Connection formulae for the radial Toda equations II
Martin A. Guest, Alexander R. Its, Maksim Kosmakov, Kenta Miyahara ∗, Ryosuke Odoi · 2025
Read Connection formulae for the radial Toda equations II by Martin A. Guest, Alexander R. Its, Maksim Kosmakov, Kenta Miyahara ∗, Ryosuke Odoi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-17783.txt.
Introduction
Martin Guest and co-authors continued in 2025 a study of the radial Toda equations, computing how all solutions behave far out. The earlier paper handled the radial Toda equation itself, whose solutions stay smooth. This sequel treats a partner equation linked to the first through shared monodromy data. The partner's solutions, unlike the first set, carry infinitely many singularities.
That difference makes the standard Riemann-Hilbert steepest descent method harder to run. The authors work through the added complications to extract the far-out asymptotics anyway. Connection formulae are the prize: explicit links between behavior in different regions. Such formulae anchor the theory of integrable equations and their special solutions. The paper is technical work for specialists in mathematical physics. It completes a two-part analysis begun in the earlier article.
Catalog record
| CREATOR | Martin A. Guest, Alexander R. Its, Maksim Kosmakov, Kenta Miyahara ∗, Ryosuke Odoi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 79 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-17783
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Connection formulae for the radial Toda equations II Thanks: ∗ Corresponding author. Martin A. Guest, Alexander R. Its, Maksim Kosmakov, Kenta Miyahara ∗, Ryosuke Odoi Abstract . This is a continuation of [ GIKMO ] in which we computed the asymptotics near x = ∞ x=\infty of all solutions of the radial Toda equation. In this article, we compute the asymptotics near x = ∞ x=\infty of all solutions of a “partner” equation. The equations are related in the sense that their respective monodromy data constitute connected components of the same “monodromy manifold”. While all solutions of the radial Toda equation are smooth, those of the partner equation have infinitely many singularities, and this makes the Riemann-Hilbert nonlinear steepest descent method (and the asymptotics of solutions) more involved. Keywords : Toda equation; Painlevé III equation; isomonodromic deformation; Riemann-Hilbert problem; steepest-descent method 1 Introduction and Main Results In [ GIKMO ] , we studied the large and small x asymptotic behaviors of the solutions of the radial Toda equation of type A_{2} , namely (1.0.1) \displaystyle(w_{0})_{xx}+\frac{1}{x}(w_{0})_{x}=2e^{-2w_{0}}-2e^{4w_{0}}, where w_{0}(x) is a real-valued function on the positive real line. We performed the desired asymptotic analysis by using the Lax-integrability of equation ( 1.0.1 ), a fact that goes back to Mikhailov [ Mik ] : (1.0.2) \displaystyle\begin{cases}\Psi_{\zeta}=\left(-\frac{1}{\zeta^{2}}W-\frac{x}{\zeta}w_{x}-x^{2}W^{T}\right)\Psi\\ \Psi_{x}=\left(-w_{x}-2x\zeta W^{T}\right)\Psi,\end{cases} where (1.0.3) \displaystyle w=\diag(w_{0},0,-w_{0})\quad\text{and}\quad W=\begin{pmatrix}0&e^{-w_{0}}&0\\ 0&0&e^{-w_{0}}\\ e^{2w_{0}}&0&0\end{pmatrix}. Our study of the radial Toda equation was motivated by the series of works on the tt*-Toda equations by Guest-Its-Lin [ GuLi12 , GuLi14 , GIL1 , GIL2 , GIL3 , GIL4 ] . The global solutions are physically important [ CV1 , CV2 , CV3 ] and we have been trying to understand the asymptotic behaviors of more general real-valued radial solutions on \mathbb{R}_{>0} . For both the radial Toda equations and the tt*-Toda equations, we have used a fusion of the Iwasawa factorization method in loop group theory and the Riemann-Hilbert nonlinear steepest descent method of Deift and Zhou [ DZ ] . In [ GIKMO ] we observed that, for a given solution of ( 1.0.1 ), the corresponding monodromy data for the first linear equation of ( 1.0.2 ) belongs to a 2 -dimensional manifold \mathcal{M} . Explicitly, this manifold can be described by four real parameters (A^{\mathbb{R}},s^{\mathbb{R}},x^{\mathbb{R}},y^{\mathbb{R}}) satisfying two algebraic equations: (1.0.4) \displaystyle(A^{\mathbb{R}})^{2}-\frac{1}{3}A^{\mathbb{R}}=(x^{\mathbb{R}})^{2}+(y^{\mathbb{R}})^{2}, \displaystyle(1+s^{\mathbb{R}})A^{\mathbb{R}}+2x^{\mathbb{R}}=\frac{1}{3}. This has two connected components, given (respectively) by A^{\mathbb{R}}>0 and A^{\mathbb{R}}<0 . (See Figure 1 for a representation of the manifold in \mathbb{R}^{3} .) Figure 1 . Monodromy manifold \mathcal{M} In [ GIKMO ] we showed that the (real) solutions w_{0}(x) of ( 1.0.1 ) correspond to the component A^{\mathbb{R}}>0 , and (based on the Riemann-Hilbert method) we observed that points in the component A^{\mathbb{R}}<0 would arise from the monodromy data of non-real solutions w_{0}(x)=v_{0}(x)+\frac{i\pi}{2} , x\in\mathbb{R}_{>0} , where v_{0}(x) is real. In other words, the function v_{0}(x) is a real-valued solution of the following “new” differential equation (1.0.5) \displaystyle(v_{0})_{xx}+\frac{1}{x}(v_{0})_{x}=-2e^{-2v_{0}}-2e^{4v_{0}}, which could be regarded as a “monodromy partner” of ( 1.0.1 ). The main result of this paper is to prove that the (real) solutions v_{0}(x) of ( 1.0.5 ) correspond precisely to the component A^{\mathbb{R}}<0 , by means of a detailed examination of the asymptotics of all solutions near x=\infty . Thus the relation between connected components of the manifold \mathcal{M} and equations ( 1.0.1 ) and ( 1.0.5 ) is not trivial and of some interest in its own right. Theorem 1.1 . The monodromy manifold corresponding to the solutions of ( 1.0.1 ) is the connected component of \mathcal{M} defined by ( 1.0.4 ) with A^{\mathbb{R}}>0 . The monodromy manifold corresponding to the solutions of ( 1.0.5 ) is the connected component of \mathcal{M} defined by ( 1.0.4 ) with A^{\mathbb{R}}<0 . The radial Toda equation ( 1.0.1 ) is known to be equivalent to a special case of the third Painlevé equation (PIII) of type D_{7} , see [ GIKMO , section 2] . The same holds for the partner equation, see Corollary 1 (as well for the tt*-Toda equation, in fact). From the Painlevé property, it follows that the radial Toda equation, its partner equation (and the tt*-Toda equation) can have only isolated logarithmic poles as “movable singularities” between zero and infinity. In the case of the radial Toda equation, it is easy to verify that these singularities cannot be real (see [ GIKMO , Proposition 2.1] ) by general ODE arguments. In the case of the partner equation, there is no such restriction, indeed our asymptotic analysis shows that they appear in the generic solutions of those equations. The equation PIII( D_{7} ) is also known as the degenerate Painlevé III equation and has been studied in several contexts. One of the important global aspects is to find the large and small x asymptotics and to connect the asymptotic parameters. In 1987, A. Kitaev [ Kitaev ] did pioneering work in this direction for a particular choice of parameters, and later Kitaev-Vartanian [ KitV1 , KitV2 ] extended the result for arbitrary choice of parameters for purely real solutions and purely imaginary solutions. They used an alternative 2\times 2 Lax pair and applied the WKB method to the associated direct monodromy problem. Although the result by Kitaev-Vartanian already captures the singular solutions of the partner equation ( 1.0.5 ), our Riemann-Hilbert approach will give another proof of the large x asymptotics, and along the way prove Theorem 1.1 . More importantly, the approach taken in [ KitV1 , KitV2 ] does not readily generalize to tackling the asymptotics of the partner equations of the radial Toda equation and the tt*-Toda equation of type A_{n} for general n>1 , which is our overarching goal. In contrast, our 3\times 3 Riemann-Hilbert approach for n=2 is more promising to generalize to the (n+1)\times(n+1) problem for general n in principle. It is important to notice that this paper deals with a 3\times 3 Riemann-Hilbert problem with singularities which gives extra complication in the Riemann-Hilbert analysis, see section 3 . In the 2\times 2 situation, there are several prior studies [ FIS , BothnerIts ] , but this paper will be the first to apply the scheme of those works to the higher rank problems. Our formulation of the large x asymptotics of ( 1.0.5 ) is as follows: Theorem 1.2 . For any (A^{\mathbb{R}},s^{\mathbb{R}},x^{\mathbb{R}},y^{\mathbb{R}}) , A^{\mathbb{R}}<0 satisfying ( 1.0.4 ), there is a real solution v_{0}(x) of \displaystyle(v_{0})_{xx}+\frac{1}{x}(v_{0})_{x}=-2e^{-2v_{0}}-2e^{4v_{0}}, such that (1.0.6) \displaystyle v_{0}(x)=\frac{1}{2}\ln\left(\frac{1+\sin\vartheta(x)}{2-\sin\vartheta(x)}\right)+\mathcal{O}(x^{-1}),\quad x\to\infty, where \displaystyle\vartheta(x)=-2\sqrt{3}x+i\nu_{0}\ln(24\sqrt{3}x)-\frac{\pi}{2}-\arg(x^{\mathbb{R}}+iy^{\mathbb{R}})-\arg\Gamma(-\nu), \displaystyle\nu=-\frac{1}{2\pi i}\ln 3A^{\mathbb{R}}=-\frac{1}{2}+\nu_{0},\quad\nu_{0}\in i\mathbb{R}. The asymptotics is uniform outside of the small neighborhood of each singularity appearing in ( 1.0.6 ). Remark . The asymptotic formula ( 1.0.6 ) implies that if we label the singularities of \frac{1}{2}\ln\left(\frac{1+\sin\vartheta(x)}{2-\sin\vartheta(x)}\right) by x_{n} in the increasing order on the real line, then for n\in\mathbb{Z} large enough, one has (1.0.7) \displaystyle x_{n}=\frac{2n\pi+\pi/3+i\nu_{0}\ln(24n\pi)-\arg\left(x^{\mathbb{R}}+iy^{\mathbb{R}}\right)-\arg\left(\Gamma(-\nu)\right)}{2\sqrt{3}}+o(1). This describes asymptotic locations of movable singularities of v_{0}(x) for x large. Corollary 1 . The change of variables (1.0.8) \displaystyle x=\frac{3}{4}s^{\frac{2}{3}}\quad\text{and}\quad\tilde{w}=s^{\frac{1}{3}}e^{-2v_{0}} transform ( 1.0.5 ) into the following \mathrm{PIII}(D_{7}) equation with (\alpha,\,\beta,\,\gamma,\,\delta)=(1,0,0,1) in the standard notation, (1.0.9) \displaystyle\tilde{w}_{ss}=\frac{\tilde{w}_{s}^{2}}{\tilde{w}}-\frac{\tilde{w}_{s}}{s}+\frac{\tilde{w}^{2}}{s}+\frac{1}{\tilde{w}}, and vice versa. Then, it follows from Theorem 1.2 that the large x asymptotics of a solution to ( 1.0.9 ) is given by the formula \displaystyle\tilde{w}(s)=s^{1/3}\,\frac{2-\sin\tilde{\vartheta}(s)}{1+\sin\tilde{\vartheta}(s)}+\mathcal{O}\left(s^{-2/3}\right),\quad s\to\infty. where \tilde{\vartheta}(s)=\vartheta\left(\frac{3}{4}s^{2/3}\right) and \vartheta is