Cauchy-Jacobi orthogonal polynomials and the discrete CKP equation
Shi-Hao Li, Satoshi Tsujimoto, Ryoto Watanabe, Guo-Fu Yu · 2025
Read Cauchy-Jacobi orthogonal polynomials and the discrete CKP equation by Shi-Hao Li, Satoshi Tsujimoto, Ryoto Watanabe, Guo-Fu Yu (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-2504-18891.txt.
Introduction
Shi-Hao Li, Satoshi Tsujimoto, Ryoto Watanabe, and Guo-Fu Yu build step-by-step changes for a special family of polynomials named for Cauchy and Jacobi. Orthogonal polynomials obey neat three-term rules and serve across physics and number work. The team crafts discrete moves that shift these polynomials while keeping their core traits. Their 2025 paper then asks which discrete equation the shift obeys.
The scale factor of the polynomials turns out to play the lead role in a known discrete equation, the CKP equation. That equation surfaces in soliton theory, cluster algebra, and shape geometry. Tying the polynomials to it opens paths to borrow tools from each side. The work adds a new worked case to the map between special functions and discrete systems.
Catalog record
| CREATOR | Shi-Hao Li, Satoshi Tsujimoto, Ryoto Watanabe, Guo-Fu Yu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-18891
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Cauchy-Jacobi orthogonal polynomials and the discrete CKP equation Shi-Hao Li, Satoshi Tsujimoto, Ryoto Watanabe, Guo-Fu Yu Abstract . This paper intends to construct discrete spectral transformations for Cauchy-Jacobi orthogonal polynomials, and find its corresponding discrete integrable systems. It turns out that the normalization factor of Cauchy-Jacobi orthogonal polynomials acts as the τ \tau -function of the discrete CKP equation, which has applications in Yang-Baxter equation, integrable geometry, cluster algebra, and so on. 2020 Mathematics Subject Classification 39A36, 15A15 1. Introduction The relation between orthogonal polynomials and integrable systems has been studied for many years, due to its deep connection with conformal field theory, random matrix theory and numerical algorithms, etc. On the one hand, orthogonal polynomials play crucial roles in the characterization of spectral problems, thus making connections with Lax representations for discrete integrable equations. On the other hand, new orthogonal polynomials were proposed when solving the spectral problem of classical integrable systems. Cauchy bi-orthogonal polynomials is one of the examples proposed during the study of Hermite-Padé approximation problem of Degasperis-Procesi peakon equation [ 15 ] . Later on, classical properties of Cauchy bi-polynomials were investigated [ 2 , 3 ] , including its zeros distribution, four-term recurrence relation, Riemann-Hilbert problem and so on. Besides, such a polynomial is related to random matrix theory. In [ 4 ] , the Cauchy two-matrix model with specific Laguerre weight was considered, with limiting behavior expressed in terms of Meijer-G function. To be precise, the inner product of Cauchy-Laguerre weight is defined from \mathbb{R}[x]\times\mathbb{R}[y]\to\mathbb{R} , by \displaystyle\langle f(x),g(y)\rangle=\int_{\mathbb{R}_{+}^{2}}\frac{x^{a}y^{b}e^{-x-y}}{x+y}f(x)g(y)dxdy. To evaluate the related bi-moments \langle x^{i},y^{j}\rangle with Cauchy-Laguerre weight, we can introduce a one-parameter generalization of the inner product [ 8 , lemma 2.3] \displaystyle\langle f(x),g(y)\rangle_{t}=\int_{\mathbb{R}_{+}^{2}}\frac{x^{a}y^{b}e^{-t(x+y)}}{x+y}f(x)g(y)dxdy. This inner product plays a dominant role when we connect Cauchy bi-orthogonal polynomials with Toda-type equation. In [ 12 ] , it has been shown that with a=b , the one-parameter deformed Cauchy-Laguerre weight is related to the CKP hierarchy, as well as a semi-discrete Toda equation of CKP type (C-Toda for brevity). If a\neq b , there is also a related discrete integrable system, with \tau -function expressed by an asymmetric Gram determinant [ 6 ] . Besides, in [ 16 ] , a full-discrete equation was constructed by introducing a discrete evolution of the inner product, while in [ 14 ] , a non-abelian C-Toda lattice was obtained by using matrix-valued Cauchy bi-orthogonal polynomials. In this paper, we mainly consider a deformed Cauchy bi-orthogonal polynomials with Jacobi weight, and its corresponding discrete integrable system. By introducing a two-parameter symmetric inner product \displaystyle\langle f(x),g(y)\rangle_{s,t}=\int_{[0,1]\times[0,1]}\frac{x^{s}y^{s}}{x+y}f(x)g(y)\left(\frac{1-x}{1+x}\right)^{t}\left(\frac{1-y}{1+y}\right)^{t}dxdy,\quad s,t\in\mathbb{N}, we show the corresponding orthogonal polynomials \{P_{n}^{s,t}(x)\}_{n\in\mathbb{N}} not only have the four-term recurrence relation, but some spectral transformations with respect to parameters s and t . Different with the discrete time evolutions considered in [ 16 ] , we consider an evolution with respect to variables x and y simultaneously. In Section 2 , we construct different spectral transformations in s -direction and t -direction, which, compatible with four-term recurrence relation, leading to an algebraic identity and nonlinear difference equations. By noting the normalization factor could be simply expressed by a discrete Gram determinant, in Section 3 , we simplify nonlinear equations by using spectral transformations, and decouple nonlinear equations into bilinear equations. It was finally found that these bilinear equations have also been derived in [ 11 ] from hexahedron recurrence relations. Besides, the bilinear equations could be expressed in terms of a single quartic equation, governed by the normalization factor of Cauchy-Jacobi orthogonal polynomials. It is nothing but the discrete CKP equation, equivalent to the star-triangle move in the Ising model studied by Kashaev [ 10 ] . To conclude, there are some highlights of this paper: (1) We introduce a two-parameter deformation for Cauchy orthogonal polynomials in Jacobi type, and study its related discrete spectral transformations; (2) We provide a determinant solution for the discrete CKP equation. Different from the soliton solution, this is a molecule solution determined by its initial values; (3) We construct a Lax pair for the discrete CKP equation. The integrability of discrete CKP equation was obtained by Yang-Baxter equation, and we show its Lax integrability by using orthogonal polynomial method. 2. Two-parameter generalization of Cauchy orthogonal polynomials Let’s consider an inner product with Cauchy weight by defining a symmetric bilinear form \langle\cdot,\cdot\rangle:\mathbb{R}[x]\times\mathbb{R}[y]\to\mathbb{R} \displaystyle\langle f(x),g(y)\rangle=\int_{\mathbb{R}_{+}\times\mathbb{R}_{+}}\frac{1}{x+y}f(x)g(y)\omega(x)\omega(y)dxdy. Here we assume that the weight function \omega(x) is supported on a subset of \mathbb{R}_{+} to avoid the singularity brought by the Cauchy kernel 1/(x+y) . With this inner product, we could obtain a family of monic orthogonal polynomials \{P_{n}(x)\}_{n\in\mathbb{N}} by requiring the orthogonality \displaystyle\langle P_{n}(x),P_{m}(y)\rangle=h_{n}\delta_{n,m}, where \delta_{n,m}=0 if n\neq m and \delta_{n,m}=1 if n=m . In this paper, we focus on a two-parameter deformation of the Cauchy inner product, reading as \displaystyle\langle f(x),g(y)\rangle_{s,t}=\int_{(0,1)\times(0,1)}\frac{x^{s}y^{s}}{x+y}f(x)g(y)\left(\frac{1-x}{1+x}\right)^{t}\left(\frac{1-y}{1+y}\right)^{t}dxdy,\quad s,t\in\mathbb{N}. In other words, we choose a deformed Jacobi weight function \omega(x)=x^{s}(1-x)^{t}(1+x)^{-t} , supported on (0,1) . With this deformed inner product, we could introduce a family of monic two-parameter Cauchy orthogonal polynomials \{P^{s,t}_{n}(x)\}_{n\in\mathbb{N}} by orthogonality \langle P_{n}^{s,t}(x),P_{m}^{s,t}(y)\rangle_{s,t}=h_{n}^{s,t}\delta_{n,m} , where h_{n}^{s,t} is a non-zero constant. Proposition 2.1 . The two-parameter deformed monic Cauchy orthogonal polynomials have the following determinant expression \displaystyle P_{n}^{s,t}(x)=\frac{1}{\tau_{n}^{s,t}}\left|\begin{array}[]{cccc}m_{0,0}^{s,t}&\cdots&m_{0,n-1}^{s,t}&1\\ \vdots&&\vdots&\vdots\\ m_{n-1,0}^{s,t}&\cdots&m_{n-1,n-1}^{s,t}&x^{n-1}\\ m_{n,0}^{s,t}&\cdots&m_{n,n-1}^{s,t}&x^{n}\end{array}\right|,\quad\tau_{n}^{s,t}=\det\left(m_{i,j}^{s,t}\right)_{i,j=0}^{n-1}, where \displaystyle m_{i,j}^{s,t}=\langle x^{i},y^{j}\rangle_{s,t}=\int_{(0,1)^{2}}\frac{x^{s+i}y^{s+j}}{x+y}\left(\frac{1-x}{1+x}\right)^{t}\left(\frac{1-y}{1+y}\right)^{t}dxdy. Using the determinant formula, it is not difficult to verify that h_{n}^{s,t}=\tau_{n+1}^{s,t}/\tau_{n}^{s,t} , and the non-zero property of normalization factor is guaranteed by the Andréief formula (for reference, see [ 2 , Equation (2.5)] ) \displaystyle\tau_{n}^{s,t}=\int_{0<x_{1}<\cdots<x_{n}<1\atop 0<y_{1}<\cdots<y_{n}<1}\prod_{i,j=1}^{n}\frac{x_{i}^{s}y_{j}^{s}}{x_{i}+y_{j}}\prod_{1\leq i<j\leq n}(x_{i}-x_{j})^{2}(y_{i}-y_{j})^{2}\prod_{i=1}^{n}\left(\frac{1-x_{i}}{1+x_{i}}\right)^{t}\left(\frac{1-y_{i}}{1+y_{i}}\right)^{t}dx_{i}dy_{i}. (2.1) Noting that every term in the integration is positive within the interval (0,1) , we can define a probability measure on the configuration space. However, different from the Cauchy-Laguerre case, it is still unknown how to evaluate this partition function in terms of hypergeometric functions. Moreover, we have the following properties of the deformed inner product. Proposition 2.2 . The two-parameter deformed inner product has the following evolution properties: (1) \langle f(x),g(y)\rangle_{s+1,t}=\langle xf(x),yg(y)\rangle_{s,t}; (2) \langle f(x),g(y)\rangle_{s,t+1}=\langle f(x),g(y)\rangle_{s,t}-\mathcal{L}_{s,t}(f)\mathcal{L}_{s,t}(g) , where \displaystyle\mathcal{L}_{s,t}(f)=\sqrt{2}\int_{0}^{1}\frac{x^{s}}{1+x}f(x)\left(\frac{1-x}{1+x}\right)^{t}dx. As a direct result, we know that moments \{m_{i,j}^{s,t}\}_{i,j\in\mathbb{N}} satisfy relations \displaystyle\begin{aligned} m_{i,j}^{s+1,t}=m_{i+1,j+1}^{s,t},\quad m_{i,j}^{s,t+1}=m_{i,j}^{s,t}-\phi_{i}^{s,t}\phi_{j}^{s,t}\end{aligned} (2.2) with \phi_{i}^{s,t}=\mathcal{L}_{s,t}(x^{i}) . Similar to ordinary Cauchy orthogonal polynomials, we have the following four-term recurrence relation for \{P_{n}^{s,t}(x)\}_{n\in\mathbb{N}} . Proposition 2.3 . For any s,\,t\in\mathbb{N} , we have \displaystyle\begin{aligned} x(P_{n}^{s,t}(x)&+a_{n}^{s,t}P_{n-1}^{s,t}(x))\\ &=P_{n+1}^{s,t}(x)+(a_{n}^{s,t}-b_{n}^{s,t})P_{n}^{s,t}(x)+(-c_{n}^{s,t}+a_{n}^{s,t}b_{n-1}^{s,t})P^{s,t}_{n-1}(x)-a_{n}^{s,t}c_{n-1}^{s,t}P^{s,t}_{n-2}(x)\end{aligned} (2.3) for some non-zero coefficients a_{n}^{s,t} , b_{n}^{s,t} and c_{n}^{s,t} , where initial values are given by P_{-1}^{s,t}(x)=0 and P_{0}^{s,t}(x)=1 . It has been noticed in [ 12 ] that variables a_{n}^{s,t} , b_{n}^{s,t} and c_{n}^{s,t} have determinant expressions. By introducing \displaystyle\tilde{\sigma}_{n}^{s,t}=\left|\begin{array}[]{cccc}m^{s,t}_{0,0}&\cdots&m^{s,t}_{0,n-1}&m^{s,t}_{0}\\ m^{s,t}_{1,0}&\cdots&m^{s,t}_{1,n-1}&m^{s,t}_{1}\\ \vdots&&\vdots&\vdots\\ m^{s,t}_{n,0}&\cdots&m^{s,t}_{n,n-1}&m_{n}^{s,t}\end{array}\right|,\quad m_{i}^{s,t}=\int_{0}^{1}x^{s+i}\left(\frac{1-x}{1+x}\right)^{t}dx and denoting p_{n}^{s,t} as the coefficient of x^{n-1} in P_{n}^{s,t}(x) , we have \displaystyle a_{n}^{s,t}=-\frac{\tilde{\sigma}^{s,t}_{n}\tau^{s,t}_{n-1}}{\tilde{\sigma}^{s,t}_{n-1}\tau^{s,t}_{n}},\quad b_{n}=p_{n+1}^{s,t}-p_{n}^{s,t},\quad c_{n}^{s,t}=\frac{\tau_{n-1}^{s,t}\tau_{n+1}^{s,t}}{(\tau_{n}^{s,t})^{2}}. (2.8) It should be remarked that p_{n}^{s,t} has determinant expression as well. If we introduce \displaystyle{\tilde{\tau}}_{n}^{s,t}=\left|\begin{array}[]{ccc}m_{0,0}^{s,t}&\cdots&m_{0,n-1}^{s,t}\\ \vdots&&\vdots\\ m_{n-2,0}^{s,t}&\cdots&m_{n-2,n-1}^{s,t}\\ m_{n,0}^{s,t}&\cdots&m_{n,n-1}^{s,t}\end{array}\right|, then we have p_{n}^{s,t}=-{\tilde{\tau}}_{n}^{s,t}/\tau_{n}^{s,t} . Remark 2.4 . The four-term recurrence relation is based on the fact that \displaystyle\int_{0}^{1}\left(P_{n}^{s,t}(x)+a_{n}^{s,t}P_{n-1}^{s,t}(x)\right)dx=0 rather than \mathcal{L}_{s,t}\left(P_{n}^{s,t}(x)+a_{n}^{s,t}P_{n-1}^{s,t}(x)\right)=0. Therefore, we need to introduce single moments m_{i}^{s,t} instead of \phi_{i}^{s,t} to express the coefficient a_{n}^{s,t} . 2.1. Discrete spectral transformation I The first discrete spectral transformation is related to another family of polynomials \{Q_{n}^{s,t}(x)\}_{n\in\mathbb{N}} . Let’s define monic polynomials \displaystyle Q_{n}^{s,t}(x)=\frac{1}{\xi_{n}^{s,t}}\left|\begin{array}[]{cccc}m_{0,1}^{s,t}&\cdots&m_{0,n}^{s,t}&1\\ \vdots&&\vdots&\vdots\\ m_{n,1}^{s,t}&\cdots&m_{n,n}^{s,t}&x^{n}\\ \end{array}\right|,\quad\xi_{n}^{s,t}=\det\left(m_{i,j+1}^{s,t}\right)_{i,j=0}^{n-1}. In the theory of orthogonal polynomials, this family of polynomials is called the adjacent family of polynomials \{P_{n}^{s,t}(x)\}_{n\in\mathbb{N}} . Similar to the