Abstract

The variational iteration algorithm combined with the exp-function method is suggested to solve the generalized Benjamin-Bona-Mahony equation (BBM) with variable coefficients. Periodic and soliton solutions are formally derived in a general form. Some particular cases are considered.

1. Introduction

The BBM equation

𝑒𝑑+𝑒𝑒π‘₯+𝑒π‘₯βˆ’πœ‡π‘’π‘₯π‘₯𝑑=0,(1.1) which describes approximately the unidirectional propagation of long waves in certain nonlinear dispersive systems, has been proposed by Benjamin et al. in 1972 [1] as a more satisfactory model than the KdV equation [2]

𝑒𝑑+𝑒𝑒π‘₯+𝑒π‘₯π‘₯π‘₯=0.(1.2) It is easy to see that (1.1) can be derived from the equal width EW-equation [3]:

𝑒𝑑+𝑒𝑒π‘₯βˆ’πœ‡π‘’π‘₯π‘₯𝑑=0,(1.3) by means of the change of variable 𝑒=𝑒+1, that is, by replacing 𝑒 with 𝑒+1. This last equation is considered as an equally valid and accurate model for the same wave phenomena simulated by (1.1) and (1.2). On the other hand, some researches analyzed the generalized KdV equation with variable coefficients

𝑒𝑑+𝜎(𝑑)𝑒𝑝𝑒π‘₯+πœ‡(𝑑)𝑒π‘₯π‘₯π‘₯=0,(1.4) because this model has important applications in several fields of science [4–7].

Motivated by these facts, we will consider here the generalized EW-equation with variable coefficients

𝑒𝑑+𝜎(𝑑)𝑒𝑝𝑒π‘₯βˆ’πœ‡(𝑑)𝑒π‘₯π‘₯𝑑=0.(1.5) Using the solutions of (1.5) we obtain exact solutions to the generalized BBM equation

𝑒𝑑+𝜎(𝑑)(𝑒+1)𝑝𝑒π‘₯βˆ’πœ‡(𝑑)𝑒π‘₯π‘₯𝑑=0,(1.6) of order 𝑝>0.

2. Exact Solutions to Generalized BBM Equation

2.1. The Variational Iteration Method

Consider the following nonlinear equation:

𝐿𝑒(π‘₯,𝑑)+𝑁𝑒(π‘₯,𝑑)=𝑔(π‘₯,𝑑),(2.1) where 𝐿 and 𝑁 are linear and nonlinear operators, respectively, and 𝑔(π‘₯,𝑑) is an inhomogeneous term. According to the variational iteration method (VIM) [8–14], a functional correction to (2.1) is given by

𝑒𝑛+1(π‘₯,𝑑)=𝑒𝑛(ξ€œπ‘₯,𝑑)+𝑑0ξ€·πœƒ(𝜏)𝐿𝑒𝑛(π‘₯,𝜏)+𝑁̃𝑒𝑛(ξ€Έπ‘₯,𝜏)βˆ’π‘”(π‘₯,𝜏)π‘‘πœ,(2.2) where πœƒ(𝜏) is a general Lagrange's multiplier, which can be identified via the variational theory; the subscript 𝑛β‰₯0 denotes the 𝑛th order approximation and ̃𝑒 is a restricted variation which means 𝛿̃𝑒=0. In this method, we first determine the Lagrange multiplier πœƒ(𝜏) that will be identified optimally via integration by parts. The successive approximation 𝑒𝑛+1 of the solution 𝑒 will be readily obtained upon using the determined Lagrangian multiplier and any selective function 𝑒0. One of the advantages of the VIM, is the free choice of the initial solution 𝑒0(π‘₯,𝑑). If we consider a special form to 𝑒0 with arbitrary parameters, using the relations

𝑒𝑛(π‘₯,𝑑)=𝑒𝑛+1πœ•(π‘₯,𝑑),π‘˜πœ•π‘‘π‘˜π‘’π‘›πœ•(π‘₯,𝑑)=π‘˜πœ•π‘‘π‘˜π‘’π‘›+1(π‘₯,𝑑),(2.3) we can obtain a set of algebraic equations in the unknowns given by the parameters that appear in 𝑒0. Solving this system, we have exact solutions to (2.1). To solve (1.5), we construct the following functional equation

𝑒𝑛+1(π‘₯,𝑑)=𝑒𝑛(ξ€œπ‘₯,𝑑)+𝑑0ξ€·πœƒ(𝜏)𝐿𝑒𝑛(π‘₯,𝜏)+𝑁̃𝑒𝑛(ξ€Έπ‘₯,𝜏)π‘‘πœ,(2.4) where

𝐿𝑒𝑛𝑒(π‘₯,𝜏)=π‘›ξ€Έπœ(π‘₯,𝜏),𝑁̃𝑒𝑛(π‘₯,𝜏)=𝜎(𝜏)(̃𝑒+1)𝑝̃𝑒π‘₯(π‘₯,𝜏)βˆ’πœ‡(𝜏)̃𝑒π‘₯π‘₯𝜏(π‘₯,𝜏).(2.5) Taking in (2.4) variation with respect to the independent variable 𝑒𝑛, and noticing that 𝛿𝑁̃𝑒𝑛=0 we have

𝛿𝑒𝑛+1(π‘₯,𝑑)=𝛿𝑒𝑛(ξ€œπ‘₯,𝑑)+𝛿𝑑0ξ€·πœƒ(𝜏)𝐿𝑒𝑛(π‘₯,𝜏)+𝑁̃𝑒𝑛(ξ€Έπ‘₯,𝜏)π‘‘πœ=𝛿𝑒𝑛(π‘₯,𝑑)+πœƒ(𝑑)π›Ώπ‘’π‘›ξ€œ(π‘₯,𝑑)βˆ’π‘‘0πœƒξ…ž(𝜏)𝛿𝑒𝑛(π‘₯,𝜏)π‘‘πœ=0.(2.6) This yields the stationary conditions

πœƒ1+πœƒ(𝑑)=0,ξ…ž(𝑑)=0.(2.7) Therefore,

πœƒ(𝑑)=βˆ’1.(2.8) Substituting this value into (2.4) we obtain the formula

𝑒𝑛+1(π‘₯,𝑑)=𝑒𝑛(ξ€œπ‘₯,𝑑)βˆ’π‘‘0𝐿𝑒𝑛(π‘₯,𝜏)+𝑁̃𝑒𝑛(ξ€Έπ‘₯,𝜏)π‘‘πœ.(2.9)

Using the wave transformation

πœ‰=π‘₯+πœ†π‘‘+πœ‰0,(2.10) setting

πœ•π‘’πœ•π‘‘1πœ•(πœ‰)=π‘’πœ•π‘‘0(πœ‰),(2.11) and performing one integration, (2.9) reduces to

πœ†π‘’0(πœ‰)+𝜎(𝑑)𝑒𝑝+10𝑝+1(πœ‰)βˆ’πœ†πœ‡(𝑑)𝑒0ξ…žξ…ž(πœ‰)=0,(2.12) where for sake of simplicity we set the constant of integration equal to zero. With the change of variable

𝑒0(πœ‰)=𝑣2/𝑝(πœ‰),(2.13) equation (2.12) converts to

πœ†π‘£2(πœ‰)βˆ’2πœ‡(𝑑)(2βˆ’π‘)𝑝2πœ†ξ€·π‘£ξ…žξ€Έ2βˆ’2πœ‡(𝑑)π‘πœ†π‘£(πœ‰)π‘£ξ…žξ…ž(πœ‰)+𝜎(𝑑)𝑝+1𝑣(πœ‰)4=0.(2.14) Observe that if 𝑣(πœ‰) is a solution to (2.14), then βˆ’π‘£(πœ‰) is also a solution to this equation.

2.2. The Exp-Function Method

Recently, He and Wu [15] have introduced the Exp-function method to solve nonlinear differential equations. In particular, the Exp-function method is an effective method for solving nonlinear equations with high nonlinearity. The method has been used in a satisfactory way by other authors to solve a great variety of nonlinear wave equations [15–21]. The Exp-function method is very simple and straightforward, and can be briefly revised as follows: Given the nonlinear partial differential equation

𝐹𝑒,𝑒π‘₯,𝑒𝑑,𝑒π‘₯π‘₯,𝑒π‘₯𝑑,𝑒𝑑𝑑,…=0,(2.15) it is transformed to ordinary differential equation

𝐹𝑒,π‘’ξ…ž,π‘’ξ…žξ…ž,π‘’ξ…žξ…žξ…ž,𝑒π‘₯𝑑,…=0,(2.16) by mean of wave transformation πœ‰=π‘₯+πœ†π‘‘+πœ‰0. Solutions to (2.16) can then be found using the expression

βˆ‘π‘’(πœ‰)=𝑑𝑛=βˆ’π‘π‘Žπ‘›exp(π‘›πœ‰)βˆ‘π‘žπ‘›=βˆ’π‘π‘π‘›,exp(π‘›πœ‰)(2.17) where 𝑐,𝑑,𝑝, and π‘ž are positive integers which are unknown to be determined later, π‘Žπ‘› and 𝑏𝑛 are unknown constants.

After balancing, we substitute (2.17) into (2.16) to obtain an algebraic systems in the variable 𝜁=exp(π‘›πœ‰). Solving the algebraic system we can obtain exact solutions to (2.16) and reversing, solutions to (2.15) in the original variables.

3. Solutions to (2.14) by the Exp-Function Method

Using the Exp-function method, we suppose that solutions to (2.14) can be expressed in the form

βˆ‘π‘£(πœ‰)=1𝑛=βˆ’1π‘Žπ‘›exp(π‘›π‘Ÿπœ‰)βˆ‘1π‘š=βˆ’1π‘π‘š=π‘Žexp(π‘šπ‘Ÿπœ‰)βˆ’1exp(βˆ’π‘Ÿπœ‰)+π‘Ž0+π‘Ž1exp(π‘Ÿπœ‰)π‘βˆ’1exp(βˆ’π‘Ÿπœ‰)+𝑏0+𝑏1exp(π‘Ÿπœ‰).(3.1) We obtain following solutions to (2.14): 𝑣1=Β±2πœ†π‘˜(𝑝+1)(𝑝+2)√2(𝑝+1)(𝑝+2)πœ†exp𝑝/2ξ‚πœ‰ξ‚πœ‡(𝑑)βˆ’π‘˜2ξ‚€βˆ’ξ‚€βˆšπœŽ(𝑑)exp𝑝/2ξ‚πœ‰ξ‚π‘£πœ‡(𝑑),πœ†=πœ†(𝑑),2=Β±2πœ†π‘˜(𝑝+1)(𝑝+2)𝜎(𝑑)π‘˜2√exp𝑝/2ξ‚πœ‰ξ‚ξ‚€βˆ’ξ‚€βˆšπœ‡(𝑑)βˆ’2πœ†(𝑝+1)(𝑝+2)exp𝑝/2ξ‚πœ‰ξ‚πœ‡(𝑑),πœ†=πœ†(𝑑).(3.2) Some special solutions are obtained if

π‘˜πœ†=πœ†(𝑑)=Β±22𝑝2ξ€Έ+3𝑝+2𝜎(𝑑).(3.3) This choice gives solutions

𝑣3π‘˜=Β±2𝑝csch2βˆšπœ‰ξƒͺπ‘˜πœ‡(𝑑),πœ†=22𝑝2𝑣+3𝑝+2𝜎(𝑑),(3.4)4=π‘˜2𝑝sech2βˆšπœ‰ξƒͺπ‘˜πœ‡(𝑑),πœ†=βˆ’22𝑝2𝑣+3𝑝+2𝜎(𝑑),(3.5)5π‘˜=Β±2𝑝csc2βˆšπœ‡πœ‰ξƒͺπ‘˜(𝑑),πœ†=βˆ’22𝑝2ξ€ΈπœŽ+3𝑝+2(𝑑).(3.6) Solution (3.6) follows from (3.4) with the identifications πœ‡(𝑑)β†’βˆ’πœ‡(𝑑) and βˆšπ‘˜β†’βˆ’π‘˜βˆ’1.

𝑣6π‘˜=βˆ’2𝑝sec2βˆšπœ‰ξƒͺπ‘˜πœ‡(𝑑),πœ†=βˆ’22𝑝2ξ€Έ+3𝑝+2𝜎(𝑑).(3.7) Solution (3.7) follows from (3.4) with the identifications πœ‡(𝑑)β†’βˆ’πœ‡(𝑑) and π‘˜β†’βˆ’π‘˜.

4. Particular Cases

4.1. Case 1: Solutions to (2.14) When 𝑝=2

Equation (2.14) takes the form

πœ†π‘£2(πœ‰)βˆ’πœ†πœ‡(𝑑)𝑣(πœ‰)π‘£ξ…žξ…ž1(πœ‰)+3𝜎(𝑑)𝑣(πœ‰)4=0.(4.1) From (3.2) with 𝑝=2:

𝑣7=Β±24πœ†π‘˜βˆš24πœ†expξ‚€ξ‚€1/πœ‡ξ‚πœ‰ξ‚(𝑑)βˆ’π‘˜2πœŽξ‚€βˆ’ξ‚€βˆš(𝑑)exp1/πœ‡ξ‚πœ‰ξ‚,𝑣(𝑑)8=Β±24πœ†π‘˜π‘˜2√𝜎(𝑑)expξ‚€ξ‚€1/ξ‚πœ‰ξ‚ξ‚€βˆ’ξ‚€βˆšπœ‡(𝑑)βˆ’24πœ†exp1/ξ‚πœ‰ξ‚.πœ‡(𝑑)(4.2) From (3.3)–(3.7) with 𝑝=2:

𝑣9π‘˜=Β±21cschβˆšπœ‰ξƒͺπ‘˜πœ‡(𝑑),πœ†=2𝑣24𝜎(𝑑),10π‘˜=Β±21sechβˆšπœ‰ξƒͺπ‘˜πœ‡(𝑑),πœ†=βˆ’2𝑣24𝜎(𝑑),11π‘˜=Β±21cscβˆšπœ‰ξƒͺπ‘˜βˆ’πœ‡(𝑑),πœ†=βˆ’2πœŽπ‘£24(𝑑),12π‘˜=Β±21secβˆšπœ‰ξƒͺπ‘˜πœ‡(𝑑),πœ†=βˆ’224𝜎(𝑑).(4.3) Other exact solutions are:

𝑣13ξ‚€=Β±3π‘Ž2ξ‚€2√expξ‚βˆšβˆ’2/πœ‡(𝑑)πœ‰+2ξ‚€βˆš55π‘Žexpξ‚ξ‚π‘˜βˆ’2/πœ‡(𝑑)πœ‰βˆ’223π‘Ž2ξ‚€2√expξ‚ξ‚€βˆšβˆ’2/πœ‡(𝑑)πœ‰+22π‘Žexp1βˆ’2/πœ‡(𝑑)πœ‰+22,πœ†=βˆ’3π‘˜2π‘£πœŽ(𝑑),14ξ‚€=Β±3π‘Ž2ñ2ξ‚€βˆš55π‘Žexp2βˆšβˆ’2/πœ‡(𝑑)πœ‰βˆ’22expπ‘˜βˆ’2/πœ‡(𝑑)πœ‰ξ‚ξ‚3π‘Ž2ξ‚€βˆš+22π‘Žexp2βˆšβˆ’2/πœ‡(𝑑)πœ‰+22exp1βˆ’2/πœ‡(𝑑)πœ‰,πœ†=βˆ’3π‘˜2π‘£πœŽ(𝑑),15βŽ›βŽœβŽœβŽξ‚€βˆš=Β±π‘˜1βˆ’448+55ξ‚€βˆš3π‘Ž11+ξ‚ξ‚€βˆš55expξ‚ξ‚€βˆšβˆ’2/πœ‡(𝑑)πœ‰+228+ξ‚βŽžβŽŸβŽŸβŽ 155,πœ†=βˆ’3π‘˜2π‘£πœŽ(𝑑),16π‘˜ξ‚€ξ‚€βˆš=Β±π‘ŽΒ±sinhβˆ’2/πœ‡(𝑑)πœ‰ξ‚ξ‚βˆšπ‘Ž2ξ‚€βˆš+1Β±cosh1βˆ’2/πœ‡(𝑑)πœ‰,πœ†=βˆ’3π‘˜2π‘£πœŽ(𝑑),17π‘˜ξ‚€ξ‚€βˆš=Β±π‘ŽΒ±coshβˆ’2/πœ‡(𝑑)πœ‰ξ‚ξ‚βˆšπ‘Ž2ξ‚€βˆš+1Β±sinh1βˆ’2/πœ‡(𝑑)πœ‰,πœ†=βˆ’3π‘˜2π‘£πœŽ(𝑑),18ξ‚€βˆš=Β±π‘˜cos2/πœ‡(𝑑)πœ‰ξ‚€βˆš1Β±sinξ‚π‘˜2/πœ‡(𝑑)πœ‰,πœ†=23𝜎(𝑑).(4.4)

4.2. Case 2: Solutions to (2.14) When 𝑝=4

Equation (2.14) takes the form

πœ†π‘£2𝑣(πœ‰)+πœ‡(𝑑)πœ†ξ…žξ€Έ2βˆ’πœ†2πœ‡(𝑑)𝑣(πœ‰)π‘£ξ…žξ…ž1(πœ‰)+5𝜎(𝑑)𝑣(πœ‰)4=0.(4.5) From (3.2) with 𝑝=4:

𝑣19=Β±60πœ†π‘˜βˆš60πœ†expξ‚€ξ‚€2/πœ‡ξ‚πœ‰ξ‚(𝑑)βˆ’π‘˜2πœŽξ‚€βˆ’ξ‚€βˆš(𝑑)exp2/πœ‡ξ‚πœ‰ξ‚π‘£(𝑑),πœ†=πœ†(𝑑),20=Β±60πœ†π‘˜πœŽ(𝑑)π‘˜2√expξ‚€ξ‚€2/ξ‚πœ‰ξ‚ξ‚€βˆ’ξ‚€βˆšπœ‡(𝑑)βˆ’60πœ†exp2/ξ‚πœ‰ξ‚πœ‡(𝑑),πœ†=πœ†(𝑑).(4.6) From (3.3)–(3.7) with 𝑝=4:

𝑣21π‘˜=Β±22cschβˆšπœ‰ξƒͺπ‘˜πœ‡(𝑑),πœ†=2𝑣60𝜎(𝑑),22π‘˜=Β±22sechβˆšπœ‰ξƒͺπ‘˜πœ‡(𝑑),πœ†=βˆ’2𝑣60𝜎(𝑑),23π‘˜=Β±22cscβˆšπœ‰ξƒͺπ‘˜βˆ’πœ‡(𝑑),πœ†=βˆ’2πœŽπ‘£60(𝑑),24π‘˜=Β±22secβˆšπœ‰ξƒͺπ‘˜βˆ’πœ‡(𝑑),πœ†=βˆ’260𝜎(𝑑).(4.7) Other exact solutions are:

𝑣25π‘˜ξ‚€βˆš=Β±π‘Žexpξ‚€ξ‚€2/ξ‚πœ‰ξ‚ξ‚βˆ’πœ‡(𝑑)βˆ’42π‘Ž2√expξ‚€ξ‚€4/ξ‚πœ‰ξ‚βˆšβˆ’πœ‡(𝑑)+16π‘Žexpξ‚€ξ‚€2/ξ‚πœ‰ξ‚1βˆ’πœ‡(𝑑)+16,πœ†=βˆ’5π‘˜2π‘£πœŽ(𝑑),26π‘˜ξ‚€ξ‚€βˆš=Β±4exp2/ξ‚ξ‚βˆ’πœ‡(𝑑)πœ‰βˆ’π‘Ž2π‘Ž2√+16π‘Žexpξ‚€ξ‚€2/ξ‚πœ‰ξ‚βˆšβˆ’πœ‡(𝑑)+16expξ‚€ξ‚€4/ξ‚πœ‰ξ‚1βˆ’πœ‡(𝑑),πœ†=βˆ’5π‘˜2π‘£πœŽ(𝑑),27βŽ›βŽœβŽœβŽ3=Β±2π‘˜1βˆ’βˆš2Β±cosξ‚€ξ‚€2/ξ‚πœ‰ξ‚βŽžβŽŸβŽŸβŽ 4πœ‡(𝑑),πœ†=βˆ’5π‘˜2π‘£πœŽ(𝑑),28βŽ›βŽœβŽœβŽ3=Β±2π‘˜1βˆ’βˆš2Β±sinξ‚€ξ‚€2/ξ‚πœ‰ξ‚βŽžβŽŸβŽŸβŽ 4πœ‡(𝑑),πœ†=βˆ’5π‘˜2𝜎(𝑑).(4.8) It is clear that using (2.13) we obtain solutions to (1.5). Finally, observe that if 𝑒0(π‘₯,𝑑) is a solution of (1.5), then the solutions 𝑒(π‘₯,𝑑) to the generalized BBM equation (1.6) are obtained as follows:

𝑒(π‘₯,𝑑)=𝑒0(π‘₯,𝑑)βˆ’1.(4.9)

5. Conclusions

We have considered the generalized EW-equation with variable coefficients and the generalized BBM-equation with variable coefficients. We obtained analytic solutions by using the variational iteration method combined with the exp-function method. With the aid of Mathematica we have derived a lot of different types of solutions for these two models. Combined formal soliton-like solutions as well as kink solutions have been formally derived. The results obtained show that the technique used here can be considered as a powerful method to analyze other types of nonlinear wave equations.

According to [22], there are alternative iteration alorithms, which might be useful for future work. Furthermore, various modifications of the exp-function method have been appeared in open literature, for example, the double exp-function method [23, 24].

Other methods for solving nonlinear differential equations may be found in [25–35].

We think that the results presented in this paper are new in the literature.