Algebraic Interpretation of discrete families of matrix valued orthogonal polynomials

Quentin Labriet, Lucia Morey and Luc Vinet · 2025

Read Algebraic Interpretation of discrete families of matrix valued orthogonal polynomials by Quentin Labriet, Lucia Morey and Luc Vinet (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-2510-25948.txt.

Introduction

Quentin Labriet, Lucia Morey, and Luc Vinet give an algebraic account of discrete families of matrix valued orthogonal polynomials. Orthogonal polynomials are families where distinct members integrate to zero against each other under a weight, and matrix valued versions replace numbers with matrices. Such families appear in probability, physics, and special function theory.

Labriet and his co-authors build the families from representations of a Lie algebra, a structure that encodes symmetry, acting on a module over matrices. Cases tied to the algebras su two and su one one, plus a q deformed algebra at a root of unity, yield matrix analogues of the Krawtchouk, Meixner, and discrete Chebyshev polynomials. The view unifies known families under one construction.

Catalog record

CREATORQuentin Labriet, Lucia Morey and Luc Vinet
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE48 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-25948

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Algebraic Interpretation of discrete families of matrix valued orthogonal polynomials Quentin Labriet, Lucia Morey and Luc Vinet Abstract . An algebraic interpretation of matrix-valued orthogonal polynomials (MVOPs) is provided. The construction is based on representations of a ( q q -deformed) Lie algebra 𝔤 \mathfrak{g} into the algebra End M n ​ ( ℂ ) ⁡ ( M ) \End_{M_{n}({\mathbb{C}})}(M) of M n ​ ( ℂ ) M_{n}({\mathbb{C}}) -linear maps over a M n ​ ( ℂ ) M_{n}({\mathbb{C}}) -module M M . Cases corresponding to the Lie algebras 𝔰 ​ 𝔲 ​ ( 2 ) \mathfrak{su}(2) and 𝔰 ​ 𝔲 ​ ( 1 , 1 ) \mathfrak{su}(1,1) as well as to the q q -deformed algebra 𝔰 ​ 𝔬 q ​ ( 3 ) \mathfrak{so}_{q}(3) at q q a root of unity are presented; they lead to matrix analogs of the Krawtchouk, Meixner and discrete Chebyshev polynomials. 1. Introduction The theory of matrix valued orthogonal polynomials (MVOPs for short) was introduced by M.G. Krein in 1949 [ 23 ] . Since then, numerous theoretical advancements have been made, closely following the development of scalar-valued orthogonal polynomials. A comprehensive overview of these developments can be found in [ 6 ] . In a manner similar to scalar orthogonal polynomials, there is considerable interest in studying families of MVOPs that also exhibit the additional property of being eigenfunctions of second-order differential operators [ 11 , 10 , 16 , 18 ] , difference operators [ 3 , 9 , 12 ] , or q -difference operators [ 1 , 2 ] . One method for constructing such examples is by utilizing the representation theory of compact Lie groups and harmonic analysis on compact symmetric spaces. In this context, scalar-valued polynomials are known to arise as matrix coefficients or spherical functions. The spherical function approach has been extended to matrix analogues and successfully applied in works such as [ 16 , 21 , 22 , 17 ] among others. Despite these advancements, there are still relatively few concrete examples of MVOPs, particularly those that satisfy a difference equation. More specifically, our goal is to identify examples of MVOPs along with their recurrence relations, orthogonality relations, and, when applicable, their corresponding differential or difference equations. The goal of this article is to provide an algebraic construction for MVOPs as transition coefficients between two specified bases. We adopt the point of view introduced in [ 15 ] and much developed by one of us often in collaboration with Zhedanov, see in particular the reviews [ 14 ] , [ 7 ] . Specifically, an approach for providing an algebraic interpretation of the polynomial families in the Askey scheme involves the algebra generated by two operators: the multiplication by the operator multiplication by the variable and the (q) -difference or differential operator. Depending on the case, the resulting algebra can either be a Lie algebra (or its q deformation), or a more general quadratic algebra (such as Hahn or Racah) which are specializations of the Askey-Wilson algebra. In [ 15 , 13 ] , scalar valued orthogonal polynomials are obtained as transition coefficients between two eigenbases of two self-adjoint operators H and P inside a Lie algebra. If H admits a discrete spectrum and P acts tridiagonally on the eigenbasis of H then the transition coefficients between both eigenbases can be seen to be orthogonal polynomials with the eigenvalues of P as variables. This method allows for the uncovering of orthogonality relations, three-term recurrence relations, and differential or difference equations satisfied by the polynomials. This known interpretation for scalar orthogonal polynomials, represents a new and intriguing research direction for MVOPs. To apply such a construction, a natural framework is M_{n}({\mathbb{C}}) modules that admits a matrix valued inner product. More precisely, we look at representations of a Lie (and q -deformed) algebra \mathfrak{g} into the algebra \End_{M_{n}({\mathbb{C}})}(M) of M_{n}({\mathbb{C}}) -linear maps over a left M_{n}({\mathbb{C}}) -module M , i.e. Lie algebra morphisms between \mathfrak{g} and \End_{M_{n}({\mathbb{C}})}(M) . Similarly as in the scalar case, we have two self-adjoint operators H and P with P acting tridiagonally on the eigenbasis of H , then MVOPs will be obtained as transition coefficients between both eigenbases. In the recent preprint [ 20 ] , the authors emphasized that naturally defined MVOP are close to their scalar valued prototypes. In some sense the present work follows the same philosophy since families of MVOP are obtained naturally from the algebra related to families of scalar valued polynomials and their representations. More precisely, since M_{n}({\mathbb{C}}) is Morita equivalent to {\mathbb{C}} there is an equivalence of categories between M_{n}({\mathbb{C}}) modules and complex vector spaces. Thus starting from a complex representation of our algebra \mathfrak{g} we get a representation in \End_{M_{n}({\mathbb{C}})}(V) . Using this equivalence, we can unfold some properties of MVOPs starting from the scalar valued case. In Section 2 we present the necessary background for the construction, i.e. Hermitian modules and Morita equivalence. Section 3 introduces the general construction for MVOPs seen as transition coefficients. In the last part of the article, we use this algebraic presentation to construct some families of discrete MVOPs together with their orthogonality, recurrence relation and difference equation. This goal is achieved in Section 4 to 6 , using three different algebras: \mathfrak{su}(2) , \mathfrak{su}(1,1) and \mathfrak{so}_{q}(3) . This leads to the construction of, respectively, Krawtchouk, Meixner and Discrete Chebyshev matrix analogues. 1.1. Notations In all the paper we use A^{*} to denote the transpose conjugate of a square matrix A . For a symmetric matrix A , we write A\geq 0 for non negative matrices and A>0 for positive definite ones. 2. M_{n}({\mathbb{C}}) -module 2.1. Morita equivalence In this subsection we discuss M_{n}({\mathbb{C}}) -modules (thought as left modules) with matrix-valued inner products, that will be called Hermitian modules. First, it is a fact that the algebras M_{n}({\mathbb{C}}) and {\mathbb{C}} are Morita equivalent, see [ 24 ] , i.e. their respective module categories \mathfrak{M}_{M_{n}({\mathbb{C}})} and \mathfrak{M}_{\mathbb{C}} are equivalent. The functors realizing this equivalence are given by F:\mathfrak{M}_{M_{n}(\mathbb{C})}\to\mathfrak{M}_{\mathbb{C}},\qquad G:\mathfrak{M}_{\mathbb{C}}\to\mathfrak{M}_{M_{n}(\mathbb{C})}, on the objects, the functors are defined by G(V)=V^{n},\qquad F(M)=e_{11}M, where e_{ij} denote the elementary matrices. Here V^{n} is a M_{n}({\mathbb{C}}) -module, the action being given, for A\in M_{n}({\mathbb{C}}) , by: (2.1) A\cdot(v_{1},\ldots,v_{n})=\left(\sum_{k}a_{1k}v_{k},\ldots,\sum_{n}a_{nk}v_{k}\right). On the morphisms, the functors are defined by \displaystyle\alpha \displaystyle:V\to V^{\prime},\quad G(\alpha):G(V)\to G(V^{\prime}),\quad G(\alpha)(v_{1},\ldots,v_{n})=(\alpha v_{1},\ldots,\alpha v_{n}), \displaystyle\beta \displaystyle:M\to M^{\prime},\quad F(\beta):F(M)\to F(M^{\prime}),\quad F(\beta)(e_{11}m)=e_{11}\beta(m). We now introduce introduce Hermitian M_{n}({\mathbb{C}}) -modules. This definition is adapted to our purpose from the one in [ 4 , Dfn II.7.1.1] for Hilbert modules. Definition 2.1 . A Hermitian M_{n}({\mathbb{C}}) -module M is a M_{n}({\mathbb{C}}) -module together with a matrix-valued inner product, i.e. a map (\cdot,\cdot):M\times M\to M_{n}({\mathbb{C}}) such that: (1) For all A\in M_{n}({\mathbb{C}}) , (A\cdot m_{1},m_{2})=A(m_{1},m_{2}) , (2) (m_{1},m_{2})=(m_{2},m_{1})^{*} , (3) (m,m)\geq 0 , (4) If (m,m)=0 then m=0 (we say that the inner product is non degenerate). In general the non degeneracy condition (4) can be omitted but in our case every inner product will satisfy it. Another remark is that as a direct consequence of the definition we get (m_{1},A\cdot m_{2})=(m_{1},m_{2})A^{*}. We will now explicitly extend the Morita equivalence from Hermitian spaces V to Hermitian M_{n}({\mathbb{C}}) -modules V^{n} . Let (V,\langle\cdot,\cdot\rangle) be a Hermitian space and endow the M_{n}({\mathbb{C}})

Read the complete text