Lax Pairs Graham W. Griffiths 1 City University, UK. oo0oo

Transcription

1 Lax Pairs Graham W. Griffiths 1 City University, UK. oo0oo Contents 1 Introduction 1 Lax pair analysis.1 Operator form Operator usage Matrix form Lax pair - operator examples The advection equation Korteweg-de Vries KdV equation a Korteweg-de Vries KdV equation b The modified KdV mkdv equation The Swada-Kotera SK equation Lax pair - matrix examples Korteweg-de Vries KdV equation Fifth order Korteweg-de Vries equation KdV Sine-Gordon equation Sinh-Gordon equation Liouville equation Introduction The term Lax Pairs refers to a set of two operators that, if they exist, indicate that a corresponding particular evolution equation, F x, t, u,... = 0, 1 is integrable. They represent a pair of differential operators having a characteristic whereby they yield a nonlinear evolution equation when they commute. The idea was originally published by Peter Lax in a seminal paper in 1968 Lax-68. A Lax pair consists of the Lax operator L which is self-adjoint may depend upon x, u x, u xx,..., etc. but not explicitly upon t the operator M that together represent a given partial differential equation such that L t = M, L = ML LM. Note: M, L = ML LM = LM ML = L, M represents the commutator of the operators L M. Operator M is required to have enough freedom in any unknown parameters or functions to enable the operator L t = M, L or L t + L, M = 0 to be chosen so that it is of degree zero, i.e. does not contain differential operator terms is thus a multiplicative operator. L M can be either scalar or matrix operators. An important characteristic of operators is that they only operate on terms to their right! The process of finding L M corresponding to a given equation is generally non-trivial. Therefore, if a clues is available, inverting the process by first postulating a given L 1 graham@griffiths1.com A self-adjoint operator is an operator that is its own adjoint or, if a matrix, one that is Hermitian, i.e a matrix that is equal to its own conjugate transpose. Graham W Griffiths 1 5 May 01

2 M then determining which partial differential equation they correspond to, can sometimes lead to good results. However, this may require the determination of many trial pairs, ultimately, may not lead to the required solution. Because the existence of a Lax pair indicates that the corresponding evolution equation is integrable 3, finding Lax pairs is a way of discovering new integrable evolution equations. This approach will not be pursued here, but detailed discussion can be found in Abl-11, Abl-91, Dra-9. In addition, if a suitable Lax pair can be found for a particular nonlinear evolutionary equation, then it is possible that they can be used to solve the associated Cauchy problem using a method such as the inverse scattering transform IST method due to Gardener, Greene, Kruskal, Miura Gar-67. However, this is generally a difficult process additional discussion can be found in Abl-11, Abl-91, Inf-00, Joh-97, Pol-07. Lax pair analysis.1 Operator form Given a linear operator L, that may depend upon the function ux, t, the spatial variable x spatial derivatives u x, u xx,, etc. but not explicitly upon the temporal variable t, such that Lψ = λψ, ψ = ψx, t,.1a the idea is to find another operator M, whereby: ψ t = Mψ..1b Now, following Ablowitz Clarkson Abl-91, we take the time derivative of eqn.1a obtain L t ψ + Lψ t =λ t ψ + λψ t Lψ t + LMψ =λ t ψ + λmψ =λ t ψ + MLψ L t + LM ML ψ =λ t ψ. 3 Hence, in order to solve for non-trivial eigenfunctions ψ x, t, we must have where L t + L, M = 0, 4 L, M := LM ML, 5 which will be true if only if λ t = 0. Equation 4 is known as the Lax representation of the given PDE, also known as the Lax equation, L, M in eqn 5 represents the commutator of the two operators L M which form a Lax pair. Thus, we have shown that a Lax pair, eqns.1a.1b, has the following properties: a. The eigenvalues λ are independent of time, i.e. λ t = 0. b. The quantity ψ t Mψ must remain a solution of eqn.1a. c. The compatibility relationship L t + LM ML = 0 must be true; this, together with property a., implies that the operator L must be self-adjoint, i.e. L is equal to its complex conjuate. 3 By integrable we mean the equation is exactly solvable; however, this begs the question: what does exactly solvable mean? There are many definitions in use, but we use it to include: i solutions obtained by linearization or direct methods Abl-11, chap. 5; ii asymptotic solutions obtained by perturbation expansion methods Dra-9, chap. 7. Graham W Griffiths 5 May 01

3 . Operator usage Because L is a composite operator which may consist of differentials, functions scalars, for consistency, each term must act as an operator. Differentials are operators by definition, therefore, non-differentials are considered to include an identity operator. Thus, as an example, an operator could be defined as, L = D n x + αux, ti, 6 where D n x represents the nth total derivative operator I the identity operator; α is a scalar u is a function of the spatial variable x temporal variable t. Alternatively, eqn 6 can be written in an equivalent form as L = n + αu, u = ux, t, 7 xn where the second term is assumed to include an implicit identity operator. Now, if L operates on an ancillary function, the result is a new function. For example, if n = the ancillary function is ψx, t, we obtain Lψ = D x + αui ψ, 8 = ψ + αuψ. 9 x However, if L operates on a second operator, the result is another operator. For example, if the second operator is M = ψi, we obtain LM = D x + αui ψi, 10 = DxψI + αuiψi, 11 ψ =D x x I + ψd x + αuψi, 1 ψ = x I + ψ x D x + ψ x D x + ψdx + αuψi, 13 = ψ x I + ψ x D x + ψd x + αuψi. 14 The above demonstrates that an operator acting upon another operator can become quite involved, particularly when either or both have multiple terms. Consequently, care has to be taken to ensure all the terms are correctly acted upon. In subsequent examples, where appropriate, we will consider the identity operator to be implicit drop the symbol I..3 Matrix form In 1974 Ablowitz, Kaup, Newell, Segur Abl-74 published a matrix formalism for Lax pairs where they introduced a construction that avoids the need to consider higher order Lax operators. This method is also referred to as the AKNS method. In their analysis they introduced the following system D x Ψ =XΨ, 15 D t Ψ =T Ψ, 16 Graham W Griffiths 3 5 May 01

4 where X T correspond to operators L M respectively, Ψ is an auxiliary vector function. Matices X T will, in general, both depend upon the time independent eigenvalue λ, the size of Ψ will depend upon the order of L. Thus, if L is of order, then vector Ψ will have two elements X T will each be a matrix. The compatibility condition for eqns is D t, D x Ψ = D t XΨ D x T Ψ = D t XΨ XD t Ψ D x T Ψ T D x Ψ = This can be written succinctly as which is known as the matrix Lax equation, where D t X D x T + X, T Ψ = 0, 18 X, T := XT T X 19 is defined as the matrix commutator. As a consequence of geometrical considerations, eqn 18 is also known as the zero-curvature equation. For a more detailed discussion of this method refer to Abl-11, Hic-1. 3 Lax pair - operator examples 3.1 The advection equation The well known 1D advection equation is defined as Now consider the Lax pair t + c x = 0, u = x, t. 0 L = ui, u = ux, t, 1 x M = c x, where again L is the Lax operator c is a constant. For this example we have included the identity operator, I, to re-inforce the earlier discussion on operator usage. Inserting eqns 1 into the Lax eqn 4, we obtain L t = = M, L = c t x, x ui, 3 where the rhs of eqn 3 is equal to LM ML = ML LM = M, L. Exping each element on the rhs of eqn 3 separately, recalling that operators operate only on terms to their right, we obtain LM = ML = x ui c x c x x ui = c 3 x 3 + cu x, 4 = c 3 x 3 + cu x + c x, 5 LM ML = c x. 6 Graham W Griffiths 4 5 May 01

5 On negating eqn 3 rearranging we arrive at t + c x = 0, 7 which is the 1D advection equation, eqn 0. If c > 0, u advects to the right, if c < 0, u advects to the left. Thus, we have shown that L M satisfy Lax eqn 4 that u satisfies the advection equation. It therefore follows that the advection equation can be thought of as the compatibility condition for the Lax pair eqns 1. Whilst this example is not particularly useful, as eqn 0 can be solved using direct methods, it does illustrate the general Lax pair idea. A Maple program that derives the above result is included in Listing 1. Listing 1: Maple program that performs a Lax pair transformation on the advection equation. # Lax pair transformation - Advection equation restart ; with DEtools : with PDEtools : alias u=ux,t; declare ux,t; _Envdiffopdomain := Dx,x; L := Dx ^ -u; M:= -c*dx; LM := exp mult L,M;ML := exp mult M,L; Commutator :=LM -ML; # Negating the Lax equation L t + LM - ML =0 we obtain pde_advec := - diff L,t+ Commutator =0; 3. Korteweg-de Vries KdV equation a The well known KdV equation is defined as Now consider the Lax pair Pol-07 t 6u x + 3 = 0, u = ux, t. 8 x3 L = x u 9 M = 4 3 x + 6u 3 x + 3 x, 30 where L is the Lax operator Lψ = λψ, eqn.1a, becomes the Sturm-Liouville equation. NOTE: we now drop the explicit inclusion of identity operators for this example, also for examples that follow. Inserting eqns 9 30 into the Lax eqn 4, we obtain L 4 t = t = 3 x + 6u 3 x + 3 x, x u, 31 where the rhs of eqn 31 is equal to LM ML = ML LM = M, L. Graham W Griffiths 5 5 May 01

6 Exping each element on the rhs of eqn 31 separately, we obtain LM ML = LM = u x5 x x x x 3u 3 x 1 x 6u x 1 x 6u x 3 ML = u x5 x x x x 9u 33 3 x 3 x + 6u x Therefore, on negating rearranging eqn 31 we finally arrive at t 6u x + 3 = 0, 35 x3 which is the familiar form of the KdV equation, eqn 8. Thus, we have shown that L M satisfy Lax eqn 4 that u satisfies the KdV equation. It therefore follows that the KdV equation can be thought of as the compatibility condition for the Lax pair eqns A Maple program that derives the above result is included in Listing. Listing : Maple program that performs a Lax pair transformation on the KdV equation. # Lax pair transformation - KdV equation restart ; with DEtools : with PDEtools : alias u=ux,t; declare ux,t; _Envdiffopdomain := Dx,x; L := Dx ^ -u; M := -4* Dx ^3+6* u*dx +3* diff u,x; LM := exp mult L,M;ML := exp mult M,L; Commutator :=LM -ML; # Negating the Lax equation L t + LM - ML =0, we obtain pde_kdv := - diff L,t+ Commutator =0; 3.3 Korteweg-de Vries KdV equation b We can also demonstrate the derivation of the previous KdV equation Lax pair example from a more general Lax pair Dra-9 - one with unknown variables. Consider the Lax pair L = u, u = ux, t, 36 x M = α 3 Bx, t Cx, t, 37 x3 x where again L is the Lax operator the variables α, B C in M have yet to be determined. Inserting eqns into the Lax eqn 4, we obtain L α t = t = 3 Bx, t x3 x Cx, t, x u, 38 Graham W Griffiths 6 5 May 01

7 where the rhs of eqn 38 is equal to LM ML = ML LM = M, L. Exping each element on the rhs of eqn 38 separately, we obtain x + C LM =α 5 3 B + αu x5 x B 3 x + ub B x C C + uc 39 x x x ML =α 5 3 B + αu x5 x 3α 3 x + C x LM ML = + ub 3α u x + uc α3 x x + B 3 x 3α 3α x B x x + x B x C x x 40 α 3 x 3 B x C x. 41 Now, for LM ML to be a multiplicative operator, the following must be true 3α x B x = 0 4 3α x B x C x = To solve for B C we integrate eqns 4 43 with respect to x. Ignoring constants of integration, we get which yields 3αu B = 0, 44 3α x B C = 0, x 45 Substituting B C into eqn 41, we obtain B = 3 αu, 46 C = 3 4 α x. 47 LM ML = 3 αu x α3 x Thus, on negating eqn 38 setting α = 4, we finally arrive at t 6u x + 3 = 0, 49 x3 which again, is the familiar form of the KdV equation, eqn 8. A Maple program that derives the above result is included in Listing 3. Graham W Griffiths 7 5 May 01

8 Listing 3: Maple program that performs a Lax pair transformation on the KdV equation. # Lax Pair transformation - KdV Equation restart ; with DEtools : with PDEtools : alias u=ux,t,b=bx,t,c=cx,t;# assume alpha = const ; declare u,b,c; _Envdiffopdomain := Dx,x; L:= Dx ^ -u; M:= alpha *Dx ^3 -B*Dx -C; LM := exp mult L,M;ML := exp mult M,L; Commutator := collect simplify LM -ML,Dx; # For Commutator to be multiplicative, the following must be true where, eqn1 := 3* alpha *ux -* Bx =0; eqn := 3* alpha *uxx -Bxx -* Cx =0; # Integrate eqn1 eqn ignoring integration constants eqn3 :=3* alpha *u -* B =0; eqn4 :=3* alpha *ux -Bx -* C =0; sol1 := solve { eqn1,eqn3, eqn4 },{B,Bx,C}; # Substitute sol1 into the Lax equation L t + LM - ML =0 LaxEqn := eval diff L,t+ subs B =3/ * alpha *u, diff B,x =3/ * alpha * diff u,x, C =3/4* alpha * diff u,x,commutator =0; # Negating LaxEqn setting alpha =-4, yields the KdV equation. -subs alpha =-4, LaxEqn ; 3.4 The modified KdV mkdv equation We have seen previously, when discussing the Bäcklund transformation, how application of the Miura transformation Miu-68a, Miu-68b given below u = v x + v, u = ux, t, v = vx, t, 50 converts the KdV equation into the modified KdV equation mkdv v v 6v t x + 3 v = x3 This suggests that it may be possible to use the Miura transformation to convert the KdV Lax pair into a mkdv Lax pair. In fact, this does turn out to be a good approach. The Miura transformation, when applied to the Lax pair for the KdV equation, given by eqns 9 30, yields the following Lax pair for the mkdv equation L = v x x + v, 5 M = 4 3 v x x + v x + 3 v x x + v. 53 Inserting eqns 5 53 into the Lax eqn 4 negating, we obtain v v + v x t t + 4 v x + v 3 v 4 x v v 3 6v x 1v 1v 3 v = 0, 54 x x which can be simplified to, v + v v 6v x t x + 3 v = x 3 Graham W Griffiths 8 5 May 01

9 Clearly, the second bracketed term of eqn 55 must be equal to zero, from which we deduce that v must be a solution to the mkdv equation, eqn 51. Thus, we have shown that L M satisfy Lax eqn 4 that v satisfies the mkdv equation. It therefore follows that the mkdv equation can be thought of as the compatibility condition for the Lax pair eqns A Maple program that derives the above result is included in Listing 4. Listing 4: Maple program that performs a Lax pair transformation on the mkdv equation. # Lax pair transformation - mkdv equation restart ; with DEtools : with PDEtools : alias u=ux,t,v=vx,t; declare u,v; _Envdiffopdomain := Dx,x; L_KdV := Dx ^ -u; M_KdV := -4* Dx ^3+6* u*dx +3* diff u,x; # Miura transformation tr :=u= diff v,x + v ^; L_mKdV := subs tr, L_KdV ; M_mKdV := subs tr, M_KdV ; LM := exp mult L_mKdV, M_mKdV ;ML := exp mult M_mKdV, L_mKdV ; Commutator :=LM -ML; # Negating the Lax equation L t + LM - ML =0, we obtain LaxEqn := - diff L_mKdV,t+ Commutator =0; pde_mkdv := diff v,t -6*v ^* diff v,x+ diff v,x,x,x; LaxEqn := exp * v* pde_mkdv + diff pde_mkdv,x; LE_Chk := simplify lhs LaxEqn -LaxEqn, size ; 3.5 The Swada-Kotera SK equation The Sawada-Kotera equation is defined as Swa-74 t + γ u 5 x + u γu 3 x + γ 3 x where a value of γ = 15 is commonly used. Now consider the Lax pair M = γu + 3γ x5 x3 x u x + 5 u = 0, u = ux, t, 56 x5 L = 3 x γu, u = ux, t, 57 x x γ u + γ u + at, 58 x x where again L is the Lax operator c is a constant. Inserting eqns into the Lax eqn 4 rearranging, we obtain γ 5 x t + γ u 5 x + γu3 x + γ u 3 x x + 5 u = x 5 Clearly, the bracketed term of eqn 59 must be equal to zero, from which we deduce that u must be a solution to the SK equation, eqn 56. Thus, we have shown that L M satisfy Lax eqn 4 that u satisfies the SK equation. It therefore follows that the SK equation can be thought of as the compatibility condition for the Lax pair eqns A Maple program that derives the above result is included in Listing 5. Graham W Griffiths 9 5 May 01

10 Listing 5: Maple program that performs a Lax pair transformation on the Sawada-Kotera equation. # Lax Pair - Sawada - Kotera Equation restart ; with DEtools : with PDEtools : alias u=ux,t: declare u; _Envdiffopdomain := Dx,x; L:= Dx ^3+1/5 * gamma *ux,t*dx; M:= 9* Dx ^5+3* gamma *ux,t*dx ^3+3* gamma * diff ux,t,x*dx ^ +1/5 * gamma ^* ux,t ^+* gamma * diff ux,t,x,x*dx + at; LM := exp mult L,M; ML := exp mult M,L; Commutator := simplify ML -LM; # Therefore from Lax1 we get : sol := diff L,t-Commutator =0; # Cancelling common terms gives SW_Eqn := simplify sol *5/ gamma *Dx,size ; 4 Lax pair - matrix examples 4.1 Korteweg-de Vries KdV equation We now demonstate the application of the matrix Lax pair method to the KdV equation t + αu x + 3 = 0, u = ux, t. 60 x3 Consider the matrix Lax pair X = 0 1 λ α 6 u 0 61 T = α 6 x 4λ + λα 3 u + α 4λ α 3 u, 6 18 u + α u α 6 x 6 x where X is the matrix equivalent of the Lax operator L, T is the matrix equivalent of the operator M. Inserting eqns 61 6 into the matrix Lax equation 18 yields cc = α 6 t + αu x + 3 = 0 3 x Therefore, the compatibility condition for X T must be t + αu x x Letting α = 6 yields the familiar KdV equation = A Maple program that derives the above result is included in Listing 6. Listing 6: Maple program that performs a Lax pair matrix transformation on the KdV equation. # Matrix Lax pair transformation - KdV equation with PDEtools : with LinearAlgebra : Graham W Griffiths 10 5 May 01

11 alias u=ux,t: declare ux,t; X:= Matrix 0, 1, lambda -1/6 * alpha *u,0 ; T:= Matrix 1/6 * alpha * diff u, x, -4* lambda -1/3 * alpha *u, -4* lambda ^+1/3 * lambda * alpha *u +1/18 * alpha ^* u ^+1/6 * alpha * diff u,x,x, -1/6 * alpha * diff u,x ; # Lax equation - the " Compatibility Condition " CC := simplify map diff,x,t-map diff,t,x+x.t-t.x= Matrix ; # Therefore, for compatibility, we must have : pde_kdv := lhs CC,1 * -6/ alpha =0; 4. Fifth order Korteweg-de Vries equation KdV5 We now demonstrate the application of the matrix Lax pair method to the fifth order KdV equation t Consider the matrix Lax pair T = + 30u x + 0 u x x + u 10u 3 x + 5 u = 0, u = ux, t x5 γ 500 γλ 3γ u γ 500 X = 30γu x + u 50 3 x 3 x 1 0 λ γλ 10 u γ u x + u 50 4 x 4 66 γ 50λ γ 500 3γu + 10 u x 30γu x + u 50 3 x 3 67 where X is the matrix equivalent of the Lax operator L, T is the matrix equivalent of the operator M. Inserting eqns into the matrix Lax equation 18 yields cc = γλ t + 3γ u u + 0γ x x x + u 10γu 3 x + u x 5 = 68 Therefore, the compatibility condition for X T must be γλ t + 3γ u u + 0γ x x x + u 10γu 3 x + u = x 5 On letting γ = 10 simplifying eqn 69 yields the KdV5 equation 65. A Maple program that derives the above result is included in Listing 7. Listing 7: Maple program that performs a Lax pair matrix transformation on the 5th order KdV equation. # KdV5 equation - Matrix Lax pair with PDEtools : with LinearAlgebra : alias u=ux,t: declare ux,t; X:= Matrix 0, 1/ lambda ^, -1/10 * gamma * lambda ^*u,0 ;,. Graham W Griffiths 11 5 May 01

12 T := gamma /500 * Matrix 10*3* gamma *u* diff u,x +5* diff u,x,x,x, -10/ lambda ^ *3* gamma *u ^+10* diff u,x,x, lambda ^*3* gamma ^* u ^3+30* gamma * diff u,x ^+40* gamma *u* diff u,x,x +50* diff u,x,x,x,x, -10*3* gamma *u* diff u,x +5* diff u,x,x,x ; # Lax equation - the " Compatibility Condition " CC := simplify map diff,x,t-map diff,t,x+x.t-t.x= Matrix ; # Therefore, for compatibility, we must have : CC1 := lhs CC,1=0; # On letting gamma =10 simplifying we obtain the # KdV5 equation in cannonical form pde_kdv5 := subs gamma =10, CC1 * -1/ lambda ^ ; 4.3 Sine-Gordon equation The Sine-Gordon equation in light-cone co-ordinates 4 is given by u sinu = 0, u = ux, t. 70 t x A matrix Lax pair for this equation is Lax-68 iλ 1 X = x 1, iλ x i = 1, 71 T = i cosu sinu 4λ sinu cosu, 7 where X is the matrix equivalent of the Lax operator L, T is the matrix equivalent of the operator M. Inserting eqns 71 7 into the matrix Lax equation 18 yields 0 u cc = t x + sinu u = t x sinu 0 Therefore, the compatibility condition for X T must be i.e the Sine-Gordon equation u sinu = 0, 74 t x A Maple program that derives the above result is included in Listing 8. Listing 8: Maple program that performs a Lax pair matrix transformation on the Sine-Gordon equation. # Sine - Gordon equation - Matrix Lax pair with PDEtools : with LinearAlgebra : alias u=ux,t: declare ux,t; X:= Matrix -I* lambda, -1/ * diff u,x, 1/ * diff u,x,i* lambda ; T := I /4* lambda * Matrix cos u, sin u, sin u, -cos u ; # Lax equation - the " Compatibility Condition " 4 A light cone is formed by all past future events that can be connected by light rays. Graham W Griffiths 1 5 May 01

13 CC := simplify map diff,x,t-map diff,t,x+x.t-t.x= Matrix ; # Therefore, for compatibility, we must have : CC_1_1 :=1/ * diff u,x,t -1/ * sin u =0; # On simplifying we obtain the Sine - Gordon equation pde_sineg := diff u,x,t-sin u =0; 4.4 Sinh-Gordon equation The Sinh-Gordon equation in light-cone co-ordinates is given by u sinhu = 0, u = ux, t. 75 t x A matrix Lax pair for this equation is Qia-0 iλ 1 X = x 1, iλ x i = 1, 76 T = i 4λ coshu sinhu sinhu coshu, 77 where X is the matrix equivalent of the Lax operator L, T is the matrix equivalent of the operator M. Inserting eqns into the matrix Lax equation 18 yields u 0 cc = t x sinhu u t x sinhu 0 Therefore, the compatibility condition for X T must be i.e the Sinh-Gordon equation 75. =. 78 u sinhu = 0, 79 t x A Maple program that derives the above result is included in Listing 9. Listing 9: Maple program that performs a Lax pair matrix transformation on the Sinh-Gordon equation. # Sinh - Gordon equation - Matrix Lax pair with PDEtools : with LinearAlgebra : alias u=ux,t: declare ux,t; X:= Matrix -I* lambda, 1/ * diff u,x, 1/ * diff u,x,i* lambda ; T := I /4* lambda * Matrix cosh u, -sinh u, sinh u, -cosh u ; # Lax equation - the " Compatibility Condition " CC := simplify map diff,x,t-map diff,t,x+x.t-t.x= Matrix ; # Therefore, for compatibility, we must have : CC_1_1 :=1/ * diff u,x,t -1/ * sinh u =0; # On simplifying we obtain the Sinh - Gordon equation pde_sinhg := diff u,x,t-sinh u =0; Graham W Griffiths 13 5 May 01

14 4.5 Liouville equation The Liouville equation in light-cone co-ordinates is given by A matrix Lax pair for this equation is Wu-9 X = x λ u expu = 0, u = ux, t. 80 t x T = λ x 0 1 λ expu 81, 8 where X is the matrix equivalent of the Lax operator L, T is the matrix equivalent of the operator M. Inserting eqns 81 8 into the matrix Lax equation 18 yields cc = u t x expu 0 0 u t x + expu Therefore, the compatibility condition for X T must be i.e the Liouville equation 80. =. 83 u expu = 0, 84 t x A Maple program that derives the above result is included in Listing 10. Listing 10: Maple program that performs a Lax pair matrix transformation on the Liouville equation. # Liouville equation - Matrix Lax pair with PDEtools : with LinearAlgebra : alias u=ux,t: declare ux,t; X:= Matrix diff u,x, lambda, lambda, -diff u,x ; T:= Matrix 0, exp * u/ lambda, 0, 0 ; # Lax equation - the " Compatibility Condition " CC := simplify map diff,x,t-map diff,t,x+x.t-t.x= Matrix ; # Therefore, for compatibility, we must have : pde_lv := lhs CC 1,1=0; References Abl-11 Ablowitz, M. J., 011. Nonlinear Dispersive Waves: Asymptonic Analysis Solitons, Cambridge University Press, Cambridge. Abl-74 Ablowitz, M. J., D. J. Kaup, A. C. Newell, H. Segur, The inverse scattering transform - Fourier analysis for nonlinear problems, Stud. Appl. Math Graham W Griffiths 14 5 May 01

15 Abl-91 Ablowitz, M. J. P. A. Clarkson Solitons, Nonlinear Evolution Equations Inverse Scattering London Mathematical Society Lecture Note Series, 149, Cambridge University Press. Dra-9 Drazin, P. G. R. S. Johnson 199. Solitons: an introduction, Cambridge University Press, Cambridge. Gar-67 Gardener, C. S., J. Greene, M. Kruskal, R. M. Miura, Method for solving the Korteweg-de Vries equation. Phys. Rev. Lett. 19, Hic-1 Hickman, M., W. Hereman, J. Larue, Ü. Gökta, 01. Scaling invariant Lax pairs of nonlinear evolution equations. Applicable Analysis, 91, Inf-00 Infield, E. G. Rowls 000. Nonlinear Waves, Solitons Chaos, nd Ed., Cambridge University Press. Joh-97 Johnson, R. S. 1999, A Modern Introduction to the Mathematical Theory of Water Waves, Cambridge University Press. Lax-68 Lax, P Integrals of nonlinear equations of evolution solitary waves, Comm. Pure Applied Math., 1 5, Miu-68a Miura, R. A Korteweg-de Vries Equation Generalizations I. A remarkable explicit transformation. J. Math. Phys., 9, pp10-4. Miu-68b Miura, R. A., C. S. Gardner M. D. Kruskal Korteweg-de Vries Equation generalizations II. Existence of conservation laws constants of motion. J. Math. Phys., 9, Pol-07 Polyanin, A. D. A. V. Manzhirov 007, Hbook of mathematics for engineers scientists, Chapman Hall/CRC Press. Qia-0 Qiao, Z. W. Strampp, 00. Negative order MKdV hierarchy a new integrable Neumann-like system, Physica A: Statistical Mechanics its Applications, 313, Swa-74 Sawada, K. T. Kotera, A method for finding N-soliton solutions of the K. d. V. equation K. d. V. -like equations. Progr. Theroret. Phys. 54, Wu-9 Wu, P., X. Guan, Z. Xiong, H. Zhou, 199. A note on infinite conservation laws in Liouville equation. Chin. Phys. Lett., 9 7, Graham W Griffiths 15 5 May 01