Introduction
In this paper we are concerned with Bäcklund transformations associated with the fifth Painlevé equation
d 2 w d z 2 = ( 1 2 w + 1 w − 1 ) ( d w d z ) 2 − 1 z d w d z + ( w − 1 ) 2 ( α w 2 + β ) z 2 w + γ w z + δ w ( w + 1 ) w − 1 (1) \frac{\mathrm{d}^{2}w}{\mathrm{d}z^{2}}=\left(\frac{1}{2w}+\frac{1}{w-1}\right)\left(\frac{\mathrm{d}w}{\mathrm{d}z}\right)^{2}-\frac{1}{z}\frac{\mathrm{d}w}{\mathrm{d}z}+\frac{(w-1)^{2}(\alpha w^{2}+\beta)}{z^{2}w}+\frac{\gamma w}{z}+\frac{\delta w(w+1)}{w-1} \tag*{(1)} d z 2 d 2 w = ( 2 w 1 + w − 1 1 ) ( d z d w ) 2 − z 1 d z d w + z 2 w ( w − 1 ) 2 ( α w 2 + β ) + z γ w + w − 1 δ w ( w + 1 ) (1)
with α \alpha α , β \beta β , γ \gamma γ and δ \delta δ constants. Primarily we study the generic case of (1) when δ ≠ 0 \delta\ne0 δ = 0 , so we set δ = − 1 2 \delta=-\frac{1}{2} δ = − 2 1 , without loss of generality (by rescaling z z z if necessary) and obtain
d 2 w d z 2 = ( 1 2 w + 1 w − 1 ) ( d w d z ) 2 − 1 z d w d z + ( w − 1 ) 2 ( α w 2 + β ) z 2 w + γ w z − w ( w + 1 ) 2 ( w − 1 ) (2) \frac{\mathrm{d}^{2}w}{\mathrm{d}z^{2}}=\left(\frac{1}{2w}+\frac{1}{w-1}\right)\left(\frac{\mathrm{d}w}{\mathrm{d}z}\right)^{2}-\frac{1}{z}\frac{\mathrm{d}w}{\mathrm{d}z}+\frac{(w-1)^{2}(\alpha w^{2}+\beta)}{z^{2}w}+\frac{\gamma w}{z}-\frac{w(w+1)}{2(w-1)} \tag*{(2)} d z 2 d 2 w = ( 2 w 1 + w − 1 1 ) ( d z d w ) 2 − z 1 d z d w + z 2 w ( w − 1 ) 2 ( α w 2 + β ) + z γ w − 2 ( w − 1 ) w ( w + 1 ) (2)
which henceforth we refer to as P V \mathrm{P}_{\mathrm{V}} P V . The case when δ = 0 \delta=0 δ = 0 in (1) ,
d 2 w d z 2 = ( 1 2 w + 1 w − 1 ) ( d w d z ) 2 − 1 z d w d z + ( w − 1 ) 2 ( α w 2 + β ) z 2 w + γ w z (3) \frac{\mathrm{d}^{2}w}{\mathrm{d}z^{2}}=\left(\frac{1}{2w}+\frac{1}{w-1}\right)\left(\frac{\mathrm{d}w}{\mathrm{d}z}\right)^{2}-\frac{1}{z}\frac{\mathrm{d}w}{\mathrm{d}z}+\frac{(w-1)^{2}(\alpha w^{2}+\beta)}{z^{2}w}+\frac{\gamma w}{z} \tag*{(3)} d z 2 d 2 w = ( 2 w 1 + w − 1 1 ) ( d z d w ) 2 − z 1 d z d w + z 2 w ( w − 1 ) 2 ( α w 2 + β ) + z γ w (3)
is known as the degenerate fifth Painlevé equation (deg-P V \mathrm{P}_{\mathrm{V}} P V ) and is related to the third Painlevé equation (P I I I \mathrm{P}_{\mathrm{III}} P III )
d 2 q d x 2 = 1 q ( d q d x ) 2 − 1 x d q d x + A q 2 + B x + C q 3 + D q (4) \frac{\mathrm{d}^{2}q}{\mathrm{d}x^{2}}=\frac{1}{q}\left(\frac{\mathrm{d}q}{\mathrm{d}x}\right)^{2}-\frac{1}{x}\frac{\mathrm{d}q}{\mathrm{d}x}+\frac{Aq^{2}+B}{x}+Cq^{3}+\frac{D}{q} \tag*{(4)} d x 2 d 2 q = q 1 ( d x d q ) 2 − x 1 d x d q + x A q 2 + B + C q 3 + q D (4)
where A A A , B B B , C C C and D D D are constants. If γ = δ = 0 \gamma=\delta=0 γ = δ = 0 in (1) , then it has the first integral
z 2 ( d w d z ) 2 = 2 ( w − 1 ) 2 ( α w 2 + κ w − β ) , z^{2}\left(\frac{\mathrm{d}w}{\mathrm{d}z}\right)^{2}=2(w-1)^{2}(\alpha w^{2}+\kappa w-\beta), z 2 ( d z d w ) 2 = 2 ( w − 1 ) 2 ( α w 2 + κ w − β ) ,
with κ \kappa κ a constant of integration, which is solvable in terms of elementary functions.
An important application of Bäcklund transformations of Painlevé equations is to generate hierarchies of solutions. Using a “seed solution”, typically a simple solution of the Painlevé equation, the successive applications of Bäcklund transformations generates a hierarchy of solutions which frequently can be expressed as a determinant, cf. [30 , 62 ] . A second application of Bäcklund transformations of Painlevé equations is to derive discrete Painlevé equations. For P V \mathrm{P}_{\mathrm{V}} P V (2) associated discrete equations are derived in [34 , 46 , 104 ] . Bäcklund transformations also provide a structure for the parameter space of Painlevé equations. Okamoto Okamoto [96 ] showed that the Bäcklund transformations of P V \mathrm{P}_{\mathrm{V}} P V (2) form the extended affine Weyl group W ~ ( A 3 ( 1 ) ) \widetilde{W}(A_{3}^{(1)}) W ( A 3 ( 1 ) ) , which is discussed further in § 3.2 .
The fifth Painlevé equation (1) arises in numerous applications including: general relativity [71 , 72 , 73 , 74 , 75 , 76 , 77 , 79 , 86 ] ; number theory [15 ] ; orthogonal polynomials [11 , 12 , 13 , 18 , 19 , 20 , 24 , 25 , 31 , 32 , 41 , 42 , 43 , 44 , 109 , 113 , 114 ] ; random matrices [2 , 4 , 5 , 6 , 7 , 10 , 14 , 21 , 22 , 26 , 27 , 28 , 49 , 50 , 51 , 52 , 53 , 54 , 106 , 112 ] ; statistical physics [3 , 39 , 64 , 68 ] ; supersymmetric quantum mechanics [16 , 17 , 40 , 80 , 85 ] ; and vortex solutions of complex sine-Gordon equations [9 , 32 , 100 ] . We remark that several of these applications involve the P V \mathrm{P}_{\mathrm{V}} P V σ \sigma σ -equation. It is the second-order, second-degree differential equation derived in [63 , 93 , 94 , 96 ] which is associated with the Hamiltonian representation of P V \mathrm{P}_{\mathrm{V}} P V given by
( z d 2 σ d z 2 ) 2 = [ 2 ( d σ d z ) 2 − z d σ d z + σ ] 2 − 4 ∏ j = 0 3 ( d σ d z + κ j ) (5) \left(z\frac{\mathrm{d}^{2}\sigma}{\mathrm{d}z^{2}}\right)^{2}=\left[2\left(\frac{\mathrm{d}\sigma}{\mathrm{d}z}\right)^{2}-z\frac{\mathrm{d}\sigma}{\mathrm{d}z}+\sigma\right]^{2}-4\prod_{j=0}^{3}\left(\frac{\mathrm{d}\sigma}{\mathrm{d}z}+\kappa_{j}\right) \tag*{(5)} ( z d z 2 d 2 σ ) 2 = [ 2 ( d z d σ ) 2 − z d z d σ + σ ] 2 − 4 j = 0 ∏ 3 ( d z d σ + κ j ) (5)
with κ 0 \kappa_{0} κ 0 , κ 1 \kappa_{1} κ 1 , κ 2 \kappa_{2} κ 2 and κ 3 \kappa_{3} κ 3 constants such that κ 0 + κ 1 + κ 2 + κ 3 = 0 \kappa_{0}+\kappa_{1}+\kappa_{2}+\kappa_{3}=0 κ 0 + κ 1 + κ 2 + κ 3 = 0 , whose solution is expressible in terms of the solution of P V \mathrm{P}_{\mathrm{V}} P V (2) .
The fifth Painlevé equation (1) also arises as a symmetry reduction of various nonlinear partial differential equations including: the Einstein-Weyl equation [102 , 103 , 105 ] ; the Maxwell-Bloch system with pumping [23 ] ; the three-wave resonant interaction system [65 , 81 ] ; the modified Yajima-Oikawa equation [69 ] ; shallow water wave equations [36 , 37 ] ; the 2 + 1 2+1 2 + 1 -dimensional Sine-Gordon equation [38 ] ; and stimulated-Raman-scattering equations [78 ] .
The outline of this paper is as follows. In § 2 we describe Bäcklund and Schlesinger transformations of P V \mathrm{P}_{\mathrm{V}} P V (2) . In § 4 we apply Bäcklund and Schlesinger transformations to rational solutions of P V \mathrm{P}_{\mathrm{V}} P V (2) . In § 5 we show that for certain values of the parameters there are non-unique solutions of P V \mathrm{P}_{\mathrm{V}} P V (2) which are expressed in terms of modified Bessel functions. Further we show that for these parameters there also is a rational solution which is not a special case of either modified Bessel function solution. In §5 we discuss the relationship between solutions of the third and fifth Painlevé equations. In § 7 we discuss our results.
Bäcklund and Schlesinger transformations
Bäcklund transformations
In this section we describe the Bäcklund transformations of PV _{\mathrm{V}} V (2) .
Definition 2.1. A Bäcklund transformation maps a solution of a given Painlevé equation to another solution of the same Painlevé equation, with different values of the parameters, or another equation.
First we note that solutions of PV _{\mathrm{V}} V (2) have the symmetries
S 1 : w 1 ( z ; α 1 , β 1 , γ 1 ) = w ( − z ; α , β , γ ) , ( α 1 , β 1 , γ 1 ) = ( α , β , − γ ) , S 2 : w 2 ( z ; α 2 , β 2 , γ 2 ) = 1 w ( z ; α , β , γ ) , ( α 2 , β 2 , γ 2 ) = ( − β , − α , − γ ) . (6) \begin{aligned}
S_1 &: \quad w_1(z;\alpha_1,\beta_1,\gamma_1)=w(-z;\alpha,\beta,\gamma), \qquad(\alpha_1,\beta_1,\gamma_1)=(\alpha,\beta,-\gamma),\\
S_2 &: \quad w_2(z;\alpha_2,\beta_2,\gamma_2)=\frac{1}{w(z;\alpha,\beta,\gamma)}, \qquad(\alpha_2,\beta_2,\gamma_2)=(-\beta,-\alpha,-\gamma). \tag*{(6)}
\end{aligned} S 1 S 2 : w 1 ( z ; α 1 , β 1 , γ 1 ) = w ( − z ; α , β , γ ) , ( α 1 , β 1 , γ 1 ) = ( α , β , − γ ) , : w 2 ( z ; α 2 , β 2 , γ 2 ) = w ( z ; α , β , γ ) 1 , ( α 2 , β 2 , γ 2 ) = ( − β , − α , − γ ) . (6)
where w ( z ; α , β , γ ) w(z;\alpha,\beta,\gamma) w ( z ; α , β , γ ) is a solution of PV _{\mathrm{V}} V (2) . If two solutions of PV _{\mathrm{V}} V are related through these symmetries then we regard them as equivalent as essentially they are the same solution.
Definition 2.2. Let w = w ( z ; α , β , γ ) w=w(z;\alpha,\beta,\gamma) w = w ( z ; α , β , γ ) be a solution of PV _{\mathrm{V}} V (2) such that
Φ ( w ; z ) = z d w d z − ε 2 a w 2 + ( ε 2 a − ε 3 b + ε 1 z ) w + ε 3 b ≠ 0 , \Phi(w;z)=z\frac{\mathrm{d}w}{\mathrm{d}z}-\varepsilon_2aw^2+(\varepsilon_2a-\varepsilon_3b+\varepsilon_1z)w+\varepsilon_3b\ne0, Φ ( w ; z ) = z d z d w − ε 2 a w 2 + ( ε 2 a − ε 3 b + ε 1 z ) w + ε 3 b = 0 , and
Φ ( w ; z ) − 2 ε 1 z w ≠ 0 , \Phi(w;z)-2\varepsilon_1zw\ne0, Φ ( w ; z ) − 2 ε 1 z w = 0 , where a = 2 α a=\sqrt{2\alpha} a = 2 α and b = − 2 β b=\sqrt{-2\beta} b = − 2 β and c = γ c=\gamma c = γ . Then the Bäcklund transformation T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 is defined by
T ε 1 , ε 2 , ε 3 : W ( z ; A , B , C ) = 1 − 2 ε 1 z w Φ ( w ; z ) = 1 − 2 ε 1 z w z w ′ − ε 2 a w 2 + ( ε 2 a − ε 3 b + ε 1 z ) w + ε 3 b , (7) \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3}:\quad W(z;A,B,C)=1-\frac{2\varepsilon_1zw}{\Phi(w;z)}=1-\frac{2\varepsilon_1zw}{zw'-\varepsilon_2aw^2+(\varepsilon_2a-\varepsilon_3b+\varepsilon_1z)w+\varepsilon_3b}, \tag*{(7)} T ε 1 , ε 2 , ε 3 : W ( z ; A , B , C ) = 1 − Φ ( w ; z ) 2 ε 1 z w = 1 − z w ′ − ε 2 a w 2 + ( ε 2 a − ε 3 b + ε 1 z ) w + ε 3 b 2 ε 1 z w , (7) with w ′ ≡ d / d z w'\equiv\mathrm{d}/\mathrm{d}z w ′ ≡ d / d z , ε j = ± 1 \varepsilon_j=\pm1 ε j = ± 1 , j = 1 , 2 , 3 j=1,2,3 j = 1 , 2 , 3 , independently, and for the parameters
T ε 1 , ε 2 , ε 3 ( a , b , c ) = ( 1 2 [ c + ε 1 ( 1 − ε 3 b − ε 2 a ) ] , 1 2 [ c − ε 1 ( 1 − ε 3 b − ε 2 a ) ] , ε 1 ( ε 3 b − ε 2 a ) ) , (8) \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3}(a,b,c)=\left(\frac{1}{2}[c+\varepsilon_1(1-\varepsilon_3b-\varepsilon_2a)],\frac{1}{2}[c-\varepsilon_1(1-\varepsilon_3b-\varepsilon_2a)],\varepsilon_1(\varepsilon_3b-\varepsilon_2a)\right), \tag*{(8)} T ε 1 , ε 2 , ε 3 ( a , b , c ) = ( 2 1 [ c + ε 1 ( 1 − ε 3 b − ε 2 a )] , 2 1 [ c − ε 1 ( 1 − ε 3 b − ε 2 a )] , ε 1 ( ε 3 b − ε 2 a ) ) , (8) so
( A , B , C ) = ( 1 8 [ c + ε 1 ( 1 − ε 3 b − ε 2 a ) ] 2 , 1 8 [ c − ε 1 ( 1 − ε 3 b − ε 2 a ) ] 2 , ε 1 ( ε 3 b − ε 2 a ) ) , (A,B,C)=\left(\frac{1}{8}[c+\varepsilon_1(1-\varepsilon_3b-\varepsilon_2a)]^2,\frac{1}{8}[c-\varepsilon_1(1-\varepsilon_3b-\varepsilon_2a)]^2,\varepsilon_1(\varepsilon_3b-\varepsilon_2a)\right), ( A , B , C ) = ( 8 1 [ c + ε 1 ( 1 − ε 3 b − ε 2 a ) ] 2 , 8 1 [ c − ε 1 ( 1 − ε 3 b − ε 2 a ) ] 2 , ε 1 ( ε 3 b − ε 2 a ) ) , cf. [58 , 59 , 60 , 61 ] , [62 §39] and [99 §32.7(v)] .
Note that we are using the same notation as in [99 §32.7(v)] which slightly differs from that used in [59 , 60 , 61 , 62 ] .
Remarks 2.3.
Since ε j = ± 1 \varepsilon_j=\pm1 ε j = ± 1 , j = 1 , 2 , 3 j=1,2,3 j = 1 , 2 , 3 , independently, there are eight distinct transformations T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 (2.2). The parameter ε 1 \varepsilon_1 ε 1 reflects that the transformation ( z , γ ) → ( − z , − γ ) (z,\gamma)\to(-z,-\gamma) ( z , γ ) → ( − z , − γ ) is a discrete symmetry of PV _{\mathrm{V}} V (2) . The parameters ε 2 \varepsilon_2 ε 2 and ε 3 \varepsilon_3 ε 3 reflect that α = 1 2 a 2 \alpha=\frac{1}{2}a^2 α = 2 1 a 2 and β = − 1 2 b 2 \beta=-\frac{1}{2}b^2 β = − 2 1 b 2 , respectively, so a a a and b b b can take either sign.
The Bäcklund transformation T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 is usually given for the parameters α \alpha α , β \beta β and γ \gamma γ , in terms of a a a , b b b and c c c . However in order to apply a sequence of Bäcklund transformations, it is preferable to define the action of Bäcklund transformation on the parameters a a a , b b b and c c c , rather than α \alpha α , β \beta β and γ \gamma γ , to avoid any ambiguity in taking a square root.
The discrete symmetries of PV _{\mathrm{V}} V (2) , corresponding to the extended affine Weyl group of type A 3 ( 1 ) A_3^{(1)} A 3 ( 1 ) , act naturally on the root variables, see, for example, [35 , 67 ] . Hence it is natural to define the Bäcklund transformations for PV _{\mathrm{V}} V (2) in terms of the parameters a = 2 α a=\sqrt{2\alpha} a = 2 α , b = − 2 β b=\sqrt{-2\beta} b = − 2 β and c = γ c=\gamma c = γ , rather than α \alpha α , β \beta β and γ \gamma γ .
When applying the Bäcklund transformation T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 followed by the Bäcklund transformation T σ 1 , σ 2 , σ 3 \mathcal{T}_{\sigma_1,\sigma_2,\sigma_3} T σ 1 , σ 2 , σ 3 we use the notation T σ 1 , σ 2 , σ 3 ∘ T ε 1 , ε 2 , ε 3 \mathcal{T}_{\sigma_1,\sigma_2,\sigma_3}\circ\mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T σ 1 , σ 2 , σ 3 ∘ T ε 1 , ε 2 , ε 3 , i.e.
T σ 1 , σ 2 , σ 3 ( T ε 1 , ε 2 , ε 3 ( w ; a , b , c ) ) = T σ 1 , σ 2 , σ 3 ∘ T ε 1 , ε 2 , ε 3 ( w ; a , b , c ) . \mathcal{T}_{\sigma_1,\sigma_2,\sigma_3}\left(\mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3}(w;a,b,c)\right)=\mathcal{T}_{\sigma_1,\sigma_2,\sigma_3}\circ\mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3}(w;a,b,c). T σ 1 , σ 2 , σ 3 ( T ε 1 , ε 2 , ε 3 ( w ; a , b , c ) ) = T σ 1 , σ 2 , σ 3 ∘ T ε 1 , ε 2 , ε 3 ( w ; a , b , c ) .
Further if
T ε 1 , ε 2 , ε 3 ( w ; a , b , c ) = ( W ( w , w ′ , a , b , c ) ; A ( a , b , c ) , B ( a , b , c ) , C ( a , b , c ) ) , \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3}(w;a,b,c)=\left(W(w,w',a,b,c);A(a,b,c),B(a,b,c),C(a,b,c)\right), T ε 1 , ε 2 , ε 3 ( w ; a , b , c ) = ( W ( w , w ′ , a , b , c ) ; A ( a , b , c ) , B ( a , b , c ) , C ( a , b , c ) ) ,
then
T σ 1 , σ 2 , σ 3 ∘ T ε 1 , ε 2 , ε 3 ( w ; a , b , c ) = T σ 1 , σ 2 , σ 3 ( W ( w , w ′ , a , b , c ) ; A ( a , b , c ) , B ( a , b , c ) , C ( a , b , c ) ) . \mathcal{T}_{\sigma_1,\sigma_2,\sigma_3}\circ\mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3}(w;a,b,c)=\mathcal{T}_{\sigma_1,\sigma_2,\sigma_3}\left(W(w,w',a,b,c);A(a,b,c),B(a,b,c),C(a,b,c)\right). T σ 1 , σ 2 , σ 3 ∘ T ε 1 , ε 2 , ε 3 ( w ; a , b , c ) = T σ 1 , σ 2 , σ 3 ( W ( w , w ′ , a , b , c ) ; A ( a , b , c ) , B ( a , b , c ) , C ( a , b , c ) ) .
Gordoa, Joshi and Pickering [55 , 56 ] use a different approach which is a truncation-type method to derive Bäcklund transformations of P V \mathrm{P}_{\mathrm{V}} P V (2) .
Example 2.4. Suppose that w ( z ) w(z) w ( z ) is a solution of P V \mathrm{P}_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 2 a 2 , − 1 2 b 2 , c ) (\alpha,\beta,\gamma)=\left(\frac{1}{2}a^2,-\frac{1}{2}b^2,c\right) ( α , β , γ ) = ( 2 1 a 2 , − 2 1 b 2 , c ) then applying the Bäcklund transformation T 1 , 1 , 1 \mathcal{T}_{1,1,1} T 1 , 1 , 1 several times gives
T 1 , 1 , 1 ( w ; a , b , c ) = z w ′ − a w 2 + ( a − b − z ) w + b z w ′ − a w 2 + ( a − b + z ) w + b = 1 − 2 z w z w ′ − a w 2 + ( a − b + z ) w + b , \mathcal{T}_{1,1,1}(w;a,b,c)=\frac{zw'-aw^2+(a-b-z)w+b}{zw'-aw^2+(a-b+z)w+b}=1-\frac{2zw}{zw'-aw^2+(a-b+z)w+b}, T 1 , 1 , 1 ( w ; a , b , c ) = z w ′ − a w 2 + ( a − b + z ) w + b z w ′ − a w 2 + ( a − b − z ) w + b = 1 − z w ′ − a w 2 + ( a − b + z ) w + b 2 z w , T 1 , 1 , 1 2 ( w ; a , b , c ) = z w ′ + ( b + c − 1 ) w 2 − ( 2 b + c − 1 + z ) w + b w [ z w ′ − a w 2 + ( 2 a + c − 1 − z ) w − a − c + 1 ] \mathcal{T}_{1,1,1}^{2}(w;a,b,c)=\frac{zw'+(b+c-1)w^2-(2b+c-1+z)w+b}{w\left[zw'-aw^2+(2a+c-1-z)w-a-c+1\right]} T 1 , 1 , 1 2 ( w ; a , b , c ) = w [ z w ′ − a w 2 + ( 2 a + c − 1 − z ) w − a − c + 1 ] z w ′ + ( b + c − 1 ) w 2 − ( 2 b + c − 1 + z ) w + b = 1 w + ( a + b + c − 1 ) ( w − 1 ) 2 w [ z w ′ − a w 2 + ( 2 a + c − 1 − z ) w − a − c + 1 ] , =\frac{1}{w}+\frac{(a+b+c-1)(w-1)^2}{w\left[zw'-aw^2+(2a+c-1-z)w-a-c+1\right]}, = w 1 + w [ z w ′ − a w 2 + ( 2 a + c − 1 − z ) w − a − c + 1 ] ( a + b + c − 1 ) ( w − 1 ) 2 , T 1 , 1 , 1 3 ( w ; a , b , c ) = 1 + 1 a { ( a + b + c − 1 ) z w z w ′ − a w 2 + ( a − b − z ) w + b + ( a − b − c + 1 ) z w z w ′ + a w 2 − ( a + b + z ) w + b } . \mathcal{T}_{1,1,1}^{3}(w;a,b,c)=1+\frac{1}{a}\left\{\frac{(a+b+c-1)zw}{zw'-aw^2+(a-b-z)w+b}+\frac{(a-b-c+1)zw}{zw'+aw^2-(a+b+z)w+b}\right\}. T 1 , 1 , 1 3 ( w ; a , b , c ) = 1 + a 1 { z w ′ − a w 2 + ( a − b − z ) w + b ( a + b + c − 1 ) z w + z w ′ + a w 2 − ( a + b + z ) w + b ( a − b − c + 1 ) z w } . with ′ = d / d z '=\mathrm{d}/\mathrm{d}z ′ = d / d z , provided that the denominators do not vanish, and the parameters ( a , b , c ) (a,b,c) ( a , b , c ) transform as
T 1 , 1 , 1 ( a , b , c ) = ( 1 2 ( c − a − b + 1 ) , 1 2 ( a + b + c − 1 ) , b − a ) , \mathcal{T}_{1,1,1}(a,b,c)=\left(\frac{1}{2}(c-a-b+1),\frac{1}{2}(a+b+c-1),b-a\right), T 1 , 1 , 1 ( a , b , c ) = ( 2 1 ( c − a − b + 1 ) , 2 1 ( a + b + c − 1 ) , b − a ) , T 1 , 1 , 1 2 ( a , b , c ) = ( 1 2 ( b − a − c + 1 ) , 1 2 ( b − a + c − 1 ) , a + b − 1 ) , \mathcal{T}_{1,1,1}^{2}(a,b,c)=\left(\frac{1}{2}(b-a-c+1),\frac{1}{2}(b-a+c-1),a+b-1\right), T 1 , 1 , 1 2 ( a , b , c ) = ( 2 1 ( b − a − c + 1 ) , 2 1 ( b − a + c − 1 ) , a + b − 1 ) , T 1 , 1 , 1 3 ( a , b , c ) = ( a , b − 1 , c − 1 ) . \mathcal{T}_{1,1,1}^{3}(a,b,c)=(a,b-1,c-1). T 1 , 1 , 1 3 ( a , b , c ) = ( a , b − 1 , c − 1 ) . We note that T 1 , 1 , 1 2 ( w ) \mathcal{T}_{1,1,1}^{2}(w) T 1 , 1 , 1 2 ( w ) is not quadratic in w ′ w' w ′ as might have been expected.
Lemma 2.5. If T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 is given by (2.2) and S 2 S_2 S 2 is the symmetry (2.1b), then
T − 1 , 1 , 1 = S 2 ∘ T 1 , 1 , 1 , T − 1 , − 1 , 1 = S 2 ∘ T 1 , − 1 , 1 , T 1 , 1 , − 1 = S 2 ∘ T − 1 , 1 , − 1 , T 1 , − 1 , − 1 = S 2 ∘ T − 1 , − 1 , − 1 . \mathcal{T}_{-1,1,1}=S_2\circ\mathcal{T}_{1,1,1},\qquad
\mathcal{T}_{-1,-1,1}=S_2\circ\mathcal{T}_{1,-1,1},\qquad
\mathcal{T}_{1,1,-1}=S_2\circ\mathcal{T}_{-1,1,-1},\qquad
\mathcal{T}_{1,-1,-1}=S_2\circ\mathcal{T}_{-1,-1,-1}. T − 1 , 1 , 1 = S 2 ∘ T 1 , 1 , 1 , T − 1 , − 1 , 1 = S 2 ∘ T 1 , − 1 , 1 , T 1 , 1 , − 1 = S 2 ∘ T − 1 , 1 , − 1 , T 1 , − 1 , − 1 = S 2 ∘ T − 1 , − 1 , − 1 .
Proof. Suppose that w ( z ) w(z) w ( z ) is a solution of P V \mathrm{P}_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 2 a 2 , − 1 2 b 2 , c ) (\alpha,\beta,\gamma)=\left(\frac{1}{2}a^2,-\frac{1}{2}b^2,c\right) ( α , β , γ ) = ( 2 1 a 2 , − 2 1 b 2 , c ) then
T 1 , 1 , 1 ( w ; a , b , c ) = ( z w ′ − a w 2 + ( a − b − z ) w + b z w ′ − a w 2 + ( a − b + z ) w + b ; 1 2 ( c − a − b + 1 ) , 1 2 ( a + b + c − 1 ) , b − a ) , \mathcal{T}_{1,1,1}(w;a,b,c)=\left(\frac{zw'-aw^2+(a-b-z)w+b}{zw'-aw^2+(a-b+z)w+b};\frac{1}{2}(c-a-b+1),\frac{1}{2}(a+b+c-1),b-a\right), T 1 , 1 , 1 ( w ; a , b , c ) = ( z w ′ − a w 2 + ( a − b + z ) w + b z w ′ − a w 2 + ( a − b − z ) w + b ; 2 1 ( c − a − b + 1 ) , 2 1 ( a + b + c − 1 ) , b − a ) ,
T − 1 , 1 , 1 ( w ; a , b , c ) = ( z w ′ − a w 2 + ( a − b + z ) w + b z w ′ − a w 2 + ( a − b − z ) w + b ; 1 2 ( a + b + c − 1 ) , 1 2 ( c − a − b + 1 ) , a − b ) . \mathcal{T}_{-1,1,1}(w;a,b,c)=\left(\frac{zw'-aw^2+(a-b+z)w+b}{zw'-aw^2+(a-b-z)w+b};\frac{1}{2}(a+b+c-1),\frac{1}{2}(c-a-b+1),a-b\right). T − 1 , 1 , 1 ( w ; a , b , c ) = ( z w ′ − a w 2 + ( a − b − z ) w + b z w ′ − a w 2 + ( a − b + z ) w + b ; 2 1 ( a + b + c − 1 ) , 2 1 ( c − a − b + 1 ) , a − b ) .
If we define
S 2 ( w ; a , b , c ) = ( 1 / w ; b , a , − c ) , S_2(w;a,b,c)=(1/w;b,a,-c), S 2 ( w ; a , b , c ) = ( 1/ w ; b , a , − c ) ,
then applying S 2 S_2 S 2 to T 1 , 1 , 1 ( w ; a , b , c ) \mathcal{T}_{1,1,1}(w;a,b,c) T 1 , 1 , 1 ( w ; a , b , c ) shows that T − 1 , 1 , 1 = S 2 ∘ T 1 , 1 , 1 \mathcal{T}_{-1,1,1}=S_2\circ\mathcal{T}_{1,1,1} T − 1 , 1 , 1 = S 2 ∘ T 1 , 1 , 1 . The other identities are proved in a similar way. □ \square □
To generate a hierarchy of rational solutions, or special function solutions, one needs to change at least one of the parameters a a a , b b b and c c c by ± 1 \pm1 ± 1 . In order to do this, it is necessary to use a combination of Bäcklund transformations. In Table 2.1 , Table 2.2 and Table 2.3 , the transformations that change two of a a a , b b b and c c c by ± 1 \pm1 ± 1 are given.
Table 2.1. Bäcklund transformations that change the parameters a a a and b b b by ± 1 \pm1 ± 1 .
Table 2.2. Bäcklund transformations that change the parameters a a a and c c c by ± 1 \pm1 ± 1 .
Table 2.3. Bäcklund transformations that change the parameters b b b and c c c by ± 1 \pm1 ± 1 .
The Bäcklund transformations given in Table 2.1 , Table 2.2 and Table 2.3 are quadratic in d w / d z dw/dz d w / d z , as illustrated by the first transformation in each table. Suppose that w w w is a solution of P V \mathrm{P}_{V} P V (2) with ( α , β , γ ) = ( 1 2 a 2 , − 1 2 b 2 , c ) (\alpha,\beta,\gamma)=(\frac{1}{2}a^2,-\frac{1}{2}b^2,c) ( α , β , γ ) = ( 2 1 a 2 , − 2 1 b 2 , c ) , then
T 1 , − 1 , 1 ∘ T 1 , 1 , 1 ( w ) = { z w ′ + ( w − 1 ) ( b w − w − b ) } 2 − z 2 w 2 + 2 c z w 2 ( w − 1 ) − ( a + b − 1 ) 2 w 2 ( w − 1 ) 2 w { z w ′ − ( w − 1 ) ( a w − a + 1 ) } 2 − z 2 w 2 + 2 c z w ( w − 1 ) − ( a + b − 1 ) 2 ( w − 1 ) 2 . \mathcal{T}_{1,-1,1}\circ\mathcal{T}_{1,1,1}(w)=\frac{\{zw'+(w-1)(bw-w-b)\}^{2}-z^{2}w^{2}+2czw^{2}(w-1)-(a+b-1)^{2}w^{2}(w-1)^{2}}{w\{zw'-(w-1)(aw-a+1)\}^{2}-z^{2}w^{2}+2czw(w-1)-(a+b-1)^{2}(w-1)^{2}}. T 1 , − 1 , 1 ∘ T 1 , 1 , 1 ( w ) = w { z w ′ − ( w − 1 ) ( a w − a + 1 ) } 2 − z 2 w 2 + 2 cz w ( w − 1 ) − ( a + b − 1 ) 2 ( w − 1 ) 2 { z w ′ + ( w − 1 ) ( b w − w − b ) } 2 − z 2 w 2 + 2 cz w 2 ( w − 1 ) − ( a + b − 1 ) 2 w 2 ( w − 1 ) 2 .
T − 1 , 1 , 1 3 ( w ) = { z w ′ − a w 2 + ( a + b − z ) w − b } { z w ′ − a w 2 + ( a − b − z ) w + b } { z w ′ − a w ( w − 1 ) } 2 − z 2 w 2 + 2 ( a + c − 1 ) z w ( w − 1 ) − b 2 ( w − 1 ) 2 = 1 − 2 z w { z w ′ − a w 2 + ( 2 a + c − z − 1 ) w − a − c + 1 } { z w ′ − a w ( w − 1 ) } 2 − z 2 w 2 + 2 ( a + c − 1 ) z w ( w − 1 ) − b 2 ( w − 1 ) 2 . \begin{aligned}
\mathcal{T}^{3}_{-1,1,1}(w)=\frac{\{zw'-aw^{2}+(a+b-z)w-b\}\{zw'-aw^{2}+(a-b-z)w+b\}}{\{zw'-aw(w-1)\}^{2}-z^{2}w^{2}+2(a+c-1)zw(w-1)-b^{2}(w-1)^{2}} \\
&=1-\frac{2zw\{zw'-aw^{2}+(2a+c-z-1)w-a-c+1\}}{\{zw'-aw(w-1)\}^{2}-z^{2}w^{2}+2(a+c-1)zw(w-1)-b^{2}(w-1)^{2}}.
\end{aligned} T − 1 , 1 , 1 3 ( w ) = { z w ′ − a w ( w − 1 ) } 2 − z 2 w 2 + 2 ( a + c − 1 ) z w ( w − 1 ) − b 2 ( w − 1 ) 2 { z w ′ − a w 2 + ( a + b − z ) w − b } { z w ′ − a w 2 + ( a − b − z ) w + b } = 1 − { z w ′ − a w ( w − 1 ) } 2 − z 2 w 2 + 2 ( a + c − 1 ) z w ( w − 1 ) − b 2 ( w − 1 ) 2 2 z w { z w ′ − a w 2 + ( 2 a + c − z − 1 ) w − a − c + 1 } .
T 1 , 1 , 1 3 ( w ) = 1 + 1 a { ( a + b + c − 1 ) z w z w ′ − a w 2 + ( a − b − z ) w + b + ( a − b − c + 1 ) z w z w ′ + a w 2 − ( a + b + z ) w + b } , \mathcal{T}^{3}_{1,1,1}(w)=1+\frac{1}{a}\left\{\frac{(a+b+c-1)zw}{zw'-aw^{2}+(a-b-z)w+b}+\frac{(a-b-c+1)zw}{zw'+aw^{2}-(a+b+z)w+b}\right\}, T 1 , 1 , 1 3 ( w ) = 1 + a 1 { z w ′ − a w 2 + ( a − b − z ) w + b ( a + b + c − 1 ) z w + z w ′ + a w 2 − ( a + b + z ) w + b ( a − b − c + 1 ) z w } ,
where ′ ' ′ ≡ d / d z \equiv d/dz ≡ d / d z . It might have been expected that the Bäcklund transformations in Table 2.2 and Table 2.3 would be cubic in w ′ w' w ′ , though frequently there are simplifications. In Table 2.1 , first apply the Bäcklund transformation T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 , then apply either T 1 , − 1 , 1 \mathcal{T}_{1,-1,1} T 1 , − 1 , 1 or T − 1 , 1 , − 1 \mathcal{T}_{-1,1,-1} T − 1 , 1 , − 1 . In Table 2.2 and Table 2.3 , first apply the Bäcklund transformation T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 , then apply one of T 1 , 1 , 1 \mathcal{T}_{1,1,1} T 1 , 1 , 1 , T − 1 , 1 , 1 \mathcal{T}_{-1,1,1} T − 1 , 1 , 1 , T 1 , − 1 , − 1 \mathcal{T}_{1,-1,-1} T 1 , − 1 , − 1 or T − 1 , − 1 , − 1 \mathcal{T}_{-1,-1,-1} T − 1 , − 1 , − 1 twice.
There are other combinations of Bäcklund transformations which will give the same effect on the parameters.
From Table 2.1 , Table 2.2 and Table 2.3 , in general two of a a a , b b b and c c c will change by ± 1 \pm1 ± 1 . If either a = ± 1 2 a=\pm\frac{1}{2} a = ± 2 1 or b = ± 1 2 b=\pm\frac{1}{2} b = ± 2 1 , then there exist Bäcklund transformations such that only one of α \alpha α , β \beta β and γ \gamma γ will change. Further, if a = ± 1 2 a=\pm\frac{1}{2} a = ± 2 1 and b = ± 1 2 b=\pm\frac{1}{2} b = ± 2 1 then there exist Bäcklund transformations such that none of α \alpha α , β \beta β and γ \gamma γ will change. This is illustrated in § 5 , see Example 4.2 .
Example 2.7. Suppose that w 0 = w ( z ; α , − α , 0 ) w_0=w(z;\alpha,-\alpha,0) w 0 = w ( z ; α , − α , 0 ) is a solution of P V \mathrm{P}_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 2 a 2 , − 1 2 a 2 , 0 ) (\alpha,\beta,\gamma)=(\frac{1}{2}a^2,-\frac{1}{2}a^2,0) ( α , β , γ ) = ( 2 1 a 2 , − 2 1 a 2 , 0 ) , then
w 1 = T 1 , 1 , 1 ( w 0 ; a , a , 0 ) = 1 T 1 , 1 , 1 ( w 0 ; a , a , 0 ) = z w ′ − a w 2 − z w + a z w ′ − a w 2 + z w + a , w_1=\mathcal{T}_{1,1,1}(w_0;a,a,0)=\frac{1}{\mathcal{T}_{1,1,1}(w_0;a,a,0)}=\frac{zw'-aw^2-zw+a}{zw'-aw^2+zw+a}, w 1 = T 1 , 1 , 1 ( w 0 ; a , a , 0 ) = T 1 , 1 , 1 ( w 0 ; a , a , 0 ) 1 = z w ′ − a w 2 + z w + a z w ′ − a w 2 − z w + a ,
w 2 = T − 1 , − 1 , − 1 ( w 0 ; a , a , 0 ) = 1 T − 1 , − 1 , − 1 ( w 0 ; a , a , 0 ) = z w ′ + a w 2 + z w − a z w ′ + a w 2 − z w − a . w_2=\mathcal{T}_{-1,-1,-1}(w_0;a,a,0)=\frac{1}{\mathcal{T}_{-1,-1,-1}(w_0;a,a,0)}=\frac{zw'+aw^2+zw-a}{zw'+aw^2-zw-a}. w 2 = T − 1 , − 1 , − 1 ( w 0 ; a , a , 0 ) = T − 1 , − 1 , − 1 ( w 0 ; a , a , 0 ) 1 = z w ′ + a w 2 − z w − a z w ′ + a w 2 + z w − a .
The solution w 1 w_1 w 1 satisfies P V \mathrm{P}_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 2 ( a − 1 2 ) 2 , − 1 2 ( a − 1 2 ) 2 , 0 ) (\alpha,\beta,\gamma)=(\frac{1}{2}(a-\frac{1}{2})^2,-\frac{1}{2}(a-\frac{1}{2})^2,0) ( α , β , γ ) = ( 2 1 ( a − 2 1 ) 2 , − 2 1 ( a − 2 1 ) 2 , 0 ) whilst the solution w 2 w_2 w 2 satisfies P V \mathrm{P}_{\mathrm{V}} P V with ( α , β , γ ) = ( 1 2 ( a + 1 2 ) 2 , − 1 2 ( a + 1 2 ) 2 , 0 ) (\alpha,\beta,\gamma)=(\frac{1}{2}(a+\frac{1}{2})^2,-\frac{1}{2}(a+\frac{1}{2})^2,0) ( α , β , γ ) = ( 2 1 ( a + 2 1 ) 2 , − 2 1 ( a + 2 1 ) 2 , 0 ) .
Schlesinger transformations
A special class of Bäcklund transformations for the Painlevé equations is generated by Schlesinger transformations , which relate two solutions of the associated monodromy problem.
Each of the Painlevé equations can be expressed as the compatibility condition of a linear system, often known as the isomonodromy problem or Lax pair,
∂ Ψ ∂ λ = A J ( z , λ ) Ψ , ∂ Ψ ∂ z = B J ( z , λ ) Ψ , J = I , I I , … , V I , (9) \frac{\partial\Psi}{\partial\lambda}=A_J(z,\lambda)\Psi,\qquad\frac{\partial\Psi}{\partial z}=B_J(z,\lambda)\Psi,\qquad J=\mathrm{I,II,\ldots,VI}, \tag*{(9)} ∂ λ ∂ Ψ = A J ( z , λ ) Ψ , ∂ z ∂ Ψ = B J ( z , λ ) Ψ , J = I , II , … , VI , (9)
where A J A_J A J and B J B_J B J are matrices whose entries depend on the solution w ( z ) w(z) w ( z ) of the Painlevé equation. The compatibility condition of (9) is given by
∂ 2 Ψ ∂ z ∂ λ = ∂ 2 Ψ ∂ λ ∂ z , \frac{\partial^2\Psi}{\partial z\partial\lambda}=\frac{\partial^2\Psi}{\partial\lambda\partial z}, ∂ z ∂ λ ∂ 2 Ψ = ∂ λ ∂ z ∂ 2 Ψ ,
which is satisfied provided that
∂ A J ∂ z − ∂ B J ∂ λ + A J B J − B J A J = 0. (10) \frac{\partial A_J}{\partial z}-\frac{\partial B_J}{\partial\lambda}+A_JB_J-B_JA_J=0. \tag*{(10)} ∂ z ∂ A J − ∂ λ ∂ B J + A J B J − B J A J = 0. (10)
Matrices A J A_J A J and B J B_J B J satisfying (10) for P I \mathrm{P}_{\mathrm{I}} P I –P V I \mathrm{P}_{\mathrm{VI}} P VI are given by Jimbo and Miwa [63 ] , see also [47 ] , though these are not unique. A Schlesinger transformation R \mathcal{R} R relates two solutions Ψ \Psi Ψ and Ψ 1 \Psi_1 Ψ 1 of (9) in the form
Ψ 1 = R Ψ . \Psi_1=\mathcal{R}\Psi. Ψ 1 = R Ψ.
For P V \mathrm{P}_{\mathrm{V}} P V (2) the matrices A V \mathbf{A}_{\mathrm{V}} A V and B V \mathbf{B}_{\mathrm{V}} B V are
A V ( z , λ ) = ( 1 2 z 0 0 − 1 2 z ) + 1 λ ( v + 1 2 θ 0 − u ( v + θ 0 ) v / u − v − 1 2 θ 0 ) + 1 λ − 1 ( − v − 1 2 ( θ 0 + θ ∞ ) u w [ v + 1 2 ( θ 0 − θ 1 + θ ∞ ) ] − v + 1 2 ( θ 0 + θ 1 + θ ∞ ) u w v + 1 2 ( θ 0 + θ ∞ ) ) , \mathbf{A}_{\mathrm{V}}(z,\lambda)=\begin{pmatrix}\frac{1}{2}z & 0 \\ 0 & -\frac{1}{2}z\end{pmatrix}+\frac{1}{\lambda}\begin{pmatrix}v+\frac{1}{2}\theta_{0} & -u(v+\theta_{0}) \\ v/u & -v-\frac{1}{2}\theta_{0}\end{pmatrix}+\frac{1}{\lambda-1}\begin{pmatrix}-v-\frac{1}{2}(\theta_{0}+\theta_{\infty}) & uw\left[v+\frac{1}{2}(\theta_{0}-\theta_{1}+\theta_{\infty})\right] \\ -\frac{v+\frac{1}{2}(\theta_{0}+\theta_{1}+\theta_{\infty})}{uw} & v+\frac{1}{2}(\theta_{0}+\theta_{\infty})\end{pmatrix}, A V ( z , λ ) = ( 2 1 z 0 0 − 2 1 z ) + λ 1 ( v + 2 1 θ 0 v / u − u ( v + θ 0 ) − v − 2 1 θ 0 ) + λ − 1 1 ( − v − 2 1 ( θ 0 + θ ∞ ) − u w v + 2 1 ( θ 0 + θ 1 + θ ∞ ) u w [ v + 2 1 ( θ 0 − θ 1 + θ ∞ ) ] v + 2 1 ( θ 0 + θ ∞ ) ) ,
B V ( z , λ ) = ( 1 2 λ 0 0 − 1 2 λ ) + 1 z ( 0 − u { v + θ 0 − w [ v + 1 2 ( θ 0 − θ 1 + θ ∞ ) ] } v u − v + 1 2 ( θ 0 + θ 1 + θ ∞ ) u w 0 ) . \mathbf{B}_{\mathrm{V}}(z,\lambda)=\begin{pmatrix}\frac{1}{2}\lambda& 0 \\ 0 & -\frac{1}{2}\lambda\end{pmatrix}+\frac{1}{z}\begin{pmatrix}0 & -u\left\{v+\theta_{0}-w\left[v+\frac{1}{2}(\theta_{0}-\theta_{1}+\theta_{\infty})\right]\right\} \\ \frac{v}{u}-\frac{v+\frac{1}{2}(\theta_{0}+\theta_{1}+\theta_{\infty})}{uw} & 0\end{pmatrix}. B V ( z , λ ) = ( 2 1 λ 0 0 − 2 1 λ ) + z 1 ( 0 u v − u w v + 2 1 ( θ 0 + θ 1 + θ ∞ ) − u { v + θ 0 − w [ v + 2 1 ( θ 0 − θ 1 + θ ∞ ) ] } 0 ) .
with θ 0 \theta_{0} θ 0 , θ 1 \theta_{1} θ 1 and θ ∞ \theta_{\infty} θ ∞ parameters, the monodromy parameters , then from (9)
z d w d z = z w − 2 v ( w − 1 ) 2 − 1 2 ( w − 1 ) [ ( θ 0 − θ 1 + θ ∞ ) w − ( 3 θ 0 + θ 1 + θ ∞ ) ] , (11) z\frac{dw}{dz}=zw-2v(w-1)^{2}-\frac{1}{2}(w-1)\left[(\theta_{0}-\theta_{1}+\theta_{\infty})w-(3\theta_{0}+\theta_{1}+\theta_{\infty})\right], \tag*{(11)} z d z d w = z w − 2 v ( w − 1 ) 2 − 2 1 ( w − 1 ) [ ( θ 0 − θ 1 + θ ∞ ) w − ( 3 θ 0 + θ 1 + θ ∞ ) ] , (11)
z d v d z = u w [ v + 1 2 ( θ 0 − θ 1 + θ ∞ ) ] − ( v + θ 0 ) [ v + 1 2 ( θ 0 + θ 1 + θ ∞ ) ] w , (12) z\frac{dv}{dz}=uw\left[v+\frac{1}{2}(\theta_{0}-\theta_{1}+\theta_{\infty})\right]-\frac{(v+\theta_{0})\left[v+\frac{1}{2}(\theta_{0}+\theta_{1}+\theta_{\infty})\right]}{w}, \tag*{(12)} z d z d v = u w [ v + 2 1 ( θ 0 − θ 1 + θ ∞ ) ] − w ( v + θ 0 ) [ v + 2 1 ( θ 0 + θ 1 + θ ∞ ) ] , (12)
z d u d z = u { − 2 v − θ 0 + w [ v + 1 2 ( θ 0 − θ 1 + θ ∞ ) ] + v + 1 2 ( θ 0 + θ 1 + θ ∞ ) w } , (13) z\frac{du}{dz}=u\left\{-2v-\theta_{0}+w\left[v+\frac{1}{2}(\theta_{0}-\theta_{1}+\theta_{\infty})\right]+\frac{v+\frac{1}{2}(\theta_{0}+\theta_{1}+\theta_{\infty})}{w}\right\}, \tag*{(13)} z d z d u = u { − 2 v − θ 0 + w [ v + 2 1 ( θ 0 − θ 1 + θ ∞ ) ] + w v + 2 1 ( θ 0 + θ 1 + θ ∞ ) } , (13)
see [63 ] , equations C.38, C.39; also [47 ] , Chapter 5, §4. Consequently, eliminating v v v shows that w w w satisfies P V \mathrm{P}_{\mathrm{V}} P V (2) with
( α , β , γ ) = ( 1 8 ( θ 0 − θ 1 + θ ∞ ) 2 , − 1 8 ( θ 0 − θ 1 − θ ∞ ) 2 , 1 − θ 0 − θ 1 ) . (14) (\alpha,\beta,\gamma)=\left(\frac{1}{8}(\theta_{0}-\theta_{1}+\theta_{\infty})^{2},-\frac{1}{8}(\theta_{0}-\theta_{1}-\theta_{\infty})^{2},1-\theta_{0}-\theta_{1}\right). \tag*{(14)} ( α , β , γ ) = ( 8 1 ( θ 0 − θ 1 + θ ∞ ) 2 , − 8 1 ( θ 0 − θ 1 − θ ∞ ) 2 , 1 − θ 0 − θ 1 ) . (14)
The main characteristic of a Schlesinger transformation for P V \mathrm{P}_{\mathrm{V}} P V (2) is that the monodromy parameters θ 0 \theta_{0} θ 0 , θ 1 \theta_{1} θ 1 and θ ∞ \theta_{\infty} θ ∞ are changed by integers.
Fokas, Muğan and Ablowitz [48 ] show that the Schlesinger transformations for P V \mathrm{P}_{\mathrm{V}} P V (2) are in three sets
(i) : { θ 0 → θ 0 ± 1 , θ 1 → θ 1 , θ ∞ → θ ∞ ± 1 , (ii) : { θ 0 → θ 0 , θ 1 → θ 1 ± 1 , θ ∞ → θ ∞ ± 1 , (iii) : { θ 0 → θ 0 ± 1 , θ 1 → θ 1 ± 1 , θ ∞ → θ ∞ , \text{(i)}:\quad\begin{cases}\theta_{0}\to\theta_{0}\pm1,\\ \theta_{1}\to\theta_{1},\\ \theta_{\infty}\to\theta_{\infty}\pm1,\end{cases}
\qquad
\text{(ii)}:\quad\begin{cases}\theta_{0}\to\theta_{0},\\ \theta_{1}\to\theta_{1}\pm1,\\ \theta_{\infty}\to\theta_{\infty}\pm1,\end{cases}
\qquad
\text{(iii)}:\quad\begin{cases}\theta_{0}\to\theta_{0}\pm1,\\ \theta_{1}\to\theta_{1}\pm1,\\ \theta_{\infty}\to\theta_{\infty},\end{cases} (i) : ⎩ ⎨ ⎧ θ 0 → θ 0 ± 1 , θ 1 → θ 1 , θ ∞ → θ ∞ ± 1 , (ii) : ⎩ ⎨ ⎧ θ 0 → θ 0 , θ 1 → θ 1 ± 1 , θ ∞ → θ ∞ ± 1 , (iii) : ⎩ ⎨ ⎧ θ 0 → θ 0 ± 1 , θ 1 → θ 1 ± 1 , θ ∞ → θ ∞ ,
where the ± \pm ± signs are independent. Fokas, Muğan and Ablowitz [48 equations (4.44)] derive the eight Schlesinger transformations R 1 , R 2 , … , R 8 \mathcal{R}_{1},\mathcal{R}_{2},\ldots,\mathcal{R}_{8} R 1 , R 2 , … , R 8 , in the first two sets where the monodromy parameter θ ∞ \theta_{\infty} θ ∞ changes by ± 1 \pm1 ± 1 and either θ 0 \theta_{0} θ 0 or θ 1 \theta_{1} θ 1 change by $\pm1; see also [87 equations (5.35)–(5.42)] .
Our interest is the effect of the Schlesinger transformations R 1 , R 2 , … , R 8 \mathcal{R}_{1},\mathcal{R}_{2},\ldots,\mathcal{R}_{8} R 1 , R 2 , … , R 8 on the parameters ( α , β , γ ) (\alpha,\beta,\gamma) ( α , β , γ ) and solutions w ( z ) w(z) w ( z ) of P V \mathrm{P}_{\mathrm{V}} P V (2) . Using (14) it is straightforward to compute what happens to α \alpha α , β \beta β and γ \gamma γ when θ 0 \theta_{0} θ 0 , θ 1 \theta_{1} θ 1 and θ ∞ \theta_{\infty} θ ∞ change by ± 1 \pm1 ± 1 . Hence using the results in Tables Table 2.1 , Table 2.2 and Table 2.3 we can deduce the combination of Bäcklund transformations T ε 1 , ε 2 , ε 3 T_{\varepsilon_{1},\varepsilon_{2},\varepsilon_{3}} T ε 1 , ε 2 , ε 3 (2.2) that are equivalent to each of the Schlesinger transformations R 1 , R 2 , … , R 8 \mathcal{R}_{1},\mathcal{R}_{2},\ldots,\mathcal{R}_{8} R 1 , R 2 , … , R 8 . These results are given in Table Table 2.4 . Consequently it is straightforward to calculate the effect of the Schlesinger transformations R 1 , R 2 , … , R 8 \mathcal{R}_{1},\mathcal{R}_{2},\ldots,\mathcal{R}_{8} R 1 , R 2 , … , R 8 to a solution w ( z ) w(z) w ( z ) of P V \mathrm{P}_{\mathrm{V}} P V (2) with parameters ( α , β , γ ) = ( 1 2 a 2 , − 1 2 b 2 , c ) (\alpha,\beta,\gamma)=\left(\frac{1}{2}a^{2},-\frac{1}{2}b^{2},c\right) ( α , β , γ ) = ( 2 1 a 2 , − 2 1 b 2 , c ) . These are given in equations (2.7) which are transformations that are quadratic in d w / d z dw/dz d w / d z .
Table 2.4. The effect of the Schlesinger transformations R 1 , R 2 , … , R 8 \mathcal{R}_1,\mathcal{R}_2,\ldots,\mathcal{R}_8 R 1 , R 2 , … , R 8 on the parameters θ 0 , θ 1 , θ ∞ , α , β \theta_0,\theta_1,\theta_\infty,\alpha,\beta θ 0 , θ 1 , θ ∞ , α , β and γ \gamma γ and the equivalent combination of Bäcklund transformations T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 (2.2).
R 1 : w 1 = { z w ′ + a w 2 − ( a + b + z ) w + b } { z w ′ + a w 2 − ( a − b + z ) w − b } { z w ′ + a w ( w − 1 ) } 2 − z 2 w 2 − 2 z ( a − c + 1 ) w ( w − 1 ) − b 2 ( w − 1 ) 2 (15) \mathcal{R}_1:\quad w_1=\frac{\left\{zw'+aw^2-(a+b+z)w+b\right\}\left\{zw'+aw^2-(a-b+z)w-b\right\}}{\left\{zw'+aw(w-1)\right\}^2-z^2w^2-2z(a-c+1)w(w-1)-b^2(w-1)^2} \tag*{(15)} R 1 : w 1 = { z w ′ + a w ( w − 1 ) } 2 − z 2 w 2 − 2 z ( a − c + 1 ) w ( w − 1 ) − b 2 ( w − 1 ) 2 { z w ′ + a w 2 − ( a + b + z ) w + b } { z w ′ + a w 2 − ( a − b + z ) w − b } (15)
R 2 : w 2 = { z w ′ − a w 2 + ( a + b + z ) w − b } { z w ′ − a w 2 + ( a − b + z ) w + b } { z w ′ − a w ( w − 1 ) } 2 − z 2 w 2 − 2 z ( a − c − 1 ) w ( w − 1 ) − b 2 ( w − 1 ) 2 (16) \mathcal{R}_2:\quad w_2=\frac{\left\{zw'-aw^2+(a+b+z)w-b\right\}\left\{zw'-aw^2+(a-b+z)w+b\right\}}{\left\{zw'-aw(w-1)\right\}^2-z^2w^2-2z(a-c-1)w(w-1)-b^2(w-1)^2} \tag*{(16)} R 2 : w 2 = { z w ′ − a w ( w − 1 ) } 2 − z 2 w 2 − 2 z ( a − c − 1 ) w ( w − 1 ) − b 2 ( w − 1 ) 2 { z w ′ − a w 2 + ( a + b + z ) w − b } { z w ′ − a w 2 + ( a − b + z ) w + b } (16)
R 3 : w 3 = 1 + 1 a { ( a + b − c + 1 ) z w z w ′ + a w 2 − ( a − b + z ) w − b + ( a − b + c − 1 ) z w z w ′ − a w 2 + ( a + b − z ) w − b } (17) \mathcal{R}_3:\quad w_3=1+\frac{1}{a}\left\{\frac{(a+b-c+1)zw}{zw'+aw^2-(a-b+z)w-b}+\frac{(a-b+c-1)zw}{zw'-aw^2+(a+b-z)w-b}\right\} \tag*{(17)} R 3 : w 3 = 1 + a 1 { z w ′ + a w 2 − ( a − b + z ) w − b ( a + b − c + 1 ) z w + z w ′ − a w 2 + ( a + b − z ) w − b ( a − b + c − 1 ) z w } (17)
R 4 : w 4 = 1 − 1 a { ( a − b + c + 1 ) z w z w ′ + a w 2 − ( a + b − z ) w + b + ( a + b − c − 1 ) z w z w ′ − a w 2 + ( a − b + z ) w + b } (18) \mathcal{R}_4:\quad w_4=1-\frac{1}{a}\left\{\frac{(a-b+c+1)zw}{zw'+aw^2-(a+b-z)w+b}+\frac{(a+b-c-1)zw}{zw'-aw^2+(a-b+z)w+b}\right\} \tag*{(18)} R 4 : w 4 = 1 − a 1 { z w ′ + a w 2 − ( a + b − z ) w + b ( a − b + c + 1 ) z w + z w ′ − a w 2 + ( a − b + z ) w + b ( a + b − c − 1 ) z w } (18)
R 5 : w 5 = 1 + 1 a { ( a − b − c + 1 ) z w z w ′ + a w 2 − ( a + b + z ) w + b + ( a + b + c − 1 ) z w z w ′ − a w 2 + ( a − b − z ) w + b } (19) \mathcal{R}_5:\quad w_5=1+\frac{1}{a}\left\{\frac{(a-b-c+1)zw}{zw'+aw^2-(a+b+z)w+b}+\frac{(a+b+c-1)zw}{zw'-aw^2+(a-b-z)w+b}\right\} \tag*{(19)} R 5 : w 5 = 1 + a 1 { z w ′ + a w 2 − ( a + b + z ) w + b ( a − b − c + 1 ) z w + z w ′ − a w 2 + ( a − b − z ) w + b ( a + b + c − 1 ) z w } (19)
R 6 : w 6 = 1 − 1 a { ( a + b + c + 1 ) z w z w ′ + a w 2 − ( a − b − z ) w − b + ( a − b − c − 1 ) z w z w ′ − a w 2 + ( a + b + z ) w − b } (20) \mathcal{R}_6:\quad w_6=1-\frac{1}{a}\left\{\frac{(a+b+c+1)zw}{zw'+aw^2-(a-b-z)w-b}+\frac{(a-b-c-1)zw}{zw'-aw^2+(a+b+z)w-b}\right\} \tag*{(20)} R 6 : w 6 = 1 − a 1 { z w ′ + a w 2 − ( a − b − z ) w − b ( a + b + c + 1 ) z w + z w ′ − a w 2 + ( a + b + z ) w − b ( a − b − c − 1 ) z w } (20)
R 7 : w 7 = { z w ′ − a w 2 + ( a − b − z ) w + b } { z w ′ − a w 2 + ( a + b − z ) w − b } { z w ′ − a w ( w − 1 ) } 2 − z 2 w 2 + 2 ( a + c − 1 ) z w ( w − 1 ) − b 2 ( w − 1 ) 2 (21) \mathcal{R}_7:\quad w_7=\frac{\left\{zw'-aw^2+(a-b-z)w+b\right\}\left\{zw'-aw^2+(a+b-z)w-b\right\}}{\left\{zw'-aw(w-1)\right\}^2-z^2w^2+2(a+c-1)zw(w-1)-b^2(w-1)^2} \tag*{(21)} R 7 : w 7 = { z w ′ − a w ( w − 1 ) } 2 − z 2 w 2 + 2 ( a + c − 1 ) z w ( w − 1 ) − b 2 ( w − 1 ) 2 { z w ′ − a w 2 + ( a − b − z ) w + b } { z w ′ − a w 2 + ( a + b − z ) w − b } (21)
R 8 : w 8 = { z w ′ + a w 2 − ( a − b − z ) w − b } { z w ′ + a w 2 − ( a + b − z ) w + b } { z w ′ + a w ( w − 1 ) } 2 − z 2 w 2 + 2 ( a + c + 1 ) z w ( w − 1 ) − b 2 ( w − 1 ) 2 (22) \mathcal{R}_8:\quad w_8=\frac{\left\{zw'+aw^2-(a-b-z)w-b\right\}\left\{zw'+aw^2-(a+b-z)w+b\right\}}{\left\{zw'+aw(w-1)\right\}^2-z^2w^2+2(a+c+1)zw(w-1)-b^2(w-1)^2} \tag*{(22)} R 8 : w 8 = { z w ′ + a w ( w − 1 ) } 2 − z 2 w 2 + 2 ( a + c + 1 ) z w ( w − 1 ) − b 2 ( w − 1 ) 2 { z w ′ + a w 2 − ( a − b − z ) w − b } { z w ′ + a w 2 − ( a + b − z ) w + b } (22)
with ′ = d / d z '=\mathrm{d}/\mathrm{d}z ′ = d / d z .
Lemma 2.8. The Schlesinger transformations R 1 , R 2 , … , R 8 \mathcal{R}_1,\mathcal{R}_2,\ldots,\mathcal{R}_8 R 1 , R 2 , … , R 8 commute pairwise, i.e.
R j R k = R k R j , j , k = 1 , 2 , … , 8. \mathcal{R}_j\mathcal{R}_k=\mathcal{R}_k\mathcal{R}_j,\qquad j,k=1,2,\ldots,8. R j R k = R k R j , j , k = 1 , 2 , … , 8.
Proof. It is straightforward to verify this using the definition of the transformations given in Table 2.4 and (2.7).
The Schlesinger transformations R 9 , … , R 12 \mathcal{R}_9,\ldots,\mathcal{R}_{12} R 9 , … , R 12 in which θ 0 \theta_0 θ 0 and θ 1 \theta_1 θ 1 change by ± 1 \pm1 ± 1 , with θ ∞ \theta_\infty θ ∞ unchanged, are obtained using a combination of two Schlesinger transformations as follows
R 9 = R 1 R 7 , R 10 = R 2 R 8 , R 11 = R 1 R 6 , R 12 = R 2 R 5 . (23) \mathcal{R}_9=\mathcal{R}_1\mathcal{R}_7,\qquad\mathcal{R}_{10}=\mathcal{R}_2\mathcal{R}_8,\qquad\mathcal{R}_{11}=\mathcal{R}_1\mathcal{R}_6,\qquad\mathcal{R}_{12}=\mathcal{R}_2\mathcal{R}_5. \tag*{(23)} R 9 = R 1 R 7 , R 10 = R 2 R 8 , R 11 = R 1 R 6 , R 12 = R 2 R 5 . (23)
Due to Lemma 2.8 , the order of the Schlesinger transformations in (23) is irrelevant. The effect of the Schlesinger transformations R 9 , … , R 12 \mathcal{R}_9,\ldots,\mathcal{R}_{12} R 9 , … , R 12 on the parameters θ 0 \theta_0 θ 0 , θ 1 \theta_1 θ 1 , θ ∞ \theta_\infty θ ∞ , α \alpha α , β \beta β and γ \gamma γ and the equivalent combination of Bäcklund transformations T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 (2.2) is given in Table 2.5 . However, from Table 2.5 we see that for the parameters a a a and b b b in α \alpha α and β \beta β respectively to change in opposite directions, then it is necessary to introduce the transformations R 13 \mathcal{R}_{13} R 13 and R 14 \mathcal{R}_{14} R 14 , which can be done in two ways
Table 2.5. The effect of the Schlesinger transformations R 9 , R 10 , … , R 14 \mathcal{R}_9,\mathcal{R}_{10},\ldots,\mathcal{R}_{14} R 9 , R 10 , … , R 14 on the parameters θ 0 , θ 1 , θ ∞ , α , β \theta_0,\theta_1,\theta_\infty,\alpha,\beta θ 0 , θ 1 , θ ∞ , α , β and γ \gamma γ and the equivalent combination of Bäcklund transformations T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 (2.2).
R 13 = R 1 R 4 = R 5 R 8 , R 14 = R 2 R 3 = R 6 R 7 . (24) \mathcal{R}_{13}=\mathcal{R}_1\mathcal{R}_4=\mathcal{R}_5\mathcal{R}_8,\qquad\mathcal{R}_{14}=\mathcal{R}_2\mathcal{R}_3=\mathcal{R}_6\mathcal{R}_7. \tag*{(24)} R 13 = R 1 R 4 = R 5 R 8 , R 14 = R 2 R 3 = R 6 R 7 . (24)
and their effect is also given in Table 2.5 . The effect of the Schlesinger transformations R 9 , R 10 , … , R 14 \mathcal{R}_9,\mathcal{R}_{10},\ldots,\mathcal{R}_{14} R 9 , R 10 , … , R 14 to a solution w ( z ) w(z) w ( z ) of P V \mathrm{P}_{\mathrm{V}} P V (1.2) with parameters ( α , β , γ ) = ( 1 2 a 2 , − 1 2 b 2 , c ) (\alpha,\beta,\gamma)=\left(\frac{1}{2}a^2,-\frac{1}{2}b^2,c\right) ( α , β , γ ) = ( 2 1 a 2 , − 2 1 b 2 , c ) is given in equations (2.10) which are also transformations that are quadratic in d w / d z \mathrm{d}w/\mathrm{d}z d w / d z .
R 9 : w 9 = 1 w + ( w − 1 ) 2 2 a w { ( a + b + c − 1 ) ( a − b + c − 1 ) z w ′ − a w 2 + ( 2 a + c − z − 1 ) w − a − c + 1 − ( a + b − c + 1 ) ( a − b − c + 1 ) z w ′ + a w 2 − ( 2 a − c + z + 1 ) w + a − c + 1 } (25) \mathcal{R}_9:\quad w_9=\frac{1}{w}+\frac{(w-1)^2}{2aw}\left\{\frac{(a+b+c-1)(a-b+c-1)}{zw'-aw^2+(2a+c-z-1)w-a-c+1}-\frac{(a+b-c+1)(a-b-c+1)}{zw'+aw^2-(2a-c+z+1)w+a-c+1}\right\} \tag*{(25)} R 9 : w 9 = w 1 + 2 a w ( w − 1 ) 2 { z w ′ − a w 2 + ( 2 a + c − z − 1 ) w − a − c + 1 ( a + b + c − 1 ) ( a − b + c − 1 ) − z w ′ + a w 2 − ( 2 a − c + z + 1 ) w + a − c + 1 ( a + b − c + 1 ) ( a − b − c + 1 ) } (25)
R 10 : w 10 = 1 w + ( w − 1 ) 2 2 a w { ( a + b − c − 1 ) ( a − b − c − 1 ) z w ′ − a w 2 + ( 2 a − c + z − 1 ) w − a + c + 1 − ( a + b + c + 1 ) ( a − b + c + 1 ) z w ′ + a w 2 − ( 2 a + c − z + 1 ) w + a + c + 1 } (26) \mathcal{R}_{10}:\quad w_{10}=\frac{1}{w}+\frac{(w-1)^2}{2aw}\left\{\frac{(a+b-c-1)(a-b-c-1)}{zw'-aw^2+(2a-c+z-1)w-a+c+1}-\frac{(a+b+c+1)(a-b+c+1)}{zw'+aw^2-(2a+c-z+1)w+a+c+1}\right\} \tag*{(26)} R 10 : w 10 = w 1 + 2 a w ( w − 1 ) 2 { z w ′ − a w 2 + ( 2 a − c + z − 1 ) w − a + c + 1 ( a + b − c − 1 ) ( a − b − c − 1 ) − z w ′ + a w 2 − ( 2 a + c − z + 1 ) w + a + c + 1 ( a + b + c + 1 ) ( a − b + c + 1 ) } (26)
R 11 : w 11 = [ z w ′ − ( w − 1 ) ( b w − b + w ) ] 2 − z 2 w 2 + 2 c z ( w − 1 ) w 2 − ( a + b + 1 ) 2 ( w − 1 ) 2 w 2 w { [ z w ′ + ( w − 1 ) ( a w − a − 1 ) ] 2 − z 2 w 2 + 2 c z ( w − 1 ) w − ( a + b + 1 ) 2 ( w − 1 ) 2 } (27) \mathcal{R}_{11}:\quad w_{11}=\frac{\left[zw'-(w-1)(bw-b+w)\right]^2-z^2w^2+2cz(w-1)w^2-(a+b+1)^2(w-1)^2w^2}{w\left\{\left[zw'+(w-1)(aw-a-1)\right]^2-z^2w^2+2cz(w-1)w-(a+b+1)^2(w-1)^2\right\}} \tag*{(27)} R 11 : w 11 = w { [ z w ′ + ( w − 1 ) ( a w − a − 1 ) ] 2 − z 2 w 2 + 2 cz ( w − 1 ) w − ( a + b + 1 ) 2 ( w − 1 ) 2 } [ z w ′ − ( w − 1 ) ( b w − b + w ) ] 2 − z 2 w 2 + 2 cz ( w − 1 ) w 2 − ( a + b + 1 ) 2 ( w − 1 ) 2 w 2 (27)
R 12 : w 12 = [ z w ′ + ( w − 1 ) ( b w − b − w ) ] 2 − z 2 w 2 + 2 c z ( w − 1 ) w 2 − ( a + b − 1 ) 2 ( w − 1 ) 2 w 2 w { [ z w ′ − ( w − 1 ) ( a w − a + 1 ) ] 2 − z 2 w 2 + 2 c z ( w − 1 ) w − ( a + b − 1 ) 2 ( w − 1 ) 2 } (28) \mathcal{R}_{12}:\quad w_{12}=\frac{\left[zw'+(w-1)(bw-b-w)\right]^2-z^2w^2+2cz(w-1)w^2-(a+b-1)^2(w-1)^2w^2}{w\left\{\left[zw'-(w-1)(aw-a+1)\right]^2-z^2w^2+2cz(w-1)w-(a+b-1)^2(w-1)^2\right\}} \tag*{(28)} R 12 : w 12 = w { [ z w ′ − ( w − 1 ) ( a w − a + 1 ) ] 2 − z 2 w 2 + 2 cz ( w − 1 ) w − ( a + b − 1 ) 2 ( w − 1 ) 2 } [ z w ′ + ( w − 1 ) ( b w − b − w ) ] 2 − z 2 w 2 + 2 cz ( w − 1 ) w 2 − ( a + b − 1 ) 2 ( w − 1 ) 2 w 2 (28)
R 13 : w 13 = [ z w ′ + ( w − 1 ) ( b w − b − w ) ] 2 − z 2 w 2 + 2 c z ( w − 1 ) w 2 − ( a − b + 1 ) 2 ( w − 1 ) 2 w 2 w { [ z w ′ + ( w − 1 ) ( a w − a − 1 ) ] 2 − z 2 w 2 + 2 c z ( w − 1 ) w − ( a − b + 1 ) 2 ( w − 1 ) 2 } (29) \mathcal{R}_{13}:\quad w_{13}=\frac{\left[zw'+(w-1)(bw-b-w)\right]^2-z^2w^2+2cz(w-1)w^2-(a-b+1)^2(w-1)^2w^2}{w\left\{\left[zw'+(w-1)(aw-a-1)\right]^2-z^2w^2+2cz(w-1)w-(a-b+1)^2(w-1)^2\right\}} \tag*{(29)} R 13 : w 13 = w { [ z w ′ + ( w − 1 ) ( a w − a − 1 ) ] 2 − z 2 w 2 + 2 cz ( w − 1 ) w − ( a − b + 1 ) 2 ( w − 1 ) 2 } [ z w ′ + ( w − 1 ) ( b w − b − w ) ] 2 − z 2 w 2 + 2 cz ( w − 1 ) w 2 − ( a − b + 1 ) 2 ( w − 1 ) 2 w 2 (29)
R 14 : w 14 = [ z w ′ − ( w − 1 ) ( b w − b + w ) ] 2 − z 2 w 2 + 2 c z ( w − 1 ) w 2 − ( a − b − 1 ) 2 ( w − 1 ) 2 w 2 w { [ z w ′ − ( w − 1 ) ( a w − a + 1 ) ] 2 − z 2 w 2 + 2 c z ( w − 1 ) w − ( a − b − 1 ) 2 ( w − 1 ) 2 } (30) \mathcal{R}_{14}:\quad w_{14}=\frac{\left[zw'-(w-1)(bw-b+w)\right]^2-z^2w^2+2cz(w-1)w^2-(a-b-1)^2(w-1)^2w^2}{w\left\{\left[zw'-(w-1)(aw-a+1)\right]^2-z^2w^2+2cz(w-1)w-(a-b-1)^2(w-1)^2\right\}} \tag*{(30)} R 14 : w 14 = w { [ z w ′ − ( w − 1 ) ( a w − a + 1 ) ] 2 − z 2 w 2 + 2 cz ( w − 1 ) w − ( a − b − 1 ) 2 ( w − 1 ) 2 } [ z w ′ − ( w − 1 ) ( b w − b + w ) ] 2 − z 2 w 2 + 2 cz ( w − 1 ) w 2 − ( a − b − 1 ) 2 ( w − 1 ) 2 w 2 (30)
Bäcklund transformations of the symmetric PV
From the works of Okamoto [95 , 96 , 97 , 98 ] , it is known that the parameter spaces of P I I \mathrm{P}_{\mathrm{II}} P II –P V I \mathrm{P}_{\mathrm{VI}} P VI all admit the action of an extended affine Weyl group; the group acts as a group of Bäcklund transformations. In a series of papers, Noumi and Yamada [88 , 89 , 91 , 92 ] have implemented this idea to derive a hierarchy of dynamical systems associated to the affine Weyl group of type A ~ N ( 1 ) \widetilde{A}_{N}^{(1)} A N ( 1 ) , often known as “symmetric forms of the Painlevé equations”. The behaviour of each dynamical system varies depending on whether N N N is even or odd. The first member of the A ~ 2 n + 1 ( 1 ) \widetilde{A}_{2n+1}^{(1)} A 2 n + 1 ( 1 ) hierarchy, i.e. A ~ 3 ( 1 ) \widetilde{A}_{3}^{(1)} A 3 ( 1 ) , usually known as symmetric-PV _{\mathrm{V}} V (sPV _{\mathrm{V}} V ), is equivalent to P V \mathrm{P}_{\mathrm{V}} P V (2) and given by
z d f 0 d z = f 0 f 2 ( f 1 − f 3 ) + ( 1 2 − ν 2 ) f 0 + ν 0 f 2 , (31) z\frac{\mathrm{d}f_0}{\mathrm{d}z}=f_0f_2(f_1-f_3)+\left(\frac{1}{2}-\nu_2\right)f_0+\nu_0f_2, \tag*{(31)} z d z d f 0 = f 0 f 2 ( f 1 − f 3 ) + ( 2 1 − ν 2 ) f 0 + ν 0 f 2 , (31)
z d f 1 d z = f 1 f 3 ( f 2 − f 0 ) + ( 1 2 − ν 3 ) f 1 + ν 1 f 3 , (32) z\frac{\mathrm{d}f_1}{\mathrm{d}z}=f_1f_3(f_2-f_0)+\left(\frac{1}{2}-\nu_3\right)f_1+\nu_1f_3, \tag*{(32)} z d z d f 1 = f 1 f 3 ( f 2 − f 0 ) + ( 2 1 − ν 3 ) f 1 + ν 1 f 3 , (32)
z d f 2 d z = f 2 f 0 ( f 3 − f 1 ) + ( 1 2 − ν 0 ) f 2 + ν 2 f 0 , (33) z\frac{\mathrm{d}f_2}{\mathrm{d}z}=f_2f_0(f_3-f_1)+\left(\frac{1}{2}-\nu_0\right)f_2+\nu_2f_0, \tag*{(33)} z d z d f 2 = f 2 f 0 ( f 3 − f 1 ) + ( 2 1 − ν 0 ) f 2 + ν 2 f 0 , (33)
z d f 3 d z = f 3 f 1 ( f 0 − f 2 ) + ( 1 2 − ν 1 ) f 3 + ν 3 f 1 , (34) z\frac{\mathrm{d}f_3}{\mathrm{d}z}=f_3f_1(f_0-f_2)+\left(\frac{1}{2}-\nu_1\right)f_3+\nu_3f_1, \tag*{(34)} z d z d f 3 = f 3 f 1 ( f 0 − f 2 ) + ( 2 1 − ν 1 ) f 3 + ν 3 f 1 , (34)
with the normalisations
f 0 ( z ) + f 2 ( z ) = z , f 1 ( z ) + f 3 ( z ) = z , (35) f_0(z)+f_2(z)=\sqrt{z},\qquad f_1(z)+f_3(z)=\sqrt{z}, \tag*{(35)} f 0 ( z ) + f 2 ( z ) = z , f 1 ( z ) + f 3 ( z ) = z , (35)
and ν 0 \nu_0 ν 0 , ν 1 \nu_1 ν 1 , ν 2 \nu_2 ν 2 and ν 3 \nu_3 ν 3 are constants such that
ν 0 + ν 1 + ν 2 + ν 3 = 1. (36) \nu_0+\nu_1+\nu_2+\nu_3=1. \tag*{(36)} ν 0 + ν 1 + ν 2 + ν 3 = 1. (36)
Here w = − f 2 / f 0 w=-f_2/f_0 w = − f 2 / f 0 satisfies P V \mathrm{P}_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 2 ν 0 2 , − 1 2 ν 2 2 , ν 3 − ν 1 ) (\alpha,\beta,\gamma)=\left(\frac{1}{2}\nu_0^2,-\frac{1}{2}\nu_2^2,\nu_3-\nu_1\right) ( α , β , γ ) = ( 2 1 ν 0 2 , − 2 1 ν 2 2 , ν 3 − ν 1 ) and so
f 0 = z 1 − w , f 1 = 1 2 z w { z w − z d w d z + ( w − 1 ) ( ν 0 w + ν 2 ) } , f_0=\frac{\sqrt{z}}{1-w},\qquad f_1=\frac{1}{2\sqrt{z}w}\left\{zw-z\frac{\mathrm{d}w}{\mathrm{d}z}+(w-1)(\nu_0w+\nu_2)\right\}, f 0 = 1 − w z , f 1 = 2 z w 1 { z w − z d z d w + ( w − 1 ) ( ν 0 w + ν 2 ) } ,
f 2 = z w w − 1 , f 3 = 1 2 z w { z w + z d w d z − ( w − 1 ) ( ν 0 w + ν 2 ) } . f_2=\frac{\sqrt{z}w}{w-1},\qquad f_3=\frac{1}{2\sqrt{z}w}\left\{zw+z\frac{\mathrm{d}w}{\mathrm{d}z}-(w-1)(\nu_0w+\nu_2)\right\}. f 2 = w − 1 z w , f 3 = 2 z w 1 { z w + z d z d w − ( w − 1 ) ( ν 0 w + ν 2 ) } .
The sP V \mathrm{P}_{\mathrm{V}} P V (2.11) was found by Adler [1 ] in the context of periodic chains of Bäcklund transformations and has applications in random matrix theory, cf. [51 ] .
The Bäcklund transformations of sP V \mathrm{P}_{\mathrm{V}} P V (2.11) are given by the extended affine Weyl group of type A 3 ( 1 ) A_3^{(1)} A 3 ( 1 ) [1 , 91 ] which has generators ⟨ s 0 , s 1 , s 2 , s 3 , π ⟩ \langle s_0,s_1,s_2,s_3,\pi\rangle ⟨ s 0 , s 1 , s 2 , s 3 , π ⟩ that satisfy the algebraic relations
s j 2 = I , s j s j + 2 = s j + 2 s j , s j s j + 1 s j = s j + 1 s j s j + 1 , s_j^2=\mathcal{I},\qquad s_js_{j+2}=s_{j+2}s_j,\qquad s_js_{j+1}s_j=s_{j+1}s_js_{j+1}, s j 2 = I , s j s j + 2 = s j + 2 s j , s j s j + 1 s j = s j + 1 s j s j + 1 ,
π 4 = I , π s j = s j + 1 π . \pi^4=\mathcal{I},\qquad\pi s_j=s_{j+1}\pi. π 4 = I , π s j = s j + 1 π .
The reflections s j s_j s j act on the parameters by
s j ( ν j ) = − ν j , s j ( ν j ± 1 ) = ν j ± 1 + ν j , s j ( ν j + 2 ) = ν j + 2 , s_j(\nu_j)=-\nu_j,\qquad s_j(\nu_{j\pm1})=\nu_{j\pm1}+\nu_j,\qquad s_j(\nu_{j+2})=\nu_{j+2}, s j ( ν j ) = − ν j , s j ( ν j ± 1 ) = ν j ± 1 + ν j , s j ( ν j + 2 ) = ν j + 2 ,
and on the variables by
s j ( f j ) = f j , s j ( f j ± 1 ) = f j ± 1 ± ν j f j , s j ( f j + 2 ) = f j + 2 . s_j(f_j)=f_j,\qquad s_j(f_{j\pm1})=f_{j\pm1}\pm\frac{\nu_j}{f_j},\qquad s_j(f_{j+2})=f_{j+2}. s j ( f j ) = f j , s j ( f j ± 1 ) = f j ± 1 ± f j ν j , s j ( f j + 2 ) = f j + 2 .
The rotation π \pi π acts
π ( ν j ) = ν j + 1 , π ( f j ) = f j + 1 . \pi(\nu_j)=\nu_{j+1},\qquad\pi(f_j)=f_{j+1}. π ( ν j ) = ν j + 1 , π ( f j ) = f j + 1 .
These Bäcklund transformations of sP V \mathrm{P}_{\mathrm{V}} P V (2.11) are summarised in Table 2.6 .
Table 2.6. The Bäcklund transformations of sP V \mathrm{P}_{\mathrm{V}} P V (2.11).
Masuda, Ohta and Kajiwara [84 ] define the translation operators T j T_j T j , j = 0 , 1 , 2 , 3 j=0,1,2,3 j = 0 , 1 , 2 , 3 , given by
T 0 = s 3 s 2 s 1 π , T 1 = π s 3 s 2 s 1 , T 2 = s 1 π s 3 s 2 , T 3 = s 2 s 1 π s 3 , (37) T_0=s_3s_2s_1\pi,\qquad T_1=\pi s_3s_2s_1,\qquad T_2=s_1\pi s_3s_2,\qquad T_3=s_2s_1\pi s_3, \tag*{(37)} T 0 = s 3 s 2 s 1 π , T 1 = π s 3 s 2 s 1 , T 2 = s 1 π s 3 s 2 , T 3 = s 2 s 1 π s 3 , (37)
which commute with each other and satisfy T 0 T 1 T 2 T 3 = I T_0T_1T_2T_3=\mathcal{I} T 0 T 1 T 2 T 3 = I , the identity. The effect of the translation operators T j T_j T j on the parameters ν 0 \nu_0 ν 0 , ν 1 \nu_1 ν 1 , ν 2 \nu_2 ν 2 , ν 3 \nu_3 ν 3 , a a a , b b b and c c c is given in Table 2.7 . Since the parameters change by ± 1 \pm1 ± 1 then using the results in Table 2.1 , 2.2 and 2.3 we can deduce the combination of Bäcklund transformations T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 (2.2) that are equivalent to the translation operators T j T_j T j (37) . The equivalent Schlesinger transformations R j R_j R j using Table 2.4 are also given in Table 2.7 .
Table 2.7. The effect of the translation operators T j T_j T j (37) on the parameters ν 0 \nu_0 ν 0 , ν 1 \nu_1 ν 1 , ν 2 \nu_2 ν 2 , ν 3 \nu_3 ν 3 , a a a , b b b and c c c together with the equivalent combination of Bäcklund transformations T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 (2.2) and Schlesinger transformation R j R_j R j .
From the Hamiltonian [66 , 111 ]
H V ( q , p , z ; ν 0 , ν 1 , ν 2 , ν 3 ) = q ( q − 1 ) p ( p + z ) − ν 3 ( q − 1 ) p + ν 0 z q − ν 1 q p , (38) \mathcal{H}_{\mathrm{V}}(q,p,z;\nu_0,\nu_1,\nu_2,\nu_3)=q(q-1)p(p+z)-\nu_3(q-1)p+\nu_0zq-\nu_1qp, \tag*{(38)} H V ( q , p , z ; ν 0 , ν 1 , ν 2 , ν 3 ) = q ( q − 1 ) p ( p + z ) − ν 3 ( q − 1 ) p + ν 0 z q − ν 1 q p , (38)
with ν 0 + ν 1 + ν 2 + ν 3 = 1 \nu_0+\nu_1+\nu_2+\nu_3=1 ν 0 + ν 1 + ν 2 + ν 3 = 1 , we obtain the system
z d q d z = ∂ H V ∂ p = q ( q − 1 ) ( 2 p + z ) − ν 3 ( q − 1 ) − ν 1 q , (39) z\frac{\mathrm{d}q}{\mathrm{d}z}=\frac{\partial\mathcal{H}_{\mathrm{V}}}{\partial p}=q(q-1)(2p+z)-\nu_3(q-1)-\nu_1q, \tag*{(39)} z d z d q = ∂ p ∂ H V = q ( q − 1 ) ( 2 p + z ) − ν 3 ( q − 1 ) − ν 1 q , (39)
z d p d z = − ∂ H V ∂ q = − ( 2 q − 1 ) p ( p + z ) + ( ν 1 + ν 3 ) p − ν 0 z . (40) z\frac{\mathrm{d}p}{\mathrm{d}z}=-\frac{\partial\mathcal{H}_{\mathrm{V}}}{\partial q}=-(2q-1)p(p+z)+(\nu_1+\nu_3)p-\nu_0z. \tag*{(40)} z d z d p = − ∂ q ∂ H V = − ( 2 q − 1 ) p ( p + z ) + ( ν 1 + ν 3 ) p − ν 0 z . (40)
Eliminating p p p in (2.14) gives
d 2 q d z 2 = 2 q − 1 2 q ( q − 1 ) ( d q d z ) 2 − 1 z d q d z + ν 3 2 ( q − 1 ) 2 − ν 1 2 q 2 2 z 2 q ( q − 1 ) + ( ν 2 − ν 0 ) q ( q − 1 ) z + 1 2 q ( q − 1 ) ( 2 q − 1 ) , (41) \frac{\mathrm{d}^2q}{\mathrm{d}z^2}=\frac{2q-1}{2q(q-1)}\left(\frac{\mathrm{d}q}{\mathrm{d}z}\right)^2-\frac{1}{z}\frac{\mathrm{d}q}{\mathrm{d}z}+\frac{\nu_3^2(q-1)^2-\nu_1^2q^2}{2z^2q(q-1)}+\frac{(\nu_2-\nu_0)q(q-1)}{z}+\frac{1}{2}q(q-1)(2q-1), \tag*{(41)} d z 2 d 2 q = 2 q ( q − 1 ) 2 q − 1 ( d z d q ) 2 − z 1 d z d q + 2 z 2 q ( q − 1 ) ν 3 2 ( q − 1 ) 2 − ν 1 2 q 2 + z ( ν 2 − ν 0 ) q ( q − 1 ) + 2 1 q ( q − 1 ) ( 2 q − 1 ) , (41)
and then letting q = w / ( w − 1 ) q=w/(w-1) q = w / ( w − 1 ) yields P V \mathrm{P}_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 2 ν 1 2 , − 1 2 ν 3 2 , ν 0 − ν 2 ) (\alpha,\beta,\gamma)=\left(\frac{1}{2}\nu_1^2,-\frac{1}{2}\nu_3^2,\nu_0-\nu_2\right) ( α , β , γ ) = ( 2 1 ν 1 2 , − 2 1 ν 3 2 , ν 0 − ν 2 ) . Similarly, eliminating q q q in (2.14) gives
d 2 p d z 2 = 2 p + z 2 p ( p + z ) ( d p d z ) 2 − 1 z ( p + z ) p d p d z + 1 − ν 2 2 2 ( p + z ) − ν 0 2 2 p + ( 2 p + 3 z + 2 ν 1 − 2 ν 3 ) p 2 2 z 2 + ( z + 2 ν 1 − 2 ν 3 ) p − ν 0 2 + ν 2 2 − 1 2 z , (42) \frac{\mathrm{d}^2p}{\mathrm{d}z^2}=\frac{2p+z}{2p(p+z)}\left(\frac{\mathrm{d}p}{\mathrm{d}z}\right)^2-\frac{1}{z(p+z)}p\frac{\mathrm{d}p}{\mathrm{d}z}+\frac{1-\nu_2^2}{2(p+z)}-\frac{\nu_0^2}{2p}+\frac{(2p+3z+2\nu_1-2\nu_3)p^2}{2z^2}+\frac{(z+2\nu_1-2\nu_3)p-\nu_0^2+\nu_2^2-1}{2z}, \tag*{(42)} d z 2 d 2 p = 2 p ( p + z ) 2 p + z ( d z d p ) 2 − z ( p + z ) 1 p d z d p + 2 ( p + z ) 1 − ν 2 2 − 2 p ν 0 2 + 2 z 2 ( 2 p + 3 z + 2 ν 1 − 2 ν 3 ) p 2 + 2 z ( z + 2 ν 1 − 2 ν 3 ) p − ν 0 2 + ν 2 2 − 1 , (42)
and then letting p = z / ( w − 1 ) p=z/(w-1) p = z / ( w − 1 ) yields P V \mathrm{P}_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 2 ν 0 2 , − 1 2 ν 2 2 , ν 3 − ν 1 ) (\alpha,\beta,\gamma)=\left(\frac{1}{2}\nu_0^2,-\frac{1}{2}\nu_2^2,\nu_3-\nu_1\right) ( α , β , γ ) = ( 2 1 ν 0 2 , − 2 1 ν 2 2 , ν 3 − ν 1 ) .
Remarks 2.9.
The system (2.14) can be obtained from the sPV _{\mathrm{V}} V (2.11) through
f 0 = − p z , f 1 = z ( 1 − q ) , f 2 = p + z z , f 3 = z q . f_0=-\frac{p}{\sqrt{z}},\qquad f_1=\sqrt{z}(1-q),\qquad f_2=\frac{p+z}{\sqrt{z}},\qquad f_3=\sqrt{z}q. f 0 = − z p , f 1 = z ( 1 − q ) , f 2 = z p + z , f 3 = z q .
We remark that (38) is not the usual Hamiltonian for P V \mathrm{P}_{\mathrm{V}} P V (2) . Okamoto [93 , 94 , 96 ] derived the Hamiltonian
H V O k ( Q , P , z ) = Q ( Q − 1 ) 2 P 2 − { b ( Q − 1 ) 2 − ( c + 1 ) Q ( Q − 1 ) + z Q } P + κ ( Q − 1 ) , (43) \mathcal{H}^{\mathrm{Ok}}_{\mathrm{V}}(Q,P,z)=Q(Q-1)^2P^2-\{b(Q-1)^2-(c+1)Q(Q-1)+zQ\}P+\kappa(Q-1), \tag*{(43)} H V Ok ( Q , P , z ) = Q ( Q − 1 ) 2 P 2 − { b ( Q − 1 ) 2 − ( c + 1 ) Q ( Q − 1 ) + z Q } P + κ ( Q − 1 ) , (43)
where κ = 1 4 ( b − c − 1 ) 2 − 1 4 a 2 \kappa= \frac{1}{4}(b-c-1)^2-\frac{1}{4}a^2 κ = 4 1 ( b − c − 1 ) 2 − 4 1 a 2 , with a a a , b b b and c c c parameters. Consequently we obtain the system
z d Q d z = ∂ H V O k ∂ P = 2 Q ( Q − 1 ) 2 P − b ( Q − 1 ) 2 + ( c + 1 ) Q ( Q − 1 ) − z Q , (44) z\frac{\mathrm{d}Q}{\mathrm{d}z}=\frac{\partial\mathcal{H}^{\mathrm{Ok}}_{\mathrm{V}}}{\partial P}=2Q(Q-1)^2P-b(Q-1)^2+(c+1)Q(Q-1)-zQ, \tag*{(44)} z d z d Q = ∂ P ∂ H V Ok = 2 Q ( Q − 1 ) 2 P − b ( Q − 1 ) 2 + ( c + 1 ) Q ( Q − 1 ) − z Q , (44)
z d P d z = − ∂ H V O k ∂ Q = − ( 3 Q − 1 ) ( Q − 1 ) P 2 + 2 ( b − c − 1 ) Q P + ( z − 2 b + c + 1 ) P − κ . (45) z\frac{\mathrm{d}P}{\mathrm{d}z}=-\frac{\partial\mathcal{H}^{\mathrm{Ok}}_{\mathrm{V}}}{\partial Q}=-(3Q-1)(Q-1)P^2+2(b-c-1)QP+(z-2b+c+1)P-\kappa. \tag*{(45)} z d z d P = − ∂ Q ∂ H V Ok = − ( 3 Q − 1 ) ( Q − 1 ) P 2 + 2 ( b − c − 1 ) QP + ( z − 2 b + c + 1 ) P − κ . (45)
Eliminating P P P then Q = w Q=w Q = w satisfies PV _{\mathrm{V}} V (2) with ( α , β , γ ) = ( 1 2 a 2 , − 1 2 b 2 , c ) (\alpha,\beta,\gamma)=(\frac{1}{2}a^2,-\frac{1}{2}b^2,c) ( α , β , γ ) = ( 2 1 a 2 , − 2 1 b 2 , c ) . The Hamiltonians H V \mathcal{H}_{\mathrm{V}} H V (38) and H V O k \mathcal{H}^{\mathrm{Ok}}_{\mathrm{V}} H V Ok (43) are related
H V ( q , p , z ) = H V O k ( Q , P , z ) + 1 4 ( c + 1 ) 2 − 1 4 ( a + b ) 2 , \mathcal{H}_{\mathrm{V}}(q,p,z)=\mathcal{H}^{\mathrm{Ok}}_{\mathrm{V}}(Q,P,z)+\frac{1}{4}(c+1)^2-\frac{1}{4}(a+b)^2, H V ( q , p , z ) = H V Ok ( Q , P , z ) + 4 1 ( c + 1 ) 2 − 4 1 ( a + b ) 2 ,
by the transformation [110 , 111 ]
( q − 1 ) ( Q − 1 ) = 1 , ( q − 1 ) p + ( Q − 1 ) P = 1 2 ( a + b − c − 1 ) , (q-1)(Q-1)=1,\qquad(q-1)p+(Q-1)P=\frac{1}{2}(a+b-c-1), ( q − 1 ) ( Q − 1 ) = 1 , ( q − 1 ) p + ( Q − 1 ) P = 2 1 ( a + b − c − 1 ) ,
where the relationship between the parameters { ν 0 , ν 1 , ν 2 , ν 3 } \{\nu_0,\nu_1,\nu_2,\nu_3\} { ν 0 , ν 1 , ν 2 , ν 3 } and { a , b , c } \{a,b,c\} { a , b , c } is given by
ν 0 = − 1 2 ( a + b − c − 1 ) , ν 1 = a , ν 2 = − 1 2 ( a + b + c − 1 ) , ν 3 = b . \nu_0=-\frac{1}{2}(a+b-c-1),\qquad\nu_1=a,\qquad\nu_2=-\frac{1}{2}(a+b+c-1),\qquad\nu_3=b. ν 0 = − 2 1 ( a + b − c − 1 ) , ν 1 = a , ν 2 = − 2 1 ( a + b + c − 1 ) , ν 3 = b .
Rational solutions
In this section we discuss applications of the Bäcklund transformations described in the previous section to rational solutions of PV _{\mathrm{V}} V (2) , which are classified in the following theorem.
Theorem 3.1. Equation (2) has a rational solution if and only if one of the following holds:
(i) α = 1 2 m 2 \alpha=\frac{1}{2}m^2 α = 2 1 m 2 , β = − 1 2 ( m + 2 n + 1 + μ ) 2 \beta=-\frac{1}{2}(m+2n+1+\mu)^2 β = − 2 1 ( m + 2 n + 1 + μ ) 2 , γ = μ \gamma=\mu γ = μ , for m ≥ 1 m\ge1 m ≥ 1 ;
(ii) α = 1 2 ( m + μ ) 2 \alpha=\frac{1}{2}(m+\mu)^2 α = 2 1 ( m + μ ) 2 , β = − 1 2 ( n + ε μ ) 2 \beta=-\frac{1}{2}(n+\varepsilon\mu)^2 β = − 2 1 ( n + ε μ ) 2 , γ = m + ε n \gamma=m+\varepsilon n γ = m + ε n , with ε = ± 1 \varepsilon=\pm1 ε = ± 1 , provided that m ≠ 0 m\ne0 m = 0 or n ≠ 0 n\ne0 n = 0 ;
(iii) α = 1 2 ( m + 1 2 ) 2 \alpha=\frac{1}{2}(m+\frac{1}{2})^2 α = 2 1 ( m + 2 1 ) 2 , β = − 1 2 ( n + 1 2 ) 2 \beta=-\frac{1}{2}(n+\frac{1}{2})^2 β = − 2 1 ( n + 2 1 ) 2 , γ ∉ Z \gamma\notin\mathbb{Z} γ ∈ / Z , provided that m ≠ 0 m\ne0 m = 0 or n ≠ 0 n\ne0 n = 0 ,
where m , n ∈ Z m,n\in\mathbb{Z} m , n ∈ Z and μ \mu μ is an arbitrary constant, together with the solutions obtained through the symmetries (6) .
Proof. See Kitaev, Law and McLeod [70 Theorem 1.1] ; also [62, Theorem 40.3.]
Remarks 3.2.
Kitaev, Law and McLeod [70 Theorem 1.2] prove that if γ ∉ Z \gamma\notin\mathbb{Z} γ ∈ / Z in case (i), then the rational solution is unique and if γ ∈ Z \gamma\in\mathbb{Z} γ ∈ Z then there are at most two rational solutions. Examples of non-unique rational solutions of PV _{\mathrm{V}} V (2) are given in [33 ] . In fact, the non-unique rational solutions are special cases of a special function solution of PV _{\mathrm{V}} V (2) in terms of the functions I n + 1 / 2 ( z ) I_{n+1/2}(z) I n + 1/2 ( z ) , where I ν ( z ) I_\nu(z) I ν ( z ) is the modified Bessel function and n ∈ Z n\in\mathbb{Z} n ∈ Z , as we illustrated in [34 ] . The functions I n + 1 / 2 ( z ) I_{n+1/2}(z) I n + 1/2 ( z ) , with n ∈ Z n\in\mathbb{Z} n ∈ Z , are proportional to modified spherical Bessel functions , cf. [99 §10.47] .
The requirement in case (iii) that γ ∉ Z \gamma\notin\mathbb{Z} γ ∈ / Z is not necessary as there exist rational solutions with γ ∈ Z \gamma\in\mathbb{Z} γ ∈ Z , see for example (74) . Further there are rational solutions in this case when m = n = 0 m=n=0 m = n = 0 , see (57) .
Rational solutions in case (i) are expressed in terms of generalised Laguerre polynomials , which are discussed in §3.1, and in cases (ii) and (iii) in terms of generalised Umemura polynomials , discussed in §3.2.
Generalised Laguerre polynomials
Rational solutions in case (i) of Theorem 3.1 are expressed in terms of generalised Laguerre polynomials , which are written in terms of a determinant of Laguerre polynomials, or equivalently as a Wronskian, properties of which are discussed in [8 , 33 ] . These rational solutions are special cases of the solutions of P V \mathrm{P}_{\mathrm{V}} P V (2) expressible in terms of the Kummer functions M ( a , b , z ) M(a,b,z) M ( a , b , z ) and U ( a , b , z ) U(a,b,z) U ( a , b , z ) discussed in [82 , 96 ] .
Definition 3.3. The generalised Laguerre polynomial T m , n ( μ ) ( z ) T_{m,n}^{(\mu)}(z) T m , n ( μ ) ( z ) , which is a polynomial of degree ( m + 1 ) n (m+1)n ( m + 1 ) n , is defined by
T m , n ( μ ) ( z ) = det [ d j + k d z j + k L m + n ( μ + 1 ) ( z ) ] j , k = 0 n − 1 , m ≥ − 1 , n ≥ 1 , (46) T_{m,n}^{(\mu)}(z)=\det\left[\frac{\mathrm{d}^{j+k}}{\mathrm{d}z^{j+k}}L_{m+n}^{(\mu+1)}(z)\right]_{j,k=0}^{n-1},\qquad m\geq-1,\quad n\geq1, \tag*{(46)} T m , n ( μ ) ( z ) = det [ d z j + k d j + k L m + n ( μ + 1 ) ( z ) ] j , k = 0 n − 1 , m ≥ − 1 , n ≥ 1 , (46) where L n ( μ ) ( z ) L_n^{(\mu)}(z) L n ( μ ) ( z ) is the Laguerre polynomial
L n ( μ ) ( z ) = z − μ e z n ! d n d z n ( z n + μ e − z ) , n ≥ 0. (47) L_n^{(\mu)}(z)=\frac{z^{-\mu}e^z}{n!}\frac{\mathrm{d}^n}{\mathrm{d}z^n}\left(z^{n+\mu}e^{-z}\right),\qquad n\geq0. \tag*{(47)} L n ( μ ) ( z ) = n ! z − μ e z d z n d n ( z n + μ e − z ) , n ≥ 0. (47)
There are three classes of rational solutions of P V \mathrm{P}_{\mathrm{V}} P V (2) which are expressed in terms of generalised Laguerre polynomials:
w m , n ( μ ) ( z ) = T m − 1 , n ( μ ) ( z ) T m − 1 , n + 1 ( μ ) ( z ) T m , n ( μ ) ( z ) T m − 2 , n + 1 ( μ ) ( z ) , ( α , β , γ ) = ( 1 2 m 2 , − 1 2 ( m + 2 n + 1 + μ ) 2 , μ ) , (48) w_{m,n}^{(\mu)}(z)=\frac{T_{m-1,n}^{(\mu)}(z)T_{m-1,n+1}^{(\mu)}(z)}{T_{m,n}^{(\mu)}(z)T_{m-2,n+1}^{(\mu)}(z)},\qquad(\alpha,\beta,\gamma)=\left(\frac{1}{2}m^2,-\frac{1}{2}(m+2n+1+\mu)^2,\mu\right), \tag*{(48)} w m , n ( μ ) ( z ) = T m , n ( μ ) ( z ) T m − 2 , n + 1 ( μ ) ( z ) T m − 1 , n ( μ ) ( z ) T m − 1 , n + 1 ( μ ) ( z ) , ( α , β , γ ) = ( 2 1 m 2 , − 2 1 ( m + 2 n + 1 + μ ) 2 , μ ) , (48)
v m , n ( μ ) ( z ) = m − n + μ m + n + 1 T m , n ( μ − 2 n ) ( z ) T m − 1 , n + 1 ( μ − 2 n − 2 ) ( z ) T m − 1 , n ( μ − 2 n ) ( z ) T m , n + 1 ( μ − 2 n − 2 ) ( z ) , ( α , β , γ ) = ( 1 2 ( m + n + 1 ) 2 , − 1 2 ( m − n + μ ) 2 , μ ) , (49) v_{m,n}^{(\mu)}(z)=\frac{m-n+\mu}{m+n+1}\frac{T_{m,n}^{(\mu-2n)}(z)T_{m-1,n+1}^{(\mu-2n-2)}(z)}{T_{m-1,n}^{(\mu-2n)}(z)T_{m,n+1}^{(\mu-2n-2)}(z)},\qquad(\alpha,\beta,\gamma)=\left(\frac{1}{2}(m+n+1)^2,-\frac{1}{2}(m-n+\mu)^2,\mu\right), \tag*{(49)} v m , n ( μ ) ( z ) = m + n + 1 m − n + μ T m − 1 , n ( μ − 2 n ) ( z ) T m , n + 1 ( μ − 2 n − 2 ) ( z ) T m , n ( μ − 2 n ) ( z ) T m − 1 , n + 1 ( μ − 2 n − 2 ) ( z ) , ( α , β , γ ) = ( 2 1 ( m + n + 1 ) 2 , − 2 1 ( m − n + μ ) 2 , μ ) , (49)
u m , n ( μ ) ( z ) = w m , n ( − μ ) ( − z ) , ( α , β , γ ) = ( 1 2 m 2 , − 1 2 ( m + 2 n + 1 − μ ) 2 , μ ) . (50) u_{m,n}^{(\mu)}(z)=w_{m,n}^{(-\mu)}(-z),\qquad(\alpha,\beta,\gamma)=\left(\frac{1}{2}m^2,-\frac{1}{2}(m+2n+1-\mu)^2,\mu\right). \tag*{(50)} u m , n ( μ ) ( z ) = w m , n ( − μ ) ( − z ) , ( α , β , γ ) = ( 2 1 m 2 , − 2 1 ( m + 2 n + 1 − μ ) 2 , μ ) . (50)
For further details see, for example, [34 ] , which also gives alternative expressions for the rational solutions and the special cases when m = 0 m=0 m = 0 or n = 0 n=0 n = 0 are discussed.
The effect of applying the Bäcklund transformation T ε 1 , ε 2 , ε 3 T_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 (2.2) to the rational solutions w m , n ( μ ) ( z ) w_{m,n}^{(\mu)}(z) w m , n ( μ ) ( z ) with ( a , b , c ) = ( m , m + 2 n + μ + 1 , μ ) (a,b,c)=(m,m+2n+\mu+1,\mu) ( a , b , c ) = ( m , m + 2 n + μ + 1 , μ ) , v m , n ( μ ) ( z ) v_{m,n}^{(\mu)}(z) v m , n ( μ ) ( z ) with ( a , b , c ) = ( m + n + 1 , m − n + μ , μ ) (a,b,c)=(m+n+1,m-n+\mu,\mu) ( a , b , c ) = ( m + n + 1 , m − n + μ , μ ) and u m , n ( μ ) ( z ) u_{m,n}^{(\mu)}(z) u m , n ( μ ) ( z ) with ( a , b , c ) = ( m , m + 2 n − μ + 1 , μ ) (a,b,c)=(m,m+2n-\mu+1,\mu) ( a , b , c ) = ( m , m + 2 n − μ + 1 , μ ) is given in Table 3.1 . We note that the Bäcklund transformation T ε 1 , ε 2 , ε 3 T_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 maps each rational solution to a rational solution in a different class. The effect of applying the Schlesinger transformations R 1 , R 2 , … , R 14 R_1,R_2,\ldots,R_{14} R 1 , R 2 , … , R 14 to the rational solutions w m , n ( μ ) ( z ) w_{m,n}^{(\mu)}(z) w m , n ( μ ) ( z ) with ( a , b , c ) = ( m , m + 2 n + μ + 1 , μ ) (a,b,c)=(m,m+2n+\mu+1,\mu) ( a , b , c ) = ( m , m + 2 n + μ + 1 , μ ) , v m , n ( μ ) ( z ) v_{m,n}^{(\mu)}(z) v m , n ( μ ) ( z ) with ( a , b , c ) = ( m + n + 1 , m − n + μ , μ ) (a,b,c)=(m+n+1,m-n+\mu,\mu) ( a , b , c ) = ( m + n + 1 , m − n + μ , μ ) and u m , n ( μ ) ( z ) u_{m,n}^{(\mu)}(z) u m , n ( μ ) ( z ) with ( a , b , c ) = ( m , m + 2 n − μ + 1 , μ ) (a,b,c)=(m,m+2n-\mu+1,\mu) ( a , b , c ) = ( m , m + 2 n − μ + 1 , μ ) is given in Table 3.2 .
Table 3.1. The effect of applying the Bäcklund transformation T ϵ 1 , ϵ 2 , ϵ 3 \mathcal{T}_{\epsilon_1,\epsilon_2,\epsilon_3} T ϵ 1 , ϵ 2 , ϵ 3 (2.2) to the rational solutions w m , n ( μ ) ( z ; m , m + 2 n + μ + 1 , μ ) w_{m,n}^{(\mu)}(z;m,m+2n+\mu+1,\mu) w m , n ( μ ) ( z ; m , m + 2 n + μ + 1 , μ ) , v m , n ( μ ) ( z ; m + n + 1 , m − n + μ , μ ) v_{m,n}^{(\mu)}(z;m+n+1,m-n+\mu,\mu) v m , n ( μ ) ( z ; m + n + 1 , m − n + μ , μ ) and u m , n ( μ ) ( z ; m , m + 2 n − μ + 1 , μ ) u_{m,n}^{(\mu)}(z;m,m+2n-\mu+1,\mu) u m , n ( μ ) ( z ; m , m + 2 n − μ + 1 , μ ) , where w ~ n , m ( z ) = 1 / w n , m ( μ ) ( z ) \widetilde{w}_{n,m}(z)=1/w_{n,m}^{(\mu)}(z) w n , m ( z ) = 1/ w n , m ( μ ) ( z ) and v ~ n , m ( z ) = 1 / v n , m ( μ ) ( z ) \widetilde{v}_{n,m}(z)=1/v_{n,m}^{(\mu)}(z) v n , m ( z ) = 1/ v n , m ( μ ) ( z ) .
Table 3.2. The effect of applying the Schlesinger transformations R 1 , R 2 , … , R 14 \mathcal{R}_1,\mathcal{R}_2,\ldots,\mathcal{R}_{14} R 1 , R 2 , … , R 14 to the rational solutions w m , n ( μ ) ( z ; m , m + 2 n + μ + 1 , μ ) w_{m,n}^{(\mu)}(z;m,m+2n+\mu+1,\mu) w m , n ( μ ) ( z ; m , m + 2 n + μ + 1 , μ ) , v m , n ( μ ) ( z ; m + n + 1 , m − n + μ , μ ) v_{m,n}^{(\mu)}(z;m+n+1,m-n+\mu,\mu) v m , n ( μ ) ( z ; m + n + 1 , m − n + μ , μ ) and u m , n ( μ ) ( z ; m , m + 2 n − μ + 1 , μ ) u_{m,n}^{(\mu)}(z;m,m+2n-\mu+1,\mu) u m , n ( μ ) ( z ; m , m + 2 n − μ + 1 , μ ) .
Generalised Umemura polynomials
Rational solutions in cases (ii) and (iii) of Theorem 3.1 are expressed in terms of generalised Umemura polynomials . Umemura [108 ] defined some polynomials through a differential-difference equation to describe rational solutions of P V \mathrm{P}_{\mathrm{V}} P V (2) ; see also [29 , 90 , 115 ] . Subsequently these were generalised by Masuda, Ohta and Kajiwara [84 ] , who defined the generalised Umemura polynomial through coupled differential-difference equations and gave a representation as a determinant of Laguerre polynomials. Some properties of generalised Umemura polynomials are discussed in [34 , 83 ] .
Rather than use the generalised Umemura polynomial defined in [84 ] we use an equivalent representation given in Definition 3.4 below, which is obtained using matrix row and column operations. We find this representation simpler to compute and useful to prove algebraic and combinatorial properties of the polynomial, though we do not pursue this further here.
Definition 3.4. The generalised Umemura polynomial U m , n ( κ ) ( z ) U_{m,n}^{(\kappa)}(z) U m , n ( κ ) ( z ) is given by the Wronskian
U m , n ( κ ) ( z ) = c m , n exp ( 1 2 m z ) Wr ( φ 1 ( κ ) , φ 3 ( κ ) , … , φ 2 m − 1 ( κ ) , ψ 1 ( κ ) , ψ 3 ( κ ) , … , ψ 2 n − 1 ( κ ) ) , m , n ≥ 0 , (51) U_{m,n}^{(\kappa)}(z)=c_{m,n}\exp\left(\frac{1}{2}mz\right)\operatorname{Wr}\left(\varphi_1^{(\kappa)},\varphi_3^{(\kappa)},\ldots,\varphi_{2m-1}^{(\kappa)},\psi_1^{(\kappa)},\psi_3^{(\kappa)},\ldots,\psi_{2n-1}^{(\kappa)}\right),\qquad m,n\geq0, \tag*{(51)} U m , n ( κ ) ( z ) = c m , n exp ( 2 1 m z ) Wr ( φ 1 ( κ ) , φ 3 ( κ ) , … , φ 2 m − 1 ( κ ) , ψ 1 ( κ ) , ψ 3 ( κ ) , … , ψ 2 n − 1 ( κ ) ) , m , n ≥ 0 , (51) where
c m , n = ( − 1 ) m 2 ( m + n ) ( m + n + 1 ) / 2 ∏ j = 1 m − 1 ( 2 j + 1 ) ! j ! ∏ k = 1 n − 1 ( 2 k + 1 ) ! k ! . (52) c_{m,n}=(-1)^m2^{(m+n)(m+n+1)/2}\prod_{j=1}^{m-1}\frac{(2j+1)!}{j!}\prod_{k=1}^{n-1}\frac{(2k+1)!}{k!}. \tag*{(52)} c m , n = ( − 1 ) m 2 ( m + n ) ( m + n + 1 ) /2 j = 1 ∏ m − 1 j ! ( 2 j + 1 )! k = 1 ∏ n − 1 k ! ( 2 k + 1 )! . (52)
which makes the polynomial monic, φ n ( κ ) ( z ) = e − z / 2 L n ( κ ) ( 1 2 z ) \varphi_n^{(\kappa)}(z)=e^{-z/2}L_n^{(\kappa)}\left(\frac{1}{2}z\right) φ n ( κ ) ( z ) = e − z /2 L n ( κ ) ( 2 1 z ) and ψ n ( κ ) ( z ) = L n ( κ ) ( − 1 2 z ) \psi_n^{(\kappa)}(z)=L_n^{(\kappa)}\left(-\frac{1}{2}z\right) ψ n ( κ ) ( z ) = L n ( κ ) ( − 2 1 z ) , with κ \kappa κ a parameter and L n ( κ ) ( z ) L_n^{(\kappa)}(z) L n ( κ ) ( z ) the Laguerre polynomial (47) and U 0 , 0 ( κ ) ( z ) = 1 U_{0,0}^{(\kappa)}(z)=1 U 0 , 0 ( κ ) ( z ) = 1 .
There are three classes of rational solutions of PV _{\mathrm{V}} V (2) which are expressed in terms of generalised Umemura polynomials
w ^ m , n ( κ ) ( z ) = − U m , n − 1 ( 2 κ ) ( z ) U m − 1 , n ( 2 κ + 2 ) ( z ) U m , n − 1 ( 2 κ + 2 ) ( z ) U m − 1 , n ( 2 κ ) ( z ) , ( α , β , γ ) = ( 1 2 ( m + κ ) 2 , − 1 2 ( n + κ ) 2 , m + n ) , (53) \widehat{w}_{m,n}^{(\kappa)}(z)=-\frac{U_{m,n-1}^{(2\kappa)}(z)U_{m-1,n}^{(2\kappa+2)}(z)}{U_{m,n-1}^{(2\kappa+2)}(z)U_{m-1,n}^{(2\kappa)}(z)},\qquad(\alpha,\beta,\gamma)=\left(\frac{1}{2}(m+\kappa)^2,-\frac{1}{2}(n+\kappa)^2,m+n\right), \tag*{(53)} w m , n ( κ ) ( z ) = − U m , n − 1 ( 2 κ + 2 ) ( z ) U m − 1 , n ( 2 κ ) ( z ) U m , n − 1 ( 2 κ ) ( z ) U m − 1 , n ( 2 κ + 2 ) ( z ) , ( α , β , γ ) = ( 2 1 ( m + κ ) 2 , − 2 1 ( n + κ ) 2 , m + n ) , (53)
v ^ m , n ( κ ) ( z ) = − U m − 1 , n − 1 ( 2 κ − 2 n + 3 ) ( z ) U m , n ( 2 κ − 2 n − 1 ) ( z ) U m − 1 , n − 1 ( 2 κ − 2 n + 1 ) ( z ) U m , n ( 2 κ − 2 n + 1 ) ( z ) , ( α , β , γ ) = ( 1 2 ( m + κ ) 2 , − 1 2 ( n − κ ) 2 , m − n ) , (54) \widehat{v}_{m,n}^{(\kappa)}(z)=-\frac{U_{m-1,n-1}^{(2\kappa-2n+3)}(z)U_{m,n}^{(2\kappa-2n-1)}(z)}{U_{m-1,n-1}^{(2\kappa-2n+1)}(z)U_{m,n}^{(2\kappa-2n+1)}(z)},\qquad(\alpha,\beta,\gamma)=\left(\frac{1}{2}(m+\kappa)^2,-\frac{1}{2}(n-\kappa)^2,m-n\right), \tag*{(54)} v m , n ( κ ) ( z ) = − U m − 1 , n − 1 ( 2 κ − 2 n + 1 ) ( z ) U m , n ( 2 κ − 2 n + 1 ) ( z ) U m − 1 , n − 1 ( 2 κ − 2 n + 3 ) ( z ) U m , n ( 2 κ − 2 n − 1 ) ( z ) , ( α , β , γ ) = ( 2 1 ( m + κ ) 2 , − 2 1 ( n − κ ) 2 , m − n ) , (54)
u ^ m , n ( κ ) ( z ) = − U m , n − 1 ( κ + 1 ) ( z ) U m , n + 1 ( κ − 1 ) ( z ) U m − 1 , n ( κ + 1 ) ( z ) U m + 1 , n ( κ − 1 ) ( z ) , ( α , β , γ ) = ( 1 2 ( m + 1 2 ) 2 , − 1 2 ( n + 1 2 ) 2 , m + n + κ ) , (55) \widehat{u}_{m,n}^{(\kappa)}(z)=-\frac{U_{m,n-1}^{(\kappa+1)}(z)U_{m,n+1}^{(\kappa-1)}(z)}{U_{m-1,n}^{(\kappa+1)}(z)U_{m+1,n}^{(\kappa-1)}(z)},\qquad(\alpha,\beta,\gamma)=\left(\frac{1}{2}\left(m+\frac{1}{2}\right)^2,-\frac{1}{2}\left(n+\frac{1}{2}\right)^2,m+n+\kappa\right), \tag*{(55)} u m , n ( κ ) ( z ) = − U m − 1 , n ( κ + 1 ) ( z ) U m + 1 , n ( κ − 1 ) ( z ) U m , n − 1 ( κ + 1 ) ( z ) U m , n + 1 ( κ − 1 ) ( z ) , ( α , β , γ ) = ( 2 1 ( m + 2 1 ) 2 , − 2 1 ( n + 2 1 ) 2 , m + n + κ ) , (55)
when m ≥ 1 m\ge1 m ≥ 1 and n ≥ 1 n\ge1 n ≥ 1 . These rational solutions were given in [84 ] ; see also [34 ] , where alternative expressions for the rational solutions are given and the special cases when m = 0 m=0 m = 0 or n = 0 n=0 n = 0 are discussed.
Further we note that for m ≥ 1 m \ge1 m ≥ 1
u ^ m , 0 ( κ ) ( z ) = − U m , 0 ( κ ) ( z ) U m , 0 ( κ − 1 ) ( z ) U m − 1 , 0 ( κ + 1 ) ( z ) U m + 1 , 0 ( κ − 1 ) ( z ) , ( α , β , γ ) = ( 1 2 ( m + 1 2 ) 2 , − 1 8 , m + κ ) , (56) \widehat{u}_{m,0}^{(\kappa)}(z)=-\frac{U_{m,0}^{(\kappa)}(z)U_{m,0}^{(\kappa-1)}(z)}{U_{m-1,0}^{(\kappa+1)}(z)U_{m+1,0}^{(\kappa-1)}(z)},\qquad(\alpha,\beta,\gamma)=\left(\frac{1}{2}\left(m+\frac{1}{2}\right)^2,-\frac{1}{8},m+\kappa\right), \tag*{(56)} u m , 0 ( κ ) ( z ) = − U m − 1 , 0 ( κ + 1 ) ( z ) U m + 1 , 0 ( κ − 1 ) ( z ) U m , 0 ( κ ) ( z ) U m , 0 ( κ − 1 ) ( z ) , ( α , β , γ ) = ( 2 1 ( m + 2 1 ) 2 , − 8 1 , m + κ ) , (56)
u ^ 0 , 0 ( κ ) ( z ) = − U 0 , 1 ( κ ) ( z ) U 1 , 0 ( κ ) ( z ) = − z + 2 κ + 2 z − 2 κ − 2 , ( α , β , γ ) = ( 1 8 , − 1 8 , κ + 1 ) . (57) \widehat{u}_{0,0}^{(\kappa)}(z)=-\frac{U_{0,1}^{(\kappa)}(z)}{U_{1,0}^{(\kappa)}(z)}=-\frac{z+2\kappa+2}{z-2\kappa-2},\qquad(\alpha,\beta,\gamma)=\left(\frac{1}{8},-\frac{1}{8},\kappa+1\right). \tag*{(57)} u 0 , 0 ( κ ) ( z ) = − U 1 , 0 ( κ ) ( z ) U 0 , 1 ( κ ) ( z ) = − z − 2 κ − 2 z + 2 κ + 2 , ( α , β , γ ) = ( 8 1 , − 8 1 , κ + 1 ) . (57)
The effect of applying the Bäcklund transformation T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 (2.2) to the rational solutions w ^ m , n ( κ ) ( z ) \widehat{w}_{m,n}^{(\kappa)}(z) w m , n ( κ ) ( z ) with ( a , b , c ) = ( m + κ , n + κ , m + n ) (a,b,c)=(m+\kappa,n+\kappa,m+n) ( a , b , c ) = ( m + κ , n + κ , m + n ) , v ^ m , n ( κ ) ( z ) \widehat{v}_{m,n}^{(\kappa)}(z) v m , n ( κ ) ( z ) with ( a , b , c ) = ( m + κ , n − κ , m − n ) (a,b,c)=(m+\kappa,n-\kappa,m-n) ( a , b , c ) = ( m + κ , n − κ , m − n ) and u ^ m , n ( κ ) ( z ) \widehat{u}_{m,n}^{(\kappa)}(z) u m , n ( κ ) ( z ) with ( a , b , c ) = ( m + 1 2 , n + 1 2 , m + n + κ ) (a,b,c)=\left(m+\frac{1}{2},n+\frac{1}{2},m+n+\kappa\right) ( a , b , c ) = ( m + 2 1 , n + 2 1 , m + n + κ ) is given in Table 3.3. As for rational solutions expressed in terms of generalised Laguerre polynomials, the Bäcklund transformation T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 maps each rational solution to a rational solution in a different class. The effect of applying the Schlesinger transformations R 1 , R 2 , … , R 14 R_1,R_2,\ldots,R_{14} R 1 , R 2 , … , R 14 to the rational solutions w ^ m , n ( κ ) ( z ) \widehat{w}_{m,n}^{(\kappa)}(z) w m , n ( κ ) ( z ) with ( a , b , c ) = ( m + κ , n + κ , m + n ) (a,b,c)=(m+\kappa,n+\kappa,m+n) ( a , b , c ) = ( m + κ , n + κ , m + n ) , v ^ m , n ( κ ) ( z ) \widehat{v}_{m,n}^{(\kappa)}(z) v m , n ( κ ) ( z ) with ( a , b , c ) = ( m + κ , n − κ , m − n ) (a,b,c)=(m+\kappa,n-\kappa,m-n) ( a , b , c ) = ( m + κ , n − κ , m − n ) and u ^ m , n ( κ ) ( z ) \widehat{u}_{m,n}^{(\kappa)}(z) u m , n ( κ ) ( z ) with ( a , b , c ) = ( m + 1 2 , n + 1 2 , m + n + κ ) (a,b,c)=\left(m+\frac{1}{2},n+\frac{1}{2},m+n+\kappa\right) ( a , b , c ) = ( m + 2 1 , n + 2 1 , m + n + κ ) is given in Table 3.4.
Table 3. The effect of applying the Bäcklund transformation T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 (2.2) to the rational solutions w ^ m , n ( κ ) ( z ; m + κ , n + κ , m + n ) \widehat{w}_{m,n}^{(\kappa)}(z;m+\kappa,n+\kappa,m+n) w m , n ( κ ) ( z ; m + κ , n + κ , m + n ) , v ^ m , n ( κ ) ( z ; m + κ , n − κ , m − n ) \widehat{v}_{m,n}^{(\kappa)}(z;m+\kappa,n-\kappa,m-n) v m , n ( κ ) ( z ; m + κ , n − κ , m − n ) and u ^ m , n ( κ ) ( z ; m + 1 2 , n + 1 2 , m + n + κ ) \widehat{u}_{m,n}^{(\kappa)}(z;m+\frac{1}{2},n+\frac{1}{2},m+n+\kappa) u m , n ( κ ) ( z ; m + 2 1 , n + 2 1 , m + n + κ ) .
Table 4. The effect of applying the Schlesinger transformations R 1 , R 2 , … , R 14 R_1,R_2,\ldots,R_{14} R 1 , R 2 , … , R 14 to the rational solutions w ^ m , n ( κ ) ( z ; m + κ , n + κ , m + n ) \widehat{w}_{m,n}^{(\kappa)}(z;m+\kappa,n+\kappa,m+n) w m , n ( κ ) ( z ; m + κ , n + κ , m + n ) , v ^ m , n ( κ ) ( z ; m + κ , n − κ , m − n ) \widehat{v}_{m,n}^{(\kappa)}(z;m+\kappa,n-\kappa,m-n) v m , n ( κ ) ( z ; m + κ , n − κ , m − n ) and u ^ m , n ( κ ) ( z ; m + 1 2 , n + 1 2 , m + n + κ ) \widehat{u}_{m,n}^{(\kappa)}(z;m+\frac{1}{2},n+\frac{1}{2},m+n+\kappa) u m , n ( κ ) ( z ; m + 2 1 , n + 2 1 , m + n + κ ) .
Non-unique modified Bessel function solutions
In this section we derive non-unique modified Bessel function solutions of P V \mathrm{P}_{\mathrm V} P V (2) , i.e. two distinct solutions in terms of modified Bessel functions, which are not related by the symmetries (6) , that satisfy P V \mathrm{P}_{\mathrm V} P V for the same values of the parameters. Further we show that there is also a rational solution expressed in terms of generalised Umemura polynomials for these parameters.
Special function solutions of P V \mathrm{P}_{\mathrm V} P V (2) are usually expressed in terms of the Whittaker functions M κ , μ ( z ) M_{\kappa,\mu}(z) M κ , μ ( z ) and W κ , μ ( z ) W_{\kappa,\mu}(z) W κ , μ ( z ) , or equivalently the Kummer functions M ( a , b , z ) M(a,b,z) M ( a , b , z ) and U ( a , b , z ) U(a,b,z) U ( a , b , z ) , cf. [99 §32.10(v)] . However if b = 2 a + n b=2a+n b = 2 a + n , with n ∈ Z n\in\mathbb{Z} n ∈ Z , then the Kummer functions M ( a , b , z ) M(a,b,z) M ( a , b , z ) and U ( a , b , z ) U(a,b,z) U ( a , b , z ) can be respectively expressed in terms of the modified Bessel functions I ν ( 1 2 z ) I_{\nu}(\frac{1}{2}z) I ν ( 2 1 z ) and K ν ( 1 2 z ) K_{\nu}(\frac{1}{2}z) K ν ( 2 1 z ) , see [99 §13.6(iii)] . Specifically, if b = 2 a = 2 ν + 1 b=2a=2\nu+1 b = 2 a = 2 ν + 1 , then it is well known that
M ( ν + 1 2 , 2 ν + 1 , 2 z ) = Γ ( ν + 1 ) e z ( 1 2 z ) − ν I ν ( z ) , (58) M\left(\nu+\frac{1}{2},2\nu+1,2z\right)=\Gamma(\nu+1)e^{z}\left(\frac{1}{2}z\right)^{-\nu}I_{\nu}(z), \tag*{(58)} M ( ν + 2 1 , 2 ν + 1 , 2 z ) = Γ ( ν + 1 ) e z ( 2 1 z ) − ν I ν ( z ) , (58)
U ( ν + 1 2 , 2 ν + 1 , 2 z ) = π − 1 / 2 e z ( 2 z ) − ν K ν ( z ) , (59) U\left(\nu+\frac{1}{2},2\nu+1,2z\right)=\pi^{-1/2}e^{z}(2z)^{-\nu}K_{\nu}(z), \tag*{(59)} U ( ν + 2 1 , 2 ν + 1 , 2 z ) = π − 1/2 e z ( 2 z ) − ν K ν ( z ) , (59)
with I ν ( z ) I_{\nu}(z) I ν ( z ) and K ν ( z ) K_{\nu}(z) K ν ( z ) modified Bessel functions. More generally, if a = ν + 1 2 a=\nu+\frac{1}{2} a = ν + 2 1 and b = 2 a + n = 2 ν + n + 1 b=2a+n=2\nu+n+1 b = 2 a + n = 2 ν + n + 1 , with n ∈ Z n\in\mathbb{Z} n ∈ Z , then for n ≥ 0 n\geq0 n ≥ 0
M ( ν + 1 2 , 2 ν + 1 + n , 2 z ) = Γ ( ν ) e z ( 1 2 z ) − ν ∑ k = 0 n ( − 1 ) k ( n k ) ( 2 ν ) k ( ν + k ) ( 2 ν + 1 + n ) k I ν + k ( z ) , (60) M\left(\nu+\frac{1}{2},2\nu+1+n,2z\right)=\Gamma(\nu)e^{z}\left(\frac{1}{2}z\right)^{-\nu}\sum_{k=0}^{n}(-1)^k\binom{n}{k}\frac{(2\nu)_k(\nu+k)}{(2\nu+1+n)_k}I_{\nu+k}(z), \tag*{(60)} M ( ν + 2 1 , 2 ν + 1 + n , 2 z ) = Γ ( ν ) e z ( 2 1 z ) − ν k = 0 ∑ n ( − 1 ) k ( k n ) ( 2 ν + 1 + n ) k ( 2 ν ) k ( ν + k ) I ν + k ( z ) , (60)
M ( ν + 1 2 , 2 ν + 1 − n , 2 z ) = Γ ( ν − n ) e z ( 1 2 z ) n − ν ∑ k = 0 n ( n k ) ( 2 ν − 2 n ) k ( ν − n + k ) ( 2 ν + 1 − n ) k I ν + k − n ( z ) , (61) M\left(\nu+\frac{1}{2},2\nu+1-n,2z\right)=\Gamma(\nu-n)e^{z}\left(\frac{1}{2}z\right)^{n-\nu}\sum_{k=0}^{n}\binom{n}{k}\frac{(2\nu-2n)_k(\nu-n+k)}{(2\nu+1-n)_k}I_{\nu+k-n}(z), \tag*{(61)} M ( ν + 2 1 , 2 ν + 1 − n , 2 z ) = Γ ( ν − n ) e z ( 2 1 z ) n − ν k = 0 ∑ n ( k n ) ( 2 ν + 1 − n ) k ( 2 ν − 2 n ) k ( ν − n + k ) I ν + k − n ( z ) , (61)
U ( ν + 1 2 , 2 ν + 1 + n , 2 z ) = 2 ( ν + 1 2 ) n e z ( 2 z ) − ν π ∑ k = 0 n ( n k ) ν + k ( 2 ν + k ) n + 1 K ν + k ( z ) , (62) U\left(\nu+\frac{1}{2},2\nu+1+n,2z\right)=\frac{2\left(\nu+\frac{1}{2}\right)_n e^{z}(2z)^{-\nu}}{\sqrt{\pi}}\sum_{k=0}^{n}\binom{n}{k}\frac{\nu+k}{(2\nu+k)_{n+1}}K_{\nu+k}(z), \tag*{(62)} U ( ν + 2 1 , 2 ν + 1 + n , 2 z ) = π 2 ( ν + 2 1 ) n e z ( 2 z ) − ν k = 0 ∑ n ( k n ) ( 2 ν + k ) n + 1 ν + k K ν + k ( z ) , (62)
U ( ν + 1 2 , 2 ν + 1 − n , 2 z ) = 2 ( − 1 ) n e z ( 2 z ) n − ν π ∑ k = 0 n ( − 1 ) k ( n k ) ν + k − n ( 2 ν + k − 2 n ) n + 1 K ν + k − n ( z ) . (63) U\left(\nu+\frac{1}{2},2\nu+1-n,2z\right)=\frac{2(-1)^n e^{z}(2z)^{n-\nu}}{\sqrt{\pi}}\sum_{k=0}^{n}(-1)^k\binom{n}{k}\frac{\nu+k-n}{(2\nu+k-2n)_{n+1}}K_{\nu+k-n}(z). \tag*{(63)} U ( ν + 2 1 , 2 ν + 1 − n , 2 z ) = π 2 ( − 1 ) n e z ( 2 z ) n − ν k = 0 ∑ n ( − 1 ) k ( k n ) ( 2 ν + k − 2 n ) n + 1 ν + k − n K ν + k − n ( z ) . (63)
It is convenient to define the function Z ν ( z ) Z_{\nu}(z) Z ν ( z ) when discussing solutions of P V \mathrm{P}_{\mathrm V} P V (2) which involve modified Bessel functions.
Definition 4.1. Define the function Z ν ( z ) Z_{\nu}(z) Z ν ( z ) by
Z ν ( z ) = { C 1 I ν ( z ) + C 2 I − ν ( z ) , if ν ∉ Z , C 1 I n ( z ) + C 2 ( − 1 ) n K n ( z ) , if n ∈ Z , n ≥ 0. (64) Z_{\nu}(z)=
\begin{cases}
C_{1}I_{\nu}(z)+C_{2}I_{-\nu}(z), & \text{if }\nu\notin\mathbb{Z},\\
C_{1}I_{n}(z)+C_{2}(-1)^{n}K_{n}(z), & \text{if }n\in\mathbb{Z},\ n\geq0.
\end{cases} \tag*{(64)} Z ν ( z ) = { C 1 I ν ( z ) + C 2 I − ν ( z ) , C 1 I n ( z ) + C 2 ( − 1 ) n K n ( z ) , if ν ∈ / Z , if n ∈ Z , n ≥ 0. (64) where I n ( z ) I_{n}(z) I n ( z ) and K n ( z ) K_{n}(z) K n ( z ) are modified Bessel functions, with C 1 C_{1} C 1 and C 2 C_{2} C 2 arbitrary constants.
If in P V \mathrm{P}_{\mathrm V} P V (2) , α = − β = 1 8 \alpha=-\beta=\frac{1}{8} α = − β = 8 1 , then a = ± 1 2 a=\pm\frac{1}{2} a = ± 2 1 and b = ± 1 2 b=\pm\frac{1}{2} b = ± 2 1 , independently. Hence if we use an appropriate Bäcklund transformation of P V \mathrm{P}_{\mathrm V} P V (2) that changes the values of a a a and/or b b b by ± 1 \pm1 ± 1 , then a a a and/or b b b can switch sign, so α \alpha α and/or β \beta β are unchanged. This is illustrated in the following example.
Example 4.2. Consider the solution
w 0 ( z ) = Z 1 ( 1 2 z ) − Z 0 ( 1 2 z ) Z 1 ( 1 2 z ) + Z 0 ( 1 2 z ) , (65) w_{0}(z)=\frac{Z_{1}\left(\frac{1}{2}z\right)-Z_{0}\left(\frac{1}{2}z\right)}{Z_{1}\left(\frac{1}{2}z\right)+Z_{0}\left(\frac{1}{2}z\right)}, \tag*{(65)} w 0 ( z ) = Z 1 ( 2 1 z ) + Z 0 ( 2 1 z ) Z 1 ( 2 1 z ) − Z 0 ( 2 1 z ) , (65) with Z ν ( z ) Z_{\nu}(z) Z ν ( z ) given by (64) , which satisfies P V \mathrm{P}_{\mathrm V} P V (2) for the parameters ( α , β , γ ) = ( 1 8 , − 1 8 , 0 ) (\alpha,\beta,\gamma)=\left(\frac{1}{8},-\frac{1}{8},0\right) ( α , β , γ ) = ( 8 1 , − 8 1 , 0 ) . Applying the Bäcklund transformations T 1 , 1 , 1 \mathcal{T}_{1,1,1} T 1 , 1 , 1 and T − 1 , − 1 , − 1 \mathcal{T}_{-1,-1,-1} T − 1 , − 1 , − 1 , with ( a , b , c ) = ( 1 2 , − 1 2 , 0 ) (a,b,c)=\left(\frac{1}{2},-\frac{1}{2},0\right) ( a , b , c ) = ( 2 1 , − 2 1 , 0 ) , which lead to the same values of ( α , β , γ ) (\alpha,\beta,\gamma) ( α , β , γ ) , to w 0 ( z ) w_{0}(z) w 0 ( z ) gives
w 1 ( z ) = T 1 , 1 , 1 ( w 0 ; 1 2 , − 1 2 , 0 ) = 1 − z + z Z 1 ( 1 2 z ) Z 0 ( 1 2 z ) , (66) w_{1}(z)=\mathcal{T}_{1,1,1}\left(w_{0};\frac{1}{2},-\frac{1}{2},0\right)=1-z+z\frac{Z_{1}\left(\frac{1}{2}z\right)}{Z_{0}\left(\frac{1}{2}z\right)}, \tag*{(66)} w 1 ( z ) = T 1 , 1 , 1 ( w 0 ; 2 1 , − 2 1 , 0 ) = 1 − z + z Z 0 ( 2 1 z ) Z 1 ( 2 1 z ) , (66) v 1 ( z ) = T − 1 , − 1 , − 1 ( w 0 ; 1 2 , − 1 2 , 0 ) = Z 0 ( 1 2 z ) z Z 1 ( 1 2 z ) + ( z + 1 ) Z 0 ( 1 2 z ) . (67) v_{1}(z)=\mathcal{T}_{-1,-1,-1}\left(w_{0};\frac{1}{2},-\frac{1}{2},0\right)=\frac{Z_{0}\left(\frac{1}{2}z\right)}{zZ_{1}\left(\frac{1}{2}z\right)+(z+1)Z_{0}\left(\frac{1}{2}z\right)}. \tag*{(67)} v 1 ( z ) = T − 1 , − 1 , − 1 ( w 0 ; 2 1 , − 2 1 , 0 ) = z Z 1 ( 2 1 z ) + ( z + 1 ) Z 0 ( 2 1 z ) Z 0 ( 2 1 z ) . (67)
These solutions both satisfy P V P_{\mathrm V} P V (2) for the parameters ( α , β , γ ) = ( 1 8 , − 1 8 , − 1 ) (\alpha,\beta,\gamma) = \left(\frac{1}{8},-\frac{1}{8},-1\right) ( α , β , γ ) = ( 8 1 , − 8 1 , − 1 ) since
T 1 , 1 , 1 ( 1 2 , − 1 2 , 0 ) = ( 1 2 , − 1 2 , − 1 ) , T − 1 , − 1 , − 1 ( 1 2 , − 1 2 , 0 ) = ( − 1 2 , 1 2 , − 1 ) . \mathcal{T}_{1,1,1}\left(\frac{1}{2},-\frac{1}{2},0\right)=\left(\frac{1}{2},-\frac{1}{2},-1\right),\qquad\mathcal{T}_{-1,-1,-1}\left(\frac{1}{2},-\frac{1}{2},0\right)=\left(-\frac{1}{2},\frac{1}{2},-1\right). T 1 , 1 , 1 ( 2 1 , − 2 1 , 0 ) = ( 2 1 , − 2 1 , − 1 ) , T − 1 , − 1 , − 1 ( 2 1 , − 2 1 , 0 ) = ( − 2 1 , 2 1 , − 1 ) .
We note that v 1 ( z ) = 1 / w 1 ( − z ) v_1(z)=1/w_1(-z) v 1 ( z ) = 1/ w 1 ( − z ) , after redefining the constants, since for n ∈ Z n\in\mathbb{Z} n ∈ Z
I n ( − z ) = ( − 1 ) n I n ( z ) , K n ( − z ) = ( − 1 ) n K n ( z ) − i π I n ( z ) , I_n(-z)=(-1)^n I_n(z),\qquad K_n(-z)=(-1)^n K_n(z)-\mathrm{i}\pi I_n(z), I n ( − z ) = ( − 1 ) n I n ( z ) , K n ( − z ) = ( − 1 ) n K n ( z ) − i π I n ( z ) ,
cf. [99 §10.34] , so the solutions are equivalent due to the symmetries (6) . The rational solution
u 1 ( z ) = T 1 , 1 , 1 ( − 1 ; 1 2 , − 1 2 , 0 ) = − z − 2 z + 2 , (68) u_1(z)=\mathcal{T}_{1,1,1}\left(-1;\frac{1}{2},-\frac{1}{2},0\right)=-\frac{z-2}{z+2}, \tag*{(68)} u 1 ( z ) = T 1 , 1 , 1 ( − 1 ; 2 1 , − 2 1 , 0 ) = − z + 2 z − 2 , (68)
which is not a special case of either w 1 ( z ) w_1(z) w 1 ( z ) or v 1 ( z ) v_1(z) v 1 ( z ) , also satisfies P V P_{\mathrm V} P V (2) for the parameters ( α , β , γ ) = ( 1 8 , − 1 8 , − 1 ) (\alpha,\beta,\gamma)=\left(\frac{1}{8},-\frac{1}{8},-1\right) ( α , β , γ ) = ( 8 1 , − 8 1 , − 1 ) . Further u 1 ( z ) u_1(z) u 1 ( z ) is the special case of a rational solution expressed in terms of Umemura polynomials given by u 1 ( z ) = u ^ 0 , 0 ( − 2 ) ( z ) u_1(z)=\widehat{u}_{0,0}^{(-2)}(z) u 1 ( z ) = u 0 , 0 ( − 2 ) ( z ) , recall (57) .
Next we apply the Bäcklund transformation T 1 , 1 , 1 3 \mathcal{T}_{1,1,1}^{3} T 1 , 1 , 1 3 , with ( a , b , c ) = ( 1 2 , 1 2 , − 1 ) (a,b,c)=\left(\frac{1}{2},\frac{1}{2},-1\right) ( a , b , c ) = ( 2 1 , 2 1 , − 1 ) , to the three solutions w 1 ( z ) w_1(z) w 1 ( z ) , v 1 ( z ) v_1(z) v 1 ( z ) and u 1 ( z ) u_1(z) u 1 ( z ) given by (4.5) and obtain the solutions
w 2 ( z ) = T 1 , 1 , 1 3 ( w 1 ; 1 2 , 1 2 , − 1 ) = − Z 1 ( 1 2 z ) − Z 0 ( 1 2 z ) Z 1 ( 1 2 z ) + Z 0 ( 1 2 z ) , (69) w_2(z)=\mathcal{T}_{1,1,1}^{3}\left(w_1;\frac{1}{2},\frac{1}{2},-1\right)=-\frac{Z_1\left(\frac{1}{2}z\right)-Z_0\left(\frac{1}{2}z\right)}{Z_1\left(\frac{1}{2}z\right)+Z_0\left(\frac{1}{2}z\right)}, \tag*{(69)} w 2 ( z ) = T 1 , 1 , 1 3 ( w 1 ; 2 1 , 2 1 , − 1 ) = − Z 1 ( 2 1 z ) + Z 0 ( 2 1 z ) Z 1 ( 2 1 z ) − Z 0 ( 2 1 z ) , (69)
v 2 ( z ) = T 1 , 1 , 1 3 ( v 1 ; 1 2 , 1 2 , − 1 ) = [ Z 1 ( 1 2 z ) + Z 0 ( 1 2 z ) ] [ z Z 1 2 ( 1 2 z ) − ( z − 2 ) Z 0 2 ( 1 2 z ) ] [ Z 1 ( 1 2 z ) − Z 0 ( 1 2 z ) ] [ z Z 1 2 ( 1 2 z ) − ( z + 2 ) Z 0 2 ( 1 2 z ) ] , (70) v_2(z)=\mathcal{T}_{1,1,1}^{3}\left(v_1;\frac{1}{2},\frac{1}{2},-1\right)=\frac{\left[Z_1\left(\frac{1}{2}z\right)+Z_0\left(\frac{1}{2}z\right)\right]\left[zZ_1^2\left(\frac{1}{2}z\right)-(z-2)Z_0^2\left(\frac{1}{2}z\right)\right]}{\left[Z_1\left(\frac{1}{2}z\right)-Z_0\left(\frac{1}{2}z\right)\right]\left[zZ_1^2\left(\frac{1}{2}z\right)-(z+2)Z_0^2\left(\frac{1}{2}z\right)\right]}, \tag*{(70)} v 2 ( z ) = T 1 , 1 , 1 3 ( v 1 ; 2 1 , 2 1 , − 1 ) = [ Z 1 ( 2 1 z ) − Z 0 ( 2 1 z ) ] [ z Z 1 2 ( 2 1 z ) − ( z + 2 ) Z 0 2 ( 2 1 z ) ] [ Z 1 ( 2 1 z ) + Z 0 ( 2 1 z ) ] [ z Z 1 2 ( 2 1 z ) − ( z − 2 ) Z 0 2 ( 2 1 z ) ] , (70)
u 2 ( z ) = T 1 , 1 , 1 3 ( u 1 ; 1 2 , 1 2 , − 1 ) = − z − 4 z + 4 , (71) u_2(z)=\mathcal{T}_{1,1,1}^{3}\left(u_1;\frac{1}{2},\frac{1}{2},-1\right)=-\frac{z-4}{z+4}, \tag*{(71)} u 2 ( z ) = T 1 , 1 , 1 3 ( u 1 ; 2 1 , 2 1 , − 1 ) = − z + 4 z − 4 , (71)
which all satisfy P V P_{\mathrm V} P V (2) for the parameters ( α , β , γ ) = ( 1 8 , − 1 8 , − 2 ) (\alpha,\beta,\gamma)=\left(\frac{1}{8},-\frac{1}{8},-2\right) ( α , β , γ ) = ( 8 1 , − 8 1 , − 2 ) since T 1 , 1 , 1 3 ( 1 2 , 1 2 , − 1 ) = ( 1 2 , − 1 2 , − 2 ) \mathcal{T}_{1,1,1}^{3}\left(\frac{1}{2},\frac{1}{2},-1\right)=\left(\frac{1}{2},-\frac{1}{2},-2\right) T 1 , 1 , 1 3 ( 2 1 , 2 1 , − 1 ) = ( 2 1 , − 2 1 , − 2 ) . Hence we have three solutions, two distinct solutions in terms of modified Bessel functions and a rational solution, which all satisfy P V P_{\mathrm V} P V (2) for the same values of the parameters.
We remark that w 2 ( z ) = − w 0 ( z ) w_2(z)=-w_0(z) w 2 ( z ) = − w 0 ( z ) and the solutions w 2 ( z ) w_2(z) w 2 ( z ) and v 2 ( z ) v_2(z) v 2 ( z ) given by (69) and (70) respectively are related by the Bäcklund transformation T − 1 , 1 , − 1 ∘ T − 1 , 1 , 1 \mathcal{T}_{-1,1,-1}\circ\mathcal{T}_{-1,1,1} T − 1 , 1 , − 1 ∘ T − 1 , 1 , 1 , with ( a , b , c ) = ( 1 2 , 1 2 , − 2 ) (a,b,c)=\left(\frac{1}{2},\frac{1}{2},-2\right) ( a , b , c ) = ( 2 1 , 2 1 , − 2 ) , since
T − 1 , 1 , − 1 ∘ T − 1 , 1 , 1 ( w 2 ; 1 2 , 1 2 , − 2 ) = ( v 2 ; − 1 2 , 1 2 , − 2 ) . \mathcal{T}_{-1,1,-1}\circ\mathcal{T}_{-1,1,1}\left(w_2;\frac{1}{2},\frac{1}{2},-2\right)=\left(v_2;-\frac{1}{2},\frac{1}{2},-2\right). T − 1 , 1 , − 1 ∘ T − 1 , 1 , 1 ( w 2 ; 2 1 , 2 1 , − 2 ) = ( v 2 ; − 2 1 , 2 1 , − 2 ) .
Further u 2 ( z ) = u ^ 0 , 0 ( − 3 ) ( z ) u_2(z)=\widehat{u}_{0,0}^{(-3)}(z) u 2 ( z ) = u 0 , 0 ( − 3 ) ( z ) , recall (57) .
Example 4.3. Applying the Bäcklund transformation T − 1 , 1 , − 1 ∘ T − 1 , − 1 , 1 \mathcal{T}_{-1,1,-1}\circ\mathcal{T}_{-1,-1,1} T − 1 , 1 , − 1 ∘ T − 1 , − 1 , 1 , with ( a , b , c ) = ( 1 2 , 1 2 , − 2 ) (a,b,c)=\left(\frac{1}{2},\frac{1}{2},-2\right) ( a , b , c ) = ( 2 1 , 2 1 , − 2 ) , to the three solutions w 2 ( z ) w_2(z) w 2 ( z ) , v 2 ( z ) v_2(z) v 2 ( z ) and u 2 ( z ) u_2(z) u 2 ( z ) given by (4.6) we obtain the solutions
w 3 ( z ) = − ( 2 z − 1 ) Z 1 ( 1 2 z ) − ( 2 z − 3 ) Z 0 ( 1 2 z ) 3 [ Z 1 ( 1 2 z ) − Z 0 ( 1 2 z ) ] = − 2 z − 1 3 − 2 Z 0 ( 1 2 z ) 3 [ Z 1 ( 1 2 z ) − Z 0 ( 1 2 z ) ] , (72) w_3(z)=-\frac{(2z-1)Z_1\left(\frac{1}{2}z\right)-(2z-3)Z_0\left(\frac{1}{2}z\right)}{3\left[Z_1\left(\frac{1}{2}z\right)-Z_0\left(\frac{1}{2}z\right)\right]}=-\frac{2z-1}{3}-\frac{2Z_0\left(\frac{1}{2}z\right)}{3\left[Z_1\left(\frac{1}{2}z\right)-Z_0\left(\frac{1}{2}z\right)\right]}, \tag*{(72)} w 3 ( z ) = − 3 [ Z 1 ( 2 1 z ) − Z 0 ( 2 1 z ) ] ( 2 z − 1 ) Z 1 ( 2 1 z ) − ( 2 z − 3 ) Z 0 ( 2 1 z ) = − 3 2 z − 1 − 3 [ Z 1 ( 2 1 z ) − Z 0 ( 2 1 z ) ] 2 Z 0 ( 2 1 z ) , (72)
v 3 ( z ) = − [ ( 2 z + 1 ) Z 1 ( 1 2 z ) + ( 2 z + 3 ) Z 0 ( 1 2 z ) ] [ z Z 1 2 ( 1 2 z ) − ( z + 2 ) Z 0 2 ( 1 2 z ) ] [ Z 1 ( 1 2 z ) + Z 0 ( 1 2 z ) ] [ z ( 4 z − 1 ) Z 1 2 ( 1 2 z ) + 8 z ( z + 1 ) Z 0 ( 1 2 z ) Z 1 ( 1 2 z ) + ( 4 z 2 + 9 z + 6 ) Z 0 2 ( 1 2 z ) ] = − 2 z + 1 4 z − 1 + 2 Z 0 ( 1 2 z ) 3 [ Z 1 ( 1 2 z ) + Z 0 ( 1 2 z ) ] + 4 z ( 2 z + 1 ) ( 2 z + 7 ) Z 1 ( 1 2 z ) + 4 ( 4 z 3 + 20 z 2 + 21 z + 3 ) Z 0 ( 1 2 z ) 3 ( 4 z − 1 ) [ z ( 4 z − 1 ) Z 1 2 ( 1 2 z ) + 8 z ( z + 1 ) Z 0 ( 1 2 z ) Z 1 ( 1 2 z ) + ( 4 z 2 + 9 z + 6 ) Z 0 2 ( 1 2 z ) ] , (73) \begin{aligned}
v_3(z)={}&-\frac{\left[(2z+1)Z_1\left(\frac{1}{2}z\right)+(2z+3)Z_0\left(\frac{1}{2}z\right)\right]\left[zZ_1^2\left(\frac{1}{2}z\right)-(z+2)Z_0^2\left(\frac{1}{2}z\right)\right]}{\left[Z_1\left(\frac{1}{2}z\right)+Z_0\left(\frac{1}{2}z\right)\right]\left[z(4z-1)Z_1^2\left(\frac{1}{2}z\right)+8z(z+1)Z_0\left(\frac{1}{2}z\right)Z_1\left(\frac{1}{2}z\right)+(4z^2+9z+6)Z_0^2\left(\frac{1}{2}z\right)\right]}\\
={}&-\frac{2z+1}{4z-1}+\frac{2Z_0\left(\frac{1}{2}z\right)}{3\left[Z_1\left(\frac{1}{2}z\right)+Z_0\left(\frac{1}{2}z\right)\right]}\\
&+\frac{4z(2z+1)(2z+7)Z_1\left(\frac{1}{2}z\right)+4(4z^3+20z^2+21z+3)Z_0\left(\frac{1}{2}z\right)}{3(4z-1)\left[z(4z-1)Z_1^2\left(\frac{1}{2}z\right)+8z(z+1)Z_0\left(\frac{1}{2}z\right)Z_1\left(\frac{1}{2}z\right)+(4z^2+9z+6)Z_0^2\left(\frac{1}{2}z\right)\right]}, \tag*{(73)}
\end{aligned} v 3 ( z ) = = − [ Z 1 ( 2 1 z ) + Z 0 ( 2 1 z ) ] [ z ( 4 z − 1 ) Z 1 2 ( 2 1 z ) + 8 z ( z + 1 ) Z 0 ( 2 1 z ) Z 1 ( 2 1 z ) + ( 4 z 2 + 9 z + 6 ) Z 0 2 ( 2 1 z ) ] [ ( 2 z + 1 ) Z 1 ( 2 1 z ) + ( 2 z + 3 ) Z 0 ( 2 1 z ) ] [ z Z 1 2 ( 2 1 z ) − ( z + 2 ) Z 0 2 ( 2 1 z ) ] − 4 z − 1 2 z + 1 + 3 [ Z 1 ( 2 1 z ) + Z 0 ( 2 1 z ) ] 2 Z 0 ( 2 1 z ) + 3 ( 4 z − 1 ) [ z ( 4 z − 1 ) Z 1 2 ( 2 1 z ) + 8 z ( z + 1 ) Z 0 ( 2 1 z ) Z 1 ( 2 1 z ) + ( 4 z 2 + 9 z + 6 ) Z 0 2 ( 2 1 z ) ] 4 z ( 2 z + 1 ) ( 2 z + 7 ) Z 1 ( 2 1 z ) + 4 ( 4 z 3 + 20 z 2 + 21 z + 3 ) Z 0 ( 2 1 z ) , (73)
u 3 ( z ) = − ( z 2 − 12 ) ( z + 4 ) z 3 + 12 z 2 + 36 z + 48 = − 1 + 8 ( z 2 + 6 z + 12 ) z 3 + 12 z 2 + 36 z + 48 , (74) u_3(z)=-\frac{(z^2-12)(z+4)}{z^3+12z^2+36z+48}=-1+\frac{8(z^2+6z+12)}{z^3+12z^2+36z+48}, \tag*{(74)} u 3 ( z ) = − z 3 + 12 z 2 + 36 z + 48 ( z 2 − 12 ) ( z + 4 ) = − 1 + z 3 + 12 z 2 + 36 z + 48 8 ( z 2 + 6 z + 12 ) , (74)
respectively, which all satisfy P V P_{\mathrm V} P V (2) for the parameters ( α , β , γ ) = ( 9 8 , − 1 8 , − 2 ) (\alpha,\beta,\gamma)=\left(\frac{9}{8},-\frac{1}{8},-2\right) ( α , β , γ ) = ( 8 9 , − 8 1 , − 2 ) . Hence we have shown that for these parameters, P V P_{\mathrm V} P V (2) has two different Bessel function solutions and a rational solution, which is expressed in terms of generalised Umemura polynomials since u 3 ( z ) = w ^ 2 , 0 ( − 1 / 2 ) ( − z ) = u ^ 1 , 0 ( − 3 ) ( z ) u_3(z)=\widehat{w}_{2,0}^{(-1/2)}(-z)=\widehat{u}_{1,0}^{(-3)}(z) u 3 ( z ) = w 2 , 0 ( − 1/2 ) ( − z ) = u 1 , 0 ( − 3 ) ( z ) , recall (56) .
The solutions w 3 ( z ) w_3(z) w 3 ( z ) , v 3 ( z ) v_3(z) v 3 ( z ) and u 3 ( z ) u_3(z) u 3 ( z ) given by (4.7) have different expansions about both z = 0 z=0 z = 0 and z = ∞ z=\infty z = ∞ . Setting C 2 = 0 C_2=0 C 2 = 0 in w 3 ( z ) w_3(z) w 3 ( z ) and v 3 ( z ) v_3(z) v 3 ( z ) , since K ν ( z ) K_\nu(z) K ν ( z ) has a singularity at z = 0 z=0 z = 0 , then as z → 0 z\to0 z → 0
w 3 ( z ) = 1 − z 2 + z 2 24 + z 3 192 − z 5 9216 + O ( z 6 ) , (75) \begin{aligned}
w_3(z) &= 1-\frac{z}{2}+\frac{z^2}{24}+\frac{z^3}{192}-\frac{z^5}{9216}+\mathcal{O}(z^6), \tag*{(75)}
\end{aligned} w 3 ( z ) = 1 − 2 z + 24 z 2 + 192 z 3 − 9216 z 5 + O ( z 6 ) , (75)
v 3 ( z ) = 1 − z 2 + z 2 24 + 19 z 3 192 − 5 z 4 64 + 95 z 5 3072 + O ( z 6 ) , (76) \begin{aligned}
v_3(z) &= 1-\frac{z}{2}+\frac{z^2}{24}+\frac{19z^3}{192}-\frac{5z^4}{64}+\frac{95z^5}{3072}+\mathcal{O}(z^6), \tag*{(76)}
\end{aligned} v 3 ( z ) = 1 − 2 z + 24 z 2 + 192 19 z 3 − 64 5 z 4 + 3072 95 z 5 + O ( z 6 ) , (76)
u 3 ( z ) = 1 − z 2 + z 2 24 + 5 z 3 96 − 5 z 4 128 + 71 z 5 4608 + O ( z 6 ) , (77) \begin{aligned}
u_3(z) &= 1-\frac{z}{2}+\frac{z^2}{24}+\frac{5z^3}{96}-\frac{5z^4}{128}+\frac{71z^5}{4608}+\mathcal{O}(z^6), \tag*{(77)}
\end{aligned} u 3 ( z ) = 1 − 2 z + 24 z 2 + 96 5 z 3 − 128 5 z 4 + 4608 71 z 5 + O ( z 6 ) , (77)
and as z → ∞ z\to\infty z → ∞
w 3 ( z ) = − 1 2 z − 3 2 z 2 − 51 8 z 3 − 273 8 z 4 − 3483 16 z 5 + O ( z − 6 ) , (78) \begin{aligned}
w_3(z) &= -\frac{1}{2z}-\frac{3}{2z^2}-\frac{51}{8z^3}-\frac{273}{8z^4}-\frac{3483}{16z^5}+\mathcal{O}(z^{-6}), \tag*{(78)}
\end{aligned} w 3 ( z ) = − 2 z 1 − 2 z 2 3 − 8 z 3 51 − 8 z 4 273 − 16 z 5 3483 + O ( z − 6 ) , (78)
v 3 ( z ) = 1 2 z + 1 2 z 2 + 3 8 z 3 + 11 8 z 4 + 83 16 z 5 + O ( z − 6 ) , (79) \begin{aligned}
v_3(z) &= \frac{1}{2z}+\frac{1}{2z^2}+\frac{3}{8z^3}+\frac{11}{8z^4}+\frac{83}{16z^5}+\mathcal{O}(z^{-6}), \tag*{(79)}
\end{aligned} v 3 ( z ) = 2 z 1 + 2 z 2 1 + 8 z 3 3 + 8 z 4 11 + 16 z 5 83 + O ( z − 6 ) , (79)
u 3 ( z ) = − 1 + 8 z − 48 z 2 + 384 z 3 − 3264 z 4 + 27648 z 5 + O ( z − 6 ) . (80) \begin{aligned}
u_3(z) &= -1+\frac{8}{z}-\frac{48}{z^2}+\frac{384}{z^3}-\frac{3264}{z^4}+\frac{27648}{z^5}+\mathcal{O}(z^{-6}). \tag*{(80)}
\end{aligned} u 3 ( z ) = − 1 + z 8 − z 2 48 + z 3 384 − z 4 3264 + z 5 27648 + O ( z − 6 ) . (80)
We note that the first three terms in the series expansions about z = 0 z=0 z = 0 are the same for all three solutions w 3 ( z ) w_3(z) w 3 ( z ) , v 3 ( z ) v_3(z) v 3 ( z ) and u 3 ( z ) u_3(z) u 3 ( z ) whereas as z → ∞ z\to\infty z → ∞ , w 3 ( z ) → 0 w_3(z)\to0 w 3 ( z ) → 0 , v 3 ( z ) → 0 v_3(z)\to0 v 3 ( z ) → 0 and u 3 ( z ) → − 1 u_3(z)\to-1 u 3 ( z ) → − 1 . A plot of the solutions w 3 ( z ) w_3(z) w 3 ( z ) , v 3 ( z ) v_3(z) v 3 ( z ) and u 3 ( z ) u_3(z) u 3 ( z ) , with C 2 = 0 C_2=0 C 2 = 0 is given in Figure 4.1 . All three solutions have the same value at z = 0 z=0 z = 0 . The solution w 3 ( z ) w_3(z) w 3 ( z ) has one zero for z > 0 z>0 z > 0 and a minimum at z ≈ 5.2855 z\approx5.2855 z ≈ 5.2855 . The solutions v 3 ( z ) v_3(z) v 3 ( z ) and u 3 ( z ) u_3(z) u 3 ( z ) are both monotonically decreasing for z > 0 z>0 z > 0 .
Figure 4.1. A plot of the solutions w 3 ( z ) w_3(z) w 3 ( z ) (blue ), v 3 ( z ) v_3(z) v 3 ( z ) (red ) and u 3 ( z ) u_3(z) u 3 ( z ) (black ) given by (4.7), with C 2 = 0 C_2=0 C 2 = 0 (since K ν ( z ) K_\nu(z) K ν ( z ) has a singularity at z = 0 z=0 z = 0 ).
Remarks
The rational solution u 3 ( z ) u_3(z) u 3 ( z ) is not a special case of either w 3 ( z ) w_3(z) w 3 ( z ) or v 3 ( z ) v_3(z) v 3 ( z ) and analogously for the solutions u 1 ( z ) u_1(z) u 1 ( z ) and u 2 ( z ) u_2(z) u 2 ( z ) in Example 4.2. Usually when a Painlevé equation has a special function solution and a rational solution for the same values of the parameters, then the rational solution is a special case of the special function solution. For example, all the rational solutions of P V \mathrm{P}_{\mathrm{V}} P V (2) in terms of generalised Laguerre polynomials given in (3.3) are special cases of the Kummer function solutions of P V \mathrm{P}_{\mathrm{V}} P V (2) , see [33 ] .
The solution w 0 ( z ) w_0(z) w 0 ( z ) given by (65) can be mapped to a solution of P I I I \mathrm{P}_{\mathrm{III}} P III (4) as shown in Example 5.7.
Relationship with the third Painlevé equation
In this section we discuss the relationship between solutions of the fifth Painlevé equation (1) and solutions of the generic case of P I I I \mathrm{P}_{\mathrm{III}} P III (4) when C D ≠ 0 CD \ne0 C D = 0 .
It is well known that solutions of the deg-P V \mathrm{P}_{\mathrm{V}} P V (3) are related to solutions of P I I I \mathrm{P}_{\mathrm{III}} P III (4) with C D ≠ 0 CD \ne0 C D = 0 , a result originally due to Gromak [57 ] ; see also [62 §34] . This relationship is described in the following lemma.
Lemma 5.1. If q ( x ) q(x) q ( x ) is a solution of
d 2 q d x 2 = 1 q ( d q d x ) 2 − 1 x d q d x + ( a − b ) q 2 x + a + b + 1 x + q 3 4 − 1 4 q , (81) \frac{\mathrm{d}^{2}q}{\mathrm{d}x^{2}}=\frac{1}{q}\left(\frac{\mathrm{d}q}{\mathrm{d}x}\right)^{2}-\frac{1}{x}\frac{\mathrm{d}q}{\mathrm{d}x}+\frac{(a-b)q^{2}}{x}+\frac{a+b+1}{x}+\frac{q^{3}}{4}-\frac{1}{4q}, \tag*{(81)} d x 2 d 2 q = q 1 ( d x d q ) 2 − x 1 d x d q + x ( a − b ) q 2 + x a + b + 1 + 4 q 3 − 4 q 1 , (81) which is P I I I \mathrm{P}_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( a − b , a + b + 1 , 1 4 , − 1 4 ) (A,B,C,D)=\left(a-b,a+b+1,\frac{1}{4},-\frac{1}{4}\right) ( A , B , C , D ) = ( a − b , a + b + 1 , 4 1 , − 4 1 ) , then
u ( z ) = 1 + 2 x q 2 2 x d q d x − x q 2 + 2 ( 2 a + 2 b + 1 ) q − x , z = 1 4 x 2 , (82) u(z)=1+\frac{2xq^{2}}{2x\frac{\mathrm{d}q}{\mathrm{d}x}-xq^{2}+2(2a+2b+1)q-x},\qquad z=\frac{1}{4}x^{2}, \tag*{(82)} u ( z ) = 1 + 2 x d x d q − x q 2 + 2 ( 2 a + 2 b + 1 ) q − x 2 x q 2 , z = 4 1 x 2 , (82) is a solution of
d 2 u d z 2 = ( 1 2 u + 1 u − 1 ) ( d u d z ) 2 − 1 z d u d z + ( u − 1 ) 2 ( a 2 u 2 − b 2 ) 2 z 2 u + u 2 z , (83) \frac{\mathrm{d}^{2}u}{\mathrm{d}z^{2}}=\left(\frac{1}{2u}+\frac{1}{u-1}\right)\left(\frac{\mathrm{d}u}{\mathrm{d}z}\right)^{2}-\frac{1}{z}\frac{\mathrm{d}u}{\mathrm{d}z}+\frac{(u-1)^{2}(a^{2}u^{2}-b^{2})}{2z^{2}u}+\frac{u}{2z}, \tag*{(83)} d z 2 d 2 u = ( 2 u 1 + u − 1 1 ) ( d z d u ) 2 − z 1 d z d u + 2 z 2 u ( u − 1 ) 2 ( a 2 u 2 − b 2 ) + 2 z u , (83) which is deg-P V \mathrm{P}_{\mathrm{V}} P V (3) with α = 1 2 a 2 \alpha=\frac{1}{2}a^{2} α = 2 1 a 2 , β = − 1 2 b 2 \beta=-\frac{1}{2}b^{2} β = − 2 1 b 2 and γ = 1 2 \gamma=\frac{1}{2} γ = 2 1 , provided that
x d q d x − 1 2 x q 2 + ( 2 a + 2 b + 1 ) q − 1 2 x ≠ 0. x\frac{\mathrm{d}q}{\mathrm{d}x}-\frac{1}{2}xq^{2}+(2a+2b+1)q-\frac{1}{2}x\ne0. x d x d q − 2 1 x q 2 + ( 2 a + 2 b + 1 ) q − 2 1 x = 0. Conversely, if u ( z ) u(z) u ( z ) is a solution of (83) , then
q ( x ) = 1 z u { z d u d z + ( u − 1 ) ( a u + b ) } , x = 2 z , (84) q(x)=\frac{1}{\sqrt{z}u}\left\{z\frac{\mathrm{d}u}{\mathrm{d}z}+(u-1)(au+b)\right\},\qquad x=2\sqrt{z}, \tag*{(84)} q ( x ) = z u 1 { z d z d u + ( u − 1 ) ( a u + b ) } , x = 2 z , (84) is a solution of (81) .
Proof. See, for example, [32 Lemma 2.1] , [45 Theorem 4.2] , [62 Theorem 34.2] .
Remark 5.2. We remark that (82) and (84) can be written as the system
x d q d x = z ( u + 1 ) q 2 u − 1 − ( 2 a + 2 b + 1 ) q + z , (85) x\frac{\mathrm{d}q}{\mathrm{d}x}=\frac{\sqrt{z}(u+1)q^{2}}{u-1}-(2a+2b+1)q+\sqrt{z}, \tag*{(85)} x d x d q = u − 1 z ( u + 1 ) q 2 − ( 2 a + 2 b + 1 ) q + z , (85)
z d u d z = − a u 2 + ( 1 2 x q + a − b ) u + b . (86) z\frac{\mathrm{d}u}{\mathrm{d}z}=-au^{2}+\left(\frac{1}{2}xq+a-b\right)u+b. \tag*{(86)} z d z d u = − a u 2 + ( 2 1 x q + a − b ) u + b . (86)
Eliminating u u u gives (81) , whilst eliminating q q q gives (83) .
Lemma 5.3. There is a relationship between solutions of P V \mathrm{P}_{\mathrm{V}} P V (2) with α = − β = λ \alpha=-\beta=\lambda α = − β = λ and γ = 0 \gamma=0 γ = 0 , and solutions of deg-P V \mathrm{P}_{\mathrm{V}} P V (3) with α = 0 \alpha=0 α = 0 , β = − 4 λ \beta=-4\lambda β = − 4 λ and γ = 1 2 \gamma=\frac{1}{2} γ = 2 1 .
Proof. Consider P V \mathrm{P}_{\mathrm{V}} P V (2) with α = − β = λ \alpha=-\beta=\lambda α = − β = λ and γ = 0 \gamma=0 γ = 0 , i.e.
d 2 w d z 2 = ( 1 2 w + 1 w − 1 ) ( d w d z ) 2 − 1 z d w d z + λ ( w − 1 ) 3 ( w + 1 ) z 2 w − w ( w + 1 ) 2 ( w − 1 ) . (87) \frac{\mathrm{d}^{2}w}{\mathrm{d}z^{2}}=\left(\frac{1}{2w}+\frac{1}{w-1}\right)\left(\frac{\mathrm{d}w}{\mathrm{d}z}\right)^{2}-\frac{1}{z}\frac{\mathrm{d}w}{\mathrm{d}z}+\frac{\lambda(w-1)^{3}(w+1)}{z^{2}w}-\frac{w(w+1)}{2(w-1)}. \tag*{(87)} d z 2 d 2 w = ( 2 w 1 + w − 1 1 ) ( d z d w ) 2 − z 1 d z d w + z 2 w λ ( w − 1 ) 3 ( w + 1 ) − 2 ( w − 1 ) w ( w + 1 ) . (87) Making the transformation
u ( x ) = 4 w ( z ) [ 1 + w ( z ) ] 2 , x = 1 4 z 2 , (88) u(x)=\frac{4w(z)}{[1+w(z)]^{2}},\qquad x=\frac{1}{4}z^{2}, \tag*{(88)} u ( x ) = [ 1 + w ( z ) ] 2 4 w ( z ) , x = 4 1 z 2 , (88) in deg-P V \mathrm{P}_{\mathrm{V}} P V (3) with α = 0 \alpha=0 α = 0 , β = − 4 λ \beta=-4\lambda β = − 4 λ and γ = 1 2 \gamma=\frac{1}{2} γ = 2 1 gives (87) . Conversely, making the transformation
w ( z ) = 1 − 1 − u ( x ) 1 + 1 − u ( x ) , z = 2 x , w(z)=\frac{1-\sqrt{1-u(x)}}{1+\sqrt{1-u(x)}},\qquad z=2\sqrt{x}, w ( z ) = 1 + 1 − u ( x ) 1 − 1 − u ( x ) , z = 2 x , in (87) gives
d 2 u d x 2 = ( 1 2 u + 1 u − 1 ) ( d u d x ) 2 − 1 x d u d x − 4 λ ( u − 1 ) 2 x 2 u + u 2 x , \frac{\mathrm{d}^{2}u}{\mathrm{d}x^{2}}=\left(\frac{1}{2u}+\frac{1}{u-1}\right)\left(\frac{\mathrm{d}u}{\mathrm{d}x}\right)^{2}-\frac{1}{x}\frac{\mathrm{d}u}{\mathrm{d}x}-\frac{4\lambda(u-1)^{2}}{x^{2}u}+\frac{u}{2x}, d x 2 d 2 u = ( 2 u 1 + u − 1 1 ) ( d x d u ) 2 − x 1 d x d u − x 2 u 4 λ ( u − 1 ) 2 + 2 x u , which is deg-P V \mathrm{P}_{\mathrm{V}} P V (3) with α = 0 \alpha=0 α = 0 , β = − 4 λ \beta=-4\lambda β = − 4 λ and γ = 1 2 \gamma=\frac{1}{2} γ = 2 1 .
From Lemmas 5.1 and 5.3 we have the following corollary.
Corollary 5.4. There is a relationship between solutions of P V P_{\mathrm{V}} P V (2) with α = − β \alpha=-\beta α = − β and γ = 0 \gamma=0 γ = 0 and solutions of P I I I P_{\mathrm{III}} P III (4) with A + B = 1 A+B=1 A + B = 1 and C = − D = 1 4 C=-D=\frac{1}{4} C = − D = 4 1 .
Proof. If w ( z ) w(z) w ( z ) is a solution of (87) with λ = 1 8 k 2 \lambda=\frac{1}{8}k^2 λ = 8 1 k 2 , then substituting (88) into (84) with a = 0 a=0 a = 0 and b = k b=k b = k gives
q ( x ) = − w ( z ) − 1 w ( z ) [ w ( z ) + 1 ] d w d z − k [ w ( z ) − 1 ] 2 2 z w ( z ) , x = z , (89) q(x)=-\frac{w(z)-1}{w(z)[w(z)+1]}\frac{\mathrm{d}w}{\mathrm{d}z}-\frac{k[w(z)-1]^2}{2zw(z)},\qquad x=z, \tag*{(89)} q ( x ) = − w ( z ) [ w ( z ) + 1 ] w ( z ) − 1 d z d w − 2 z w ( z ) k [ w ( z ) − 1 ] 2 , x = z , (89)
which satisfies P I I I P_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( − k , k + 1 , 1 4 , − 1 4 ) (A,B,C,D)=(-k,k+1,\frac{1}{4},-\frac{1}{4}) ( A , B , C , D ) = ( − k , k + 1 , 4 1 , − 4 1 ) .
To derive the inverse transformation, from (82) and (88)
4 w ( z ) [ 1 + w ( z ) ] 2 = 1 + 2 x q 2 ( x ) 2 x d q d x − x q 2 ( x ) + 2 ( 2 k + 1 ) q ( x ) − x , \frac{4w(z)}{[1+w(z)]^2}=1+\frac{2xq^2(x)}{2x\frac{\mathrm{d}q}{\mathrm{d}x}-xq^2(x)+2(2k+1)q(x)-x}, [ 1 + w ( z ) ] 2 4 w ( z ) = 1 + 2 x d x d q − x q 2 ( x ) + 2 ( 2 k + 1 ) q ( x ) − x 2 x q 2 ( x ) ,
we obtain
w ( z ) = Q ( x ) + 1 Q ( x ) − 1 , Q ( x ) = − 1 q 2 ( x ) d q d x + q 2 ( x ) + 1 2 q 2 ( x ) − 2 k + 1 x q ( x ) , z = x . (90) w(z)=\frac{\sqrt{Q(x)}+1}{\sqrt{Q(x)}-1},\qquad Q(x)=-\frac{1}{q^2(x)}\frac{\mathrm{d}q}{\mathrm{d}x}+\frac{q^2(x)+1}{2q^2(x)}-\frac{2k+1}{xq(x)},\qquad z=x. \tag*{(90)} w ( z ) = Q ( x ) − 1 Q ( x ) + 1 , Q ( x ) = − q 2 ( x ) 1 d x d q + 2 q 2 ( x ) q 2 ( x ) + 1 − x q ( x ) 2 k + 1 , z = x . (90)
Therefore if q ( x ) q(x) q ( x ) satisfies P I I I P_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( − k , k + 1 , 1 4 , − 1 4 ) (A,B,C,D)=(-k,k+1,\frac{1}{4},-\frac{1}{4}) ( A , B , C , D ) = ( − k , k + 1 , 4 1 , − 4 1 ) , then w ( z ) w(z) w ( z ) given by (90) satisfies (87) with λ = 1 8 k 2 \lambda=\frac{1}{8}k^2 λ = 8 1 k 2 , which is P V P_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 8 k 2 , − 1 8 k 2 , 0 ) (\alpha,\beta,\gamma)=(\frac{1}{8}k^2,-\frac{1}{8}k^2,0) ( α , β , γ ) = ( 8 1 k 2 , − 8 1 k 2 , 0 ) .
Using ideas due to Tsegel’nik [107 ] , there is another case when solutions of P V P_{\mathrm{V}} P V (2) with γ = 0 \gamma=0 γ = 0 are related to solutions of P I I I P_{\mathrm{III}} P III (4) .
Lemma 5.5. Suppose that q ( x ) q(x) q ( x ) is a solution of P I I I P_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( 1 2 λ , − 1 2 λ , 1 4 , − 1 4 ) (A,B,C,D)=(\frac{1}{2}\lambda,-\frac{1}{2}\lambda,\frac{1}{4},-\frac{1}{4}) ( A , B , C , D ) = ( 2 1 λ , − 2 1 λ , 4 1 , − 4 1 ) , where λ \lambda λ is an arbitrary constant, then
w ( z ) = 1 + 2 [ q 2 ( x ) − 1 ] 2 d q d x − q 2 ( x ) + 1 , z = 2 x , (91) w(z)=1+\frac{2[q^2(x)-1]}{2\frac{\mathrm{d}q}{\mathrm{d}x}-q^2(x)+1},\qquad z=2x, \tag*{(91)} w ( z ) = 1 + 2 d x d q − q 2 ( x ) + 1 2 [ q 2 ( x ) − 1 ] , z = 2 x , (91) satisfies P V P_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 8 ( λ − 1 ) 2 , − 1 8 ( λ + 1 ) 2 , 0 ) (\alpha,\beta,\gamma)=(\frac{1}{8}(\lambda-1)^2,-\frac{1}{8}(\lambda+1)^2,0) ( α , β , γ ) = ( 8 1 ( λ − 1 ) 2 , − 8 1 ( λ + 1 ) 2 , 0 ) . The inverse transformation is given by
q 2 ( x ) + 1 4 q ( x ) = 1 2 w ( z ) d w d z + ( λ − 1 ) [ w ( z ) − 1 ] 4 z − ( λ + 1 ) [ w ( z ) − 1 ] 4 z w ( z ) , x = 1 2 z . (92) \frac{q^2(x)+1}{4q(x)}=\frac{1}{2w(z)}\frac{\mathrm{d}w}{\mathrm{d}z}+\frac{(\lambda-1)[w(z)-1]}{4z}-\frac{(\lambda+1)[w(z)-1]}{4zw(z)},\qquad x=\frac{1}{2}z. \tag*{(92)} 4 q ( x ) q 2 ( x ) + 1 = 2 w ( z ) 1 d z d w + 4 z ( λ − 1 ) [ w ( z ) − 1 ] − 4 z w ( z ) ( λ + 1 ) [ w ( z ) − 1 ] , x = 2 1 z . (92)
Proof. It is straightforward to show that if w ( z ) w(z) w ( z ) is given by (91) , with q ( x ) q(x) q ( x ) a solution of P I I I P_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( 1 2 λ , − 1 2 λ , 1 4 , − 1 4 ) (A,B,C,D)=(\frac{1}{2}\lambda,-\frac{1}{2}\lambda,\frac{1}{4},-\frac{1}{4}) ( A , B , C , D ) = ( 2 1 λ , − 2 1 λ , 4 1 , − 4 1 ) , then it satisfies
d 2 w d z 2 = ( 1 2 w + 1 w − 1 ) ( d w d z ) 2 − 1 z d w d z + ( w − 1 ) 2 { ( λ − 1 ) 2 w 2 − ( λ + 1 ) 2 } 8 z 2 w − w ( w + 1 ) 2 ( w − 1 ) , (93) \frac{\mathrm{d}^2w}{\mathrm{d}z^2}=\left(\frac{1}{2w}+\frac{1}{w-1}\right)\left(\frac{\mathrm{d}w}{\mathrm{d}z}\right)^2-\frac{1}{z}\frac{\mathrm{d}w}{\mathrm{d}z}+\frac{(w-1)^2\{(\lambda-1)^2w^2-(\lambda+1)^2\}}{8z^2w}-\frac{w(w+1)}{2(w-1)}, \tag*{(93)} d z 2 d 2 w = ( 2 w 1 + w − 1 1 ) ( d z d w ) 2 − z 1 d z d w + 8 z 2 w ( w − 1 ) 2 {( λ − 1 ) 2 w 2 − ( λ + 1 ) 2 } − 2 ( w − 1 ) w ( w + 1 ) , (93)
which is P V P_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 8 ( λ − 1 ) 2 , − 1 8 ( λ + 1 ) 2 , 0 ) (\alpha,\beta,\gamma)=(\frac{1}{8}(\lambda-1)^2,-\frac{1}{8}(\lambda+1)^2,0) ( α , β , γ ) = ( 8 1 ( λ − 1 ) 2 , − 8 1 ( λ + 1 ) 2 , 0 ) .
To prove the inverse transformation (92) , consider
v ( z ) = 1 2 w ( z ) d w d z + ( λ − 1 ) [ w ( z ) − 1 ] 4 z − ( λ + 1 ) [ w ( z ) − 1 ] 4 z w ( z ) , v(z)=\frac{1}{2w(z)}\frac{\mathrm{d}w}{\mathrm{d}z}+\frac{(\lambda-1)[w(z)-1]}{4z}-\frac{(\lambda+1)[w(z)-1]}{4zw(z)}, v ( z ) = 2 w ( z ) 1 d z d w + 4 z ( λ − 1 ) [ w ( z ) − 1 ] − 4 z w ( z ) ( λ + 1 ) [ w ( z ) − 1 ] ,
where w ( z ) w(z) w ( z ) satisfies (93) , then it is straightforward to show that v ( z ) v(z) v ( z ) satisfies
d 2 v d z 2 = 4 v 4 v 2 − 1 ( d v d z ) 2 − 1 z d v d z + ( z v + λ ) ( 4 v 2 − 1 ) 4 z . (94) \frac{\mathrm{d}^2v}{\mathrm{d}z^2}=\frac{4v}{4v^2-1}\left(\frac{\mathrm{d}v}{\mathrm{d}z}\right)^2-\frac{1}{z}\frac{\mathrm{d}v}{\mathrm{d}z}+\frac{(zv+\lambda)(4v^2-1)}{4z}. \tag*{(94)} d z 2 d 2 v = 4 v 2 − 1 4 v ( d z d v ) 2 − z 1 d z d v + 4 z ( z v + λ ) ( 4 v 2 − 1 ) . (94)
Consequently making the transformation
v ( z ) = q 2 ( x ) + 1 4 q ( x ) , z = 2 x , v(z)=\frac{q^2(x)+1}{4q(x)},\qquad z=2x, v ( z ) = 4 q ( x ) q 2 ( x ) + 1 , z = 2 x ,
in equation (94) gives
d 2 q d x 2 = 1 q ( d q d x ) 2 − 1 x d q d x + λ ( q 2 − 1 ) 2 x + q 3 4 − 1 4 q , (95) \frac{\mathrm{d}^2q}{\mathrm{d}x^2}=\frac{1}{q}\left(\frac{\mathrm{d}q}{\mathrm{d}x}\right)^2-\frac{1}{x}\frac{\mathrm{d}q}{\mathrm{d}x}+\frac{\lambda(q^2-1)}{2x}+\frac{q^3}{4}-\frac{1}{4q}, \tag*{(95)} d x 2 d 2 q = q 1 ( d x d q ) 2 − x 1 d x d q + 2 x λ ( q 2 − 1 ) + 4 q 3 − 4 q 1 , (95)
which is P I I I P_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( 1 2 λ , − 1 2 λ , 1 4 , − 1 4 ) (A,B,C,D)=(\frac{1}{2}\lambda,-\frac{1}{2}\lambda,\frac{1}{4},-\frac{1}{4}) ( A , B , C , D ) = ( 2 1 λ , − 2 1 λ , 4 1 , − 4 1 ) .
Remarks 5.6.
We remark that making the transformation
v ( z ) = w ( z ) + 1 2 [ w ( z ) − 1 ] , v(z)=\frac{w(z)+1}{2[w(z)-1]}, v ( z ) = 2 [ w ( z ) − 1 ] w ( z ) + 1 ,
in equation (94) gives
d 2 w d z 2 = ( 1 2 w + 1 w − 1 ) ( d w d z ) 2 − 1 z d w d z − λ w z − w ( w + 1 ) 2 ( w − 1 ) , \frac{\mathrm{d}^{2}w}{\mathrm{d}z^{2}}=\left(\frac{1}{2w}+\frac{1}{w-1}\right)\left(\frac{\mathrm{d}w}{\mathrm{d}z}\right)^{2}-\frac{1}{z}\frac{\mathrm{d}w}{\mathrm{d}z}-\frac{\lambda w}{z}-\frac{w(w+1)}{2(w-1)}, d z 2 d 2 w = ( 2 w 1 + w − 1 1 ) ( d z d w ) 2 − z 1 d z d w − z λ w − 2 ( w − 1 ) w ( w + 1 ) ,
which is P V \mathrm{P}_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 0 , 0 , − λ ) (\alpha,\beta,\gamma)=(0,0,-\lambda) ( α , β , γ ) = ( 0 , 0 , − λ ) .
Painlevé [101, pp. 39–40] showed that P V \mathrm{P}_{\mathrm{V}} P V (2) with α = β = 0 \alpha=\beta=0 α = β = 0 and γ = c \gamma=c γ = c can be transformed into P I I I \mathrm{P}_{\mathrm{III}} P III (4) through a quadratic transformation. Substituting
w ( z ) = { q ( x ) + 1 q ( x ) − 1 } 2 , z = 2 x , w(z)=\left\{\frac{q(x)+1}{q(x)-1}\right\}^{2},\qquad z=2x, w ( z ) = { q ( x ) − 1 q ( x ) + 1 } 2 , z = 2 x ,
into P V \mathrm{P}_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 0 , 0 , c ) (\alpha,\beta,\gamma)=(0,0,c) ( α , β , γ ) = ( 0 , 0 , c ) gives equation (95) with λ = − c \lambda=-c λ = − c .
Example 5.7. In Example 4.2 we saw that the solution
w 0 ( z ) = Z 1 ( 1 2 z ) − Z 0 ( 1 2 z ) Z 1 ( 1 2 z ) + Z 0 ( 1 2 z ) , w_{0}(z)=\frac{\mathcal{Z}_{1}\left(\frac{1}{2}z\right)-\mathcal{Z}_{0}\left(\frac{1}{2}z\right)}{\mathcal{Z}_{1}\left(\frac{1}{2}z\right)+\mathcal{Z}_{0}\left(\frac{1}{2}z\right)}, w 0 ( z ) = Z 1 ( 2 1 z ) + Z 0 ( 2 1 z ) Z 1 ( 2 1 z ) − Z 0 ( 2 1 z ) ,
with Z ν ( z ) \mathcal{Z}_{\nu}(z) Z ν ( z ) given by (64) , satisfies P V \mathrm{P}_{\mathrm{V}} P V (2) for the parameters ( α , β , γ ) = ( 1 8 , − 1 8 , 0 ) (\alpha,\beta,\gamma)=\left(\frac{1}{8},-\frac{1}{8},0\right) ( α , β , γ ) = ( 8 1 , − 8 1 , 0 ) . Hence we can use (89) , with κ = ± 1 \kappa=\pm1 κ = ± 1 , to derive a solution of P I I I \mathrm{P}_{\mathrm{III}} P III (4) . Substituting w 0 ( z ) w_{0}(z) w 0 ( z ) into (89) with κ = − 1 \kappa=-1 κ = − 1 gives
q ( x ) = − Z 0 ( 1 2 x ) Z 1 ( 1 2 x ) , q(x)=-\frac{\mathcal{Z}_{0}\left(\frac{1}{2}x\right)}{\mathcal{Z}_{1}\left(\frac{1}{2}x\right)}, q ( x ) = − Z 1 ( 2 1 x ) Z 0 ( 2 1 x ) ,
which satisfies P I I I \mathrm{P}_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( 1 , 0 , 1 4 , − 1 4 ) (A,B,C,D)=\left(1,0,\frac{1}{4},-\frac{1}{4}\right) ( A , B , C , D ) = ( 1 , 0 , 4 1 , − 4 1 ) , whilst substituting w 0 ( z ) w_{0}(z) w 0 ( z ) into (89) with κ = 1 \kappa=1 κ = 1 gives
q ( x ) = − Z 0 ( 1 2 x ) Z 1 ( 1 2 x ) − 4 Z 0 2 ( 1 2 x ) x { Z 1 2 ( 1 2 x ) − Z 0 2 ( 1 2 x ) } , q(x)=-\frac{\mathcal{Z}_{0}\left(\frac{1}{2}x\right)}{\mathcal{Z}_{1}\left(\frac{1}{2}x\right)}-\frac{4\mathcal{Z}_{0}^{2}\left(\frac{1}{2}x\right)}{x\left\{\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)-\mathcal{Z}_{0}^{2}\left(\frac{1}{2}x\right)\right\}}, q ( x ) = − Z 1 ( 2 1 x ) Z 0 ( 2 1 x ) − x { Z 1 2 ( 2 1 x ) − Z 0 2 ( 2 1 x ) } 4 Z 0 2 ( 2 1 x ) ,
which satisfies P I I I \mathrm{P}_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( − 1 , 2 , 1 4 , − 1 4 ) (A,B,C,D)=\left(-1,2,\frac{1}{4},-\frac{1}{4}\right) ( A , B , C , D ) = ( − 1 , 2 , 4 1 , − 4 1 ) .
Example 5.8. If w 0 w_{0} w 0 is given by (65) , then consider the function
w ( z ) = T 1 , − 1 , 1 2 ( w 0 ; 1 2 , − 1 2 , 0 ) = − { Z 1 ( 1 2 z ) + Z 0 ( 1 2 z ) } { ( z − 2 ) Z 1 2 ( 1 2 z ) + 4 Z 1 ( 1 2 z ) Z 0 ( 1 2 z ) − z Z 0 2 ( 1 2 z ) } { Z 1 ( 1 2 z ) − Z 0 ( 1 2 z ) } { ( z + 2 ) Z 1 2 ( 1 2 z ) + 4 Z 1 ( 1 2 z ) Z 0 ( 1 2 z ) − z Z 0 2 ( 1 2 z ) } , w(z)=\mathcal{T}_{1,-1,1}^{2}\left(w_{0};\frac{1}{2},-\frac{1}{2},0\right)=-\frac{\left\{\mathcal{Z}_{1}\left(\frac{1}{2}z\right)+\mathcal{Z}_{0}\left(\frac{1}{2}z\right)\right\}\left\{(z-2)\mathcal{Z}_{1}^{2}\left(\frac{1}{2}z\right)+4\mathcal{Z}_{1}\left(\frac{1}{2}z\right)\mathcal{Z}_{0}\left(\frac{1}{2}z\right)-z\mathcal{Z}_{0}^{2}\left(\frac{1}{2}z\right)\right\}}{\left\{\mathcal{Z}_{1}\left(\frac{1}{2}z\right)-\mathcal{Z}_{0}\left(\frac{1}{2}z\right)\right\}\left\{(z+2)\mathcal{Z}_{1}^{2}\left(\frac{1}{2}z\right)+4\mathcal{Z}_{1}\left(\frac{1}{2}z\right)\mathcal{Z}_{0}\left(\frac{1}{2}z\right)-z\mathcal{Z}_{0}^{2}\left(\frac{1}{2}z\right)\right\}}, w ( z ) = T 1 , − 1 , 1 2 ( w 0 ; 2 1 , − 2 1 , 0 ) = − { Z 1 ( 2 1 z ) − Z 0 ( 2 1 z ) } { ( z + 2 ) Z 1 2 ( 2 1 z ) + 4 Z 1 ( 2 1 z ) Z 0 ( 2 1 z ) − z Z 0 2 ( 2 1 z ) } { Z 1 ( 2 1 z ) + Z 0 ( 2 1 z ) } { ( z − 2 ) Z 1 2 ( 2 1 z ) + 4 Z 1 ( 2 1 z ) Z 0 ( 2 1 z ) − z Z 0 2 ( 2 1 z ) } ,
with Z ν ( z ) \mathcal{Z}_{\nu}(z) Z ν ( z ) given by (64) , satisfies P V \mathrm{P}_{\mathrm{V}} P V (2) for the parameters ( α , β , γ ) = ( 9 8 , − 9 8 , 0 ) (\alpha,\beta,\gamma)=\left(\frac{9}{8},-\frac{9}{8},0\right) ( α , β , γ ) = ( 8 9 , − 8 9 , 0 ) . Substituting this into (89) with κ = 3 \kappa=3 κ = 3 gives
q ( x ) = Z 1 ( 1 2 x ) { ( x 2 + 8 ) Z 1 2 ( 1 2 x ) + 2 x Z 1 ( 1 2 x ) Z 0 ( 1 2 x ) − x 2 Z 0 2 ( 1 2 x ) } x { 2 Z 1 2 ( 1 2 x ) − x Z 1 2 ( 1 2 x ) Z 0 ( 1 2 x ) − 4 Z 1 ( 1 2 x ) Z 0 2 ( 1 2 x ) + x Z 0 3 ( 1 2 x ) } − 8 Z 1 2 ( 1 2 x ) { ( x 2 + 2 ) Z 1 2 ( 1 2 x ) + 2 x Z 1 ( 1 2 x ) Z 0 ( 1 2 x ) − x 2 Z 0 2 ( 1 2 x ) } x { 4 Z 1 2 ( 1 2 x ) − [ x Z 1 2 ( 1 2 x ) + 4 Z 1 ( 1 2 x ) Z 0 ( 1 2 x ) − x Z 0 2 ( 1 2 x ) ] } , q(x)=\frac{\mathcal{Z}_{1}\left(\frac{1}{2}x\right)\left\{(x^{2}+8)\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)+2x\mathcal{Z}_{1}\left(\frac{1}{2}x\right)\mathcal{Z}_{0}\left(\frac{1}{2}x\right)-x^{2}\mathcal{Z}_{0}^{2}\left(\frac{1}{2}x\right)\right\}}{x\left\{2\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)-x\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)\mathcal{Z}_{0}\left(\frac{1}{2}x\right)-4\mathcal{Z}_{1}\left(\frac{1}{2}x\right)\mathcal{Z}_{0}^{2}\left(\frac{1}{2}x\right)+x\mathcal{Z}_{0}^{3}\left(\frac{1}{2}x\right)\right\}}-\frac{8\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)\left\{(x^{2}+2)\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)+2x\mathcal{Z}_{1}\left(\frac{1}{2}x\right)\mathcal{Z}_{0}\left(\frac{1}{2}x\right)-x^{2}\mathcal{Z}_{0}^{2}\left(\frac{1}{2}x\right)\right\}}{x\left\{4\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)-\left[x\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)+4\mathcal{Z}_{1}\left(\frac{1}{2}x\right)\mathcal{Z}_{0}\left(\frac{1}{2}x\right)-x\mathcal{Z}_{0}^{2}\left(\frac{1}{2}x\right)\right]\right\}}, q ( x ) = x { 2 Z 1 2 ( 2 1 x ) − x Z 1 2 ( 2 1 x ) Z 0 ( 2 1 x ) − 4 Z 1 ( 2 1 x ) Z 0 2 ( 2 1 x ) + x Z 0 3 ( 2 1 x ) } Z 1 ( 2 1 x ) { ( x 2 + 8 ) Z 1 2 ( 2 1 x ) + 2 x Z 1 ( 2 1 x ) Z 0 ( 2 1 x ) − x 2 Z 0 2 ( 2 1 x ) } − x { 4 Z 1 2 ( 2 1 x ) − [ x Z 1 2 ( 2 1 x ) + 4 Z 1 ( 2 1 x ) Z 0 ( 2 1 x ) − x Z 0 2 ( 2 1 x ) ] } 8 Z 1 2 ( 2 1 x ) { ( x 2 + 2 ) Z 1 2 ( 2 1 x ) + 2 x Z 1 ( 2 1 x ) Z 0 ( 2 1 x ) − x 2 Z 0 2 ( 2 1 x ) } ,
which satisfies P I I I \mathrm{P}_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( − 3 , 4 , 1 4 , − 1 4 ) (A,B,C,D)=\left(-3,4,\frac{1}{4},-\frac{1}{4}\right) ( A , B , C , D ) = ( − 3 , 4 , 4 1 , − 4 1 ) , and substituting it into (89) with κ = − 3 \kappa=-3 κ = − 3 gives
q ( x ) = 4 Z 1 2 ( 1 2 x ) x { Z 0 2 ( 1 2 x ) − Z 1 2 ( 1 2 x ) } + Z 1 ( 1 2 x ) { ( x 2 + 8 ) Z 1 2 ( 1 2 x ) + 2 x Z 1 ( 1 2 x ) Z 0 ( 1 2 x ) − x 2 Z 0 2 ( 1 2 x ) } x { 2 Z 1 2 ( 1 2 x ) − x Z 1 2 ( 1 2 x ) Z 0 ( 1 2 x ) − 4 Z 1 ( 1 2 x ) Z 0 2 ( 1 2 x ) + x Z 0 3 ( 1 2 x ) } , q(x)=\frac{4\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)}{x\left\{\mathcal{Z}_{0}^{2}\left(\frac{1}{2}x\right)-\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)\right\}}+\frac{\mathcal{Z}_{1}\left(\frac{1}{2}x\right)\left\{(x^{2}+8)\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)+2x\mathcal{Z}_{1}\left(\frac{1}{2}x\right)\mathcal{Z}_{0}\left(\frac{1}{2}x\right)-x^{2}\mathcal{Z}_{0}^{2}\left(\frac{1}{2}x\right)\right\}}{x\left\{2\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)-x\mathcal{Z}_{1}^{2}\left(\frac{1}{2}x\right)\mathcal{Z}_{0}\left(\frac{1}{2}x\right)-4\mathcal{Z}_{1}\left(\frac{1}{2}x\right)\mathcal{Z}_{0}^{2}\left(\frac{1}{2}x\right)+x\mathcal{Z}_{0}^{3}\left(\frac{1}{2}x\right)\right\}}, q ( x ) = x { Z 0 2 ( 2 1 x ) − Z 1 2 ( 2 1 x ) } 4 Z 1 2 ( 2 1 x ) + x { 2 Z 1 2 ( 2 1 x ) − x Z 1 2 ( 2 1 x ) Z 0 ( 2 1 x ) − 4 Z 1 ( 2 1 x ) Z 0 2 ( 2 1 x ) + x Z 0 3 ( 2 1 x ) } Z 1 ( 2 1 x ) { ( x 2 + 8 ) Z 1 2 ( 2 1 x ) + 2 x Z 1 ( 2 1 x ) Z 0 ( 2 1 x ) − x 2 Z 0 2 ( 2 1 x ) } ,
which satisfies P I I I \mathrm{P}_{\mathrm{III}} P III (4) with ( A , B , C , D ) = ( 3 , − 2 , 1 4 , − 1 4 ) (A,B,C,D)=\left(3,-2,\frac{1}{4},-\frac{1}{4}\right) ( A , B , C , D ) = ( 3 , − 2 , 4 1 , − 4 1 ) .
Theorem 5.9. The equation P V P_{\mathrm{V}} P V (2) reduces to a special case of P I I I P_{\mathrm{III}} P III (4) when the parameters α \alpha α , β \beta β , γ \gamma γ and δ \delta δ of P V P_{\mathrm{V}} P V take one of the following forms:
(i) if − 2 β = 2 α + m \sqrt{-2\beta}=\sqrt{2\alpha}+m − 2 β = 2 α + m , γ = n \gamma=n γ = n and δ = − 1 2 \delta=-\frac{1}{2} δ = − 2 1 , where m , n ∈ Z m,n\in\mathbb{Z} m , n ∈ Z and m + n m+n m + n is even;
(ii) if α = 1 8 ( 2 m − 1 ) 2 \alpha=\frac{1}{8}(2m-1)^2 α = 8 1 ( 2 m − 1 ) 2 , β = − 1 8 ( 2 n − 1 ) 2 \beta=-\frac{1}{8}(2n-1)^2 β = − 8 1 ( 2 n − 1 ) 2 , γ ∉ Z \gamma\notin\mathbb{Z} γ ∈ / Z and δ = − 1 2 \delta=-\frac{1}{2} δ = − 2 1 , where m , n ∈ Z m,n\in\mathbb{Z} m , n ∈ Z .
Proof. See [62 Theorem 41.5] .
Discussion
In this paper we studied the Bäcklund transformations of the fifth Painlevé equation P V P_{\mathrm{V}} P V (2) defining their action on the parameters a = 2 α a=\sqrt{2\alpha} a = 2 α , b = − 2 β b=\sqrt{-2\beta} b = − 2 β and c = γ c=\gamma c = γ , rather than on α \alpha α , β \beta β and γ \gamma γ . This removes the ambiguity in the choice of square roots and makes it straightforward to compose transformations. Using compositions of the eight elementary transformations T ε 1 , ε 2 , ε 3 \mathcal{T}_{\varepsilon_1,\varepsilon_2,\varepsilon_3} T ε 1 , ε 2 , ε 3 we gave explicitly the combinations which shift two of the parameters a a a , b b b and c c c by ± 1 \pm1 ± 1 . We also identified the Schlesinger transformations R 1 \mathcal{R}_1 R 1 , R 2 , … , R 14 \mathcal{R}_2,\ldots,\mathcal{R}_{14} R 2 , … , R 14 obtained from the isomonodromy problem, and the translation operators of the symmetric form s P V s_{\mathrm{PV}} s PV (2.11) associated with the extended affine Weyl group W ~ ( A 3 ( 1 ) ) \widetilde{W}(A_3^{(1)}) W ( A 3 ( 1 ) ) , with specific compositions of these Bäcklund transformations. This gives a unified description of the Bäcklund transformations of P V P_{\mathrm{V}} P V (2) that appear in the literature in several different notations. As an application, we applied these Bäcklund and Schlesinger transformations to the rational solutions of P V P_{\mathrm{V}} P V (2) , which are expressed in terms of generalised Laguerre polynomials and generalised Umemura polynomials.
An interesting observation in the paper concerns the case when α = − β = 1 8 \alpha=-\beta=\frac{1}{8} α = − β = 8 1 , so that a = ± 1 2 a=\pm\frac{1}{2} a = ± 2 1 and b = ± 1 2 b=\pm\frac{1}{2} b = ± 2 1 independently. In this case a Bäcklund transformation can change the sign of a a a and/or b b b without changing α \alpha α and β \beta β . Consequently two different transformations applied to the same seed solution can produce distinct solutions for the same values of the parameters. In this way we obtained for the same parameters, two distinct solutions expressed in terms of the modified Bessel functions I 0 ( z ) I_0(z) I 0 ( z ) , I 1 ( z ) I_1(z) I 1 ( z ) , K 0 ( z ) K_0(z) K 0 ( z ) and K 1 ( z ) K_1(z) K 1 ( z ) . Further, for the same parameters, we showed that there is also a rational solution, expressible in terms of generalised Umemura polynomials, which is not a special case of either Bessel function solution. This contrasts with the usual situation for Painlevé equations, when there is both a special function solution and a rational solution for the same parameters, then the rational solution is a special case of the special function solution.
Finally, we discussed the relationship between P V P_{\mathrm{V}} P V (2) and P I I I P_{\mathrm{III}} P III (4) with C D ≠ 0 CD\ne0 C D = 0 . Using the known relationship between P I I I P_{\mathrm{III}} P III and deg \operatorname{deg} deg -P V P_{\mathrm{V}} P V (3) , we clarified a direct relationship between P V P_{\mathrm{V}} P V (2) with α = − β \alpha=-\beta α = − β , γ = 0 \gamma=0 γ = 0 and P I I I P_{\mathrm{III}} P III (4) with A + B = − 1 A+B=-1 A + B = − 1 , C = − D = 1 4 C=-D=\frac{1}{4} C = − D = 4 1 . Motivated by the work of Tsegel’nik [107 ] , we obtained a second relationship for P V P_{\mathrm{V}} P V (2) with ( α , β , γ ) = ( 1 8 ( λ − 1 ) 2 , − 1 8 ( λ + 1 ) 2 , 0 ) (\alpha,\beta,\gamma)=\left(\frac{1}{8}(\lambda-1)^2,-\frac{1}{8}(\lambda+1)^2,0\right) ( α , β , γ ) = ( 8 1 ( λ − 1 ) 2 , − 8 1 ( λ + 1 ) 2 , 0 ) .
Declarations
Acknowledgements. BM is grateful to EPSRC for the support of Doctoral Training Partnerships grant EP/W524050/1.
Data availability statement. No data were used for the research described in the article.
Declaration of competing interest. The authors declare no conflicts of interest.
ORCID iDs
Peter A. Clarkson 0000-0002-8777-5284
Clare Dunning 0000-0003-0535-9891