<?xml version="1.0" encoding="utf-8"?>
<?xml-stylesheet type="text/css" href="math.css"?>
<!DOCTYPE article [<!ATTLIST article xmlns CDATA "http://xml-maiden.com"><!ATTLIST a xmlns CDATA "http://www.w3.org/1999/xhtml">]>
<article>
<title>Bi-Hamiltonian structure as a shadow of non-Noether symmetry</title>
<author>George Chavchanidze</author>
<affiliation>Department of Theoretical Physics,
A. Razmadze Institute of Mathematics,
1 Aleksidze Street, Tbilisi 0193, Georgia</affiliation>
<abstract>In the present paper correspondence between non-Noether symmetries and bi-Hamiltonian structures
is disscussed. We show that in regular Hamiltonian systems presence of the global bi-Hamiltonian
structure is caused by symmetry of the space of solution. As an example well known bi-Hamiltonian
realisation of Korteweg-De Vries equation is disscussed.</abstract>
<keywords>Bi-Hamiltonian system; Non-Noether symmetry; Non-Cartan symmetry; Korteweg- De Vries equation.</keywords>
<msc> 70H33, 70H06, 53Z05</msc>
<reference>Georgian Math. J. 10 (2003) 057-061</reference>

<paragraph>
Noether theorem, Lutzky's theorem, bi-Hamiltonian formalism and bidifferential calculi are often used 
in generating conservation laws and all
this approaches are unified by the single idea — to construct conserved quantities out of some invariant
geometric object (generator of the symmetry — Hamiltonian vector field in Noether theorem, 
non-Hamiltonian one in Lutzky's approach, closed 2-form in bi-Hamiltonian formalism and auxiliary
differential in case of bidifferential calculi). There is close relationship between later three approaches.
Some aspects of this relationship has been uncovered in <a href="#r3">[3]</a>,<a href="#r4">[4]</a>. In the present paper it is
discussed how bi-Hamiltonian structure can be interpreted as a manifestation of symmetry of space of
solutions. Good candidate for this role is non-Noether symmetry. Such a symmetry is a group of
transformation that maps the space of solutions of equations of motion onto itself, but unlike the
Noether one, does not preserve action. </paragraph>
<paragraph>
In the case of regular Hamiltonian system phase space is equipped with symplectic form <math>ω</math>
(closed <math>dω = 0</math> and nondegenerate <math>i<sub>X</sub>ω = 0 ⇒ X = 0</math> 2-form) and time
evolution is governed by Hamilton's equation
<formula xml:id="e1">
i<sub>X<sub>h</sub></sub>ω + dh = 0
</formula>
where <math>X<sub>h</sub></math> is Hamiltonian vector field that defines time evolution
<formula>
<fraction><num>df</num><den>dt</den></fraction> = X<sub>h</sub>(f)
</formula>
 for any function <math>f</math> and <math>i<sub>X<sub>h</sub></sub>ω</math> denotes contraction of
<math>X<sub>h</sub></math> and <math>ω</math>. Vector field is said to be (locally) Hamiltonian if it preserves <math>ω</math>.
According to the Liouville's theorem <math>X<sub>h</sub></math> defined by <a href="#e1">(1)</a> automatically preserves <math>ω</math> 
due to relation
<formula>
L<sub>X<sub>h</sub></sub>ω = di<sub>X<sub>h</sub></sub>ω + i<sub>X<sub>h</sub></sub>dω = − ddh = 0
</formula>
</paragraph>
<paragraph>
One can show that group of transformations of phase space generated by any non-Hamiltonian vector
field <math>E</math>
<formula>
g(a) = e<sup>aL<sub>E</sub></sup>
</formula>
does not preserve action
<formula>
g<sub>*</sub>(A) = g<sub>*</sub>(<ope>∫</ope> pdq − hdt) = <ope>∫</ope> g<sub>*</sub>(pdq − hdt) ≠ 0
</formula>
because <math>d(L<sub>E</sub>(pdq − hdt)) = L<sub>E</sub>ω −  dE(h) ∧ dt ≠ 0</math> (first term in r.h.s. does not vanish
since <math>E</math> is non-Hamiltonian and as far as <math>E</math> is time independent <math>L<sub>E</sub>ω</math> and 
<math>dE(h)  ∧ dt</math> are linearly independent 2-forms). As a result every non-Hamiltonian vector field <math>E</math>
commuting with <math>X<sub>h</sub></math> leads to the non-Noether symmetry (since <math>E</math> preserves vector field tangent
to solutions <math>L<sub>E</sub>(X<sub>h</sub>) = [E , X<sub>h</sub>] = 0</math> it maps the space of solutions onto itself). Any such
symmetry yields the following integrals of motion <a href="#r1">[1]</a>,<a href="#r2">[2]</a>,<a href="#r4">[4]</a>,<a href="#r5">[5]</a>
<formula>
I<sup>(k)</sup> = Tr(R<sup>k</sup>)         k = 1,2 ... n
</formula>
where <math>R = ω<sup>−1</sup>L<sub>E</sub>ω</math> and <math>n</math> is half-dimension of phase space.
</paragraph>
<paragraph>
It is interesting that for any non-Noether symmetry, triple <math>(h, ω, ω<sub>E</sub>)</math> carries 
bi-Hamiltonian structure (§4.12 in <a href="#r6">[6]</a>,<a href="#r7">[7]</a>-<a href="#r9">[9]</a>). 
Indeed <math>ω<sub>E</sub></math> is closed 
(<math>dω<sub>E</sub> = dL<sub>E</sub>ω = L<sub>E</sub>dω = 0</math>) and invariant 
(<math>L<sub>X<sub>h</sub></sub>ω<sub>E</sub> = L<sub>X<sub>h</sub></sub>L<sub>E</sub>ω = L<sub>E</sub>L<sub>X<sub>h</sub></sub>ω = 0</math>) 
2-form (but generic <math>ω<sub>E</sub></math> is degenerate). So every non-Noether
symmetry quite naturally endows dynamical system with bi-Hamiltonian structure. 
</paragraph>
<paragraph>
Now let's discuss how non-Noether symmetry can be recovered from bi-Hamiltonian system. Generic 
bi-Hamiltonian structure on phase space consists of Hamiltonian system <math>h, ω</math> and auxiliary
closed 2- form <math>ω<sup>∗</sup></math> satisfying <math>L<sub>X<sub>h</sub></sub>ω<sup>∗</sup> = 0</math>. Let us call it global 
bi-Hamiltonian structure whenever <math>ω<sup>∗</sup></math> is exact (there exists 1-form <math>θ<sup>∗</sup></math> such that
<math>ω<sup>∗</sup> = dθ<sup>∗</sup></math>) and <math>X<sub>h</sub></math> is (globally) Hamiltonian vector field with respect to
<math>ω<sup>∗</sup></math> (<math>i<sub>X<sub>h</sub></sub>ω<sup>∗</sup> + dh<sup>∗</sup> = 0</math>). 
As far as <math>ω</math> is nondegenerate there exists vector field 
<math>E<sup>∗</sup></math> such that 
<math>i<sub>E<sup>∗</sup></sub>ω = θ<sup>∗</sup></math>. 
By construction
<formula>
L<sub>E<sup>∗</sup></sub>ω = ω<sup>∗</sup>
</formula>
Indeed 
<formula>
L<sub>E<sup>∗</sup></sub>ω = di<sub>E<sup>∗</sup></sub>ω + i<sub>E<sup>∗</sup></sub>dω
= dθ<sup>∗</sup> = ω<sup>∗</sup></formula>
</paragraph>
<paragraph>
And
<formula>
i<sub>[E<sup>∗</sup>,X<sub>h</sub>]</sub>ω = 
L<sub>E<sup>∗</sup></sub>(i<sub>X<sub>h</sub></sub>ω)  −  i<sub>X<sub>h</sub></sub>L<sub>E<sup>∗</sup></sub>ω
=  −  d(E<sup>∗</sup>(h)  −  h<sup>∗</sup>) =  −  dh'
</formula>
In other words <math>[X<sub>h</sub> , E<sup>∗</sup>]</math> is Hamiltonian vector field, i. e., <math>[X<sub>h</sub> , E] = X<sub>h'</sub></math>. So
<math>E<sup>∗</sup></math> is not generator of symmetry since it does not commute with <math>X<sub>h</sub></math> but one can
construct (locally) Hamiltonian counterpart of <math>E<sup>∗</sup></math> (note that <math>E<sup>∗</sup></math> itself is 
non-Hamiltonian) —  <math>X<sub>g</sub></math> with 
<formula xml:id="e10">
g(z) =<under><over><sat>t</sat><ope>∫</ope></over><sat>0</sat></under> h'dτ
</formula>
Here integration along solution of Hamilton's equation, with fixed origin and end point in <math>z(t) = z</math>,
is assumed. Note that <a href="#e10">(10)</a> defines <math>g(z)</math> only locally and, as a result, <math>X<sub>g</sub></math> is a locally
Hamiltonian vector field, satisfying, by construction, the same commutation relations as 
<math>E<sup>∗</sup></math> (namely <math>[X<sub>h</sub> , X<sub>g</sub>] = X<sub>h'</sub></math>). 
Finally one recovers generator of non-Noether symmetry — non-Hamiltonian vector field 
<math>E = E<sup>∗</sup>  −  X<sub>g</sub></math> commuting with <math>X<sub>h</sub></math> and satisfying
<formula>
L<sub>E</sub>ω = L<sub>E<sup>∗</sup></sub>ω −  L<sub>X<sub>g</sub></sub>ω = L<sub>E<sup>∗</sup></sub>ω = ω<sup>∗</sup>
</formula>
(thanks to Liouville's theorem <math>L<sub>X<sub>g</sub></sub>ω = 0</math>). So in case of regular Hamiltonian system every
global bi-Hamiltonian structure is naturally associated with (non-Noether) symmetry of space of
solutions.</paragraph>
<example>
 As a toy example one can consider free particle
<formula>
h = ½ <under><ope>∑</ope><sat>m</sat></under> p<sub>m</sub><sup>2</sup>
       ω = <under><ope>∑</ope><sat>m</sat></under> dp<sub>m</sub> ∧ dq<sub>m</sub>
</formula>
this Hamiltonian system can be extended to the bi-Hamiltonian one
<formula>
h, ω, ω<sup>∗</sup> = <under><ope>∑</ope><sat>m</sat></under> p<sub>m</sub>dp<sub>m</sub> ∧ dq<sub>m</sub>
</formula>
clearly <math>dω<sup>∗</sup> = 0</math> and <math>X<sub>h</sub></math> preserves 
<math>ω<sup>∗</sup></math>. Conserved quantities <math>p<sub>m</sub></math> are associated with this simple 
bi-Hamiltonian structure.
This system can be obtained from the following (non-Noether) symmetry (infinitesimal form)
<formula>
q<sub>m</sub>        →        (1 + ap<sub>m</sub>)q<sub>m</sub><line />
p<sub>m</sub>         →        (1 + ap<sub>m</sub>)p<sub>m</sub>
</formula>
</example>
<example>
The earliest and probably the most well known bi-Hamiltonian structure is the one
discovered by F. Magri and assosiated with Korteweg- De Vries integrable hierarchy. The KdV equation
<formula>
u<sub>t</sub> + u<sub>xxx</sub> + uu<sub>x</sub> = 0
</formula>
(zero boundary conditions for <math>u</math> and its derivatives are assumed) appears to be Hamilton's equation
<formula>
i<sub>X<sub>h</sub></sub>ω+ dh = 0
</formula>
where 
<formula>
X<sub>h</sub> =  <under><over><sat>+ ∞</sat><ope>∫</ope></over><sat>− ∞</sat></under> dx u<sub>t</sub><fraction><num>δ</num><den>δu</den></fraction></formula> 
(here <math><fraction><num>δ</num><den>δu</den></fraction></math> 
denotes variational derivative with respect to the field <math>u(x)</math>) is the vector field tangent to the
solutions,
<formula>
ω =  <under><over><sat>+ ∞</sat><ope>∫</ope></over><sat>− ∞</sat></under> dx du ∧ dv
</formula>
is the symplectic form (here <math>v</math> is defined by <math>v<sub>x</sub> = u</math>) and the function
<formula>
h =  <under><over><sat>+ ∞</sat><ope>∫</ope></over><sat>− ∞</sat></under> dx (<fraction><num>u<sup>3</sup></num><den>3</den></fraction>  −  u<sub>x</sub><sup>2</sup>)
</formula>
plays the role of Hamiltonian. This dynamical system possesses non-trivial symmetry — one-parameter
group of non-cannonical transformations <math>g(a) = e<sup>L<sub>E</sub></sup></math> generated by the non-Hamiltonian vector
field
<formula>
E =  <under><over><sat>+ ∞</sat><ope>∫</ope></over><sat>− ∞</sat></under> dx (u<sub>xx</sub> + <fraction><num>u<sup>2</sup></num><den>2</den></fraction>)<fraction><num>∂</num><den>∂u</den></fraction> + X<sub>F</sub>
</formula>
here first term represents non-Hamiltonian part of the generator of the symmetry, while the second one
is its Hamiltonian counterpart assosiated with
<formula>
F =  <under><over><sat>+ ∞</sat><ope>∫</ope></over><sat>− ∞</sat></under>(<fraction><num>u<sup>2</sup>v</num><den>12</den></fraction> + <fraction><num>G</num><den>4</den></fraction> + <fraction><num>3vI⁽<sup>2</sup>⁾</num><den>4I⁽<sup>3</sup>⁾</den></fraction>)dx
</formula>
(<math>I<sup>(2,3)</sup></math> are defined in <a href="#e22">(22)</a>, while 
<math>G</math> is defined by <math>G<sub>x</sub> = <fraction><num>u<sup>3</sup></num><den>3</den></fraction>  −  u<sub>x</sub><sup>2</sup></math> . 
The physical origin of this symmetry is unclear, however the
symmetry seems to be very important since it leads to the celebrated infinite sequence of conservation
laws in involution:
<formula xml:id="e22">
I<sup>(1)</sup> =  <under><over><sat>+ ∞</sat><ope>∫</ope></over><sat>− ∞</sat></under> u dx<line />
I<sup>(2)</sup> =  <under><over><sat>+ ∞</sat><ope>∫</ope></over><sat>− ∞</sat></under> u<sup>2</sup> dx<line />
I<sup>(3)</sup> =  <under><over><sat>+ ∞</sat><ope>∫</ope></over><sat>− ∞</sat></under> (<fraction><num>u<sup>3</sup></num><den>3</den></fraction>  −  u<sub>x</sub><sup>2</sup>) dx<line />
I<sup>(4)</sup> =  <under><over><sat>+ ∞</sat><ope>∫</ope></over><sat>− ∞</sat></under> (<fraction><num>5</num><den>36</den></fraction>u<sup>4</sup>  −  <fraction><num>5</num><den>3</den></fraction>uu<sub>x</sub><sup>2</sup> 
+ u<sub>xx</sub><sup>2</sup>) dx<line />
⋯
</formula>
and ensures integrability of KdV equation. Second Hamiltonian realization of KdV equation discovered
by F. Magri <a href="#r7">[7]</a>
<formula>
i<sub>X<sub>h<sup>∗</sup></sub></sub>ω<sup>∗</sup> + dh<sup>∗</sup> = 0
</formula>
(where <math>ω<sup>∗</sup> = L<sub>E</sub>ω</math> and <math>h<sup>∗</sup> = L<sub>E</sub>h</math>) is a result of 
invariance of KdV under aforementioned transformations <math>g(a)</math>.
</example>
<acknowledgements>
 Author is grateful to Z. Giunashvili for constructive discussions and to G.
Jorjadze for support. This work was supported by INTAS (00-00561) and Scholarship from World
Federation of Scientists.
</acknowledgements>



<references>
<cite xml:id="r1">
	<who>F. González-Gascón</who>
	<what>Geometric foundations of a new conservation law discovered by Hojman</what>
	<where>J. Phys. A: Math. Gen. 27 L59-60</where>
	<when>1994</when>
</cite>
<cite xml:id="r2">
	<who>M. Lutzky</who>
	<what>New derivation of a conserved quantity for Lagrangian systems</what>
	<where>J. Phys. A: Math. Gen. 15 L721-722</where>
	<when>1998</when>
</cite>
<cite xml:id="r3">
	<who>M. Crampin, W. Sarlet, G. Thompson</who>
	<what>Bi-differential calculi and bi-Hamiltonian systems</what>
	<where>J. Phys. A: Math. Gen. 33 No. 22 L177-180</where>
	<when>2000</when>
</cite>
<cite xml:id="r4">
	<who>P. Guha</who>
	<what>A Note on Bidifferential Calculi and Bihamiltonian systems</what>
	<where>IHÉS preprint M/64</where>
	<when>2000</when>
</cite>
<cite xml:id="r5">
	<who>G. Chavchanidze</who>
	<what>Non-Noether symmetries in singular dynamical systems</what>
	<where>Georgian Math. J. 8 (2001) 027-032</where>
	<when>2001</when>
</cite>
<cite xml:id="r6">
	<who>N.M.J. Woodhouse</who>
	<what>Geometric Quantization</what>
	<where>Claredon, Oxford</where>
	<when>1992</when>
</cite>
<cite xml:id="r7">
	<who>F. Magri</who>
	<what>A simple model of the integrable Hamiltonian equation</what>
	<where>J. Math. Phys. 19 no.5, 1156-1162</where>
	<when>1978</when>
</cite>
<cite xml:id="r8">
	<who>A. Das</who>
	<what>Integrable models</what>
	<where>World Scientific Lecture Notes in Physics, vol. 30</where>
	<when>1989</when>
</cite>
<cite xml:id="r9">
	<who>R. Brouzet</who>
	<what>Sur quelques propriétés géométriques des variétés bihamiltoniennes</what>
	<where>C. R. Acad. Sci. Paris 308, série I, 287-92</where>
	<when>1989</when>
</cite>
</references>
</article>