• No results found

Fourth-order recursion operators for third-order evolution equations

N/A
N/A
Protected

Academic year: 2022

Share "Fourth-order recursion operators for third-order evolution equations"

Copied!
5
0
0

Loading.... (view fulltext now)

Full text

(1)

Fourth-order recursion operators for third-order evolution equations

Marianna EULER

Department of Mathematics, Lule˚a University of Technology SE-971 87 Lule˚a, Sweden

Received February 20, 2008; Accepted April 10, 2008

AbstractWe report the recursion operators for a class of symmetry integrable evolution equations of third order which admit fourth-order recursion operators. Under the given assumptions we obtain the complete list of equations, one of which is the well-known Krichever-Novikov equation.

We consider the third-order evolution equation

ut= uxxx+ F (u, ux, uxx) (1)

and require that (1) admits a recursion operator of the form

R[u] = Dx4+ G3D3x+ G2Dx2+ G1Dx+ G0+ I1Dx−1◦ J1+ I2Dx−1◦ J2, (2) where Gj = Gj(u, ux, uxx, . . .), Ii are Lie point symmetries and Ji integrating factors for (1), with

∂Ji

∂u6x 6= 0 (3)

for J1 and/or J2 such that Ji = ˆEuΦti.

Here ˆEu is the Euler operator Eˆu =

∂u− Dx

∂ux

+ Dx2

∂u2x − · · ·

and Φt a conserved density for the evolution equation. We recall that, if (1) admits a recursion operator and an infinite hierarchy of conservation laws,

DtΦti+ DxΦxi = 0,

Copyright c 2008 by M Euler

(2)

then (1) is said to be symmetry integrable [3].

Proposition: Equations of the form (1) which admit recursion operators of the form (2) under condition (3) are exhausted by the following two cases:

1. The equation ut= u3x3

2u−1x u22x+ P (u)u−1x + Q(u)u3x, (4) where P and Q satisfy the relation

P(5)+ 10P(3)Q+ 15P′′Q+ P 9Q′′+ 16Q2 + 2P

Q(3)+ 8QQ

= 0 (5) 2. The equation

ut= u3x3 2

 ux

u2x− c



u22x 3

2c u2x− c uxP(u), (6)

where c is an arbitrary but nonzero constant and P satisfies the equation P2

= 4

cP3+ a1P+ a2 (7)

with a1 and a2 arbitrary constants.

Remarks:

1. For Q = 0, equation (4) reduces to the well-known Krichever-Novikov equation [4].

The condition on P then reduces to

P(5)= 0. (8)

A fourth-order recursion operator for the Krichever-Novikov equation was reported in [7]. Note further that (4) can be obtained from the Krichever-Novikov equation by a change of variables of the form u 7→ G(u).

2. The equation (6) with condition (7) and c = −1 has been reported in ([5]). As far as we know, no recursion operator has been reported before for this equation.

3. We stress that in our current classification we require the equation of the form (1) to admit a recursion operator of order four and an integrating factor of order six.

If these conditions are relaxed, a much larger class of equations emerges such as linearizable equations (which have first-order recursion operators and zero-order in- tegrating factors [2]) and the class of semilinear third-order evolution equations which includes the the well-known Korteweg-de Vries equation (with second order recursion operators and second-order integrating factors [6]), as well as the three equations reported in [1] which admit second-order recursion operators with fourth-order inte- grating factors.

(3)

4. Equations (4) and (6) do not admit second-order recursion operators of the form R[u] = Dx2+ G1Dx+ G0+ I1Dx−1◦ J1+ I2Dx−1◦ J2 (9) for any order of J1 and J2. The equations do, however, admit local Lie-B¨acklund symmetries of order five, seven, nine etc., which indicates that there exists a second- order nonlocal recursion operator.

The coefficients G1j, integrating factors J1jand symmetries I1jof the recursion operator R1[u] = Dx4+ G13Dx3+ G12D2x+ G11Dx+ G10+ I11D−1x ◦ J11+ I12Dx−1◦ J12 (10) for equation (4) under condition (5) take the form

G13= −4u2x

ux (11a)

G12= −2u3x

ux + 6u22x

u2x + 4Qu2x4 3

P

u2x (11b)

G11= −2u4x ux

+ 8u3xu2x

u2x 6u32x

u3x + 4Pu2x

u3x 4Quxu2x+ 2Qu3x2 3

P ux

(11c)

G10= u5x ux

4u4xu2x

u2x 2u23x u2x +

 8u22x

u3x + 8Qux



u3x3u42x u4x 4

3 3Qu4x− P u22x u4x +8



Qu2x1 3

P u2x



u2x+ 4Q2u4x+ 2Q′′u4x+4 9

P2 u4x +8

9P Q+10

9 P′′ (11d)

J11= u6x

u2x 6u2xu5x u3x +



−10u3x u3x +45

2 u2xx

u4x + 5Q − 5 3

P u4x



u4x+ 30u23xu2x u4x +



−60u32x u5x +40

3 P u2x

u5x + 10Qux10 3

P u3x



u3x+ 9

4u52x 5

3P u32x 10 u6x +15

2

 P u4x + Q

 u22x+



9Q′′u2x+ 6Q2u2x5 3

P′′

u2x +10 9

P2 u6x

 u2x

+5

9 P′′′− P Q− PQ −5 9

P P

u4x + Q′′′+ 3QQ u4x (12a) J12= u4x

u2x 4u2xu3x

u3x + 3u32x u4x +



2Q −2P u4x



u2x+ Qu2x+P

u2x (12b)

I11= −ux (13a)

I12= −



u3x3

2u−1x u22x+ P (u)u−1x + Q(u)u3x



. (13b)

(4)

The coefficients G2j, integrating factors J2jand symmetries I2jof the recursion operator R2[u] = Dx4+ G23Dx3+ G22D2x+ G21Dx+ G20+ I21D−1x ◦ J21+ I22Dx−1◦ J22 (14) for equation (6) under condition (7) take the form

G23= −4uxu2x

u2x− c (15a)

G22= −2uxu3x

u2x− c+ 2(3u2x+ 2c)u22x (u2x− c)2 6

cP u2x+ 2P (15b)

G21= −2uxu4x

u2x− c + 4(2u2x+ c)u2xu3x

(ux− c)2 2(3u2x+ 7c)uxu32x (u2x− c)3 +2

c

(3u2x+ c)uxu2x

(u2x− c) P −3(u2x− c)ux

c P (15c)

G20= uxu5x

u2x− c 4u2xu2xu4x

(u2x− c)2 (c + 2u2x)u23x

(u2x− c)2 + 4(2c − 3u2x)uxu3x c(u2x− c) P +2(5c + 4u2x)uxu22xu3x

(u2x− c)3 2(c − 2u2x)(c − 3u2x)u2x

c(u2x− c) P2(c − 3u2x)u2xu22x c(u2x− c)2 P

−3(4c + u2x)u2xu42x

(u2x− c)4 +(c − 3u2x)2

c2 P2+(c − 3u2x)u2x

c P′′ (15d)

J21= u6x

u2x− c 6uxu2xu5x

(u2x− c)2  10uxu3x

(u2x− c)2 23c + 45u2x

2(u2x− c)3 u22x 7c − 15u2x 2c(u2x− c)P

 u4x

+6(5u2x+ 3c)u2xu23x

(u2x− c)3 + 16uxu2xu3x

(u2x− c)2 P − (15u2x7c)uxu3x c(u2x− c) P

12(9c + 5u2x)uxu32xu3x

(u2x− c)4 4(3u2x+ c)u32x

(u2x− c)3 P −3(−7c + 3u2x)(−c + 5u2x)u22x 4c(u2x− c)2 P +3(58cu2x+ 15u4x+ 7c2)

2(u2x− c)5 u52x+ c2+ 18cu2x27u4x

2c(u2x− c) P′′ 3(−9u2x+ c) 2c2 P2

 u2x

9u4x2cu2x+ 3c2

6c P′′′+27u4x34cu2x+ 23c2

4c2 P P (16a)

J22= − u4x

u2x− c +4uxu2xu3x

(u2x− c)2 (3u2x+ c)u32x

(u2x− c)3 +3u2x

c P+ 3u2x+ c

2c P (16b)

I21= −ux (16c)

I22= u3x3 2

 ux

u2x− c



u22x 3

2c u2x− c uxP(u). (16d)

(5)

Acknowledgement

The author acknowledges the financial research support provided under LTU grant nr.

2557-05.

References

[1] Euler M and Euler N, Second-order recursion operators of third-order evolution equations with fourth-order integrating factors, J. Nonlinear Math. Phys. 14 (2007), 313-315.

[2] Euler M, Euler N and Petersson N, Linearizable hierarchies of evolution equations in (1 + 1)- dimensional evolution equations, Stud. Appl. Math 111 (2003), 315-337.

[3] Fokas A S, Symmetries and integrability, Stud. Appl. Math. 77 (1987), 253-299.

[4] Krichever I M and Novikov S P Holomorphic bundles over algebraic curves, and nonlinear equations, Russ. Math. Surv. 35 (1980), 53-80. English translation of Uspekhi Mat. Nauk 35 (1980), 47-68.

[5] Mikhailov A V, Shabat A B and Sokolov V V, The symmetry approach to classification of integrable equations, in What is Integrability?, Editor Zhakarov E V, Springer Heidelberg 1991, 115-184.

[6] Petersson N, Euler N and Euler M, Recursion operators for a class of integrable evolution equations, Stud. Appl. Math 112 (2003), 201-225.

[7] Sokolov V V, Hamiltonian property of the Krichever-Novikov equation Dokl. Akad. Nauk SSSR 277 (1984), 48-50.

References

Related documents

In the second part we switch to the framework of control the- ory and dynamical systems to prove upper bounds of convergence rates in some special cases, in particular the

We furthermore write the equations in potential form and use the pure hodograph transformation and corresponding recursion operators to construct hierarchies of autonomous

We present a list of (1 + 1)-dimensional second-order evolution equations all connected via a proposed generalised hodograph transformation, resulting in a tree of

It would be useful to ‘invert’ the conditions on the Λ’s given in each Proposition, to obtain necessary and sufficient conditions for a given second- or third-order ordinary

Torkning och lagring, alternativ Leverans till central tork Egen tork och lagring Värde vid skördeleverans Värde vid leverans i Pool 2 Arbets- och maskinkostnad Arbets-

In order to prove that the schemes work well we applied it to the nonlinear Black-Scholes equation with transaction costs in case of American option which was presented in Chapter

Many of these methods such as high order local NRBCs and the perfectly matched layer (PML) which we will discuss in Chapter 2 and Chap- ter 3 respectively, are efficient for

In this paper, we study the stability of the perfectly matched layer (PML) for symmetric second order hyperbolic partial differential equations on the up- per half plane, with