Sei sulla pagina 1di 9

Available online at www.sciencedirect.

com

Applied Mathematics and Computation 193 (2007) 446454 www.elsevier.com/locate/amc

Charpit and Adomian for solving integral equations


Alice Gorguis
Mathematics Department, North Park University, 3225 W. Foster Avenue, Chicago, IL 60625, United States

Abstract In this paper, we consider the nonlinear partial dierential equations of rst order involving two independent variables. Geometrically the complete solution exists in an innite variety. Applying Charpits Method, we can get a two-parameter family of integral surfaces by passing integral surfaces through a suitably chosen two-parameter family of curves. Also, briey, we will present an easier method of solution called Adomian decomposition method. 2007 Elsevier Inc. All rights reserved.
Keywords: Nonlinear partial dierential equations; Charpits Method; Adomian method; Complete integrals

1. Introduction In this work, we will discus nonlinear partial dierential equations of the rst order, and various important types of integrals possessed by them. We will consider the general method, due to Charpit, for solving nonlinear equations involving two independent variables, and also will discuss the method of solving these equations using the decomposition method, which can be used generally for all types of dierential and integral equations. We will see how the decomposition method can be applied in a straightforward manner and how it provides a rapidly convergent series. 2. Analysis of Charpits method Consider a rst order nonlinear, and homogeneous partial dierential equation of the form, F x ; y ; z ; p ; q 0;
oz oz where, x, y are independent variables, and z is a dependent variable, p  o ,qo . x y To solve (1), we will determine a second equation of the same type,

G x ; y ; z ; p ; q 0; such that (1) and (2) can be solved for p and q, in terms of x, y, and z.

E-mail address: agorguis@northpark.edu 0096-3003/$ - see front matter 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.amc.2007.03.078

A. Gorguis / Applied Mathematics and Computation 193 (2007) 446454

447

The resulting expression will make px; y ; z dx qx; y ; z dy dz 0; 3 integrable. In order that (1) and (2) be solvable for p and q, these relations must be independent, and therefore their Jacobian J oF ; G F p Gq F q Gp o p ; q 4

cannot vanish identically. In what follows we will assume that J 5 0. The necessary and sucient condition that the total dierential equation P x; y ; z dx Qx; y ; z dy Rx; y ; z dz 0 be integrable is that       oQ oR oR oP oP oQ P Q R oz oy ox oz oy ox vanish identically. Hence if (3) is to be integrable, where (P = p, Q = q, R = 1), we must have p oq op op o q q 0 oz oz oy ox 7 5

assuming that both p and q in (1) and (2) are functions of x, y, and z. Dierentiating (1) partially with respect to x gives oF oF op oF oq 0: ox op ox o q ox Dierentiating (2) partially with respect to x gives oG oG op oG oq 0: ox op ox o q ox
q Now, solving (8) and (9) for o , then substituting in Eq. (7) gives ox

oq F x G p F p G x ; ox J op F q Gy F y Gq ; oy J

J 6 0:

10

In the same manner, we will dierentiate (1) and (2) with respect to y and solve for J 6 0: 11

Also, dierentiating (1) and (2) partially with respect to z gives Fz Fp op oq Fq 0; oz oz op oq Gz Gp Gq 0: oz oz


op oz

12

Then solving (12) for

and

oq oz

gives

op F q Gz F z Gq ; oz J oq F z Gp F p Gz ; oz J

J 6 0; 13 J 6 0:
op oz

q op Inserting the expressions for o , , ox oy

and

oq oz

in Eq. (7) and multiplying throughout by J we get 14

pF z Gp F p Gz qF q Gz F z Gq F q Gy F y Gq F x Gp F p Gx 0:

448

A. Gorguis / Applied Mathematics and Computation 193 (2007) 446454

Rearranging (14) results in F x pF z oG oG oG oG oG F y qF z pF p qF q Fp Fq 0: op oq oz ox oy 15

Eq. (15) is a six-dimensional homogeneous linear partial dierential equation for the determination of (G) as a function of x, y, z, p, and q. The subsidiary equations for (15) are dp dq dz dx dy ; F x pF z F y qF z pF p qF q F p F q dG 0: 16

Eq. (16) is called Charpit subsidiary equation. Therefore, to solve Eq. (1) we have to form the subsidiary equation (16) rst, then try to nd one integral of (16), the simpler the better, that contains p and q or both, say u x ; y ; z ; p ; q a; 17 where a = arbitrary constant. This gives the desired relation (2), namely, G  u a = 0. We then solve (1) and (17) simultaneously for p and q as a function of x, y, z, and a, then substitute in Eq. (3), and integrate to get f x; y ; z; a; b 0; 18 where b is a second arbitrary constant. Eq. (18) with the two arbitrary constants will serve as a Complete Integral for Eq. (1). In what follows, we shall examine some equations of special form for which standard methods, obtainable from the general Charpits procedure, may be formulated. 2.1. Equations of the form F(p, q) = 0 Consider a nonlinear equation of the form F p ; q 0; 19 containing p and q but none of the variables, x, y, and z. Since Fx, Fy, and Fz = 0, the denominators of the rst two ratios in the Charpits subsidiary equation (16) will vanish identically. Therefore, we have dp = 0 and dq = 0, we may thus use either value of p = a, together with q obtained from F(a, q) = 0, or use q = a together with the value of p obtained from F(p, a) = 0, whichever is more convenient. 2.2. Equations of the form F(z, p, q) = 0 Consider a nonlinear equation of the form F z ; p ; q 0; in which the independent variables x and y do not appear. Since Fx = 0 and Fy = 0, Charpits subsidiary equations (16) will reduce to dp dq dz dx dy : pF z qF z pF p qF q F p F q From the rst two ratios, we get q ap : Combining (20) and (22), we can nd p and q in terms of z, p f z; q ap af z: Therefore, dz = p dx + q dy yields dz dx a dy ; f z Z dz b x ay : f z 20

21

22 23 24 25

A. Gorguis / Applied Mathematics and Computation 193 (2007) 446454

449

Thus it appears that, in a complete integral (25), z will be a function of the combination x + ay. We may use this fact to formulate a method of solving an equation of type (20) by setting z g x ay g u ; so that p og dg ; ox du q og dg a : oy du 27 26

Then the partial dierential Eq. (20) is transformed into the following ordinary dierential equation of rst order:   dz dz F z; ; a 0: 28 du du The substitution u = x + ay has the Geometric interpretation of rotating the coordinate axes in the xy p plane through the angle arctan a, and stretching the x and y-coordinates in the ratio 1 a2 : 1. Since Eq. (25) represents a cylinder, the surfaces given by a complete integral of an equation of type (20) will therefore be cylinders with elements at an angle arctan a with the x-axes and parallel to the xy-plane. 2.3. Equations of the form f(x, p) = g(y, q) Consider a nonlinear partial dierential equation of the form f x ; p g y ; q : Letting F = f(x, p) g(y, q), we will then have F x fx ; F y gy ; F z 0; F y fp ; F q g q ; 29

so that Charpit subsidiary equation (16) becomes dp dq dz dx dy : fx gy pfp qgq fp gq From the rst and fourth of these ratios, we have fx dx fp dp 0; and from the second and fth gy dy gq dq 0: 32 31 30

But the left sides of (31) and (32) are respectively, the total dierentials of df and dq of the functions f(x, p) and g(y, q). Hence f(x, p) = constant and f(y, q) = constant, and by the original Eq. (29) these constants are equal. Therefore, f x ; p a; g y ; q a: 33

The procedure of solving an equation of type (29) is thus clear, we set each side of (29) equal to an arbitrary constant a, and solve the resulting pair of Eq. (33) for p in terms of x and a and for q in terms of y and a. We then insert these expressions for p and q in p dx + q dy dz = 0, to get a total dierential equation dz px; a dx qy ; a dy : Then by integration we get Z Z z px; a dx qy ; a dy b; which is the complete integral of the problem (29). 34

35

450

A. Gorguis / Applied Mathematics and Computation 193 (2007) 446454

2.4. Equations of the form z = xp + yq + f(p, q) If we let F  xp yq f p; q z 0; then F x p; so that F x pF z 0 and F y qF z 0: 38 F y q; F z 1; F p x fp ; f q y fq ; 37 36

Consequently the Charpit subsidiary equation (16) yields dp = 0 and dq = 0, so that we may take either p = constant or q = constant in conjunction with the given equation. 3. Applications In this section, we shall illustrate Charpits method through dierent examples. Example 1. Find a complete integral of the nonlinear partial dierential equation q2 2 q 3 p 1 : 39

Since the second denominator of the subsidiary equation (16) is Fy + qFz = 0, therefore we have dq = 0 and q = a. Then substituting q = a in Eq. (38), we get a2 2 a 3 p 1 : Solving for p gives p 1 2a a2 : 3 41 40

Substituting in p dx q dy dz 0; then integrating and solving for z gives 3 z 1 2 a a 2 x 3 a y b; which is the required complete integral equation. Example 2. Find a complete integral for the nonlinear partial dierential equation pq p q 0: 44 This equation is of the form F(p, q) = 0. From the rst ratio of Charpit subsidiary equation, we get dp = 0 and p = a. Substitute this in (43) and solve for q to get a : 45 q a1 Then substitute in p dx q dy dz 0; which can be simply integrated and solved for z, a 1z aa 1x ay b: Eq. (47) is the complete integral equation. 47 46 43 42

A. Gorguis / Applied Mathematics and Computation 193 (2007) 446454

451

Example 3. Solve the nonlinear equation for a complete integral p 2 q2 z 4 : 48

To solve this example which is of the type F(z, p, q), we chose z = g(x + ay) = g(u), where u = x + ay. This gives og dz ; ox du og dz a : q oy du p By substitution of (52) and (53) into the original problem(51), we get  2  2 dz 2 dz a z4 ; du du  2 dz 1 a 2 z4 ; du  2 dz z4 ; du 1a dz z2 p ; du 1 a2 dz z2 p : z2 1 a2 Integrating both sides of Eq. (58) gives p 2 z2 x ay b 1 a 1 a2 : Replacing u with x + ay in Eq. (59) gives the desired complete integral z2 x ay b 1 a2 : Example 4. Find the complete integral of the given nonlinear partial dierential equation yp x2 q2 x2 y : Rewriting the equation into the form f(x, p) = g(y, q), we get p q2 y a: 2 x y Solving for p and q separately to get p ax 2 and q p y a 1 : 60 61 62 59 58
2

49 50

51 52 53 54 55

56

57

Substituting both Eqs. (53) and (54) into dz p dx p dy ; and integrating gives the complete integral equation 3z ax3 3b2 4a 1y 3 : Example 5. Solve for the integral equation z xp yq 2p2 3q2 : 64 63

452

A. Gorguis / Applied Mathematics and Computation 193 (2007) 446454

This equation is of the form z = xp + yq + f(p, q). To solve this type of equations, let F  xp + yq 2p2 3q2 z = 0. We have Fx = p, Fy = , Fz = 1, Fp = x 4p , Fq = y 6q. Whence Fx + pFz = 0 and Fy + qFz = 0. Consequently Charpits subsidiary equation (16) yields dp = 0, p = a, and dq = 0, q = b. Substituting in (57) gives the complete integral z ax by 2a2 3b2 : Dierentiating with respect to a and b gives x y and b : a 4 6 Substitute in Eq. (58) and simplify to get the integral equation 24z 3x2 2y 2 : 67 65

66

Notice that one can easily verify that Eq. (67) satises Eq. (65), and consequently Eq. (67) is a singular integral equation. Even though Charpits method seems to be simple, but that is after the subsidiary equation is formed, which requires a lengthy step. Next, we will see how Adomian method can prove that the problems of nonlinear differential equations can be solved without using any helping substituting equation. 3.1. Adomians method and its application In this section we will give a general description to the Adomian decomposition method applied to a rst order nonlinear partial dierential equation, then it will be followed by an example to illustrate the method. Consider the nonlinear partial dierential equation of the form ux x; y uy x; y Rux; y F ux; y 0: In operator form, Eq. (68) can be written as Lx ux; y Ly ux; y Rux; y F ux; y 0; 69 68

where Lx is the highest order dierential in x, Ly is the highest order in y, R contains the remaining terms of lower derivatives and F(u(x, y)) is an analytic nonlinear term. The solutions for u(x, y) obtained from the operator equations Lx and Ly are called partial solutions, these partial solutions are equivalent and each converges to the exact solution [2]. However, to solve the problem, a decision on selecting one of the two operators should be made, that is the one which minimizes the size of computation, and has the best known condition, to accelerate the evaluation of the components of the solution. Suppose the best operator that meets the conditions is Lx, then Eq. (69) can be written as Lx ux; y Ly ux; y Rux; y F ux; y : Assuming that the operator is invertible, the inverse operator can be applied,
1 1 1 1 L x Lx ux; y Lx Ly ux; y RLx ux; y Lx F ux; y ; Rx 1 where Lx oox and L x 0 dx, then Eq. (71) can be written as 1 1 ux; y u0; y L x Ly ux; y Lx fRux; y F ux; y g:

70

71

72

In the series form, the solution u(x, y) is 1 X un x ; y : u x ; y


n 0

73

A. Gorguis / Applied Mathematics and Computation 193 (2007) 446454

453

Applying (73) to Eq. (71) gives


1 X n0

un x; y u0; y

1 L x Ly

1 X n0

! un x ; y
1 L x

1 X n0

! un x ; y
1 L x

1 X n0

! An ; 74

where F un x; y
1 X n0

An :

75

The components un(x, y), n P 0 of the solution u(x, y) can be recursively determined, u0 x; y u0; y ;
1 1 1 uk1 x; y L x Ly uk Lx Ruk Lx Ak ;

k P 0:

76

Then, using the Algorithm for the Adomian polynomials An [1] for the nonlinear term F(u(x, y)), the components can be identied by u0 x; y u0; y ;
1 1 1 u1 x; y L x Ly u0 x; y Lx Ru0 x; y Lx A0 ; 1 1 1 u2 x; y L x Ly u1 x; y Lx Ru1 x; y Lx A1 ; ::::

77

where each component can be determined by using the preceding component. Then the solution in series form will be obtained as
1 X n0

un x; y u0 u1 u2

78

Example. Solve the nonlinear, non-homogeneous rst order partial dierential equation given as q2 2q 3p 0; and p where q Then Eq. (79) can be written as u2 y 2uy ux 0: Then Eq. (80) in operator form will be 3Lx u 2Ly u Ly u;
1 where Lx oox ; Ly ooy . Then applying the inverse operator L x to both sides of Eq. (81), we get 1 1 2 82 ux; y u0; y 2=3L x Ly u 1=3Lx Ly u : R x 1 where L x 0 dx. The decomposition method identies the unknown function u(x, y) as an innite number of components, un(x, y), n P 0 given by 1 X un x ; y : 83 u x ; y n0 o , oy o . ox

79

80

81

In terms of (83), Eq. (82) will be written as


1 1 u0 u 1 u 2 u 0 ; y 2 = 3 L x fLy u0 u1 u2 g 1=3Lx fLy u0 u1 u2 g:

84

454

A. Gorguis / Applied Mathematics and Computation 193 (2007) 446454

4. Discussion The main goal of this work was to propose Charpits method for solving the integral equations in a very simple and nice way. Charpits method has worked eectively to handle the presented standard models. But using Adomian Decomposition Method to solve theses problems is even easier. References
[1] D. Zwillinger, Hand Book of Dierential Equations, Academic Press, Inc., New York, 1989. [2] Wazwaz, Partial Dierential Equations, Methods and Applications, Balkema Publishers, 2002.

Further reading
[1] L.R. Ford, Dierential Equations, McGraw and Hill, 1955. [2] F.H. Miller, Partial Dierential Equations, John Willy and Sons, Inc., 1956. [3] C.R. Chester, Techniques in Partial Dierential Equations, McGraw-Hill, New York, 1970, p. 212 and Chapter 15 (pp. 315337).

Potrebbero piacerti anche