Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Keywords: In this paper, a numerical power series algorithm which is based on the improved Taylor
Nonlinear differential equations matrix method is introduced for the approximate solution of Abel-type differential equations
Taylor matrix method
and also, Riccati differential equations. The technique is defined and illustrated with some
Power series method
numerical examples. The obtained results reveal that the method is very effective, simple and
Abel-type equations
valid high accuracy. The method can be easily extended to other nonlinear equations.
© 2015 Elsevier Inc. All rights reserved.
1. Introduction
Nonlinear differential equations are frequently used to model a wide class of problems in many scientific fields such as
engineering, chemical reactions, mathematical physics, astrophysics, biology, ecology and economics. Most of these equations
have no analytic solution and numerical techniques may be required to obtain approximate solutions. Thus, methods of solution
for nonlinear differential equations are of great importance to engineers and scientists. In recent years, the attention has been
focused on nonlinear problems; one of the most fundamental equations in the study of the nonlinear problems is nonlinear
Abel-type first order differential equation which is defined in the general form [11–16],
A1 (x)y (x) + A2 (x)y(x)y (x) + A3 (x)y(x) + A4 (x)y2 (x) + A5 (x)y3 (x) = G(x), a≤x≤b (1)
∗
Corresponding author. Tel.: +90 (236) 2013202.
E-mail addresses: bbulbul@mu.edu.tr (B. Bülbül), mehmet.sezer@cbu.edu.tr, msezer54@gmail.com (M. Sezer).
http://dx.doi.org/10.1016/j.amc.2015.04.057
0096-3003/© 2015 Elsevier Inc. All rights reserved.
170 B. Bülbül, M. Sezer / Applied Mathematics and Computation 262 (2015) 169–177
Chebyshev (matrix and collocation) methods to solve linear and nonlinear differential, integral and integro-differential equations
have been presented in many articles by Sezer and co-workers [17-27].
Our purpose in this study is to develop the above mentioned methods and apply to the generalized nonlinear Abel-type
differential equation (1), which is a generalized case of the first order nonlinear Abel equations given in [11–15], with the initial
condition
y(0) = α (2)
and to find the solution in terms of power series (Taylor polynomial form in origin)
N
y(n)(0)
y(x) = yn xn , yn = , 0≤x≤b (3)
n!
n=0
which is Taylor polynomial of degree N at x = 0, where yn , n = 0, 1, . . . , N are unknown Taylor coefficients to be determined.
2. Fundamental relations
In this section we convert the expressions defined in (1)–(3) to the matrix forms by the following procedure: Firstly, the
function y(x) defined by (3) can be written in the matrix form
y(x) = X(x)Y, (4)
where
X(x) = 1 x x2 . . . xN , Y = [y0 y1 . . . yN ]T .
On the other hand, it is clearly seen that the relation between the matrix X(x) and its derivative X (x) is
X (x) = X(x)B, (5)
where
⎡ ⎤
0 1 0 ··· 0
⎢0 0 2 ··· 0⎥
⎢ ⎥
⎢ .. ⎥ .
B = ⎢ ... ..
.
..
.
..
. . ⎥
⎢ ⎥
⎣0 0 0 ··· N⎦
0 0 0 ··· 0
From the matrix equations (4) and (5), it follows that
y (x) = X (x)Y = X(x)BY. (6)
By using the production of two series, the matrix form of expression y2 (x) is obtained as
y2 (x) = X̃(x)Ỹ, (7)
where
X̃(x) = X̃0 (x)X̃1 (x) . . . X̃N (x) ,
1×(N+1)2
X̃i (x) = xi x0 xi x1 · · · xi xN or X̃i (x) = xi X(x),
and
T
Ỹ = Ỹ0 Ỹ1 . . . ỸN , Ỹi = [yi y0 yi y1 · · · yi yN ]T , i = 0, 1, . . . , N.
(N+1)2 ×1
In a similar way, we will have
≈ ≈
y3 (x) = X(x) Y (8)
where
≈ ≈ ≈ ≈ ≈
X = X0 (x) X1 (x) . . . XN (x) X(x) = x X̃(x).
i
,
1×(N+1)3 i
≈ ≈ ≈ ≈ T ≈
Y = Y0 Y1 · · · YN , Y = yi Ỹ, i = 0, 1, . . . , N.
(N+1) ×1
3 i
where
X̃ (x) = X̃0 (x) X̃1 (x) . . . X̃N (x) , (10)
⎡ ⎤
0 1 0 ··· 0 ⎡ ⎤
0+i 0 ··· 0
⎢0 0 2 ··· 0⎥
⎢ ⎥ ⎢0 1 + i ··· 0 ⎥
⎢ .. ⎥ ; ⎢ ⎥
and B0 = ⎢ ... ..
.
..
.
..
. . ⎥ Bi = ⎢ . .. .. .. ⎥ i = 1, . . . , N.
⎢ ⎥ ⎣ .. . . . ⎦
⎣0 0 0 ··· N⎦
0 0 ··· N+i
0 0 0 ··· 0
Note that here X̃0 (x) = X(x) and B0 = B. Thus expression (10) becomes
where
K(x) = X̃0 (x) X̃0 (x) X̃1 (x) . . . X̃N−1 (x) ,
1×(N+1)2
⎡ ⎤
B0 0 ··· 0
⎢0 B1 ··· 0 ⎥
⎢ ⎥
B̃ = ⎢ . .. .. .. ⎥ .
⎣ .. . . . ⎦
0 0 ··· BN (N+1)2 ×
(N+1)2
Consequently, by substituting the matrix relation (11) into Eq. (9), we obtain the matrix relation for y (x)y(x) as
1
y (x)y(x) = K(x)B̃Ỹ (12)
2
which corresponds to a system of (N + 1) nonlinear algebraic equations with the unknown Taylor coefficientsyn , n = 0, 1, . . . , N.
The matrices in Eq. (14) are as follows:
W = [wp,q ] = A1 XB + A3 X, p, q = 0, 1, . . . , N,
1
V = [vr,s ] = A4 X̃ + A2 KB̃, r = 0, 1, . . . , N, s = 0, 1, . . . , (N + 1)2 − 1,
2
≈
Z = [zm,n ] = A5 X, m = 0, 1, . . . , N, n = 0, 1, . . . , (N + 1)3 − 1
and
⎡ ⎤ ⎡ ⎤
Aj (x0 ) 0 ··· 0 K(x0 )
⎢0 Aj (x1 ) ··· 0 ⎥ ⎢ K(x1 ) ⎥
⎢ ⎥ ⎢ ⎥
Aj = ⎢ . .. .. .. ⎥, ( j = 1, 2, 3, 4, 5), K = ⎢ . ⎥,
⎣ .. . . . ⎦ ⎣ .. ⎦
0 0 ··· Aj (xN ) K(xN )
172 B. Bülbül, M. Sezer / Applied Mathematics and Computation 262 (2015) 169–177
⎡ ⎤
⎡ ⎤ ⎡ ⎤ ≈
⎡ ⎤
X(x0 ) X̃(x0 ) ⎢≈X (x 0 ) ⎥ G(x0 )
⎢ X(x1 ) ⎥ ⎢ X̃(x ) ⎥ ⎢ ⎥ ⎢ G(x1 ) ⎥
⎢ ⎥ ⎢ 1 ⎥ ≈ ⎢ X(x1 ) ⎥ ⎢ ⎥
X = ⎢ . ⎥, X̃ = ⎢
⎢ .. ⎥ ,
⎥ X=⎢ . ⎥
⎢
⎥, G = ⎢ . ⎥,
⎣ .. ⎦ ⎣ . ⎦ ⎢ .. ⎥ ⎣ .. ⎦
⎣ ⎦
X(xN ) X̃(xN ) ≈ G(xN )
X(xN )
The augmented matrix for Eq. (14) becomes, clearly,
⎡ ⎤
w0,0 w0,1 ··· w0,N ; v0,0 v0,1 ··· v0,(N+1)2 −1 ; z0,0 z0,1 ··· z0,(N+1)3 −1 ; G(x0 )
⎢ w1,0 w1,1 ··· w1,N ; v1,0 v1,1 ··· v1,(N+1)2 −1 ; z1,0 z1,1 ··· z1,(N+1)3 −1 ; G(x1 ) ⎥
⎢ ⎥
[W; V; Z : G] = ⎢
⎢ .. .. .. .. .. .. .. .. .. .. .. .. .. .. ⎥⎥ (15)
⎣ . . . . . . . ··· . . . . ··· . . . ⎦
wN,0 wN,1 ··· wN,N ; vN,0 vN,1 ··· vN,(N+1)2 −1 ; zN,0 zN,1 ··· zN,(N+1)3 −1 ; G(xN )
Next, by means of relation (4), we can obtain the corresponding matrix form for the initial condition (2) as
X(0)Y = α or [1 0 · · · 0] Y = α (16)
Consequently, to obtain the Taylor coefficients yn , n = 0, 1, . . . , N, related with the approximate solution of the problem
consisting of Eq. (1) and condition (2), by replacing the row matrix (16) by the last row of the augmented matrix (15), we have
new augmented matrix
⎡ ⎤
w0,0 w0,1 · · · w0,N ; v0,0 v0,1 ··· v0,(N+1)2 −1 ; z0,0 z0,1 ··· z0,(N+1)3 −1 ; G(x0 )
⎢ w1,0 w · · · w1,N ; v1,0 v1,1 ··· v1,(N+1)2 −1 ; z1,0 z1,1 ··· z1,(N+1)3 −1 ; G(x1 ) ⎥
⎢ ⎥
1,1
⎢ ⎥
W̃; Ṽ; Z̃ : G̃ = ⎢ . . . . . . . . . . . ⎥.
⎢ . . . . . . . . . . . ⎥
⎢ . . . . ; . . ··· . ; . . ··· . ; . ⎥
⎣ w(N−1),0 w(N−1),1 · · · w(N−1),N ; v(N−1),0 v(N−1),1 · · · v(N−1),(N+1)2 −1 ; z(N−1),0 z(N−1),1 · · · z(N−1),(N+1)3 −1 ; G(xN−1 ) ⎦
1 0 0 0 ; 0 0 ··· 0 ; 0 0 0 0 ; α
(17)
4. Accuracy of solution
We can easily check the accuracy of this solution as follows. Since the Taylor polynomial (3) is an approximate solution
of Eq. (1), then the solution y(x) is substituted in Eq. (1), the resulting equation must be satisfied approximately; that is for
x = xi ∈ [a, b], i = 0, 1, 2, . . .
E(xi ) = A1 (xi )y (xi ) + A2 (xi )y(xi )y (xi ) + A3 (xi )y(xi ) + A4 (xi )y2 (xi ) + A5 (xi )y3 (xi ) − G(xi ) ∼
=0
EN (x) = A1 (x)y (x) + A2 (x)y(x)y (x) + A3 (x)y(x) + A4 (x)y2 (x) + A5 (x)y3 (x) − G(x)
As EN (x) → 0 when N is sufficiently large enough, then the error decreases.
Also if we know the exact solution of problem we can find error bound of method.
5. Illustrative examples
In this section, several numerical examples are given to illustrate the properties of the method and all of them were performed
on the computer using a program written in Maple 9. The absolute errors in Tables 1–3 are the values of |y(x) − yN (x)| at selected
points.
Table 1
Comparison of the absolute errors of Example 1.
0 0 0 0 0 0
0.1 1.00E-10 1.40E-9 4.66E-11 3.98E-14 2.43E-17
0.2 7.50E-9 8.64E-8 6.80E-11 7.72E-14 2.73E-17
0.3 7.93E-8 9.70E-7 4.28E-11 3.45E-14 4.03E-17
0.4 4.11E-7 5.37E-6 2.29E-11 4.52E-14 6.03E-17
0.5 1.45E-6 2.02E-5 3.23E-11 3.65E-14 9.41E-17
0.6 4.00E-6 5.96E-5 4.03E-11 8.80E-14 1.75E-17
0.7 9.34E-6 1.48E-4 1.28E-12 1.23E-14 5.29E-18
0.8 1.92E-5 3.26E-4 4.41E-11 2.48E-15 2.35E-18
0.9 3.62E-5 6.52E-4 1.87E-11 5.68E-14 1.08E-17
1 6.33E-5 1.21E-3 1.43E-9 3.12E-12 4.56E-15
(a) (b)
Fig. 1. (a) Error function for N = 12. (b) Comparison of the approximate solution for N = 12 with exact solution.
Table 1 shows absolute errors of Example 1 with N = 8, 10, 12 by presented method. The previous results in ref. [15] are also
given in Table 1. The current method seems more rapidly convergent than the method in [15]. We display a plot of error function
in Fig. 1(a). Fig. 1(b) is for comparison of numerical solution with the exact solution.
The solutions obtained for N = 5, 8, 10 are compared with the exact solution y(x) = sin(x) [20] (see Table 1 and Fig. 1).
Table 2
Comparison of the absolute errors of Example 2.
N=5 N=8
N ≥ 10
Fig. 2. Comparison of exact and numerical solution for different values of N.
B. Bülbül, M. Sezer / Applied Mathematics and Computation 262 (2015) 169–177 175
Table 3
Comparison of the absolute errors of Example 3.
xi Euler method Runga–Kutta Adomian The method by [21] Present method (N = 20)
(a) (b)
Fig. 3. (a) Absolute error for Example 3. (b) Error function for Example 3.
6. Conclusion
In this paper, we have suggested a power series approach method for solving the Abel equations by using an improved Taylor
matrix technique. Our method is very simple as compared with other methods. Also all examples show that the method is a
powerful mathematical tool for solving Abel equations. The proposed numerical approach has many applications, and thus may
be applied in a similar ways to other nonlinear problems to get very accurate and satisfactory results.
Appendix
> #example
∗
> Diff(y(x),x) y(x) + x∗ y(x) + y(x)^2 + x^2∗ y(x)^3 = x∗ exp(-x) + x^2∗ exp(-3∗ x);
> Diff(y(x),x)∗ y(x) + x∗ y(x) + y(x)^2 + (yx^2)(x)^3 = x∗ exp(-x) + x^2∗ exp(-3∗ x);
> #initial condition
> y(0) = 1;
> restart;
> with(linalg);
> Digits: = 25;
> n: = 8;
> x: = array(0..n):
> for i from 0 to n do x[i]: = i/n od:
> L: = array(1..n + 1,1..n + 1);
> for i to n + 1 do for j to n + 1 do L[i,j]: = 1 od; od;
> for i to n + 1 do for j to n + 1 do L[i,j]: = (x[i-1])^(j-1) od; od;
> print(L);
> p1: = x->x;
> P1: = diag(seq(evalf(p1(x[i])),i = 0..n));
> p2: = x->x^2;
176 B. Bülbül, M. Sezer / Applied Mathematics and Computation 262 (2015) 169–177
> print(A2K);
> A3K: = array(1..n + 1,1..1);
> for i to n do A3K[i,1]: = A3[i,1] od;A3K[n + 1,1]: = 0;
> print(A3K);
> A4K: = array(1..n + 1,1..1);
> for i to n do A4K[i,1]: = A4[i,1] od;A4K[n + 1,1]: = 0;
> print(A4K);
> print(WK);
> s: = evalm(A1 + A2K + A3K + A4K);
> fsolve({ seq( s[i,1] = GG[i,1], i = 1..n + 1)}, {seq(y[i], i = 0..n)});
> u: = {%};
> K: = eval(YN,u);
> evalm(T&∗ K);
> approcimatesolution: = (x)->1.000000000000000000000000-1.000000000000000000000000∗ x + .4999999810695402
689844478∗ x^2-.1666663975425214493374945∗ x^3 + .4166501013618614115230115e-1∗ x^4-
.832774256154972190877
8264e-2∗ x^5 + .1377770669897782807599873e-2∗ x^6-.1853100486209861248858419e-3∗ x^7 +
.1613088592772372209
789256e-4∗ x^8
References
[1] M.K. Mak, New method for generating general solution of Abel differential equation, Comput. Math. Appl. 43 (2002) 91–94.
[2] J.P.M. Lebrun, On two coupled Abel-type differential equations arising in a magnetostatic problem, Il Nuovo Cimento 103A (1990) 1369–1379.
[3] F. Borghero, A. Melis, On Szebehely’s problem for holonomic systems involving generalized potential functions, Celestial Mech. Dyn. Astron. 49 (1990)
273–284.
[4] S.Ş. Bayin, Solutions of Einstein’s field equations for static fluid spheres, Phys. Rev. D 18 (1978) 2745–2751.
[5] A. Garcia, A. Macias, E.W. Mielke, Stewart-Lyth second order approach as an Abel equation for reconstructing inflationary dynamics, Phys. Lett. A 229 (1997)
32–36.
[6] V.R. Gavrilov, V.D. Ivashchuk, V.N. Melnikov, Multidimensional integrable vacuum cosmology with two curvatures, Class. Quantum Gravity 13 (1996)
3039–3056.
[7] G. Haager, M. Mars, A self-similar inhomogeneous dust cosmology, Class. Quantum Gravity 15 (1998) 1567–1580.
[8] M.K. Mak, T. Harko, Full causal bulk-viscous cosmological models, J. Math. Phys 39 (1998) 5458–5476.
[9] M.K. Mak, T. Harko, Exact causal viscous cosmologies, Gen. Relativ. Gravit 30 (1998) 1171–1186.
[10] Y. Matsuno, Two-dimensional dynamical system associated with Abel’s nonlinear differential equation, J. Math. Phys 33 (1992) 412–421.
[11] M.K. Mak, Solutions generating technique for Abel-type nonlinear ordinary differential equations, Comput. Math. Appl. 41 (2001) 1395–1401.
[12] T. Harko, M.K. Mak, Relativistic dissipative cosmological models and differential equation, Comput. Math. Appl. 46 (2003) 849–853.
[13] A. Garcia, A. Macias, E.W. Mielke, Stewart-Lyth second-order approach as Abel equation for reconstructing inflationary dynamics, Phys. Lett. A 229 (1997)
32–36.
[14] M.J. Alvarez, A. Gasull, H. Giacomini, A new uniqueness criterion for the number of periodic orbits Abel equations, J. Diff. Eq. 234 (2007) 161–176.
[15] C. Güler, A new numerical algorithm for the Abel equation of the second kind, Int. J. Comput. Math 84 (2007) 109–119.
[16] M.A.M. Alwash, The composition conjecture for Abel equation, Expo. Math. 27 (2009) 241–250.
[17] M. Sezer, A method for approximate solution of the second order linear differential equations in terms of Taylor polynomials, Int. J. Math. Educ. Sci. Technol.
27 (1996) 821–834.
[18] H. Sorkun, S. Yalçinbaş, Approximate solutions of linear Volterra integral equation systems with variable coefficients, Appl. Math. Model. 34 (11) (2010)
3451–3464.
[19] M. Gulsu, M. Sezer, The approximate solution of high-order linear difference equation with variable coefficients in terms of Taylor polynomials, Appl. Math.
Comput. 168 (2005) 76–88.
[20] S. Yalçinbaş, Taylor polynomial solutions of nonlinear Volterra-Fredholm integral equations, Appl. Math. Comput. 127 (2002) 196–206.
[21] M. Gülsu, M. Sezer, On the solution of the Riccati equation by the Taylor matrix method, Appl. Math. Comput. 188 (2007) 446–449.
[22] N. Kurt, M. Sezer, Polynomial solution of high-order linear Fredholm integro-differential equations with constant coefficients, J. Franklin Inst. 345 (2008)
839–850.
[23] M. Sezer, M. Gülsu, B. Tanay, A Taylor collocation method for the numerical solution a complex differential equations with mixed conditions in elliptic
domains, Appl. Math. Comput. 182 (2006) 498–508.
[24] S. Yalçinbaş, M. Sezer, The approximate solution of high-order linear Volterra-Fredholm integro-differential equations in terms of Taylor polynomials, Appl.
Math. Comput. 112 (2000) 291–308.
[25] A. Akyüz, M. Sezer, A Chebyshev collocation method for the solution of linear integro differential equations, Int. J. Comput. Math 72 (1999) 491–507.
[26] M. Gülsu, M. Sezer, A method for the approximate solution of the high-order linear difference equations in terms of Taylor polynomials, Int. J. Comput.
Math. 82 (5) (2005) 629–642.
[27] B. Bülbül, M. Gülsu, M. Sezer, A new Taylor collocation method for nonlinear Fredholm-Volterra integro-differential equations, Numer. Methods Partial Diff.
Eq. 26 (2009) 1006–1020.
[28] S.A. Khuri, Laplace decomposition algorithm applied to a class of nonlinear differential equations, J. Appl. Math. 1 (4) (2001) 141–155.