Research Article | Open Access

Volume 2016 |Article ID 8565821 | https://doi.org/10.1155/2016/8565821

Grégory Antoni, "An Efficient and Straightforward Numerical Technique Coupled to Classical Newton’s Method for Enhancing the Accuracy of Approximate Solutions Associated with Scalar Nonlinear Equations", International Journal of Engineering Mathematics, vol. 2016, Article ID 8565821, 12 pages, 2016. https://doi.org/10.1155/2016/8565821

# An Efficient and Straightforward Numerical Technique Coupled to Classical Newton’s Method for Enhancing the Accuracy of Approximate Solutions Associated with Scalar Nonlinear Equations

Accepted14 Aug 2016
Published02 Nov 2016

#### Abstract

This study concerns the development of a straightforward numerical technique associated with Classical Newton’s Method for providing a more accurate approximate solution of scalar nonlinear equations. The proposed procedure is based on some practical geometric rules and requires the knowledge of the local slope of the curve representing the considered nonlinear function. Therefore, this new technique uses, only as input data, the first-order derivative of the nonlinear equation in question. The relevance of this numerical procedure is tested, evaluated, and discussed through some examples.

#### 1. Introduction

The resolution of nonlinear problems is an issue frequently encountered in several scientific fields such as mathematics, physics, or many engineering branches, for example, mechanics of solids . In most cases, these problems are governed by nonlinear equations not having any analytical solution. In this regard, the introduction of iterative methods is therefore needed in order to provide a numerical approximate solution associated with any type of nonlinear equation . Among these iterative algorithms, Classical Newton’s Method (CNM) [24, 25] is one of the most used mainly for the following reasons: (i) the simplicity for numerical implementation in any scientific computation software; (ii) the only knowledge of the first-order derivative of the considered function; (iii) the quadratic rate of convergence. In this paper, we propose a New Numerical Technique (NNT) based on geometric considerations which enable providing a more accurate approximate solution than that obtained by CNM. The present study is organized as follows: (i) in the first part, Section 2.1, we outline the scientific framework of this study, then, in second part, Section 2.2, we recall CNM including some convergence results, and, finally, in the third part, Section 2.3, we present NNT which uses only the first-order derivative of the considered nonlinear equation in order to enhance the predictive abilities of CNM; (ii) in Section 3, the numerical relevance of the proposed procedure is addressed, assessed, and discussed on some specific examples.

#### 2. A New Numerical Technique (NNT) Combined with Classical Newton’s Method (CNM)

##### 2.1. Problem Statement

We consider scalar-valued nonlinear equation (with ) in the following form (see Figure 1):where denotes the class of infinitely differentiable functions in domain , is the simple solution (so-called “simple true zero” or “simple root”) on interval , that is, with .

##### 2.2. Classical Newton’s Method (CNM)
###### 2.2.1. Iterative Algorithm

For using Classical Newton’s Method (CNM) [24, 25], we consider only the first-order term in Taylor series expansion associated with function (i.e., the linearization of the considered function):where denotes the first-order derivative of functionat point.

The equation of tangent straight line passing at point associated with function is (; see Figure 2) as follows:where (resp.,) denotes th (resp., th) iteration associated with variable (with ).

Using (3), the linearization of (1) must check the following relation ( and ):

Based on (4),th iterative point provided by CNM is solution () such as

###### 2.2.2. Convergence Results

Considering Taylor series expansion of functionand truncation error () between th iterative approximate solution () and true zero (), that is, with and , we havewithwhere denotes -order derivative of function at point with , is -factorial, represent the Higher-Order Terms which check: or , and are Landau notations associated with the asymptotic behavior of function , and is the square norm or Euclidean-norm associated with the quantity (here, Euclidean distance reduces to absolute value , i.e., ).

Since that , (6)-(7) can be written as follows (with):

The first-order derivative associated with function at point (i.e., ) checks:

Combining (5), (8)-(9), and leads to the following:

An iterative numerical scheme has th order of convergence with associated convergence rate to number (i.e., “true zero”) if there exists and  such aswhere is the absolute-value function of variable(such as when , when , and when ) and “” is the limit operator. It is important to emphasize that (i) if and then the convergence is linear (or “first-order” type, e.g., bisection or false position method); (ii) ifandthen the convergence is sublinear; (iii) ifand (or) then the convergence is superlinear (e.g., secant method); (iv) if(with) then the convergence is quadratic (or “second-order” type, e.g., CNM); (v) if (with) then the convergence is cubic (or “third-order” type, e.g., TMNM; see Section 3.1). th order of convergence means that the number of significant digits is-fold at each iteration, for example, in the case of CNM (resp., TMNM), where (resp.,), the number of exact decimals doubles (resp., triples) at each iteration .

Using (10)-(11), we can see that CNM has quadratic convergence () with a rate of convergence :

##### 2.3. New Numerical Technique (NNT)
###### 2.3.1. Proposed Iterative Procedure

In this section, we propose a New Numerical Technique (NNT) for improving the accuracy associated with the approximate solution provided by CNM (see Section 2.2, which is used for finding the roots of scalar nonlinear equations. NNT is based on some practical geometric rules and requires only the determination of the local slope of the curve representing the considered nonlinear equation taking into account the first-order derivation (see ).

Here, we present the main steps associated with the development of NNT:(i)We consider normal straight line associated with the curve representing nonlinear equation at point (see [26, 27] and Figure 2):(ii)We introduce straight line having direction vector which depends on the sum of direction vectors associated with normal and tangent straight lines passing by point (with ; see Figure 2), that is:withwhere and are the components of direction vector associated with straight line in any orthonormal basis .(iii)Using CNM, we define th iterative point (with ; see (5) and also Figure 2):(iv)We introduce also straight line in the following form (with ; see Figure 2):with(v)Combining (14)-(15) and (17)-(18), we adopt the solution of the equation (with ; see Figure 2), that is:According to (16) and (19), th iterative point associated with NNT (see Figure 2) can be rewritten (with ):

The iterative numerical scheme associated with NNT is coupled with CNM and therefore th iterative solution can be written as follows (with, ; see Figure 2):with the different conditions associated with the iterative solution (with ):(i)First condition [A1]: and and .(ii)Second condition [A2]:

###### 2.3.2. Convergence Analysis

Similar to that in Section 2.2.2, we analyze the convergence associated with NNT which is presented in Section 2.2.

Using (9) leads to the following:

In line with (22), we have the following:

On the other hand, we have the following:

Combining (8) and (24) leads to the following:

According to (19) and using (23) and (25), it holds that

In line with (11) and (26), we can see that the rate of convergenceassociated with NNT isand the order of convergence is of linear-type (i.e., ) if and quadratic-type (i.e., ) if .

By taking (12) and (27), the rate of convergenceof NNT combined with CNM is

It is important to emphasize that the associated convergence order is linear-type () if with condition A (i.e., [A1] or [A2]) and quadratic-type () if with condition (i.e., [A1] or [A2]) and elsewhere with .

#### 3. Numerical Examples

##### 3.1. Some Preliminary Remarks

In this section, we propose to test, evaluate, and analyze the iterative numerical procedure presented in Section 2.3 on some particular examples. This New Numerical Technique (NNT) is coupled with Classical Newton’s Method (CNM) in order to provide a more accurate approximate solution associated with scalar nonlinear equations. The numerical predictions obtained by combining both NNT and CNM are compared with those provided by Third-order Modified Newton’s Method (TMNM) [22, 26]. All numerical results presented here have been made with Matlab software (see [25, 2932]).

For stopping the iterative process associated with each considered algorithm, we adopt three coupled types of Convergence Criterion (CC):where denotes the maximum number of iterations and (resp.) is the tolerance parameter associated with the residue (resp., approximation) error criterion. Here, the considered values for each CC are , , and .

The iterative numerical scheme associated with TMNM (see [22, 26]) iswhere denotes the second-order derivative of function at point . It should be noted that order of convergence is cubic (i.e., ) and rate of convergence is (see [22, 26])

##### 3.2. Examples

We consider the following scalar nonlinear equations:where , , and represent exponential, natural logarithm, and cosinus functions.

The approximate numerical solution of roots associated with functions , , and are, respectively,

##### 3.3. Results and Discussion

All the numerical results associated with scalar nonlinear functions to are presented in Tables 16 and also in Figures 320. For each scalar nonlinear function (with ), we test two guest points for starting iterative procedure: (i) in the case of function , we adopt the first-guest point (resp., second-guest point ) and we show different approximate solutions (with ) and also the evolution of both residue error and approximation error provided by Classical Newton’s Method (CNM; see Section 2.2), CNM coupled to the proposed New Numerical Technique (NNT) with both conditions [A1] and [A2] (CNM + NNT; see Section 2.3), and Third-order Modified Newton’s Method (TMNM; see Section 3.1) in Table 1 (resp., Table 2) and Figures 35 (resp., Figures 68); (ii) in the case of function , we adopt first-guest point (resp., second-guest point ) and we show different approximate solutions (with ) provided by CNM, CNM + NNT with both conditions [A1] and [A2], and TMNM in Table 3 (resp., Table 4) and Figures 911 (resp., Figures 1214); (iii) in the case of function, we adopt first-guest point (resp., second-guest point ) and we show different approximate solutions (with ) provided by CNM, CNM + NNT with both conditions [A1] and [A2], and TMNM in Table 5 (resp., Table 6) and Figures 1517 (resp., Figures 1820). The obtained numerical results emphasize clearly that NNT combined with CNM is able to provide in the vast majority of cases a better approximate solution than that supplied only by CNM. In addition, on all results presented, condition [A2] seems to be more convenient than condition [A1]. For summary, NNT presents the main advantages of being a procedure: (i) “accurate” in terms of achieved approximate solution; (ii) “straightforward” to implement in computation software.

 Iteration () CNM CNM + NNT with condition [A1] CNM + NNT with condition [A2] TMNM 0 3.0000000000 3.0000000000 3.0000000000 3.0000000000 1 2.2471426305 1.4942904056 () 1.4942904056 () 1.9644439737 2 1.6823826593 1.1286469200 () 0.7631581663 () 1.2884809024 3 1.2337388302 0.8906377962 () 0.7721173336 () 0.8906149334 4 0.9723200229 0.7878458749 () 0.7720250356 () 0.7746164630 5 0.8146707520 0.7723183427 () 0.7720250256 () 0.7720250503 6 0.7741430851 0.7720251266 () 0.7720250256 7 0.7720302909 0.7720250256 () 8 0.7720250256 9 0.7720250256 10
 Iteration () CNM CNM + NNT with condition [A1] CNM + NNT with condition [A2] TMNM 0 0.0010000000 0.0010000000 0.0010000000 0.0010000000 1 0.0082410886 0.0154821756 () 0.0154821756 () 0.0344577770 2 0.0505339927 0.1548613133 () 0.1548613133 () 0.3980194000 3 0.2182202963 0.4938846484 () 0.8311197179 () 0.9031891893 4 0.6163033790 0.8261543800 () 0.7760693275 () 0.7755098327 5 0.7975366151 0.7754244611 () 0.7720442199 () 0.7720250858 6 0.7727863680 0.7720385875 () 0.7720250260 () 0.7720250256 7 0.7720257060 0.7720250258 () 0.7720250256 () 8 0.7720250256 0.7720250256 () 9 10
 Iteration () CNM CNM + NNT with condition [A1] CNM + NNT with condition [A2] TMNM 0 10.000000000 10.000000000 10.000000000 10.000000000 1 8.9811766535 7.9623533071 () 7.9623533071 () 8.4655416573 2 7.9456740785 5.8372149905 () 5.8372149905 () 6.8799380466 3 6.8825672128 3.5600762341 () 3.5600762341 () 5.2048513181 4 5.7803883041 1.4821734803 () 1.4821734803 () 3.4865001565 5 4.6404493806 0.9609028311 () 0.4402289741 () 2.0197617679 6 3.5056070182 0.6928062817 () 0.6483081663 () 1.0685736902 7 2.4744259573 0.6233620739 () 0.6199084714 () 0.6632122684 8 1.6461067325 0.6192585092 () 0.6192453069 () 0.6193191836 9 1.0635449602 0.6192449541 () 0.6192449540 () 0.6192449540 10 0.7354406002 0.6192449510 () 0.6192449540
 Iteration () CNM CNM + NNT with condition [A1] CNM + NNT with condition [A2] TMNM 0 −4.0000000000 −4.0000000000 −4.0000000000 −4.0000000000 1 −2.5987493208 −1.1976336149 () −1.1976336149 () −2.1088093909 2 −1.5803235380 −0.2513579056 () 0.6858578709 () −0.7930763263 3 −0.6950023819 1.1303884977 () 0.6226379112 () 1.4022343921 4 0.5489290102 0.7665675348 () 0.6192541648 () 0.7698587570 5 0.6234126286 0.6349082108 () 0.6192449540 () 0.6217981372 6 0.6192588436 0.6194395336 () 0.6192449540 () 0.6192449694 7 0.6192449541 0.6192449843 () 0.6192449540 8 0.6192449540 0.6192449540 () 0.6192449540 9 10
 Iteration () CNM CNM + NNT with condition [A1] CNM + NNT with condition [A2] TMNM 0 10.000000000 10.000000000 10.000000000 10.000000000 1 9.0104267607 8.0208535216 () 8.0208535216 () 8.5171506533 2 8.0146859130 5.9615544966 () 5.9615544966 () 6.9859301990 3 6.9847503720 3.9493415483 () 3.9493415483 () 5.3851997390 4 5.9189110435 2.4996523082 () 2.4996523082 () 4.0778432768 5 4.9180876708 1.7226271036 () 0.9467658294 () 3.0067861343 6 4.0738875311 1.3881795157 () 1.4273487664 () 1.8098899052 7 3.3404588095 1.3598140013 () 1.3607723321 () 1.3648044141 8 2.6091363141 1.3595986336 () 1.3595989925 () 1.3595986468 9 1.8194405679 1.3595986211 () 1.3595986211 () 1.3595986211 10 1.4045600984
 Iteration () CNM CNM + NNT with condition [A1] CNM + NNT with condition [A2] TMNM 0 −1.0000000000 −1.0000000000 −1.0000000000 −1.0000000000 1 6.1409303723 6.1409303723 () 6.1409303723 () 14.5343055115 2 5.1148208484 4.0887116616 () 4.0887116616 () 13.0337347682 3 4.2379703013 2.6194618995 () 2.6194618995 () 11.5318478151 4 3.4892766889 1.8291595323 () 1.0397597354 () 10.0370297219 5 2.7676801225 1.4064394475 () 1.3966765848 () 8.5556896133 6 1.9776897481 1.3601688047 () 1.3599586541 () 7.0203746257 7 1.4417009128 1.3595987088 () 1.3595986561 () 5.3974911835 8 1.3613036994 1.3595986211 () 1.3595986211 () 4.1321703085 9 1.3595994046 3.1258556869 10 1.3595986211 1.9485648236

This study is devoted to a New Numerical Technique (NNT) to improve the accuracy of approximate solution provided by Classical Newton’s Method (CNM) and afford to have better numerical evaluation of the roots associated with the scalar nonlinear equations. As in CNM, this NNT requires only the determination of the first-order derivative of the nonlinear function under consideration. The predictive capabilities associated with NNT are shown on some examples.

#### Competing Interests

The author declares that they have no competing interests.

1. Q. S. Nguyen, Stability and Nonlinear Solid Mechanics, John Wiley & Sons, New York, NY, USA, 2000.
2. I. Doghri, Mechanics of Deformable Solids: linear, Nonlinear, Analytical And Computational Aspects, Springer, Berlin, Germany, 2000. View at: Publisher Site | MathSciNet
3. A. Curnier, Méthodes Numériques en Mécanique des Solides, Presses Polytechniques et Universitaires Romandes, 2000.
4. R. Parnes, Solid Mechanics in Engineering, Wiley, Hoboken, NJ, USA, 2001.
5. J. Besson, G. Cailletaud, J.-L. Chaboche, and S. Forest, Non-Linear Mechanics of Materials, vol. 167 of Solid Mechanics and Its Applications, Springer, 2010.
6. R. De Borst, M. A. Crisfield, J. J. C. Remmers, and C. V. Verhoosel, Non-Linear Finite Element Analysis of Solids and Structures. Computational Mechanics, Wiley-Blackwell, Hoboken, NJ, USA, 2012.
7. M. Bonnet, A. Frangi, and C. Rey, The Finite Element Method in Solid Mechanics, McGraw-Hill Education, 2014.
8. M. Kojić and K. J. Bathe, Inelastic Analysis of Solids and Structures, Computational Fluid and Solid Mechanics, Springer, 2005.
9. S. Abbasbandy, “Improving Newton-Raphson method for nonlinear equations by modified Adomian decomposition method,” Applied Mathematics and Computation, vol. 145, no. 2-3, pp. 887–893, 2003. View at: Publisher Site | Google Scholar | MathSciNet
10. M. A. Noor, “New family of iterative methods for nonlinear equations,” Applied Mathematics and Computation, vol. 190, no. 1, pp. 553–558, 2007. View at: Publisher Site | Google Scholar | MathSciNet
11. K. I. Noor, M. A. Noor, and S. Momani, “Modified Householder iterative method for nonlinear equations,” Applied Mathematics and Computation, vol. 190, no. 2, pp. 1534–1539, 2007. View at: Publisher Site | Google Scholar | MathSciNet
12. M. A. Noor, “Some iterative methods free from second derivatives for nonlinear equations,” Applied Mathematics and Computation, vol. 192, no. 1, pp. 101–106, 2007. View at: Publisher Site | Google Scholar | MathSciNet
13. M. A. Noor, “New classes of iterative methods for nonlinear equations,” Applied Mathematics and Computation, vol. 191, no. 1, pp. 128–131, 2007. View at: Publisher Site | Google Scholar | MathSciNet
14. C. Chun, “A family of composite fourth-order iterative methods for solving nonlinear equations,” Applied Mathematics and Computation, vol. 187, no. 2, pp. 951–956, 2007. View at: Publisher Site | Google Scholar | MathSciNet
15. B. Neta and A. N. Johnson, “High-order nonlinear solver for multiple roots,” Computers & Mathematics with Applications, vol. 55, no. 9, pp. 2012–2017, 2008. View at: Publisher Site | Google Scholar | MathSciNet
16. A. Cordero and J. R. Torregrosa, “A class of multi-point iterative methods for nonlinear equations,” Applied Mathematics and Computation, vol. 197, no. 1, pp. 337–344, 2008.
17. T.-F. Li, D.-S. Li, Z.-D. Xu, and Y.-l. Fang, “New iterative methods for non-linear equations,” Applied Mathematics and Computation, vol. 197, no. 2, pp. 755–759, 2008. View at: Publisher Site | Google Scholar | MathSciNet
18. C. Chun, “A simply constructed third-order modifications of Newton's method,” Journal of Computational and Applied Mathematics, vol. 219, no. 1, pp. 81–89, 2008. View at: Publisher Site | Google Scholar | MathSciNet
19. Y. M. Ham, C. Chun, and S.-G. Lee, “Some higher-order modifications of Newton's method for solving nonlinear equations,” Journal of Computational and Applied Mathematics, vol. 222, no. 2, pp. 477–486, 2008. View at: Publisher Site | Google Scholar | MathSciNet
20. L. Fang and G. He, “Some modifications of Newton's method with higher-order convergence for solving nonlinear equations,” Journal of Computational and Applied Mathematics, vol. 228, no. 1, pp. 296–303, 2009. View at: Publisher Site | Google Scholar | MathSciNet
21. C. Chun and B. Neta, “Some modification of Newton's method by the method of undetermined coefficients,” Computers & Mathematics with Applications, vol. 56, no. 10, pp. 2528–2538, 2008. View at: Publisher Site | Google Scholar
22. G. Antoni, “A new Newton-type method for solving both non-linear equations and systems,” Asian Journal of Mathematics and Computer Research, vol. 6, no. 3, pp. 193–212, 2015. View at: Google Scholar
23. G. Antoni, “A new class of two-step iterative algorithms for finding roots of nonlinear equations,” Asian Journal of Mathematics and Computer Research, vol. 7, no. 3, pp. 175–189, 2015. View at: Google Scholar
24. C. T. Kelley, Solving Nonlinear Equations with Newton's Method, vol. 1 of Fundamental Algorithms for Numerical Calculations, Society for Industrial and Applied Mathematics, Philadelphia, Pa, USA, 2003.
25. A. Quarteroni, R. Sacco, and F. Saleri, Méthodes Numériques pour le Calcul Scientifique: Programmes en MATLAB, Springer, Berlin, Germany, 2000.
26. G. Antoni, “A new iterative algorithm for approximating zeros of nonlinear scalar equations: geometry-based solving procedure,” Asian Journal of Mathematics and Computer Research, vol. 10, no. 2, pp. 78–97, 2015. View at: Google Scholar
27. G. Antoni, “A new accurate and efficient iterative numerical method for solving the scalar and vector nonlinear equations: approach based on geometric considerations,” International Journal of Engineering Mathematics, vol. 2016, Article ID 6390367, 18 pages, 2016. View at: Publisher Site | Google Scholar
28. G. Antoni, “A simple and efficient root-finding algorithm for dealing with scalar nonlinear equations: iterative procedure based on geometric considerations,” Asian Journal of Mathematics and Computer Research, In press View at: Google Scholar
29. J. H. Mathews and K. K. Fink, Numerical Methods Using Matlab, Pearson, 2004.
30. W. Y. Yang, W. Cao, T. S. Chung, and J. Morris, Applied Numerical Methods Using MATLAB, Wiley-Interscience, New York, NY, USA, 2005.
31. J. Kiusalaas, Numerical Methods in Engineering with MATLAB, Cambridge University Press, Cambridge, UK, 2009.
32. S. C. Chapra, Applied Numerical Methods with MATLAB for Engineers and Scientists, McGraw-Hill Higher Education, 2011.

#### More related articles

We are committed to sharing findings related to COVID-19 as quickly as possible. We will be providing unlimited waivers of publication charges for accepted research articles as well as case reports and case series related to COVID-19. Review articles are excluded from this waiver policy. Sign up here as a reviewer to help fast-track new submissions.