Matrix Bessel Biorthogonal Polynomials: A Riemann–Hilbert approach
Amílcar Branquinho 1, Ana Foulquié-Moreno 2, Assil Fradi 3, Manuel Mañas 4 · 2025
Read Matrix Bessel Biorthogonal Polynomials: A Riemann–Hilbert approach by Amílcar Branquinho 1, Ana Foulquié-Moreno 2, Assil Fradi 3, Manuel Mañas 4 (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-2502-19326.txt.
Introduction
Amílcar Branquinho, Ana Foulquié-Moreno, Assil Fradi, and Manuel Mañas studied in 2025 matrix-valued orthogonal polynomials tied to Bessel-type weights. These polynomials, whose entries are matrices rather than numbers, arise in random matrix theory and related fields. Their weights are defined through a matrix version of a classic differential equation.
Starting from a Riemann-Hilbert problem, the authors derive first and second order differential relations for the polynomials and their companion functions. They show the recurrence coefficients obey matrix versions of a known family of discrete Painlevé equations, which govern many integrable systems. Nontrivial worked examples of Bessel-type matrix polynomials close the paper.
Catalog record
| CREATOR | Amílcar Branquinho 1, Ana Foulquié-Moreno 2, Assil Fradi 3, Manuel Mañas 4 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 60 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-19326
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Matrix Bessel Biorthogonal Polynomials: A Riemann–Hilbert approach Thanks: 1 Acknowledges Centre for Mathematics of the University of Coimbra (funded by the Portuguese Government through FCT/MCTES, Multi-Annual Financing Program for R&D Units, doi: 10.54499/UIDB/00324/2020). Thanks: 2 and 3 Acknowledge the CIDMA Center for Research and Development in Mathematics and Applications (University of Aveiro) and the Portuguese Foundation for Science and Technology (FCT) Multi-Annual Financing Program for R&D Units, for their support within projects doi: 10.54499/UIDB/04106/2020 & doi: 10.54499/UIDP/04106/2020 Thanks: 4 Acknowledge PID2021-122154NB-I00, entitled “Ortogonalidad y Aproximación con Aplicaciones en Machine Learning y Teoría de la Probabilidad”, funded by MICIU/AEI/10.13039 /501100011033 and by “RDF A Way of making Europe”. Amílcar Branquinho 1, Ana Foulquié-Moreno 2, Assil Fradi 3, Manuel Mañas 4 Abstract . We consider matrix orthogonal polynomials related to Bessel type matrices of weights that can be defined in terms of a given matrix Pearson equation. From a Riemann–Hilbert problem we derive first and second order differential relations for the matrix orthogonal polynomials and functions of second kind. It is shown that the corresponding matrix recurrence coefficients satisfy a non-Abelian extensions of a family of discrete Painlevé d-PIV equations. We present some nontrivial examples of matrix orthogonal polynomials of Bessel type. 2020 Mathematics Subject Classification 33C45, 33C47, 42C05, 47A56. 1. Introduction In [ 13 ] , Krall and Frink introduced the generalized Bessel polynomials (GBPs), (1) \displaystyle y_{n}(x,a,b)=\,_{2}F_{0}\left(-n,a+n-1;-;-\frac{x}{b}\right), \displaystyle n\in\mathbb{N}\coloneqq\{0,1,2,\ldots\}. These polynomials satisfy an orthogonality property on the unit circle, for all m,n\in\mathbb{N} , (2) \displaystyle\int_{|z|=1}\rho^{(a,b)}(z)y_{m}(z,a,b)y_{n}(z,a,b)\frac{\operatorname{d}\hskip-1.46527ptz}{2\pi\hskip-1.28032pt\operatorname{i}}=\frac{(-1)^{n+1}n!\,b\,\Gamma(a)\,\delta_{m,n}}{(a+2n-1)\Gamma(a+n-1)}, where \delta_{m,n} is the Kronecker delta, \Gamma denotes the gamma function, and the weight function \rho^{(a,b)} is given by (3) \displaystyle\rho^{(a,b)}(z)\coloneqq\sum_{n=0}^{\infty}\frac{\Gamma(a)}{\Gamma(a+n-1)}\Big(-\frac{b}{z}\Big)^{n}. As noted in [ 15 ] , the integral in ( 2 ) is not an inner product with the kernel \displaystyle\rho^{(a,b)}(z)y_{m}(z,a,b)\overline{y_{n}(z,a,b)}, which is unexpected for polynomials orthogonal on curves in the complex plane. In the special case where a\in\mathbb{N} and m,n\in\mathbb{N} , Burchnall provided a simpler orthogonality property over any closed contour C surrounding the origin z=0 in the complex plane (see [ 5 ] ), for all m,n\in\mathbb{N} , \displaystyle\int_{C}z^{a-2}\operatorname{e}^{-\frac{b}{z}}y_{m}(z,a,b)y_{n}(z,a,b)\frac{\operatorname{d}\hskip-1.46527ptz}{2\pi\hskip-1.28032pt\operatorname{i}}=\frac{(-1)^{a+n-1}b^{a-1}n!}{(a+n-2)!(a+2n-1)}\delta_{m,n}. Later, an orthogonality relation on the positive real axis for all m,n\in\mathbb{N} , with \operatorname{Re}(a)<1-m-n , was given in [ 9 ] , where the scaling factor b has been put equal to unity without loss of generality, for all m,n\in\mathbb{N} , \displaystyle\int_{0}^{\infty}x^{a-2}\operatorname{e}^{-\frac{b}{x}}y_{m}(x,a,b)y_{n}(x,a,b)\operatorname{d}\hskip-1.46527ptx=\frac{(-1)^{n}b^{a-1}n!}{(a+2n-1)\Gamma(a+n-1)}\frac{\pi}{\sin(\pi a)}\delta_{m,n}, and in the same work, the following was established, for all m,n\in\mathbb{N} , \displaystyle\int_{0}^{(\infty+)}z^{a-2}\operatorname{e}^{-\frac{b}{z}}y_{m}(z;a,b)y_{n}(z;a,b)\frac{\operatorname{d}\hskip-1.46527ptz}{2\pi\hskip-1.28032pt\operatorname{i}}=\frac{(-1)^{a+n}b^{a-1}n!}{(a+2n-1)\Gamma(a+n-1)}\delta_{m,n}, where the contour is taken along a simple loop starting at the origin (z=0) , encircling the point at infinity once in the positive (counterclockwise) direction, and then returning to the origin. For this choice of contour, \sigma(z)=z^{a-2}\operatorname{e}^{-b/z} , satisfies the Pearson equation \big(z^{2}\sigma\big)^{\prime}=(az+b)\sigma , which is crucial for deriving the main explicit formulas in this work. On the other hand, the weight function \rho^{(a,b)} , defined in ( 3 ) and satisfying the orthogonality relation ( 2 ), does not satisfy a Pearson equation except in the cases a=1 or a=2 . Specifically, \displaystyle\big(z^{2}\rho^{(a,b)}\big)^{\prime}=(az+b)\rho^{(a,b)}-(a-1)(a-2)z. For further details, see [ 13 , equation 40] and [ 15 ] for an overview on this subject. Our study of matrix valued orthogonal polynomials of Bessel type is motivated by the fact that the systems of orthogonal polynomials associated with the names of Hermite, Laguerre, Jacobi (including the special cases named after Chebyshev, Legendre, and Gegenbauer), and Bessel are among the most widely studied and applied systems. However, while a Riemann–Hilbert approach has been used to discuss matrix biorthogonal polynomials (MBPs) of Hermite type [ 3 , 6 , 7 , 12 ] , Laguerre type [ 4 ] , and Jacobi type [ 2 ] , to our knowledge, this approach has only been applied to scalar Bessel-type orthogonal polynomials [ 8 ] . Therefore, in this work, we focus on matrix valued orthogonal polynomials of Bessel type. Our aim is to investigate the MBPs associated with Bessel-type matrices of weights, which are constructed in terms of matrix Pearson equations. In this work, we deal with regular matrix of weights W(z) , i.e., their moments W_{n}=\int_{\gamma}z^{n}W(z)\frac{\operatorname{d}\hskip-1.46527ptz}{2\pi\hskip-1.28032pt\operatorname{i}} , n\in\mathbb{N} , are such that \det\big[W_{j+k}\big]_{j,k=0,\ldots n}\not=0 , n\in\mathbb{N} . Definition 1 . We say that a N\times N matrix of weights W=\begin{bmatrix}W^{(1,1)}&\cdots&W^{(1,N)}\\ \vdots&\ddots&\vdots\\ W^{(N,1)}&\cdots&W^{(N,N)}\end{bmatrix} with support \gamma , taken along a simple loop starting at the origin, encircling the point at infinity once in the positive direction, and then returning to the origin, is of Bessel type if the entries W^{(j,k)} of the matrix measure W can be written as (4) \displaystyle W^{(j,k)}(z)=\sum_{m\in I_{j,k}}\varphi_{m}(z)z^{p_{m}}\log^{q_{m}}(z)\operatorname{e}^{-\frac{\lambda_{m}}{z}}, \displaystyle z\in\gamma, where I_{j,k} denotes a finite set of indexes, j,k\in\{1,2,\ldots,N\} , \operatorname{Re}(\lambda_{m})\,\geq 0 , q_{m}\in\mathbb{N} and \varphi_{m} is Hölder continuous, bounded and nonvanishing on \gamma . We will request, in the development of the theory, that the functions \varphi_{m} have a holomorphic extension to the whole complex plane. Given a regular matrix of weights W , we define two sequences of matrix monic polynomials , \big\{P_{n}^{\mathsf{L}}(z)\big\}_{n\in\mathbb{Z}_{+}} and \big\{P_{n}^{\mathsf{R}}(z)\big\}_{n\in\mathbb{N}} , which are left, respectively right-orthogonal. These polynomials satisfy the degree conditions \deg P_{n}^{\mathsf{L}}(z)=n , \deg P_{n}^{\mathsf{R}}(z)=n , n\in\mathbb{N} , and are defined by the orthogonality relations, (5) \displaystyle\int_{\gamma}{P}_{n}^{\mathsf{L}}(z)W(z)z^{k}\frac{\operatorname{d}\hskip-1.46527ptz}{2\pi\hskip-1.28032pt\operatorname{i}}=\delta_{n,k}C_{n}^{-1}, \displaystyle\int_{\gamma}z^{k}W(z){P}_{n}^{\mathsf{R}}(z)\frac{\operatorname{d}\hskip-1.46527ptz}{2\pi\hskip-1.28032pt\operatorname{i}}=\delta_{n,k}C_{n}^{-1}, for all k\in\{0,1,\ldots,n\} and n\in\mathbb{N} , where C_{n} is a nonsingular matrix. The weight matrix W induces a sesquilinear form on the set of matrix polynomials \mathbb{C}^{N\times N}[z] given by \displaystyle\langle P,Q\rangle_{W}\coloneqq\int_{\gamma}{P}(z)W(z)Q(z)\frac{\operatorname{d}\hskip-1.46527ptz}{2\pi\hskip-1.28032pt\operatorname{i}}, with respect to which the sequences \big\{P_{n}^{\mathsf{L}}(z)\big\}_{n\in\mathbb{N}} and \big\{P_{n}^{\mathsf{R}}(z)\big\}_{n\in\mathbb{N}} are biorthogonal, \big\langle P_{n}^{\mathsf{L}},{P}_{m}^{\mathsf{R}}\big\rangle_{W}=\delta_{n,m}C_{n}^{-1} , n,m\in\mathbb{N} . Since the polynomials are monic, they can be expressed as follows, \displaystyle{P}_{n}^{\mathsf{L}}(z) \displaystyle=\operatorname{I}z^{n}+p_{\mathsf{L},n}^{1}z^{n-1}+p_{\mathsf{L},n}^{2}z^{n-2}+\cdots+p_{\mathsf{L},n}^{n}, \displaystyle{P}_{n}^{\mathsf{R}}(z) \displaystyle=\operatorname{I}z^{n}+p_{\mathsf{R},n}^{1}z^{n-1}+p_{\mathsf{R},n}^{2}z^{n-2}+\cdots+p_{\mathsf{R},n}^{n}, where p_{\mathsf{L},n}^{k},p_{\mathsf{R},n}^{k}\in\mathbb{C}^{N\times N} , k\in\{0,\ldots,n\} and n\in\mathbb{N} , with the condition p_{\mathsf{L},n}^{0}=p_{\mathsf{R},n}^{0}=\operatorname{I} , n\in\mathbb{N} and \operatorname{I}\in\mathbb{C}^{N\times N} denotes the identity matrix. We define the sequence of second kind matrix functions , for all n\in\mathbb{N} , by \displaystyle Q^{\mathsf{L}}_{n}(z)\coloneqq\int_{\gamma}\frac{P^{\mathsf{L}}_{n}(z^{\prime})}{z^{\prime}-z}{W(z^{\prime})}\frac{\operatorname{d}\hskip-1.46527ptz^{\prime}}{2\pi\hskip-1.28032pt\operatorname{i}}, \displaystyle{Q}_{n}^{\mathsf{R}}(z)\coloneqq\int_{\gamma}W(z^{\prime})\frac{P^{\mathsf{R}}_{n}(z^{\prime})}{z^{\prime}-z}\frac{\operatorname{d}\hskip-1.46527ptz^{\prime}}{2\pi\hskip-1.28032pt\operatorname{i}}. From the orthogonality conditions ( 5 ), the following asymptotic expansions hold for all n\in\mathbb{N} , as |z|\to\infty (6) \displaystyle\hskip-3.84096ptQ^{\mathsf{L}}_{n}(z)=-C_{n}^{-1}z^{-n}\big(\operatorname{I}z^{-1}+q_{\mathsf{L},n}^{1}z^{-2}+\cdots\big),Q^{\mathsf{R}}_{n}(z)=-z^{-n}\big(\operatorname{I}z^{-1}+q_{\mathsf{R},n}^{1}z^{-2}+\cdots\big)C_{n}^{-1}. The sequences \big\{{P}^{\mathsf{L}}_{n}(z)\big\}_{n\in\mathbb{N}} and \big\{{Q}^{\mathsf{L}}_{n}(z)\big\}_{n\in\mathbb{N}} satisfy the following three-term recurrence relations, for all n\in\mathbb{N} , \displaystyle z{P}^{\mathsf{L}}_{n}(z) \displaystyle={P}^{\mathsf{L}}_{n+1}(z)+\xi^{\mathsf{L}}_{n}{P}^{\mathsf{L}}_{n}(z)+\eta^{\mathsf{L}}_{n}{P}^{\mathsf{L}}_{n-1}(z), \displaystyle zQ_{n}^{\mathsf{L}}(z)=Q_{n+1}^{\mathsf{L}}(z)+\xi_{n}^{\mathsf{L}}Q_{n}^{\mathsf{L}}(z)+\eta^{\mathsf{L}}_{n}Q_{n-1}^{\mathsf{L}}(z), where \eta^{\mathsf{L}}_{n}=C_{n}^{-1}C_{n-1} , with the initial conditions {P}^{\mathsf{L}}_{-1}={\bf\mathbf{0}} , {P}^{\mathsf{L}}_{0}=\operatorname{I} , Q_{-1}^{\mathsf{L}}(z)=-C_{-1}^{-1} , and Q_{0}^{\mathsf{L}}(z)=S_{W}(z) , and S_{W}(z) is the Stieltjes–Markov transform given by \displaystyle S_{W}(z)\coloneqq\int_{\gamma}\frac{W\left(z^{\prime}\right)}{z^{\prime}-z}\frac{\operatorname{d}\hskip-1.46527ptz^{\prime}}{2\pi\hskip-1.28032pt\operatorname{i}}. Similarly, the sequences of monic polynomials \big\{{P}^{\mathsf{R}}_{n}(z)\big\}_{n\in\mathbb{N}} and second kind matrix functions \big\{{Q}^{\mathsf{R}}_{n}(z)\big\}_{n\in\mathbb{N}} satisfy the relations, for all n\in\mathbb{N} , \displaystyle zP^{\mathsf{R}}_{n}(z) \displaystyle=P^{\mathsf{R}}_{n+1}(z)+P^{\mathsf{R}}_{n}(z)\xi^{\mathsf{R}}_{n}+P^{\mathsf{R}}_{n-1}(z)\eta^{\mathsf{R}}_{n}, \displaystyle zQ^{\mathsf{R}}_{n}(z)=Q^{\mathsf{R}}_{n+1}(z)+Q^{\mathsf{R}}_{n}(z)\xi^{\mathsf{R}}_{n}+Q^{\mathsf{R}}_{n-1}(z)\eta^{\mathsf{R}}_{n}, where \eta^{\mathsf{R}}_{n}=C_{n}\eta^{\mathsf{L}}_{n}C_{n}^{-1}=C_{n-1}C_{n}^{-1} , with initial conditions {P}^{\mathsf{R}}_{-1}={\bf\mathbf{0}} , {P}^{\mathsf{R}}_{0}=\operatorname{I} , Q_{-1}^{\mathsf{R}}(z)=-C_{-1}^{-1} and Q_{0}^{\mathsf{R}}(z)=S_{W}(z) , and the relation \xi_{n}^{\mathsf{R}}=C_{n}\xi^{\mathsf{L}}_{n}C_{n}^{-1} holds. These objects are collected in the matrices, (7) \displaystyle Y^{\mathsf{L}}_{n}(z) \displaystyle\coloneqq\begin{bmatrix}{P}^{\mathsf{L}}_{n}(z)&Q^{\mathsf{L}}_{n}(z)\\[1.42271pt] -C_{n-1}{P}^{\mathsf{L}}_{n-1}(z)&-C_{n-1}Q^{\mathsf{L}}_{n-1}(z)\end{bmatrix},\ {Y}^{\mathsf{R}}_{n}(z)\coloneqq\begin{bmatrix}P^{\mathsf{R}}_{n}(z)&-P^{\mathsf{R}}_{n-1}(z)C_{n-1}\\[1.42271pt] {Q}^{\mathsf{R}}_{n}(z)&-{Q}^{\mathsf{R}}_{n-1}(z)C_{n-1}\end{bmatrix}. The three term recurrence relations for P^{\mathsf{L}}_{n} , P^{\mathsf{R}}_{n} , Q^{\mathsf{L}}_{n} and Q^{\mathsf{R}}_{n} can be expressed as \displaystyle Y_{n+1}^{\mathsf{L}}(z) \displaystyle=T^{\mathsf{L}}_{n}(z)Y^{\mathsf{L}}_{n}(z), \displaystyle Y_{n+1}^{\mathsf{R}}(z)=Y^{\mathsf{R}}_{n}(z)T^{\mathsf{R}}_{n}(z), \displaystyle n\in\mathbb{N}, where the transfer matrices T^{\mathsf{L}}_{n} and T^{\mathsf{R}}_{n} are given by, \displaystyle T^{\mathsf{L}}_{n}=\begin{bmatrix}z\operatorname{I}-\xi_{n}^{\mathsf{L}}&C_{n}^{-1}\\ -C_{n}&{\bf\mathbf{0}}\end{bmatrix}, \displaystyle T^{\mathsf{R}}_{n}=\begin{bmatrix}z\operatorname{I}-\xi_{n}^{\mathsf{R}}&-C_{n}\\ C_{n}^{-1}&{\bf\mathbf{0}}\end{bmatrix}. Moreover, it is known, cf. [ 3 ] , that \displaystyle\big({Y}^{\mathsf{L}}_{n}(z)\big)^{-1}=\begin{bmatrix}{\bf\mathbf{0}}&\operatorname{I}\\ -\operatorname{I}&{\bf\mathbf{0}}\end{bmatrix}{Y}^{\mathsf{R}}_{n}(z)\begin{bmatrix}{\bf\mathbf{0}}&-\operatorname{I}\\ \operatorname{I}&{\bf\mathbf{0}}\end{bmatrix}. The structure of this work is as follows: In § 2 we formulate the Riemann–Hilbert problem for the matrix Bessel weights. Next, we study the analytic properties of the constant jump fundamental matrix associated with a matrix weight function that satisfies a Pearson-type equation of Bessel type. In § 3 we employ a Riemann–Hilbert approach to derive first and second-order differential equations from the structure matrix. We show how these equations reduce to well-known second-order differential