On recurrence coefficients of classical orthogonal polynomials
K. Castillo, G. Gordillo-Núñez · 2025
Read On recurrence coefficients of classical orthogonal polynomials by K. Castillo, G. Gordillo-Núñez (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-2507-20646.txt.
Introduction
Castillo and Gordillo-Núñez show that a claimed new test for classical orthogonal polynomials was already covered by older general theorems. Their note appeared in two thousand twenty-five. Two recent papers proposed judging a polynomial family from its first recurrence coefficients. The note checks those claims against the record.
The authors show both papers' results follow from two theorems published years earlier. They also supply a symbolic algorithm that automates the check. The program runs in a standard algebra system and works on quadratic lattices. As a test case, they treat a family said to live on a double lattice. It turns out to be a classical family on a plain linear lattice after all. Its properties then drop out of the older theorem. The note restores priority and simplifies the theory.
Catalog record
| CREATOR | K. Castillo, G. Gordillo-Núñez |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-20646
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On recurrence coefficients of classical orthogonal polynomials K. Castillo, G. Gordillo-Núñez Abstract . In Lett. Math. Phys. 114 , 54 (2024) and 115 , 70 (2025), the author introduces what is presented as a novel method for determining whether a sequence of orthogonal polynomials is “classical”, based solely on its initial recurrence coefficients. This note demonstrates that all the results contained in those works are already encompassed by two general theorems previously established in J. Math. Anal. Appl. 515 (2022), Article 126390. A symbolic algorithm, implemented in Mathematica , is also provided to enable automated verification of the classical character of orthogonal polynomial sequences on quadratic lattices. As an application, it is shown that the so-called para-Krawtchouk polynomials on bi-lattices, discussed in Lett. Math. Phys. 115 , 70 (2025), constitute a particular instance of a classical orthogonal family on a linear lattice. Consequently, their algebraic properties follow as a specific case of one of the main theorems established in J. Math. Anal. Appl. 515 (2022), Article 126390. 2010 Mathematics Subject Classification 42C05, 33C45 1. Introduction To ensure self-containment, the essential elements required to follow the core developments are briefly reviewed below, with topological considerations and other technical subtleties intentionally omitted. A quick reading of Chapters 1 and 9 of [ 4 ] —which lay the foundations of the so-called algebraic theory of orthogonal polynomials, originally developed by Maroni —may offer the reader a deeper perspective on these topics. Nonetheless, such familiarity is not strictly necessary, given the specific focus of the present work. A sequence of polynomials (P_{n})_{n\in\mathbb{N}} , with each P_{n} having exact degree n , is said to be orthogonal with respect to a linear functional \mathbf{u} on \mathbb{C}[\cdot] , the space of all polynomial functions (or simply polynomials—in our context, we shall make no distinction between the two) with complex coefficients, whenever (1) \displaystyle\langle\mathbf{u},P_{n}\,P_{m}\rangle\begin{cases}\neq 0,&\text{if }n=m,\\[7.0pt] =0,&\text{if }n\neq m.\end{cases} In a completely analogous fashion, one may construct a finite sequence of orthogonal polynomials satisfying the orthogonality relations for all degrees n\in\mathbb{N} such that n\leq M , for some fixed natural number M . A linear functional \mathbf{v}:\mathbb{C}[\cdot]\to\mathbb{C} is said to be regular if there exists a sequence of orthogonal polynomials with respect to \mathbf{v} . In this context, a lattice is defined as any mapping x:\mathbb{C}\to\mathbb{C} of the form [ 4 , (1.9)] (2) \displaystyle x(s)=\begin{cases}\mathfrak{c}_{1}q^{-s}+\mathfrak{c}_{2}q^{s}+\mathfrak{c}_{3},&q\neq 1,\\[7.0pt] \mathfrak{c}_{4}s^{2}+\mathfrak{c}_{5}s+\mathfrak{c}_{6},&q=1,\end{cases} where q is a fixed positive parameter and \mathfrak{c}_{1},\dots,\mathfrak{c}_{6} are complex constants such that (\mathfrak{c}_{1},\mathfrak{c}_{2})\neq(0,0) for q\neq 1 and (\mathfrak{c}_{4},\mathfrak{c}_{5},\mathfrak{c}_{6})\neq(0,0,0) for q=1 . Observe that \displaystyle x\left(s+\frac{1}{2}\right)+x\left(s-\frac{1}{2}\right)=2\,\alpha\,x(s)+2\,\beta, where \alpha and \beta are given by \displaystyle\alpha=\frac{q^{1/2}+q^{-1/2}}{2},\quad\beta=\begin{cases}(1-\alpha)\,\mathfrak{c}_{3},&q\neq 1,\\[7.0pt] \displaystyle\frac{1}{4}\,\mathfrak{c}_{4},&q=1.\end{cases} Given a lattice x and a polynomial p , we define the x -derivative and the x -average [ 4 , Definition 1.2] , denoted respectively by \mathrm{D} and \mathrm{S} , as the linear operators satisfying \deg(\mathrm{D}p)=\deg p-1 , \deg(\mathrm{S}p)=\deg p , and \displaystyle\mathrm{D}p(x(s)) \displaystyle=\frac{p\left(x\left(s+\displaystyle\frac{1}{2}\right)\right)-p\left(x\left(s-\displaystyle\frac{1}{2}\right)\right)}{x\left(s+\displaystyle\frac{1}{2}\right)-x\left(s-\displaystyle\frac{1}{2}\right)}, \displaystyle\mathrm{S}p(x(s)) \displaystyle=\frac{1}{2}\,p\left(x\left(s+\displaystyle\frac{1}{2}\right)\right)-\frac{1}{2}\,p\left(x\left(s-\displaystyle\frac{1}{2}\right)\right), where \mathrm{D}p=p^{\prime} and \mathrm{S}p=p whenever x(s)=\mathfrak{c}_{6} . The reader should bear in mind that these operators depend intrinsically on the fixed lattice, even though their dependence on the “variable” is often suppressed for simplicity—much like it is occasionally done in the context of the standard derivative. By defining, via transposition, certain operations on a given linear functional \mathbf{v}:\mathbb{C}[\cdot]\to\mathbb{C} —such as multiplication by a polynomial p [ 4 , Definition 1.1] , x -differentiation, and x -averaging [ 4 , Definition 1.3] — denoted respectively by p\mathbf{v} , \mathbf{Dv} , and \mathbf{Sv} , and by applying standard principles of duality, it can be shown that all the families in the Askey scheme of hypergeometric orthogonal polynomials and their q-analogues are orthogonal with respect to a regular linear functional \mathbf{u}:\mathbb{C}[\cdot]\to\mathbb{C} satisfying the equation (3) \displaystyle\mathbf{D}(\phi\,\mathbf{u})=\mathbf{S}(\psi\,\mathbf{u}), where \phi is a nonzero polynomial of degree at most two, and \psi is a polynomial of degree exactly one, both independent of n . In this note, we follow the definition given in [ 3 , Definition 3.1] and [ 4 , Definition 9.1] , according to which a sequence of orthogonal polynomials with respect to \mathbf{u} is said to be classical if it satisfies the equation ( 3 ). A detailed understanding of the origin of ( 3 ) is not required to follow the present note; the reader may simply rely on the previously stated assertions and the theorems that follow. Theorem 1.1 . [ 3 , Theorem 4.1] [ 4 , Theorem 9.2] Consider the lattice x(s)=\mathfrak{c}_{1}q^{-s}+\mathfrak{c}_{2}q^{s}+\mathfrak{c}_{3}. Let {\bf u}:\mathbb{C}[\cdot]\to\mathbb{C} be a nontrivial linear functional, and suppose that there exist polynomials \phi and \psi , of degree at most two and one respectively, not both identically zero, such that \mathbf{D}(\phi\,{\bf u})=\mathbf{S}(\psi\,{\bf u}). Let \phi(z)=az^{2}+bz+c,\quad\psi(z)=dz+e. Then, {\bf u} is regular if and only if (4) d_{n}\neq 0\quad\text{and}\quad\phi^{[n]}\left(\mathfrak{c}_{3}-\frac{e_{n}}{d_{2n}}\right)\neq 0,\quad\text{for all }n\in\mathbb{N}, where \displaystyle\phi^{[n]}(z) \displaystyle=\left(d(\alpha^{2}-1)\gamma_{2n}+a\alpha_{2n}\right)\left((z-\mathfrak{c}_{3})^{2}-2\mathfrak{c}_{1}\mathfrak{c}_{2}\right) \displaystyle\quad+\left(\phi^{\prime}(\mathfrak{c}_{3})\alpha_{n}+\psi(\mathfrak{c}_{3})(\alpha^{2}-1)\gamma_{n}\right)(z-\mathfrak{c}_{3})+\phi(\mathfrak{c}_{3})+2a\mathfrak{c}_{1}\mathfrak{c}_{2}, and \displaystyle d_{n}=a\gamma_{n}+d\alpha_{n},\quad e_{n}=(2a\mathfrak{c}_{3}+b)\gamma_{n}+(d\mathfrak{c}_{3}+e)\alpha_{n}, with \alpha_{n}=\frac{q^{n/2}+q^{-n/2}}{2},\quad\gamma_{n}=\displaystyle\frac{q^{n/2}-q^{-n/2}}{q^{1/2}-q^{-1/2}}. Only under the aforementioned regularity conditions does there exist a monic orthogonal polynomial sequence (P_{n})_{n\in\mathbb{N}} —possibly only up to some finite degree, if condition ( 4 ) is satisfied for n\in\mathbb{N} with n\leq M , for some fixed natural number M —that satisfies the three-term recurrence relation P_{n+1}(z)=(z-B_{n})\,P_{n}(z)-C_{n}\,P_{n-1}(z), with initial conditions P_{-1}(z)=0 , P_{0}(z)=1 , and recurrence coefficients given by \displaystyle B_{n} \displaystyle=\mathfrak{c}_{3}+\frac{\gamma_{n}e_{n-1}}{d_{2n-2}}-\frac{\gamma_{n+1}e_{n}}{d_{2n}}, \displaystyle C_{n+1} \displaystyle=-\frac{\gamma_{n+1}d_{n-1}}{d_{2n-1}d_{2n+1}}\phi^{[n]}\left(\mathfrak{c}_{3}-\frac{e_{n}}{d_{2n}}\right). Theorem 1.2 . [ 3 , Theorem 4.2] [ 4 , Theorem 9.3] Consider the quadratic lattice x(s)=\mathfrak{c}_{4}s^{2}+\mathfrak{c}_{5}s+\mathfrak{c}_{6}. Let {\bf u}:\mathbb{C}[\cdot]\to\mathbb{C} be a nontrivial linear functional, and suppose that there exist polynomials \phi and \psi , of degree at most two and one respectively, not both identically zero, such that \mathbf{D}(\phi\,{\bf u})=\mathbf{S}(\psi\,{\bf u}). Define \phi(z)=az^{2}+bz+c,\quad\psi(z)=dz+e. Then, {\bf u} is regular if and only if (5) d_{n}\neq 0,\quad\phi^{[n]}\left(-\frac{1}{4}\mathfrak{c}_{4}n^{2}-\frac{e_{n}}{d_{2n}}\right)\neq 0,\quad\text{for all }n\in\mathbb{N}, where d_{n}=an+d,\quad e_{n}=bn+e+\frac{1}{2}\mathfrak{c}_{4}dn^{2}, and \displaystyle\phi^{[n]}(z) \displaystyle=az^{2}+\left(b+\frac{3}{2}\mathfrak{c}_{4}nd_{n}\right)z+\phi\left(\frac{1}{4}\mathfrak{c}_{4}n^{2}\right)+\frac{1}{2}\mathfrak{c}_{4}n\psi\left(\frac{1}{4}\mathfrak{c}_{4}n^{2}\right) \displaystyle\quad-\frac{n}{4}\left(16\mathfrak{c}_{4}\mathfrak{c}_{6}-\mathfrak{c}_{5}^{2}\right)d_{n}. Only under the aforementioned regularity conditions does there exist a monic orthogonal polynomial sequence (P_{n})_{n\in\mathbb{N}} —possibly only up to some finite degree, if condition ( 4 ) is satisfied for n\in\mathbb{N} with n\leq M , for some fixed natural number M —that satisfies the three-term recurrence relation P_{n+1}(z)=(z-B_{n})\,P_{n}(z)-C_{n}\,P_{n-1}(z), with initial conditions P_{-1}(z)=0 , P_{0}(z)=1 , and recurrence coefficients given by \displaystyle B_{n} \displaystyle=\frac{ne_{n-1}}{d_{2n-2}}-\frac{(n+1)e_{n}}{d_{2n}}-\mathfrak{c}_{4}n(n-1), \displaystyle C_{n+1} \displaystyle=-\frac{(n+1)d_{n-1}}{d_{2n-1}d_{2n+1}}\,\phi^{[n]}\left(-\frac{1}{4}\mathfrak{c}_{4}n^{2}-\frac{e_{n}}{d_{2n}}\right). In Section 2 , it will be shown, in a single step, that all the results of [ 5 , 6 ] are already contained in Theorems 1.1 and 1.2 . Section 3 presents an algorithm for determining whether a given pair of sequences, interpreted as the recurrence coefficients of a monic orthogonal polynomial sequence, corresponds to a classical family on x(s)=\mathfrak{c}_{4}s^{2}+\mathfrak{c}_{5}s+\mathfrak{c}_{6} . Section 4 demonstrates —both formally and through the algorithm presented in the preceding section — that the so-called para-Krawtchouk polynomials on bi-lattices are simply a special case of classical orthogonal polynomials on a linear lattice, though conveyed in an inadvertently obscure and intricate manner. Consequently, their regularity, recurrence coefficients, and all associated algebraic properties follow directly from previously established theorems, rendering the main theorem of [ 1 ] largely superfluous. 2. A “Proof” of the Main Results of [ 5 , 6 ] Assume that (P_{n})_{n\geq 0} is a sequence of classical orthogonal polynomials in the sense of ( 1 ). Consequently, the recurrence coefficients are determined by Theorem 1.1 or Theorem 1.2 . By evaluating B_{n} and C_{n+1} at n=0 and n=1 in these theorems, one obtains explicit expressions for the polynomials \phi and \psi in