Académique Documents
Professionnel Documents
Culture Documents
D.G. Marshall.
1
Keywords 2
Keywords
Integrability
Nijenhuis tensor
Haantjes tensor
Hydrodynamic chains
Benney Chain
Diagonalizability
Generating function of conservations laws
Abstract 3
Abstract
Acknowledgements
I would like to express my gratitude to the following people for their assistance
during my work on this thesis:
Contents
1 Introduction 7
3 Hydrodynamic Chains 38
3.1 The Benney chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.2 Hydrodynamic chains and the Haantjes tensor . . . . . . . . . . . . . 43
3.3 Another chain considered by Benney . . . . . . . . . . . . . . . . . . 53
3.4 Hydrodynamic reductions and diagonalizability . . . . . . . . . . . . 54
4 Conservative chains 65
4.1 Egorov case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.2 Method of hydrodynamic reductions . . . . . . . . . . . . . . . . . . 67
4.3 Integration of the Egorov case . . . . . . . . . . . . . . . . . . . . . . 72
4.4 More general conservative chains . . . . . . . . . . . . . . . . . . . . 76
4.5 Integration of the general conservative chains . . . . . . . . . . . . . . 79
5 Kupershmidt’s brackets 89
5.1 Integration of the system of integrability conditions . . . . . . . . . . 91
5.2 The vanishing of the Haantjes tensor . . . . . . . . . . . . . . . . . . 93
Contents 6
1 Introduction
ut = v(u)ux (1)
Some systems of the form (1) are diagonalizable, i.e. reducible to the Riemann
invariant form, more details follow in Sect. 2. The diagonalizability of (1) is a
necessary condition for integrability via the generalized hodograph method [61].
There exists an efficient tensor criterion of the diagonalizability which does not
require the computation of eigenvalues and eigenvectors of the matrix v(u). This is
Introduction 8
i
Njk = vjp ∂up vki − vkp ∂up vji − vpi (∂uj vkp − ∂uk vjp ), (2)
i i p r p i r p i r p i r
Hjk = Npr vj vk − Njr vp vk − Nrk vp vj + Njk vr vp . (3)
For strictly hyperbolic systems, i.e. when the eigenvalues are positive and distinct,
the condition of diagonalizability is given by the following theorem:
The focus of this work is hydrodynamic chains, these are infinite component systems
of form (1). More formally a hydrodynamic chain is,
ut = V (u)ux
Definition 1 [20] An infinite matrix V (u) is said to belong to the class C (chain
class) if it satisfies the following two properties:
(a) each row of V (u) contains finitely many nonzero elements;
(b) each matrix element of V (u) depends on finitely many variables ui .
Introduction 9
All the chains considered throughout the paper are of the class C. Below is the
Benney chain written in matrix form, the matrix is of class C, it has only two
nonzero elements in every row apart from the first where there is only one. The sole
element in the first row is constant while of the two elements on every other row
one is constant and the other is only a function of one variable. The Benney chain
has the form
A0 0 1 0 0 0 0 . A0
1 0 1
A A 0 1 0 0 0 . A
A2 + 2A1 0 0 1 0 0 . A2 = 0.
3
3A2
3
A 0 0 0 1 0 . A
· . . . . . . . ·
t x
The Benney chain is a classical example of a hydrodynamic chain. This chain was
derived in [3] and was shown to consist of an infinite number of conservation laws
for the system of equations that describe fluid under the action of gravity. The
derivation of the Benney chain is shown in Sect. 3.1.
unt = un+1
x + u1 unx + cn un u1x
where cn = const. It can be verified that the vanishing of the Haantjes tensor implies
the relation cn+2 = 2cn+1 − cn . Since the form of this chain is relatively simple it
is straightforward to calculate the Haanjtes tensor by hand though it is quicker to
using symbolic computations. Throughout this thesis the Mathematica [43] package
has been used, the three files contained in the Appendices can be downloaded from
http://www-staff.lboro.ac.uk/%7emakk/Marshall.html. The PDF file of this theis
is also available online at the fore mentioned address.
If the spectrum of the infinite matrix V is simple, that is, for any λ there exists a
unique eigenvector ξ such that V ξ = λξ, the following stronger result is obtained:
Theorem 3 In the simple spectrum case the vanishing of the Haantjes tensor H is
necessary and sufficient for the existence of two-component reductions parametrized
by two arbitrary functions of a single variable.
Introduction 11
The necessary part of this conjecture, the claim that the integrability implies that
both H and P vanish identically, is proved in Sect. 3.4. The sufficiency is a much
more difficult property to show, and this is not yet established in general. We point
out that the vanishing of H (in fact, the vanishing of the first few components of
H), is already sufficiently restrictive and implies the integrability in many cases (e.g.
for conservative chains, Hamiltonian chains, etc).
The chain (4) was investigated in [47], where a class of new integrable examples
was found, see also [6], based on the symmetry approach. These papers provide a
classification of conservative chains of the form (4). It turns out that the condi-
1
tions Hjk = 0 are already sufficiently restrictive and give an over-determined system
expressing all second order partial derivatives of h(u1 , u2 , u3 , u4 ) in terms of g, see
(103). The consistency conditions of these equations lead to a system of equations
expressing all third order partial derivatives of g in terms of its lower order deriva-
tives, see (104). Computer algebra was used to calculate the Haantjes tensor and to
verify the involutivity by calculating the consistency conditions. It must be empha-
sized that the same system of equations for g was found in [47] using the symmetry
approach, as well as in [17] based on the method of hydrodynamic reductions. So,
for conservative chains (4) the condition of diagonalizability is equivalent to the
i
integrability. The requirement of the vanishing of other components Hjk , i ≥ 2,
imposes no additional constraints on h and g: these conditions reconstruct the re-
m
maining equations of the chain (4). Furthermore, the conditions Hjk = 0 specify
the right hand side of the mth equation um
t = ..., etc.
Also contained in Sect. 4 is the classification of conservative chains where the flux
of the first equation is f (u1 , u2 ), i.e. chains of the type
1
As in the Egorov case, the conditions Hjk = 0 lead to expressions for all second
order partial derivatives of h in terms of g and f . The consistency conditions of
these equations result in a system of equations expressing all third order partial
derivatives of g and f in terms of lower order derivatives.
Sect. 5 and Sect. 6 are devoted to the classification of Hamiltonian chains which
Introduction 13
According to the results of Tsarev [61], the vanishing of the Haantjes tensor is nec-
essary and sufficient for the integrability of finite-component Hamiltonian systems
of hydrodynamic type by the generalized hodograph method. Thus, we formulate
our main conjecture regarding Hamiltonian hydrodynamic chains:
Conjecture 2 [20] The vanishing of the Haantjes tensor is a necessary and suffi-
cient condition for the integrability of Hamiltonian hydrodynamic chains. In par-
ticular, it implies the existence of infinitely many Poisson commuting conservation
laws, and infinitely many hydrodynamic reductions.
The necessity part of this conjecture follows from Theorem 2, which is proved in
Sect 3.4, which gives the vanishing of the Haantjes tensor as a necessary condition
Introduction 14
for the integrability of hydrodynamic chains (not necessarily Hamiltonian). The suf-
ficiency is more difficult to establish. The conjecture is supported by all examples
of integrable Hamiltonian chains known to us. Since components of the Haantjes
tensor can be calculated using computer algebra, this approach provides an effective
classification criterion. We are going to demonstrate that the conjecture is indeed
true for Hamiltonian chains of the type (6) with Kupershmidt’s bracket and Manin-
Kupershmidt’s bracket, respectively.
We also investigate the existence of higher order (than the hamiltonian, h) con-
servation laws for the two types of Hamiltonian systems in question; recall that
the existence of infinitely many conservation laws is one of the main features of
the integrability. We show that the requirement of the existence of one additional
conservation law of the form,
1
leads to the same relations as requiring Hjk = 0 for the respective systems. Further-
1
more the system of equations obtained from setting Hjk = 0, imply the existence
of a generating function of conservation laws and, hence, an infinity of conservation
laws. This establishes the integrability of all examples constructed in Sect. 5 and
Sect. 6.
ut = v(u)ux , (7)
Consider (7), such a system possesses Riemann invariants if there exists a change of
variables
If (7) possesses Riemann invariants then it is said that the matrix v(u) is diagonal-
izable.
1. Write the equations in the form (7) and find the eigenvalues of matrix v(u),
det v(u) − λi I = 0.
2. Require the gradients of the Riemann invariants are left eigenvectors of matrix
v corresponding to eigenvalues λ1 , λ2 .
grad Ri v(u) − λi I = 0.
3. Solve the resulting two pairs of PDE’s to obtain the Riemann invariants
R1 , R2 ,
∂R1 ∂R1 1
( , ) v(u) − λ I = 0,
∂u1 ∂u2
∂R2 ∂R2
) v(u) − λ2 I = 0.
( 1, 2
∂u ∂u
ρ u ρ ρ
+ = 0. (11)
γ−2
u γρ u u
t x
Hydrodynamic type systems in 1+1 dimensions 17
u1 ρ u ρ
= , vji = .
u2 u γργ−2 u
the same is done for λ2 , R2 . In the third step, multiply out and solving we see that
1 2γ 1/2 ρ(γ−1)/2
R =u+ .
γ−1
γ−1 γ−1
1 2γ 1/2 ρ 2 2 2γ 1/2 ρ 2
R =u+ , R =u− . (13)
γ−1 γ−1
1 γ−1
R1 + R2 1 4γ 2 ρ 2
2
= u, R −R = . (15)
2 γ−1
Hydrodynamic type systems in 1+1 dimensions 18
R1 + R2 (γ − 1)(R1 − R2 ) R1 + R2 (γ − 1)(R2 − R1 )
λ1 = + , λ2 = + . (16)
2 4 2 4
R1 + R2 (γ − 1)(R1 − R2 )
Rt1+ + Rx1 = 0,
2 4
1 2
(γ − 1)(R2 − R1 )
2 R +R
Rt + + Rx2 = 0. (17)
2 4
One can verify that (11) indeed goes to (17) under the change of variables
1 γ−1
R1 + R2 1 4γ 2 ρ 2
2
= u, R −R = .
2 γ−1
Example [38] This is a much more involved example of determining the Riemann
invariants for a given system. Equations of Chromatography are
Different models are obtained via the choice of ai (u), called isotherms of adsorption.
In this example we are going to obtain the Riemann invariants for the Langmuir
isotherm, thus,
n
i Γi ui X
a (u) = , ∆=1+ Γs us . (19)
∆ s=1
Hydrodynamic type systems in 1+1 dimensions 19
It was shown in [38] that the characteristic equation, det(A − λI) = 0, has the form,
n
!
∆Γ1 − Γ21 u1 − α1 ∆2 X Γ2s un (∆λ1 − Γ1 )
det(A − λI) = +
∆2 s=2
∆2 (Γs − ∆λs )
Γ2 − ∆λ2 Γn − ∆λn
...
∆ ∆
n
! n
Γ1 − ∆λ1 (∆λ1 − Γ1 ) X Γ2s un Γp − ∆λp
Y
= + +
∆ ∆2 Γ − ∆λs
s=1 s p=2
∆
n n n
Γp − ∆λp 1 Y Γs − ∆λs X Γ2s us
Y
= − ,
p=1
∆ ∆ s=1 ∆ s=1
Γs − ∆λs
so,
n
X Γ2s us
∆= , (21)
s=1
Γs − ∆λs
n n
X
s
X Γ2s us
1+ Γs u = (22)
s=1 s=1
Γs − ∆λ
R
and then substitue to show Rx + R
∆ t
= 0.
X Γs R Γs us p upt Γp Γs us
P
t
Rx = − 2
− ,
s
Γs − R ∆ ∆
P Γs ust
s Γs −R
Rt = −R P up Γ2p
.
p (Γp −R)2
n n n
X Γ2s ust X p
X Γs ust
− + Γp ut + R = 0, (23)
Γ − R p=1
s=1 s
Γ −R
s=1 s
R
one can see that coefficients at ukt cancel out. Thus, Rx + ∆ Rt = 0 modulo R = λ∆,
as required.
Qn
Rs i
Rxi i
+ R Qs=1
n Rt = 0. (24)
s=1 Γs
n
X Γ2s us
= ∆,
s=1
Γs − R
write as a polynomial
n
Y n
Y n
Y
2 1 2 n
∆ (Γs − R) − Γ1 u (Γs − R) − · · · − Γn u (Γs − R) = 0.
s=1 s6=1 s6=n
Hydrodynamic type systems in 1+1 dimensions 21
n n
! n
Y X Y
i
∆ Γs − Γi u Γs = 0,
s=1 i=1 s=1
Q
Γs
Rn + · · · + (−1)n s
= 0.
∆
n Q
n
Y
s n Γs
s
(−1) R = (−1) ,
s=1
∆
Γs
Q
thus, ∆ = s Rs , so we have (24) as required.
Now it is claimed that the formulas that connect Riemann invariants Ri with vari-
ables ui are,
Γi
Q
1 s6=q Rs
−1
ui = Q Γi
. (25)
Γi s6=i Γs
−1
Indeed, this can be shown to be true as follows. From (24) we know that
n
i Ri i
Y Rs
λ = =R .
∆ s=1
Γs
Qn Γs
This implies that ∆ = s=1 Rs . Next, take (21) and substitute (25) in to eliminate
Hydrodynamic type systems in 1+1 dimensions 22
n n
Γi s ( RΓis − 1)
Q
X Γ2s us X
∆ = =
Γ − Rk (Γi − Rk ) s6=i ( ΓΓsi − 1)
Q
s=1 s i=1
n Qn n Q s
Γi − Rs s6=k Γi − R
Y Γs X Y Γs X
s=1Q
= =
Rs i−1 (Γi − Rk ) ns6=i (Γi − Γs )
Q
s s
Rs i=1 s6=i (Γi − Γs )
Y Γs
= s
.
s
R
So, (25) is consistent as it gives the correct expression for ∆ when substituted into
(21). Thus, it has been shown that the system (19) has Riemann invariants and the
diagonal form is given in (24).
For two systems to commute means that they are consistent with each other. Take
two systems of type (7), ut = V (u)ux , uy = W (u)ux , for them to commute means
uty = uyt .
here t and y are the corresponding ’times’. It is claimed that for these equations to
be consistent then the following condition must be true,
∂j λi ∂j µi
= , i 6= j. (27)
(λj − λi ) (µj − µi )
∂
Where ∂j = ∂Rj
.
Hydrodynamic type systems in 1+1 dimensions 23
i i
Rty = Ryt . (28)
i
Rty = (λi Rxi )y = ∂j λi Ryj Rxi + λi Rxy
i
i
Ryt = (µi Rxi )t = ∂j µi Rtj Rxi + µi Rxt
i
= ∂j µi (λj Rxj )Rxi + µi (λi Rxi )x
i
coefficients of Rxx cancel, leaving
∂j λi (µj − µi ) = ∂j µi (λj − λi ).
Which gives (27). (27) shall be referred to as the commutativity condition [61]. Fur-
thermore, if (27) is satisified then the systems (26) are said to be commuting flows.
Rt1 = R2 Rx1 ,
Rt2 = R1 Rx2 ,
in the form,
1 ∂2 µ1 1 ∂1 µ2
= , = . (33)
R1 − R2 µ2 − µ1 R2 − R1 µ1 − µ2
which must hold identically by the virtue of (9). The quantities f (R) and g(R) are
called the conserved density and the flux, respectively. By substituting (9) into (35)
we see that
∂i f λi Rxi = ∂i gRxi
∂j ∂i g = ∂i ∂j g.
∂j λi ∂i λj
∂i ∂j f = ∂i f + ∂j f, i 6= j.
λj − λi λi − λj
here λ1 = R2 , λ2 = R1 . The equation (36) for conserved densities takes the form
1 1 ∂1 f − ∂2 f
∂2 ∂1 f = ∂1 f + 2 ∂2 f = 1 . (37)
R1 −R 2 R −R 1 R − R2
which can be verified by a straightfoward differentiation. Here p(R1 ) and q(R2 ) are
arbitrary functions of one variable.
Remark The condition (35) is equivalent to the 1-form f dx+gdt being closed. This
Hydrodynamic type systems in 1+1 dimensions 26
d2 F = fx d2 x + ft dxdt + gx dtdx + gt d2 t
d2 F = (ft − gx )dxdt.
∂j λi ∂k λi
∂k = ∂j (38)
λj − λi λk − λi
∂j λi
for any i 6= j 6= k 6= i. Introducing aij = λj −λi
one can rewrite the previous equation
in the simpler form
∂k aij = ∂j aik . (39)
Work out ∂j ∂k µi and ∂k ∂j µi and use (40) and (41) to eliminate ∂n µm and ∂m anp ,
∂j ∂i f = aij ∂i f + aji ∂j f,
Set these last two equations equal to each other and collect coefficients of fi , fj , fk .
The coefficients of fi immediately cancel to give 0, while the coefficients of fj , fk are
0 due to (40). So we know all mixed partial derivatives of all f , we also know that
all these partial derivatives are consistent. Thus conserved densities depend on n
arbitrary functions of one variable, indeed one can define f arbitrarily on any of the
coordinate lines.
The generalized hodograph method allows the general solution of a system of the
form (7) to be found. First we will demonstrate with a scalar example and then give
a more general outline of the generalized hodograph method. Consider the scalar
Hopf equation
Rt = RRx . (46)
It is known that the general solution of (46) is given by the implicit formula
µ(R) = x + Rt (47)
1
µ0 R x = 1 + R x t ⇒ R x = ,
µ0
−t
R
µ0 R t = R + R t t ⇒ R t = 0 . (48)
µ −t
Hydrodynamic type systems in 1+1 dimensions 29
By substituting (48) into (46) we obtain the identity. Thus, (47) is the general
solution.
is given by
where λi (R) satisifes (38) and µi (R) are characteristic speeds of commuting flows:
∂j µi ∂j λi
= j i 6= j. (51)
µj − µi λ − λi
∂j µi = ∂j λi t. (53)
1 λi
Rxi = , Rti = . (54)
∂i µ − ∂i λi t
i ∂i µi − ∂i λi t
Hydrodynamic type systems in 1+1 dimensions 30
Thus by substituing (54) into (49) we obtain the identity so (50) is the general so-
lution. General solution means that (50) has the same amount of freedom as the
system (46) being solved. Considering the initial value problem, t = 0, of (49), all
Ri can be defined arbitrarily at t = 0, thus we have the freedom of n arbitrary
functions of one variable in the system being solved. In (51) all derivatives of the
solution, µi , are defined apart from the ith one, thus µi can be defined arbitrarily on
the Ri − axis, thus there are n arbitrary functions of one variable.
The Haantjes tensor (3) is used to verify the diagonalizability of the system (7).
Calculating the Haantjes tensor using computer algebra is relatively straightforward
for a well defined matrix and thus is an efficient way to find out if the system is
diagonalizable whereas calculating the eigenvalues and eigenvectors, in some cases,
is extremely difficult. As defined previously the Nijenhuis tensor and the Haantjes
tensor are,
i
Njk = vjp ∂p vki − vkp ∂p vji − vpi ∂j vkp + vpi ∂k vjp .
i i p r p i r p i r p i r
Hjk = Npr vj vk − Njr vp vk − Nrk vp vj + Njk vr vp ,
∂
where ∂s = ∂us
. Both the Nijenhuis tensor and the Haantjes tensor are skew sym-
metric in lower indices, this means reversing the two lower indices changes the sign,
i i i
Hjk = −Hkj . A direct corollary of this is Hjj = 0 for j = 1 · · · n.
i
Theorem 6 The matrix v(u) from (7) can be diagonalized if and only if Hjk ≡ 0.
Hydrodynamic type systems in 1+1 dimensions 31
Proof To prove this we first derive the necessary condition for the Haanjtes tensor
to be identically zero. Then we obtain the condition for v(u) to be diagonalizable
and see that they are equivalent. Evaluated on pairs of vector fields X, Y , the
Nijenhuis tensor and the Haantjes tensor can be written in the form,
j j
where [Z, Y ] = (Z i δY
δui
− Y i δZ ) δ . Let Xi , Xj be eigenvectors of v, thus,
δui δuj
vXi = λi Xi , (57)
= (λi λji − λj λji )Xj + (λi λij − λj λij )Xi + (λj λi + v 2 − (λi + λj )v)[Xi , Xj ].
Then using (58) calculating the Haantjes tensor and collecting terms at Xi , Xj , [Xi , Xj ],
we obtain
The coefficients of Xi , Xj cancel identically due to the fact vXk = λk Xk . Next, let
the expansions of commutators [Xi , Xj ] be nk=1 Cijk Xk . Thus,
P
Cijk = 0, i 6= j, i 6= k, k 6= j. (59)
The matrix v is diagonalizable if and only if each distribution < Xi , Xj > is inte-
grable, i.e [Xi , Xj ] ∈ span(Xi , Xj ), which is exactly (59). For an explanation let
us consider 3D space. Take X1 (u1 , u2 , u3 ), X2 (u1 , u2 , u3 ), X3 (u1 , u2 , u3 ) as eigenvec-
tors. We are looking for a change of variables of type (8). The Riemann invariant
R3 (u1 , u2 , u3 ) is defined by the equations:
X1 R3 = 0, X2 R3 = 0,
X1 R2 = 0, X3 R2 = 0,
Hydrodynamic type systems in 1+1 dimensions 33
Is (60) diagonalizable and under what restrictions on q(a)? To answer this ques-
tion we calculate all components of the Haanjtes tensor for the system (60). First
i i i
calculate Njk . Note that Njk = −Nkj .
1 1 1 2
N12 = 0, N13 = −1, N23 = 0, N21 = 0,
2
N23 2
= −1, N13 3
= 0, N31 3
= 0, N32 3
= 0, N12 = −aq 0 (a). (61)
i i i
Now we can calculate Hjk . Note that Hjk = −Hkj
1
H23 1
= 0, H13 1
= 0, H12 = −aq 0 (a), H13
2
= −aq 0 (a), H23
2
= 0,
2
H21 = −avq 0 (a), H12
3
= av 2 q 0 (a), H32
3
= a2 q 0 (a), H31
3
= avq 0 (a).
i
Thus, Hjk ≡ 0 if and only if q 0 (a) = 0. So, q(a) must be a constant. Thus (60) is
diagonalizable if q(a) is constant.
Hydrodynamic type systems in 1+1 dimensions 34
Let B(u)dx + A(u)dt and N (u)dx + M (u)dt be two conservation laws of the system
(7), equivalently we can view them as 1-form closed by virtue of (7). One can
introduce new variables X and T defined by
dX = Adx + Bdt, dT = M dx + N dt
substitue these into (7) and write in the same form we get,
uT = w(u)uX ,
where w(u) = (Av − BI)(N I − M v)−1 . These transformations originate from gas
dynamics.
(Rs + αβ
Q
β Rs
Q
n+1
dT = α Q s dt + Q s dx ,
R α Γ
Q s δ
(R + γ δ Rs
Q Y
n+1
dX = γ Q s dt + Q s dx Γs ,
R γ Γ
Hydrodynamic type systems in 1+1 dimensions 35
αRi +β
and the new dependent variable R̃ = γRi +δ
. So, (24) becomes
!
s β Y Rs Y Rs + β
δ nR + α
Y
i n s i i
RX γ δ (R + + RT βα Q +R RTi αn+1 α
s
γ Γs s
Γs s
Rs
!
Y (Rs + γδ ) Y
i n+1
+RX γ Γs = 0.
Rs s
After simplification,
! !
Y Y Ri
(δ + γRi ) (γRs + δ) RX
i
+ (αRi + β) (αRs + β) Q T = 0.
s s
Γs
s
Q
Divide by s (γR + δ) to get
Y R̃s
i i i
(δ + γR )RX + (αR + β) RTi = 0.
s
Γs
Y R̃s
i
RX + R̃i RTi = 0.
s
Γs
αδ−γβ
Multiply both sides by (γRi +δ)2
to obtain
Y R̃s
i
ṘX + R̃i R̃Ti = 0.
s
Γs
dX = Adx + Bdt,
dT = M dx + N dt
to (9) we obtain
i λi A − B i
RX = R
N − λi M T
λi A − B
i
µ = .
N − λi M
µij
We will start by expanding µ −µi
j and using the fact that Bj = λj Aj , Nj = λj Mj ,
k
µij
=
µj − µi k
(N − λi M )(λij A + (λi − λj )Aj ) − (λi A − B)(λij M + (λi − λj )Mj )
(N − λj M )
(N − λi M )(λj − λi )(AN − M B)
λij λij M (N − λi M )Aj − (λi A − B)Mj N − λj M
= − −
N − λi M N − λi M AN − M B N − λi M
i i i
λj Nj − λ j M − λ M j Nj − λ i M j
= + − −
N − λi M N − λi M N − λi M
((N − λi M )Aj − (λi A − B)Mj ) (N − λj M )
(AN − M B)(N − λi M )
λij (AN )j − (BM )j
+ ln(N − λi M ) j −
= i
N −λM AN − BM
i i
!
λj
N −λM
= + ln .
λj − λi AN − M B j
k
Hydrodynamic type systems in 1+1 dimensions 37
µij µik
=
µj − µi k µk − µi j
as required.
Hydrodynamic Chains 38
3 Hydrodynamic Chains
ut = V (u)ux
A0t + A1x = 0,
...
Hydrodynamic Chains 39
0 0
A 0 1 0 0 0 0 . A
1 0 1
A A 0 1 0 0 0 .
A
A2 + 2A1
A = 0.
0 0 1 0 0 . 2
3
3A2
3
A 0 0 0 1 0 . A
· . . . . . . . ·
t x
Let us take a moment to look at how and where the (63) first appeared. This chain
was derived in [3] and was shown to consist of an infinite number of conservation
laws for the system of equations that describe long waves on shallow ideal fluid
under the action of gravity. These equations are,
ux + vy = 0, (64)
v = 0, y = 0, (66)
ht + uhx − v = 0, y = h. (67)
Here h(x, t, ) is the height of the water, u(x, y, t), v(x, y, t) are the horizontal and
vertical velocities of the water, g is gravity. The conservation laws of mass, momen-
tum and energy are all straightforward to derive. Equation (67) can be rewritten
as
Z h
∂
ht + udy = 0; (68)
∂x o
Hydrodynamic Chains 40
(64) is used to eliminate v. This is the mass conservation law. Integrate (65) w.r.t
y from 0 to h to obtain the momentum conservation law:
h h
gh2
Z Z
∂ ∂ 2
udy + u dy + = 0; (69)
∂t o ∂x o 2
(64), (65), (67) have each been used. Finally multiply (65) by u and integrate w.r.t
y from 0 to h to obtain the energy conservation law:
Z h
∂ 1 h 3
Z h
gh2
Z
∂ 1 2
u dy + + u dy + gh udy = 0. (70)
∂t 2 o 2 ∂x 2 o o
has been used. Benney went on to show that there are an infinte number of such
conservation laws as follows. Let us introduce the notation
Z h
An = un (x, y, t)dy,
0
Rh
so A0 = h, A1 = 0
u(x, y, t)dy, · · · , etc, then by multiplying (65) by un−1 and then
integrating from y = 0 to y = h. Using (64), (65), (66) we obtain an unclosed set of
equations for the moments:
These equations are not conservative and nor is it in any way clear that they can
be brought into such a form. The existence of an infinity of conservation laws was
Hydrodynamic Chains 41
demonstrated in [3, 32]. In our discussion we follow [22] which extends these ideas.
Instead of dealing directly with (71) introduce a generating function
u1 u2 u3
λ=p+ + 2 + 3 + ... (72)
p p p
λt − pλx = λp pt − (p2 /2 + u1 )x .
(73)
Let us show that (72) does indeed satisfy (73) modulo (63). Using (73) and differ-
entiating we obtain,
λt − pλx = λp pt − (p2 /2 + u1 )x ,
where k = 1, 2, 3, . . . , now collect t derivatives on the left hand side and x derivatives
on the right hand side,
ukt uk+1
x kuk u1x
= + , k = 1, 2, 3, . . .
pk pk pk+1
ukt = uk+1
x + (k − 1)uk−1 u1x , k = 1, 2, 3, . . .
pt = (p2 /2 + u1 )x
where one has to substitute the expression for p(λ) obtained from (72): p = λ −
H1 H2 H3
λ
− λ2
− λ3
− .... Explicitly, one gets H 1 = u1 , H 2 = u2 , H 3 = u3 + (u1 )2 , etc:
Ht1 = (H 2 )x ,
H1
Ht2 = (H 3 − )x , (77)
2
Ht3 = (H 4 − H 2 H 1 )x ,
(H 2 )2
4 5 3 1
Ht = H −H H − .
2 x
Hydrodynamic Chains 43
Both the tensors (2) and (3) make perfect sense for infinite matrices which are of
class C as stated in the introduction. For matrices from the class C all contractions
in the expressions (2) and (3) reduce to finite summations so that each particular
i
component Hjk is a well-defined object which can be effectively computed. More-
over, for a fixed value of the upper index i there exist only finitely many non-zero
i
components Hjk .
Although we have stated the above definition it must be noted that the Riemann
invariants of a hydrodynamic chain cannot be calculated because the chain is infinite.
Example In this example we illustrate how the Nijenhuis tensor and the Haantjes
tensor can be calculated for a general hydrodynamic chain of class C. First, calculate
the Nijenhuis tensor as this is required to calculate the Haantjes tensor. Then
1
calculate the Haanjtes tensor, in this example the upper index is fixed at 1, Hjk .
Consider the infinite system
u3t + u4x + v33 (u1 , u2 , u3 )u3x + v23 (u1 , u2 , u3 )u2x + v13 (u1 , u2 , u3 )u1x = 0,
...
Hydrodynamic Chains 44
1
u v11 1 0 0 0 0 . u 1
2 2 2
u
v1
v22 1 0 0 u
0 .
u3 + v 3 v23 v33 u = 0.
1 0 0 . 3
1
4 4
v24 v34 v44
4
u v1 1 0 . u
· ··· ··· ··· ·
t x
(79)
v i (u1 , . . . , ui ) k <i+1
k
vki = 1 k =i+1
0
k > i + 1.
i
To calculate Njk the index i will be considered fixed, this implies that we need only
i
consider j ≤ i + 1, k ≤ i + 1, 1 ≤ p ≤ i + 1. Similarly when calculating Hjk with
i fixed, implies that we need only consider j ≤ i + 4, k ≤ i + 4, 1 ≤ p, r ≤ i + 3.
In general if vki = 0 ∀k > i + l, then to calculate Njk
i
we only need to consider
i
j ≤ i + l, k ≤ i + l, 1 ≤ p ≤ i + l. To calculate Hjk we only need to consider
j ≤ i + 4l, k ≤ i + 4l, 1 ≤ p, r ≤ i + 3l.
i ∂
Calculate Njk , i ≤ 3, j, k ≤ 5, 1 ≤ p ≤ 4. The notation ∂j = ∂uj
is used in the
Hydrodynamic Chains 45
following evaluation.
∂1 v11 + ∂1 v22 − ∂2 v12 j = 2, k = 1
1
Njk = −∂1 v11 − ∂1 v22 + ∂2 v12 j = 1, k = 2
0
else
(v22 − v11 )∂1 v22 + ∂1 v12 + ∂1 v23 − v12 ∂2 v22 − ∂2 v13 j = 2, k = 1
∂1 v33 + ∂2 v12 − ∂3 v13 j = 3, k = 1
(v 1 − v22 )∂1 v22 − ∂1 v12 − ∂1 v23 + v12 ∂2 v22 + ∂2 v13 j = 1, k = 2
1
2
Njk = ∂2 v22 + ∂2 v33 − ∂3 v23 j = 3, k = 2
−∂1 v33 − ∂2 v12 + ∂3 v13 j = 1, k = 3
−∂2 v22 − ∂2 v33 + ∂3 v23 j = 2, k = 3
0 else
Hydrodynamic Chains 46
∂1 (v13 + v24 ) + (v33 − v11 )∂1 v23 + v23 ∂1 v22 + (v22 − v33 )∂2 v13
−∂2 (v12 v23 ) − ∂2 v14 + v23 ∂3 v13 − v13 ∂3 v23
j = 2, k = 1
(v33 − v11 )∂1 v33 + ∂1 v34 + ∂2 v13 − v12 ∂2 v33 − v13 ∂3 v33 − ∂3 v14 j = 3, k = 1
∂1 v44 + ∂3 v13 − ∂4 v14 j = 4, k = 1
(v11 − v33 )∂1 v23 − v23 ∂1 v22 − ∂1 (v24 + v13 ) + ∂2 (v12 v23 )
+∂2 (v33 − v22 ) + ∂2 v14 + v13 ∂3 v23 − v23 ∂3 v13 j = 1, k = 2
−∂1 v33 + ∂2 v23 + (v33 − v22 )∂2 v33 + ∂2 v34 − v23 ∂3 v33 − ∂3 v24 j = 3, k = 2
3
Njk = ∂2 v44 + ∂2 v23 − ∂4 v24 j = 4, k = 2
(v11 − v33 )∂1 v33 − ∂1 v34 + v12 ∂2 v33 − ∂2 v13 + v13 ∂3 v33 + ∂3 v14 j = 1, k = 3
∂1 v33 + v22 ∂2 v33 − ∂2 v23 − v33 ∂2 v33 − ∂2 v34 + v23 ∂3 v33 + ∂3 v24 j = 2, k = 3
∂3 (v44 + v33 ) − ∂4 v34
j = 4, k = 3
−∂1 v44 − ∂3 v13 + ∂4 v14 j = 1, k = 4
−∂2 v44 − ∂3 v23 + ∂4 v24 j = 2, k = 4
−∂3 (v44 + v33 ) + ∂4 v34
j = 3, k = 4
0 otherwise
Hydrodynamic Chains 47
1
Now Hjk can be calculated,
v11 ∂1 v23 − v13 (∂2 v22 − 2∂3 v23 + ∂2 v33 ) + v23 (2∂2 v12 −
∂1 v22 − 2∂3 v13 + ∂1 v33 ) − v22 ∂2 v13 + v33 ∂2 v13 +
v12 ∂1 v23 − ∂1 v13 − v33 ∂1 v23 + ∂2 v14 − ∂1 v24
j = 1, k = 2
v11 ∂1 (v33 − v22 ) + v22 (∂1 (v22 − v33 ) − ∂2 v12 + ∂3 v13 )+
v12 (2∂2 v33 − ∂3 v23 ) + v33 (∂2 v12 − ∂3 v13 ) + ∂1 (v12 + v23 − v34 )
−2∂2 v13 + ∂3 v14 + v13 ∂3 v33 j = 1, k = 3
−∂1 v44 − 2∂3 v13 + ∂4 v14 + ∂2 v12 + ∂1 v33
j = 1, k = 4
v23 (−2∂2 v12 + ∂1 v22 + 2∂3 v13 − ∂1 v33 ) + v13 (∂2 v22 +
1
Hjk = ∂2 v33 − 2∂3 v23 ) + v22 ∂2 v13 − v33 ∂2 v13 − v12 ∂2 v23 + ∂1 v13 − v11 ∂1 v23 +
v33 ∂1 v23 − ∂2 v14 + ∂1 v24 j = 2, k = 1
v11 (−∂2 v22 − ∂2 v33 + ∂3 v23 ) + v33 (∂2 v22 − ∂3 v23 )
+v23 ∂3 v33 ) + ∂1 (−v11 − v22 + 2v33 ) + ∂2 (2v12 − v23 − v34 )
+v22 ∂2 v33 + ∂3 (v24 − v13 ) j = 2, k = 3
∂2 (v22 + v33 − v44 ) − 2∂3 v23 + ∂4 v24 j = 2, k = 4
v11 ∂1 (v22 − v33 ) + v12 (∂3 v23 − 2∂2 v33 ) + v22 (∂1 (v33 − v22 ) + ∂2 v12 −
∂3 v13 ) + v33 (∂3 v13 − ∂2 v12 ) + ∂1 (v34 − v12 − v23 ) + 2∂2 v13
−∂3 v14 − v13 ∂3 v33 j = 3, k = 1
Hydrodynamic Chains 48
v11 (∂2 (v22 + v33 ) − ∂3 v23 ) + v33 (−∂2 v22 + ∂3 v23 ) − 2∂1 v33 +
∂1 (v11 + v22 ) + ∂3 (v13 − v24 ) − v22 ∂2 v33 + ∂2 (v23 − v34 − 2v12 )
−v23 ∂3 v33 j = 3, k = 2
∂3 (−v44 − v33 ) + ∂4 v34 j = 3, k = 4
1
Hjk = ∂1 v44 + 2∂3 v13 − ∂4 v14 − ∂2 v12 − ∂1 v33 j = 4, k = 1
∂2 (v44 − v22 − v33 ) + 2∂3 v23 − ∂4 v24 j = 4, k = 2
∂3 (v44 + v33 ) − ∂4 v34 j = 4, k = 3
0 otherwise.
1
It is a necessary condition that each component of Hjk is identically zero for the
matrix to be diagonalizable.
Example In this example we show how to calculate every component of the Haantjes
tensor for the Benney chain (63). As in the previous example the Nijenhuis tensor
is first calculated. Since the system under consideration is infinite, it means that in
theory the summation for Nijenhuis and Haantjes tensor may be infinite. However,
the matrix is of the class C so the summation is in fact finite, as is now shown. It
is clear that
m
vm+1 = 1, v1m = (m − 1)Am−2 ,
Hydrodynamic Chains 49
all other vnm = 0. Now we calculate the Nijenhuis tensor, first consider N1k
i
,
i
N1k = v1p ∂p vki − vkp ∂p v1i − vpi ∂1 vkp + vpi ∂k v1p .
Let us consider each term separately for a moment, the first term:
(i − 1)(i − 2)Ai−3
k=1
p i
v1 ∂p vk =
0
otherwise
0 0 0 0 0 0 . . .
0 1 0 0 0 0 .
0 0 1 0 0 0 .
1 i = k, i 6= 1
i
N1k = = . .
. 1
0
otherwise
. . .
. . .
. . .
Now we calculate the Nijenhuis tensor with j 6= 1, by the skew symmetry property,
i i
Njk = −Nkj , it is clear that,
1 j = 1, i = k, i 6= 1
i
Njk = 1 k = 1, i = j, i 6= 1
0
otherwise
i i p r p i r p i r p i r
H1k = Npr v1 vk − N1r vp vk − Nrk vp v1 + N1k vr vp .
i
Thus, H1k = 0, ∀i, k. Now work on
i i p r p i r p i r p i r
H2k = (Npr v2 vk ) − (N2r vp vk ) − (Nrk vp v2 ) + (N2k vr vp ).
i
Thus, H2k = 0, ∀i, k. Now work on
i i p r p i r p i r p i r
Hjk = (Npr vj vk ) − (Njr vp vk ) − (Nrk vp vj ) + (Njk vr vp ), j 6= 1, 2.
i i i
Thus, Hjk = 0, ∀ i, k, j ≥ 3. We have previously shown that H1k = 0, H2k = 0.
i
Thus, Hjk = 0, ∀ i, j, k.
an bn+1
= constant. (81)
n+1
Calculating the Haanjtes tensor of system (80) and setting each component equal
to zero implies that
This implies
an bn+1 = αn + β (83)
where α, β are constants. It is clear that (81) just (83) with α = β = 1, thus
requiring that the Haantjes tensor is zero has produced a slightly more general
result than obtained in [4]. Under the change of variables An → a0 · · · an−1 C n , the
Benney chain, (63), is transformed to (80) modulo (83).
u1t = u2x ,
etc. Following the approach of [24, 25] let us seek solutions in the form ui =
ui (R1 , . . . , Rm ) where the Riemann invariants R1 , . . . , Rm solve the diagonal system
(9),
Rti = λi (R)Rxi .
Hydrodynamic Chains 55
Substituting this ansatz into the Benney equations and equating to zero coefficients
at Rxi we arrive at the following relations:
∂i u2 = λi ∂i u, (84)
∂j λi ∂i λj
∂i ∂j u = ∂i u + ∂j u,
λj − λi λi − λj
∂j λi ∂i u + ∂i λj ∂j u = 0,
λi ∂j λi ∂i u + λj ∂i λj ∂j u + ∂i u∂j u = 0,
It is a truly remarkable fact that all other consistency conditions (e.g., of the relation
(87), etc), are satisfied identically modulo (88). Moreover, the semi-Hamiltonian
property (38) is also automatically satisfied. Thus, the system (88) governs m-
component reductions of the Benney chain. Up to reparametrizations Ri → f i (Ri )
these reductions depend on m arbitrary functions of a single variable. Solutions
arising within this approach are known as multiple waves, or nonlinear interactions
of planar simple waves.
Hydrodynamic Chains 56
The main result of this section is the proof of Theorem 2 formulated in the intro-
duction: The vanishing of the Haantjes tensor H is a necessary condition for the
integrability of hydrodynamic chains from the class C [20].
Proof This proof is computational. Writing down the equations of the chain in the
form um m n i i 1 m
t = Vn ux and substituting the ansatz u = u (R , ..., R ) we arrive at an
we point out that all summations here and below involve finitely many nonzero
terms. Applying the operator ∂j , j 6= i, we obtain
m
Vn,k ∂i un ∂j uk + Vnm ∂i ∂j un = ∂j λi ∂i um + λi ∂i ∂j um . (89)
Interchanging the indices i and j and subtracting the results we arrive at the ex-
pression for ∂i ∂j um in the form
m m
∂j λi ∂i λj Vn,k − Vk,n
∂i ∂j um = ∂i u m
+ ∂j u m
+ ∂i un ∂j uk .
λj − λi λi − λj λi − λj
m
Nnk ∂i un ∂j uk
∂j λi ∂i um + ∂i λj ∂j um =
λi − λj
where N is the Nijenhuis tensor of V . This can be rewritten in the invariant form
N (∂i u, ∂j u)
∂j λi ∂i u + ∂i λj ∂j u = (90)
λi − λj
V 2 N (∂i u, ∂j u)
(λi )2 ∂j λi ∂i u + (λj )2 ∂i λj ∂j u = , (91)
λi − λj
V N (V ∂i u, ∂j u)
(λi )2 ∂j λi ∂i u + λi λj ∂i λj ∂j u = , (92)
λi − λj
V N (∂i u, V ∂j u)
λi λj ∂j λi ∂i u + (λj )2 ∂i λj ∂j u = , (93)
λi − λj
N (V ∂i u, V ∂j u)
λi λj ∂j λi ∂i u + λi λj ∂i λj ∂j u = . (94)
λi − λj
The first relation can be obtained by applying the operator V 2 to (90) and using
V ∂i u = λi ∂i u. Similarly by applying V λi , V λj , λi λj respectively, we can obtain
(92),(93),(94). Thus combining all four relations (91), (92), (93), (94),
Relation (95) can be rewritten in the form H(∂i u, ∂j u) = 0 where H is the Haantjes
tensor, indeed, a coordinate-free form of the relation (3) is
we conclude that H(X, Y ) = 0 for any two formal eigenvectors of the matrix V .
Hydrodynamic Chains 58
By formal eigenvectors we mean vectors X defined by AX = λX. So, for the Benney
chain (63) we get, in matrix form, the relations,
0 1 0 0 0
1 1
1
x x
u 0 1 0 0
x2 x2
u2 = λ .
0 0 1 0
3
x3
3
x
u 0 0 0 1
. .
. . . . .
V X = λX,
0 1 0 0 0 0 . x1 x1
1
u 0 1 0 0 0 . x2 x2
u2
0 0 1
.
0 0 . x3 = λ x3
3
u 0 0 0 1 0 . x4 x4
. . . . . . . . .
x2 = λx1 ,
x3 + u1 x1 = λx2 ,
x4 + u2 x1 = λx3 ,
...
x1 = 1,
x2 = λ,
x 3 = λ2 − u 1 ,
x 4 = λ3 − u 1 λ − u 2 ,
...
Notice that we have proved a more general result, namely, that the existence of
sufficiently many two-component reductions already implies the vanishing of the
Haantjes tensor H. Indeed, nothing changes in the proof if we set i = 1, j = 2
in the formula (90). As demonstrated in the Theorem 5 below, one can further
strengthen the result by reversing the above proof under the additional assumption
of the simplicity of the spectrum of the matrix V . This will require the relations
N (V ∂i u, ∂j u) − V N (∂i u, ∂j u) N (V ∂j u, ∂i u) − V N (∂j u, ∂i u)
∂i λj ∂j u = i j 2
, ∂j λi ∂i u = ,
(λ − λ ) (λi − λj )2
which can be obtained by applying V to both sides of (90) and solving for ∂i λj ∂j u
and ∂j λi ∂i u.
etc. Substituting these expressions into the first m equations of the chain we ob-
tain an m-component system Sm for u1 , . . . , um , while the remaining equations will
be satisfied identically (by the definition of a reduction). Notice that the Haantjes
tensor of the reduced system Sm is identically zero since the reduction is diagonal-
Hydrodynamic Chains 61
izable. Let us now choose the number m sufficiently large so that the first C(i, j, k)
equations of the chain do not contain variables um+1 , um+2 , . . . (one can always do
so since any equation of the chain depends on finitely many u’s, and m can be ar-
bitrarily large). Then the first C(i, j, k) equations of the reduced system Sm will
be identical to the first C(i, j, k) equations of the original infinite chain. Hence, the
i
corresponding components Hjk for the reduced system and for the infinite chain will
also coincide. This proves that all components of the Haantjes tensor of the chain
must be zero.
A straightforward modification of the second proof allows one to show that the ex-
istence of an infinity of semi-Hamiltonian reductions implies the vanishing of the
tensor P . This establishes the necessity of the conjecture formulated in the Intro-
duction.
We emphasize that the condition of diagonalizability alone is not sufficient for the
integrability in general. This can be seen as follows.
etc, which is obtained from the Benney chain (63) ut = V (u)ux by the transfor-
mation V → V + p(u1 )E where E is an infinite identity matrix and p is a function
of u1 . One can verify that the corresponding Haantjes tensor is zero (which is not
at all surprising since the addition of a multiple of the identity does not effect the
Hydrodynamic Chains 62
The semi-Hamiltonian property is satisfied if and only if p00 = 0. Thus, we have con-
structed examples which possess infinitely many diagonal hydrodynamic reductions
none of which are semi-Hamiltonian (if p00 6= 0).
We now prove Theorem 3 from the Introduction: in the simple spectrum case the
vanishing of the Haantjes tensor H is necessary and sufficient for the existence
of two-component reductions parametrized by two arbitrary functions of a single
argument.
Proof The necessity part is contained in the first proof of the previous theorem. To
establish the sufficiency one has to show that the vanishing of the Haantjes tensor
implies the solvability of the equations
V ∂1 u = λ1 ∂1 u, V ∂2 u = λ2 ∂2 u, (96)
m m
m ∂2 λ1 m ∂1 λ2 m
Vn,k − Vk,n
∂1 ∂2 u = 2 1
∂1 u + 1 2
∂2 u + 1 2
∂1 un ∂2 uk (97)
λ −λ λ −λ λ −λ
and
N (V ∂1 u, ∂2 u)−V N (∂1 u, ∂2 u)
∂1 λ2 ∂2 u = (λ1 −λ2 )2
,
(98)
N (V ∂2 u, ∂1 u)−V N (∂2 u, ∂1 u)
∂2 λ1 ∂1 u = (λ1 −λ2 )2
,
which govern two-component reductions. Our first observation is that the vanishing
Hydrodynamic Chains 63
are the eigenvectors of V with the eigenvalues λ2 and λ1 , respectively. Let us show
that N (V ∂1 u, ∂2 u) − V N (∂1 u, ∂2 u) is indeed the eigenvector of eigenvalue λ2 .
V X = λX (100)
V N (V ∂1 u, ∂2 u) − V 2 N (∂1 u, ∂2 u) = λ2 N (V ∂1 u, ∂2 u) − λ2 N (∂1 u, ∂2 u)
V N (V ∂1 u, ∂2 u) − V 2 N (∂1 u, ∂2 u) = N (V ∂1 u, V ∂2 u) − V N (∂1 u, V ∂2 u)
(100) is satisfied modulo Haantjes tensor identically zero. The same process can
be carried out for the other eigenvector in question. By the assumption of the
simplicity of the spectrum, that is the eigenvectors are distinct, it is known that
(99) is proportional to ∂2 u and ∂1 u. Thus, equations (98) reduce to a pair of first
order PDEs for λ1 and λ2 ,
k1 k2
∂1 λ2 = , ∂2 λ1 = , (101)
(λ1 − λ2 )2 (λ1 − λ2 )2
1 1
1 Vn,k − Vk,n
∂1 ∂2 u1 = k ∂
1 2 u 1
− k ∂
2 1 u 1
+ ∂1 un ∂2 uk . (102)
(λ1 − λ2 )3 λ1 − λ2
Hydrodynamic Chains 64
Thus, relations (96) – (98) reduce to a pair of first order equations (101) plus one
second order PDE for u1 . Up to reparametrizations R1 → f 1 (R1 ), R2 → f 2 (R2 ),
their general solution depends on two arbitrary functions of a single variable.
Conservative chains 65
4 Conservative chains
u1t = f (u1 , u2 )x ,
u2t = g(u1 , u2 , u3 )x ,
u3t = h(u1 , u2 , u3 , u4 )x ,
... .
All known integrable hydrodynamic chains can be written in conservative form, for
example we showed that the Benney chain, (63), can be written in the conservative
form (77).
The Egorov case is when f (u1 , u2 ) = u2 in (5) thus becoming (4). There are ten
i
nonzero components of Hjk and the requirement that they vanish leads to all second
order partial derivatives of h(u1 , u2 , u3 , u4 ) in terms of first order partial derivatives
Conservative chains 66
of h and g:
The consistency conditions for the equations (103) lead to expressions for all third
order partial derivatives of the function g(u1 , u2 , u3 ) in terms of its lower order
derivatives:
2
2g33 2g13 g33 2g23 g33
g333 = , g133 = , g233 = ,
g3 g3 g3
2
2g13 2g13 g23 2g 2
g113 = , g123 = , g223 = 23 ,
g3 g3 g3
2 2
g222 = g2 g23 + g 23 (g3 g22 + 2g13 ) − g 33 (g2 g22 + 2g12 ) ,
g32
2 2
g122 = 2
g1 g23 + g13 (g3 g22 + g13 ) − g33 (g1 g22 + g11 ) , (104)
g3
2
g112 = (g33 (g2 g11 − 2g1 g12 ) − g13 (g2 g13 − 2g3 g12 ) − g23 (g3 g11 − 2g1 g13 )) ,
g32
2
g111 = (g1 + g22 )g13
2
+ g12 g23
2
+ g32 (g12
2
− g11 g22 ) − g22 g33 g12
g32
+g13 g3 (g11 + 2(g1 g22 − g2 g12 )) + 2g23 (g2 (g3 g11 − g1 g13 ) − g1 g3 g12 )
This systems general solution depends on ten integration constants, indeed, the
values of g and its partial derivatives up to the second order can be prescribed
arbitrarily at any fixed point u10 , u20 , u30 . The system (104) was first derived in [47]
from the requirement that the chain (4) is embedded into a hierarchy of commuting
hydrodynamic chains of Egorov’s type. Exactly the same equations for g were
obtained in [17] by applying the method of hydrodynamic reductions to the (2+1)-
dimensional PDE
utt = g(uxx , uxt , uxy ) (105)
which is naturally associated with the chain (4); here the function g is the same
as in (4), (104). Thus, for hydrodynamic chains of the type (4) the condition of
diagonalizability is necessary and sufficient for the integrability. Let us summarize
the work referred to in [17].
First, introduce the notation uxx = a, uxt = b, uxy = c. Thus (105) becomes
Now we write the PDE (106) in quasilinear form, that is in the form,
at = bx ,
ay = cx ,
by = c t , (107)
bt = G(a, b, c)x .
c = c(R1 , · · · , Rn ), where the Riemann invariants satisfy the equations (26). We ar-
rive at the following identities
∂i b = λi ∂i a, (108)
∂i c = µi ∂i a, (109)
(λi )2 = Ga + Gb λi + Gc µi . (110)
rearrange (111) to make λij the subject. Use (26) to eliminate µij , and (110) to
eliminate µi . Thus (111) becomes
aj
λij = i j
[Gaa Gc + (λi + λj )Gab Gc + λi λj Gbb Gc +
λ −λ
(Gac + λi Gbc )((λj )2 − λj Gb − Ga )
Gcc j 2
((λi )2 − λi Gb − Ga )(Gac + λj Gbc + ((λ ) − λj Gb − Ga ))]. (112)
Gc
The compatibility conditions of (108), (109), that is (bi )j = (bj )i and (ci )j = (cj )i
respectively, imply that
λij λji
∂i ∂j a = j ∂i a + i ∂j a. (113)
λ − λi λ − λj
X∂j a∂k a.
Conservative chains 69
Returning to the the Haantjes tensor, it can be shown that the vanishing of other
components of the Haantjes tensor does not impose any additional constraints on the
derivatives of g and h. Thus, writing the fourth equation of the chain (4) in the form
u4t = s(u1 , u2 , u3 , u4 , u5 )x and setting Hjk
2
= 0, one obtains the expressions for all
second order partial derivatives of s in terms of h and g, which are analogous to (103).
The consistency conditions are satisfied identically modulo (103), (104). Similarly,
3
the condition Hjk = 0 specifies the right hand side of the fifth equation of the chain,
etc. Although we know no direct way to show that this recursive procedure works in
general, there exists an alternative direct approach to the reconstruction of a chain
from the function g(u1 , u2 , u3 ). To illustrate this procedure we consider a simple
example g = u3 − 12 (u1 )2 which automatically satisfies (104). The corresponding h,
as specified by (103), is given by h = µ + αu1 + βu2 + γu3 + δu4 − u1 u2 , which can
be reduced to a canonical form h = u4 − u1 u2 by redefining u4 appropriately (this
transformation freedom allows one to absorb arbitrary integration constants arising
Conservative chains 70
at each step of the construction). Thus, the first three equations of the chain are
1
u1t = u2x , u2t = (u3 − (u1 )2 )x , u3t = (u4 − u1 u2 )x , ..., (115)
2
etc. Equations (115) are nothing but the first three equations in the conservative
representation of the Benney chain (63). The recostruction of the remaining equa-
tions of the chain consists of three steps.
(i) One introduces the corresponding PDE (105). To find the required PDE we
proceed as follows, introduce the commuting chain:
substitute (116) into (115) to eliminate u3x , thus obtaining two equations depending
only on u1 , u2 , that depend on independent variables x, t, y,
u1t = u2x ,
1 2
vtt = vxy − vxx , (118)
2
1
utt = uxy − u2xx . (119)
2
It was demonstrated in [17] that the general PDE (105) is integrable by the method
of hydrodynamic reductions if and only if the function g satisfies the relations (104).
Conservative chains 71
(ii) One constructs a dispersionless Lax pair for the PDE (105), the procedure is as
follows. Look for functions R and Q such that,
Differentiate both equations w.r.t x and then make the substitution ψx = p, thus
we have
p t = Rx , (122)
py = Qx (123)
the consistency conditions pty = pyt are satisfied identically modulo (105), or more
specifically in our example (119). The existence of such Lax pairs was established
in [47, 17] for any equation (105) provided g satisfies the compatibility conditions
(104). In our example it takes the form
1 1
pt = ( p2 + uxx )x , py = ( p3 + uxx p + uxt )x ;
2 3
u1 u2 u3
p=λ− − 2 − 3 − ....; (124)
λ λ λ
the substitution of this ansatz into the first equation of the Lax pair implies an
infinite hydrodynamic chain for the variables ui . The first three equations of this
chain identically coincide with (115). The substitution into the second equation of
the Lax pair produces a commuting chain (one has to set u1 = uxx , u2 = uxt ). Both
chains possess infinitely many hydrodynamic reductions since this is the case for
Conservative chains 72
the generating equation (105). Thus, the Haantjes tensor will automatically vanish.
Notice that (124) does not work in general, and a more complicated dependance of
p on λ is required.
Now let us return to the work in question, solving the system (104). To explicitly
calculate g(u1 , u2 , u3 ) we will follow [47]. The main observation is that the first six
equations in (104) imply that the function 1/g3 is linear, 1/g3 = α + βu1 + γu2 + δu3 .
If δ 6= 0 then, up to a linear change of variables, one can assume that 1/g3 = u3 .
Similarly, if δ = 0, γ 6= 0, one can set 1/g3 = u2 . If δ = γ = 0, β 6= 0 one has
1/g3 = u1 . The last possibility is 1/g3 = 1. Thus, we have four cases to consider:
u3 u3
g = u3 + p(u1 , u2 ), g= + p(u1 , u2 ), g= + p(u1 , u2 ), g = ln u3 + p(u1 , u2 );
u1 u2
here the function p(u1 , u2 ) can be recovered after the substitution into the remaining
four equations (104). In each of these cases the resulting equations for p(u1 , u2 )
integrate explicitly, see [47], leading to four different forms:
1 1
g = u3 + (Au2 + 2Bu1 )2 + Ce−Au , (125)
4A
u3 B2
1 A B 2 1
g = 1+ 1
− 1 2
(u 2 2
) + 1 2
u − 1 2
− Ce−A/u , (126)
u u 4(u ) (u ) A(u )
3
u 1
g = 2 + η(u1 )(u2 )2 , (127)
u 6 Z
3 1 2
1
g = ln u − ln σ u , u − η(u1 )du1 . (128)
4
2
η 000 + 2ηη 00 − 3η 0 = 0, (129)
Conservative chains 73
1
p = A(u2 )2 + (Bu1 + D)u2 + q(u1 ),
4
and the substitution into the first equation results in the linear ODE q 000 +Aq 00 = 2B 2 .
Up to a transformation of the form u3 → u3 + αu2 + βu1 + γ this leads to (125).
Case II. Substituting the ansatz g = u3 /u1 + p(u1 , u2 ) into (104) one arrives at the
equations
2 2
p111 = 2(p212 − p11 p22 ) + 1 2
(p1 + p22 ) − 1 (p11 + 2p1 p22 − 2p2 p12 ),
(u ) u
2 4
p112 = − 1 2
p2 − 1 p12 ,
(u ) u
2 2
p122 = 1 2 − 1 p22 , p222 = 0.
(u ) u
B2
1 A 2 2 D B
p= 1
− (u ) + + 1 2 u2 − + q(u1 ),
u 4(u1 )2 u 1 (u ) A(u1 )2
Conservative chains 74
and the substitution into the first equation results in the linear ODE (u1 )3 q 000 +
u1 (6u1 − A)q 00 + (6u1 − 2A)q 0 = 0 whose basis of solutions consists of 1, 1/u1
1
and e−A/u . Up to a transformation of the form u3 → u3 + αu2 + βu1 + γ this
implies (126). It was observed in [47] that the cases I and II are reciprocally re-
lated: under the change from x, t to the new independent variables X, T defined as
dX = u1 dx+u2 dt, T = t, and the introduction of the new dependent variables U 1 =
1 2 3
u1
, U 2 = − uu1 , U 3 = − uu1 , etc, the chains from the case I transform to the chains
from the case II, and vice versa. On the level of the corresponding equations this
1 2 (u2 )2 (u2 )2
means that the change of variables U 1 = u1
, U 2 = − uu1 , P = u1
− p, G = u1
−g
transforms the equations for p from the Case I to the equations for p from the Case
II. Equivalently, (125) goes to (126). The details of such a change of variables will
be demonstrated in a more general context later on.
Case III. Substituting the ansatz g = u3 /u2 + p(u1 , u2 ) into (104) one arrives at
the equations
2 2 2 4
p111 = p 1 + 2(p 12 − p 11 p 22 ) + (p1 p12 − p2 p11 ),
(u2 )2 u2
2 2
p112 = 2 p11 , p122 = 2 2 p1 ,
u (u )
2 2
p222 = 2 2 p2 − 2 p22 .
(u ) u
B + Cu1 1
p=A+ 2
+ η(u1 )(u2 )2 ,
u 6
and the substitution into the first relation results in the Chazy equation (129) for
η. Elimination of the constants A, B, C leads to (127).
Conservative chains 75
Case IV. Substituting the ansatz g = ln u3 + p(u1 , u2 ) into (104) one arrives at the
equations
p111 = 2(p212 − p11 p22 ) + 2(p1 + p22 )p11 − 4p1 p2 p12 + 2p21 p22 ,
The general solution of the fourth equation can be represented in the form
Z
1 2 1
η(u1 )du1 ;
p = − ln σ u , u −
4
here σ solves the heat equation 4σ1 = σ22 and η is a function of u1 . It is convenient
to introduce the new variable v(u1 , u2 ) by the formula v = −(ln σ)22 . Taking into
account the heat equation for σ one has
1 1 1
v1 = v22 − v 2 + (ln σ)2 v2 .
4 2 2
0
≡ d/du1 , the second equation is satisfied identically and the first takes the form
8
v22 = 4v 3 − 4v 2 η − 8vη 0 − η 00 . (131)
3
This shows that v is a shift of the Weierstrass elliptic function in the variable u2 .
Since v = −(ln σ)22 , the function σ is the corresponding sigma function. Notice that
Conservative chains 76
1
v1 = v 2 − vη − η 0 + (ln σ)2 v2 ,
2
8
v2 = 4v − 4v η − 8vη 0 − η 00 .
2 3 2
3
These chain are an extension of those considered in the previous section. The general
chain has been altered by only the first flux, f = f (u1 , u2 ), which is now a function
of u1 , u2 , rather than just u2 . The conditions Hjk
1
= 0 lead to expressions for all
second order partial derivatives of h in terms of g and f . The consistency conditions
of these equations result in a system expressing all third order partial derivatives
of g and f in terms of lower order derivatives. Appendix 2 shows the Mathematica
1 1
file used to calculate the components of Hjk . The condition Hjk = 0 implies the
Conservative chains 77
By calculating the consistency conditions for the above equations we obtain the ex-
pressions for all third order partial derivatives of g and f .
Equations for g:
2
2g33 2g13 g33 2g23 g33
g333 = , g133 = , g233 = ,
g3 g3 g3
2g 2 2g13 g23 2g 2
g113 = 13 , g123 = , g223 = 23 ,
g3 g3 g3
−2g23 (g1 g3 g12 + g2 (g1 g13 − g3 g11 ))f12 (g33 g12 + g3 (−2g1 g13 + g3 g11 ))
+2 , (134)
f2 g32
(g33 g12 + g3 (g3 g11 − 2g1 g13 ))f22 + f2 (f11 (g33 g1 − g3 g13 ) − 2g13 (g2 g13 − 2g3 g12 ))
g112 =
f2 g32
2
g33 (2g1 g12 − g2 g11 ) + g23 (g3 g11 − 2g1 g13 ) + f1 (g33 g11 − g13 )
−2 ,
g32
(g1 g2 g33 + (g3 g13 − g1 g33 )f1 + (g3 g12 − g23 g1 − g2 g13 )g3 )
g122 = f22
f2 g32
2(g 2 g1 − g33 g22 g1 + g3 g22 g13 + f2 g13
2
+ (g33 g1 − g3 g13 )f12 − f2 g33 g11 )
+ 23 2
,
g3
(((f1 − 2g2 )g33 + 2g3 g23 )f1 + (g22 + f2 g1 )g33 + (g3 g22 − 2g2 g23 − f2 g13 )g3 )f22
g222 =
f2 g32
2 2
−f1 (f12 g33 + g23 − g33 g22 ) + f12 (g33 g2 − g3 g23 ) + g3 g23 g22 + g2 (g23 − g22 g33 )
+2 2
g3
f2 (g33 (f11 − 4g12 ) + 4g23 g13 )
+ .
g32
Equations for f :
f11 (g33 (f12 + g22 + f2 g1 ) − 2g2 g3 g23 + g32 g22 − f2 g3 g13 ) − 2f1 f11 (g33 g2 − g3 g23 )
f111 = −
f2 g32
g32 f11 + 2(g1 g2 g33 + f1 (−g33 g1 + g3 g13 ) + g3 (−g23 g1 − g2 g13 + g3 g12 ))
+f12
f2 g32
f22 (g33 g12 − 2g1 g3 g13 + g32 g11 )
− , (135)
f2 g32
f22 f11 g32 − f2 (f1 f11 g33 + f11 (g3 g23 − g33 g2 ) + 2f12 (g33 g1 − g3 g13 ))
f112 = ,
f2 g32
2 2
f22 g3 − 2f22 f12 g33 + f2 f22 ((f1 − g2 )g33 + g3 g23 )
f222 = .
f2 g32
Conservative chains 79
We have verified that the system (134), (135) is in involution. We claim that the
functions f and g contain all the necessary information about conservative chains
(5). In particular, equations (132) allow one to reconstruct the function h and, hence,
the right hand side of the third equation of the chain. Similarly, the requirement
2
Hjk = 0 reconstructs the fourth equation (i.e., provides an involutive second order
system for the right hand side of the fourth equation), etc. Although we have verified
directly that our procedure is non-obstructed up to order 5, we could not establish
this property in general.
Notice that the first six equations in (134) are exactly the same as when (104) was
being investigated. Thus, there are four essentially different subcases to consider:
Case I: g = u3 + p(u1 , u2 ). Substituting this ansatz into (134), (135) one arrives at
two cases to consider,
Subcase I1 :
2 2
f = s(u1 )eu , p = q(u1 )eu .
The substitution of these expressions into the remaining equations for p111 and f111
leads to a system of coupled ODEs for s(u1 ) and q(u1 ):
2((q 0 )2 − qq 00 ) − q 0 s00 + 2q 00 s0
q 000 = ,
s
s00 s0 + 2s0 q 0 − qs00 − q 00 s
s000 = .
s
Conservative chains 80
Setting q = −s0 , the second equation will be satisfied identically while the first
implies a fourth order ODE
2 2
f = σ(u1 )eu , p = −σ 0 (u1 )eu .
Subcase I2 :
1
f = (au1 + b)u2 + s(u1 ), p = A(u2 )2 + Bu1 u2 + q(u1 ),
2
1 B2 1 2 B 1 1
s = αe−Au , q= (u ) − α e−Au + βe−2Au .
2A A
B B 2 (A − 2) 1 2 αB
s= (u1 )2 + α(u1 )3−A , q= 2
(u ) + (u1 )3−A + β(u1 )2(2−A) .
A−1 2(A − 1) 1−A
Case II: g = u3 /u1 +p(u1 , u2 ). Substituting this ansatz into (134), (135) one arrives
at the following set of relations.
Conservative chains 81
equations for p:
equations for f :
The cases I and II are reciprocally related (we thank Maxim Pavlov for point-
ing out this equivalence): under the change from x, t to the new independent
variables X, T defined as dX = u1 dx + f dt, T = t, and the introduction of
1 2 3
the new dependent variables U 1 = u1
, U 2 = − uu1 , U 3 = − uu1 , etc, the chains
from the case I transform to the chains from the case II, and vice versa. At
the level of the corresponding equations this means that the change of variables
2 u2
U1 = 1
u1
, U 2 = − uu1 , F = − uf1 , P = u1
f − p transforms the equations for p, f
from the Case I to the equations for P, F from the Case II. Lets now demonstrate
that the change of variables described previously does indeed link the cases I and
II. Introduce the new independent variables X, T, described by
dX = u1 dx + f dt,
T = t. (136)
Conservative chains 82
where f is from the first equation of (5). Also, introduce the new dependent vari-
ables,
1 u2 u3
U1 = , U 2
= − , U 3
= − , ···
u1 u1 u1
dT dX
uix = uiT + uiX
dx dx
= u1 uiX , (137)
dT dX
uit = uiT + uiX
dt dx
= uiT + f uiX . (138)
u1t = f (u1 , u2 )x ,
u2t = g(u1 , u2 , u3 )x ,
make the change of variables described by (136) by substituting (137), (138) and
proceed to show that u1t = f (u1 , u2 )x becomes UT1 = F (U 1 , U 2 )X ,
u1t = f (u1 , u2 )x ,
u2
1 −1 1 2
u1T + f u1X = f1 (− 1 2 ) + f2 1 2 u1 u1X + f2 u uX ,
(u ) (u ) u1
collect T derivatives on the left hand side, X derivatives on the right, cancel u1
when it is contained as a common factor in both the numerator and denominator.
Collect terms on the right hand side so it starts to resemble a total derivative,
u2 1
f1
u1T =− +f u1X + f2 2
u − uX ,
u1 u1 X
1
multiply the whole equation by a factor (u1 )2
so that the left hand side becomes a
total derivative,
2
u1T u1X u
using the fact that UT1 = (u1 )2
, UX1 = (u1 )2
and UX2 =− u1
we have,
X
f f2 2
UT1 =− +f UX1 − U ,
u1 u1 X
1
eliminate u1 using the fact that U 1 = u1
,
now, it is clear to see that the right hand side is a total derivative
UT1 = − f U 1
X
Conservative chains 84
UT1 = F (U 1 , U 2 )X ,
where F (U 1 , U 2 ) = − (f U 1 ) .
Similaraly, let us show the second equation in Case I, u2t = (u3 + p(u1 , u2 ))x , can be
3
expressed in the form UX2 = UU 1 + P (U 1 , U 2 , U 3 )X .
X
Starting with
u2t = u3 + p(u1 , u2 ) x ,
u2
1 p2 1 2
u2T + f u2X = −p1 1 2 + p2 1 2 u1 u1X − u uX + u1 u3X ,
(u ) (u ) u1
u2 u1T
collect T derivatives on the left hand side, X derivatives on the right. Subtract u1
1
from each side of the equation and multiply the equation by a factor u1
, this makes
the left hand side a total derivative,
now work on the right hand side, start to arrange terms so that it looks like a total
derivative, this is slight more complicated that the first equation due to the UT1 term,
U3
−UT2 = p1 Ux1 + p2 UX2 − − f U 1 u2X − u2 (f U 1 )X ,
U1 X
it can now be seen that the left hand side is indeed a total derivative,
U3 u2
−UX2 =− + p−f 1 ,
U1 X u X
U3
UX2 = + P (U 1 , U 2 )X
U1 X
2
where P (U 1 , U 2 ) = p − f uu1 . Thus it has been shown that cases I and II are
reciprocally related. So, this case need not be discussed any further here.
Case III: g = u3 /u2 + p(u1 , u2 ). Substituting this ansatz into (134), (135) one
arrives at the following set of relations for f and p.
Equations for p:
2p21 p1 (4p12 − f11 ) + 4p11 (f1 − p2 ) p12 (2p12 − f11 ) + 2p11 (f12 − p22 )
p111 = 2 2
+ + ,
f2 (u ) u2 f2 f2
f22 p11 p11 2f2 p1 + u2 f22 (p1 + u2 p12 )
p112 = + 2 2 , p122 = ,
f2 u f2 (u2 )2
f22 (p2 − f1 ) − f2 p22 p22 f22 f12 2
p222 =2 2
+ + 2 2 − 2 2 (f1 − p2 );
u f2 f2 u (u )
Equations for f :
The last three equations for f and the last three equations for p lead to the two
essentially different possibilities:
Subcase III1 :
f = s(u1 )(u2 )k , p = q(u1 )(u2 )k+1 ,
k=const. The substitution of these expressions into the remaining equations for p111
and f111 leads to the coupled system of ODEs for s(u1 ) and q(u1 ):
k+2
q 000 = (2(k + 2)(q 0 )2 − 2(k + 1)qq 00 − q 0 s00 + 2q 00 s0 ),
ks
k + 2 00 0
s000 = (s s − k(k + 1)s00 q + 2ks0 q 0 ) − (k − 1)q 00 .
ks
Notice that under the substitution s = k = 1 the equation for q reduces to the
Chazy equation (129) for q(u1 ) = 61 η(u1 ).
Subcase III2 :
a, b=const. The substitution of these expressions into the remaining equations for
p111 and f111 leads to the coupled system of ODEs for s(u1 ) and q(u1 ):
Case IV: g = ln u3 +p(u1 , u2 ). Substituting this ansatz into (134), (135) one obtains
Conservative chains 87
equations for p:
equations for f :
f11 2 f12
f111 = (f1 + p22 + f2 p1 − p22 − 2f1 p2 + f12 ) + 2 (p1 (f1 − p2 ) + p12 ) +
f2 f2
f22 2
(p − p11 ),
f2 1
f22 f11 f22 f12
f112 = + f11 (f1 − p2 ) + 2f12 p1 , f122 = + f2 f11 + f22 p1 ,
f2 f2
f2
f222 = 22 + 2f2 f12 + f22 (p2 − f1 ).
f2
2
θ fxx fyy − fxy + ψfxx + ηfxy + σfyy + Θ = 0.
2 2
(ii) f = s(u1 )eu , p = −s0 (u1 )eu , where s satisfies the equation ss00 − (s0 )2 = 0. Up
1
to a linear transformation of u1 one has s = eu .
Kupershmidt’s brackets 89
5 Kupershmidt’s brackets
d d ∂h
ut = B + Bt (139)
dx dx ∂u
1 2 3
βu βu βu ....
(α + β)u2 (α + β)u3 (α + β)u4 ....
B= .
(2α + β)u3 (2α + β)u4 (2α + β)u5 ....
.... .... .... ....
u1t = 2βu1 h11 + (α + 2β)u2 h12 + βh1 ) u1x + 2βu1 h12 + (α + 2β)u2 h22 +
(α + β)h2 ) u2x ,
u3t = (2α + 2β)u3 h11 + (3α + 2β)u4 h12 u1x + (2α + 2β)u3 h12 +
...
where α and β are arbitrary constants. We impose the restraint that the components
1
Hjk of the Haantjes tensor are equal to zero. There are only finitely many non-
Kupershmidt’s brackets 90
zero components, in fact there are just enough non-zero coefficients to provide the
expressions for all third order partial derivatives of h(u1 , u2 ) in terms of lower order
derivatives,
(2α + 3β)h222
h222 = ,
(α + β)h2
here 4 = (α + β)h2 + (α + 2β)u2 h22 + 2βu1 h12 . We have verified the consistency of
this system. There are two ‘trivial’ cases which are to be excluded from the further
analysis:
(i) α + β = 0. In this case the Haantjes tensor vanishes identically for any Hamilto-
nian density h(u1 , u2 ), indeed, the first two equations for u1 , u2 form an independent
subsystem, while the remaining equations become strictly lower-triangular.
(ii) 4 = (α + β)h2 + (α + 2β)u2 h22 + 2βu1 h12 = 0. In this case the first equation
of the chain decouples from the rest, taking the form u1t = λ(u1 )u1x where λ(u1 ) is a
function of u1 . The Haantjes tensor is also identically zero.
Kupershmidt’s brackets 91
β α+2β α+2β
• h(u1 , u2 ) = (u2 + f (u1 )) α+2β , where f (u1 ) = a(u1 + c) β + b(u1 − c) β .
α+β
• h(u1 , u2 ) = (u1 + c)− β u2 + a(u1 − c)2+α/β (u1 + c)−1−α/β .
• h(u1 , u2 ) = u2 + c(u1 )2 .
In the case β = 0 we have two subcases h22 = 0 and h22 6= 0. First if h22 6= 0 then
we have
1 1
h(u1 , u2 ) = ecu u2 + e2cu , (143)
or,
Finally, the case α + 2β = 0 again has two subcases, h22 = 0 and h22 6= 0. Beginning
with h22 6= 0, the densities obtained are,
2
h(u1 , u2 ) = f (u1 )eau , (145)
Kupershmidt’s brackets 92
1
h√ 1
i √b
2 b2 +1
here f (u1 ) = (b2 + 1 − c2 (u1 )2 ) 2 √b +1+cu 2
.
b +1−cu1
2
u1 + c
h(u1 , u2 ) = (u1 + c)u2 + b(u1 + c) ln , (146)
u1 − c
or
c
h(u1 , u2 ) = u1 u2 + . (147)
u1
In Sect. 5.2 it is proved that all remaining components of the Haantjes tensor of the
corresponding Hamiltonian chains vanish identically by virtue of (141). In Sect. 5.3
we demonstrate that the requirement of the existence of an additional conservation
law of the form p(u1 , u2 , u3 )t = q(u1 , u2 , u3 , u4 )x leads to the same relations (141).
As shown in Sect. 5.4, equations (141) imply the existence of a generating function
of conservation laws and, hence, the infinity of conservation laws. This establishes
the integrability of all examples constructed in this Section.
the addition of terms linear in u1 to the Hamiltonian density h effects neither the
integrability of the chain, nor the vanishing of the Haantjes tensor because the
Kupershmidt’s brackets 93
addition of the identity martrix to a diagonalizable matrix does not affect the diag-
onalizability. Thus, the classification below is carried out up to transformations of
the form
h → ah + bu1 + c (148)
In this section we demonstrate that the relations (141), which were obtained from
1
the requirement of the vanishing of the first few components Hjk of the Haantjes
tensor, are already sufficiently restrictive and imply the vanishing of all other com-
ponents. The proof utilizes an important property of Hamiltonian chains (139),
namely, the existence of finite-component reductions for any (not necessarily inte-
grable) Hamiltonian density h [52]. Following [52], let us parametrize ui in terms of
finitely many ‘moments’ v a , a = 1, ..., n, as follows:
n
1 X α
k
u = α (v a )2+ β (k−1) ,
2 + β (k − 1) 1
k = 1, 2, 3, .... (we consider the generic case when all expressions in the denominators
are non-vanishing, see [52] for a discussion of the exceptional cases). Thus,
n n
1X a 2 1 X α
1
u = (v ) , 2
u = α (v a )2+ β , (149)
2 1 2+ β 1
etc. One can verify that under this substitution the infinite chain (139),
a 1+ α
vta a
= β v h1 + (v ) β h2 , (150)
x
a = 1, ..., n, see [53]. Here h(u1 , u2 ) is an arbitrary Hamiltonian density, not necessar-
ily satisfying the integrability conditions (141), and u1 , u2 are given by (149). Notice
that the system (150) is Hamiltonian: vta = β(∂h/∂v a )x . We have the following
Proof The proof follows from the explicit formulas for the Nijenhuis tensor,
1 α α α α
N23 = βv v v ((v ) − (v ) ) 2β 2 u1 (h11 h212 − h211 h22 ) + β(2 + )(αu2 − 2β(v 1 ) β u1 )
1 2 3 2 β 3 β
β
α α α
(h11 h22 h12 − h312 ) + (α + β)(2(1 + )u3 − β(2 + )2 (v 1 ) β u2 )(h11 h222 − h212 h22 )
β β
α α α α
+(α + β)h2 ([(α + 2β)h11 h12 + (v 1 v 2 v 3 ) β h222 ) + ((v 1 v 2 ) β + (v 1 v 3 ) β + (v 2 v 3 ) β )h12 h22 +
α α α α α α α
((α + β)((v 1 ) β + (v 2 ) β + (v 3 ) β ) − (2α + β)((v 1 ) β )h11 h22 + (β((v 1 ) β + (v 2 ) β + (v 3 ) β )
α
i
+(2α + β)(v 1 ) β )h212 )] ,
Kupershmidt’s brackets 95
and
1 α α α α
N12 = β (v ) v ((v ) − (v ) ) (2 + )(αu2 − 2β(v 1 ) β u1 )(h11 h22 h12 − h312 )+
2 1 2 2 1 β 2 β
β
α α α
2β 2 u1 (h11 h212 − h211 h22 ) + ((2 + )(α − 2β)(v 1 ) β u2 − 2(α + β)(1 + )u3 )(h212 h22 −
β β
α α α α
h222 h11 ) − β(α + β)h2 v 2 (v 1 v 2 ) β (2(α + β)(1 + )u3 + (α + 2β)((v 2 ) β − (v 1 ) β )
β
α α α α α α
(v 1 )2+ β )h222 + (−α(v 1 ) β (−(2 + )u2 − 2(v 2 ) β u1 + 2((v 1 ) β − (v 2 ) β )(v 1 )2 )+
β
α 1 α α α α α α α
β((2 + )((v ) β + (v ) β )u + 2(v 1 v 2 ) β u1 + 2(1 + )u3 + (v 1 )2+ β (2(v 2 ) β − 3(v 1 ) β )+
2 2
β β
2 α α α α α
(v ) β (v ) ))h12 + 2βu h11 + (2((α + β)(v ) β + β(v 2 ) β )u1 + β(2 + )(((v 2 ) β −
1 2 2 1 2 1
β
α α α α α α
(v 1 ) β )(v 1 )2 + u2 ))h11 h12 + (2(1 + )((α + β)(v 1 ) β + β(v 2 ) β )u3 + (2 + )(v 1 ) β
β β
α 2α 2α α
i
(2(v 2 ) β (α + β)u3 + β(2(v 2 ) β − (v 1 ) β − (v 1 v 2 ) β )(v 1 )2 ))h12 h22 − (α + β)2 h2 v 2
h α α α α α α
((α − β)(v 1 ) β + β(v 2 ) β )h11 + ((α + β)(v 2 ) β − β(v 1 ) β )(v 1 v 2 ) β h22 + (2α(v 1 v 2 ) β +
i
2 2α 1 2α
β((v ) − (v ) ))h12 ,
β β
which were obtained using the integrability conditions (141). Notice that, since (150)
1 1
is invariant under permutations of v’s, it is sufficient to specify N23 and N12 only.
Note that that these expressions do not explicitly depend on the size n of the reduc-
tion (150): this dependence is contained in the variables u1 , u2 , u3 . The vanishing of
the Haantjes tensor (3) is now a straightforward algebraic calculation. Thus, for the
Hamiltonian densities h(u1 , u2 ) which satisfy the integrability conditions (141), the
corresponding hydrodynamic chains possess diagonalizable n-component reductions
(150) for any value of n. According to the results of [20], this implies that the full
Haantjes tensor of the Hamiltonian chain vanishes identically.
Kupershmidt’s brackets 96
For any Hamiltonian density h(u1 , u2 ) the system (139) necessarily possesses two
conservation laws, namely,
and
which correspond to the conservation of the momentum and the Hamiltonian, re-
spectively. Let us require that there exists an ‘extra’ conservation law of the form
Theorem 8 The integrability conditions (141) are necessary and sufficient for the
existence of an additional conservation law of the form (151).
Proof The proof is computational: substituting in for u1t , u2t , u3t , u4t into the left
hand side of (151), we collect coefficients at u1x , ..., u4x and equate them to zero. This
results in a system of first order partial differential equations for the flux q, note
that q is linear in u4 ,
q1 = (3α + 2β)u4 p3 h12 + (α + 2β)u2 (p1 h12 + p2 h11 ) + 2(α + β)u3 (p3 h11 + p2 h12 ) +
q3 = βh1 p3 + (α + β)h2 p2 ,
q4 = (α + β)h2 p3 .
Calculating the consistency conditions for the flux, qij = qji , we obtain all second
order partial derivatives of the density p,
2(2α + β)((α + β)u3 h212 + βu1 h211 + (α + 2β)u2 h12 h11 )p3 + (α + β)2 p2 h2 h11
p11 = ,
(α + β)2 h22
1
p12 = (α + β)2 h2 h12 p2 + (2α + β)((α + β)(2u3 h12 + h2 h22 ) + 2βh11 h12 +
(α + β)2 h22
(α + 2β)(h212 + h11 h22 ))p3 ,
1
p22 = 2
(2α + β)((α + β)(2u3 p3 h222 + 2h12 h2 p3 ) + 2((α + 2β)u2 h22 +
(α + β) h2
βu1 h12 )p3 h12 ) + (α + β)2 h22 h2 p2 ,
(2α + β)h12 p3
p13 = ,
(α + β)h2
(2α + β)h22 p3
p23 = ,
(α + β)h2
p33 = 0.
Next, it will be demonstrated in Sect. 5.4 that equations (141) imply the existence
of a generating function for conservation laws. Thus, we can claim that the exis-
tence of one additional conservation law is already very restrictive and implies the
existence of an infinity of conservation laws, thus confirming the integrability.
Kupershmidt’s brackets 98
Theorem 9 The integrability conditions (141) are necessary and sufficient for the
existence of a generating function for conservation laws of the Hamiltonian systems
of the type (139).
Proof The structure of reductions (150) suggests that a generating function for
conservation laws should be sought in the form of
α
1+ α
λt − β h1 + (1 + α/β)p h2 λx = λp pt − β(ph1 + p
β β h2 )x ; (153)
∞
X β α
λ= q 1−2 α −k uk + s(u1 , q), q = pβ , (154)
k=2
while the function s(u1 , q), which specifies the dependence of λ on u1 , satisfies the
equations
β 1 −1−2 β
s1 = q −1−2 α F, sq = q α G. (155)
α
Here F and G are the following rational expressions in q depending on the Hamil-
tonian density h:
(α+β)2 qh2 +2β(α+2β)u1 u2 (h212 −h11 h22 )+(α+β)h2 [(α+2β)qu2 h22 −2βu1 h11 ]
G= (α+β)h2 [q 2 h22 +2qh12 +h11 ]+(2βqu1 −(α+2β)u2 )[h212 −h11 h22 ]
.
With the ansatz (154), all terms in (153) containing pt , px , u3x , u4x , ..., cancel identi-
cally, while the requirement of cancellation of the coefficients at u1x and u2x results
in (155). The functions F and G satisfy the relations
which are the consistency conditions of the equations (155). These relations are sat-
isfied identically modulo the integrability conditions (141). Conversely, the relations
(156) imply the integrability conditions (141). Note that, although the variable u2
is seemingly present in the expressions for F and G, the right hand sides of (155)
do not depend on u2 . This lack of dependance on u2 can be seen from the fact that
F2 = G2 = 0. For each of the cases arising in the classification in Sect. 5.1, the
equations (155) for s(u1 , q) can be solved explicitly. Thus, the generating function
λ can be reconstructed in closed form. This is mainly due to a simple dependence
Kupershmidt’s brackets 100
Here we consider densities (5.1)–(5.1). For the Hamiltonian density (5.1) the func-
tion s(u1 , q) is given by
βq(u1 + c)−α/β
i
1 2− α
+b((c + u ) β F − + q −2β/α u1 ;
2ac(α + 2β)
here F is the same hypergeometric function as above. Finally, the case of (5.1) leads
to
α+β
1
s(u , q) = 1
u + q q −2β/α . (157)
2c(α − 2β)
Let us now show the generating functions for the case β = 0, densities (142)–(144).
Here the relation (153) takes the form
λt − ep h2 λx = λp (pt − (h1 + ep h2 )x )
Kupershmidt’s brackets 101
where
∞
X
λ= q 1−k uk + s(u1 , q), q = ep .
k=2
which are consistent and define s(u1 , q) explicitly: for the Hamiltonian density (142)
one has
q + f0
1 1
s(u , q) = √ Arctanh √ ;
c cf
1 1 1
s(u1 , q) = u1 + log(2cecu + q).
2 2c
Let’s look at densities when α + 2β = 0, (145)–(147). Here the relation (153) takes
the form (set β = 1, α = −2):
where
∞
X
λ= q 2−k uk + s(u1 , q), q = p−2 .
k=2
1
2u h11 +qh2
sq = − h22 (h11 +qh12 )(2u1 h12 −h 1
2 )−(h12 +qh22 )(2u h11 +qh2 )
,
which are consistent and define s(u1 , q) explicitly: for the Hamiltonian density (145)
one has
1
2q(1+b2 −c2 (u1 )2 )2 −cb(b−cu1 )2 −cb+2c2 u1
bArctanh √cu
1+b2
bArctanh √
c 1+b2 (1+(b−cu1 )2 )
s(u1 , q) = − √ + √
2 1 + b2 4 1 + b2
1
+ log −4cbq(1+b2 )+4q 2 (1+b2 )2 −4c3 bq(u1 )2 +4c4 q 2 (u1 )4 −c2 +8c2 qu1 (1+b2 )(1−qu1 ) .
8
1 b
s(u1 , q) = q(u1 + c)2 + 1 − log(q(u1 − c)2 + 2cb) .
4c 2
1 2 1 4
s(u1 , q) = qu1 + q (u ) .
8c
1 2
s(u1 , q) = qu1 + q ;
8c
6 Manin-Kupershmidt’s bracket
d d ∂h
ut = B + Bt , (158)
dx dx ∂u
0 0 0 0 0···
u1 u2 u3 u4 u5 · · ·
2u2 2u3 2u4 2u5 2u6 · · ·
B ij = 3u3 3u4 3u5 3u6 3u · · · .
7
. . . . .
. . . . .
. . . . .
u1t = um h1m + nun h1(n+1) + h2 u1x + 2um+1 h1m + nun h2(n+1) + 2h3 u2x
u2t = um h2m + nun+1 h1(n+1) u1x + 2um+1 h2m + nun+1 h2(n+1) + h2 u2x
u3t = um h3m + nun+2 h1(n+1) u1x + 2um+1 h3m + nun+2 h2(n+1) u2x
u4t = nun+3 h1(n+1) u1x + nun+3 h2(n+1) u2x + nun+3 h3(n+1) u3x + h2 u4x + 2h3 u5x ,
...
the remaining expressions for h111 , h112 , h122 , h222 are not written out explicitly due to
their complexity (we have used Mathematica to manipulate with these expressions).
It was verified by using Mathematica that the system for h is in involution. It was
demonstrated in [26] that the relations (161) imply the vanishing of all remaining
components of the Haantjes tensor, that is,
1 i
Hjk = 0 =⇒ Hjk = 0 for any i ≥ 2.
Theorem 10 [20] The relations (161) imply the existence of a generating function
which gives rise to an infinity of conservation laws. To be more precise, we claim
that for any n there exist n linearly independent conserved densities which are func-
tions of the first n coordinates ui . All higher conservation laws are polynomial in
u4 , u5 , u6 , ....
u3
h= 1 2 2
+ p(u1 , u2 ).
(c + au + bu )
1 3 1 4
ũ1 = λu1 , ũ2 = u2 , ũ3 = u, ũ4 = u ,..., (162)
λ λ2
etc, s=const. Both transformations preserve the Poisson bracket specified by (158).
The transformations do not alter the chain. Hence, they can be used to simplify the
Hamiltonian. Suppose, for instance, that b 6= 0. Then, up to the second canonical
transformation, one can assume the ansatz h = u3 /(c + u2 )2 + p(u1 , u2 ). If b = 0
then h = u3 /(c + u1 )2 + p(u1 , u2 ). Thus, there are two subcases:
1+αc(c+u1 ) βu2
The last three equations lead to p(u1 , u2 ) = c(c+u1 )2
(u2 )2 + c+u1
+ q(u1 ) where α
and β are arbitrary constants. The substitution into the first equation gives a linear
ODE for q,
c2 β 2
(αc2 − 3u1 αc − 4)q 00 − (c + u1 )(1 + cu1 α)q 000 = .
2(c + u1 )3
m n
q(u1 ) = 1 2
+
(c + u ) c + u1
1 + αc(c + u1 ) 2 2 βu2 m n
h = u3 /(c + u1 )2 + 1 2
(u ) + 1
+ 1 2
+ ,
c(c + u ) c+u (c + u ) c + u1
(u1 )3 (c + u2 )p111 − 2c(u1 )2 p11 − 2cu1 (c − u2 )p12 − (5c − 3u2 )(c − u2 )2 p22
p222 = ,
(c − u2 )3 (c + u2 )
(u1 )2 (c + u2 )p111 − 2cu1 p11 − 2(2c2 − 3cu2 + (u2 )2 )p12
p122 = ,
(c − u2 )2 (c + u2 )
u1 (c + u2 )p111 + (u2 − 3c)p11
p112 = ,
(c − u2 )(c + u2 )
along with one more relation for p111 which we do not write out due to its complexity.
u1 u2 −αu2 − αc +(c−u2 )3 g(η)
The first three relations imply p(u1 , u2 ) = 3
(c+u2 )2
where g(η) is an
u1
arbitrary function of a single variable η = c−u2
. The result of the substitution of
this ansatz into the remaining equation for p111 factors into a product of two terms,
leading to the cases (i) and (ii) below:
(i) The function g satisfies a first order ODE
η α 1
the solution of this equation is g(η) = µη 3 + 4c
− 12c2
− 16c2 η
. This results in the
Hamiltonian densities of the form
u3 u1 u2 u2 + c/3 (u1 )3
h= + −α +µ +
(c + u2 )2 (c + u2 )2 (c + u2 )2 (c + u2 )2
1 (c − u2 )3
1
c − u2
u α
− − .
4c (c + u2 )2 c − u2 3c 4cu1
1
(4c2 η 2 g 0 −12c2 ηg−1−αη+2cη 2 )g 000 +(α−4cη+12c2 g+4c2 ηg 0 −2c2 η 2 g 00 )g 00 +(4c−8c2 g 0 )g 0 = .
2
Since the possibility when the coefficient at g 0000 equals zero was considered above in
the case (i), we conclude that g must be a cubic polynomial,
g = µ + νη + γη 2 + δη 3 ,
where the constants satisfy a single relation 12δ − 8cν + 16c2 (ν 2 − 3γµ) − 4γα + 1 = 0
which can be obtained by substituting back into the original ODE. This leads to
Hamiltonian densities of the form
αc
u3 u1 u2 − αu2 − 3
+ µ(c − u2 )3 + νu1 (c − u2 )2 + γ(u1 )2 (c − u2 ) + δ(u1 )3
h= + .
(c + u2 )2 (c + u2 )2
Notice that the particular case c = 0 results (up to obvious equivalence transforma-
tions and relabeling of constants) in Hamiltonian densities of the form
u3 (u1 )2 u1 1 (u1 )3
h= + α + β + γ + δ (164)
(u2 )2 u2 u2 u2 (u2 )2
Remark. The apparent similarity of the cases (160) and (164) is not accidental and
highlights a reciprocal invariance of the class of Hamiltonian chains (158). Recall
that the conservation of momentum reads
h 1 u1 2 1 3 u3 n un
H= , U = , U = , U = , . . . , U = , (165)
u2 u2 u2 (u2 )3 (u2 )n
one arrives at the system which takes the original form (158) in the variables
X, T, U n , H. Thus, the above reciprocal transformation is canonical. One can verify
directly that the change of variables (165) indeed identifies (160) and (164).
and the substitution into the last three equations (161) implies that q is linear. Up
to the equivalence h → αh + au1 + bu2 + c we thus have
the substitution of this ansatz into the remaining equations for h implies a compli-
cated system of third order PDEs for p(u1 , u2 ). A useful observation is that this
system is invariant under a 3-parameter group of point symmetries which is gener-
ated by the two canonical transformations (162), (163) and the reciprocal transfor-
mation (165). We want to find a change of variables which simplifies the system
Manin-Kupershmidt’s bracket 111
(161) so it is easier to work with. To find this change of variables we first find the
three infinitesimal generators of (162), (163) and (165), we then look for a change
of variables to simplify the infinitesimal generators, it is change of variables we use
on (161). Let us calculate the infinitesimal generators of these symmetries, starting
with the canonical transformation (162). From (162) we have
1 3
ũ1 = λu1 , ũ2 = u2 , ũ3 = u , ... ,
λ
making the substitution λ = er so that the change of variables is a group over addi-
tion not multiplication. For the form of (166) to be preserved in the new variables
ũi then we must have
thus we have p̃ = e−r p. We now have all the necessary information for the infinites-
imal generator
∂ ũ1 ∂ ∂ ũ2 ∂ ∂ p̃ ∂
X1 = |r=0 1 + |r=0 2 + |r=0
∂r ∂u ∂r ∂u ∂r ∂p
= u1 ∂u1 − p∂p , (167)
∂
where ∂K = ∂K
.
Now carry out the same procedure for the second canonical transformation. From
(163) we have
Applying this change of variables to (166) and insisting that the form is preserved
then p must transform as
p̃ = p − 2su2 − s2 u1 ,
∂ ũ1 ∂ ∂ ũ2 ∂ ∂ p̃ ∂
X2 = |r=0 1 + |r=0 2 + |r=0
∂r ∂u ∂r ∂u ∂r ∂p
= u1 ∂u2 − 2u2 ∂p . (168)
The method of how to calculate the infinitesimal generator of the reciprocal transfor-
mation (165) is different. We proceed as follows, substitute the change of variables
(165) into (167), (168), this should provide us with the third infinitesimal generator
up to a linear combination of (167) and (168). Making the change of variables
X1 = u1 ∂u1 − p∂p
X2 = u1 ∂u2 − 2u2 ∂p
The change of variables on X2 has provided the third generator. Thus, we have the
three required infinitesimal generators, which are,
[X1 , X2 ] = X1 X2 − X2 X1
= [u1 ∂u1 − p∂p ][u1 ∂u2 − 2u2 ∂p ] − [u1 ∂u2 − 2u2 ∂p ][u1 ∂u1 − p∂p ]
= u1 ∂u2 − 2u2 ∂p = X2 ,
the same can be done for [X1 , X3 ] and [X1 , X3 ] to obtain the relations
∂ ∂ ∂ξ ∂ ∂η ∂ ∂s
1
= 1
+ 1
+
∂u ∂ξ ∂u ∂η ∂u ∂s ∂u1
u2 ∂ 4(u2 )2 ∂
1 ∂ 3p
= − − + ,
(u1 )2 ∂ξ (u1 )2 ∂η (u1 )4 (u1 )5 ∂s
∂ ∂ ∂ξ ∂ ∂η ∂ ∂s
2
= 2
+ 2
+ (172)
∂u ∂ξ ∂u ∂η ∂u ∂s ∂u2
1 ∂ 2u2 ∂
= + ,
(u1 )−2 ∂ξ (u1 )4 ∂s
∂ ∂ ∂ξ ∂ ∂η ∂ ∂s
= + +
∂p ∂ξ ∂p ∂η ∂p ∂s ∂p
1 ∂
= .
(u1 )3 ∂s
Manin-Kupershmidt’s bracket 114
X1 = u1 ∂u1 − p∂p
1 1 −2 ∂ 2 1 −2 ∂ 1 −4 2 2 1 −5 ∂ 1 ∂
= u (u ) − u (u ) − 3p(u ) + 4(u ) (u ) −p 1 3
∂ξ ∂η ∂s (u ) ∂s
u2 ∂ 4(u2 )2 ∂
1 ∂ 4p
= 1 − 1 − +
u ∂ξ u ∂η (u1 )3 (u1 )4 ∂s
= −ξ∂ξ − η∂η − 4s∂s , (173)
the same change of variables on the other two equations in (171) results in
2 2
Setting s = s(ξ, η) we have p = s(ξ, η)(u1 )3 − (uu1) . In terms of s(ξ, η) the equations
for p assume a remarkable symmetric form
sηηηη = q, sξηηη = qr, sξξηη − 4 = qr2 , sξξξη = qr3 , sξξξξ = qr4 , (180)
s = ξ 2 η 2 + aξ 3 + bξ 2 η + cξη 2 + dη 3 + αξ 2 + βξη + γη 2 + µξ + νη + .
The substitution into the remaining equations for s implies the following relations
among the coefficients:
u2 (u2 )2
Notice that the corresponding p(u1 , u2 ) = s(− u11 , u1
)(u1 )3 − u1
will be a cubic
polynomial in u1 , u2 . A particular example from this class with the Hamiltonian
density h = (u3 + τ )1/3 was discussed in [34]. It corresponds to the case where
a = −τ and all other coefficients of s(ξ, η) are zero.
Subcase II2 : q 6= 0. Then the consistency conditions of (180) imply the relations
qξ = (qr)η , rξ = rrη . Taking into account that r = −sξ /sη we have sξξ s2η −sηη s2ξ = 0.
Manin-Kupershmidt’s bracket 116
α =const.
s
Subcase II4 : sξξ s2η − sηη s2ξ = 0. Setting r = − sηξ one obtains
rξ = rrη , sξ = −rsη .
Calculating partial derivatives of s with the help of the above relations and sub-
1
stituting them into the conditions Hjk = 0 we obtain rηη = 0; without any loss of
generality one can set r = − ηξ . This implies ξsξ − ηsη = 0, therefore, s = s(z), z =
ξη = −u2 /(u1 )2 . For s(z) we obtain the ODE
which linearizes after being differentiated by z once. Its general solution is given by
the formula
s(z) = z 2 + α + β(−z) + γ(−z)1/2 + δ(−z)3/2
1/3
h = u3 + α(u1 )3 + βu1 u2 + γ(u1 )2 (u2 )1/2 + δ(u2 )3/2 .
Manin-Kupershmidt’s bracket 117
The result of this Section is to show that Hamiltonian chains of the type (158)
possess a conservation law of higher order than the Hamiltonian density, h. All
Hamiltonian chain (158) possess three conservation laws of the form
and
ht = (u1 h1 h2 + 2u2 h1 h3 + u2 h22 + 3u3 h2 h3 + 2u4 h23 )x ,
which correspond to the Casimir (or Annihilator), Momentum and the Hamiltonian
respectively. Let us require the existence of an extra ‘higher’ conservation law of
the form
P (u1 , u2 , u3 , u4 )t = Q(u1 , u2 , u3 , u4 , u5 )x (181)
Theorem 11 [20] The relations (161) are necessary and sufficient for the existence
of an extra conservation law of the form (181).
Proof Let us assume that there exists a conservation law of the form
P (u1 , u2 , u3 , u4 )t = Q(u1 , u2 , u3 , u4 , u5 )x .
Manin-Kupershmidt’s bracket 118
Using (158) and collecting coefficients at uix we obtain the expressions for Qi =
∂Q/∂ui ,
2
X 3
X
5 4 1
Q1 = 5u P4 h13 + (4u P5−i h1,(i+1) + u P3−i h1,i ) + (3u3 P5−j h1,j + 2u2 P4−j h1,j ) +
i=1 j=1
P1 h2 ,
2
X 3
X
3 4 1
Q2 = 5u P4 h23 + (4u P5−i h2,(i+1) + u P3−i h2,i ) + (3u3 P5−j h2,j + 2u2 P4−j h2,j ) +
i=1 j=1
P2 h2 + 2P1 h3 ,
X 2 3
X
5 4 1
Q3 = 5u P4 h33 + (4u P5−i h3,(i+1) + u P3−i h3,i ) + (3u3 P5−j h3,j + 2u2 P4−j h3j ) +
i=1 j=1
P3 h2 + 2P2 h3 ,
Q4 = h2 P4 + 2h3 P3 ,
Q5 = 2h3 P4 .
The consistency conditions Qij = Qji imply the following expressions for second
order partial derivatives of P ,
6u4 P4 h213 + 9u3 P4 h12 h13 + 3u2 P4 (h212 + 2h13 h11 ) + 3u1 P4 h11 h12 + 2P3 h3 h11
P11 = ,
2h23
12u4 P4 h23 h13 + 9u3 P4 (h22 h13 + h23 h12 ) + 6u2 P4 (h12 (h22 + h13 ) + h23 h11 )
P12 = +
4h23
3u1 P4 (h212 + h22 h11 ) + 2h3 (3p4 h11 + 2p3 h12 )
,
4h23
12u4 P4 h13 h33 + 9u3 P4 (h23 h13 + h33 h12 ) + 6u2 P4 (h213 + h23 h12 + h33 h11 )
P13 = +
4h23
3u1 P4 (h13 h12 + h23 h11 ) + 2h3 (3P4 h12 + 2P3 h13 )
,
4h23
3P4 h13
P14 = ,
2h3
Manin-Kupershmidt’s bracket 119
6u4 P4 h223 + 9u3 P4 h22 h23 + 3u2 P4 (h222 + 2h23 h12 ) + 3u1 P4 h12 h22
P22 = +
2h23
2h3 (3P4 h12 + P3 h22 )
2h23
12u4 P4 h23 h33 + 9u3 P4 (h223 + h22 h33 ) + 6u2 P4 (h23 (h22 + h13 ) + h33 h12 )
P23 = +
4h23
3u1 P4 (h13 h22 + h23 h12 ) + 2h3 (3P4 h13 + 2P3 h23 + 3P4 h22 )
,
4h23
3P4 h23
P24 = ,
2h3
6u4 P4 h233 + 3P4 (3u3 h33 h23 + u2 (h223 + 2h33 h13 ) + u1 h23 h13 ) + 2h3 (3P4 h23 + P3 h33 )
P33 = ,
2h23
3P4 h33
P34 = ,
2h3
P44 = 0.
By ensuring the consistency of this system we obtain the set of relations (161). We
believe that the same process applied to any higher order conservation law
will results in the same relations (161). The Mathematica program that was used
to collect coefficients and calculate the consistency conditions for this proof is con-
tained in Appendix 4.
In this section we demonstrate that any Hamiltonian chain (158) whose density
h = h(u1 , u2 , u3 ) satisfies the system (161), possesses a generating function which
provides an infinity of conservation laws. In Sect. 3.1 we showed how to obtain
Manin-Kupershmidt’s bracket 120
the generating function for conservation laws for the Benney chain, we now out-
line the process in general. Given an arbitrary Hamiltonian density h(u1 , u2 , u3 ), a
generating function is sought in the form
∞
1 2
X uk
λ = ψ(p, u , u ) + ; (182)
k=1
pk
notice that the dependence on higher moments u3 , u4 , u5 , etc, is exactly the same
as in (72). As demonstrated in [51], the generating function (182) has to satisfy the
Gibbons-type relation
which reduces to (73) for the Benney Hamiltonian, h = (u3 + (u1 )2 )/2. Substituting
(182) into (183) and using (158), one obtains the following relations among the first
order partial derivatives ψ1 , ψ2 , ψp :
3u3 2
(ψp − 4
)(p h13 + ph12 + h11 ) + ψ1 (2u2 h13 + u1 h12 − 2ph3 ) + ψ2 (3u3 h13 +
p
1 3u3
2u2 h12 + u1 h11 ) + 3 (3u3 h12 + (2u2 + )h11 ) = 0,
p p
3u3
(ψp − 4 )(p2 h23 + ph22 + h12 ) + ψ1 (2u2 h23 + u1 h22 + 2h3 ) + ψ2 (3u3 h23 +
p
1 3u3
2u2 h22 + u1 h12 − 2ph3 ) + 3 (3u3 h22 + (2u2 + )h12 ) = 0, (184)
p p
3u3
(ψp − 4 )(p2 h33 + ph23 + h13 ) + ψ1 (2u2 h33 + u1 h23 ) + ψ2 (3u3 h33 + 2u2 h23 +
p
1 3u3
u1 h13 + 2h3 ) + 3 (3u3 h23 + (2u2 + )h13 − 2ph3 ) = 0.
p p
Manin-Kupershmidt’s bracket 121
ψ1 =
1 2
− 4p h3 (ph3 h33 + h23 (2h3 + u1 h13 ) − h33 (u1 h12 + u2 h22 ))
4
p(2h3 (3u3 (h33 h22 − h223 ) + 2h3 (h22 + h13 ) + u1 (h13 (2h22 + h13 ) − 2h12 h23 − h11 h33 ))
+(u1 )2 (h12 (h12 + h33 − h13 h23 ) + h23 (h11 h23 − h12 h13 ) + h22 (h213 − h11 h33 )))
h3 (6u3 (h13 h23 − h12 h33 ) − 2h3 h12 + u1 (h11 h23 − h12 h13 )) + 2u2 u1 (h12 (h12 h33 −
h13 h23 ) + h23 (h11 h23 − h12 h13 ) + h22 (h212 − h11 h33 )) ,
ψ2 =
1
− 4ph23 h33 p2 + 4h3 (h23 (h3 + u2 h23 + u1 h13 ) − h33 (u2 h22 + u1 h12 ))p+
4
(u1 )2 (h12 (h12 h33 − h13 h23 ) + h23 (h11 h23 − h12 h13 ) + h22 (h213 − h11 h33 )) +
2h3 (h13 (2h3 + 2u2 h23 + u1 (h13 + h22 )) − h12 (2u2 h33 + u1 h23 ) − u1 h11 h33 ) ,
ψp =
1 2
− 4h3 h33 u1 p2 + 4h3 p(2h3 (u2 h33 + u1 h23 ) + u2 u1 (h223 − h22 h33 ) − (u1 )2 (h13 h23
4
−h12 h33 )) + 2h3 (2h3 (3u3 h33 + 4u2 h23 + u1 (2h13 + h22 ) + (u1 )3 (h12 (h12 h33 − h13 h23 ) +
h23 (h23 h11 − h12 h13 ) + h22 (h213 − h11 h33 ))) + 4(u2 )2 (h223 − h22 h33 ) + 4u1 u2 (h13 h23 −
h12 h33 ) − 3u1 u3 (h223 − h22 h33 ) + (u1 )2 (h13 (2h22 + h13 ) − 2h23 h12 − h11 h33 )) .
Manin-Kupershmidt’s bracket 122
Here
4=
4p4 h23 h33 + 4p3 h2 (2h3 h23 + u2 (h223 − h22 h33 ) + u1 (h13 h23 − h12 h33 )) + ((u1 )2 p2 +
2u2 u1 p)(2h3 (h13 (h13 + 2h22 ) − 2h12 h23 − h11 h33 ) + (h12 (h12 h33 − h13 h23 ) + h23 (h11 h23
−h12 h13 ) + h22 (h213 − h11 h33 ))) + 2h3 p2 (3u3 (h22 h33 − h223 ) + 2h3 (2h13 + h22 )) + 4h3 p
(3u3 (h13 h23 − h12 h33 ) − 2h3 h12 ) + 2h11 (h3 (2h3 + u1 h22 + 4u2 h23 ) + 2(u2 )2 (h223 −
h22 h33 )) − 3u3 ((h213 − h11 h33 )(2h3 + u1 h22 ) + u1 h23 (h12 h33 − h13 h23 ) + u1 h23 (h11 h23
Mathematica was used to verify the consistency of these equations modulo the sys-
tem (161). This proves Theorem 10: indeed, an infinite sequence of conservation
laws results from the equation
pt = (p2 h3 + ph2 + h1 )x
where one has to substitute the expansion for p in terms of λ obtained from (182).
The calculation of ψ(p, u1 , u2 ) can be summarized as follows:
— one first integrates the equation for ψp , which appears to be rational in p, with
respect to p. This defines ψ up to a function of u1 , u2 .
— one fixes this function by a substitution into the equations for ψ1 , ψ2 .
This procedure leads to the following explicit formulae for generating functions
for the Hamiltonian densities obtained in Sect. 6.1.
Linear case. This is the simplest generalization of the Benney Hamiltonian with
Manin-Kupershmidt’s bracket 123
the density
where the constants satisfy the condition β 2 − 4αγ + 12δ = 0. Even in this case the
function ψ is quite nontrivial,
" #
2 β + 2γp
ψ(p, u1 , u2 ) = 1 arctan 1 .
(4αγ − β 2 ) 2 (1 + γu1 )(4αγ − β 2 ) 2
Below we show some of the ψ functions obtained for some of the subcases from Sect
6.1.
u3 1 + αc(c + u1 ) 2 2 βu2 m n
h(u1 , u2 , u3 ) = + (u ) + + + ,
(c + u1 )2 c(c + u1 )2 c + u1 (c + u1 )2 c + u1
where the constants satisfy the single relation, 2(1 − c2 α)n − 6cαm + 12 c2 β 2 = 0. The
corresponding function ψ is given by
3
" #
c2 2(c2 α − 1)u2 + c(2(c2 α − 1)p − βc) − cβu1
ψ(p, u1 , u2 ) = − 1 tanh−1 1 .
(3m) 2 2(3cm) 2 (1 + cαu1 )
u3 + u1 u2 − α(u2 + 3c ) + µ(u1 )3 (c − u2 )3 u1 c − u2
1 2 3 α
h(u , u , u ) = + − − ,
(c + u2 )2 4c(c + u2 )2 c − u2 3c 4cu1
Manin-Kupershmidt’s bracket 124
(u2 )2 1
h(u1 , u2 , u3 ) = (u3 + α(u1 )3 − )3 ,
u1
Other examples obtained in Sect. 6.1 lead to more complicated expressions for ψ.
Conclusions and further work 125
In this chapter we outline possibilities to extend the work presented in this thesis.
First, we summarize the ideas of this thesis.
We have given a simple and easy-to-verify necessary condition for the integrability
of hydrodynamic chains based on the vanishing of the Haantjes tensor; it must be
emphasized that the chain must be of class C. The utilization of Mathematica was
crucial in our ability to calculate the Haantjes tensor. We remember that vanishing
of the Haantjes tensor is not sufficient for the integrability in general, although zero
Haantjes tensor is already sufficiently restrictive and implies the integrability in the
cases of conservative chains and Hamiltonian chains. For finite-component systems
of hydrodynamic type, the following is a known result: any diagonalizable semi-
Hamiltonian system possesses infinitely many conservation laws and commuting
flows of hydrodynamic type, and can be solved by the generalized hodograph method
[61]. The work of this thesis has been to extend this idea to infinite-component
systems of hydrodynamic type, known as hydrodynamic chains. We have classified
conservative chains of the form,
first with f (u1 , u2 ) = u1 finding cases (125), (126), (127), (128) thus agreeing with
[47]. Then with f (u1 , u2 ) again finding four subcases for consideration.
We showed that for the two Hamiltonian systems (139), (158) that the conditions
imposed by zero Haantjes tensor, (141) and (161) respectively, not only leads to one
additional conservation law of higher order than the Hamiltonian density, but in
fact leads to a generating function which provides an infinity of conservation laws.
During the time spent working towards this thesis I have had two joint publications,
Conclusions and further work 126
[18], [20].
Each example known to us support the evidence that these approaches are equiv-
alent, by equivalent we mean they lead to the same integrability conditions and
thus the same classification results. The approach using the Haantjes tensor in our
experience has been relatively trouble free when using Mathematica, leading to the
required integrability conditions in a straightforward way.
Questions that have arisen during this work and are possible areas of further inves-
tigation include:
doing it iteratively;
— Integrating the remaining examples in the case of the general conservative chain.
In Cases III and IV of Sect. 4.5 we were unable to complete the integration of the
systems of equations, thus we could not explicitly show the form of the chains.
Appendix 1 128
Hamiltonian property
For m-component systems (1) there exists a tensor object which is responsible for
the semi-Hamiltonian property. First of all one computes (1, 3)-tensors M and K,
s p q s p q p q p q
s
Mkij = Nkp vq Nij + Npq s
vk Nij − Npq Nik s
vj − Npq Nkj s
vi − Nkp Niqp vjq − Nkp
s p q
Nqj vi
and
s
Kkij = bsp ∂uk Nijp − bpk ∂up Nijs + Nijp ∂up bsk − Nkp
s
∂ui bpj + Nkp
s
∂uj bpi
p
+bsp ∂ui Njk − bpi ∂up Njk
s p
+ Njk ∂up bsi − Nips ∂uj bpk + Nips ∂uk bpj
p
+bsp ∂uj Nki − bpj ∂up Nki
s p
+ Nki ∂up bsj − Njp
s
∂uk bpi + Njp
s
∂ui bpk ;
here b = v 2 , that is, bij = vpi vjp . Using M and K one defines a (1, 3)-tensor Q as
+4vkp Mpij
s s
− 2Mkpj vip − 2Mkip
s
vjp .
s
Pkij = vps Qpkqj viq + vps Qpkiq vjq − vqs vpq Qpkij − Qskpq vip vjq . (185)
These objects can be calculated using computer algebra. Notice that they are well-
defined for hydrodynamic chains from the class C: all tensor operations will involve
Appendix 1 129
It was pointed out in [56] that any strictly hyperbolic conservative system
with the zero Haantjes tensor is necessarily semi-Hamiltonian. The shortest proof
of this statement known to us can be summarized as follows. Let us first rewrite
(186) in the Riemann invariants, Rti = λi (R)Rxi , where λi 6= λj due to the strict
hyperbolicity. The conserved densities u(R) satisfy an over-determined system of
second order linear PDEs [61],
∂j λi ∂i λj
∂i ∂j u = ∂i u + ∂j u, i 6= j. (187)
λj − λi λi − λj
It would be of interest to obtain a direct tensor proof of the above result by showing
that the vanishing of the Haantjes tensor H for a conservative system (186) implies
the vanishing of the tensor P . Such a proof would then generalize to hydrodynamic
chains, for which we have yet no rigorous definition of Riemann invariants.
Appendix 2 131
;
2 conservative chain for thesis.nb
d@ sD
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m
Format@Derivative@i_, j_, k_, l_, m_, n_D@V51D@a_, b_, c_, d_, e_, f_D D :=
d@ V51D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n
Format@Derivative@i_, j_, k_, l_, m_, n_D@V52D@a_, b_, c_, d_, e_, f_D D :=
d@ V52D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n
Format@Derivative@i_, j_, k_, l_, m_, n_D@V53D@a_, b_, c_, d_, e_, f_D D :=
d@ V53D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n
Format@Derivative@i_, j_, k_, l_, m_, n_D@V54D@a_, b_, c_, d_, e_, f_D D :=
d@ V54D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n
Format@Derivative@i_, j_, k_, l_, m_, n_D@V55D@a_, b_, c_, d_, e_, f_D D :=
d@ V55D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n
Format@Derivative@i_, j_, k_, l_, m_, n_D@V56D@a_, b_, c_, d_, e_, f_D D :=
d@ V56D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n
Format@Derivative@i_, j_, k_, l_, m_, n_D@V57D@a_, b_, c_, d_, e_, f_D D :=
d@ V57D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V61D@a_, b_, c_, d_, e_, f_, g_D D :=
d@ V61D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ p
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V62D@a_, b_, c_, d_, e_, f_, g_D D :=
d@ V62D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ p
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V63D@a_, b_, c_, d_, e_, f_, g_D D :=
d@ V63D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ p
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V64D@a_, b_, c_, d_, e_, f_, g_D D :=
d@ V64D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ p
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V65D@a_, b_, c_, d_, e_, f_, g_D D :=
d@ V65D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ p
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V76D@a_, b_, c_, d_, e_, f_, g_D D :=
d@ V66D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ p
conservative chain for thesis.nb 3
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V67D@a_, b_, c_, d_, e_, f_, g_D D :=
d@ V67D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ p
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V68D@a_, b_, c_, d_, e_, f_, g_D D :=
d@ V68D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ p
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V71D@a_, b_, c_, d_, e_, f_, g_, q_D D :=
d@ V71D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ pdu8 ^ q
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V72D@a_, b_, c_, d_, e_, f_, g_, q_D D :=
d@ V72D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ pdu8 ^ q
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V73D@a_, b_, c_, d_, e_, f_, g_, q_D D :=
d@ V73D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ pdu8 ^ q
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V74D@a_, b_, c_, d_, e_, f_, g_, q_D D :=
d@ V74D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ pdu8 ^ q
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V75D@a_, b_, c_, d_, e_, f_, g_, q_D D :=
d@ V75D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ pdu8 ^ q
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V76D@a_, b_, c_, d_, e_, f_, g_, q_D D :=
d@ V76D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ pdu8 ^ q
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V77D@a_, b_, c_, d_, e_, f_, g_, q_D D :=
d@ V77D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ pdu8 ^ q
Format@Derivative@i_, j_, k_, l_, m_, n_, p_D@V78D@a_, b_, c_, d_, e_, f_, g_, q_D D :=
d@ V78D
;
du1 ^ i du2 ^ j du3 ^ k du4 ^ l du5 ^ m du6 ^ n du7 ^ pdu8 ^ q
$HistoryLength = 3
3
4 conservative chain for thesis.nb
H* Matrix V *L
H**********************L
8V41@u1, u2, u3, u4, u5D, V42@u1, u2, u3, u4, u5D, V43@u1, u2, u3, u4, u5D,
V33@u1, u2, u3, u4D, V34@u1, u2, u3, u4D, 0, 0, 0, 0, 0<,
8V51@u1, u2, u3, u4, u5, u6D, V52@u1, u2, u3, u4, u5, u6D,
V44@u1, u2, u3, u4, u5D, V45@u1, u2, u3, u4, u5D, 0, 0, 0, 0<,
V53@u1, u2, u3, u4, u5, u6D, V54@u1, u2, u3, u4, u5, u6D,
8V61@u1, u2, u3, u4, u5, u6, u7D, V62@u1, u2, u3, u4, u5, u6, u7D, V63@u1, u2, u3,
V55@u1, u2, u3, u4, u5, u6D, V56@u1, u2, u3, u4, u5, u6D, 0, 0, 0<,
u4, u5, u6, u7D, V64@u1, u2, u3, u4, u5, u6, u7D, V65@u1, u2, u3, u4, u5, u6, u7D,
8V71@u1, u2, u3, u4, u5, u6, u7, u8D, V72@u1, u2, u3, u4, u5, u6, u7, u8D,
V66@u1, u2, u3, u4, u5, u6, u7D, V67@u1, u2, u3, u4, u5, u6, u7D, 0, 0<,
V73@u1, u2, u3, u4, u5, u6, u7, u8D, V74@u1, u2, u3, u4, u5, u6, u7, u8D,
V75@u1, u2, u3, u4, u5, u6, u7, u8D, V76@u1, u2, u3, u4, u5, u6, u7, u8D,
V77@u1, u2, u3, u4, u5, u6, u7, u8D, V78@u1, u2, u3, u4, u5, u6, u7, u8D, 0<<;
V41@u1, u2, u3, u4, u5D = D@s@u1, u2, u3, u4, u5D, u1D;
V42@u1, u2, u3, u4, u5D = D@s@u1, u2, u3, u4, u5D, u2D;
V43@u1, u2, u3, u4, u5D = D@s@u1, u2, u3, u4, u5D, u3D;
V44@u1, u2, u3, u4, u5D = D@s@u1, u2, u3, u4, u5D, u4D;
V45@u1, u2, u3, u4, u5D = D@s@u1, u2, u3, u4, u5D, u5D;
conservative chain for thesis.nb 5
MatrixForm@VD
d@fD d@fD
du1 du2
0 0 0 0 0 0 0
d@gD d@gD d@gD
du1 du2 du3
0 0 0 0 0 0
d@hD d@hD d@hD d@hD
du1 du2 du3 du4
0 0 0 0 0
d@sD d@sD d@sD d@sD d@sD
du1 du2 du3 du4 du5
0 0 0 0
V51 V52 V53 V54 V55 V56 0 0 0
V61 V62 V63 V64 V65 V66 V67 0 0
V71 V72 V73 V74 V75 V76 V77 V78 0
H*Setting up u@iD so it can be used in the summation to calulate the two tensors*L
u@1D = u1;
u@2D = u2;
u@3D = u3;
u@4D = u4;
u@5D = u5;
u@6D = u6;
u@7D = u7;
t=1
t=1 w=1
H**L
Appendix 3 137
Manin-Kupershmidt bracket
H*Mathematica Programme for calculating the Haantjes tensor for a hamiltonian chain*L
Format@V81@u1, u2, u3, u4, u5, u6, u7, u8, u9DD := V81;
Format@V82@u1, u2, u3, u4, u5, u6, u7, u8, u9DD := V82;
Format@V83@u1, u2, u3, u4, u5, u6, u7, u8, u9DD := V83;
Format@V84@u1, u2, u3, u4, u5, u6, u7, u8, u9DD := V84;
Format@V85@u1, u2, u3, u4, u5, u6, u7, u8, u9DD := V85;
Format@V86@u1, u2, u3, u4, u5, u6, u7, u8, u9DD := V86;
Format@V87@u1, u2, u3, u4, u5, u6, u7, u8, u9DD := V87;
Format@V88@u1, u2, u3, u4, u5, u6, u7, u8, u9DD := V88;
Format@V89@u1, u2, u3, u4, u5, u6, u7, u8, u9DD := V89;
Format@V91@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V91;
2 Hamiltonian Systems Classification. for thesis.nb
Format@V91@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V91;
Format@V92@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V92;
Format@V93@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V93;
Format@V94@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V94;
Format@V95@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V95;
Format@V96@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V96;
Format@V97@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V97;
Format@V98@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V98;
Format@V99@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V99;
Format@V910@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10DD := V910;
Format@V101@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V101;
Format@V102@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V102;
Format@V103@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V103;
Format@V104@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V104;
Format@V105@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V105;
Format@V106@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V106;
Format@V107@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V107;
Format@V108@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V108;
Format@V109@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V109;
Format@V1010@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V1010;
Format@V1011@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11DD := V1011;
$HistoryLength = 3
3
Hamiltonian Systems Classification. for thesis.nb 3
H* Matrix VL *L
H**********************L
V = 88V11@u1, u2, u3D , V12@u1, u2, u3D, V13@u1, u2, u3D, 0, 0, 0, 0, 0, 0, 0, 0, 0<,
8V21@u1, u2, u3D, V22@u1, u2, u3D, V23@u1, u2, u3D, 0, 0, 0, 0, 0, 0, 0, 0, 0<,
8V41@u1, u2, u3, u4, u5D, V42@u1, u2, u3, u4, u5D, V43@u1, u2, u3, u4, u5D,
V44@u1, u2, u3, u4, u5D, V45@u1, u2, u3, u4, u5D, 0, 0, 0, 0, 0, 0, 0<,
8V51@u1, u2, u3, u4, u5, u6D, V52@u1, u2, u3, u4, u5, u6D,
V53@u1, u2, u3, u4, u5, u6D, V54@u1, u2, u3, u4, u5, u6D,
V55@u1, u2, u3, u4, u5, u6D, V56@u1, u2, u3, u4, u5, u6D, 0, 0, 0, 0, 0, 0<,
8V61@u1, u2, u3, u4, u5, u6, u7D, V62@u1, u2, u3, u4, u5, u6, u7D,
V63@u1, u2, u3, u4, u5, u6, u7D, V64@u1, u2, u3, u4, u5, u6, u7D,
V65@u1, u2, u3, u4, u5, u6, u7D, V66@u1, u2, u3, u4, u5, u6, u7D,
V67@u1, u2, u3, u4, u5, u6, u7D, 0, 0, 0, 0, 0<,
8V81@u1, u2, u3, u4, u5, u6, u7, u8, u9D, V82@u1, u2, u3, u4, u5, u6, u7, u8, u9D,
V83@u1, u2, u3, u4, u5, u6, u7, u8, u9D, V84@u1, u2, u3, u4, u5, u6, u7, u8, u9D,
V85@u1, u2, u3, u4, u5, u6, u7, u8, u9D, V86@u1, u2, u3, u4, u5, u6, u7, u8, u9D,
V87@u1, u2, u3, u4, u5, u6, u7, u8, u9D, V88@u1, u2, u3, u4, u5, u6, u7, u8, u9D,
V89@u1, u2, u3, u4, u5, u6, u7, u8, u9D, 0, 0, 0<,
8V91@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D,
V92@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D, V93@u1, u2, u3, u4, u5,
u6, u7, u8, u9, u10D, V94@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D,
V95@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D, V96@u1, u2, u3, u4, u5,
u6, u7, u8, u9, u10D, V97@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D,
V98@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D, V99@u1, u2, u3, u4, u5, u6,
u7, u8, u9, u10D, V910@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D, 0, 0<,
8V101@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11D,
V102@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11D,
V103@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11D, V104@u1, u2, u3, u4, u5,
u6, u7, u8, u9, u10, u11D, V105@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11D,
V106@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11D, V107@u1, u2, u3, u4, u5,
u6, u7, u8, u9, u10, u11D, V108@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11D,
V109@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11D, V1010@u1, u2, u3, u4, u5,
u6, u7, u8, u9, u10, u11D, V1011@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11D, 0<
<;
u@1D = u1;
u@2D = u2;
u@3D = u3;
u@4D = u4;
u@5D = u5;
u@6D = u6;
u@7D = u7;
u@8D = u8;
u@9D = u9;
u@10D = u10;
u@11D = u11;
4 Hamiltonian Systems Classification. for thesis.nb
p = P@u1, u2D
h = Hu3 + pL ^ H1 ê 3L;
V11@u1, u2, u3D = D@h, u2D + u1 * D@h, u1, u2D + 2 * u2 * D@h, u1, u3D;
V12@u1, u2, u3D = u1 * D@h, u2, u2D + 2 * u2 * D@h, u2, u3D + 2 * D@h, u3D;
V13@u1, u2, u3D = u1 * D@h, u2, u3D + 2 * u2 * D@h, u3, u3D;
V31@u1, u2, u3, u4D = 2 * Hu2 * D@h, u1, u1D + u3 * D@h, u2, u1D + u4 * D@h, u3, u1DL +
V32@u1, u2, u3, u4D = 2 * Hu2 * D@h, u1, u2D + u3 * D@h, u2, u2D + u4 * D@h, u3, u2DL +
2 * u4 * D@h, u1, u3D + u3 * D@h, u1, u2D;
V33@u1, u2, u3, u4D = 2 * Hu2 * D@h, u1, u3D + u3 * D@h, u2, u3D + u4 * D@h, u3, u3DL +
2 * u4 * D@h, u2, u3D + u3 * D@h, u2, u2D;
V41@u1, u2, u3, u4, u5D = 3 * Hu3 * D@h, u1, u1D + u4 * D@h, u2, u1D + u5 * D@h, u3, u1DL +
V42@u1, u2, u3, u4, u5D = 3 * Hu3 * D@h, u1, u2D + u4 * D@h, u2, u2D + u5 * D@h, u3, u2DL +
2 * u5 * D@h, u1, u3D + u4 * D@h, u1, u2D;
V43@u1, u2, u3, u4, u5D = 3 * Hu3 * D@h, u1, u3D + u4 * D@h, u2, u3D + u5 * D@h, u3, u3DL +
2 * u5 * D@h, u2, u3D + u4 * D@h, u2, u2D;
V51@u1, u2, u3, u4, u5, u6D = 4 * Hu4 * D@h, u1, u1D + u5 * D@h, u2, u1D + u6 * D@h, u3, u1DL +
V52@u1, u2, u3, u4, u5, u6D = 4 * Hu4 * D@h, u1, u2D + u5 * D@h, u2, u2D + u6 * D@h, u3, u2DL +
2 * u6 * D@h, u1, u3D + u5 * D@h, u1, u2D;
V53@u1, u2, u3, u4, u5, u6D = 4 * Hu4 * D@h, u1, u3D + u5 * D@h, u2, u3D + u6 * D@h, u3, u3DL +
2 * u6 * D@h, u2, u3D + u5 * D@h, u2, u2D;
V61@u1, u2, u3, u4, u5, u6, u7D = 5 * Hu5 * D@h, u1, u1D + u6 * D@h, u2, u1D + u7 * D@h, u3, u1DL +
V62@u1, u2, u3, u4, u5, u6, u7D = 5 * Hu5 * D@h, u1, u2D + u6 * D@h, u2, u2D + u7 * D@h, u3, u2DL +
2 * u7 * D@h, u1, u3D + u6 * D@h, u1, u2D;
V63@u1, u2, u3, u4, u5, u6, u7D = 5 * Hu5 * D@h, u1, u3D + u6 * D@h, u2, u3D + u7 * D@h, u3, u3DL +
2 * u7 * D@h, u2, u3D + u6 * D@h, u2, u2D;
V71@u1, u2, u3, u4, u5, u6, u7, u8D = 6 * Hu6 * D@h, u1, u1D + u7 * D@h, u2, u1D + u8 * D@h, u3, u1DL +
V72@u1, u2, u3, u4, u5, u6, u7, u8D = 6 * Hu6 * D@h, u1, u2D + u7 * D@h, u2, u2D + u8 * D@h, u3, u2DL +
2 * u8 * D@h, u1, u3D + u7 * D@h, u1, u2D;
V73@u1, u2, u3, u4, u5, u6, u7, u8D = 6 * Hu6 * D@h, u1, u3D + u7 * D@h, u2, u3D + u8 * D@h, u3, u3DL +
2 * u8 * D@h, u2, u3D + u7 * D@h, u2, u2D;
V78@u1, u2, u3, u4, u5, u6, u7, u8D = 2 * D@h, u3D;
7 * Hu7 * D@h, u1, u1D + u8 * D@h, u2, u1D + u9 * D@h, u3, u1DL +
V81@u1, u2, u3, u4, u5, u6, u7, u8, u9D =
7 * Hu7 * D@h, u1, u2D + u8 * D@h, u2, u2D + u9 * D@h, u3, u2DL +
V82@u1, u2, u3, u4, u5, u6, u7, u8, u9D =
7 * Hu7 * D@h, u1, u3D + u8 * D@h, u2, u3D + u9 * D@h, u3, u3DL +
V83@u1, u2, u3, u4, u5, u6, u7, u8, u9D =
8 * Hu8 * D@h, u1, u1D + u9 * D@h, u2, u1D + u10 * D@h, u3, u1DL +
V91@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D =
8 * Hu8 * D@h, u1, u2D + u9 * D@h, u2, u2D + u10 * D@h, u3, u2DL +
V92@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D =
8 * Hu8 * D@h, u1, u3D + u9 * D@h, u2, u3D + u10 * D@h, u3, u3DL +
V93@u1, u2, u3, u4, u5, u6, u7, u8, u9, u10D =
MatrixForm@VD
2J N
d@PD d@PD d@PD d@PD d@PD 2 d@PD
2 u3 2
2J N
d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD 2 d@PD
4 u4 2 2 u4 2
2J N
d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD 2 d@PD
4 u5 2 2 u5 2
2J N
d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD 2 d@PD
4 u6 2 2 u6 2
2J N
d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD 2 d@PD
4 u7 2 2 u7 2
2J N
d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD 2 d@PD
4 u8 2 2 u8 2
2J N
d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD 2 d@PD
4 u9 2 2 u9 2
2J N
d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD d@PD 2 d@PD
4 u10 2 2 u10 2
V101
Hamiltonian Systems Classification. for thesis.nb 7
t=1
t=1 w=1
Simplify@K11D
Simplify@K22D
I-b2 + 4 a gM
1
12
Manin-Kupershmidt bracket
H*Mathematica Programme for calculating the conditions imposed on a
particular chain when insisting that there exists one extra conservation law*L
H*Define the output form for functions of the chain*L
Format@H@u1, u2, u3DD := H
Format@P@u1, u2, u3, u4DD := P
Format@Q@u1, u2, u3, u4, u5DD := Q
Format@U1@x, tDD := U1
Format@U2@x, tDD := U2
Format@U3@x, tDD := U3
Format@U4@x, tDD := U4
Format@U5@x, tDD := U5
Format@U6@x, tDD := U6
d@U5D d@HD d@U4D d@HD d@U3D d@HD d@U2D d@HD d@U1D d@HD
2 + + 5 U5 + + +
dx dU3 dx dU2 dx dU3 2 dx dU2 dU3 dx dU1 dU3
d@U3D d@HD d@U2D d@HD d@U1D d@HD
4 U4 + + +
dx dU2 dU3 dx dU2 2 dx dU1 dU2
d@U3D d@HD d@U2D d@HD d@U1D d@HD d@PD
3 U3 + + +
dx dU1 dU3 dx dU1 dU2 dx dU1 2 dU4
d@U4D d@HD d@U3D d@HD d@U3D d@HD d@U2D d@HD d@U1D d@HD
2 + + 4 U4 + + +
dx dU3 dx dU2 dx dU3 2 dx dU2 dU3 dx dU1 dU3
d@U3D d@HD d@U2D d@HD d@U1D d@HD
3 U3 + + +
dx dU2 dU3 dx dU2 2 dx dU1 dU2
d@U3D d@HD d@U2D d@HD d@U1D d@HD d@PD
2 U2 + + +
dx dU1 dU3 dx dU1 dU2 dx dU1 2 dU3
d@U3D d@HD d@U2D d@HD d@U3D d@HD d@U2D d@HD d@U1D d@HD
2 + + 3 U3 + + +
dx dU3 dx dU2 dx dU3 2 dx dU2 dU3 dx dU1 dU3
d@U3D d@HD d@U2D d@HD d@U1D d@HD
2 U2 + + +
dx dU2 dU3 dx dU2 2 dx dU1 dU2
d@U3D d@HD d@U2D d@HD d@U1D d@HD d@PD
U1 + + +
dx dU1 dU3 dx dU1 dU2 dx dU1 2 dU2
d@U1D d@HD d@U3D d@HD d@U2D d@HD d@HD d@HD
+ U1 + 2 + 2 U2 + U1 +
dx dU2 dx dU2 dU3 dx dU3 dU2 dU3 dU22
d@U3D d@HD d@U1D d@HD d@U1D d@HD d@PD
2 U2 + + U1
dx dU32 dx dU1 dU3 dx dU1 dU2 dU1
D@q, xD
d@U5D d@QD d@U4D d@QD d@U3D d@QD d@U2D d@QD d@U1D d@QD
+ + + +
dx dU5 dx dU4 dx dU3 dx dU2 dx dU1
d@U5D d@HD d@U4D d@HD d@U3D d@HD d@U2D d@HD d@U1D d@HD
2 + + 5 U5 + + +
dx dU3 dx dU2 dx dU3 2 dx dU2 dU3 dx dU1 dU3
d@U3D d@HD d@U2D d@HD d@U1D d@HD
4 U4 + + +
dx dU2 dU3 dx dU2 2 dx dU1 dU2
d@U3D d@HD d@U2D d@HD d@U1D d@HD d@PD
3 U3 + + +
dx dU1 dU3 dx dU1 dU2 dx dU12 dU4
d@U4D d@HD d@U3D d@HD d@U3D d@HD d@U2D d@HD d@U1D d@HD
2 + + 4 U4 + + +
dx dU3 dx dU2 dx dU3 2 dx dU2 dU3 dx dU1 dU3
d@U3D d@HD d@U2D d@HD d@U1D d@HD
3 U3 + + +
dx dU2 dU3 dx dU2 2 dx dU1 dU2
d@U3D d@HD d@U2D d@HD d@U1D d@HD d@PD
2 U2 + + +
dx dU1 dU3 dx dU1 dU2 dx dU12 dU3
d@U3D d@HD d@HD d@HD d@HD
2 + 3 U3 + 2 U2 + U1 +
dx dU3 dU3 2 dU2 dU3 dU1 dU3
d@U2D d@HD d@HD d@HD d@HD
+ 3 U3 + 2 U2 + U1 +
dx dU2 dU2 dU3 dU22 dU1 dU2
d@U1D d@HD d@HD d@HD d@PD
3 U3 + 2 U2 + U1 +
dx dU1 dU3 dU1 dU2 dU1 2 dU2
d@U1D d@HD d@U3D d@HD d@U2D d@HD d@HD d@HD
+ U1 + 2 + 2 U2 + U1 +
dx dU2 dx dU2 dU3 dx dU3 dU2 dU3 dU22
d@U3D d@HD d@U1D d@HD d@U1D d@HD d@PD
2 U2 + + U1 -
dx dU3 2 dx dU1 dU3 dx dU1 dU2 dU1
d@U5D d@QD d@U4D d@QD d@U3D d@QD d@U2D d@QD d@U1D d@QD
- - - -
dx dU5 dx dU4 dx dU3 dx dU2 dx dU1
q2 = Simplify@D@q, u2DD;
q3 = Simplify@D@q, u3DD;
q4 = Simplify@D@q, u4DD;
q5 = Simplify@D@q, u5DD;
I dU3 M
6 U4 +
d@HD 2 dU3 2 dU4
2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@PD
3 3 U3 + U1 + U2 +2 +
dU3 2 dU2 dU3 dU2 dU3 dU1 dU3 dU2 dU3 dU3 2 dU1 dU3 dU4
d@HD d@HD d@PD d@HD d@PD
2 3 +
dU3 dU2 dU3 dU4 dU32 dU3
I dU3 M
d@HD 2
4
2
d@HD d@HD d@PD d@HD d@PD d@HD d@HD d@PD d@HD d@HD
12 U4 + 6 U2 +6 + 6 U2
dU3 2 dU1 dU3 dU4 dU1 dU3 dU4 dU3 dU1 dU2 dU4 dU2 dU3 dU1 dU2
d@PD d@HD d@HD d@PD d@HD d@HD d@HD d@HD d@PD
+ 3 U1 + 9 U3 + +
dU4 dU1 dU3 dU1 dU2 dU4 dU2 dU3 dU1 dU3 dU32 dU1 dU2 dU4
d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@HD d@PD
6 U2 + 3 U1 +4
dU3 2
dU1 2 dU4 dU2 dU3 dU12 dU4 dU3 dU1 dU3 dU3
4I M
12 U4 +6 +
d@HD 2
dU3 2 dU2 dU3 dU4 dU3 dU22 dU4
dU3
2
d@HD d@HD d@PD d@HD d@HD d@HD d@PD
6 U2 + 9 U3 + +
dU2 dU3 dU22 dU4 dU2 dU3 dU32
dU2 2 dU4
d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@HD d@PD
6 + 6 U2 + 3 U1 +
dU3 dU1 dU3 dU4 dU2 dU3 dU1 dU3 dU4 dU22 dU1 dU3 dU4
d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@HD d@PD
6 U2 + 3 U1 +4
dU32 dU1 dU2 dU4 dU2 dU3 dU1 dU2 dU4 dU3 dU2 dU3 dU3
4I M J2 N
27 U32 - + -
d@HD 2 d@HD d@HD d@HD dU2 dU3 dU2 2 dU1 dU3 dU2 dU3 dU1 dU2 dU4
dU3 dU3
+ 2 U2 dU2 dU3
+ U1 2
dU2
2 2
d@HD d@HD d@HD d@HD d@PD
3 U3 6 U2 - -
dU22 dU1 dU3 dU2 dU3 dU1 2 dU4
2
d@HD d@HD d@HD d@HD d@PD d@HD d@HD d@HD d@PD d@HD d@PD
3 U1 - +2 2 - +
dU2 dU3 dU1 dU2 dU2 2
dU1 2 dU4 dU3 dU1 dU3 dU22 dU3 dU3 dU22
d@HD d@HD d@PD d@HD d@PD d@HD d@PD
-3 -2 +2 +
dU2 dU3 dU1 2 dU4 dU1 dU2 dU3 dU3 dU1 dU2
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 -6 U22 - + -
dU1 dU3 dU1 dU2 dU2 dU3 dU1 dU3 dU12 dU2 dU3 dU1 dU2 dU22 dU12
2 2 2
d@PD d@HD d@HD d@HD d@HD d@HD d@HD d@PD
+ 2 3 U4 2 + U1 - +
dU4 dU3 dU1 dU3 dU2 2 dU1 dU3 dU2 dU3 dU1 2 dU4
2
d@HD d@HD d@PD d@PD
2 + U1 +
dU3 dU1 2 dU3 dU22
d@HD d@HD d@HD d@HD d@HD d@PD
U2 -12 U4 - + -
dU2 dU3 dU2 2 dU1 dU3 dU2 dU3 dU1 dU2 dU4
2 2
d@HD d@HD d@HD d@HD d@PD d@HD d@HD d@PD
3 U1 - + -6 -
dU1 dU3 dU1 dU2 dU2 2
dU1 2 dU4 dU3 dU1 dU2 dU4
d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@PD d@HD d@PD
4 +2 3 +3 +2 +
dU1 dU3 dU1 dU2 dU3 dU1 2
dU2 2 dU4 dU1 dU3 dU4 dU2 dU3 dU3
2
d@HD d@HD d@PD d@HD d@HD d@PD
4 + - +
dU3 dU1 dU2 dU22 dU2 2 dU1 dU3 dU1 dU2
4I M J2 U2 N
d@HD 2 d@HD d@HD
dU3
+ U1 dU2 dU3
dU32
2 2
d@HD d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@HD d@PD
54 U32 + 12 U2 + 12 U22 +
dU3 2 dU2 dU3 dU22 dU4 dU3 dU2 2 dU4 dU2 dU3 dU2 2 dU4
2
d@HD d@HD d@PD d@HD d@HD d@HD d@HD d@HD
6 U1 U2 + 12 U4 2 + 3 U3 +
dU2 2 dU1 dU3 dU4 dU2 dU3 dU3 dU2 dU3 dU3 2 dU2 dU3
2
d@HD d@HD d@HD d@HD d@HD d@PD
2 U2 + 2 U2 + 2 U1 +
dU2 dU3 dU3 2 dU1 dU3 dU2 dU3 dU1 dU3 dU4
2 2
d@HD d@HD d@PD d@HD d@HD d@HD d@PD d@HD d@HD d@PD
24 + 48 U2 + 24 U22 +
dU3 dU1 dU2 dU4 dU3 dU2 dU3 dU1 dU2 dU4 dU2 dU3 dU1 dU2 dU4
d@HD d@HD d@HD d@PD d@HD d@HD d@HD d@PD
12 U1 + 12 U22 +
dU3 dU22 dU1 dU2 dU4 dU3 2
dU2 2 dU1 dU2 dU4
d@HD d@HD d@HD d@PD d@HD d@HD d@HD d@PD
18 U1 U2 + 12 U1 +
dU2 dU3 dU22 dU1 dU2 dU4 dU3 dU1 dU3 dU1 dU2 dU4
d@HD d@HD d@HD d@PD d@HD d@HD d@HD d@PD
12 U22 + 18 U1 U2 +
dU3 2 dU1 dU3 dU1 dU2 dU4 dU2 dU3 dU1 dU3 dU1 dU2 dU4
2
d@HD d@HD d@HD d@PD d@HD d@HD d@PD
6 U12 + 6 U1 U2 +
dU2 2 dU1 dU3 dU1 dU2 dU4 dU3 2 dU1 dU2 dU4
2
d@HD d@HD d@PD d@HD d@HD d@HD d@PD d@HD d@HD d@HD d@PD
3 U12 + 12 U2 + 6 U1 +
dU2 dU3 dU1 dU2 dU4 dU3 dU3 2
dU1 2 dU4 dU3 dU2 dU3 dU12 dU4
2
d@HD d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@HD d@HD d@PD
12 U22 + 6 U1 U2 + 6 U1 U2 +
dU3 2 dU2 dU3 dU1 2 dU4 dU2 dU3 dU1 2 dU4 dU32 dU22 dU12 dU4
2
d@HD d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@HD d@HD d@PD
3 U12 +8 + 8 U2 +
dU2 dU3 dU22 dU12 dU4 dU3 dU2 2 dU3 dU3 dU2 dU3 dU22 dU3
d@HD d@HD d@HD d@PD d@HD d@HD d@HD d@PD d@HD d@HD d@HD d@PD
4 U1 + 8 U2 + 4 U1 -
dU3 dU22 dU1 dU3 dU3 dU3 dU32 dU1 dU2 dU3 dU3 dU2 dU3 dU1 dU2 dU3
3 2 2
d@HD d@PD d@HD d@HD d@PD d@HD d@HD d@PD
8 - 8 U2 - 4 U1 +
dU3 dU2 2 dU3 dU2 dU3 dU22 dU3 dU1 dU3 dU22
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 U3 3 2 U2 2 + + +3 +
dU2 dU3 dU2 2
dU3 2
dU2 2
dU2 2 dU1 dU3 dU3 2 dU2 dU3 dU1 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@PD d@HD
U1 3 + +2 +4
dU2 dU3 dU22 dU1 dU3 dU2 dU3 dU1 dU2 dU3 2
dU2 2 dU1 dU2 dU4 dU3
2
d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@PD d@HD d@HD d@PD
3 + 3 + -4
dU2 dU3 dU22 dU4 dU3 2 dU1 dU2 dU4 dU2 2 dU3 dU3 dU32 dU22
I dU3 M
6 U4 + 9 U3 + 3 U2 +
d@HD 2 dU2 dU3 dU4 dU2 dU3 dU22 dU4 dU22 dU4
2
d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@HD d@PD d@HD d@HD d@PD
6 + 6 U2 + 3 U1 +2
dU3 dU1 dU2 dU4 dU2 dU3 dU1 dU2 dU4 dU22 dU1 dU2 dU4 dU3 dU22 dU3
I dU3 M
-
d@HD 3
8
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 -2 U2 + 2 U2 +2 + 6 U2
dU2 2 dU1 dU3 dU1 dU2 dU1 dU3 dU1 dU2 dU3 dU1 dU2 dU2 dU3
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
+ U1 - 4 U2 -2 -
dU1 dU2 dU1 dU3 dU1 dU2 dU3 dU1 dU2 dU1 dU22 dU3 dU22 dU12
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 U2 - 10 - 2 U2 -
dU2 dU3 dU22 dU12 dU3 dU1 dU3 dU12 dU2 dU3 dU1 dU3 dU12
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 U1 + 2 U1 - 4 U2 -
dU22 dU1 dU3 dU12 dU2 dU3 dU1 dU2 dU12 dU3 dU1 dU2 dU3 dU12
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U1 + 4 U2 + 8 U4
dU3 dU1 dU22 dU12 dU3 dU1 dU2 dU12 dU3 dU32
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
- + - + +
dU1 dU3 dU1 dU2 dU2 dU3 dU1 2 dU3 dU1 dU3 dU1 dU2 dU3 dU2 dU3 dU12 dU3
d@HD d@HD d@HD d@HD d@HD d@HD
4 U2 - 4 U2 -
dU3 dU2 2 2
dU1 dU2 dU3 dU1 dU3 dU12 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 U3 2 - + + +
dU3 2 dU1 dU2 dU3 dU1 dU2 dU1 dU2 dU3 dU1 dU3 dU1 dU22
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 + +2 -2 -
dU2 dU3 dU12 dU22 dU1 dU3 dU32 dU12 dU3 dU12 dU3
d@HD d@HD d@HD d@HD d@HD
3 +2 +
dU2 dU3 dU1 dU3 dU1 dU2 dU3 dU12 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@PD
4 + 4 U2 + 2 U1
dU3 dU1 3 dU3 dU2 dU3 dU13 dU3 dU22 dU13 dU4
12 Extra conservation law .nb
3
1 d@HD d@HD d@HD d@HD
8I M
- 3 2 U2 -4 +
d@HD 3 dU1 dU3 dU3 dU1 dU3 dU1 dU2
dU3
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U2 + U1 + 2 U2 -
dU2 dU3 dU1 dU3 dU1 dU2 dU1 dU3 dU1 dU2 dU32 dU1 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 U2 -6 - 4 U2 -
dU3 dU1 dU2 dU1 dU2 dU3 dU3 dU2 dU3 dU1 2 dU2 dU3 dU12
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U2 - 3 U1 - 4 U2 +
dU3 2 dU1 dU3 dU1 2 dU2 dU3 dU1 dU3 dU1 2 dU3 dU1 dU3 2
dU12
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U1 - 2 U1 - 2 U1 -
dU3 2 dU1 dU2 dU1 2 dU3 dU1 dU2 dU3 dU1 2 dU3 dU1 dU2 dU12 dU3
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
8 U4 + - + +
dU3 dU1 dU3 dU1 dU3 2
dU3 2
dU1 2
dU3 2 dU1 dU3 dU3 dU12 dU3
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 + 4 U2 + 2 U1 +
dU3 2
dU1 dU2 dU3 dU2 dU3 dU12 dU2 dU3 dU1 dU3 dU12 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 U3 -2 + + -
dU3 dU1 dU32 dU1 dU2 dU1 dU3 dU1 dU2 dU3 dU2 dU3 dU1 dU3
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 +2 + 3 +2 +
dU3 2
dU1 2 dU3 dU1 dU3 2
dU3 2 dU1 dU3 dU1 dU2 dU3 dU12 dU2
d@HD d@HD d@HD d@HD d@HD d@HD d@PD
4 U2 + 2 U1
dU3 dU32 dU13 dU3 dU2 dU3 dU13 dU4
4I M
d@HD 2
dU3
Extra conservation law .nb 13
2 2
1 d@HD d@HD d@HD d@HD d@HD d@HD d@HD
8I M
- 3 -2 U2 + 4 U2 + 2 U2 -
d@HD 3
dU2 2 dU1 dU3 dU2 2 dU1 dU3 dU2 dU3 dU2 2 dU1 dU2
dU3
2
1 d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
8I M
- 3 -2 + 4 U2 +8 + 2 U2
d@HD 3 dU3 dU2 2 dU1 dU3 dU2 dU3 dU2 2 dU1 dU3 dU3 dU1 dU3 dU2 dU3
dU3
2 2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
+ 2 U1 -8 - 6 U2 +
dU1 dU3 dU2 2 dU1 dU3 dU3 dU2 dU3 dU1 dU2 dU2 dU3 dU1 dU2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U2 - 2 U2 - 3 U1 -
dU3 2
dU2 2 dU1 dU2 dU3 2 dU1 dU3 dU1 dU2 dU2 dU3 dU1 dU3 dU1 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 U2 + U1 - 4 U2 +
dU3 dU1 dU3 2 dU1 dU2 dU3 2 dU1 dU2 dU3 dU2 2 dU1 dU2 dU3
d@HD d@HD d@HD d@HD d@HD d@HD
4 U2 - 2 U1 - 8 U4
dU3 dU1 dU3 dU1 dU2 dU3 dU3 dU1 dU2 dU1 dU2 dU3
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
+ - + +
dU3 dU2 dU3 dU1 dU3 2
dU3 2 dU1 dU2 dU3 2 dU2 dU3 dU1 dU3 dU3 dU1 dU2 dU3
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 + 4 U2 + 2 U1 +
dU3 dU1 dU2 2 dU3 dU2 dU3 dU1 dU2 2 dU3 dU1 dU3 dU1 dU22
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 U3 -2 -4 +
dU2 dU3 dU1 dU3 dU3 dU22 dU1 dU32 dU3 2 dU2 dU3 dU1 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 +2 +2 - U1 +
dU3 2
dU2 2 dU1 dU3 dU3 dU1 dU22 dU3 dU32 dU12 dU2 dU3 dU12
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
U1 -4 - 4 U2 -
dU3 2
dU2 2
dU1 2 dU3 dU1 dU3 2 dU3 dU2 dU3 dU12 dU3
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@PD
2 U1 + 4 U2 + 2 U1
dU3 dU22 dU12 dU3 dU3 dU32 dU12 dU2 dU3 dU2 dU3 dU12 dU2 dU4
Extra conservation law .nb 15
2
1 d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
8I M
- 3 -2 - 2 U2 +2 +
d@HD 3 dU3 dU22 dU1 dU3 dU2 dU3 dU22 dU1 dU3 dU3 dU1 dU3
dU3
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U2 - U1 -8 +
dU2 dU3 dU1 dU3 dU2 2 dU1 dU3 dU3 dU2 dU3 dU1 dU2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U2 - 4 U2 - 8 U2 -
dU32
dU22 dU1 dU2 dU3 dU22 dU3 dU1 dU2 dU3 2 dU1 dU3 dU1 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 U1 + U1 - 2 U1 - 8 U4
dU2 dU3 dU1 dU3 dU1 dU2 dU3 2 dU1 dU2 dU3 dU1 dU2 dU1 dU2 dU3
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
+ - + +
dU3 dU2 dU32 dU1 dU3 dU3 2 dU1 dU2 dU3 2 dU2 dU3 dU1 dU3 dU3 dU1 dU2 dU3
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 + 4 U2 + 2 U1 +
dU3 dU1 dU2 2 dU3 dU2 dU3 dU1 dU22 dU3 dU1 dU3 dU1 dU22
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 U3 + - +2 -
dU2 dU3 dU1 dU3 dU2 dU3 dU3 2 dU1 dU2 dU3 dU1 dU2 dU3
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 + - -
dU3 dU2 dU3 dU1 dU3
2
dU2 dU3 2 dU1 dU2 dU32
dU1 dU22
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 + 6 U2 + 2 U1 -
dU3 dU32 dU12 dU3 2 dU2 dU3 dU12 dU2 dU3 dU12
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 U2 + U1 - 2 U1 +
dU3 dU2 dU32 dU12 dU3 2
dU2 2
dU1 2 dU3 dU22 dU3 dU12
d@HD d@HD d@HD d@HD d@HD d@HD d@PD
4 U2 + 2 U1
dU3 dU3 2
dU1 dU2 2 dU3 dU2 dU3 dU1 dU2 2 dU4
3 J3 N dU4
d@HD d@HD d@HD d@HD d@HD d@HD d@PD
dU2 dU3 dU1 dU3
+2 -2
dU32 dU1 dU2 dU3 dU1 dU2 dU3
4I M
d@HD 2
dU3
3 J3 N dU4
d@HD d@HD d@HD d@HD d@HD d@HD d@PD
dU2 dU3 dU1 dU3
+2 -2
dU32 dU1 dU2 dU3 dU1 dU2 dU3
4I M
d@HD 2
dU3
16 Extra conservation law .nb
I dU3 M
- 3 8 U4 - + +
d@HD 3 dU3 dU33 dU1 dU3 dU3 2
dU1 dU32
8
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 + 3 3 U3 - + -
dU3 dU1 dU2 dU3 dU3 2 dU2 dU3 dU1 dU3 dU3 2 dU1 dU2
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U2 - + - +
dU2 dU3 dU1 dU3 dU3 2 dU2 dU3 dU1 dU2 dU3 2 dU1 dU3 dU32 dU12
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
U1 + -2 + -
dU3 2 dU1 dU3 dU1 dU2 dU2 dU3 dU1 dU3 dU32 dU12
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 2 + 2 U2 + U1 -
dU3 dU3 2 dU1 dU2 dU2 dU3 2 dU1 dU2 dU1 dU3 2 dU1 dU2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
U1 + 3 U3 + - +
dU1 dU3 dU1 dU2 dU3 dU2 dU3 2 dU1 dU3 dU3 3 dU1 dU2 dU3 2 dU1 dU2 dU3
d@HD d@HD d@HD d@HD d@HD d@HD
2 U2 + U1 - 2 U2 +
3 2 2 2 2
dU3 dU1 dU2 dU3 dU1 dU3 dU12 dU3
d@HD d@HD d@HD d@HD d@HD d@PD
3 - 3 U3 - 2 U2 - U1
dU2 dU3 dU1 dU3 dU1 dU3 2 dU1 dU2 dU3 2
dU1 dU3 dU4
3 J5 N
d@HD d@HD d@HD d@HD d@PD
-2
dU32 dU1 dU3 dU3 dU1 dU32 dU4
4I M
d@HD 2
dU3
3 J5 N
d@HD d@HD d@HD d@HD d@PD
-2
dU32 dU1 dU3 dU3 dU1 dU32 dU4
4I M
d@HD 2
dU3
0
Extra conservation law .nb 17
I dU3 M
-
d@HD 3
8
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 -10 - 2 U2 + 2 U2 - 4 U2 -
dU3 dU2 dU3 dU22 dU2 dU3 dU2 2
dU3 2
dU2 2 dU3 dU22 dU22 dU3
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
8 U4 + - + +
dU3 dU2 dU3 dU2 dU32 dU3 2
dU2 2
dU3 2 dU2 dU3 dU3 dU22 dU3
2
d@HD d@HD d@HD d@HD d@HD
4 + 4 U2 +
dU3 dU2 3 dU3 dU2 dU3 dU23
3
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 U3 - +2 - + +
dU2 dU3 dU3 2 dU2 dU3 dU2 2 dU3 dU2 dU3 2
dU2 2
dU32 dU23
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
10 + 2 U2 - 8 U2 -
dU3 dU2 dU3 dU1 dU3 dU2 dU3 dU1 dU3 dU3 2
dU2 2 dU1 dU3
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
3 U1 + 4 U2 + 2 U1 +
dU2 dU3 dU2 2 dU1 dU3 dU3 dU2 dU3 dU1 dU3 2 dU3 dU2 3 dU1 dU3
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
6 U2 + U1 - 4 U2 +
dU3 2 dU2 dU3 dU1 dU2 dU2 dU3 dU1 dU2 dU3 dU2 dU32 dU1 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U1 - 2 U1 -4 -
dU3 2
dU22 dU1 dU2 dU3 dU2 dU3 dU1 dU2
2 dU3 dU1 dU2 dU3
d@HD d@HD d@HD d@HD d@HD d@HD
4 U2 - 2 U1 +
dU3 dU2 dU3 dU1 dU2 dU3 dU3 dU22 dU1 dU2 dU3
d@HD d@HD d@HD d@HD d@HD d@HD d@PD
4 U2 + 2 U1
dU3 dU3 2
dU1 dU2 2 dU3 dU2 dU3 dU1 dU2 2 dU4
4I M
d@HD 2
dU3
18 Extra conservation law .nb
I dU3 M
- 3 8 U4 - + +
d@HD 3 dU3 dU3 3 dU2 dU3 dU3 2
dU2 dU32
8
2 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
4 - + 3 3 U3 - + -
dU3 2
dU2 dU3 dU1 dU3 2
dU3 2 dU2 dU3 dU32 dU22
3 2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 U2 + - + - +
dU2 dU3 dU3 2 dU2 dU3 dU2 2 dU1 dU3 dU3 2 dU1 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
U1 -2 + + -
dU2 dU3 dU1 dU3 dU3 2
dU2 2 dU1 dU3 dU3 2 dU2 dU3 dU1 dU2
2
d@HD d@HD d@HD d@HD d@HD d@HD d@HD d@HD
2 3 + 3 U3 + 2 U2 - 2 U2 -
dU3 dU2 dU3 dU3 3
dU2 2
dU2 dU3 2
dU2 2
dU2 dU3 2 dU1 dU3
d@HD d@HD d@HD d@HD d@HD d@HD
U1 + U1 + 2 U2 +
dU2 dU3 dU1 dU3
2
dU2 2
dU1 dU3 2
dU3 3 dU1 dU2
d@HD d@HD d@HD d@HD d@HD d@HD
U1 - 2 U2 - + U1 +
dU2 dU3 2 dU1 dU2 dU2 dU3 dU2 dU3 2
dU1 dU3 2 dU1 dU2 dU3
d@HD d@HD d@HD d@HD d@HD d@PD
2 - 3 U3 -5 - 2 U2
dU3 2
dU2 2
dU2 dU3 2 dU1 dU3 dU1 dU2 dU3 dU4
3 J5 N
d@HD d@HD d@HD d@HD d@PD
-2
dU32 dU2 dU3 dU3 dU2 dU32 dU4
4I M
d@HD 2
dU3
3 J5 N
d@HD d@HD d@HD d@HD d@PD
-2
dU32 dU2 dU3 dU3 dU2 dU32 dU4
4I M
d@HD 2
dU3
3 K5 J N -2 O dU4
d@HD 2 d@HD d@HD d@PD
dU32 dU3 dU33
4I M
d@HD 2
dU3
0
Extra conservation law .nb 19
3 K5 J N -2 O dU4
d@HD 2 d@HD d@HD d@PD
dU32 dU3 dU33
4I M
d@HD 2
dU3
3 I dU1 dU3 M + 2
d@HD 2 d@HD d@HD
dU32 dU12
d@HD
2 dU3
d@HD d@HD
5
dU32 dU1 dU3
d@HD
2 dU3
3 I dU2 dU3 M + 2
d@HD 2 d@HD d@HD
dU32 dU22
d@HD
2 dU3
Extra conservation law .nb 21
d@HD d@HD
5
dU32 dU2 dU3
d@HD
2 dU3
5J N
d@HD 2
dU32
d@HD
2 dU3
References 168
References
[1] S.J. Alber, Associated integrable systems, J. Math. Phys. 32 (1991) 916-922.
[2] M.J.Ablowitz and H.Segur, Solitons and the Inverse scattering transform. Siam
(1981).
[3] D.J. Benney, Some properties of long nonlinear waves, Stud. Appl. Math. 52
(1973) 45-50.
[4] D.J. Benney, Conservation laws associated with long surface waves, Journal of
Computational and Applied Mathematics 190 (2006) 136-141.
[6] V.M. Bukhshtaber, D.V. Leikin and M.V. Pavlov, Egorov hydrodynamic chains,
the Chazy equation, and the group SL(2, C), Funktsional. Anal. i Prilozhen.
37, no. 4 (2003) 13-26.
[7] J. Chazy, Sur les équations différentielles dont l’intégrale générale poss ède un
coupure essentielle mobile, C.R. Acad. Sci. Paris 150 (1910) 456-458.
[8] B.A. Dubrovin and S.P. Novikov, Hydrodynamics of weakly deformed soliton
lattices: differential geometry and Hamiltonian theory, Russian Math. Surveys,
44 , no. 6 (1989) 35-124.
[9] B.A. Dubrovin, A.T.Fomenko and S.P. Novikov, Geometry-Methods and Ap-
plications, Part I: The Geometry of Surfaces of Transformation Groups, and
Fields. Springer (1984)
[10] B.A. Dubrovin, Geometry of 2D topological field theories, Lect. Notes in Math.
1620, Springer-Verlag (1996) 120-348.
References 169
[17] E.V. Ferapontov, K.R. Khusnutdinova and M.V. Pavlov, Classification of in-
tegrable (2+1)-dimensional quasilinear hierarchies, Theor. Math. Phys. 144
(2005) 35-43.
[18] E.V. Ferapontov, K.R. Khusnutdinova, D.G. Marshall and M.V. Pavlov, Clas-
sification of integrable Hamiltonian hydrodynamic chains associated with Ku-
pershmidt’s brackets, J. Math. Phys. 47, no.1 (2006).
References 170
[19] E.V. Ferapontov, K.R. Khusnutdinova and S.P. Tsarev, On a class of three-
dimensional integrable Lagrangians, Comm. Math. Phys. 261, N1 (2006) 225-
243.
[23] J. Gibbons and Y. Kodama, A method for solving the dispersionless KP hier-
archy and its exact solutions. II, Phys. Lett. A 135, no. 3 (1989) 167-170.
[24] J. Gibbons and S. P. Tsarev, Reductions of the Benney equations, Phys. Lett.
A 211 (1996) 19-24.
[25] J. Gibbons and S. P. Tsarev, Conformal maps and reductions of the Benney
equations, Phys. Lett. A 258 (1999) 263-271.
[31] I.M. Krichever, The τ -function of the universal Whitham hierarchy, matrix
models and topological field theories, Comm. Pure Appl. Math. 47, no. 4 (1994)
437-475.
[32] B.A. Kupershmidt and Yu.I. Manin, Long wave equations with a free surface.
I. Conservation laws and solutions. (Russian) Funktsional. Anal. i Prilozhen.
11, no. 3 (1977) 31-42.
[33] B.A. Kupershmidt and Yu.I. Manin, Long wave equations with a free surface.
II. The Hamiltonian structure and the higher equations. (Russian) Funktsional.
Anal. i Prilozhen. 12, no. 1 (1978) 25-37.
[34] B.A. Kupershmidt, Deformations of integrable systems. Proc. Roy. Irish Acad.
Sect. A 83, no. 1 (1983) 45-74.
[39] D.R. Lebedev and Yu.I. Manin, Conservation laws and Lax representation for
Benney’s long wave equations, Phys. lett. 74A (1979) 154-156.
[41] M. Mañas, S-functions, reductions and hodograph solutions of the rth disper-
sionless modified KP and Dym hierarchies, J. Phys. A: Math. Gen. 37 (2004)
11191-11221.
[45] R.M. Miura, Conservation laws for fully nonlinear long wave equations, Stud.
Appl. Math. 53 (1974) 45-56.
[46] M.V. Pavlov, Integrable hydrodynamic chains, J. Math. Phys. 44, no. 9 (2003)
4134-4156.
[48] M.V. Pavlov, S.I. Svinolupov and R.A. Sharipov, An invariant criterion for
hydrodynamic integrability, Funktsional. Anal. i Prilozhen. 30 (1996) 18-29;
translation in Funct. Anal. Appl. 30 (1996) 15-22.
[52] M.V. Pavlov, The Hamiltonian approach in the classification of integrable hy-
drodynamic chains, nlin.SI/0603057.
[54] M.V. Pavlov, Hydrodynamic chains and a classification of their Poisson brack-
ets, nlin.SI/0603056.
[55] Z. Peradzyński, Nonlinear plane k-waves and Riemann invariants, Bull. Acad.
Polon. Sci. Sr. Sci. Tech. 19 (1971) 625-632.
[57] L. Martinez Alonso and A.B. Shabat, Towards a theory of differential con-
straints of a hydrodynamic hierarchy, J. Nonlinear Math. Phys. 10, no. 2 (2003)
229-242.
References 174
[59] V.M. Teshukov, On the hyperbolicity of long wave equations, Dokl. Akad. Nauk
SSSR 284, no. 3 (1985) 555-559.
[60] V.M. Teshukov, Characteristics, conservation laws and symmetries of the ki-
netic equations of motion of bubbles in a fluid, Prikl. Mekh. Tekhn. Fiz. 40,
no. 2 (1999) 86-100; translation in J. Appl. Mech. Tech. Phys. 40, no. 2 (1999)
263-275.
[62] V.E. Zakharov, Benney equations and quasiclassical approximation in the in-
verse problem method, Funktsional. Anal. i Prilozhen. 14, no. 2 (1980) 15-24.
[63] V.E. Zakharov, On the Benney equations, Physica 3D, no. 1,2 (1981) 193-202.