Professional Documents
Culture Documents
Research Article
New Family of Iterative Methods for Solving Nonlinear Models
Received 6 December 2017; Revised 1 March 2018; Accepted 22 March 2018; Published 30 April 2018
Copyright © 2018 Faisal Ali et al. This is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
We introduce a new family of iterative methods for solving mathematical models whose governing equations are nonlinear in nature.
The new family gives several iterative schemes as special cases. We also give the convergence analysis of our proposed methods. In
order to demonstrate the improved performance of newly developed methods, we consider some nonlinear equations along with
two complex mathematical models. The graphical analysis for these models is also presented.
2 (𝑓 (𝛾) + 𝑔 (𝑥))
𝑥=𝛾− 𝑝 𝑝 (6)
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥 − 𝛾)) + (𝑥 − 𝛾) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥 − 𝛾))
= 𝑐 + 𝑁 (𝑥) , (7)
where
𝑐 = 𝛾, (8)
2 (𝑓 (𝛾) + 𝑔 (𝑥))
𝑁 (𝑥) = − 𝑝 𝑝 . (9)
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥 − 𝛾)) + (𝑥 − 𝛾) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥 − 𝛾))
The purpose of the above decomposition is to find the solu- 𝑥2 = 𝑁 (𝑥0 + 𝑥1 ) − 𝑁 (𝑥0 ) ,
tion of (2) in the form of a series: (13)
..
.
∞
𝑥 = ∑𝑥𝑖. (11) 𝑥𝑚+1 = 𝑁 (𝑥0 + 𝑥1 + ⋅ ⋅ ⋅ + 𝑥𝑚 )
𝑖=0
− 𝑁 (𝑥0 + 𝑥1 + ⋅ ⋅ ⋅ + 𝑥𝑚−1 ) , 𝑚 = 1, 2, . . . .
Combining (7), (10), and (11), we get
Therefore, we have
∞
𝑥 = ∑𝑥𝑖
𝑖=0 𝑥1 + 𝑥2 + ⋅ ⋅ ⋅ + 𝑥𝑚+1 = 𝑁 (𝑥0 + 𝑥1 + ⋅ ⋅ ⋅ + 𝑥𝑚 ) ,
Discrete Dynamics in Nature and Society 3
2 (𝑓 (𝛾) + 𝑔 (𝑥0 ))
𝑥1 = 𝑁 (𝑥0 ) = − 𝑝 𝑝 . (18)
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 − 𝛾)) + (𝑥0 − 𝛾) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 − 𝛾))
Using condition (5) and (17), the above equation yields which is the well-known Newton’s method for solving non-
linear (2). Now from (21), we have
𝑓 (𝛾) 𝑓 (𝛾)
𝑥1 = − . (19) 𝑥0 + 𝑥1 − 𝛾 = − . (23)
𝑓 (𝛾) 𝑓 (𝛾)
From (4) and using (23), the following can easily be obtained:
We note that 𝑥 is approximated as
𝑔 (𝑥0 + 𝑥1 ) = 𝑓 (𝑥0 + 𝑥1 ) − 𝑓 (𝛾)
𝑝
𝑥 = lim 𝑋𝑚 , where 𝑋𝑚 = 𝑥0 + 𝑥1 + ⋅ ⋅ ⋅ + 𝑥𝑚 .
𝑚→∞ (20) − [∑𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 − 𝛾))] (𝑥0 + 𝑥1 − 𝛾)
𝑖=1
2 (𝑓 (𝛾) + 𝑔 (𝑥0 + 𝑥1 ))
𝑥1 + 𝑥2 = 𝑁 (𝑥0 + 𝑥1 ) = − 𝑝 𝑝
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 − 𝛾)) + (𝑥0 + 𝑥1 − 𝛾) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 − 𝛾))
(25)
𝑓 (𝛾) 2𝑓 (𝑥0 + 𝑥1 )
=− − .
𝑓 (𝛾) 2 ∑𝑝𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 − 𝛾)) + (𝑥0 + 𝑥1 − 𝛾) ∑𝑝𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 − 𝛾))
For 𝑚 = 2, in (20) and by using (13), (14), (16), and (25), we get
𝑥 ≈ 𝑋2 = 𝑥0 + 𝑥1 + 𝑥2 = 𝑐 + 𝑁 (𝑥0 + 𝑥1 )
Using above relation, we can suggest the following two-step Algorithm 2. For a given 𝑥0 , compute the approximate solu-
iterative method for solving nonlinear equation (2). tion 𝑥𝑛+1 by the following iterative scheme:
2𝑓 (𝑦𝑛 )
𝑥𝑛+1 = 𝑦𝑛 − 𝑝 𝑝 ,
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑦𝑛 − 𝑥𝑛 )) + (𝑦𝑛 − 𝑥𝑛 ) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑦𝑛 − 𝑥𝑛 ))
(27)
𝑓 (𝑥 )
𝑦𝑛 = 𝑥𝑛 − 𝑛 𝑛 = 0, 1, 2, . . . .
𝑓 (𝑥𝑛 )
𝑓 (𝛾) 2𝑓 (𝑥0 + 𝑥1 )
𝑥0 + 𝑥1 + 𝑥2 − 𝛾 = − . (28)
𝑓 (𝛾) 2 ∑𝑝𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 − 𝛾)) + (𝑥0 + 𝑥1 − 𝛾) ∑𝑝𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 − 𝛾))
𝑝
From (4) and by using (28), we have
− [∑𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 + 𝑥2 − 𝛾))]
𝑖=1
𝑔 (𝑥0 + 𝑥1 + 𝑥2 ) = 𝑓 (𝑥0 + 𝑥1 + 𝑥2 ) − 𝑓 (𝛾)
2
(𝑥0 + 𝑥1 + 𝑥2 − 𝛾)
𝑝 ⋅ .
2
− [∑𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 + 𝑥2 − 𝛾))] (29)
𝑖=1
𝑥1 + 𝑥2 + 𝑥3 = 𝑁 (𝑥0 + 𝑥1 + 𝑥2 )
2 (𝑓 (𝛾) + 𝑔 (𝑥0 + 𝑥1 + 𝑥2 ))
=− 𝑝 𝑝
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 + 𝑥2 − 𝛾)) + (𝑥0 + 𝑥1 + 𝑥2 − 𝛾) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 + 𝑥2 − 𝛾))
(30)
= (𝑥0 + 𝑥1 + 𝑥2 − 𝛾)
2𝑓 (𝑥0 + 𝑥1 + 𝑥2 )
− 𝑝 𝑝 .
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 + 𝑥2 − 𝛾)) + (𝑥0 + 𝑥1 + 𝑥2 − 𝛾) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 + 𝑥2 − 𝛾))
𝑥 ≈ 𝑋3 = 𝑥0 + 𝑥1 + 𝑥2 + 𝑥3 = 𝑐 + 𝑁 (𝑥0 + 𝑥1 + 𝑥2 )
𝑓 (𝛾) 2𝑓 (𝑥0 + 𝑥1 )
=𝛾− −
𝑓 (𝛾) 2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 − 𝛾)) + (𝑥0 + 𝑥1 − 𝛾) ∑𝑝𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 − 𝛾))
𝑝 (31)
2𝑓 (𝑥0 + 𝑥1 + 𝑥2 )
− 𝑝 𝑝 .
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 + 𝑥2 − 𝛾)) + (𝑥0 + 𝑥1 + 𝑥2 − 𝛾) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑥0 + 𝑥1 + 𝑥2 − 𝛾))
Using this formulation, we suggest the following three-step Algorithm 3. For a given 𝑥0 , compute the approximate solu-
iterative method for solving nonlinear equations (2). tion 𝑥𝑛+1 by the following iterative scheme:
Discrete Dynamics in Nature and Society 5
2𝑓 (𝑧𝑛 )
𝑥𝑛+1 = 𝑧𝑛 − 𝑝 𝑝 ,
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑧𝑛 − 𝑥𝑛 )) + (𝑧𝑛 − 𝑥𝑛 ) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑧𝑛 − 𝑥𝑛 ))
2𝑓 (𝑦𝑛 )
𝑧𝑛 = 𝑦𝑛 − 𝑝 𝑝 , (32)
2 ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑦𝑛 − 𝑥𝑛 )) + (𝑦𝑛 − 𝑥𝑛 ) ∑𝑖=1 𝑤𝑖 𝑓 (𝛾 + 𝜏𝑖 (𝑦𝑛 − 𝑥𝑛 ))
𝑓 (𝑥𝑛 )
𝑦𝑛 = 𝑥𝑛 − 𝑛 = 0, 1, 2, . . .
𝑓 (𝑥𝑛 )
8𝑓 (𝑦𝑛 )
𝑥𝑛+1 = 𝑦𝑛 −
2𝑓 (𝑥𝑛 ) + 6𝑓 ((𝑥𝑛 + 2𝑦𝑛 ) /3) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 (𝑥𝑛 ) + 3 (𝑦𝑛 − 𝑥𝑛 ) 𝑓 ((𝑥𝑛 + 2𝑦𝑛 ) /3)
(36)
𝑓 (𝑥 )
𝑦𝑛 = 𝑥𝑛 − 𝑛 𝑛 = 0, 1, 2, . . . .
𝑓 (𝑥𝑛 )
2𝑓 (𝑧𝑛 ) 2𝑓 (𝑧𝑛 )
𝑥𝑛+1 = 𝑧𝑛 − 𝑥𝑛+1 = 𝑧𝑛 −
2𝑓 (𝑥𝑛 ) + (𝑧𝑛 − 𝑥𝑛 ) 𝑓 (𝑥𝑛 ) 2𝑓 (𝑧𝑛 ) + (𝑧𝑛 − 𝑥𝑛 ) 𝑓 (𝑧𝑛 )
6 Discrete Dynamics in Nature and Society
2𝑓 (𝑦𝑛 ) 𝑧𝑛 = 𝑦𝑛
𝑧𝑛 = 𝑦𝑛 −
2𝑓 (𝑦𝑛 ) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 (𝑦𝑛 )
𝑓 (𝑥𝑛 ) 2𝑓 (𝑦𝑛 )
− ,
𝑦𝑛 = 𝑥𝑛 − , 𝑛 = 0, 1, 2, . . . . 2𝑓 ((𝑦𝑛 + 𝑥𝑛 ) /2) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 ((𝑦𝑛 + 𝑥𝑛 ) /2)
𝑓 (𝑥𝑛 )
(38) 𝑓 (𝑥𝑛 )
𝑦𝑛 = 𝑥𝑛 − , 𝑛 = 0, 1, 2, . . . .
Now, using 𝑝 = 1, 𝑤1 = 1 and 𝜏1 = 1/2, Algorithm 3 reduces 𝑓 (𝑥𝑛 )
to the following iterative method for solving 𝑓(𝑥) = 0. (39)
8𝑓 (𝑧𝑛 )
𝑥𝑛+1 = 𝑧𝑛 − ,
2𝑓 (𝑥𝑛 ) + 6𝑓 ((𝑥𝑛 + 2𝑧𝑛 ) /3) + (𝑧𝑛 − 𝑥𝑛 ) 𝑓 (𝑥𝑛 ) + 3 (𝑧𝑛 − 𝑥𝑛 ) 𝑓 ((𝑥𝑛 + 2𝑧𝑛 ) /3)
8𝑓 (𝑦𝑛 )
𝑧𝑛 = 𝑦𝑛 − , (40)
2𝑓 (𝑥𝑛 ) + 6𝑓 ((𝑥𝑛 + 2𝑦𝑛 ) /3) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 (𝑥𝑛 ) + 3 (𝑦𝑛 − 𝑥𝑛 ) 𝑓 ((𝑥𝑛 + 2𝑦𝑛 ) /3)
𝑓 (𝑥𝑛 )
𝑦𝑛 = 𝑥𝑛 − , 𝑛 = 0, 1, 2, . . . .
𝑓 (𝑥𝑛 )
To the best of our knowledge, Algorithms 8–11 are new From (41) and (42), we get
iterative methods for solving nonlinear equation (2).
𝑓 (𝑥𝑛 )
3. Convergence Analysis = 𝑒𝑛 − 𝑐2 𝑒𝑛2 + 2 (𝑐22 − 𝑐3 ) 𝑒𝑛3
𝑓 (𝑥𝑛 )
In this section, convergence criteria of newly suggested meth- + (−4𝑐23 + 7𝑐2 𝑐3 − 3𝑐4 ) 𝑒𝑛4
ods are studied in the form of the following theorem. (44)
+ (8𝑐24 − 20𝑐22 𝑐3 + 10𝑐2 𝑐4 + 6𝑐32 − 4𝑐5 ) 𝑒𝑛5
Theorem 12. Assume that the function 𝑓 : 𝐼 ⊂ R → R on
an open interval 𝐼 has a simple root 𝛼 ∈ 𝐼. Let 𝑓(𝑥) be suffi-
+ 𝑂 (𝑒𝑛6 ) .
ciently differentiable in the neighborhood of 𝛼; then the con-
vergence orders of the methods defined by Algorithms 2–11 are
3, 4, 3, 3, 4, 4, 4, 4, 6 and 6, respectively. Using (44), we have
𝑝 𝑝 𝑝
Now, the Taylor’s series expansions of ∑𝑖=1 𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑧𝑛 −
𝑝 ∑𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑦𝑛 − 𝑥𝑛 )) = 2𝑐2 + ∑𝑤𝑖 (−6𝜏𝑖 𝑐3
𝑥𝑛 )) and ∑𝑖=1 𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑧𝑛 − 𝑥𝑛 )) are given by 𝑖=1 𝑖=1
𝑝 𝑝 𝑝
∑𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑦𝑛 − 𝑥𝑛 )) = 1 − 2∑𝑤𝑖 (𝜏𝑖 𝑐2 − 𝑐2 ) 𝑒𝑛 + 6𝑐3 ) 𝑒𝑛 + ∑𝑤𝑖 (6𝜏𝑖 𝑐2 𝑐3 + 24𝜏𝑖2 𝑐4 − 48𝜏𝑖 𝑐4 + 24𝑐4 )
𝑖=1 𝑖=1 𝑖=1
𝑝 (48)
𝑝
⋅ 𝑒𝑛2 + ∑𝑤𝑖 (−12𝜏𝑖 𝑐22 𝑐3 − 48𝜏𝑖2 𝑐2 𝑐4 − 120𝜏𝑖3 𝑐5
+ ∑𝑤𝑖 (2𝜏𝑖 𝑐22 + 3𝜏𝑖2 𝑐3 − 6𝜏𝑖 𝑐3 + 3𝑐3 ) 𝑒𝑛2 𝑖=1
𝑖=1 (47)
+ 48𝜏𝑖 𝑐2 𝑐4 + 12𝜏𝑖 𝑐32 + 360𝜏𝑖2 𝑐5 − 360𝜏𝑖 𝑐5 + 120𝑐5 ) 𝑒𝑛3
𝑝
+ ∑𝑤𝑖 (−4𝜏𝑖 𝑐23 − 6𝜏𝑖2 𝑐2 𝑐3 − 4𝜏𝑖3 𝑐4 + 10𝜏𝑖 𝑐2 𝑐3 + 𝑂 (𝑒𝑛4 ) .
𝑖=1
Using (45), (47), and (48), we obtain the error term for
+ 12𝜏𝑖2 𝑐4 − 12𝜏𝑖 𝑐4 + 4𝑐4 ) 𝑒𝑛3 + 𝑂 (𝑒𝑛4 ) , Algorithm 2 as follows:
2𝑓 (𝑦𝑛 )
𝑧𝑛 = 𝑦𝑛 − 𝑝 𝑝
∑𝑖=1 𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑦𝑛 − 𝑥𝑛 )) + (𝑦𝑛 − 𝑥𝑛 ) ∑𝑖=1 𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑦𝑛 − 𝑥𝑛 ))
(49)
𝑝 𝑝
= 𝛼 + ∑𝑤𝑖 (−2𝑐22 𝜏𝑖 + 𝑐22 ) 𝑒𝑛3 + ∑𝑤𝑖 (−4𝜏𝑖2 𝑐23 + 10𝜏𝑖 𝑐23 + 3𝜏𝑖2 𝑐2 𝑐3 − 3𝑐23 − 7𝜏𝑖 𝑐2 𝑐3 + 2𝑐2 𝑐3 ) 𝑒𝑛4 + 𝑂 (𝑒5) .
𝑖=1 𝑖=1
𝑝 𝑝
From (49), we have
+ ∑𝑤𝑖 (3𝜏𝑖2 𝑐3 − 6𝜏𝑖 𝑐3 + 3𝑐3 ) 𝑒𝑛2 + ∑𝑤𝑖 (−4𝜏𝑖2 𝑐23
𝑖=1 𝑖=1
𝑝 𝑝
𝑓 (𝑧𝑛 ) = ∑𝑤𝑖 (−2𝑐22 𝜏𝑖 + 𝑐22 ) 𝑒𝑛3 + ∑𝑤𝑖 (−4𝜏𝑖2 𝑐23 + 2𝜏𝑖 𝑐23 − 4𝜏𝑖3 𝑐4 + 12𝜏𝑖2 𝑐4 − 12𝜏𝑖 𝑐4 + 4𝑐4 ) 𝑒𝑛3
𝑖=1 𝑖=1 + 𝑂 (𝑒𝑛4 ) ,
(50) 𝑝 𝑝
+ 10𝜏𝑖 𝑐23 + 3𝜏𝑖2 𝑐2 𝑐3 − 3𝑐23 − 7𝜏𝑖 𝑐2 𝑐3 + 2𝑐2 𝑐3 ) 𝑒𝑛4
∑𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑧𝑛 − 𝑥𝑛 )) = 2 + ∑𝑤𝑖 (−4𝜏𝑖 𝑐2 − 2𝑐2 )
𝑖=1 𝑖=1
+ 𝑂 (𝑒5) . 𝑝 𝑝
⋅ 𝑒𝑛 + ∑𝑤𝑖 (6𝜏𝑖2 𝑐3 − 6𝜏𝑖 𝑐3 ) 𝑒𝑛2 + ∑𝑤𝑖 (−8𝜏𝑖2 𝑐23
𝑝
The Taylor’s series expansions of 𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑧𝑛 − 𝑥𝑛 ))
∑𝑖=1 𝑖=1 𝑖=1
𝑝
and ∑𝑖=1 𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑧𝑛 − 𝑥𝑛 )) are given by + 2𝜏𝑖 𝑐23 − 8𝜏𝑖3 𝑐4 + 2𝑐23 + 24𝜏𝑖 𝑐4 − 16𝑐4 ) 𝑒𝑛3 + 𝑂 (𝑒𝑛4 ) .
(51)
𝑝 𝑝
∑𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑧𝑛 − 𝑥𝑛 )) = 1 − 2∑𝑤𝑖 (𝜏𝑖 𝑐2 − 𝑐2 ) 𝑒𝑛 Using (49) and (51), the error term for Algorithm 3 is obtained
𝑖=1 𝑖=1 as
2𝑓 (𝑧𝑛 )
𝑥𝑛+1 = 𝑧𝑛 − 𝑝 𝑝
∑𝑖=1 𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑧𝑛 − 𝑥𝑛 )) + (𝑧𝑛 − 𝑥𝑛 ) ∑𝑖=1 𝑤𝑖 𝑓 (𝑥𝑛 + 𝜏𝑖 (𝑧𝑛 − 𝑥𝑛 ))
(52)
𝑝
= 𝛼 + ∑𝑤𝑖 (4𝜏𝑖2 𝑐23 − 4𝜏𝑖 𝑐23 + 𝑐23 ) 𝑒𝑛4 + 𝑂 (𝑒𝑛5 ) .
𝑖=1
Now, we prove the convergence orders of the special cases of 2𝑓 (𝑥𝑛 ) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 (𝑥𝑛 )
Algorithms 2 and 3.
= 2 + 2𝑐2 𝑒𝑛 + 2𝑐22 𝑒𝑛2 + (−4𝑐23 + 10𝑐2 𝑐3 − 4𝑐4 ) 𝑒𝑛3
The Taylor’s series expansion of 2𝑓 (𝑥𝑛 )+(𝑦𝑛 −𝑥𝑛 )𝑓 (𝑥𝑛 )
is given by + (8𝑐24 − 26𝑐22 𝑐3 + 18𝑐2 𝑐4 + 12𝑐32 − 10𝑐5 ) 𝑒𝑛4
8 Discrete Dynamics in Nature and Society
+ 𝑂 (𝑒𝑛5 ) . 𝑥𝑛 + 𝑦𝑛
𝑓 ( ) = 2𝑐2 + 3𝑐3 𝑒𝑛 + (3𝑐2 𝑐3 + 6𝑐4 ) 𝑒𝑛2
2
(53)
+ (−6𝑐22 𝑐3 + 12𝑐2 𝑐4 + 6𝑐32 + 15𝑐5 ) 𝑒𝑛3 + (12𝑐22 𝑐3
Using (45), (46), and (53), the error term for Algorithm 4 is (58)
obtained as − 18𝑐22 𝑐4 − 21𝑐2 𝑐32 + 45𝑐2 𝑐5 + 33𝑐3 𝑐4 + 45𝑐6 ) 𝑒𝑛4
2𝑓 (𝑦𝑛 ) + 𝑂 (𝑒𝑛5 ) .
𝑧𝑛 = 𝑦𝑛 −
2𝑓 (𝑥𝑛 ) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 (𝑥𝑛 )
(54) Using (45), (46), (57), and (58), we obtain the error term for
=𝛼+ 𝑐22 𝑒𝑛3 + (−3𝑐23 + 2𝑐2 𝑐3 ) 𝑒𝑛4 + 𝑂 (𝑒𝑛5 ) . Algorithm 6 as follows:
𝑧𝑛 = 𝑦𝑛
Now, expending 2𝑓 (𝑦𝑛 ) + (𝑦𝑛 − 𝑥𝑛 )𝑓 (𝑦𝑛 ) by Taylor’s series,
we get 2𝑓 (𝑦𝑛 )
−
2𝑓 ((𝑥𝑛 + 𝑦𝑛 ) /2) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 ((𝑥𝑛 + 𝑦𝑛 ) /2) (59)
2𝑓 (𝑦𝑛 ) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 (𝑦𝑛 )
3
= 𝛼 + (𝑐23 − 𝑐2 𝑐3 ) 𝑒𝑛4 + 𝑂 (𝑒𝑛5 ) .
4
= 2 − 2𝑐2 𝑒𝑛 + 6𝑐22 𝑒𝑛2 + (−12𝑐23 + 6𝑐2 𝑐3 ) 𝑒𝑛3 (55)
Expanding 𝑓 ((𝑥𝑛 + 2𝑦𝑛 )/3) and 𝑓 ((𝑥𝑛 + 2𝑦𝑛 )/3) by Taylor’s
+ (24𝑐24 − 18𝑐22 𝑐3 + 18𝑐2 𝑐4 − 12𝑐32 ) 𝑒𝑛4 + 𝑂 (𝑒𝑛5 ) . series, we have
𝑥𝑛 + 2𝑦𝑛 2 4 1
Using equations (45), (46), and (55), the error term for 𝑓 ( ) = 1 + 𝑐2 𝑒𝑛 + ( 𝑐22 + 𝑐3 ) 𝑒𝑛2
3 3 3 3
Algorithm 5 is obtained as
8 4 16 32
+ (− 𝑐23 + 4𝑐2 𝑐3 + 𝑐4 ) 𝑒𝑛3 + ( 𝑐24 − 𝑐22 𝑐3
2𝑓 (𝑦𝑛 ) 3 27 3 3
𝑧𝑛 = 𝑦𝑛 −
2𝑓 (𝑦𝑛 ) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 (𝑦𝑛 ) 44 8 5
(56) + 𝑐 𝑐 + 𝑐2 + 𝑐 ) 𝑒4 + 𝑂 (𝑒𝑛5 ) ,
9 2 4 3 3 81 5 𝑛
= 𝛼 − 𝑐22 𝑒𝑛3 + (3𝑐23 − 2𝑐2 𝑐3 ) 𝑒𝑛4 + 𝑂 (𝑒𝑛5 ) .
𝑥𝑛 + 2𝑦𝑛 8
𝑓 ( ) = 2𝑐2 + 2𝑐3 𝑒𝑛 + (4𝑐2 𝑐3 + 𝑐4 ) 𝑒𝑛2 (60)
3 3
The Taylor’s series expansions of 𝑓 ((𝑥𝑛 +𝑦𝑛 )/2) and 𝑓 ((𝑥𝑛 +
𝑦𝑛 )/2) are given by 32 40
+ (−8𝑐22 𝑐3 + 8𝑐32 + 𝑐 𝑐 + 𝑐 ) 𝑒3 + (16𝑐23 𝑐3
3 24 9 5 𝑛
𝑥𝑛 + 𝑦𝑛 3
𝑓 ( ) = 1 + 𝑐2 𝑒𝑛 + (𝑐22 + 𝑐3 ) 𝑒𝑛2 + (−2𝑐23 100 32 80 80
2 4 − 28𝑐2 𝑐32 + 𝑐 𝑐 − 𝑐2 𝑐 + 𝑐 𝑐 + 𝑐 ) 𝑒4
3 34 3 2 4 3 25 9 6 𝑛
7 1 37 9
+ 𝑐2 𝑐3 + 𝑐4 ) 𝑒𝑛3 + (4𝑐24 − 𝑐22 𝑐3 + 𝑐2 𝑐4 + 3𝑐32 (57) + 𝑂 (𝑒𝑛5 ) .
2 2 4 2
5 Using (42), (43), (45), (46), and (60), the error term for
+ 𝑐 ) 𝑒4 + 𝑂 (𝑒𝑛5 ) ,
16 5 𝑛 Algorithm 7 is obtained as
8𝑓 (𝑦𝑛 )
𝑧𝑛 = 𝑦𝑛 −
2𝑓 (𝑥𝑛 ) + 6𝑓 ((𝑥𝑛 + 2𝑦𝑛 ) /3) + (𝑦𝑛 − 𝑥𝑛 ) 𝑓 (𝑥𝑛 ) + 3 (𝑦𝑛 − 𝑥𝑛 ) 𝑓 ((𝑥𝑛 + 2𝑦𝑛 ) /3) (61)
1
= 𝛼 + (𝑐23 − 𝑐2 𝑐3 ) 𝑒𝑛4 + 𝑂 (𝑒𝑛5 ) .
2
𝑓 (𝑧𝑛 ) = 1 − 2𝑐23 𝑒𝑛3 + (6𝑐24 − 4𝑐22 𝑐3 ) 𝑒𝑛4 + 𝑂 (𝑒𝑛5 ) , Therefore, using (59), (66), and (67), the error term for
Algorithm 10 is obtained as
𝑓 (𝑧𝑛 ) = 2𝑐2 − 6𝑐22 𝑐3 𝑒𝑛3 + (18𝑐23 𝑐3 − 12𝑐2 𝑐32 ) 𝑒𝑛4
𝑥𝑛+1 = 𝑧𝑛
+ 𝑂 (𝑒𝑛5 ) .
(64) 2𝑓 (𝑧𝑛 )
−
2𝑓 ((𝑧𝑛 + 𝑥𝑛 ) /2) + (𝑧𝑛 − 𝑥𝑛 ) 𝑓 ((𝑧𝑛 + 𝑥𝑛 ) /2) (68)
Using (56) and (64), the error term for Algorithm 9 is
3 9
obtained as = 𝛼 + (− 𝑐23 𝑐3 + 𝑐2 𝑐32 ) 𝑒𝑛6 + 𝑂 (𝑒𝑛7 ) .
4 16
2𝑓 (𝑧𝑛 )
𝑥𝑛+1 = 𝑧𝑛 −
2𝑓 (𝑧𝑛 ) + (𝑧𝑛 − 𝑥𝑛 ) 𝑓 (𝑧𝑛 ) (65)
Using (61), we have
8𝑓 (𝑧𝑛 )
𝑥𝑛+1 = 𝑧𝑛 −
2𝑓 (𝑥𝑛 ) + 6𝑓 ((𝑥𝑛 + 2𝑧𝑛 ) /3) + (𝑧𝑛 − 𝑥𝑛 ) 𝑓 (𝑥𝑛 ) + 3 (𝑧𝑛 − 𝑥𝑛 ) 𝑓 ((𝑥𝑛 + 2𝑧𝑛 ) /3)
(71)
3 7 1
= 𝛼 + ( 𝑐25 − 𝑐23 𝑐3 + 𝑐2 𝑐32 ) 𝑒𝑛6 + 𝑂 (𝑒𝑛7 ) .
4 8 4
435,000 𝜆
This complete the proof. 1,564,000 = 1,000,000𝑒𝜆 + (𝑒 − 1) , (72)
𝜆
𝜔𝑡 −𝜔𝑡
𝑔 𝑒 −𝑒
4. Numerical Examples 𝑥 (𝑡) = − 2 ( − sin 𝜔𝑡) . (73)
2𝜔 2
In this section, we demonstrate the validity and efficiency of
our proposed iterative schemes by considering the nonlinear We take 𝑥0 = 1.5 for (72) and 𝑥0 = −1.2 for (73) as initial guess
equations obtained from population model and the motion for computer program. Tables 1 and 2 and Figures 1 and 2 give
of a particle on an inclined plane, that is, the comparison of our newly proposed iterative methods, that
10 Discrete Dynamics in Nature and Society
20 (NR 1), ([18], Algorithm 2.7) (NR 2), ([18], Algorithm 2.12)
(NR 3), and ([18], Algorithm 2.13) (NR 4). In Table 3, we
15 present numerical results for various nonlinear equations
which also indicate that our methods are more efficient.
10 We use Maple18 for all numerical computations and
MATLAB 9 to plot the graphs. For the computer programs,
5 we use the following stopping criteria:
Log of residuals
0 (𝑥𝑛+1 − 𝑥𝑛 + 𝑓 (𝑥𝑛 )) < 𝜖, where 𝜖 = 10−15 . (74)
−5
5. Conclusions
−10
We have introduced a new family of iterative methods
(Algorithms 2 and 3), based on decomposition technique, for
−15
solving nonlinear equations using coupled system of equa-
tions. Several new iterative methods have been established as
−20
1 2 3 4 5 6 7 8 9 10 special cases of newly established family. We have explored
Iterations the convergence criteria of our new methods and investigated
for convergence order and efficiency index. We present the
AG1 NR2 comparative study both numerically and graphically (Tables
AG2 NR3
AG3 NR4
1 and 2 and Figures 1 and 2) of our newly constructed
AG4 NM methods with some known methods by considering two
NR1 real life models. In Table 3, we present numerical results by
considering various nonlinear equations.
Figure 1: Log of residuals for equation (72).
Conflicts of Interest
is, Algorithm 6 (AG 1), Algorithm 7 (AG 2), Algorithm 10
(AG 3), and Algorithm 11 (AG 4) with standard Newton’s The authors declare that there are no conflicts of interest
method (NM, (1)), Noor et al. methods ([18], Algorithm 2.6) regarding the publication of this paper.
Discrete Dynamics in Nature and Society 11
Table 3
𝑓(𝑥) 𝑥∘ Method Number of iterations 𝑥[𝑘] 𝑓(𝑥 𝑛) 𝛿
NM 7 3.0000000000000000007 9.100000𝑒 − 18 3.263396𝑒 − 10
NR1 7 3.0000000000000000007 9.105537𝑒 − 18 3.263396𝑒 − 10
NR2 7 3.0000000000000000007 9.105537𝑒 − 18 3.263396𝑒 − 10
NR3 5 3.0000000000000000000 0.000000𝑒 + 00 1.415161𝑒 − 15
𝑒( 𝑥2 + 7𝑥 − 30) − 1 2.9 NR4 5 3.0000000000000000000 0.000000𝑒 + 00 3.366886𝑒 − 27
AG1 4 3.0000000000000000000 2.921644𝑒 − 79 3.555566𝑒 − 21
AG2 4 3.0000000000000000000 8.793703𝑒 − 51 4.351958𝑒 − 14
AG3 3 3.0000000000000000000 0.000000𝑒 + 00 1.645921𝑒 − 14
AG4 3 3.0000000000000000000 0.000000𝑒 + 00 3.392389𝑒 − 14
NM 7 2.2041177331716202960 3.305103𝑒 − 20 1.295280𝑒 − 10
NR1 7 2.2041177331716202960 1.864247𝑒 − 22 7.456850𝑒 − 12
NR2 7 2.2041177331716202960 3.743730𝑒 − 20 1.033112𝑒 − 10
NR3 5 2.2041177331716202960 3.305103𝑒 − 20 1.194680𝑒 − 24
𝑥2 − 𝑒𝑥 + 𝑥 + 2 1.65 NR4 5 2.2041177331716202960 3.305103𝑒 − 20 7.482519𝑒 − 36
AG1 4 2.2041177331716202960 2.534748𝑒 − 56 1.035700𝑒 − 14
AG2 4 2.2041177331716202960 3.405915𝑒 − 48 1.073168𝑒 − 12
AG3 3 2.2041177331716203000 1.464912𝑒 − 17 3.775506𝑒 − 15
AG4 3 2.2041177331716203000 1.464912𝑒 − 17 2.401455𝑒 − 11
NM 8 −1.8414056604369606378 2.472561𝑒 − 19 8.481033𝑒 − 21
NR1 7 −1.8414056604369606378 1.469668𝑒 − 22 5.161458𝑒 − 12
NR2 8 −1.8414056604369606378 1.750316𝑒 − 40 5.508062𝑒 − 21
NR3 5 −1.8414056604369606378 2.472561𝑒 − 19 5.885759𝑒 − 25
(𝑥 + 2)𝑒( − 𝑥) − 1 −1.35 NR4 5 −1.8414056604369606378 2.472561𝑒 − 19 3.160149𝑒 − 36
AG1 4 −1.8414056604369606378 1.123854𝑒 − 58 2.235422𝑒 − 15
AG2 4 −1.8414056604369606378 2.782588𝑒 − 49 4.782839𝑒 − 13
AG3 3 −1.8414056604369606000 2.007912𝑒 − 16 6.603435𝑒 − 17
AG4 3 −1.8414056604369606000 2.007912𝑒 − 16 1.073115𝑒 − 11
NM 6 0.9977265554138634538 8.696898𝑒 − 21 7.838761𝑒 − 13
NR1 5 0.9977265554138634538 1.302709𝑒 − 21 3.286611𝑒 − 11
NR2 6 0.9977265554138634538 3.176527𝑒 − 37 4.745782𝑒 − 19
NR3 4 0.9977265554138634538 8.696898𝑒 − 21 5.066551𝑒 − 25
2𝑥2 − 1/2 sin𝑥 − 1/2𝜋 0.5 NR4 4 0.9977265554138634538 8.696898𝑒 − 21 3.934532𝑒 − 25
AG1 4 0.9977265554138634538 1.775184𝑒 − 99 2.198399𝑒 − 25
AG2 4 0.9977265554138634538 3.133051𝑒 − 99 2.528325𝑒 − 25
AG3 3 0.9977265554138634500 1.410674𝑒 − 17 5.046616𝑒 − 21
AG4 3 0.9977265554138634500 1.410674𝑒 − 17 4.537399𝑒 − 21
NM 8 0.9716662861175176121 2.044833𝑒 − 20 2.890826𝑒 − 20
NR1 7 0.9716662861175176121 1.855259𝑒 − 30 3.505396𝑒 − 16
NR2 7 0.9716662861175176121 3.234098𝑒 − 23 1.380792𝑒 − 12
NR3 5 0.9716662861175176121 2.044833𝑒 − 20 5.272663𝑒 − 24
8𝑥3 − 3𝑥2 sin𝑥 − 5 0.5 NR4 5 0.9716662861175176121 2.044833𝑒 − 20 4.036720𝑒 − 35
AG1 4 0.9716662861175176121 1.832541𝑒 − 52 5.721104𝑒 − 14
AG2 4 0.9716662861175176121 7.018615𝑒 − 46 2.443339𝑒 − 12
AG3 3 0.9716662861175176100 3.426161𝑒 − 17 5.515718𝑒 − 14
AG4 3 0.9716662861175176100 3.426161𝑒 − 17 4.503689𝑒 − 11
12 Discrete Dynamics in Nature and Society
International
Journal of
Mathematics and
Mathematical
Sciences
Journal of
Hindawi
Optimization
Hindawi
www.hindawi.com Volume 2018 www.hindawi.com Volume 2018
International Journal of
Engineering International Journal of
Mathematics
Hindawi
Analysis
Hindawi
www.hindawi.com Volume 2018 www.hindawi.com Volume 2018