Semilocal Convergence Analysis for Inexact Newton Method under Weak Condition
Xiubin Xu,1Yuan Xiao,1and Tao Liu1
Academic Editor: Jen-Chih Yao
Received29 May 2012
Accepted05 Aug 2012
Published30 Aug 2012
Abstract
Under the hypothesis that the
first derivative satisfies some kind of weak Lipschitz conditions, a new
semilocal convergence theorem for inexact Newton method is presented.
Unified convergence criteria ensuring the convergence of inexact Newton method
are also established. Applications to some special cases such as the Kantorovich
type conditions and -Conditions are provided and some well-known
convergence theorems for Newton's method are obtained as corollaries.
There is a huge literature on local as well as semilocal convergence for Newton's method under various assumptions (see [1β9]). Besides, there are a lot of works on the weakness of the hypotheses made on the underlying operators, see for example [2, 3, 5β9] and references therein. In particular, Wang in [7, 8] introduced the notions of Lipschitz conditions with average, under which Kantorovich like convergence criteria and Smale's point estimate theory can be put together to be investigated.
However, Newton's method has two disadvantages. One is to evaluate involved, the other is to solve the exact solution of Newton equations:
In many applications, for example, those in Euclidean spaces, computing the exact solutions using a direct method such as Gaussian elimination can be expensive if the number of unknowns is large and may not be justified when is far from the searched solution. While using linear iterative methods to approximate the solutions of (1.3) instead of solving it exactly can reduce some of the costs of Newton's method. One of the methods is inexact Newton method which can be found in [10] and takes the following form:
where is a sequence in .
As is well known, the convergence behavior of the inexact Newton method depends on the residual controls of under the hypothesis that satisfies different conditions. Some relative results can be found in [10β24], for example.
Under the Lipschitz continuity assumption on , different residual controls were used. For example, the residual controls were adopted in [10, 12]; in [15] the affine invariant conditions were considered; while in [21] Shen has analyzed the semilocal convergence behavior in some manner such that the relative residuals satisfy
Assuming that the residuals satisfy , where is a sequence of invertible operators from to , and that satisfies the HΓΆlder condition around , Li and Shen established the local and semilocal convergence in [16, 20], respectively. Besides, the -condition was also introduced into inexact Newton method in [22] by considering residual controls (1.5) with , that is,
Smale's -theory for the inexact Newton method was established there.
In the present paper, by considering the residual controls (1.6), we will study the convergence of inexact Newton method under the assumption that has a continuous derivative in a closed ball , exists and satisfies the weak Lipschitz condition:
where is a positive number, , and is a positive integrable nondecreasing function on . We also establish the unified convergence criteria, which include Kantorovich type and Smale type convergence criteria as special cases. In particular, in the special case when , (1.4) reduces to Newton's method and our result extends the corresponding one in [7].
The paper is organized as follows. Section 2 gives some lemmas which are used in the proof of our main theorem. In Section 3, the semilocal convergence of inexact Newton method is studied under the weak Lipschitz condition (1.7). Its applications to some special cases are provided in Section 4.
2. Preliminaries
Let and be Banach spaces. Throughout this paper, are two positive numbers, is a positive integrable nondecreasing function on any involved intervals, and is an open ball in with center and radius . Let , and . Define
Obviously,
Set
Write . Then
where is such that . Furthermore, it follows that
Let
The following two lemmas describe some properties about the majorizing function and the convergence property of .
Lemma 2.1. Suppose that and is defined by (2.1). Then the function is strictly decreasing and has exact one zero on satisfying .
Proof. By (2.4) and (2.5), we know is strictly increasing on and has the values and . This implies that is strictly decreasing on . Note that and by the definition of . Thus, has exact one solution on . Since
we have . The proof is complete.
Lemma 2.2. Let be the positive solution of equation on . Suppose that and the sequence is defined by (2.9). Then
Consequently, is strictly increasing and converges to .
Proof. We prove the lemma by mathematical induction. Note that . For , assume that
Since is strictly decreasing on . Hence,
Moreover, by of Lemma 2.1. It follows that
Define a function on by
Note that , unless , and , for which we adopt the convention that and . Hence, the function is well defined and continuous on . Moreover, by (2.2) and (2.3), we have
Hence, is monotonically increasing on . This together with (2.9) and (2.14) implies that
Therefore, by mathematical induction, (2.11) holds. Consequently, is increasing, bounded, and converges to a point , which satisfies . Hence, . The proof is complete.
To prove our main result, we need two more lemmas. The first can be found in [23] and the second in [7].
Lemma 2.3. Suppose that has a continuous derivative satisfying the weak Lipschitz condition (1.7). Let satisfy . Then is invertible in the ball and
Lemma 2.4. Let and define
Then, is increasing on .
Lemma 3.1. Let be a sequence generated by (1.4). Suppose that F satisfies the weak Lipschitz condition (1.7) on and that . For an integer , if
hold for each , then the following assertions hold:
Proof. Assume that (3.4) holds for each . Write . Applying (1.4), we have
Hence,
To estimate , by (3.4), we notice that
In particular,
Thus, by the weak Lipschitz condition (1.7), we obtain
Below we estimate . We firstly notice that (3.1) and (3.4) yield
Since
we have
Combining this with (3.1) implies that
Consequently, by (3.7), (3.10), (3.14) and Lemma 2.4, we get
Noting that and , we have
Moreover, since is decreasing on [0, ], one has
And therefore
That is, (3.5) holds, and the proof is complete.
We now give the main result.
Theorem 3.2. Suppose that and , and that satisfies the weak Lipschitz condition (1.7) on . Then the sequence generated by the inexact Newton method (1.4) converges to a solution of (1.1). Moreover,
Proof. We firstly use mathematical induction to prove that (3.4) holds for each . For , by the above condition and (3.2), the first inequality in (3.4) holds trivially. While the second one can be proved as follows:
Assume that (3.4) holds for all . Then, Lemma 3.1 is applicable to concluding that
Hence, by (3.5), together with the weak Lipschitz condition (1.7) and Lemma 2.3, one has
Therefore, (3.4) holds for and so for each . Consequently, for and ,
This together with Lemma 2.2 means that is a Cauchy sequence and so converges to some . While taking in (3.23), we obtain
The proof is complete.
In the special case when , inexact Newton method (1.4) reduces to Newton's method. Moreover, . Thus, Theorem 3.2 reduces to the related theorem of Newton's method.
Corollary 3.3. Assume that and , where and satisfying . Suppose that satisfies the weak Lipschitz condition (1.7) on . Then the sequence generated by Newton's method (1.2) converges to a solution of (1.1). Moreover,
where and are defined in Lemma 2.2 for .
In more particular, suppose that and . Then Corollary 3.3 reduces to the following result given in (Theorem 3.1, [7]).
Corollary 3.4. Assume that , where and . Suppose that satisfies weak Lipschitz condition (1.7) on . Then the sequence generated by Newton's method (1.2) converges to a solution of (1.1). Moreover,
where and are defined in Lemma 2.2 for and .
4. Application
This section is divided into two subsections: we consider the applications of our main results specializing, respectively, in Kantorovich type condition and in -condition. In particular, our results reduce some of the corresponding results of Newton's method.
4.1. Kantorovich-Type Condition
Throughout this subsection, let be a positive constant. By (2.1), we have
By (2.5) and (2.6), we get
The convergence criterion becomes
Moreover, suppose that and . Then criterion (4.3) reduces to the well-known Kantorovich type criterion of Newton's method in [7].
Corollary 4.1. Let be a positive constant, and , where and . Assume that satisfies the condition:
where . Then the sequence generated by Newton's method (1.2) converges to a solution of (1.1), and satisfies
4.2. -Condition
Throughout this subsection, we assume that and has continuous second derivative and satisfies
Let
Then, by (2.1), we have
By (2.5) and (2.6), and satisfy
The convergence criterion becomes
In the more special case, when and , we obtain the criterion the same with Newton's method in [7].
Corollary 4.2. Let be a positive constant, and , where and . Assume that F satisfies the condition:
where . Then the sequence generated by Newton's method (1.2) converges to a solution of (1.1), and satisfies
Acknowledgment
Supported in part by the National Natural Science Foundation of China (Grants no. 61170109 and no. 10971194) and Zhejiang Innovation Project (Grant no. T200905).
References
I. K. Argyros, βOn the Newton-Kantorovich hypothesis for solving equations,β Journal of Computational and Applied Mathematics, vol. 169, no. 2, pp. 315β332, 2004.
J. A. Ezquerro and M. A. Hernández, βGeneralized differentiability conditions for Newton's method,β IMA Journal of Numerical Analysis, vol. 22, no. 2, pp. 187β205, 2002.
J. A. Ezquerro and M. A. Hernández, βOn an application of Newton's method to nonlinear operators with -conditioned second derivative,β BIT. Numerical Mathematics, vol. 42, no. 3, pp. 519β530, 2002.
J. M. Gutiérrez, βA new semilocal convergence theorem for Newton's method,β Journal of Computational and Applied Mathematics, vol. 79, no. 1, pp. 131β145, 1997.
J. M. Gutiérrez and M. A. Hernández, βNewton's method under weak Kantorovich conditions,β IMA Journal of Numerical Analysis, vol. 20, no. 4, pp. 521β532, 2000.
M. A. Hernández, βThe Newton method for operators with Hölder continuous first derivative,β Journal of Optimization Theory and Applications, vol. 109, no. 3, pp. 631β648, 2001.
X. H. Wang, βConvergence of Newton's method and inverse function theorem in Banach space,β Mathematics of Computation, vol. 68, no. 225, pp. 169β186, 1999.
X. H. Wang, βConvergence of Newton's method and uniqueness of the solution of equations in Banach space,β IMA Journal of Numerical Analysis, vol. 20, no. 1, pp. 123β134, 2000.
X. H. Wang and C. Li, βConvergence of Newton's method and uniqueness of the solution of equations in Banach spaces. II,β Acta Mathematica Sinica, English Series, vol. 19, no. 2, pp. 405β412, 2003.
I. K. Argyros, βA new convergence theorem for the inexact Newton methods based on assumptions involving the second Fréchet derivative,β Computers & Mathematics with Applications, vol. 37, no. 7, pp. 109β115, 1999.
Z. Z. Bai and P. L. Tong, βAffine invariant convergence of the inexact Newton method and Broyden's method,β Journal of University of Electronic Science and Technology of China, vol. 23, no. 5, pp. 535β540, 1994.
J. Chen and W. Li, βConvergence behaviour of inexact Newton methods under weak Lipschitz condition,β Journal of Computational and Applied Mathematics, vol. 191, no. 1, pp. 143β164, 2006.
M. G. Gasparo and G. Morini, βInexact methods: forcing terms and conditioning,β Journal of Optimization Theory and Applications, vol. 107, no. 3, pp. 573β589, 2000.
C. Li and W. P. Shen, βLocal convergence of inexact methods under the Hölder condition,β Journal of Computational and Applied Mathematics, vol. 222, no. 2, pp. 544β560, 2008.
I. Moret, βA Kantorovich-type theorem for inexact Newton methods,β Numerical Functional Analysis and Optimization, vol. 10, no. 3-4, pp. 351β365, 1989.
J. M. Martínez and L. Q. Qi, βInexact Newton methods for solving nonsmooth equations,β Journal of Computational and Applied Mathematics, vol. 60, no. 1-2, pp. 127β145, 1995.
W. P. Shen and C. Li, βConvergence criterion of inexact methods for operators with Hölder continuous derivatives,β Taiwanese Journal of Mathematics, vol. 12, no. 7, pp. 1865β1882, 2008.
W. P. Shen and C. Li, βKantorovich-type convergence criterion for inexact Newton methods,β Applied Numerical Mathematics, vol. 59, no. 7, pp. 1599β1611, 2009.
W. P. Shen and C. Li, βSmale's -theory for inexact Newton methods under the -condition,β Journal of Mathematical Analysis and Applications, vol. 369, no. 1, pp. 29β42, 2010.
M. Wu, βA new semi-local convergence theorem for the inexact Newton methods,β Applied Mathematics and Computation, vol. 200, no. 1, pp. 80β86, 2008.