AMP Advances in Mathematical Physics 1687-9139 1687-9120 Hindawi Publishing Corporation 10.1155/2014/694580 694580 Research Article A Matrix Method Based on the Fibonacci Polynomials to the Generalized Pantograph Equations with Functional Arguments Koç Ayşe Betül 1 Çakmak Musa 2 Kurnaz Aydın 1 Kurasov Pavel 1 Department of Mathematics, Faculty of Science, Selcuk University, 42003 Konya Turkey selcuk.edu.tr 2 Yayladağı Vocational School, Mustafa Kemal University, 31550 Hatay Turkey mku.edu.tr 2014 1382014 2014 12 04 2014 31 07 2014 13 8 2014 2014 Copyright © 2014 Ayşe Betül Koç 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.

A pseudospectral method based on the Fibonacci operational matrix is proposed to solve generalized pantograph equations with linear functional arguments. By using this method, approximate solutions of the problems are easily obtained in form of the truncated Fibonacci series. Some illustrative examples are given to verify the efficiency and effectiveness of the proposed method. Then, the numerical results are compared with other methods.

1. Introduction

Many phenomena in applied branches that fail to be modeled by the ordinary differential equations can be described by the delay differential equations. Many researchers have studied different applications of those equations in variety of applied sciences such as biology, physics, economy, and electrodynamics (see ). Pantograph equations with proportional delays play an important role in this context. The existence and uniqueness of the analytic solutions of the multipantograph equation are investigated in . A numerical approach to multipantograph equations with variable coefficients is also studied in . An extension of the multipantograph equation is known to be the generalized pantograph equation with functional arguments defined as (1) y ( m ) ( x ) = j = 0 J k = 0 m - 1 p j k ( x ) y ( k ) ( α j k x + β j k ) + g ( x ) , 00000000000000000000000 x I = [ a , b ] under the mixed conditions (2) k = 0 m - 1 c r k y ( k ) ( μ r k ) = λ r , r = 0 ( 1 ) ( m - 1 ) , 000000000000 a μ r k b , where proportional delay- α j k , constant delay- β j k , c r k , μ r k , and λ r are real and/or complex coefficients, and the coefficients of k th order unknown function- p j k and known g ( x ) are the analytical functions defined in the interval a x b .

In recent years, many researchers have developed different numerical approaches to the generalized pantograph equations as variational iteration method , differential transform approach , Taylor method , collocation method based on Bernoulli matrix , and Bessel collocation method . In this study, we investigate a collocation method based on the Fibonacci polynomial operational matrix for the numerical solution of the generalized pantograph equation (1). Even the Fibonacci numbers have been known for a long time; the Fibonacci polynomials are very recently defined to be an important agent in the world of polynomials [12, 13]. Compared to the methods of the orthogonal polynomials, the Fibonacci approach has proved to give more precise and reliable results in the solution of differential equations .

This study is organized as follows. In the second part, a short review of the Fibonacci polynomials is presented. A Fibonacci operational matrix for the solution of the pantograph equation is developed in Section 3. Some numerical examples are given in Section 4 to illustrate efficiency and effectiveness of the method.

2. Operational Matrices of the Fibonacci Polynomials

The Fibonacci polynomials F r ( x ) are determined by following general formula [12, 13]: (3) F r + 1 ( x ) = x F r ( x ) + F r - 1 ( x ) , for    r > 1 , with F 1 ( x ) = 1 and F 2 ( x ) = x . Now, we will mention some matrix relations in terms of Fibonacci polynomials.

2.1. Fibonacci Series Expansions

To obtain an expansion form of the analytic solution of the pantograph equation, we use the Fibonacci collocation method as follows.

Suppose that (1) has a continuous function solution that can be expressed in the Fibonacci polynomials (4) y ( x ) = r = 1 a r F r ( x ) . Then, a truncated expansion of N -Fibonacci polynomials can be written in the vector form (5) y N ( x ) = r = 1 N a r F r ( x ) = F ( x ) A , where the Fibonacci row vector F ( x ) and the unknown Fibonacci coefficients column vector A are given, respectively, by (6) F ( x ) = [ F 1 ( x ) F 2 ( x ) F N ( x ) ] 1 × N , A = [ a 1 a 2 a N ] T N × 1 .

2.2. Matrix Relations of the Derivatives

The k th order derivative of (5) can be written as (7) y N ( k ) ( x ) = r = 1 N a r ( k ) F r ( x ) = F ( x ) A ( k ) , 000000000000 k = 0,1 , . . . , n , where a r ( 0 ) = a r , y ( 0 ) ( x ) = y ( x ) , and (8) A ( k ) = [ a 1 ( k ) a 2 ( k ) a N ( k ) ] T is the coefficient vector of the polynomial approximation of k th order derivative. Then, there exists a relation between the Fibonacci coefficients as (9) A ( k + 1 ) = D k A , k = 0,1 , , n , where D is N × N operational matrix for the derivative defined by  (10) D = [ d i , j ] = { i · sin ( j - i ) π 2 ,    j > i 0 , j i . Making use of (7) and (9) yields (11) y N ( k ) ( x ) = F ( x ) D k A , k = 0,1 , , n .

3. Solution Procedure for the Pantograph Differential Equations

Let us recall the m th order linear pantograph differential equation, (12) y ( m ) ( x ) = j = 0 J k = 0 m - 1 p j k ( x ) y ( k ) ( α j k x + β j k ) + g ( x ) , 00000000000000000000000 x I = [ a , b ] . The first step in the solution procedure is to define the collocation points in the domain, so that (13) x i = a + b - a N - 1 ( i - 1 ) , i = 1,2 , , N , a x i b . Then, collocating problem (12) at the points in (13) yields (14) y ( m ) ( x i ) = j = 0 J k = 0 m - 1 p j k ( x i ) y ( k ) ( α j k x i + β j k ) + g ( x i ) , 000000000000000 i = 1 ( 1 ) N , m + 1 N . The system (14) can, alternatively, be rewritten in the matrix form (15) Y ( m ) - j = 0 J k = 0 m - 1 P j k Y - j k ( k ) = G , where (16) P j k = [ P j k ( x 1 ) 0 0 0 P j k ( x 2 ) 0 0 0 P j k ( x N ) ] , G = [ g ( x 1 ) g ( x 2 ) g ( x N ) ] T . Therefore, the k th order derivative of the unknown function at the collocation points can be written in the matrix form as (17) [ y N ( k ) ( x i ) ] = F ( x i ) D k A , i = 1 ( 1 ) N , or, equivalently, (18) Y ( k ) = [ y N ( k ) ( x 1 ) y N ( k ) ( x 2 ) y N ( k ) ( x N ) ] = F D k A . To express the functional terms of (1) as in the form (5), let we put α j k x i + β j k instead of x in the relation (18) and then obtain (19) [ y N ( k ) ( α j k x i + β j k ) ] = F - ( α j k x i + β j k ) D k A , i = 1 ( 1 ) N , or (20) Y - j k ( k ) = [ y N ( k ) ( α j k x 1 + β j k ) y N ( k ) ( α j k x 2 + β j k ) y N ( k ) ( α j k x N + β j k ) ]    = F - j k    D k A , where F - j k are Fibonacci operational matrices corresponding to the coefficients α j k . Therefore, replacing (18) and (20) in (15) gives the fundamental matrix equation for the problem (12) as (21) { F D m - j = 0 J k = 0 m - 1 P j k F - j k D k } A = G , which corresponds to a system of N algebraic equations for the unknown Fibonacci coefficients a r ,   r = 1,2 , . . . , N . In other words, when we denote the expression in the sum by W = [ w s , t ] , for s = 1,2 , . . . , N and    t = 1,2 , . . . , N , we get (22) W A = G . Thus, the augmented matrix of (22) becomes (23) [ W ; G ] .

On the other hand, in view of (11), the conditions (2) can be taken into account by forming the following matrix equation: (24) U j = [ k = 0 m - 1 c r k y ( k ) ( μ r k ) ] = [ λ r ] , where (25) U j = [ u ρ , σ ] = k = 0 m - 1 c r k F μ r k D k A , F μ r k = [ F 1 ( μ r k ) F 2 ( μ r k ) F N ( μ r k ) ]    . Therefore, the augmented matrix of the specified conditions is (26) [ U j    ; λ j ] = [ u j 1 u j 2 u j N : γ j ] . Consequently, (23) together with (26) can be written in the new augmented matrix form (27) [ W * : G * ] . This form can also be achieved by replacing some rows of the matrix (23) by the rows of (26) or adding those rows to the matrix (23) provided that det ( W * ) 0 . Finally, the vector A (thereby vector of the coefficients a r ) is determined by applying some numerical methods designed especially to solve the system of linear equations. On the other hand, when the singular case det ( W * ) = 0 appears, the least square methods are inevitably available to reach the best possible approximation. Therefore, the approximated solution can be obtained. This would be the Fibonacci series expansion of the solution to the problem (12) with the specified conditions.

3.1. Accuracy of the Results

We can, now, proceed with a short accuracy analysis of the problem in a similar way to . As the truncated Fibonacci series expansion is an approximate solution of (1) with (2), it must satisfy the following equality for x = x r [ a , b ] , r = 1 ( 1 ) N : (28) E ( x r ) = | y ( m ) ( x r ) - j = 0 J k = 0 m - 1 p j k ( x r ) y ( k ) ( α j k x r + β j k ) 000 - g ( x r ) j = 0 J k = 0 m - 1 p j k ( x r ) y ( k ) ( α j k x r + β j k ) | 0 , or (29) E ( x r ) 1 0 - k r ( k r is    any    positive    integer ) . When max ( 1 0 - k r ) = 1 0 - k ( k is any integer) is prescribed, the truncation limit N is increased until the difference E ( x r ) at each of the collocation points becomes smaller than the desired value 1 0 - k .

4. Numerical Results

In this part, three illustrative examples are given in order to clarify the findings of the previous section. The errors of the proposed method are compared with those of the errors that occurred in the solutions by some other methods in Tables 13 for two sample examples. It is noted here that the number of collocation points in the examples is indicated by the capital letter N .

Comparison of the absolute errors of different approximation techniques to Example 2.

x Present method Exponential approach  Taylor polynomial approach 
N = 5 N = 5 N = 5
0.2 0.2553 E - 05 0.32778 E - 02 0.271 E - 06
0.4 0.1965 E - 05 0.32081 E - 02 0.882 E - 05
0.6 0.3874 E - 05 0.44444 E - 02 0.682 E - 04
0.8 0.4833 E - 05 0.46898 E - 02 0.293 E - 03
1.0 0.2690 E - 04 0.12864 E - 01 0.912 E - 03

Comparison of the absolute errors of different approximation techniques to Example 2.

x Present method Taylor polynomial approach  Taylor method 
N = 9 N = 9 N = 8
0.2 0 0.271 E - 06 1.44 E - 12
0.4 0 0.882 E - 05 7.524 E - 10
0.6 0 0.682 E - 04 2.953 E - 08
0.8 0.1 E - 08 0.293 E - 03 4.018 E - 07
1.0 0.1 E - 08 0.912 E - 03 3.059 E - 06

Comparison of the absolute errors of different approximation techniques to Example 3.

x Present method Taylor matrix method  Boubaker matrix method 
N = 5 N = 9 N = 12 N = 5 N = 9 N = 9 N = 12
0.2 0.18903 E - 5 0.12102 E - 10 0.18730 E - 14 0.69082 E - 6 0.1300 E - 8 0.121 E - 10 0
0.4 0.62395 E - 6 0.96855 E - 11 0.14897 E - 14 0.42924 E - 4 0.1434 E - 6 0.968 E - 11 0.310 E - 14
0.6 0.13542 E - 5 0.71954 E - 11 0.11151 E - 14 0.47443 E - 3 0.2058 E - 5 0.719 E - 11 0.110 E - 14
0.8 0.15097 E - 5 0.68229 E - 11 0.68964 E - 15 0.25855 E - 2 0.1212 E - 4 0.682 E - 11 0.200 E - 14
1.0 0.47735 E - 4 0.75830 E - 9 0.13256 E - 12 0.95631 E - 2 0.4003 E - 4 0.758 E - 9 0.562 E - 12
Example 1 (see [<xref ref-type="bibr" rid="B10">10</xref>, <xref ref-type="bibr" rid="B11">11</xref>]).

Consider the following linear pantograph type problem equation: (30) y ′′ ( x ) = 3 4 y ( x ) + y ( x 2 ) - x 2 + 2 , 0 x 1 , with the initial conditions (31) y ( 0 ) = y ( 0 ) = 0 .

The exact solution of this problem is known to be y ( x ) = x 2 . When the solution procedure in Section 3 is applied to the problem, the solution of the linear algebraic system gives the numerical approximation of the solution to the problem. It is noteworthy that the method reaches the exact solution y ( x ) = x 2 even for N = 3 .

Example 2 (see [<xref ref-type="bibr" rid="B9">9</xref>, <xref ref-type="bibr" rid="B17">15</xref>, <xref ref-type="bibr" rid="B16">16</xref>]).

Now, consider the following equation with variable coefficient given by (32) y ( x ) = 1 2 e x / 2 y ( x 2 ) + 1 2 y ( x )    , 0 x 1 , and the condition (33) y ( 0 ) = 1 .

The exact solution is also known to be y ( x ) = exp ( x ) . A comparison of the absolute errors of the proposed approach Taylor method  and the exponential approach  is given in Table 1 for N = 5 . Another comparison of the present method with the methods of Taylor polynomials [9, 16] is also given for N = 9 and N = 8 in Table 2. These results show that the Fibonacci approach has better accuracy, at least one decimal place, than the other methods.

Example 3 (see [<xref ref-type="bibr" rid="B6">6</xref>, <xref ref-type="bibr" rid="B18">17</xref>]).

Finally, let us consider the pantograph equation with variable coefficients (34) y ( x ) = - y ( x ) - e - 0.5 x sin ( 0.5 x ) y ( 0.5 x ) - 2 e - 0.75 x cos ( 0.5 x ) sin ( 0.25 x ) y ( 0.25 x ) , 000000000000000000000000000 0 x 1 , and the condition (35) y ( 0 ) = 1 , which has the exact solution y ( x ) = e - x cos x . Computed results are compared with the results of the Taylor  and Boubaker  matrix methods in Table 3.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

Acknowledgments

This study was supported by Research Projects Center (BAP) of Selcuk University. The authors would like to thank Selcuk University and TUBITAK for their support. Also, the authors would like to thank the editor and referees for their valuable comments and remarks which led to a great improvement of the article. They have denoted here that a minor part of this study was presented orally at the “2nd International Eurasian Conference on Mathematical Sciences and Applications (IECMSA-2013),” Sarajevo, August, 2013.

Ajello W. G. Freedman H. I. Wu J. A model of stage structured population growth with density dependent time delay SIAM Journal on Applied Mathematics 1992 52 855 869 Kuang Y. Delay Differential Equations with Applications in Population Dynamics 1993 Academic Press Boston, Mass, USA Dehghan M. Shakeri F. The use of the decomposition procedure of Adomian for solving a delay differential equation arising in electrodynamics Physica Scripta 2008 78 6 065004 10.1088/0031-8949/78/06/065004 ZBL1159.78319 2-s2.0-58149468214 Ockendon J. R. Tayler A. B. The dynamics of a current collection system for an electric locomotive Proceedings of the Royal Society of London A 1971 322 447 468 10.1098/rspa.1971.0078 Liu M. Z. Li D. Properties of analytic solution and numerical solution of multi-pantograph equation Applied Mathematics and Computation 2004 155 3 853 871 10.1016/j.amc.2003.07.017 MR2078696 ZBL1059.65060 2-s2.0-3242732129 Sezer M. Yalcinbas S. Sahin N. Approximate solution of multi pantograph equation with variable coefficients Journal of Computational and Applied Mathematics 2008 214 2 406 416 10.1016/j.cam.2007.03.024 Saadatmandi A. Dehghan M. Variational iteration method for solving a generalized pantograph equation Computers & Mathematics with Applications 2009 58 11-12 2190 2196 10.1016/j.camwa.2009.03.017 MR2557348 2-s2.0-70350575223 Keskin Y. Kurnaz A. Kiris M. E. Oturanc G. Approximate solutions of generalized pantograph equations by the differential transform method International Journal of Nonlinear Sciences and Numerical Simulation 2007 8 2 159 164 2-s2.0-34249875994 Sezer M. Akyüz-Daşcıoğlu A. A Taylor method for numerical solution of generalized pantograph equations with linear functional argument Journal of Computational and Applied Mathematics 2007 200 1 217 225 10.1016/j.cam.2005.12.015 MR2276827 2-s2.0-33750956351 Tohidi E. Bhrawy A. H. Erfani K. A collocation method based on Bernoulli operational matrix for numerical solution of generalized pantograph equation Applied Mathematical Modelling. Simulation and Computation for Engineering and Environmental Systems 2013 37 6 4283 4294 10.1016/j.apm.2012.09.032 MR3020571 ZBL1273.34082 2-s2.0-84872607486 Yuzbası S. Sahin N. Sezer M. A Bessel collocation method for numerical solution of generalized pantograph equations Numerical Methods for Partial Differential Equations 2011 28 4 1105 1123 10.1002/num.20660 MR2914784 2-s2.0-84860129388 Falcón S. Plaza Á. The k -Fibonacci sequence and the Pascal 2-triangle Chaos, Solitons & Fractals 2007 33 1 38 49 10.1016/j.chaos.2006.10.022 MR2301846 2-s2.0-33846818865 Falcon S. Plaza A. On K-Fibonacci sequences and polynomials and their derivatives Chaos, Solitons & Fractals 2009 39 3 1005 1019 10.1016/j.chaos.2007.03.007 MR2512912 2-s2.0-62949102109 Koç A. B. Çakmak M. Kurnaz M. Uslu K. A new Fibonacci type collocation procedure for boundary value problems Advances in Difference Equations 2013 2013, article 262 10.1186/1687-1847-2013-262 MR3116124 Yuzbası S. Sezer M. An exponential approximation for solutions of generalized pantograph-delay differential equations Applied Mathematical Modelling 2013 37 22 9160 9173 10.1016/j.apm.2013.04.028 MR3115390 Sezer M. Yalçınbas S. Gülsu M. A Taylor polynomial approach for solving generalized pantograph equations with nonhomogenous term International Journal of Computer Mathematics 2008 85 7 1055 1063 10.1080/00207160701466784 MR2431785 Akkaya T. Yalçinbaş S. Sezer M. Numeric solutions for the pantograph type delay differential equation using First Boubaker polynomials Applied Mathematics and Computation 2013 219 17 9484 9492 10.1016/j.amc.2013.03.021 MR3047844 2-s2.0-84876574103 Baykus N. Sezer M. Solution of high-order linear Fredholm integro-differential equations with piecewise intervals Numerical Methods for Partial Differential Equations 2011 27 5 1327 1339 10.1002/num.20587 MR2833941 2-s2.0-79960810578