<?xml version="1.0" encoding="utf-8"?>
<?xml-stylesheet type="text/css" href="article.css"?>
<!DOCTYPE article PUBLIC "ISO 12083:1994//DTD Article//EN" "article.dtd">
<article>
<front>
<title>Involutive orbits of non-Noether symmetry groups</title>
<author>
	<fname>George</fname>
	<surname>Chavchanidze</surname>
</author>
<aff>
	<orgdiv>Department of Theoretical Physics</orgdiv>
	<orgname>A. Razmadze Institute of Mathematics</orgname>
	<street>1 Aleksidze Street</street>
	<city>Tbilisi</city>
	<postcode>0193</postcode>
	<country>Georgia</country>
</aff>
<abstract>
<title>Abstract</title>
<p>We consider set of functions on Poisson manifold related by continues one-parameter group 
of transformations. Class of vector fields that produce involutive families of functions
is investigated and relationship between these vector fields and non-Noether 
symmetries of Hamiltonian dynamical systems is outlined. Theory is illustrated 
with sample models: modified Boussinesq system and Broer-Kaup system.</p>
</abstract>
<supmatl>
<p>
	<emph>Keywords:</emph>
	<keyphras>Non-Noether symmetry</keyphras>,
	<keyphras>Conservation laws</keyphras>,
	<keyphras>Modified Boussinesq system</keyphras>,
	<keyphras>Broer-Kaup system</keyphras>
</p>
<p><emph>MSC 2000:</emph> 70H33; 70H06; 58J70; 53Z05; 35A30</p>
<!--J. Phys. A: Math. Gen. 38 (2005) 6517-6524-->
</supmatl>
</front>
<chapter>
<p>
In Hamiltonian integrable models, conservation laws often form involutive orbit of
one-parameter symmetry group. Such a symmetry carries important information about 
integrable model and its bi-Hamiltonian structure. The present paper is an attempt to
describe class of one-parameter group of transformations of Poisson manifold 
that possess involutive orbits and may be related to Hamiltonian integrable systems.
</p>
<p>
Let <formula>C<sup>∞</sup>(M)</formula> be algebra of smooth functions on manifold <formula>M</formula> equipped with Poisson bracket
<dformula>
{f , g} = W(df ∧ dg)
</dformula> where <formula>W</formula> is Poisson bivector satisfying property <formula>[W , W] = 0</formula>.
Each vector field <formula>E</formula> on manifold <formula>M</formula> gives rise to one-parameter group of transformations of 
<formula>C<sup>∞</sup>(M)</formula> algebra
<dformula>
g<inf>z</inf> = e<sup>zL<inf>E</inf></sup>
</dformula>
where <formula>L<inf>E</inf></formula> denotes Lie derivative along the vector field <formula>E</formula>.
To any smooth function <formula>J ∈ C<sup>∞</sup>(M)</formula> this group assigns orbit that goes through <formula>J</formula>
<dformula>
J(z) = g<inf>z</inf>(J) = e<sup>zL<inf>E</inf></sup>(J) = J + zL<inf>E</inf>J + ½z<sup>2</sup>(L<inf>E</inf>)<sup>2</sup>J + ⋯
</dformula>
the orbit <formula>J(z)</formula> is called involutive if 
<dformula>
{J(x) , J(y)} = 0          ∀x, y ∈ ℝ
</dformula>
Involutive orbits are often related to integrable models where <formula>J(z)</formula>
plays the role of involutive family of conservation laws.
</p>
<p>
Involutivity of orbit <formula>J(z)</formula> depends on nature of vector field  <formula>E</formula> and function <formula>J = J(0)</formula> 
and in general it is hard to describe all pairs <formula>(E , J)</formula> that produce involutive orbits
however one interesting class of involutive orbits can be outlined by the following theorem:
</p>
<p><emph>Theorem 1.</emph> 
For any non-Poisson <formula>[E , W] ≠ 0</formula> vector field <formula>E</formula> 
satisfying property 
<dformula id="e5">
[E , [E , W]] = 0
</dformula> and any function <formula>J</formula> such that
<dformula id="e6">
W(dL<inf>E</inf>J) = c[E , W](dJ)          c ∈ ℝ∖(0∪ℕ)
</dformula>
one-parameter family of functions <formula>J(z) = e<sup>zL<inf>E</inf></sup>(J)</formula> is involutive.
</p>
<p><emph>Proof.</emph> 
By taking Lie derivative of  property <formref rid="e6">(6)</formref> along the vector field <formula>E</formula> we get
<dformula>
[E , W](dL<inf>E</inf>J) +  W(d(L<inf>E</inf>)<sup>2</sup>J) = c[E,[E , W]](dJ) + c[E , W](dL<inf>E</inf>J)
</dformula>
where <formula>c</formula> is real constant which is neither zero nor positive integer.
Taking into account <formref rid="e5">(5)</formref> one can rewrite result as follows
<dformula>
W(d(L<inf>E</inf>)<sup>2</sup>J) =  (c − 1)[E , W](dL<inf>E</inf>J)
</dformula>
that after <formula>m</formula> iterations produces
<dformula id="e9">
W(d(L<inf>E</inf>)<sup>m + 1</sup>J) =  (c − m)[E , W](d(L<inf>E</inf>)<sup>m</sup>J)
</dformula>
Now using this property let us prove that functions  
<formula>J<sup>(m)</sup> = (L<inf>E</inf>)<sup>m</sup>J</formula> are in involution.
Indeed
<dformula>
{J<sup>(k)</sup>, J<sup>(m)</sup>} = W(dJ<sup>(k)</sup> ∧ dJ<sup>(m)</sup>)
</dformula>
Suppose that <formula>k &gt; m</formula> and let us rewrite Poisson bracket as follows
<dformula>
W(dJ<sup>(k)</sup> ∧ dJ<sup>(m)</sup>) = W(d(L<inf>E</inf>)<sup>k</sup>J ∧ dJ<sup>(m)</sup>) 
= L<inf>W(d(L<inf>E</inf>)<sup>k</sup>J)</inf>J<sup>(m)</sup><break/>
= (c − k + 1)L<inf>[E , W](d(L<inf>E</inf>)<sup>k − 1</sup>J)</inf>J<sup>(m)</sup> 
= (c −  k + 1)[E , W](dJ<sup>(k − 1)</sup> ∧ dJ<sup>(m)</sup>)<break/>
= − (c −  k + 1)L<inf>[E , W](d(L<inf>E</inf>)<sup>m</sup>J)</inf>J<sup>(k − 1)</sup>
= − <fraction><num>c − k + 1</num><den>c − m</den></fraction>L<inf>W(d(L<inf>E</inf>)<sup>m + 1</sup>J)</inf>J<sup>(k − 1)</sup><break/>
= <fraction><num>c − k + 1</num><den>c − m</den></fraction>W(dJ<sup>(k − 1)</sup> ∧ dJ<sup>(m + 1)</sup>)
</dformula>
Thus we have
<dformula>
(c − m){J<sup>(k)</sup>, J<sup>(m)</sup>} = (c − k + 1){J<sup>(k  − 1)</sup>, J<sup>(m + 1)</sup>} 
</dformula>
Using this property <formula>2(m − k)</formula> times produces
<dformula>
{J<sup>(k)</sup>, J<sup>(m)</sup>} = {J<sup>(m)</sup>, J<sup>(k)</sup>}
</dformula>
and since Poisson bracket is skew-symmetric we finally get
<dformula>
{J<sup>(k)</sup>, J<sup>(m)</sup>} = 0
</dformula>
So we showed that functions <formula>J<sup>(m)</sup> = (L<inf>E</inf>)<sup>m</sup>J</formula> are in involution.
In the same time orbit <formula>J(z)</formula> is linear combination of functions <formula>J<sup>(m)</sup></formula>
and thus it is involutive as well.
 ∎</p>
<p><!--Remark--> 
Property <formref rid="e9">(9)</formref> implies that vector field 
<dformula>
S = (c − m)E + t(c − m + 1)W(dJ<sup>(m + 1)</sup>)
</dformula> 
is non-Noether symmetry <citeref rid="r1">[1]</citeref> of Hamiltonian dynamical system
<dformula>
<fraction><num>d</num><den>dt</den></fraction>f = {J<sup>(m)</sup>, f}
</dformula>
in other words non-Poisson vector field <formula>S</formula> 
commutes with time evolution defined by Hamiltonian vector field
<dformula>
X = <fraction><num>∂</num><den>∂t</den></fraction> + W(dJ<sup>(m)</sup>)
</dformula>
This fact can be checked directly
<dformula>
[S , X] = (c − m)[E , X] + t(c − m + 1)[W(dJ<sup>(m + 1)</sup>), W(dJ<sup>(m)</sup>)] <break/>
− (c − m + 1)W(dJ<sup>(m + 1)</sup>) = (c − m)[E , W](dJ<sup>(m)</sup>) + (c − m)W(dL<inf>E</inf>J<sup>(m)</sup>) <break/>
+ t(c − m + 1)W(d{J<sup>(m + 1)</sup>,J<sup>(m)</sup>}) − (c − m + 1)W(dJ<sup>(m + 1)</sup>) <break/>
= W(dJ<sup>(m + 1)</sup>) + (c − m)W(dJ<sup>(m + 1)</sup>) − (c − m + 1)W(dJ<sup>(m + 1)</sup>) = 0
</dformula>
In the same time property <formref rid="e9">(9)</formref> means that functions 
<formula>J<sup>(m)</sup> = (L<inf>E</inf>)<sup>m</sup>J</formula> form 
Lenard scheme with respect to bi-Hamiltonian structure formed by Poisson bivector 
fields <formula>W</formula> and <formula>[E , W]</formula> (see <citeref rid="r1">[1]</citeref>,<citeref rid="r4">[4]</citeref>).
</p>
<p>
In many infinite dimensional integrable Hamiltonian systems Poisson bivector has nontrivial kernel,
and set of conservation laws belongs to orbit of non-Noether symmetry group that goes through
centre of Poisson algebra. This fact is reflected in the following theorem:
</p>
<p><emph>Theorem 2.</emph> 
If non-Poisson vector field <formula>E</formula> satisfies property
<dformula>
[E, [E , W]] = 0
</dformula> then every orbit derived from centre <formula>I</formula> 
of Poisson algebra <formula>C<sup>∞</sup>(M)</formula> is involutive. 
</p>
<p><emph>Proof.</emph> 
If function <formula>J</formula> belongs to centre <formula>J ∈ I</formula> of Poisson algebra <formula>C<sup>∞</sup>(M)</formula>
then by definition <formula>W(dJ) = 0</formula>. By taking Lie derivative of this condition along vector field <formula>E</formula>
one gets 
<dformula>
W(dL<inf>E</inf>J) = − [E , W](dJ)
</dformula>
that according to Theorem 1 ensures involutivity of <formula>J(z)</formula> orbit.
 ∎</p>
<p><emph>Example.</emph> 
The theorems proved above may have interesting applications in theory of infinite dimensional
Hamiltonian models where they provide simple way to construct involutive family of conservation laws.
One non-trivial example of such a model is modified Boussinesq system 
<citeref rid="r2">[2]</citeref>,<citeref rid="r5">[5]</citeref>,<citeref rid="r6">[6]</citeref> described by the following
set of partial differential equations
<dformula>
u<inf>t</inf> = cv<inf>xx</inf> + u<inf>x</inf>v + uv<inf>x</inf><break/>
v<inf>t</inf> = − cu<inf>xx</inf> + uu<inf>x</inf> + 3vv<inf>x</inf>
</dformula>
where <formula>u = u(x, t), v = v(x, t)</formula> are smooth functions on <formula>ℝ<sup>2</sup></formula>
subjected to zero boundary conditions <formula>u(±∞, t) = v(±∞, t) = 0</formula>
This system can be rewritten in Hamiltonian form
<dformula>
<fraction><num>d</num><den>dt</den></fraction>f = {h, f} = W(dh ∧ df)
</dformula>
with the following Hamiltonian
<dformula>
h = ½<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (u<sup>2</sup>v + v<sup>3</sup> + 2cuv<inf>x</inf>)dx
</dformula>
and Poisson bracket defined by Poisson bivector field
<dformula id="e24">
W = ½<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (A ∧ A<inf>x</inf> + B ∧ B<inf>x</inf>)dx
</dformula>
where <formula>A, B</formula> are vector fields that for every smooth functional <formula>R = R(u)</formula> are defined
via variational derivatives <formula>A(R) = δR/δu</formula> and <formula>B(R) = δR/δv</formula>.
For Poisson bivector <formref rid="e24">(24)</formref> there exist vector field <formula>E</formula> such that
<dformula>
[E,[E,W]] = 0
</dformula>
this vector field has the following form
<dformula>
E = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (uvA<inf>x</inf> − cvA<inf>xx</inf> + 
(uu<inf>x</inf> + vv<inf>x</inf>)B + (u<sup>2</sup> + 2v<sup>2</sup>)B<inf>x</inf> + cuB<inf>xx</inf>)xdx<break/>
= − <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> [(uv + 2cv<inf>x</inf> + x((uv)<inf>x</inf> + cv<inf>xx</inf>))A<break/>
+ (u<sup>2</sup> + 2v<sup>2</sup> − 2cu<inf>x</inf> + x(uu<inf>x</inf> + 3vv<inf>x</inf> − cu<inf>xx</inf>))B]dx
</dformula>
Applying one-parameter group of transformations generated by this vector field to centre of Poisson algebra 
which in our case is formed by functional 
<dformula>
J = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (ku + mv)dx
</dformula>
where <formula>k, m</formula> are arbitrary constants, produces involutive orbit that recovers 
infinite sequence of conservation laws of modified Boussinesq hierarchy
<dformula>
J<sup>(0)</sup> = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (ku + mv)dx<break/>
J<sup>(1)</sup> = L<inf>E</inf>J<sup>(0)</sup>
=  <fraction><num>m</num><den>2</den></fraction><subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>(u<sup>2</sup> + v<sup>2</sup>)dx<break/>
J<sup>(2)</sup> = (L<inf>E</inf>)<sup>2</sup>J<sup>(0)</sup> 
= m<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (u<sup>2</sup>v + v<sup>3</sup> + 2cuv<inf>x</inf>)dx<break/>
J<sup>(3)</sup> = (L<inf>E</inf>)<sup>3</sup>J<sup>(0)</sup> = 
<fraction><num>3m</num><den>4</den></fraction><subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (u<sup>4</sup> + 5v<sup>4</sup> + 6u<sup>2</sup>v<sup>2</sup><break/>
 − 12cv<sup>2</sup>u<inf>x</inf> + 4c<sup>2</sup>u<inf>x</inf><sup>2</sup> + 4c<sup>2</sup>v<inf>x</inf><sup>2</sup>)dx<break/>
J<sup>(m)</sup> = (L<inf>E</inf>)<sup>m</sup>J<sup>(0)</sup> = L<inf>E</inf>J<sup>(m − 1)</sup>
</dformula>
</p>
<p><emph>Example.</emph> 
Another interesting model that has infinite sequence of conservation laws lying on single
orbit of non-Noether symmetry group is Broer-Kaup system 
<citeref rid="r2">[3]</citeref>,<citeref rid="r5">[5]</citeref>,<citeref rid="r6">[6]</citeref>, or more precisely special case
of Broer-Kaup system formed by the following partial differential equations
<dformula id="e29">
u<inf>t</inf> = cu<inf>xx</inf> + 2uu<inf>x</inf><break/>
v<inf>t</inf> = − cv<inf>xx</inf> + 2uv<inf>x</inf> + 2u<inf>x</inf>v
</dformula>
where <formula>u = u(x, t), v = v(x, t)</formula> are again smooth functions on <formula>ℝ<sup>2</sup></formula>
subjected to zero boundary conditions <formula>u(±∞, t) = v(±∞, t) = 0</formula>
Equations <formref rid="e29">(29)</formref> can be rewritten in Hamiltonian form
<dformula>
<fraction><num>d</num><den>dt</den></fraction>f = {h, f} = W(dh ∧ df)
</dformula>
with the Hamiltonian equal to
<dformula>
h = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (u<sup>2</sup>v + cu<inf>x</inf>v)dx
</dformula>
and Poisson bracket defined by
<dformula>
W = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> A ∧ B<inf>x</inf>dx
</dformula>
One can show that the following vector field <formula>E</formula> 
<dformula>
E = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>(u<sup>2</sup>A<inf>x</inf> − cuA<inf>xx</inf> 
+ (uv)<inf>x</inf>B +  3uvB<inf>x</inf> + cvB<inf>xx</inf>)xdx<break/>
= − <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>   [(u<sup>2</sup> + 2cu<inf>x</inf> + x(2uu<inf>x</inf> + cu<inf>xx</inf>))A<break/>
+ (3uv − 2cv<inf>x</inf> + x(2(uv)<inf>x</inf> − cv<inf>xx</inf>))B]dx
</dformula>
has property 
<dformula>
[E,[E,W]] = 0
</dformula>
and thus  group of transformations generated by this vector field transforms centre of Poisson algebra 
formed by functional 
<dformula>
J = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (ku + mv)dx
</dformula>
into involutive orbit that reproduces well known infinite set of conservation laws 
of modified Broer-Kaup hierarchy
<dformula id="e36">
J<sup>(0)</sup> =<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (ku + mv)dx<break/>
J<sup>(1)</sup> = L<inf>E</inf>J<sup>(0)</sup> = m<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> uvdx<break/>
J<sup>(2)</sup> = (L<inf>E</inf>)<sup>2</sup>J<sup>(0)</sup>
 = 2m<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (u<sup>2</sup>v + cu<inf>x</inf>v)dx<break/>
J<sup>(3)</sup> = (L<inf>E</inf>)<sup>3</sup>J<sup>(0)</sup> 
= 3m<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform> (2u<sup>3</sup>v − 3cu<sup>2</sup>v<inf>x</inf> − 2c<sup>2</sup>u<inf>x</inf>v<inf>x</inf>)dx<break/>
J<sup>(m)</sup> = (L<inf>E</inf>)<sup>m</sup>J<sup>(0)</sup> = L<inf>E</inf>J<sup>(m − 1)</sup>
</dformula>
</p>
<p>
Two samples discussed above are representatives of one interesting family of infinite dimensional 
Hamiltonian systems formed by <formula>D</formula> partial differential equations of the following type
<dformula id="e37">
U<inf>t</inf> = − 2FGU<inf>xx</inf> + 〈U , GU<inf>x</inf>〉C 
+ 〈C , GU<inf>x</inf>〉U + 〈C , GU〉U<inf>x</inf><break/>
detG ≠ 0,          G<sup>T</sup> = G,       F<sup>T</sup> = − F<break/>
F<inf>mn</inf>C<inf>k</inf> + F<inf>km</inf>C<inf>n</inf> + F<inf>nk</inf>C<inf>m</inf> = 0
</dformula>
where <formula>U</formula> is vector with components <formula>u<inf>m</inf></formula>
that are smooth functions on <formula>ℝ<sup>2</sup></formula> subjected to zero boundary conditions
<dformula>
u<inf>m</inf> = u<inf>m</inf>(x, t);          u<inf>m</inf>(±∞, t) = 0;          m = 1 ... D
</dformula> 
<formula>G</formula> is constant symmetric nondegenerate matrix, <formula>F</formula> is constant skew-symmetric matrix,
<formula>C</formula> is constants vector that satisfies condition
<dformula>
F<inf>mn</inf>C<inf>k</inf> + F<inf>km</inf>C<inf>n</inf> + F<inf>nk</inf>C<inf>m</inf> = 0
</dformula>
and <formula>〈 · , · 〉</formula> denotes scalar product 
<dformula>
〈X , Y〉 = <subform><roman>∑</roman><bottom>m=1</bottom><top>D</top></subform>X<inf>m</inf>Y<inf>m</inf>.
</dformula>
System of equations <formref rid="e37">(37)</formref> is Hamiltonian with respect to Poisson bivector equal to
<dformula>
W = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>〈A , G<sup>−1</sup>A<inf>x</inf>〉dx
</dformula>
where <formula>A</formula> is vector with components <formula>A<inf>m</inf></formula> that are vector fields defined
for every smooth functional <formula>R(u)</formula> via variational derivatives <formula>A<inf>m</inf>(R) = δR/δu<inf>m</inf></formula>.
Moreover this model is actually bi-Hamiltonian as there exist another invariant Poisson bivector
<dformula>
Ŵ = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>{〈C , A〉〈U , A<inf>x</inf>〉 + 〈A<inf>x</inf> , FA<inf>x</inf>〉}dx
</dformula>
that is compatible with <formula>W</formula> or in other words
<dformula>
[W , W] = [W , Ŵ] = [Ŵ , Ŵ] = 0
</dformula>
Corresponding Hamiltonians that produce Hamiltonian realization 
<dformula>
<fraction><num>d</num><den>dt</den></fraction>U = Ŵ(dĤ ∧ dU) = W(dH ∧ dU)
</dformula>
of the evolution equations <formref rid="e37">(37)</formref> are
<dformula>
Ĥ = ½<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>〈U , GU〉dx
</dformula>
and
<dformula>
H = ½<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>{〈C , GU〉〈U , GU〉 + 2〈FGU<inf>x</inf> , GU〉}dx
</dformula>
The most remarkable property of system <formref rid="e37">(37)</formref> is that it possesses 
set of conservation laws that belong to single orbit obtained from 
centre of Poisson algebra via one-parameter
group of transformations generated by the following vector field
<dformula>
E = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>{〈C , GU〉〈U , A<inf>x</inf>〉 + 〈U , GU〉〈C , A<inf>x</inf>〉<break/>
+ 〈U , GU<inf>x</inf>〉〈C , A〉 +  2〈FGU , A<inf>xx</inf>〉}xdx<break/>
= <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>{〈C , GU〉〈U , A〉 + 〈U , GU〉〈C , A〉 + 4〈FGU<inf>x</inf> , A〉<break/>
+ x (〈C , GU<inf>x</inf>〉〈U , A〉 + 〈C , GU〉〈U<inf>x</inf> , A〉<break/>
+ 〈U , GU<inf>x</inf>〉〈C , A〉 + 2〈FGU<inf>xx</inf> , A〉)}dx
</dformula>
Note that centre of Poisson algebra (with respect to bracket defined by <formula>W</formula>) is formed by
functionals of the following type
<dformula>
J = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>〈K , U〉dx
</dformula>
where <formula>K</formula> is arbitrary constant vector and applying group of transformations generated by <formula>E</formula>
to this functional <formula>J</formula> yields the infinite sequence of functionals 
<dformula>
J<sup>(0)</sup> = <subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>〈K , U〉dx<break/>
J<sup>(1)</sup> = L<inf>E</inf>J<sup>(0)</sup> = ½〈C , K〉<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>〈U , GU〉dx<break/>
J<sup>(2)</sup> = (L<inf>E</inf>)<sup>2</sup>J<sup>(0)</sup> 
= 〈C , K〉<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>{〈C , GU〉〈U , GU〉 + 2〈FGU<inf>x</inf> , GU〉}dx<break/>
J<sup>(3)</sup> = (L<inf>E</inf>)<sup>3</sup>J<sup>(0)</sup> = ¼〈C , K〉<subform><roman>∫</roman><bottom>− ∞</bottom><top>+ ∞</top></subform>{3〈C , GC〉〈U , GU〉<sup>2</sup><break/>
 
+ 12〈C , GU〉<sup>2</sup>〈U , GU〉 + 32〈C , GU〉〈GU , FGU<inf>x</inf>〉<break/>
+ 24〈U , GC〉〈GU , FGU<inf>x</inf>〉 + 48〈FGU<inf>x</inf> , GFGU<inf>x</inf>〉}dx<break/>
J<sup>(m)</sup> = (L<inf>E</inf>)<sup>m</sup>J<sup>(0)</sup> = L<inf>E</inf>J<sup>(m − 1)</sup>
</dformula>
One can check that the vector field <formula>E</formula> satisfies condition
<dformula>
[E , [E , W]] = 0
</dformula>
and according to Theorem 2 the sequence <formula>J<sup>(m)</sup></formula> is involutive. 
So <formula>J<sup>(m)</sup></formula> are conservation laws of bi-Hamiltonian dynamical system <formref rid="e37">(37)</formref>
and vector field <formula>E</formula> is related to non-Noether symmetries of evolutionary equations
(see Remark 1).
</p>
<p>
Note that in special case when <formula>C, F, G, K</formula> have the following form
<dformula>
D = 2,      F<inf>12</inf> = − F<inf>21</inf> = ½c,      C = K = (0 , 1),      G = 1
</dformula>
model <formref rid="e37">(37)</formref> reduces to modified Boussinesq system discussed above.
Another choice of constants <formula>C, F, G, K</formula> 
<dformula>
D = 2,     F<inf>12</inf> = − F<inf>21</inf> = ½c,      C = K = (0 , 1)<break/>
G<inf>12</inf> = G<inf>21</inf> = 1,      G<inf>11</inf> = G<inf>22</inf> = 0
</dformula>
gives rise to Broer-Kaup system described in previous sample.
</p>
<p><emph>Summary.</emph> 
Groups of transformations of Poisson manifold that possess involutive orbits play important
role in some integrable models where  conservation laws form orbit of non-Noether symmetry group. 
Therefore classification of vector fields that generate such a groups would create good  background
for description of remarkable class of integrable system that have interesting geometric origin.
The present paper is an attempt to outline one particular class of vector fields that are
related to non-Noether symmetries of Hamiltonian dynamical systems 
and produce involutive families of conservation laws.
</p>
</chapter>
<back>
<ack>
<p><emph>Acknowledgements.</emph> 
The research described in this publication was made possible in part by
Award No. GEP1-3327-TB-03 of  the Georgian Research and Development Foundation (GRDF) 
and the U.S. Civilian Research &amp; Development Foundation for the 
Independent States of the Former Soviet Union (CRDF).
</p>
</ack>
<biblist>
<head>References</head>
<citation id="r1">
	<title>Non-Noether symmetries and their influence on phase space geometry</title>
	<author>
		<fname>G.</fname>
		<surname>Chavchanidze</surname>
	</author>
	<othinfo>J. Geom. Phys. 48, 190-202</othinfo>
	<date>2003</date>
</citation>
<citation id="r2">
	<title>Factorization of operators. II</title>
	<author>
		<fname>A.</fname>
		<surname>Fordy</surname>
	</author>
	<author>
		<fname> J.</fname>
		<surname>Gibbons</surname>
	</author>
	<othinfo>J. Math. Phys. 22, No. 6, 1170–1175</othinfo>
	<date>1981</date>
</citation>
<citation id="r3">
	<title>Algorithmic Computation of Symmetries, Invariants and Recursion Operators for Systems of Nonlinear Evolution and Differential-Difference Equations</title>
	<author>
		<fname>Ü.</fname>
		<surname>Göktaş</surname>
	</author>
	<othinfo>Ph.D. thesis, Colorado School of Mines, Golden, CO</othinfo>
	<date>1998</date>
</citation>
<citation id="r4">
	<title>A simple way of making a Hamiltonian system into a bi-Hamiltonian one</title>
	<author>
		<fname>A.</fname>
		<surname>Sergyeyev</surname>
	</author>
	<othinfo>Acta Appl. Math. 83, 183-197</othinfo>
	<date>2004</date>
</citation>
<citation id="r5">
	<title>A list of 1 + 1 dimensional integrable equations and their properties</title>
	<author>
		<fname>J.P.</fname>
		<surname>Wang</surname>
	</author>
	<othinfo>J. Nonlinear Math. Phys. 9, suppl. 1, 213–233</othinfo>
	<date>2002</date>
</citation>
<citation id="r6">
	<title>Symmetries and conservation laws of evolution equations</title>
	<author>
		<fname>J.P.</fname>
		<surname>Wang</surname>
	</author>
	<othinfo>Ph.D. Thesis, Vrije Universiteit van Amsterdam</othinfo>
	<date>1998</date>
</citation>
</biblist>
</back>
</article>