Startseite Similarity Reduction and Exact Solutions of a Boussinesq-like Equation
Artikel Öffentlich zugänglich

Similarity Reduction and Exact Solutions of a Boussinesq-like Equation

  • Bo Zhang EMAIL logo und Hengchun Hu
Veröffentlicht/Copyright: 6. Februar 2018

Abstract

The similarity reduction and similarity solutions of a Boussinesq-like equation are obtained by means of Clarkson and Kruskal (CK) direct method. By using Lie symmetry method, we also obtain the similarity reduction and group invariant solutions of the model. Further, we compare the results obtained by the CK direct method and Lie symmetry method, and we demonstrate the connection of the two methods.

1 Introduction

In 1989, a method called Clarkson and Kruskal (CK) direct method was firstly proposed by Clarkson and Kruskal to seek similarity reduction of the nonlinear partial differential equation, and they obtained many similarity solutions different from ones obtained by the classical Lie group approach [1] and the nonclassical Lie group approach [2]. CK direct method is an ansatz-based method and it does not involve group transformation, so it is widely loved and used by non-mathematicians. By using CK direct method, one can directly obtain the symmetry reduction of high dimensional system and nonlinear system. Up to now, a lot of nonlinear partial differential equations have been reduced to ODEs or low order PDEs, such as the generalised Burgers equations [3], the (2+1)-dimensional KdV equation [4], the (2+1)-dimensional Boiti-Leon-Manna-Pempinelli equation [5], and the (2+1)-dimensional Lax-Kadomtsev-Petviashvili equation [6].

On the other hand, Lie symmetry analysis is also an important subject and it has been applied to many fields of mathematics and physics, such as Lie algebras [7], classical mechanics [8], and other nonlinear partial differential equations. Lie symmetry is a point transformation. And it was proposed to reduce the order of a given differential equation. With the development of computers, a lot of mathematical and physical equations are solved by the method. And it becomes one of the most common methods with Darboux transformation [9], [10], [11], Bäcklund transformation [12], [13], Painlevé analysis [14], [15], CK direct method and Hirota bilinear approach [16], [17], [18]. Recently, Lie symmetry analysis has been successfully used to find some group invariant solutions for a number of integrable systems such as the Heisenberg equation [19], the Sawada-Kotera equation [20], the (3+1)-dimensional Yu-Toda-Sasa-Fukuyama equation [21], and the modified KdV-Zakharov-Kuznetsov equation [22].

In this paper, we focus on the following Boussinesq-like equation

(1)utt12uux26u2uxxuxxxx=0,

which was proposed in [23]. As we all know, the Boussinesq equation is a very famous nonlinear evolution equation in describing the propagation of long waves in shallow water. Since then, Boussinesq-like equations were proposed to study variable water depth problems [24], and now the applications of Boussinesq-like equation involve a wide range of fields such as the dynamics of the thin inviscid layer with free surface, the surface waves, the nonlinear string, the shape-memory alloys or other nonlinear phenomena [25], [26], [27], [28]. Different from the regularised Boussinesq equation, (1) retains the fourth spatial derivative uxxxx but eliminates the dissipative term uxx. Then it is difficult to analyse because the model is no more fully integrable. In this paper, we study the Boussinesq-like equation with the CK direct method to find the similarity reduction equation and the corresponding exact solutions. We also compare the results obtained by the CK direct method and Lie symmetry method.

In the second section of this paper, we discuss the similarity reduction and the corresponding similarity solutions of the Boussinesq-like equation with the CK direct method. In the third section, the similarity reduction and the corresponding group invariant solutions of the Boussinesq-like equation are discussed by using Lie symmetry method. The group explanation of reduction is introduced to demonstrate the connection of the two methods in the fourth section. And in the last section, we give summary and discussions.

2 Clarkson and Kruskal Direct Method and Similarity Reduction

In order to find similarity solutions of a (1+1)-dimensional partial differential equation, Clarkson and Kruskal have indicated the solution in the form of [29]

(2)u(x,t)=U[x,t,P(ξ(x,t))].

On this base, one can prove it is sufficient that the solution u to a nonlinear system may commonly be simplified as the following linear ansatz

(3)u(x,t)=α(x,t)+β(x,t)P(ξ(x,t)).

Substituting the ansatz (3) into the (1) yields

(4)βξx4P(4)(4βxξx3+6βξx2ξxx)P+(βξt26α2βξx212βxξxξxx4βξxξxxx6βxxξx23βξxx2)P(12α2βxξx+6α2βξxx+4βxxxξx+6βxxξxx+4βxξxxx+βξxxxxβξtt+24ααxβξx2βtξt)P(48αβxβξx+24β2αxξx+12αβ2ξxx)PP(24ααxβx+12αβαxx+βxxxxβtt+6α2βxx+12βαx2)P12αβ2ξx2P2+(24βαxβx12αββxx6β2αxx12αβx2)P212β3ξx2PP26β3ξx2P2P(6β3ξxx+36β2βxξx)P2P(6β2βxx+12ββx2)P312αβ2ξx2PP+αttαxxxx12ααx26α2αxx=0,

where primes represent the derivatives with respect to ξ. In order to make (4) be an ODE of P(ξ), the ratio of the coefficients of the derivatives of P(ξ), including the combinations of their products, cannot be anything but the function of ξ. In (4), we use the coefficient of P(4) (i.e. βξx4) as the normalizing coefficient and we have

(5)(4βxξx3+6βξx2ξxx)=βξx4Γ1(ξ),
(6)βξt26α2βξx212βxξxξxx4βξxξxxx6βxxξx23βξxx2=βξx4Γ2(ξ),
(7)(12α2βxξx+6α2βξxx+4βxxxξx+6βxxξxx+4βxξxxx+βξxxxxβξtt+24ααxβξx2βtξt)=βξx4Γ3(ξ),
(8)(48αβxβξx+24β2αxξx+12αβ2ξxx)=βξx4Γ4(ξ),
(9)(24ααxβx+12αβαxx+βxxxxβtt+6α2βxx+12βαx2)=βξx4Γ5(ξ),
(10)12αβ2ξx2=βξx4Γ6(ξ),
(11)24βαxβx12αββxx6β2αxx12αβx2=βξx4Γ7(ξ),
(12)12β3ξx2=βξx4Γ8(ξ),
(13)6β3ξx2=βξx4Γ9(ξ),
(14)(6β3ξxx+36β2βxξx)=βξx4Γ10(ξ),
(15)(6β2βxx+12ββx2)=βξx4Γ11(ξ),
(16)12αβ2ξx2=βξx4Γ12(ξ),
(17)αttαxxxx12ααx26α2αxx=βξx4Γ13(ξ),

where Γ1(ξ), Γ2(ξ),…,Γ13(ξ) are functions of ξ to be determined. Before dealing with the equations above, we propose the following remarks.

Remark 1: If α(x, t) has the form α=α0(x, t)+ β(x, t)Ω(ξ), then one can take Ω(ξ)=0;

Remark 2: If β(x, t) has the form β=β0(x, t)Ω(ξ), then one can take Ω(ξ)=1;

Remark 3: If ξ(x, t) can be determined by an equation of the form Γ(ξ)=ξ0(x, t), then one can take Γ(ξ)=ξ;

Remark 4: In order to facilitate calculation, one can take the original letter for the undermined function obtained by integrating, differentiating, taking the index or logarithm, and scaling transformation.

By using Remarks 1–4, it is easy to obtain the similarity reductions of the Boussinesq-like equation. If ξx=0, then ξ and P(ξ) are only the functions of t and the similarity reduction equation is only obviously reduced to the second order ODE. So here, we consider only the ξx≠0 case.

For ξx≠0, it is easy to obtain

(18)β=±ξx,   Γ8(ξ)=12,

from (12) by using the Remark 2. Here we take

(19)β=ξx,

and substitute (19) into (5), by scaling transformation for Γ1(ξ), we can obtain

(20)ξxxξx+Γ(ξ)ξx=0.

Integrating (20) respect to x and using the Remark 4, we have

(21)Γ(ξ)+lnξx=θ(t),

where θ(t)≡θ are functions of t.

By taking the index on both sides of the (21) and using the Remark 4 again, we can obtain

(22)ξxΓ(ξ)=θ(t).

Integrating (22) respect to x and using the Remark 4, we have

(23)Γ(ξ)=θ(t)x+σ(t),

where σ(t)≡σ are functions of t.

Using the Remark 3, we can obtain

(24)ξ=θ(t)x+σ(t),   Γ1(ξ)=0,

Obviously,

(25)β=θ.

From (10), we can get

(26)α=0,   Γ6(ξ)=0,

by using the Remark 1.

Substituting (24)~(26) into (4), we can obtain

(27)θ5(P(4)+6P2P+12PP2)θ(dθdtx+dσdt)2P[θ(d2θdt2x+d2σdt2)+2dθdt(dθdtx+dσdt)]Pd2θdt2P=0.

Using the coefficient of P(4) as the normalizing coefficient again, we have

(28)θ5γ1(ξ)=θ(dθdtx+dσdt)2,
(29)θ5γ2(ξ)=θ(d2θdt2x+d2σdt2)+2dθdt(dθdtx+dσdt),
(30)θ5γ3(ξ)=d2θdt2,

where γ1(ξ), γ2(ξ), γ3(ξ) are functions of ξ to be determined. From (28), we have

(31)θ2γ(ξ)=dθdtx+dσdt,

where γ(ξ)=(γ1(ξ))12.

First, because ξ=θ(t)x+σ(t) and the right-hand side of the (31) is a linear function related to x, we can assume

(32)γ(ξ)=Aξ+B,

with A and B being arbitrary constants.

It is easy to obtain

(33)γ1(ξ)=(Aξ+B)2.

Substituting (32) into (31), we can obtain

(34)θ2[A(xθ+σ)+B]=dθdtx+dσdt.

Comparing the coefficients of x on both sides of the (34), we have

(35)ddtθ=Aθ3,
(36)ddtσ=θ2(Aσ+B).

Substituting (35) and (36) into (29) and (30), we can obtain

(37)γ2(ξ)=5A(Aξ+B),   γ3(ξ)=3A2.

Now the similarity reduction of the Boussinesq-like equation reads

(38)u=θP(ξ),

with (24), (35), and (36).

Finally, by substituting (28)~(30), (33) and (37) into (27), we can obtain the corresponding similarity reduction equation for P,

(39)P(4)+6P2P+12PP2(Aξ+B)2P5A(Aξ+B)P3A2P=0.

If A=0, from (35) and (36), we can obtain the general solution for θ and σ:

(40)θ(t)=a0,   σ(t)=Ba02t+b0,

with a0 and b0 being arbitrary constants, and the similarity reduction of the Boussinesq-like equation has the form

(41)u=a0P(ξ),   ξ=a0x+Ba02t+b0,

then the similarity reduction (41) is a travelling reduction and (39) is reduced to

(42)P(4)+6P2P+12PP2B2P=0.

By integrating (42) twice, we can obtain the general ODE

(43)P+2P3B2P=c1ξ+c0,

with c0 and c1 being arbitrary constants. Without loss of generality, we take c0=c1=0, then the (43) becomes

(44)P+2P3B2P=0,

and (44) is the Duffing equation which has been studied extensively. Here we give only some solutions of the Duffing equation:

  1. Bell-shaped soliton solution

    (45)P1=±B[sech(±Bξ)].
  2. Jacobi elliptic sine function solution

    (46)P2=a[sn(±a2B2ξ,m)],

    where a is a constant and m=±aB2a2.

  3. Jacobi elliptic cosine function solution

    (47)P3=a[cn(±amξ,m)],

    where a is a constant and m=±a2a2B2.

  4. The third kind of Jacobi elliptic function solution

    (48)P4=a[dn(±aξ,m)],

where a is a constant and m=±2a2B2a.

There are different types of the exact solutions of the Duffing equation, thus we can obtain the corresponding solutions of the Boussinesq-like equation from (41).

If A≠0, from (35) and (36), we can obtain the general solution for θ and σ:

(49)θ(t)=(2At+a0)12,   σ(t)=BA+b0(2At+a0)12,

with a0 and b0 being arbitrary constants, and the similarity reduction of the Boussinesq-like equation has the form

(50)u=(2At+a0)12P(ξ),   ξ=(2At+a0)12x+b0(2At+a0)12BA,

with P(ξ) being determined by (39).

3 Lie Symmetry Analysis and Group Invariant Solutions

In this section, we perform Lie symmetry analysis on the Boussinesq-like equation in order to obtain similarity reduction. In the classical Lie group theory, we know that if the Boussinesq-like equation is invariant under one-parameter Lie group of transformation,

(51)xx+εX(x,t,u),
(52)tt+εT(x,t,u),
(53)uu+εU(x,t,u),

then, the group invariants ξ, P(ξ) can be obtained by dealing with characteristic equation,

(54)dxX(x,t,u)=dtT(x,t,u)=duU(x,t,u).

Now we take uu+εσ into (1) and extract only coefficients of ε, then we can obtain the following symmetry equation:

(55)σtt6u2σxx12σux212uσuxx24uuxσxσxxxx=0,

and the corresponding symmetry is

(56)σ=X(x,t,u)ux+T(x,t,u)utU(x,t,u).

Substituting (56) into (55) and using (1) again, we can obtain

(57)X(x,t,u)=C1x+C2,
(58)T(x,t,u)=2C1t+C3,
(59)U(x,t,u)=C1u,

with C1, C2 and C3 being arbitrary constants. we will study three cases by the different selections of the arbitrary constants.

Case 1. C1=1, C2=C3=0.

Considering the characteristic equation,

(60)dxx=dt2t=duu.

From equation dxx=dt2t, we can obtain

(61)xt12=c1,

and from equation dt2t=duu, we can obtain

(62)ut12=c2,

with c1, c2 being arbitrary constants. In the classical Lie group theory, we can define

(63)xt12=z,
(64)ut12=V(z),

then we can obtain a similarity reduction of the Boussinesq-like equation

(65)u=V(z)t12,   z=xt12.

The final corresponding reduction equation for V can be obtained by substituting (65) into (1),

(66)V(4)+6V2V+12VV214z2V54zV34V=0.

It is easy to see that (66) is just a special case of (39).

Case 2. C1=0, C2=k, C3=1.

Considering the characteristic equation,

(67)dxk=dt1=du0.

Similarly, we can obtain a travelling wave reduction of the Boussinesq-like equation

(68)u=V(z),   z=xkt.

The final corresponding reduction equation for V can be obtained by substituting (68) into (1),

(69)V(4)+6V2V+12VV2k2V=0.

It is obvious that (69) is equivalent to (42).

Case 3. C2=1, C1=C3=0.

Considering the characteristic equation,

(70)dx1=dt0=du0.

Similarly, we can obtain a similarity reduction of the Boussinesq-like equation

(71)u=V(z),   z=t.

The corresponding reduction equation for V can be obtained by substituting (71) into (1),

(72)V=0,

i.e.

(73)V(z)=az+b,

with a, b being arbitrary constants. Then a similarity solution of the Boussinesq-like equation can be obtained by substituting (73) into (71),

(74)u=at+b.

4 The Connection of the CK Direct Method and Lie Approach

In the second section and third section, we can see that CK direct method is efficient and convenient because of not using group theory. And the similarity reductions by using it contain the similarity reductions obtained by the classical Lie symmetry analysis.

In the following paper, we will demonstrate that the similarity reduction obtained by CK direct method is the same as that one by the Lie symmetry analysis [30].

Now we give a group explanation of the reduction (38) with (24). In reality, from (24)

(75)ξ=θ(t)x+σ(t),

we have

(76)dxdt=X(x,t,u)T(x,t,u)=θ(t)x+σ(t)θ(t),

i.e.

(77)X(x,t,u)=θ(t)x+σ(t)θ(t)T(x,t,u).

In the same way, we have

(78)dudt=U(x,t,u)T(x,t,u)=θ(t)uθ(t),

i.e.

(79)U(x,t,u)=θ(t)uθ(t)T(x,t,u).

Therefore, the corresponding symmetry of the reduction in the sense of CK direct method has the form

(80)σ=X(x,t,u)ux+T(x,t,u)utU(x,t,u)=Tθ(t)(θ(t)uxxσ(t)ux+θ(t)utθ(t)u).

Substituting (35), (36) and (80) into (55), we can obtain

(81)dTdt=2Aθ2T,   T=T(t),

i.e. T=θ−2 for A≠0 and T=1 for A=0. If T is determined by (81), then σ given by (80) is really a solution of (55). That means the similarity reduction obtained by CK direct method also can be obtained by the classical Lie approach.

5 Summary and Discussion

Clarkson and Kruskal direct method is performed on the Boussinesq-like equation, the similarity reduction and similarity solutions of the Boussinesq-like equation have been studied. On the other hand, the similarity reduction and group invariant solutions of the Boussinesq-like equation also have been obtained with Lie symmetry analysis. Further, the group explanation of the reduction has been discussed to compare the results obtained by the two methods.

From the analysis of this paper, we establish the relation between the Boussinesq-like equation and the Duffing equation which has been studied extensively, and get some new forms of solution of the Boussinesq-like equation. We also demonstrate the similarity reduction obtained by CK direct method is the same as that one by the Lie symmetry analysis. In addition, these methods are also convenient to solve high dimensional nonlinear systems and many coupled integrable systems.

Acknowledgments

The work described in this paper was supported by National Natural Science Foundation of China (11471215), Shanghai Leading Academic Discipline Project (XTKX2012) and the Hujiang Foundation of China (B14005).

References

[1] X. Y. Tang and J. Lin, Commun. Theor. Phys. 39, 6 (2003).10.1088/0253-6102/39/1/6Suche in Google Scholar

[2] S. P. Qian and L. X. Tian, Commun. Theor. Phys. 48, 399 (2007).10.1088/0253-6102/48/3/003Suche in Google Scholar

[3] B. M. Vaganan and R. Asokan, Stud. Appl. Math. 111, 435 (2003).10.1111/1467-9590.t01-1-00041Suche in Google Scholar

[4] S. Y. Lou, J. Math. Phys. 41, 8286 (2000).10.1063/1.1320859Suche in Google Scholar

[5] Y. Li and D. S. Li, Appl. Math. Sci. 6, 579 (2012).10.1007/978-3-642-18293-8_15Suche in Google Scholar

[6] H. C. Hu, J. B. Wang, and H. D. Zhu, Commun. Theor. Phys. 63, 136 (2015).10.1088/0253-6102/63/2/03Suche in Google Scholar

[7] M. Boldyreva and A. Magazev, Russ. Phys. J. 10, 1671 (2017).10.1007/s11182-017-0959-0Suche in Google Scholar

[8] N. H. Ibragimovab, V. F. Kovalevcd, S. V. Meleshkoe, and V. Yu. Bychenkovdf, Int. J. Nonlinear Mech. 71, 1 (2015).10.1016/j.ijnonlinmec.2014.12.011Suche in Google Scholar

[9] Y. S. Zhang, D. Q. Qiu, Y. Cheng, and J. S. He, Theor. Math. Phys. 191, 710 (2017).10.1134/S0040577917050117Suche in Google Scholar

[10] U. Saleem and M. Hassan, J. Math. Anal. Appl. 447, 1080 (2017).10.1016/j.jmaa.2016.10.045Suche in Google Scholar

[11] H. Q. Zhang, M. Y. Zhang, and R. Hu, Appl. Math. Lett. 76, 170 (2018).10.1016/j.aml.2017.09.002Suche in Google Scholar

[12] M. Singh and R. K. Gupta, Commun. Nonlinear Sci. Numer. Simul. 37, 362 (2016).10.1016/j.cnsns.2016.01.023Suche in Google Scholar

[13] Y. Sun, Y. P. Wang, Y. Jiang, H. L. Zhen, B. Tian, et al., Nonlinear Dynam. 80, 1 (2015).10.1007/s11071-014-1321-5Suche in Google Scholar

[14] S. Erbay and E. S. uhubi, Int. J. Eng. Sci. 27, 915 (1989).10.1016/0020-7225(89)90032-3Suche in Google Scholar

[15] C. L. Zheng and Y. Li, Chin. Phys. B. 21, 1 (2012).10.5246/jcps.2012.06.075Suche in Google Scholar

[16] H. H. Hao and L. Lü, J. Shanghai University. 11, 241 (2007).10.1007/s11741-007-0310-3Suche in Google Scholar

[17] J. H. Lee and O. K. Pashaev, Theor. Math. Phys. 152, 991 (2007).10.1007/s11232-007-0083-3Suche in Google Scholar

[18] Y. N. Tang, W. J. Zai, S. Q. Tao, and Q. Guan, Appl. Mathe. Comput. 293, 565 (2017).10.1016/j.amc.2016.08.022Suche in Google Scholar

[19] Z. L. Zhao and B. Han, Commun. Nonlinear Sci. Numer. Simul. 45, 220 (2017).10.1016/j.cnsns.2016.10.008Suche in Google Scholar

[20] Y. W. Zhang, Turkish J. Math. 41, 158 (2017).10.3906/mat-1504-29Suche in Google Scholar

[21] S. Sahoo and S. S. Ray, Comput. Math. Appl. 73, 253 (2017).10.1016/j.camwa.2016.11.016Suche in Google Scholar

[22] S. Sahoo, G. Garai, and S. S. Ray, Nonlinear Dynam. 87, 1995 (2017).10.1007/s11071-016-3169-3Suche in Google Scholar

[23] A. M. Wazwaz, Ocean Eng. 53, 1 (2012).10.1016/j.oceaneng.2012.06.012Suche in Google Scholar

[24] D. H. Peregrinea, J. Fluid Mech. 27, 815 (1967).10.1017/S0022112067002605Suche in Google Scholar

[25] A. Biswas, D. Milovic, and A. Ranasinghe, Commun. Nonlinear Sci. Numer Simul. 14, 3738 (2009).10.1016/j.cnsns.2009.02.021Suche in Google Scholar

[26] C. I. Christov, G. A. Maugin, and M. G. Velarde, Phys. Rev. E Stat. Phys. Plasmas Fluids Relat. Interdiscip. Topics. 54, 3621 (1996).10.1103/PhysRevE.54.3621Suche in Google Scholar

[27] M. Dehghan and R. Salehi, Appl. Math. Model. 36, 1939 (2012).10.1016/j.apm.2011.07.075Suche in Google Scholar

[28] A. Esfahani and L. G. Farah, J. Math. Anal. Appl. 385, 230 (2012).10.1016/j.jmaa.2011.06.038Suche in Google Scholar

[29] P. A. Clarkson and M. D. Kruskal, J. Math. Phys. 30, 2201 (1989).10.1063/1.528613Suche in Google Scholar

[30] M. Jia, Commun. Theor. Phys. 49, 275 (2008).10.1088/0253-6102/49/2/04Suche in Google Scholar

Received: 2017-12-09
Accepted: 2018-01-15
Published Online: 2018-02-06
Published in Print: 2018-03-28

©2018 Walter de Gruyter GmbH, Berlin/Boston

Heruntergeladen am 28.9.2025 von https://www.degruyterbrill.com/document/doi/10.1515/zna-2017-0442/html
Button zum nach oben scrollen