Download as pdf or txt
Download as pdf or txt
You are on page 1of 47

Some General Solutions to the

Painlev PIV Equation



Solomon M. Antoniou
SKEMSYS
Scientific Knowledge Engineering
and Management Systems

Corinthos 20100, Greece
solomon_antoniou@yahoo.com



Abstract
We initiate a solution procedure to the Painlev equations. The Riccati equation
method with variable expansion coefficients is used to find closed-form solutions
to the Painlev-PIV equation. The solutions are expressed in terms of Whittaker's
W and M functions.


Keywords: Painlev equations, Painlev-PIV equation, extended Riccati equation
method, nonlinear equations, exact solutions, Whittaker's functions.




2
1. Introduction.
The Painlev equations (numbered as PI-PVI) is a special class of second order
nonlinear ODEs which have no movable critical points (branch points or essential
singularities). These equations were discovered under a number of assumptions at
the end of nineteenth century/beginning of twentieth century by Painlev and
Gambier. They also appear in many physical applications. Some reviews and
further developments the reader can consult, are the classical book by E. L. Ince
(Ince [10], Chapter XIV) and the articles by A. S. Fokas and M. J. Ablowitz
(Fokas and Ablowitz [9]) and P. A. Clarkson (Clarkson [7]). Closely related to
Painlev equations is the so-called ARS conjecture (after Ablowitz, Ramani and
Segur [1], [2] and [11]) which is an integrability test for ordinary differential
equations. The ARS conjecture was extended to PDEs by J. Weiss, M. Tabor and
G. Carnevale (Weiss, Tabor and Carnevale [13]) who introduced the so-called
singular manifold method (for a review and further examples see Ramani,
Grammaticos and Bountis [12]). This method (also named as Painlev truncation
method) serve also as a solution method (Weiss [14], [15] and [16]) in the sense
that it can determine the Lax pairs and the Bcklund transformations.
No explicit solutions of the Painlev equations have been found so far. In some
cases only rational solutions are available (Airlaut [4] and Wenjum and Yezhou
[17]). Some relations have also been established between solutions (Fokas and
Ablowitz [9], Clarkson [7]). We have been able to find explicit solutions for the
Painlev PII equation (Antoniou [6]) expressed in terms of the Airy functions.
In this paper we introduce a solution method and find some closed-form solutions
to the Painlev PIV equation using the extended Riccati equation method with
variable expansion coefficients. The solutions are expressed in terms of
Whittaker's W and M functions. The paper is organized as follows: In Section 2
we describe the extended Riccati equation method with variable expansion
coefficients. This method was introduced in Antoniou ([5]). In Section 3 we apply
our method to the Painleve PIV equation
3
w
w ) x ( 2 w x 4 w
2
3
dx
dw
w 2
1
dx
w d
2 2 3
2
2
2

+ + + +
|

\
|
=
and we set up a system of nine ordinary differential equations. In solving that
system we consider four cases. This is done in the next four Sections (Section 4-
Section 7). In Sections 4 and 5 we find two families of solutions which are
expressed in terms of Whittaker's functions (modulo some compatibility
conditions). To each family there correspond an infinite number of subfamilies. In
Sections 6 and 7 we find two more families of solutions, expressed again in terms
of Whittaker's functions. However it is not known if these two solutions have to be
accepted or not because it is not clear if they satisfy the compatibility conditions.
In Section 8 we consider another method of solution, the ) G / G ( expansion
method with variable expansion coefficients. We obtain a system of five
differential equations which, when solved, can in principle determine the unknown
function G and then ) x ( u .
2. The Method.
We suppose that a nonlinear ordinary differential equation
0 ) , u , u , u , x ( F
xx x
= L (2.1)
with unknown function ) x ( u admits solution expressed in the form


= =
+ =
n
1 k
k
k
n
0 k
k
k
Y
b
Y a ) x ( u (2.2)
where all the expansion coefficients depend on the variable x:
) x ( a a
k k
, ) x ( b b
k k
for every n , , 2 , 1 , 0 k L =
The function ) x ( Y Y satisfies Riccatis equation

2
Y B A ) x ( Y + = (2.3)
where the coefficients A and B depend on the variable x as well.
In solving the nonlinear ODE (2.1), we substitute the expansion (2.2) in (2.1) and
then we balance the nonlinear term with the highest derivative term of the function
4
) x ( u which determines n (the number of the expansion terms). Equating the
coefficients of the different powers of the function ) x ( Y to zero, we find a
system of nonlinear ordinary differential equations from which we can determine
the various expansion coefficients ) x ( a
k
, ) x ( b
k
and the functions ) x ( A , ) x ( B .
We finally solve Riccati's equation and then find the solutions of the equation
considered.
3. The Painleve PIV equation and its solutions.
We consider the Painleve PIV equation

w
w ) x ( 2 w x 4 w
2
3
dx
dw
w 2
1
dx
w d
2 2 3
2
2
2

+ + + +
|

\
|
= (3.1)
where ) x ( w w is the unknown function and x the independent variable,
considered complex in general. We shall use the extended Riccati equation method
in solving equation (3.1). In this case we consider the expansion

= =
+ =
n
1 k
k
k
n
0 k
k
k
Y
b
Y a ) z ( w (3.2)
and balance the second order derivative term with the second order nonlinear term
of (3.1). We then find 1 n = and thus

Y
b
Y a a ) x ( w
1
1 0
+ + = (3.3)
where all the coefficients
0
a ,
1
a and
1
b depend on x, and Y satisfies Riccatis
equation
) x ( Y ) x ( B ) x ( A ) x ( Y
2
+ = (3.4)
The prime will always denote derivative with respect to the variable x. From
equation (3.3) we obtain, taking into account
2
BY A Y + =

2
2
1 1 2
1 1 0
Y
) BY A ( b
Y
b
) BY A ( a Y a a ) x ( w
+

+ + + + = (3.5)
) BY A ( Y B 2 Y B A ( a ) BY A ( a 2 Y a a ) x ( w
2 2
1
2
1 1 0
+ + + + + + + =
5
2
2
1
2
1
2
1 1
Y
) BY A ( BY b 2 ) Y B A ( b ) BY A ( b 2
Y
b + + + + +


+
3
2 2
1
Y
) BY A ( b 2 +
+ (3.6)
Therefore equation (3.1), written as
0 2 w ) x ( 4 w x 8 w 3 ) w ( w w 2
2 2 3 4 2
=
under the substitutions (3.3), (3.5) and (3.6), becomes

3
2 2
1 1 2
1 1 0
1
1 0
Y
) BY A ( b 2
Y
b
) BY A ( a 2 Y a a
Y
b
Y a a 2
+
+


+ + + +
|

\
|
+ +
) BY A ( Y B 2 Y B A ( a
2 2
1
+ + + +

+ + + + +

2
2
1
2
1
2
1
Y
) BY A ( BY b 2 ) Y B A ( b ) BY A ( b 2


4
1
1 0
2
2
2
1 1 2
1 1 0
Y
b
Y a a 3
Y
) BY A ( b
Y
b
) BY A ( a Y a a
|

\
|
+ +

+ + + +
0 2
Y
b
Y a a ) x ( 4
Y
b
Y a a x 8
2
1
1 0
2
3
1
1 0
=
|

\
|
+ +
|

\
|
+ + (3.7)
Upon expanding and equating the coefficients of Y to zero, we obtain from the
above equation a system of nine nonlinear ordinary differential equations from
which we can determine the various expansion coefficients. We obtain
coefficient of
4
Y :
0 a 3 B a 3
4
1
2 2
1
= (3.8)
coefficient of
3
Y :
0 a x 8 ) B a 2 B a ( a 2 B a a 4 B a a 2 a a 12
3
1 1 1 1
2
1 0 1 1
3
1 0
= + + + (3.9)
coefficient of
2
Y :

2
1 1 1 0 1
2 2
1
) a ( B a ) B b a A a ( 2 ) a x ( a 4 +
6

2
1 1
2
1 0
2
1
2
0
2
1
2
0 1 1
B b a 4 a a x 24 a a 12 a ) a a b 2 ( 6 + +
0 ) AB a 2 a ( a 2 ) B a 2 B a ( a 2
1 1 1 1 1 0
= + + + + (3.10)
coefficient of Y:

1 0
2
0 1 1 1
2
1 0
2
1 0
a a ) a b a 2 ( 12 b a a 12 ) x ( a a 8 +
) A a A a 2 a B b 2 B b ( a 2 ) B a 2 B a ( b 2
1 1 0 1 1 1 1 1 1
+ + + + + +
) B b a A a ( a 2 B b a 2 ) B A a 2 a ( a 2
1 0 1 1 1 1 1 1 0
+ + +
0 ) a a a b ( x 24
1
2
0
2
1 1
= + (3.11)
coefficient of
0
Y :
) A a A a 2 a B b 2 B b ( a 2 ) B A a 2 a ( b 2
1 1 0 1 1 0 1 1 1
+ + + + +
2
1 0 1 1 1 1 1 1 1 1
) B b a A a ( a b 2 AB b a 2 ) AB b 2 b ( a 2 + + + +
2
1
2
1
3
0 1 1 0
2 2
0 1 1
b a 6 ) a b a a 6 ( x 8 ) x ( ) a b a 2 ( 4 + +
0 2 ) a b a 2 ( 3 b a a 24
2 2
0 1 1 1 1
2
0
= + (3.12)
coefficient of
1
Y

:

1 0
2
0 1 1
2
1 1 0
2
1 0
b a ) a b a 2 ( 12 b a a 12 ) x ( b a 8 +
) A a A a 2 a B b 2 B b ( b 2 ) A b 2 A b ( a 2
1 1 0 1 1 1 1 1 1
+ + + + +
) B b a A a ( b 2 A a b 2 ) AB b 2 b ( a 2
1 0 1 1 1 1 1 1 0
+ + +
0 ) b a b a ( x 24
1
2
0
2
1 1
= + (3.13)
coefficient of
2
Y

:

2
1 1 1 0 1
2 2
1
) b ( A b ) B b a A a ( 2 ) x ( b 4 + +

2
1 1
2
1 0
2
1
2
0
2
1
2
0 1 1
A b a 4 b a x 24 b a 12 b ) a a b 2 ( 6 + +
0 ) AB b 2 b ( b 2 ) A b 2 A b ( a 2
1 1 1 1 1 0
= + + + (3.14)
coefficient of
3
Y

:
0 b x 8 ) A b 2 A b ( b 2 A b a 4 A b b 2 b a 12
3
1 1 1 1
2
1 0 1 1
3
1 0
= + + + (3.15)
7
coefficient of
4
Y

:
0 b 3 A b 3
4
1
2 2
1
= (3.16)
We are to solve the system of equations (3.8)-(3.16), supplemented by Riccati's
equation
2
BY A Y + = . From equations (3.8) and (3.16), ignoring the trivial
solutions, we obtain
B a
1
= and A b
1
= (3.17)
respectively. We thus consider the following four cases.
4. Case I. First Solution of the Painlev PIV Equation.
We first consider the case
B a
1
= and A b
1
= (4.1)
We then obtain from (3.9) and (3.15)
x 2
B
B
a 2
0

= and x 2
A
A
a 2
0

= (4.2)
respectively. Equating the two different expressions of
0
a we have 0
A
A
B
B
=


and by integration, we find that
p B A = (4.3)
where p is a constant to be determined.
From equation (3.10), using (4.1) we obtain
0 ) x ( 4 a x 24 a 2 a 18 AB 4
B
B
B
B
a 6
B
B
2
2
0 0
2
0
2
0
=
|

\
|

+


From the above equation, using the first of (4.2) and (4.3), we get
0 ) x ( 2
B
B
2
3
B
B
2
2
= + +
|

\
|


(4.4)
where
1 p 2 2 + = (4.5)
From equation (3.14), using (4.1) we get
8
0 ) a x ( 4 a x 24 a 2 a 18 B A 4
A
A
A
A
a 6
A
A
2
2
0 0
2
0
2
0
= +
|

\
|



From the above equation, using the second of (4.2) and (4.3), we obtain
0 ) x ( 2
A
A
2
3
A
A
2
2
= + +
|

\
|


(4.6)
where
1 p 2 2 = (4.7)
From equation (3.11) we get, using (4.1)

3
0 0 0 0 0
a 12 B A 4 B A 8 B A a 32 a 2
B
B
a 2
B
B
a 2 + +




0 ) a AB ( x 24 ) a x ( a 8
2
0
2
0
= +
From the above equation, using the first of (4.2) and (4.3), we get

(
(

\
|

\
|

\
|

\
|

2 3
B
B
2
3
B
B
x 2
B
B
2
3
B
B
B
B
B
B
B
B
B
B

0 ) p 2 2 x ( x 4
B
B
) 2 p 4 4 x 2 (
2 2
= +

+ + + (4.8)
From equation (3.13) we get, using (4.1)

3
0 0 0 0 0
a 12 B A 8 B A 4 B A a 32 a 2
A
A
a 2
A
A
a 2 + +




0 ) a AB ( x 24 ) a x ( a 8
2
0
2
0
= +
From the above equation, using the second of (4.2) and (4.3), we get

(
(

\
|


+
|

\
|

+


2 3
2
A
A
2
3
A
A
x 2
A
A
2
3
A
A A
3
A
A

0 ) p 2 2 x ( x 4
A
A
) 1 p 2 2 x ( 2
2 2
= + +

+ + (4.9)
From equation (3.12), using (4.1) and the first of (4.2), we obtain

4
2
2 2
B
B
16
9
B
B B
2
1
B
B
B
B
B
B
4
1
B
B
|

\
|

+

+
|

\
|

\
|



9

|

\
|

+

+

\
|

+

A
A
B
B
) B A ( 2
B
B
2
3
B
B B
3
B
B
x
3
2


|

\
|

+
|

\
|

+

+
B
B
A
A
) AB ( 3
B
B
) 2 x ( x 2 ) AB (
B
B
5
A
A
x 2
2

) AB ( 8 x 4 x
B
B
1 x
2
1
B
B
) AB ( 8
2 4
2
2
2
+ + +
|

\
|

|

\
|
+ +
|

\
|


0 1 2 ) B A ( 8 ) AB ( x 4
2 2
= + (4.10)
The previous equation can be further transformed taking into account
B
B
A
A
=

,
p B A = and
2
B
B
2
A
A
B
B
|

\
|

=

+

. We also express the derived equation in terms of
the function
B
B
F

= (4.11)
taking into account the identities
2
F F
B
B
+ =

and
3
F F F 3 F
B
B
+ + =

. We thus
have that (4.10) takes the form
F ) p 4 2 x ( x 2 F
2
1
F x F
2
1
F F
2
1
2 3 3
+
|

\
|

|

\
|

F ) F (
4
1
F
16
1
F ) p 2 2 x (
2
1
2 4 2 2
+ + + +
0 ) 1 2 p 8 p 8 x p 4 x 4 x (
2 2 2 4
= + + + + (4.12)
Before we go on to solving equation (4.4), we first prove that equations (4.4) and
(4.8) are compatible each other: every solution of (4.4) satisfies equation (4.8). To
prove that, we express both equations in terms of F (introduced in (4.11)). We
find that (4.4) and (4.8) are equivalent to
0 ) 1 p 2 2 x ( 2 F
2
1
F
2 2
= + + + (4.13)
and
10
0 ) p 2 2 x ( x 4 F ) 1 p 2 2 x ( 2 ) F
2
1
F ( x 2 F
2
1
F
2 2 2 3
= + + + + (4.14)
respectively. Upon substituting the expression
2
F
2
1
F derived from (4.13) into
(4.14) we obtain the equation
0 x 4 F ) 1 p 2 2 x ( 2 F
2
1
F
2 3
= + + + + (4.15)
Multiplying (4.13) by F and rearranging terms, we obtain
F F F ) 1 p 2 2 x ( 2 F
2
1
2 3
= + + +
Substituting the above expression into (4.15) we get
0 x 4 F F F = + (4.16)
Differentiation of (4.13) gives 0 x 4 F F F = + which is (4.16). We have thus
proved that (4.4) and (4.8) are compatible each other.
We now come into solving equation (4.4), i.e. equation (4.13):
0 ) x ( 2 F
2
1
F
2 2
= + + (4.17)
The above equation is a Riccati differential equation and thus under the standard
substitution

u
u
2 F

= (4.18)
takes on the form
0 u ) x ( u
2
= + (4.19)
The above is a Weber differential equation (Abramowitz and Stegun [3], 19.1-
19.3) and its solution can be expressed in terms of Parabolic Cylinder functions,
Kummer's confluent hypergeometric functions or Whittaker's functions. We prefer
to express the solution in terms of Whittaker's W and M functions (Abramowitz
and Stegun [3], 13.1.31-13.1.34) since the notation is more compact:
} ) x ( M C ) x ( W C {
x
1
) x ( u
2
, 2
2
, 1
+ = (4.20)
11
where
4

= ,
4
1
= (4.21)
The function ) x ( B B= can easily be calculated. Since
u
u
2
B
B
=

, we obtain by
integration

) x ( u
C
) x ( B
2
= (4.22)
where C is a constant. We come next to evaluate the function Y solving
Riccati's equation
2
BY A Y + = . Under the substitution

v
v
B
1
Y

= (4.23)
Riccati's equation takes on the form
0 v ) AB ( v
B
B
v = +
|

\
|

(4.24)
Since
u
u
2
B
B
=

and p AB= , equation (4.24) becomes 0 v u p v u 2 v u = + +


which can also be written as 0 ) v u ( p v u ) v u ( = + . Under the substitution
v u U = (4.25)
we get the differential equation 0 U
u
u
p U =
|

\
|

+ and since + =

2
x
u
u
(see
equation (4.19)) we get the equation
0 U ) p x ( U
2
= + (4.26)
The above differential equation admits the general solution
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( U
2
, 2
2
, 1
+ = (4.27)
where
4
p
= ,
4
1
= (4.28)
12
We have further
u
u
U
U
v
v

and then we obtain the solution of Riccati's equation


from (4.23), using also (4.22):

|
|

\
|

=
) x ( u
) x ( u
) x ( U
) x ( U
) x ( u
C
1
Y
2
(4.29)
where ) x ( u and ) x ( U are given by (4.20) and (4.27) respectively.
We now come to consider the differential equations (4.6) and (4.9) satisfied by the
function A. These equations can be expressed in terms of the function

A
A
H

= (4.30)
as
0 ) 1 p 2 2 x ( 2 H
2
1
H
2 2
= + + (4.31)
and
H ) 1 p 2 2 x ( 2 H
2
1
H x 2 H
2
1
H
2 2 3
+ +
|

\
|
+
0 ) p 2 2 x ( x 4
2
= + + (4.32)
The above two equations are compatible each other. The proof is along the lines of
proving that (4.13) and (4.14) are compatible. We thus have to combine (4.4) with
(4.6) and establish a compatibility condition. Adding (4.4) and (4.6) and taking
into account that
B
B
A
A
=

and
2
B
B
2
A
A
B
B
|

\
|

=

+

, we obtain the equation
0 ) p 2 2 x ( 4
B
B
2
2
= + +
|

\
|

, which in view of
u
u
2
B
B
=

, takes the form


0 ) p 2 2 x (
u
u
2
2
= +
|

\
|

(4.33)
This is the first compatibility condition.
We consider next the compatibility issue between (4.4), (4.8) and (4.10). We could
instead examine the compatibility between (4.13), (4.15) and (4.12).
13
Solving (4.13) with respect to
2
F
2
1
F squaring both members and rearranging,
we obtain
2 2 4 2 2
) 1 p 2 2 x ( F
16
1
F F
4
1
) F (
4
1
+ + = + (4.34)
Adding (4.12), (4.34) and substituting
3
F
2
1
F derived from (4.15) into (4.12)
simultaneously, we get the equation

2 2 2 2 2
F ) 1 p 2 2 x (
2
1
F x p 4 F
2
1
F F
2
1
F F
4
1
+ + +
|

\
|
+
|

\
|

0 p 4 4 4 2 2 p 16 p 12 x 2 x p 8
2 2 2 2
= + + + + (4.35)
Upon substituting
2
F
2
1
F derived from (4.13) into (4.35) we obtain the equation
0 2 p 4 4 2 p 8 p 6 ) x 2 F ( x p 2
2 2
= + + + (4.36)
Since
u
u
2 F

= with u given by (4.20), equation (4.36) takes the form
0 ) 2 p 4 4 2 p 8 p 6 (
) x ( u
) x ( u x ) x ( u
x p 4
2 2
= + + + +
|
|

\
| +
(4.37)
The above equation can be written as
0 ) x ( u k ] ) x ( u x ) x ( u [ x p 4 = + (4.38)
where
2 p 4 4 2 p 8 p 6 k
2 2
+ + + + = (4.39)
Equation (4.38) will be called the second compatibility condition.
Both compatibility conditions are examined in Appendix A. According to the
results of Appendix A, the compatibility conditions hold for every x if
n 4 3 = + ( L , 3 , 2 , 1 , 0 n = ) and 0 C
4
1
2
C
1 2
=
|

\
| +

+ (4.40)
14
We thus see that to the first family of solutions there correspond an infinite
number of subfamilies. Since is given by (4.5), we can build a Table of the
various values of the parameters. The values of the parameters affect both
functions ) x ( u and ) x ( U given by the pairs (4.20), (4.21) and (4.27), (4.28).
We consider below a Table (Table 1) for the various parameters corresponding to
the first four values of n ( 3 , 2 , 1 , 0 n = ).
n

p

|

\
|
+

4
1

0 C
4
1
2
C
1 2
=
|

\
| +

+
0 3 2 +
4
5 +

2
0 C C
1 2
= +
1 7 4 +
4
11 +

3
4

0 C
2
3
C
1 2
= +
2
11 6 +
4
17 +

5
8

0 C
4
5
C
1 2
= +
3 15 8 +
4
23 +

105
16

0 C
8
105
C
1 2
= +

Table 1
Therefore we have the following
For 0 n = , 2 p + = and
)} x ( M ) x ( W {
x
C
) x ( u
2
,
2
,
1

= where
4
3
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( U
2
, 2
2
, 1
+ = where
4
5 +
= ,
4
1
=
For 1 n = , 4 p + = and

)
`

+ =

) x ( M
2
3
) x ( W
x
C
) x ( u
2
,
2
,
1
where
4
7
= ,
4
1
=
15
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( U
2
, 2
2
, 1
+ = where
4
11 +
= ,
4
1
=
For 2 n = , 6 p + = and

)
`

=

) x ( M
4
5
) x ( W
x
C
) x ( u
2
,
2
,
1
where
4
11
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( U
2
, 2
2
, 1
+ = where
4
17 +
= ,
4
1
=
For 3 n = , 8 p + = and

)
`

+ =

) x ( M
8
105
) x ( W
x
C
) x ( u
2
,
2
,
1
where
4
15
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( U
2
, 2
2
, 1
+ = where
4
23 +
= ,
4
1
=
Conclusion. The solution of Painlev PIV equation

w
w ) x ( 2 w x 4 w
2
3
dx
dw
w 2
1
dx
w d
2 2 3
2
2
2

+ + + +
|

\
|
=
is given by
Y
b
Y a a ) x ( w
1
1 0
+ + = where

|
|

\
|
+

= x
) x ( u
) x ( u
x
B
B
2
1
a
0
,
) x ( u
C
) x ( B a
2
1
= = , ) x ( u
C
p
A b
2
1
= =
and
|
|

\
|

=
) x ( u
) x ( u
) x ( U
) x ( U
) x ( u
C
1
Y
2
. Therefore we have the solution

1
) x ( u
) x ( u
) x ( U
) x ( U
p x
) x ( U
) x ( U
) x ( w

|
|

\
|


|
|

\
|
+

= (4.41)
where ) x ( U and ) x ( u are given by (4.27) and (4.20) respectively.
Below we list the following four solutions only.
(I) For 0 n = ,
1
) x ( u
) x ( u
) x ( U
) x ( U
) 2 ( x
) x ( U
) x ( U
) x ( w

|
|

\
|

+
|
|

\
|
+

=
16
)} x ( M ) x ( W {
x
C
) x ( u
2
,
2
,
1

= with
4
3
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( U
2
, 2
2
, 1
+ = with
4
5 +
= ,
4
1
=
(II) For 1 n = ,
1
) x ( u
) x ( u
) x ( U
) x ( U
) 4 ( x
) x ( U
) x ( U
) x ( w

|
|

\
|

+
|
|

\
|
+

=

)
`

+ =

) x ( M
2
3
) x ( W
x
C
) x ( u
2
,
2
,
1
with
4
7
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( U
2
, 2
2
, 1
+ = with
4
11 +
= ,
4
1
=
(III) For 2 n = ,
1
) x ( u
) x ( u
) x ( U
) x ( U
) 6 ( x
) x ( U
) x ( U
) x ( w

|
|

\
|

+
|
|

\
|
+

=

)
`

=

) x ( M
4
5
) x ( W
x
C
) x ( u
2
,
2
,
1
with
4
11
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( U
2
, 2
2
, 1
+ = with
4
17 +
= ,
4
1
=
(IV) For 3 n = ,
1
) x ( u
) x ( u
) x ( U
) x ( U
) 8 ( x
) x ( U
) x ( U
) x ( w

|
|

\
|

+
|
|

\
|
+

=

)
`

+ =

) x ( M
8
105
) x ( W
x
C
) x ( u
2
,
2
,
1
with
4
15
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( U
2
, 2
2
, 1
+ = with
4
23 +
= ,
4
1
=
5. Case II. Second Solution of the Painlev PIV Equation.
We next consider the case
B a
1
= and A b
1
= (5.1)
We then obtain from (3.9) and (3.15)
x 2
B
B
a 2
0

= and x 2
A
A
a 2
0

= (5.2)
17
respectively. Equating the two different expressions of
0
a and integrating, we find
that
p B A = (5.3)
where p is a constant. From equation (3.10), using (3.26), we obtain
0 ) x ( 4 B A 4 a x 24 a 18 a 2
B
B
B
B
a 6
B
B
2
2
0
2
0 0
2
0
= +
|

\
|



From the above equation, using the first of (3.27) and (3.28), we get
0 ) 1 p 2 2 x ( 2
B
B
2
3
B
B
2
2
= + +
|

\
|


(5.4)
From equation (3.14) we get
0 ) x ( 4 a x 24 a 2 a 18 B A 4
A
A
A
A
a 6
A
A
2
2
0 0
2
0
2
0
=
|

\
|

+


From the above equation, using the second of (5.2) and (5.3) we obtain
0 ) 1 p 2 2 x ( 2
A
A
2
3
A
A
2
2
= + + +
|

\
|


(5.5)
From equation (3.11) we get

3
0 0 0 0 0
a 12 B A 4 B A 8 B A a 32 a 2
B
B
a 2
B
B
a 2 + +




0 ) a B A ( x 24 ) x ( a 8
2
0
2
0
= +
From the above equation, using the first of (5.2) and (5.3) we obtain

B
B
) 1 p 2 2 x ( 2
B
B
2
3
B
B
x 2
B
B
2
3
B
B B
3
B
B
2
2 3
2

+ +
(
(

\
|


+
|

\
|

+



0 ) p 2 2 x ( x 4
2
= + + (5.6)
From equation (3.13) we obtain

2
0 0
3
0 0 0 0
a x 12 B A x 12 B A a 16 a 6 a
A
A
a
A
A
a +




0 ) x ( a 4 B A 2 B A 4
2
0
= +
18
From the above equation, using the first of (5.2) and (5.3) we obtain

A
A
x 2
A
A
x 3
A
A
2
3
A
A A
3
A
A
2 3
2

\
|

+
|

\
|

+



0 ) p 2 2 x ( x 4
A
A
) 1 p 2 2 x ( 2
2 2
= +

+ + + (5.7)
From equation (3.12), using (5.1) and the first of (5.2), we obtain

4
2
2 2
B
B
16
9
B
B B
2
1
B
B
B
B
B
B
4
1
B
B
|

\
|

+

+
|

\
|

\
|

\
|

+

+

\
|

+


+
A
A
B
B
) B A ( 2
B
B
2
3
B
B B
3
B
B
x
3
2


|

\
|

+ +
|

\
|

+

B
B
A
A
) AB ( 3
B
B
) 2 x ( x 2 ) AB (
B
B
5
A
A
x 2
2

) AB ( 8 x 4 x
B
B
1 x
2
1
B
B
) AB ( 8
2 4
2
2
2
+ + +
|

\
|

|

\
|
+ + +
|

\
|


0 1 2 ) B A ( 8 ) AB ( x 4
2 2
= + (5.8)
The previous equation can be further transformed taking into account
B
B
A
A
=

,
p B A = ,
2
B
B
2
A
A
B
B
|

\
|

=

+

. We also express the derived equation in terms of the
function F defined by
B
B
F

= (5.9)
taking into account the identities
2
F F
B
B
+ =

and
3
F F F 3 F
B
B
+ + =

. We thus
have that (5.8) takes the form
F ) p 4 2 x ( x 2 F
2
1
F x F
2
1
F F
2
1
2 3 3
+ +
|

\
|
+
|

\
|

F ) F (
4
1
F
16
1
F ) p 2 2 x (
2
1
2 4 2 2
+ + +
19
0 ) 1 2 p 8 p 8 x p 4 x 4 x (
2 2 2 4
= + + + + (5.10)
The above equation can be simplified using equations (5.4) and (5.6). First of all,
equations (5.4) and (5.6) can be expressed in terms of F and they are equivalent
to
0 ) 1 p 2 2 x ( 2 F
2
1
F
2 2
= + + (5.11)
and
0 ) p 2 2 x ( x 4 F ) 1 p 2 2 x ( 2 ) F
2
1
F ( x 2 F
2
1
F
2 2 2 3
= + + + + + (5.12)
respectively. Substituting
2
F
2
1
F derived from (5.11) into (5.12) we obtain the
equation
0 x 4 F ) 1 p 2 2 x ( 2 F
2
1
F
2 3
= + + + (5.13)
Equations (5.12) and (5.13) are compatible each other. The proof goes along the
lines of the previous Section. Solving (5.11) with respect to
2
F
2
1
F , squaring
both members and rearranging, we obtain

2 2 4 2 2
) 1 p 2 2 x ( F
16
1
F F
4
1
) F (
4
1
+ = + (5.14)
Adding (5.14) to (5.10) and substituting
3
F
2
1
F derived from (5.13) into (5.10)
simultaneously, we obtain the equation
F x p 4 F ) 1 p 2 2 x (
2
1
F
2
1
F F
2
1
F F
4
1
2 2 2 2 2
+
|

\
|

|

\
|

0 2 2 p 4 4 4 p 16 p 12 x 2 x p 8
2 2 2 2
= + + + (5.15)
We substitute
2
F
2
1
F derived from (5.11) into the above equation and we obtain
0 ) 2 p 4 4 2 p 8 p 6 ( ) x 4 F 2 ( x p
2 2
= + + + + + (5.16)
20
The above equation is the second compatibility condition between (5.4), (5.6) and
(5.8).
We come next to the solution of (5.4), which is equivalent to (5.11). Under the
substitution

f
f
2 F

= (5.17)
equation (5.11) becomes
0 f ) x ( f
2
= + (5.18)
where
1 p 2 2 = (5.19)
The general solution of (5.18) is given by
)} x ( M C ) x ( W C {
x
1
) x ( f
2
, 2
2
, 1
+ = (5.20)
where

4

= ,
4
1
= (5.21)
Since
f
f
2
B
B
=

, we obtain by integration

) x ( f
C
) x ( B
2
= (5.22)
The above relation expresses the function B in terms of ) x ( f .
Equations (5.5) and (5.7) are compatible each other. Every solution of (5.5)
satisfies equation (5.7). These two equations can be expressed in terms of the
function H defined by

A
A
H

= (5.23)
as
0 ) 1 p 2 2 x ( 2 H
2
1
H
2 2
= + + + (5.24)
and
21
H ) 1 p 2 2 x ( 2 H
2
1
H x 2 H
2
1
H
2 2 3
+ + +
|

\
|

0 ) p 2 2 x ( x 4
2
= + (5.25)
In fact, multiplying (5.24) by F and rearranging, we obtain
H ) 1 p 2 2 x ( 2 H H H
2
1
2 3
+ + = (5.26)
Upon substitution of
3
H
2
1
using (5.26) and
2
H
2
1
H derived from (5.24) into
(5.25) we obtain the equation 0 x 4 H H H = + . This last equation however
comes by differentiation of (5.24).
We next examine the compatibility between equations (5.4) and (5.5). Adding
these two equations and taking into account
B
B
A
A
=

and
2
B
B
2
B
B
A
A
|

\
|

=

+

, we
obtain the equation 0 ) p 2 2 x ( 4
B
B
2
2
= + +
|

\
|

, which in view of
f
f
2
B
B
=


takes on the form
0 ) p 2 2 x (
f
f
2
2
= +
|

\
|

(5.27)
The above equation will be called the first compatibility condition.
In Appendix B we consider both equations (5.27) and (5.16) and we end up with a
pair of compatibility conditions.
We are now ready to solve Riccati's equation
2
BY A Y + = . This equation under
the substitution

v
v
B
1
Y

= (5.28)
takes the form 0 v ) AB ( v
B
B
v = +
|

\
|

. This equation, since
f
f
2
B
B
=

and
p AB= , becomes 0 v f p v f 2 v f = + + , which is equivalent to
22
0 ) v f ( p v f ) v f ( = + (5.29)
The substitution
v f X = (5.30)
transforms (5.29) into 0 X
f
f
p X =
|

\
|

+ and since + =

2
x
f
f
, we get the
equation
0 X ) p x ( X
2
= + (5.31)
The general solution of (5.31) is given by
)} x ( M C
~
) x ( W C
~
{
x
1
) x ( X
2
, 2
2
, 1
+ = (5.32)
where

4
p
= ,
4
1
= (5.33)
Since
f
f
X
X
v
v

, we have that

|

\
|

=
f
f
X
X
C
) x ( f
Y
2
(5.34)
Both compatibility conditions (5.27) and (5.16) are examined in Appendix B.
According to the results of Appendix B, the compatibility conditions hold for
every x if
n 4 3 = + ( L , 3 , 2 , 1 , 0 n = ) and 0 C
4
1
2
C
1 2
=
|

\
| +

(5.35)
We thus see that to the first family of solutions there correspond an infinite
number of subfamilies. Since is given by (5.19), we can build a Table of the
various values of the parameters. The values of the parameters affect both
functions ) x ( w and ) x ( X given by the pairs (5.20), (5.21) and (5.32), (5.33).
We consider below a Table (Table 2) for the various parameters corresponding to
the first four values of n ( 3 , 2 , 1 , 0 n = ).
23
n p

|

\
|
+

4
1

0 C
4
1
2
C
1 2
=
|

\
| +


0 3 1 +
4
4 +

2
0 C C
1 2
= +
1 7 3 +
4
10 +

3
4

0 C
2
3
C
1 2
=
2
11 5 +
4
16 +

5
8

0 C
4
5
C
1 2
= +
3 15 7 +
4
22 +

105
16

0 C
8
105
C
1 2
=

Table 2
Therefore we have the following
For 0 n = , we have 1 p + = and
)} x ( M ) x ( W {
x
C
) x ( f
2
,
2
,
1

= where
4
3
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( X
2
, 2
2
, 1
+ = where
4
4 +
= ,
4
1
=
For 1 n = , we have 3 p + = and

)
`

+ =

) x ( M
2
3
) x ( W
x
C
) x ( f
2
,
2
,
1
where
4
7
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( X
2
, 2
2
, 1
+ = where
4
10 +
= ,
4
1
=
For 2 n = , we have 5 p + = and

)
`

=

) x ( M
4
5
) x ( W
x
C
) x ( f
2
,
2
,
1
where
4
11
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( X
2
, 2
2
, 1
+ = where
4
16 +
= ,
4
1
=
For 3 n = , we have 7 p + = and
24

)
`

+ =

) x ( M
8
105
) x ( W
x
C
) x ( f
2
,
2
,
1
where
4
15
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( X
2
, 2
2
, 1
+ = where
4
22 +
= ,
4
1
=
Conclusion. The solution of the Painleve PIV equation

w
w ) x ( 2 w x 4 w
2
3
dx
dw
w 2
1
dx
w d
2 2 3
2
2
2

+ + + +
|

\
|
=
is given by
Y
b
Y a a ) x ( w
1
1 0
+ + = where
x
) x ( f
) x ( f
x
B
B
2
1
a
0

= ,
) x ( f
C
) x ( B a
2
1
= = , ) x ( f
C
p
) x ( A b
2
1
= =
and
|
|

\
|

=
) x ( f
) x ( f
) x ( X
) x ( X
C
) x ( f
Y
2
. Therefore we have the solution

1
) x ( f
) x ( f
) x ( X
) x ( X
p x
) x ( X
) x ( X
) x ( w

|
|

\
|

+
|
|

\
|

=
Below we list the following four solutions only.
For 0 n = we have the solution
1
) x ( f
) x ( f
) x ( X
) x ( X
) 1 ( x
) x ( X
) x ( X
) x ( w

|
|

\
|

+ +
|
|

\
|

=
where
)} x ( M ) x ( W {
x
C
) x ( f
2
,
2
,
1

= with
4
3
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( X
2
, 2
2
, 1
+ = with
4
4 +
= ,
4
1
=
For 1 n = we have the solution
1
) x ( f
) x ( f
) x ( X
) x ( X
) 3 ( x
) x ( X
) x ( X
) x ( w

|
|

\
|

+ +
|
|

\
|

=
where

)
`

+ =

) x ( M
2
3
) x ( W
x
C
) x ( f
2
,
2
,
1
with
4
7
= ,
4
1
=
25
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( X
2
, 2
2
, 1
+ = with
4
10 +
= ,
4
1
=
For 2 n = we have the solution
1
) x ( f
) x ( f
) x ( X
) x ( X
) 5 ( x
) x ( X
) x ( X
) x ( w

|
|

\
|

+ +
|
|

\
|

=
where

)
`

=

) x ( M
4
5
) x ( W
x
C
) x ( f
2
,
2
,
1
with
4
11
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( X
2
, 2
2
, 1
+ = with
4
16 +
= ,
4
1
=
For 3 n = we have the solution
1
) x ( f
) x ( f
) x ( X
) x ( X
) 7 ( x
) x ( X
) x ( X
) x ( w

|
|

\
|

+ +
|
|

\
|

=
where

)
`

+ =

) x ( M
8
105
) x ( W
x
C
) x ( f
2
,
2
,
1
with
4
15
= ,
4
1
=
} ) x ( M C
~
) x ( W C
~
{
x
1
) x ( X
2
, 2
2
, 1
+ = with
4
22 +
= ,
4
1
=
6. Case III. Third Solution of the Painlev PIV Equation.
We continue considering the case
B a
1
= and A b
1
= (6.1)
From (3.9) and (3.15) we obtain
x 2
B
B
a 2
0

= and x 2
A
A
a 2
0

= (6.2)
respectively. Equating the two different expressions of
0
a and integrating, we find
that
B r A
2
= (6.3)
where r is real. From equation (3.10), using (6.1), we obtain
0 ) x ( 4 a x 24 a 2 a 18 B A 8
B
B
B
B
a 6
B
B
2
2
0 0
2
0
2
0
= +
|

\
|

+


26
The above equation, using the first of (6.2) and the relation (6.3), becomes
0 ) 1 2 x ( 2 B r 8
B
B
2
3
B
B
2 2 2
2
= + + +
|

\
|


(6.4)
From equation (3.14), using (6.1), we obtain
0 ) x ( 4 a x 24 a 2 a 18 B A 8
A
A
A
A
a 6
A
A
2
2
0 0
2
0
2
0
= +
|

\
|

+


The above equation, using the second of (6.2) and (6.3), takes on the form
0 ) 1 2 x ( 2 B r 8
A
A
2
3
A
A
2 2 2
2
= + + +
|

\
|


(6.5)
Equations (6.4) and (6.5) are equivalent in view of (6.3).
From equation (3.11), using (6.1), we obtain

3
0 0 0 0 0
a 12 B A 8 B A 4 B A a 40 a 2
B
B
a 2
B
B
a 2 + + +




0 ) a B A ( x 24 ) x ( a 8
2
0
2
0
= +
The previous equation, taking into account the first of (6.2) and (6.3) becomes

(
(

\
|

\
|

\
|

\
|

2 3
2
B
B
2
3
B
B
x 2
B
B
2
3
B
B
B
B
B
B B
B
B

0 ) 2 x ( x 4 B x r 16 B B r 24
B
B
) 1 2 x ( 2
2 2 2 2 2
= + +

+ + + (6.6)
From equation (3.13) we obtain

3
0 0 0 0 0
a 12 B A 4 B A 8 B A a 40 a 2
A
A
a 2
A
A
a 2 + + +




0 ) a B A ( x 24 ) a x ( a 8
2
0
2
0
= +
The above equation using the second of (6.2) and (6.3) takes on the form

A
A
) 1 2 x ( 2
A
A
2
3
A
A
x 2
A
A
2
3
A
A A
3
A
A
2
2 3
2

+ + +
(
(

\
|

\
|

+



0 ) 2 x ( x 4 B x r 16 B B r 24
2 2 2 2
= + + (6.7)
27
We now prove the compatibility of (6.4) and (6.6). These two equations expressed
in terms of

B
B
G

= (6.8)
take the form
0 ) 1 2 x ( 2 B r 8 G
2
1
G
2 2 2 2
= + + + (6.9)
and
B B r 24 G ) 1 2 x ( 2 G
2
1
G x 2 G
2
1
G
2 2 2 3
+ + +
|

\
|

0 ) 2 x ( x 4 B x r 16
2 2 2
= + + (6.10)
respectively. Substituting
2
G
2
1
G obtained from (6.9) into (6.10) we get the
equation
0 x 4 G ) 1 2 x ( 2 B B r 24 G
2
1
G
2 2 3
= + + + + (6.11)
Multiplying (6.9) by G and rearranging we obtain the equation
G ) 1 2 x ( 2 G B r 8 G G G
2
1
2 2 2 3
+ + + =
We substitute in (6.11) the above expression and get the equation
0 x 4 B B r 16 G G G
2
= + (6.12)
Upon differentiation of (6.9) we obtain
x 4 B B r 16 G G G
2
=
which when combined with (6.12) gives an identity. Therefore (6.9) and (6.10),
i.e. (6.4) and (6.6) are compatible each other.
Equations (6.5) and (6.7) expressed in terms of
A
A
H

= as
0 ) 1 2 x ( 2 B r 8 H
2
1
H
2 2 2 2
= + + +
and
28
H ) 1 2 x ( 2 H
2
1
H x 2 H
2
1
H
2 2 3
+ + +
|

\
|

0 ) 2 x ( x 4 B B r 24 B x r 16
2 2 2 2
= + +
respectively, are proved similarly to be compatible each other.
We now come into solving equation (6.4). Under the substitution

) x ( g
1
B
2
= (6.13)
since
g
g
2
B
B
=

and
2
g
g
6
g
g
2
B
B
|
|

\
|
+

=

, equation (6.4) becomes

3
2
2
g
r 4
g ) 1 2 x ( g = + + (6.14)
which is Ermakov's equation. Two linearly independent solutions of the equation
0 g ) 1 2 x ( g
2
= + +
are given by Whittaker's W and M functions
) x ( W
x
1
g
2
, 1
= and ) x ( M
x
1
g
2
, 2
= (6.15)
where

4
1 2 +
= ,
4
1
= (6.16)
According to the standard procedure (Ermakov [8]) the solutions of Ermakov's
equation (6.14) are given by

|
|

\
|
+ + =



2
2 2
,
1 2
2 2 2
,
2
1
dx
)) x ( W (
x
C C r 4 )) x ( W (
x
1
) x ( g C (6.17)
or

|
|

\
|
+ + =



2
2 2
,
1 2
2 2 2
,
2
1
dx
)) x ( M (
x
C C r 4 )) x ( M (
x
1
) x ( g C (6.18)
29
Equation (3.12) takes the form, because of (6.1) and the first of (6.2)

4
2
2 2
B
B
16
9
B
B B
2
1
B
B
B
B
B
B
4
1
B
B
|

\
|

+

+
|

\
|

\
|

\
|

+

\
|

+

A
A
B
B
2 ) B A ( 2
B
B
2
3
B
B B
3
B
B
x
3
2


|

\
|

+
|

\
|

+

B
B
A
A
) AB ( 5
B
B
) 2 x ( x 2 ) AB (
B
B
3
A
A
x 6
2

) AB ( 8 x 4 x
B
B
1 x
2
1
B
B
) AB ( 14
2 4
2
2
2
+ +
|

\
|

|

\
|
+ +
|

\
|

+
0 1 2 ) B A ( 32 ) AB ( 4 ) AB ( x 4
2 2
= +
The above equation, taking into account B r A
2
= and introducing the function
G by
B
B
G

= , takes the form

4 2 3 3
G
16
1
G ) G (
4
1
G
2
1
G x G
2
1
G G
2
1
+ +
|

\
|

|

\
|


2 2 2 2 2 2 2
G ) 2 x (
2
1
G ) 2 x ( x 2 B B r 6 B B x r 24 ) B ( r 19 + + + + +
0 1 2 x 4 x B r 32 B ) 1 2 x ( r 4
2 4 4 4 2 2 2
= + + + + (6.19)
Let us now determine the function Y which is the solution of Riccati's equation
2
Y B A Y + = . Under the standard substitution
u
u
B
1
Y

= Riccati's equation
takes the form 0 u AB u
B
B
u = +
|

\
|

which because of B r A
2
= becomes
30
0 u B r u
B
B
u
2 2
=
|

\
|

or dividing by B, 0 u B r
B
B u
B
u
2
2
=


. The last
equation can be written as 0 u B r
B
u
2
=

\
|

and then multiplying by
|

\
|

B
u
we get
the equation 0 u u r
B
u
B
u
2
=

\
|

|

\
|

which is equivalent to 0 ) u ( r
B
u
2 2
2
=

(
(

\
|


from which we get 0 u r
B
u
2 2
2
=
|

\
|

(the integration constant has been set equal
to zero). We thus have u r
B
u
=

and then B r
u
u
=

. Because of the last equality,


we obtain r ) B r (
B
1
u
u
B
1
Y = =

= .
We come next to consider the compatibility between the equations (6.6), (6.4) (i.e.
(6.9), (6.10)) and (6.19). Solving (6.9) with respect to
2
G
2
1
G , squaring both
members and rearranging, we have
= +
4 2 2
G
16
1
G G
4
1
) G (
4
1

2 2 2 2 2 4 4
) 1 2 x ( B ) 1 2 x ( r 8 B r 16 + + + + + = (6.20)
We add (6.19), (6.20) and simultaneously substitute
3
G
2
1
G derived from
(6.11) into (6.19). We then obtain the equation
B B r 6 ) B ( r 7 G
2
1
G G
2
1
G G
4
1
2 2 2 2 2 2
+
|

\
|
+
|

\
|

4 4 2 2 2 2 2
B r 48 B ) 1 2 x ( r 12 G ) 1 2 x (
2
1
+ + + + +
0 1 2 ) 1 2 x ( x 4 x 4 x
2 2 2 2 4
= + + + + + (6.21)
31
We substitute
2
G
2
1
G derived from (6.9) into the previous equation and we get
the equation

2 2 2 2 2 4 4
2
2 2
B r 8 B ) 1 2 x ( r 12 B r 48
B
B
2
3
B
B
B r 6 + + + +

\
|



0 ) 2 4 2 ( 2
2
= + + + (6.22)
In the above expression we substitute
2
B
B
2
3
B
B
|

\
|


derived from (6.4) and we get
the equation
0 ) 2 4 2 ( B r 4
2 2 2
= + + + (6.23)
which is the compatibility condition.
It is rather hard to find relations between the various parameters and the constants
due to the complexity of the solutions. It is not even known if there are any such
relations. One way out of this problem is to substitute the found expressions to the
original equation and check if they satisfy the equation, a task which may require
supercomputing facilities with symbolic capabilities.
Conclusion. The solution of the Painleve PIV equation

w
w ) x ( 2 w x 4 w
2
3
dx
dw
w 2
1
dx
w d
2 2 3
2
2
2

+ + + +
|

\
|
=
is given by
Y
b
Y a a ) x ( w
1
1 0
+ + = where
|
|

\
|
+

= x
) x ( g
) x ( g
x
B
B
2
1
a
0
,
) x ( g
1
) x ( B a
2
1
= = ,
) x ( g
r
B r ) x ( A b
2
2
2
1
= = =
and r Y = . Therefore we have the solutions

) x ( g
r 2
x
) x ( g
) x ( g
) x ( w
2
+
|
|

\
|
+

= if r Y =
32

) x ( g
r 2
x
) x ( g
) x ( g
) x ( w
2

|
|

\
|
+

= if r Y =
In both cases the function ) x ( g is given by (6.17) or (6.18) provided that the
compatibility condition (6.23) is satisfied.
7. Case IV. Fourth Solution of the Painlev PIV Equation.
We finally consider the case
B a
1
= and A b
1
= (7.1)
From (3.9) and (3.15) we obtain
x 2
B
B
a 2
0

= and x 2
A
A
a 2
0

= (7.2)
respectively. Equating the two different expressions of
0
a and integrating, we find
that
B r A
2
= (7.3)
where r is real. From equation (3.10), using (7.1), we obtain
0 ) x ( 4 a x 24 a 2 a 18 B A 8
B
B
B
B
a 6
B
B
2
2
0 0
2
0
2
0
= + +
|

\
|



From the above equation, taking into account the first of (7.2) and (7.3), we obtain
0 ) 1 2 x ( 2 B r 8
B
B
2
3
B
B
2 2 2
2
= + +
|

\
|


(7.4)
The previous equation can also be written in terms of
B
B
G

= as
0 ) 1 2 x ( 2 B r 8 G
2
1
G
2 2 2 2
= + + (7.5)
From equation (3.14), using (7.1), we obtain
0 ) x ( 4 a x 24 a 2 a 18 B A 8
A
A
A
A
a 6
A
A
2
2
0 0
2
0
2
0
= + +
|

\
|



From the above equation, taking into account the second of (7.2) and (7.3), we
obtain
33
0 ) 1 2 x ( 2 B r 8
A
A
2
3
A
A
2 2 2
2
= + +
|

\
|


(7.6)
Equations (7.4) and (7.6) are equivalent in view of (7.3).
From equation (3.11) we get
3
0 0 0 0 0
a 12 B A 8 B A 4 B A a 40 a 2
B
B
a 2
B
B
a 2 + + +




0 ) a B A ( x 24 ) x ( a 8
2
0
2
0
= +
From the above equation, taking into account the first of (7.2) and (7.3), we obtain

B
B
) 1 2 x ( 2
B
B
2
3
B
B
x 2
B
B
2
3
B
B B
3
B
B
2
2 3
2

+ +
(
(

\
|


+
|

\
|

+



0 ) 2 x ( x 4 B x r 16 B B r 24
2 2 2 2
= + + (7.7)
The previous equation can be expressed in terms of
B
B
G

= as
G ) 1 2 x ( 2 G
2
1
G x 2 G
2
1
G
2 2 3
+ +
|

\
|
+
0 ) 2 x ( x 4 B x r 16 B B r 24
2 2 2 2
= + + (7.8)
Substituting
|

\
|

2
G
2
1
G derived from (7.5) into (7.8) we obtain the equation
0 x 4 B B r 24 G ) 1 2 x ( 2 G
2
1
G
2 2 3
= + + + (7.9)
We can prove - using also (7.9) - that equations (7.5) and (7.8) are compatible
each other, i.e. every solution of (7.5) satisfies (7.8). The proof goes along the
lines of the previous Section.
From equation (3.13) we obtain using (7.1)

2
0 0
3
0 0 0 0
a x 12 B A a 20 a 6 a
A
A
a
A
A
a + +




0 ) x ( 4 B A 2 B A 4 B A x 12
2
= + +
The above equation, because of the second of (7.2) and (7.3) becomes
34

A
A
) 1 2 x ( 2
A
A
2
3
A
A
x 2
A
A
2
3
A
A A
3
A
A
2
2 3
2

+ +
(
(

\
|


+
|

\
|

+



0 ) 2 x ( x 4 B x r 16 B B r 24
2 2 2 2
= + + (7.10)
Equations (7.6) and (7.10) expressed in terms of
A
A
H

= as
0 ) 1 2 x ( 2 B r 8 H
2
1
H
2 2 2 2
= + +
and
H ) 1 2 x ( 2 H
2
1
H x 2 H
2
1
H
2 2 3
+ +
|

\
|
+
0 ) 2 x ( x 4 B B r 24 B x r 16
2 2 2 2
= + +
respectively, are proved similarly to be compatible each other.
We now come into solving equation (7.4). Under the substitution

) x ( g
1
B
2
= (7.11)
since
g
g
2
B
B
=

and
2
g
g
6
g
g
2
B
B
|
|

\
|
+

=

, equation (7.4) becomes

3
2
2
g
r 4
g ) 1 2 x ( g = + (7.12)
which is Ermakov's equation. Two linearly independent solutions of the equation
0 g ) 1 2 x ( g
2
= +
are given by Whittaker's W and M functions
) x ( W
x
1
g
2
, 1
= and ) x ( M
x
1
g
2
, 2
= (7.13)
where

4
1 2
= ,
4
1
= (7.14)
35
According to the standard procedure (Ermakov [8]) the solutions of Ermakov's
equation (7.12) are given by

|
|

\
|
+ + =



2
2 2
,
1 2
2 2 2
,
2
1
dx
)) x ( W (
x
C C r 4 )) x ( W (
x
1
) x ( g C (7.15)
or

|
|

\
|
+ + =



2
2 2
,
1 2
2 2 2
,
2
1
dx
)) x ( M (
x
C C r 4 )) x ( M (
x
1
) x ( g C (7.16)
We find, using the method of the previous Section that the solution of Riccati's
equation is given by
r Y = (7.17)
Equation (3.12) takes the form, because of (7.1) and the first of (7.2)

4
2
2 2
B
B
16
9
B
B B
2
1
B
B
B
B
B
B
4
1
B
B
|

\
|

+

+
|

\
|

\
|

\
|

+

\
|

+


+
A
A
B
B
2 ) B A ( 2
B
B
2
3
B
B B
3
B
B
x
3
2


|

\
|

+ +
|

\
|

+

+
B
B
A
A
) AB ( 5
B
B
) 2 x ( x 2 ) AB (
B
B
3
A
A
x 6
2

) AB ( 8 x 4 x
B
B
1 x
2
1
B
B
) AB ( 14
2 4
2
2
2
+ +
|

\
|

|

\
|
+ + +
|

\
|

+
0 1 2 ) B A ( 32 ) AB ( 4 ) AB ( x 4
2 2
=
The above equation, taking into account B r A
2
= and introducing the function G
by
B
B
G

= , takes the form

4 2 3 3
G
16
1
G ) G (
4
1
G
2
1
G x G
2
1
G G
2
1
+
|

\
|
+
|

\
|

36

2 2 2 2 2 2 2
G ) 2 x (
2
1
G ) 2 x ( x 2 B B r 6 B B x r 24 ) B ( r 19 + + + + +
0 1 2 x 4 x B r 32 B ) 1 2 x ( r 4
2 4 4 4 2 2 2
= + + + + (7.18)
Solving (7.5) with respect to
2
G
2
1
G , squaring both members and rearranging,
we have
= +
4 2 2
G
16
1
G G
4
1
) G (
4
1

2 2 2 2 2 4 4
) 1 2 x ( B ) 1 2 x ( r 8 B r 16 + + + = (7.19)
We add (7.18), (7.19) and simultaneously substitute
3
G
2
1
G derived from (7.9)
into (7.18). We then obtain the equation
B B r 6 ) B ( r 7 G
2
1
G G
2
1
G G
4
1
2 2 2 2 2 2
+
|

\
|

|

\
|

4 4 2 2 2 2 2
B r 48 B ) 1 2 x ( r 12 G ) 1 2 x (
2
1
+ + +
0 1 2 ) 1 2 x ( x 4 x 4 x
2 2 2 2 4
= + + +
In the above equation we substitute
2
G
2
1
G derived from (7.5) and we obtain
the equation

4 4 2 2 2 2 2
2
2 2
B r 48 B r 8 B ) 1 2 x ( r 12
B
B
2
3
B
B
B r 6 + +

\
|



0 ) 2 4 2 ( 2
2
= + +
Upon substituting
2
B
B
2
3
B
B
|

\
|


derived from (7.4) into the above equation we
obtain
0 ) 2 4 2 ( B r 4
2 2 2
= + + + (7.20)
which is the compatibility condition.
37
It is rather hard to find relations between the various parameters and the constants
due to the complexity of the solutions. It is not even known if there are any such
relations. One way out of this problem is to substitute the found expressions to the
original equation and check if they satisfy the equation, a task which may require
supercomputing facilities with symbolic capabilities.
Conclusion. The solution of the Painleve PIV equation

w
w ) x ( 2 w x 4 w
2
3
dx
dw
w 2
1
dx
w d
2 2 3
2
2
2

+ + + +
|

\
|
=
is given by
Y
b
Y a a ) x ( w
1
1 0
+ + = where
x
) x ( g
) x ( g
x
B
B
2
1
a
0

= ,
) x ( g
1
) x ( B a
2
1
= = ,
) x ( g
r
B r ) x ( A b
2
2
2
1
= = =
and r Y = . Therefore we have the solutions

) x ( g
r 2
x
) x ( g
) x ( g
) x ( w
2
+

= if r Y =

) x ( g
r 2
x
) x ( g
) x ( g
) x ( w
2

= if r Y =
In both cases the function ) x ( g is given by (7.15) or (7.16) provided that the
compatibility condition (7.20) is satisfied.
8. The
|

\
|

G
G
expansion method with variable expansion
coefficients.
We consider that equation (3.1) admits a solution of the form

|
|

\
|
+ =
) x ( G
) x ( G
) x ( b ) x ( b ) x ( w
1 0
(8.1)
Upon substituting (8.1) into (3.1) we obtain an equation, which when arranged in
powers of G, takes the form
2 b ) x ( 4 b x 8 b 3 ) b ( b b 2
2
0
2 3
0
4
0
2
0 0 0

38

+ + +


+ + ) x ( b b 4 b b 6 b b x 12
G
G
) b b 4 b b (
) x ( G
2
2
0 1
3
0 1
2
0 1 1 0 0 1

) x ( G b b 2
G
G
b b 2 b b 2 b
1 0 1 0 0 1 1

(

(
+


+ + +

\
|


+ +
2
1 1
2 2
1
2
1 0 1 0 0 1
2
G
G
b b ) x ( b 4 ) b b ( 18 b b 4 b b 2
] ) x ( G [
1


2 2
1 0 1 1
2
1 0 1 1
)) x ( G (
G
G
b 2
G
G
) b 3 b 2 ( b 2 b b x 24 b b 2
(


+ +

3 2
1 1 1
3
1 0
3
1 0 1
3
)) x ( G (
G
G
b 4 b b 2 b b 12 b x 8 b b 4
] ) x ( G [
1


+

4 2
1
2
1
4
)) x ( G ( ) b 1 ( b
] ) x ( G [
3
+ (8.2)
Equating to zero all the coefficients of ) x ( G in the above equation, we obtain the
following system of ordinary differential equations
0 2 b ) x ( 4 b x 8 b 3 ) b ( b b 2
2
0
2 3
0
4
0
2
0 0 0
= (8.3)
) x ( b b 4 b b 6 b b x 12
G
G
) b b 4 b b (
2
0 1
3
0 1
2
0 1 1 0 0 1
+ + +


+ +
0 b b 2
G
G
b b 2 b b 2 b
1 0 1 0 0 1 1
= +


+ + + (8.4)

2
1 1
2 2
1
2
1 0 1 0 0 1
G
G
b b ) x ( b 4 ) b b ( 18 b b 4 b b 2
|

\
|


+
0
G
G
b 2
G
G
) b 3 b 2 ( b 2 b b x 24 b b 2
2
1 0 1 1
2
1 0 1 1
=


+ + (8.5)
0
G
G
b 4 b b 2 b b 12 b x 8 b b 4
2
1 1 1
3
1 0
3
1 0 1
=


(8.6)
0 b 1
2
1
= (8.7)
Equation (8.3) is essentially equation (3.1) and thus admits all the solutions found
in previous Sections. From equation (8.7), we obtain that
39
1 b
1
= (8.8)
From equation (8.6), taking into account the above values of
1
b , we derive the
equations
) b x ( b 2
G
G
0 1
+ =


(8.9)
From equation (8.4), we obtain
0 b 3 b x 6 ) x ( 2
b
b
G
G
b 2
b
G
G
2
0 0
2
0
0
0
0
= + + +

+


(8.10)
From equation (8.5), we obtain
0 b 9 b x 12 ) x ( 2
b
b
G
G
2
1
G
G
b
b
3
G
G
2
0 0
2
1
0
2
1
0
=

+
|

\
|


(8.11)
From equations (8.10) and (8.11), we obtain because of (8.9), the following two
expressions for the ratio
G
G



+
+
+

=


2 x 2 b 3 b x 6
b
b ) b x (
b
b
b
G
G
2 2
0 0
0
0 0
1
0
0
(8.12)
and
+ + +


2 x 4 b x 10 b 5
b
b
G
G
2
0
2
0
1
0
(8.13)
respectively.
Upon equating the two expressions (8.12) and (8.13), we obtain the following
differential equation for the coefficient
0
b :
0 ) 4 x 6 b x 16 b 8 ( b b b x b b 2 b b
2
0
2
0 0 1 0 0 0 0 1
= + + + (8.14)
We thus have to solve the following two systems depending on the value of the
coefficient
1
b :
(I) For 1 b
1
= , we have
) b x ( 2
G
G
0
+ =


(8.15)
40
+ + + =


2 x 4 b x 10 b 5 b
G
G
2
0
2
0 0
(8.16)
0 b 8 b x 16 b ) 2 x 3 ( 2 b x b b 2 b
3
0
2
0 0
2
0 0 0 0
= + + + (8.17)
(II) For 1 b
1
= , we have
) b x ( 2
G
G
0
+ =


(8.18)
+ + + =


2 x 4 b x 10 b 5 b
G
G
2
0
2
0 0
(8.19)
0 b 8 b x 16 b ) 2 x 3 ( 2 b x b b 2 b
3
0
2
0 0
2
0 0 0 0
= + + + + + (8.20)
Each one of the above systems should be supplemented by equation (8.3).
Upon elimination of
0
b between (8.3) and (8.17), (8.20) we obtain the equations
0 2 b ) 3 x 4 ( 4 b x 40 b 19 b ) x b 2 ( b 2 ) b (
2
0
2 3
0
4
0 0 0 0
2
0
= + + + + + (8.21)
and
0 2 b ) 3 x 4 ( 4 b x 40 b 19 b ) x b 2 ( b 2 ) b (
2
0
2 3
0
4
0 0 0 0
2
0
= + + + + + + (8.22)
respectively. These two equations are the compatibility conditions. Every solution
which satisfies (8.3) should satisfy each one of (8.21) and (8.22).
We thus have the two solutions:
Solution 1. This solution corresponds to 1 b
1
= . Dividing (8.16) by (8.15) we
obtain

) b x ( 2
2 x 4 b x 10 b 5 b
G
G
0
2
0
2
0 0
+
+ + +
=


(8.23)
where
0
b satisfies the compatibility condition (8.21). Solving (8.23) we determine
) x ( G .
Solution 2. This solution corresponds to 1 b
1
= . Dividing (8.19) by (8.18) we
obtain

) b x ( 2
2 x 4 b x 10 b 5 b
G
G
0
2
0
2
0 0
+
+ + +
=


(8.24)
41
where
0
b satisfies the compatibility condition (8.22). Solving (8.24) we determine
) x ( G .
Because of the complexity of the expressions for
0
b , the integration of the
differential equations (8.23) and (8.24) might require supercomputing facilities
with symbolic capabilities. A remarkable feature of the ) G / G ( expansion with
variable expansion coefficients is that a repeated use of the method in determining
0
b leads to a proliferation of solutions, a rather unique feature of the Riccati
) G / G ( expansion.
Appendix A.
We examine the first compatibility condition given by equation (4.33). This
equation splits into a pair of equations
0 ) x ( u p 2 2 x ) x ( u
2
= + (A.1)
and
0 ) x ( u p 2 2 x ) x ( u
2
= + + (A.2)
where ) x ( u is given by (4.20).
Expanding (A.1) into power series we obtain

|
|
|
|

\
|
|

\
| +

+
|

\
| +


1 2 1
C
4
1
2
C C
4
3
) p 2 2 (
x C
4
1
2
C p 2 2 C
4
3
1 2 1

|
|
|
|

\
|
|

\
| +

\
| +

+
42
2
1 2 1
x C
4
1
2
C p 2 2 C
4
3
] 1 ) p ( 2 [
p 2 2 2
1

|
|
|
|

\
|
|

\
| +

\
| +

) x ( O
3
+ (A.3)
Expanding (A.2) into power series we obtain

|
|
|
|

\
|
|

\
|
+

+
|

\
|
+


1 2 1
C
4
1
2
C C
4
3
) p 2 2 (
x C
4
1
2
C p 2 2 C
4
3
1 2 1

|
|
|
|

\
|
|

\
| +

\
| +

+
2
1 2 1
x C
4
1
2
C p 2 2 C
4
3
] 1 ) p ( 2 [
p 2 2 2
1

|
|
|
|

\
|
|

\
| +

+
|

\
| +

+

+
) x ( O
3
+ (A.4)
We next consider the second compatibility condition (4.38):
0 ) x ( u k ] ) x ( u x ) x ( u [ x p 4 = + (A.5)
Taking into account the solution ) x ( u given by (4.20) and expanding, we get from
(A.5)
2
1 1 2 1
x C
4
3
2
] p 8 ) p 8 k [(
x C
4
1
2
C ) p 4 k ( C
4
3
k
|

\
| +

|
|
|
|

\
|
|

\
| +

+ + +
|

\
| +


3
1 2
x C
4
1
2
C ] p 24 ) p 12 k [(
6
1
|
|
|
|

\
|
|

\
|
+

+ + +
43
4
1
2
x C
4
3
24
] k 2 p 32 p 48 ) p 16 k [(
|

\
| +

+ + +

) x ( O x C
4
1
2
C ] k 6 p 120 p 80 ) p 20 k [(
120
1
6 5
1 2
2
+
|
|
|
|

\
|
|

\
| +

+ + + + + (A.6)
A remarkable feature of the above compatibility conditions, expressed by the
expansions (A.3), (A.4) and (A.6) is that there are two patterns appearing
repeatedly:
|

\
| +

4
3
1
and
|
|
|
|

\
|
|

\
|
+

+
1 2
C
4
1
2
C
Therefore the compatibility conditions hold for every x if
n 4 3 = + ( L , 3 , 2 , 1 , 0 n = ) and 0 C
4
1
2
C
1 2
=
|

\
| +

+ .
It is 0
4
3
1
=
|

\
|
+

for n 4 3 = + ( L , 3 , 2 , 1 , 0 n = ) in view of the fact that the


function ) z ( has a pole at 0 z = and to the set of negative integers (Abramowitz
and Stegun [3], 6.1.7):
0
) z (
1
lim
n z
=

, L , 3 , 2 , 1 , 0 n =
Appendix B. We examine the first consistency condition (5.27) written as a
couple of equations
0 ) x ( f p 2 2 x ) x ( f
2
= + (B.1)
and
44
0 ) x ( f p 2 2 x ) x ( f
2
= + + (B.2)
where ) x ( f is given by (5.20).
Expanding (B.1) into power series we obtain

|
|
|
|

\
|
|

\
| +

+
|

\
| +


1 2 1
C
4
1
2
C C
4
3
) p 2 2 (
x C
4
1
2
C p 2 2 C
4
3
1 2 1

|
|
|
|

\
|
|

\
| +

\
| +

+
2
1 2 1
x C
4
1
2
C p 2 2 C
4
3
] 1 ) p ( 2 [
p 2 2 2
1

|
|
|
|

\
|
|

\
| +

\
| +

) x ( O
3
+ (B.3)
Expanding (B.2) into power series we obtain

|
|
|
|

\
|
|

\
| +

+
|

\
| +


1 2 1
C
4
1
2
C C
4
3
) p 2 2 (
x C
4
1
2
C p 2 2 C
4
3
1 2 1

|
|
|
|

\
|
|

\
| +

\
| +

+
2
1 2 1
x C
4
1
2
C p 2 2 C
4
3
] 1 ) p ( 2 [
p 2 2 2
1

|
|
|
|

\
|
|

\
| +

+
|

\
| +

+

+
) x ( O
3
+ (B.4)
45
We next consider the consistency condition (5.16). This condition in view of
f
f
2 F

= , takes the form
0 ) x ( f k )] x ( f x ) x ( f [ x p 4 = + (B.5)
where
2 4 2 p ) 8 4 ( p 6 k
2 2
+ + + + = (B.6)
and ) x ( f is given by (5.20).
Upon expanding ) x ( f we obtain from (B.5)
2 1
1 2
1
x
4
3
C
] p 8 ) p 8 k ( [
2
1
x C
4
1
2
C ) k p 4 (
4
3
C
k
|

\
| +


|
|
|
|

\
|
|

\
| +

+
|

\
| +


3
1 2
x C
4
1
2
C ] p 24 ) k p 12 [(
6
1
|
|
|
|

\
|
|

\
| +

+ +
4 1 2
x
4
3
C
] p 32 k 2 p 48 ) p 16 k [(
24
1
|

\
| +

+
) x ( O x C
4
1
2
C ] k 6 p 120 p 80 ) k p 20 [(
120
1
6 5
1 2
2
+
|
|
|
|

\
|
|

\
| +

+ + + (B.6)
We see that there are two patterns appearing repeatedly:
|

\
|
+

4
3
1
and
|
|
|
|

\
|
|

\
| +

1 2
C
4
1
2
C
Therefore expression (B.5) holds for every x if
46
n 4 3 = + ( L , 3 , 2 , 1 , 0 n = ) and 0 C
4
1
2
C
1 2
=
|

\
|
+



References
[1] M. J. Ablowitz, A. Ramani and H. Segur: "A connection between
nonlinear evolution equations and ordinary differential equations of
P-type. I" J. Math. Phys. 21 (1980) 715-721
[2] M. J. Ablowitz, A. Ramani and H. Segur: "A connection between
nonlinear evolution equations and ordinary differential equations of
P-type. II" J. Math. Phys. 21 (1980) 1006-1015
[3] M. Abramowitz and I. A. Stegun: "Handbook of Mathematical
Functions". Dover 1972
[4] H. Airault: "Rational solutions of Painlev equations"
Stud. Appl. Math. 61 (1979) 31-53
[5] S. Antoniou: The Riccati equation method with variable expansion
coefficients. I. Solving the Burgers equation. submitted for publication
[6] S. Antoniou: Some General Solutions to the Painlev PII Equation.
submitted for publication
[7] P. A. Clarkson: "Painlev equations-nonlinear special functions"
J. Comp. Appl. Math. 153 (2003) 127-140
[8] V. P. Ermakov: Second-Order Differential Equations: Conditions of
Complete Integrability. Appl. Anal. Discr. Math. 2 (2008) 123-145
Translation from the original Russian article:
Universitetskiye Izvestiya Kiev No. 9 (1880) 1-25
[9] A. S. Fokas and M. J. Ablowitz: "On a unified approach to
transformations and elementary solutions of Painlev equations"
J. Math. Phys. 23 (1982) 2033-2042

47
[10] E. L. Ince: "Ordinary Differential Equations". Dover 1956
[11] A. Ramani, B. Dorizzi and B. Grammaticos: "Painlev conjecture
revisited". J. Math. Phys. 21 (1980) 715-721
[12] A. Ramani, B. Grammaticos and T. Bountis: "The Painleve Property and
Singularity Analysis of Integrable and Non-Integrable Systems".
Phys. Rep. 180 (1989) 159-245
[13] J. Weiss, M. Tabor and G. Carnevale: "The Painlev property for partial
differential equations". J. Math. Phys. 24 (1983) 522-526
[14] J. Weiss: "The Painlev property for partial differential equations. II:
Bcklund transformation, Lax pairs, and the Schwarzian derivative"
J. Math. Phys. 24 (1983) 1405-1413
[15] J. Weiss: "On classes of integrable systems and the Painlev property "
J. Math. Phys. 25 (1984) 13-24
[16] J. Weiss: "The Painlev property and Bcklund transformations for the
sequence of Boussinesq equations".
J. Math. Phys. 26 (1985) 258-269
[17] Y. Wenjum and Li Yezhou: "Rational Solutions of Painlev Equations"
Canadian J. Math. 54 (2002) 648-670

You might also like