IJDEInternational Journal of Differential Equations1687-96511687-9643Hindawi Publishing Corporation73850910.1155/2011/738509738509Research ArticleGeometric Integrability of Some Generalizations of the Camassa-Holm EquationChristovOgnyanYurkoV. A.Faculty of Mathematics and InformaticsSofia University5 J. Bouchier boulevard., 1164 SofiaBulgariauni-sofia.bg20114102011201127052011170720112011Copyright © 2011 Ognyan Christov.This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

We study the Camassa-Holm (CH) equation and recently introduced μCH equation from the geometric point of view. We show that Kupershmidt deformations of these equations describe pseudospherical surfaces and hence are geometrically integrable.

1. Introduction

The modern theory of integrable nonlinear partial differential equations arose as a result of the inverse scattering method (ISM) discovered by Gardner et al.  for Korteweg de Vries (KdV) equation. Soon after, it was realized that this method can be applied to several important nonlinear equations like nonlinear Shrödinger equation, sine-Gordon, and so forth.

Sasaki  gave a natural geometric interpretation for ISM in terms of pseudospherical surfaces. Motivated by Sasaki, Chern and Tenenblat  introduce the notion of a scalar equation of pseudospherical type and study systematically the evolution equations that describe pseudospherical surfaces. It appears that almost all important equations and systems in mathematical physics enjoy this property . The advantage of this geometric treatment is that most of the ingredients connected with the integrable equations such as Lax pair, zero curvature representation, conservation laws, and symmetries come naturally.

Let us recall some facts about the equations we study in this paper.

The Camassa-Holm equation (CH)ut-uxxt+2ωux+3uux-2uxuxx-uuxxx=0, where ω, has appeared in  as an equation with bi-Hamiltonian structure. Later, in , it was considered as a model, describing the unidirectional propagation of shallow water waves over a flat bottom. CH is a completely integrable equation; see, for example, . Furthermore, CH is geometrically integrable . Here, we consider the case ω=0. Then, (1.1) ismt+umx+2mux,m=uxx-u.

The bi-Hamiltonian form of (1.2) is [8, 9]mt=-B1δH2[m]δm=-B2δH1[m]δm, where 1=-3,2=m+m are the two compatible Hamiltonian operators ( stands for /x), and the corresponding Hamiltonians areH1[m]=12mudx,H2[m]=12(u3+uux2)dx.

There exists an infinite sequence of conservation laws (multi-Hamiltonian structure) Hn[m],  n=0,±1,±2, including (1.4) such that B1δHn[m]δm=B2δHn-1[m]δm. Very recently a “modified” CH equation (mCH) was introduced by analogy of “Miura transform” from the theory of KdV equation in .

The μCH equation was derived recently in [13, 14] as-uxxt=-2μ(u)ux+2uxuxx+uuxxx, where μ(u)=01udx and u(t,x) is a spatially periodic real-valued function of time variable t and space variable xS1=[0,1). In order to keep a certain symmetry and analogy with CH, one can write the above equation in the form (see also )μ(ut)-uxxt=-2μ(u)ux+2uxuxx+uuxxx. Note that μ(ut)=0 in the periodic case. Introducing m=Au=μ(u)-uxx (1.7) becomesmt+umx+2mux=0,m=μ(u)-uxx. The μCH is an integrable equation and arises as an asymptotic rotator equation in a liquid crystal with a preferred direction if one takes into account the reciprocal action of dipoles on themselves [13, 14].

The bi-Hamiltonian form of (1.8) ismt=-B1δH2[m]δm=-B2δH1[m]δm, where 1=A=-3,2=m+m are the two compatible Hamiltonian operators, and the corresponding Hamiltonians areH1[m]=12mudx,H2[m]=(μ(u)u2+12uux2)dx. Also the μCH equation is geometrically integrable  (see there for other geometric descriptions of this equation).

Recently in , a new 6th-order wave equation, named KdV6, was derived. After some rescaling, this equation can be presented as the following system; ut=6uux+uxxx-wx,wxxx+4uwx+2uxw=0. This system gives a perturbation to KdV equation (w=0), and since the constrain on w is differential, this is a nonholonomic deformation.

Kupershmidt  suggested a general construction applicable to any bi-Hamiltonian systems providing a nonholonomic perturbation on it. This perturbation is conjectured to preserve integrability. In the case of KdV6, the system (1.11) can be converted intout=  B1δHn+1δu-B1(w)=B2δHnδu-B1(w),B2(w)=0, where 1=,2=3+2(m+m) are the two standard Hamiltonian operators of the KdV hierarchy andH1=udx,H2=12u2dx,. In the same paper, Kupershmidt verifies integrability of KdV6, as well as integrability of such nonholonomic deformations for some representative cases: the classical long-wave equation, the Toda lattice (both continuous and discrete), and the Euler top.

In fact, Kersten et al.  prove that the Kupershmidt deformation of every bi-Hamiltonian equation is again bi-Hamiltonian system and every hierarchy of conservation laws of the original bi-Hamiltonian system, gives rise to a hierarchy of conservation laws of the Kupershmidt deformation.

The aim of this paper is to show that the CH and the μCH equations have geometrically integrable Kupershmidt deformations. We also show that the KdV6 equation and two-component CH system  are also geometrically integrable.

As a matter of fact, Yao and Zeng  propose a generalized Kupershmidt deformation and verify that this generalized deformation also preserves integrability in few representative cases: KdV equation, Boussinesq equation, Jaulent-Miodek equation, and CH equation.

Kundu et al.  consider slightly generalized form of deformation for the KdV equation and extend this approach to mKdV equation and to AKNS system.

Guha  uses Kirillov's theory of coadjoint representation of Virasoro algebra to obtain a large class of KdV6-type equations, equivalent to the original one. Also, applying the Adler-Konstant-Symes scheme, he constructs a new nonholonomic deformation of the coupled KdV equation.

The paper is organized as follows. In Section 2, we recall some facts about equations that describe pseudospherical surfaces. The main results are in Section 3. There we show that the Kupershmidt deformations for the CH equation and for the μCH equation are of pseudospherical type and hence are geometrically integrable. We also derive the corresponding quadratic pseudopotentials which turn out to be very useful in obtaining conservation laws and symmetries. At the end, we make some speculations about “modified” μCH equation.

2. Equations of Pseudospherical Type and Pseudopotentials

In this section, we recall some definitions and facts. One can consult, for example,  for more details.

Definition 2.1.

A scalar differential equation Ξ(x,t,u,ux,,uxntm)=0 in two independent variables x,t is of pseudospherical type (or, it describes pseudospherical surfaces) if there exist one-forms ωα0ωα=fα1(x,t,u,,uxrtp)dx+fα2(x,t,u,,uxstq)dt,α=1,2,3, whose coefficients fαβ are smooth functions which depend on x,t and finite number of derivatives of u, such that the 1-forms ω¯α=ωα(u(x,t)) satisfy the structure equations dω¯1=ω¯3ω¯2,dω¯2=ω¯1ω¯3,dω¯3=ω¯1ω¯2, whenever u=u(x,t) is a solution of Ξ=0.

Equations (2.2) can be interpreted as follows. The graphs of local solutions of equations of pseudospherical type can be equipped with structure of pseudospherical surface (see [3, 6, 7]): if ω¯1ω¯20, the tensor ω¯1ω¯1+ω¯2ω¯2 defines a Riemannian metric of constant Gaussian curvature −1 on the graph of solution u(x,t), and ω¯3 is the corresponding metric connection one-form.

An equation of pseudospherical type is the integrability condition for a sl(2,)-valued problem dψ=Ωψ, where Ω is the matrix-valued one-form Ω=Xdx+Tdt=12(ω2ω1-ω3ω1+ω3-ω2).

Definition 2.2.

An equation Ξ=0 is geometrically integrable if it describes a nontrivial one-parameter family of pseudospherical surfaces.

Hence, if Ξ=0 is geometrically integrable, it is the integrability condition of one-parameter family of linear problems ψx=Xψ,  ψt=Tψ. In fact, this is equivalent to the zero curvature equationXt-Tx+[X,T]=0, which is an essential ingredient of integrable equations.

Another important property of equations of pseudospherical type is that they admit quadratic pseudopotentials. Pseudopotentials are a generalization of conservation laws.

Proposition 2.3 (see [<xref ref-type="bibr" rid="B6">6</xref>]).

Let Ξ=0 be a differential equation describing pseudospherical surfaces with associated one-forms ωα. The following two Pfaffian systems are completely integrable whenever u(x,t) is a solution of Ξ=0: -2dΓ=ω3+ω2-2Γω1+Γ2(ω3-ω2),2dγ=ω3-ω2-2γω1+γ2(ω3+ω2). Moreover, the one-forms Θ=ω1-Γ(ω3-ω2),Θ̂=-ω1+γ(ω3+ω2) are closed whenever u(x,t) is a solution of Ξ=0 and Γ (resp. γ) is a solution of (2.6) (resp. (2.7)).

Geometrically, Pfaffian systems (2.6) and (2.7) determine geodesic coordinates on the pseudospherical surfaces associated with the equation Ξ=0 [3, 6].

3. Results

In this section, we consider the nonholonomic deformation of CH equation and μCH equation. We show that they are geometrically integrable and consider their quadratic pseudopotentials. The nonlocal symmetries will be studied elsewhere.

3.1. CH Equation

Recall from Introduction the CH equation (1.2) and its bi-Hamiltonian form (1.3) with the corresponding Hamiltonian operators 1 and 2.

Following Kupershmidt's construction, we introduce the nonholonomic deformation of the CH equationmt=-umx-2mux-B1(w),B2(w)=0.

Proposition 3.1.

The system (3.1) describes pseudospherical surface and hence is geometrically integrable.

Let us give the corresponding 1-formsω1=(m+1)dx+[-um+η+1η(ux-u)+1η-wxx+(η+1)wx-ηw(m+1)]dt,ω2=(η+1)dx+[η+1η-(η+1)u+ux-η(η+1)w+ηwx]dt,ω3=(m+1+η)dx+[-um+η-(η+1)2ηu+η+1ηux+η+1η-wxx+(η+1)wx-ηw(m+1+η)η-(η+1)2η]dt.

For the proof of Proposition 3.1, we need only to verify the structure equations (2.2) and to check that the parameter η is intrinsic.

For the matrices X and T in, we getX=12(η+1-η2(m+1)+η-(η+1)),T=12(T11T12T21-T11), whereT11=η+1η-(η+1)u+ux-η(η+1)w+ηwx,T12=ηu+η2w-1,T21=-2um-(η+1)2+1ηu+2(η+1)ηux+η+2η-2wxx+2(η+1)wx-η(η+2)w-2ηwm. Hence, we have a zero curvature representation Xt-Tx+[X,T]=0 for the system (3.1). From (3.3), it is straightforward to obtain the corresponding scalar linear problemψxx=(14-η2m)ψ,ψt=(-u-ηw+1η)ψx+ux+ηwx2ψ.

In order to apply Proposition 2.3 to nonholonomic deformation of the CH equation, we consider new 1-forms ωnewαωnew1=ω2,ωnew2=-ω1,ωnew3=ω3. With these forms the Pfaffian system (2.7) becomes2γx=ηγ2-2(η+1)γ+2(m+1)+η,2γt=γ2-2wxx-2[ηγ-(η+1)]wx-2wηγx-2[γ-η+1η]ux-2u(γx+1η)+η+2η-2γη+1η. Applying the transform γγ+η+1η, after some algebraic manipulations and setting λ=-1/η, we obtain the following result.

Proposition 3.2.

The nonholonomic deformation of the CH equation (3.1) admits a quadratic pseudopotential γ, defined by the equations m=γ22λ+γx-λ2,γt=γ22[1+1λ(u-wλ)]-γ(u-wλ)x-(u-wλ)(m+λ2)+λu-λ22, where λ0,  m=uxx-u. Moreover, (3.1) possesses the parameter-dependent conservation law γt=λ[(u+w)x-γ-1λ(u-wλ)γ]x.

Conservation densities can be obtained by expanding (3.9) and (3.11) in powers of λ. Note that the left hand side of (3.11) and (3.9) does not depend on w as it should be. The corresponding expansions are performed in .

3.2. <inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M95"><mml:mrow><mml:mi>μ</mml:mi></mml:mrow></mml:math></inline-formula>CH Equation

Consider now the μCH equation (1.8), its bi-Hamiltonian form (1.9) with the corresponding Hamiltonian operators 1,2. Applying Kupershmidt's procedure to (1.8), we obtain the nonholonomic deformation of the μCH equationmt=-umx-2mux-B1(w),B2(w)=0, ormt=-umx-2mux+wxxx,2mwx+wmx=0,m=μ(u)-uxx.

Proposition 3.3.

The nonholonomic deformation of the μCH equation (3.13) describes pseudospherical surfaces and, hence, is geometrically integrable.

For validation of Proposition 3.3, we give the 1-forms associated with (3.13) ω1=12(ηm-η22+2)dx+12[η22u-η(ux+um+12)+μ(u)-2u+2η+(η32-2η)w-η2wx+ηwxx-η2mw]dt,ω2=ηdx+(1-ηu+ux-η2w+ηwx)dt,ω3=12(ηm-η22-2)dx+12[η22u-η(ux+um+12)+μ(u)+2u-2η+(η32+2η)w-η2wx+ηwxx-η2mw]dt.

For the matrices X and T in, we getX=12(η2ηm-η22-η),T=12(T11T12T21-T11), whereT11=1-ηu+ux-η2w+ηwx,T12=2(-u+1η-ηw),T21=η22u-η(ux+um+12)+μ(u)-η2wx+ηwxx-η2mw+η32w. Hence, we have a zero curvature representation Xt-Tx+[X,T]=0 for the system (3.13). From (3.3), it is straightforward to obtain the corresponding scalar linear problemψxx=(η2m)ψ,ψt=(-u-ηw+1η)ψx+ux+ηwx2ψ, which coincides with those in  upon setting w=0 and λ=η/2.

In order to find pseudopotentials for the nonholonomic deformation of the μCH equation, we proceed as before denoting ωnew1=ω2,ωnew2=-ω1,ωnew3=ω3. With these forms, the Pfaffian system (2.7) becomes2γx=-2γ2-2ηγ+ηm-η22,2γt=-2γ2η+2γ2(u+ηw)-2γ(1-ηu+ux-η2w+ηwx)+[η22u-η(ux+m+12)+μ(u)+η32w-η2wx+ηwxx-η2mw]. After some manipulations, the above system obtains the form 2γx=-2γ2-2ηγ+ηm-η22,2γt=-2ηγ2+ηwxx-[(2γ+η)(u+ηw)]x+μ(u)-2γ-η2. Applying the transform γγ-η/2, we getγx=-γ2+η2m,γt=-γ2η+η2wxx-[γ(u+ηw)]x+μ(u)2. Multiplying the first equation (3.22) by -1/η and then adding the result to the second equation (3.23), we get the following result denoting λ=η/2.

Proposition 3.4.

The nonholonomic deformation of the μCH equation (3.13) admits a quadratic pseudopotential γ, defined by the equations m=γ2λ+γxλ,γt=-2γ2λλwxx-[γ(u+2λw)]x+μ(u)2, where λ0,  m=μ(u)-uxx. Moreover, (3.13) possesses the parameter-dependent conservation law γt=12λ[γ+λ(u+2λw)x-2λ(u+2λw)γ]x.

As the conserved densities for the nonholonomic deformation are the same as for the original bi-Hamiltonian system, we make use of the pseudopotentials to obtain them for the μCH equation. One possible expansion of γ isγ=λ1/2γ1+γ0+j=1λ-j/2γ-j. Substituting this into (3.24) yields γ1=m,γ0=-mx4m,γ-1=132mx2m5/2+18(mxm3/2)x,and so forth. In this way, we can obtain local functionals; see .

We finish this section with the geometric integrability of one of the most popular two-component generalization of CH equation and of KdV6 equation.

Another generalization of the Camassa-Holm equation is the following integrable two-component CH system :ut-uxxt=-3uux+2uxuxx+uuxxx+σρρx,ρt+(uρ)x=0, where σ=±1. Introducing a new variable v=ρ2/2, the above system becomesut-uxxt=-3uux+2uxuxx+uuxxx+σvx,vt+2vux+uvx=0. The system (3.30) is geometrically integrable. The corresponding 1-forms, satisfying the structure equations (2.2), are the following: ω1=(uxx-u-σηv+1)dx+[u2-uuxx+η+1η(ux-u)+1η-σv+ησuv]dt,ω2=(η+1)dx+(η+1η-(η+1)u+ux)dt,ω3=(uxx-u-σηv+η+1)dx+[u2-uuxx+η-(η+1)2ηu+η+1η(ux+1)-σv+ησuv]dt. We could easily include two-component Hunter-Saxon system  into this picture (see also ).

Finally, we note that nonholonomic perturbation of KdV equation, known as KdV6 equation, is also of pseudospherical type, that is, KdV6 equation is geometrically integrable. We just give the corresponding 1-formsω1=(1-u)dx+[-uxx+ηux-2u2+1ηwx-1η2(wxx+2uw)+(2-η2)u+2η2w+η2]dt,ω2=ηdx+(η3+2ηu+2ηw-2ux-2η2wx)dt,ω3=-(1+u)dx+[-uxx+ηux-2u2+1ηwx-1η2(wxx+2uw)-(2+η2)u-2η2w-η2]dt, which coincide with those for KdV equation  when w0.

4. Concluding Remarks

In this paper, we study the CH equation and some of its generalizations from the geometric point of view. We show that Kupershmidt deformations for CH and μCH equations preserve integrability and derive some important objects like quadratic pseudopotentials which turn out to be useful for obtaining conservation laws and nonlocal symmetries. It is also shown that the KdV6 equation and two-component CH system are also geometrically integrable.

Having at hand these examples of geometrically integrable Kupershmidt deformations, it is natural to think that maybe there exists a general link in this sense: a Kupershmidt deformation of geometrically integrable system is again geometrically integrable. We have not succeeded in establishing such a link up to now, but we believe that this is true at least for the systems with local Hamiltonian pair of operators as in the above examples.

Let us return, however, to the μCH equation (w=0). It is obvious that pseudopotentials for the μCH equation (1.8) and parameter-dependent conservation law are obtained fromm=γ2λ+γxλ,γt=12λ(γ+λux-2λuγ)x=γx2λ+uxx2-(γu). Equation (4.1) is an analogue of the Miura transformation of KdV theory. We can repeat, purely formally, the procedure for obtaining the “modified” CH (mCH) equation  in this case. However, it is clear that since μCH contains nonlocal term, one can expect that the “modified” equation also will have nonlocal terms.

Denote by A the operator A=μ-2,A(u)=m=μ(u)-uxx. The operators A-1 and commute and μ(u)=μ(Au).

We haveu=A-1m,        ux=A-1mx,        uxx=A-1mxx, in which m is determined by (4.1). Then, the second equation (4.2) takes the form γt=γx2λ+A-1mxx2-γA-1mx-γxA-1m, or Aγt=Aγx2λ+mxx2-A(γA-1m). Formally, this equation can be named as a “modified” μCH equation. One can simplify further this equation using (4.1) or even to present it as a system as in , it remains nonlocal and, hence, it is of no immediate advantage.

Acknowledgments

This work is partially supported by Grant no. 169/2010 of Sofia University and by Grant no. DD VU 02/90 with NSF of Bulgaria.

GardnerC.GreenJ.KruskalM.MiuraR.Method for solving the Korteweg-de Vriez equationPhysical Review Letters19671910951097SasakiR.Soliton equations and pseudospherical surfacesNuclear Physics. B1979154234335753750910.1016/0550-3213(79)90517-0ChernS. S.TenenblatK.Pseudospherical surfaces and evolution equationsStudies in Applied Mathematics19867415583827492ZBL0605.35080DingQ.TenenblatK.On differential systems describing surfaces of constant curvatureJournal of Differential Equations20021841185214192915210.1006/jdeq.2001.4141ZBL1022.58005GórkaP.ReyesE. G.The modified Camassa-Holm equationIMRN20112011122617264910.1093/imrn/rnq163ReyesE.Geometric integrability of the Camassa-Holm equationLetters in Mathematical Physics200259211713110.1023/A:10149333161691894812ZBL0997.35081ReyesE.Pseudo-potentials, nonlocal symmetries and integrability of some shallow water equationsSelecta Mathematica. New Series2006122241270228126410.1007/s00029-006-0024-2ZBL1117.35072FokasA. S.FuchssteinerB.Symplectic structures, their Bäcklund transformations and hereditary symmetriesPhysica D. Nonlinear Phenomena198141476663647010.1016/0167-2789(81)90004-XCamassaR.HolmD.An integrable shallow water equation with peaked solitonsPhysical Review Letters1993711116611664123445310.1103/PhysRevLett.71.1661ZBL0972.35521ConstantinA.On the inverse spectral problem for the Camassa-Holm equationJournal of Functional Analysis19981552352363162455310.1006/jfan.1997.3231ZBL0907.35009ConstantinA.McKeanH.A shallow water equation on the circleCommunications on Pure and Applied Mathematics1999528949982168696910.1002/(SICI)1097-0312(199908)52:8&lt;949::AID-CPA3&gt;3.0.CO;2-DZBL0940.35177ConstantinA.IvanovR.Poisson structure and action-angle variables for the Camassa-Holm equationLetters in Mathematical Physics20067619310810.1007/s11005-006-0063-92223766ZBL1137.37032KhesinB.LenellsJ.MisiołekG.Generalized Hunter-Saxton equation and the geometry of the group of circle diffeomorphismsMathematische Annalen20083423617656243099310.1007/s00208-008-0250-3ZBL1156.35082LenellsJ.MisiołekG.TiğlayF.Integrable evolution equations on spaces of tensor densities and their peakon solutionsCommunications in Mathematical Physics20102991129161267280010.1007/s00220-010-1069-9ZBL1214.35059YingF.YueL.ChougzhengO.On the blow-up structure for the generalized periodic Camassa-Holm and Degasperis-Procesi equationsarXiv: 1009.2466v2Karasu-KalkanlıA.KarasuA.SakovichA.SakovichS.TurhanR.A new integrable generalization of the Korteweg-de Vries equationJournal of Mathematical Physics2008497, article 07351610.1063/1.29534742432054ZBL1152.35473KupershmidtB.KdV6: an integrable systemPhysics Letters. A20083721526342639240599310.1016/j.physleta.2007.12.019KerstenP.Krasil'shchikI.VerbovetskyA.VitoloR.Integrability of Kupershmidt deformationsActa Applicandae Mathematicae20101091758610.1007/s10440-009-9442-42579883ZBL1187.37091ConstantinA.IvanovR.On an integrable two-component Camassa-Holm shallow water systemPhysics Letters. A20083724871297132247460810.1016/j.physleta.2008.10.050YaoY.ZengY.The generalized Kupershmidt deformation for constructing new integrable systems from integrable bi-Hamiltonian systemsJournal of Mathematical Physics201051610.1063/1.34319672676480KunduA.SahadevanR.NalinideviR.Nonholonomic deformation of KdV and mKdV equations and their symmetries, hierarchies and integrabilityJournal of Physics. A: Mathematical and Theoretical20094211, article 11521310.1088/1751-8113/42/11/1152132485847ZBL1181.35240GuhaP.Nonholonomic deformation of generalized KdV-type equationsJournal of Physics. A: Mathematical and Theoretical20094234, article 34520110.1088/1751-8113/42/34/3452012530242ZBL1179.35286