, then termination on small steps becomes more
attractive.
5.3. Implementation of Newton’s method
In this section we assume that direct methods will be used to solve the linear
equation for the Newton step. Our examples and discussions of operation
counts make the implicit assumption that the Jacobian is dense. However,
the issues for sparse Jacobians when direct methods are used to solve linear
systems are very similar. In Chapter 6 we consider algorithms in which iterative
methods are used to solve the equation for the Newton step.
In order to compute the Newton iterate x
+
from a current point x
c
one
must ﬁrst evaluate F(x
c
) and decide whether to terminate the iteration. If
one decides to continue, the Jacobian F
(x
c
) must be computed and factored.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
74 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Then the step is computed as the solution of F
(x
c
)s = −F(x
c
) and the iterate
is updated x
+
= x
c
+s. Of these steps, the evaluation and factorization of F
are the most costly. Factorization of F
in the dense case costs O(N
3
) ﬂoating
point operations. Evaluation of F
by ﬁnite diﬀerences should be expected to
cost N times the cost of an evaluation of F because each column in F
requires
an evaluation of F to form the diﬀerence approximation. Hence the cost of a
Newton step may be roughly estimated as N + 1 evaluations of F and O(N
3
)
ﬂoatingpoint operations. In many cases F
can be computed more eﬃciently,
accurately, and directly than with diﬀerences and the analysis above for the
cost of a Newton iterate is very pessimistic. See Exercise 5.7.21 for an example.
For general nonlinearities or general purpose codes such as the MATLAB code
nsol from the collection, there may be little alternative to diﬀerence Jacobians.
In the future, automatic diﬀerentiation (see [94] for a collection of articles on
this topic) may provide such an alternative.
For deﬁniteness in the description of Algorithm newton we use an LU
factorization of F
. Any other appropriate factorization such as QR or
Cholesky could be used as well. The inputs of the algorithm are the initial
iterate x, the nonlinear map F, and a vector of termination tolerances
τ = (τ
r
, τ
a
) ∈ R
2
.
Algorithm 5.3.1. newton(x, F, τ)
1. r
0
= F(x)
2. Do while F(x) > τ
r
r
0
+τ
a
(a) Compute F
(x)
(b) Factor F
(x) = LU.
(c) Solve LUs = −F(x)
(d) x = x +s
(e) Evaluate F(x).
One approach to reducing the cost of items 2a and item 2b in the Newton
iteration is to move them outside of the main loop. This means that the linear
approximation of F(x) = 0 that is solved at each iteration has derivative
determined by the initial iterate.
x
+
= x
c
−F
(x
0
)
−1
F(x
c
).
This method is called the chord method. The inputs are the same as for
Newton’s method.
Algorithm 5.3.2. chord(x, F, τ)
1. r
0
= F(x)
2. Compute F
(x)
3. Factor F
(x) = LU.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 75
4. Do while F(x) > τ
r
r
0
+τ
a
(a) Solve LUs = −F(x)
(b) x = x +s
(c) Evaluate F(x).
The only diﬀerence in implementation from Newton’s method is that the
computation and factorization of the Jacobian are done before the iteration
is begun. The diﬀerence in the iteration itself is that an approximation to
F
(x
c
) is used. Similar diﬀerences arise if F
(x
c
) is numerically approximated
by diﬀerences. We continue in the next section with an analysis of the eﬀects
of errors in the function and Jacobian in a general context and then consider
the chord method and diﬀerence approximation to F
as applications.
5.4. Errors in the function and derivative
Suppose that F and F
are computed inaccurately so that F + and F
+ ∆
are used instead of F and F
in the iteration. If ∆ is suﬃciently small, the
resulting iteration can return a result that is an O() accurate approximation
to x
∗
. This is diﬀerent from convergence and was called “local improvement”
in [65]. These issues are discussed in [201] as well. If, for example, is entirely
due to ﬂoatingpoint roundoﬀ, there is no reason to expect that F(x
n
) will
ever be smaller than in general. We will refer to this phase of the iteration
in which the nonlinear residual is no longer being reduced as stagnation of the
iteration.
Theorem 5.4.1. Let the standard assumptions hold. Then there are
¯
K > 0, δ > 0, and δ
1
> 0 such that if x
c
∈ B(δ) and ∆(x
c
) < δ
1
then
x
+
= x
c
−(F
(x
c
) + ∆(x
c
))
−1
(F(x
c
) +(x
c
))
is deﬁned (i.e., F
(x
c
) + ∆(x
c
) is nonsingular) and satisﬁes
e
+
 ≤
¯
K(e
c

2
+∆(x
c
)e
c
 +(x
c
)). (5.5)
Proof. Let δ be small enough so that the conclusions of Lemma 4.3.1 and
Theorem 5.1.1 hold. Let
x
N
+
= x
c
−F
(x
c
)
−1
F(x
c
)
and note that
x
+
= x
N
+
+ (F
(x
c
)
−1
−(F
(x
c
) + ∆(x
c
))
−1
)F(x
c
) −(F
(x
c
) + ∆(x
c
))
−1
(x
c
).
Lemma 4.3.1 and Theorem 5.1.1 imply
e
+
 ≤ Ke
c

2
+ 2F
(x
c
)
−1
−(F
(x
c
) + ∆(x
c
))
−1
)F
(x
∗
)e
c

+(F
(x
c
) + ∆(x
c
))
−1
(x
c
).
(5.6)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
76 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
If
∆(x
c
) ≤ F
(x
∗
)
−1

−1
/4
then Lemma 4.3.1 implies that
∆(x
c
) ≤ F
(x
c
)
−1

−1
/2
and the Banach Lemma implies that F
(x
c
) + ∆(x
c
) is nonsingular and
(F
(x
c
) + ∆(x
c
))
−1
 ≤ 2F
(x
c
)
−1
 ≤ 4F
(x
∗
)
−1
.
Hence
F
(x
c
)
−1
−(F
(x
c
) + ∆(x
c
))
−1
 ≤ 8F
(x
∗
)
−1

2
∆(x
c
).
We use these estimates and (5.6) to obtain
e
+
 ≤ Ke
c

2
+ 16F
(x
∗
)
−1

2
F
(x
∗
)∆(x
c
)e
c
 + 4F
(x
∗
)
−1
(x
c
).
Setting
¯
K = K + 16F
(x
∗
)
−1

2
F
(x
∗
) + 4F
(x
∗
)
−1

completes the proof.
The remainder of this section is devoted to applications of Theorem 5.4.1.
5.4.1. The chord method. Recall that the chord method is given by
x
+
= x
c
−F
(x
0
)
−1
F(x
c
).
In the language of Theorem 5.4.1
(x
c
) = 0, ∆(x
c
) = F
(x
0
) −F
(x
c
).
Hence, if x
c
, x
0
∈ B(δ) ⊂ Ω
∆(x
c
) ≤ γx
0
−x
c
 ≤ γ(e
0
 +e
c
). (5.7)
We apply Theorem 5.4.1 to obtain the following result.
Theorem 5.4.2. Let the standard assumptions hold. Then there are
K
C
> 0 and δ > 0 such that if x
0
∈ B(δ) the chord iterates converge qlinearly
to x
∗
and
e
n+1
 ≤ K
C
e
0
e
n
. (5.8)
Proof. Let δ be small enough so that B(δ) ⊂ Ω and the conclusions of
Theorem 5.4.1 hold. Assume that x
n
∈ B(δ). Combining (5.7) and (5.5)
implies
e
n+1
 ≤
¯
K(e
n
(1 +γ) +γe
0
)e
n
 ≤
¯
K(1 + 2γ)δe
n
.
Hence if δ is small enough so that
¯
K(1 + 2γ)δ = η < 1
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 77
then the chord iterates converge qlinearly to x
∗
. Qlinear convergence implies
that e
n
 < e
0
 and hence (5.8) holds with K
C
=
¯
K(1 + 2γ).
Another variation of the chord method is
x
+
= x
c
−A
−1
F(x
c
),
where A ≈ F
(x
∗
). Methods of this type may be viewed as preconditioned
nonlinear Richardson iteration. Since
∆(x
c
) = A−F
(x
c
) ≤ A−F
(x
∗
) +F
(x
∗
) −F
(x
c
),
if x
c
∈ B(δ) ⊂ Ω then
∆(x
c
) ≤ A−F
(x
∗
) +γe
c
 ≤ A−F
(x
∗
) +γδ.
Theorem 5.4.3. Let the standard assumptions hold. Then there are
K
A
> 0, δ > 0, and δ
1
> 0, such that if x
0
∈ B(δ) and A−F
(x
∗
) < δ
1
then
the iteration
x
n+1
= x
n
−A
−1
F(x
n
)
converges qlinearly to x
∗
and
e
n+1
 ≤ K
A
(e
0
 +A−F
(x
∗
))e
n
. (5.9)
5.4.2. Approximate inversion of F
. Another way to implement chord
type methods is to provide an approximate inverse of F
. Here we replace
F
(x)
−1
by B(x), where the action of B on a vector is less expensive to compute
than a solution using the LU factorization. Rather than express the iteration
in terms of B
−1
(x) −F
(x) and using Theorem 5.4.3 one can proceed directly
from the deﬁnition of approximate inverse.
Note that if B is constant (independent of x) then the iteration
x
+
= x
c
−BF(x
c
)
can be viewed as a preconditioned nonlinear Richardson iteration.
We have the following result.
Theorem 5.4.4. Let the standard assumptions hold. Then there are
K
B
> 0, δ > 0, and δ
1
> 0, such that if x
0
∈ B(δ) and the matrixvalued
function B(x) satisﬁes
I −B(x)F
(x
∗
) = ρ(x) < δ
1
(5.10)
for all x ∈ B(δ) then the iteration
x
n+1
= x
n
−B(x
n
)F(x
n
)
converges qlinearly to x
∗
and
e
n+1
 ≤ K
B
(ρ(x
n
) +e
n
)e
n
. (5.11)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
78 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Proof. On one hand, this theorem could be regarded as a corollary of the
Banach Lemma and Theorem 5.4.1. We give a direct proof.
First, by (5.10) we have
B(x) = B(x)F
(x
∗
)F
(x
∗
)
−1
 ≤ B(x)F
(x
∗
)F
(x
∗
)
−1

≤ M
B
= (1 +δ
1
)F
(x
∗
)
−1
.
(5.12)
Using (5.12) and
e
+
= e
c
−B(x
c
)F(x
c
) =
_
1
0
(I −B(x
c
)F
(x
∗
+te
c
))e
c
dt
= (I −B(x
c
)F
(x
∗
))e
c
+B(x
c
)
_
1
0
(F
(x
∗
) −F
(x
∗
+te
c
))e
c
dt
we have
e
+
 ≤ ρ(x
c
)e
c
 +M
B
γe
c

2
/2.
This completes the proof with K
B
= 1 +M
B
γ/2.
5.4.3. The Shamanskii method. Alternation of a Newton step with a
sequence of chord steps leads to a class of highorder methods, that is, methods
that converge qsuperlinearly with qorder larger than 2. We follow [18] and
name this method for Shamanskii [174], who considered the ﬁnite diﬀerence
Jacobian formulation of the method from the point of view of eﬃciency. The
method itself was analyzed in [190]. Other highorder methods, especially for
problems in one dimension, are described in [190] and [146].
We can describe the transition from x
c
to x
+
by
y
1
= x
c
−F
(x
c
)
−1
F(x
c
),
y
j+1
= y
j
−F
(x
c
)
−1
F(y
j
) for 1 ≤ j ≤ m−1,
x
+
= y
m
.
(5.13)
Note that m = 1 is Newton’s method and m = ∞ is the chord method with
¦y
j
¦ playing the role of the chord iterates.
Algorithmically a second loop surrounds the factor/solve step. The inputs
for the Shamanskii method are the same as for Newton’s method or the chord
method except for the addition of the parameter m. Note that we can overwrite
x
c
with the sequence of y
j
’s. Note that we apply the termination test after
computation of each y
j
.
Algorithm 5.4.1. sham(x, F, τ, m)
1. r
0
= F(x)
2. Do while F(x) > τ
r
r
0
+τ
a
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 79
(a) Compute F
(x)
(b) Factor F
(x) = LU.
(c) for j = 1, . . . m
i. Solve LUs = −F(x)
ii. x = x +s
iii. Evaluate F(x).
iv. If F(x) ≤ τ
r
r
0
+τ
a
exit.
The convergence result is a simple application of Theorem 5.4.2.
Theorem 5.4.5. Let the standard assumptions hold and let m ≥ 1 be
given. Then there are K
S
> 0 and δ > 0 such that if x
0
∈ B(δ) the Shamanskii
iterates converge qsuperlinearly to x
∗
with qorder m+ 1 and
e
n+1
 ≤ K
S
e
n

m+1
. (5.14)
Proof. Let δ be small enough so that B(δ) ⊂ Ω and that the conclusions
of Theorem 5.4.2 hold. Then if x
n
∈ B(δ) all the intermediate iterates ¦y
j
¦
are in B(δ) by Theorem 5.4.2. In fact, if we set y
1
= x
n
, (5.8) implies that for
1 ≤ j ≤ m
y
j
−x
∗
 ≤ K
C
x
n
−x
∗
y
j−1
−x
∗
 = K
C
e
n
y
j−1
−x
∗
 ≤ . . . ≤ K
j
C
e
n

j+1
.
Hence x
n+1
∈ B(δ). Setting j = m in the above inequality completes the
proof.
The advantage of the Shamanskii method over Newton’s method is
that high qorders can be obtained with far fewer Jacobian evaluations or
factorizations. Optimal choices of m as a function of N can be derived
under assumptions consistent with dense matrix algebra [18] by balancing the
O(N
3
) cost of a matrix factorization with the cost of function and Jacobian
evaluation. The analysis in [18] and the expectation that, if the initial iterate is
suﬃciently accurate, only a small number of iterations will be taken, indicate
that the chord method is usually the best option for very large problems.
Algorithm nsol is based on this idea and using an idea from [114] computes
and factors the Jacobian only until an estimate of the qfactor for the linear
convergence rate is suﬃciently low.
5.4.4. Diﬀerence approximation to F
. Assume that we compute
F(x) + (x) instead of F(x) and attempt to approximate the action of F
on a vector by a forward diﬀerence. We would do this, for example, when
building an approximate Jacobian. Here the jth column of F
(x) is F
(x)e
j
where e
j
is the unit vector with jth component 1 and other components 0.
A forward diﬀerence approximation to F
(x)w would be
F(x +hw) +(x +hw) −F(x) −(x)
h
.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
80 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Assume that (x) ≤ ¯ . Then
F
(x)w −
F(x +hw) +(x +hw) −F(x) −(x)
h
= O(h + ¯ /h).
The quantity inside the Oterm is minimized when h =
√
¯ . This means, for
instance, that very small diﬀerence steps h can lead to inaccurate results. In
the special case where (x) is a result of ﬂoatingpoint roundoﬀ in full precision
(¯ ≈ 10
−15
), the analysis here indicates that h ≈ 10
−7
is a reasonable choice.
The choice h ≈ 10
−15
in this case can lead to disaster (try it!). Here we have
made the implicit assumption that x and w are of roughly the same size. If
this is not the case, h should be scaled to reﬂect that. The choice
h = ¯
1/2
x/w
reﬂects all the discussion above.
A more important consequence of this analysis is that the choice of step
size in a forward diﬀerence approximation to the action of the Jacobian on
a vector must take into account the error in the evaluation of F. This can
become very important if part of F is computed from measured data.
If F(x) has already been computed, the cost of a forward diﬀerence
approximation of F
(x)w is an additional evaluation of F (at the point x+hw).
Hence the evaluation of a full ﬁnite diﬀerence Jacobian would cost N function
evaluations, one for each column of the Jacobian. Exploitation of special
structure could reduce this cost.
The forward diﬀerence approximation to the directional derivative is not
a linear operator in w. The reason for this is that the derivative has been
approximated, and so linearity has been lost. Because of this we must
carefully specify what we mean by a diﬀerence approximation to the Jacobian.
Deﬁnition 5.4.1 below speciﬁes the usual choice.
Definition 5.4.1. Let F be deﬁned in a neighborhood of x ∈ R
N
.
(∇
h
F)(x) is the N N matrix whose jth column is given by
(∇
h
F)(x)
j
=
_
¸
¸
¸
¸
_
¸
¸
¸
¸
_
F(x +hxe
j
) −F(x)
hx
x ,= 0
F(he
j
) −F(x)
h
x = 0
We make a similar deﬁnition of the diﬀerence approximation of the
directional derivative.
Definition 5.4.2. Let F be deﬁned in a neighborhood of x ∈ R
N
and let
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 81
w ∈ R
N
. We have
D
h
F(x : w) =
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
0, w = 0,
w
F(x +hxw/w) −F(x)
hx
, w, x ,= 0,
w
F(hw/w) −F(x)
h
, x = 0, w ,= 0.
(5.15)
The standard assumptions and the Banach Lemma imply the following
result.
Lemma 5.4.1. Let the standard assumptions hold. Then there are δ, ¯ ,
¯
h >
0 such that if x ∈ B(δ), h ≤
¯
h, and (x) ≤ ¯ for all x ∈ B(δ), then
∇
h
(F +)(x) is nonsingular for all x ∈ B(δ) and there is M
F
> 0 such that
F
(x) −∇
h
(F +)(x) ≤ M
F
(h + ¯ /h).
If we assume that the forward diﬀerence step has been selected so that
h = O(
√
¯ ), then ∆(x) = O(
√
¯ ) by Lemma 5.4.1. Theorem 5.4.1 implies the
following result.
Theorem 5.4.6. Let the standard assumptions hold. Then there are
positive δ, ¯ , and K
D
such that if x
c
∈ B(δ), (x) ≤ ¯ for all x ∈ B(δ),
and there is M
−
> 0 such that
h
c
> M
−
√
¯
then
x
+
= x
c
−∇
hc
(F +)(x
c
)
−1
(F(x
c
) +(x
c
))
satisﬁes
e
+
 ≤ K
D
(¯ + (e
c
 +h
c
)e
c
).
Note that Theorem 5.4.6 does not imply that an iteration will converge to
x
∗
. Even if is entirely due to ﬂoatingpoint roundoﬀ, there is no guarantee
that (x
n
) → 0 as x
n
→ x
∗
. Hence, one should expect the sequence ¦e
n
¦
to stagnate and cease to decrease once e
n
 ≈ ¯ . Therefore in the early
phases of the iteration, while e
n
 >>
√
¯ , the progress of the iteration will
be indistinguishable from Newton’s method as implemented with an exact
Jacobian. When e
n
 ≤
√
¯ , Theorem 5.4.6 says that e
n+1
 = O(¯ ). Hence
one will see quadratic convergence until the error in F admits no further
reduction.
A diﬀerence approximation to a full Jacobian does not use any special
structure of the Jacobian and will probably be less eﬃcient than a hand coded
analytic Jacobian. If information on the sparsity pattern of the Jacobian is
available it is sometimes possible [47], [42], to evaluate ∇
h
F with far fewer than
N evaluations of F. If the sparsity pattern is such that F
can be reduced to
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
82 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
a block triangular form, the entire solution process can take advantage of this
structure [59].
For problems in which the Lipschitz constant of F
is large, the error in
the diﬀerence approximation will be large, too. Moreover, D
h
F(x : w) is not,
in general, a linear function of w and it is usually the case that
(∇
h
F(x))w ,= D
h
F(x : w).
If F
(x
∗
) is ill conditioned or the Lipschitz constant of F
is large, the
diﬀerence between them may be signiﬁcant and a diﬀerence approximation to
a directional derivative, which uses the size of w, is likely to be more accurate
than ∇
h
F(x)w.
While we used full matrix approximations to numerical Jacobians in all
the examples reported here, they should be used with caution.
5.4.5. The secant method. The results in this section are for single
equations f(x) = 0, where f is a realvalued function of one real variable.
We assume for the sake of simplicity that there are no errors in the evaluation
of f, i.e., (x) = 0. The standard assumptions, then, imply that f
(x
∗
) ,= 0.
There is no reason to use a diﬀerence approximation to f
for equations in a
single variable because one can use previously computed data to approximate
f
(x
n
) by
a
n
=
f(x
n
) −f(x
n−1
)
x
n
−x
n−1
.
The resulting method is called the secant method and is due to Newton
[140], [137]. We will prove that the secant method is locally qsuperlinearly
convergent if the standard assumptions hold.
In order to start, one needs two approximations to x
∗
, x
0
and x
−1
. The
local convergence theory can be easily described by Theorem 5.4.1. Let x
c
be
the current iterate and x
−
the previous iterate. We have (x
c
) = 0 and
∆(x
c
) =
f(x
c
) −f(x
−
)
x
c
−x
−
−f
(x
c
). (5.16)
We have the following theorem.
Theorem 5.4.7. Let the standard assumptions hold with N = 1. Then
there is δ > 0 such that if x
0
, x
−1
∈ B(δ) and x
0
,= x
−1
then the secant iterates
converge qsuperlinearly to x
∗
.
Proof. Let δ be small enough so that B(δ) ⊂ Ω and the conclusions of
Theorem 5.4.1 hold. Assume that x
−1
, x
0
, . . . , x
n
∈ B(δ). If x
n
= x
∗
for some
ﬁnite n, we are done. Otherwise x
n
,= x
n−1
. If we set s = x
n
−x
n−1
we have
by (5.16) and the fundamental theorem of calculus
∆(x
n
) =
_
1
0
(f
(x
n−1
+ts) −f
(x
n
)) dt
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 83
and hence
[∆(x
n
)[ ≤ γ[x
n−1
−x
n
[/2 ≤
γ([e
n
[ +[e
n−1
[)
2
.
Applying Theorem 5.4.1 implies
[e
n+1
[ ≤
¯
K((1 +γ/2)[e
n
[
2
+γ[e
n
[[e
n−1
[/2). (5.17)
If we reduce δ so that
¯
K(1 +γ)δ = η < 1
then x
n
→ x
∗
qlinearly. Therefore
[e
n+1
[
[e
n
[
≤
¯
K((1 +γ/2)[e
n
[ +γ[e
n−1
[/2) → 0
as n → ∞. This completes the proof.
The secant method and the classical bisection (see Exercise 5.7.8) method
are the basis for the popular method of Brent [17] for the solution of single
equations.
5.5. The Kantorovich Theorem
In this section we present the result of Kantorovich [107], [106], which states
that if the standard assumptions “almost” hold at a point x
0
, then there
is a root and the Newton iteration converges rquadratically to that root.
This result is of use, for example, in proving that discretizations of nonlinear
diﬀerential and integral equations have solutions that are near to that of the
continuous problem.
We require the following assumption.
Assumption 5.5.1. There are β, η, ¯ r, and γ with βηγ ≤ 1/2 and x
0
∈ R
N
such that
1. F is diﬀerentiable at x
0
,
F
(x
0
)
−1
 ≤ β, and F
(x
0
)
−1
F(x
0
) ≤ η.
2. F
is Lipschitz continuous with Lipschitz constant γ in a ball of radius
¯ r ≥ r
−
about x
0
where
r
−
=
1 −
√
1 −2βηγ
βγ
.
We will let B
0
be the closed ball of radius r
−
about x
0
B
0
= ¦x[ x −x
0
 ≤ r
−
¦.
We do not prove the result in its full generality and refer to [106], [145],
[57], and [58] for a complete proof and for discussions of related results.
However, we give a simpliﬁed version of a Kantorovichlike result for the chord
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
84 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
method to show how existence of a solution can be derived from estimates
on the function and Jacobian. Unlike the chord method results in [106],
[145], and [57], Theorem 5.5.1 assumes a bit more than Assumption 5.5.1 but
has a simple proof that uses only the contraction mapping theorem and the
fundamental theorem of calculus and obtains qlinear convergence instead of
rlinear convergence.
Theorem 5.5.1. Let Assumption 5.5.1 hold and let
βγη < 1/2. (5.18)
Then there is a unique root x
∗
of F in B
0
, the chord iteration with x
0
as the
initial iterate converges to x
∗
qlinearly with qfactor βγr
−
, and x
n
∈ B
0
for
all n. Moreover x
∗
is the unique root of F in the ball of radius
min
_
¯ r,
1 +
√
1 −2βηγ
βγ
_
about x
0
.
Proof. We will show that the map
φ(x) = x −F
(x
0
)
−1
F(x)
is a contraction on the set
o = ¦x[ x −x
0
 ≤ r
−
¦.
By the fundamental theorem of calculus and Assumption 5.5.1 we have for
all x ∈ o
F
(x
0
)
−1
F(x) = F
(x
0
)
−1
F(x
0
)
+F
(x
0
)
−1
_
1
0
F
(x
0
+t(x −x
0
))(x −x
0
) dt
= F
(x
0
)
−1
F(x
0
) +x −x
0
+F
(x
0
)
−1
_
1
0
(F
(x
0
+t(x −x
0
)) −F
(x
0
))(x −x
0
) dt
(5.19)
and
φ(x) −x
0
= −F
(x
0
)
−1
F(x
0
)
−F
(x
0
)
−1
_
1
0
(F
(x
0
+t(x −x
0
)) −F
(x
0
))(x −x
0
) dt.
Hence,
φ(x) −x
0
 ≤ η +βγr
2
−
/2 = r
−
.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 85
and φ maps o into itself.
To show that φ is a contraction on o and to prove the qlinear convergence
we will show that
φ(x) −φ(y) ≤ βγr
−
x −y
for all x, y ∈ o. If we do this, the contraction mapping theorem (Theo
rem 4.2.1) will imply that there is a unique ﬁxed point x
∗
of φ in o and
that x
n
→ x
∗
qlinearly with qfactor βγr
−
. Clearly x
∗
is a root of F because
it is a ﬁxed point of φ. We know that βγr
−
< 1 by our assumption that
βηγ < 1/2.
Note that for all x ∈ B
0
φ
(x) = I −F
(x
0
)
−1
F
(x) = F
(x
0
)
−1
(F
(x
0
) −F
(x))
and hence
φ
(x) ≤ βγx −x
0
 ≤ βγr
−
< 1.
Therefore, for all x, y ∈ B
0
,
φ(x) −φ(y) =
_
_
_
_
_
1
0
φ
(y +t(x −y))(x −y) dt
_
_
_
_
≤ βγr
−
x −y.
This proves the convergence result and the uniqueness of x
∗
in B
0
.
To prove the remainder of the uniqueness assertion let
r
+
=
1 +
√
1 −2βηγ
βγ
and note that r
+
> r
−
because βηγ < 1/2. If ¯ r = r
−
, then we are done by the
uniqueness of x
∗
in B
0
. Hence we may assume that ¯ r > r
−
. We must show
that if x is such that
r
−
< x −x
0
 < min(¯ r, r
+
)
then F(x) ,= 0. Letting r = x −x
0
 we have by (5.19)
F
(x
0
)
−1
F(x) ≥ r −η −βγr
2
/2 > 0
because r
−
< r < r
+
. This completes the proof.
The Kantorovich theorem is more precise than Theorem 5.5.1 and has a
very diﬀerent proof. We state the theorem in the standard way [145], [106],
[63], using the rquadratic estimates from [106].
Theorem 5.5.2. Let Assumption 5.5.1 hold. Then there is a unique root
x
∗
of F in B
0
, the Newton iteration with x
0
as the initial iterate converges to
x
∗
, and x
n
∈ B
0
for all n. x
∗
is the unique root of F in the ball of radius
min
_
¯ r,
1 +
√
1 −2βηγ
βγ
_
about x
0
and the errors satisfy
e
n
 ≤
(2βηγ)
2
n
2
n
βγ
. (5.20)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
86 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
If βηγ < 1/2 then (5.20) implies that the convergence is rquadratic (see
Exercise 5.7.16).
5.6. Examples for Newton’s method
In the collection of MATLAB codes we provide an implementation nsol
of Newton’s method, the chord method, the Shamanskii method, or a
combination of all of them based on the reduction in the nonlinear residual.
This latter option decides if the Jacobian should be recomputed based on the
ratio of successive residuals. If
F(x
+
)/F(x
c
)
is below a given threshold, the Jacobian is not recomputed. If the ratio is too
large, however, we compute and factor F
(x
+
) for use in the subsequent chord
steps. In addition to the inputs of Algorithm sham, a threshold for the ratio
of successive residuals is also input. The Jacobian is recomputed and factored
if either the ratio of successive residuals exceeds the threshold 0 < ρ < 1 or
the number of iterations without an update exceeds m. The output of the
MATLAB implementation of nsol is the solution, the history of the iteration
stored as the vector of l
∞
norms of the nonlinear residuals, and an error ﬂag.
nsol uses diffjac to approximate the Jacobian by ∇
h
F with h = 10
−7
.
While our examples have dense Jacobians, the ideas in the implementation
of nsol also apply to situations in which the Jacobian is sparse and direct
methods for solving the equation for the Newton step are necessary. One would
still want to minimize the number of Jacobian computations (by the methods
of [47] or [42], say) and sparse matrix factorizations. In Exercise 5.7.25, which
is not easy, the reader is asked to create a sparse form of nsol.
Algorithm 5.6.1. nsol(x, F, τ, m, ρ)
1. r
c
= r
0
= F(x)
2. Do while F(x) > τ
r
r
0
+τ
a
(a) Compute ∇
h
F(x) ≈ F
(x)
(b) Factor ∇
h
F(x) = LU.
(c) i
s
= 0, σ = 0
(d) Do While i
s
< m and σ ≤ ρ
i. i
s
= i
s
+ 1
ii. Solve LUs = −F(x)
iii. x = x +s
iv. Evaluate F(x)
v. r
+
= F(x), σ = r
+
/r
c
, r
c
= r
+
vi. If F(x) ≤ τ
r
r
0
+τ
a
exit.
In the MATLAB code the iteration is terminated with an error condition
if σ ≥ 1 at any point. This indicates that the local convergence analysis in this
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 87
section is not applicable. In Chapter 8 we discuss how to recover from this.
nsol is based on the assumption that a diﬀerence Jacobian will be used. In
Exercise 5.7.21 you are asked to modify the code to accept an analytic Jacobian.
Many times the Jacobian can be computed eﬃciently by reusing results that
are already available from the function evaluation. The parameters m and ρ
are assigned the default values of 1000 and .5. With these default parameters
the decision to update the Jacobian is made based entirely on the reduction in
the norm of the nonlinear residual. nsol allows the user to specify a maximum
number of iterations; the default is 40.
The Chandrasekhar Hequation. The Chandrasekhar Hequation, [41],
[30],
F(H)(µ) = H(µ) −
_
1 −
c
2
_
1
0
µH(ν) dν
µ +ν
_
−1
= 0, (5.21)
is used to solve exit distribution problems in radiative transfer.
We will discretize the equation with the composite midpoint rule. Here we
approximate integrals on [0, 1] by
_
1
0
f(µ) dµ ≈
1
N
N
j=1
f(µ
j
),
where µ
i
= (i −1/2)/N for 1 ≤ i ≤ N. The resulting discrete problem is
F(x)
i
= x
i
−
_
_
1 −
c
2N
N
j=1
µ
i
x
j
µ
i
+µ
j
_
_
−1
. (5.22)
It is known [132] that both (5.21) and the discrete analog (5.22) have two
solutions for c ∈ (0, 1). Only one of these solutions has physical meaning,
however. Newton’s method, with the 0 function or the constant function with
value 1 as initial iterate, will ﬁnd the physically meaningful solution [110].
The standard assumptions hold near either solution [110] for c ∈ [0, 1). The
discrete problem has its own physical meaning [41] and we will attempt to
solve it to high accuracy. Because H has a singularity at µ = 0, the solution of
the discrete problem is not even a ﬁrstorder accurate approximation to that
of the continuous problem.
In the computations reported here we set N = 100 and c = .9. We used the
function identically one as initial iterate. We computed all Jacobians with the
MATLAB code diffjac. We set the termination parameters τ
r
= τ
a
= 10
−6
.
We begin with a comparison of Newton’s method and the chord method.
We present some iteration statistics in both tabular and graphical form. In
Table 5.1 we tabulate the iteration counter n, F(x
n
)
∞
/F(x
0
)
∞
, and the
ratio
R
n
= F(x
n
)
∞
/F(x
n−1
)
∞
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
88 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Table 5.1
Comparison of Newton and chord iteration.
n F(x
n
)
∞
/F(x
0
)
∞
R
n
F(x
n
)
∞
/F(x
0
)
∞
R
n
0 1.000e+00 1.000e+00
1 1.480e01 1.480e01 1.480e01 1.480e01
2 2.698e03 1.823e02 3.074e02 2.077e01
3 7.729e07 2.865e04 6.511e03 2.118e01
4 1.388e03 2.132e01
5 2.965e04 2.136e01
6 6.334e05 2.136e01
7 1.353e05 2.136e01
8 2.891e06 2.136e01
for n ≥ 1 for both Newton’s method and the chord method. In Fig. 5.1
the solid curve is a plot of F(x
n
)
∞
/F(x
0
)
∞
for Newton’s method and
the dashed curve a plot of F(x
n
)
∞
/F(x
0
)
∞
for the chord method. The
concave iteration track is the signature of superlinear convergence, the linear
track indicates linear convergence. See Exercise 5.7.15 for a more careful
examination of this concavity. The linear convergence of the chord method
can also be seen in the convergence of the ratios R
n
to a constant value.
For this example the MATLAB flops command returns an fairly accurate
indicator of the cost. The Newton iteration required over 8.8 million ﬂoating
point operations while the chord iteration required under 3.3 million. In
Exercise 5.7.18 you are asked to obtain timings in your environment and will
see how the ﬂoatingpoint operation counts are related to the actual timings.
The good performance of the chord method is no accident. If we think
of the chord method as a preconditioned nonlinear Richardson iteration with
the initial Jacobian playing the role of preconditioner, then the chord method
should be much more eﬃcient than Newton’s method if the equations for the
steps can be solved cheaply. In the case considered here, the cost of a single
evaluation and factorization of the Jacobian is far more than the cost of the
entire remainder of the chord iteration.
The ideas in Chapters 6 and 7 show how the low iteration cost of the
chord method can be combined with the small number of iterations of a
superlinearly convergent nonlinear iterative method. In the context of this
particular example, the cost of the Jacobian computation can be reduced by
analytic evaluation and reuse of data that have already been computed for the
evaluation of F. The reader is encouraged to do this in Exercise 5.7.21.
In the remaining examples, we plot, but do not tabulate, the iteration
statistics. In Fig. 5.2 we compare the Shamanskii method with m = 2 with
Newton’s method. The Shamanskii computation required under 6 million
ﬂoatingpoint operations, as compared with over 8.8 for Newton’s method.
However, for this problem, the simple chord method is the most eﬃcient.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 89
0 1 2 3 4 5 6 7 8
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
Iterations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 5.1. Newton and chord methods for Hequation, c = .9.
0 0.5 1 1.5 2 2.5 3 3.5 4
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
Iterations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 5.2. Newton and Shamanskii method for Hequation, c = .9.
The hybrid approach in nsol with the default parameters would be the chord
method for this problem.
We now reconsider the Hequation with c = .9999. For this problem the
Jacobian at the solution is nearly singular [110]. Aside from c, all iteration
parameters are the same as in the previous example. While Newton’s method
converges in 7 iterations and requires 21 million ﬂoatingpoint operations, the
chord method requires 188 iterations and 10.6 million ﬂoatingpoint operations.
The qfactor for the chord method in this computation was over .96, indicating
very slow convergence. The hybrid method in nsol with the default parameters
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
90 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
computes the Jacobian four times during the course of the iteration, converges
in 14 iterations, and requires 12 million ﬂoatingpoint operations. In Fig. 5.3
we compare the Newton iteration with the hybrid.
0 2 4 6 8 10 12 14
10
6
10
5
10
4
10
3
10
2
10
1
10
0
Iterations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 5.3. Newton and hybrid methods for Hequation c = .9999.
One should approach plots of convergence histories with some caution. The
theory for the chord method asserts that the error norms converge qlinearly to
zero. This implies only that the nonlinear residual norms ¦F(x
n
)¦ converge
rlinearly to zero. While the plots indicate qlinear convergence of the nonlinear
residuals, the theory does not, in general, support this. In practice, however,
plots like those in this section are common.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 91
5.7. Exercises on Newton’s method
In some of the exercises here, and in the rest of this part of the book, you will
be asked to plot or tabulate iteration statistics. When asked to do this, for
each iteration tabulate the iteration counter, the norm of F, and the ratio
of F(x
n
)/F(x
n−1
) for n ≥ 1. A better alternative in the MATLAB
environment is to use the semilogy command to plot the norms. When
one does this, one can visually inspect the plot to determine superlinear
convergence (concavity) without explicitly computing the ratios. Use the l
∞
norm.
5.7.1. A function G is said to be H¨older continuous with exponent α in Ω if
G(x) −G(y) ≤ Kx −y
α
for all x, y ∈ Ω. Show that if the Lipschitz
continuity condition on F
in the standard assumptions is replaced by
H¨older continuity with exponent α > 0 that the Newton iterates converge
with qorder 1 +α.
5.7.2. Can the performance of the Newton iteration be improved by a linear
change of variables? That is, for nonsingular N N matrices A and
B, can the Newton iterates for F(x) = 0 and AF(Bx) = 0 show any
performance diﬀerence when started at the same initial iterate? What
about the chord method?
5.7.3. Let ∇
h
F be given by Deﬁnition 5.4.1. Given the standard assumptions,
prove that the iteration given by
x
n+1
= x
n
−(∇
hn
F(x
n
))
−1
F(x
n
)
is locally qsuperlinearly convergent if h
n
→ 0. When is it locally q
quadratically convergent?
5.7.4. Suppose F
(x
n
) is replaced by ∇
hn
F(x
n
) in the Shamanskii method.
Discuss how h
n
must be decreased as the iteration progresses to preserve
the qorder of m+ 1 [174].
5.7.5. In what way is the convergence analysis of the secant method changed if
there are errors in the evaluation of f?
5.7.6. Prove that the secant method converges rsuperlinearly with rorder
(
√
5 + 1)/2. This is easy.
5.7.7. Show that the secant method converges qsuperlinearly with qorder
(
√
5 + 1)/2.
5.7.8. The bisection method produces a sequence of intervals (x
k
l
, x
k
r
) that
contain a root of a function of one variable. Given (x
k
l
, x
k
r
) with
f(x
k
l
)f(x
k
r
) < 0 we replace one of the endpoints with y = (x
k
l
+ x
k
r
)/2
to force f(x
k+1
l
)f(x
k+1
r
) < 0. If f(y) = 0, we terminate the iteration.
Show that the sequence x
k
= (x
k
l
+ x
k
r
)/2 is rlinearly convergent if f is
continuous and there exists (x
0
l
, x
0
r
) such that f(x
0
l
)f(x
0
r
) < 0.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
92 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
5.7.9. Write a program that solves single nonlinear equations with Newton’s
method, the chord method, and the secant method. For the secant
method, use x
−1
= .99x
0
. Apply your program to the following
function/initial iterate combinations, tabulate or plot iteration statistics,
and explain your results.
1. f(x) = cos(x) −x, x
0
= .5
2. f(x) = tan
−1
(x), x
0
= 1
3. f(x) = sin(x), x
0
= 3
4. f(x) = x
2
, x
0
= .5
5. f(x) = x
2
+ 1, x
0
= 10
5.7.10. For scalar functions of one variable a quadratic model of f about a point
x
c
is
m
c
(x) = f(x
c
) +f
(x
c
)(x −x
c
) +f
(x
c
)(x −x
c
)
2
/2.
Consider the iterative method that deﬁnes x
+
as a root of m
c
. What
problems might arise in such a method? Resolve these problems and
implement your algorithm. Test it on the problems in Exercise 5.7.9. In
what cases is this algorithm superior to Newton’s method? For hints and
generalizations of this idea to functions of several variables, see [170]. Can
the ideas in [170] be applied to some of the highorder methods discussed
in [190] for equations in one unknown?
5.7.11. Show that if f is a real function of one real variable, f
is Lipschitz
continuous, and f(x
∗
) = f
(x
∗
) = 0 but f
(x
∗
) ,= 0 then the iteration
x
n+1
= x
n
−2f(x
n
)/f
(x
n
)
converges locally qquadratically to x
∗
provided x
0
is suﬃciently near to
x
∗
, but not equal to x
∗
[171].
5.7.12. Show that if f is a real function of one real variable, the standard
assumptions hold, f
(x
∗
) = 0 and f
(x) is Lipschitz continuous, then
the Newton iteration converges cubically (qsuperlinear with qorder 3).
5.7.13. Show that if f is a real function of one real variable, the standard
assumptions hold, x
0
is suﬃciently near x
∗
, and f(x
0
)f
(x
0
) > 0, then
the Newton iteration converges monotonically to x
∗
. What if f
(x
∗
) = 0?
How can you extend this result if f
(x
∗
) = f
(x
∗
) = . . . f
(k)
(x
∗
) = 0 ,=
f
(k+1)
(x
∗
)? See [131].
5.7.14. How would you estimate the qorder of a qsuperlinearly convergent
sequence? Rigorously justify your answer.
5.7.15. If x
n
→ x
∗
qsuperlinearly with qorder α > 1, show that log(e
n
) is a
concave function of n in the sense that
log(e
n+1
) −log(e
n
)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
NEWTON’S METHOD 93
is a decreasing function of n for n suﬃciently large. Is this still the case
if the convergence is qsuperlinear but there is no qorder? What if the
convergence is qlinear?
5.7.16. Show that the sequence on the right side of (5.20) is rquadratically (but
not qquadratically) convergent to 0 if βηγ < 1/2.
5.7.17. Typically (see [89], [184]) the computed LU factorization of F
(x) is the
exact LU factorization of a nearby matrix.
LU = F
(x) +E.
Use the theory in [89], [184] or another book on numerical linear algebra
and Theorem 5.4.1 to describe how this factorization error aﬀects the
convergence of the Newton iteration. How does the analysis for the QR
factorization diﬀer?
5.7.18. Duplicate the results on the Hequation in ¸ 5.6. Try diﬀerent values of
N and c. How does the performance of the methods diﬀer as N and c
change? Compare your results to the tabulated values of the H function
in [41] or [14]. Compare execution times in your environment. Tabulate
or plot iteration statistics. Use the MATLAB cputime command to see
how long the entire iteration took for each N, c combination and the
MATLAB flops command to get an idea for the number of ﬂoating
point operations required by the various methods.
5.7.19. Solve the Hequation with c = 1 and N = 100. Explain your results
(see [50] and [157]). Does the trick in Exercise 5.7.11 improve the
convergence? Try one of the approaches in [51], [49], or [118] for
acceleration of the convergence. You might also look at [90], [170], [75],
[35] and [155] for more discussion of this. Try the chord method and
compare the results to part 4 of Exercise 5.7.9 (see [52]).
5.7.20. Solve the Hequation with N = 100, c = .9 and c = .9999 with ﬁxedpoint
iteration. Compare timings and ﬂop counts with Newton’s method, the
chord method, and the algorithm in nsol.
5.7.21. Compute by hand the Jacobian of F, the discretized nonlinearity from
the H equation in (5.22). Prove that F
can be computed at a cost of
less than twice that of an evaluation of F. Modify nsol to accept this
analytic Jacobian rather than evaluate a diﬀerence Jacobian. How do
the execution times and ﬂop counts change?
5.7.22. Modify dirder by setting the forward diﬀerence step h = 10
−2
. How
does this change aﬀect the convergence rate observations that you have
made?
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
94 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
5.7.23. Add noise to your function evaluation with a random number generator,
such as the the MATLAB rand function. If the size of the noise is
O(10
−4
) and you make the appropriate adjustments to dirder, how are
the convergence rates diﬀerent? What is an appropriate termination
criterion? At what point does the iteration stagnate?
5.7.24. Assume that the standard assumptions hold, that the cost of a function
evaluation is O(N
2
) ﬂoatingpoint operations, the cost of a Jacobian is
O(N) function evaluations, and that x
0
is near enough to x
∗
so that the
Newton iteration converges qquadratically to x
∗
. Show that the number
n of Newton iterations needed to obtain e
n
 ≤ e
0
 is O(log()) and
that the number of ﬂoatingpoint operations required is O(N
3
log()).
What about the chord method?
5.7.25. Develop a sparse version of nsol for problems with tridiagonal Jacobian.
In MATLAB, for example, you could modify diffjac to use the
techniques of [47] and nsol to use the sparse matrix factorization in
MATLAB. Apply your code to the central diﬀerence discretization of the
twopoint boundary value problem
−u
= sin(u) +f(x), u(0) = u(1) = 0.
Here, the solution is u
∗
(x) = x(1 −x), and
f(x) = 2 −sin(x(1 −x)).
Let u
0
= 0 be the initial iterate.
5.7.26. Suppose you try to ﬁnd an eigenvectoreigenvalue pair (φ, λ) of an NN
matrix A by solving the system of N + 1 nonlinear equations
F(φ, λ) =
_
Aφ −λφ
φ
T
φ −1
_
=
_
0
0
_
(5.23)
for the vector x = (φ
T
, λ)
T
∈ R
N+1
. What is the Jacobian of this
system? If x = (φ, λ) ∈ R
N+1
is an eigenvectoreigenvalue pair, when is
F
(x) nonsingular? Relate the application of Newton’s method to (5.23)
to the inverse power method. See [150] for more development of this
idea.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Chapter 6
Inexact Newton Methods
Theorem 5.4.1 describes how errors in the derivative/function aﬀect the
progress in the Newton iteration. Another way to look at this is to ask how
an approximate solution of the linear equation for the Newton step aﬀects the
iteration. This was the view taken in [55] where inexact Newton methods in
which the step satisﬁes
F
(x
c
)s +F(x
c
) ≤ η
c
F(x
c
) (6.1)
are considered. Any approximate step is accepted provided that the relative
residual of the linear equation is small. This is quite useful because conditions
like (6.1) are precisely the small linear residual termination conditions for
iterative solution of the linear system for the Newton step. Such methods are
not new. See [145] and [175], for example, for discussion of these ideas in the
context of the classical stationary iterative methods. In most of this chapter
we will focus on the approximate solution of the equation for the Newton step
by GMRES, but the other Krylov subspace methods discussed in Chapters 2
and 3 and elsewhere can also be used. We follow [69] and refer to the term η
c
on the right hand side of (6.1) as the forcing term.
6.1. The basic estimates
We will discuss speciﬁc implementation issues in ¸ 6.2. Before that we will
give the basic result from [55] to show how a step that satisﬁes (6.1) aﬀects
the convergence. We will present this result in two parts. In ¸ 6.1.1 we give a
straightforward analysis that uses the techniques of Chapter 5. In ¸ 6.1.2 we
show how the requirements of the simple analysis can be changed to admit a
more aggressive (i.e., larger) choice of the parameter η.
6.1.1. Direct analysis. The proof and application of Theorem 6.1.1 should
be compared to that of Theorem 5.1.1 and the other results in Chapter 5.
Theorem 6.1.1. Let the standard assumptions hold. Then there are δ and
K
I
such that if x
c
∈ B(δ), s satisﬁes (6.1), and x
+
= x
c
+s then
e
+
 ≤ K
I
(e
c
 +η
c
)e
c
. (6.2)
95
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
96 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Proof. Let δ be small enough so that the conclusions of Lemma 4.3.1 and
Theorem 5.1.1 hold. To prove the ﬁrst assertion (6.2) note that if
2
r = −F
(x
c
)s −F(x
c
)
is the linear residual then
s +F
(x
c
)
−1
F(x
c
) = −F
(x
c
)
−1
r
and
e
+
= e
c
+s = e
c
−F
(x
c
)
−1
F(x
c
) −F
(x
c
)
−1
r. (6.3)
Now, (6.1), (4.7), and (4.6) imply that
s +F
(x
c
)
−1
F(x
c
) ≤ F
(x
c
)
−1
η
c
F(x
c
)
≤ 4κ(F
(x
∗
))η
c
e
c
.
Hence, using (6.3) and Theorem 5.1.1
e
+
 ≤ e
c
−F
(x
c
)
−1
F(x
c
) + 4κ(F
(x
∗
))η
c
e
c

≤ Ke
c

2
+ 4κ(F
(x
∗
))η
c
e
c
,
where K is the constant from (5.2). If we set
K
I
= K + 4κ(F
(x
∗
)),
the proof is complete.
We summarize the implications of Theorem 6.1.1 for iterative methods in
the next result, which is a direct consequence of (6.2) and will suﬃce to explain
most computational observations.
Theorem 6.1.2. Let the standard assumptions hold. Then there are δ and
¯ η such that if x
0
∈ B(δ), ¦η
n
¦ ⊂ [0, ¯ η], then the inexact Newton iteration
x
n+1
= x
n
+s
n
,
where
F
(x
n
)s
n
+F(x
n
) ≤ η
n
F(x
n
)
converges qlinearly to x
∗
. Moreover
• if η
n
→ 0 the convergence is qsuperlinear, and
• if η
n
≤ K
η
F(x
n
)
p
for some K
η
> 0 the convergence is qsuperlinear
with qorder 1 +p.
2
Our deﬁnition of r is consistent with the idea that the residual in a linear equation is b −Ax. In
[55] r = F
(xc)s +F(xc).
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
INEXACT NEWTON METHODS 97
Proof. Let δ be small enough so that (6.2) holds for x
c
∈ B(δ). Reduce δ
and ¯ η if needed so that
K
I
(δ + ¯ η) < 1,
where K
I
is from (6.2). Then if n ≥ 0 and x
n
∈ B(δ) we have
e
n+1
 ≤ K
I
(e
n
 +η
n
)e
n
 ≤ K
I
(δ + ¯ η)e
n
 < e
n
.
This proves qlinear convergence with a qfactor of K
I
(δ + ¯ η).
If η
n
→ 0 then qsuperlinear convergence follows from the deﬁnition. If
η
n
≤ K
η
F(x
n
)
p
then we may use (4.7) and (6.2) to conclude
e
n+1
 ≤ K
I
(e
n

1−p
+K
η
2
p
F
(x
∗
)
p
)e
n

1+p
which completes the proof.
6.1.2. Weighted norm analysis. Since K
I
= O(κ(F
(x
∗
))), one might
conclude from Theorem 6.1.2 that if F
(x
∗
) is ill conditioned very small forcing
terms must be used. This is not the case and the purpose of this subsection
is to describe more accurately the necessary restrictions on the forcing terms.
The results here diﬀer from those in ¸ 6.1.1 in that no restriction is put on
the sequence ¦η
n
¦ ⊂ [0, 1) other than requiring that 1 not be an accumulation
point.
Theorem 6.1.3. Let the standard assumptions hold. Then there is δ such
that if x
c
∈ B(δ), s satisﬁes (6.1), x
+
= x
c
+s, and η
c
≤ η < ¯ η < 1, then
F
(x
∗
)e
+
 ≤ ¯ ηF
(x
∗
)e
c
. (6.4)
Proof. To prove (6.4) note that Theorem 4.0.1 implies that
F(x
c
) ≤ F
(x
∗
)e
c
 +
γe
c

2
2
.
Since
e
c
 = F
(x
∗
)
−1
F
(x
∗
)e
c
 ≤ F
(x
∗
)
−1
F
(x
∗
)e
c

we have, with
M
0
=
γF
(x
∗
)
−1

2
F(x
c
) ≤ (1 +M
0
δ)F
(x
∗
)e
c
. (6.5)
Now,
F
(x
∗
)e
+
= F
(x
∗
)(e
c
+s)
= F
(x
∗
)(e
c
−F
(x
c
)
−1
F(x
c
) −F
(x
c
)
−1
r).
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
98 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
By Theorem 5.1.1
F
(x
∗
)(e
c
−F
(x
c
)
−1
F(x
c
)) ≤ KF
(x
∗
)e
c

2
.
Hence,
F
(x
∗
)e
+
 ≤ F
(x
∗
)F
(x
c
)
−1
r +KF
(x
∗
)e
c

2
. (6.6)
Since
F
(x
∗
)F
(x
c
)
−1
r ≤ r +(F
(x
∗
) −F
(x
c
))F
(x
c
)
−1
r
≤ (1 + 2γF
(x
∗
)
−1
e
c
)r,
we may set
M
1
= 2γF
(x
∗
)
−1
, and M
2
= KF
(x
∗
)
and obtain, using (4.7) and (6.5),
F
(x
∗
)e
+
 ≤ (1 +M
1
δ)r +M
2
δe
c

≤ (1 +M
1
δ)(1 +M
0
δ)η
c
F
(x
∗
)e
c

+M
2
δF
(x
∗
)
−1
F
(x
∗
)e
c

≤ ((1 +M
1
δ)(1 +M
0
δ)η +M
2
δF
(x
∗
)
−1
)F
(x
∗
)e
c
.
(6.7)
Now let δ be small enough so that
(1 +M
1
δ)(1 +M
0
δ)η +M
2
δF
(x
∗
)
−1
 ≤ ¯ η
and the proof is complete.
Note that the distance δ from the initial iterate to the solution may have
to be smaller for (6.4) to hold than for (6.2). However (6.4) is remarkable in its
assertion that any method that produces a step with a linear residual less than
that of the zero step will reduce the norm of the error if the error is measured
with the weighted norm
 
∗
= F
(x
∗
) .
In fact, Theorem 6.1.3 asserts qlinear convergence in the weighted norm if the
initial iterate is suﬃciently near x
∗
. This is made precise in Theorem 6.1.4.
The application of Theorem 6.1.3 described in Theorem 6.1.4 diﬀers from
Theorem 6.1.2 in that we do not demand that ¦η
n
¦ be bounded away from 1
by a suﬃciently large amount, just that 1 not be an accumulation point. The
importance of this theorem for implementation is that choices of the sequence
of forcing terms ¦η
n
¦ (such as η
n
= .5 for all n) that try to minimize the
number of inner iterations are completely justiﬁed by this result if the initial
iterate is suﬃciently near the solution. We make such a choice in one of the
examples considered in ¸ 6.4 and compare it to a modiﬁcation of a choice from
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
INEXACT NEWTON METHODS 99
[69] that decreases η
n
rapidly with a view toward minimizing the number of
outer iterations.
Theorem 6.1.4. Let the standard assumptions hold. Then there is δ such
that if x
0
∈ B(δ), ¦η
n
¦ ⊂ [0, η] with η < ¯ η < 1, then the inexact Newton
iteration
x
n+1
= x
n
+s
n
,
where
F
(x
n
)s
n
+F(x
n
) ≤ η
n
F(x
n
)
converges qlinearly with respect to  
∗
to x
∗
. Moreover
• if η
n
→ 0 the convergence is qsuperlinear, and
• if η
n
≤ K
η
F(x
n
)
p
for some K
η
> 0 the convergence is qsuperlinear
with qorder 1 +p.
Proof. The proof uses both (6.4) and (6.2). Our ﬁrst task is to relate the
norms   and  
∗
so that we can estimate δ. Let δ
0
be such that (6.4) holds
for e
c
 < δ
0
.
For all x ∈ R
N
,
x
∗
≤ F
(x
∗
)x ≤ κ(F
(x
∗
))x
∗
, (6.8)
Note that e < δ
0
if
e
∗
< δ
∗
= F
(x
∗
)δ
0
Set δ = F
(x
∗
)
−1
δ
∗
. Then e
0

∗
< δ
∗
if e
0
 < δ.
Our proof does not rely on the (possibly false) assertions that e
n
 < δ for
all n or that x
n
→ x
∗
qlinearly with respect to the unweighted norm. Rather
we note that (6.4) implies that if e
n

∗
< δ
∗
then
e
n+1

∗
≤ ¯ ηe
n

∗
< δ
∗
(6.9)
and hence x
n
→ x
∗
qlinearly with respect to the weighted norm. This proves
the ﬁrst assertion.
To prove the assertions on qsuperlinear convergence, note that since
x
n
→ x
∗
, eventually e
n
 will be small enough so that the conclusions of
Theorem 6.1.1 hold. The superlinear convergence results then follow exactly
as in the proof of Theorem 6.1.2.
We close this section with some remarks on the two types of results in
Theorems 6.1.2 and 6.1.4. The most signiﬁcant diﬀerence is the norm in which
qlinear convergence takes place. qlinear convergence with respect to the
weighted norm  
∗
is equivalent to qlinear convergence of the sequence of
nonlinear residuals ¦F(x
n
)¦. We state this result as Proposition 6.1.1 and
leave the proof to the reader in Exercise 6.5.2. The implications for the choice of
the forcing terms ¦η
n
¦ are also diﬀerent. While Theorem 6.1.2 might lead one
to believe that a small value of η
n
is necessary for convergence, Theorem 6.1.4
shows that the sequence of forcing terms need only be kept bounded away from
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
100 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
1. A constant sequence such as η
n
= .5 may well suﬃce for linear convergence,
but at a price of a greater number of outer iterations. We will discuss other
factors in the choice of forcing terms in ¸ 6.3.
Proposition 6.1.1. Let the standard assumptions hold and let x
n
→ x
∗
.
Then F(x
n
) converges qlinearly to 0 if and only if e
n

∗
does.
6.1.3. Errors in the function. In this section, we state two theorems on
inexact local improvement. We leave the proofs, which are direct analogs to
the proofs of Theorems 6.1.1 and 6.1.3, to the reader as exercises. Note that
errors in the derivative, such as those arising from a diﬀerence approximation
of the action of F
(x
c
) on a vector, can be regarded as part of the inexact
solution of the equation for the Newton step. See [55] or Proposition 6.2.1 and
its proof for an illustration of this point.
Theorem 6.1.5. Let the standard assumptions hold. Then there are δ and
K
I
such that if x
c
∈ B(δ), s satisﬁes
F
(x
c
)s +F(x
c
) +(x
c
) ≤ η
c
F(x
c
) +(x
c
) (6.10)
and x
+
= x
c
+s, then
e
+
 ≤ K
I
((e
c
 +η
c
)e
c
 +(x
c
)). (6.11)
Theorem 6.1.6. Let the standard assumptions hold. Then there is δ such
that if x
c
∈ B(δ), s satisﬁes (6.10), x
+
= x
c
+ s, and η
c
≤ η < ¯ η < 1, then
there is K
E
such that
F
(x
∗
)e
+
 ≤ ¯ ηF
(x
∗
)e
c
 +K
E
(x
c
). (6.12)
6.2. Newtoniterative methods
A Newtoniterative method realizes (6.1) with an iterative method for the linear
system for the Newton step, terminating when the relative linear residual is
smaller than η
c
(i.e, when (6.1) holds). The name of the method indicates
which linear iterative method is used, for example, NewtonSOR, NewtonCG,
NewtonGMRES, would use SOR, CG, or GMRES to solve the linear system.
This naming convention is taken from [175] and [145]. These methods have also
been called truncated Newton methods [56], [136] in the context of optimization.
Typically the nonlinear iteration that generates the sequence ¦x
n
¦ is called
the outer iteration and the linear iteration that generates the approximations
to the steps is called the inner iteration.
In this section we use the l
2
norm to measure nonlinear residuals. The
reason for this is that the Krylov methods use the scalar product and the
estimates in Chapters 2 and 3 are in terms of the Euclidean norm. In the
examples discussed in ¸ 6.4 we will scale the l
2
norm by a factor of 1/N so
that the results for the diﬀerential and integral equations will be independent
of the computational mesh.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
INEXACT NEWTON METHODS 101
6.2.1. Newton GMRES. We provide a detailed analysis of the Newton
GMRES iteration. We begin by discussing the eﬀects of a forward diﬀerence
approximation to the action of the Jacobian on a vector.
If the linear iterative method is any one of the Krylov subspace methods
discussed in Chapters 2 and 3 then each inner iteration requires at least one
evaluation of the action of F
(x
c
) on a vector. In the case of CGNR and CGNE,
an evaluation of the action of F
(x
c
)
T
on a vector is also required. In many
implementations [20], [24], the action of F
(x
c
) on a vector w is approximated
by a forward diﬀerence, (5.15), D
h
F(x : w) for some h. It is important to note
that this is entirely diﬀerent from forming the ﬁnite diﬀerence Jacobian ∇
h
F(x)
and applying that matrix to w. In fact, as pointed out in [20], application
of GMRES to the linear equation for the Newton step with matrixvector
products approximated by ﬁnite diﬀerences is the same as the application of
GMRES to the matrix G
h
F(x) whose last k −1 columns are the vectors
v
k
= D
h
F(x : v
k−1
)
The sequence ¦v
k
¦ is formed in the course of the forwarddiﬀerence GMRES
iteration.
To illustrate this point, we give the forwarddiﬀerence GMRES algorithm
for computation of the Newton step, s = −F
(x)
−1
F(x). Note that matrix
vector products are replaced by forward diﬀerence approximations to F
(x), a
sequence of approximate steps ¦s
k
¦ is produced, that b = −F(x), and that the
initial iterate for the linear problem is the zero vector. We give a MATLAB
implementation of this algorithm in the collection of MATLAB codes.
Algorithm 6.2.1. fdgmres(s, x, F, h, η, kmax, ρ)
1. s = 0, r = −F(x), v
1
= r/r
2
, ρ = r
2
, β = ρ, k = 0
2. While ρ > ηF(x)
2
and k < kmax do
(a) k = k + 1
(b) v
k+1
= D
h
F(x : v
k
)
for j = 1, . . . k
i. h
jk
= v
T
k+1
v
j
ii. v
k+1
= v
k+1
−h
jk
v
j
(c) h
k+1,k
= v
k+1

2
(d) v
k+1
= v
k+1
/v
k+1

2
(e) e
1
= (1, 0, . . . , 0)
T
∈ R
k+1
Minimize βe
1
−H
k
y
k

R
k+1 to obtain y
k
∈ R
k
.
(f) ρ = βe
1
−H
k
y
k

R
k+1.
3. s = V
k
y
k
.
In our MATLAB implementation we solve the least squares problem in
step 2e by the Givens rotation approach used in Algorithm gmres.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
102 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
As should be clear from inspection of Algorithm gmres, the diﬀerence
between ∇
h
F and G
h
F is that the columns are computed by taking directional
derivatives based on two diﬀerent orthonormal bases: ∇
h
F using the basis of
unit vectors in coordinate directions and G
h
F the basis for the Krylov space /
k
constructed by algorithm fdgmres. The action of G
h
F on /
⊥
k
, the orthogonal
complement of /
k
, is not used by the algorithm and, for the purposes of
analysis, could be speciﬁed by the action of ∇
h
F on any basis for /
⊥
k
. Hence
G
h
F is also ﬁrst order accurate in h. Assuming that there is no error in the
evaluation of F we may summarize the observations above in the following
proposition, which is a special case of the more general results in [20].
Proposition 6.2.1. Let the standard assumptions hold. Let η ∈ (0, 1).
Then there are C
G
,
¯
h, and δ such that if x ∈ B(δ), h ≤
¯
h, and Algo
rithm fdgmres terminates with k < kmax then the computed step s satisﬁes
F
(x)s +F(x)
2
< (η +C
G
h)F(x)
2
. (6.13)
Proof. First let δ be small enough so that B(δ) ⊂ Ω and the conclusions
to Lemma 4.3.1 hold. Let ¦u
j
¦ be any orthonormal basis for R
N
such that
u
j
= v
j
for 1 ≤ j ≤ k, where ¦v
j
¦
k
j=1
is the orthonormal basis for /
k
generated
by Algorithm fdgmres.
Consider the linear map deﬁned by its action on ¦u
j
¦
Bu
j
= D
h
F(x : u
j
).
Note that
D
h
F(x : u
j
) =
_
1
0
F
(x +thx
2
u
j
)u
j
dt
= F
(x)u
j
+
_
1
0
(F
(x +thx
2
u
j
) −F
(x))u
j
dt.
Since F
is Lipschitz continuous and ¦u
j
¦ is an orthonormal basis for R
N
we
have, with ¯ γ = γ(x
∗

2
+δ),
B −F
(x)
2
≤ hx
2
γ/2 ≤ h¯ γ/2 (6.14)
Since the linear iteration (fdgmres) terminates in k < kmax iterations, we
have, since B and G
h
F agree on /
k
,
Bs +F(x)
2
≤ ηF(x)
2
and therefore
F
(x)s +F(x)
2
≤ ηF(x)
2
+h¯ γs
2
/2. (6.15)
Assume that
¯
h¯ γ ≤ F
(x
∗
)
−1

−1
2
/2
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
INEXACT NEWTON METHODS 103
Then Lemma 4.3.1 and (6.15) imply
F
(x
∗
)
−1

−1
2
s
2
/2 ≤ F
(x)
−1

−1
2
s
2
≤ F
(x)s
2
≤ (1 +η)F(x)
2
+
¯
h¯ γs
2
/2.
Therefore,
s
2
≤ 4(1 +η)F
(x
∗
)
−1

2
F(x)
2
. (6.16)
Combining (6.16) with (6.15) completes the proof with C
G
= 4¯ γ(1 +
η)F
(x
∗
)
−1

2
.
Proposition 6.2.1 implies that a ﬁnite diﬀerence implementation will not
aﬀect the performance of NewtonGMRES if the steps in the forward diﬀerence
approximation of the derivatives are suﬃciently small. In an implementation,
therefore, we must specify not only the sequence ¦η
n
¦, but also the steps ¦h
n
¦
used to compute forward diﬀerences. Note also that Proposition 6.2.1 applies
to a restarted GMRES because upon successful termination (6.15) will hold.
We summarize our results so far in the analogs of Theorem 6.1.4 and
Theorem 6.1.2. The proofs are immediate consequences of Theorems 6.1.2,
6.1.4, and Proposition 6.2.1 and are left as (easy) exercises.
Theorem 6.2.1. Assume that the assumptions of Proposition 6.2.1 hold.
Then there are δ, ¯ σ such that if x
0
∈ B(δ) and the sequences ¦η
n
¦ and ¦h
n
¦
satisfy
σ
n
= η
n
+C
G
h
n
≤ ¯ σ
then the forward diﬀerence NewtonGMRES iteration
x
n+1
= x
n
+s
n
where s
n
is computed by Algorithm fdgmres with arguments
(s
n
, x
n
, F, h
n
, η
n
, kmax, ρ)
converges qlinearly and s
n
satisﬁes
F
(x
n
)s
n
+F(x
n
)
2
< σ
n
F(x
n
)
2
. (6.17)
Moreover,
• if σ
n
→ 0 the convergence is qsuperlinear, and
• if σ
n
≤ K
η
F(x
n
)
p
2
for some K
η
> 0 the convergence is qsuperlinear
with qorder 1 +p.
Theorem 6.2.2. Assume that the assumptions of Proposition 6.2.1 hold.
Then there is δ such that if x
0
∈ B(δ) and the sequences ¦η
n
¦ and ¦h
n
¦ satisfy
σ
n
= η
n
+C
G
h
n
∈ [0, ¯ σ]
with 0 < ¯ σ < 1 then the forward diﬀerence NewtonGMRES iteration
x
n+1
= x
n
+s
n
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
104 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
where s
n
is computed by Algorithm fdgmres with arguments
(s
n
, x
n
, F, h
n
, η
n
, kmax, ρ)
converges qlinearly with respect to  
∗
and s
n
satisﬁes (6.17). Moreover
• if σ
n
→ 0 the convergence is qsuperlinear, and
• if σ
n
≤ K
η
F(x
n
)
p
2
for some K
η
> 0 the convergence is qsuperlinear
with qorder 1 +p.
In the case where h
n
is approximately the square root of machine roundoﬀ,
σ
n
≈ η
n
if η
n
is not too small and Theorem 6.2.2 states that the observed
behavior of the forward diﬀerence implementation will be the same as that of
an implementation with exact derivatives.
6.2.2. Other Newtoniterative methods. Other iterative methods may
be used as the solver for the linear equation for the Newton step. Newton
multigrid [99] is one approach using a stationary iterative method. See also
[6], [111]. If F
is symmetric and positive deﬁnite, as it is in the context of
unconstrained minimization, [63], NewtonCG is a possible approach.
When storage is restricted or the problem is very large, GMRES may not
be practical. GMRES(m), for a small m, may not converge rapidly enough
to be useful. CGNR and CGNE oﬀer alternatives, but require evaluation of
transposevector products. BiCGSTAB and TFQMR should be considered in
all cases where there is not enough storage for GMRES(m) to perform well.
If a transpose is needed, as it will be if CGNR or CGNE is used as the
iterative method, a forward diﬀerence formulation is not an option because
approximation of the transposevector product by diﬀerences is not possible.
Computation of ∇
h
F(x) and its transpose analytically is one option as is
diﬀerentiation of F automatically or by hand. The reader may explore this
further in Exercise 6.5.13.
The reader may also be interested in experimenting with the other Krylov
subspace methods described in ¸ 3.6, [78], and [12].
6.3. NewtonGMRES implementation
We provide a MATLAB code nsolgm in the collection that implements
a forwarddiﬀerence NewtonGMRES algorithm. This algorithm requires
diﬀerent inputs than nsol. As input we must give the initial iterate, the
function, the vector of termination tolerances as we did for nsol. In addition,
we must provide a method for forming the sequence of forcing terms ¦η
n
¦.
We adjust η as the iteration progresses with a variation of a choice from
[69]. This issue is independent of the particular linear solver and our discussion
is in the general inexact Newton method setting. Setting η to a constant for
the entire iteration is often a reasonable strategy as we see in ¸ 6.4, but the
choice of that constant depends on the problem. If a constant η is too small,
much eﬀort can be wasted in the initial stages of the iteration. The choice in
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
INEXACT NEWTON METHODS 105
[69] is motivated by a desire to avoid such over solving. Over solving means
that the linear equation for the Newton step is solved to a precision far beyond
what is needed to correct the nonlinear iteration. As a measure of the degree
to which the nonlinear iteration approximates the solution we begin with
η
A
n
= γF(x
n
)
2
/F(x
n−1
)
2
,
where γ ∈ (0, 1] is a parameter. If η
A
n
is uniformly bounded away from 1,
then setting η
n
= η
A
n
for n > 0 would guarantee qquadratic convergence by
Theorem 6.1.1. In this way, the most information possible would be extracted
from the inner iteration. In order to specify the choice at n = 0 and bound
the sequence away from 1 we set
η
B
n
=
_
¸
_
¸
_
η
max
, n = 0,
min(η
max
, η
A
n
), n > 0.
(6.18)
In (6.18) the parameter η
max
is an upper limit on the sequence ¦η
n
¦. In [69]
the choices γ = .9 and η
max
= .9999 are used.
It may happen that η
B
n
is small for one or more iterations while x
n
is still
far from the solution. A method of safeguarding was suggested in [69] to avoid
volatile decreases in η
n
. The idea is that if η
n−1
is suﬃciently large we do not
let η
n
decrease by much more than a factor of η
n−1
.
η
C
n
=
_
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
_
η
max
, n = 0,
min(η
max
, η
A
n
), n > 0, γη
2
n−1
≤ .1,
min(η
max
, max(η
A
n
, γη
2
n−1
)), n > 0, γη
2
n−1
> .1.
(6.19)
The constant .1 is somewhat arbitrary. This safeguarding does improve the
performance of the iteration.
There is a chance that the ﬁnal iterate will reduce F far beyond the
desired level and that the cost of the solution of the linear equation for the last
step will be more accurate than is really needed. This oversolving on the ﬁnal
step can be controlled comparing the norm of the current nonlinear residual
F(x
n
) to the nonlinear residual norm at which the iteration would terminate
τ
t
= τ
a
+τ
r
F(x
0
)
and bounding η
n
from below by a constant multiple of τ
t
/F(x
n
). The
algorithm nsolgm does this and uses
η
n
= min(η
max
, max(η
C
n
, .5τ
t
/F(x
n
))). (6.20)
Exercise 6.5.9 asks the reader to modify nsolgm so that other choices of
the sequence ¦η
n
¦ can be used.
We use dirder to approximate directional derivatives and use the default
value of h = 10
−7
. In all the examples in this book we use the value γ = .9 as
recommended in [69].
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
106 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Algorithm 6.3.1. nsolgm(x, F, τ, η)
1. r
c
= r
0
= F(x)
2
/
√
N
2. Do while F(x)
2
/
√
N > τ
r
r
0
+τ
a
(a) Select η.
(b) fdgmres(s, x, F, η)
(c) x = x +s
(d) Evaluate F(x)
(e) r
+
= F(x)
2
/
√
N, σ = r
+
/r
c
, r
c
= r
+
(f) If F(x)
2
≤ τ
r
r
0
+τ
a
exit.
Note that the cost of an outer iterate is one evaluation of F to compute
the value at the current iterate and other evaluations of F to compute the
forward diﬀerences for each inner iterate. Hence if a large value of η can be
used, the cost of the entire iteration can be greatly reduced. If a maximum
of m GMRES iterations is needed (or GMRES(m) is used as a linear solver)
the storage requirements of Algorithm nsolgm are the m+ 5 vectors x, F(x),
x +hv, F(x +hv), s, and the Krylov basis ¦v
k
¦
m
k=1
.
Since GMRES forms the inner iterates and makes termination decisions
based on scalar products and l
2
norms, we also terminate the outer iteration on
small l
2
norms of the nonlinear residuals. However, because of our applications
to diﬀerential and integral equations, we scale the l
2
norm of the nonlinear
residual by a factor of 1/
√
N so that constant functions will have norms that
are independent of the computational mesh.
For large and poorly conditioned problems, GMRES will have to be
restarted. We illustrate the consequences of this in ¸ 6.4 and ask the reader to
make the necessary modiﬁcations to nsolgm in Exercise 6.5.9.
The preconditioners we use in ¸ 6.4 are independent of the outer iteration.
Since the Newton steps for MF(x) = 0 are the same as those for F(x) = 0, it
is equivalent to precondition the nonlinear equation before calling nsolgm.
6.4. Examples for NewtonGMRES
In this section we consider two examples. We revisit the Hequation and solve
a preconditioned nonlinear convectiondiﬀusion equation.
In all the ﬁgures we plot the relative nonlinear residual F(x
n
)
2
/F(x
0
)
2
against the number of function evaluations required by all inner and outer
iterations to compute x
n
. Counts of function evaluations corresponding to
outer iterations are indicated by circles. From these plots one can compare
not only the number of outer iterations, but also the total cost. This enables
us to directly compare the costs of diﬀerent strategies for selection of η. Note
that if only a single inner iteration is needed, the total cost of the outer iteration
will be two function evaluations since F(x
c
) will be known. One new function
evaluation will be needed to approximate the action of F
(x
c
) on a vector
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
INEXACT NEWTON METHODS 107
and then F(x
+
) will be evaluated to test for termination. We also count
the evaluation of F(x
0
) as an additional cost in the evaluation of x
1
, which
therefore has a minimum cost of three function evaluations. For each example
we compare a constant value of η
n
with the choice given in (6.20) and used as
the default in nsolgm.
6.4.1. Chandrasekhar Hequation. We solve the Hequation on a 100
point mesh with c = .9 using two schemes for selection of the parameter η.
The initial iterate is the function identically one. We set the parameters in
(6.20) to γ = .9 and η
max
= .25.
We used τ
r
= τ
a
= 10
−6
in this example.
In Fig. 6.1 we plot the progress of the iteration using η = .1 with the solid
line and using the sequence given by (6.20) with the dashed line. We set the
parameters in (6.20) to γ = .9 and η
max
= .25. This choice of η
max
was the
one that did best overall in our experiments on this problem.
We see that the constant η iteration terminates after 12 function evalua
tions, 4 outer iterations, and roughly 275 thousand ﬂoatingpoint operations.
This is a slightly higher overall cost than the other approach in which ¦η
n
¦ is
given by (6.20), which terminated after 10 function evaluations, 3 outer itera
tions, and 230 thousand ﬂoatingpoint operations. The chord method with the
diﬀerent (but very similar for this problem) l
∞
termination condition for the
same problem, reported in ¸ 5.6 incurred a cost of 3.3 million ﬂoatingpoint
operations, slower by a factor of over 1000. This cost is entirely a result of the
computation and factorization of a single Jacobian.
0 2 4 6 8 10 12
10
8
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 6.1. NewtonGMRES for the Hequation, c = .9.
In Fig. 6.2 the results change for the more illconditioned problem with
c = .9999. Here the iteration with η
n
= .1 performed slightly better and
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
108 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
0 5 10 15 20 25
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 6.2. NewtonGMRES for the Hequation, c = .9999
0 2 4 6 8 10 12 14 16 18
10
4
10
3
10
2
10
1
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 6.3. NewtonGMRES for the PDE, C = 20.
terminated after 22 function evaluations and 7 outer iterations, and roughly
513 thousand ﬂoatingpoint operations. The iteration with the decreasing ¦η
n
¦
given by (6.20) required 23 function evaluations, 7 outer iterations, and 537
thousand ﬂoatingpoint operations.
6.4.2. Convectiondiﬀusion equation. We consider the partial diﬀeren
tial equation
−∇
2
u +Cu(u
x
+u
y
) = f (6.21)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
INEXACT NEWTON METHODS 109
with homogeneous Dirichlet boundary conditions on the unit square (0, 1)
(0, 1). f has been constructed so that the exact solution was the discretization
of
10xy(1 −x)(1 −y) exp(x
4.5
).
We set C = 20 and u
0
= 0. As in ¸ 3.7 we discretized (6.21) on a 31 31 grid
using centered diﬀerences. To be consistent with the secondorder accuracy of
the diﬀerence scheme we used τ
r
= τ
a
= h
2
, where h = 1/32.
We compare the same possibilities for ¦η
n
¦ as in the previous example and
report the results in Fig. 6.3. In (6.20) we set γ = .9 and η
max
= .5. The
computation with constant η terminated after 19 function evaluations and 4
outer iterates at a cost of roughly 3 million ﬂoatingpoint operations. The
iteration with ¦η
n
¦ given by (6.20) required 16 function evaluations, 4 outer
iterations, and 2.6 million ﬂoatingpoint operations. In the computation we
preconditioned (6.21) with the fast Poisson solver fish2d.
In this case, the choice of ¦η
n
¦ given by (6.20) reduced the number of inner
iterations in the early stages of the iteration and avoided oversolving. In cases
where the initial iterate is very good and only a single Newton iterate might be
needed, however, a large choice of η
max
or constant η may be the more eﬃcient
choice. Examples of such cases include (1) implicit time integration of nonlinear
parabolic partial diﬀerential equations or ordinary diﬀerential equations in
which the initial iterate is either the converged solution from the previous
time step or the output of a predictor and (2) solution of problems in which
the initial iterate is an interpolant of the solution from a coarser mesh.
Preconditioning is crucial for good performance. When preconditioning
was not done, the iteration for C = 20 required more than 80 function
evaluations to converge. In Exercise 6.5.9 the reader is asked to explore this.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
110 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
6.5. Exercises on inexact Newton methods
6.5.1. Verify (6.8).
6.5.2. Prove Proposition 6.1.1 by showing that if the standard assumptions
hold, 0 < < 1, and x is suﬃciently near x
∗
then
F(x)/(1 +) ≤ F
(x
∗
)e ≤ (1 +)F(x).
6.5.3. Verify (6.14).
6.5.4. Prove Theorems 6.2.2 and 6.2.1.
6.5.5. Prove Theorems 6.1.5 and 6.1.6.
6.5.6. Can anything like Proposition 6.2.1 be proved for a ﬁnitediﬀerence Bi
CGSTAB or TFQMR linear solver? If not, what are some possible
implications?
6.5.7. Give an example in which F(x
n
) → 0 qlinearly but x
n
does not converge
to x
∗
qlinearly.
6.5.8. In the DASPK code for integration of diﬀerential algebraic equations,
[23], preconditioned NewtonGMRES is implemented in such a way
that the data needed for the preconditioner is only computed when
needed, which may be less often that with each outer iteration. Discuss
this strategy and when it might be useful. Is there a relation to the
Shamanskii method?
6.5.9. Duplicate the results in ¸ 6.4. For the Hequation experiment with
various values of c, such as c = .5, .99999, .999999, and for the convection
diﬀusion equation various values of C such as C = 1, 10, 40 and how
preconditioning aﬀects the iteration. Experiment with the sequence of
forcing terms ¦η
n
¦. How do the choices η
n
= 1/(n + 1), η
n
= 10
−4
[32], η
n
= 2
1−n
[24], η
n
= min(F(x
n
)
2
, (n + 2)
−1
) [56], η
n
= .05, and
η
n
= .75 aﬀect the results? Present the results of your experiments as
graphical iteration histories.
6.5.10. For the Hequation example in ¸ 6.4, vary the parameter c and see how
the performance of NewtonGMRES with the various choices of ¦η
n
¦ is
aﬀected. What happens when c = 1? See [119] for an explanation of
this.
6.5.11. Are the converged solutions for the preconditioned and unpreconditioned
convectiondiﬀusion example in ¸ 6.4 equally accurate?
6.5.12. For the convectiondiﬀusion example in ¸ 6.4, how is the performance
aﬀected if GMRES(m) is used instead of GMRES?
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
INEXACT NEWTON METHODS 111
6.5.13. Apply NewtonCGNR, NewtonBiCGSTAB, or NewtonTFQMR to the
two examples in ¸ 6.4. How will you deal with the need for a transpose in
the case of CGNR? How does the performance (in terms of time, ﬂoating
point operations, function evaluations) compare to NewtonGMRES?
Experiment with the forcing terms. The MATLAB codes fdkrylov,
which allows you to choose from a variety of forward diﬀerence Krylov
methods, and nsola, which we describe more fully in Chapter 8, might
be of use in this exercise.
6.5.14. If you have done Exercise 5.7.25, apply nsolgm to the same twopoint
boundary value problem and compare the performance.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
112 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Chapter 7
Broyden’s method
QuasiNewton methods maintain approximations of the solution x
∗
and the
Jacobian at the solution F
(x
∗
) as the iteration progresses. If x
c
and B
c
are
the current approximate solution and Jacobian, then
x
+
= x
c
−B
−1
c
F(x
c
). (7.1)
After the computation of x
+
, B
c
is updated to form B
+
. The construction of
B
+
determines the quasiNewton method.
The advantages of such methods, as we shall see in ¸ 7.3, is that solution of
equations with the quasiNewton approximation to the Jacobian is often much
cheaper than using F
(x
n
) as the coeﬃcient matrix. In fact, if the Jacobian is
dense, the cost is little more than that of the chord method. The method we
present in this chapter, Broyden’s method, is locally superlinearly convergent,
and hence is a very powerful alternative to Newton’s method or the chord
method.
In comparison with Newtoniterative methods, quasiNewton methods
require only one function evaluation for each nonlinear iterate; there is no cost
in function evaluations associated with an inner iteration. Hence, if a good
preconditioner (initial approximation to F
(x
∗
)) can be found, these methods
could have an advantage in terms of function evaluation cost over Newton
iterative methods.
Broyden’s method [26] computes B
+
by
B
+
= B
c
+
(y −B
c
s)s
T
s
T
s
= B
c
+
F(x
+
)s
T
s
T
s
. (7.2)
In (7.2) y = F(x
+
) −F(x
c
) and s = x
+
−x
c
.
Broyden’s method is an example of a secant update. This means that the
updated approximation to F
(x
∗
) satisﬁes the secant equation
B
+
s = y. (7.3)
In one dimension, (7.3) uniquely speciﬁes the classical secant method. For
equations in several variables (7.3) alone is a system of N equations in N
2
113
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
114 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
unknowns. The role of the secant equation is very deep. See [62] and [64] for
discussion of this topic. In this book we focus on Broyden’s method and our
methods of analysis are not as general as those in [62] and [64]. In this chapter
we will assume that the function values are accurate and refer the reader to [66],
[65], and [114] for discussions of quasiNewton methods with inaccurate data.
Locally superlinearly convergent secant methods can be designed to maintain
the sparsity pattern, symmetry, or positive deﬁniteness, of the approximate
Jacobian, and we refer the reader to [28], [63], [62], and [64] for discussion and
analysis of several such methods. More subtle structural properties can also
be preserved by secant methods. Some examples can be found in [101], [96],
[95], [113], and [116].
Broyden’s method is also applicable to linear equations [29], [82], [84],
[104], [143], [130], Ax = b, where B ≈ A is updated. One can also apply the
method to linear least squares problems [84], [104]. The analysis is somewhat
clearer in the linear case and the plan of this chapter is to present results for
the linear case for both the theory and the examples. One interesting result,
which we will not discuss further, is that for linear problems Broyden’s method
converges in 2N iterations [29], [82], [84], [143].
We will express our results in terms of the error in the Jacobian
E = B −F
(x
∗
) (7.4)
and the step s = x
+
−x
c
. While we could use E = B−F
(x), (7.4) is consistent
with our use of e = x − x
∗
. When indexing of the steps is necessary, we will
write
s
n
= x
n+1
−x
n
.
In this chapter we show that if the data x
0
and B
0
are suﬃciently good,
the Broyden iteration will converge qsuperlinearly to the root.
7.1. The Dennis–Mor´e condition
In this section we consider general iterative methods of the form
x
n+1
= x
n
−B
−1
n
F(x
n
) (7.5)
where B
n
= F
(x
∗
)+E
n
≈ F
(x
∗
) is generated by some method (not necessarily
Broyden’s method).
Veriﬁcation of qsuperlinear convergence cannot be done by the direct
approaches used in Chapters 5 and 6. We must relate the superlinear
convergence condition that η
n
→ 0 in Theorem 6.1.2 to a more subtle, but
easier to verify condition in order to analyze Broyden’s method. This technique
is based on veriﬁcation of the Dennis–Mor´e condition [61], [60] on the sequence
of steps ¦s
n
¦ and errors in the Jacobian ¦E
n
¦
lim
n→∞
E
n
s
n

s
n

= 0. (7.6)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 115
The main result in this section, Theorem 7.1.1 and its corollary for linear
problems, are used to prove superlinear convergence for Broyden’s method in
this chapter and many other quasiNewton methods. See [61], [28], [62], and
[64] for several more examples. Our formulation of the Dennis–Mor´e result is
a bit less general than that in [60] or [63].
Theorem 7.1.1. Let the standard assumptions hold, let ¦B
n
¦ be a sequence
of nonsingular N N matrices, let x
0
∈ R
N
be given and let ¦x
n
¦
∞
n=1
be given
by (7.5). Assume that x
n
,= x
∗
for any n. Then x
n
→ x
∗
qsuperlinearly if
and only if x
n
→ x
∗
and the Dennis–Mor´e condition (7.6) holds.
Proof. Since
−F(x
n
) = B
n
s
n
= F
(x
∗
)s
n
+E
n
s
n
we have
E
n
s
n
= −F
(x
∗
)s
n
−F(x
n
) = −F
(x
∗
)e
n+1
+F
(x
∗
)e
n
−F(x
n
). (7.7)
We use the fundamental theorem of calculus and the standard assumptions to
obtain
F
(x
∗
)e
n
−F(x
n
) =
_
1
0
(F
(x
∗
) −F
(x
∗
+te
n
))e
n
dt
and hence
F
(x
∗
)e
n
−F(x
n
) ≤ γe
n

2
/2.
Therefore, by (7.7)
E
n
s
n
 ≤ F
(x
∗
)e
n+1
 +γe
n

2
/2. (7.8)
Now, if x
n
→ x
∗
qsuperlinearly, then for n suﬃciently large
s
n
/2 ≤ e
n
 ≤ 2s
n
. (7.9)
The assumption that x
n
,= x
∗
for any n implies that the sequence
ν
n
= e
n+1
/e
n
 (7.10)
is deﬁned and and ν
n
→ 0 by superlinear convergence of ¦x
n
¦. By (7.9) we
have
e
n+1
 = ν
n
e
n
 ≤ 2ν
n
s
n
,
and hence (7.8) implies that
E
n
s
n
 ≤ (2F
(x
∗
)ν
n
+γe
n
)s
n

which implies the DennisMor´e condition (7.6).
Conversely, assume that x
n
→ x
∗
and that (7.6) holds. Let
µ
n
=
E
n
s
n

s
n

.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
116 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
We have, since
E
n
s
n
= (B
n
−F
(x
∗
))s
n
= −F(x
n
) −F
(x
∗
)s
n
, (7.11)
s
n
F
(x
∗
)
−1

−1
≤ F
(x
∗
)s
n
 ≤ E
n
s
n
 +F(x
n
)
= µ
n
s
n
 +F(x
n
).
(7.12)
Since µ
n
→ 0 by assumption,
µ
n
≤ F
(x
∗
)
−1

−1
/2, (7.13)
for n suﬃciently large. We assume now that n is large enough so that (7.13)
holds. Hence,
s
n
 ≤ 2F
(x
∗
)
−1
F(x
n
). (7.14)
Moreover, using the standard assumptions and (7.11) again,
F(x
n
) +F
(x
n
)s
n
 ≤ F(x
n
) +F
(x
∗
)s
n
 +(F
(x
∗
) −F
(x
n
))s
n

≤ (µ
n
+γe
n
)s
n
.
(7.15)
Since x
n
→ x
∗
by assumption, η
n
→ 0 where
η
n
= 2F
(x
∗
)
−1
(µ
n
+γe
n
).
Combining (7.14) and (7.15) implies that
F(x
n
) +F
(x
n
)s
n
 ≤ η
n
F(x
n
), (7.16)
which is the inexact Newton condition. Since η
n
→ 0, the proof is complete
by Theorem 6.1.2.
7.2. Convergence analysis
Our convergence analysis is based on an approach to inﬁnitedimensional
problems from [97]. That paper was the basis for [168], [104], and [115], and
we employ notation from all four of these references. We will use the l
2
norm
throughout our discussion. We begin with a lemma from [104].
Lemma 7.2.1. Let 0 <
ˆ
θ < 1 and let
¦θ
n
¦
∞
n=0
⊂ (
ˆ
θ, 2 −
ˆ
θ).
Let ¦
n
¦
∞
n=0
⊂ R
N
be such that
n

n

2
< ∞,
and let ¦η
n
¦
∞
n=0
be a set of vectors in R
N
such that η
n

2
is either 1 or 0 for
all n. Let ψ
0
∈ R
N
be given. If ¦ψ
n
¦
∞
n=1
is given by
ψ
n+1
= ψ
n
−θ
n
(η
T
n
ψ
n
)η
n
+
n
(7.17)
then
lim
n→∞
η
T
n
ψ
n
= 0. (7.18)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 117
Proof. We ﬁrst consider the case in which
n
= 0 for all n. In that case, we
can use the fact that
θ
n
(2 −θ
n
) >
ˆ
θ
2
> 0
to show that the sequence ¦ψ
n
¦ is bounded in l
2
norm by ψ
0

2
and, in fact,
ψ
n+1

2
2
≤ ψ
n

2
2
−θ
n
(2 −θ
n
)(η
T
n
ψ
n
)
2
≤ ψ
n

2
2
−
ˆ
θ
2
(η
T
n
ψ
n
)
2
≤ ψ
n

2
2
.
Therefore, for any M > 0,
M
n=0
(η
T
n
ψ
n
)
2
≤
ψ
0

2
2
−ψ
M+1

2
2
ˆ
θ
2
≤
ψ
0

2
ˆ
θ
2
.
We let M → ∞ to obtain
∞
n=0
(η
T
n
ψ
n
)
2
< ∞. (7.19)
Convergence of the series in (7.19) implies convergence to zero of the terms in
the series, which is (7.18).
To prove the result for
n
,= 0 we use the inequality
_
a
2
−b
2
≤ a −
b
2
2a
, (7.20)
which is valid for a > 0 and [b[ ≤ a. This inequality is used often in the analysis
of quasiNewton methods [63]. From (7.20) we conclude that if ψ
n
,= 0 then
ψ
n
−θ
n
(η
T
n
ψ
n
)η
n

2
≤
_
ψ
n

2
2
−θ
n
(2 −θ
n
)(η
T
n
ψ
n
)
2
≤ ψ
n

2
−
θ
n
(2 −θ
n
)(η
T
n
ψ
n
)
2
2ψ
n

2
.
Hence if ψ
n
,= 0
ψ
n+1

2
≤ ψ
n

2
−
θ
n
(2 −θ
n
)(η
T
n
ψ
n
)
2
2ψ
n

2
+
n

2
. (7.21)
Hence
(η
T
n
ψ
n
)
2
≤
2ψ
n

2
θ
n
(2 −θ
n
)
(ψ
n

2
−ψ
n+1

2
+
n

2
), (7.22)
which holds even if ψ
n
= 0.
From (7.21) we conclude that
ψ
n+1

2
≤ µ,
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
118 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
where
µ =
∞
i=0

i

2
+ψ
0

2
.
Hence
M
n=0
(η
T
n
ψ
n
)
2
≤
2µ
ˆ
θ
2
M
n=0
(ψ
n

2
−ψ
n+1

2
+
n

2
)
=
2µ
ˆ
θ
2
_
ψ
0

2
−ψ
M+1

2
+
M
n=0

n

2
_
≤
2µ
2
ˆ
θ
2
.
Hence (7.19) holds and the proof is complete.
7.2.1. Linear problems. We may use Lemma 7.2.1 to prove a result from
[130] on the convergence of Broyden’s method for linear problems. The role of
the parameter
ˆ
θ and the sequence ¦θ
n
¦ in Lemma 7.2.1 is nicely illustrated by
this application of the lemma.
In this section F(x) = Ax−b and B
n
= A+E
n
. The standard assumptions
in this case reduce to nonsingularity of A. Our ﬁrst task is to show that
the errors in the Broyden approximations to A do not grow. This property,
called bounded deterioration [28], is an important part of the convergence
analysis. The reader should compare the statement of the result with that
in the nonlinear case.
In the linear case, the Broyden update has a very simple form. We will
consider a modiﬁed update
B
+
= B
c
+θ
c
(y −B
c
s)s
T
s
T
s
= B
c
+θ
c
(Ax
+
−b)s
T
s
T
s
. (7.23)
If x
c
and B
c
are the current approximations to x
∗
= A
−1
b and A and
s = x
+
−x
c
then
y −B
c
s = (Ax
+
−Ax
c
) −B
c
(x
+
−x
c
) = −E
c
s
and therefore
B
+
= B
c
+θ
c
(y −B
c
s)s
T
s
2
2
= B
c
−θ
c
(E
c
s)s
T
s
2
2
. (7.24)
Lemma 7.2.2. Let A be nonsingular, θ
c
∈ [0, 2], and x
c
∈ R
N
be given. Let
B
c
be nonsingular and let B
+
be formed by the Broyden update (7.23). Then
E
+

2
≤ E
c

2
. (7.25)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 119
Proof. By subtracting A from both sides of (7.24) we have
E
+
= E
c
−θ
c
(E
c
s)s
T
s
2
2
= E
c
(I −θ
c
P
s
),
(7.26)
where P
s
is the orthogonal projector
P
s
=
ss
T
s
2
2
. (7.27)
This completes the proof as
E
+

2
≤ E
c

2
I −θ
c
P
s

2
and I −θ
c
P
s

2
≤ 1 by orthogonality of P
s
and the fact that 0 ≤ θ
c
≤ 2.
Now, θ
c
can always be selected to make B
+
nonsingular. In [130] one
suggestion was
θ
c
=
_
_
_
1, [γ
c
[ ≥ σ,
1 −sign(γ
c
)σ
1 −γ
c
, otherwise,
where
γ
c
=
(B
−1
c
y)
T
s
s
2
2
=
(B
−1
c
As)
T
s
s
2
2
and σ ∈ (0, 1) is ﬁxed. However the results in [130] assume only that the
sequence ¦θ
n
¦ satisﬁes the hypotheses of Lemma 7.2.1 for some
ˆ
θ ∈ (0, 1) and
that θ
c
is always chosen so that B
+
is nonsingular.
For linear problems one can show that the Dennis–Mor´e condition implies
convergence and hence the assumption that convergence takes place is not
needed. Speciﬁcally
Proposition 7.2.1. Let A be nonsingular. Assume that ¦θ
n
¦ satisﬁes
the hypotheses of Lemma 7.2.1 for some
ˆ
θ ∈ (0, 1). If ¦θ
n
¦ is such that
the matrices B
n
obtained by (7.23) are nonsingular and ¦x
n
¦ is given by the
modiﬁed Broyden iteration
x
n+1
= x
n
−B
−1
n
(Ax
n
−b) (7.28)
then the Dennis–Mor´e condition implies convergence of ¦x
n
¦ to x
∗
= A
−1
b.
We leave the proof as Exercise 7.5.1.
The superlinear convergence result is remarkable in that the initial iterate
need not be near the root.
Theorem 7.2.1. Let A be nonsingular. Assume that ¦θ
n
¦ satisﬁes the
hypotheses of Lemma 7.2.1 for some
ˆ
θ ∈ (0, 1). If ¦θ
n
¦ is such the matrices B
n
obtained by (7.23) are nonsingular then the modiﬁed Broyden iteration (7.28)
converges qsuperlinearly to x
∗
= A
−1
b for every initial iterate x
0
.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
120 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Proof. Our approach is to show that for every φ ∈ R
N
that
lim
n→∞
φ
T
_
E
n
s
n
s
n

2
_
= 0. (7.29)
Assuming that (7.29) holds, setting φ = e
j
, the unit vector in the jth
coordinate direction, shows that the j component of
E
n
s
n
s
n

2
has limit zero.
As j was arbitrary, we have
lim
n→∞
E
n
s
n
s
n

2
= 0
which is the Dennis–Mor´e condition. This will imply qsuperlinear convergence
by Theorem 7.1.1 and Proposition 7.2.1.
Let φ ∈ R
N
be given. Let
P
n
=
s
n
s
T
n
s
n

2
2
.
Since for all v ∈ R
N
and all n
v
T
(E
T
n
φ) = φ
T
(E
n
v)
and, by (7.26)
E
T
n+1
φ = (I −θ
n
P
n
)E
T
n
φ,
we may invoke Lemma 7.2.1 with
η
n
= s
n
/s
n

2
, ψ
n
= E
T
n
φ, and
n
= 0
to conclude that
η
T
n
ψ
n
=
(E
T
n
φ)
T
s
n
s
n

2
= φ
T
E
n
s
n
s
n

2
→ 0. (7.30)
This completes the proof.
The result here is not a local convergence result. It is global in the sense
that qsuperlinear convergence takes place independently of the initial choices
of x and B. It is known [29], [82], [84], [143], that the Broyden iteration will
terminate in 2N steps in the linear case.
7.2.2. Nonlinear problems. Our analysis of the nonlinear case is diﬀerent
from the classical work [28], [63]. As indicated in the preface, we provide a
proof based on ideas from [97], [115], and [104], that does not use the Frobenius
norm and extends to various inﬁnitedimensional settings.
For nonlinear problems our analysis is local and we set θ
n
= 1. However
we must show that the sequence ¦B
n
¦ of Broyden updates exists and remains
nonsingular. We make a deﬁnition that allows us to compactly express this
fact.
Definition 7.2.1. We say that the Broyden sequence (¦x
n
¦, ¦B
n
¦) for the
data (F, x
0
, B
0
) exists if F is deﬁned at x
n
for all n, B
n
is nonsingular for all
n, and x
n+1
and B
n+1
can be computed from x
n
and B
n
using (7.1) and (7.2).
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 121
We will show that if the standard assumptions hold and x
0
and B
0
are
suﬃciently good approximations to x
∗
and F
(x
∗
) then the Broyden sequence
exists for the data (F, x
0
, B
0
) and that the Broyden iterates converge q
superlinearly to x
∗
. The development of the proof is complicated. We ﬁrst
prove a bounded deterioration result like Lemma 7.2.2. However in the
nonlinear case, the errors in the Jacobian approximation can grow as the
iteration progresses. We then show that if the initial data is suﬃciently good,
this growth can be limited and therefore the Broyden sequence exists and
the Broyden iterates converges to x
∗
at least qlinearly. Finally we verify the
Dennis–Mor´e condition (7.6), which, together with the qlinear convergence
proved earlier completes the proof of local qsuperlinear convergence.
Bounded deterioration. Our ﬁrst task is to show that bounded deteriora
tion holds. Unlike the linear case, the sequence ¦E
n
¦ need not be monotone
nonincreasing. However, the possible increase can be bounded in terms of the
errors in the current and new iterates.
Theorem 7.2.2. Let the standard assumptions hold. Let x
c
∈ Ω and a
nonsingular matrix B
c
be given. Assume that
x
+
= x
c
−B
−1
c
F(x
c
) = x
c
+s ∈ Ω
and B
+
is given by (7.2). Then
E
+

2
≤ E
c
 +γ(e
c

2
+e
+

2
)/2 (7.31)
Proof. Note that (7.7) implies that
E
c
s = −F(x
+
) + (F(x
+
) −F(x
c
) −F
(x
∗
)s)
= −F(x
+
) +
_
1
0
(F
(x
c
+ts) −F
(x
∗
))s dt.
(7.32)
We begin by writing (7.32) as
F(x
+
) = −E
c
s +
_
1
0
(F
(x
c
+ts) −F
(x
∗
))s dt
and then using (7.2) to obtain
E
+
= E
c
(I −P
s
) +
(∆
c
s)s
T
s
2
2
where P
s
is is given by (7.27) and
∆
c
=
_
1
0
(F
(x
c
+ts) −F
(x
∗
)) dt.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
122 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Hence, by the standard assumptions
∆
c

2
≤ γ(e
c

2
+e
+

2
)/2.
Just as in the proof of Lemma 7.2.2 for the linear case
E
+

2
≤ E
c

2
+∆
c

2
and the proof is complete.
Local qlinear convergence. Theorem 7.2.3 states that Broyden’s method
is locally qlinearly convergent. The proof is more complex than similar results
in Chapters 5 and 6. We must explore the relation between the size of E
0

needed to enforce qlinear convergence and the qfactor.
Theorem 7.2.3. Let the standard assumptions hold. Let r ∈ (0, 1) be
given. Then there are δ and δ
B
such that if x
0
∈ B(δ) and E
0

2
< δ
B
the
Broyden sequence for the data (F, x
0
, B
0
) exists and x
n
→ x
∗
qlinearly with
qfactor at most r.
Proof. Let δ and δ
1
be small enough so that the conclusions of Theo
rem 5.4.3 hold. Then, reducing δ and δ
1
further if needed, the qfactor is at
most
K
A
(δ +δ
1
) ≤ r,
where K
A
is from the statement of Theorem 5.4.3.
We will prove the result by choosing δ
B
and reducing δ if necessary so that
E
0

2
≤ δ
B
will imply that E
n

2
≤ δ
1
for all n, which will prove the result.
To do this reduce δ if needed so that
δ
2
=
γ(1 +r)δ
2(1 −r)
< δ
1
(7.33)
and set
δ
B
= δ
1
−δ
2
. (7.34)
We show that E
n

2
≤ δ
1
by induction. Since E
0

2
< δ
B
< δ
1
we may
begin the induction. Now assume that E
k

2
< δ
1
for all 0 ≤ k ≤ n. By
Theorem 7.2.2,
E
n+1

2
≤ E
n

2
+γ(e
n+1

2
+e
n

2
)/2.
Since E
n

2
< δ
1
, e
n+1

2
≤ re
n

2
and therefore
E
n+1

2
≤ E
n

2
+γ(1 +r)e
n

2
/2 ≤ E
n

2
+γ(1 +r)r
n
δ/2.
We can estimate E
n

2
, E
n−1

2
, . . . in the same way to obtain
E
n+1

2
≤ E
0

2
+
(1 +r)γδ
2
n
j=0
r
j
≤ δ
B
+
(1 +r)γδ
2(1 −r)
≤ δ
1
,
which completes the proof.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 123
Veriﬁcation of the Dennis–Mor´e condition. The ﬁnal task is to verify
the Dennis–Mor´e condition.
Theorem 7.2.4. Let the standard assumptions hold. Then there are δ and
δ
B
such that if x
0
∈ B(δ) and E
0

2
< δ
B
the Broyden sequence for the data
(F, x
0
, B
0
) exists and x
n
→ x
∗
qsuperlinearly.
Proof. Let δ and δ
B
be such that the conclusions of Theorem 7.2.3 hold.
By Theorem 7.1.1 we need only show that Dennis–Mor´e condition (7.6) holds
to complete the proof.
As in the proof of Theorem 7.2.1 let
P
n
=
s
n
s
T
n
s
n

2
2
.
Set
∆
n
=
_
1
0
(F
(x
n
+ts
n
) −F
(x
∗
)) dt.
Let φ ∈ R
N
be arbitrary. Note that
E
T
n+1
φ = (I −P
n
)E
T
n
φ +P
n
∆
T
n
φ.
We wish to apply Lemma 7.2.1 with
ψ
n
= E
T
n
φ, η
n
= s
n
/s
n

2
, and
n
= P
n
∆
T
n
φ.
The hypothesis in the lemma that
n

n

2
< ∞
holds since Theorem 7.2.3 and the standard assumptions imply
∆
n

2
≤ γ(1 +r)r
n
δ/2.
Hence Lemma 7.2.1 implies that
η
T
n
ψ
n
=
(E
T
n
φ)
T
s
n
s
n

2
= φ
T
E
n
s
n
s
n

2
→ 0.
This, as in the proof of Theorem 7.2.1, implies the Dennis–Mor´e condition and
completes the proof.
7.3. Implementation of Broyden’s method
The implementation of Broyden’s method that we advocate here is related
to one that has been used in computational ﬂuid mechanics [73]. Some
such methods are called limited memory formulations in the optimization
literature [138], [142], [31]. In these methods, after the storage available for
the iteration is exhausted, the oldest of the stored vectors is replaced by the
most recent. Another approach, restarting, clears the storage and starts over.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
124 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
This latter approach is similar to the restarted GMRES iteration. Our basic
implementation is a nonlinear version of the implementation for linear problems
in [67]. We use a restarted approach, though a limited memory approach could
be used as well.
We begin by showing how the initial approximation to F
(x
∗
), B
0
, may be
regarded as a preconditioner and incorporated into the deﬁntion of F just as
preconditioners were in the implementation of NewtonGMRES in ¸ 6.4.
Lemma 7.3.1. Assume that the Broyden sequence (¦x
n
¦, ¦B
n
¦) for the data
(F, x
0
, B
0
) exists. Then the Broyden sequence (¦y
n
¦, ¦C
n
¦) exists for the data
(B
−1
0
F, x
0
, I) and
x
n
= y
n
, B
n
= B
0
C
n
(7.35)
for all n.
Proof. We proceed by induction. Equation (7.35) holds by deﬁnition when
n = 0. If (7.35) holds for a given n, then
y
n+1
= y
n
−C
−1
n
B
−1
0
F(y
n
)
= x
n
−(B
0
C
n
)
−1
F(x
n
) = x
n
−B
−1
n
F(x
n
)
= x
n+1
.
(7.36)
Now s
n
= x
n+1
−x
n
= y
n+1
−y
n
by (7.36). Hence
B
n+1
= B
n
+
F(x
n+1
)s
T
n
s
n

2
2
= B
0
C
n
+B
0
B
−1
0
F(y
n+1
)s
T
n
s
n

2
2
= B
0
C
n+1
.
This completes the proof.
The consequence of this proof for implementation is that we can assume
that B
0
= I. If a better choice for B
0
exists, we can incorporate that into the
function evaluation. Our plan will be to store B
−1
n
in a compact and easy to
apply form. The basis for this is the following simple formula [68], [176], [177],
[11].
Proposition 7.3.1. Let B be a nonsingular N N matrix and let
u, v ∈ R
N
. Then B + uv
T
is invertible if and only if 1 + v
T
B
−1
u ,= 0. In
this case,
(B +uv
T
)
−1
=
_
I −
(B
−1
u)v
T
1 +v
T
B
−1
u
_
B
−1
. (7.37)
The expression (7.37) is called the Sherman–Morrison formula. We leave
the proof to Exercise 7.5.2.
In the context of a sequence of Broyden updates ¦B
n
¦ we have for n ≥ 0,
B
n+1
= B
n
+u
n
v
T
n
,
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 125
where
u
n
= F(x
n+1
)/s
n

2
and v
n
= s
n
/s
n

2
.
Setting
w
n
= (B
−1
n
u
n
)/(1 +v
T
n
B
−1
n
u
n
)
we see that, if B
0
= I,
B
−1
n
= (I −w
n−1
v
T
n−1
)(I −w
n−2
v
T
n−2
) . . . (I −w
0
v
T
0
)
=
n−1
j=0
(I −w
j
v
T
j
).
(7.38)
Since the empty matrix product is the identity, (7.38) is valid for n ≥ 0.
Hence the action of B
−1
n
on F(x
n
) (i.e., the computation of the Broyden
step) can be computed from the 2n vectors ¦w
j
, v
j
¦
n−1
j=0
at a cost of O(Nn)
ﬂoatingpoint operations. Moreover, the Broyden step for the following
iteration is
s
n
= −B
−1
n
F(x
n
) = −
n−1
j=0
(I −w
j
v
T
j
)F(x
n
). (7.39)
Since the product
n−2
j=0
(I −w
j
v
T
j
)F(x
n
)
must also be computed as part of the computation of w
n−1
we can combine
the computation of w
n−1
and s
n
as follows:
w =
n−2
j=0
(I −w
j
v
T
j
)F(x
n
)
w
n−1
= C
w
w where C
w
= (s
n−1

2
+v
T
n−1
w)
−1
s
n
= −(I −w
n−1
v
T
n−1
)w.
(7.40)
One may carry this one important step further [67] and eliminate the need
to store the sequence ¦w
n
¦. Note that (7.40) implies that for n ≥ 1
s
n
= −w +C
w
w(v
T
n−1
w) = w(−1 +C
w
(C
−1
w
−s
n−1

2
))
= −C
w
ws
n−1

2
= −s
n−1

2
w
n−1
.
Hence, for n ≥ 0,
w
n
= −s
n+1
/s
n

2
. (7.41)
Therefore, one need only store the steps ¦s
n
¦ and their norms to construct
the sequence ¦w
n
¦. In fact, we can write (7.38) as
B
−1
n
=
n−1
j=0
_
I +
s
j+1
s
T
j
s
j

2
2
_
. (7.42)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
126 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
We cannot use (7.42) and (7.39) directly to compute s
n+1
because s
n+1
appears
on both sides of the equation
s
n+1
= −
_
I +
s
n+1
s
T
n
s
n

2
2
_
n−1
j=0
_
I +
s
j+1
s
T
j
s
j

2
2
_
F(x
n+1
)
= −
_
I +
s
n+1
s
T
n
s
n

2
2
_
B
−1
n
F(x
n+1
).
(7.43)
Instead, we solve (7.43) for s
n+1
to obtain
s
n+1
= −
B
−1
n
F(x
n+1
)
1 +s
T
n
B
−1
n
F(x
n+1
)/s
n

2
2
. (7.44)
By Proposition 7.3.1, the denominator in (7.44)
1 +s
T
n
B
−1
n
F(x
n+1
)/s
n

2
2
= 1 +v
T
n
B
−1
n
u
n
is nonzero unless B
n+1
is singular.
The storage requirement, of n +O(1) vectors for the nth iterate, is of the
same order as GMRES, making Broyden’s method competitive in storage with
GMRES as a linear solver [67]. Our implementation is similar to GMRES(m),
restarting with B
0
when storage is exhausted.
We can use (7.42) and (7.44) to describe an algorithm. The termination
criteria are the same as for Algorithm nsol in ¸ 5.6. We use a restarting
approach in the algorithm, adding to the input list a limit nmax on the number
of Broyden iterations taken before a restart and a limit maxit on the number
of nonlinear iterations. Note that B
0
= I is implicit in the algorithm. Our
looping convention is that the loop in step 2(e)ii is skipped if n = 0; this is
consistent with setting the empty matrix product to the identity. Our notation
and algorithmic framework are based on [67].
Algorithm 7.3.1. brsol(x, F, τ, maxit, nmax)
1. r
0
= F(x)
2
, n = −1,
s
0
= −F(x), itc = 0.
2. Do while itc < maxit
(a) n = n + 1; itc = itc + 1
(b) x = x +s
n
(c) Evaluate F(x)
(d) If F(x)
2
≤ τ
r
r
0
+τ
a
exit.
(e) if n < nmax then
i. z = −F(x)
ii. for j = 0, n −1
z = z +s
j+1
s
T
j
z/s
j

2
2
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 127
iii. s
n+1
= z/(1 −s
T
n
z/s
n

2
2
)
(f) if n = nmax then
n = −1; s = −F(x);
If n < nmax, then the nth iteration of brsol, which computes x
n
and
s
n+1
, requires O(nN) ﬂoatingpoint operations and storage of n + 3 vectors,
the steps ¦s
j
¦
n
j=0
, x, z, and F(x), where z and F(x) may occupy the same
storage location. This requirement of one more vector than the algorithm
in [67] needed is a result of the nonlinearity. If n = nmax, the iteration
will restart with x
nmax
, not compute s
nmax+1
, and therefore need storage for
nmax+2 vectors. Our implementation of brsol in the collection of MATLAB
codes stores z and F separately in the interest of clarity and therefore requires
nmax + 3 vectors of storage.
The Sherman–Morrison approach in Algorithm brsol is more eﬃcient, in
terms of both time and storage, than the dense matrix approach proposed
in [85]. However the dense matrix approach makes it possible to detect
ill conditioning in the approximate Jacobians. If the data is suﬃciently
good, bounded deterioration implies that the Broyden matrices will be well
conditioned and superlinear convergence implies that only a few iterates will
be needed. In view of all this, we feel that the approach of Algorithm brsol
is the best approach, especially for large problems.
A MATLAB implementation of brsol is provided in the collection of
MATLAB codes.
7.4. Examples for Broyden’s method
In all the examples in this section we use the l
2
norm multiplied by 1/
√
N.
We use the MATLAB code brsol.
7.4.1. Linear problems. The theory in this chapter indicates that Broy
den’s method can be viewed as an alternative to GMRES as an iterative method
for nonsymmetric linear equations. This point has been explored in some detail
in [67], where more elaborate numerical experiments that those given here are
presented, and in several papers on inﬁnitedimensional problems as well [91],
[104], [120]. As an example we consider the linear PDE (3.33) from ¸ 3.7. We
use the same mesh, righthand side, coeﬃcients, and solution as the example
in ¸ 3.7.
We compare Broyden’s method without restarts (solid line) to Broyden’s
method with restarts every three iterates (dashed line). For linear problems,
we allow for increases in the nonlinear residual. We set τ
a
= τ
r
= h
2
where
h = 1/32. In Fig. 7.1 we present results for (3.33) preconditioned with the
Poisson solver fish2d.
Broyden’s method terminated after 9 iterations at a cost of roughly 1.5
million ﬂoating point operations, a slightly higher cost than the GMRES
solution. The restarted iteration required 24 iterations and 3.5 million ﬂoating
point operations. One can also see highly irregular performance from the
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
128 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
restarted method.
In the linear case [67] one can store only the residuals and recover the
terminal iterate upon exit. Storage of the residual, the vector z, and the
sequence of steps saves a vector over the nonlinear case.
0 5 10 15 20 25
10
3
10
2
10
1
10
0
10
1
Iterations
R
e
l
a
t
i
v
e
R
e
s
i
d
u
a
l
Fig. 7.1. Broyden’s method for the linear PDE.
7.4.2. Nonlinear problems. We consider the Hequation from Chapters 5
and 6 and a nonlinear elliptic partial diﬀerential equation. Broyden’s method
and NewtonGMRES both require more storage as iterations progress, Newton
GMRES as the inner iteration progresses and Broyden’s method as the outer
iteration progresses. The cost in function evaluations for Broyden’s method is
one for each nonlinear iteration and for NewtonGMRES one for each outer
iteration and one for each inner iteration. Hence, if function evaluations are
very expensive, the cost of a solution by Broyden’s method could be much less
than that of one by NewtonGMRES. If storage is at a premium, however,
and the number of nonlinear iterations is large, Broyden’s method could be at
a disadvantage as it might need to be restarted many times. The examples
in this section, as well as those in ¸ 8.4 illustrate the diﬀerences in the two
methods. Our general recommendation is to try both.
Chandrasekhar Hequation. As before we solve the Hequation on a 100
point mesh with c = .9 (Fig. 7.2) and c = .9999 (Fig. 7.3). We compare
Broyden’s method without restarts (solid line) to Broyden’s method with
restarts every three iterates (dashed line). For c = .9 both methods required six
iterations for convergence. The implementation without restarts required 153
thousand ﬂoatingpoint operations and the restarted method 150, a substantial
improvement over the NewtonGMRES implementation.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 129
For c = .9999, the restarted iteration has much more trouble. The un
restarted iteration converges in 10 iterations at a cost of roughly 249 thousand
ﬂoatingpoint operations. The restarted iteration requires 18 iterations and
407 ﬂoatingpoint operations. Even so, both implementations performed bet
ter than NewtonGMRES.
0 1 2 3 4 5 6
10
8
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
Iterations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 7.2. Broyden’s method for the Hequation, c = .9.
0 2 4 6 8 10 12 14 16 18
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
Iterations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 7.3. Broyden’s method for the Hequation, c = .9999.
One should take some care in drawing general conclusions. While Broyden’s
method on this example performed better than NewtonGMRES, the storage
in NewtonGMRES is used in the inner iteration, while that in Broyden in the
nonlinear iteration. If many nonlinear iterations are needed, it may be the case
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
130 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
that Broyden’s method may need restarts while the GMRES iteration for the
linear problem for the Newton step may not.
Convectiondiﬀusion equation. We consider the nonlinear convection
diﬀusion equation from ¸ 6.4.2. We use the same mesh width h = 1/32 on
a uniform square grid. We set C = 20, u
0
= 0, and τ
r
= τ
a
= h
2
. We
preconditioned with the Poisson solver fish2d.
We allowed for an increase in the residual, which happened on the second
iterate. This is not supported by the local convergence theory for nonlinear
problems; however it can be quite successful in practice. One can do this in
brsol by setting the third entry in the parameter list to 1, as one would if the
problem were truly linear. In ¸ 8.4.3 we will return to this example.
In Fig. 7.4 we compare Broyden’s method without restarts (solid line) to
Broyden’s method with restarts every 8 iterates (dashed line). The unrestarted
Broyden iteration converges in 12 iterations at a cost of roughly 2.2 million
ﬂoatingpoint operations. This is a modest improvement over the Newton
GMRES cost (reported in ¸ 6.4.2) of 2.6 million ﬂoatingpoint operations.
However, the residual norms do not monotonically decrease in the Broyden
iteration (we ﬁx this in ¸ 8.3.2) and more storage was used.
0 2 4 6 8 10 12 14 16
10
4
10
3
10
2
10
1
10
0
10
1
Iterations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 7.4. Broyden’s method for nonlinear PDE.
At most ﬁve inner GMRES iterations were required. Therefore the GMRES
inner iteration for the NewtonGMRES approach needed storage for at most
10 vectors (the Krylov basis, s, x
c
, F(x
c
), F(x
c
+ hv)) to accommodate the
NewtonGMRES iteration. The Broyden iteration when restarted every n
iterates, requires storage of n+2 vectors ¦s
j
¦
n−1
j=0
, x, and z (when stored in the
same place as F(x)). Hence restarting the Broyden iteration every 8 iterations
most closely corresponds to the NewtonGMRES requirements. This approach
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 131
took 15 iterations to terminate at a cost of 2.4 million ﬂoatingpoint operations,
but the convergence was extremely irregular after the restart.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
132 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
7.5. Exercises on Broyden’s method
7.5.1. Prove Proposition 7.2.1.
7.5.2. Prove Proposition 7.3.1.
7.5.3. Extend Proposition 7.3.1 by showing that if A is a nonsingular N N
matrix, U is N M, V is M N, then A + UV is nonsingular if and
only if I + V A
−1
U is a nonsingular M M matrix. If this is the case,
then
(A+UV )
−1
= A
−1
−A
−1
U(I +V A
−1
U)
−1
V A
−1
.
This is the Sherman–Morrison–Woodbury formula [68], [199]. See [102]
for further generalizations.
7.5.4. Duplicate the results reported in ¸ 7.4. Vary the parameters in the
equations and the number of vectors stored by Broyden’s method and
report on their eﬀects on performance. What happens in the diﬀerential
equation examples if preconditioning is omitted?
7.5.5. Solve the Hequation with Broyden’s method and c = 1. Set τ
a
= τ
r
=
10
−8
. What qfactor for linear convergence do you see? Can you account
for this by applying the secant method to the equation x
2
= 0? See [53]
for a complete explanation.
7.5.6. Try to solve the nonlinear convectiondiﬀusion equation in ¸ 6.4 with
Broyden’s method using various values for the parameter C. How does
the performance of Broyden’s method diﬀer from NewtonGMRES?
7.5.7. Modify nsol to use Broyden’s method instead of the chord method after
the initial Jacobian evaluation. How does this compare with nsol on the
examples in Chapter 5.
7.5.8. Compare the performance of Broyden’s method and NewtonGMRES on
the Modiﬁed Bratu Problem
−∇
2
u +du
x
+e
u
= 0
on (0, 1) (0, 1) with homogeneous Dirichlet boundary conditions.
Precondition with the Poisson solver fish2d. Experiment with various
mesh widths, initial iterates, and values for the convection coeﬃcient d.
7.5.9. The bad Broyden method [26], so named because of its inferior perfor
mance in practice [63], updates an approximation to the inverse of the
Jacobian at the root so that B
−1
≈ F
(x
∗
)
−1
satisﬁes the inverse secant
equation
B
−1
+
y = s (7.45)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BROYDEN’S METHOD 133
with the rankone update
B
−1
+
= B
−1
c
+
(s −B
−1
c
y)y
T
y
2
2
.
Show that the bad Broyden method is locally superlinearly convergent
if the standard assumptions hold. Try to implement the bad Broyden
method in a storageeﬃcient manner using the ideas from [67] and ¸ 7.3.
Does anything go wrong? See [67] for more about this.
7.5.10. The Schubert or sparse Broyden algorithm [172], [27] is a quasiNewton
update for sparse matrices that enforces both the secant equation and the
sparsity pattern. For problems with tridiagonal Jacobian, for example,
the update is
(B
+
)
ij
= (B
c
)
ij
+
(y −B
c
s)
i
s
j
i+1
k=i−1
s
2
k
for 1 ≤ i, j ≤ N and [i − j[ ≤ 1. Note that only the subdiagonal, su
perdiagonal, and main diagonal are updated. Read about the algorithm
in [172] and prove local superlinear convergence under the standard as
sumptions and the additional assumption that the sparsity pattern of B
0
is the same as that of F
(x
∗
). How would the Schubert algorithm be
aﬀected by preconditioning? Compare your analysis with those in [158],
[125], and [189].
7.5.11. Implement the bad Broyden method, apply it to the examples in ¸ 7.4,
and compare the performance to that of Broyden’s method. Discuss the
diﬀerences in implementation from Broyden’s method.
7.5.12. Implement the Schubert algorithm on an unpreconditioned discretized
partial diﬀerential equation (such as the Bratu problem from Exer
cise 7.5.8) and compare it to NewtonGMRES and Broyden’s method.
Does the relative performance of the methods change as h is decreased?
This is interesting even in one space dimension where all matrices are
tridiagonal. References [101], [93], [92], and [112], are relevant to this
exercise.
7.5.13. Use your result from Exercise 5.7.14 in Chapter 5 to numerically estimate
the qorder of Broyden’s method for some of the examples in this
section (both linear and nonlinear). What can you conclude from your
observations?
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
134 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Chapter 8
Global Convergence
By a globally convergent algorithm we mean an algorithm with the property
that for any initial iterate the iteration either converges to a root of F or fails
to do so in one of a small number of ways. Of the many such algorithms we
will focus on the class of line search methods and the Armijo rule [3], [88],
implemented inexactly [24], [25], [70].
Other methods, such as trust region [153], [154], [181], [129], [63], [24],
[25], [173], [70], and continuation/homotopy methods [1], [2], [109], [159],
[198], can be used to accomplish the same objective. We select the line search
paradigm because of its simplicity, and because it is trivial to add a line search
to an existing locally convergent implementation of Newton’s method. The
implementation and analysis of the line search method we develop in this
chapter do not depend on whether iterative or direct methods are used to
compute the Newton step or upon how or if the Jacobian is computed and
stored. We begin with single equations, where the algorithms can be motivated
and intuition developed with a very simple example.
8.1. Single equations
If we apply Newton’s method to ﬁnd the root x
∗
= 0 of the function
f(x) = arctan(x) with initial iterate x
0
= 10 we ﬁnd that the initial iterate
is too far from the root for the local convergence theory in Chapter 5 to be
applicable. The reason for this is that f
(x) = (1 + x
2
)
−1
is small for large
x; f(x) = arctan(x) ≈ ±π/2, and hence the magnitude of the Newton step
is much larger than that of the iterate itself. We can see this eﬀect in the
sequence of iterates:
10, −138, 2.9 10
4
, −1.5 10
9
, 9.9 10
17
,
a failure of Newton’s method that results from an inaccurate initial iterate.
If we look more closely, we see that the Newton step s = −f(x
c
)/f
(x
c
) is
pointing toward the root in the sense that sx
c
< 0, but the length of s is too
large. This observation motivates the simple ﬁx of reducing the size of the step
until the size of the nonlinear residual is decreased. A prototype algorithm is
given below.
135
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
136 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Algorithm 8.1.1. lines1(x, f, τ)
1. r
0
= [f(x)[
2. Do while [f(x)[ > τ
r
r
0
+τ
a
(a) If f
(x) = 0 terminate with failure.
(b) s = −f(x)/f
(x) (search direction)
(c) x
t
= x +s (trial point)
(d) If [f(x
t
)[ < [f(x)[ then x = x
t
(accept the step)
else
s = s/2 goto 2c (reject the step)
When we apply Algorithm lines1 to our simple example, we obtain the
sequence
10, −8.5, 4.9, −3.8, 1.4, −1.3, 1.2, −.99, .56, −0.1, 9 10
−4
, −6 10
−10
.
The quadratic convergence behavior of Newton’s method is only apparent
very late in the iteration. Iterates 1, 2, 3, and 4 required 3, 3, 2, and 2
steplength reductions. After the fourth iterate, decrease in the size of the
nonlinear residual was obtained with a full Newton step. This is typical of the
performance of the algorithms we discuss in this chapter.
We plot the progress of the iteration in Fig. 8.1. We plot
[arctan(x)/arctan(x
0
)[
as a function of the number of function evaluations with the solid line. The
outer iterations are identiﬁed with circles as in ¸ 6.4. One can see that most of
the work is in the initial phase of the iteration. In the terminal phase, where
the local theory is valid, the minimum number of two function evaluations per
iterate is required. However, even in this terminal phase rapid convergence
takes place only in the ﬁnal three outer iterations.
Note the diﬀerences between Algorithm lines1 and Algorithm nsol. One
does not set x
+
= x +s without testing the step to see if the absolute value of
the nonlinear residual has been decreased. We call this method a line search
because we search along the line segment
(x
c
, x
c
−f(x
c
)/f
(x
c
))
to ﬁnd a decrease in [f(x
c
)[. As we move from the right endpoint to the left,
some authors [63] refer to this as backtracking.
Algorithm lines1 almost always works well. In theory, however, there is
the possibility of oscillation about a solution without convergence [63]. To
remedy this and to make analysis possible we replace the test for simple
decrease with one for suﬃcient decrease. The algorithm is to compute a search
direction d which for us will be the Newton direction
d = −f(x
c
)/f
(x
c
)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GLOBAL CONVERGENCE 137
0 5 10 15 20 25 30 35
10
10
10
9
10
8
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 8.1. Newton–Armijo iteration for inverse tangent.
and then test steps of the form s = λd, with λ = 2
−j
for some j ≥ 0, until
f(x +s) satisﬁes
[f(x
c
+λd)[ < (1 −αλ)[f(x
c
)[. (8.1)
The condition in (8.1) is called suﬃcient decrease of [f[. The parameter
α ∈ (0, 1) is a small, but positive, number intended to make (8.1) as easy
as possible to satisfy. We follow the recent optimization literature [63] and
set α = 10
−4
. Once suﬃcient decrease has been obtained we accept the step
s = λd. This strategy, from [3] (see also [88]) is called the Armijo rule. Note
that we must now distinguish between the step and the Newton direction,
something we did not have to do in the earlier chapters.
Algorithm 8.1.2. nsola1(x, f, τ)
1. r
0
= [f(x)[
2. Do while [f(x)[ > τ
r
r
0
+τ
a
(a) If f
(x) = 0 terminate with failure.
(b) d = −f(x)/f
(x) (search direction)
(c) λ = 1
i. x
t
= x +λd (trial point)
ii. If [f(x
t
)[ < (1 −αλ)[f(x)[ then x = x
t
(accept the step)
else
λ = λ/2 goto 2(c)i (reject the step)
This is essentially the algorithm implemented in the MATLAB code nsola.
We will analyze Algorithm nsola in ¸ 8.2. We remark here that there is
nothing critical about the reduction factor of 2 in the line search. A factor of
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
138 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
10 could well be better in situations in which small values of λ are needed for
several consecutive steps (such as our example with the arctan function). In
that event a reduction of a factor of 8 would require three passes through the
loop in nsola1 but only a single reduction by a factor of 10. This could be
quite important if the search direction were determined by an inexact Newton
method or an approximation to Newton’s method such as the chord method
or Broyden’s method.
On the other hand, reducing λ by too much can be costly as well. Taking
full Newton steps ensures fast local convergence. Taking as large a fraction as
possible helps move the iteration into the terminal phase in which full steps
may be taken and fast convergence expected. We will return to the issue of
reduction in λ in ¸ 8.3.
8.2. Analysis of the Armijo rule
In this section we extend the onedimensional algorithm in two ways before
proceeding with the analysis. First, we accept any direction that satisﬁes the
inexact Newton condition (6.1). We write this for the Armijo sequence as
F
(x
n
)d
n
+F(x
n
) ≤ η
n
F(x
n
). (8.2)
Our second extension is to allow more ﬂexibility in the reduction of λ. We
allow for any choice that produces a reduction that satisﬁes
σ
0
λ
old
≤ λ
new
≤ σ
1
λ
old
,
where 0 < σ
0
< σ
1
< 1. One such method, developed in ¸ 8.3, is to minimize
an interpolating polynomial based on the previous trial steps. The danger is
that the minimum may be too near zero to be of much use and, in fact, the
iteration may stagnate as a result. The parameter σ
0
safeguards against that.
Safeguarding is an important aspect of many globally convergent methods and
has been used for many years in the context of the polynomial interpolation
algorithms we discuss in ¸ 8.3.1 [54], [63], [86], [133], [87]. Typical values of σ
0
and σ
1
are .1 and .5. Our algorithm nsola incorporates these ideas.
While (8.2) is motivated by the Newtoniterative paradigm, the analysis
here applies equally to methods that solve the equations for the Newton step
directly (so η
n
= 0), approximate the Newton step by a quasiNewton method,
or even use the chord method provided that the resulting direction d
n
satisﬁes
(8.2). In Exercise 8.5.9 you are asked to implement the Armijo rule in the
context of Algorithm nsol.
Algorithm 8.2.1. nsola(x, F, τ, η).
1. r
0
= F(x)
2. Do while F(x) > τ
r
r
0
+τ
a
(a) Find d such that F
(x)d +F(x) ≤ ηF(x)
If no such d can be found, terminate with failure.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GLOBAL CONVERGENCE 139
(b) λ = 1
i. x
t
= x +λd
ii. If F(x
t
) < (1 −αλ)F(x) then x = x
t
else
Choose σ ∈ [σ
0
, σ
1
]
λ = σλ
goto 2(b)i
Note that step 2a must allow for the possibility that F
is illconditioned
and that no direction can be found that satisﬁes (8.2). If, for example, step
2a were implemented with a direct factorization method, illconditioning of F
might be detected as part of the factorization.
Let ¦x
n
¦ be the iteration produced by Algorithm nsola with initial iterate
x
0
. The algorithm can fail in some obvious ways:
1. F
(x
n
) is singular for some n. The inner iteration could terminate with
a failure.
2. x
n
→ ¯ x, a local minimum of F which is not a root.
3. x
n
 → ∞.
Clearly if F has no roots, the algorithm must fail. We will see that if F has no
roots then either F
(x
n
) has a limit point which is singular or ¦x
n
¦ becomes
unbounded.
The convergence theorem is the remarkable statement that if ¦x
n
¦ and
¦F
(x
n
)
−1
¦ are bounded (thereby eliminating premature failure and local
minima of F that are not roots) and F
is Lipschitz continuous in a
neighborhood of the entire sequence ¦x
n
¦, then the iteration converges to a
root of F at which the standard assumptions hold, full steps are taken in the
terminal phase, and the convergence is qquadratic.
We begin with a formal statement of our assumption that F
is uniformly
Lipschitz continuous and bounded away from zero on the sequence ¦x
n
¦.
Assumption 8.2.1. There are r, γ, m
f
> 0 such that F is deﬁned, F
is
Lipschitz continuous with Lipschitz constant γ, and F
(x)
−1
 ≤ m
f
on the
set
Ω(¦x
n
¦, r) =
∞
_
n=0
¦x[ x −x
n
 ≤ r¦.
Assumption 8.2.1 implies that the line search phase (step 2b in Algorithm
nsola) will terminate in a number of cycles that is independent of n. We show
this, as do [25] and [70], by giving a uniform lower bound on λ
n
(assuming
that F(x
0
) ,= 0). Note how the safeguard bound σ
0
enters this result. Note
also the very modest conditions on η
n
.
Lemma 8.2.1. Let x
0
∈ R
N
and α ∈ (0, 1) be given. Assume that ¦x
n
¦ is
given by Algorithm nsola with
¦η
n
¦ ⊂ (0, ¯ η] ⊂ (0, 1 −α) (8.3)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
140 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
and that Assumption 8.2.1 holds. Then either F(x
0
) = 0 or
λ
n
≥
¯
λ = σ
0
min
_
r
m
f
F(x
0
)
,
2(1 −α − ¯ η)
m
2
f
F(x
0
)(1 + ¯ η)
2
γ
_
. (8.4)
Proof. Let n ≥ 0. By the fundamental theorem of calculus and (8.2) we
have, for any λ ∈ [0, 1]
F(x
n
+λd
n
) = F(x
n
) +λF
(x
n
)d
n
+
_
1
0
(F
(x
n
+tλd
n
) −F
(x
n
))λd
n
dt
= (1 −λ)F(x
n
) +λξ
n
+
_
1
0
(F
(x
n
+tλd
n
) −F
(x
n
))λd
n
dt,
where, by the bound η
n
≤ ¯ η,
ξ
n
 ≤ η
n
F(x
n
) ≤ ¯ ηF(x
n
).
The step acceptance condition in nsola implies that ¦F(x
n
)¦ is a decreasing
sequence and therefore
d
n
 = F
(x
n
)
−1
(ξ
n
−F(x
n
)) ≤ m
f
(1 + ¯ η)F(x
n
).
Therefore, by Assumption 8.2.1 F
is Lipschitz continuous on the line segment
[x
n
, x
n
+λd
n
] whenever
λ ≤
¯
λ
1
= r/(m
f
F(x
0
)).
Lipschitz continuity of F
implies that if λ ≤
¯
λ
1
then
F(x
n
+λd
n
) ≤ (1 −λ)F(x
n
) +λ¯ ηF(x
n
) +
γ
2
λ
2
d
n

2
≤ (1 −λ)F(x
n
)[ +λ¯ ηF(x
n
) +
m
2
f
γ(1 + ¯ η)
2
λ
2
F(x
n
)
2
2
≤ (1 −λ + ¯ ηλ)F(x
n
) +λF(x
n
)
m
2
f
γλ(1 + ¯ η)
2
F(x
0
)
2
< (1 −αλ)F(x
n
),
which will be implied by
λ ≤
¯
λ
2
= min
_
¯
λ
1
,
2(1 −α − ¯ η)
(1 + ¯ η)
2
m
2
f
γF(x
0
)
_
.
This shows that λ can be no smaller than σ
0
¯
λ
2
, which completes the proof.
Now we can show directly that the sequence of Armijo iterates either is
unbounded or converges to a solution. Theorem 8.2.1 says even more. The
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GLOBAL CONVERGENCE 141
sequence converges to a root at which the standard assumptions hold, so in the
terminal phase of the iteration the step lengths are all 1 and the convergence
is qquadratic.
Theorem 8.2.1. Let x
0
∈ R
N
and α ∈ (0, 1) be given. Assume that ¦x
n
¦
is given by Algorithm nsola, ¦η
n
¦ satisﬁes (8.3), ¦x
n
¦ is bounded, and that
Assumption 8.2.1 holds. Then ¦x
n
¦ converges to a root x
∗
of F at which the
standard assumptions hold, full steps are taken for n suﬃciently large, and the
convergence behavior in the ﬁnal phase of the iteration is that given by the local
theory for inexact Newton methods (Theorem 6.1.2).
Proof. If F(x
n
) = 0 for some n, then we are done because Assump
tion 8.2.1 implies that the standard assumptions hold at x
∗
= x
n
. Otherwise
Lemma 8.2.1 implies that F(x
n
) converges qlinearly to zero with qfactor at
most (1 −α
¯
λ).
The boundedness of the sequence ¦x
n
¦ implies that a subsequence ¦x
n
k
¦
converges, say to x
∗
. Since F is continuous, F(x
∗
) = 0. Eventually
[x
n
k
− x
∗
[ < r, where r is the radius in Assumption 8.2.1 and therefore the
standard assumptions hold at x
∗
.
Since the standard assumptions hold at x
∗
, there is δ such that if x ∈ B(δ),
the Newton iteration with x
0
= x will remain in B(δ) and converge q
quadratically to x
∗
. Hence as soon as x
n
k
∈ B(δ), the entire sequence, not
just the subsequence, will remain in B(δ) and converge to x
∗
. Moreover,
Theorem 6.1.2 applies and hence full steps can be taken.
At this stage we have shown that if the Armijo iteration fails to converge
to a root either the continuity properties of F or nonsingularity of F
break
down as the iteration progresses (Assumption 8.2.1 is violated) or the iteration
becomes unbounded. This is all that one could ask for in terms of robustness
of such an algorithm.
Thus far we have used for d the inexact Newton direction. It could well be
advantageous to use a chord or Broyden direction, for example. All that one
needs to make the theorems and proofs in ¸ 8.2 hold is (8.2), which is easy to
verify, in principle, via a forward diﬀerence approximation to F
(x
n
)d.
8.3. Implementation of the Armijo rule
The MATLAB code nsola is a modiﬁcation of nsolgm which provides for a
choice of several Krylov methods for computing an inexact Newton direction
and globalizes Newton’s method with the Armijo rule. We use the l
2
norm in
that code and in the numerical results reported in ¸ 8.4. If GMRES is used
as the linear solver, then the storage requirements are roughly the same as for
Algorithm nsolgm if one reuses the storage location for the ﬁnite diﬀerence
directional derivative to test for suﬃcient decrease.
In Exercise 8.5.9 you are asked to do this with nsol and, with each
iteration, numerically determine if the chord direction satisﬁes the inexact
Newton condition. In this case the storage requirements are roughly the same
as for Algorithm nsol. The iteration can be very costly if the Jacobian must be
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
142 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
evaluated and factored many times during the phase of the iteration in which
the approximate solution is far from the root.
We consider two topics in this section. The ﬁrst, polynomial interpolation
as a way to choose λ, is applicable to all methods. The second, how Broyden’s
method must be implemented, is again based on [67] and is more specialized.
Our Broyden–Armijo code does not explicitly check the inexact Newton
condition and thereby saves an evaluation of F. Evaluation of F
(x
c
)d
c
if the
full step is not accepted (here d
c
is the Broyden direction) would not only verify
the inexact Newton condition but provide enough information for a twopoint
parabolic line search. The reader is asked to pursue this in Exercise 8.5.10.
8.3.1. Polynomial line searches. In this section we consider an elabora
tion of the simple line search scheme of reduction of the steplength by a ﬁxed
factor. The motivation for this is that some problems respond well to one or
two reductions in the steplength by modest amounts (such as 1/2) and others
require many such reductions, but might respond well to a more aggressive
steplength reduction (by factors of 1/10, say). If one can model the decrease
in F
2
as the steplength is reduced, one might expect to be able to better
estimate the appropriate reduction factor. In practice such methods usually
perform better than constant reduction factors.
If we have rejected k steps we have in hand the values
F(x
n
)
2
, F(x
n
+λ
1
d
n
)
2
, . . . F(x
n
+λ
k−1
d
n
)
2
.
We can use this iteration history to model the scalar function
f(λ) = F(x
n
+λd
n
)
2
2
with a polynomial and use the minimum of that polynomial as the next
steplength. We consider two ways of doing this that use second degree
polynomials which we compute using previously computed information.
After λ
c
has been rejected and a model polynomial computed, we compute
the minimum λ
t
of that polynomial analytically and set
λ
+
=
_
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
_
σ
0
λ
c
if λ
t
< σ
0
λ
c
,
σ
1
λ
c
if λ
t
> σ
1
λ
c
,
λ
t
otherwise.
(8.5)
Twopoint parabolic model. In this approach we use values of f and f
at λ = 0 and the value of f at the current value of λ to construct a 2nd degree
interpolating polynomial for f.
f(0) = F(x
n
)
2
2
is known. Clearly f
(0) can be computed as
f
(0) = 2(F
(x
n
)
T
d
n
)
T
F(x
n
) = 2F(x
n
)
T
(F
(x
n
)d
n
). (8.6)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GLOBAL CONVERGENCE 143
Use of (8.6) requires evaluation of F
(x
n
)d
n
, which can be obtained by
examination of the ﬁnal residual from GMRES or by expending an additional
function evaluation to compute F
(x
n
)d
n
with a forward diﬀerence. In any
case, our approximation to f
(0) should be negative. If it is not, it may
be necessary to compute a new search direction. This is a possibility with
directions other than inexact Newton directions, such as the Broyden direction.
The polynomial p(λ) such that p, p
agree with f, f
at 0 and p(λ
c
) = f(λ
c
)
is
p(λ) = f(0) +f
(0)λ +cλ
2
,
where
c =
f(λ
c
) −f(0) −f
(0)λ
c
λ
2
c
.
Our approximation of f
(0) < 0, so if f(λ
c
) > f(0), then c > 0 and p(λ) has a
minimum at
λ
t
= −f
(0)/(2c) > 0.
We then compute λ
+
with (8.5).
Threepoint parabolic model. An alternative to the twopoint model that
avoids the need to approximate f
(0) is a threepoint model, which uses f(0)
and the two most recently rejected steps to create the parabola. The MATLAB
code parab3p implements the threepoint parabolic line search and is called
by the two codes nsola and brsola.
In this approach one evaluates f(0) and f(1) as before. If the full step is
rejected, we set λ = σ
1
and try again. After the second step is rejected, we
have the values
f(0), f(λ
c
), and f(λ
−
),
where λ
c
and λ
−
are the most recently rejected values of λ. The polynomial
that interpolates f at 0, λ
c
, λ
−
is
p(λ) = f(0) +
λ
λ
c
−λ
−
_
(λ −λ
−
)(f(λ
c
) −f(0))
λ
c
+
(λ
c
−λ)(f(λ
−
) −f(0))
λ
−
_
.
We must consider two situations. If
p
(0) =
2
λ
c
λ
−
(λ
c
−λ
−
)
(λ
−
(f(λ
c
) −f(0)) −λ
c
(f(λ
−
) −f(0)))
is positive, then we set λ
t
to the minimum of p
λ
t
= −p
(0)/p
(0)
and apply the safeguarding step (8.5) to compute λ
+
. If p
(0) ≤ 0 one could
either set λ
+
to be the minimum of p on the interval [σ
0
λ, σ
1
λ] or reject the
parabolic model and simply set λ
+
to σ
0
λ
c
or σ
1
λ
c
. We take the latter approach
and set λ
+
= σ
1
λ
c
. In the MATLAB code nsola from the collection, we
implement this method with σ
0
= .1, σ
1
= .5.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
144 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Interpolation with a cubic polynomial has also been suggested [87], [86],
[63], [133]. Clearly, the polynomial modeling approach is heuristic. The
safeguarding and Theorem 8.2.1 provide insurance that progress will be made
in reducing F.
8.3.2. Broyden’s method. In Chapter 7 we implemented Broyden’s
method at a storage cost of one vector for each iteration. The relation
y − B
c
s = F(x
+
) was not critical to this and we may also incorporate a line
search at a cost of a bit of complexity in the implementation. As in ¸ 7.3, our
approach is a nonlinear version of the method in [67].
The diﬀerence from the development in ¸ 7.3 is that the simple relation
between the sequence ¦w
n
¦ and ¦v
n
¦ is changed. If a line search is used, then
s
n
= −λ
n
B
−1
n
F(x
n
)
and hence
y
n
−B
n
s
n
= F(x
n+1
) −(1 −λ
n
)F(x
n
). (8.7)
If, using the notation in ¸ 7.3, we set
u
n
=
y
n
−B
n
s
n
s
n

2
, v
n
=
s
n
s
n

2
, and w
n
= (B
−1
n
u
n
)/(1 +v
T
n
B
−1
n
u
n
),
we can use (8.7) and (7.38) to obtain
w
n
= λ
n
_
−d
n+1
s
n

2
+ (λ
−1
n
−1)
s
n
s
n

2
_
= λ
n
_
−s
n+1
/λ
n+1
s
n

2
+ (λ
−1
n
−1)
s
n
s
n

2
_
,
(8.8)
where
d
n+1
= −B
−1
n+1
F(x
n+1
)
is the search direction used to compute x
n+1
. Note that (8.8) reduces to (7.41)
when λ
n
= 1 and λ
n+1
= 1 (so d
n+1
= s
n+1
).
We can use the ﬁrst equality in (8.8) and the relation
d
n+1
= −
_
I −
w
n
s
T
n
s
n

2
_
B
−1
n
F(x
n+1
)
to solve for d
n+1
and obtain an analog of (7.44)
d
n+1
= −
s
n

2
2
B
−1
n
F(x
n+1
) −(1 −λ
n
)s
T
n
B
−1
n
F(x
n+1
)s
n
s
n

2
2
+λ
n
s
T
n
B
−1
n
F(x
n+1
)
(8.9)
We then use the second equality in (8.8) to construct B
−1
c
F(x
+
) as we did in
Algorithm brsol. Veriﬁcation of (8.8) and (8.9) is left to Exercise 8.5.5.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GLOBAL CONVERGENCE 145
Algorthm brsola uses these ideas. We do not include details of the line
search or provide for restarts or a limited memory implementation. The
MATLAB implementation, provided in the collection of MATLAB codes, uses
a threepoint parabolic line search and performs a restart when storage is
exhausted like Algorithm brsol does.
It is possible that the Broyden direction fails to satisfy (8.2). In fact, there
is no guarantee that the Broyden direction is a direction of decrease for F
2
.
The MATLAB code returns with an error message if more than a ﬁxed number
of reductions in the step length are needed. Another approach would be to
compute F
(x
c
)d numerically if a full step is rejected and then test (8.2) before
beginning the line search. As a side eﬀect, an approximation of f
(0) can be
obtained at no additional cost and a parabolic line search based on f(0), f
(0),
and f(1) can be used. In Exercise 8.5.10 the reader is asked to fully develop
and implement these ideas.
Algorithm 8.3.1. brsola(x, F, τ, maxit, nmax)
1. Initialize:
F
c
= F(x) r
0
= F(x)
2
, n = −1,
d = −F(x), compute λ
0
, s
0
= λ
0
d
2. Do while n ≤ maxit
(a) n = n + 1
(b) x = x +s
n
(c) Evaluate F(x)
(d) If F(x)
2
≤ τ
r
r
0
+τ
a
exit.
(e) i. z = −F(x)
ii. for j = 0, n −1
a = −λ
j
/λ
j+1
, b = 1 −λ
j
z = z + (as
j+1
+bs
j
)s
T
j
z/s
j

2
2
(f) d = (s
n

2
2
z + (1 −λ
n
)s
T
n
zs
n
)/(s
n

2
2
−λ
n
s
T
n
z)
(g) Compute λ
n+1
with a line search.
(h) Set s
n+1
= λ
n+1
d, store s
n+1

2
and λ
n+1
.
Since d
n+1
and s
n+1
can occupy the same location, the storage require
ments of Algorithm brsola would appear to be essentially the same as those
of brsol. However, computation of the step length requires storage of both the
current point x and the trial point x + λs before a step can be accepted. So,
the storage of the globalized methods exceeds that of the local method by one
vector. The MATLAB implementation of brsola in the collection illustrates
this point.
8.4. Examples for Newton–Armijo
The MATLAB code nsola is a modiﬁcation of nsolgm that incorporates the
threepoint parabolic line search from ¸ 8.3.1 and also changes η using (6.20)
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
146 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
once the iteration is near the root. We compare this strategy, using GMRES
as the linear solver, with the constant η method as we did in ¸ 6.4 with γ = .9.
As before, in all the ﬁgures we plot F(x
n
)
2
/F(x
0
)
2
against the number
of function evaluations required by all line searches and inner and outer
iterations to compute x
n
. Counts of function evaluations corresponding to
outer iterations are indicated by circles. We base our absolute error criterion
on the norm  
2
/
√
N as we did in ¸ 6.4.2 and 7.4.
In ¸ 8.4.3 we compare the Broyden–Armijo method with NewtonGMRES
for those problems for which both methods were successful. We caution the
reader now, and will repeat the caution later, that if the initial iterate is far
from the solution, an inexact Newton method such as NewtonGMRES can
succeed in many case in which a quasiNewton method can fail because the
quasiNewton direction may not be an inexact Newton direction. However,
when both methods succeed, the quasiNewton method, which requires a single
function evaluation for each outer iteration when full steps can be taken, may
well be most eﬃcient.
8.4.1. Inverse tangent function. Since we have easy access to analytic
derivatives in this example, we can use the twopoint parabolic line search.
We compare the twopoint parabolic line search with the constant reduction
search (σ
0
= σ
1
= .5) for the arctan function. In Fig. 8.2 we plot the iteration
history corresponding to the parabolic line search with the solid line and that
for the constant reduction with the dashed line. We use x
0
= 10 as the initial
iterate with τ
r
= τ
a
= 10
−8
. The parabolic line search required 7 outer iterates
and 21 function evaluations in contrast with the constant reduction searches 11
outer iterates and 33 function evaluations. In the ﬁrst outer iteration, both line
searches take three steplength reductions. However, the parabolic line search
takes only one reduction in the next three iterations and none thereafter. The
constant reduction line search took three reductions in the ﬁrst two outer
iterations and two each in following two.
8.4.2. Convectiondiﬀusion equation. We consider a much more diﬃ
cult problem. We take (6.21) from ¸ 6.4.2,
−∇
2
u +Cu(u
x
+u
y
) = f
with the same right hand side f, initial iterate u
0
= 0, 31 31 mesh and
homogeneous Dirichlet boundary conditions on the unit square (0, 1) (0, 1)
that we considered before. Here, however, we set C = 100. This makes a
globalization strategy critical (Try it without one!). We set τ
r
= τ
a
= h
2
/10.
This is a tighter tolerance that in ¸ 6.4.2 and, because of the large value of C
and resulting illconditioning, is needed to obtain an accurate solution.
In Fig. 8.3 we show the progress of the iteration for the unpreconditioned
equation. For this problem we plot the progress of the iteration using η = .25
with the solid line and using the sequence given by (6.20) with the dashed
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GLOBAL CONVERGENCE 147
0 5 10 15 20 25 30 35
10
12
10
10
10
8
10
6
10
4
10
2
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 8.2. Newton–Armijo for the arctan function.
line. We set the parameters in (6.20) to γ = .9 and η
max
= .25. This problem
is very poorly conditioned and much tighter control on the inner iteration is
needed than for the other problems. The two approaches to selection of the
forcing term performed very similarly. The iterations required 75.3 (constant
η) and 75.4 (varying η) million ﬂoating point operations, 759 (constant η) and
744 (varying η) function evaluations, and 25 (constant η) and 22 (varying η)
outer iterations.
0 100 200 300 400 500 600 700 800
10
5
10
4
10
3
10
2
10
1
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 8.3. Newton–Armijo for the PDE, C = 100.
We consider the preconditioned problem in Fig. 8.4. In the computation we
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
148 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
preconditioned (6.21) with the fast Poisson solver fish2d. In Fig. 8.4 we show
the progress of the iteration for the preconditioned equation. For this problem
we plot the progress of the iteration using η = .25 with the solid line and using
the sequence given by (6.20) with the dashed line. We set the parameters in
(6.20) to γ = .9 and η
max
= .99.
The constant η iteration terminated after 79 function evaluations, 9 outer
iterations, and roughly 13.5 million ﬂoatingpoint operations. The line search
in this case reduced the step length three times on the ﬁrst iteration and twice
on the second and third.
The iteration in which ¦η
n
¦ is given by (6.20) terminated after 70 function
evaluations, 9 outer iterations, and 12.3 million ﬂoatingpoint operations. The
line search in the nonconstant η case reduced the step length three times on
the ﬁrst iteration and once on the second. The most eﬃcient iteration, with
the forcing term given by (6.20), required at most 16 inner iterations while the
constant η approach needed at most 10.
0 10 20 30 40 50 60 70 80
10
5
10
4
10
3
10
2
10
1
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 8.4. Newton–Armijo for the PDE, C = 100.
8.4.3. Broyden–Armijo. We found that the Broyden–Armijo line search
failed on all the unpreconditioned convection diﬀusion equation examples. In
these cases the Jacobian is not well approximated by a low rank perturbation
of the identity [112] and the ability of the inexact Newton iteration to ﬁnd a
good search direction was the important factor.
We begin by revisiting the example in ¸ 6.4.2 with C = 20. As reported
in ¸ 6.4.2, the NewtonGMRES iteration always took full steps, terminating
in 4 outer iterations, 16 function evaluations, and 2.6 million ﬂoatingpoint
operations. When compared to the Broyden’s method results in ¸ 6.4.2,
when increases in the residual were allowed, the Broyden–Armijo costs reﬂect
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GLOBAL CONVERGENCE 149
an improvement in eﬃciency. The Broyden iteration in ¸ 6.4.2 took 12
iterations and 2.8 million ﬂoatingpoint operations while the Broyden–Armijo
took 9 iterations and required 2 million ﬂoatingpoint operations. We set
τ
a
= τ
r
= h
2
, η
max
= .5, and γ = .9 as we did in ¸ 6.4.2.
In our implementation of brsola the threepoint parabolic line search was
used. The steplength was reduced on the second iteration (twice) and the third
(once). Figure 8.5 compares the Broyden–Armijo iteration (solid line) to the
GMRES iteration (dashed line).
0 2 4 6 8 10 12 14 16
10
3
10
2
10
1
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 8.5. NewtonGMRES and Broyden–Armijo for the PDE, C = 20.
For the more diﬃcult problem with C = 100, the performance of the
two methods is more similar. In Fig. 8.6 we compare our implementation
of Newton–GMRES–Armijo (dashed line) to Broyden–Armijo (solid line). We
set τ
a
= τ
r
= h
2
/10, η
max
= .99, and γ = .9. The Broyden–Armijo iteration
required 34 nonlinear iterations, roughly 16 million ﬂoatingpoint operations,
and 85 function evaluations. In terms of storage, the Broyden–Armijo iteration
required storage for 37 vectors while the NewtonGMRES iteration needed at
most 16 inner iterations, needing to store 21 vectors, and therefore was much
more eﬃcient in terms of storage.
Restarting the Broyden iteration every 19 iterations would equalize the
storage requirements with Newton–GMRES–Armijo. Broyden’s method suf
fers under these conditions as one can see from the comparison of Broyden–
Armijo (solid line) and Newton–GMRES–Armijo (dashed line) in Fig. 8.7. The
restarted Broyden–Armijo iteration took 19.5 million ﬂoatingpoint operations,
42 outer iterates, and 123 function evaluations.
From these examples we see that Broyden–Armijo can be very eﬃcient pro
vided only a small number of iterations are needed. The storage requirements
increase as the nonlinear iteration progresses and this fact puts Broyden’s
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
150 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
0 10 20 30 40 50 60 70 80 90
10
5
10
4
10
3
10
2
10
1
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 8.6. NewtonGMRES and Broyden–Armijo for the PDE, C = 100.
0 20 40 60 80 100 120 140
10
5
10
4
10
3
10
2
10
1
10
0
Function Evaluations
R
e
l
a
t
i
v
e
N
o
n
l
i
n
e
a
r
R
e
s
i
d
u
a
l
Fig. 8.7. NewtonGMRES and restarted Broyden–Armijo for the PDE, C = 100.
method at a disadvantage to NewtonGMRES when the number of nonlinear
iterations is large.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
GLOBAL CONVERGENCE 151
8.5. Exercises on global convergence
8.5.1. How does the method proposed in [10] diﬀer from the one implemented in
nsola? What could be advantages and disadvantages of that approach?
8.5.2. In what ways are the results in [25] and [70] more general than those in
this section? What ideas do these papers share with the analysis in this
section and with [3] and [88]?
8.5.3. Implement the Newton–Armijo method for single equations. Apply your
code to the following functions with x
0
= 10. Explain your results.
1. f(x) = arctan(x) (i.e., Duplicate the results in ¸ 8.1.)
2. f(x) = arctan(x
2
)
3. f(x) = .9 + arctan(x)
4. f(x) = x(1 + sin(1/x))
5. f(x) = e
x
6. f(x) = 2 + sin(x)
7. f(x) = 1 +x
2
8.5.4. A numerical analyst buys a German sports car for $50,000. He puts
$10,000 down and takes a 7 year installment loan to pay the balance. If
the monthly payments are $713.40, what is the interest rate? Assume
monthly compounding.
8.5.5. Verify (8.8) and (8.9).
8.5.6. Show that if F
is Lipschitz continuous and the iteration ¦x
n
¦ produced
by Algorithm nsola converges to x
∗
with F(x
∗
) ,= 0, then F
(x
∗
) is
singular.
8.5.7. Use nsola to duplicate the results in ¸ 8.4.2. Vary the convection
coeﬃcient in the convectiondiﬀusion equation and the mesh size and
report the results.
8.5.8. Experiment with other linear solvers such as GMRES(m), BiCGSTAB,
and TFQMR. This is interesting in the locally convergent case as well.
You might use the MATLAB code nsola to do this.
8.5.9. Modify nsol, the hybrid Newton algorithm from Chapter 5, to use the
Armijo rule. Try to do it in such a way that the chord direction is used
whenever possible.
8.5.10. Modify brsola to test the Broyden direction for the descent property
and use a twopoint parabolic line search. What could you do if the
Broyden direction is not a descent direction? Apply your code to the
examples in ¸ 8.4.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
152 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
8.5.11. Modify nsola to use a cubic polynomial and constant reduction line
searches instead of the quadratic polynomial line search. Compare the
results with the examples in ¸ 8.4.
8.5.12. Does the secant method for equations in one variable always give a
direction that satisﬁes (8.2) with η
n
bounded away from 1? If not, when
does it? How would you implement a secantArmijo method in such a
way that the convergence theorem 8.2.1 is applicable?
8.5.13. Under what conditions will the iteration given by nsola converge to a
root x
∗
that is independent of the initial iterate?
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Bibliography
[1] E. L. Allgower and K. Georg, Simplicial and continuation methods for
approximating ﬁxed points and solutions to systems of equations, SIAM Rev., 22
(1980), pp. 28–85.
[2] , Numerical Continuation Methods : An Introduction, SpringerVerlag, New
York, 1990.
[3] L. Armijo, Minimization of functions having Lipschitzcontinuous ﬁrst partial
derivatives, Paciﬁc J. Math., 16 (1966), pp. 1–3.
[4] W. E. Arnoldi, The principle of minimized iterations in the solution of the
matrix eigenvalue problem, Quart. Appl. Math., 9 (1951), pp. 17–29.
[5] S. F. Ashby, T. A. Manteuffel, and J. S. Otto, A comparison of
adaptive Chebyshev and least squares polynomial preconditioning for Hermetian
positive deﬁnite linear systems, SIAM J. Sci. Statist. Comput., 13 (1992), pp. 1–
29.
[6] K. E. Atkinson, Iterative variants of the Nystr¨ om method for the numerical
solution of integral equations, Numer. Math., 22 (1973), pp. 17–31.
[7] , An Introduction to Numerical Analysis, 2nd. ed., John Wiley, New York,
1989.
[8] O. Axelsson, Iterative Solution Methods, Cambridge University Press, Cam
bridge, 1994.
[9] S. Banach, Sur les op´erations dans les ensembles abstraits et leur applications
aux ´equations int´egrales, Fund. Math, 3 (1922), pp. 133–181.
[10] R. E. Bank and D. J. Rose, Global approximate Newton methods, Numer.
Math., 37 (1981), pp. 279–295.
[11] M. S. Barlett, An inverse matrix adjustment arising in discriminant analysis,
Ann. Math. Statist., 22 (1951), pp. 107–111.
[12] R. Barrett, M. Berry, T. F. Chan, J. Demmel, J. Donato, J. Don
garra, V. Eijkhout, R. Pozo, C. Romine, and H. van der Vorst, Templates
for the Solution of Linear Systems: Building Blocks for Iterative Methods, Society
for Industrial and Applied Mathematics, Philadelphia, PA, 1993.
[13] N. Bi´ cani´ c and K. H. Johnson, Who was ‘Raphson’?, Internat. J. Numer.
Methods. Engrg., 14 (1979), pp. 148–152.
[14] P. B. Bosma and W. A. DeRooij, Eﬃcient methods to calculate Chan
drasekhar’s Hfunctions, Astron. Astrophys., 126 (1983), p. 283.
[15] H. Brakhage,
¨
Uber die numerische Behandlung von Integralgleichungen nach
der Quadraturformelmethode, Numer. Math., 2 (1960), pp. 183–196.
[16] K. E. Brenan, S. L. Campbell, and L. R. Petzold, Numerical Solution
153
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
154 BIBLIOGRAPHY
of Initial Value Problems in DiﬀerentialAlgebraic Equations, no. 14 in Classics in
Applied Mathematics, SIAM, Philadelphia, 1996.
[17] R. Brent, Algorithms for Minimization Without Deriviatives, PrenticeHall,
Englewood Cliﬀs, NJ, 1973.
[18] , Some eﬃcient algorithms for solving systems of nonlinear equations, SIAM
J. Numer. Anal., 10 (1973), pp. 327–344.
[19] W. Briggs, A Multigrid Tutorial, Society for Industrial and Applied Mathemat
ics, Philadelphia, PA, 1987.
[20] P. N. Brown, A local convergence theory for combined inexact–Newton/ ﬁnite–
diﬀerence projection methods, SIAM J. Numer. Anal., 24 (1987), pp. 407–434.
[21] P. N. Brown, G. D. Byrne, and A. C. Hindmarsh, VODE: A variable
coeﬃcient ode solver, SIAM J. Sci. Statist. Comput., 10 (1989), pp. 1038–1051.
[22] P. N. Brown and A. C. Hindmarsh, Reduced storage matrix methods in stiﬀ
ODE systems, J. Appl. Math. Comput., 31 (1989), pp. 40–91.
[23] P. N. Brown, A. C. Hindmarsh, and L. R. Petzold, Using Krylov methods
in the solution of largescale diﬀerentialalgebraic systems, SIAM J. Sci. Comput., 15
(1994), pp. 1467–1488.
[24] P. N. Brown and Y. Saad, Hybrid Krylov methods for nonlinear systems of
equations, SIAM J. Sci. Statist. Comput., 11 (1990), pp. 450–481.
[25] , Convergence theory of nonlinear NewtonKrylov algorithms, SIAM J.
Optim., 4 (1994), pp. 297–330.
[26] C. G. Broyden, A class of methods for solving nonlinear simultaneous equa
tions, Math. Comput., 19 (1965), pp. 577–593.
[27] , The convergence of an algorithm for solving sparse nonlinear systems,
Math. Comput., 25 (1971), pp. 285–294.
[28] C. G. Broyden, J. E. Dennis, and J. J. Mor´ e, On the local and superlinear
convergence of quasiNewton methods, J. Inst. Maths. Applic., 12 (1973), pp. 223–
246.
[29] W. Burmeister, Zur Konvergenz einiger verfahren der konjugierten Richtun
gen, in Proceedings of Internationaler Kongreß ¨ uber Anwendung der Mathematik in
dem Ingenieurwissenschaften, Weimar, 1975.
[30] I. W. Busbridge, The Mathematics of Radiative Transfer, Cambridge Tracts,
No. 50, Cambridge Univ. Press, Cambridge, 1960.
[31] R. H. Byrd, J. Nocedal, and R. B. Schnabel, Representation of quasi
Newton matrices and their use in limited memory methods, Math. Programming, 63
(1994), pp. 129–156.
[32] X.C. Cai, W. D. Gropp, D. E. Keyes, and M. D. Tidriri, NewtonKrylov
Schwarz methods in CFD, in Proceedings of the International Workshop on the
NavierStokes Equations, R. Rannacher, ed., Notes in Numerical Fluid Mechanics,
Braunschweig, 1994, Vieweg Verlag.
[33] S. L. Campbell, I. C. F. Ipsen, C. T. Kelley, and C. D. Meyer, GMRES
and the minimal polynomial, Tech. Report CRSCTR9410, North Carolina State
University, Center for Research in Scientiﬁc Computation, July 1994. BIT, to appear.
[34] S. L. Campbell, I. C. F. Ipsen, C. T. Kelley, C. D. Meyer, and Z. Q.
Xue, Convergence estimates for solution of integral equations with GMRES, Tech.
Report CRSCTR9513, North Carolina State University, Center for Research in
Scientiﬁc Computation, March 1995. Journal of Integral Equations and Applications,
to appear.
[35] R. Cavanaugh, Diﬀerence Equations and Iterative Processes, PhD thesis,
University of Maryland, 1970.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BIBLIOGRAPHY 155
[36] F. ChaitinChatelin, Is nonnormality a serious diﬃculty ?, Tech. Report
TR/PA/94/18, CERFACS, December 1994.
[37] T. Chan, E. Gallopoulos, V. Simoncini, T. Szeto, and C. Tong, A quasi
minimal residual variant of the BiCGSTAB algorithm for nonsymmetric systems,
SIAM J. Sci. Comput., 15 (1994), p. 338.
[38] T. Chan, R. Glowinski, J. P´ eriaux, and O. Widlund, eds., Domain
Decomposition Methods,Proceedings of the Second International Symposium on
Domain Decomposition Methods, Los Angeles, CA, January 14–16, 1988; Society
for Industrial and Applied Mathematics, Philadelphia, PA, 1989.
[39] , eds., Domain Decomposition Methods, Proceedings of the Third Interna
tional Symposium on Domain Decomposition Methods, Houston, TX, 1989; Society
for Industrial and Applied Mathematics, Philadelphia, PA, 1990.
[40] , eds., Domain Decomposition Methods, Proceedings of the Fourth Interna
tional Symposium on Domain Decomposition Methods, Moscow, USSR, 1990; Soci
ety for Industrial and Applied Mathematics, Philadelphia, PA, 1991.,
[41] S. Chandrasekhar, Radiative Transfer, Dover, New York, 1960.
[42] T. F. Coleman and J. J. Mor´ e, Estimation of sparse Jacobian matrices and
graph coloring problems, SIAM J. Numer. Anal., 20 (1983), pp. 187–209.
[43] T. F. Coleman and C. VanLoan, Handbook for Matrix Computations,
Frontiers in Appl. Math., No. 4, Society for Industrial and Applied Mathematics,
Philadelphia, PA, 1988.
[44] P. Concus, G. H. Golub, and G. Meurant, Block preconditioning for the
conjugate gradient method, SIAM J. Sci. Statist. Comput., 6 (1985), pp. 220–252.
[45] P. Concus, G. H. Golub, and D. P. O’Leary, A generalized conjugate
gradient method for the numerical solution of elliptic partial diﬀerential equations,
in Sparse Matrix Computations, J. R. Bunch and D. J. Rose, eds., Academic Press,
1976, pp. 309–332.
[46] E. J. Craig, The Nstep iteration procedures, J. Math. Phys., 34 (1955), pp. 64–
73.
[47] A. R. Curtis, M. J. D. Powell, and J. K. Reid, On the estimation of sparse
Jacobian matrices, J. Inst. Math. Appl., 13 (1974), pp. 117–119.
[48] J. W. Daniel, The conjugate gradient method for linear and nonlinear operator
equations, SIAM J. Numer. Anal., 4 (1967), pp. 10–26.
[49] D. W. Decker, H. B. Keller, and C. T. Kelley, Convergence rates for
Newton’s method at singular points, SIAM J. Numer. Anal., 20 (1983), pp. 296–314.
[50] D. W. Decker and C. T. Kelley, Newton’s method at singular points I, SIAM
J. Numer. Anal., 17 (1980), pp. 66–70.
[51] , Convergence acceleration for Newton’s method at singular points, SIAM J.
Numer. Anal., 19 (1982), pp. 219–229.
[52] , Sublinear convergence of the chord method at singular points, Numer.
Math., 42 (1983), pp. 147–154.
[53] , Broyden’s method for a class of problems having singular Jacobian at the
root, SIAM J. Numer. Anal., 22 (1985), pp. 566–574.
[54] T. J. Dekker, Finding a zero by means of successive linear interpolation, in
Constructive Aspects of the Fundamental Theorem of Algebra, P. Henrici, ed., 1969,
pp. 37–48.
[55] R. Dembo, S. Eisenstat, and T. Steihaug, Inexact Newton methods, SIAM
J. Numer. Anal., 19 (1982), pp. 400–408.
[56] R. Dembo and T. Steihaug, Truncated Newton algorithms for largescale
optimization, Math. Programming, 26 (1983), pp. 190–212.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
156 BIBLIOGRAPHY
[57] J. E. Dennis, On the Kantorovich hypothesis for Newton’s method, SIAM J.
Numer. Anal., 6 (1969), pp. 493–507.
[58] , Toward a uniﬁed convergence theory for Newtonlike methods, in Nonlinear
Functional Analysis and Applications, L. B. Rall, ed., Academic Press, New York,
1971, pp. 425–472.
[59] J. E. Dennis, J. M. Martinez, and X. Zhang, Triangular decomposition
methods for solving reducible nonlinear systems of equations, SIAM J. Optim., 4
(1994), pp. 358–382.
[60] J. E. Dennis and J. J. Mor´ e, A characterization of superlinear convergence
and its application to quasiNewton methods, Math. Comput., 28 (1974), pp. 549–560.
[61] , QuasiNewton methods, methods, motivation and theory, SIAM Rev., 19
(1977), pp. 46–89.
[62] J. E. Dennis and R. B. Schnabel, Least change secant updates for quasi
Newton methods, SIAM Rev., 21 (1979), pp. 443–459.
[63] , Numerical Methods for Unconstrained Optimization and Nonlinear Equa
tions, no. 16 in Classics in Applied Mathematics, SIAM, Philadelphia, 1996.
[64] J. E. Dennis and H. F. Walker, Convergence theorems for least change secant
update methods, SIAM J. Numer. Anal., 18 (1981), pp. 949–987.
[65] , Inaccuracy in quasiNewton methods: Local improvement theorems, in
Mathematical Programming Study 22: Mathematical programming at Oberwolfach
II, North–Holland, Amsterdam, 1984, pp. 70–85.
[66] , Leastchange sparse secant updates with inaccurate secant conditions,
SIAM J. Numer. Anal., 22 (1985), pp. 760–778.
[67] P. Deuflhard, R. W. Freund, and A. Walter, Fast secant methods for
the iterative solution of large nonsymmetric linear systems, Impact of Computing in
Science and Engineering, 2 (1990), pp. 244–276.
[68] W. J. Duncan, Some devices for the solution of large sets of simultaneous
linear equations (with an appendix on the reciprocation of partitioned matrices), The
London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science,
Seventh Series, 35 (1944), pp. 660–670.
[69] S. C. Eisenstat and H. F. Walker, Choosing the forcing terms in an inexact
Newton method, SIAM J. Sci. Comput., 17 (1996), pp. 16–32.
[70] , Globally convergent inexact Newton methods, SIAM J. Optim., 4 (1994),
pp. 393–422.
[71] H. C. Elman, Iterative Methods for Large, Sparse, Nonsymmetric Systems of
Linear Equations, PhD thesis, Yale University, New Haven, CT, 1982.
[72] H. C. Elman, Y. Saad, and P. E. Saylor, A hybrid ChebyshevKrylov
subspace algorithm for solving nonsymmetric systems of linear equations, SIAM J.
Sci. Statist. Comput., 7 (1986), pp. 840–855.
[73] M. Engelman, G. Strang, and K. J. Bathe, The application of quasi
Newton methods in ﬂuid mechanics, Internat. J. Numer. Methods Engrg., 17 (1981),
pp. 707–718.
[74] V. Faber and T. A. Manteuffel, Necessary and suﬃcient conditions for the
existence of a conjugate gradient method, SIAM J. Numer. Anal., 21 (1984), pp. 352–
362.
[75] D. Feng, P. D. Frank, and R. B. Schnabel, Local convergence analysis of
tensor methods for nonlinear equations, Math. Programming, 62 (1993), pp. 427–459.
[76] R. Fletcher, Conjugate gradient methods for indeﬁnite systems, in Numerical
Analysis Dundee 1975, G. Watson, ed., SpringerVerlag, Berlin, New York, 1976,
pp. 73–89.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BIBLIOGRAPHY 157
[77] R. W. Freund, A transposefree quasiminimal residual algorithm for non
Hermitian linear systems, SIAM J. Sci. Comput., 14 (1993), pp. 470–482.
[78] R. W. Freund, G. H. Golub, and N. M. Nachtigal, Iterative solution of
linear systems, Acta Numerica, 1 (1991), pp. 57–100.
[79] R. W. Freund, M. H. Gutknecht, and N. M. Nachtigal, An implemen
tation of the lookahead Lanczos algorithm for nonHermitian matrices, SIAM J. Sci.
Comput., 14 (1993), pp. 137–158.
[80] R. W. Freund and N. M. Nachtigal, QMR: a quasiminimal residual
algorithm for nonhermitian linear systems, Numer. Math., 60 (1991), pp. 315–339.
[81] , An implementation of the QMR method based on coupled twoterm
recurrences, SIAM J. Sci. Comput., 15 (1994), pp. 313–337.
[82] D. M. Gay, Some convergence properties of Broyden’s method, SIAM J. Numer.
Anal., 16 (1979), pp. 623–630.
[83] C. W. Gear, Numerical Initial Value Problems in Ordinary Diﬀerential Equa
tions, PrenticeHall, Englewood Cliﬀs, NJ, 1971.
[84] R. R. Gerber and F. T. Luk, A generalized Broyden’s method for solving
simultaneous linear equations, SIAM J. Numer. Anal., 18 (1981), pp. 882–890.
[85] P. E. Gill and W. Murray, QuasiNewton methods for unconstrained
optimization, J. I. M. A., 9 (1972), pp. 91–108.
[86] , Safeguarded steplength algorithms for optimization using descent methods,
Tech. Report NAC 37, National Physical Laboratory Report, Teddington, England,
1974.
[87] P. E. Gill, W. Murray, and M. H. Wright, Practical Optimization,
Academic Press, London, 1981.
[88] A. A. Goldstein, Constructive Real Analysis, Harper and Row, New York,
1967.
[89] G. H. Golub and C. G. VanLoan, Matrix Computations, Johns Hopkins
University Press, Baltimore, 1983.
[90] A. Griewank, Analysis and modiﬁcation of Newton’s method at singularities,
PhD thesis, Australian National University, 1981.
[91] , Rates of convergence for secant methods on nonlinear problems in Hilbert
space, in Numerical Analysis, Proceedings Guanajuato , Mexico 1984, Lecture Notes
in Math., No, 1230, J. P. Hennart, ed., SpringerVerlag, Heidelberg, 1986, pp. 138–
157.
[92] , The solution of boundary value problems by Broyden based secant methods,
in Computational Techniques and Applications: CTAC 85, Proceedings of CTAC,
Melbourne, August 1985, J. Noye and R. May, eds., North Holland, Amsterdam,
1986, pp. 309–321.
[93] , On the iterative solution of diﬀerential and integral equations using secant
updating techniques, in The State of the Art in Numerical Analysis, A. Iserles and
M. J. D. Powell, eds., Clarendon Press, Oxford, 1987, pp. 299–324.
[94] A. Griewank and G. F. Corliss, eds., Automatic Diﬀerentiation of Algo
rithms: Theory, Implementation, and Application, Society for Industrial and Applied
Mathematics, Philadelphia, PA, 1991.
[95] A. Griewank and P. L. Toint, Local convergence analysis for partitioned
quasinewton updates, Numer. Math., 39 (1982), pp. 429–448.
[96] , Partitioned variable metric methods for large sparse optimization problems,
Numer. Math., 39 (1982), pp. 119–137.
[97] W. A. Gruver and E. Sachs, Algorithmic Methods In Optimal Control,
Pitman, London, 1980.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
158 BIBLIOGRAPHY
[98] M. H. Gutknecht, Variants of BICGSTAB for matrices with complex spectrum,
SIAM J. Sci. Comput., 14 (1993), pp. 1020–1033.
[99] W. Hackbusch, MultiGrid Methods and Applications, vol. 4 of Springer Series
in Computational Mathematics, SpringerVerlag, New York, 1985.
[100] , Multigrid methods of the second kind, in Multigrid Methods for Integral
and Diﬀerential Equations, Oxford University Press, Oxford, 1985.
[101] W. E. Hart and S. O. W. Soul, QuasiNewton methods for discretized
nonlinear boundary problems, Journal of the Institute of Applied Mathematics, 11
(1973), pp. 351–359.
[102] H. V. Henderson and S. R. Searle, On deriving the inverse of a sum of
matrices, SIAM Rev., 23 (1981), pp. 53–60.
[103] M. R. Hestenes and E. Steifel, Methods of conjugate gradient for solving
linear systems, J. of Res. Nat. Bureau Standards, 49 (1952), pp. 409–436.
[104] D. M. Hwang and C. T. Kelley, Convergence of Broyden’s method in Banach
spaces, SIAM J. Optim., 2 (1992), pp. 505–532.
[105] E. Isaacson and H. B. Keller, Analysis of numerical methods, John Wiley,
New York, 1966.
[106] L. Kantorovich and G. Akilov, Functional Analysis, 2nd ed., Pergamon
Press, New York, 1982.
[107] L. V. Kantorovich, Functional analysis and applied mathematics, Uspehi Mat.
Nauk., 3 (1948), pp. 89–185. translation by C. Benster as Nat. Bur. Standards Report
1509. Washington, D. C., 1952.
[108] H. B. Keller, Newton’s method under mild diﬀerentiability conditions, J.
Comput. Sys. Sci, 4 (1970), pp. 15–28.
[109] , Lectures on Numerical Methods in Bifurcation Theory, Tata Institute of
Fundamental Research, Lectures on Mathematics and Physics, SpringerVerlag, New
York, 1987.
[110] C. T. Kelley, Solution of the Chandrasekhar Hequation by Newton’s method,
J. Math. Phys., 21 (1980), pp. 1625–1628.
[111] , A fast multilevel algorithm for integral equations, SIAM J. Numer. Anal.,
32 (1995), pp. 501–513.
[112] C. T. Kelley and E. W. Sachs, A quasiNewton method for elliptic boundary
value problems, SIAM J. Numer. Anal., 24 (1987), pp. 516–531.
[113] , A pointwise quasiNewton method for unconstrained optimal control
problems, Numer. Math., 55 (1989), pp. 159–176.
[114] , Fast algorithms for compact ﬁxed point problems with inexact function
evaluations, SIAM J. Sci. Statist. Comput., 12 (1991), pp. 725–742.
[115] , A new proof of superlinear convergence for Broyden’s method in Hilbert
space, SIAM J. Optim., 1 (1991), pp. 146–150.
[116] , Pointwise Broyden methods, SIAM J. Optim., 3 (1993), pp. 423–441.
[117] , Multilevel algorithms for constrained compact ﬁxed point problems, SIAM
J. Sci. Comput., 15 (1994), pp. 645–667.
[118] C. T. Kelley and R. Suresh, A new acceleration method for Newton’s method
at singular points, SIAM J. Numer. Anal., 20 (1983), pp. 1001–1009.
[119] C. T. Kelley and Z. Q. Xue, Inexact Newton methods for singular problems,
Optimization Methods and Software, 2 (1993), pp. 249–267.
[120] , GMRES and integral operators, SIAM J. Sci. Comput., 17 (1996), pp. 217–
226.
[121] T. Kerkhoven and Y. Saad, On acceleration methods for coupled nonlinear
elliptic systems, Numerische Mathematik, 60 (1992), pp. 525–548.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BIBLIOGRAPHY 159
[122] C. Lanczos, Solution of linear equations by minimized iterations, J. Res. Natl.
Bur. Stand., 49 (1952), pp. 33–53.
[123] T. A. Manteuffel, Adaptive procedure for estimating parameters for the
nonsymmetric Tchebychev iteration, Numer. Math., 31 (1978), pp. 183–208.
[124] T. A. Manteuffel and S. Parter, Preconditioning and boundary conditions,
SIAM J. Numer. Anal., 27 (1990), pp. 656–694.
[125] E. S. Marwil, Convergence results for Schubert’s method for solving sparse
nonlinear equations, SIAM J. Numer. Anal., (1979), pp. 588–604.
[126] S. McCormick, Multilevel Adaptive Methods for Partial Diﬀerential Equations,
Society for Industrial and Applied Mathematics, Philadelphia, PA, 1989.
[127] J. A. Meijerink and H. A. van der Vorst, An iterative solution method
for linear systems of which the coeﬃcient matrix is a symmetric Mmatrix, Math.
Comput., 31 (1977), pp. 148–162.
[128] C. D. Meyer, Matrix Analysis and Applied Linear Algebra, forthcoming.
[129] J. J. Mor´ e, Recent developments in algorithms and software for trust region
methods, in Mathematical Programming: The State of the Art, A. Bachem,
M. Gr¨ oschel, and B. Korte, eds., SpringerVerlag, Berlin, 1983, pp. 258–287.
[130] J. J. Mor´ e and J. A. Trangenstein, On the global convergence of Broyden’s
method, Math. Comput., 30 (1976), pp. 523–540.
[131] T. E. Mott, Newton’s method and multiple roots, Amer. Math. Monthly, 64
(1957), pp. 635–638.
[132] T. W. Mullikin, Some probability distributions for neutron transport in a half
space, J. Appl. Probab., 5 (1968), pp. 357–374.
[133] W. Murray and M. L. Overton, Steplength algorithms for minimizing a
class of nondiﬀerentiable functions, Computing, 23 (1979), pp. 309–331.
[134] N. M. Nachtigal, S. C. Reddy, and L. N. Trefethen, How fast are
nonsymmetric matrix iterations?, SIAM J. Matrix Anal. Appl., 13 (1992), pp. 778–
795.
[135] N. M. Nachtigal, L. Reichel, and L. N. Trefethen, A hybrid gmres
algorithm for nonsymmetric linear systems, SIAM J. Matrix Anal. Appl., 13 (1992).
[136] S. G. Nash, Preconditioning of truncated Newton methods, SIAM J. Sci. Statist.
Comput., 6 (1985), pp. 599–616.
[137] , Who was Raphson? (an answer). Electronic Posting to NADigest,
v92n23, 1992.
[138] J. L. Nazareth, Conjugate gradient methods less dependent on conjugacy,
SIAM Rev., 28 (1986), pp. 501–512.
[139] O. Nevanlinna, Convergence of Iterations for Linear Equations, Birkh¨ auser,
Basel, 1993.
[140] I. Newton, The Mathematical Papers of Isaac Newton (7 volumes), D. T.
Whiteside, ed., Cambridge University Press, Cambridge, 1967–1976.
[141] B. Noble, Applied Linear Algebra, Prentice Hall, Englewood Cliﬀs, NJ, 1969.
[142] J. Nocedal, Theory of algorithms for unconstrained optimization, Acta Numer
ica, 1 (1991), pp. 199–242.
[143] D. P. O’Leary, Why Broyden’s nonsymmetric method terminates on linear
equations, SIAM J. Optim., 4 (1995), pp. 231–235.
[144] J. M. Ortega, Numerical Analysis A Second Course, Classics in Appl. Math.,
No. 3, Society for Industrial and Applied Mathematics, Philadelphia, PA, 1990.
[145] J. M. Ortega and W. C. Rheinboldt, Iterative Solution of Nonlinear
Equations in Several Variables, Academic Press, New York, 1970.
[146] A. M. Ostrowski, Solution of Equations and Systems of Equations, Academic
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
160 BIBLIOGRAPHY
Press, New York, 1960.
[147] B. N. Parlett, The Symmetric Eigenvalue Problem, Prentice Hall, Englewood
Cliﬀs, NJ, 1980.
[148] B. N. Parlett, D. R. Taylor, and Z. A. Liu, A lookahead Lanczos
algorithm for unsymmetric matrices, Math. Comput., 44 (1985), pp. 105–124.
[149] D. W. Peaceman and H. H. Rachford, The numerical solution of parabolic
and elliptic diﬀerential equations, J. Soc. Indus. Appl. Math., 11 (1955), pp. 28–41.
[150] G. Peters and J. Wilkinson, Inverse iteration, illconditioned equations and
Newton’s method, SIAM Rev., 29 (1979), pp. 339–360.
[151] L. R. Petzold, A description of DASSL: a diﬀerential/algebraic system solver,
in Scientiﬁc Computing, R. S. Stepleman et al., eds., North Holland, Amsterdam,
1983, pp. 65–68.
[152] E. Picard, M´emoire sur la th´eorie des ´equations aux d´eriv´ees partielles et la
m´ethode des approximations successives, J. de Math. ser 4, 6 (1890), pp. 145–210.
[153] M. J. D. Powell, A hybrid method for nonlinear equations, in Numerical
Methods for Nonlinear Algebraic Equations, Gordon and Breach, New York, 1970,
pp. 87–114.
[154] , Convergence properties of a class of minimization algorithms, in Nonlinear
Programming 2, O. L. Mangasarian, R. R. Meyer, and S. M. Robinson, eds.,
Academic Press, New York, 1975, pp. 1–27.
[155] L. B. Rall, Convergence of the Newton process to multiple solutions, Numer.
Math., 9 (1961), pp. 23–37.
[156] J. Raphson, Analysis aequationum universalis seu ad aequationes algebraicas
resolvendas methodus generalis, et expedita, ex nova inﬁnitarum serierum doctrina,
deducta ac demonstrata. Original in British Library, London, 1690.
[157] G. W. Reddien, On Newton’s method for singular problems, SIAM J. Numer.
Anal., 15 (1978), pp. 993–986.
[158] J. K. Reid, Least squares solution of sparse systems of nonlinear equations by a
modiﬁed Marquardt algorithm, in Proc. NATO Conf. at Cambridge, July 1972, North
Holand, Amsterdam, 1973, pp. 437–445.
[159] W. C. Rheinboldt, Numerical Analysis of Parametrized Nonlinear Equations,
John Wiley, New York, 1986.
[160] J. R. Rice, Experiments on GramSchmidt orthogonalization, Math. Comput.,
20 (1966), pp. 325–328.
[161] T. J. Rivlin, The Chebyshev Polynomials, John Wiley, New York, 1974.
[162] H. L. Royden, Real Analysis, 2nd ed., Macmillan, New York, 1968.
[163] Y. Saad, Practical use of polynomial preconditionings for the conjugate gradient
method, SIAM J. Sci. Statist. Comput., 6 (1985), pp. 865–881.
[164] , Least squares polynomials in the complex plane and their use for solving
nonsymmetric linear systems, SIAM J. Numer. Anal., 24 (1987), pp. 155–169.
[165] , ILUT: A dual threshold incomplete LU factorization, Tech. Report 9238,
Computer Science Department, University of Minnesota, 1992.
[166] , A ﬂexible innerouter preconditioned GMRES algorithm, SIAM J. Sci.
Comput., (1993), pp. 461–469.
[167] Y. Saad and M. Schultz, GMRES a generalized minimal residual algorithm
for solving nonsymmetric linear systems, SIAM J. Sci. Statist. Comput., 7 (1986),
pp. 856–869.
[168] E. Sachs, Broyden’s method in Hilbert space, Math. Programming, 35 (1986),
pp. 71–82.
[169] P. E. Saylor and D. C. Smolarski, Implementation of an adaptive algorithm
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
BIBLIOGRAPHY 161
for Richardson’s method, Linear Algebrra Appl., 154/156 (1991), pp. 615–646.
[170] R. B. Schnabel and P. D. Frank, Tensor methods for nonlinear equations,
SIAM J. Numer. Anal., 21 (1984), pp. 815–843.
[171] E. Schr¨ oder,
¨
Uber unendlich viele Algorithmen zur Auﬂosung der Gleichungen,
Math. Ann., 2 (1870), pp. 317–365.
[172] L. K. Schubert, Modiﬁcation of a quasiNewton method for nonlinear equations
with sparse Jacobian, Math. Comput., 24 (1970), pp. 27–30.
[173] G. A. Schultz, R. B. Schnabel, and R. H. Byrd, A family of trustregion
based algorithms for unconstrained minimization with strong global convergence
properties, SIAM J. Numer. Anal., 22 (1985), pp. 47–67.
[174] V. E. Shamanskii, A modiﬁcation of Newton’s method, Ukran. Mat. Zh., 19
(1967), pp. 133–138. (In Russian.)
[175] A. H. Sherman, On Newtoniterative methods for the solution of systems of
nonlinear equations, SIAM J. Numer. Anal., 14 (1978), pp. 755–774.
[176] J. Sherman and W. J. Morrison, Adjustment of an inverse matrix corre
sponding to changes in the elements of a given column or a given row of the original
matrix (abstract), Ann. Math. Statist., 20 (1949), p. 621.
[177] , Adjustment of an inverse matrix corresponding to a change in one element
of a given matrix, Ann. Math. Statist., 21 (1950), pp. 124–127.
[178] K. Sigmon, MATLAB Primer, Fourth Edition, CRC Press, Boca Raton, FL,
1994.
[179] D. C. Smolarski and P. E. Saylor, An optimum iterative method for solving
any linear system with a square matrix, BIT, 28 (1988), pp. 163–178.
[180] P. Sonneveld, CGS, a fast Lanczostype solver for nonsymmetric linear
systems, SIAM J. Sci. Statist. Comput., 10 (1989), pp. 36–52.
[181] D. C. Sorensen, Newton’s method with a model trust region modiﬁcation, SIAM
J. Numer. Anal., 19 (1982), pp. 409–426.
[182] G. Starke, Alternating direction preconditioning for nonsymmetric systems of
linear equations, SIAM J. Sci. Comput., 15 (1994), pp. 369–385.
[183] G. Starke and R. S. Varga, A hybrid ArnoldiFaber iterative method for
nonsymmetric systems of linear equations, Numer. Math., 64 (1993), pp. 213–239.
[184] G. W. Stewart, Introduction to matrix computations, Academic Press, New
York, 1973.
[185] E. L. Stiefel, Kernel polynomials in linear algebra and their numerical
applications, U.S. National Bureau of Standards, Applied Mathematics Series, 49
(1958), pp. 1–22.
[186] P. N. Swarztrauber, The methods of cyclic reduction, Fourier analysis and
the FACR algorithm for the discrete solution of Poisson’s equation on a rectangle,
SIAM Rev., 19 (1977), pp. 490–501.
[187] , Approximate cyclic reduction for solving Poisson’s equation, SIAM J. Sci.
Statist. Comput., 8 (1987), pp. 199–209.
[188] P. N. Swarztrauber and R. A. Sweet, Algorithm 541: Eﬃcient FORTRAN
subprograms for the solution of elliptic partial diﬀerential equations, ACM Trans.
Math. Software, 5 (1979), pp. 352–364.
[189] P. L. Toint, On sparse and symmetric matrix updating subject to a linear
equation, Math. Comput., 31 (1977), pp. 954–961.
[190] J. F. Traub, Iterative Methods for the Solution of Equations, Prentice Hall,
Englewood Cliﬀs, NJ, 1964.
[191] H. A. van der Vorst, BiCGSTAB: A fast and smoothly converging variant to
BiCG for the solution of nonsymmetric systems, SIAM J. Sci. Statist. Comput., 13
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
162 BIBLIOGRAPHY
(1992), pp. 631–644.
[192] H. A. van der Vorst and C. Vuik, The superlinear convergence behaviour
of GMRES, Journal Comput. Appl. Math., 48 (1993), pp. 327–341.
[193] R. S. Varga, Matrix Iterative Analysis, Prentice Hall, Englewood Cliﬀs, NJ,
1962.
[194] E. L. Wachspress, Iterative Solution of Elliptic Systems and Applications to
the Neutron Diﬀusion Equations of Reactor Physics, Prentice Hall, Englewood Cliﬀs,
NJ, 1966.
[195] H. F. Walker, Implementation of the GMRES method using Householder
transformations, SIAM J. Sci. Statist. Comput., 9 (1989), pp. 815–825.
[196] , Residual smoothing and peak/plateau behavior in Krylov subspace methods,
Applied Numer. Math., 19 (1995), pp. 279–286.
[197] H. F. Walker and L. Zhou, A simpler GMRES, J. Numer. Lin. Alg. Appl.,
6 (1994), pp. 571–581.
[198] L. T. Watson, S. C. Billups, and A. P. Morgan, Algorithm 652:
HOMPACK: A suite of codes for globally convergent homotopy algorithms, ACM
Trans. Math. Software, 13 (1987), pp. 281–310.
[199] M. A. Woodbury, Inverting modiﬁed matrices, Memorandum Report 42,
Statistical Research Group, Princeton NJ, 1950.
[200] D. M. Young, Iterative Solution of Large Linear Systems, Academic Press, New
York, 1971.
[201] T. J. Ypma, The eﬀect of rounding errors on Newtonlike methods, IMA J.
Numer. Anal., 3 (1983), pp. 109–118.
[202] L. Zhou and H. F. Walker, Residual smoothing techniques for iterative
methods, SIAM J. Sci. Comput., 15 (1994), pp. 297–312.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Index
Aconjugacy, 20
Anorm, 12
Approximate inverse, 6, 77
Armijo rule, 137
Arnoldi process, 37
Backtracking, 136
Bad Broyden method, 132
Banach Lemma, 6
BiCG, 47
algorithm, 47
BiCGSTAB, 50
algorithm, 50
ﬁnite diﬀerence, 110
Biconjugacy, 47
BiConjugate Gradient, 47
algorithm, 47
Biorthogonality, 47
Bounded deterioration, 118, 121
Breakdown, 47
Broyden sequence, 120
Broyden’s method, 113
convergence
linear equations, 118
nonlinear equations, 120
implementation, 123
limited memory, 123
restarting, 123
sparse, 133
Brsol
algorithm, 126
Brsola
algorithm, 145
Cauchy sequence, 67
CGNE, 25
CGNR, 25
CGS, 48
algorithm, 49
Chandrasekhar Hequation, 87
solution
Broyden’s method, 128
Newton’s method, 87
NewtonGMRES, 107
Characteristic polynomial, 34
Chord method, 74
algorithm, 74
convergence, 76
Condition number, 3
Conjugate Gradient
algorithm, 22
ﬁnite termination, 14
minimization property, 12
use of residual polynomials, 14
Conjugate Gradient Squared, 48
algorithm, 49
Conjugate transpose, 34
Contraction mapping, 67
Contraction Mapping Theorem, 67
ConvectionDiﬀusion Equation
Broyden’s method, 130
Newton–Armijo, 146
NewtonGMRES, 108
solution
Broyden’s method, 130
Convergence theorem
Broyden’s method
linear equations, 119
163
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
164 INDEX
nonlinear equations, 123
chord method, 76, 77
contraction mapping, 67
ﬁnite diﬀerence Newton, 81
inexact Newton’s method, 96,
99
Newton’s method, 71
Newton–Armijo, 141
NewtonGMRES, 103
secant method, 82
Shamanskii method, 79
DASPK, 110
Dennis–Mor´e Condition, 114
superlinear convergence, 115
Diagonalizable matrix, 34
Diagonalizing transformation, 34
Diﬀerence approximation, 79
choice of h, 80
directional derivative, 80
Jacobian, 80
nonlinearity, 80
Elementary Jordan block, 60
Fixed point, 66
Fixedpoint iteration, 66
Forcing term, 95
choice of, 104
Gauss–Seidel
algorithm, 9
Givens rotation, 43
Globally convergent algorithm, 135
GMRES
algorithm, 40, 45
diagonalizable matrices, 34
ﬁnite diﬀerences, 101
ﬁnite termination, 34
GMRES(m), 39, 46
algorithm, 46
loss of orthogonality, 41
minimization property, 33
restarts, 39, 46
use of residual polynomials, 33
Gram–Schmidt process, 37
loss of orthogonality, 40, 41
modiﬁed, 40
Hequation, 87
solution
Broyden’s method, 128
Newton’s method, 87
NewtonGMRES, 107
H¨older Continuity, 91
High order methods, 78
Hybrid algorithms, 36
Induced matrix norm
deﬁnition, 3
Inexact Newton method, 95
Inner iteration, 100
Inverse power method, 94
Inverse secant equation, 132
Inverse Tangent
Newton–Armijo, 146
Iteration matrix, 5
Iteration statistics, 91
Jacobi
algorithm, 8
convergence result, 8
preconditioning, 9
Jacobi preconditioning, 24
Jordan block, 60
Kantorovich Theorem, 83
Krylov subspace, 11
Left preconditioning, 36
Limited memory, 123
Line Search
parabolic
threepoint, 143
twopoint, 142
polynomial, 142
Line search, 135, 136
Linear PDE
BiCGSTAB, 57
Broyden’s method, 127
CGNR, 55
GMRES, 54
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
INDEX 165
solution
Broyden’s method, 127
TFQMR, 58
Lipschitz continuous
deﬁnition, 67
Local improvement, 75
inexact, 100
Matrix norm, 3
Matrix splitting, 7
Modiﬁed Bratu Problem, 132
Newton direction, 136
Newton’s method, 71
algorithm, 74
convergence rate, 71
ﬁnite diﬀerences, 81
stagnation, 81
inexact, 95
monotone convergence, 92
termination, 72
NewtonGMRES, 101
convergence rates, 103
ﬁnite diﬀerences, 101
safeguarding, 105
Newtoniterative method, 100
Normal matrix, 34
Nsol
algorithm, 86
Nsola
algorithm, 138
Nsolgm
algorithm, 105
Outer iteration, 100
Over solving, 104
Picard iteration, 66
Poisson solver, 24, 27
preconditioner
for CG, 24, 27
for GMRES, 36, 54
Polynomial preconditioning, 24, 36
Richardson iteration, 7
Positive deﬁnite matrix, 11
Preconditioned CG
algorithm, 23
Preconditioning
conjugate gradient, 19
GMRES, 36
incomplete
factorization, 24, 36
Jacobi, 24
Poisson solver, 27, 36, 54
polynomial, 24, 36
Richardson iteration, 6
Quasiminimization, 52
QuasiNewton methods, 113
Quasiresidual, 51
Quasiresidual norm, 52
Relaxation parameter, 9
Reorthogonalization, 41
Residual
linear, 11
as error estimate, 4
nonlinear
as error estimate, 72
relative nonlinear, 72
Residual polynomials
application to CG, 14
application to CGNR, 25
application to GMRES, 33
application to TFQMR, 52
deﬁnition, 14
Richardson iteration, 5
nonlinear, 66
Right preconditioning, 36
Schubert algorithm, 133
Search direction, 136
Secant equation, 113
Secant method, 82
convergence, 82
qorder, 91
rorder, 91
Secant update, 113
Shamanskii method, 78
algorithm, 78
convergence, 79
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
166 INDEX
ﬁnite diﬀerences, 91
Sherman–Morrison
formula, 124, 132
Sherman–Morrison–Woodbury for
mula, 132
Simple decrease, 136
SOR
algorithm, 9
SPD matrix, 12
Spectral radius, 7
Stagnation, 94
Newton’s method, 75
Stationary iterative methods, 5
as preconditioners, 24
Successive substitution, 66
Suﬃcient decrease, 137
Symmetric Gauss–Seidel
algorithm, 9
preconditioning, 9
Symmetric matrix, 11
Symmetric positive deﬁnite matrix,
12
Termination
CG, 16
GMRES, 35, 38
linear equations, 4
nonlinear equations, 72
TFQMR, 52
TFQMR, 51
algorithm, 53
ﬁnite diﬀerence, 110
weights, 52
Weighted norm, 98
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
To Polly H. Thomas, 19061994, devoted mother and grandmother
1
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
2
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
. . . . . . . . . . . . . . . . . . . . . . .1 Review and notation . . .2 CGS. . . . . . . . . . . 5 1. .5 Exercises on stationary iterative methods . . . . . . . . 7 1. . . . . Conjugate Gradient Iteration 2. . . . 3. . 10 CHAPTER 2. . . . . .6 CGNR and CGNE .2 Consequences of the minimization property .1 BiCG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. .3 Termination of the iteration . 7 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CHAPTER 3. . . . . . . . . . . .4 Matrix splittings and classical stationary iterative methods . . .Copyright ©1995 by the Society for Industrial and Applied Mathematics. . . .1 Krylov methods and the minimization property . . . . . . . . . . . . . 2. This electronic version is for personal use and may not be duplicated or distributed. . . . . . . . . . .com/SIAM/FR16. . . . 3. . . . . . .8 Exercises on conjugate gradient . . .6. 3. . . . 2.4 GMRES implementation: Basic ideas . . . . . . . . . . . . . . . . . . . Basic Concepts and Stationary Iterative Methods 3 1. 2. . . . . . . . . . . vii Buy this book from SIAM at http://www.1 The minimization property and its consequences 3. . 3. . . . . . . . . . . . .7 Examples for preconditioned conjugate iteration 2. . 11 11 13 15 19 22 25 26 30 33 33 35 36 37 43 46 47 48 50 .3 BiCGSTAB. . . . 3. 3. .3 Preconditioning . 3. . . . . . .6. . . . 2. . .6. . . . . . . . . . . . . .6 Other methods for nonsymmetric systems . GMRES Iteration 3. . . . . . . . . . . . . . . . .ecsecurehost. . . . . . . . . . . . .5 Implementation: Givens rotations . . . .4 Implementation . . . . . . . . .2 Termination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Preconditioning . . . . . . . . . .2 The Banach Lemma and approximate inverses . . . . . . . . . . . . .3 The spectral radius . . . . . . . . . . . . . . . . . .html. . 3 1. . . . . . . . . . . . . 2. . . . . . . . Contents Preface xi How to Get the Software xiii CHAPTER 1. . . . . .
. . . . . . . . . .7 3. . . .1. . . . . . . . . . . . . . . . . Exercises on GMRES . . . . . . . . . 5.4. . . . . . . . . . . . . . . . . . . . . 5. . . . . . . . . . . .1 Local convergence of Newton’s method 5. . . . . .4. . . . . 6. . . . . . . . . .3 The Shamanskii method. . . . . . viii 3. .2 Approximate inversion of F . . . . . 7.2 Nonlinear problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. . . . . . . . . . . . . . . . . CHAPTER 5. . . . . . . . . .2 Convectiondiﬀusion equation. . . . . . . . . . . . . . . .2. . 6. . . .1 The chord method. 6.1 The Dennis–Mor´ condition . . . Buy this book from SIAM at http://www. . . . . . . . . . . . . . . .2 Weighted norm analysis. . . . . . . . . . . . .html. . . . . . . . . . .4. . . . . . . . . . . . . . . .1 Linear problems.5 The Kantorovich Theorem . . . . .3 Implementation of Broyden’s method 7. . . . . 5. . . . . . . . . CHAPTER 7.6 Examples for Newton’s method . . . . . . . . .2. . . . . . .com/SIAM/FR16. . . . . . . . . .4 Examples for Broyden’s method . . . . . . . . . .9 . 6. Newton’s Method 5. . . . . . . . . . . . 6. . . .4 Errors in the function and derivative . .1 Direct analysis. . . . .4 Diﬀerence approximation to F . .4. . . . . . . . . . . . . . . .3 Implementation of Newton’s method . . . . .2. . . . . . .8 3. . . . . . . . . . . . . 4. . . Broyden’s method 7. . . . . . . . . .1. . . .2. . . . 5. . . . . .3 Errors in the function. . Basic Concepts and FixedPoint Iteration 4. . . . BiCGSTAB.1 The basic estimates . . . . . . . . . . . . . . . . . . . . . . . . . . Inexact Newton Methods 6. . . . . . . . .Copyright ©1995 by the Society for Industrial and Applied Mathematics. . . . . . . . . . 6. . . . . . . . . . . . . . . . . . . . . . .2 Other Newtoniterative methods.4. . . . . . . . . . . . . 5. . . . . . . . .1. . . . . . . . .1 Newton GMRES. . . . . . 6. . . . . . . . . . . . . . . . . . .4. . . .1 Chandrasekhar Hequation. . . e 7. Examples for GMRES iteration .6. . . . . . . . . . . Examples for CGNR. .2 Termination of the iteration . . . . 6. . . . . . . . . 5. . . . . . . . . . . . . . .3 The standard assumptions . . . . . . . . 5.4 Examples for NewtonGMRES . . . . . . . .5 Exercises on inexact Newton methods . . .7 Exercises on Newton’s method . . . . . . . . . . . . . . . . . . . . . . . . .2 Fixedpoint iteration . . . . . . . . . . .ecsecurehost. . . . . . 4. 5. . . 51 54 55 60 65 65 66 68 71 71 72 73 75 76 77 78 79 82 83 86 91 95 95 95 97 100 100 101 104 104 106 107 108 110 113 114 116 118 120 123 127 CHAPTER 4. . . . . . . . 6. . . .4. . .5 The secant method. 5. . . . . . . . . . . . . . 6. . . . . . . . . . CHAPTER 6. . . . . . . . .3 NewtonGMRES implementation . 7. and TFQMR iteration . . . . . . . . .1 Types of convergence . . . . . . . . . . . This electronic version is for personal use and may not be duplicated or distributed. . . . .2 Convergence analysis . . . . 6. 7. . . . . .2 Newtoniterative methods .4 TFQMR. . . . . . . CONTENTS 3. . . . . . . . . . . . . . . . . . . . . .
.4. . .2 Nonlinear problems. . 135 135 138 141 142 144 146 146 146 148 151 153 163 CHAPTER 8. . . . . . . . .3 Broyden–Armijo. .1 Single equations . . . . . . . . . . . . . . . . . . . . . . . . 8. . . . .com/SIAM/FR16. . . . . . . . . . . .1 Inverse tangent function. . . . . . . . . . . .4 Examples for Newton–Armijo . . . . . . . .5 7. . . . . . . . . . .4. .2 Convectiondiﬀusion equation. . . . . . .1 Linear problems. . . .ecsecurehost. . CONTENTS ix 7. . . . . . . . . . Global Convergence 8. . . 127 7. . .4. 8. . . . . . . . . . . . . 8. . . . 8. . 132 . . .4. . . . .3. . . .2 Analysis of the Armijo rule . . . . . . .3. . . . . . . Bibliography Index Buy this book from SIAM at http://www.Copyright ©1995 by the Society for Industrial and Applied Mathematics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Exercises on global convergence . .3 Implementation of the Armijo rule . . .2 Broyden’s method. . . . . . . . . . . . . . . . . . . 128 Exercises on Broyden’s method . . . 8. .html. . . . 8. . . . . . . .4. . . 8. . . . . . . . . . . . 8. . . . . . . . 8. . . . . .1 Polynomial line searches. . . . . This electronic version is for personal use and may not be duplicated or distributed. . . .
x CONTENTS Buy this book from SIAM at http://www.com/SIAM/FR16. .Copyright ©1995 by the Society for Industrial and Applied Mathematics.html. This electronic version is for personal use and may not be duplicated or distributed.ecsecurehost.
and the central ideas of direct methods for the numerical solution of dense linear systems as described in standard texts such as [7]. A good introduction to the latest 1 MATLAB is a registered trademark of The MathWorks. [105]. and for smalltomedium scale production work. we have selected for coverage mostly algorithms and methods of analysis which extend directly to the inﬁnitedimensional case and whose convergence can be thoroughly analyzed. This electronic version is for personal use and may not be duplicated or distributed.0a on various SUN SPARCstations and version 4. . The examples are not designed to give a complete picture of performance or to be a suite of test problems. linear algebra.com/SIAM/FR16. The computational examples and exercises focus on discretizations of inﬁnitedimensional problems such as integral and diﬀerential equations.Copyright ©1995 by the Society for Industrial and Applied Mathematics. It may also be used as a textbook for introductory courses in nonlinear equations or iterative methods or as source material for an introductory course in numerical analysis at the graduate level.html. The analysis of Broyden’s method presented in Chapter 7 and the implementations presented in Chapters 7 and 8 are diﬀerent from the classical ones and also extend directly to an inﬁnitedimensional setting. xi Buy this book from SIAM at http://www.1 MATLAB codes for many of the algorithms are available by anonymous ftp. or [184]. Preface This book on iterative methods for linear and nonlinear equations can be used as a tutorial and a reference by anyone who needs to solve nonlinear systems of equations or large linear systems. Our approach is to focus on a small number of methods and treat them in depth. Though this book is written in a ﬁnitedimensional setting. The computational examples in this book were done with MATLAB (version 4. These examples are intended to provide results that can be used to validate the reader’s own implementations and to give a sense of how the algorithms perform. We assume that the reader is familiar with elementary numerical analysis. Inc.1 on an Apple Macintosh Powerbook 180) and the MATLAB environment is an excellent one for getting experience with the algorithms. for doing the exercises. For example.ecsecurehost. We present a limited number of computational examples. the matrixfree formulation and analysis for GMRES and conjugate gradient is almost unchanged in an inﬁnitedimensional setting.
Tony Choi.com/SIAM/FR16. Jeﬀ Scroggs. Gordon Wade. Parts of this book are based upon work supported by the National Science Foundation and the Air Force Oﬃce of Scientiﬁc Research over several years. I thank ChungWei Ng and my parents for over one hundred and ten years of patience and support. Ekkehard Sachs. suggestions. Howard Elman. most recently under National Science Foundation Grant Nos. Andreas Griewank. Moody Chu. 1998 Buy this book from SIAM at http://www. [43] is also a useful resource. I am especially indebted to Jim Banoczi. Mike Tocci. Belinda King. Steve Wright. Lea Jenkins. Joseph Skudlarek. Carl Meyer. Jim Epperson. Steve Campbell. This electronic version is for personal use and may not be duplicated or distributed.2) of MATLAB is the MATLAB Primer [178]. T. and an anonymous reviewer for their contributions and encouragement. C. Yue Zhang. If the reader has no access to MATLAB or will be solving very large problems. Homer Walker. Laura Helfrich.Copyright ©1995 by the Society for Industrial and Applied Mathematics. .ecsecurehost. Vickie Kearn. Casey Miller. and conclusions or recommendations expressed in this material are those of the author and do not necessarily reﬂect the views of the National Science Foundation or of the Air Force Oﬃce of Scientiﬁc Research. Most importantly. Jeﬀ Butera. Kelley Raleigh. DMS9024622 and DMS9321938. Ilse Ipsen. Debbie Lockhart. the general algorithmic descriptions or even the MATLAB codes can easily be translated to another language. Any opinions. and pointers to the literature. ideas. ﬁndings.html. Many of my students and colleagues discussed various aspects of this project with me and provided important corrections. Zhaqing Xue. North Carolina January. xii PREFACE version (version 4.
com or from SIAM’s World Wide Web site http://www.Copyright ©1995 by the Society for Industrial and Applied Mathematics. . http://www.siam. from the MathWorks World Wide Web site.com/SIAM/FR16. Phone: (508) 6531415 Fax: (508) 6532997 Email: info@mathworks.mathworks.org/books/kelley/kelley.mathworks. How to get the software A collection of MATLAB codes has been written to accompany this book. The MATLAB codes can be obtained by anonymous ftp from the MathWorks server ftp.com in the directory pub/books/kelley. Inc. This electronic version is for personal use and may not be duplicated or distributed.com WWW: http://www. MA 01760. 24 Prime Park Way Natick.html.com xiii Buy this book from SIAM at http://www.mathworks.ecsecurehost.html One can obtain MATLAB from The MathWorks.
1. This electronic version is for personal use and may not be duplicated or distributed. . Throughout this chapter x will denote a potential solution and {xk }k≥0 the sequence of iterates. Chapter 1 Basic Concepts and Stationary Iterative Methods 1. We will denote the ith component of a vector x by (x)i (note the parentheses) and the ith component of xk by (xk )i .1.Copyright ©1995 by the Society for Industrial and Applied Mathematics.1. If · · is is the lp norm x p = N j=1 1/p (x)i  p 3 Buy this book from SIAM at http://www. x =1 where A is a nonsingular N × N matrix. and Induced norms have the important property that Ax ≤ A x . Review and notation We begin by setting notation and reviewing some ideas from numerical linear algebra that we expect the reader to be familiar with. b ∈ RN is given. We will write linear equations as (1. x∗ = A−1 b ∈ RN is to be found. Definition 1. An excellent reference for the basic ideas of numerical linear algebra and direct methods for linear equations is [184].1) Ax = b.com/SIAM/FR16. The induced matrix norm of an N × N matrix A is deﬁned by A = max Ax .html. Recall that the condition number of A relative to the norm κ(A) = A A−1 . where κ(A) is understood to be inﬁnite if A is singular.ecsecurehost. We will rarely need to refer to individual components of vectors. In this chapter · will denote a norm on RN as well as the induced matrix norm. Let · be a norm on RN .
e0 r0 rk < τ.4) < τ.2) and (1.4) are the same when x0 = 0. which is a common choice. . Let A be nonsingular and let x∗ = A−1 b. particularly when the linear iteration is being used as part of a nonlinear solver. The termination criterion (1. r0 as asserted.2) depends on the initial iterate and may result in unnecessary work when the initial iterate is good and a poor result when the initial iterate is far from the solution. Buy this book from SIAM at http://www.1. This electronic version is for personal use and may not be duplicated or distributed. One termination criterion is (1.ecsecurehost. 4 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS we will write the condition number as κp . e0 A −1 r0 r0 Ae = A−1 r e r ≤ κ(A) .1. (1.com/SIAM/FR16. Most iterative methods terminate when the residual r = b − Ax is suﬃciently small. x. Since r = b − Ax = −Ae we have e = A−1 Ae ≤ A−1 and r0 = Ae0 ≤ A e0 . Let b.3) Proof. For this reason we prefer to terminate the iteration when rk (1.html. x0 ∈ RN . Lemma 1. b The two conditions (1.2) which can be related to the error e = x − x∗ in terms of the condition number.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Hence e A−1 r r ≤ = κ(A) .
This proves that I − M is nonsingular and that S = (I − M )−1 . 1− M Proof. The Banach Lemma and approximate inverses The most straightforward approach to an iterative solution of a linear system is to rewrite (1. Iterative methods of this form are called stationary iterative methods because the transition from xk to xk+1 does not depend on the history of the iteration. say to S. we must have M S + I = S and hence (I − M )S = I. and to deﬁne the Richardson iteration (1. Hence the sequence Sk converges.2. l=0 The partial sums Sk = form a Cauchy sequence in RN ×N . Since M Sk + I = Sk+1 .7) xk+1 = M xk + c. This electronic version is for personal use and may not be duplicated or distributed.1) as a linear ﬁxedpoint iteration. We will discuss more general methods in which {xk } is given by (1. We will show that I − M is nonsingular and that (1. Buy this book from SIAM at http://www. k l=0 Ml To see this note that for all m > k m l=k+1 Sk − Sm ≤ Now.2.7) M is an N ×N matrix called the iteration matrix.6) xk+1 = (I − A)xk + b. .8) .8) holds by showing that the series ∞ M l = (I − M )−1 . One way to do this is to write Ax = b as (1.5) x = (I − A)x + b. The Krylov methods discussed in Chapters 2 and 3 are not stationary iterative methods. All our results are based on the following lemma. Hence l Ml .com/SIAM/FR16. BASIC CONCEPTS 5 1. If M is an N × N matrix with M < 1 then I − M is nonsingular and 1 (I − M )−1 ≤ (1.ecsecurehost. Lemma 1. because m l=k+1 · is a matrix norm that is induced by a vector 1 − M m−k 1− M Sk − Sm ≤ M l = M k+1 →0 as m. M l ≤ M norm. k → ∞.html. In (1.1.Copyright ©1995 by the Society for Industrial and Applied Mathematics.
1 − I − BA Since A−1 = (I − M )−1 B.2. It is sometimes possible to precondition a linear equation by multiplying both sides of (1. If A and B are N × N matrices and B is an approximate inverse of A.1 indicates.2. (1. Richardson iteration.2. A − B −1 = B −1 (I − BA).ecsecurehost. and use (1.8) and completes the proof. 1 − I − BA B I − BA . Corollary 1. Definition 1. By Lemma 1. the matrices B that allow us to apply the Banach lemma and its corollary are called approximate inverses.1 is that Richardson iteration (1. 1 − I − BA B −1 ≤ A .1. Hence both A and B are nonsingular. 6 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Noting that (I − M )−1 ≤ ∞ l=0 M l = (1 − M )−1 .9). then not only will the iteration converge rapidly.com/SIAM/FR16. A consequence of Corollary 1. This electronic version is for personal use and may not be duplicated or distributed.2.12) If the norm of I − BA is small.8) (1.11) implies the ﬁrst part of (1. The following corollary is a direct consequence of Lemma 1. preconditioned with approximate inversion. but. proves (1.1.10) A−1 − B ≤ A−1 ≤ B . Theorem 1. By (1. 1 − I − BA A − B −1 ≤ Proof.6) will converge if I − A < 1.1) by a matrix B BAx = Bb so that convergence of iterative methods is improved.9). Let M = I − BA.1. 1 − I − BA A I − BA .1. then A and B are both nonsingular and (1. as Lemma 1. In the context of Richardson iteration.2. termination decisions based on the Buy this book from SIAM at http://www.7) converges to x = (I − M )−1 c for all initial iterates x0 .11) A−1 B −1 = (I − M )−1 ≤ 1 1− M = 1 .1. B is an approximate inverse of A if I − BA < 1.Copyright ©1995 by the Society for Industrial and Applied Mathematics. has the form xk+1 = (I − BA)xk + Bb. The second part follows in a similar way from B −1 = A(I − M )−1 . The following theorem is often referred to as the Banach Lemma. inequality (1.9) and (1. To complete the proof note that A−1 − B = (I − BA)A−1 .1 I − M = I − (I − BA) = BA is nonsingular.html.2. If M < 1 then the iteration (1. .
This electronic version is for personal use and may not be duplicated or distributed. and related problems [15]. Multigrid methods [19]. which tries to approximate A−1 by a polynomial in A [123]. The spectral radius The analysis in § 1. Hence the performance of the iteration is not completely described by M .1. The iteration (1. [99].3. We state that converse as a theorem and refer to [105] for a proof. Definition 1.3. The spectral radius of M is independent of any particular matrix norm of M . can also be interpreted in this light. Methods such as Jacobi. [117]. The term on the righthand side of the second equality in (1.13) ρ(A) = max λ = lim An λ∈σ(A) n→∞ 1/n . . Lemma 1. Then for any > 0 there is a norm · on RN such that ρ(A) > A − . Matrix splittings and classical stationary iterative methods There are ways to convert Ax = b to a linear ﬁxedpoint iteration that are diﬀerent from (1.3. Theorem 1.14) ρ(A) ≤ A for any induced matrix norm. [100]. in fact. It is clear. This method is a very eﬀective technique for solving diﬀerential equations. Buy this book from SIAM at http://www. We let σ(A) denote the set of eigenvalues of A. The concept of spectral radius allows us to make a complete description.1. 1. integral equations.4.1 is a characterization of convergent stationary iterative methods.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Let A be an N × N matrix.3. BASIC CONCEPTS 7 preconditioned residual Bb − BAx will better reﬂect the actual error. that (1.7) converges for all c ∈ RN if and only if ρ(M ) < 1. [6].html.5. The proof is left as an exercise.13) is the limit used by the radical test for convergence of the series An .7) in terms of spectral radius. Gauss–Seidel. [169]. However the norm of M could be small in some norms and quite large in others. A consequence of Theorem 1.2.2 related convergence of the iteration (1. The inequality (1.2. 1.5). The spectral radius of an N × N matrix A is (1. [111].ecsecurehost.1. [126].3. We mention one other approach. Let M be an N ×N matrix. and Exercise 1.7) to the norm of the matrix M . Theorem 1. polynomial preconditioning.14) has a partial converse that allows us to completely describe the performance of iteration (1. [179].com/SIAM/FR16. and sucessive overrelaxation (SOR) iteration are based on splittings of A of the form A = A1 + A2 .1.
1 The analysis of the method is based on an estimation of the spectral radius of the iteration matrix M = −A−1 A2 . Let A be an N × N matrix and assume that for all 1≤i≤N (1. However the classical methods have a role as preconditioners.15) (xk+1 )i = a−1 bi ii − j=i aij (xk )j .4. A2 = L + U. This electronic version is for personal use and may not be duplicated or distributed. Hence MJAC ∞ < 1 and the iteration converges to the unique solution of x = M x + D−1 b. Also I − M = D−1 A is nonsingular and therefore A is nonsingular. Note that the ith row sum of M = MJAC satisﬁes N j=1 mij  = j=i aij  aii  < 1.ecsecurehost. 8 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS where A1 is a nonsingular matrix constructed so that equations with A1 as coeﬃcient matrix are easy to solve. Then Ax = b is converted to the ﬁxedpoint problem x = A−1 (b − A2 x). As a ﬁrst example we consider the Jacobi iteration that uses the splitting A1 = D. Proof.Copyright ©1995 by the Society for Industrial and Applied Mathematics. or [200]. Theorem 1. This leads to the iteration matrix MJAC = −D−1 (L + U ).com/SIAM/FR16. Buy this book from SIAM at http://www.15) converges to x∗ = A−1 b for all b. 1 For a detailed description of the classical stationary iterative methods the reader may consult [89]. The limited description in this section is intended as a review that will set some notation to be used later.html. 0< j=i Then A is nonsingular and the Jacobi iteration (1.1. [144]. . We present only one convergence result for the classical stationary iterative methods. [105]. These methods are usually less eﬃcient than the Krylov methods discussed in Chapters 2 and 3 or the more modern stationary methods based on multigrid ideas.16) aij  < aii . Letting (xk )i denote the ith component of the kth iterate we can express Jacobi iteration concretely as (1. [193]. where D is the diagonal of A and L and U are the (strict) lower and upper triangular parts. Note that A1 is diagonal and hence trivial to invert.
[8]. This results in the iteration (xk+1 )i = a−1 bi − ii the splitting and iteration matrix aij (xk+1 )j − j<i j>i aij (xk )j . (1. The successive overrelaxation iteration modiﬁes Gauss–Seidel by adding a relaxation parameter ω to construct an iteration with iteration matrix MSOR = (D + ωL)−1 ((1 − ω)D − ωU ).17) For symmetric Gauss–Seidel (1. BJAC = D−1 . From the point of view of preconditioning. A2 = U. Backward Gauss–Seidel begins the update of x with the N th coordinate rather than the ﬁrst. That means that one wants to ﬁnd B such that M = I − BA and then use B as an approximate inverse. In that event MSGS = (D + U )−1 L(D + L)−1 U = (D + LT )−1 L(D + L)−1 LT . References [200]. BASIC CONCEPTS 9 Gauss–Seidel iteration overwrites the approximate solution with the new value as soon as it is computed. [193]. resulting in the splitting A1 = D + U. MGS = −(D + L)−1 U. This leads to the iteration matrix MSGS = MBGS MGS = (D + U )−1 L(D + L)−1 U.html. [89]. A2 = L. Note that A1 is lower triangular. A further disadvantage is that the choice of ω is often diﬃcult to make. and hence A−1 y is easy to compute for vectors 1 y. and the papers cited therein provide additional reading on this topic. the iteration depends on the ordering of the unknowns. Buy this book from SIAM at http://www. unlike Jacobi iteration. This electronic version is for personal use and may not be duplicated or distributed.18) BSGS = (D + LT )−1 D(D + L)−1 .Copyright ©1995 by the Society for Industrial and Applied Mathematics. For the Jacobi iteration. one wants to write the stationary method as a preconditioned Richardson iteration. If A is symmetric then U = LT . A symmetric Gauss–Seidel iteration is a forward Gauss–Seidel iteration followed by a backward Gauss–Seidel iteration. The performance can be dramatically improved with a good choice of ω but still is not competitive with Krylov methods. . and iteration matrix MBGS = −(D + U )−1 L. A1 = D + L.ecsecurehost. Note also that.com/SIAM/FR16.
5. 1.7) fails to converge. Show that if ρ(M ) ≥ 1 then there are x0 and c such that the iteration (1.5. 1.1.5.ecsecurehost.3.com/SIAM/FR16. Show that if A is symmetric and positive deﬁnite (that is AT = A and xT Ax > 0 for all x = 0) that BSGS is also symmetric and positive deﬁnite. . Exercises on stationary iterative methods 1. 1.Copyright ©1995 by the Society for Industrial and Applied Mathematics.18).5.4. Buy this book from SIAM at http://www.5. This electronic version is for personal use and may not be duplicated or distributed.3. Prove Theorem 1.2.2.html. 10 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 1. Verify equality (1.
com/SIAM/FR16.html. . The conjugate gradient (CG) iteration was invented in the 1950s [103] as a direct method. we assume that A is a nonsingular N × N matrix and let x∗ = A−1 b. .6. Unlike the stationary iterative methods. at the kth iteration.ecsecurehost. This electronic version is for personal use and may not be duplicated or distributed. Brief descriptions of several of these methods and their properties are in § 3. conjugate gradient and GMRES. Ak−1 r0 ) for k ≥ 1. The two such methods that we’ll discuss in depth. The residual is r = b − Ax. Recall that A is symmetric if A = AT and positive deﬁnite if xT Ax > 0 for all x = 0. where x0 is the initial iterate and the kth Krylov subspace Kk is Kk = span(r0 . some measure of error over the aﬃne space x0 + Kk . . There are other Krylov methods that are not as well understood as CG or GMRES. It has come into wide use over the last 15 years as an iterative method and has generally superseded the Jacobi–Gauss–Seidel–SOR family of methods. Chapter 2 Conjugate Gradient Iteration 2. Ar0 . Krylov methods do not have an iteration matrix. .Copyright ©1995 by the Society for Industrial and Applied Mathematics. So {rk }k≥0 will denote the sequence of residuals rk = b − Axk . Krylov methods and the minimization property In the following two chapters we describe some of the Krylov space methods for linear equations. and [78]. 11 Buy this book from SIAM at http://www. minimize. CG is intended to solve symmetric positive deﬁnite (spd) systems. [12].1. As in Chapter 1. .
ecsecurehost.1. Since A is spd we may deﬁne a norm (you should check that this is a norm) by (2. Since (x∗ )T Ax∗ is independent of x. we begin with a description of what the algorithm does and the consequences of the minimization property of the iterates. preconditioning. Since A is symmetric and Ax∗ = b −xT Ax∗ − (x∗ )T Ax = −2xT Ax∗ = −2xT b. This electronic version is for personal use and may not be duplicated or distributed. The development in these notes is diﬀerent from the classical work and more like the analysis for GMRES and CGNR in [134]. the implementation.com/SIAM/FR16. A If e = x − x∗ then e 2 A = eT Ae = (A(x − x∗ ))T A−1 (A(x − x∗ )) = b − Ax 2 A−1 and hence the Anorm of the error is also the A−1 norm of the residual.2) 1 φ(x) = xT Ax − xT b 2 over x0 + Kk . · A is called the Anorm.html. performance. Therefore x − x∗ 2 A = 2φ(x) + (x∗ )T Ax∗ . Note that if φ(˜) is the minimal value (in RN ) then x ∇φ(˜) = A˜ − b = 0 x x and hence x = x∗ .1. Buy this book from SIAM at http://www. . The kth iterate xk of CG minimizes (2. We will use this lemma in the particular case that S = x0 + Kk for some k. We state this as a lemma. Let S ⊂ RN . After that we describe termination criterion. Note that x − x∗ 2 A = (x − x∗ )T A(x − x∗ ) = xT Ax − xT Ax∗ − (x∗ )T Ax + (x∗ )T Ax∗ .1) x A = √ xT Ax. and in the section on GMRES that follows. In this section. Lemma 2. minimizing φ is equivalent to minimizing x − x∗ 2 and hence to minimizing x − x∗ A . If xk minimizes φ over S then xk also minimizes x∗ − x A = r A−1 over S. ˜ Minimizing φ over any subset of RN is the same as minimizing x − x∗ A over that subset. Proof.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 12 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS In this section we assume that A is spd. and at the very end.
Since any w ∈ x0 + Kk can be written as w= k−1 j=0 γj Aj r0 + x0 for some coeﬃcients {γj }.Copyright ©1995 by the Society for Industrial and Applied Mathematics. . where U is an orthogonal matrix whose columns are the eigenvectors of A and Λ is a diagonal matrix with the positive eigenvalues of A on the diagonal.4) x∗ − xk A = p∈Pk . we have Aj = U Λj U T . 2 Buy this book from SIAM at http://www. Hence p(A) = U p(Λ)U T . 2 A Deﬁne A1/2 = U Λ1/2 U T and note that (2. we can express x∗ − w as x∗ − w = x∗ − x0 − Since Ax∗ = b we have r0 = b − Ax0 = A(x∗ − x0 ) and therefore x∗ − w = x∗ − x0 − where the polynomial p(z) = 1 − k−1 j=0 k−1 j=0 k−1 j=0 γj Aj r0 . CONJUGATE GRADIENT ITERATION 13 2.4) Pk denotes the set of polynomials of degree k.1.p(0)=1 min p(A)(x∗ − x0 ) A. The spectral theorem for spd matrices asserts that A = U ΛU T .html. γj z j+1 has degree k and satisﬁes p(0) = 1. In (2. This electronic version is for personal use and may not be duplicated or distributed.ecsecurehost.3) x∗ − xk A ≤ x∗ − w A for all w ∈ x0 + Kk .2. γj Aj+1 (x∗ − x0 ) = p(A)(x∗ − x0 ). Hence (2. Consequences of the minimization property Lemma 2. Since U U T = U T U = I by orthogonality of U .5) x = xT Ax = A1/2 x 2 .com/SIAM/FR16.1 implies that since xk minimizes φ over x0 + Kk (2.
since in most applications the number of unknowns N is very large.1.8) Pk = {p  p is a polynomial of degree k and p(0) = 1.7) easy to evaluate. Proof. This leads to an upper estimate for the number of CG iterations required to reduce the Anorm of the error to a given tolerance.html. based on information on σ(A). Hence. Then the CG algorithm will ﬁnd the solution within N iterations. In that way we showed (in the absence of all roundoﬀ error!) that CG terminated in ﬁnitely many iterations with the exact solution.1. Let A be spd and let {xk } be the CG iterates.2.com/SIAM/FR16. 14 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Hence.2.4) implies that (2.7) xk − x∗ x0 − x∗ A A ≤ max ¯k (z). The set of kth degree residual polynomials is (2. This is not as good as it sounds. ¯ ¯ p ∈ PN because p has degree N and p(0) = 1. Buy this book from SIAM at http://www. together with (2. Theorem 2. Let k be given and let {¯k } be any kth degree polynomial such that pk (0) = 1. Corollary 2. Then p ¯ (2. ¯ Definition 2.6).ecsecurehost. and one cannot aﬀord to perform N iterations. Let A be spd. by (2.7) and the fact ¯ that p vanishes on σ(A). p Note that our test polynomial had the eigenvalues of A as its roots. p z∈σ(A) We will refer to the polynomial pk as a residual polynomial [185]. let i=1 p(z) = ¯ N i=1 (λi − z)/λi . Let {λi }N be the eigenvalues of A. As a test polynomial.6) xk − x∗ A ≤ x0 − x∗ A p∈Pk . for any x ∈ RN and p(A)x A = A1/2 p(A)x 2 ≤ p(A) 2 A1/2 x 2 ≤ p(A) 2 x A. . When doing that we seek to terminate the iteration when some speciﬁed error tolerance is reached.} In speciﬁc contexts we try to construct sequences of residual polynomials. This electronic version is for personal use and may not be duplicated or distributed. One simple application of (2.p(0)=1 z∈σ(A) min max p(z).Copyright ©1995 by the Society for Industrial and Applied Mathematics.2. This. The following corollary is an important consequence of (2. ¯ xN − x∗ A ≤ x0 − x∗ A z∈σ(A) max ¯(z) = 0. It is best to regard CG as an iterative method.1.7) is to show how the CG algorithm can be viewed as a direct method. Here σ(A) is the set of all eigenvalues of A. that make either the middle or the right term in (2.
com/SIAM/FR16.3.9) b − Axk 2 ≤ η b 2 . CONJUGATE GRADIENT ITERATION 15 In the two examples that follow we look at some other easy consequences of (2. ¯ One can easily verify that p ∈ Pk . Buy this book from SIAM at http://www.2.3. Let {λil } be the eigenvalues of A associated with the eigenvectors {uil }k . .4) and the fact that x0 = 0 that xk − x∗ A ≤ p(A)x∗ ¯ A = 0. One typical criterion is small (say ≤ η) relative residuals. Termination of the iteration In practice we do not run the CG iteration until an exact solution is found.7). Let b be a linear i=1 combination of k of the eigenvectors of A b= k l=1 γl uil .2. but rather terminate once some criterion has been satisﬁed. Moreover.html.ecsecurehost. Theorem 2. we can use a similar technique to prove the following theorem. Then the CG iteration for Ax = b with x0 = 0 will terminate in at most k iterations. however. This electronic version is for personal use and may not be duplicated or distributed. are based on (2. Assume that there are exactly k ≤ N distinct eigenvalues of A. Theorem 2. Then the CG iteration terminates in at most k iterations. The error estimates that come from the minimization property.7) and therefore estimate the reduction in the relative Anorm of the error. 2.Copyright ©1995 by the Society for Industrial and Applied Mathematics. By the spectral theorem l=1 x∗ = k l=1 (γl /λil )uil . ¯ So. Let A be spd with eigenvectors {ui }N .2. If the spectrum of A has fewer than N points. We use the residual polynomial. p(z) = ¯ k l=1 (λil − z)/λil . p(λil ) = 0 for 1 ≤ l ≤ k and ¯ hence p(A)x∗ = ¯ k l=1 p(λil )γl /λil uil = 0. Let A be spd. we have by (2. Proof. This means that we terminate the iteration after (2. This completes the proof.
12).3.10) and (2.7) to predict when the relative Buy this book from SIAM at http://www. Lemma 2. using (2. . Then for all z ∈ RN . Let A be spd with eigenvalues λ1 ≥ λ2 ≥ .12) and (2. This electronic version is for personal use and may not be duplicated or distributed. to predict the performance of the CG iteration based on termination on small relative residuals. Clearly z 2 A λN 1/2 z A ≤ Az 2 ≤ λ1 1/2 z A. A1/2 z 2 = z A (2.13) b r0 2 2 b − Axk b 2 b − Axk b 2 2 = b − Axk b − Ax0 2 2 ≤ κ2 (A) xk − x∗ x∗ − x0 xk − x∗ x∗ − x0 A A A A 2 ≤ κ2 (A) r0 b 2 2 .12).ecsecurehost. Lemma 2.2. (2. . So. We may write A = U ΛU T as Az = Hence λN A1/2 z 2 2 N i=1 λi (uT z)ui .com/SIAM/FR16. .3.html. ﬁrst recall that if A = U ΛU T is the spectral decomposition of A and we order the eigenvalues such that λ1 ≥ λ2 ≥ .1. we must not only use (2.10) complete the proof.12) is clear and (2. We will do this in the next two lemmas and then illustrate the point with an example. Let ui be a unit eigenvector corresponding to λi . Therefore. The equality on the left of (2. i = λN ≤ Az ≤ λ1 N T 2 i=1 λi (ui z) 2 2 = N 2 T 2 i=1 λi (ui z) N T 2 i=1 λi (ui z) = λ1 A1/2 z 2 . λN . = z T Az = (A1/2 z)T (A1/2 z) = A1/2 z 2 2 which proves (2. .10) and (2.11) twice.13) follows directly from (2. So κ2 (A) = λ1 /λN .10). λN > 0. 2 Taking square roots and using (2. Proof. 16 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Our next task is to relate the relative residual in the Euclidean norm to the relative error in the Anorm. . then A 2 = λ1 and A−1 2 = 1/λN .Copyright ©1995 by the Society for Industrial and Applied Mathematics. To obtain the inequality on the right of (2.11) Proof. b − Axk b − Ax0 2 2 = A(x∗ − xk ) A(x∗ − x0 ) 2 2 ≤ λ1 x∗ − xk λN x∗ − x0 A A as asserted.
7).2 we simply note that κ2 (A) ≤ 11/9. In the case of the above example. Hence. after k iterations (2.3. One can obtain a more precise estimate by using a polynomial other than pk in the upper estimate for the righthand side of (2. ≤√ κ2 (A) + 1 11 + 3 Buy this book from SIAM at http://www. when k ≥ 3. Note that it is always the case that the spectrum of a spd matrix is contained in the interval [λN . This means that we may apply (2. ∞). A result from [48] (see also [45]) that is. √ κ2 (A) − 1 11 − 3 ≈ . that is. To use Lemma 2. Assume that x0 = 0 and that the eigenvalues of A are contained in the interval (9. that is.15) xk − x ∗ A ≤ 2 x0 − x ∗ A κ2 (A) − 1 κ2 (A) + 1 k . b 2 So. but also use Lemma 2. the sharpest possible. p ≤ x∗ A 10 −k . 11). If we let pk (z) = ¯ (10 − z)k /10k .14) xk − x∗ A A ≤ x∗ A 9≤z≤11 max ¯k (z). then pk ∈ Pk .html. √ √ Hence. in one sense. after k iterations we have Axk − b 2 √ ≤ 11 × 10−k /3. p 9≤z≤11 max ¯k (z) = 10−k . since ( x − 1)/( x + 1) is an increasing function of x on the interval (1.05. is (2.7) to get ¯ xk − x∗ It is easy to see that Hence.3. So. This electronic version is for personal use and may not be duplicated or distributed. the size of the Anorm of the error will be reduced by a factor of 10−3 when 10−k ≤ 10−3 .ecsecurehost. .com/SIAM/FR16. CONJUGATE GRADIENT ITERATION 17 Anorm error is small. when k ≥ 4.2 to relate small Anorm errors to small relative residuals. We consider a very simple example. λ1 ] and that κ2 (A) = λ1 /λN .Copyright ©1995 by the Society for Industrial and Applied Mathematics. we can estimate κ2 (A) by κ2 (A) ≤ 11/9. the size of the relative residual will be reduced by a factor of 10−3 when √ 10−k ≤ 3 × 10−3 / 11.
Exercise 2. This would indicate fairly slow convergence. When we do. the condition number can be quite large. the estimate in (2.5 also covers this point. (2.com/SIAM/FR16.15) can be very pessimistic.05) ≈ 3. The estimate based on the clustering gives convergence in 3k iterations when (.91) ≈ 3.Copyright ©1995 by the Society for Industrial and Applied Mathematics. when inserted into (2.25/1. .9).25)k × 4002k max ¯3k (z) ≤ (.3.5. In fact. if we use as a residual polynomial p3k ∈ P3k ¯ p3k (z) = ¯ It is easy to see that z∈σ(A) (1. Even if the condition number Buy this book from SIAM at http://www. However. but CG can perform very well.8. Hence (2. 18 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Therefore (2. Assume that x0 = 0 and the eigenvalues of A lie in the two intervals (1. This electronic version is for personal use and may not be duplicated or distributed. 1. the CG iteration will converge very rapidly.15) would predict a reduction in the size of the Anorm error by a factor of 10−3 when 2 × .2) = 4.ecsecurehost.3 = 12.91)k < 10−3 or when k > − log10 (2000)/ log10 (. p which is a sharper estimate on convergence.2)k .5) and (399. From the results above one can see that if the condition number of A is near one.91)k .6.04 = 82. which also predicts termination within three iterations.3/.2)k ≤ 10−3 or when k > −3/ log10 (.15) would predict that xk − x∗ when 2 × (.3 ≈ 2. which. (1.15) predicts 83 iterations and the clustering analysis 15 (the smallest integer multiple of 3 larger than 3 × 4.15) gives xk − x∗ x∗ A A ≤ 2 × (19/21)k ≈ 2 × (. We may have more precise information than a single interval containing σ(A). If the eigenvalues cluster in a small number of intervals.3/1. A ≤ 10−3 x∗ A.25)k = (.25 − z)k (400 − z)2k . Based on this information the best estimate of the condition number of A is κ2 (A) ≤ 400. We will illustrate this with an example. 400).05k < 10−3 or when k > − log10 (2000)/ log10 (.html.
In the context of CG. We show that this only happens if xk is the solution. (2.Copyright ©1995 by the Society for Industrial and Applied Mathematics. we have. for any ξ ∈ Kk dφ(xk + tξ) = ∇φ(xk + tξ)T ξ = 0 dt at t = 0.18) Proof. pT Apk+1 pk+1 Apk+1 k+1 If xk = xk+1 then the above analysis implies that α = 0. unless the l iteration terminates prematurely.16) =0 dα for the correct choice of α = αk+1 . Let A be spd and let {xk } be the conjugate gradient iterates.1). 2 The next lemma characterizes the search direction and.1 then implies that T T rk 2 = rk rk = rk rk+1 = 0 and hence xk = x∗ . .17) αk+1 = pT (b − Axk ) p T rk k+1 = T k+1 . Then T rk rl = 0 for all 0 ≤ l < k. Buy this book from SIAM at http://www. Equation (2. Implementation The implementation of CG depends on the amazing fact that once xk has been determined. such a simple approach can destroy the symmetry of the coeﬃcient matrix and a more subtle implementation is required. 2.5. Recalling that ∇φ(x) = Ax − b = −r we have (2.19) T ∇φ(xk )T ξ = −rk ξ = 0 for all ξ ∈ Kk . Since rl ∈ Kk for all l < k (see Exercise 2. Since xk minimizes φ on x0 + Kk . if xk = xk+1 . This electronic version is for personal use and may not be duplicated or distributed. either xk = x∗ or a search direction pk+1 = 0 can be found very cheaply so that xk+1 = xk + αk+1 pk+1 for some scalar αk+1 .8. this proves (2.com/SIAM/FR16. Lemma 2.4.18). as a side eﬀect.. easier to solve) is called preconditioning. The transformation of the problem into one with eigenvalues clustered near one (i. In fact dφ(xk + αpk+1 ) (2. then rk = rk+1 . Now.1. Once pk+1 has been found.4.html. We discuss this in § 2. Lemma 2.4.ecsecurehost. CONJUGATE GRADIENT ITERATION 19 is large. proves that (if we deﬁne p0 = 0) pT rk = 0 for all 0 ≤ l < k ≤ n. αk+1 is easy to compute from the minimization property of the iteration. the iteration will perform well if the eigenvalues are clustered in a few small intervals.e.16) can be written as pT Axk + αpT Apk+1 − pT b = 0 k+1 k+1 k+1 leading to (2. We used this term before in the context of Richardson iteration and accomplished the goal by multiplying A by an approximate inverse.
4.4.com/SIAM/FR16.2 and the fact that Kk = span(r0 . Then pk+1 = rk + βk+1 pk for some βk+1 and k ≥ 0. pT Aξ = 0 for all ξ ∈ Kk . Deﬁne p0 = 0. then rl ∈ Kl+1 ⊂ Kk−1 . Now.1. Theorem 2.e. a subspace of dimension one less than Kk+1 . .Copyright ©1995 by the Society for Industrial and Applied Mathematics. Since k rk = rk−1 − αk Apk Buy this book from SIAM at http://www. We have the following theorem. The condition pT Aξ = 0 is called Aconjugacy of pk+1 to Kk . in the scalar product (x.23) Proof. Lemma 2. Let pk+1 be given by (2.24) T βk+1 = −rk Ark−1 /pT Ark−1 k provided pT Ark−1 = 0.ecsecurehost. up to a scalar multiple.. Let A be spd and let {xk } be the conjugate gradient iterates.19) and (2. . Let A be spd and assume that rk = 0. Trivially k+1 (2. in fact. While one might think that wk would be hard to compute. y) = xT Ay) to Kk . it is. . (2. (2. Then for any l ≤ k T pT Arl = rk Arl + βk+1 pT Arl .4. k+1 This uniquely speciﬁes the direction of pk+1 as (2.20) pk+1 ∈ Kk+1 . This electronic version is for personal use and may not be duplicated or distributed. k+1 Proof.22) αk+1 pT Aξ = −(Axk − b)T ξ = −∇φ(xk )T ξ = 0.21) then imply that for all ξ ∈ Kk . rk−1 ). .4. By Lemma 2. be expressed as pk+1 = rk + wk with wk ∈ Kk . (2. Since Kk ⊂ Kk+1 . k+1 k If l ≤ k − 2. we need only verify that a βk+1 can be found so that if pk+1 is given by (2.23) then pT Arl = 0 k+1 for all 0 ≤ l ≤ k − 1. (2. any k+1 pk+1 satisfying (2.20) can. trivial.23).2. . k+1 It only remains to solve for βk+1 so that pT Ark−1 = 0.21) ∇φ(xk+1 )T ξ = (Axk + αk+1 Apk+1 − b)T ξ = 0 for all ξ ∈ Kk .2 then implies that pT Arl = 0 for 0 ≤ l ≤ k − 2. If xk = x∗ then xk+1 = xk + αk+1 pk+1 and pk+1 is determined up to a scalar multiple by the conditions (2. 20 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Lemma 2.html.22) implies that pk+1 ∈ Kk+1 is Aorthogonal (i.
30). Let A be spd and assume that rk = 0. k+1 which is equivalent to (2.4. Note that for k ≥ 0 (2. To get (2.ecsecurehost. CONJUGATE GRADIENT ITERATION 21 2 2 we have T rk rk−1 = rk−1 − αk pT Ark−1 .com/SIAM/FR16. An immediate consequence of (2. k k k βk+1 = T −rk Apk . This electronic version is for personal use and may not be duplicated or distributed. (2. k T Since rk rk−1 = 0 by Lemma 2.31). The common implementation of conjugate gradient uses a diﬀerent form for αk and βk than given in (2.27) note that pT Apk = 0 and hence (2.28) T pT rk+1 = rk rk+1 + βk+1 pT rk+1 = 0 k+1 k αk+1 = rk 2 2 pT Apk+1 k+1 rk rk−1 2 2 2. (2.17) and (2.25) to get βk+1 = T −rk Apk αk . 2 βk+1 = by Lemma 2.28) is that pT rk = 0 and k hence pT rk = (rk + βk+1 pk )T rk = rk 2 .27) Proof.26).html.2.1 we have (2. . rk−1 2 2 Buy this book from SIAM at http://www.29) k+1 2 Taking scalar products of both sides of rk+1 = rk − αk+1 Apk+1 with pk+1 and using (2.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Then (2. Lemma 2.3.29) gives T 0 = pT rk − αk+1 pT Apk+1 = rk k+1 k+1 2 2 − αk+1 pT Apk+1 .23) implies that k+1 (2.26) and (2.25) pT Ark−1 = rk−1 2 /αk = 0.4.4.30) Also note that (2.31) pT Apk = pT A(rk−1 + βk pk−1 ) k k = pT Ark−1 + βk pT Apk−1 = pT Ark−1 .24). pT Apk k Now combine (2. k 2 This completes the proof. and (2.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
22
ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Now take scalar products of both sides of rk = rk−1 − αk Apk with rk and use Lemma 2.4.1 to get rk
2 2 T = −αk rk Apk .
Hence (2.27) holds. The usual implementation reﬂects all of the above results. The goal is to ﬁnd, for a given , a vector x so that b−Ax 2 ≤ b 2 . The input is the initial iterate x, which is overwritten with the solution, the right hand side b, and a routine which computes the action of A on a vector. We limit the number of iterations to kmax and return the solution, which overwrites the initial iterate x and the residual norm. Algorithm 2.4.1. cg(x, b, A, , kmax) 1. r = b − Ax, ρ0 = r 2 , k = 1. 2 √ 2. Do While ρk−1 > b 2 and k < kmax (a) if k = 1 then p = r else β = ρk−1 /ρk−2 and p = r + βp (b) w = Ap (c) α = ρk−1 /pT w (d) x = x + αp (e) r = r − αw (f) ρk = r
2 2
(g) k = k + 1 Note that the matrix A itself need not be formed or stored, only a routine for matrixvector products is required. Krylov space methods are often called matrixfree for that reason. Now, consider the costs. We need store only the four vectors x, w, p, and r. Each iteration requires a single matrixvector product (to compute w = Ap), two scalar products (one for pT w and one to compute ρk = r 2 ), and three 2 operations of the form ax + y, where x and y are vectors and a is a scalar. It is remarkable that the iteration can progress without storing a basis for the entire Krylov subspace. As we will see in the section on GMRES, this is not the case in general. The spd structure buys quite a lot. 2.5. Preconditioning To reduce the condition number, and hence improve the performance of the iteration, one might try to replace Ax = b by another spd system with the same solution. If M is a spd matrix that is close to A−1 , then the eigenvalues
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
CONJUGATE GRADIENT ITERATION
23
of M A will be clustered near one. However M A is unlikely to be spd, and hence CG cannot be applied to the system M Ax = M b. In theory one avoids this diﬃculty by expressing the preconditioned problem in terms of B, where B is spd, A = B 2 , and by using a twosided preconditioner, S ≈ B −1 (so M = S 2 ). Then the matrix SAS is spd and its eigenvalues are clustered near one. Moreover the preconditioned system SASy = Sb has y ∗ = S −1 x∗ as a solution, where Ax∗ = b. Hence x∗ can be recovered from y ∗ by multiplication by S. One might think, therefore, that computing S (or a subroutine for its action on a vector) would be necessary and that a matrixvector multiply by SAS would incur a cost of one multiplication by A and two by S. Fortunately, this is not the case. If y k , rk , pk are the iterate, residual, and search direction for CG applied ˆ ˆ to SAS and we let ˆ ˆ ˆ ˆ xk = S y k , rk = S −1 rk , pk = S pk , and zk = S rk , then one can perform the iteration directly in terms of xk , A, and M . The reader should verify that the following algorithm does exactly that. The input is the same as that for Algorithm cg and the routine to compute the action of the preconditioner on a vector. Aside from the preconditioner, the arguments to pcg are the same as those to Algorithm cg. Algorithm 2.5.1. pcg(x, b, A, M, , kmax) 1. r = b − Ax, ρ0 = r 2 , k = 1 2 √ 2. Do While ρk−1 > b 2 and k < kmax (a) z = M r (b) τk−1 = z T r (c) if k = 1 then β = 0 and p = z else β = τk−1 /τk−2 , p = z + βp (d) w = Ap (e) α = τk−1 /pT w (f) x = x + αp (g) r = r − αw (h) ρk = rT r (i) k = k + 1 Note that the cost is identical to CG with the addition of • the application of the preconditioner M in step 2a and • the additional inner product required to compute τk in step 2b.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
24
ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Of these costs, the application of the preconditioner is usually the larger. In the remainder of this section we brieﬂy mention some classes of preconditioners. A more complete and detailed discussion of preconditioners is in [8] and a concise survey with many pointers to the literature is in [12]. Some eﬀective preconditioners are based on deep insight into the structure of the problem. See [124] for an example in the context of partial diﬀerential equations, where it is shown that certain discretized secondorder elliptic problems on simple geometries can be very well preconditioned with fast Poisson solvers [99], [188], and [187]. Similar performance can be obtained from multigrid [99], domain decomposition, [38], [39], [40], and alternating direction preconditioners [8], [149], [193], [194]. We use a Poisson solver preconditioner in the examples in § 2.7 and § 3.7 as well as for nonlinear problems in § 6.4.2 and § 8.4.2. One commonly used and easily implemented preconditioner is Jacobi preconditioning, where M is the inverse of the diagonal part of A. One can also use other preconditioners based on the classical stationary iterative methods, such as the symmetric Gauss–Seidel preconditioner (1.18). For applications to partial diﬀerential equations, these preconditioners may be somewhat useful, but should not be expected to have dramatic eﬀects. Another approach is to apply a sparse Cholesky factorization to the matrix A (thereby giving up a fully matrixfree formulation) and discarding small elements of the factors and/or allowing only a ﬁxed amount of storage for the factors. Such preconditioners are called incomplete factorization preconditioners. So if A = LLT + E, where E is small, the preconditioner is (LLT )−1 and its action on a vector is done by two sparse triangular solves. We refer the reader to [8], [127], and [44] for more detail. One could also attempt to estimate the spectrum of A, ﬁnd a polynomial p such that 1 − zp(z) is small on the approximate spectrum, and use p(A) as a preconditioner. This is called polynomial preconditioning. The preconditioned system is p(A)Ax = p(A)b and we would expect the spectrum of p(A)A to be more clustered near z = 1 than that of A. If an interval containing the spectrum can be found, the residual polynomial q(z) = 1 − zp(z) of smallest L∞ norm on that interval can be expressed in terms of Chebyshev [161] polynomials. Alternatively q can be selected to solve a least squares minimization problem [5], [163]. The preconditioning p can be directly recovered from q and convergence rate estimates made. This technique is used to prove the estimate (2.15), for example. The cost of such a preconditioner, if a polynomial of degree K is used, is K matrixvector products for each application of the preconditioner [5]. The performance gains can be very signiﬁcant and the implementation is matrixfree.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
the analysis for CGNE is essentially the same.32) AT Ax = AT b. [134]. There are three disadvantages that may or may not be serious. . [134]. The ﬁrst is that the condition number of the coeﬃcient matrix AT A is the square of that of A. The second is that two matrixvector products are needed for each CG iterate since w = AT Ap = AT (Ap) in CGNR and w = AAT p = A(AT p) in CGNE.html. The advantages of this approach are that all the theory for CG carries over and the simple implementation for both CG and PCG can be used. ¯ (2.33) AAT y = b and then set x = AT y to solve Ax = b. The third. 2 Hence. one might consider solving Ax = b by applying CG to the normal equations (2. p Buy this book from SIAM at http://www. The reason for this name is that the minimization property of CG as applied to (2. We consider the case for CGNR. there are situations where this is not possible.Copyright ©1995 by the Society for Industrial and Applied Mathematics.com/SIAM/FR16.ecsecurehost. more important. when we consider the AT A norm of the error. one could solve (2. CGNR and CGNE If A is nonsingular and nonsymmetric. The reason for this name is that the minimization property of CG as applied to (2. The analysis with residual polynomials is similar to that for CG. we have x∗ − x 2 AT A = (x∗ − x)T AT A(x∗ − x) = A(x∗ − x) 2 2 = r 2. As above. This approach [103] is called CGNR [71].33) then y∗ − y 2 AAT = (y ∗ − y)T (AAT )(y ∗ − y) = (AT y ∗ − AT y)T (AT y ∗ − AT y) = x∗ − x 2 2 is minimized over y0 + Kk at each iterate. disadvantage is that one must compute the action of AT on a vector as part of the matrixvector product involving AT A.34) rk 2 ≤ pk (AT A)r0 ¯ 2 ≤ r0 2 z∈σ(AT A) max ¯k (z). This approach [46] is now called CGNE [78]. for any residual polynomial pk ∈ Pk . CONJUGATE GRADIENT ITERATION 25 2. Hence the name Conjugate Gradient on the Normal equations to minimize the Residual. Alternatively. [78]. This electronic version is for personal use and may not be duplicated or distributed.33) asserts that if y ∗ is the solution to (2. As we will see in the chapter on nonlinear problems. Conjugate Gradient on the Normal equations to minimize the Error.32) asserts that x∗ − x 2 AT A = (x∗ − x)T AT A(x∗ − x) = (Ax∗ − Ax)T (Ax∗ − Ax) = (b − Ax)T (b − Ax) = r 2 is minimized over x0 + Kk at each iterate.6.
7). y < 1 subject to homogeneous Dirichlet boundary conditions u(x.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 1) = u(0. The inputs.8. described in the comment lines.2.34) and (2. the right hand side vector b. Examples for preconditioned conjugate iteration In the collection of MATLAB codes we provide a code for preconditioned conjugate gradient iteration. We express the discrete matrixvector product as (Au)ij (2. The unknowns are uij ≈ u(xi . Hence the performance of CGNR and CGNE is determined by the distribution of singular values. y)∇u) = f (x. The estimate is in terms of the l2 norm of the residual.com/SIAM/FR16.3. y < 1. to reﬂect the boundary conditions. 2. which is the set of the squares of the singular values of A. the residual polynomial is to be maximized over the eigenvalues of AT A. and deﬁne αij = −a(xi . One can verify [105] that the diﬀerential operator is positive deﬁnite in the Hilbert space sense and that the ﬁvepoint discretization described below is positive deﬁnite if a > 0 for all 0 ≤ x. .ecsecurehost. x0 . 0 < x. y) on 0 < x. and iteration parameters to specify the maximum number of iterations and the termination criterion. This electronic version is for personal use and may not be duplicated or distributed. are the initial iterate.7. We consider the discretization of the partial diﬀerential equation (2.html. Most signiﬁcantly. We discretize with a ﬁvepoint centered diﬀerence scheme with n2 points and mesh width h = 1/(n + 1).10).36) = (αij + α(i+1)j )(u(i+1)j − uij ) −(α(i−1)j + αij )(uij − u(i−1)j ) + (αi(j+1) + αij )(ui(j+1) − uij ) −(αij + αi(j−1) )(uij − ui(j−1) ) Buy this book from SIAM at http://www. y ≤ 1 (Exercise 2. y) = 0. xj )h−2 /2. MATLAB functions for the matrixvector product and (optionally) the preconditioner. 0) = u(x. y) = u(1. 26 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS There are two major diﬀerences between (2.35) ˙ −∇(a(x. On return the code supplies the approximate solution x and the history of the iteration as the vector of residual norms. hence we need not prove a result like Lemma 2. We set u0j = u(n+1)j = ui0 = ui(n+1) = 0. which corresponds exactly to the termination criterion. xj ) where xi = ih for 1 ≤ i ≤ n.
[186].5 ).1 the solid line is a plot of rk / b and the dashed line a plot of u∗ −uk A / u∗ −u0 A .8.9.ecsecurehost.Copyright ©1995 by the Society for Industrial and Applied Mathematics. which is what pcgsol expects to see. For the MATLAB implementation we convert freely from the representation of u as a twodimensional array (with the boundary conditions added). .com/SIAM/FR16. This electronic version is for personal use and may not be duplicated or distributed. Note that the reduction in r is not monotone. plotting rk 2 / b 2 on a semilog scale. The reader is asked to quantify this in terms of execution times in Exercise 2. Note that the unpreconditioned iteration is slowly convergent. See the routine fish2d in collection of MATLAB codes for an example of how to do this in MATLAB. subject to homogeneous Dirichlet boundary conditions. designed to take advantage of fast MATLAB builtin functions. The initial iterate was u0 = 0. Because of this their execution time is less than a simple count of ﬂoatingpoint operations would indicate. The eﬀectiveness of such a preconditioner has been analyzed in [124] and some of the many ways to implement the solver eﬃciently are discussed in [99].html.15) indicates that convergence will be slow. which predicts decrease in e A but not necessarily in r as the iteration progresses. is more eﬃcient that the classical stationary iterative methods. By this we mean an operator G such that v = Gw is the solution of the discrete form of −vxx − vyy = w. This example illustrates the importance of a good preconditioner. [188]. For many types of domains and boundary conditions. in the case of the MATLAB environment used in this book. The properties of CG on the preconditioned problem in the continuous case have been analyzed in [48]. CONJUGATE GRADIENT ITERATION 27 for 1 ≤ i. and [187]. We expect secondorder accuracy from the method and accordingly we set termination parameter = h2 = 1/1024. y) = cos(x) and took the right hand side so that the exact solution was the discretization of 10xy(1 − x)(1 − y) exp(x4. which is useful for computing the action of A on u and applying fast solvers. We will report our results graphically. We allowed up to 100 CG iterations. j ≤ n. This can be explained by the fact that the eigenvalues are not clustered and κ(A) = O(1/h2 ) = O(n2 ) = O(N ) and hence (2. and the representation as a onedimensional array. For a preconditioner we use a Poisson solver. The initial iterate was the zero vector. The fast Poisson solver in the collection of Buy this book from SIAM at http://www. This is consistent with the theory. In the results reported here we took n = 31 resulting in a system with N = n2 = 961 unknowns. Poisson solvers can be designed to take advantage of vector and/or parallel architectures or. In Fig. For the computations reported in this section we took a(x. however. 2. Even the unpreconditioned iteration.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. 2. This electronic version is for personal use and may not be duplicated or distributed. but the number of iterations required to reduce the relative residual by a given amount is independent of the mesh spacing [124]. 10 1 10 0 Relative Residual 10 1 10 2 10 3 10 4 0 10 20 30 Iterations 40 50 60 Fig.1.2 the solid line is the graph of rk 2 / b 2 for the preconditioned iteration and the dashed line for the unpreconditioned. codes fish2d is based on the MATLAB fast Fourier transform.ecsecurehost. PCG for 2D elliptic equation. We caution the reader that the preconditioned iteration is not as much faster than the Buy this book from SIAM at http://www. the builtin function fft. 2. The preconditioned iteration required 5 iterations for convergence and the unpreconditioned iteration 52. 2. Not only does the preconditioned iteration converge more rapidly.com/SIAM/FR16.html.2. In Fig. CG for 2D elliptic equation. 28 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 1 10 Relative Residual and Anorm of Error 10 0 10 1 10 2 10 3 10 4 0 10 20 30 Iterations 40 50 60 Fig. .
ecsecurehost. .com/SIAM/FR16.2 million ﬂoatingpoint operations while the preconditioned iteration required .87 million ﬂoatingpoint operations.11 you are asked to compare execution times for your own environment. In the MATLAB environment we used. As we remarked above. Hence.html. the execution time of the preconditioned iteration was about 60% of that of the unpreconditioned. Buy this book from SIAM at http://www. CONJUGATE GRADIENT ITERATION 29 unpreconditioned one as the iteration count would suggest. This electronic version is for personal use and may not be duplicated or distributed. this speed is a result of the eﬃciency of the MATLAB fast Fourier transform. In Exercise 2.8.Copyright ©1995 by the Society for Industrial and Applied Mathematics. the cost of the preconditioner is considerable. The MATLAB flops command indicates that the unpreconditioned iteration required roughly 1.
8. 2.2. Prove Theorem 2. Assume that A is spd and that σ(A) ⊂ (1. 2. y)? Try a(x. nonsingular.1 + x and examine the accuracy of the result. 30 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 2.3. give a rough estimate of the number of CG iterates required to reduce the relative residual to O(1/N ). Let A be a nonsingular matrix with all singular values in the interval (1.6) for the number of CG iterations required to reduce the A norm of the error by a factor of 10−3 and for the number of CG iterations required to reduce the residual by a factor of 10−3 . Compare the execution times on your computing environment (using the cputime command in MATLAB. Explain your ﬁndings. Show that if A has constant diagonal then PCG with Jacobi preconditioning produces the same iterates as CG with no preconditioning.7. . Give upper estimates based on (2. assume that the cost of a matrix vector multiply is 4N ﬂoatingpoint multiplies. Let A be spd. For the matrix A in problem 5. and spd.9.8.10. Prove that rl ∈ Kk for all l < k.com/SIAM/FR16. Duplicate the results in § 2. y) = .8.1) ∪ (2.5.8. Is B unique? 2.4. 2. Prove that the linear transformation given by (2. What happens as N is increased? How are the performance √ and accuracy aﬀected by changes in a(x. y) > 0 for all 0 ≤ x.8. Buy this book from SIAM at http://www. 2.8. 2. Estimate the number of CGNR/CGNE iterations required to reduce the relative residual by a factor of 10−4 .html.8. Assume that A is N × N .11. Let {xk } be the conjugate gradient iterates.Copyright ©1995 by the Society for Industrial and Applied Mathematics.8. 2. Show that there is a spd B such that B 2 = A. Prove that p(Λ) = maxi p(λi ) where · is any induced matrix norm.8. for instance). This electronic version is for personal use and may not be duplicated or distributed. 2). y ≤ 1.8.3.8.7 for example. Exercises on conjugate gradient 2. 2.6.8.2.ecsecurehost.36) is symmetric and 2 positive deﬁnite on Rn if a(x. If κ(A) = O(N ). in MATLAB by writing the matrixvector product routines and using the MATLAB codes pcgsol and fish2d.8.2). Estimate the number of ﬂoatingpoint operations reduce the A norm of the error by a factor of 10−3 using CG iteration. 2. Let Λ be a diagonal matrix with Λii = λi and let p be a polynomial. 2. 1.1. 2.
y) = 1. 2. 0 < x. Use the Jacobi and symmetric Gauss–Seidel iterations from Chapter 1 to solve the elliptic boundary value problem in § 2. Modify pcgsol so that φ(x) is computed and stored at each iterate and returned on output. Experiment with diﬀerent mesh sizes and preconditioners (Fast Poisson solver. 0) = u(x.7.8. u(0. subject to the inhomogeneous Dirichlet boundary conditions u(x.Copyright ©1995 by the Society for Industrial and Applied Mathematics. y) + ex+y u(x. Apply CG and PCG to solve the ﬁvepoint discretization of −uxx (x. 1) = u(1. Plot φ(xn ) as a function of n for each of the examples.18) preconditioners for the elliptic boundary value problem in § 2. Buy this book from SIAM at http://www. y < 1.8. and symmetric Gauss–Seidel).15.8.17) and symmetric Gauss–Seidel (1. y) = 0.com/SIAM/FR16.html. Implement Jacobi (1. Compare the performance with respect to both computer time and number of iterations to preconditioning with the Poisson solver.13.14.7.12. 0 < x. Jacobi. How does the performance compare to CG and PCG? 2. y. . y) − uyy (x. < 1. CONJUGATE GRADIENT ITERATION 31 2. This electronic version is for personal use and may not be duplicated or distributed. y) = 1. 2.8.ecsecurehost.
.html.Copyright ©1995 by the Society for Industrial and Applied Mathematics.com/SIAM/FR16. 32 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Buy this book from SIAM at http://www.ecsecurehost. This electronic version is for personal use and may not be duplicated or distributed.
The minimization property and its consequences The GMRES (Generalized Minimum RESidual) was proposed in 1986 in [167] as a Krylov subspace method for nonsymmetric systems. Corollary 3. We have just proved the following result. The kth (k ≥ 1) iteration of GMRES is the solution to the least squares problem minimizex∈x0 +Kk b − Ax 2 .1. GMRES does not require computation of the action of AT on a vector. . ¯ ¯ Hence if x ∈ x0 + Kk then r = p(A)r0 where p ∈ Pk is a residual polynomial.1. Moreover. ¯ From this we have the following corollary. This electronic version is for personal use and may not be duplicated or distributed. Chapter 3 GMRES Iteration 3. ¯ 33 Buy this book from SIAM at http://www. This is a signiﬁcant advantage in many cases. Then for all pk ∈ Pk ¯ (3.3) rk r0 2 2 ≤ pk (A) 2 .com/SIAM/FR16. one must store a basis for Kk .2) rk 2 = min p(A)r0 ¯ p∈Pk 2 ≤ pk (A)r0 2 .1. Theorem 3. The use of residual polynomials is made more complicated because we cannot use the spectral theorem to decompose A.ecsecurehost. Then for all pk ∈ Pk ¯ (3. and therefore storage requirements increase as the iteration progresses.1.1. Let A be nonsingular and let xk be the kth GMRES iteration. Let A be nonsingular and let xk be the kth GMRES iteration.1) The beginning of this section is much like the analysis for CG.html. Unlike CGNR. (3.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Note that if x ∈ x0 + Kk then x = x0 + and so b − Ax = b − Ax0 − k−1 j=0 k−1 j=0 k j=1 γj Aj r0 γj A j+1 r0 = r0 − γj−1 Aj r0 .
Recall that A is diagonalizable if there is a nonsingular (possibly complex!) matrix V such that A = V ΛV −1 . we will use the l2 norm in C N . Proof. Then for all pk ∈ Pk ¯ (3. It is well known [141] that p(A) = pN (A) = 0. is xH y. Let pk ∈ Pk . Then the GMRES algorithm will ﬁnd the solution within N iterations. In particular. If A is a diagonalizable matrix and p is a polynomial then p(A) = V p(Λ)V −1 A is normal if the diagonalizing transformation V is orthogonal. We can use the structure of a diagonalizable matrix to prove the following result. Hence we will switch to complex matrices and vectors. the space of complex N vectors. Let xk be the kth GMRES iterate. 34 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS We can apply the corollary to prove ﬁnite termination of the GMRES iteration.3). In Chapter 2 we applied the spectral theorem to obtain more precise information on convergence rates.ecsecurehost. However.3. The characteristic polynomial of A is p(z) = det(A − zI). In that case the columns of V are the eigenvectors of A and V −1 = V H .2. rN = b − AxN = 0 and hence xN is the solution.4) rk r0 2 2 ≤ κ2 (V ) max ¯k (z). and so pN (z) = p(z)/p(0) ∈ PN ¯ is a residual polynomial.html. We pay a price in that we must use complex arithmetic for the only time in this book. . Recall that the scalar product in C N .Copyright ©1995 by the Society for Industrial and Applied Mathematics. p z∈σ(A) 2 ¯ Proof. Here Λ is a (possibly complex!) diagonal matrix with the eigenvalues of A on the diagonal. In the remainder of this section we must use complex arithmetic to analyze the convergence. We can easily estimate pk (A) ¯ pk (A) ¯ as asserted. Let A be nonsingular. Theorem 3.1. By ¯ (3.com/SIAM/FR16. 2 by ≤ V 2 V −1 2 pk (Λ) ¯ 2 ≤ κ2 (V ) max ¯k (z). if A is diagonalizable we may use (3.2) to get information from clustering of the spectrum just like we did for CG. p z∈σ(A) Buy this book from SIAM at http://www.1. Our use of complex arithmetic will be implicit for the most part and is needed only so that we may admit the possibility of complex eigenvalues of A. Let A = V ΛV −1 be a nonsingular diagonalizable matrix. Theorem 3. This electronic version is for personal use and may not be duplicated or distributed. This is not an option for general nonsymmetric matrices. Here V H is the complex conjugate transpose of V . p(0) = det(A) = 0 since A is nonsingular. p has degree N .
[33]. Again we look at examples. [192]. rate estimates have been derived in [139]. Hence we conﬁne our attention to diagonalizable matrices. that the eigenvalues of A lie in the interval (9.1. [134].4. We can use (3.6) 2 ≤ ρ < 1. Buy this book from SIAM at http://www.5.1.2.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Then the GMRES iteration. Let b be a linear combination of k of the eigenvectors of A b= k l=1 γl uil . and [34]. for Ax = b will terminate in at most k iterations.3) and (3. The estimate (3. Termination As is the case with CG. As we did for CG. We assume that x0 = 0 and hence r0 = b. Let A be a nonsingular diagonalizable matrix. with x0 = 0. we look at some easy consequences of (3. we terminate the iteration when (3. This is particularly so for the ﬁnite diﬀerence Jacobian matrices we consider in Chapters 6 and 8. It is a direct ¯ rk 2 ≤ ρk r0 2 . Then GMRES will terminate in at most k iterations. GMRES ITERATION 35 It is not clear how one should estimate the condition number of the diagonalizing transformation if it exists. Theorem 3.ecsecurehost.5) holds without requiring a lemma like Lemma 2. Let A be a nonsingular normal matrix. If A is not diagonalizable.3) and (3. 3. 11).2. Assume that A has only k distinct eigenvalues. . The convergence rate estimates for the diagonalizable case will involve κ2 (V ). As the set of nondiagonalizable matrices has measure zero in the space of N × N matrices. Assume that I − A consequence of (3.3. Assume that A = V ΛV −1 is diagonalizable.2) that (3. and that κ2 (V ) = 100. Hence (3.4) directly to estimate the ﬁrst k such that (3. but will otherwise resemble those for CG. Let pk (z) = (1 − z)k . If A is normal. of course.5) holds when 102−k < η or when k > 2 + log10 (η). As was the case with CG.html.com/SIAM/FR16.6) illustrates the potential beneﬁts of a good approximate inverse preconditioner. the chances are very high that a computed matrix will be diagonalizable. Using the residual polynomial pk (z) = (10 − z)k /10k we ﬁnd ¯ rk r0 2 2 ≤ (100)10−k = 102−k .4). This electronic version is for personal use and may not be duplicated or distributed. Theorem 3. κ2 (V ) = 1.5) rk 2 ≤η b 2 for the purposes of this example. GMRES is best thought of as an iterative method.
6) to the preconditioned system. Hence. Preconditioning Preconditioning for GMRES and other methods for nonsymmetric problems is diﬀerent from that for CG.7 we use the Poisson solver preconditioner for nonsymmetric partial diﬀerential equations.ecsecurehost. one can write a single matrixvector product routine for M A (or AM ) that includes both the application of A to a vector and that of the preconditioner. However there are two diﬀerent ways to view preconditioning. If r = M Ax − M b is the residual for the preconditioned system. One important aspect of implementation is that. then applying GMRES to M Ax = M b allows one to apply (3. This is called right preconditioning. Incomplete factorization (LU in this case) preconditioners can be used [165] as can polynomial preconditioners. [33].Copyright ©1995 by the Society for Industrial and Applied Mathematics. . by Lemma 1. [120]. Multigrid [99] and alternating direction [8]. Again we mention [8] and [12] as a good general references for preconditioning. and [36] for discussions of and pointers to additional references to several questions related to nonnormal matrices. This is a very active area of research and we refer to [134]. 3. [183]. If I − AM 2 = ρ < 1. Hence. Buy this book from SIAM at http://www. CGNR. [182] methods have similar performance and may be more generally applicable. CGNE. This electronic version is for personal use and may not be duplicated or distributed.com/SIAM/FR16.5 are useful for GMRES as well. 36 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS The convergence estimates for GMRES in the nonnormal case are much less satisfying that those for CG. [164]. ek e0 2 2 ≤ κ2 (M A) rk r0 2 2 . one can solve the system AM y = b with GMRES and then set x = M y. unlike PCG. If one can ﬁnd M such that I − MA 2 = ρ < 1. [72]. or GMRES in the normal case. There is no concern that the preconditioned system be spd and hence (3.6) essentially tells the whole story. Most of the preconditioning ideas mentioned in § 2. In the examples in § 3. we have.1. Some hybrid algorithms use the GMRES/Arnoldi process itself to construct polynomial preconditioners for GMRES or for Richardson iteration [135]. Preconditioning done in this way is called left preconditioning. one can apply the algorithm directly to the system M Ax = M b (or AM y = b).html. Hence. Right preconditioning has been used as the basis for a method that changes the preconditioner as the iteration progresses [166]. the value of the relative residuals as estimators of the relative error is unchanged. if the product M A can be formed without error.1. we might expect the relative residual of the preconditioned system to be a better indicator of the relative error than the relative residual of the original system. if M A has a smaller condition number than A. [34].3. The residual of the preconditioned problem is the same as that of the unpreconditioned problem.
The Gram–Schmidt procedure for formation of an orthonormal basis for Kk is called the Arnoldi [4] process. The data are vectors x0 and b. a map that computes the action of A on a vector.ecsecurehost. Hence we can convert (3. V ) 1. The algorithm computes an orthonormal basis for Kk and stores it in the columns of V . For i = 1. we must have xk = x0 + Vk y where y minimizes b − A(x0 + Vk y) 2 = r0 − AVk y 2 . . is the lth column of Vk .Copyright ©1995 by the Society for Industrial and Applied Mathematics. Hence. GMRES implementation: Basic ideas Recall that the least squares problem deﬁning the kth GMRES iterate is minimizex∈x0 +Kk b − Ax 2 . GMRES ITERATION 37 3.1. our least squares problem in Rk is (3.7) very eﬃciently and the resulting least squares problem requires no extra multiplications of A with vectors. b. . say. This electronic version is for personal use and may not be duplicated or distributed. If one uses Gram–Schmidt orthogonalization. and a dimension k. Then any z ∈ Kk can be written as z= vlk k l=1 yl vlk .com/SIAM/FR16.html.7) minimizey∈Rk r0 − AVk y 2 . arnoldi(x0 . Suppose one had an orthogonal projector Vk onto Kk . . Since x − x0 = Vk y for some y ∈ Rk . . however. A division by zero is referred to as breakdown and happens only if the solution to Ax = b is in x0 + Kk−1 . Buy this book from SIAM at http://www. then the columns of the matrix Vk are an orthonormal basis for Kk .1) to a least squares where problem in Rk for the coeﬃcient vector y of z = x − x0 . . The problem with such a direct approach is that the matrix vector product of A with the basis matrix Vk must be taken at each iteration. k − 1 vi+1 = Avi − Avi − i T j=1 ((Avi ) vj )vj i T j=1 ((Avi ) vj )vj 2 If there is never a division by zero in step 2 of Algorithm arnoldi. 2. Deﬁne r0 = b − Ax0 and v1 = r0 / r0 2 .4. k. This is a standard linear least squares problem that could be solved by a QR factorization. A. Algorithm 3. one can represent (3.4.
and let i be the smallest integer for which Avi − i j=1 ((Avi )T vj )vj = 0. we can use it to implement GMRES in an eﬃcient way. Set hji = (Avj )T vi . If the Arnoldi process does not break down. Setting y = βH −1 e1 proves that ri = 0 by the minimization property. 0)T ∈ Ri we have ri 2 2 = b − Axi 2 = r0 − A(xi − x0 ) 2 . .8) Hence. which overwrites the initial iterate x and the residual norm. Since xi − x0 ∈ Ki there is y ∈ Ri such that xi − x0 = Vi y. Then x = A−1 b ∈ x0 + Ki . where · Rk+1 denotes the Euclidean norm in Rk+1 . Let A be nonsingular.ecsecurehost.9) The goal of the iteration is to ﬁnd. Since the columns of Vi are an orthonormal basis for Ki . hij = 0 if i > j + 1. ≤ b 2. The Arnoldi process (unless it terminates prematurely with a solution) produces matrices {Vk } with orthonormal columns such that (3. .. Buy this book from SIAM at http://www.1.com/SIAM/FR16. Setting β = r0 and e1 = (1. 0. The upper Hessenberg structure can be exploited to make the solution of the least squares problems very eﬃcient [167]. We limit the number of iterations to kmax and return the solution. (3. x. the norm of rk can be found without explicitly forming xk and computing rk = b − Axk . for a given .e.html. . the k +1×k matrix Hk whose entries are hij is upper Hessenberg. H is nonsingular since A is. the righthand side b. a vector x so that b − Ax 2 AVk = Vk+1 Hk . rk = b − Axk = r0 − A(xk − x0 ) = Vk+1 (βe1 − Hk y k ). i. We have. The input is the initial iterate. Since r0 = βVi e1 and Vi is an orthogonal matrix ri 2 = Vi (βe1 − Hy) 2 = βe1 − Hy Ri+1 . using the orthogonality of Vk+1 . Hence xk = x0 + Vk y k . . This electronic version is for personal use and may not be duplicated or distributed. Note that when y k has been computed. 38 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Lemma 3.4. By hypothesis Avi ∈ Ki and hence AKi ⊂ Ki . . Proof. and a map that computes the action of A on a vector. let the vectors vj be generated by Algorithm arnoldi. By the Gram–Schmidt construction. for some y k ∈ Rk .Copyright ©1995 by the Society for Industrial and Applied Mathematics. we have AVi = Vi H. rk 2 = Vk+1 (βe1 − Hk y k ) 2 = βe1 − Hk y k Rk+1 . where y k minimizes βe1 − Hk y 2 over Rk . where H is an i × i matrix.
[167]. This electronic version is for personal use and may not be duplicated or distributed. when it works it can signiﬁcantly reduce the storage costs of the iteration. and we use the method of [167] in the collection of MATLAB codes. b. There are more eﬃcient ways to solve the least squares problem in step 2f. Note that xk is only computed upon termination and is not needed within the iteration. The savings are slight if k is small relative to N . We discuss implementation of GMRES(m) later in this section. v1 = r/ r 2 . Buy this book from SIAM at http://www. which is often the case for large problems. the result from the previous call. . Algorithm gmresa can be implemented very straightforwardly in MATLAB. . which depends on this orthogonality. . xk = x0 + Vk y k . If the norm of the residual is larger than . k hjk = (Avk )T vj (c) vk+1 = Avk − (d) hk+1. call GMRES and explicitly test the residual b−Axk if k = m upon termination. This means that in order to perform k GMRES iterations one must store k vectors of length N . It is an important property of GMRES that the basis for the Krylov space must be stored as the iteration progress. ρ) 1. Step 2f can be done with a single MATLAB command. . A more serious problem with the implementation proposed in Algorithm gmresa is that the vectors vj may become nonorthogonal as a result of cancellation errors.2. [197].9).ecsecurehost. . . A partial remedy is to replace the classical Gram–Schmidt orthogonalization in Algorithm gmresa with modiﬁed Gram–Schmidt orthogonalization.k = vk+1 k j=1 hjk vj 2 2 (e) vk+1 = vk+1 / vk+1 (1.com/SIAM/FR16. will not hold and the residual and approximate solution could be inaccurate. . r = b − Ax. There is no general convergence theorem for the restarted algorithm and restarting will slow the convergence down.4. the backward division operator. This restarted version of the algorithm is called GMRES(m) in [167]. While ρ > b 2 and k < kmax do (a) k = k + 1 (b) for j = 1.html. (3. (g) ρ = βe1 − Hk y k 3. For very large problems this becomes prohibitive and the iteration is restarted when the available room for basis vectors is exhausted. gmresa(x. If this happens. However. 0)T (f) e1 = ∈ Rk+1 Minimize βe1 − Hk y k Rk+1 over Rk to obtain y k . . A. 0. . β = ρ. GMRES ITERATION 39 Algorithm 3. k = 0 2. call GMRES again with x0 = xk . kmax. . and the simple oneline MATLAB approach can be eﬃcient for such problems. ρ = r 2 . at a cost of O(k 3 ) ﬂoatingpoint operations. One way to implement this is to set kmax to the maximum number m of vectors that one can store. We replace Rk+1 .Copyright ©1995 by the Society for Industrial and Applied Mathematics.
β = ρ. The versions we implement in the collection of MATLAB codes use modiﬁed Gram–Schmidt.0000e + 00 4.0000e + 00 1.0000e − 07 1.3568e − 02 T The columns of VU are not orthogonal at all. 40 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS the loop in step 2c of Algorithm gmresa with vk+1 = Avk for j = 1.0000e − 07 −1.com/SIAM/FR16. Algorithm 3. We illustrate this point with a simple example from [128]. One can take a second pass through the modiﬁed Gram–Schmidt process and restore lost orthogonality [147].0456e − 14 1.0000e + 00 1.0000e + 00 4. While ρ > b 2 and k < kmax do (a) k = k + 1 Buy this book from SIAM at http://www. k = 0 2. r = b − Ax. V = 1. b. This implementation solves the upper Hessenberg least squares problem using the MATLAB backward division operator.html. A.004.4. . For modiﬁed Gram–Schmidt 1. there is an opportunity to compensate for a loss of orthogonality in the basis vectors for the Krylov space.ecsecurehost. [160]. In fact v2 v3 ≈ −. k T vk+1 = vk+1 − (vk+1 vj )vj . the modiﬁed form is much more likely in practice to maintain orthogonality of the basis. and is not particularly eﬃcient. ρ) 1. j. This electronic version is for personal use and may not be duplicated or distributed.Copyright ©1995 by the Society for Industrial and Applied Mathematics. The outline of our implementation is Algorithm gmresb. We present a better implementation in Algorithm gmres.3565e − 16 T Here vi vj − δij  ≤ 10−8 for all i.0000e − 07 −1.0436e − 07 9. gmresb(x. kmax. .0000e + 00 .0456e − 14 1. ρ = r 2 .9715e − 08 −9. Second. δ 0 δ We orthogonalize the columns of A with classical Gram–Schmidt to obtain 1.0000e − 07 1. However. this version is very simple and illustrates some important ideas. v1 = r/ r 2 . Let δ = 10−7 and deﬁne 1 1 1 A = δ δ 0 . While modiﬁed Gram–Schmidt and classical Gram–Schmidt are equivalent in inﬁnite precision. First.0436e − 07 −1.9905e − 01 . . . we see that xk need only be computed after termination as the least squares residual ρ can be used to approximate the norm of the residual (they are identical in exact arithmetic). V = 1.3. doing the computations in MATLAB. .0436e − 07 1.
One can test for loss of orthogonality [22]. k T hjk = vk+1 vj vk+1 = vk+1 − hjk vj • hk+1. .k = vk+1 2 2 • vk+1 = vk+1 / vk+1 Buy this book from SIAM at http://www. For example. (f) ρ = βe1 − Hk y k 3. one can augment the modiﬁed Gram–Schmidt process • vk+1 = Avk for j = 1.com/SIAM/FR16. . .k = vk+1 2 • If loss of orthogonality is detected For j = 1.ecsecurehost. . xk = x0 + Vk y k . 0)T (e) e1 = ∈ Rk+1 Minimize βe1 − Hk y k Rk+1 to obtain y k ∈ Rk . One can reorthogonalize in every iteration or only if a test [147] detects a loss of orthogonality. . There is nothing to be gained by reorthogonalizing more than once [147]. . . • vk+1 = vk+1 / vk+1 2 with a second pass (reorthogonalization).html. and reorthogonalize if needed or use a more stable means to create the matrix V [195]. . k T htmp = vk+1 vj hjk = hjk + htmp vk+1 = vk+1 − htmp vj • hk+1. .k = vk+1 2 2 (d) vk+1 = vk+1 / vk+1 (1. . one can still lose orthogonality in the columns of V . vk+1 = vk+1 − hjk vj (c) hk+1. k T i. [147]. These more complex implementations are necessary if A is ill conditioned or many iterations will be taken. . hjk = vk+1 vj ii. GMRES ITERATION 41 (b) vk+1 = Avk for j = 1. Even if modiﬁed Gram–Schmidt orthogonalization is used. 0. . . . k T hjk = vk+1 vj vk+1 = vk+1 − hjk vj • hk+1. .Copyright ©1995 by the Society for Industrial and Applied Mathematics. . This electronic version is for personal use and may not be duplicated or distributed.k = vk+1 2 Rk+1 . . . The modiﬁed Gram–Schmidt process with reorthogonalization looks like • vk+1 = Avk for j = 1. .
Hence the total cost of the m GMRES iterations is m matrixvector products and O(m4 + m2 N ) ﬂoatingpoint operations. The method based on (3. 42 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS One approach to reorthogonalization is to reorthogonalize in every step.com/SIAM/FR16.1 we tabulate relative residuals as a function of the iteration counter for classical Gram–Schmidt without reorthogonalization (CGM). k scalar products. especially when implemented in an environment like MATLAB. In Table 3. For large k. x0 = (0. the bruteforce method can be very costly.10) (MGMPO). say. and (3. in step 2e of gmresb is O(k3 ) ﬂoatingpoint operations. however.10) is almost as eﬀective as the much more expensive approach of reorthogonalizing in every step. When k is not too large and the cost of matrixvector products is high. . modiﬁed Gram–Schmidt without reorthogonalization (MGM). To illustrate the eﬀects of loss of orthogonality and those of reorthogonalization we apply GMRES to the diagonal system Ax = b where b = (1.Copyright ©1995 by the Society for Industrial and Applied Mathematics.10) Avk 2 + δ vk+1 2 = Avk 2 to working precision. A variation on a method from [147] is used in [22].ecsecurehost. Reorthogonalization is done after the Gram–Schmidt loop and before vk+1 is normalized if (3. 1. This doubles the cost of the computation of V and is usually unnecessary. by QR factorization or the MATLAB backward division operator. 1)T . and reorthogonalization in every iteration (MGMFO). the reorthogonalization strategy based on (3. A= 0 0 0 104 While in inﬁnite precision arithmetic only three iterations are needed to solve the system exactly. While classical Gram–Schmidt fails. We employ this test in the MATLAB code gmres with δ = 10−3 . More eﬃcient and equally eﬀective approaches are based on other ideas. This electronic version is for personal use and may not be duplicated or distributed. reorthogonalization based on the test (3. because of its simplicity.0011 0 . Buy this book from SIAM at http://www. 0.001 0 0 . This is an appealing algorithm. We provide an implementation of Algorithm gmresb in the collection of MATLAB codes.html. a bruteforce solution to the least squares problem using the MATLAB backward division operator is not terribly ineﬃcient.10) is the default in the MATLAB code gmresa. The idea is that if the new vector is very small relative to Avk then information may have been lost and a second pass through the modiﬁed Gram–Schmidt process is needed. we ﬁnd in the MATLAB environment that a solution to full precision requires more than three iterations unless reorthogonalization is applied after every iteration. and the solution of the Hessenberg least squares problem in step 2e. 0)T . The k scalar products require O(kN ) ﬂoatingpoint operations and the cost of a solution of the Hessenberg least squares problem. The kth GMRES iteration requires a matrixvector product.11) .
The orthogonal matrix G rotates the vector (c. where c = cos(θ). 43 k 0 1 2 3 4 5 6 7 8 9 10 CGM 1.ecsecurehost.88e02 6.04e18 MGMPO 1.12) G= .16e01 3. An N × N Givens rotation replaces a 2 × 2 block on the diagonal of the Buy this book from SIAM at http://www.88e02 6.1 Eﬀects of reorthogonalization. . This electronic version is for personal use and may not be duplicated or distributed.53e05 2. The O(k 2 ) cost of the triangular solve and the O(kN ) cost of the construction of xk are incurred after termination.88e02 6. [22].87e05 3. s = sin(θ) for θ ∈ [−π.00e+00 8. This implementation maintains the QR factorization of Hk in a clever way so that the cost for a single GMRES iteration is O(N k) ﬂoatingpoint operations.00e+00 8.00e05 2.00e+00 8.Copyright ©1995 by the Society for Industrial and Applied Mathematics. π].00e+00 8.html. −s). Implementation: Givens rotations If k is large.70e08 3. Householder reﬂections [195].16e01 3.69e05 4.16e01 3. implementations using Givens rotations [167].5. The implementation in Algorithm gmres and in the MATLAB code collection is from [167]. G c −s = 1 0 .35e05 3.42e08 3. or a shifted Arnoldi process [197] are much more eﬃcient than the bruteforce approach in Algorithm gmresb.com/SIAM/FR16.88e02 6.04e24 MGMFO 1. GMRES ITERATION Table 3.34e34 3.42e08 5.37e05 MGM 1.74e05 2.16e01 3. which makes an angle of −θ with the xaxis through an angle θ so that it overlaps the xaxis. A 2 × 2 Givens rotation is a matrix of the form c −s s c (3.74e05 3.
Buy this book from SIAM at http://www. . (3.. 0 . and noting that βe1 − Hk y k Rk+1 = Q(βe1 − Hk y k ) Rk+1 = g − Rk y k Rk+1 . We deﬁne G1 = G1 (c1 . 1 .13) 0 . s c 1 0 . . s) is an N × N givens rotation of the form (3. . This reduction can be accomplished in O(N ) ﬂoatingpoint operations and hence is far more eﬃcient than a solution by a singular value decomposition or a reduction based on Householder transformations. . G1 we see that QH = R is upper triangular. setting g = βQe1 . of course. 0 .14) c1 = h11 / h2 + h2 and s1 = −h21 / h2 + h2 ..Copyright ©1995 by the Society for Industrial and Applied Mathematics.html. . This electronic version is for personal use and may not be duplicated or distributed. G= . 22 32 22 32 and annihilate h32 . 0 .13) with a 2 × 2 Givens rotation in rows and columns j and j + 1. Let H be an N × M (N ≥ M ) upper Hessenberg matrix with rank M . c −s . Similarly. where Rk is the k + 1 × k triangular factor of the QR factorization of Hk . s2 ) to H. we can now apply G2 (c2 . Our notation is that Gj (c. . ... 11 21 11 21 If we replace H by G1 H. [184]. We reduce H to triangular form by ﬁrst multiplying the matrix by a Givens rotation that annihilates h21 (and. 44 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS N × N identity matrix with a 2 × 2 Givens rotation... where (3. .. . changes h11 and the subsequent columns)...ecsecurehost. They are of particular value in reducing Hessenberg matrices to triangular form and thereby solving Hessenberg least squares problems such as the ones that arise in GMRES. . 0 1 0 0 . . This method is also used in the QR algorithm for computing eigenvalues [89]. then the ﬁrst column of H now has only a single nonzero element h11 . Givens rotations are used to annihilate single nonzero elements of matrices in reduction to triangular form [89]. .com/SIAM/FR16. s1 ) by (3.. Note that G2 does not aﬀect the ﬁrst column of H. . . A straightforward application of these ideas to Algorithm gmres would solve the least squares problem by computing the product of k Givens rotations Q. . Continuing in this way and setting Q = GN .15) c2 = h22 / h2 + h2 and s2 = −h32 / h2 + h2 .
k = vk+1 2 (d) Test for loss of orthogonality and reorthogonalize if necessary. v1 = r/ r 2 . 3. sk ) . (g) ρ = (g)k+1 . however. iii. 0)T ∈ Rkmax+1 2. then computing the Givens rotation Gk+1 that annihilates the (k + 2)nd element of Qk hk+2 . The MATLAB implementation of Algorithm gmres uses (3. and ﬁnally setting Qk+1 = Gk+1 Qk and forming Rk+1 by augmenting Rk with Gk+1 Qk hk+2 . GMRES ITERATION 45 In the context of GMRES iteration. A. .com/SIAM/FR16. . xk = x0 + Vk y k .k k+1. (e) vk+1 = vk+1 / vk+1 (f) ii. . h2 + h2 k. ρ = r 2 .10) to test for loss of orthogonality.k /ν. .k = 0 iv. sk = −hk+1. note that if Rk = Qk Hk and. . . Algorithm 3. β = ρ. If k > 1 apply Qk−1 to the kth column of H. . 4. The MATLAB implementation of Algorithm gmres stores Qk by storing the sequences {cj } and {sj } and then computing the action of Qk on a vector x ∈ Rk+1 by applying Gj (cj . .j = hi.1. 0. Solve the upper triangular system Ry k = w.k − sk hk+1. k T i. hjk = vk+1 vj ii. we can incrementally perform the QR factorization of H as the GMRES iteration progresses [167]. vk+1 = vk+1 − hjk vj (c) hk+1. we can update both Qk and Rk by ﬁrst multiplying hk+2 by Qk (that is applying the ﬁrst k Givens rotations to hk+2 ).html.j for 1 ≤ i. . after orthogonalization. ν = 2 i. g = ρ(1.Copyright ©1995 by the Society for Industrial and Applied Mathematics. sk )g. To see this. sj ) in turn to obtain Qk x = Gk (ck .ecsecurehost.k = ck hk. . This electronic version is for personal use and may not be duplicated or distributed. . j ≤ k. While ρ > b 2 and k < kmax do (a) k = k + 1 (b) vk+1 = Avk for j = 1. Buy this book from SIAM at http://www.5. k = 0.k . G1 (c1 . We overwrite the upper triangular part of Hk with the triangular part of the QR factorization of Hk in the MATLAB code. ρ) 1. we add the new column hk+2 to Hk . s1 )x.k . hk+1. b. Set ri.k /ν hk. r = b − Ax. kmax. ck = hk. Set (w)i = (g)i for 1 ≤ i ≤ k. gmres(x. g = Gk (ck .
Copyright ©1995 by the Society for Industrial and Applied Mathematics. CGNR and CGNE have the disadvantages that a transposevector product must be computed for each iteration and that the coeﬃcient matrix AT A or AAT has condition number the square of that of the original matrix. This electronic version is for personal use and may not be duplicated or distributed. Algorithm 3. as we have seen. the storage requirements increase as the iteration progresses. share the properties that they are easy to implement. can be analyzed by a common residual polynomial approach. k=1 3.2.com/SIAM/FR16. b. A. ρ) (b) k = k + m The storage costs of m iterations of gmres or of gmresm are the m + 2 vectors b. All the methods we present in this section require two matrixvector products for each iteration. Consistently with our implementation of GMRES.html. For large and illconditioned problems. methods based on shortterm recurrences such as CG that also satisfy minimization or conjugacy conditions cannot be constructed for general matrices. and {vk }m .8 we give an example of how this squaring of the condition number can lead to failure of the iteration. . only need matrixvector products. GMRES needs only matrixvector products and uses A alone. m. and only terminate if an acceptable approximate solution has been found. b. An ideal method would. Hence. m. and converge in N iterations for all nonsingular A. but can still be quite useful. 46 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS We close with an example of an implementation of GMRES(m) . However. While ρ > b 2 and k < kmax do (a) gmres(x. ρ) 1. x. Restarting can seriously degrade performance. m. and CGNE. it may be impossible to store enough basis vectors and the iteration may have to be restarted.5. However. a basis for Kk must be stored to compute xk .6. GMRES. the arguments to Algorithm gmresm are the same as those for Algorithm gmres. [74]. . In § 3. be based on some kind of minimization principle or conjugacy. kmax. therefore.ecsecurehost. as with CG. . The methods we describe in this section fall short of the ideal. A. these methods can also be implemented in a manner that incorporates Buy this book from SIAM at http://www. . gmresm(x. This implementation does not test for success and may. k = m 3. fail to terminate. b. like CG. Aside from the parameter m. A. We discuss only a small subset of these methods and refer the reader to [12] and [78] for pointers to more of the literature on this subject. ρ) 2. gmres(x. but. we take the view that preconditioners will be applied externally to the iteration. have modest storage requirements that do not depend on the number of iterations needed for convergence. You are asked to repair this in exercise 7. Other methods for nonsymmetric systems The methods for nonsymmetric linear systems that receive most attention in this book. CGNR.
6. p = r + β p ˆ ˆ ˆ (d) v = Ap (e) α = ρk /(ˆT v) p (f) x = x + αp (g) r = r − αv. BiCG produces the same iterations as CG (but computes everything except x twice). Do While r 2 > b 2 and k < kmax (a) k = k + 1 (b) ρk = rT r. r0 is a usersupplied vector and is ˆ ˆ often set to r0 . . Kk = span(ˆ0 .html.16) where T rk w = 0 for all w ∈ Kk . . will not ˆ T Ap = 0 we say that a breakdown has taken vanish. If either ρk−1 = 0 or p ˆ place. there is no guarantee that ρk in step 2b or pT Ap. the kth residual must satisfy the biorthogonality condition (3. however. The earliest such method BiCG (Biconjugate gradient) [122]. {ˆk } such that biorthogonality r p holds. This electronic version is for personal use and may not be duplicated or distributed. b. and [191]. GMRES ITERATION 47 the preconditioner and uses the residual for the original system to control termination. r = r. AT r0 . Algorithm 3. [76]. [78]. BiCG. the algorithm can become unstable or produce inaccurate results. . ρ0 = 1. kmax) ˆ 1. 3. does not enforce a minimization principle. This algorithmic description is explained and ˆ motivated in more detail in [122]. A is not spd. bicg(x. r = b − Ax.1. (AT )k−1 r0 ) r ˆ ˆ is the Krylov space for AT and the vector r0 . [76].2) for CG [89].16) is equivalent to the minimization principle (2.ecsecurehost. Using the notation of Chapter 2 and [191] we give an implementation of BiCG making the choice r0 = r0 . p = p = 0. instead. If. We do not recommend use of BiCG and present this algorithm only as a basis for discussion of some of the properties of this class of algorithms.Copyright ©1995 by the Society for Industrial and Applied Mathematics. A.6. The algorithm gets its name because it produces sequences of residuals {rk }. While there are approaches Buy this book from SIAM at http://www. e. ˆk In the symmetric positive deﬁnite case (3. k = 0 ˆ 2. . . rk rl = 0 if k = l and the search directions {pk } and {ˆk } satisfy ˆT p the biconjugacy property pT Apl = 0 if k = l. {ˆk } and search directions {pk }. β = ρk /ρk−1 ˆ (c) p = r + βp.com/SIAM/FR16. i. . r = r − αAT p ˆ ˆ ˆ Note that if A = AT is spd. Even if these quantities are nonzero but very small.1. the denominator in step 2e.
A remedy for one of the problems with BiCG is the Conjugate Gradient Squared (CGS) algorithm [180].Copyright ©1995 by the Society for Industrial and Applied Mathematics. Conjugate Gradient Squared. [78]. ˆ p p r p ˆ The other references to AT can be eliminated in a similar fashion and an iteration satisfying rk = pk (A)2 r0 (3. Once breakdown has occurred. A transposevector product is needed. 48 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS [80]. Note that there is pk ∈ Pk such that both ¯ (3. We can compare the bestcase performance of BiCG with that of GMRES. one can restart the BiCG iteration or pass the computation to another algorithm. to avoid some breakdowns in many variations of BiCG. All of the methods we present in this section can break down and should be implemented with that possibility in mind.2. ¯ ˆ ¯ r Hence. CGS.18) ¯ is produced. This explains the name. will always reduce the residual more rapidly than BiCG (in terms of iterations. where pk is the same polynomial produced by BiCG and used in ¯ (3. The performance of the algorithm can be erratic or even unstable with residuals increasing by several orders of magnitude from one iteration to the next. reminding us that GMRES. if suﬃcient storage is available. One should also keep in mind that a single BiCG iteration requires two matrixvector products and a GMRES iterate only one.ecsecurehost. Breakdowns aside.com/SIAM/FR16.6.17). by the minimization property for GMRES GM rk RES 2 Bi−CG ≤ rk 2.17) implies that the scalar product rT r in BiCG (step 2b of ˆ Algorithm bicg) can be represented without using AT as T rk rk = (¯k (A)r0 )T (¯k (AT )ˆ0 ) = (¯k (A)2 r0 )T r0 .17) rk = pk (A)r0 and rk = pk (AT )ˆ0 . The algorithm takes advantage of the fact that (3. such as GMRES. However. Buy this book from SIAM at http://www. . there are other problems with BiCG. [81].html. there are no methods that both limit storage and completely eliminate the possibility of breakdown [74]. [148]. This electronic version is for personal use and may not be duplicated or distributed. 3. ˆ BiCG sometimes performs extremely well and the remaining algorithms in this section represent attempts to capture this good performance and damp the erratic behavior when it occurs. [79]. Moreover. The possibility of breakdown is small and certainly should not eliminate the algorithms discussed below from consideration if there is not enough storage for GMRES to perform well. the eﬀort in computing r ˆ at each iteration is wasted in that r makes no contribution to x. but not necessarily in terms of computational work). which at the least will require additional programming and may not be possible at all. but that the cost of the GMRES iteration increases (in terms of ﬂoatingpoint operations) as the iteration progresses.
k = 0 ˆT 4. This may.18) then ¯ (3. ¯ We provide no MATLAB implementation of BiCG or CGS. b. CGS ˆ replaces the transposevector product with an additional matrixvector product and applies the square of the BiCG polynomial to r0 to produce rk . BiCGSTAB is the most eﬀective representative of this class of algorithms.6. Since p2 ∈ P2k we have ¯k (3. [191]. kmax) 1. change the convergence properties for the worse and either improve good convergence or magnify erratic behavior [134].19) GM r2k RES 2 CGS ≤ rk 2. . . v0 = Ap0 . βk = ρk /ρk−1 ˆT uk = rk + βk qk pk = uk + βk (qk + βk pk−1 ) vk = Apk Breakdowns take place when either ρk−1 or σk−1 vanish. If the algorithm does not break down.21) ¯ ¯ qk (z) = pk (z) + βk qk−1 (z). In our description of CGS we will index the vectors and scalars that would be overwritten in an implementation so that we can describe some of the ideas in TFQMR later on. GMRES ITERATION 49 The work used in BiCG to compute r is now used to update x. Algorithm 3. that if pk is the residual polynomial from (3. p0 = u0 = r0 = b − Ax 2. We present an algorithmic description of CGS that we will refer to in our discussion of TFQMR in § 3. This description is taken from [77]. then αk−1 = 0 for all k.html. CGS has the same potential for breakdown as BiCG. This electronic version is for personal use and may not be duplicated or distributed. cgs(x.ecsecurehost. Do While r 2 > b 2 and k < kmax (a) k = k + 1 ˆT (b) σk−1 = r0 vk−1 (c) αk−1 = ρk−1 /σk−1 (d) qk = uk−1 − αk−1 vk−1 (e) xk = xk−1 + αk−1 (uk−1 + qk ) rk = rk−1 − αk−1 A(uk−1 + qk ) (f) ρk = r0 rk . x0 = x.Copyright ©1995 by the Society for Industrial and Applied Mathematics. r0 = r0 ˆ 3. [77].20) ¯ ¯ pk (z) = pk−1 (z) − αk−1 z qk−1 (z) ¯ ¯ where the auxiliary polynomials qk ∈ Pk are given by q0 = 1 and for k ≥ 1 by ¯ (3.com/SIAM/FR16. Buy this book from SIAM at http://www. A.4. ρ0 = r0 r0 .17) and (3.2. of course. One can show [180].6.
(3. letting s overwrite r when needed.19). r. v = p = 0.18) with rk = qk (A)pk (A)r0 (3. which is described and motivated fully in [191]. There is no convergence theory for BiCGSTAB and no estimate of the residuals that is better than that for CGS.22) where qk (z) = k i=1 (1 − ωi z). Buy this book from SIAM at http://www. p. ρ1 = r0 r ˆ 2. t = As rT (g) ω = tT s/ t 2 . k = 0. r = b − Ax. A single iteration reˆ quires four scalar products. kmax) ˆ ˆT 1. bicgstab(x. which can be thought of [12] as a blend of BiCG and GMRES(1).Copyright ©1995 by the Society for Industrial and Applied Mathematics.8. The performance in cases where the spectrum has a signiﬁcant imaginary part can be improved by constructing qk to have complex conjugate pairs of roots. r0 = r = r. we use r0 = r0 ˆ and follow the notation of [191].6. We provide an implementation of Algorithm bicgstab in the collection of MATLAB codes. 50 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 3. BiCGSTAB can have a much lower average cost per iterate than GMRES. b. The reason for this is that the cost of orthogonalization in GMRES can be much more than that of a matrixvector product if the dimension of the Krylov space is large. The examples in [191] indicate the eﬀectiveness of this approach. In our description of the implementation. In a situation where many GMRES iterations are needed and matrixvector product is fast.3. . b. This electronic version is for personal use and may not be duplicated or distributed.html.6. r0 . ρ0 = α = ω = 1. The constant ωi is selected to minimize ri = qi (A)pi (A)r0 as a function of ωi . t).com/SIAM/FR16. ρk+1 = −ωˆ0 t 2 (h) x = x + αp + ωs (i) r = s − ωt Note that the iteration can break down in steps 2b and 2e. . We will present such a case in § 3. Do While r 2 > b 2 and k < kmax (a) k = k + 1 (b) β = (ρk /ρk−1 )(α/ω) (c) p = r + β(p − ωv) (d) v = Ap rT (e) α = ρk /(ˆ0 v) (f) s = r − αv. One must store seven vectors (x. The Biconjugate gradient stabilized method (BiCGSTAB) [191] attempts to smooth the convergence of CGS by replacing (3. v. A. [98]. BiCGSTAB.ecsecurehost.3. The cost in storage and in ﬂoatingpoint operations per iteration remains bounded for the entire iteration. Algorithm 3.
0 −1 −1 0 diag(α0 .ecsecurehost.. .. . TFQMR. and Bm is the (m + 1) × m matrix given by j=1 −1 Bm = 0 . The quasiresidual is related to the true residual by a fullrank linear transformation r = Lq and reduction of the residual can be approximately measured by reduction in q. . (¯k (A))2 r0 p p (A)¯ ¯k pk−1 (A)r0 q Here the sequences {¯k } and {¯k } are given by (3. . (3. Therefore. [37]. . All algorithms in this family minimize the norm of an easily computable quantity q = f − Hz. Wm+1 the N × (m + 1) j=1 matrix with columns {wj }m+1 . Now we consider the quasiminimal residual (QMR) family of algorithms [80].21). .com/SIAM/FR16.6. (3. If we let r be the nearest integer less than or equal to a real r we have Aym = (wm − wm+1 )/α (m−1)/2 .23) where the denominator is nonzero by assumption. We assume that CGS does not break down and. . −1 . This electronic version is for personal use and may not be duplicated or distributed. α0 . . [202].25) for some z ∈ Rm . .24) where Ym is the N × m matrix with columns {yj }m . . (3. . The speciﬁc formulation of q and L depends on the algorithm. to illustrate the quasiminimization idea.20) and (3. . . Returning to Algorithm cgs we deﬁne sequences ym = and wm = uk−1 q k if m = 2k − 1 is odd. [81]. 1 0 1 . .html. [77].. 0 . Hence. α (m−1)/2 ) . We express (3. . Buy this book from SIAM at http://www. if m = 2k is even. . therefore.Copyright ©1995 by the Society for Industrial and Applied Mathematics. the p CGS residual satisﬁes CGS rk = w2k+1 .4. a quasiresidual over z ∈ RN .26) rm = r0 − AYm z = Wm+1 (e1 − Bm z). if m = 2k − 1 is odd. GMRES ITERATION 51 3. 0 . if m = 2k is even. . . We will focus on the transposefree algorithm TFQMR proposed in [77]. 1 Our assumptions that no breakdown takes place imply that Km is the span of {yj }m and hence j=1 xm = x0 + Ym z (3. .23) in matrix form as AYm = Wm+1 Bm . .. αk = 0 for all k ≥ 0.
as in our discussion of GMRES. the conditioning of the matrix Wm+1 can be improved by scaling.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Instead we have the quasiresidual norm τm = fm+1 − Hm zm 2 . each k produces two approximate solutions (corresponding to m = 2k − 1 and m = 2k). 52 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS where. Let rm RES be the residual for the mth GMRES iteration. . b ∈ RN be given. . Hence (3.ecsecurehost. As one can see from that description. 0.29) can be done by using Givens rotations to update the factorization of the upper Hessenberg matrix Hm [77] and this is reﬂected in the implementation description below. Let ξm be the smallest singular value of Wm+1 Ω−1 . and we do this in Algorithm tfqmr. . The quasiminimization idea is to solve (3. we could minimize rm by solving m+1 the least squares problem (3.28) rm = r0 − AYm z = Wm+1 Ω−1 (fm+1 − Hm z) m+1 Ω = diag(ω1 . ωm ) where fm+1 = Ωm+1 e1 = ωm+1 e1 and Hm = Ωm+1 Bm is bidiagonal and hence upper Hessenberg. Note that if k corresponds to the iteration counter for CGS.26) as (3.html. Now if we pick the weights ωi = wi norm one. Then if ξm > 0 m+1 (3.32) GM τm ≤ rm RES 2 /ξm . [78]. Buy this book from SIAM at http://www. The solution to the least squares problem (3. We can. One can [80]. base our termination criterion on τm . ω2 .27) and rewrite (3. The weights make it possible to derive straightforward estimates of the residual in terms of easily computed terms in the iteration. Theorem 3. 0)T ∈ Rm . also use the quasiminimization condition to estimate the TFQMR residual in terms of residual polynomials. In a sense.1.com/SIAM/FR16.6. . .30) xm = x0 + Ym zm . Assume that the TFQMR iteration does not break down or terminate with GM the exact solution for k iterations and let 1 ≤ m ≤ 2k.29) (thereby quasiminimizing the residual) to obtain zm and then set (3. therefore. . This electronic version is for personal use and may not be duplicated or distributed. The diagonal entries in the matrix Ωm+1 are called weights. . approximate residuals are not computed in Algorithm tfqmr.29) minimizez∈Rm fm+1 − Hm z 2 .31) rm 2 2 then the columns of Wm+1 Ω−1 have m+1 2 = Wm+1 Ω−1 (fm+1 − Hm z) m+1 √ ≤ τm m + 1. . If Wm+1 Ω−1 were an orthogonal matrix. Let (3. e1 = (1. Let A be an N × N matrix and let x0 . .
β = ρk /ρk−1 Buy this book from SIAM at http://www.31) to obtain the following result. j = 1.html.32) and use (3. ii. u2 = Ay2 (c) For i. Then. Hence. Then within (N + 1)/2 iterations the TFQMR iteration will either break down or terminate with the solution. η = c2 αk−1 x = x + ηd √ If τ m + 1 ≤ b 2 terminate successfully T (d) ρk = r0 w. As a corollary we have a ﬁnite termination result and a convergence estimate for diagonalizable matrices similar to Theorem 3.30). iii. iv. .6. Corollary 3.1.3 to (3. c = 1/ 1 + θ2 τ = τ θc. y2 = y1 − αk−1 v. if ξm > 0. vi. A. b ∈ RN be given.4. Assume that TFQMR does not break down or terminate with the solution for k iterations. rm r0 2 2 ≤ √ −1 m + 1ξm κ(V ) max φ(z) z∈σ(A) for all φ ∈ Pm .Copyright ©1995 by the Society for Industrial and Applied Mathematics.ecsecurehost. kmax) 1.3. The implementation below follows [77] with r0 = r0 used throughout. . θ = 0. η = 0 2. tfqmr(x. Theorem 3. Let A be an N × N matrix and let x0 . v. Let A = V ΛV −1 be a nonsingular diagonalizable matrix. d = 0 T ρ0 = r0 r0 . b. This electronic version is for personal use and may not be duplicated or distributed.2. τ = r 2 . ˆ Algorithm 3. GMRES ITERATION 53 Proof.6.1. by the quasiminimization property GM rm RES 2 ≥ ξm τm as desired. 2 (m = 2k − 2 + j) w = w − αk−1 uj d = yj + (θ2 η/αk−1 )d √ θ = w 2 /τ .com/SIAM/FR16. w1 = y1 = r0 = b − Ax. We may write GM GM rm RES = r0 + Ym zm RES and obtain GM rm RES 2 GM = Wm+1 Ω−1 (fm+1 −Hm zm RES ) m+1 2 GM ≥ ξm fm+1 −Hm zm RES 2. Do While k < kmax (a) k = k + 1 T (b) σk−1 = r0 v. k = 0.1. u1 = v = Ay1 .6. α = ρk−1 /σk−1 . For 1 ≤ m ≤ 2k let ξm be the smallest singular value of Wm+1 Ω−1 m+1 and let xm be given by (3. We apply Theorem 3.
we expect secondorder accuracy and terminate the iteration when rk 2 ≤ h2 b 2 ≈ 9. 54 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS (e) y1 = w + βy2 .7. and iteration parameters to specify the maximum number of iterations and the termination criterion. rather we pass the preconditioned problem. y) +a2 (x. whose storage requirements do not increase with the number of iterations. We do not pass the preconditioner to the GMRES code. 0) = u(x. and TFQMR.html. y)) + a1 (x. y) = 20y. As a preconditioner we use the fast Poisson solver fish2d described in § 2. BiCGSTAB. the righthand side vector b. y) = 1. a2 (x. In all the GMRES examples.8 × 10−4 b 2 . y) = f (x. For methods like GMRES(m). y < 1. u1 = Ay1 (f) v = u1 + β(u2 + βv) Note that y2 and u2 = Ay2 need only be computed if the loop in step 2c does not terminate when j = 1. For general coeﬃcients {aj } the operator L is not selfadjoint and its discretization is not symmetric. This electronic version is for personal use and may not be duplicated or distributed. y)ux (x. xj ) where xi = ih for 1 ≤ i ≤ n. we allow for more iterations. We compute Lu in a matrixfree manner as we did in § 2.33) (Lu)(x. y)u(x. 3. For the computations reported in this section we took a1 (x.7 we discretize with a ﬁvepoint centered difference scheme with n2 points and mesh width h = 1/(n + 1). y) = u(1. y) + a3 (x. y) on 0 < x. and [139]. [34]. described in the comment lines.ecsecurehost. [121]. CGNR. As in § 2. we limit the number of GMRES iterations to 60. 0 < x. y) + uyy (x.7.com/SIAM/FR16. . and a3 (x. We let n = 31 to create a system with 961 unknowns. y) = −(uxx (x. The unknowns are uij ≈ u(xi . y) = 0. Buy this book from SIAM at http://www.Copyright ©1995 by the Society for Industrial and Applied Mathematics. The motivation for this choice is similar to that for the conjugate gradient computations and has been partially explained theoretically in [33]. We take advantage of this in the MATLAB code tfqmr to potentially reduce the matrixvector product cost by one. y < 1 subject to homogeneous Dirichlet boundary conditions u(x. y)uy (x. are the initial iterate x0 . As before. We consider the discretization of the partial diﬀerential equation (3. Examples for GMRES iteration These examples use the code gmres from the collection of MATLAB codes. a MATLAB function for the matrixvector product. y) = 1.7. 1) = u(0. The inputs. the preconditioned system is GLu = Gf . If we let Gu denote the action of the Poisson solver on u.
The execution times in our environment indicate that the preconditioned iterations are even faster than the diﬀerence in operation counts indicates. the restarted iteration terminated successfully.1 and 3. Note the importance of preconditioning. this eﬀect has serious consequences as Fig. Restarting after 3 iterations is more costly in terms of the iteration count than not restarting. As in § 2. However. CGNR has the eﬀect of squaring the condition number. the transpose (adjoint) of L is given by L∗ u = −uxx − uyy − a1 ux − a2 uy + a3 u as one can derive by integration by parts. CGNR (or CGNE) is an alternative to GMRES that does not require storage of the iteration history. The cost of the application of the transpose of L or GL is the same as that of the application of L or GL itself. a single CGNR iteration costs roughly the same as two GMRES iterations. we require the transpose of GL.7 million ﬂoating point operations and the preconditioned iteration required 13 iterations and 2. In other computing environments preconditioners which vectorize or parallelize particularly well might perform in the same way. Examples for CGNR. The preconditioned iteration required 1.8.6 million ﬂoating point operations and converged in 56 iterations. In this case the MATLAB flops command indicates that the unpreconditioned iteration terminated after 223 iterations and 9. we restarted the algorithm after three iterations. in the case where the cost of the matrixvector product dominates the computation. Hence. In Figs.33).com/SIAM/FR16. In both ﬁgures the solid line is the history of the preconditioned iteration and the dashed line that of the unpreconditioned.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 3.6 million ﬂoating point operations. BiCGSTAB.5 ). This electronic version is for personal use and may not be duplicated or distributed.7 we took the right hand side so that the exact solution was the discretization of 10xy(1 − x)(1 − y) exp(x4. . For elliptic problems like (3. To apply CGNR to the preconditioned problem. even in the unpreconditioned case. which is given by (GL)∗ u = L∗ G∗ u = L∗ Gu.ecsecurehost. 3. However.2 we plot iteration histories corresponding to preconditioned or unpreconditioned GMRES. The MATLAB flops command indicates that the unpreconditioned iteration required 7. Note that the plots are semilog plots of r 2 / b 2 and that the righthand sides in the preconditioned and unpreconditioned cases are not the same. For the restarted iteration. The reason for this is that the FFT routine is a MATLAB intrinsic and is not interpreted MATLAB code. The dashed line corresponds to Buy this book from SIAM at http://www. GMRES ITERATION 55 u0 = 0 was the initial iterate. and TFQMR iteration When a transpose is available. GMRES(3) in the terminology of § 3.3 million ﬂoating point operations and terminated after 8 iterations.3 shows. 3.html.4.
The MATLAB flops command indicates that the unpreconditioned iteration required 13.2. We limited the number of iterations to 310 = 10m and the unpreconditioned formulation of CGNR had made very little progress after 310 iterations.1.. The preconditioned formulation did quite well. This is a result of squaring the large condition number.ecsecurehost.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 3. 3.e.. CG applied to L∗ Lu = L∗ f .7 million ﬂoatingpoint operations Buy this book from SIAM at http://www. . GMRES for 2D elliptic equation. CGNR on the unpreconditioned problem. The solid line corresponds to the CGNR applied to the preconditioned problem. i. 56 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 0 10 10 Relative Residual 1 10 2 10 3 10 4 0 10 20 30 Iterations 40 50 60 Fig. 10 0 10 Relative Residual 1 10 2 10 3 10 4 0 50 100 Iterations 150 200 250 Fig. The matrix corresponding to the discretization of L∗ L has a large condition number. converging in eight iterations. This electronic version is for personal use and may not be duplicated or distributed.e.com/SIAM/FR16. GMRES(3) for 2D elliptic equation. CG applied to L∗ G2 Lu = L∗ G2 f . i.html.
CGNE has similar performance. This code has the same input/output arguments as gmres. 3. GMRES ITERATION 57 and the preconditioned iteration 2. since our unpreconditioned matrixvector product is so inexpensive. We applied BiCGSTAB to both the preconditioned and unpreconditioned forms of (3.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 3. possibly. We plot only full iterations (m = 2k) except.9.8.html. As in the previous examples. CGNR for 2D elliptic equation. In Fig. We see a considerable improvement in cost over CGNR and. The examples for BiCGSTAB use the code bicgstab from the collection of MATLAB codes.5 we plot the convergence histories in terms of the approximate residual 2k + 1τ2k given by (3. the behavior of even the preconditioned iteration can also be poor if the coeﬃcients of the ﬁrst derivative terms are too large. for the ﬁnal iterate.4.ecsecurehost. and the graphs are monotonic. with the unpreconditioned iteration being very irregular.33) with the same data and termination criterion as for the GMRES example. 10 0 10 Relative Residual 1 10 2 10 3 10 4 0 50 100 150 200 Iterations 250 300 350 Fig.9.√3.3. see Exercise 3. The unpreconditioned iteration took 40 iterations and 2. . However.2 million ﬂoatingpoint operations. the solid line corresponds to the preconditioned iteration and the dashed line to the unpreconditioned iteration. The approximate residual is given in terms of the quasiresidual norm τm rather than the true residual. as you are asked to investigate in Exercise 3. The preconditioned iteration terminated in six iterations (12 matrixvector products) and required roughly 1. We repeat the experiment with TFQMR as our linear solver. This electronic version is for personal use and may not be duplicated or distributed.9. a better cost/iterate than GMRES in the unpreconditioned case. Our Buy this book from SIAM at http://www. We use the code tfqmr that is included in our collection.7 million ﬂoatingpoint operations. Neither convergence history is monotonic. We plot the iteration histories in Fig.com/SIAM/FR16.31). but not varying by many orders of magnitude as a CGS or BiCG iteration might.4 million.
31) as described in √ § 3. 10 0 10 Relative Approximate Residual 1 10 2 10 3 10 4 10 5 0 10 20 30 40 Iterations 50 60 70 Fig.6. The preconditioned iteration was much more eﬃcient. 3.html. stopping the iteration when τm m + 1/ b 2 ≤ h−2 . BiCGSTAB for 2D elliptic equation.com/SIAM/FR16. 3.Copyright ©1995 by the Society for Industrial and Applied Mathematics.ecsecurehost. indicating that it is reasonable to use the quasiresidual estimate (3. . The plateaus in the graph of the convergence history for the Buy this book from SIAM at http://www.31) to make termination decisions. TFQMR for 2D elliptic equation.4. This electronic version is for personal use and may not be duplicated or distributed. The unpreconditioned iteration required roughly 4. 58 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 1 10 10 0 Relative Residual 10 1 10 2 10 3 10 4 0 5 10 15 20 Iterations 25 30 35 40 Fig.5. termination criterion is based on the quasiresidual and (3.4.1 million ﬂoatingpoint operations and 67 iterations for termination.9 million ﬂoatingpoint operations and 7 iterations. taking 1. This led to actual relative residuals of 7 × 10−5 for the unpreconditioned iteration and 2 × 10−4 for the preconditioned iteration.
html. This electronic version is for personal use and may not be duplicated or distributed.ecsecurehost. GMRES ITERATION 59 unpreconditioned iteration are related (see [196]) to spikes in the corresponding graph for a CGS iteration.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Buy this book from SIAM at http://www.com/SIAM/FR16. .
That is. How would you test for failure in a restarted algorithm? What kinds of failures can occur? Repeat Exercise 6 with a limit of three GMRES iterations before a restart. and λ ∈ R such that the GMRES residuals satisfy rk 2 = r0 2 for all k < N . Duplicate Table 3.1. . Let A = Jλ be an elementary Jordan block of size N corresponding to an eigenvalue λ = 0. Give an example of a matrix that is not diagonalizable.9.4. Solve this problem with GMRES (without preconditioning) and then apply as a preconditioner the map M deﬁned by −(M f ) = f. 100. Buy this book from SIAM at http://www. .com/SIAM/FR16.1 and add two more columns corresponding to classical Gram–Schmidt orthogonalization with reorthogonalization at every step and determine when the test in (3. Prove that the Arnoldi process (unless it terminates prematurely with a solution) produces a basis for Kk+1 that satisﬁes (3.5. b ∈ RN .5... (M f )(0) = (M f )(1) = 0. See [167] for another example. λ 0 .Copyright ©1995 by the Society for Industrial and Applied Mathematics. 0 0 .4 and Theorem 3. . How do other methods such as CGNR. 3.2. This means that Jλ = λ 0 1 λ . u(0) = u(1) = 0. and TFQMR perform? 3.6. Prove Theorem 3. . 3. How does the performance of the iteration depend on the mesh? Try other preconditioners such as Gauss–Seidel and Jacobi.9.9.1.8).ecsecurehost. One way to do this is to call the gmres code with another code which has an outer loop to control the restarting as in Algorithm gmresm. How do your conclusions compare with those in [160]? 3. CGNE. 60 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 3. Exercises on GMRES 3.7. .html. . Consider the centered diﬀerence discretization of −u + u + u = 1.9. Modify the MATLAB GMRES code to allow for restarts. 200.9. Try this for meshes with 50..9. points.9.10) detects loss of orthogonality.. 3.9. 1 λ 0 λ . precondition with a solver for the highorder term in the diﬀerential equation using the correct boundary conditions. 3. This electronic version is for personal use and may not be duplicated or distributed. .1.3. 0 1 Give an example of x0 . BiCGSTAB. . 0 1 .
Apply CGNE to the PDE in § 3.8. Try ay = 5y.9. and/or CGNE. How does the performance diﬀer from CGNR and GMRES? 3.10.12.7 and § 3. apply GMRES. Buy this book from SIAM at http://www. CGNR. Duplicate the results in § 3.7.8. Implement BiCG and CGS and try them on some of the examples in this chapter.com/SIAM/FR16. Compare the execution times in your computing environment. Why is the performance of the methods sensitive to ay and ax ? 3. Consider preconditioning from the right.7 and § 3.9. Compare timings and solution accuracy with those obtained by left preconditioning. BiCGSTAB. Experiment with diﬀerent values of m.9. Why might GMRES(m) be faster than GMRES? Vary the dimension and the coeﬃcient ay . This electronic version is for personal use and may not be duplicated or distributed.Copyright ©1995 by the Society for Industrial and Applied Mathematics. GMRES ITERATION 61 3.7. For the PDE in § 3. . TFQMR. Are the solutions to (3.ecsecurehost.9.9. 3.html.8 equally accurate? Compare the ﬁnal results with the known solution.11.33) in § 3.9. 3. to the equation LGw = f and then recover the solution from u = Gw.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
62
ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
Chapter 4
Basic Concepts and FixedPoint Iteration
This part of the book is about numerical methods for the solution of systems of nonlinear equations. Throughout this part, we will let · denote a norm on RN as well as the induced matrix norm. We begin by setting the notation. We seek to solve (4.1) F (x) = 0. Here F : RN → RN . We denote the ith component of F by fi . If the components of F are diﬀerentiable at x ∈ RN we deﬁne the Jacobian matrix F (x) by ∂fi F (x)ij = (x). ∂xj The Jacobian matrix is the vector analog of the derivative. We may express the fundamental theorem of calculus as follows. Theorem 4.0.1. Let F be diﬀerentiable in an open set Ω ⊂ RN and let ∗ ∈ Ω. Then for all x ∈ Ω suﬃciently near x∗ x F (x) − F (x∗ ) = 4.1. Types of convergence Iterative methods can be classiﬁed by the rate of convergence. Definition 4.1.1. Let {xn } ⊂ RN and x∗ ∈ RN . Then • xn → x∗ qquadratically if xn → x∗ and there is K > 0 such that xn+1 − x∗ ≤ K xn − x∗ 2 . • xn → x∗ qsuperlinearly with qorder α > 1 if xn → x∗ and there is K > 0 such that xn+1 − x∗ ≤ K xn − x∗ α . • xn → x∗ qsuperlinearly if
n→∞ 1 0
F (x∗ + t(x − x∗ ))(x − x∗ ) dt.
lim
xn+1 − x∗ = 0. xn − x∗
65
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
66
ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
• xn → x∗ qlinearly with qfactor σ ∈ (0, 1) if xn+1 − x∗ ≤ σ xn − x∗ for n suﬃciently large. Definition 4.1.2. An iterative method for computing x∗ is said to be locally (qquadratically, qsuperlinearly, qlinearly, etc.) convergent if the iterates converge to x∗ (qquadratically, qsuperlinearly, qlinearly, etc.) given that the initial data for the iteration is suﬃciently good. Note that a qsuperlinearly convergent sequence is also qlinearly convergent with qfactor σ for any σ > 0. A qquadratically convergent sequence is qsuperlinearly convergent with qorder 2. In general, a qsuperlinearly convergent method is preferable to a qlinearly convergent one if the cost of a single iterate is the same for both methods. We will see that often a method that is more slowly convergent, in terms of the convergence types deﬁned above, can have a cost/iterate so low that the slow iteration is more eﬃcient. Sometimes errors are introduced into the iteration that are independent of the errors in the approximate solution. An example of this would be a problem in which the nonlinear function F is the output of another algorithm that provides error control, such as adjustment of the mesh size in an approximation of a diﬀerential equation. One might only compute F to low accuracy in the initial phases of the iteration to compute a solution and tighten the accuracy as the iteration progresses. The rtype convergence classiﬁcation enables us to describe that idea. Definition 4.1.3. Let {xn } ⊂ RN and x∗ ∈ RN . Then {xn } converges to x∗ r(quadratically, superlinearly, linearly) if there is a sequence {ξn } ⊂ R converging q(quadratically, superlinearly, linearly) to zero such that xn − x∗ ≤ ξn . We say that {xn } converges rsuperlinearly with rorder α > 1 if ξn → 0 qsuperlinearly with qorder α. 4.2. Fixedpoint iteration Many nonlinear equations are naturally formulated as ﬁxedpoint problems (4.2) x = K(x) where K, the ﬁxedpoint map, may be nonlinear. A solution x∗ of (4.2) is called a ﬁxed point of the map K. Such problems are nonlinear analogs of the linear ﬁxedpoint problems considered in Chapter 1. In this section we analyze convergence of ﬁxedpoint iteration, which is given by (4.3) xn+1 = K(xn ).
This iterative method is also called nonlinear Richardson iteration, Picard iteration, or the method of successive substitution.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
3) converges qlinearly to x∗ with qfactor γ for all initial iterates x0 ∈ Ω. The standard result for ﬁxedpoint iteration is the Contraction Mapping Theorem [9]. Let x0 ∈ Ω.2. Hence lim xn+k − xn = 0 n. Proof. and the iteration deﬁned by (4. for all n. . Let Ω ⊂ RN and let G : Ω → RM .2.2. G is Lipschitz continuous on Ω with Lipschitz constant γ if G(x) − G(y) ≤ γ x − y for all x. Buy this book from SIAM at http://www. Let Ω ⊂ RN . Definition 4. .html. k ≥ 0. Compare the proof to that of Lemma 1. Theorem 4. This electronic version is for personal use and may not be duplicated or distributed.com/SIAM/FR16.1. Then there is a unique ﬁxed point of K. and therefore xn − x0 = ≤ n−1 i=0 xi+1 n−1 i=0 − xi n−1 i i=0 γ xi+1 − xi ≤ x1 − x0 ≤ x1 − x0 /(1 − γ).2. FIXEDPOINT ITERATION 67 Before discussing convergence of ﬁxedpoint iteration we make two deﬁnitions. then x∗ − y ∗ = K(x∗ ) − K(y ∗ ) ≤ γ x∗ − y ∗ . Note that {xn } ⊂ Ω because x0 ∈ Ω and K(x) ∈ Ω whenever x ∈ Ω. ≤ γ n xk − x0 ≤ γ n x1 − x0 /(1 − γ).k→∞ and therefore the sequence {xn } is a Cauchy sequence and has a limit x∗ [162].1. y ∈ Ω. .2. .1. Now. x∗ ∈ Ω. If K has two ﬁxed points x∗ and y ∗ in Ω. Definition 4. xn+k − xn = K(xn+k−1 ) − K(xn−1 ) ≤ γ xn+k−1 − xn−1 ≤ γ K(xn+k−2 ) − K(xn−2 ) ≤ γ 2 xn+k−2 − xn−2 ≤ . Let Ω be a closed subset of RN and let K be a contraction mapping on Ω with Lipschitz constant γ < 1 such that K(x) ∈ Ω for all x ∈ Ω. The sequence {xn } remains bounded since for all i ≥ 1 xi+1 − xi = K(xi ) − K(xi−1 ) ≤ γ xi − xi−1 .1 and Corollary 1.2. ≤ γ i x1 − x0 . .Copyright ©1995 by the Society for Industrial and Applied Mathematics. K : Ω → RN is a contraction mapping on Ω if K is Lipschitz continuous on Ω with Lipschitz constant γ < 1.ecsecurehost.
consider the initial value problem y = f (y). Hence the ﬁxed point is unique. Nonlinear variations of the classical stationary iterative methods such as Gauss–Seidel have also been used. e.com/SIAM/FR16.4) y n+1 − y n = hf (y n+1 ). 2.3. One simple application of this theorem is to numerical integration of ordinary diﬀerential equations by implicit methods. (4. .3. If hM < 1. say f (y) ≤ M and f (x) − f (y) ≤ M x − y for all x. x∗ = y ∗ . y. y(t0 ) = y0 and its solution by the backward Euler method with time step h. F : Ω → RN ×N is Lipschitz continuous with Lipschitz constant γ. Equation 4. Advancing in time from y n to y n+1 requires solution of the ﬁxedpoint problem y = K(y) = y n + hf (y). see [145] for a more complete discussion of these algorithms. and the methods based on Newton’s method which we present in the subsequent chapters are used much more often [83].1 has a solution x∗ . This type of analysis was used by Picard to prove existence of solutions of initial value problems [152]. This electronic version is for personal use and may not be duplicated or distributed. The standard assumptions We will make the standard assumptions on F . In (4. i. 68 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Since γ < 1 this can only hold if x∗ − y ∗ = 0. Buy this book from SIAM at http://www.html. From this we conclude that for h suﬃciently small we can integrate forward in time and that the time step h can be chosen independently of y. we may apply the contraction mapping theorem with Ω = R since for all x. 1. This time step may well be too small to be practical. Assumption 4. 4. Finally we note that xn+1 − x∗ = K(xn ) − K(x∗ ) ≤ γ xn − x∗ .1. For simplicity we assume that f is bounded and Lipschitz continuous.ecsecurehost.4) y k denotes the approximation of y(t0 + kh).Copyright ©1995 by the Society for Industrial and Applied Mathematics. y K(x) − K(y) = hf (x) − f (y) ≤ hM x − y. which completes the proof. For example.
This will follow from δ< since then (4.3.7.5) holds if γδ < F (x∗ ) .com/SIAM/FR16. Assume that the standard assumptions hold. Lemma 4. Throughout this part.1 to obtain F (x) ≤ 1 0 F (x∗ + te) e dt ≤ 2 F (x∗ ) e Buy this book from SIAM at http://www. Exercise 5. If xn is the nth iterate of a sequence.5) (4. we note that if x ∈ B(δ) then x∗ + te ∈ B(δ) for all 0 ≤ t ≤ 1. We use (4. These assumptions can be weakened [108] without sacriﬁcing convergence of the methods we consider here. FIXEDPOINT ITERATION 69 3.1 illustrates one way to weaken the standard assumptions. Lemma 4. where e = x − x∗ .6) and (4.Copyright ©1995 by the Society for Industrial and Applied Mathematics.3. F (x∗ ) is nonsingular.ecsecurehost. For all x ∈ B(δ) we have F (x) ≤ F (x∗ ) + γ e . F (x)−1 ≤ 2 F (x∗ )−1 .5) and Theorem 4.8) I − F (x∗ )−1 F (x) = F (x∗ )−1 (F (x∗ ) − F (x)) ≤ γ F (x∗ )−1 e ≤ γδ F (x∗ )−1 < 1/2. The next result (4. en = xn − x∗ is the error in that iterate.6) is a direct consequence of the Banach Lemma if I − F (x∗ )−1 F (x) < 1/2. F (x∗ )−1 2γ −1 To prove the ﬁnal inequality (4.7) F (x∗ )−1 F (x) ≤ 2 F (x∗ ) . However the classical result on quadratic convergence of Newton’s method requires them and we make them throughout.0. we will always denote a root of F by x∗ .1.1 is an important consequence of the standard assumptions. This electronic version is for personal use and may not be duplicated or distributed. .html. Then there is δ > 0 so that for all x ∈ B(δ) (4. Hence (4. −1 e /2 ≤ F (x) ≤ 2 F (x∗ ) e .7). The notation introduced above will be used consistently. Let δ be small enough so that B(δ) ⊂ Ω. We let B(r) denote the ball of radius r about x∗ B(r) = {x  e < r}. Proof.
This electronic version is for personal use and may not be duplicated or distributed. I − F (x∗ )−1 F (x∗ + te)dt ) ≥ e /2.Copyright ©1995 by the Society for Industrial and Applied Mathematics.html. 70 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS which is the right inequality in (4.7) note that F (x∗ )−1 F (x) = F (x∗ )−1 =e− and hence. Buy this book from SIAM at http://www.8) F (x∗ )−1 F (x) ≥ e (1 − Therefore 1 0 1 0 1 0 F (x∗ + te)e dt (I − F (x∗ )−1 F (x∗ + te))e dt.ecsecurehost.7). . by (4. which completes the proof.com/SIAM/FR16. e /2 ≤ F (x∗ )−1 F (x) ≤ F (x∗ )−1 F (x) . To prove the left inequality in (4.
The proof of convergence of the complete Newton iteration will be complete if we reduce δ if needed so that Kδ < 1.3.2) with K = γ F (x∗ )−1 . By Theorem 4.1.2.0. Theorem 5. Let the standard assumptions hold. 71 Buy this book from SIAM at http://www.1. Chapter 5 Newton’s Method 5. (5.com/SIAM/FR16. Let δ be small enough so that the conclusions of Lemma 4.1) We may also view x+ as the root of the twoterm Taylor expansion or linear model of F about xc Mc (x) = F (xc ) + F (xc )(x − xc ). This completes the proof of (5.html.3. (5.1. [156].Copyright ©1995 by the Society for Industrial and Applied Mathematics. In the context of single systems this method appeared in the 17th century [140].1 hold. [13]. In this language. Then there is δ such that if x0 ∈ B(δ) the Newton iteration xn+1 = xn − F (xn )−1 F (xn ) converges qquadratically to x∗ . Local convergence of Newton’s method We will describe iterative methods for nonlinear equations in terms of the transition from a current iterate xc to a new iterate x+ . Let the standard assumptions hold. This electronic version is for personal use and may not be duplicated or distributed.1 e+ = ec − F (xc )−1 F (xc ) = F (xc )−1 1 0 (F (xc ) − F (x∗ + tec ))ec dt.1) satisﬁes e+ ≤ K ec 2 . [137].3. By Lemma 4.ecsecurehost. Then there are K > 0 and δ > 0 such that if xc ∈ B(δ) the Newton iterate from xc given by (5. The convergence result on Newton’s method follows from Lemma 4.1. .1 and the Lipschitz continuity of F e+ ≤ (2 F (x∗ )−1 )γ ec 2 /2. Newton’s method is x+ = xc − F (xc )−1 F (xc ).1. Theorem 5.2) Proof.
3. where the initial iterate is derived from the solution at the previous time step. Let δ > 0 be small enough so that the conclusions of Lemma 4. [126]. From Lemma 5.2. there is no “righthand side” to use as a scaling factor and we must balance the relative and absolute size of the nonlinear residuals in some other way.com/SIAM/FR16.3. when the initial iterate is far from the root and methods such as those discussed in Chapter 8 are needed. [16]. In all Buy this book from SIAM at http://www. a termination decision based on the relative residual may be made too late in the iteration or. Assume that the standard assumptions hold.1 hold. the entire sequence {xn } ⊂ B(δ). 5. if there is error in the evaluation of F or the initial iterate is near a root.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Lemma 5.2) and (1. [105].1 hold. local convergence results like those in this and the following two chapters describe the terminal phase of the iteration. Then for all x ∈ B(δ) e F (x) 4κ(F (x∗ )) e ≤ ≤ .1. [83]. The assumption that the initial iterate be “suﬃciently near” the solution (x0 ∈ B(δ)) may seem artiﬁcial at ﬁrst look. However. then if F (x∗ ) is well conditioned. Let δ be small enough so that the conclusions of Theorem 5.1.1 implies that (5. where the initial iterate is an interpolation of a solution from a coarser computational mesh [99]. as we will see in the next section. 72 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Proof.1 we conclude that if F (x∗ ) is well conditioned. the iteration may not terminate at all. however.1.ecsecurehost. .4).3) then implies that xn → x∗ qquadratically.1.1 hold for x ∈ B(δ). we may terminate the iteration when the relative nonlinear residual F (x) / F (x0 ) is small. as a direct consequence of applying Lemma 4.3) en+1 ≤ K en 2 ≤ η en < en and hence xn+1 ∈ B(ηδ) ⊂ B(δ).2. If x ∈ B(δ). Therefore if xn ∈ B(δ) we may continue the iteration. ∗ )) 4 e0 κ(F (x F (x0 ) e0 where κ(F (x∗ )) = F (x∗ ) F (x∗ )−1 is the condition number of F (x∗ ) relative to the norm · .1. many situations in which the initial iterate is very near the root. There are. where δ is small enough so that the conclusions of Lemma 4. Termination of the iteration We can base our termination decision on Lemma 4. Moreover.3.1. This electronic version is for personal use and may not be duplicated or distributed. Two examples are implicit integration of ordinary diﬀerential equations and diﬀerential algebraic equations. Then if n ≥ 0 and xn ∈ B(δ).2. We have. and solution of discretizations of partial diﬀerential equations. Since x0 ∈ B(δ) by assumption. a nonlinear form of Lemma 1. Theorem 5. (5. Reduce δ if needed so that Kδ = η < 1.html. In the nonlinear case. We raised this issue in the context of linear equations in Chapter 1 when we compared (1.3.1 twice. the size of the relative nonlinear residual is a good indicator of the size of the error.
[151]. and [151]. [21]. whereas by terminating on small relative residuals or on a condition like (5. the relation between the step and the error is less clear. then e+ ≤ σ ec implies that (1 − σ) ec ≤ s ≤ (1 + σ) ec . then termination on small steps becomes more attractive. If computation and factorization of the Jacobian are inexpensive. [23]. NEWTON’S METHOD 73 of our MATLAB codes and algorithms for nonlinear equations our termination criterion is to stop the iteration if (5. In order to compute the Newton iterate x+ from a current point xc one must ﬁrst evaluate F (xc ) and decide whether to terminate the iteration. However. 5. where the relative error tolerance τr and absolute error tolerance τa are input to the algorithm.3. and terminate the iteration when s is suﬃciently small. [23]. Hence. Our examples and discussions of operation counts make the implicit assumption that the Jacobian is dense. . Another way to decide whether to terminate is to look at the Newton step s = −F (xc )−1 F (xc ) = x+ − xc .com/SIAM/FR16. If the iteration is qlinearly convergent. For methods other than Newton’s method. the issues for sparse Jacobians when direct methods are used to solve linear systems are very similar. in order to estimate ec this way.1. This electronic version is for personal use and may not be duplicated or distributed.4) the decision to stop the iteration can be made before computation and factorization of F (xc ). the Jacobian F (xc ) must be computed and factored. Buy this book from SIAM at http://www. which implies that ec = s + O( ec 2 ).html. near the solution s and ec are essentially the same size. The diﬀerential equations codes discussed in [16]. This criterion is based on Theorem 5. However. [21].4) F (x) ≤ τr F (x0 ) + τa . Combinations of relative and absolute error tolerances are commonly used in numerical methods for ordinary diﬀerential equations or diﬀerential algebraic equations [16].1. one must do most of the work needed to compute x+ . Implementation of Newton’s method In this section we assume that direct methods will be used to solve the linear equation for the Newton step. Hence the step is a reliable indicator of the error provided σ is not too near 1. If one decides to continue. make use of this in the special case of the chord method. In Chapter 6 we consider algorithms in which iterative methods are used to solve the equation for the Newton step.ecsecurehost. say.Copyright ©1995 by the Society for Industrial and Applied Mathematics.
1.com/SIAM/FR16. the evaluation and factorization of F are the most costly. and a vector of termination tolerances τ = (τr .ecsecurehost. and directly than with diﬀerences and the analysis above for the cost of a Newton iterate is very pessimistic. In the future. One approach to reducing the cost of items 2a and item 2b in the Newton iteration is to move them outside of the main loop.2. The inputs of the algorithm are the initial iterate x. 74 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Then the step is computed as the solution of F (xc )s = −F (xc ) and the iterate is updated x+ = xc + s. accurately. r0 = F (x) 2. (c) Solve LU s = −F (x) (d) x = x + s (e) Evaluate F (x). automatic diﬀerentiation (see [94] for a collection of articles on this topic) may provide such an alternative.html. Hence the cost of a Newton step may be roughly estimated as N + 1 evaluations of F and O(N 3 ) ﬂoatingpoint operations. the nonlinear map F . x+ = xc − F (x0 )−1 F (xc ). τa ) ∈ R2 . newton(x. Evaluation of F by ﬁnite diﬀerences should be expected to cost N times the cost of an evaluation of F because each column in F requires an evaluation of F to form the diﬀerence approximation. τ ) 1.3. r0 = F (x) 2. Factor F (x) = LU .3. For general nonlinearities or general purpose codes such as the MATLAB code nsol from the collection. τ ) 1.7. Algorithm 5. chord(x.21 for an example. Do while F (x) > τr r0 + τa (a) Compute F (x) (b) Factor F (x) = LU . Algorithm 5. For deﬁniteness in the description of Algorithm newton we use an LU factorization of F . F. This method is called the chord method. there may be little alternative to diﬀerence Jacobians. Buy this book from SIAM at http://www. This electronic version is for personal use and may not be duplicated or distributed. Of these steps. In many cases F can be computed more eﬃciently. Factorization of F in the dense case costs O(N 3 ) ﬂoatingpoint operations. See Exercise 5. Any other appropriate factorization such as QR or Cholesky could be used as well. This means that the linear approximation of F (x) = 0 that is solved at each iteration has derivative determined by the initial iterate. F.Copyright ©1995 by the Society for Industrial and Applied Mathematics. The inputs are the same as for Newton’s method. . Compute F (x) 3.
δ > 0.5) ¯ e+ ≤ K( ec 2 + ∆(xc ) ec + (xc ) ).3.4. + Lemma 4.3.1 and Theorem 5. F (xc ) + ∆(xc ) is nonsingular) and satisﬁes (5. We continue in the next section with an analysis of the eﬀects of errors in the function and Jacobian in a general context and then consider the chord method and diﬀerence approximation to F as applications. Let the standard assumptions hold. Then there are ¯ K > 0.4. These issues are discussed in [201] as well. NEWTON’S METHOD 75 4. Proof. for example.1. Let xN = xc − F (xc )−1 F (xc ) + and note that x+ = xN + (F (xc )−1 − (F (xc ) + ∆(xc ))−1 )F (xc ) − (F (xc ) + ∆(xc ))−1 (xc ).1 and Theorem 5. We will refer to this phase of the iteration in which the nonlinear residual is no longer being reduced as stagnation of the iteration. + (F (xc ) + ∆(xc ))−1 Buy this book from SIAM at http://www. Theorem 5. and δ1 > 0 such that if xc ∈ B(δ) and ∆(xc ) < δ1 then x+ = xc − (F (xc ) + ∆(xc ))−1 (F (xc ) + (xc )) is deﬁned (i. The diﬀerence in the iteration itself is that an approximation to F (xc ) is used. Similar diﬀerences arise if F (xc ) is numerically approximated by diﬀerences. This electronic version is for personal use and may not be duplicated or distributed. the resulting iteration can return a result that is an O( ) accurate approximation to x∗ .1 hold. Let δ be small enough so that the conclusions of Lemma 4.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Errors in the function and derivative Suppose that F and F are computed inaccurately so that F + and F + ∆ are used instead of F and F in the iteration.. 5. Do while F (x) > τr r0 + τa (a) Solve LU s = −F (x) (b) x = x + s (c) Evaluate F (x).html.6) e+ ≤ K ec 2 + 2 F (xc )−1 − (F (xc ) + ∆(xc ))−1 ) F (x∗ ) ec (xc ) . This is diﬀerent from convergence and was called “local improvement” in [65]. is entirely due to ﬂoatingpoint roundoﬀ. If ∆ is suﬃciently small.1 imply (5.e. there is no reason to expect that F (xn ) will ever be smaller than in general.1. If.com/SIAM/FR16.ecsecurehost.1. The only diﬀerence in implementation from Newton’s method is that the computation and factorization of the Jacobian are done before the iteration is begun. .
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
76 If
ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
∆(xc ) ≤ F (x∗ )−1 ∆(xc ) ≤ F (xc )−1
−1
/4
then Lemma 4.3.1 implies that
−1
/2
and the Banach Lemma implies that F (xc ) + ∆(xc ) is nonsingular and (F (xc ) + ∆(xc ))−1 ≤ 2 F (xc )−1 ≤ 4 F (x∗ )−1 . Hence F (xc )−1 − (F (xc ) + ∆(xc ))−1 ≤ 8 F (x∗ )−1 We use these estimates and (5.6) to obtain e+ ≤ K ec Setting
2 2
∆(xc ) .
+ 16 F (x∗ )−1
2
F (x∗ ) ∆(xc ) ec + 4 F (x∗ )−1
2
(xc ) .
¯ K = K + 16 F (x∗ )−1
F (x∗ ) + 4 F (x∗ )−1
completes the proof. The remainder of this section is devoted to applications of Theorem 5.4.1. 5.4.1. The chord method. Recall that the chord method is given by
x+ = xc − F (x0 )−1 F (xc ). In the language of Theorem 5.4.1 (xc ) = 0, ∆(xc ) = F (x0 ) − F (xc ). Hence, if xc , x0 ∈ B(δ) ⊂ Ω (5.7) ∆(xc ) ≤ γ x0 − xc ≤ γ( e0 + ec ).
We apply Theorem 5.4.1 to obtain the following result. Theorem 5.4.2. Let the standard assumptions hold. Then there are KC > 0 and δ > 0 such that if x0 ∈ B(δ) the chord iterates converge qlinearly to x∗ and (5.8) en+1 ≤ KC e0 en . Proof. Let δ be small enough so that B(δ) ⊂ Ω and the conclusions of Theorem 5.4.1 hold. Assume that xn ∈ B(δ). Combining (5.7) and (5.5) implies ¯ ¯ en+1 ≤ K( en (1 + γ) + γ e0 ) en ≤ K(1 + 2γ)δ en . Hence if δ is small enough so that ¯ K(1 + 2γ)δ = η < 1
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
NEWTON’S METHOD
77
then the chord iterates converge qlinearly to x∗ . Qlinear convergence implies ¯ that en < e0 and hence (5.8) holds with KC = K(1 + 2γ). Another variation of the chord method is x+ = xc − A−1 F (xc ), where A ≈ F (x∗ ). Methods of this type may be viewed as preconditioned nonlinear Richardson iteration. Since ∆(xc ) = A − F (xc ) ≤ A − F (x∗ ) + F (x∗ ) − F (xc ) , if xc ∈ B(δ) ⊂ Ω then ∆(xc ) ≤ A − F (x∗ ) + γ ec ≤ A − F (x∗ ) + γδ. Theorem 5.4.3. Let the standard assumptions hold. Then there are KA > 0, δ > 0, and δ1 > 0, such that if x0 ∈ B(δ) and A − F (x∗ ) < δ1 then the iteration xn+1 = xn − A−1 F (xn ) converges qlinearly to x∗ and (5.9) en+1 ≤ KA ( e0 + A − F (x∗ ) ) en .
5.4.2. Approximate inversion of F . Another way to implement chordtype methods is to provide an approximate inverse of F . Here we replace F (x)−1 by B(x), where the action of B on a vector is less expensive to compute than a solution using the LU factorization. Rather than express the iteration in terms of B −1 (x) − F (x) and using Theorem 5.4.3 one can proceed directly from the deﬁnition of approximate inverse. Note that if B is constant (independent of x) then the iteration x+ = xc − BF (xc ) can be viewed as a preconditioned nonlinear Richardson iteration. We have the following result. Theorem 5.4.4. Let the standard assumptions hold. Then there are KB > 0, δ > 0, and δ1 > 0, such that if x0 ∈ B(δ) and the matrixvalued function B(x) satisﬁes (5.10) I − B(x)F (x∗ ) = ρ(x) < δ1
for all x ∈ B(δ) then the iteration xn+1 = xn − B(xn )F (xn ) converges qlinearly to x∗ and (5.11) en+1 ≤ KB (ρ(xn ) + en ) en .
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
78
ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS
Proof. On one hand, this theorem could be regarded as a corollary of the Banach Lemma and Theorem 5.4.1. We give a direct proof. First, by (5.10) we have (5.12) B(x) = B(x)F (x∗ )F (x∗ )−1 ≤ B(x)F (x∗ ) F (x∗ )−1 ≤ MB = (1 + δ1 ) F (x∗ )−1 .
Using (5.12) and e+ = ec − B(xc )F (xc ) =
1 0
(I − B(xc )F (x∗ + tec ))ec dt
1 0
= (I − B(xc )F (x∗ ))ec + B(xc ) we have
(F (x∗ ) − F (x∗ + tec ))ec dt
e+ ≤ ρ(xc ) ec + MB γ ec 2 /2.
This completes the proof with KB = 1 + MB γ/2. 5.4.3. The Shamanskii method. Alternation of a Newton step with a sequence of chord steps leads to a class of highorder methods, that is, methods that converge qsuperlinearly with qorder larger than 2. We follow [18] and name this method for Shamanskii [174], who considered the ﬁnite diﬀerence Jacobian formulation of the method from the point of view of eﬃciency. The method itself was analyzed in [190]. Other highorder methods, especially for problems in one dimension, are described in [190] and [146]. We can describe the transition from xc to x+ by y1 (5.13) = xc − F (xc )−1 F (xc ),
yj+1 = yj − F (xc )−1 F (yj ) for 1 ≤ j ≤ m − 1, x+ = ym .
Note that m = 1 is Newton’s method and m = ∞ is the chord method with {yj } playing the role of the chord iterates. Algorithmically a second loop surrounds the factor/solve step. The inputs for the Shamanskii method are the same as for Newton’s method or the chord method except for the addition of the parameter m. Note that we can overwrite xc with the sequence of yj ’s. Note that we apply the termination test after computation of each yj . Algorithm 5.4.1. sham(x, F, τ, m) 1. r0 = F (x) 2. Do while F (x) > τr r0 + τa
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
≤ KC en j+1 . Setting j = m in the above inequality completes the proof. This electronic version is for personal use and may not be duplicated or distributed.html. (c) for j = 1. 5. Let the standard assumptions hold and let m ≥ 1 be given.4. h Buy this book from SIAM at http://www. Then there are KS > 0 and δ > 0 such that if x0 ∈ B(δ) the Shamanskii iterates converge qsuperlinearly to x∗ with qorder m + 1 and (5. . Hence xn+1 ∈ B(δ). Optimal choices of m as a function of N can be derived under assumptions consistent with dense matrix algebra [18] by balancing the O(N 3 ) cost of a matrix factorization with the cost of function and Jacobian evaluation.2. Then if xn ∈ B(δ) all the intermediate iterates {yj } are in B(δ) by Theorem 5. Diﬀerence approximation to F . m i. if the initial iterate is suﬃciently accurate. only a small number of iterations will be taken. The advantage of the Shamanskii method over Newton’s method is that high qorders can be obtained with far fewer Jacobian evaluations or factorizations. NEWTON’S METHOD 79 (a) Compute F (x) (b) Factor F (x) = LU .2 hold.4. If F (x) ≤ τr r0 + τa exit. .Copyright ©1995 by the Society for Industrial and Applied Mathematics.4.ecsecurehost. . In fact. Proof.8) implies that for 1≤j≤m yj −x∗ ≤ KC xn −x∗ yj−1 −x∗ = KC en j yj−1 −x∗ ≤ .4. .4.2. iv. Assume that we compute F (x) + (x) instead of F (x) and attempt to approximate the action of F on a vector by a forward diﬀerence. A forward diﬀerence approximation to F (x)w would be F (x + hw) + (x + hw) − F (x) − (x) . . (5.5. Here the jth column of F (x) is F (x)ej where ej is the unit vector with jth component 1 and other components 0.14) en+1 ≤ KS en m+1 . for example. The convergence result is a simple application of Theorem 5. Evaluate F (x). Solve LU s = −F (x) ii.4. Let δ be small enough so that B(δ) ⊂ Ω and that the conclusions of Theorem 5. . x = x + s iii. Algorithm nsol is based on this idea and using an idea from [114] computes and factors the Jacobian only until an estimate of the qfactor for the linear convergence rate is suﬃciently low. indicate that the chord method is usually the best option for very large problems. We would do this. Theorem 5. when building an approximate Jacobian. The analysis in [18] and the expectation that. if we set y1 = xn .com/SIAM/FR16.
80 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Assume that (x) ≤ ¯.ecsecurehost. the cost of a forward diﬀerence approximation of F (x)w is an additional evaluation of F (at the point x+hw).4. Because of this we must carefully specify what we mean by a diﬀerence approximation to the Jacobian. the analysis here indicates that h ≈ 10−7 is a reasonable choice. Definition 5.Copyright ©1995 by the Society for Industrial and Applied Mathematics. This can become very important if part of F is computed from measured data.html. Here we have made the implicit assumption that x and w are of roughly the same size. h F (x)w − √ The quantity inside the Oterm is minimized when h = ¯.com/SIAM/FR16.1 below speciﬁes the usual choice. The forward diﬀerence approximation to the directional derivative is not a linear operator in w. This electronic version is for personal use and may not be duplicated or distributed. The reason for this is that the derivative has been approximated. If F (x) has already been computed. The choice h ≈ 10−15 in this case can lead to disaster (try it!).4. Deﬁnition 5. A more important consequence of this analysis is that the choice of step size in a forward diﬀerence approximation to the action of the Jacobian on a vector must take into account the error in the evaluation of F . Hence the evaluation of a full ﬁnite diﬀerence Jacobian would cost N function evaluations. and so linearity has been lost. Let F be deﬁned in a neighborhood of x ∈ RN and let Buy this book from SIAM at http://www. for instance.1. . (∇h F )(x) is the N × N matrix whose jth column is given by F (x + h x ej ) − F (x) h x F (he ) − F (x) j x=0 x=0 (∇h F )(x)j = h We make a similar deﬁnition of the diﬀerence approximation of the directional derivative. Exploitation of special structure could reduce this cost. h should be scaled to reﬂect that.4. If this is not the case. Then F (x + hw) + (x + hw) − F (x) − (x) = O(h + ¯/h). one for each column of the Jacobian. The choice h = ¯1/2 x / w reﬂects all the discussion above. Let F be deﬁned in a neighborhood of x ∈ RN .2. that very small diﬀerence steps h can lead to inaccurate results. In the special case where (x) is a result of ﬂoatingpoint roundoﬀ in full precision (¯ ≈ 10−15 ). Definition 5. This means.
. and KD such that if xc ∈ B(δ). (x) ≤ ¯ for all x ∈ B(δ). A diﬀerence approximation to a full Jacobian does not use any special structure of the Jacobian and will probably be less eﬃcient than a hand coded analytic Jacobian. Therefore in the early phases of the iteration. ¯ Lemma 5. h ≤ h. x = 0.4. then ∆(x) = O( ¯) by Lemma 5. then ∇h (F + )(x) is nonsingular for all x ∈ B(δ) and there is MF > 0 such that F (x) − ∇h (F + )(x) ≤ MF (h + ¯/h). Let the standard assumptions hold. there is no guarantee that (xn ) → 0 as xn → x∗ . F (x + h x w/ w ) − F (x) . Then there are δ. h x = 0.4. ¯.ecsecurehost. Theorem 5.html. We have (5.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Let the standard assumptions hold. When en ≤ ¯.6. [42]. If we assume that the forward diﬀerence step has been selected so that √ √ h = O( ¯). while en >> ¯. to evaluate ∇h F with far fewer than N evaluations of F . w w The standard assumptions and the Banach Lemma imply the following result. Hence one will see quadratic convergence until the error in F admits no further reduction. one should expect the sequence { en } to stagnate and cease to decrease once√ en ≈ ¯.6 says that en+1 = O(¯). If information on the sparsity pattern of the Jacobian is available it is sometimes possible [47]. h x F (hw/ w ) − F (x) . Theorem 5.15)Dh F (x : w) = 0.4. If the sparsity pattern is such that F can be reduced to x∗ .1.4. Note that Theorem 5. h > ¯ 0 such that if x ∈ B(δ). w = 0. This electronic version is for personal use and may not be duplicated or distributed. NEWTON’S METHOD 81 w ∈ RN .4.6 does not imply that an iteration will converge to Even if is entirely due to ﬂoatingpoint roundoﬀ. x+ = xc − ∇hc (F + )(xc )−1 (F (xc ) + (xc )) Buy this book from SIAM at http://www. and (x) ≤ ¯ for all x ∈ B(δ). w = 0. Then there are positive δ. and there is M− > 0 such that √ hc > M− ¯ then satisﬁes e+ ≤ KD (¯ + ( ec + hc ) ec ). the progress of the iteration will be indistinguishable from Newton’s method as implemented with an exact √ Jacobian. w.4. Theorem 5. ¯.1 implies the following result.1. Hence.com/SIAM/FR16.
Dh F (x : w) is not.Copyright ©1995 by the Society for Industrial and Applied Mathematics.16) and the fundamental theorem of calculus ∆(xn ) = 1 0 (f (xn−1 + ts) − f (xn )) dt Buy this book from SIAM at http://www. xc − x− We have the following theorem.4. In order to start.5. the error in the diﬀerence approximation will be large. .1. too. If we set s = xn − xn−1 we have by (5. If xn = x∗ for some ﬁnite n. where f is a realvalued function of one real variable. x−1 ∈ B(δ) and x0 = x−1 then the secant iterates converge qsuperlinearly to x∗ . There is no reason to use a diﬀerence approximation to f for equations in a single variable because one can use previously computed data to approximate f (xn ) by f (xn ) − f (xn−1 ) . The standard assumptions. an = xn − xn−1 The resulting method is called the secant method and is due to Newton [140]. [137]. . a linear function of w and it is usually the case that (∇h F (x))w = Dh F (x : w). the entire solution process can take advantage of this structure [59].4. in general. . 5. Proof. While we used full matrix approximations to numerical Jacobians in all the examples reported here. then. (x) = 0. imply that f (x∗ ) = 0. Otherwise xn = xn−1 . which uses the size of w. The local convergence theory can be easily described by Theorem 5. Theorem 5. x0 and x−1 . We assume for the sake of simplicity that there are no errors in the evaluation of f .e. the diﬀerence between them may be signiﬁcant and a diﬀerence approximation to a directional derivative. Let δ be small enough so that B(δ) ⊂ Ω and the conclusions of Theorem 5. We will prove that the secant method is locally qsuperlinearly convergent if the standard assumptions hold.ecsecurehost.html. xn ∈ B(δ). .4. they should be used with caution. Then there is δ > 0 such that if x0 . 82 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS a block triangular form. .7.16) ∆(xc ) = f (xc ) − f (x− ) − f (xc ).1 hold.com/SIAM/FR16. one needs two approximations to x∗ . If F (x∗ ) is ill conditioned or the Lipschitz constant of F is large. For problems in which the Lipschitz constant of F is large. The results in this section are for single equations f (x) = 0. we are done. Moreover. is likely to be more accurate than ∇h F (x)w. We have (xc ) = 0 and (5.4.. This electronic version is for personal use and may not be duplicated or distributed. Assume that x−1 . x0 . i. The secant method. Let the standard assumptions hold with N = 1. Let xc be the current iterate and x− the previous iterate.
[106]. η.com/SIAM/FR16.17) ¯ en+1  ≤ K((1 + γ/2)en 2 + γen en−1 /2). ¯ K(1 + γ)δ = η < 1 If we reduce δ so that then xn → x∗ qlinearly. r− = βγ We will let B0 be the closed ball of radius r− about x0 B0 = {x  x − x0 ≤ r− }.1. [145].1 implies (5. and [58] for a complete proof and for discussions of related results. We do not prove the result in its full generality and refer to [106].Copyright ©1995 by the Society for Industrial and Applied Mathematics.5. which states that if the standard assumptions “almost” hold at a point x0 . This result is of use.8) method are the basis for the popular method of Brent [17] for the solution of single equations.5. Assumption 5. This completes the proof. There are β. we give a simpliﬁed version of a Kantorovichlike result for the chord Buy this book from SIAM at http://www. 5.html. Therefore en+1  ¯ ≤ K((1 + γ/2)en  + γen−1 /2) → 0 en  as n → ∞.7. This electronic version is for personal use and may not be duplicated or distributed. r. The secant method and the classical bisection (see Exercise 5. F is diﬀerentiable at x0 . for example. then there is a root and the Newton iteration converges rquadratically to that root.4. The Kantorovich Theorem In this section we present the result of Kantorovich [107]. in proving that discretizations of nonlinear diﬀerential and integral equations have solutions that are near to that of the continuous problem. 2 and hence ∆(xn ) ≤ γxn−1 − xn /2 ≤ Applying Theorem 5. F (x0 )−1 ≤ β. and F (x0 )−1 F (x0 ) ≤ η. However.ecsecurehost. NEWTON’S METHOD 83 γ(en  + en−1 ) . F is Lipschitz continuous with Lipschitz constant γ in a ball of radius r ≥ r− about x0 where ¯ √ 1 − 1 − 2βηγ . 2. [57]. . We require the following assumption. and γ with βηγ ≤ 1/2 and x0 ∈ RN ¯ such that 1.
Let Assumption 5. [145].1 assumes a bit more than Assumption 5. and [57]. the chord iteration with x0 as the initial iterate converges to x∗ qlinearly with qfactor βγr− . Buy this book from SIAM at http://www.5.com/SIAM/FR16. 2 φ(x) − x0 ≤ η + βγr− /2 = r− .1. Theorem 5.1 hold and let (5. Then there is a unique root x∗ of F in B0 .1 we have for all x ∈ S F (x0 )−1 F (x) = F (x0 )−1 F (x0 ) +F (x0 )−1 (5.18) βγη < 1/2. By the fundamental theorem of calculus and Assumption 5. Unlike the chord method results in [106].5. 84 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS method to show how existence of a solution can be derived from estimates on the function and Jacobian. and xn ∈ B0 for all n.1 but has a simple proof that uses only the contraction mapping theorem and the fundamental theorem of calculus and obtains qlinear convergence instead of rlinear convergence.Copyright ©1995 by the Society for Industrial and Applied Mathematics. We will show that the map φ(x) = x − F (x0 )−1 F (x) is a contraction on the set S = {x  x − x0 ≤ r− }. . This electronic version is for personal use and may not be duplicated or distributed. Moreover x∗ is the unique root of F in the ball of radius √ 1 + 1 − 2βηγ ¯ min r. Theorem 5.5. βγ about x0 . Proof.ecsecurehost.5.html. 1 0 (F (x0 + t(x − x0 )) − F (x0 ))(x − x0 ) dt.19) 1 0 F (x0 + t(x − x0 ))(x − x0 ) dt = F (x0 )−1 F (x0 ) + x − x0 +F (x0 )−1 1 0 (F (x0 + t(x − x0 )) − F (x0 ))(x − x0 ) dt and φ(x) − x0 = −F (x0 )−1 F (x0 ) −F (x0 )−1 Hence.5.
1 and has a very diﬀerent proof. Let Assumption 5. y ∈ S. [63]. We state the theorem in the standard way [145].html. This electronic version is for personal use and may not be duplicated or distributed.Copyright ©1995 by the Society for Industrial and Applied Mathematics.ecsecurehost.19) F (x0 )−1 F (x) ≥ r − η − βγr2 /2 > 0 because r− < r < r+ . To show that φ is a contraction on S and to prove the qlinear convergence we will show that φ(x) − φ(y) ≤ βγr− x − y for all x.2.1) will imply that there is a unique ﬁxed point x∗ of φ in S and that xn → x∗ qlinearly with qfactor βγr− . r+ ) r then F (x) = 0. This proves the convergence result and the uniqueness of x∗ in B0 . and xn ∈ B0 for all n. y ∈ B0 .com/SIAM/FR16. Hence we may assume that r > r− . the Newton iteration with x0 as the initial iterate converges to x∗ . φ(x) − φ(y) = φ (y + t(x − y))(x − y) dt ≤ βγr− x − y . ≤ 2n βγ n Buy this book from SIAM at http://www. . Theorem 5. Then there is a unique root x∗ of F in B0 . [106]. This completes the proof. If we do this. We must show ¯ that if x is such that r− < x − x0 < min(¯.20) en (2βηγ)2 .1 hold. using the rquadratic estimates from [106].5. Letting r = x − x0 we have by (5.2. then we are done by the uniqueness of x∗ in B0 .5. NEWTON’S METHOD 85 and φ maps S into itself. the contraction mapping theorem (Theorem 4.5. We know that βγr− < 1 by our assumption that βηγ < 1/2. for all x. βγ about x0 and the errors satisfy (5. Note that for all x ∈ B0 φ (x) = I − F (x0 )−1 F (x) = F (x0 )−1 (F (x0 ) − F (x)) and hence φ (x) ≤ βγ x − x0 ≤ βγr− < 1. x∗ is the unique root of F in the ball of radius √ 1 + 1 − 2βηγ ¯ min r. The Kantorovich theorem is more precise than Theorem 5. Clearly x∗ is a root of F because it is a ﬁxed point of φ. If r = r− . To prove the remainder of the uniqueness assertion let √ 1 + 1 − 2βηγ r+ = βγ ¯ and note that r+ > r− because βηγ < 1/2. 1 0 Therefore.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. Algorithm 5. we compute and factor F (x+ ) for use in the subsequent chord steps. The Jacobian is recomputed and factored if either the ratio of successive residuals exceeds the threshold 0 < ρ < 1 or the number of iterations without an update exceeds m. Examples for Newton’s method In the collection of MATLAB codes we provide an implementation nsol of Newton’s method. If the ratio is too large.6. however. τ.1.html.25. Solve LU s = −F (x) iii. rc = r0 = F (x) 2. In Exercise 5. The output of the MATLAB implementation of nsol is the solution. . If F (x+ ) / F (xc ) is below a given threshold. which is not easy.7.com/SIAM/FR16. Evaluate F (x) v. rc = r+ vi. a threshold for the ratio of successive residuals is also input.16). the chord method. or a combination of all of them based on the reduction in the nonlinear residual. is = is + 1 ii. say) and sparse matrix factorizations. While our examples have dense Jacobians. This electronic version is for personal use and may not be duplicated or distributed. σ = 0 (d) Do While is < m and σ ≤ ρ i. In the MATLAB code the iteration is terminated with an error condition if σ ≥ 1 at any point. ρ) 1. the Jacobian is not recomputed. and an error ﬂag. 5.6. 86 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS If βηγ < 1/2 then (5. nsol(x. In addition to the inputs of Algorithm sham. x = x + s iv. the Shamanskii method.ecsecurehost. This indicates that the local convergence analysis in this Buy this book from SIAM at http://www. the reader is asked to create a sparse form of nsol. Do while F (x) > τr r0 + τa (a) Compute ∇h F (x) ≈ F (x) (b) Factor ∇h F (x) = LU . One would still want to minimize the number of Jacobian computations (by the methods of [47] or [42].7. σ = r+ /rc . F. r+ = F (x) . the ideas in the implementation of nsol also apply to situations in which the Jacobian is sparse and direct methods for solving the equation for the Newton step are necessary. the history of the iteration stored as the vector of l∞ norms of the nonlinear residuals. (c) is = 0. This latter option decides if the Jacobian should be recomputed based on the ratio of successive residuals. If F (x) ≤ τr r0 + τa exit. nsol uses diffjac to approximate the Jacobian by ∇h F with h = 10−7 . m.20) implies that the convergence is rquadratic (see Exercise 5.
21) and the discrete analog (5.html. In Table 5. This electronic version is for personal use and may not be duplicated or distributed. and the ratio Rn = F (xn ) ∞ / F (xn−1 ) ∞ Buy this book from SIAM at http://www.9. The standard assumptions hold near either solution [110] for c ∈ [0. The parameters m and ρ are assigned the default values of 1000 and . the default is 40. is used to solve exit distribution problems in radiative transfer. however. NEWTON’S METHOD 87 section is not applicable. We begin with a comparison of Newton’s method and the chord method. 1] by 1 0 f (µ) dµ ≈ 1 N N j=1 f (µj ). will ﬁnd the physically meaningful solution [110]. We set the termination parameters τr = τa = 10−6 . (5. We present some iteration statistics in both tabular and graphical form. We computed all Jacobians with the MATLAB code diffjac. In Chapter 8 we discuss how to recover from this. Here we approximate integrals on [0. Because H has a singularity at µ = 0.7.22) have two solutions for c ∈ (0.ecsecurehost. Newton’s method. In the computations reported here we set N = 100 and c = . F (xn ) ∞ / F (x0 ) ∞ .21) F (H)(µ) = H(µ) − 1 − The Chandrasekhar Hequation.1 we tabulate the iteration counter n. With these default parameters the decision to update the Jacobian is made based entirely on the reduction in the norm of the nonlinear residual. The discrete problem has its own physical meaning [41] and we will attempt to solve it to high accuracy. nsol is based on the assumption that a diﬀerence Jacobian will be used.21 you are asked to modify the code to accept an analytic Jacobian. We used the function identically one as initial iterate.com/SIAM/FR16. In Exercise 5. The Chandrasekhar Hequation. c 2 1 µH(ν) dν 0 −1 µ+ν = 0. 1).5. [30]. where µi = (i − 1/2)/N for 1 ≤ i ≤ N . 1). We will discretize the equation with the composite midpoint rule. Many times the Jacobian can be computed eﬃciently by reusing results that are already available from the function evaluation.Copyright ©1995 by the Society for Industrial and Applied Mathematics.22) F (x)i = xi − 1 − c 2N µi xj µi + µj j=1 N −1 . nsol allows the user to specify a maximum number of iterations. . Only one of these solutions has physical meaning. with the 0 function or the constant function with value 1 as initial iterate. The resulting discrete problem is (5. [41]. the solution of the discrete problem is not even a ﬁrstorder accurate approximation to that of the continuous problem. It is known [132] that both (5.
21.480e01 2.18 you are asked to obtain timings in your environment and will see how the ﬂoatingpoint operation counts are related to the actual timings.Copyright ©1995 by the Society for Industrial and Applied Mathematics.000e+00 1. .118e01 2.511e03 1.965e04 6.7.074e02 6.136e01 2. we plot.15 for a more careful examination of this concavity. If we think of the chord method as a preconditioned nonlinear Richardson iteration with the initial Jacobian playing the role of preconditioner.334e05 1. the cost of a single evaluation and factorization of the Jacobian is far more than the cost of the entire remainder of the chord iteration. The concave iteration track is the signature of superlinear convergence.136e01 2.8 for Newton’s method.353e05 2. the simple chord method is the most eﬃcient.823e02 2. The ideas in Chapters 6 and 7 show how the low iteration cost of the chord method can be combined with the small number of iterations of a superlinearly convergent nonlinear iterative method.com/SIAM/FR16. The Newton iteration required over 8.ecsecurehost. See Exercise 5. 5.8 million ﬂoatingpoint operations while the chord iteration required under 3. The Shamanskii computation required under 6 million ﬂoatingpoint operations.136e01 2. but do not tabulate.388e03 2. The reader is encouraged to do this in Exercise 5. This electronic version is for personal use and may not be duplicated or distributed. However. In the context of this particular example.480e01 2.077e01 2. as compared with over 8.480e01 1. Buy this book from SIAM at http://www.480e01 3. the cost of the Jacobian computation can be reduced by analytic evaluation and reuse of data that have already been computed for the evaluation of F . In the case considered here.html. The good performance of the chord method is no accident.7.132e01 2.2 we compare the Shamanskii method with m = 2 with Newton’s method.865e04 F (xn ) ∞ / F (x0 ) 1. In the remaining examples.729e07 ∞ Rn 1. the linear track indicates linear convergence. 88 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Table 5.1 Comparison of Newton and chord iteration. The linear convergence of the chord method can also be seen in the convergence of the ratios Rn to a constant value. then the chord method should be much more eﬃcient than Newton’s method if the equations for the steps can be solved cheaply.1 the solid curve is a plot of F (xn ) ∞ / F (x0 ) ∞ for Newton’s method and the dashed curve a plot of F (xn ) ∞ / F (x0 ) ∞ for the chord method. for this problem. 5. In Fig.698e03 7. the iteration statistics. In Exercise 5. In Fig. n 0 1 2 3 4 5 6 7 8 F (xn ) ∞ / F (x0 ) 1. For this example the MATLAB flops command returns an fairly accurate indicator of the cost.891e06 ∞ Rn 1.000e+00 1.136e01 for n ≥ 1 for both Newton’s method and the chord method.7.3 million.
all iteration parameters are the same as in the previous example. The qfactor for the chord method in this computation was over . indicating very slow convergence. 5. While Newton’s method converges in 7 iterations and requires 21 million ﬂoatingpoint operations.5 1 1. the chord method requires 188 iterations and 10. The hybrid approach in nsol with the default parameters would be the chord method for this problem.9999. 5.2.96. We now reconsider the Hequation with c = .1. c = .html.9. Aside from c.6 million ﬂoatingpoint operations. Newton and Shamanskii method for Hequation.5 3 3. NEWTON’S METHOD 0 89 10 10 Relative Nonlinear Residual 1 10 2 10 3 10 4 10 5 10 6 10 7 0 1 2 3 4 Iterations 5 6 7 8 Fig. c = .5 4 Fig. .com/SIAM/FR16.9. The hybrid method in nsol with the default parameters Buy this book from SIAM at http://www.ecsecurehost. This electronic version is for personal use and may not be duplicated or distributed. Newton and chord methods for Hequation. For this problem the Jacobian at the solution is nearly singular [110].5 2 Iterations 2.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 10 0 10 Relative Nonlinear Residual 1 10 2 10 3 10 4 10 5 10 6 10 7 0 0.
3. support this. . In practice.com/SIAM/FR16. however. plots like those in this section are common. 10 0 10 Relative Nonlinear Residual 1 10 2 10 3 10 4 10 5 10 6 0 2 4 6 8 Iterations 10 12 14 Fig. While the plots indicate qlinear convergence of the nonlinear residuals.3 we compare the Newton iteration with the hybrid. converges in 14 iterations. This implies only that the nonlinear residual norms { F (xn ) } converge rlinearly to zero. One should approach plots of convergence histories with some caution.ecsecurehost. The theory for the chord method asserts that the error norms converge qlinearly to zero. the theory does not. 5.html. This electronic version is for personal use and may not be duplicated or distributed. and requires 12 million ﬂoatingpoint operations. in general. Buy this book from SIAM at http://www.Copyright ©1995 by the Society for Industrial and Applied Mathematics.9999. 90 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS computes the Jacobian four times during the course of the iteration. Newton and hybrid methods for Hequation c = . In Fig. 5.
Exercises on Newton’s method In some of the exercises here. 5.7.7. Prove that the secant method converges rsuperlinearly with rorder √ ( 5 + 1)/2. r r l l Buy this book from SIAM at http://www. for each iteration tabulate the iteration counter.8. can the Newton iterates for F (x) = 0 and AF (Bx) = 0 show any performance diﬀerence when started at the same initial iterate? What about the chord method? 5.ecsecurehost. x0 ) such that f (x0 )f (x0 ) < 0.3. Can the performance of the Newton iteration be improved by a linear change of variables? That is.7. and the ratio of F (xn ) / F (xn−1 ) for n ≥ 1.7.7. Use the l∞ norm.4. one can visually inspect the plot to determine superlinear convergence (concavity) without explicitly computing the ratios. r l Show that the sequence xk = (xk + xk )/2 is rlinearly convergent if f is r l continuous and there exists (x0 . 5. A function G is said to be H¨lder continuous with exponent α in Ω if o α for all x. If f (y) = 0. When one does this.6. prove that the iteration given by xn+1 = xn − (∇hn F (xn ))−1 F (xn ) is locally qsuperlinearly convergent if hn → 0. 5. Given (xk . 5. we terminate the iteration. xk ) that r l contain a root of a function of one variable. 5. you will be asked to plot or tabulate iteration statistics. A better alternative in the MATLAB environment is to use the semilogy command to plot the norms.7. Let ∇h F be given by Deﬁnition 5. for nonsingular N × N matrices A and B. The bisection method produces a sequence of intervals (xk . When asked to do this. Suppose F (xn ) is replaced by ∇hn F (xn ) in the Shamanskii method. This is easy.2.7. Given the standard assumptions.Copyright ©1995 by the Society for Industrial and Applied Mathematics.com/SIAM/FR16.1. y ∈ Ω. .7.7. In what way is the convergence analysis of the secant method changed if there are errors in the evaluation of f ? 5.4. Show that if the Lipschitz G(x) − G(y) ≤ K x − y continuity condition on F in the standard assumptions is replaced by H¨lder continuity with exponent α > 0 that the Newton iterates converge o with qorder 1 + α. Discuss how hn must be decreased as the iteration progresses to preserve the qorder of m + 1 [174]. NEWTON’S METHOD 91 5. This electronic version is for personal use and may not be duplicated or distributed.7. When is it locally qquadratically convergent? 5.5.html. the norm of F .1. Show that the secant method converges qsuperlinearly with qorder √ ( 5 + 1)/2. xk ) with r l f (xk )f (xk ) < 0 we replace one of the endpoints with y = (xk + xk )/2 r r l l to force f (xk+1 )f (xk+1 ) < 0. and in the rest of this part of the book.
use x−1 = . show that log( en ) is a concave function of n in the sense that log( en+1 ) − log( en ) Buy this book from SIAM at http://www. see [170]. x0 = 3 4. f (x∗ ) = 0 and f (x) is Lipschitz continuous. This electronic version is for personal use and may not be duplicated or distributed. If xn → x∗ qsuperlinearly with qorder α > 1. 5. x0 = 10 5. the standard assumptions hold.13.7.7. For the secant method.7. and f (x∗ ) = f (x∗ ) = 0 but f (x∗ ) = 0 then the iteration xn+1 = xn − 2f (xn )/f (xn ) converges locally qquadratically to x∗ provided x0 is suﬃciently near to x∗ . . For scalar functions of one variable a quadratic model of f about a point xc is mc (x) = f (xc ) + f (xc )(x − xc ) + f (xc )(x − xc )2 /2. Can the ideas in [170] be applied to some of the highorder methods discussed in [190] for equations in one unknown? 5. f (x) = sin(x). f is Lipschitz continuous. 5. x0 = . f (x) = x2 + 1. tabulate or plot iteration statistics.12.15.5 5. and explain your results.html. but not equal to x∗ [171]. How would you estimate the qorder of a qsuperlinearly convergent sequence? Rigorously justify your answer. 5. then the Newton iteration converges cubically (qsuperlinear with qorder 3).Copyright ©1995 by the Society for Industrial and Applied Mathematics. Consider the iterative method that deﬁnes x+ as a root of mc . Apply your program to the following function/initial iterate combinations.7. f (x) = cos(x) − x.11.5 2. 92 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 5. the standard assumptions hold. .7. In what cases is this algorithm superior to Newton’s method? For hints and generalizations of this idea to functions of several variables. What if f (x∗ ) = 0? How can you extend this result if f (x∗ ) = f (x∗ ) = . f (x) = x2 . 1.9. Write a program that solves single nonlinear equations with Newton’s method. and the secant method. f (x) = tan−1 (x).9. and f (x0 )f (x0 ) > 0. Show that if f is a real function of one real variable. Show that if f is a real function of one real variable.7.7. Show that if f is a real function of one real variable.14.ecsecurehost. x0 = 1 3. x0 is suﬃciently near x∗ .10. x0 = . then the Newton iteration converges monotonically to x∗ .99x0 . Test it on the problems in Exercise 5. f (k) (x∗ ) = 0 = f (k+1) (x∗ )? See [131]. What problems might arise in such a method? Resolve these problems and implement your algorithm. 5. the chord method.7.com/SIAM/FR16. .
4.7. How does this change aﬀect the convergence rate observations that you have made? Buy this book from SIAM at http://www. [49]. [75]. .7.20.18. Use the theory in [89].22.7.ecsecurehost.16.com/SIAM/FR16. [35] and [155] for more discussion of this. Use the MATLAB cputime command to see how long the entire iteration took for each N.7. How do the execution times and ﬂop counts change? 5.22). LU = F (x) + E. 5. or [118] for acceleration of the convergence. 5. [184]) the computed LU factorization of F (x) is the exact LU factorization of a nearby matrix. Explain your results (see [50] and [157]). 5. the discretized nonlinearity from the H equation in (5. Try diﬀerent values of N and c. Compare execution times in your environment.1 to describe how this factorization error aﬀects the convergence of the Newton iteration.7. Does the trick in Exercise 5.9 (see [52]). NEWTON’S METHOD 93 is a decreasing function of n for n suﬃciently large. c combination and the MATLAB flops command to get an idea for the number of ﬂoatingpoint operations required by the various methods.21.6. Compute by hand the Jacobian of F . Show that the sequence on the right side of (5. and the algorithm in nsol. [184] or another book on numerical linear algebra and Theorem 5.html. Solve the Hequation with N = 100.11 improve the convergence? Try one of the approaches in [51]. [170].Copyright ©1995 by the Society for Industrial and Applied Mathematics. the chord method.7. Try the chord method and compare the results to part 4 of Exercise 5. Modify dirder by setting the forward diﬀerence step h = 10−2 .7. Is this still the case if the convergence is qsuperlinear but there is no qorder? What if the convergence is qlinear? 5. Solve the Hequation with c = 1 and N = 100.7.17. How does the analysis for the QR factorization diﬀer? 5. Typically (see [89].20) is rquadratically (but not qquadratically) convergent to 0 if βηγ < 1/2. You might also look at [90].9999 with ﬁxedpoint iteration. Prove that F can be computed at a cost of less than twice that of an evaluation of F . This electronic version is for personal use and may not be duplicated or distributed. How does the performance of the methods diﬀer as N and c change? Compare your results to the tabulated values of the H function in [41] or [14]. 5. c = . Tabulate or plot iteration statistics.9 and c = .19. Duplicate the results on the Hequation in § 5.7. Modify nsol to accept this analytic Jacobian rather than evaluate a diﬀerence Jacobian. Compare timings and ﬂop counts with Newton’s method.
html.7. Here. Develop a sparse version of nsol for problems with tridiagonal Jacobian. the solution is u∗ (x) = x(1 − x).7. λ) of an N ×N matrix A by solving the system of N + 1 nonlinear equations (5. 94 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 5. how are the convergence rates diﬀerent? What is an appropriate termination criterion? At what point does the iteration stagnate? 5.23) F (φ.Copyright ©1995 by the Society for Industrial and Applied Mathematics. that the cost of a function evaluation is O(N 2 ) ﬂoatingpoint operations.23. .26.25. Suppose you try to ﬁnd an eigenvectoreigenvalue pair (φ.24. Buy this book from SIAM at http://www. What about the chord method? 5. What is the Jacobian of this system? If x = (φ.ecsecurehost. λ) ∈ RN +1 is an eigenvectoreigenvalue pair. for example. you could modify diffjac to use the techniques of [47] and nsol to use the sparse matrix factorization in MATLAB. Add noise to your function evaluation with a random number generator. Show that the number n of Newton iterations needed to obtain en ≤ e0 is O(log( )) and that the number of ﬂoatingpoint operations required is O(N 3 log( )). λ) = Aφ − λφ φT φ − 1 = 0 0 for the vector x = (φT . u(0) = u(1) = 0. λ)T ∈ RN +1 .23) to the inverse power method. when is F (x) nonsingular? Relate the application of Newton’s method to (5.7. and f (x) = 2 − sin(x(1 − x)). In MATLAB. the cost of a Jacobian is O(N ) function evaluations. See [150] for more development of this idea. Let u0 = 0 be the initial iterate. Apply your code to the central diﬀerence discretization of the twopoint boundary value problem −u = sin(u) + f (x). If the size of the noise is O(10−4 ) and you make the appropriate adjustments to dirder. 5. such as the the MATLAB rand function.7. Assume that the standard assumptions hold.com/SIAM/FR16. and that x0 is near enough to x∗ so that the Newton iteration converges qquadratically to x∗ . This electronic version is for personal use and may not be duplicated or distributed.
1) F (xc )s + F (xc ) ≤ ηc F (xc ) are considered. The basic estimates We will discuss speciﬁc implementation issues in § 6.1).1 should be compared to that of Theorem 5.1.1) are precisely the small linear residual termination conditions for iterative solution of the linear system for the Newton step.com/SIAM/FR16. We follow [69] and refer to the term ηc on the right hand side of (6. for discussion of these ideas in the context of the classical stationary iterative methods. Theorem 6.1 we give a straightforward analysis that uses the techniques of Chapter 5.2) e+ ≤ KI ( ec + ηc ) ec . In most of this chapter we will focus on the approximate solution of the equation for the Newton step by GMRES. .1. See [145] and [175].1 and the other results in Chapter 5. We will present this result in two parts. s satisﬁes (6.2. Let the standard assumptions hold.ecsecurehost. Then there are δ and KI such that if xc ∈ B(δ).html. and x+ = xc + s then (6. This was the view taken in [55] where inexact Newton methods in which the step satisﬁes (6.1. In § 6. Chapter 6 Inexact Newton Methods Theorem 5.1.1. Another way to look at this is to ask how an approximate solution of the linear equation for the Newton step aﬀects the iteration. 6.2 we show how the requirements of the simple analysis can be changed to admit a more aggressive (i.1 describes how errors in the derivative/function aﬀect the progress in the Newton iteration. 95 Buy this book from SIAM at http://www.Copyright ©1995 by the Society for Industrial and Applied Mathematics.1. but the other Krylov subspace methods discussed in Chapters 2 and 3 and elsewhere can also be used. This is quite useful because conditions like (6. The proof and application of Theorem 6. Before that we will give the basic result from [55] to show how a step that satisﬁes (6.1) as the forcing term..1) aﬀects the convergence. Such methods are not new. larger) choice of the parameter η.1. In § 6. 6. Any approximate step is accepted provided that the relative residual of the linear equation is small.e.1. This electronic version is for personal use and may not be duplicated or distributed. Direct analysis.4.1. for example.
To prove the ﬁrst assertion (6. Now. Let the standard assumptions hold. then the inexact Newton iteration ¯ ¯ xn+1 = xn + sn . the proof is complete. 96 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Proof. In [55] r = F (xc )s + F (xc ).1).html.1.com/SIAM/FR16.1.Copyright ©1995 by the Society for Industrial and Applied Mathematics.1 for iterative methods in the next result.3.ecsecurehost. Let δ be small enough so that the conclusions of Lemma 4.7). Theorem 6.1 e+ ≤ ec − F (xc )−1 F (xc ) + 4κ(F (x∗ ))ηc ec ≤ K ec 2 + 4κ(F (x∗ ))ηc ec .1. Buy this book from SIAM at http://www. Hence. where K is the constant from (5. and • if ηn ≤ Kη F (xn ) with qorder 1 + p. Then there are δ and η such that if x0 ∈ B(δ). using (6. If we set KI = K + 4κ(F (x∗ )). η ].1 hold.3) and Theorem 5.1 and Theorem 5.2). where F (xn )sn + F (xn ) ≤ ηn F (xn ) converges qlinearly to x∗ . .3) e+ = ec + s = ec − F (xc )−1 F (xc ) − F (xc )−1 r. which is a direct consequence of (6. This electronic version is for personal use and may not be duplicated or distributed.1. (4.2) and will suﬃce to explain most computational observations. Moreover • if ηn → 0 the convergence is qsuperlinear.2.2) note that if2 r = −F (xc )s − F (xc ) is the linear residual then s + F (xc )−1 F (xc ) = −F (xc )−1 r and (6. p for some Kη > 0 the convergence is qsuperlinear 2 Our deﬁnition of r is consistent with the idea that the residual in a linear equation is b − Ax. {ηn } ⊂ [0.6) imply that s + F (xc )−1 F (xc ) ≤ F (xc )−1 ηc F (xc ) ≤ 4κ(F (x∗ ))ηc ec . We summarize the implications of Theorem 6. (6. and (4.
¯ 1−p p + Kη 2p F (x∗ ) p ) en 1+p Proof. F (x∗ )e+ = F (x∗ )(ec + s) = F (x∗ )(ec − F (xc )−1 F (xc ) − F (xc )−1 r). INEXACT NEWTON METHODS 97 Proof. 1) other than requiring that 1 not be an accumulation point. Then there is δ such that if xc ∈ B(δ).3. Theorem 6.4) F (x∗ )e+ ≤ η F (x∗ )ec . ¯ where KI is from (6.ecsecurehost. s satisﬁes (6. and ηc ≤ η < η < 1.1. then ¯ (6.1 implies that F (xc ) ≤ F (x∗ )ec + Since we have.7) and (6.html. To prove (6. If ηn ≤ Kη F (xn ) then we may use (4. x+ = xc + s. F (x∗ )ec ec = F (x∗ )−1 F (x∗ )ec ≤ F (x∗ )−1 γ F (x∗ )−1 2 F (xc ) ≤ (1 + M0 δ) F (x∗ )ec . Weighted norm analysis.Copyright ©1995 by the Society for Industrial and Applied Mathematics.0.2).4) note that Theorem 4. Reduce δ and η if needed so that ¯ KI (δ + η ) < 1. 6. Then if n ≥ 0 and xn ∈ B(δ) we have en+1 ≤ KI ( en + ηn ) en ≤ KI (δ + η ) en < en . one might conclude from Theorem 6.com/SIAM/FR16. Let the standard assumptions hold. Since KI = O(κ(F (x∗ ))). ¯ This proves qlinear convergence with a qfactor of KI (δ + η ).1.1. This is not the case and the purpose of this subsection is to describe more accurately the necessary restrictions on the forcing terms. with γ ec 2 2 . Buy this book from SIAM at http://www.2) to conclude en+1 ≤ KI ( en which completes the proof. ¯ If ηn → 0 then qsuperlinear convergence follows from the deﬁnition.2. .2 that if F (x∗ ) is ill conditioned very small forcing terms must be used. This electronic version is for personal use and may not be duplicated or distributed.2) holds for xc ∈ B(δ). Let δ be small enough so that (6.5) Now.1. M0 = (6. The results here diﬀer from those in § 6.1).1 in that no restriction is put on the sequence {ηn } ⊂ [0.
1. Note that the distance δ from the initial iterate to the solution may have to be smaller for (6. .5). Now let δ be small enough so that ¯ (1 + M1 δ)(1 + M0 δ)η + M2 δ F (x∗ )−1 ≤ η and the proof is complete.1 F (x∗ )(ec − F (xc )−1 F (xc )) ≤ K F (x∗ ) ec 2 .1.4 and compare it to a modiﬁcation of a choice from Buy this book from SIAM at http://www. This is made precise in Theorem 6. F (x∗ )e+ ≤ F (x∗ )F (xc )−1 r + K F (x∗ ) ec 2 .6) Since F (x∗ )F (xc )−1 r ≤ r + (F (x∗ ) − F (xc ))F (xc )−1 r ≤ (1 + 2γ F (x∗ )−1 we may set ec ) r . This electronic version is for personal use and may not be duplicated or distributed.1.ecsecurehost. The application of Theorem 6. Hence.1.html.Copyright ©1995 by the Society for Industrial and Applied Mathematics. and M2 = K F (x∗ ) and obtain.3 described in Theorem 6.com/SIAM/FR16.4) is remarkable in its assertion that any method that produces a step with a linear residual less than that of the zero step will reduce the norm of the error if the error is measured with the weighted norm · ∗ = F (x∗ ) · . just that 1 not be an accumulation point. 98 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS By Theorem 5. The importance of this theorem for implementation is that choices of the sequence of forcing terms {ηn } (such as ηn = .3 asserts qlinear convergence in the weighted norm if the initial iterate is suﬃciently near x∗ . using (4.2). M1 = 2γ F (x∗ )−1 .5 for all n) that try to minimize the number of inner iterations are completely justiﬁed by this result if the initial iterate is suﬃciently near the solution.7) and (6.1. However (6.4.4) to hold than for (6. F (x∗ )e+ (6.4 diﬀers from Theorem 6. We make such a choice in one of the examples considered in § 6.1.2 in that we do not demand that {ηn } be bounded away from 1 by a suﬃciently large amount. Theorem 6. (6. In fact.7) ≤ (1 + M1 δ) r + M2 δ ec ≤ (1 + M1 δ)(1 + M0 δ)ηc F (x∗ )ec +M2 δ F (x∗ )−1 F (x∗ )ec ≤ ((1 + M1 δ)(1 + M0 δ)η + M2 δ F (x∗ )−1 ) F (x∗ )ec .
1. e < δ∗ = F (x∗ ) δ0 ∗ Set δ = F (x∗ ) −1 δ∗ . Moreover • if ηn → 0 the convergence is qsuperlinear. While Theorem 6.2). This electronic version is for personal use and may not be duplicated or distributed.1.5. (6. Let the standard assumptions hold. eventually en will be small enough so that the conclusions of Theorem 6.html.1. {ηn } ⊂ [0.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Rather we note that (6.1.1.4.1. The most signiﬁcant diﬀerence is the norm in which qlinear convergence takes place.4) holds for ec < δ0 .4) and (6. We close this section with some remarks on the two types of results in Theorems 6. This proves the ﬁrst assertion.1 hold.2.1 and leave the proof to the reader in Exercise 6.4 shows that the sequence of forcing terms need only be kept bounded away from Buy this book from SIAM at http://www.2.2 might lead one to believe that a small value of ηn is necessary for convergence. The implications for the choice of the forcing terms {ηn } are also diﬀerent. η] with η < η < 1. Our ﬁrst task is to relate the norms · and · ∗ so that we can estimate δ. Let δ0 be such that (6. Then there is δ such that if x0 ∈ B(δ). Theorem 6. qlinear convergence with respect to the weighted norm · ∗ is equivalent to qlinear convergence of the sequence of nonlinear residuals { F (xn ) }. The superlinear convergence results then follow exactly as in the proof of Theorem 6. . then the inexact Newton ¯ iteration xn+1 = xn + sn . and • if ηn ≤ Kη F (xn ) p for some Kη > 0 the convergence is qsuperlinear with qorder 1 + p.com/SIAM/FR16. For all x ∈ RN . Theorem 6. Then e0 ∗ < δ∗ if e0 < δ.4) implies that if en ∗ < δ∗ then (6. INEXACT NEWTON METHODS 99 [69] that decreases ηn rapidly with a view toward minimizing the number of outer iterations. where F (xn )sn + F (xn ) ≤ ηn F (xn ) converges qlinearly with respect to · ∗ to x∗ .4.1. We state this result as Proposition 6. note that since xn → x∗ . To prove the assertions on qsuperlinear convergence. Our proof does not rely on the (possibly false) assertions that en < δ for all n or that xn → x∗ qlinearly with respect to the unweighted norm.9) en+1 ∗ ≤ η en ¯ ∗ < δ∗ and hence xn → x∗ qlinearly with respect to the weighted norm. The proof uses both (6. Proof.1.2 and 6.8) Note that e < δ0 if x ∗ ≤ F (x∗ ) x ≤ κ(F (x∗ )) x ∗ .ecsecurehost.
ecsecurehost. Errors in the function. Let the standard assumptions hold. This naming convention is taken from [175] and [145].html.2. Buy this book from SIAM at http://www. such as those arising from a diﬀerence approximation of the action of F (xc ) on a vector. In the examples discussed in § 6. . but at a price of a greater number of outer iterations. 6. s satisﬁes (6. These methods have also been called truncated Newton methods [56].1) with an iterative method for the linear system for the Newton step.com/SIAM/FR16. CG. We will discuss other factors in the choice of forcing terms in § 6. ¯ 6. Typically the nonlinear iteration that generates the sequence {xn } is called the outer iteration and the linear iteration that generates the approximations to the steps is called the inner iteration. In this section. Theorem 6. In this section we use the l2 norm to measure nonlinear residuals. or GMRES to solve the linear system.6. we state two theorems on inexact local improvement.4 we will scale the l2 norm by a factor of 1/N so that the results for the diﬀerential and integral equations will be independent of the computational mesh.3.10). would use SOR. A constant sequence such as ηn = . Theorem 6.1) holds). Let the standard assumptions hold and let xn → x∗ . NewtonGMRES. The name of the method indicates which linear iterative method is used.1. The reason for this is that the Krylov methods use the scalar product and the estimates in Chapters 2 and 3 are in terms of the Euclidean norm. then ¯ there is KE such that (6. when (6.10) F (xc )s + F (xc ) + (xc ) ≤ ηc F (xc ) + (xc ) and x+ = xc + s.1. 100 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 1.5 may well suﬃce for linear convergence.11) e+ ≤ KI (( ec + ηc ) ec + (xc ) ).5.3.e.1. Newtoniterative methods A Newtoniterative method realizes (6.1 and its proof for an illustration of this point. to the reader as exercises. NewtonSOR. for example.12) F (x∗ )e+ ≤ η F (x∗ )ec + KE (xc ) . [136] in the context of optimization. terminating when the relative linear residual is smaller than ηc (i. can be regarded as part of the inexact solution of the equation for the Newton step. See [55] or Proposition 6.3. then (6. Proposition 6. Then there are δ and KI such that if xc ∈ B(δ). This electronic version is for personal use and may not be duplicated or distributed.2.Copyright ©1995 by the Society for Industrial and Applied Mathematics.1. Then F (xn ) converges qlinearly to 0 if and only if en ∗ does. x+ = xc + s.1. s satisﬁes (6.1. NewtonCG.1. Let the standard assumptions hold. Then there is δ such that if xc ∈ B(δ). Note that errors in the derivative. and ηc ≤ η < η < 1. We leave the proofs.1 and 6. which are direct analogs to the proofs of Theorems 6.
(5. In fact. Dh F (x : w) for some h. Note that matrixvector products are replaced by forward diﬀerence approximations to F (x). [24]. In the case of CGNR and CGNE. vk+1 = vk+1 − hjk vj (c) hk+1. 0. 0)T (e) e1 = ∈ Rk+1 Minimize βe1 − Hk y k Rk+1 to obtain y k ∈ Rk . x. This electronic version is for personal use and may not be duplicated or distributed. While ρ > η F (x) (a) k = k + 1 (b) vk+1 = Dh F (x : vk ) for j = 1. β = ρ. we give the forwarddiﬀerence GMRES algorithm for computation of the Newton step. ρ) 1. k T i. kmax.ecsecurehost. application of GMRES to the linear equation for the Newton step with matrixvector products approximated by ﬁnite diﬀerences is the same as the application of GMRES to the matrix Gh F (x) whose last k − 1 columns are the vectors vk = Dh F (x : vk−1 ) The sequence {vk } is formed in the course of the forwarddiﬀerence GMRES iteration. . s = 0. In many implementations [20].1. In our MATLAB implementation we solve the least squares problem in step 2e by the Givens rotation approach used in Algorithm gmres. r = −F (x). Algorithm 6.2. Rk+1 . . ρ = r 2 . hjk = vk+1 vj ii. . s = Vk y k . It is important to note that this is entirely diﬀerent from forming the ﬁnite diﬀerence Jacobian ∇h F (x) and applying that matrix to w.15). If the linear iterative method is any one of the Krylov subspace methods discussed in Chapters 2 and 3 then each inner iteration requires at least one evaluation of the action of F (xc ) on a vector. s = −F (x)−1 F (x). Newton GMRES. that b = −F (x). . INEXACT NEWTON METHODS 101 6. η. We give a MATLAB implementation of this algorithm in the collection of MATLAB codes. To illustrate this point. fdgmres(s. and that the initial iterate for the linear problem is the zero vector. . We begin by discussing the eﬀects of a forward diﬀerence approximation to the action of the Jacobian on a vector. h.Copyright ©1995 by the Society for Industrial and Applied Mathematics. . k = 0 2. . v1 = r/ r 2 . F. a sequence of approximate steps {sk } is produced. We provide a detailed analysis of the NewtonGMRES iteration.1. .com/SIAM/FR16. an evaluation of the action of F (xc )T on a vector is also required. Buy this book from SIAM at http://www. (f) ρ = βe1 − Hk y k 3. the action of F (xc ) on a vector w is approximated by a forward diﬀerence.html. as pointed out in [20].k = vk+1 2 2 2 and k < kmax do (d) vk+1 = vk+1 / vk+1 (1.2.
com/SIAM/FR16. Let η ∈ (0.1 hold. First let δ be small enough so that B(δ) ⊂ Ω and the conclusions to Lemma 4.2.3. we have. ¯ ¯ Then there are CG . Hence Gh F is also ﬁrst order accurate in h.1.14) B − F (x) 2 ≤ h x 2 γ/2 ≤ h¯ /2 γ Since the linear iteration (fdgmres) terminates in k < kmax iterations. The action of Gh F on Kk . Proof. ¯ (6. This electronic version is for personal use and may not be duplicated or distributed.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Let the standard assumptions hold. the diﬀerence between ∇h F and Gh F is that the columns are computed by taking directional derivatives based on two diﬀerent orthonormal bases: ∇h F using the basis of unit vectors in coordinate directions and Gh F the basis for the Krylov space Kk ⊥ constructed by algorithm fdgmres. could be speciﬁed by the action of ∇h F on any basis for Kk . h ≤ h. since B and Gh F agree on Kk . and δ such that if x ∈ B(δ). for the purposes of ⊥ analysis. 1).15) Assume that F (x)s + F (x) 2 2 ≤ η F (x) 2 ≤ η F (x) 2 + h¯ s 2 /2.ecsecurehost. which is a special case of the more general results in [20]. with γ = γ( x∗ 2 + δ). γ ¯γ h¯ ≤ F (x∗ )−1 −1 2 /2 Buy this book from SIAM at http://www. where {vj }k is the orthonormal basis for Kk generated j=1 by Algorithm fdgmres.html. h. Let {uj } be any orthonormal basis for RN such that uj = vj for 1 ≤ j ≤ k. . and Algorithm fdgmres terminates with k < kmax then the computed step s satisﬁes (6. 102 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS As should be clear from inspection of Algorithm gmres. Bs + F (x) and therefore (6.13) F (x)s + F (x) 2 < (η + CG h) F (x) 2 . the orthogonal complement of Kk . Consider the linear map deﬁned by its action on {uj } Buj = Dh F (x : uj ). is not used by the algorithm and. Proposition 6. Note that Dh F (x : uj ) = 1 0 F (x + th x 2 uj )uj dt 1 0 (F = F (x)uj + (x + th x 2 uj ) − F (x))uj dt. Since F is Lipschitz continuous and {uj } is an orthonormal basis for RN we have. Assuming that there is no error in the evaluation of F we may summarize the observations above in the following proposition.
com/SIAM/FR16.1 hold. (6.2.1 implies that a ﬁnite diﬀerence implementation will not aﬀect the performance of NewtonGMRES if the steps in the forward diﬀerence approximation of the derivatives are suﬃciently small. Assume that the assumptions of Proposition 6.1. hn . and Proposition 6. . INEXACT NEWTON METHODS 103 Then Lemma 4. Assume that the assumptions of Proposition 6. In an implementation. F.4 and Theorem 6. Theorem 6.1 and are left as (easy) exercises. Then there is δ such that if x0 ∈ B(δ) and the sequences {ηn } and {hn } satisfy σn = ηn + CG hn ∈ [0.1.1 and (6. σ ] ¯ with 0 < σ < 1 then the forward diﬀerence NewtonGMRES iteration ¯ xn+1 = xn + sn Buy this book from SIAM at http://www.16) with (6. η) F (x 2 Proposition 6.15) imply F (x∗ )−1 −1 2 s 2 /2 ≤ F (x)−1 ≤ F (x)s 2 −1 2 s 2 2 ≤ (1 + η) F (x) 2 ¯γ + h¯ s 2 /2.17) F (xn )sn + F (xn ) 2 < σn F (xn ) 2 . xn . γ Combining (6.2. This electronic version is for personal use and may not be duplicated or distributed.Copyright ©1995 by the Society for Industrial and Applied Mathematics.16) s 2 ≤ 4(1 + η) F (x∗ )−1 F (x) 2 .ecsecurehost.2. but also the steps {hn } used to compute forward diﬀerences. ρ) converges qlinearly and sn satisﬁes (6. Moreover.1.2. we must specify not only the sequence {ηn }. and • if σn ≤ Kη F (xn ) p for some Kη > 0 the convergence is qsuperlinear 2 with qorder 1 + p. Theorem 6.html. Note also that Proposition 6.15) completes the proof with CG = 4¯ (1 + ∗ )−1 . • if σn → 0 the convergence is qsuperlinear.1 applies to a restarted GMRES because upon successful termination (6.2. therefore.2.15) will hold. ηn .2. Therefore.2. Then there are δ.1.1. σ such that if x0 ∈ B(δ) and the sequences {ηn } and {hn } ¯ satisfy σn = ηn + CG hn ≤ σ ¯ then the forward diﬀerence NewtonGMRES iteration xn+1 = xn + sn where sn is computed by Algorithm fdgmres with arguments (sn .2.4. We summarize our results so far in the analogs of Theorem 6.1 hold.2.3. kmax. The proofs are immediate consequences of Theorems 6. 6.
2.6. ρ) converges qlinearly with respect to · ∗ and sn satisﬁes (6. We adjust η as the iteration progresses with a variation of a choice from [69]. a forward diﬀerence formulation is not an option because approximation of the transposevector product by diﬀerences is not possible. In addition. Computation of ∇h F (x) and its transpose analytically is one option as is diﬀerentiation of F automatically or by hand. The choice in Buy this book from SIAM at http://www. xn . See also [6]. As input we must give the initial iterate. Newtonmultigrid [99] is one approach using a stationary iterative method. If a constant η is too small.Copyright ©1995 by the Society for Industrial and Applied Mathematics. If a transpose is needed. CGNR and CGNE oﬀer alternatives. GMRES(m). ηn .13.2. . as it is in the context of unconstrained minimization. In the case where hn is approximately the square root of machine roundoﬀ.2. NewtonGMRES implementation We provide a MATLAB code nsolgm in the collection that implements a forwarddiﬀerence NewtonGMRES algorithm. and [12]. If F is symmetric and positive deﬁnite. we must provide a method for forming the sequence of forcing terms {ηn }. Setting η to a constant for the entire iteration is often a reasonable strategy as we see in § 6. and • if σn ≤ Kη F (xn ) p for some Kη > 0 the convergence is qsuperlinear 2 with qorder 1 + p.5. for a small m. When storage is restricted or the problem is very large.2 states that the observed behavior of the forward diﬀerence implementation will be the same as that of an implementation with exact derivatives. but require evaluation of transposevector products. Moreover • if σn → 0 the convergence is qsuperlinear.html. may not converge rapidly enough to be useful. but the choice of that constant depends on the problem. This algorithm requires diﬀerent inputs than nsol. [78]. the vector of termination tolerances as we did for nsol. GMRES may not be practical. [111]. σn ≈ ηn if ηn is not too small and Theorem 6. 6. 6. This issue is independent of the particular linear solver and our discussion is in the general inexact Newton method setting. kmax. F. [63].ecsecurehost. 104 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS where sn is computed by Algorithm fdgmres with arguments (sn . the function. Other iterative methods may be used as the solver for the linear equation for the Newton step. Other Newtoniterative methods. NewtonCG is a possible approach. as it will be if CGNR or CGNE is used as the iterative method. This electronic version is for personal use and may not be duplicated or distributed. The reader may also be interested in experimenting with the other Krylov subspace methods described in § 3.4. hn . much eﬀort can be wasted in the initial stages of the iteration.com/SIAM/FR16. BiCGSTAB and TFQMR should be considered in all cases where there is not enough storage for GMRES(m) to perform well. The reader may explore this further in Exercise 6.3.17).
9999 are used. ηmax . In order to specify the choice at n = 0 and bound the sequence away from 1 we set (6. (6. n > 0.ecsecurehost. γηn−1 ≤ .1. 1] is a parameter. If ηn is uniformly bounded away from 1.9 asks the reader to modify nsolgm so that other choices of the sequence {ηn } can be used. n = 0. min(η A max . B It may happen that ηn is small for one or more iterations while xn is still far from the solution. There is a chance that the ﬁnal iterate will reduce F far beyond the desired level and that the cost of the solution of the linear equation for the last step will be more accurate than is really needed.5τt / F (xn ) )). ηn ). This safeguarding does improve the performance of the iteration.1 is somewhat arbitrary.1. As a measure of the degree to which the nonlinear iteration approximates the solution we begin with A ηn = γ F (xn ) 2 / F (xn−1 ) 2 . This electronic version is for personal use and may not be duplicated or distributed. A for n > 0 would guarantee qquadratic convergence by then setting ηn = ηn Theorem 6.1. ηn ).9 and ηmax = . 2 n > 0.com/SIAM/FR16. . A method of safeguarding was suggested in [69] to avoid volatile decreases in ηn .20) C ηn = min(ηmax .19) C ηn = A min(ηmax . Buy this book from SIAM at http://www. 2 A 2 min(ηmax . In [69] the choices γ = . This oversolving on the ﬁnal step can be controlled comparing the norm of the current nonlinear residual F (xn ) to the nonlinear residual norm at which the iteration would terminate τt = τa + τr F (x0 ) and bounding ηn from below by a constant multiple of τt / F (xn ) . The algorithm nsolgm does this and uses (6. γηn−1 > . The constant . max(ηn . γηn−1 )). . We use dirder to approximate directional derivatives and use the default value of h = 10−7 .5. In this way.html. the most information possible would be extracted from the inner iteration. INEXACT NEWTON METHODS 105 [69] is motivated by a desire to avoid such over solving. Over solving means that the linear equation for the Newton step is solved to a precision far beyond what is needed to correct the nonlinear iteration. n > 0.18) B ηn = ηmax . In (6. Exercise 6. The idea is that if ηn−1 is suﬃciently large we do not let ηn decrease by much more than a factor of ηn−1 . In all the examples in this book we use the value γ = .18) the parameter ηmax is an upper limit on the sequence {ηn }. max(ηn .Copyright ©1995 by the Society for Industrial and Applied Mathematics.1. A where γ ∈ (0.9 as recommended in [69]. n = 0.
η) √ 1. η) (c) x = x + s (d) Evaluate F (x) (f) If F (x) 2 √ (e) r+ = F (x) 2 / N .4 are independent of the outer iteration. we scale the l2 norm of the nonlinear √ residual by a factor of 1/ N so that constant functions will have norms that are independent of the computational mesh. (b) fdgmres(s. However. τ.Copyright ©1995 by the Society for Industrial and Applied Mathematics. rc = r0 = F (x) 2 / N √ 2. Counts of function evaluations corresponding to outer iterations are indicated by circles. but also the total cost. 6. For large and poorly conditioned problems.1. F. the cost of the entire iteration can be greatly reduced.5.4. One new function evaluation will be needed to approximate the action of F (xc ) on a vector Buy this book from SIAM at http://www.com/SIAM/FR16. x + hv. F (x). and the Krylov basis {vk }m .4 and ask the reader to make the necessary modiﬁcations to nsolgm in Exercise 6. F. Note that if only a single inner iteration is needed. We revisit the Hequation and solve a preconditioned nonlinear convectiondiﬀusion equation.html. Since the Newton steps for M F (x) = 0 are the same as those for F (x) = 0. s. because of our applications to diﬀerential and integral equations. In all the ﬁgures we plot the relative nonlinear residual F (xn ) 2 / F (x0 ) 2 against the number of function evaluations required by all inner and outer iterations to compute xn .ecsecurehost. This electronic version is for personal use and may not be duplicated or distributed. k=1 Since GMRES forms the inner iterates and makes termination decisions based on scalar products and l2 norms.3. it is equivalent to precondition the nonlinear equation before calling nsolgm. x. . Do while F (x) 2 / N > τr r0 + τa (a) Select η. 106 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Algorithm 6. From these plots one can compare not only the number of outer iterations. Hence if a large value of η can be used. nsolgm(x. GMRES will have to be restarted.9. the total cost of the outer iteration will be two function evaluations since F (xc ) will be known. Examples for NewtonGMRES In this section we consider two examples. σ = r+ /rc . F (x + hv). We illustrate the consequences of this in § 6. This enables us to directly compare the costs of diﬀerent strategies for selection of η. If a maximum of m GMRES iterations is needed (or GMRES(m) is used as a linear solver) the storage requirements of Algorithm nsolgm are the m + 5 vectors x. Note that the cost of an outer iterate is one evaluation of F to compute the value at the current iterate and other evaluations of F to compute the forward diﬀerences for each inner iterate. we also terminate the outer iteration on small l2 norms of the nonlinear residuals. The preconditioners we use in § 6. rc = r+ ≤ τr r0 + τa exit.
1 we plot the progress of the iteration using η = . In Fig.9. Chandrasekhar Hequation.1 performed slightly better and Buy this book from SIAM at http://www.9 using two schemes for selection of the parameter η. We also count the evaluation of F (x0 ) as an additional cost in the evaluation of x1 .20) to γ = . c = .20) to γ = . 6. This electronic version is for personal use and may not be duplicated or distributed.9 and ηmax = .20) and used as the default in nsolgm.25.4.ecsecurehost.6 incurred a cost of 3.2 the results change for the more illconditioned problem with c = .1 with the solid line and using the sequence given by (6. This is a slightly higher overall cost than the other approach in which {ηn } is given by (6. and 230 thousand ﬂoatingpoint operations. We set the parameters in (6.20). NewtonGMRES for the Hequation. 6.com/SIAM/FR16. This cost is entirely a result of the computation and factorization of a single Jacobian. We used τr = τa = 10−6 in this example. For each example we compare a constant value of ηn with the choice given in (6. The initial iterate is the function identically one. slower by a factor of over 1000. . This choice of ηmax was the one that did best overall in our experiments on this problem.9 and ηmax = .Copyright ©1995 by the Society for Industrial and Applied Mathematics. INEXACT NEWTON METHODS 107 and then F (x+ ) will be evaluated to test for termination. 3 outer iterations. and roughly 275 thousand ﬂoatingpoint operations. 6. We see that the constant η iteration terminates after 12 function evaluations. which terminated after 10 function evaluations. 10 10 10 10 10 10 10 10 10 0 1 2 Relative Nonlinear Residual 3 4 5 6 7 8 0 2 4 6 8 Function Evaluations 10 12 Fig. which therefore has a minimum cost of three function evaluations. Here the iteration with ηn = . In Fig.html. We set the parameters in (6.1. reported in § 5. 4 outer iterations.3 million ﬂoatingpoint operations.1. 6. The chord method with the diﬀerent (but very similar for this problem) l∞ termination condition for the same problem. We solve the Hequation on a 100point mesh with c = .25.9999.20) with the dashed line.
ecsecurehost.com/SIAM/FR16. 6. 7 outer iterations.21) We consider the partial diﬀeren − ∇2 u + Cu(ux + uy ) = f Buy this book from SIAM at http://www. tial equation (6. NewtonGMRES for the PDE.9999 10 0 Relative Nonlinear Residual 10 1 10 2 10 3 10 4 0 2 4 6 8 10 12 Function Evaluations 14 16 18 Fig. 6.20) required 23 function evaluations.2. NewtonGMRES for the Hequation. Convectiondiﬀusion equation. 6.2. c = . The iteration with the decreasing {ηn } given by (6.html. and roughly 513 thousand ﬂoatingpoint operations.Copyright ©1995 by the Society for Industrial and Applied Mathematics.4. C = 20. . 108 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 0 10 10 Relative Nonlinear Residual 1 10 2 10 3 10 4 10 5 10 6 10 7 0 5 10 15 Function Evaluations 20 25 Fig. terminated after 22 function evaluations and 7 outer iterations. and 537 thousand ﬂoatingpoint operations.3. This electronic version is for personal use and may not be duplicated or distributed.
20) reduced the number of inner iterations in the early stages of the iteration and avoided oversolving. We compare the same possibilities for {ηn } as in the previous example and report the results in Fig. 4 outer iterations.20) required 16 function evaluations.5 ). and 2.3.ecsecurehost. In the computation we preconditioned (6.21) with the fast Poisson solver fish2d. In (6. Buy this book from SIAM at http://www.Copyright ©1995 by the Society for Industrial and Applied Mathematics. however. Preconditioning is crucial for good performance.9 the reader is asked to explore this. INEXACT NEWTON METHODS 109 with homogeneous Dirichlet boundary conditions on the unit square (0.9 and ηmax = . In this case. 1) × (0.5. Examples of such cases include (1) implicit time integration of nonlinear parabolic partial diﬀerential equations or ordinary diﬀerential equations in which the initial iterate is either the converged solution from the previous time step or the output of a predictor and (2) solution of problems in which the initial iterate is an interpolant of the solution from a coarser mesh. To be consistent with the secondorder accuracy of the diﬀerence scheme we used τr = τa = h2 .5. .20) we set γ = . In cases where the initial iterate is very good and only a single Newton iterate might be needed. 6. In Exercise 6. We set C = 20 and u0 = 0.21) on a 31 × 31 grid using centered diﬀerences. a large choice of ηmax or constant η may be the more eﬃcient choice.com/SIAM/FR16. As in § 3. f has been constructed so that the exact solution was the discretization of 10xy(1 − x)(1 − y) exp(x4. This electronic version is for personal use and may not be duplicated or distributed. When preconditioning was not done. The iteration with {ηn } given by (6.7 we discretized (6. the choice of {ηn } given by (6. 1).html. the iteration for C = 20 required more than 80 function evaluations to converge. The computation with constant η terminated after 19 function evaluations and 4 outer iterates at a cost of roughly 3 million ﬂoatingpoint operations. where h = 1/32.6 million ﬂoatingpoint operations.
6. Can anything like Proposition 6.ecsecurehost. 6. and x is suﬃciently near x∗ then F (x) /(1 + ) ≤ F (x∗ )e ≤ (1 + ) F (x) . . how is the performance aﬀected if GMRES(m) is used instead of GMRES? Buy this book from SIAM at http://www.10. and ηn = . 6.1. Are the converged solutions for the preconditioned and unpreconditioned convectiondiﬀusion example in § 6.14).9.75 aﬀect the results? Present the results of your experiments as graphical iteration histories.5.1. Give an example in which F (xn ) → 0 qlinearly but xn does not converge to x∗ qlinearly.4. In the DASPK code for integration of diﬀerential algebraic equations.11.1 be proved for a ﬁnitediﬀerence BiCGSTAB or TFQMR linear solver? If not. Prove Proposition 6.5.5.5.5. 0 < < 1.1. Prove Theorems 6. ηn = 10−4 [32].4.5.5.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Verify (6. vary the parameter c and see how the performance of NewtonGMRES with the various choices of {ηn } is aﬀected.5. and for the convectiondiﬀusion equation various values of C such as C = 1. 110 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 6. [23].8.2. Experiment with the sequence of forcing terms {ηn }.8).5. (n + 2)−1 ) [56]. This electronic version is for personal use and may not be duplicated or distributed.4 equally accurate? 6. 6. what are some possible implications? 6. 6. 40 and how preconditioning aﬀects the iteration. ηn = 21−n [24].html.5.2.2. Prove Theorems 6. Verify (6.4.99999.com/SIAM/FR16.5 and 6. 6. For the convectiondiﬀusion example in § 6.6. 10. preconditioned NewtonGMRES is implemented in such a way that the data needed for the preconditioner is only computed when needed. . What happens when c = 1? See [119] for an explanation of this.1. which may be less often that with each outer iteration. 6. ηn = min( F (xn ) 2 . How do the choices ηn = 1/(n + 1). .3. such as c = . Duplicate the results in § 6. 6.1 by showing that if the standard assumptions hold. For the Hequation example in § 6. ηn = .2 and 6.5.05. Exercises on inexact Newton methods 6.12.6. For the Hequation experiment with various values of c. Is there a relation to the Shamanskii method? 6.1.2. Discuss this strategy and when it might be useful.5.5.5.999999.7.4.5.
com/SIAM/FR16.7.5. ﬂoatingpoint operations.html.Copyright ©1995 by the Society for Industrial and Applied Mathematics.14.4. NewtonBiCGSTAB. How will you deal with the need for a transpose in the case of CGNR? How does the performance (in terms of time. Apply NewtonCGNR. function evaluations) compare to NewtonGMRES? Experiment with the forcing terms.25. which allows you to choose from a variety of forward diﬀerence Krylov methods. Buy this book from SIAM at http://www. which we describe more fully in Chapter 8.5. and nsola. apply nsolgm to the same twopoint boundary value problem and compare the performance. If you have done Exercise 5. or NewtonTFQMR to the two examples in § 6.13. 6. .ecsecurehost. The MATLAB codes fdkrylov. This electronic version is for personal use and may not be duplicated or distributed. might be of use in this exercise. INEXACT NEWTON METHODS 111 6.
112 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Buy this book from SIAM at http://www. . This electronic version is for personal use and may not be duplicated or distributed.Copyright ©1995 by the Society for Industrial and Applied Mathematics.com/SIAM/FR16.ecsecurehost.html.
1) −1 x+ = xc − Bc F (xc ). these methods could have an advantage in terms of function evaluation cost over Newtoniterative methods. This means that the updated approximation to F (x∗ ) satisﬁes the secant equation (7. Broyden’s method [26] computes B+ by (7.3) B+ s = y.2) B+ = Bc + (y − Bc s)sT F (x+ )sT = Bc + . The advantages of such methods. If xc and Bc are the current approximate solution and Jacobian. In fact. In comparison with Newtoniterative methods. then (7. and hence is a very powerful alternative to Newton’s method or the chord method. the cost is little more than that of the chord method. Broyden’s method. if the Jacobian is dense. is that solution of equations with the quasiNewton approximation to the Jacobian is often much cheaper than using F (xn ) as the coeﬃcient matrix.Copyright ©1995 by the Society for Industrial and Applied Mathematics. (7. Bc is updated to form B+ .3) uniquely speciﬁes the classical secant method. For equations in several variables (7. is locally superlinearly convergent. . Hence.3.ecsecurehost. as we shall see in § 7. After the computation of x+ . This electronic version is for personal use and may not be duplicated or distributed. there is no cost in function evaluations associated with an inner iteration. Chapter 7 Broyden’s method QuasiNewton methods maintain approximations of the solution x∗ and the Jacobian at the solution F (x∗ ) as the iteration progresses. In one dimension. Broyden’s method is an example of a secant update. sT s sT s In (7. The method we present in this chapter. The construction of B+ determines the quasiNewton method. quasiNewton methods require only one function evaluation for each nonlinear iterate.3) alone is a system of N equations in N 2 113 Buy this book from SIAM at http://www.html. if a good preconditioner (initial approximation to F (x∗ )) can be found.2) y = F (x+ ) − F (xc ) and s = x+ − xc .com/SIAM/FR16.
5) −1 xn+1 = xn − Bn F (xn ) where Bn = F (x∗ )+En ≈ F (x∗ ) is generated by some method (not necessarily Broyden’s method). [96]. is that for linear problems Broyden’s method converges in 2N iterations [29].4) E = B − F (x∗ ) and the step s = x+ −xc .6) n→∞ lim En sn = 0. and [116]. In this chapter we show that if the data x0 and B0 are suﬃciently good. and [64] for discussion and analysis of several such methods. [60] on the sequence e of steps {sn } and errors in the Jacobian {En } (7. [63]. where B ≈ A is updated.Copyright ©1995 by the Society for Industrial and Applied Mathematics. We must relate the superlinear convergence condition that ηn → 0 in Theorem 6. but easier to verify condition in order to analyze Broyden’s method. When indexing of the steps is necessary. Locally superlinearly convergent secant methods can be designed to maintain the sparsity pattern. or positive deﬁniteness. [84]. the Broyden iteration will converge qsuperlinearly to the root. We will express our results in terms of the error in the Jacobian (7. This technique is based on veriﬁcation of the Dennis–Mor´ condition [61]. The role of the secant equation is very deep. The Dennis–Mor´ condition e In this section we consider general iterative methods of the form (7. [143]. [82]. [130].4) is consistent with our use of e = x − x∗ . Ax = b. 114 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS unknowns. One can also apply the method to linear least squares problems [84]. 7.1. Broyden’s method is also applicable to linear equations [29]. [82]. This electronic version is for personal use and may not be duplicated or distributed. [95]. See [62] and [64] for discussion of this topic. More subtle structural properties can also be preserved by secant methods. [143]. Veriﬁcation of qsuperlinear convergence cannot be done by the direct approaches used in Chapters 5 and 6. and we refer the reader to [28]. we will write sn = xn+1 − xn . of the approximate Jacobian. One interesting result. In this chapter we will assume that the function values are accurate and refer the reader to [66]. [62]. [104]. sn Buy this book from SIAM at http://www. In this book we focus on Broyden’s method and our methods of analysis are not as general as those in [62] and [64]. [65]. . and [114] for discussions of quasiNewton methods with inaccurate data. (7.2 to a more subtle.html. [84].1. Some examples can be found in [101]. While we could use E = B −F (x). symmetry. [104].ecsecurehost. which we will not discuss further.com/SIAM/FR16. The analysis is somewhat clearer in the linear case and the plan of this chapter is to present results for the linear case for both the theory and the examples. [113].
8) implies that En sn ≤ (2 F (x∗ ) νn + γ en ) sn which implies the DennisMor´ condition (7. We use the fundamental theorem of calculus and the standard assumptions to obtain F (x∗ )en − F (xn ) = and hence Therefore.6).7) En sn = −F (x∗ )sn − F (xn ) = −F (x∗ )en+1 + F (x∗ )en − F (xn ). 1 0 (F (x∗ ) − F (x∗ + ten ))en dt F (x∗ )en − F (xn ) ≤ γ en 2 /2.6) holds.Copyright ©1995 by the Society for Industrial and Applied Mathematics. The assumption that xn = x∗ for any n implies that the sequence (7.8) En sn ≤ F (x∗ )en+1 + γ en 2 /2.html.1 and its corollary for linear problems. This electronic version is for personal use and may not be duplicated or distributed. Our formulation of the Dennis–Mor´ result is e a bit less general than that in [60] or [63]. assume that xn → x∗ and that (7. and hence (7. [62]. Let the standard assumptions hold.com/SIAM/FR16. Let µn = En sn . Since −F (xn ) = Bn sn = F (x∗ )sn + En sn we have (7. let x0 ∈ RN be given and let {xn }∞ be given n=1 by (7. let {Bn } be a sequence of nonsingular N × N matrices. Assume that xn = x∗ for any n. by (7.9) sn /2 ≤ en ≤ 2 sn . are used to prove superlinear convergence for Broyden’s method in this chapter and many other quasiNewton methods.1.1. BROYDEN’S METHOD 115 The main result in this section. and [64] for several more examples. if xn → x∗ qsuperlinearly. then for n suﬃciently large (7.ecsecurehost. By (7. sn Buy this book from SIAM at http://www.9) we have en+1 = νn en ≤ 2νn sn .5). Theorem 7.1. . See [61].10) νn = en+1 / en is deﬁned and and νn → 0 by superlinear convergence of {xn }. Now.6) holds. e Proof. Theorem 7. [28].7) (7. Then xn → x∗ qsuperlinearly if and only if xn → x∗ and the Dennis–Mor´ condition (7. e Conversely.
This electronic version is for personal use and may not be duplicated or distributed. the proof is complete by Theorem 6.15) implies that (7. F (x∗ )−1 −1 ≤ F (x∗ )sn ≤ En sn + F (xn ) = µn sn + F (xn ) .13) µn ≤ F (x∗ )−1 −1 /2.ecsecurehost. n=0 Let { n }∞ ⊂ RN be such that n=0 n n 2 Since xn → x∗ by assumption. That paper was the basis for [168]. Buy this book from SIAM at http://www. We assume now that n is large enough so that (7. .1. sn ≤ 2 F (x∗ )−1 F (xn ) . and [115]. [104]. Since ηn → 0.html.13) holds.11) (7. We will use the l2 norm throughout our discussion. Hence.14) Moreover. ηn = 2 F (x∗ )−1 (µn + γ en ).12) sn En sn = (Bn − F (x∗ ))sn = −F (xn ) − F (x∗ )sn .2.14) and (7. 7.2. Let 0 < θ < 1 and let ˆ ˆ {θn }∞ ⊂ (θ.11) again. F (xn ) + F (xn )sn (7.1. which is the inexact Newton condition. ηn → 0 where < ∞. using the standard assumptions and (7. 2 and let {ηn }∞ be a set of vectors in RN such that ηn n=0 all n.16) F (xn ) + F (xn )sn ≤ ηn F (xn ) . Combining (7.17) then (7. 116 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS We have. We begin with a lemma from [104]. ˆ Lemma 7.15) ≤ F (xn ) + F (x∗ )sn + (F (x∗ ) − F (xn ))sn ≤ (µn + γ en ) sn . Since µn → 0 by assumption. (7. Let ψ0 ∈ RN be given.18) T ψn+1 = ψn − θn (ηn ψn )ηn + n is either 1 or 0 for lim η T ψn n→∞ n = 0.2.com/SIAM/FR16. If {ψn }∞ is given by n=1 (7. Convergence analysis Our convergence analysis is based on an approach to inﬁnitedimensional problems from [97]. for n suﬃciently large. and we employ notation from all four of these references. (7. since (7.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 2 − θ).
html. In that case. 2 Therefore. Buy this book from SIAM at http://www.21) Hence (7. ≤ ψn ≤ ψn 2 2 2 2 T − θn (2 − θn )(ηn ψn )2 ˆ T − θ2 (ηn ψn )2 ≤ ψn 2 .21) we conclude that ψn+1 2 ≤ µ.20) a2 − b2 ≤ a − b2 .20) we conclude that if ψn = 0 then T ψn − θn (ηn ψn )ηn 2 ≤ ψn 2 2 2 T − θn (2 − θn )(ηn ψn )2 T θn (2 − θn )(ηn ψn )2 . From (7. BROYDEN’S METHOD 117 Proof.ecsecurehost. . in fact. which holds even if ψn = 0.22) T (ηn ψn )2 ≤ − ψn+1 2 ≤ ψn 2 − T θn (2 − θn )(ηn ψn )2 + 2 ψn 2 n 2. This electronic version is for personal use and may not be duplicated or distributed.19) implies convergence to zero of the terms in the series. 2 ψn 2 ≤ ψn Hence if ψn = 0 (7. This inequality is used often in the analysis of quasiNewton methods [63]. 2 ψn 2 ( ψn θn (2 − θn ) 2 − ψn+1 2 + n 2 ). We ﬁrst consider the case in which n = 0 for all n. which is (7. To prove the result for n = 0 we use the inequality (7. 2a which is valid for a > 0 and b ≤ a. ˆ θ2 We let M → ∞ to obtain (7.19) ∞ n=0 T (ηn ψn )2 < ∞.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Convergence of the series in (7. From (7.18).com/SIAM/FR16. we can use the fact that ˆ θn (2 − θn ) > θ2 > 0 to show that the sequence {ψn } is bounded in l2 norm by ψ0 ψn+1 2 2 2 and. M n=0 T (ηn ψn )2 ≤ ψ0 2 2 − ψM +1 ˆ θ2 2 2 ≤ ψ0 2 . for any M > 0.
1. This property. the Broyden update has a very simple form. We will consider a modiﬁed update (7.Copyright ©1995 by the Society for Industrial and Applied Mathematics. In this section F (x) = Ax−b and Bn = A+En . Let A be nonsingular.1 is nicely illustrated by this application of the lemma. 2]. ˆ θ2 Hence (7.2. is an important part of the convergence analysis.2. Let Bc be nonsingular and let B+ be formed by the Broyden update (7. This electronic version is for personal use and may not be duplicated or distributed.25) E+ 2 ≤ Ec 2 . We may use Lemma 7. called bounded deterioration [28]. Our ﬁrst task is to show that the errors in the Broyden approximations to A do not grow.2. sT s sT s If xc and Bc are the current approximations to x∗ = A−1 b and A and s = x+ − xc then y − Bc s = (Ax+ − Axc ) − Bc (x+ − xc ) = −Ec s and therefore (7.ecsecurehost. Linear problems.2.2.html. 7.1 to prove a result from [130] on the convergence of Broyden’s method for linear problems. The role of ˆ the parameter θ and the sequence {θn } in Lemma 7. Then (7. Buy this book from SIAM at http://www.23) B+ = Bc + θc (y − Bc s)sT (Ax+ − b)sT = Bc + θc . and xc ∈ RN be given. θc ∈ [0.24) B+ = Bc + θc (y − Bc s)sT (Ec s)sT = Bc − θc . 118 where ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS ∞ i 2 i=0 µ= Hence M T 2 n=0 (ηn ψn ) + ψ0 2 .com/SIAM/FR16. In the linear case. ≤ 2µ ˆ θ2 2µ ˆ θ2 M n=0 ( ψn 2 − ψn+1 2 + M n 2) = ψ0 2 − ψM +1 2+ n 2 n=0 ≤ 2µ2 . The standard assumptions in this case reduce to nonsingularity of A. The reader should compare the statement of the result with that in the nonlinear case. s 2 s 2 2 2 Lemma 7. .23).19) holds and the proof is complete.
1.26) = Ec (I − θc Ps ).2. Assume that {θn } satisﬁes ˆ the hypotheses of Lemma 7.1 for some θ ∈ (0. If {θn } is such that the matrices Bn obtained by (7.27) This completes the proof as E+ 2 Ps = ssT . In [130] one suggestion was γc  ≥ σ. By subtracting A from both sides of (7. where Ps is the orthogonal projector (7.ecsecurehost. For linear problems one can show that the Dennis–Mor´ condition implies e convergence and hence the assumption that convergence takes place is not needed. 1). Let A be nonsingular.28) −1 xn+1 = xn − Bn (Axn − b) then the Dennis–Mor´ condition implies convergence of {xn } to x∗ = A−1 b. 1) is ﬁxed.5. Buy this book from SIAM at http://www.1.1 for some θ ∈ (0. Speciﬁcally Proposition 7. If {θn } is such the matrices Bn obtained by (7.1. Now. e We leave the proof as Exercise 7.html. 1 − sign(γc )σ θc = . BROYDEN’S METHOD 119 Proof.2. 1). 1 − γc where γc = −1 −1 (Bc y)T s (Bc As)T s = 2 s 2 s 2 2 and σ ∈ (0. However the results in [130] assume only that the ˆ sequence {θn } satisﬁes the hypotheses of Lemma 7.2.23) are nonsingular then the modiﬁed Broyden iteration (7. The superlinear convergence result is remarkable in that the initial iterate need not be near the root. Let A be nonsingular. Theorem 7.28) converges qsuperlinearly to x∗ = A−1 b for every initial iterate x0 .2. . Assume that {θn } satisﬁes the ˆ hypotheses of Lemma 7.com/SIAM/FR16.1 for some θ ∈ (0. otherwise. s 2 2 ≤ Ec 2 I − θ c Ps 2 and I − θc Ps 2 ≤ 1 by orthogonality of Ps and the fact that 0 ≤ θc ≤ 2. θc can always be selected to make B+ nonsingular. 1) and that θc is always chosen so that B+ is nonsingular.24) we have E+ = E c − θ c (Ec s)sT s 2 2 (7.2.Copyright ©1995 by the Society for Industrial and Applied Mathematics.1. This electronic version is for personal use and may not be duplicated or distributed.23) are nonsingular and {xn } is given by the modiﬁed Broyden iteration (7.
However we must show that the sequence {Bn } of Broyden updates exists and remains nonsingular. by (7. that does not use the Frobenius norm and extends to various inﬁnitedimensional settings.1 and Proposition 7. [84]. This electronic version is for personal use and may not be duplicated or distributed.30) T ηn ψ n = T En sn (En φ)T sn = φT → 0. . Assuming that (7. sn 2 As j was arbitrary.com/SIAM/FR16. sn 2 2 and. ψn = En φ.1) and (7. we have n→∞ lim En sn =0 sn 2 which is the Dennis–Mor´ condition. that the Broyden iteration will terminate in 2N steps in the linear case.1. [143]. the unit vector in the jth En sn coordinate direction. T ηn = sn / sn 2 . sn 2 sn 2 This completes the proof. Let Pn = Since for all v ∈ RN and all n T v T (En φ) = φT (En v) sn sT n . The result here is not a local convergence result. 120 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Proof. Our approach is to show that for every φ ∈ RN that (7. For nonlinear problems our analysis is local and we set θn = 1. Bn is nonsingular for all n. [82].html.26) T T En+1 φ = (I − θn Pn )En φ. Our analysis of the nonlinear case is diﬀerent from the classical work [28].1 with n =0 to conclude that (7. B0 ) exists if F is deﬁned at xn for all n. and we may invoke Lemma 7.29) holds. x0 .29) n→∞ lim φT En sn sn 2 = 0. It is global in the sense that qsuperlinear convergence takes place independently of the initial choices of x and B.2.2. {Bn }) for the data (F. Buy this book from SIAM at http://www. It is known [29]. we provide a proof based on ideas from [97].ecsecurehost.1.2. Definition 7. 7. [115]. As indicated in the preface.2.1. [63]. Nonlinear problems.2. This will imply qsuperlinear convergence e by Theorem 7. and [104].2). setting φ = ej .Copyright ©1995 by the Society for Industrial and Applied Mathematics. We say that the Broyden sequence ({xn }. We make a deﬁnition that allows us to compactly express this fact. shows that the j component of has limit zero. Let φ ∈ RN be given. and xn+1 and Bn+1 can be computed from xn and Bn using (7.
The development of the proof is complicated. together with the qlinear convergence e proved earlier completes the proof of local qsuperlinear convergence.7) implies that Ec s = −F (x+ ) + (F (x+ ) − F (xc ) − F (x∗ )s) (7.6).ecsecurehost. We ﬁrst prove a bounded deterioration result like Lemma 7. Theorem 7. Bounded deterioration. the errors in the Jacobian approximation can grow as the iteration progresses. However. B0 ) and that the Broyden iterates converge qsuperlinearly to x∗ .32) as F (x+ ) = −Ec s + and then using (7. BROYDEN’S METHOD 121 We will show that if the standard assumptions hold and x0 and B0 are suﬃciently good approximations to x∗ and F (x∗ ) then the Broyden sequence exists for the data (F.2.2.2) to obtain E+ = Ec (I − Ps ) + where Ps is is given by (7. We then show that if the initial data is suﬃciently good. However in the nonlinear case.32) = −F (x+ ) + We begin by writing (7. x0 .Copyright ©1995 by the Society for Industrial and Applied Mathematics. Unlike the linear case. (F (xc + ts) − F (x∗ ))s dt (∆c s)sT s 2 2 (F (xc + ts) − F (x∗ )) dt.31) E+ 2 ≤ Ec + γ( ec 2 + e+ 2 )/2 Proof.2). Note that (7. which. Let the standard assumptions hold. .html. Buy this book from SIAM at http://www.2. This electronic version is for personal use and may not be duplicated or distributed. the sequence { En } need not be monotone nonincreasing.com/SIAM/FR16. this growth can be limited and therefore the Broyden sequence exists and the Broyden iterates converges to x∗ at least qlinearly. Let xc ∈ Ω and a nonsingular matrix Bc be given.27) and ∆c = 1 0 1 0 1 0 (F (xc + ts) − F (x∗ ))s dt. Our ﬁrst task is to show that bounded deterioration holds. Finally we verify the Dennis–Mor´ condition (7.2. Assume that −1 x+ = xc − Bc F (xc ) = xc + s ∈ Ω and B+ is given by (7. the possible increase can be bounded in terms of the errors in the current and new iterates. Then (7.
2. Then. (1 + r)γδ ≤ δ1 .2. 122 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Hence. Let r ∈ (0. . Theorem 7.33) δ2 = 2(1 − r) and set (7. Let the standard assumptions hold.2. the qfactor is at most KA (δ + δ1 ) ≤ r. . which will prove the result. Theorem 7. en+1 2 2 and therefore 2 En+1 ≤ En 2 + γ(1 + r) en 2 /2 ≤ En (1 + r)γδ 2 We can estimate En 2 . We will prove the result by choosing δB and reducing δ if necessary so that E0 2 ≤ δB will imply that En 2 ≤ δ1 for all n. x0 .ecsecurehost. This electronic version is for personal use and may not be duplicated or distributed.Copyright ©1995 by the Society for Industrial and Applied Mathematics. By Theorem 7. Proof. Local qlinear convergence.com/SIAM/FR16. + γ(1 + r)rn δ/2. 2 2 2 We show that En 2 ≤ δ1 by induction. Since E0 2 < δB < δ1 we may begin the induction.2 for the linear case E+ and the proof is complete.3 hold. 1) be given. En−1 2 . Let δ and δ1 be small enough so that the conclusions of Theorem 5. n j=0 < δ1 . Then there are δ and δB such that if x0 ∈ B(δ) and E0 2 < δB the Broyden sequence for the data (F. where KA is from the statement of Theorem 5. To do this reduce δ if needed so that γ(1 + r)δ < δ1 (7.4. 2(1 − r) Buy this book from SIAM at http://www. in the same way to obtain En+1 2 ≤ E0 2+ rj ≤ δB + which completes the proof. Now assume that Ek 2 < δ1 for all 0 ≤ k ≤ n.2. We must explore the relation between the size of E0 needed to enforce qlinear convergence and the qfactor.3. reducing δ and δ1 further if needed.3 states that Broyden’s method is locally qlinearly convergent. B0 ) exists and xn → x∗ qlinearly with qfactor at most r.34) δB = δ1 − δ 2 . En+1 Since En 2 2 2 ≤ En ≤ r en 2 + γ( en+1 2 + en 2 )/2. .4. . The proof is more complex than similar results in Chapters 5 and 6.2.3. by the standard assumptions ∆c 2 ≤ γ( ec ≤ Ec 2 + e+ 2 )/2.html. + ∆c Just as in the proof of Lemma 7.
2. as in the proof of Theorem 7. This electronic version is for personal use and may not be duplicated or distributed. sn 2 2 (F (xn + tsn ) − F (x∗ )) dt.3 and the standard assumptions imply ∆n 2 ≤ γ(1 + r)rn δ/2. ηn = sn / sn 2 .1.1 with T ψn = En φ. B0 ) exists and xn → x∗ qsuperlinearly.1.1 implies that T ηn ψn = T En sn (En φ)T sn = φT → 0.com/SIAM/FR16. [31]. the oldest of the stored vectors is replaced by the most recent.Copyright ©1995 by the Society for Industrial and Applied Mathematics. BROYDEN’S METHOD 123 Veriﬁcation of the Dennis–Mor´ condition.2. Note that T T En+1 φ = (I − Pn )En φ + Pn ∆T φ.6) holds e to complete the proof. and n = Pn ∆T φ. implies the Dennis–Mor´ condition and e completes the proof. Let δ and δB be such that the conclusions of Theorem 7.1 we need only show that Dennis–Mor´ condition (7. As in the proof of Theorem 7.2. Another approach. Then there are δ and δB such that if x0 ∈ B(δ) and E0 2 < δB the Broyden sequence for the data (F. Proof. Let the standard assumptions hold. Hence Lemma 7. Implementation of Broyden’s method The implementation of Broyden’s method that we advocate here is related to one that has been used in computational ﬂuid mechanics [73].html.1 let Pn = Set ∆n = 1 0 sn sT n . x0 . Let φ ∈ RN be arbitrary.2. The ﬁnal task is to verify e the Dennis–Mor´ condition. 7. [142].2. In these methods.2. e Theorem 7. Buy this book from SIAM at http://www.3. sn 2 sn 2 This.2. Some such methods are called limited memory formulations in the optimization literature [138]. n We wish to apply Lemma 7. .ecsecurehost. after the storage available for the iteration is exhausted. clears the storage and starts over. By Theorem 7.4.3 hold. n The hypothesis in the lemma that n n 2 <∞ holds since Theorem 7. restarting.
B0 . [177]. Lemma 7.4.36) −1 = xn − (B0 Cn )−1 F (xn ) = xn − Bn F (xn ) = xn+1 . . The basis for this is the following simple formula [68]. {Cn }) exists for the data −1 (B0 F.Copyright ©1995 by the Society for Industrial and Applied Mathematics.35) holds by deﬁnition when n = 0. Bn = B0 Cn (7. Our plan will be to store Bn in a compact and easy to apply form. If (7. We leave the proof to Exercise 7.ecsecurehost. The consequence of this proof for implementation is that we can assume that B0 = I.1. We begin by showing how the initial approximation to F (x∗ ).35) holds for a given n. In the context of a sequence of Broyden updates {Bn } we have for n ≥ 0. {Bn }) for the data (F. B0 ) exists. [176]. Proof.1. This electronic version is for personal use and may not be duplicated or distributed.36). may be regarded as a preconditioner and incorporated into the deﬁntion of F just as preconditioners were in the implementation of NewtonGMRES in § 6. sn 2 2 = B0 Cn + B0 This completes the proof. x0 . we can incorporate that into the −1 function evaluation. Buy this book from SIAM at http://www. Hence Bn+1 = Bn + F (xn+1 )sT n sn 2 2 −1 B0 F (yn+1 )sT n = B0 Cn+1 . If a better choice for B0 exists. We proceed by induction.com/SIAM/FR16. Our basic implementation is a nonlinear version of the implementation for linear problems in [67]. though a limited memory approach could be used as well. (B −1 u)v T (7. x0 . Now sn = xn+1 − xn = yn+1 − yn by (7.37) is called the Sherman–Morrison formula. We use a restarted approach.3. T Bn+1 = Bn + un vn .37) B −1 . Equation (7. then −1 −1 yn+1 = yn − Cn B0 F (yn ) (7.3. Let B be a nonsingular N × N matrix and let u. Then B + uv T is invertible if and only if 1 + v T B −1 u = 0.2. Assume that the Broyden sequence ({xn }.5.html. 124 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS This latter approach is similar to the restarted GMRES iteration. I) and xn = yn . v ∈ RN . Proposition 7. Then the Broyden sequence ({yn }. [11]. (B + uv T )−1 = I − 1 + v T B −1 u The expression (7.35) for all n. In this case.
38) as (7. One may carry this one important step further [67] and eliminate the need to store the sequence {wn }.ecsecurehost. we can write (7. . −1 T −1 wn = (Bn un )/(1 + vn Bn un ) (7. Buy this book from SIAM at http://www. the computation of the Broyden step) can be computed from the 2n vectors {wj .com/SIAM/FR16.42) −1 Bn = n−1 j=0 I+ sj+1 sT j sj 2 2 . vj }n−1 at a cost of O(N n) j=0 ﬂoatingpoint operations.. the Broyden step for the following iteration is n−1 −1 T (7. Therefore.40) = n−2 j=0 T (I − wj vj )F (xn ) 2 T + vn−1 w)−1 wn−1 = Cw w where Cw = ( sn−1 sn T = −(I − wn−1 vn−1 )w. (I − w0 v0 ) 2 and vn = sn / sn 2 . Note that (7. In fact. This electronic version is for personal use and may not be duplicated or distributed. Moreover. BROYDEN’S METHOD 125 where un = F (xn+1 )/ sn Setting we see that. for n ≥ 0.html.38) = n−1 j=0 T (I − wj vj ).Copyright ©1995 by the Society for Industrial and Applied Mathematics. (7.40) implies that for n ≥ 1 T −1 sn = −w + Cw w(vn−1 w) = w(−1 + Cw (Cw − sn−1 2 )) = −Cw w sn−1 Hence.38) is valid for n ≥ 0.e. sn = −Bn F (xn ) = − j=0 Since the product n−2 j=0 T (I − wj vj )F (xn ) must also be computed as part of the computation of wn−1 we can combine the computation of wn−1 and sn as follows: w (7. . Since the empty matrix product is the identity. wn = −sn+1 / sn 2 . −1 = (I − w T T T Bn n−1 vn−1 )(I − wn−2 vn−2 ) .41) 2 = − sn−1 2 wn−1 .39) (I − wj vj )F (xn ). one need only store the steps {sn } and their norms to construct the sequence {wn }. if B0 = I. (7. −1 Hence the action of Bn on F (xn ) (i. .
42) and (7.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Note that B0 = I is implicit in the algorithm. nmax) 1. the denominator in (7. By Proposition 7. 2. s0 = −F (x).44) to describe an algorithm. Do while itc < maxit (a) n = n + 1.3. restarting with B0 when storage is exhausted.39) directly to compute sn+1 because sn+1 appears on both sides of the equation sn+1 = − I + (7.42) and (7. The termination criteria are the same as for Algorithm nsol in § 5. This electronic version is for personal use and may not be duplicated or distributed. itc = itc + 1 (b) x = x + sn (c) Evaluate F (x) (d) If F (x) 2 ≤ τr r0 + τa exit. z = −F (x) ii. brsol(x. 126 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS We cannot use (7. We can use (7.43) =− I+ sn+1 sT n sn 2 2 sn+1 sT n sn 2 2 n−1 j=0 I+ sj+1 sT j 2 sj 2 F (xn+1 ) −1 Bn F (xn+1 ). F. itc = 0.44) −1 1 + sT Bn F (xn+1 )/ sn n 2 2 T −1 = 1 + vn Bn un is nonzero unless Bn+1 is singular. τ. We use a restarting approach in the algorithm. is of the same order as GMRES. r0 = F (x) 2 . The storage requirement.6. Our looping convention is that the loop in step 2(e)ii is skipped if n = 0. this is consistent with setting the empty matrix product to the identity.com/SIAM/FR16. Instead. (e) if n < nmax then i. adding to the input list a limit nmax on the number of Broyden iterations taken before a restart and a limit maxit on the number of nonlinear iterations. of n + O(1) vectors for the nth iterate.1.ecsecurehost. Algorithm 7.43) for sn+1 to obtain (7. for j = 0. we solve (7.1. Our implementation is similar to GMRES(m).3. n − 1 z = z + sj+1 sT z/ sj j 2 2 Buy this book from SIAM at http://www. . Our notation and algorithmic framework are based on [67]. maxit.html. n = −1. making Broyden’s method competitive in storage with GMRES as a linear solver [67].44) sn+1 = − −1 Bn F (xn+1 ) −1 1 + sT Bn F (xn+1 )/ sn n 2 2 .
If the data is suﬃciently good. z.5 million ﬂoating point operations. Examples for Broyden’s method Buy this book from SIAM at http://www. The restarted iteration required 24 iterations and 3.html. BROYDEN’S METHOD 127 iii. especially for large problems. Linear problems.1 we present results for (3.ecsecurehost. If n < nmax. and therefore need storage for nmax + 2 vectors. and solution as the example in § 3. We set τa = τr = h2 where h = 1/32. [120]. One can also see highly irregular performance from the 7. As an example we consider the linear PDE (3. coeﬃcients.1.7. In Fig.Copyright ©1995 by the Society for Industrial and Applied Mathematics. than the dense matrix approach proposed in [85]. requires O(nN ) ﬂoatingpoint operations and storage of n + 3 vectors.7. sn+1 = z/(1 − sT z/ sn 2 ) n 2 (f) if n = nmax then n = −1.4. the iteration will restart with xnmax . where more elaborate numerical experiments that those given here are presented. We use the same mesh. we allow for increases in the nonlinear residual. Our implementation of brsol in the collection of MATLAB codes stores z and F separately in the interest of clarity and therefore requires nmax + 3 vectors of storage. In view of all this. then the nth iteration of brsol. in terms of both time and storage. If n = nmax. which computes xn and sn+1 . For linear problems. where z and F (x) may occupy the same j=0 storage location.com/SIAM/FR16. A MATLAB implementation of brsol is provided in the collection of MATLAB codes. a slightly higher cost than the GMRES solution. We use the MATLAB code brsol. √ In all the examples in this section we use the l2 norm multiplied by 1/ N . bounded deterioration implies that the Broyden matrices will be well conditioned and superlinear convergence implies that only a few iterates will be needed. The theory in this chapter indicates that Broyden’s method can be viewed as an alternative to GMRES as an iterative method for nonsymmetric linear equations. However the dense matrix approach makes it possible to detect ill conditioning in the approximate Jacobians. and in several papers on inﬁnitedimensional problems as well [91]. This requirement of one more vector than the algorithm in [67] needed is a result of the nonlinearity. x. 7. This point has been explored in some detail in [67]. the steps {sj }n . righthand side. s = −F (x). This electronic version is for personal use and may not be duplicated or distributed. [104]. Broyden’s method terminated after 9 iterations at a cost of roughly 1.5 million ﬂoating point operations. We compare Broyden’s method without restarts (solid line) to Broyden’s method with restarts every three iterates (dashed line). not compute snmax+1 .33) preconditioned with the Poisson solver fish2d. we feel that the approach of Algorithm brsol is the best approach.4.33) from § 3. The Sherman–Morrison approach in Algorithm brsol is more eﬃcient. 7. and F (x). .
Broyden’s method for the linear PDE. As before we solve the Hequation on a 100point mesh with c = . however. We consider the Hequation from Chapters 5 and 6 and a nonlinear elliptic partial diﬀerential equation. the cost of a solution by Broyden’s method could be much less than that of one by NewtonGMRES. If storage is at a premium. The examples in this section. 7. . Broyden’s method could be at a disadvantage as it might need to be restarted many times. Broyden’s method and NewtonGMRES both require more storage as iterations progress. NewtonGMRES as the inner iteration progresses and Broyden’s method as the outer iteration progresses.ecsecurehost. In the linear case [67] one can store only the residuals and recover the terminal iterate upon exit. Our general recommendation is to try both. Storage of the residual. the vector z. as well as those in § 8. This electronic version is for personal use and may not be duplicated or distributed. 7. We compare Broyden’s method without restarts (solid line) to Broyden’s method with restarts every three iterates (dashed line). For c = .2.3).Copyright ©1995 by the Society for Industrial and Applied Mathematics.9 both methods required six iterations for convergence. if function evaluations are very expensive. 7. The cost in function evaluations for Broyden’s method is one for each nonlinear iteration and for NewtonGMRES one for each outer iteration and one for each inner iteration. and the number of nonlinear iterations is large. Nonlinear problems.9 (Fig. 10 1 10 Relative Residual 0 10 1 10 2 10 3 0 5 10 Iterations 15 20 25 Fig. and the sequence of steps saves a vector over the nonlinear case. 7. 128 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS restarted method. a substantial improvement over the NewtonGMRES implementation. The implementation without restarts required 153 thousand ﬂoatingpoint operations and the restarted method 150. Hence.4.html.2) and c = . Chandrasekhar Hequation. Buy this book from SIAM at http://www.4 illustrate the diﬀerences in the two methods.9999 (Fig.com/SIAM/FR16.1.
9999. One should take some care in drawing general conclusions. it may be the case Buy this book from SIAM at http://www. The unrestarted iteration converges in 10 iterations at a cost of roughly 249 thousand ﬂoatingpoint operations. 10 0 10 Relative Nonlinear Residual 1 10 2 10 3 10 4 10 5 10 6 10 7 0 2 4 6 8 10 Iterations 12 14 16 18 Fig. while that in Broyden in the nonlinear iteration. Broyden’s method for the Hequation. BROYDEN’S METHOD 129 For c = . The restarted iteration requires 18 iterations and 407 ﬂoatingpoint operations. 7. 10 10 10 10 10 10 10 10 10 0 1 2 Relative Nonlinear Residual 3 4 5 6 7 8 0 1 2 3 Iterations 4 5 6 Fig.html. 7.com/SIAM/FR16. the storage in NewtonGMRES is used in the inner iteration. both implementations performed better than NewtonGMRES. Even so. .9999.2. the restarted iteration has much more trouble. While Broyden’s method on this example performed better than NewtonGMRES.3.9.Copyright ©1995 by the Society for Industrial and Applied Mathematics. c = .ecsecurehost. Broyden’s method for the Hequation. This electronic version is for personal use and may not be duplicated or distributed. c = . If many nonlinear iterations are needed.
We allowed for an increase in the residual.2) of 2.3. . F (xc + hv)) to accommodate the NewtonGMRES iteration. In Fig. This approach Buy this book from SIAM at http://www. F (xc ). This is not supported by the local convergence theory for nonlinear problems. One can do this in brsol by setting the third entry in the parameter list to 1.4. 7.4 we compare Broyden’s method without restarts (solid line) to Broyden’s method with restarts every 8 iterates (dashed line). We use the same mesh width h = 1/32 on a uniform square grid.ecsecurehost.Copyright ©1995 by the Society for Industrial and Applied Mathematics.4. xc .6 million ﬂoatingpoint operations.3 we will return to this example. The unrestarted Broyden iteration converges in 12 iterations at a cost of roughly 2.html. This is a modest improvement over the NewtonGMRES cost (reported in § 6.2) and more storage was used. s. However.4. u0 = 0. 10 1 10 Relative Nonlinear Residual 0 10 1 10 2 10 3 10 4 0 2 4 6 8 Iterations 10 12 14 16 Fig. We set C = 20. as one would if the problem were truly linear. This electronic version is for personal use and may not be duplicated or distributed. and z (when stored in the j=0 same place as F (x)).2 million ﬂoatingpoint operations. Hence restarting the Broyden iteration every 8 iterations most closely corresponds to the NewtonGMRES requirements. however it can be quite successful in practice. Therefore the GMRES inner iteration for the NewtonGMRES approach needed storage for at most 10 vectors (the Krylov basis. Convectiondiﬀusion equation. which happened on the second iterate. 7. The Broyden iteration when restarted every n iterates. We preconditioned with the Poisson solver fish2d. requires storage of n + 2 vectors {sj }n−1 . We consider the nonlinear convectiondiﬀusion equation from § 6.2. x.com/SIAM/FR16. the residual norms do not monotonically decrease in the Broyden iteration (we ﬁx this in § 8. Broyden’s method for nonlinear PDE.4. At most ﬁve inner GMRES iterations were required. and τr = τa = h2 . 130 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS that Broyden’s method may need restarts while the GMRES iteration for the linear problem for the Newton step may not. In § 8.
com/SIAM/FR16. but the convergence was extremely irregular after the restart. BROYDEN’S METHOD 131 took 15 iterations to terminate at a cost of 2.4 million ﬂoatingpoint operations. Buy this book from SIAM at http://www. .Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.ecsecurehost.html.
3. Extend Proposition 7. and values for the convection coeﬃcient d. so named because of its inferior performance in practice [63]. then A + U V is nonsingular if and only if I + V A−1 U is a nonsingular M × M matrix. then (A + U V )−1 = A−1 − A−1 U (I + V A−1 U )−1 V A−1 .4. Try to solve the nonlinear convectiondiﬀusion equation in § 6. 1) with homogeneous Dirichlet boundary conditions.5.5. How does the performance of Broyden’s method diﬀer from NewtonGMRES? 7.5.5. Compare the performance of Broyden’s method and NewtonGMRES on the Modiﬁed Bratu Problem −∇2 u + dux + eu = 0 on (0. Modify nsol to use Broyden’s method instead of the chord method after the initial Jacobian evaluation. This electronic version is for personal use and may not be duplicated or distributed. 7. 7. 7. If this is the case. 7. Solve the Hequation with Broyden’s method and c = 1.2. . [199]. 7.3.1. What qfactor for linear convergence do you see? Can you account for this by applying the secant method to the equation x2 = 0? See [53] for a complete explanation.html. Set τa = τr = 10−8 . Prove Proposition 7.3. U is N × M . Exercises on Broyden’s method 7. How does this compare with nsol on the examples in Chapter 5.5. What happens in the diﬀerential equation examples if preconditioning is omitted? 7.5. V is M × N . Duplicate the results reported in § 7.1 by showing that if A is a nonsingular N × N matrix. updates an approximation to the inverse of the Jacobian at the root so that B −1 ≈ F (x∗ )−1 satisﬁes the inverse secant equation −1 B+ y = s (7.9. This is the Sherman–Morrison–Woodbury formula [68]. 7.7.5.Copyright ©1995 by the Society for Industrial and Applied Mathematics. initial iterates. Prove Proposition 7.5. Experiment with various mesh widths.5.8.6. See [102] for further generalizations. Precondition with the Poisson solver fish2d. 132 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 7.5.ecsecurehost.1.45) Buy this book from SIAM at http://www. The bad Broyden method [26]. 1) × (0.4 with Broyden’s method using various values for the parameter C.com/SIAM/FR16.2.4.1.5. Vary the parameters in the equations and the number of vectors stored by Broyden’s method and report on their eﬀects on performance.
Implement the Schubert algorithm on an unpreconditioned discretized partial diﬀerential equation (such as the Bratu problem from Exercise 7.14 in Chapter 5 to numerically estimate the qorder of Broyden’s method for some of the examples in this section (both linear and nonlinear). Try to implement the bad Broyden method in a storageeﬃcient manner using the ideas from [67] and § 7. [27] is a quasiNewton update for sparse matrices that enforces both the secant equation and the sparsity pattern. and main diagonal are updated. BROYDEN’S METHOD 133 with the rankone update −1 −1 B+ = Bc + −1 (s − Bc y)y T .11. Note that only the subdiagonal.8) and compare it to NewtonGMRES and Broyden’s method. The Schubert or sparse Broyden algorithm [172].5.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 7.5. and compare the performance to that of Broyden’s method.7. Does the relative performance of the methods change as h is decreased? This is interesting even in one space dimension where all matrices are tridiagonal. . Discuss the diﬀerences in implementation from Broyden’s method. This electronic version is for personal use and may not be duplicated or distributed. 7. 7. Implement the bad Broyden method. What can you conclude from your observations? Buy this book from SIAM at http://www. How would the Schubert algorithm be aﬀected by preconditioning? Compare your analysis with those in [158].12.10. References [101]. apply it to the examples in § 7. are relevant to this exercise.4.13. Does anything go wrong? See [67] for more about this.html. For problems with tridiagonal Jacobian. 7.com/SIAM/FR16. and [189]. j ≤ N and i − j ≤ 1.5.5. [93]. [125].3. [92]. the update is (y − Bc s)i sj (B+ )ij = (Bc )ij + i+1 2 k=i−1 sk for 1 ≤ i. Read about the algorithm in [172] and prove local superlinear convergence under the standard assumptions and the additional assumption that the sparsity pattern of B0 is the same as that of F (x∗ ). and [112]. superdiagonal.ecsecurehost. Use your result from Exercise 5. y 2 2 Show that the bad Broyden method is locally superlinearly convergent if the standard assumptions hold. for example.5.
134 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Buy this book from SIAM at http://www.com/SIAM/FR16.ecsecurehost. . This electronic version is for personal use and may not be duplicated or distributed.Copyright ©1995 by the Society for Industrial and Applied Mathematics.html.
8.9 × 1017 . and hence the magnitude of the Newton step is much larger than that of the iterate itself. This observation motivates the simple ﬁx of reducing the size of the step until the size of the nonlinear residual is decreased. 9. [25]. [70]. We begin with single equations. Of the many such algorithms we will focus on the class of line search methods and the Armijo rule [3]. [63]. We select the line search paradigm because of its simplicity. The implementation and analysis of the line search method we develop in this chapter do not depend on whether iterative or direct methods are used to compute the Newton step or upon how or if the Jacobian is computed and stored. [70]. such as trust region [153]. This electronic version is for personal use and may not be duplicated or distributed. but the length of s is too large. 2. The reason for this is that f (x) = (1 + x2 )−1 is small for large x. [109]. a failure of Newton’s method that results from an inaccurate initial iterate. [25]. [129]. Single equations If we apply Newton’s method to ﬁnd the root x∗ = 0 of the function f (x) = arctan(x) with initial iterate x0 = 10 we ﬁnd that the initial iterate is too far from the root for the local convergence theory in Chapter 5 to be applicable. Chapter 8 Global Convergence By a globally convergent algorithm we mean an algorithm with the property that for any initial iterate the iteration either converges to a root of F or fails to do so in one of a small number of ways.1. where the algorithms can be motivated and intuition developed with a very simple example.com/SIAM/FR16. [2]. If we look more closely.Copyright ©1995 by the Society for Industrial and Applied Mathematics. we see that the Newton step s = −f (xc )/f (xc ) is pointing toward the root in the sense that sxc < 0. . −138. [88]. f (x) = arctan(x) ≈ ±π/2. [24].ecsecurehost. Other methods. implemented inexactly [24]. [173].9 × 104 .5 × 109 . can be used to accomplish the same objective. [159]. and because it is trivial to add a line search to an existing locally convergent implementation of Newton’s method.html. A prototype algorithm is given below. [154]. [198]. −1. and continuation/homotopy methods [1]. 135 Buy this book from SIAM at http://www. [181]. We can see this eﬀect in the sequence of iterates: 10.
We call this method a line search because we search along the line segment (xc . We plot the progress of the iteration in Fig. there is the possibility of oscillation about a solution without convergence [63]. and 4 required 3. and 2 steplength reductions.8. Iterates 1.ecsecurehost.56. where the local theory is valid. τ ) 1. −1.5.html. we obtain the sequence 10. the minimum number of two function evaluations per iterate is required. 136 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Algorithm 8.1.99. lines1(x. We plot arctan(x)/arctan(x0 ) as a function of the number of function evaluations with the solid line. f. The outer iterations are identiﬁed with circles as in § 6. One does not set x+ = x + s without testing the step to see if the absolute value of the nonlinear residual has been decreased. 8. To remedy this and to make analysis possible we replace the test for simple decrease with one for suﬃcient decrease. 3.3. The quadratic convergence behavior of Newton’s method is only apparent very late in the iteration. −8. (b) s = −f (x)/f (x) (search direction) (c) xt = x + s (trial point) (d) If f (xt ) < f (x) then x = xt (accept the step) else s = s/2 goto 2c (reject the step) When we apply Algorithm lines1 to our simple example. Do while f (x) > τr r0 + τa (a) If f (x) = 0 terminate with failure. −0. 1. As we move from the right endpoint to the left. xc − f (xc )/f (xc )) to ﬁnd a decrease in f (xc ). decrease in the size of the nonlinear residual was obtained with a full Newton step.Copyright ©1995 by the Society for Industrial and Applied Mathematics.1. 1. . 3. This electronic version is for personal use and may not be duplicated or distributed. In theory. The algorithm is to compute a search direction d which for us will be the Newton direction d = −f (xc )/f (xc ) Buy this book from SIAM at http://www. even in this terminal phase rapid convergence takes place only in the ﬁnal three outer iterations. 4. however. −3. One can see that most of the work is in the initial phase of the iteration. some authors [63] refer to this as backtracking. −6 × 10−10 . r0 = f (x) 2.1. In the terminal phase. After the fourth iterate. 2. 9 × 10−4 . However. 2.9.4.2.com/SIAM/FR16. .4. Algorithm lines1 almost always works well. Note the diﬀerences between Algorithm lines1 and Algorithm nsol. −. This is typical of the performance of the algorithms we discuss in this chapter.1.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. 8. r0 = f (x) 2.1) is called suﬃcient decrease of f .2.1. f.1) as easy as possible to satisfy. with λ = 2−j for some j ≥ 0. This electronic version is for personal use and may not be duplicated or distributed. If f (xt ) < (1 − αλ)f (x) then x = xt (accept the step) else λ = λ/2 goto 2(c)i (reject the step) This is essentially the algorithm implemented in the MATLAB code nsola. We follow the recent optimization literature [63] and set α = 10−4 . GLOBAL CONVERGENCE 0 137 10 10 10 Relative Nonlinear Residual 10 10 10 10 10 10 10 10 1 2 3 4 5 6 7 8 9 10 0 5 10 15 20 Function Evaluations 25 30 35 Fig.ecsecurehost. and then test steps of the form s = λd. something we did not have to do in the earlier chapters. xt = x + λd (trial point) ii. until f (x + s) satisﬁes (8. . nsola1(x. Once suﬃcient decrease has been obtained we accept the step s = λd.com/SIAM/FR16.1. Algorithm 8. The parameter α ∈ (0. We remark here that there is nothing critical about the reduction factor of 2 in the line search.1) f (xc + λd) < (1 − αλ)f (xc ). τ ) 1. (b) d = −f (x)/f (x) (search direction) (c) λ = 1 i.html. Newton–Armijo iteration for inverse tangent. number intended to make (8. We will analyze Algorithm nsola in § 8. Note that we must now distinguish between the step and the Newton direction. but positive. A factor of Buy this book from SIAM at http://www. Do while f (x) > τr r0 + τa (a) If f (x) = 0 terminate with failure. This strategy.2. from [3] (see also [88]) is called the Armijo rule. 1) is a small. The condition in (8.
In Exercise 8.1.2) is motivated by the Newtoniterative paradigm.3. where 0 < σ0 < σ1 < 1. [87]. [63]. or even use the chord method provided that the resulting direction dn satisﬁes (8.1 and .3. Do while F (x) > τr r0 + τa (a) Find d such that F (x)d + F (x) ≤ η F (x) If no such d can be found.1 [54]. While (8. Analysis of the Armijo rule In this section we extend the onedimensional algorithm in two ways before proceeding with the analysis. [133]. Typical values of σ0 and σ1 are . 1. in fact. reducing λ by too much can be costly as well.2).2) F (xn )dn + F (xn ) ≤ ηn F (xn ) . On the other hand. Taking full Newton steps ensures fast local convergence. is to minimize an interpolating polynomial based on the previous trial steps. Buy this book from SIAM at http://www. nsola(x. This could be quite important if the search direction were determined by an inexact Newton method or an approximation to Newton’s method such as the chord method or Broyden’s method. 138 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 10 could well be better in situations in which small values of λ are needed for several consecutive steps (such as our example with the arctan function).2.9 you are asked to implement the Armijo rule in the context of Algorithm nsol. The danger is that the minimum may be too near zero to be of much use and. we accept any direction that satisﬁes the inexact Newton condition (6. Our algorithm nsola incorporates these ideas.1). r0 = F (x) 2.5. One such method.3.ecsecurehost. This electronic version is for personal use and may not be duplicated or distributed. F. Our second extension is to allow more ﬂexibility in the reduction of λ. 8.5. the iteration may stagnate as a result. Safeguarding is an important aspect of many globally convergent methods and has been used for many years in the context of the polynomial interpolation algorithms we discuss in § 8. approximate the Newton step by a quasiNewton method.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Taking as large a fraction as possible helps move the iteration into the terminal phase in which full steps may be taken and fast convergence expected. In that event a reduction of a factor of 8 would require three passes through the loop in nsola1 but only a single reduction by a factor of 10. . η). We write this for the Armijo sequence as (8. First. developed in § 8. terminate with failure. Algorithm 8. the analysis here applies equally to methods that solve the equations for the Newton step directly (so ηn = 0). [86]. We allow for any choice that produces a reduction that satisﬁes σ0 λold ≤ λnew ≤ σ1 λold .html. τ. We will return to the issue of reduction in λ in § 8. The parameter σ0 safeguards against that.2.com/SIAM/FR16.
n=0 Assumption 8. r) = ∞ {x  x − xn ≤ r}. If F (xt ) < (1 − αλ) F (x) then x = xt else Choose σ ∈ [σ0 . Let x0 ∈ RN and α ∈ (0.1. the algorithm must fail. and F (x)−1 ≤ mf on the set Ω({xn }. Lemma 8. There are r. This electronic version is for personal use and may not be duplicated or distributed.2. by giving a uniform lower bound on λn (assuming that F (x0 ) = 0). xn → ∞. mf > 0 such that F is deﬁned.html. The algorithm can fail in some obvious ways: 1.3) ¯ {ηn } ⊂ (0. illconditioning of F might be detected as part of the factorization. The convergence theorem is the remarkable statement that if {xn } and { F (xn )−1 } are bounded (thereby eliminating premature failure and local minima of F that are not roots) and F is Lipschitz continuous in a neighborhood of the entire sequence {xn }. We begin with a formal statement of our assumption that F is uniformly Lipschitz continuous and bounded away from zero on the sequence {xn }. Assumption 8.2). Note also the very modest conditions on ηn . a local minimum of F which is not a root. xt = x + λd ii. Clearly if F has no roots.Copyright ©1995 by the Society for Industrial and Applied Mathematics. then the iteration converges to a root of F at which the standard assumptions hold. full steps are taken in the terminal phase.2.ecsecurehost.2. as do [25] and [70]. The inner iteration could terminate with a failure. Assume that {xn } is given by Algorithm nsola with (8. . If. GLOBAL CONVERGENCE 139 (b) λ = 1 i. 2. We will see that if F has no roots then either F (xn ) has a limit point which is singular or {xn } becomes unbounded.1 implies that the line search phase (step 2b in Algorithm nsola) will terminate in a number of cycles that is independent of n. F is Lipschitz continuous with Lipschitz constant γ. 1) be given. 1 − α) Buy this book from SIAM at http://www. γ. F (xn ) is singular for some n. Let {xn } be the iteration produced by Algorithm nsola with initial iterate x0 . xn → x. Note how the safeguard bound σ0 enters this result. and the convergence is qquadratic. η ] ⊂ (0. step 2a were implemented with a direct factorization method.1. σ1 ] λ = σλ goto 2(b)i Note that step 2a must allow for the possibility that F is illconditioned and that no direction can be found that satisﬁes (8. for example. We show this.com/SIAM/FR16. ¯ 3.
. ¯ Therefore. xn + λdn ] whenever ¯ λ ≤ λ1 = r/(mf F (x0 ) ).2. By the fundamental theorem of calculus and (8. for any λ ∈ [0. Now we can show directly that the sequence of Armijo iterates either is unbounded or converges to a solution.1 F is Lipschitz continuous on the line segment [xn .2. Theorem 8. which will be implied by ¯ ¯ λ ≤ λ2 = min λ1 . Proof. ¯ Lipschitz continuity of F implies that if λ ≤ λ1 then F (xn + λdn ) ≤ (1 − λ) F (xn ) + λ¯ F (xn ) + η ≤ (1 − λ) F (xn ) + λ¯ F (xn ) + η γ 2 λ dn 2 2 1 0 (F (xn + tλdn ) − F (xn ))λdn dt (F (xn + tλdn ) − F (xn ))λdn dt.html.2. This electronic version is for personal use and may not be duplicated or distributed. Let n ≥ 0. ¯ ξn ≤ ηn F (xn ) ≤ η F (xn ) . which completes the proof. mf F (x0 ) m2 F (x0 ) (1 + η )2 γ ¯ f .1 says even more. The Buy this book from SIAM at http://www.4) ¯ λn ≥ λ = σ0 min r 2(1 − α − η ) ¯ .Copyright ©1995 by the Society for Industrial and Applied Mathematics. by Assumption 8.ecsecurehost.com/SIAM/FR16. 1] F (xn + λdn ) = F (xn ) + λF (xn )dn + = (1 − λ)F (xn ) + λξn + where. 140 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS and that Assumption 8.1 holds. by the bound ηn ≤ η . Then either F (x0 ) = 0 or (8.2) we have. 1 0 ¯ m2 γ(1 + η )2 λ2 F (xn ) f 2 ¯ m2 γλ(1 + η )2 F (x0 ) f 2 2 ≤ (1 − λ + η λ) F (xn ) + λ F (xn ) ¯ < (1 − αλ) F (xn ) . ¯ This shows that λ can be no smaller than σ0 λ2 . ¯ The step acceptance condition in nsola implies that { F (xn ) } is a decreasing sequence and therefore dn = F (xn )−1 (ξn − F (xn )) ≤ mf (1 + η ) F (xn ) . 2(1 − α − η ) ¯ (1 + η )2 m2 γ F (x0 ) ¯ f .
1 implies that the standard assumptions hold at x∗ = xn . All that one needs to make the theorems and proofs in § 8. and that Assumption 8.html.2. The boundedness of the sequence {xn } implies that a subsequence {xnk } converges. {xn } is bounded. Theorem 6. the entire sequence. In Exercise 8. the Newton iteration with x0 = x will remain in B(δ) and converge qquadratically to x∗ . Let x0 ∈ RN and α ∈ (0. Assume that {xn } is given by Algorithm nsola. GLOBAL CONVERGENCE 141 sequence converges to a root at which the standard assumptions hold.9 you are asked to do this with nsol and. It could well be advantageous to use a chord or Broyden direction.1.2. full steps are taken for n suﬃciently large. Since the standard assumptions hold at x∗ . so in the terminal phase of the iteration the step lengths are all 1 and the convergence is qquadratic. This electronic version is for personal use and may not be duplicated or distributed.2 applies and hence full steps can be taken.4.1 and therefore the standard assumptions hold at x∗ . then we are done because Assumption 8. Thus far we have used for d the inexact Newton direction. Proof. Moreover.2. then the storage requirements are roughly the same as for Algorithm nsolgm if one reuses the storage location for the ﬁnite diﬀerence directional derivative to test for suﬃcient decrease. This is all that one could ask for in terms of robustness of such an algorithm.2). where r is the radius in Assumption 8. If GMRES is used as the linear solver.1 implies that F (xn ) converges qlinearly to zero with qfactor at ¯ most (1 − αλ). Then {xn } converges to a root x∗ of F at which the standard assumptions hold.2.1 holds. F (x∗ ) = 0. If F (xn ) = 0 for some n. not just the subsequence. which is easy to verify.1. numerically determine if the chord direction satisﬁes the inexact Newton condition. The iteration can be very costly if the Jacobian must be Buy this book from SIAM at http://www. Implementation of the Armijo rule The MATLAB code nsola is a modiﬁcation of nsolgm which provides for a choice of several Krylov methods for computing an inexact Newton direction and globalizes Newton’s method with the Armijo rule. 1) be given.1. Otherwise Lemma 8. with each iteration. Since F is continuous.2.3).Copyright ©1995 by the Society for Industrial and Applied Mathematics. Theorem 8.2 hold is (8. 8. for example.1 is violated) or the iteration becomes unbounded. . will remain in B(δ) and converge to x∗ . We use the l2 norm in that code and in the numerical results reported in § 8. In this case the storage requirements are roughly the same as for Algorithm nsol.ecsecurehost.com/SIAM/FR16. Eventually xnk − x∗  < r.2).5. via a forward diﬀerence approximation to F (xn )d. Hence as soon as xnk ∈ B(δ). {ηn } satisﬁes (8. in principle. and the convergence behavior in the ﬁnal phase of the iteration is that given by the local theory for inexact Newton methods (Theorem 6. there is δ such that if x ∈ B(δ). say to x∗ .3. At this stage we have shown that if the Armijo iteration fails to converge to a root either the continuity properties of F or nonsingularity of F break down as the iteration progresses (Assumption 8.2.
We can use this iteration history to model the scalar function f (λ) = F (xn + λdn ) 2 2 with a polynomial and use the minimum of that polynomial as the next steplength. In practice such methods usually perform better than constant reduction factors. say). Buy this book from SIAM at http://www. .com/SIAM/FR16. polynomial interpolation as a way to choose λ. This electronic version is for personal use and may not be duplicated or distributed. one might expect to be able to better estimate the appropriate reduction factor. If one can model the decrease in F 2 as the steplength is reduced. how Broyden’s method must be implemented. but might respond well to a more aggressive steplength reduction (by factors of 1/10. Clearly f (0) can be computed as 2 (8. we compute the minimum λt of that polynomial analytically and set σ0 λc if λt < σ0 λc . Our Broyden–Armijo code does not explicitly check the inexact Newton condition and thereby saves an evaluation of F . . The motivation for this is that some problems respond well to one or two reductions in the steplength by modest amounts (such as 1/2) and others require many such reductions. is again based on [67] and is more specialized. 142 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS evaluated and factored many times during the phase of the iteration in which the approximate solution is far from the root. After λc has been rejected and a model polynomial computed.3.html. .Copyright ©1995 by the Society for Industrial and Applied Mathematics.1.5) λ+ = σ 1 λc λt Twopoint parabolic model. The second. The reader is asked to pursue this in Exercise 8.5. In this section we consider an elaboration of the simple line search scheme of reduction of the steplength by a ﬁxed factor. f (0) = F (xn ) 2 is known. is applicable to all methods. If we have rejected k steps we have in hand the values F (xn ) 2 . 8. The ﬁrst.10. otherwise. Polynomial line searches. In this approach we use values of f and f at λ = 0 and the value of f at the current value of λ to construct a 2nd degree interpolating polynomial for f . . F (xn + λk−1 dn ) 2 .6) f (0) = 2(F (xn )T dn )T F (xn ) = 2F (xn )T (F (xn )dn ). We consider two ways of doing this that use second degree polynomials which we compute using previously computed information. We consider two topics in this section. Evaluation of F (xc )dc if the full step is not accepted (here dc is the Broyden direction) would not only verify the inexact Newton condition but provide enough information for a twopoint parabolic line search.ecsecurehost. if λt > σ1 λc . (8. F (xn + λ1 dn ) 2 .
Threepoint parabolic model. The polynomial that interpolates f at 0.com/SIAM/FR16.html.5). where c= f (λc ) − f (0) − f (0)λc .5) to compute λ+ . The polynomial p(λ) such that p. This is a possibility with directions other than inexact Newton directions. Buy this book from SIAM at http://www. λ− is p(λ) = f (0) + λ λc − λ− (λ − λ− )(f (λc ) − f (0)) (λc − λ)(f (λ− ) − f (0)) + . λ2 c Our approximation of f (0) < 0. After the second step is rejected. our approximation to f (0) should be negative. where λc and λ− are the most recently rejected values of λ. If the full step is rejected. GLOBAL CONVERGENCE 143 Use of (8.1. If p (0) = 2 (λ− (f (λc ) − f (0)) − λc (f (λ− ) − f (0))) λc λ− (λc − λ− ) is positive. we implement this method with σ0 = . we have the values f (0).6) requires evaluation of F (xn )dn . we set λ = σ1 and try again. In the MATLAB code nsola from the collection. which uses f (0) and the two most recently rejected steps to create the parabola. An alternative to the twopoint model that avoids the need to approximate f (0) is a threepoint model. σ1 = . We take the latter approach and set λ+ = σ1 λc . λc λ− We must consider two situations. σ1 λ] or reject the parabolic model and simply set λ+ to σ0 λc or σ1 λc . λc . In any case. then we set λt to the minimum of p λt = −p (0)/p (0) and apply the safeguarding step (8. p agree with f. then c > 0 and p(λ) has a minimum at λt = −f (0)/(2c) > 0. If it is not. . This electronic version is for personal use and may not be duplicated or distributed. We then compute λ+ with (8.ecsecurehost. such as the Broyden direction. and f (λ− ). In this approach one evaluates f (0) and f (1) as before. so if f (λc ) > f (0). which can be obtained by examination of the ﬁnal residual from GMRES or by expending an additional function evaluation to compute F (xn )dn with a forward diﬀerence. f at 0 and p(λc ) = f (λc ) is p(λ) = f (0) + f (0)λ + cλ2 .Copyright ©1995 by the Society for Industrial and Applied Mathematics.5. If p (0) ≤ 0 one could either set λ+ to be the minimum of p on the interval [σ0 λ. it may be necessary to compute a new search direction. The MATLAB code parab3p implements the threepoint parabolic line search and is called by the two codes nsola and brsola. f (λc ).
. The diﬀerence from the development in § 7.html. This electronic version is for personal use and may not be duplicated or distributed.9) dn+1 = − −1 −1 sn 2 Bn F (xn+1 ) − (1 − λn )sT Bn F (xn+1 )sn n 2 −1 sn 2 + λn sT Bn F (xn+1 ) n 2 −1 We then use the second equality in (8. In Chapter 7 we implemented Broyden’s method at a storage cost of one vector for each iteration.3.44) (8.7) and (7. we set un = we can use (8. Clearly. sn 2 sn 2 If.com/SIAM/FR16.2.8) reduces to (7.3 is that the simple relation between the sequence {wn } and {vn } is changed. Buy this book from SIAM at http://www.8) and the relation dn+1 = − I − wn sT n sn 2 −1 Bn F (xn+1 ) to solve for dn+1 and obtain an analog of (7. We can use the ﬁrst equality in (8. using the notation in § 7.7) yn − Bn sn = F (xn+1 ) − (1 − λn )F (xn ). [86]. vn = .9) is left to Exercise 8.41) when λn = 1 and λn+1 = 1 (so dn+1 = sn+1 ). [133].8) to construct Bc F (x+ ) as we did in Algorithm brsol. is the search direction used to compute xn+1 .3. The safeguarding and Theorem 8.ecsecurehost.Copyright ©1995 by the Society for Industrial and Applied Mathematics.38) to obtain wn = λn (8. As in § 7. Veriﬁcation of (8. 144 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS Interpolation with a cubic polynomial has also been suggested [87]. The relation y − Bc s = F (x+ ) was not critical to this and we may also incorporate a line search at a cost of a bit of complexity in the implementation.8) and (8. [63].1 provide insurance that progress will be made in reducing F . If a line search is used. 8. our approach is a nonlinear version of the method in [67].5.2. then −1 sn = −λn Bn F (xn ) and hence (8.3.5. yn − Bn sn sn −1 T −1 .8) = λn where −dn+1 sn + (λ−1 − 1) n sn 2 sn 2 −sn+1 /λn+1 sn + (λ−1 − 1) n sn 2 sn −1 dn+1 = −Bn+1 F (xn+1 ) 2 . Broyden’s method. and wn = (Bn un )/(1 + vn Bn un ). Note that (8. the polynomial modeling approach is heuristic.
4.1. an approximation of f (0) can be obtained at no additional cost and a parabolic line search based on f (0). We do not include details of the line search or provide for restarts or a limited memory implementation. Algorithm 8. uses a threepoint parabolic line search and performs a restart when storage is exhausted like Algorithm brsol does. brsola(x.1 and also changes η using (6. the storage requirements of Algorithm brsola would appear to be essentially the same as those of brsol.2). So. z = −F (x) ii. As a side eﬀect. for j = 0. maxit. f (0). d = −F (x). store sn+1 2 − λn sT z) n and λn+1 . F. provided in the collection of MATLAB codes. It is possible that the Broyden direction fails to satisfy (8. i. In Exercise 8. 8. The MATLAB implementation of brsola in the collection illustrates this point. compute λ0 . there is no guarantee that the Broyden direction is a direction of decrease for F 2 . GLOBAL CONVERGENCE 145 Algorthm brsola uses these ideas. n = −1. This electronic version is for personal use and may not be duplicated or distributed. Do while n ≤ maxit (a) n = n + 1 (b) x = x + sn (c) Evaluate F (x) (d) If F (x) (e) 2 ≤ τr r0 + τa exit. Another approach would be to compute F (xc )d numerically if a full step is rejected and then test (8. nmax) 1. Since dn+1 and sn+1 can occupy the same location.5. In fact. and f (1) can be used.ecsecurehost. s0 = λ0 d 2.Copyright ©1995 by the Society for Industrial and Applied Mathematics.3. Examples for Newton–Armijo The MATLAB code nsola is a modiﬁcation of nsolgm that incorporates the threepoint parabolic line search from § 8.html. Initialize: Fc = F (x) r0 = F (x) 2 . the storage of the globalized methods exceeds that of the local method by one vector.2) before beginning the line search. computation of the step length requires storage of both the current point x and the trial point x + λs before a step can be accepted. b = 1 − λj z = z + (asj+1 + bsj )sT z/ sj j 2 2 2 2 (f) d = ( sn 2 z + (1 − λn )sT zsn )/( sn n 2 (g) Compute λn+1 with a line search.10 the reader is asked to fully develop and implement these ideas. The MATLAB implementation. The MATLAB code returns with an error message if more than a ﬁxed number of reductions in the step length are needed. However. (h) Set sn+1 = λn+1 d.3. n − 1 a = −λj /λj+1 .20) Buy this book from SIAM at http://www. .com/SIAM/FR16. τ.
In Fig.4.2. 1) that we considered before. we can use the twopoint parabolic line search. We set τr = τa = h2 /10. that if the initial iterate is far from the solution. 8. the parabolic line search takes only one reduction in the next three iterations and none thereafter. is needed to obtain an accurate solution.2 and. Inverse tangent function.4. We use x0 = 10 as the initial iterate with τr = τa = 10−8 . 8.4.21) from § 6. with the constant η method as we did in § 6. We compare the twopoint parabolic line search with the constant reduction search (σ0 = σ1 = . 31 × 31 mesh and homogeneous Dirichlet boundary conditions on the unit square (0. both line searches take three steplength reductions.4 with γ = . an inexact Newton method such as NewtonGMRES can succeed in many case in which a quasiNewton method can fail because the quasiNewton direction may not be an inexact Newton direction.4.ecsecurehost. In Fig. We base our absolute error criterion on the norm · 2 / N as we did in § 6.9.2 we plot the iteration history corresponding to the parabolic line search with the solid line and that for the constant reduction with the dashed line. 1) × (0. 8. The parabolic line search required 7 outer iterates and 21 function evaluations in contrast with the constant reduction searches 11 outer iterates and 33 function evaluations. 146 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS once the iteration is near the root. and will repeat the caution later. However. Convectiondiﬀusion equation.html.1. however. Here. We compare this strategy. We consider a much more diﬃ −∇2 u + Cu(ux + uy ) = f with the same right hand side f .4. .20) with the dashed Buy this book from SIAM at http://www. may well be most eﬃcient. Since we have easy access to analytic derivatives in this example. This electronic version is for personal use and may not be duplicated or distributed. In the ﬁrst outer iteration.25 with the solid line and using the sequence given by (6.2.4. For this problem we plot the progress of the iteration using η = .4. we set C = 100. This is a tighter tolerance that in § 6. 8. We caution the reader now. which requires a single function evaluation for each outer iteration when full steps can be taken. in all the ﬁgures we plot F (xn ) 2 / F (x0 ) 2 against the number of function evaluations required by all line searches and inner and outer iterations to compute xn . However. because of the large value of C and resulting illconditioning. when both methods succeed.3 we show the progress of the iteration for the unpreconditioned equation. This makes a globalization strategy critical (Try it without one!). using GMRES as the linear solver.2 and 7.3 we compare the Broyden–Armijo method with NewtonGMRES for those problems for which both methods were successful.Copyright ©1995 by the Society for Industrial and Applied Mathematics. cult problem. the quasiNewton method. The constant reduction line search took three reductions in the ﬁrst two outer iterations and two each in following two. In § 8.5) for the arctan function. We take (6. Counts of function evaluations corresponding to outer iterations are√ indicated by circles. initial iterate u0 = 0.com/SIAM/FR16. As before.
9 and ηmax = .3.Copyright ©1995 by the Society for Industrial and Applied Mathematics. We set the parameters in (6. Newton–Armijo for the PDE.html. 759 (constant η) and 744 (varying η) function evaluations. 8. GLOBAL CONVERGENCE 0 147 10 10 Relative Nonlinear Residual 2 10 4 10 6 10 8 10 10 10 12 0 5 10 15 20 Function Evaluations 25 30 35 Fig.4. The iterations required 75.com/SIAM/FR16. 10 0 10 Relative Nonlinear Residual 1 10 2 10 3 10 4 10 5 0 100 200 300 400 500 Function Evaluations 600 700 800 Fig. Newton–Armijo for the arctan function. We consider the preconditioned problem in Fig.3 (constant η) and 75. .4 (varying η) million ﬂoating point operations.ecsecurehost. The two approaches to selection of the forcing term performed very similarly. This electronic version is for personal use and may not be duplicated or distributed. In the computation we Buy this book from SIAM at http://www. C = 100.20) to γ = . and 25 (constant η) and 22 (varying η) outer iterations.2.25. line. 8. 8. This problem is very poorly conditioned and much tighter control on the inner iteration is needed than for the other problems.
20) with the dashed line.4.3 million ﬂoatingpoint operations.2 with C = 20.25 with the solid line and using the sequence given by (6. terminating in 4 outer iterations. The constant η iteration terminated after 79 function evaluations.2. 16 function evaluations. Broyden–Armijo. C = 100. The line search in this case reduced the step length three times on the ﬁrst iteration and twice on the second and third. The iteration in which {ηn } is given by (6.9 and ηmax = . Newton–Armijo for the PDE. For this problem we plot the progress of the iteration using η = . and 12. 148 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS preconditioned (6.4 we show the progress of the iteration for the preconditioned equation.2. 8. required at most 16 inner iterations while the constant η approach needed at most 10. and roughly 13. 9 outer iterations. . We set the parameters in (6. In these cases the Jacobian is not well approximated by a low rank perturbation of the identity [112] and the ability of the inexact Newton iteration to ﬁnd a good search direction was the important factor. the Broyden–Armijo costs reﬂect Buy this book from SIAM at http://www.20) terminated after 70 function evaluations.4.20). 9 outer iterations. and 2.4.4.com/SIAM/FR16. This electronic version is for personal use and may not be duplicated or distributed.5 million ﬂoatingpoint operations.html.20) to γ = . The line search in the nonconstant η case reduced the step length three times on the ﬁrst iteration and once on the second.Copyright ©1995 by the Society for Industrial and Applied Mathematics. the NewtonGMRES iteration always took full steps.99. with the forcing term given by (6.3. The most eﬃcient iteration.ecsecurehost. 8. In Fig.21) with the fast Poisson solver fish2d. 8. when increases in the residual were allowed. 10 0 10 Relative Nonlinear Residual 1 10 2 10 3 10 4 10 5 0 10 20 30 40 50 Function Evaluations 60 70 80 Fig. We begin by revisiting the example in § 6. When compared to the Broyden’s method results in § 6. We found that the Broyden–Armijo line search failed on all the unpreconditioned convection diﬀusion equation examples. As reported in § 6.4.6 million ﬂoatingpoint operations.
This electronic version is for personal use and may not be duplicated or distributed. and γ = .2 took 12 iterations and 2.8 million ﬂoatingpoint operations while the Broyden–Armijo took 9 iterations and required 2 million ﬂoatingpoint operations. 42 outer iterates. From these examples we see that Broyden–Armijo can be very eﬃcient provided only a small number of iterations are needed. C = 20. and 123 function evaluations. 8. The restarted Broyden–Armijo iteration took 19. the Broyden–Armijo iteration required storage for 37 vectors while the NewtonGMRES iteration needed at most 16 inner iterations.9.com/SIAM/FR16. The steplength was reduced on the second iteration (twice) and the third (once).4. 8. 10 0 Relative Nonlinear Residual 10 1 10 2 10 3 0 2 4 6 8 10 Function Evaluations 12 14 16 Fig. The Broyden–Armijo iteration required 34 nonlinear iterations. For the more diﬃcult problem with C = 100. Broyden’s method suffers under these conditions as one can see from the comparison of Broyden– Armijo (solid line) and Newton–GMRES–Armijo (dashed line) in Fig.5 compares the Broyden–Armijo iteration (solid line) to the GMRES iteration (dashed line).5. and γ = . The Broyden iteration in § 6. . ηmax = .9 as we did in § 6. needing to store 21 vectors. 8. Figure 8. the performance of the two methods is more similar.5. and 85 function evaluations.6 we compare our implementation of Newton–GMRES–Armijo (dashed line) to Broyden–Armijo (solid line).99.5 million ﬂoatingpoint operations. roughly 16 million ﬂoatingpoint operations.4. In Fig.ecsecurehost. In our implementation of brsola the threepoint parabolic line search was used.html.7.2. We set τa = τr = h2 /10. and therefore was much more eﬃcient in terms of storage. NewtonGMRES and Broyden–Armijo for the PDE. Restarting the Broyden iteration every 19 iterations would equalize the storage requirements with Newton–GMRES–Armijo. The storage requirements increase as the nonlinear iteration progresses and this fact puts Broyden’s Buy this book from SIAM at http://www. In terms of storage. GLOBAL CONVERGENCE 149 an improvement in eﬃciency.Copyright ©1995 by the Society for Industrial and Applied Mathematics. We set τa = τr = h2 . ηmax = .
Copyright ©1995 by the Society for Industrial and Applied Mathematics. method at a disadvantage to NewtonGMRES when the number of nonlinear iterations is large.com/SIAM/FR16. This electronic version is for personal use and may not be duplicated or distributed. 150 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 0 10 10 Relative Nonlinear Residual 1 10 2 10 3 10 4 10 5 0 10 20 30 40 50 60 Function Evaluations 70 80 90 Fig.ecsecurehost.html. NewtonGMRES and restarted Broyden–Armijo for the PDE. Buy this book from SIAM at http://www.6. NewtonGMRES and Broyden–Armijo for the PDE. 8. C = 100. . 10 0 10 Relative Nonlinear Residual 1 10 2 10 3 10 4 10 5 0 20 40 60 80 Function Evaluations 100 120 140 Fig. 8.7. C = 100.
.8) and (8. Show that if F is Lipschitz continuous and the iteration {xn } produced by Algorithm nsola converges to x∗ with F (x∗ ) = 0. You might use the MATLAB code nsola to do this.com/SIAM/FR16. How does the method proposed in [10] diﬀer from the one implemented in nsola? What could be advantages and disadvantages of that approach? 8. 1. Buy this book from SIAM at http://www.ecsecurehost. What could you do if the Broyden direction is not a descent direction? Apply your code to the examples in § 8. to use the Armijo rule.html. If the monthly payments are $713. Vary the convection coeﬃcient in the convectiondiﬀusion equation and the mesh size and report the results. f (x) = arctan(x2 ) 3. Modify brsola to test the Broyden direction for the descent property and use a twopoint parabolic line search. f (x) = 1 + x2 8.5.40. 8. f (x) = . In what ways are the results in [25] and [70] more general than those in this section? What ideas do these papers share with the analysis in this section and with [3] and [88]? 8.3. 8.4.7.1.8. Apply your code to the following functions with x0 = 10. Duplicate the results in § 8.10. Implement the Newton–Armijo method for single equations.4.000 down and takes a 7 year installment loan to pay the balance. 8. This electronic version is for personal use and may not be duplicated or distributed. 8.2. f (x) = ex 6. f (x) = x(1 + sin(1/x)) 5. Try to do it in such a way that the chord direction is used whenever possible. He puts $10.5. f (x) = arctan(x) (i.6.5. Explain your results. GLOBAL CONVERGENCE 151 8.) 2.9. BiCGSTAB. and TFQMR. . Experiment with other linear solvers such as GMRES(m). Verify (8.5.4. Exercises on global convergence 8.2. then F (x∗ ) is singular. f (x) = 2 + sin(x) 7. A numerical analyst buys a German sports car for $50.e.1. what is the interest rate? Assume monthly compounding. Use nsola to duplicate the results in § 8.5.5.5.5. This is interesting in the locally convergent case as well. the hybrid Newton algorithm from Chapter 5. 8. 8.000.9 + arctan(x) 4.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Modify nsol.5.9).5.5.5.
13.1 is applicable? 8.5.com/SIAM/FR16. Under what conditions will the iteration given by nsola converge to a root x∗ that is independent of the initial iterate? Buy this book from SIAM at http://www. . when does it? How would you implement a secantArmijo method in such a way that the convergence theorem 8.5. Modify nsola to use a cubic polynomial and constant reduction line searches instead of the quadratic polynomial line search. 8.2.html.12. This electronic version is for personal use and may not be duplicated or distributed. 152 ITERATIVE METHODS FOR LINEAR AND NONLINEAR EQUATIONS 8. Does the secant method for equations in one variable always give a direction that satisﬁes (8.2) with ηn bounded away from 1? If not.4.11.Copyright ©1995 by the Society for Industrial and Applied Mathematics.5. Compare the results with the examples in § 8.ecsecurehost.
S. [6] K. SpringerVerlag.. E. V. Arnoldi. F. 1993. H. R. Uber die numerische Behandlung von Integralgleichungen nach der Quadraturformelmethode. Numer. 2nd. 13 (1992). T. Fund. [2] .. [16] K. Bibliography [1] E. New York. and L. pp. Comput. New York. Cambridge University Press. SIAM J. 1994. Georg. SIAM Rev. Numerical Solution 153 Buy this book from SIAM at http://www. Templates for the Solution of Linear Systems: Building Blocks for Iterative Methods. Numer. 183–196. 28–85. Methods. pp. pp. pp. pp. Romine. Barlett. pp. Chan. [7] . An inverse matrix adjustment arising in discriminant analysis.. e e [10] R. Global approximate Newton methods. Quart.. Simplicial and continuation methods for approximating ﬁxed points and solutions to systems of equations. 1–3. E. Bank and D. Petzold. 14 (1979). Math. pp. 22 (1980). F. J. Dongarra. and J. M.. 22 (1951). ¨ [15] H. Otto. Minimization of functions having Lipschitzcontinuous ﬁrst partial derivatives. Paciﬁc J. 133–181. This electronic version is for personal use and may not be duplicated or distributed. J. Appl.html. Armijo. Pozo.Copyright ©1995 by the Society for Industrial and Applied Mathematics. A comparison of adaptive Chebyshev and least squares polynomial preconditioning for Hermetian positive deﬁnite linear systems. pp. Brakhage. Barrett. Atkinson. John Wiley.. E. J. Bicanic and K. A. ´ ´ [13] N. Donato. J. Philadelphia. R. . [11] M. S. and H. Math. p. Bosma and W. 2 (1960). Allgower and K. pp. [9] S. 17–31. 17–29.. B. pp. The principle of minimized iterations in the solution of the matrix eigenvalue problem. [14] P. Society for Industrial and Applied Mathematics. Berry. Iterative Solution Methods. E. Astron. Sci.. S. 9 (1951). 1990. 126 (1983). Iterative variants of the Nystr¨m method for the numerical o solution of integral equations. Engrg. J. Statist. Sur les op´rations dans les ensembles abstraits et leur applications e aux ´quations int´grales. DeRooij. Math. van der Vorst. Rose. PA. Brenan. Axelsson. 1989. 16 (1966). [5] S. 148–152. Math. Numer. [8] O. L. Internat. Johnson. C... Cambridge. 279–295. A. Eijkhout. Astrophys. 1– 29. Demmel. Numerical Continuation Methods : An Introduction. Math.ecsecurehost. [12] R. ed. Banach. [4] W. Statist. Who was ‘Raphson’?. Eﬃcient methods to calculate Chandrasekhar’s Hfunctions. Math. 22 (1973). Numer.. Ashby. Campbell. 283. Ann. Math.com/SIAM/FR16. T. An Introduction to Numerical Analysis. 107–111. [3] L. Manteuffel. L. 37 (1981). 3 (1922).
Weimar. Nocedal. 1960. H. pp. A Multigrid Tutorial. Xue. 129–156. 1038–1051. SIAM J. E. and M. 1467–1488. [27] . Report CRSCTR9410. SIAM. [22] P. Brown. pp. SIAM J. 11 (1990). N. R.. 1975. C. Sci. Campbell. Dennis. BIT. Math. Englewood Cliﬀs. Busbridge. Meyer. D. Buy this book from SIAM at http://www. Byrne. 407–434. pp. F. C. pp. A class of methods for solving nonlinear simultaneous equations. Applic. J. Numer. Using Krylov methods in the solution of largescale diﬀerentialalgebraic systems. in Proceedings of the International Workshop on the NavierStokes Equations. Journal of Integral Equations and Applications. . C. Some eﬃcient algorithms for solving systems of nonlinear equations. Braunschweig. 40–91. NewtonKrylovSchwarz methods in CFD. [34] S. in Proceedings of Internationaler Kongreß uber Anwendung der Mathematik in ¨ dem Ingenieurwissenschaften. North Carolina State University. Campbell. 297–330.. Society for Industrial and Applied Mathematics. March 1995. [35] R. Burmeister. Philadelphia. Saad.com/SIAM/FR16. [29] W. 1987. Byrd. pp. Kelley. SIAM J. pp. Statist. Tidriri. Philadelphia.ecsecurehost. Petzold. Center for Research in Scientiﬁc Computation. Cavanaugh. Math. J. and A. A local convergence theory for combined inexact–Newton/ ﬁnite– diﬀerence projection methods. W. Comput. The Mathematics of Radiative Transfer. and C.Copyright ©1995 by the Society for Industrial and Applied Mathematics. T. pp. Reduced storage matrix methods in stiﬀ ODE systems. F. N. [24] P. 50. Hindmarsh. 12 (1973). [18] . Cambridge. R. N. B. Q. 1970. D. More. Optim. 31 (1989). Convergence theory of nonlinear NewtonKrylov algorithms. N. Briggs. Algorithms for Minimization Without Deriviatives. [31] R. to appear. C. D.. Brown. Zur Konvergenz einiger verfahren der konjugierten Richtungen. Maths. 1996. [25] . Center for Research in Scientiﬁc Computation. Press.. I. 10 (1989). 14 in Classics in Applied Mathematics. Brown and Y. to appear. Brent. Brown. C. L. Comput. On the local and superlinear convergence of quasiNewton methods. [17] R. C. Broyden. Math. No. PrenticeHall. Hybrid Krylov methods for nonlinear systems of equations.. Cai. Comput. Ipsen. This electronic version is for personal use and may not be duplicated or distributed. Convergence estimates for solution of integral equations with GMRES. ed. 4 (1994). Ipsen. Tech. Sci. Math. [33] S.. University of Maryland. 63 (1994). 1994. 1973. Cambridge Tracts. Rannacher. Programming. Meyer. J. pp. Hindmarsh. Anal. N. The convergence of an algorithm for solving sparse nonlinear systems. and R. Comput. Broyden. 450–481. T.. 24 (1987). J. Keyes. ´ [28] C. 10 (1973). SIAM J. VODE: A variable coeﬃcient ode solver. D.html. PA. Vieweg Verlag. [20] P. E. 327–344. [21] P...C. Cambridge Univ. and J. G. Notes in Numerical Fluid Mechanics. A. G. pp. 19 (1965). pp. GMRES and the minimal polynomial. [19] W. Report CRSCTR9513. Schnabel. 15 (1994). L. pp. 223– 246. [23] P. 577–593. Brown and A. SIAM J. NJ. 285–294. Sci. G. Appl. PhD thesis. no. [26] C. Kelley. North Carolina State University. and Z. Comput. I. Anal. J. Representation of quasiNewton matrices and their use in limited memory methods. Hindmarsh. Diﬀerence Equations and Iterative Processes.. W. C. and L. July 1994. D. 154 BIBLIOGRAPHY of Initial Value Problems in DiﬀerentialAlgebraic Equations.. Gropp. [32] X. Statist. C. 25 (1971). Inst. Comput. Tech. [30] I. SIAM J. Numer. D.
New York. F.ecsecurehost. pp.. P. [48] J. p. TX. SIAM J. Anal. [39] . Rose. pp. [53] . Block preconditioning for the conjugate gradient method. Proceedings of the Fourth International Symposium on Domain Decomposition Methods. 13 (1974). Chandrasekhar. Convergence acceleration for Newton’s method at singular points. Decker and C. 190–212. Programming. 1969. CERFACS. eds. [47] A. ´ [38] T. H. Anal. Philadelphia. Kelley. pp.. PA. Numer. Phys. December 1994. The conjugate gradient method for linear and nonlinear operator equations. Chan. J. 187–209. Proceedings of the Third International Symposium on Domain Decomposition Methods. Academic Press. SIAM J. H. O’Leary. [41] S. 1989. Buy this book from SIAM at http://www. On the estimation of sparse Jacobian matrices. J. Handbook for Matrix Computations. 20 (1983). 19 (1982). pp. Appl. Moscow. Steihaug. pp. Math. Steihaug. PA.. 10–26. W. E. More. Golub. 219–229. Domain Decomposition Methods.. Numer. T. Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed. PA. Dembo.. Comput. pp. [56] R.com/SIAM/FR16. Society for Industrial and Applied Mathematics. 19 (1982). and G. Numer. T. Henrici. Glowinski. Broyden’s method for a class of problems having singular Jacobian at the root. Truncated Newton algorithms for largescale optimization. Numer. 37–48. Anal. Is nonnormality a serious diﬃculty ?. Domain Decomposition Methods. pp. G. SIAM J. 15 (1994). 17 (1980). P. Statist. Math. Curtis. D. F. Eisenstat. 4 (1967).. 117–119. Decker.html.. J. 1988. SIAM J. 20 (1983). A quasiminimal residual variant of the BiCGSTAB algorithm for nonsymmetric systems. [37] T. pp. Meurant. G. Newton’s method at singular points I. USSR. Philadelphia. Math. W. in Sparse Matrix Computations. Tong. Coleman and J. 147–154. 309–332. Estimation of sparse Jacobian matrices and graph coloring problems.Copyright ©1995 by the Society for Industrial and Applied Mathematics. [43] T.. [44] P. eds. pp. Houston. [45] P. 1988. Math. Dover. pp. Society for Industrial and Applied Mathematics. Powell. 6 (1985). SIAM J. [50] D. J. J. Anal. Szeto. Anal. 64– 73. 26 (1983). and C. 1990. 1960. and D. and T. Reid. S. W. [54] T. Numer.. 296–314. Dekker.. J. Bunch and D. 566–574. Radiative Transfer. Philadelphia. 4. Periaux. Inst. and C.. [51] . 338. ChaitinChatelin..Proceedings of the Second International Symposium on Domain Decomposition Methods. . J. K. pp. Anal. Keller. Sci. Gallopoulos. 220–252. V. 42 (1983). [55] R. Domain Decomposition Methods. Craig. Kelley. R. 1989. Concus. Golub. in Constructive Aspects of the Fundamental Theorem of Algebra. BIBLIOGRAPHY 155 [36] F.. Finding a zero by means of successive linear interpolation. pp. B. Sci. eds. and J. R. [49] D. [52] . 400–408. 22 (1985). Comput. SIAM J. pp. PA. Inexact Newton methods. A generalized conjugate gradient method for the numerical solution of elliptic partial diﬀerential equations. Sublinear convergence of the chord method at singular points. SIAM J. and O. Coleman and C. Numer. SIAM J.. Frontiers in Appl. J. 1976. T. 1991. H.. 1990. Philadelphia.. 34 (1955). Chan. SIAM J. R. ed. ´ [42] T. M.. No. The Nstep iteration procedures. Los Angeles. Tech... Numer. pp. [46] E. January 14–16. Anal. VanLoan. Simoncini. Daniel. Widlund. Numer. Society for Industrial and Applied Mathematics. Concus. J. 66–70. CA. Convergence rates for Newton’s method at singular points. Math. Dembo and T. Report TR/PA/94/18. eds. [40] .
[64] J. Local convergence analysis of tensor methods for nonlinear equations. Triangular decomposition methods for solving reducible nonlinear systems of equations. Leastchange sparse secant updates with inaccurate secant conditions. SIAM Rev. Some devices for the solution of large sets of simultaneous linear equations (with an appendix on the reciprocation of partitioned matrices).. 493–507. pp. Freund. SIAM J. Dennis. [68] W. [66] . E. Programming. SIAM J. Saylor.. Engelman. E. Academic Press. 443–459. in Mathematical Programming Study 22: Mathematical programming at Oberwolfach II. [70] . The application of quasiNewton methods in ﬂuid mechanics. SIAM Rev. Anal.html. North–Holland. Least change secant updates for quasiNewton methods. no. Numerical Methods for Unconstrained Optimization and Nonlinear Equations.. . G.. and A. 21 (1979). Numer. Dennis. [65] . Amsterdam. pp. A characterization of superlinear convergence and its application to quasiNewton methods. Anal. A. C. 660–670. Manteuffel. 393–422. J. Anal. pp. PhD thesis. 156 BIBLIOGRAPHY [57] J. Walter. 427–459. J. Iterative Methods for Large. Yale University. Numer. Zhang. E. pp. Comput. Dennis and H. Y. E. [63] . New Haven. Inaccuracy in quasiNewton methods: Local improvement theorems. Buy this book from SIAM at http://www. 21 (1984). Philadelphia. L.. B. Walker. E. 358–382. 18 (1981). 352– 362. Statist. and R. pp.. ed. B. Fast secant methods for the iterative solution of large nonsymmetric linear systems. Optim. Comput. Strang.ecsecurehost. 17 (1996). CT. A hybrid ChebyshevKrylov subspace algorithm for solving nonsymmetric systems of linear equations. This electronic version is for personal use and may not be duplicated or distributed. 28 (1974). in Numerical Analysis Dundee 1975. 70–85. Toward a uniﬁed convergence theory for Newtonlike methods. Sci. F. The London. Numer. and K. [72] H. [69] S. Feng. Frank. Numer. pp. pp. Walker. 1976. 1971. 6 (1969). 4 (1994)... 17 (1981). Dennis and R. pp. ed. Methods Engrg. pp. Internat. P.. SIAM J. Dennis and J. J. Sci. More.. W. Comput. 425–472. 7 (1986). J. pp. 244–276. Faber and T. and X. pp. 19 (1977).. pp. SIAM. 949–987. D. [74] V. [62] J.Copyright ©1995 by the Society for Industrial and Applied Mathematics.com/SIAM/FR16. Math. [61] . SIAM J. Fletcher. E. Martinez. 22 (1985). [67] P. 16 in Classics in Applied Mathematics.. Impact of Computing in Science and Engineering. Necessary and suﬃcient conditions for the existence of a conjugate gradient method. 1984. Edinburgh. R. SIAM J. pp. Anal. Convergence theorems for least change secant update methods. 549–560. 2 (1990). Math. and Dublin Philosophical Magazine and Journal of Science. QuasiNewton methods.. Schnabel. 1996. 1982. 73–89. Eisenstat and H. [59] J. Schnabel. Berlin. G. Bathe. 46–89. ´ [60] J. J. [75] D. On the Kantorovich hypothesis for Newton’s method. [71] H. New York. 4 (1994). Globally convergent inexact Newton methods. SpringerVerlag. F. SIAM J. C. [76] R. [58] . pp. 16–32. 35 (1944). 840–855. motivation and theory. pp. Numer. Watson. pp. Deuflhard. 62 (1993). pp. Nonsymmetric Systems of Linear Equations. Sparse. SIAM J. pp. [73] M. Elman. Duncan. 707–718.. Seventh Series. Optim. in Nonlinear Functional Analysis and Applications. 760–778. New York. Choosing the forcing terms in an inexact Newton method. Elman. M. B. methods. SIAM J. and P. Conjugate gradient methods for indeﬁnite systems. Saad. C. Rall.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
BIBLIOGRAPHY
157
[77] R. W. Freund, A transposefree quasiminimal residual algorithm for nonHermitian linear systems, SIAM J. Sci. Comput., 14 (1993), pp. 470–482. [78] R. W. Freund, G. H. Golub, and N. M. Nachtigal, Iterative solution of linear systems, Acta Numerica, 1 (1991), pp. 57–100. [79] R. W. Freund, M. H. Gutknecht, and N. M. Nachtigal, An implementation of the lookahead Lanczos algorithm for nonHermitian matrices, SIAM J. Sci. Comput., 14 (1993), pp. 137–158. [80] R. W. Freund and N. M. Nachtigal, QMR: a quasiminimal residual algorithm for nonhermitian linear systems, Numer. Math., 60 (1991), pp. 315–339. [81] , An implementation of the QMR method based on coupled twoterm recurrences, SIAM J. Sci. Comput., 15 (1994), pp. 313–337. [82] D. M. Gay, Some convergence properties of Broyden’s method, SIAM J. Numer. Anal., 16 (1979), pp. 623–630. [83] C. W. Gear, Numerical Initial Value Problems in Ordinary Diﬀerential Equations, PrenticeHall, Englewood Cliﬀs, NJ, 1971. [84] R. R. Gerber and F. T. Luk, A generalized Broyden’s method for solving simultaneous linear equations, SIAM J. Numer. Anal., 18 (1981), pp. 882–890. [85] P. E. Gill and W. Murray, QuasiNewton methods for unconstrained optimization, J. I. M. A., 9 (1972), pp. 91–108. [86] , Safeguarded steplength algorithms for optimization using descent methods, Tech. Report NAC 37, National Physical Laboratory Report, Teddington, England, 1974. [87] P. E. Gill, W. Murray, and M. H. Wright, Practical Optimization, Academic Press, London, 1981. [88] A. A. Goldstein, Constructive Real Analysis, Harper and Row, New York, 1967. [89] G. H. Golub and C. G. VanLoan, Matrix Computations, Johns Hopkins University Press, Baltimore, 1983. [90] A. Griewank, Analysis and modiﬁcation of Newton’s method at singularities, PhD thesis, Australian National University, 1981. [91] , Rates of convergence for secant methods on nonlinear problems in Hilbert space, in Numerical Analysis, Proceedings Guanajuato , Mexico 1984, Lecture Notes in Math., No, 1230, J. P. Hennart, ed., SpringerVerlag, Heidelberg, 1986, pp. 138– 157. [92] , The solution of boundary value problems by Broyden based secant methods, in Computational Techniques and Applications: CTAC 85, Proceedings of CTAC, Melbourne, August 1985, J. Noye and R. May, eds., North Holland, Amsterdam, 1986, pp. 309–321. [93] , On the iterative solution of diﬀerential and integral equations using secant updating techniques, in The State of the Art in Numerical Analysis, A. Iserles and M. J. D. Powell, eds., Clarendon Press, Oxford, 1987, pp. 299–324. [94] A. Griewank and G. F. Corliss, eds., Automatic Diﬀerentiation of Algorithms: Theory, Implementation, and Application, Society for Industrial and Applied Mathematics, Philadelphia, PA, 1991. [95] A. Griewank and P. L. Toint, Local convergence analysis for partitioned quasinewton updates, Numer. Math., 39 (1982), pp. 429–448. [96] , Partitioned variable metric methods for large sparse optimization problems, Numer. Math., 39 (1982), pp. 119–137. [97] W. A. Gruver and E. Sachs, Algorithmic Methods In Optimal Control, Pitman, London, 1980.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
158
BIBLIOGRAPHY
[98] M. H. Gutknecht, Variants of BICGSTAB for matrices with complex spectrum, SIAM J. Sci. Comput., 14 (1993), pp. 1020–1033. [99] W. Hackbusch, MultiGrid Methods and Applications, vol. 4 of Springer Series in Computational Mathematics, SpringerVerlag, New York, 1985. [100] , Multigrid methods of the second kind, in Multigrid Methods for Integral and Diﬀerential Equations, Oxford University Press, Oxford, 1985. [101] W. E. Hart and S. O. W. Soul, QuasiNewton methods for discretized nonlinear boundary problems, Journal of the Institute of Applied Mathematics, 11 (1973), pp. 351–359. [102] H. V. Henderson and S. R. Searle, On deriving the inverse of a sum of matrices, SIAM Rev., 23 (1981), pp. 53–60. [103] M. R. Hestenes and E. Steifel, Methods of conjugate gradient for solving linear systems, J. of Res. Nat. Bureau Standards, 49 (1952), pp. 409–436. [104] D. M. Hwang and C. T. Kelley, Convergence of Broyden’s method in Banach spaces, SIAM J. Optim., 2 (1992), pp. 505–532. [105] E. Isaacson and H. B. Keller, Analysis of numerical methods, John Wiley, New York, 1966. [106] L. Kantorovich and G. Akilov, Functional Analysis, 2nd ed., Pergamon Press, New York, 1982. [107] L. V. Kantorovich, Functional analysis and applied mathematics, Uspehi Mat. Nauk., 3 (1948), pp. 89–185. translation by C. Benster as Nat. Bur. Standards Report 1509. Washington, D. C., 1952. [108] H. B. Keller, Newton’s method under mild diﬀerentiability conditions, J. Comput. Sys. Sci, 4 (1970), pp. 15–28. [109] , Lectures on Numerical Methods in Bifurcation Theory, Tata Institute of Fundamental Research, Lectures on Mathematics and Physics, SpringerVerlag, New York, 1987. [110] C. T. Kelley, Solution of the Chandrasekhar Hequation by Newton’s method, J. Math. Phys., 21 (1980), pp. 1625–1628. [111] , A fast multilevel algorithm for integral equations, SIAM J. Numer. Anal., 32 (1995), pp. 501–513. [112] C. T. Kelley and E. W. Sachs, A quasiNewton method for elliptic boundary value problems, SIAM J. Numer. Anal., 24 (1987), pp. 516–531. [113] , A pointwise quasiNewton method for unconstrained optimal control problems, Numer. Math., 55 (1989), pp. 159–176. [114] , Fast algorithms for compact ﬁxed point problems with inexact function evaluations, SIAM J. Sci. Statist. Comput., 12 (1991), pp. 725–742. [115] , A new proof of superlinear convergence for Broyden’s method in Hilbert space, SIAM J. Optim., 1 (1991), pp. 146–150. [116] , Pointwise Broyden methods, SIAM J. Optim., 3 (1993), pp. 423–441. [117] , Multilevel algorithms for constrained compact ﬁxed point problems, SIAM J. Sci. Comput., 15 (1994), pp. 645–667. [118] C. T. Kelley and R. Suresh, A new acceleration method for Newton’s method at singular points, SIAM J. Numer. Anal., 20 (1983), pp. 1001–1009. [119] C. T. Kelley and Z. Q. Xue, Inexact Newton methods for singular problems, Optimization Methods and Software, 2 (1993), pp. 249–267. [120] , GMRES and integral operators, SIAM J. Sci. Comput., 17 (1996), pp. 217– 226. [121] T. Kerkhoven and Y. Saad, On acceleration methods for coupled nonlinear elliptic systems, Numerische Mathematik, 60 (1992), pp. 525–548.
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
Copyright ©1995 by the Society for Industrial and Applied Mathematics. This electronic version is for personal use and may not be duplicated or distributed.
BIBLIOGRAPHY
159
[122] C. Lanczos, Solution of linear equations by minimized iterations, J. Res. Natl. Bur. Stand., 49 (1952), pp. 33–53. [123] T. A. Manteuffel, Adaptive procedure for estimating parameters for the nonsymmetric Tchebychev iteration, Numer. Math., 31 (1978), pp. 183–208. [124] T. A. Manteuffel and S. Parter, Preconditioning and boundary conditions, SIAM J. Numer. Anal., 27 (1990), pp. 656–694. [125] E. S. Marwil, Convergence results for Schubert’s method for solving sparse nonlinear equations, SIAM J. Numer. Anal., (1979), pp. 588–604. [126] S. McCormick, Multilevel Adaptive Methods for Partial Diﬀerential Equations, Society for Industrial and Applied Mathematics, Philadelphia, PA, 1989. [127] J. A. Meijerink and H. A. van der Vorst, An iterative solution method for linear systems of which the coeﬃcient matrix is a symmetric Mmatrix, Math. Comput., 31 (1977), pp. 148–162. [128] C. D. Meyer, Matrix Analysis and Applied Linear Algebra, forthcoming. ´ [129] J. J. More, Recent developments in algorithms and software for trust region methods, in Mathematical Programming: The State of the Art, A. Bachem, M. Gr¨schel, and B. Korte, eds., SpringerVerlag, Berlin, 1983, pp. 258–287. o ´ [130] J. J. More and J. A. Trangenstein, On the global convergence of Broyden’s method, Math. Comput., 30 (1976), pp. 523–540. [131] T. E. Mott, Newton’s method and multiple roots, Amer. Math. Monthly, 64 (1957), pp. 635–638. [132] T. W. Mullikin, Some probability distributions for neutron transport in a half space, J. Appl. Probab., 5 (1968), pp. 357–374. [133] W. Murray and M. L. Overton, Steplength algorithms for minimizing a class of nondiﬀerentiable functions, Computing, 23 (1979), pp. 309–331. [134] N. M. Nachtigal, S. C. Reddy, and L. N. Trefethen, How fast are nonsymmetric matrix iterations?, SIAM J. Matrix Anal. Appl., 13 (1992), pp. 778– 795. [135] N. M. Nachtigal, L. Reichel, and L. N. Trefethen, A hybrid gmres algorithm for nonsymmetric linear systems, SIAM J. Matrix Anal. Appl., 13 (1992). [136] S. G. Nash, Preconditioning of truncated Newton methods, SIAM J. Sci. Statist. Comput., 6 (1985), pp. 599–616. [137] , Who was Raphson? (an answer). Electronic Posting to NADigest, v92n23, 1992. [138] J. L. Nazareth, Conjugate gradient methods less dependent on conjugacy, SIAM Rev., 28 (1986), pp. 501–512. [139] O. Nevanlinna, Convergence of Iterations for Linear Equations, Birkh¨user, a Basel, 1993. [140] I. Newton, The Mathematical Papers of Isaac Newton (7 volumes), D. T. Whiteside, ed., Cambridge University Press, Cambridge, 1967–1976. [141] B. Noble, Applied Linear Algebra, Prentice Hall, Englewood Cliﬀs, NJ, 1969. [142] J. Nocedal, Theory of algorithms for unconstrained optimization, Acta Numerica, 1 (1991), pp. 199–242. [143] D. P. O’Leary, Why Broyden’s nonsymmetric method terminates on linear equations, SIAM J. Optim., 4 (1995), pp. 231–235. [144] J. M. Ortega, Numerical Analysis A Second Course, Classics in Appl. Math., No. 3, Society for Industrial and Applied Mathematics, Philadelphia, PA, 1990. [145] J. M. Ortega and W. C. Rheinboldt, Iterative Solution of Nonlinear Equations in Several Variables, Academic Press, New York, 1970. [146] A. M. Ostrowski, Solution of Equations and Systems of Equations, Academic
Buy this book from SIAM at http://www.ecsecurehost.com/SIAM/FR16.html.
SIAM J. Statist. 11 (1955). Convergence of the Newton process to multiple solutions.. Comput. [159] W. The Chebyshev Polynomials. [168] E. Computer Science Department. [158] J. [160] J. 65–68. 87–114. Englewood Cliﬀs. pp. J. Petzold. Convergence properties of a class of minimization algorithms. Schultz.com/SIAM/FR16. A lookahead Lanczos algorithm for unsymmetric matrices. On Newton’s method for singular problems. University of Minnesota. [166] . 1975. Sachs. New York. 24 (1987). Original in British Library. Comput.. 105–124. and Z. A ﬂexible innerouter preconditioned GMRES algorithm. Sci. NJ. K. Liu. Mangasarian. 1–27. C.. [147] B. Saylor and D. ILUT: A dual threshold incomplete LU factorization. illconditioned equations and Newton’s method. N. [149] D. W. pp.. Math. Real Analysis. Peaceman and H. 461–469.. pp. [151] L. Soc. Reid. Indus. Picard. pp. Math.. 145–210. R. Report 9238. Rheinboldt. [165] . John Wiley. at Cambridge. New York. 35 (1986). 325–328. SIAM J. O. 44 (1985). R. pp.html. Least squares polynomials in the complex plane and their use for solving nonsymmetric linear systems. B. D. Numer. 1992. C. [154] . 9 (1961). in Nonlinear Programming 2. pp. H. Numerical Analysis of Parametrized Nonlinear Equations. Appl. eds. 1974. Gordon and Breach. [155] L. pp. [163] Y. Amsterdam.. [164] . Numer. Inverse iteration. . 71–82. North Holland. 28–41. R.. 155–169. Practical use of polynomial preconditionings for the conjugate gradient method. New York. NATO Conf. S. Analysis aequationum universalis seu ad aequationes algebraicas resolvendas methodus generalis. pp. 29 (1979). in Numerical Methods for Nonlinear Algebraic Equations. Comput. Numer. Prentice Hall. 1973. J. Saad. [156] J. 7 (1986). North Holand. eds. Comput. Macmillan. (1993). pp.ecsecurehost. pp. ser 4.. A hybrid method for nonlinear equations. 160 BIBLIOGRAPHY Press. Peters and J. SIAM J. ex nova inﬁnitarum serierum doctrina. N.. Broyden’s method in Hilbert space. Parlett. Academic Press. Sci. New York. 20 (1966). SIAM J. E. Least squares solution of sparse systems of nonlinear equations by a modiﬁed Marquardt algorithm. 1983.. [150] G. M´moire sur la th´orie des ´quations aux d´riv´es partielles et la e e e e e e m´thode des approximations successives. [169] P. [153] M. GMRES a generalized minimal residual algorithm for solving nonsymmetric linear systems. D. Raphson. July 1972. [157] G. Math. 1690. Sci. W. The Symmetric Eigenvalue Problem. [148] B. pp. Rall. Implementation of an adaptive algorithm Buy this book from SIAM at http://www. Taylor. 865–881. deducta ac demonstrata. Tech. Math. 437–445. R. 2nd ed. 15 (1978). Rachford. pp. 23–37. 1960. Smolarski. [161] T. The numerical solution of parabolic and elliptic diﬀerential equations. R. Programming. [167] Y. A. 1968. 1986. 6 (1985). R. and S. Reddien. This electronic version is for personal use and may not be duplicated or distributed. Robinson. 6 (1890). [152] E. New York. Stepleman et al. 339–360. 993–986.. pp. Parlett. in Proc. Wilkinson. L. SIAM J.. L. J. Anal. New York.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Rivlin. 1980. Powell. Amsterdam. pp. de Math.. Saad and M. Comput. Anal. Royden. [162] H. Statist. John Wiley. Experiments on GramSchmidt orthogonalization. Meyer. 1970. in Scientiﬁc Computing. SIAM Rev. M. J. London. A description of DASSL: a diﬀerential/algebraic system solver. pp. et expedita. Rice. 856–869. pp. Math.
Numer. SIAM J. 815–843.. New York. SIAM J. U. 1994. Sigmon. Saylor. [188] P. pp. 21 (1984).. Math. 409–426. BIT. Alternating direction preconditioning for nonsymmetric systems of linear equations. A hybrid ArnoldiFaber iterative method for nonsymmetric systems of linear equations. B. Schnabel and P. pp. 28 (1988). [185] E. Swarztrauber and R. SIAM J. Varga. Anal. NJ.. 1964. On sparse and symmetric matrix updating subject to a linear equation. Schroder. 31 (1977). 154/156 (1991). pp. Numer. 36–52.) [175] A. 317–365. A. [179] D. Sweet.. Numer. 163–178. Anal... Sci. Comput.. 369–385.ecsecurehost. Linear Algebrra Appl. SIAM Rev. Smolarski and P.S. [183] G. H. pp. N. Academic Press. Adjustment of an inverse matrix corresponding to changes in the elements of a given column or a given row of the original matrix (abstract). Math. Fourier analysis and the FACR algorithm for the discrete solution of Poisson’s equation on a rectangle. L. a fast Lanczostype solver for nonsymmetric linear systems. pp. ¨ ¨ [171] E. Shamanskii. Applied Mathematics Series. Numer. Ann. B. A family of trustregionbased algorithms for unconstrained minimization with strong global convergence properties. SIAM J. Software. 490–501. and R. [177] . Math. 352–364. Sci. Algorithm 541: Eﬃcient FORTRAN subprograms for the solution of elliptic partial diﬀerential equations. [180] P. National Bureau of Standards. [174] V. Sci. pp. 954–961. Math.. pp. [182] G. 213–239. [170] R. Statist. A modiﬁcation of Newton’s method. Starke and R. Toint. pp. 621. CRC Press. Ann. H. pp. A. Comput. 49 (1958). [186] P. Comput. [187] . Uber unendlich viele Algorithmen zur Auﬂosung der Gleichungen. pp..html. On Newtoniterative methods for the solution of systems of nonlinear equations. van der Vorst. Math.. pp. [172] L. pp. ACM Trans. Newton’s method with a model trust region modiﬁcation. Sonneveld. 14 (1978). SIAM J. Math. [176] J. (In Russian. 2 (1870). Comput. Starke. 8 (1987). An optimum iterative method for solving any linear system with a square matrix. 21 (1950). [184] G. 755–774. 615–646. Adjustment of an inverse matrix corresponding to a change in one element of a given matrix. 199–209. pp. Schultz. 20 (1949). Schnabel. Math. J. SIAM J. E.. Stewart. SIAM J.. SIAM J. pp. 64 (1993). D. 19 (1967). Schubert. Sherman and W. pp. 124–127. 22 (1985). 1–22. Statist. [181] D. CGS. C. Statist. Englewood Cliﬀs. The methods of cyclic reduction.. S.. Traub. 15 (1994). [178] K. 24 (1970). 13 Buy this book from SIAM at http://www. 133–138. Frank. Sorensen.Copyright ©1995 by the Society for Industrial and Applied Mathematics. Sherman. Numer. A. K. [189] P. Modiﬁcation of a quasiNewton method for nonlinear equations with sparse Jacobian. Fourth Edition. BIBLIOGRAPHY 161 for Richardson’s method.. [190] J. FL. Tensor methods for nonlinear equations. N. [173] G. Mat. Swarztrauber. 1973. Byrd. W. 27–30. R.com/SIAM/FR16. Kernel polynomials in linear algebra and their numerical applications. pp. Boca Raton. C. E. Morrison. Anal. pp.. Comput. Ann. MATLAB Primer. BiCGSTAB: A fast and smoothly converging variant to BiCG for the solution of nonsymmetric systems. Stiefel. 5 (1979). Approximate cyclic reduction for solving Poisson’s equation. . This electronic version is for personal use and may not be duplicated or distributed. Statist. 10 (1989). Zh. Comput. 47–67. 19 (1982). Anal. 19 (1977). p. Introduction to matrix computations. [191] H. Ukran. Sci. Statist. L. Prentice Hall. F.. Iterative Methods for the Solution of Equations. pp.
van der Vorst and C. Comput. Comput. Alg. Iterative Solution of Elliptic Systems and Applications to the Neutron Diﬀusion Equations of Reactor Physics. [202] L. 3 (1983). A simpler GMRES. Math. Sci. T. Walker and L. SIAM J. Ypma. [199] M. M. 327–341. 1966. 631–644. Wachspress. . Englewood Cliﬀs. 9 (1989). NJ. S. [200] D. Anal. Sci. pp. A. pp. Appl. NJ. F. 15 (1994). Math. Residual smoothing and peak/plateau behavior in Krylov subspace methods. [197] H. A. Memorandum Report 42. Walker. Vuik. 571–581. Math. Zhou and H. The superlinear convergence behaviour of GMRES. The eﬀect of rounding errors on Newtonlike methods. Numer. IMA J. [201] T. Academic Press. Journal Comput. pp. Software. S. ACM Trans. Iterative Solution of Large Linear Systems. Young. pp. Statist..com/SIAM/FR16. Residual smoothing techniques for iterative methods. 279–286. F. pp.. 48 (1993). Varga. 815–825. Walker. 1971. P. New York. F.. Woodbury.. 297–312. Morgan. [198] L. Prentice Hall. Statistical Research Group. 6 (1994). Algorithm 652: HOMPACK: A suite of codes for globally convergent homotopy algorithms. J. Princeton NJ. [194] E.html.ecsecurehost.. 109–118. 1962. pp. [192] H. 1950. Appl. pp. 281–310. This electronic version is for personal use and may not be duplicated or distributed. C. J. L. 13 (1987).Copyright ©1995 by the Society for Industrial and Applied Mathematics.. Englewood Cliﬀs. Watson. Zhou. [195] H. and A. [196] . Implementation of the GMRES method using Householder transformations. 19 (1995). Matrix Iterative Analysis. [193] R. pp. Applied Numer. Prentice Hall. Numer. Buy this book from SIAM at http://www. Billups. Lin. 162 BIBLIOGRAPHY (1992). SIAM J. Inverting modiﬁed matrices.
87 NewtonGMRES. 47 BiConjugate Gradient. 25 CGNR. 137 Arnoldi process. 74 convergence. 20 Anorm. .ecsecurehost. 37 Backtracking.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 67 CGNE. 48 algorithm. 76 Condition number. 14 minimization property. 107 Characteristic polynomial. 50 algorithm. 47 Bounded deterioration. 130 Newton–Armijo. This electronic version is for personal use and may not be duplicated or distributed. 34 Contraction mapping. 48 algorithm. 123 restarting. 130 Convergence theorem Broyden’s method linear equations. 12 Approximate inverse. Index Aconjugacy. 49 Chandrasekhar Hequation. 6. 126 Brsola algorithm. 47 Broyden sequence. 47 algorithm. 25 CGS. 67 ConvectionDiﬀusion Equation Broyden’s method. 47 algorithm. 22 ﬁnite termination. 50 ﬁnite diﬀerence. 108 solution Broyden’s method. 119 Buy this book from SIAM at http://www. 87 solution Broyden’s method. 113 convergence linear equations.html. 120 implementation. 120 Broyden’s method. 47 Biorthogonality. 6 BiCG. 123 limited memory. 110 Biconjugacy. 123 sparse. 133 Brsol algorithm. 74 algorithm. 136 Bad Broyden method. 146 NewtonGMRES. 12 use of residual polynomials.com/SIAM/FR16. 14 Conjugate Gradient Squared. 128 Newton’s method. 132 Banach Lemma. 3 Conjugate Gradient algorithm. 34 Chord method. 121 Breakdown. 47 BiCGSTAB. 77 Armijo rule. 145 163 Cauchy sequence. 118 nonlinear equations. 49 Conjugate transpose. 67 Contraction Mapping Theorem. 118.
95 Inner iteration. 5 Iteration statistics. 101 ﬁnite termination. 115 Diagonalizable matrix. 34 GMRES(m). 103 secant method. 34 Diﬀerence approximation. 55 GMRES. 142 Line search. 135. 40 Hequation.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 110 Dennis–Mor´ Condition. 3 Inexact Newton method. 83 Krylov subspace. 79 DASPK. 60 Fixed point. 41 modiﬁed. 123 chord method. 54 Buy this book from SIAM at http://www. This electronic version is for personal use and may not be duplicated or distributed. 8 convergence result. 136 Linear PDE BiCGSTAB. 33 Gram–Schmidt process. 81 inexact Newton’s method. 46 loss of orthogonality. 41 minimization property. 94 Inverse secant equation. 143 twopoint. 128 Newton’s method. 66 Forcing term. 34 Diagonalizing transformation. 95 choice of. 123 Line Search parabolic threepoint. 71 Newton–Armijo. 78 Hybrid algorithms. 107 H¨lder Continuity. 34 ﬁnite diﬀerences. 9 Jacobi preconditioning. 91 Jacobi algorithm. 60 Kantorovich Theorem. 132 Inverse Tangent Newton–Armijo. 80 Elementary Jordan block. 79 choice of h. 141 NewtonGMRES. 8 preconditioning. 40. 127 CGNR. 57 Broyden’s method. 43 Globally convergent algorithm. 37 INDEX loss of orthogonality. 80 Jacobian. 100 Inverse power method. 80 directional derivative. 87 solution Broyden’s method. 164 nonlinear equations. 40. 80 nonlinearity. 142 polynomial. 9 Givens rotation. 96. 114 e superlinear convergence. 24 Jordan block. 36 Induced matrix norm deﬁnition.html. 99 Newton’s method. 36 Limited memory. 87 NewtonGMRES. 39. 46 algorithm. 66 Fixedpoint iteration. 46 use of residual polynomials. 104 Gauss–Seidel algorithm. 45 diagonalizable matrices. 11 Left preconditioning. 91 o High order methods.ecsecurehost. 77 contraction mapping. 76. 146 Iteration matrix. 33 restarts. 67 ﬁnite diﬀerence Newton. . 82 Shamanskii method. 39.com/SIAM/FR16. 135 GMRES algorithm.
91 rorder. 67 Local improvement. 82 convergence. 6 Quasiminimization. 24. 36 Schubert algorithm.com/SIAM/FR16. 113 Quasiresidual. 24 Poisson solver.ecsecurehost. 78 convergence. 104 Picard iteration. 36. 24. 95 monotone convergence. 58 Lipschitz continuous deﬁnition. 52 Relaxation parameter. 52 deﬁnition. 101 safeguarding. 74 convergence rate. 3 Matrix splitting. 41 Residual linear. 72 Residual polynomials application to CG. 92 termination. 24. 14 application to CGNR. 7 Modiﬁed Bratu Problem.html. 105 Outer iteration. . 79 solution Broyden’s method. 14 Richardson iteration. 27 preconditioner for CG. 86 Nsola algorithm. INDEX 165 algorithm. 7 Positive deﬁnite matrix. 34 Nsol algorithm. 82 qorder. 81 inexact. 54 polynomial. 101 convergence rates.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 133 Search direction. 113 Shamanskii method. 71 ﬁnite diﬀerences. 105 Newtoniterative method. 27 for GMRES. 51 Quasiresidual norm. 36 Richardson iteration. 127 TFQMR. 33 application to TFQMR. 36 incomplete factorization. 72 relative nonlinear. 78 algorithm. 36 Richardson iteration. 27. 36. 36 Jacobi. 5 nonlinear. 72 NewtonGMRES. 23 Preconditioning conjugate gradient. 100 Matrix norm. 25 application to GMRES. 4 nonlinear as error estimate. 81 stagnation. 19 GMRES. 132 Newton direction. 9 Reorthogonalization. 11 as error estimate. 100 Over solving. 91 Secant update. 66 Right preconditioning. 52 QuasiNewton methods. 103 ﬁnite diﬀerences. 136 Newton’s method. 24. 136 Secant equation. 138 Nsolgm algorithm. 113 Secant method. 24. This electronic version is for personal use and may not be duplicated or distributed. 71 algorithm. 54 Polynomial preconditioning. 66 Poisson solver. 100 Normal matrix. 75 inexact. 11 Preconditioned CG Buy this book from SIAM at http://www.
24 Successive substitution. 72 TFQMR. 51 algorithm. 132 Sherman–Morrison–Woodbury formula. 9 preconditioning. 53 ﬁnite diﬀerence. . 35. 12 Spectral radius. 16 GMRES. 132 Simple decrease. 38 linear equations. 66 Suﬃcient decrease. 136 SOR algorithm.ecsecurehost. 75 Stationary iterative methods. 166 ﬁnite diﬀerences.html.Copyright ©1995 by the Society for Industrial and Applied Mathematics. 4 nonlinear equations. 98 INDEX Buy this book from SIAM at http://www. 124. 52 TFQMR.com/SIAM/FR16. 11 Symmetric positive deﬁnite matrix. 110 weights. 91 Sherman–Morrison formula. 52 Weighted norm. 12 Termination CG. 9 Symmetric matrix. 5 as preconditioners. 9 SPD matrix. This electronic version is for personal use and may not be duplicated or distributed. 94 Newton’s method. 137 Symmetric Gauss–Seidel algorithm. 7 Stagnation.