[Paper Review] A characterization of Leonard pairs using the parameters $\{a_i\}_{i=0}^{d}$

Edward Hanson|arXiv (Cornell University)|May 20, 2012

graph theory and CDMA systems13 references3 citations

TL;DR

This paper provides an algebraic characterization of Leonard pairs using the parameters $\{a_i\}_{i=0}^d$, extending Pascasio's combinatorial results on $Q$-polynomial distance-regular graphs to a purely linear algebraic framework. It establishes that a Leonard pair is fully determined by the $a_i$ parameters and their associated structural constraints, offering a new criterion for identifying such pairs without relying on graph-theoretic assumptions.

ABSTRACT

Let $V$ denote a vector space with finite positive dimension. We consider an ordered pair of linear transformations $A: V o V$ and $A^*: V o V$ that satisfy (i) and (ii) below. (i) There exists a basis for $V$ with respect to which the matrix representing $A$ is irreducible tridiagonal and the matrix representing $A^*$ is diagonal. (ii) There exists a basis for $V$ with respect to which the matrix representing $A^*$ is irreducible tridiagonal and the matrix representing $A$ is diagonal. We call such a pair a Leonard pair on $V$. Arlene Pascasio recently obtained a characterization of the $Q$-polynomial distance-regular graphs using the intersection numbers $a_i$. In this paper, we extend her results to a linear algebraic level and obtain a characterization of Leonard pairs. Pascasio's argument appears to rely on the underlying combinatorial assumptions, so we take a different approach that is algebraic in nature.

Motivation & Objective

• To generalize Pascasio's characterization of $Q$-polynomial distance-regular graphs using intersection numbers to a purely algebraic setting.
• To provide a characterization of Leonard pairs that depends only on the parameters $\{a_i\}_{i=0}^d$ and their structural constraints.
• To eliminate reliance on combinatorial assumptions by developing an algebraic framework based on tridiagonal and diagonal matrix representations.
• To introduce and analyze the concept of 'leaves' in the context of Leonard systems, using the $a_i$ parameters to identify structural features.
• To present two algorithms for recognizing leaves in the associated diagram $\Delta$, based on the $a_i$ parameters and eigenvalue sequences.

Proposed method

• Uses the definition of a Leonard pair via two bases: one where $A$ is irreducible tridiagonal and $A^*$ is diagonal, and vice versa.
• Introduces the parameters $\{a_i\}_{i=0}^d$ from the diagonal entries of $A$ in a feasible basis, forming the core of the characterization.
• Applies the theory of Leonard systems and orthogonal polynomials to derive recurrence relations involving $a_i$, $\theta_i$, and $\theta_i^*$.
• Employs a diagram $\Delta$ to represent adjacency relations between eigenvalues, with 'leaves' defined as vertices adjacent to only one other vertex.
• Develops two algorithms to test whether a vertex $r$ is a leaf adjacent only to $s$, based on the $a_i$ parameters and eigenvalue sequences.
• Uses recurrence relations and matrix equations to verify whether the $a_i$ parameters satisfy necessary and sufficient conditions for a Leonard pair.

Experimental results

Research questions

• RQ1 Can the characterization of $Q$-polynomial distance-regular graphs via intersection numbers be generalized to a purely algebraic setting without graph-theoretic assumptions?
• RQ2 How can the parameters $\{a_i\}_{i=0}^d$ be used to fully characterize a Leonard pair?
• RQ3 What algebraic conditions on $\{a_i\}_{i=0}^d$ ensure that a pair of linear transformations forms a Leonard pair?
• RQ4 How can the concept of a 'leaf' in the eigenvalue diagram $\Delta$ be algorithmically identified using only the $a_i$ parameters?
• RQ5 What conditions on the $a_i$ parameters and eigenvalue sequences $\{\theta_i^*\}$ ensure that a vertex in $\Delta$ is a leaf?

Key findings

• A Leonard pair is characterized by the $a_i$ parameters and their compatibility with a system of recurrence relations involving $\theta_i^*$ and $\theta_i$.
• The condition $\theta_i^* \neq \theta_0^*$ for $1 \leq i \leq d$ ensures the connectedness of the diagram $\Delta$, which is necessary for the characterization.
• A vertex $r$ is a leaf adjacent only to $s$ if and only if the sequence $u_i(\theta_s)$ satisfies a specific rational function identity involving $\theta_i^*$ and $a_r^*$.
• The first algorithm for recognizing a leaf uses a recursive computation of $\alpha_j$ and checks a matrix equation at each step for consistency with the eigenvalue $\theta_s$.
• The second algorithm, assuming constant row sum for $A$, simplifies the verification by eliminating the need to compute $\alpha_j$ and instead checks a single matrix equation per index $j$.
• The paper proves that the $a_i$ parameters fully determine the Leonard pair up to isomorphism, provided the eigenvalue and parameter constraints are satisfied.