Abstract
Based on Lipschitz continuity condition, we study the local convergence analysis for the fifth-order convergent modified Weerakoon’s method for solving nonlinear equations in Banach spaces. Lipschitz continuity condition on the first derivative is assumed to extend the applicability of the scheme. This analysis enables us to solve such problems for which previous studies based on higher-order derivatives unable to find the solution. A theorem showing the existence and uniqueness of the solution along with computable error bounds is established. Standard numerical examples like nonlinear integral equation and system of nonlinear equations are solved to demonstrate the productiveness of our theoretical outcomes.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
The study presented in this paper is based on the problem of finding a locally unique solution \(x^{*}\) of the equation
where \(F:\varOmega \subseteq X \rightarrow Y\) is a Fréchet differentiable function and \(\varOmega\) is a convex subset of X. X and Y are Banach spaces. In the field of applied science and engineering, a large number of problems can be solved by transforming them into nonlinear equations of the form (1). For instance, the boundary value problems occur in Kinetic theory of gases, the integral equations related to radiative transfer theory, problems in optimization and many others can be reduced to the problem of solving nonlinear equations. Usually, the solutions of these nonlinear equations can be found in closed form. So, the most frequently used solution techniques are iterative in nature.
A commonly used iterative technique for solving (1) is Newton’s scheme, which can be expressed as:
Evaluation of second and more order derivatives is a major drawback of higher-order iterative schemes and are not appropriate for practical use. Due to the calculation of \(F''\) in each iteration, the cubically convergent classical schemes are not suitable in terms of computational cost. Some classical third-order algorithms include Chebyshev’s, the Halley’s and Super-Halley’s schemes are produced by putting (\(\alpha =0\)), (\(\alpha =\frac{1}{2}\)) and (\(\alpha =1\)) respectively in
where \(H_{F}(x_{n})=F'(x_{n})^{-1}F''(x_{n})F'(x_{n})^{-1}F(x_{n})\).
The local convergence analysis of many varieties of the methods defined in (3) has been studied by numerous authors in Refs. [1,2,3,4,5,6]. Also, the local convergence analysis for various iterative algorithms is studied in Banach spaces in Refs. [7,8,9,10,11,12]. In this paper, we use the Lipschitz continuity condition on the first derivative only to enhance the applicability of modified Weerakoon’s method in Banach spaces.
In Ref. [13], the authors studied the modification of Weerakoon’s method [14] with fifth-order convergence to solve systems of nonlinear equations in \({\mathbb {R}}^{n}\). The method is given as:
In this method, only the first-order derivative occurs in the iteration function but the convergence is proved with the assumption on higher-order derivatives for which the applicability of the method is restricted. For instance, consider a function F defined on \(\varOmega =\left[ -\frac{1}{2}, \frac{5}{2}\right]\) by
Notice that \(F'''\) is unbounded on \(\varOmega\). Therefore, the previous studies [13,14,15] based on higher-order derivatives fail to solve this problem. Also, no information is mentioned regarding the radius of convergence ball in these studies. The local convergence analysis of iterative algorithms provides essential information about the radius of convergence ball. In this paper, we provide the local convergence analysis of the method (4) using the hypotheses only on \(F'\) to avoid the use of higher-order derivatives. Particularly, it is assumed that the first derivative is Lipschitz continuous. This study extends the applicability of the method (4) and helps in obtaining the solution of such problems for which previous studies fail.
The rest portion of this paper is arranged as follows: The local convergence analysis of the method (4) is placed in Sect. 2. Section 3 is devoted to demonstrating the applications of our theoretical outcomes on some numerical examples. Conclusions are discussed in the last section.
2 Local convergence analysis
The local convergence analysis of modified Weerakoon’s method (4) is studied in this section. Let the open and closed balls in X are denoted as \(B(c, \rho )\) and \({\bar{B}}(c, \rho )\) respectively with center c and radius \(\rho >0\). Suppose the parameters \(k_0>0\) and \(k>0\) be given with \(k_0 \le k\). To study the local convergence of the scheme (4), we introduce the function \(J_1\) on the interval \(\left[ 0, \frac{1}{k_0}\right)\) by
and the parameter
Observe that \(J_1(R_1)=1\). Again, we define functions \(J_2\) and \(K_2\) on \(\left[ 0, \frac{1}{k_0}\right)\) by
and
Now, \(K_2(0)=-1<0\) and \(\displaystyle {\lim _{s \rightarrow \left( \frac{1}{k_0}\right) ^{-}}K_2(s)=+\infty }\). According to the intermediate value theorem, the interval \((0,\frac{1}{k_0})\) contains the zeros of the function \(K_2(s)\). Let the smallest zero of \(K_2(s)\) in \(\left( 0, \frac{1}{k_0}\right)\) is \(R_2\). Also, we introduce functions \(J_3\) and \(K_3\) on \([0, R_2)\) by
and
Now, \(K_3(0)=-1<0\) and \(\displaystyle {\lim _{s \rightarrow R_2^{-}}K_3(s)=+\infty }\). The intermediate value theorem confirms that the interval \((0, R_2)\) contains the zeros of the function \(K_3(s)\). Let the smallest zero of \(K_3(s)\) in \((0, R_2)\) is \(R_3\). Again, we define \(J_4\) and \(K_4\) on \(\left[ 0, \frac{1}{k_0}\right)\) by
and
Now, \(K_4(0)=-1<0\) and \(\displaystyle {\lim _{s \rightarrow \left( \frac{1}{k_0}\right) ^{-}}K_4(s)=+\infty }\). According to the intermediate value theorem, the interval \(\left( 0, \frac{1}{k_0}\right)\) contains the zeros of the function \(K_4(s)\). Let the smallest zero of \(K_4(s)\) in \(\left( 0, \frac{1}{k_0}\right)\) is \(R_4\). Finally, let us define \(J_5\) and \(K_5\) on \([0, R_4)\) by
and
Now, \(K_5(0)=-1<0\) and \(\displaystyle {\lim _{s \rightarrow R_4^{-}}K_5(s)=+\infty }\). The intermediate value theorem confirms that the interval \((0, R_4)\) contains the zeros of the function \(K_5(s)\). Let the smallest zero of \(K_5(s)\) in \((0, R_4)\) is \(R_5\). Consider
Now, we have
and
for each \(s \in [0, R)\). Furthermore, let us assume the followings hold for the Fréchet differentiable function \(F:\varOmega \subseteq X \rightarrow Y\).
and
where BL(Y, X) is the set of all bounded linear operators from Y to X.
In several studies [1, 2, 9, 16, 17], a third condition assumed is
This assumption is not taken in our study. We use the following results to avoid this extra condition.
Lemma 1
If F obeys (16) and \({\bar{B}}(x^{*}, R) \subseteq \varOmega\), then \(\forall x \in B(x^{*}, R)\), we get
and
Proof
Applying (16), we obtain
For \(\theta \in [0, 1]\),
The mean value theorem is used to obtain
\(\square\)
Next, the local convergence analysis of the method (4) is presented in Theorem 1.
Theorem 1
Let \(F:\varOmega \subseteq X \rightarrow Y\)be a Fréchet differentiable function. Suppose \(x^{*} \in \varOmega, F(x^{*})=0, F'(x^{*})^{-1} \in BL(Y, X)\), F obeys (16), (17) and
where R is defined in (10). Starting from \(x_0 \in B(x^{*}, R)\)the method (4) generates the sequence of iterates \(\{x_n\}\)which is well defined, \(\{x_n\}_{n \ge 0} \in B(x^{*}, R)\)and converges to the solution \(x^{*}\)of (1). Moreover, the following estimations hold \(\forall n \ge 0\)
and
where the functions \(J_1\), \(J_3\)and \(J_5\)are given in (5), (7) and (9) respectively. Furthermore, the solution \(x^{*}\)of the equation \(F(x)=0\)is unique in \({\bar{B}}(x^{*}, \varDelta ) \cap \varOmega\), where \(\varDelta \in [R, \frac{2}{k_0})\).
Proof
Using the definition of R, the equation (16) and the assumption \(x_0 \in B(x^{*}, R)\), we find
Now, Banach Lemma on invertible functions [18,19,20,21,22,23] confirms that \(F'(x_0)^{-1} \in BL(Y, X)\) and
Hence, it follows from the first step of the method (4) for \(n=0\) that \(y_0\) is well defined. Again,
Using (5), (10), (11) and (17), we find
and this shows (22) for \(n=0\). Then we show \([F'(x_0)+F'(y_0)]^{-1} \in BL(Y, X)\). The equations (6), (10), (12), (16) and (28) are used to obtain
Now, we obtain \([F'(x_0)+F'(y_0)]^{-1} \in BL(Y, X)\) using Banach Lemma on invertible functions. Also,
Now, it follows from the second step of the method (4) for \(n=0\) that \(z_0\) is well defined. Using the definition of R, (13), (17), (28) and (29), we get
Hence, we establish (23) for \(n=0\). Again,
So, \(F'(y_0)^{-1} \in BL(Y, X)\) with
Now, it follows from the last step of the method (4) for \(n=0\) that \(x_1\) is well define. Finally, we use (10), (15), (20), (30) and (32) to get
Thus, we show the estimate (24) for \(n=0\). We get the estimates (22)-(24) by substituting \(x_n\), \(y_n\), \(z_n\) and \(x_{n+1}\) in place of \(x_0\), \(y_0\), \(z_0\) and \(x_1\) respectively in the previous estimations. Using the fact \(||x_{n+1}-x^{*}|| \le J_5(R)||x_n-x^{*}||<R\), we derive that \(x_{n+1} \in B(x^{*}, R)\) and \(\displaystyle {\lim _{n \rightarrow \infty } x_n=x^{*}}\). Now, we want to show the uniqueness of the solution \(x^{*}\). Suppose there exist another solution \(y^{*}\) of \(F(x)=0\) in \({B}(x^{*}, \varDelta )\). Consider \(T=\int _{0}^{1} F'(y^{*}+\theta (x^{*}-y^{*})) d\theta\). From equation (16), we get
Applying Banach Lemma, we find \(T^{-1} \in BL(Y, X)\). Now, Using the identity \(0=F(x^{*})-F(y^{*})=T(x^{*}-y^{*})\), it is concluded that \(x^{*}=y^{*}\). This ends the proof. \(\square\)
3 Numerical examples
Example 1
Define F on \(\varOmega =[-\frac{1}{2}, \frac{5}{2}]\) by
We have \(x^{*}=1\). Also, \(k_0=k=96.6628\). The value of R is determined using the definitions of \(``J''\) functions (Table 1).
Example 2
Let us define F on \({\bar{B}}(0, 1)\) for \((x_1, x_2, x_3)^{t}\) by
We have \(x^{*}=(0, 0, 0)^t\). Also, we have \(k_0=e-1\) and \(k=e\). We determine the value of R using \(``J''\) functions (Table 2).
Example 3
Let us define F on \(\varOmega =[-1, 1]\) by
We have \(x^{*}=0\). Also, we have \(k_0=k=1\). R is determined using \(``J''\) functions (Table 3).
Example 4
Consider the nonlinear Hammerstein type integral equation given by
where \(x(s) \in C[0, 1]\). We have \(x^{*}=0\). Also, \(k_0=7.5\) and \(k=15\). Using the definitions of \(``J''\) functions the value of R is determined (Table 4).
4 Conclusions
We studied the local convergence analysis of the method (4) to find a locally unique solution of a nonlinear equation in Banach spaces. The Lipschitz continuity condition on the first derivative is used to enhance the applicability of these methods. This study helps in solving those problems for which higher-order derivative based previous studies fail. Lastly, the theoretical outcomes are applied on standard numerical examples like Hammerstein equation and system of nonlinear equations.
References
Argyros, I.K., and Á.A. Magreñán. 2015. A study on the local convergence and the dynamics of Chebyshev–Halley-type methods free from second derivative. Numerical Algorithms 71 (1): 1–23.
Argyros, I.K., and S. George. 2015. Local convergence of deformed Halley method in Banach space under Holder continuity conditions. Journal of Nonlinear Sciences and Applications 8: 246–254.
Argyros, I.K., S. George, and Á.A. Magreñán. 2015. Local convergence for multi-point-parametric Chebyshev–Halley-type methods of higher convergence order. Journal of Computational and Applied Mathematics 282: 215–224.
Argyros, I.K., and S. George. 2015. Local convergence of modified Halley-like methods with less computation of inversion. Novi Sad Journal of Mathematics 45 (2): 47–58.
Argyros, I.K., Y.J. Cho, and H.M. Ren. 2013. Convergence of Halley’s method for operators with the bounded second derivative in Banach spaces. Journal of Inequalities and Applications 2013: 260.
Candela, V., and A. Marquina. 1990. Recurrence relations for rational cubic methods I: The Halley method. Computing 44 (2): 169–184.
Singh, S., D.K. Gupta, R.P. Badoni, E. Martínez, and J.L. Hueso. 2016. Local convergence of a parameter based iteration with Holder continuous derivative in Banach spaces. Calcolo. 54 (2): 527–539. https://doi.org/10.1007/s10092-016-0197-9
Martínez, E., S. Singh, J.L. Hueso, and D.K. Gupta. 2016. Enlarging the convergence domain in local convergence studies for iterative methods in Banach spaces. Applied Mathematics and Computation 281: 252–265.
Argyros, I.K., and S. Hilout. 2013. On the local convergence of fast two-step Newton-like methods for solving nonlinear equations. Journal of Computational and Applied Mathematics 245: 1–9.
Argyros, I.K., and S.K. Khattri. 2015. Local convergence for a family of third order methods in Banach spaces. Applied Mathematics and Computation 251: 396–403.
Cordero, A., J.A. Ezquerro, and M.A. Hernandez-Veron. 2014. On the local convergence of a fifth-order iterative method in Banach spaces. Journal of Mathematics 46: 53–62.
Argyros, I.K., Y.J. Cho, and S. George. 2014. On the “Terra incognita” for the Newton–Kantrovich method. Journal of the Korean Mathematical Society 51 (2): 251–266.
Cordero, A., J.L. Hueso, E. Martínez, and J.R. Toregrossa. 2012. Increasing the convergence order of an iterative method for nonlinear systems. Applied Mathematics Letters 25: 2369–2374.
Weerakoon, S., and T.G.I. Fernando. 2000. A variant of Newton’s method with accelerated third-order convergence. Applied Mathematics Letters 13 (8): 87–93.
Nishani, H.P.S., S. Weerakoon, T.G.I. Fernando, and M. Liyanag. 2018. Weerakoon–Fernando method with accelerated third-order convergence for systems of nonlinear equations. International Journal of Mathematical Modelling and Numerical Optimisation 8 (13): 287–304.
Argyros, I.K., Y.J. Cho, and S. George. 2016. Local convergence for some third order iterative methods under weak conditions. Journal of the Korean Mathematical Society 53 (4): 781–793.
Argyros, I.K., and D. González. 2015. Local convergence for an improved Jarratt-type method in Banach space. International Journal of Artificial Intelligence 3: 20–25. (3(Special Issue on Teaching Mathematics Using New and Classic Tools)).
Argyros, I.K. 2008. Convergence and application of Newton-type iterations. Berlin: Springer.
Argyros, I.K., and S. Hilout. 2013. Computational methods in nonlinear Analysis. New Jersey: World Scientific Publishing House.
Petković, M.S., B. Neta, L. Petković, and D. Dzunić. 2013. Multipoint methods for solving nonlinear equations. Amsterdam: Elsevier.
Rall, L.B. 1979. Computational solution of nonlinear operator equations. New York: Robert E. Krieger.
Traub, J.F. 1964. Iterative methods for solution of equations. Englewood Cliffs: Prentice-Hall.
Argyros, I.K., Y.J. Cho, and S. Hilout. 2012. Numerical methods for equations and its applications. New York: Taylor & Francis.
Acknowledgements
This research is supported by University Grants Commission of India (Ref. no.: 988/(CSIR-UGC NET JUNE 2017)).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they do not have conflict of interests.
Ethical standards
This research complies with ethical standards.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Sharma, D., Parhi, S.K. On the local convergence of modified Weerakoon’s method in Banach spaces. J Anal 28, 867–877 (2020). https://doi.org/10.1007/s41478-019-00216-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s41478-019-00216-x