We find exact solutions of the Generalized Modified Boussinesq (GMB) equation, the Kuromoto-Sivashinsky (KS) equation the and, Camassa-Holm (CH) equation by utilizing the double reduction theory related to conserved vectors. The fourth order GMB equation involves the arbitrary function and mixed derivative terms in highest derivative. The partial Noether’s approach yields seven conserved vectors for GMB equation and one conserved for vector KS equation. Due to presence of mixed derivative term the conserved vectors for GMB equation derived by the Noether like theorem do not satisfy the divergence relationship. The extra terms that constitute the trivial part of conserved vectors are adjusted and the resulting conserved vectors satisfy the divergence property. The double reduction theory yields two independent solutions and one reduction for GMB equation and one solution for KS equation. For CH equation two independent solutions are obtained elsewhere by double reduction theory with the help of conserved Vectors.
1. Introduction
Nonlinear differential equations have many significant implications for the mathematical models and have been of great interest in the last few decades. In the literature various techniques are used to construct exact and numerical solutions of differential equations [1–6]. Different approximations and numerical methods can be implemented to reduce nonlinear partial differential equations (PDEs) to ordinary differential equations (ODEs) but the problem occurs in the convergence of the solutions. Symmetry analysis and conservation laws play a vital role in the analysis of differential equations and have many important applications in numerical methods, linearization, and integrability. For variational type differential equations standard Lagrangian always exists and conservation laws can be computed from the well-known Noether’s formula [7]. The standard Lagrangian does not exist for nonvariational differential equations. Kara and Mahomed [8] introduced the notion of partial Lagrangians for such differential equations. A Noether like theorem was invoked for the derivation of conservation laws associated with a partial Lagrangian. This technique is referred as to Noether’s like approach or partial Lagrangian approach. There are some approaches for the construction of conservation laws which do not use the knowledge of standard or pariah Lagrangians. The different methods to compute the conservation laws and the comparison of approaches were studied in [9] (also see references therein).
The relationship between Noether symmetries and conservation laws is well known and is useful to reduce the numbers of variables and order of differential equations. In [10], a conserved vector is associated with the Lie-Bäcklund symmetries. Sjöberg and Mahomed [11, 12] have generalized association of a conserved vector to nonlocal symmetries. The association of Lie-Backlund symmetries or nonlocal symmetries with a conserved vector led to the development of the double reduction theory for the nonvariational type PDEs or system of PDEs with two independent variables [13, 14]. Bokhari et al. [15] generalized the theory of double reduction to find the invariant solutions for a nonlinear system of PDEs with several independent variables. This theory is helpful in finding invariant solution of PDEs having a nontrivial conserved vector and at least one symmetry associated with it. Recently, Narain and Kara [16] redefine the variational and nonvariational approaches for a class of PDE involving mixed derivative terms. Due to the presence of mixed derivative term a conserved vector computed by Noether’s theorem does not satisfy the divergence relationship. A number of extra terms contributing to the trivial part of conserved vector arise and need to be adjusted to satisfy the divergence relationship.
The objective of this paper is to find the exact solutions of GMB, KS, and CH equations using the double reduction theory. The conservation laws of GMB and KS equations are constructed via partial lagrangian approach. For GMB equation the derived conserved vector failed to satisfy the divergence property. The extra terms which constitute the trivial part of the conserved vectors are adjusted to satisfy the divergence relationship. After construction of conservation laws the theory of double reduction is applied to compute the solutions of GMB and KS equations. The conserved vectors of Camassa-Holm equation are derived in [17]. Two exact solutions are computed for the Camassa-Holm equation by utilization of double reduction theory.
This paper is arranged in the following manner. In Section 2, basic definitions, fundamental operators, and theorem for double reduction theory are invoked. In Section 3, the conservation laws and exact solutions for GMB equation are presented. The conservation laws and solution for KS equation are derived in Section 4. Section 5 deals with the exact solutions of CH equation using double reduction theory. Concluding remarks are summarized in Section 6.
2. Fundamental Operators
Assume an nth-order system of m partial differential equations of p independent variables x=(x1,x2,x3,…,xp) and q dependent variables u=(u1,u2,…,uq):
(1)Eα=(x,u,u(1),u(2),…,u(n)),α=1,2,3,…,m,
where u(1),u(2),…,u(n) symbolize the set of all first, second,…, nth-order partial derivatives; that is, uiα=Di(uα), uijα=DjDi(uα),…, in which Di, the total differential operator corresponding to xi, is given by
(2)Di=∂∂xi+uiα∂∂uα+uijα∂∂ujα+⋯,i=1,2,…,p.
The summation convention is used whenever appropriate.
Definition 1.
The nth-order system (1) can be expressed as
(3)Eα=E0α+E1α=0,α=1,2,…,m.
Suppose there exists a function L=L(x,u,u(1),…,u(n)) such that (1) can be expressed as
(4)δLδuα=FβαE1β,α=1,2,…,m.
In (4) Fβα is the invertible matrix and δ/δuα is the Euler operator defined by
(5)δδuα=∂∂uα+∑s≥1(-1)sDi1⋯Dis∂∂ui1⋯isα.
If E1β≠0 in (4), then L is said to be a partial Lagrangian of (1); otherwise, it is known as a standard Lagrangian.
Definition 2.
A Lie-Bäcklund or generalized operator is
(6)X=ξi∂∂xi+ηα∂∂uα+ζiα∂∂uiα+ζi1i2α∂∂ui1i2α+⋯,
where ξi, ηα∈𝒜 (space of differential function) and the additional coefficients can be determined from the formula
(7)ζiα=Di(Wα)+ξjuijα,ζi1⋯isα=Di1⋯Dis(Wα)+ξjuji1⋯isα,s>1,
where Wα is the Lie characteristic function described by
(8)Wα=ηα-ξjujα.
Definition 3.
A vector T=(T1,T2,…,Tn), Ti∈𝒜, i=1,2,…,n, is called a conserved vector if DiTi=0 holds for all solutions of (1).
Definition 4.
A Lie-Bäcklund operator X defined in (6) is said to be a Noether symmetry generator corresponding to a Lagrangian L=L(x,u,u(1),…,u(n)) of (1) if it satisfies
(9)X(L)+LDi(ξi)=DiBi,
where Di is the total derivative operator defined in (2) and Bi are the guage terms.
The generator X in (6) is said to be a partial Noether operator corresponding to a partial Lagrangian L=L(x,u,u(1),…,u(n)) of (1) if
(10)X(L)+LDi(ξi)=WαδLδuα+DiBi.
Definition 5.
If X in (6) is a Noether symmetry with respect to a Lagrangian or partial Noether operator with respect to a partial Lagrangian, then the conserved vector can be constructed from
(11)Ti=Bi-[ξi+Wαδδuiα+∑s≥1Dii1⋯Dis(Wα)δδuii1⋯isα]L,
in which
(12)δδuiα=∂∂uiα+∑s≥1(-1)sDj1⋯Djs∂∂uij1…jsα,ddi=1,2,…,n,α=1,2,…,m.
Definition 6.
Assume that X is a symmetry of the system (1) and T is a conserved vector of (1). If X and T satisfy
(13)X(Ti)+TiDj(ξj)-TjDj(ξi)=0,i=1,2,…,n.
Then X is associated with T.
The following theorem illustrates the construction of new conservation laws from symmetries and known conservation laws [18, 19].
Theorem 7.
Let X be any Lie-Bäcklund symmetry of (1) and Ti, i=1, 2,…, n comprise the components of a conserved vector of (1); then
(14)T*i=X(Ti)+TiDj(ξj)-TjDj(ξi)
forms a conserved vector of (1).
We now describe some results which are used in our work (see [13, 14]).
Consider a scalar partial differential equation E=0 with α=2, (x1,x2)=(t,x) such that it has conserved vector (Tt,Tx) and admits symmetry generator X associated with conserved vectors. In terms of canonical variables r,s with the symmetry X=∂/∂s the conservation laws can be expressed as DrTr+DsTs=0. The vectors Tr and Ts in terms of (t,x) are
(15)Tr=TtDt(r)+TxDx(r)Dt(r)Dx(s)-Dx(r)Dt(s),(16)Ts=TtDt(s)+TxDx(s)Dt(r)Dx(s)-Dx(r)Dt(s).
Theorem 8.
A PDE of order n with two independent variables, which admits a symmetry X that is associated with a conserved vector T, is reduced to an ODE of order n-1; namely, Tr=k, where Tr is defined in (15) for solutions invariant under X.
3. Conservation Laws and Exact Solutions of GMB Equation
The generalized modified Boussinesq (GMB) equation is
(17)utt-δuttxx-(f(u))xx=0,
where δ is a constant and f(u) is an arbitrary function. GMB equation describes nonlinear model of wave propagation of elastic rods and also arises in nonlinear lattice waves, iron sound waves, and vibrations in a nonlinear string and is thus important to study.
Equation (17) admits a partial Lagrangian
(18)L=-ut22-δ2utx2+12fuux2,
and the corresponding partial Euler-Lagrange equation is
(19)δLδu=12fuuux2.
Substituting these values in (10) and comparing the coefficients of like monomials of u we obtain
(20)τu=0,τx=0,ξu=0,ξt=0,ηuu=0,ηtu=0,ηtx=0,ηxu-τtx=0,τfuu-τufu=0,τt-ξx-2ηu=0,fu(2ηu+τt-ξx)=0,Bu1=-ηt,Bu2=fuηx,Bt1+Bx2=0.
In order to solve the system (20), we discuss the following cases.
Case 1.
fuu=0.
In this case we obtain f(u)=au+b, τ=d2, ξ=d4, η=d1t+d3, B1=-ud1, and B2=0.
Substituting the above values of τ, ξ, η, B1, and B2 in (11) we obtain
(21)Tt=-ud1+d2(ut22+δ2utx2-12aux2)+(d1t+d3-d2ut-d4ux)(ut-δutxx)-(d2utx+d4uxx)(δutx),Tx=d4(ut22+δ2utx2-12aux2)-(d1t+d3-d2ut-d4ux)(aux+δuttx)+(d1-d2utt-d4utx)(δutx),
where d1,…,d4 are arbitrary constants. The choice of constants one by one equal to one and the rest to zero yields the following conserved vectors:
(22)T1t=-u+tut-δtutxx,T1x=-atux-δtuttx+δutx,T2t=-ut22-δ2utx2-12aux2+δututxx,T2x=autux+δututtx-δuttutx,T3t=ut-δutxx,T3x=-aux-δuttx,T4t=-utux+δuxutxx-δutxuxx,T4x=ut22-δ2utx2+12aux2+δuxuttx.
The divergence of (22) becomes
(23)DtT1t+DxT1x=-tδuttxx,DtT2t+DxT2x=δututtxx-δuttxutx,DtT3t+DxT3x=-δuttxx,DtT4t+DxT4x=δuxuttxx-δutxutxx.
The conserved vectors in (22) fail to satisfy the divergence properties. Narain and Kara [16] prove that Ti can be adjusted to T~i such that DiT~i=0.
Following the same line we find that the modified conserved vectors T~ji(i=1,2, j=1,2,3,4) are
(24)T~1t=-u+tut-tδutxx,T~1x=-atux+δutx,T~2t=-ut22-δ2utx2-12aux2+δututxx,T~2x=autux-δuttutx,T~3t=ut,T~3x=-aux-δuttx,T~4t=-utux+δuxutxx,T~4x=ut22-δ2utx2+12aux2.
Case 2.
fuu≠0.
In this case after some simple but lengthy manipulations result in τ=0,ξ=0,η=a1x+a2+a3t,B1=-ua3 and B2=(u)1. Substitutig τ, ξ, η, B1, and B2 in (11), we arrive at
(25)Tt=-ua3+(a1x+a2+a3t)(ut-δutxx)+δa1utx,Tx=a1f(u)-(a1x+a2+a3t)(fuux+δuttx)+δa3utx.
The choice of constants gives rise to
(26)T5t=xut-δxutxx+δutx,T5x=f(u)-xfuux-δxuttx,T6t=ut-δutxx,T6x=-fuux-δuttx,T7t=-u+tut-δtutxx,T7x=-tfuux-δtuttx+δutx.
The conserved vectors in (26) do not satisfy the relation DiTi=0; that is,
(27)DtT5t+DxT5x=-δxuttxx,DtT6t+DxT6x=-δuttxx,DtT7t+DxT7x=-δtuttxx.
The new conserved vectors T~i after adjustment result in
(28)T~5t=xut+δutx,T~5x=f(u)-xfuux-δxuttx,T~6t=ut,T~6x=-fuux-δuttx,T~7t=-u+tut-δtutxx,T~7x=-tfuux+δutx.
Now, we apply the double reduction theory to associate the Lie point symmetries with the conservation laws. Equation (17) admits the following Lie point symmetries when fuu=0; that is, f(u)=au+b. Consider
(29)X1=∂∂x,X2=∂∂t,X3=u∂∂u,X4=f(t,x)∂∂u,
where f(t,x) satisfies (17).
The symmetries X1 and X2 in (29) are associated with the conservation laws of GMB equation given below when fuu=0:
(30)T1=(-ut22-δ2utx2-12aux2+δututxx,autux-δuttutx),T2=(ut,-aux-δuttx).
The combination of these symmetries X=X1+αX2 yields the generator, X in canonical form X=∂/∂q if
(31)dx1=dtα=du0=dr0=ds1
or
(32)s=x,r=αx-t,u=u(r).
Using relation (15), we determine T1r and T2r corresponding to T1 and T2 given in (30):
(33)T1r=-ur22-δα22urr2+a2α2ur2+δα2ururrr,(34)T2r=-ur+δα2urrr+aα2ur.
Since T1=(T1r,T1s) is associated with X, so T1r=k in (33) implies
(35)-ur22-δα22urr2+a2α2ur2+δα2ururrr=k.
The solution of (35) is
(36)u(r)=c1+mr+c2sin(α2a-1rαδ)+c3cos(α2a-1rαδ),
where
(37)m=(+α4a2c32-2α2ac32+α4a2c22)(α2a-1)dd×δ(c22+c32+2δkα2-2α2ac22dddsdd+α4a2c32-2α2ac32+α4a2c22))1/2dd×((α2a-1)δα)-1
and c1, c2, and c3 are arbitrary constants. Hence
(38)u(t,x)=c1+m(αx-t)+c2sin(α2a-1(αx-t)αδ)+c3cos(α2a-1(αx-t)αδ)
is a solution of (17) invariant under X.
Similarly from (34), one can easily find that
(39)u(t,x)=((c1αδsin(α2a-1(αx-t)αδ)ddd-c2αδcos(α2a-1(αx-t)αδ))dd×(α2a-1)-1(c1αδsin(α2a-1(αx-t)αδ))+k(αx-t)α2a-1+c3
is also a solution of (17).
Equation (17) admits the trivial symmetry generators X1=∂/∂x and X2=∂/∂t in Case 2 when f(u) is an arbitrary function. In this case the symmetry generators X1=∂/∂x and X2=∂/∂t are associated with the conserved vectors
(40)T3=(ut,-fuux-δuttx)
of GMB equation. The similarity variables are defined in (32) and with the help of conserved vectors (40), (17) reduces to
(41)δα2urr-ur+α2f(u(r))=kr-c1δα2.
Equation (41) cannot be reduced further due to an arbitrary function f(u).
4. Conservation Laws and Exact Solution of the Kuramoto-Sivashinsky Equation
The Kuramoto-Sivashinsky (KS) equation plays a dominant role in stability of flame fronts, reaction diffusion, and other physical phenomena. The KS equation is
(42)ut+auux+buxx+kuxxxx=0,
where a, b, and k are constants.
Equation (42) does not admit a standard Lagrangian but the partial Lagrangian is
(43)L=kuxx22-b2ux2,
and the corresponding partial Euler-Lagrange equation is
(44)δLδu=-ut-auux.
Substituting these values in (10) and after some lengthy manipulation results in τ=0, ξ=0, η=c, B1=cu, and B2=acu. Setting c=1, we obtain τ=0, ξ=0, η=1, B1=u, and B2=au.
Equation (11) with the substitution of τ, ξ, η, B1, and B2 gives rise to
(45)Tt=u,Tx=au22+bux+kuxxx,
which satisfies
(46)DtTt+DxTx=0.
We reduce (42) using the double reduction theory, that is, association of symmetries with the conserved vectors.
Equation (42) possesses three Lie point symmetry generators:
(47)X1=∂∂x,X2=∂∂t,X3=∂∂u+at∂∂x.
Using (13) one can easily verify that X1,X2 in (47) are associated with the conserved vector
(48)T=(u,au22+bux+kuxxx)
of KS equation. We set X=αX1+X2; then, the canonical coordinates of X are s=t, r=αt-x, and u=u(r). Using (15) and with the use of the above canonical coordinates, we find that
(49)Tr=αu-au22+bur+kurrr.
Replacing Tr=m (arbitrary constant), (49) becomes
(50)αu-au22+bur+kurrr=m.
Equation (50) admits the symmetry generator X=∂/∂r and the similarity variables are v=u, w=ur. After using the similarity variables, (50) reduces to
(51)k=αv-av22+bw+kw2w′′+kww′2,w′=dwdv.
The solution of (51) gives rise to (52)w(v)=RootOf[[∫1(-1+2bv+8kα2v3)b-ln(2αv-av2-2c)+4[∫1(-1+2bv+8kα2v3)b×[[∫b3kα2v3+b3v-b3dvkα2v[[∫b3kα2v3+b3v-b3dv-2bv+RootOf[∫b3kα2v3+b3v-b3dv+∫[-2b-1+2bv+8kα2v3]dv+c1]]]dv]+c2]×(-2αv+av2+2c).
Equation (52) in terms of w=ur and v=u yields(53)ur=RootOf[[∫1(-1+2bu+8kα2u3)b[kα2u[-2bu-ln(2αu-au2-2c)+4[∫1(-1+2bu+8kα2u3)b×[kα2u[[∫b3kα2u3+b3u-b3du-2bu+RootOf[∫b3kα2u3+b3u-b3du+∫[-2b-1+2bu+8kα2u3]du+c1[∫b3kα2u3+b3u-b3du]]]du]+c2]×(-2αu+au2+2c), which in turn results in(54)r-[∫1×[RootOf[[∫1(-1+2bu+8kα2u3)b-ln(2αu-au2-2c)+4[∫1(-1+2bu+8kα2u3)b×[kα2u[[∫b3kα2u3+b3u-b3du-2bu+RootOf[∫b3kα2u3+b3u-b3du+∫[-2b-1+2bu+8kα2u3]du+c1]]]du]+c2]×(-2αu+au2+2c)[[∫1(-1+2bu+8kα2u3)b-ln(2αu-au2-2c)]-1du]+c3,r=αt-x, which is a solution of (42).
5. Exact Solutions of the Camassa-Holm Equation
The Camassa Holm (CH) equation is described by Camassa and Holm as a bi-Hamiltonian model for waves in shallow water. It is prominent for turbulent flows and waves in a hyperelastic rod. The CH equation is
(55)ut+2ωux+3uux-α2(uxxt+2uxuxx+uuxxx)=0,
where α and ω are constants.
Conservation laws of CH equation are derived in [17]:
(56)T1t=ut-α2uxx,T1x=3u22+2ωu-α2uuxx-α22ux2,T2t=u22-α2uuxx-α22ux2,T2x=ωu2+u3-α2u2uxx+α2utux.
Equation (55) admits the following symmetries:
(57)X1=∂∂t,X2=∂∂x,X3=t∂∂t-ωt∂∂x-(u+ω)∂∂u.
From (13) it can be easily shown that the conserved flow(T1,T2) for CH equation are associated with only X1,X2. We define the combination of these symmetries X=X1+αX2. The generator X has the canonical form X=∂/∂q if(58)dt1=dxα=du0=dr0=ds1=dw0,
or
(59)s=t,r=αt-x,u=u(r).
Equation (15) with the use of similarity variables defined above reduces (56) to
(60)T1r=(α-2ω)u+α2(u-α)urr-3u22+α2ur22,(61)T2r=αu22-α3uurr+α2u2urr+α3ur22-ωu2-u3.
Setting T1r=k in (60), we obtain
(62)(α-2ω)u+α2(u-α)urr-3u22+α2ur22=k.
The solution of (62) yields
(63)±∫(u-α)α(u-α)(2u2ω-u2α+u3+2ku+c1α2)du=r+c2,ddddddddddddddddddddd.dddddddddddr=αt-x,
which is a solution of (55).
Similarly from (61), applying the same procedure, we have
(64)±∫(u-α)α(u-α)(2ωu2-αu2+u3-2k+c3α2u)du=αt-x+c4
which constitutes the solution of (55).
6. Conclusions
Exact solutions of the GMB equation, the KS equation, and the CH equation were constructed by utilizing the conservation laws. Firstly GMB equation was considered and conservation laws were computed by partial Noether’s approach. Two cases arise; namely, Case 1: fuu=0 and Case 2: fuu≠0. In Case 1, when f(u) was a linear function four conserved vectors were obtained, whereas in Case 2 for arbitrary f(u) three conserved vectors were reported. The derived conserved vectors failed to satisfy divergence condition. The extra terms arising in conserved vectors were absorbed and the new forms of conserved vectors satisfying the divergence property were found. When f(u) is linear only two conserved vectors (30) satisfy the symmetry conservation laws relationship. The double reduction theory was applied to these two conserved vectors and two independent solutions were constructed. The symmetry was associated with only one conserved vector (40) when f(u) is arbitrary. For this case GMB equation was reduced to a second order ODE. The partial Noether approach for KS equation yielded one conserved vector which satisfies the symmetry conservation laws relation. The conserved vector reduced the KS equation to a third order ODE (50) which further reduced to a second order ODE (51) which in turn results in the exact solution (54) of KS equation. A similar procedure is carried out to obtain two exact solutions of CH equation. These solutions are new and not obtained in the literature. The derived solutions cannot be interpreted physically due to deficiency of experimental sources; however these are important for numerical simulations.
OlverP. J.1986107New York, NY, USASpringerGraduate Texts in Mathematics10.1007/978-1-4684-0274-2MR836734NayfehA. H.1973New York, NY, USAJohn Wiley & SonsMR0404788BlumanG. W.KumeiS.1989New York, NY, USASpringerMR1006433IbragimovN. H.CRC Handbook of Lie group analysis of differential equations19941Boca Raton, Fla, USACRCStephaniH.1989Cambridge, UKCambridge University PressMR1041800WazwazA.-M.New solitary wave solutions to the Kuramoto-Sivashinsky and the Kawahara equations200618221642165010.1016/j.amc.2006.06.002MR2282604ZBL1107.65094NoetherE.Invariante Variationsprobleme19182235257(English translation in Transport Theory and Statistical Physiscs, vol. 1, no. 3, pp. 186–207, 1971)MR0406752ZBL0292.49008KaraA. H.MahomedF. M.Noether-type symmetries and conservation laws via partial Lagrangians2006453-436738310.1007/s11071-005-9013-9MR2250141ZBL1121.70014NazR.MahomedF. M.MasonD. P.Comparison of different approaches to conservation laws for some partial differential equations in fluid mechanics2008205121223010.1016/j.amc.2008.06.042MR2466625ZBL1153.76051KaraA. H.MahomedF. M.Relationship between symmetries and conservation laws2000391234010.1023/A:1003686831523MR1755015ZBL0962.35009SjöbergA.MahomedF. M.Non-local symmetries and conservation laws for one-dimensional gas dynamics equations2004150237939710.1016/S0096-3003(03)00259-5MR2040029ZBL1102.76059SjöbergA.MahomedF. M.The association of non-local symmetries with conservation laws: applications to the heat and Burgers' equations200516821098110810.1016/j.amc.2004.10.006MR2171765ZBL1084.35075SjöbergA.Double reduction of PDEs from the association of symmetries with conservation laws with applications2007184260861610.1016/j.amc.2006.06.059MR2294874ZBL1116.35004SjöbergA.On double reductions from symmetries and conservation laws20091063472347710.1016/j.nonrwa.2008.09.029MR2561361ZBL1179.35038BokhariA. H.Al-DweikA. Y.MahomedF. M.ZamanF. D.Conservation laws of a nonlinear ( n+1) wave equation20101142862287010.1016/j.nonrwa.2009.10.009MR2661950ZBL1197.35174NarainR.KaraA. H.On the redefinition of the variational and “partial” variational conservation laws in a class of nonlinear PDEs with mixed derivatives2010154732741MR2807511ZBL05943178NazR.NaeemI.AbelmanS.Conservation laws for Camassa-Holm equation, Dullin-Gottwald-Holm equation and generalized Dullin-Gottwald-Holm equation20091063466347110.1016/j.nonrwa.2008.09.028MR2561360ZBL1179.35275SteebW. H.StramppW.Diffusion equations and Lie-Backlund transformation groups19821149599KaraA.MahomedF. M.Action of Lie Bäcklund symmetries on conservation lawsModern Group Analysis, vol. 7, 1997