## Study of Integrability and Exact Solutions for Nonlinear Evolution Equations

View this Special IssueResearch Article | Open Access

Junquan Song, Yujian Ye, Danda Zhang, Jun Zhang, "Conditional Lie-Bäcklund Symmetries and Reductions of the Nonlinear Diffusion Equations with Source", *Abstract and Applied Analysis*, vol. 2014, Article ID 898032, 17 pages, 2014. https://doi.org/10.1155/2014/898032

# Conditional Lie-Bäcklund Symmetries and Reductions of the Nonlinear Diffusion Equations with Source

**Academic Editor:**Weiguo Rui

#### Abstract

Conditional Lie-Bäcklund symmetry approach is used to study the invariant subspace of the nonlinear diffusion equations with source , . We obtain a complete list of canonical forms for such equations admit multidimensional invariant subspaces determined by higher order conditional Lie-Bäcklund symmetries. The resulting equations are either solved exactly or reduced to some finite-dimensional dynamic systems.

#### 1. Introduction

The classical symmetry theory for studying differential equations is presented firstly by Lie, which has been universally used and proved to be very effective in similarity reductions and group classifications [1–7]. However, there exist some important equations with very small Lie point symmetry groups. For example, the Fisher equation and Fitzhugh-Nagumo equation, which are widely used in mathematical biology, are invariant only under the time and space translations. This means that the classical symmetry reduction method is not a proper tool for dealing with these equations. To overcome this difficulty, several generalized methods have been developed and established, including the nonclassical symmetry method (or referred to as the conditional symmetry method) [8], the weak symmetry method [9, 10], iteration of the nonclassical method [11], the Clarkson-Kruskal direct method [12, 13], and the conditional Lie-Bäklund symmetry (CLBS) method (or referred to as the generalized conditional symmetry method) [14–16].

CLBS can be regarded as a natural generalization of the nonclassical symmetry. Therefore, the procedure for computing the CLBSs is about the same as for the nonclassical method. Furthermore, Galaktionov and Svirshchevski have shown that the CLBS method is closely related to the invariant subspace (IS) method; namely, exact solutions defined on ISs for differential equations or their variant forms can be obtained by using the CLBS method [17–19]. For nonlinear diffusion equations (NLDEs), symmetry-related methods, especially the CLBS method, have been proved to be very powerful to classify and reduce the considered equations [20–34]. For example, NLDEs can be used to describe not only the process by which matter is transported from one part of a system to another, as a result of random molecular motion, but they can also represent many other apparently unrelated phenomena such as heat conduction in solids or even the stationary notion of a boundary layer of fluid over a flat plate [35]. In [36], the Lie point symmetry method has been used to obtain the similarity solutions of the inhomogeneous NLDEs belonging to the above equations, where nonzero constants and have several applications such as propagation of a thermal wave in an exponential atmosphere. A complete classification of the symmetry reductions of these equations using the nonclassical method is given by Saied in [37]. The second-order CLBSs of these equations have also been studied in [33]. Furthermore, the generalized porous medium equations are considered by using the CLBS method in [34]. Some exact solutions, defined on the polynomial, trigonometric, and exponential ISs determined by the CLBSs, are constructed.

In this paper, we mainly discuss the following NLDEs: by means of the CLBS method. Here, and are, respectively, referred to as the diffusion and source terms. Equation (4) has a wide range of applications in physics, diffusion process, and engineering sciences and has been applied to describe several situations such as heat conduction by electrons in a plasma, heat conduction by radiation in a fully ionized gas, axisymmetric flow of a very viscous fluid, and turbulent diffusion [38, 39].

The remainder of this paper is organized as follows. In the following section, we review some necessary notations, definitions, and fundamental theorems on the CLBS method. Equations of the form (4) admitting CLBSs and the corresponding ISs are classified in Section 3. Exact solutions and reductions of some examples in the resulting equations are obtained in Section 4. The last section is devoted to conclusions and discussions.

#### 2. Preliminaries

Let us give a brief discussion on the CLBS method. For the th-order equation we set as an evolutionary vector field with characteristic . Here, we use the following notations:

*Definition 1. *The evolutionary vector field (6) is said to be a Lie-Bäcklund symmetry of (5) if and only if
where is the set of all differential consequences of the equation; that is,

*Definition 2. *The evolutionary vector field (6) is said to be a CLBS of (5) if and only if
where denotes the set of all differential consequences of equation with respect to ; that is, , .

Proposition 3 (Zhdanov [14] and Fokas and Liu [15, 16]). *Equation (5) admits the CLBS (6) if there exists a function such that
**
where , the prime denotes the Fréchet derivative, and is an analytic function of , , , , and , , .*

An obvious conclusion of this proposition is that (5) admits the CLBS with the characteristic if Here, and are given as in Definitions 1 and 2.

For (4), we set the characteristic where , , and . It is important to note that if (4) admits CLBS (13), then equation admits CLBS In fact, (4) and (14) are related as where denotes the inverse function of .

From (12), we can see that (14) admitting of the CLBS with the characteristic (15) is equivalent to ; namely, where and is given as in Definition 2. Thus, the linear solution space of linear ordinary differential equation (ODE) is invariant with respect to the above operator ; that is, It follows that if (14) admits CLBS (15), then (14) has an exact solution of the generalized separation of variables form where the coefficients satisfy the -dimensional dynamic system

Thus, we will determine the forms of (14) so that (14) admits the CLBS (15), which is equivalent to classify (14) in terms of the IS (19), or it is equivalent to study the CLBS (13) of (4).

The following theorem provides us with the estimate to the maximal dimension of the IS admitted by an th-order nonlinear differential operator.

Theorem 4 (Galaktionov and Svirshchevski [19]). *If a linear space is invariant with respect to the nonlinear differential operator of order , then there exists an inequality
*

It follows that the order of linear ODE is not greater than five if (14) admits the CLBS (15). This allows us to classify (14) based on the existence of the generalized variable separation solutions (GVSSs) (21), which are generated by the solution space (19) determined by the linear ODE , or it is equivalent to study the GVSSs of (4) generated by .

#### 3. CLBSs and Corresponding ISs of (14)

In view of Theorem 4, it suffices to consider CLBSs (15) of (14) with . We first consider the case . It implies from (17) that (14) admits the CLBS (15) if there holds where the primes and subscripts denote the derivatives and the partial derivatives with respect to the indicated variables, respectively. To vanish all the coefficients of (24), we will have the following overdetermined system: For general , it is apparent that and . Substituting into the second of the above system, we arrive at To solve (26), we can derive three possibilities: , , and .

Assume that the diffusion coefficient takes power or exponential forms. From (16), without loss of generality, it is reasonable to consider the following four cases: (i) , ; (ii) , ; (iii) , ; and (iv) , , where and are arbitrary constants.

*Case 1 (). *In this case, we can derive . Correspondingly, there exist the following cases: (a) , ; (b) , ; and (c) , . If , , the third and fourth equations are simplified as
Then, we derive or . However, if , function turns to zero, which should be omitted. By the similar calculation, we obtain or with case (b), while we derive with case (c). Therefore, we have results listed as the 1–6th entries in Table 1 with Case 1.