Next: The Kadomtsev-Petviashvili Equation
Up: Modified equations, rational solutions,
Previous: Modified equations, rational solutions,
In [1] we have formulated a procedure for calculating the Lax
pair and rational solutions of partial differential equations that
possess the Painlevé property. That is, for an equation with the
Painlevé property, a Bäcklund transformation is defined in terms
of an expansion about the ``singular manifold". This Bäcklund
transformation obtains (1) a type of ``modified equation" that can be
expressed in terms of Schwarzian derivatives and (2) a Miura
transformation from the modified to the original equation. By
linearizing the Ricati-type Miura transformation (and the modified
equations), the Lax pair is found. Then, further consideration of the
Bäcklund transformations for the modified equations provides a
method for the iterative construction of ``rational" solutions, and
finds the Lax pair for the modified equations as well.
We recall that the partial differential equation is said to possess
the Painlevé property [2]-[7] when the solutions
of the partial differential equation (pde) are ``single valued" about
the movable, singularity manifold and the singularity manifold is
``noncharacteristic." To be precise, if the singularity manifold is
determined by
and
is a solution of the pde, then we require that
where
,
, and
are analytic functions of
in a neighborhood of the
manifold (1.1) and
(the leading-order exponent) is a
(negative) rational number. The requirement that the manifold (1.1) be
noncharacteristic (for the pde) insures that the expansion (1.2) will
be well defined, in the sense of the Cauchy-Kovalevskaya theorem
[8]. Substitution of(1.2) into the pde determines that
value(s) of
, and defines the recursion relations for
,
. When the expansion (1.2) is well defined and
contains the maximum number of arbitrary functions allowed at the
``resonances" [2], [9], [10], the pde is
said to possess the Painlevé property and is conjectured to be
integrable. Informally, the resonances are the values of
for which
the
are not ``fixed" by the recursion relations (i.e., are
arbitrary).
The Bäcklund transformation is defined by truncating the expansion
(1.2) at the constant level term. That is, we set
and find, from the recursion relations for
and the condition that
vanish for
, a system of equations for
), where
will satisfy the (original) pde. This
system of equations will, in general (depending on the values of the
resonances), be overdetermined. Upon solving this system, it is found,
for those equations considered, the
satisfies an equation
formulated in terms of Schwarzian derivatives [3]:
This equation, or system of equations, we regard as a type of modified
equation. By the invariance of (1.4) under the Moebius group,
the ``modified" equations allow the Bäcklund transformation (1.5).
The above procedure may now be reapplied to the ``modified" (or
equivalent) equations to find different forms of Bäcklund
transformations. These Bäcklund transformations may take the form of
discrete symmetries [1],[5], [6],
reductions [1], or, as we shall see, more complicated
structures. The group of Bäcklund transformations for the modified
equations may be conveniently employed to iteratively construct
sequences of rational solutions. Also, by linearizing the Miura
transformation from modified to original equation we propose to
calculate the Lax pair [1], [6].
In this paper we consider the Kadomtsev-Petviashvili (KP) equation and
the Hirota-Satsuma equations. The modified equations are derived,
their (modified) Bäcklund transformations are calculated, and the
sequences of rational solutions are found.
Next: The Kadomtsev-Petviashvili Equation
Up: Modified equations, rational solutions,
Previous: Modified equations, rational solutions,
John Edward Weiss
2002-03-31