Zeros and Orthogonality of Generalized Fibonacci Polynomials

Basic Information

• Paper ID: 2510.00074
• Title: Zeros and Orthogonality of Generalized Fibonacci Polynomials
• Authors: Cristian F. Coletti, Rigoberto Flórez, Robinson A. Higuita, Sandra Z. Yepes
• Classification: math.CO (Combinatorics)
• Publication Date: October 11, 2025 (arXiv v2)
• Paper Link: https://arxiv.org/abs/2510.00074

Abstract

This paper analyzes the concept of orthogonality for second-order polynomial sequences with Binet formulas similar to Fibonacci and Lucas numbers, referred to as generalized Fibonacci polynomials (GFP). The paper provides techniques for finding GFP roots and, as a corollary, offers an alternative proof of a special case of Favard's theorem. While the general Favard theorem guarantees the existence of a measure determining whether a second-order polynomial sequence is orthogonal, it does not provide an explicit measure. The special case presented here provides both an explicit measure and establishes the relationship between second-order recurrence relations and orthogonality, enabling determination of whether GFP polynomials are orthogonal. This allows classification of which familiar GFPs are orthogonal and which are not. Some familiar orthogonal polynomials include Fermat, Fermat-Lucas, both classes of Chebyshev polynomials, both classes of Morgan-Voyce polynomials, and Vieta and Vieta-Lucas polynomials. However, the paper proves that Fibonacci, Lucas, Pell, and Pell-Lucas sequences are not orthogonal.

Research Background and Motivation

Problem Background

1. Core Problem: Determining which generalized Fibonacci polynomials (GFP) possess orthogonality properties and under what conditions they are orthogonal or non-orthogonal.
2. Significance:
   • Orthogonal polynomials have important applications in mathematical analysis, numerical computation, probability theory, and stochastic processes
   • Fibonacci-type polynomials appear widely in combinatorics and number theory
   • Research connecting orthogonal polynomial theory with classical recurrence sequences
3. Limitations of Existing Methods:
   • While the general case of Favard's theorem guarantees the existence of an orthogonal measure, it does not provide an explicit form
   • Lack of systematic methods to determine orthogonality for specific GFP families
   • Insufficient research on the roots of GFPs
4. Research Motivation:
   • Provide explicit orthogonal measures
   • Establish clear connections between second-order recurrence relations and orthogonality
   • Classify orthogonality properties of common GFPs
   • Explore connections between orthogonal polynomials and Markov processes

Core Contributions

1. Developed systematic techniques for finding GFP roots: Determining roots of general GFPs through roots of classical Fibonacci and Lucas polynomials
2. Provided alternative proof of special case of Favard's theorem: Offering explicit orthogonal measures and discriminant conditions
3. Completed orthogonality classification of common GFPs: Clearly distinguishing which 8 of 13 common polynomials are orthogonal and which 5 are not
4. Established connections between orthogonal GFPs and Markov chains: Determining sufficient conditions for orthogonal polynomials to induce random walks
5. Provided explicit weight function constructions: Giving explicit weight functions for orthogonal GFPs

Methodology Details

Task Definition

Input: Generalized Fibonacci polynomial sequences defined by recurrence relations

• Fibonacci-type: $F_0(x) = 0, F_1(x) = 1, F_n(x) = d(x)F_{n-1}(x) + g(x)F_{n-2}(x)$
• Lucas-type: $L_0(x) = p_0, L_1(x) = p_1(x), L_n(x) = d(x)L_{n-1}(x) + g(x)L_{n-2}(x)$

Output: Determining whether the sequence is orthogonal; if orthogonal, providing the weight function

Constraints: $d(x), g(x) \in \mathbb{Q}[x]$, with $\deg(d(x)) > \deg(g(x))$, and $d^2(x) + 4g(x) \neq 0$

Core Theoretical Framework

1. Binet Formula Representation

For $n \geq 0$ with $d^2(x) + 4g(x) \neq 0$, the Binet formula is: $F_n(x) = \frac{a^n(x) - b^n(x)}{a(x) - b(x)}, \quad L_n(x) = \frac{a^n(x) + b^n(x)}{\alpha}$

where: $a(x) = \frac{d(x) + \sqrt{d^2(x) + 4g(x)}}{2}, \quad b(x) = \frac{d(x) - \sqrt{d^2(x) + 4g(x)}}{2}$

2. Root Determination Method

Theorem 4.3: Let $F_n(x)$ be a GFP as defined in (2.1). If $r \in \mathbb{C}$ satisfies $\frac{d(r)}{\sqrt{g(r)}} = \gamma_j$, where $\gamma_j = 2i\cos\frac{j\pi}{n}$ ($j = 1,2,\ldots,n-1$) and $g(r) \neq 0$, then $r$ is a root of $F_n(x)$.

Theorem 4.4: Similarly, for Lucas-type GFP, if $\frac{d(t)}{\sqrt{g(t)}} = \tau_j$, where $\tau_j = 2i\cos\frac{(2j+1)\pi}{2n}$, then $t$ is a root of $L_n(x)$.

3. Orthogonality Discriminant Conditions

Proposition 3.6: Let $g(x) = -4k$ ($k \in \mathbb{R}_{<0}$), and $d(x)$ as defined in (2.1) and (2.2). If there exist constants $s_1, s_2$ such that $d(s_1) = -\sqrt{-4k}$, $d(s_2) = \sqrt{-4k}$, and for every $x$ in the interval $[s_1, s_2]$ we have $4k + d^2(x) \leq 0$, then for $n \neq m$:

$\int_{s_1}^{s_2} F_n(x)F_m(x)\sqrt{-4k - d^2(x)}d'(x)dx = 0$

Technical Innovations

1. Unified root-solving method: Reducing the root problem for general GFPs to that of classical Fibonacci/Lucas polynomials
2. Explicit weight function construction: Providing concrete orthogonal weight functions for specific forms of $d(x)$ and $g(x)$
3. Systematic classification framework: Determining orthogonality based on properties of $d(x)$ and $g(x)$

Experimental Setup

Research Objects

The paper studied 13 common GFPs, including:

• Non-orthogonal: Fibonacci, Lucas, Pell, Pell-Lucas, Jacobsthal
• Orthogonal: Fermat, Fermat-Lucas, Chebyshev (both classes), Morgan-Voyce (both classes), Vieta, Vieta-Lucas

Analysis Methods

1. Theoretical analysis: Using Favard's theorem and orthogonal polynomial theory
2. Root computation: Based on known results for classical polynomial roots
3. Weight function verification: Verifying orthogonal relationships through integral calculations

Verification Standards

Using the standard definition of orthogonality: $\langle f_i(x), f_j(x) \rangle = \int_{\mathbb{R}} f_i(x)f_j(x)d\mu(x) = \delta(i,j)$

Experimental Results

Main Results

1. Orthogonality Classification

Orthogonal GFPs (8 types):

• Fermat: $\Phi_n(x) = 3x\Phi_{n-1}(x) - 2\Phi_{n-2}(x)$
• Fermat-Lucas: $\vartheta_n(x) = 3x\vartheta_{n-1}(x) - 2\vartheta_{n-2}(x)$
• Chebyshev polynomials of the second kind: $U_n(x) = 2xU_{n-1}(x) - U_{n-2}(x)$
• Chebyshev polynomials of the first kind: $T_n(x) = 2xT_{n-1}(x) - T_{n-2}(x)$
• Morgan-Voyce polynomials (both classes)
• Vieta and Vieta-Lucas polynomials

Non-orthogonal GFPs (5 types):

• Fibonacci, Lucas, Pell, Pell-Lucas, Jacobsthal

2. Weight Function Results

Corollary 3.7: For $d(x) = cx^t + h$, $g(x) = -k/4$ (where $c,h,k,t \in \mathbb{Z}$, $c \neq 0$, $k,t > 0$, and $t$ is odd), the weight function is: $\omega(x) = \sqrt{k - d^2(x)}x^{t-1}$

3. Markov Chain Connection

Proposition 5.2: Under specific conditions, Lucas-type GFPs determine a random walk with transition matrix: