See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/266137900

The Generalized Weierstrass System and an Application to the Study of

Deformations of Surfaces by Means of Integrable Hierarchies

Article in Dynamical Systems An International Journal · June 2009

DOI: 10.1080/14689360802609351

CITATIONS READS

0 26

1 author:

Paul Bracken
University of Texas Rio Grande Valley
175 PUBLICATIONS 737 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Dynamical Systems View project

All content following this page was uploaded by Paul Bracken on 26 September 2014.

The user has requested enhancement of the downloaded file.

This article was downloaded by: [Bracken, Paul]
On: 27 April 2009
Access details: Access Details: [subscription number 910734460]
Publisher Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Dynamical Systems
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713414890

The generalized Weierstrass system and an application to the study of

deformations of surfaces by means of integrable hierarchies
Paul Bracken a
a
Department of Mathematics, University of Texas, Edinburg, Tx, USA

Online Publication Date: 01 June 2009

To cite this Article Bracken, Paul(2009)'The generalized Weierstrass system and an application to the study of deformations of
surfaces by means of integrable hierarchies',Dynamical Systems,24:2,223 — 235
To link to this Article: DOI: 10.1080/14689360802609351
URL: http://dx.doi.org/10.1080/14689360802609351

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
 Dynamical Systems
Vol. 24, No. 2, June 2009, 223–235

The generalized Weierstrass system and an application to the study

of deformations of surfaces by means of integrable hierarchies
Paul Bracken*

Department of Mathematics, University of Texas, Edinburg, Tx, USA

(Received 11 July 2008; final version received 23 October 2008)

Generalized Weierstrass representations can be used to investigate deformations

of surfaces under the action of integrable hierarchies. It will be shown that the
generalized Weierstrass representation of Konopelchenko leads to a linear system
in a single independent space variable in the case of surfaces of revolution.
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

This result is combined with a second linear problem with an unknown second
matrix in terms of a parameter. It is then shown how an integrable hierarchy can
be obtained. Solutions to this hierarchy generate deformations which preserve the
surface-inducing means of generalized Weierstrass representation.
Keywords: integrable hierarchy; surface of revolution; deformation of surfaces
PACS: 02.30.Ik, 05.45.-a, 65.35.Ja

1. Introduction
Equations that lead to minimal surfaces embedded in three-dimensional Euclidean space
were first formulated by Enneper and Weierstrass [1,2] in the nineteenth century. More
recently, this kind of idea has been substantially generalized by Konopelchenko [3].
He established a connection between certain classes of constant mean curvature surfaces
and the trajectories of an infinite-dimensional Hamiltonian system. Konoplechenko
and Taimanov [4] considered a non-linear Dirac system which consists of partial
differential equations which are given in terms of two complex-valued functions 1 and 2
written as

@ 1 ¼ p 2, @ 2 ¼ p 1
ð1:1Þ
@  1 ¼ p  2 , @ 2 ¼ p  1 :


 is a real-valued function which is referred to as the potential function.

Here, pðz, zÞ
The derivatives are abbreviated to read @ ¼ @/@z and @ ¼ @=@z,  and bar denoted complex
conjugation. For the case of minimal surfaces p ¼ 0, and for the case of constant mean
curvature surfaces, p is given by p ¼ H(j 1j2 þ j 2j2), where H is a constant in this instance.
A solution ( 1, 2) to (1.1) induces a surface by means of the integrals
ð
X1 þ iX2 ¼ 2i ð  21 dz   22 dzÞ,



*Email: [email protected]

ISSN 1468–9367 print/ISSN 1468–9375 online

 2009 Taylor & Francis
DOI: 10.1080/14689360802609351
http://www.informaworld.com
 224 P. Bracken
ð
2 2
X1  iX2 ¼ 2i ð 2 dz  1 
dzÞ, ð1:2Þ


ð
X3 ¼ 2 ð 1 2 dz þ 1
 2 dzÞ:



By virtue of (1.1), the integrals in (1.2) do not depend on the choice of the curve  of
integration, which starts at z0 and ends at z. One then treats z and z as local parameters for
a surface such that (X1, X2, X3) are coordinates of its immersion in R3. Substitution of
( 1, 2) into (1.2) yields the explicit coordinates of a surface in R3 by means of the
solutions to system (1.1). The first fundamental form is
2 2
 ¼ 4ðj 1j þj 2j 
Þdz dz, ð1:3Þ

and the Gaussian and mean curvatures are, respectively,

@@ logðj 1 j2 þ j 2 j2 Þ 
pðz, zÞ
K¼ , H¼ : ð1:4Þ
ðj 1 j2 þ j 2 j2 Þ2 2
j 2 j2
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

j 1 j þ
It has also been shown that (1.1) is an integrable system [5] and by introducing a new
complex variable  ¼ 1 =  2 , it can be shown that solutions of (1.1) are related to solutions
of the integrable two-dimensional non-linear Euclidean sigma model and conversely [6].
The symmetries of these two systems have also been studied [7] and a Bäcklund
transformation has been determined [8]. Konopelchenko and Landolfi [9,10] have
discussed how this formalism can be extended to higher dimensional spaces.
Surfaces and integrable systems have been discussed in [11–13] recently. The primary
objective of these authors is the problem of the immersion of a two-dimensional surface
into a three-dimensional Euclidean space, as well as the n-dimensional generalization. It is
shown that this is related to the study of Lie groups and surfaces in Lie algebras. It is
shown in [11] that all compact surfaces generated by the infinite number of symmetries of
the sine-Gordon equation are homeomorphic to a sphere. Under the rigid SU(2) rotations
all integrable equations are mapped to a sphere. A mapping from each symmetry of
integrable equation to surface in R3 can be established. In contrast, what is done here is
different. A generalized Weierstrass representation is used to produce a linear problem
based on surfaces of revolution. When this is combined with a second linear problem in
a different parameter, an integrable hierarchy is obtained.
One of the reasons for the interest in this type of representation is that due to the linear
character of (1.1), it allows the description of integrable deformations of induced surfaces
in a straightforward manner. To consider integrable deformations suppose that the
functions p, 1 and 2 in (1.1) also depend on an additional real parameter t. Consider
then those deformations of ( 1, 2) for which differential operators A, B, C and D exist,
such that the t evolution is given by

1,t ¼A 1 þB 2, 2,t ¼C 1 þD 2: ð1:5Þ

This system in (1.5) can be written in the form of a single matrix operator M in the
following way
   
1 1
¼M : ð1:6Þ
2 t 2
 Dynamical Systems 225

Here, M is a differential 2  2 matrix differential operator. The requirement that (1.1)

and (1.6) be compatible, guarantees the preservation of the generalized Weierstrass
inducing in terms of t. Thus, as t is varied, a deformation of the surface will result. In
applications in which the surface is meant to model a string or an interface between phases
or a shock front, such a parameter dependence can be used to model the evolution of the
given object. This procedure will lead to non-linear evolution equations for the function p.
By varying the operators A, B, C and D an infinite hierarchy of integrable equations for
the function p can be generated. For example, choosing A, B, C and D as first-order
differential operators, a linear equation for p would result.
One of the remarkable properties of equations which are integrable by the inverse
scattering method is that they are grouped into hierarchies of such equations which are
defined recursively [14]. For example, the KdV equation and its modified version, the
mKdV equation, are only the first members of hierarchies of equations which can be
written collectively in the form

t ¼ N2kþ1 ðÞ,
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

where the N2kþ1 are a sequence of non-linear operators. These equations can be
represented by pairs of Lax operators, which are referred to as (L, A)-pairs, such that one
of the operators pertains to the KdV or mKdV hierarchy, respectively.
The Novikov–Veselov equation and the modified Novikov–Veselov equation can be
placed together with associated A operators into hierarchies attached to particular
L-operators. These are of course two-dimensional generalizations of the KdV and mKdV
systems, respectively. System (1.1) is two-dimensional and so it is appropriate for one of
the operators in the (L, A)-pair. System (1.1) can be written in the form of the following
linear problem
    
@ p 1 0
L ¼ ¼ : ð1:7Þ
p @ 2 0
However, it is important to realize that (2 þ 1) equations, and the Novikov–Veselov
equation among them, do not admit a simple representation in terms of a local operator.
Some work exists in this direction, and a method based on bilocal operators [15], but its
realization is more difficult than the case of (1 þ 1) equations.
The aim of this work is to consider the representation of surfaces of revolution by
means of representation (1.1) [16]. It will be seen that this will effectively lead to
a reduction in dimension from two to one in terms of the space variables which appear in
the equations. It will yield an operator which can be used to investigate (1 þ 1) equations.
Such an operator can be used as one of the elements of an (L, A)-pair. It will be shown how
the corresponding element of the pair can be generated based on this postulated matrix in
a novel way. The general equations from the compatibility condition will be derived and
one general solution for the elements of the matrix will be given. Finally, some general
results implied by these facts with regard to these surfaces will be discussed [17].

2. Surfaces of revolution
Let  be a surface of revolution in three-dimensional Euclidean space which has
coordinates (z1, z2, z3). Here,  will be generated by revolving a curve in the z2 ¼ 0 plane
 226 P. Bracken

about the z3-axis. The surface can then be parametrized in terms of two parameters which
we will call x and y as follows

z ¼ ðz1 , z2 , z3 Þ ¼ ððxÞ cos y, ðxÞ sin y, ’ðxÞÞ: ð2:1Þ

By calculating the first derivatives of z in (2.1), the first fundamental form is obtained

I ¼ ðx2 þ ’2x Þ dx2 þ ðxÞ2 dy2 : ð2:2Þ

A normal vector to the surface is given by zx  zy ¼ (’x cos y, ’x sin y, ’x).
This determines a unit normal vector to the surface
1
N ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð’x cos y,  ’x sin y, x Þ: ð2:3Þ
x þ ’2x
2

Using N, the second fundamental form is determined to be

 
’x xx x ’xx
II ¼  þ dx2 þ ’x dy2 : ð2:4Þ
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

 

Theorem 1: A surface of revolution can be obtained from the generalized Weierstrass

representation (1.1) by taking the components of  ¼ ( 1, 2) to have the form

1 ¼ rðxÞeiy=2 , 2 ¼ sðxÞeiy=2 , ð2:5Þ

with the corresponding potential function p ¼ p(x) a function of x. System (1.1) reduces to the
coupled pair of first-order equations

1 1
rx ¼  r þ 2ps, sx ¼ s  2pr, ð2:6Þ
2 2
for the functions (r, s).
Proof The derivatives in terms of x and y can be written in terms of the variables x and y
as @ ¼ 12 ð@x  i@y Þ and @ ¼ 12 ð@x þ i@y Þ. Therefore,

1 r 1 s
@ 1 ¼ rx eiy=2 þ eiy=2 , @ 2 ¼ sx eiy=2  eiy=2 :
2 4 2 4
Substituting these derivatives into (1.1), the pair of (2.6) follows.
System (2.6) can also be written in the form of the following linear matrix problem

L ¼ 0, ð2:7Þ

where the operator L is defined to be

   
1 1 4p 1 1 q
L ¼ @x I  ¼ @x I  : ð2:8Þ
2 4p 1 2 q 1

Here, the new function q ¼ 4p has been introduced, and (2.8) amounts to writing (2.6) in
matrix form. A surface is represented by (2.7), (2.8) and (1.2).
 Dynamical Systems 227

If we multiply the first equation in (2.6) by sx and the second by rx and subtract the
resulting pair, we obtain

1
ðsx r þ srx Þ  2pðssx þ rrx Þ ¼ 0:
2
Integrating this, it is found that r and s must satisfy

sr ¼ 2pðs2 þ r2 Þ þ C, ð2:9Þ

where C is a constant of integration. Thus, (2.9) takes the form

1 C
s2  rs þ r2 þ ¼ 0: ð2:10Þ
2p 2p

This determines one element of the pair (r, s) in terms of the other. Solving for s and
substituting into (2.6 i), a differential equation for r is obtained
 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

rx ¼ ð1  16p2 Þr2  8pC,  ¼ 1: ð2:11Þ

2
If we put C ¼ 0, then (2.11) is separable and r can be obtained explicitly in the form
of an integral
 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi
ðlog rÞx ¼ 1  16p : ð2:12Þ
2
The profile curve can be found by putting y ¼ 0 into (2.1) to give (x) ¼ ((x), 0, ’(x)),
and the functions  and ’ are related to the components of the pair ( 1, 2) in (2.5)
through the relations

x2 ¼ s2  r2 , ’x ¼ 2rs: ð2:13Þ

œ
Lemma 1: Subject to system (2.6) and (2.13), the functions (r, s) satisfy the following
six equations

ðs2 þ r2 Þx ¼ x , ’2x ¼ 2  x2 ,

xx ¼ s2 þ r2  q’x , ’xx ¼ qx ,

1 1
s2 ¼ ð þ x Þ, r2 ¼ ð  x Þ:
2 2
These can be shown by differentiating and using (2.6) and (2.13). In particular, the
expression ’2x ¼ 2  x2 implies that xx ¼ (x  ’x’xx)/x and allows us to write the first
and second fundamental forms as follows
 

I ¼ ðxÞ2 ðdx2 þ dy2 Þ, II ¼ ’xx  ’x dx2 þ ’x dy2 : ð2:14Þ
x
 228 P. Bracken

The Gaussian and mean curvatures can be determined from (2.14) using these results and
they are

x2  xx ’xx

K¼ , H¼ : ð2:15Þ
4 2x
If (x) is periodic, then (x) ¼ (x þ T ), and the surface of revolution is closed if and only if
ð T qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2  x2 dx ¼ 0: ð2:16Þ
0

This is what is required with regard to surfaces of revolution considered here.

3. The linear problems and generation of hierarchies

The two linear problems which will be used here are now formulated. The first is the
matrix which relates to the generation of surfaces of revolution obtained from (1.1).
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

A spectral parameter can be written along the main diagonal of the matrix in (2.8) to give
the related linear problem
     
1  q 1 0
@x  ¼ : ð3:1Þ
2 q  2 0

Now let us write a linear system which can be regarded as a sequence of linear systems
which are indexed by n in terms of a parameter tn given by
" ! #   
1 AðnÞ BðnÞ 1 0
@ tn  ¼ : ð3:2Þ
2 CðnÞ
A ðnÞ
2 0

The Lax matrix in (3.2) is a standard form and will be denoted Qn(), since the elements
will come to depend on the spectral parameter in the development. For convenience,
we will often denote tn simply by t. The compatibility condition in matrix form is
 
1
LðÞ, @t  Qn ðÞ ¼ 0: ð3:3Þ
2

Consider the derivatives of 1 first. Differentiating, we obtain

   
 ðnÞ q ðnÞ 1 1 ðnÞ q ðnÞ
@t @x 1 ¼ A þ C 1 þ q t þ B  A 2,
4 4 2 4 4

   
1 ðnÞ 1 ðnÞ 1 ðnÞ q ðnÞ 1 
@x @t 1 ¼ @x A þ A  qB 1þ A þ @x BðnÞ  BðnÞ 2:
2 4 4 4 2 4

Equating the coefficients of 1 and 2, the following pair of equations result

1
@x AðnÞ ¼ qðBðnÞ þ CðnÞ Þ, qt ¼ @x BðnÞ  BðnÞ þ qAðnÞ : ð3:4Þ
2
 Dynamical Systems 229

Next, consider the derivatives of 2, which are

   
1 q  q ðnÞ  ðnÞ
@t @x 2 ¼  qt  AðnÞ  CðnÞ 1 þ  B þ A 2,
2 4 4 4 4
   
1  q q ðnÞ 1 
@x @t 2 ¼ @x CðnÞ þ CðnÞ þ AðnÞ 1þ C  @x AðnÞ þ AðnÞ 2:
2 4 4 4 2 4
Equating coefficients of the i on both sides of these two expressions yields the following
pair of equations
1
@x AðnÞ ¼ qðBðnÞ þ CðnÞ Þ, qt ¼ @x CðnÞ  CðnÞ  qAðnÞ : ð3:5Þ
2
This completes the proof of the following theorem.
Theorem 2: Systems (3.1) and (3.2) will satisfy zero-curvature condition (3.3) provided the
functions in the corresponding matrices satisfy the following set of equations
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

q
@x AðnÞ ¼ ðBðnÞ þ CðnÞ Þ,
2

qt ¼ @x BðnÞ  BðnÞ þ qAðnÞ , ð3:6Þ

qt ¼ @x CðnÞ  CðnÞ  qAðnÞ :

Now we want to look for solutions to system (3.6). In particular, consider solutions in
the form of ascending series in powers of . It will prove useful to define the following
series expansions in powers of 
X
n X
n1 X
n
AðnÞ ¼ AðnÞ
2kþ1 
2kþ1
, BðnÞ þ CðnÞ ¼ 2 SðnÞ
2kþ1 
2kþ1
, BðnÞ  CðnÞ ¼ 2 TðnÞ 2k
2k  :
k¼0 k¼0 k¼0
ð3:7Þ
Let us now make use of the series defined in (3.7) by substituting into (3.6). It will be
required that the coefficients of  match each other on both sides of each respective
equation. Substituting (3.7(i)) and (3.7(ii)) into (3.6(i)), we obtain
X
n X
n1
@x AðnÞ
2kþ1 
2kþ1
¼q SðnÞ
2kþ1 
2kþ1
: ð3:8Þ
k¼0 k¼0

The power 2nþ1 can be matched by taking AðnÞ 2nþ1 to be constant with respect to x and
in particular, normalized so that AðnÞ
2nþ1 ¼ 1. If we equate the remaining coefficients of the
powers of  on both sides of (3.8), there results

@x AðnÞ ðnÞ
2kþ1 ¼ qS2kþ1 ,
ð3:9Þ

where k ¼ 0, . . . , n  1. Now, adding (3.6(ii)) and (3.6(iii)), we obtain that

1 1
qt ¼ @x ðBðnÞ  CðnÞ Þ  ðBðnÞ þ CðnÞ Þ: ð3:10Þ
2 2
 230 P. Bracken

Substituting the series (3.7) into (3.10), we obtain that

X
n
qt ¼ @x TðnÞ
0 þ ð@x TðnÞ ðnÞ 2k
2k  S2k1 Þ : ð3:11Þ
k¼1

The coefficients of powers of  will vanish provided that

@x TðnÞ ðnÞ
2k ¼ S2k1 :
ð3:12Þ

This means that (3.11) reduces to the following hierarchy of equations

qt ¼ @x TðnÞ
0 :
ð3:13Þ

This will lead to the n-th equation in the hierarchy once a specific form for TðnÞ
0
has been specified.
If we now subtract the second and third equations of (3.6), we find that
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

@x ðBðnÞ þ CðnÞ Þ  ðBðnÞ  CðnÞ Þ þ 2qAðnÞ ¼ 0:

Using (3.7), this takes the form

X
n1
ð@x SðnÞ ðnÞ ðnÞ
2kþ1  T2k þ qA2kþ1 Þ
2kþ1
 ðTðnÞ ðnÞ
2n  qA2nþ1 Þ
2nþ1
¼ 0: ð3:14Þ
k¼0

Since we have taken AðnÞ

2nþ1 ¼ 1, the coefficient of 
2nþ1
in (3.14) implies that

TðnÞ
2n ¼ q:
ð3:15Þ

The remaining terms of the powers of  will vanish provided that

@x SðnÞ ðnÞ ðnÞ

2kþ1  T2k þ qA2kþ1 ¼ 0:
ð3:16Þ

This finishes the proof of the following theorem.

Theorem 3: The expression (3.7) for the elements A(n), B(n) and C(n) will be solutions to (3.6)
provided that the coefficient functions AðnÞ ðnÞ ðnÞ
2kþ1 , T2kþ1 and S2kþ1 satisfy the following system
of equations

AðnÞ
2nþ1 ¼ 1, TðnÞ
2n ¼ q,
ð3:17Þ

and the set

@x AðnÞ ðnÞ
2kþ1 ¼ qS2kþ1 ,
ð3:18Þ

@x TðnÞ ðnÞ
2k ¼ S2k1 ,
ð3:19Þ

@x SðnÞ ðnÞ ðnÞ

2kþ1  T2k þ qA2kþ1 ¼ 0:
ð3:20Þ

The choice for the constant AðnÞ

2nþ1 is of course not unique. Let us show that a formal
solution to the system (3.17)–(3.20) can be developed. At this point, the operator @1x
 Dynamical Systems 231

is introduced which satisfies @1 1 1

x @x ¼ @x @x ¼ I. In terms of the operator @x , we can write
ðnÞ ðnÞ
solutions for A2kþ1 and T2kþ1 based on (3.18) and (3.19) as follows

AðnÞ 1 ðnÞ
2kþ1 ¼ @x ðqS2kþ1 Þ, TðnÞ 1 ðnÞ
2k ¼ @x S2k1 :
ð3:21Þ

Substituting (3.21) into (3.20), we find that the coefficients of S(n) satisfy

@x SðnÞ 1 ðnÞ 1 ðnÞ

2kþ1  @x S2k1 þ q@x ðqS2kþ1 Þ ¼ 0:
ð3:22Þ

Moving the second term of (3.22) to the right-hand side and differentiating both sides of
the result with respect to @x, we have
ðnÞ ðnÞ
ð@2x þ q2 þ qx @1
x qÞS2kþ1 ¼ S2k1 : ð3:23Þ

Define the operator D which appears in (3.23) to be

D ¼ @2x þ q2 þ qx @1
x q: ð3:24Þ
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

In terms of the operator D, (3.23) takes the form

DSðnÞ ðnÞ
2kþ1 ¼ S2k1 , SðnÞ 1 ðnÞ
2kþ1 ¼ D S2k1 :
ð3:25Þ

In (3.25), the operator D1 is the formal inverse of D. The operators D and D1 can be
interpreted as lowering or raising operators, as the case may be, with respect to the k-index
of SðnÞ
2kþ1 . Using the relations (3.25) and induction, we have proved the following result.

Theorem 4: System (3.18)–(3.20) has a solution of the form

SðnÞ
2kþ1 ¼ D
nk1
qx , ð3:26Þ

where AðnÞ ðnÞ

2kþ1 and T2k are given by (3.21) as

AðnÞ 1
2kþ1 ¼ @x ðqD
nk1
qx Þ, TðnÞ 1 nk
2k ¼ @x D qx , ð3:27Þ

for k ¼ 0, . . . , n  1.
Thus, from (3.27) with k ¼ 0, we have TðnÞ 1 n
0 ¼ @x D qx and so the hierarchy (3.13) is
given explicitly in the form

@t q ¼ Dn qx : ð3:28Þ

The first two equations of the hierarchy generated by (3.28) are given by
3
qt ¼ qxxx þ q2 qx ,
2 ð3:29Þ
5 5 15
qt ¼ qxxxxx þ q2 qxxx þ 10qqx qxx þ q3x þ q4 qx :
2 2 8
As an example, there are two distinct solutions to (3.29(i)) in terms of both (x, t) of
the form
1 1
qðx, tÞ ¼ t , qðx, tÞ ¼ : ð3:30Þ
4 coshðx8 þ 512Þ x
16 coshð32 t
þ 32768Þ
 232 P. Bracken

Each of these functions has a standard soliton profile and generate a potential function
through p ¼ q/4. Substituting this p into (2.12), a function r can be obtained and finally s.
Putting (2.5) into (1.2), the coordinates of a surface in R3 are obtained upon integration
such that the coordinates depend on the remaining evolution parameter t.

4. The Willmore functional

Several properties are preserved under these deformations, but only the conservation of
the Willmore functional with respect to tn will be discussed. The following results will hold
whenever the surface terms produced by integration by parts vanish, which will be
assumed here. This will certainly be the case when p is a periodic function.
Lemma 2: Define the operators
D ¼ @2x þ q2 þ qx @1
x q, Dy ¼ @2x þ q2  q@1
x qx : ð4:1Þ

(a) The operator Dy is formally coadjoint to D, that is,

ð ð
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

f  Dg dx ¼ Dy f  g dx: ð4:2Þ

(b) D and Dy satisfy @xDy ¼ D@x.

Proof
(a) Starting with the expression on the right-hand side and integrating by parts
repeatedly,
ð ð
Dy f  g dx ¼ ð@2x þ q2  q@1
x qx Þfg dx
ð ð ð ð
2 2 2 2
¼ ð f@x g þ fq gÞdx  q@x ðqx fÞg dx ¼ ð f@x g þ fq gÞdx  @1
1
x ðqx f Þqg dx
ð ð ð ð
¼ ð f@2x g þ fq2 gÞdx þ qx f @1
x ðqgÞdx ¼ f ð @ 2
x þ q 2
þ q @
x x
1
qÞgdx ¼ f  Dg dx

as required.
(b) It is required to show that for an arbitrary function h, @x Dy h ¼ D@x h ¼
y
@2x hx þ q2 hx þ qx @1
x qhx . To do this, we write @xD h as follows

@x Dy h ¼ @x ð@2x h þ q2 h  q@1
x ðqx hÞÞ
¼ @3x h þ 2qqx h þ q2 @x h  qx @1 3 2 1
x ðqx hÞ  qðqx hÞ ¼ @x h þ qqx h þ q @x h  qx @x ðqx hÞ:

Integrating by parts, we have @1 1

x ðqx hÞ ¼ hq þ @x ðqhx Þ, hence

1
@x Dy h ¼ @3x h þ ðq2 Þx h þ q2 h  qx hq þ qx @1 3 2 1
x ðqhx Þ ¼ @x h þ q hx þ qx @x ðqhx Þ ¼ D@x h:
2
The first integral of the mKdV equation is proportional to the Willmore functional
defined as
ðT
W ¼ q2 dx: ð4:3Þ
0
œ
 Dynamical Systems 233

Theorem 5: The Willmore functional W is conserved provided that q satisfies the hierarchy
of equations (3.28)

qtn ¼ Dn qx :
Proof Applying Lemma 2 repeatedly, D must satisfy

Dn @x q ¼ Dn1 ð@x Dy qÞ ¼ Dn2 @x ðDy Þ2 q ¼    ¼ @x ðDy Þn q: ð4:4Þ

Differentiating W in (4.2) with respect to tn and using (3.28), we obtain

ðT ðT ðT
@W @q
¼2 q qðD qx Þdx ¼ 2 qDðDn1 qx Þdx
dx ¼ 2 n
@tn 0 @tn 0 0
ðT ðT
¼ 2 Dy q  Dn1 qx dx ¼ 2 ððDy Þn qÞqx dx:
0 0

Integrating by parts once and using (4.4), it follows that

Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

ðT
@W
¼ 2½qðDy Þn qT0  2 q@x ððDy Þn qÞdx
@tn 0
ðT
¼ 2½qðDy Þn qT0  2 qðDn qx Þdx:
0

Therefore,
ðT ðT
y n
2 n
qðD qx Þ dx ¼ 2½qðD Þ qT0 2 qðDn qx Þ dx:
0 0

Putting all the terms with an integral on the left-hand side, the surface term is required to
vanish, and hence we obtain
ðT
1
qðDn qx Þ dx ¼ ½qðDy Þn qT0 ¼ 0:
0 2
This is the case when the function q(Dy)nq is periodic, for example, or if q vanishes at the
endpoints. From this, it follows that @W/@tn ¼ 0. œ
Another way of expressing this result in a specific context is to say that the squared
mean curvature W of a torus of revolution is preserved by mKdV deformations.

5. Summary and conclusions

It might be hoped that such a straightforward scheme for obtaining an integrable
hierarchy can be extended to the (2 þ 1) case as well. As a brief example, consider linear
problem (1.7) and the following analogue of (3.2)
3
@t 1 þ @3 1 þ @3 1  3ð@pÞ@ 2 þ ð@! 1 þ 2p! 2 Þ ¼ 0,
þ 3! @ 1
2 ð5:1Þ
3
@t 2 þ @3 1 þ @3  
2 þ 3!@ 2 þ 3ð@pÞ@ 1 þ ð2p!  1 þ @! 2 Þ ¼ 0:
2
 234 P. Bracken

In (5.1), ! is an unspecified complex function which will be determined in due course. By

using symbolic manipulation, it can be verified directly that the following theorem holds.
Theorem 6: The compatibility condition for the linear problem (1.7) and (5.1) generates the
following pair of distinct equations in terms of the function p

ð@t p þ @3 p þ @3 p þ 3!ð@pÞ þ 3!ð  þ 3 pð@!Þ þ 3 pð@!ÞÞ

 @pÞ  2
2 2
3  2 Þ 1 þ 3ð@!  @p
 2 Þ@ 1 ¼ 0,
þ @ð@ !  @p
2
and

ð@t p þ @3 p þ @3 p þ 3!ð  þ 3!ð@pÞ þ 3 pð@!Þ þ 3 pð@!ÞÞ

 @pÞ  1
2 2
3  
þ @ð@p2  @!Þ 2
2 þ 3ð@p  @!Þ@ 2 ¼ 0:
2
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

If it is required that ! satisfy the constraints

 ¼ @p2 ,
@!  2,
@! ¼ @p ð5:2Þ

and their complex conjugate equations, the modified Novikov–Vesolov equation results

 þ 3 p@! þ 3 p@! ¼ 0:
@t p þ @3 p þ @3 p þ 3!@p þ 3! @p ð5:3Þ
2 2
Equation (5.3) is the associated non-linear integrable equation for the potential function p.
However, a link between the linear spectral problem and the generation of an
integrable hierarchy related to the mKdV equation has been clearly established which
yields the idea of mKdV deformations. The n-th mKdV equation is the compatibility
condition (3.3) for (3.1) and (3.2). The deformation of the potential q ¼ 4p by means of
(3.28) induces the deformation of  by (3.2). Using the deformed functions 1(x, y, t) and
2(x, y, t) of (2.5) in (1.2), the n-th mKdV deformation of a surface of revolution is
obtained. It might also be considered what additional hierarchies might be generated
by altering (3.1).

References

[1] K. Weierstrass, Mathematische Werke, Vol. 3, Verlagsbuchhandlung, Hillesheim, 1866, p. 219.

[2] A. Enneper, Nachr. Königl. Gesell., Wissenschaft, Georg-Augustus University, Göttingen 12
1868, p. 258.
[3] B.G. Konopelchenko, Induced Surfaces and their Integrable Dynamics, Stud. Appl. Math. 96
(1996), p. 9.
[4] B.G. Konopelchenko and I. Taimanov, Constant Mean Curvature Surfaces via an Integrable
Dynamical System, J. Phys. A 29 (1996), p. 1261.
[5] P. Bracken and A.M. Grundland, On Complete Integrability of the Generalized Weierstrass
System, J. Nonlin. Math. Phys. 9 (2002), p. 229.
[6] P. Bracken, A.M. Grundland, and L. Martina, The Weierstrass-Enneper System for Constant
Mean Curvature Surfaces and the Completely Integrable Sigma Model, J. Math. Phys. 40 (1999),
p. 3379.
 Dynamical Systems 235

[7] P. Bracken and A.M. Grundland, Symmetry Properties and Explicit Solutions of the Generalized
Weierstrass System, J. Math. Phys. 42 (2001), p. 1250.
[8] P. Bracken and A.M. Grundland, On the Backlund Transformation and the Generalized
Weierstrass System, Inverse Problems 16 (2000), p. 145.
[9] B.G. Konopelchenko and G. Landolfi, Generalized Weierstrass Representations for Surfaces in
Multi-Dimensional Riemann Spaces, J. Geom. Phys. 29 (1999), p. 319.
[10] P. Bracken and A.M. Grundland, Solutions of the Generalized Weierstrass Representation in
Four-Dimensional Euclidean Space, J. Nonlin. Math. Phys. 9 (2002), p. 357.
[11] O. Ceyhan, A.S. Fokas, and M. Gürses, Deformations of Surfaces Associated with Integrable
Gauss-Mainardi-Codazzi Equations, J. Math. Phys. 41 (2000), p. 2251.
[12] A.S. Fokas and I.M. Gelfand, Surfaces on Lie Groups, on Lie Algebras, and Their Integrability,
Commun. Math. Phys. 177 (1996), p. 203.
[13] F. Finkel and A.S. Fokas, A New Immersion Formula for Surfaces on Lie Algebras and
Integrable Equations, Vol. 29, CRM Proceedings and Lecture Notes, 2001.
[14] I. Taimanov, Modified Novikov-Vesolov Equation and Differential Geometry of Surfaces,
Translations Amer. Math. Soc., Ser 2 179 (1997), p. 133.
[15] I.A. Taimanov, Surfaces in four-space and the Davey Stewartson equations, J. Geom. Phys. 56
(2006), pp. 1235–1256.
Downloaded By: [Bracken, Paul] At: 17:50 27 April 2009

[16] I. Taimanov, Surfaces of Revolution in Terms of Solitons, Ann. Global Anal. Geom. 15 (1997),
p. 419.
[17] R. Osserman, A Survey of Minimal Surfaces, Dover, New York, 1996.

View publication stats