1. Introduction

Partial differential equations describe various nonlinear phenomena in natural and applied sciences such as fluid dynamics, plasma physics, solid state physics, optical fibers, acoustics, biology, and mathematical finance. It is of significant importance to solve nonlinear partial differential equations (NLPDEs) from both theoretical and practical points of view. In the past decades, many powerful methods for solving NLPDEs have been developed, such as the inverse scattering method [1], Bäcklund and Darboux transform [2, 3], Hirota's bilinear method [4], Painlevé analysis method [5], variable separation method [6–8], tanh-function method [9–11], variational iteration method [12], homotopy perturbation method [13], Jacobian elliptic function expansion [14–20], F-expansion method [21, 22], and Fan subequation method [23, 24] and its various extensions [25–29].

Sirendaoreji [30, 31] presented a new auxiliary equation method by introducing a new first-order nonlinear ordinary differential equation: where are real parameters to be determined. The key idea of the auxiliary equation method is to use the solutions of (1) instead of in tanh-function method and extended tanh-function method [23]. Zhang and his coworker [32], Huang et al. [33], and Yomba [34] have improved the Sirendaoreji method in different manners.

The structure of this paper is organized as follows. In Section 2, an extended auxiliary equation method is described to construct exact solutions of NLPDEs. In Section 3, we apply this improved method to three generalized NLS equations with cubic-quintic terms. Some conclusions are given in Section 4.

2. An Extended Auxiliary Equation Method

For the sake of simplicity, we assume , , and . The derivatives of the above three kinds of Jacobian elliptic functions satisfy [35]: where , and is the modulus of Jacobian elliptic functions ().

For a given NLPDE, with independent variables () and dependent variable .

Step 1. Use the travelling wave transformation , , and reduce the given NLPDE: to the following ordinary differential equation where denotes .

Step 2. The solutions of (4) can be supposed as where is a suitable variable transformation, and satisfies the following first-order differential equation: which is slightly different from (1) by adding one constant term . The crucial step is to give the solutions of (6). It is difficult to give the general solution of (6); here we only consider twelve solutions expressed by various kinds of Jacobian elliptic functions, which read where the function could be expressed through elliptic functions , , , their inverse and different ratios like , and so on.
Type I. If ,  , and , in (7) takes the form
Type II. If ,  , and , in (7) takes the form
Type III. If ,  , and , in (7) takes the form
Type IV. If ,  , and , in (7) takes the form
Type V. If ,  , and , in (7) takes the form
Type VI. If ,  , and , in (7) takes the form

Step 3. Substituting (7) together with (8)–(13) into (5), some new types of Jacobian elliptic function solutions of (3) can be obtained in a unified way.

With the aid of the computer algebraic software Maple, the solutions given by (7)–(13) have been verified by putting them back to the original equation (6). To our knowledge, these twelve solutions are firstly reported here. When the modulus approaches 1 or 0, the Jacobian elliptic functions degenerate to hyperbolic functions and trigonometric functions, respectively.

3. Applications

In this section, three generalized NLS equations with physical interests are chosen to illustrate the effectiveness of the above method.

3.1. The RKL Model

Let us first consider the third-order generalized NLS equation [36], which is proposed by Radhakrishnan, Kundu, and Lakshmanan (RKL). The normalized RKL model can be written as which describes the propagation of femtosecond optical pulses. In (14), represents a normalized complex slowly varying amplitude of the pulse envelope, and are real constants. Some solitary wave solutions and combined Jacobian elliptic function solution were constructed by different methods [37–39].

In order to solve (14), its solutions may be supposed as where , , , and are real constants. Substituting (15) into (14) and taking real and imaginary parts separately, we have There are two cases to discuss.

Case 1. When .
Integrating (17) and setting the integration constant to zero, we obtain Equations (16) and (18) will be equivalent, provided that From which we get
Under the parametric constraints (20), integrating (18) and setting the integration constant to zero, we have where is an arbitrary integration constant, and are given by
Up to now, we can obtain twelve Jacobi elliptic function solutions of (14), which can be expressed as the unified form: where , , are given by (19), , , and are given by (22), and     () are given by (8)–(13).
Some solitary wave solutions can be obtained if the modulus approaches 1. The solution given by (23) degenerates to the kink-type solitary wave solution: where , and the parameter is determined by .
When , the solution given by (23) degenerates to the bell-type solitary wave solution: where , and is determined by .
When , the solution degenerates to where , and is determined by .
Some trigonometric function solutions can be obtained if the modulus ; for example, the solution becomes where , and is determined by .
When , the solution becomes where , and is determined by .

Case 2. When , we have then setting the coefficients of , , and in (16) to zero, respectively, yields
Substituting (29) and (30) into (17) and integrating it with respect to , we have where is an arbitrary integration constant, and are given by
Similar to Case 1, we can also get twelve elliptic function solutions of (14), which read where the parameters , , , are given by (29)-(30), , , and are given by (32), and are given by (8)–(13).
When the modulus , the solution given by (33) degenerates to the solitary wave solution: where , and .
When , the solution degenerates to where , and .
When , the solution degenerates to the solitary wave solution: where , and .
All solutions given above have been checked with Maple by putting them back into the original equation (14). Among them, only the solutions , , , and had been found in [38]. To the author's knowledge, the other solutions have not been found before.

3.2. The Generalized Derivative NLS Equation

Next we consider the generalized derivative NLS equation: where are real constants. Both pulse and fronts that existed in perturbed system (37) were studied by van Saarloos and Hohenberg [40]. Some solitary wave solutions of (37) were obtained by Huang et al. through a generalized auxiliary expansion method [33]. However, the Jacobian elliptic function solutions of (37) have not been reported in literature.

The solutions of (37) may be supposed as where and are constants to be determined. Substituting (38) into (37) and then separating the real and imaginary parts yield the ordinary differential equations about and :

Under the constraint equation (39) is satisfied identically, and (40) becomes

Multiplying (42) by and integrating it with respect to , we have where is an arbitrary integration constant, and are given by

Up to now, we can obtain abundant Jacobi elliptic function solutions of (37), which can be expressed as the unified form: where , , and are given by (44), satisfies the constraint given by (41), and are given by (8)–(13).

When , the solution given by (45) degenerates to the kink-type solitary wave solution: where is determined by .

When , the solution degenerates to the bell-type solitary wave solution: where is determined by . To the authors’ knowledge, these fourteen solutions of (37) are firstly reported here.

3.3. The Kundu-Eckhaus Equation

Finally we consider the Kundu-Eckhaus equation with important physical interests: which was derived by Kundu [41, 42] and Eckhaus [43–45] independently and then known as Kundu-Eckhaus equation. In (48), the parameters , are real constants. To the author's knowledge, the integrable nonlinear equation (48) has not been investigated for possible exact solutions through elliptic function.

The solutions of (48) may be supposed as where and are constants to be determined. Substituting (49) into (48) and then separating the real and imaginary parts yield the ordinary differential equations about and :