---
title: Sobolev Orthogonal Polynomials
url: https://www.emergentmind.com/topics/sobolev-orthogonal-polynomials
type: topic
---

# Sobolev Orthogonal Polynomials

Sobolev orthogonal polynomials are orthogonal polynomial systems with respect to a Sobolev-type inner product—one in which derivatives of the polynomials (possibly of various orders) and function values at discrete points are coupled, sometimes involving matrix-valued measures. In the fully discrete Jacobi–Sobolev setting, these polynomials may satisfy finite-order differential equations and arise as eigenfunctions of higher-order self-adjoint operators. Their construction, recurrence structure, and spectral properties are central both in approximation theory and in the spectral analysis of differential equations.

## 1. Discrete Jacobi–Sobolev Bilinear Forms and Defining Orthogonality

The discrete Jacobi–Sobolev bilinear form central to the study in "Differential equations for discrete Jacobi-Sobolev orthogonal polynomials" [1510.02570] is defined for parameters α, β ∈ ℝ, m₁, m₂ ∈ ℕ₀ (m = m₁ + m₂ ≥ 1), and positive semi-definite real matrices M, N. Let the generalized Jacobi weight be:
\[
w_{α−m₂,β−m₁}(x) = (1 - x)^{α-m_2}(1 + x)^{β-m_1},\quad x \in (-1,1).
\]
For polynomials \(p, q\), the Sobolev bilinear form is:
\[
(p, q) = \int_{-1}^{1} p(x)\,q(x)\,w_{α-m_2, β-m_1}(x)\,dx
+ T_{m_1}(p)\,M\,T_{m_1}(q)^T
+ T_{m_2}(p)\,N\,T_{m_2}(q)^T
\]
where \(T_{m_1}(p) = (p(-1), p'(-1), \dots, p^{(m_1-1)}(-1))\), and \(T_{m_2}(p)\) similar at \(x=+1\).

Orthogonal polynomials with respect to this form are sequences \((Q_n)\) with
\[
(Q_n, x^k) = 0, \quad k=0,\dots,n-1,
\]
and unique (up to scalars) of exact degree \(n\).

## 2. Construction via Casorati Determinants

Discrete Jacobi–Sobolev orthogonal polynomials are constructed using Casorati (quasi-determinant) structures:

- **Jacobi base:** Use classical Jacobi polynomials \(J_n^{(α,β)}\) normalized so \(D_{α,β} J_n^{(α,β)} = n(n+α+β+1) J_n^{(α,β)}\) for the hypergeometric-type operator
\[
D_{α,β} = (1-x^2) \frac{d^2}{dx^2} + [(β-α) - (α+β+2)x] \frac{d}{dx}.
\]

- **Auxiliary sequences:** From the matrices \(M, N\), define \(m=m_1+m_2\) sequences \(z_\ell(n)\), each as explicit sums involving Jacobi polynomials and matrix entries.

- **Casorati determinant:** Let \(p(n), q(n)\) be rising and falling Pochhammer-type products, then assemble
\[
A(n) = \det[p(n-j) q(n-j);\, z_\ell(n-j)]_{1\le \ell,j\le m}
\]
and form
\[
Q_n(x) = \frac{1}{A(n)}\; \det
\begin{pmatrix}
J_n^{(α,β)}(x) & \dots & J_{n-m}^{(α,β)}(x) \\
z_1(n) & \dots & z_1(n-m) \\
\vdots & & \vdots \\
z_m(n) & \dots & z_m(n-m)
\end{pmatrix}.
\]
If \(A(n)\neq0\) for all \(n\), then these \(Q_n\) are orthogonal (of exact degree \(n\)) for the Sobolev bilinear form [1510.02570].

## 3. Differential Operators and Finite-Order Eigenproblems

A central result is that discrete Jacobi–Sobolev orthogonal polynomials are eigenfunctions of finite-order differential operators. For integer parameters α, β, the eigenoperator is explicitly constructed:

- **D-operator scheme:** Using first-order operators
\[
\mathcal{D}_1 = (1-x)\frac{d}{dx} + \frac{β+1}{2};\qquad
\mathcal{D}_2 = (1+x)\frac{d}{dx} - \frac{α+1}{2},
\]
any rational function \(S(x)\), and auxiliary polynomials \(Y_i(\theta)\), the constructed operator
\[
\mathbb{D}_{Q,S} = P_S(D_{α,β}) + \sum_{i=1}^m M_i(D_{α,β})\,\mathcal{D}_i\,Y_i(D_{α,β})
\]
yields
\[
\mathbb{D}_{Q,S} Q_n = \Lambda_n Q_n,
\quad \Lambda_n = P_S(n(n+α+β+1)).
\]
For the Sobolev case, a specific rational \(S(x)\) in terms of the matrices and combinatorial factors ensures that the Casorati determinant and the operator have the required algebraic properties [1510.02570].

- **Order bound:** The order of the differential operator is determined via the weighted ranks of \(M,N\): it does not exceed
\[
\text{order}(\mathbb{D}_{Q,S}) \le 2\left[\beta-wr(M) + \alpha-wr(N) + 1\right],
\]
where wr(M), wr(N) are weighted ranks defined from defect sequences arising from the structure of the Casorati determinant.

## 4. Algebraic Structure and Spectral Properties

Discrete Jacobi–Sobolev polynomials exhibit banded recurrence relations and generalized eigenvalue structures.

- **Recurrence:** In the generic case, a recurrence with bandwidth \(m+2\), where \(m=m_1+m_2\) is the total number of mass points, arises:
\[
x Q_n(x) = Q_{n+1}(x) + \sum_{k=0}^{m} \beta_{n,k} Q_{n-k}(x).
\]
The precise coefficients are computable via the structure of the Casorati determinant and the matrices \(M,N\).

- **Spectral Theory:** The polynomials
are generalized eigenfunctions corresponding to a higher-order differential operator, satisfying
\[
\mathbb{D}_{Q,S}(Q_n) = \Lambda_n Q_n,
\]
where the eigenvalues \(\Lambda_n\) and the operator are determined by the constructed symbol \(S(x)\) [1510.02570].

- **Limiting Cases:** When both \(M=N=0\), the polynomials reduce to classical Jacobi polynomials, yielding Bochner’s second-order self-adjoint operator. With one Dirac mass (e.g., \(M>0\), \(N=0\) or vice versa) and integer α or β, one recovers higher-order Jacobi–Koornwinder (“generalized Jacobi”) operators.

- **Unification and Extension:** The construction encompasses all previously known finite-order Sobolev-type ODEs, including those for Laguerre–Sobolev and discrete Jacobi–Sobolev polynomials, and supplies a route to extend to discrete–difference analogues (e.g., for Hahn/Meixner families) via the same D-operator methods.

## 5. Connection to Broader Theory and Open Problems

- **Generality of Construction:** The approach unifies the treatment of Sobolev orthogonal polynomials with arbitrary mass-point structures, providing explicit (and minimal, up to open conjectures in low-dimensional or symmetric settings) order computations for the associated differential operators [1510.02570].

- **Relation to Coherent Pair

Source: https://www.emergentmind.com/topics/sobolev-orthogonal-polynomials