Mixed-Type MOP and Riemann-Hilbert Analysis

• Mixed-type multiple orthogonal polynomials are defined by satisfying orthogonality conditions across two or more weight systems, capturing both type I and type II configurations.
• The approach utilizes LU factorization, a 'snake' Jacobi matrix, and tau-function determinants to explicitly formulate recurrence relations, Christoffel-Darboux kernels, and projection operators.
• The Riemann-Hilbert framework recasts the polynomial system into a matrix analytic problem, enabling rigorous asymptotic analysis and revealing deep connections to integrable systems and random matrix theory.

Mixed-Type Multiple Orthogonal Polynomial/Riemann-Hilbert Description

A mixed-type multiple orthogonal polynomial is a polynomial (or vector of polynomials) that satisfies orthogonality conditions distributed among two or more systems of weights, possibly of differing types—such as type I and type II, or combinations thereof. The Riemann-Hilbert (RH) description refers to encoding both the polynomial system and its properties (orthogonality, recurrence, and asymptotics) as the solution to a matrix RH problem, which is a crucial tool for analysis, especially in random matrix theory, approximation theory, and integrable systems. This area synthesizes algebraic, analytic, and spectral perspectives within a unified framework.

1. Algebraic Structure of Mixed-Type Multiple Orthogonal Polynomials

Mixed-type multiple orthogonality generalizes classical single-weight/one-dimensional orthogonality and traditional multiple orthogonal polynomials by considering systems involving polynomials and linear forms with respect to different weight families, which may be acted on by distinct or combined sets of orthogonality conditions.

For given measures $\mu$ (possibly vector or matrix-valued) and two families of weights $\vec{w}_1 = (w_{1,1}, ..., w_{1,p_1})$, $\vec{w}_2 = (w_{2,1}, ..., w_{2,p_2})$, the mixed-type (MT) orthogonality problem seeks vector polynomials $\vec{A} = (A_1, ..., A_{p_1})$ of prescribed degrees, not all zero, such that

$$\sum_{a=1}^{p_1}\int A_a(x) w_{1,a}(x)w_{2,b}(x)x^j\, d\mu(x) = 0, \quad j=0,\ldots,n_{2,b}-1,\;\; b=1,\ldots,p_2.$$

This encompasses type I $(p_1 \geq 1, p_2 = 1)$, type II $(p_1=1, p_2\geq 1)$, and genuinely mixed settings $(p_1,p_2\geq2)$ (Álvarez-Fernández et al., 2010).

The existence and uniqueness of solutions are guaranteed for "perfect" weight systems, notably Nikishin and M-Nikishin systems, for which all minors of the associated moment matrix are nonvanishing. In such systems, the moment matrix can be LU (Gauss-Borel) factorized, which provides an explicit algebraic construction of mixed-type polynomials and their duals.

A central feature is the determinant (tau-function) representation: all polynomial and kernel objects can be written as ratios of minors of the (possibly infinite) moment matrix, generalizing the scalar case.

2. LU Factorization, Recurrence, and Jacobi Structures

The moment matrix associated with mixed-type multiple orthogonality is given in block form

$$g_{i,j} = \int x^{k_1(i)+k_2(j)} w_{1,a_1(i)}(x) w_{2,a_2(j)}(x) d\mu(x),$$

where the mapping $i \rightarrow (a_1(i),k_1(i))$ arises from "compositions" defined by the weight structure (Álvarez-Fernández et al., 2010).

For perfect systems, all principal minors are nonzero, yielding a unique LU factorization $g = S^{-1} \bar{S}$, with $S$ lower unitriangular, $\bar{S}$ upper triangular. The columns of $S$ (respectively, $(\bar{S}^{-1})^\top$) encode the coefficients of type II (resp., type I) polynomials; their linear forms and duals.

The action of the shift operator (multiplication by $x$ in the monomial basis) translates, via this factorization, into a banded "snake" Jacobi-type matrix $J$ with bandwidth determined by the mixed multi-index structure. The recurrence

$$zA^{(\ell)}(z) = \sum_{s=-N_2}^{N_1} J_s^{(\ell)}A^{(\ell+s)}(z)$$

demonstrates that higher-degree mixed-type MOPs decompose into lower-degree polynomials with coefficients structured by the LU data. The shape and bandwidth of $J$ encode the combinatorial interleaving or "snaking" of degree assignments, inherited from the step-line or graded lexicographic multi-indexing (Mañas et al., 23 Jul 2025).

3. Christoffel-Darboux Kernel and Formulas

The Christoffel-Darboux (CD) kernel plays a fundamental role in projection, reproducing properties, and the computation of correlation functions. For mixed-type systems, the kernel is explicitly

$$K_{[\ell]}(x, y) = \sum_{k=0}^{\ell-1} Q_k^{(1)}(y) Q_k^{(2)}(x)$$

where $Q_k^{(1)}$ and $Q_k^{(2)}$ are associated biorthogonal linear forms (Álvarez-Fernández et al., 2010, Ariznabarreta et al., 2013).

Alternative explicit CD formulas are available, constructed purely in terms of the LU factorization and the generalized Jacobi matrix $J$: $$(y-x)K^{[l]}(x, y) = \sum_{(i,j)\in \mathscr{O}_1^{[l]}} Q^{[l]}(x)^{(j)} J_{j,i} Q^{[l]}(y)^{(i)} -\sum_{(i,j)\in \mathscr{O}_2^{[l]}} Q^{[l]}(x)^{(j)} J_{j,i} Q^{[l]}(y)^{(i)},$$ where the index sets $\mathscr{O}_1^{[l]}$, $\mathscr{O}_2^{[l]}$ capture the precise structure of the moment matrix and measure composition (Ariznabarreta et al., 2013).

These formulas generalize the scalar Christoffel-Darboux identity and provide explicit, finite-sum representations of the kernel for computational and theoretical purposes. They make the relationship between the recurrence (Jacobi matrix) and the projection kernel manifest.

4. Riemann-Hilbert Problem for Mixed-Type Multiple Orthogonality

The Riemann-Hilbert framework provides a robust analytic characterization of mixed-type multiple orthogonal polynomials. In general, the associated RH problem is a matrix factorization problem: seek a matrix-valued function (size depending on the number of weights and the mixed structure) analytic off the support of the underlying measures, subject to prescribed jumps (reflecting the distribution and weights), and prescribed asymptotics at infinity (encoding the degree structure of the polynomials).

Explicit examples include:

• For planar orthogonal polynomials with weight $W(z) = \prod_{j=1}^p(z-a_j)^{c_j}$ (with all $c_j\in\mathbb{N}$), the type I perspective yields a $(p+2)\times(p+2)$ RH problem with a block jump matrix reflecting the weights $W(z)$ and exponential functions $e^{\overline{a_j}z}$, together with explicit asymptotics tied to the degree of $P_n$ and the exponents $c_j$ (Berezin et al., 2022).
• For Bessel-type mixed orthogonality, a $4\times4$ RHP is formulated whose solution encodes the linear forms and Cauchy transforms, with the correlation kernel assembled as a block of suitable products and integrals of this solution (Zhang, 2016).
• In the context of random matrix theory, RH problems of size $(r+1)\times(r+1)$ or larger naturally encode correlation kernels and recurrence, with the jump structure mirroring the weights and mixed-type structure (Kuijlaars, 2010, Filipuk et al., 2012).
• For finite-band step-line bivariate problems, the block structure of the LU factorization and recurrence matrices closely mirrors that of the RH construction, even if the RH form is not always explicitly written (Mañas et al., 23 Jul 2025).

In all these cases, the main feature is that the orthogonality relations (possibly mixed, possibly on different supports/contours) dictate the structure of the jump matrices, and the polynomial/exponential growth at infinity encodes the degree assignments to each component of the mixed system.

5. Explicit Examples and Applications

Planar Orthogonal Polynomials

For the ensemble determined by $|W(z)|^2 e^{-|z|^2}$ over $\mathbb{C}$ with $W(z)$ as above and all exponents integer, the monic planar orthogonal polynomial $P_n(z)$ is both a type II and type I multiple orthogonal polynomial associated with $W(z)$ and $e^{\overline{a_j}z}$, obeying explicit Hermite-Padé type conditions. The RH problem for this system is explicit, with blocks determined by $W(z)$ and exponential functions; several directly equivalent RH problems are derived (Berezin et al., 2022).

Bessel-Function Mixed Weights

For systems $(\omega_{\mu,a}, \omega_{\mu+1,a})$, $(\rho_{\nu,b}, \rho_{\nu+1,b})$ formed by modified Bessel functions, all basic properties of the mixed-type MOPs—explicit expressions, determinantal representations, Mellin-Barnes integrals, five-term recursions, and a $4\x4$ RH problem—are available and can be linked directly to eigenvalue statistics of products of coupled random matrices (Zhang, 2016).

Multicomponent 2D Toda Hierarchies

For "perfect" systems of Nikishin or M-Nikishin type, the mixed-type orthogonality, the moment matrix LU factorization, and the Riemann-Hilbert structure establish a tau-function and integrable hierarchy perspective: all biorthogonal functions, Cauchy transforms, and projection kernels are encoded by minors of the moment matrix, and the associated flows correspond to deformations of the weights or orthogonality supports (Álvarez-Fernández et al., 2010).

6. Connections to Integrable Systems, Recursion, and Asymptotics

The algebraic and analytic structures outlined above naturally interface with integrable hierarchies (e.g., multicomponent 2D Toda, non-Abelian versions), as the moment matrix factorization and the recurrence "snake" Jacobi matrix correspond to Lax pairs, wave operators, and discrete flows (Darboux/Miwa shifts). There are associated tau-function (Wick/Pfaffian) representations, and the Christoffel-Darboux kernel realizes projection operators essential in integrable systems and random matrix theory (Álvarez-Fernández et al., 2010, Ariznabarreta et al., 2013, Mañas et al., 23 Jul 2025).

The Riemann-Hilbert problem enables a rigorous route to obtaining strong (uniform) asymptotics of the mixed-type MOPs, via the Deift-Zhou steepest descent method, even in settings featuring vector equilibrium problems, coupled supports, or interacting measures (Lagomasino et al., 2016, Berezin et al., 2022).

7. Overview Table: Structural Summary

Feature	Mixed-Type MOP Formulation	Riemann-Hilbert Description
Algebraic backbone	Moment matrix, LU factorization, perfectness	Jump/asymptotics reflect weights and degree structure
Recurrence	"Snake" banded Jacobi matrix; explicit structure	Encoded via expansions/commutators of the RH solution
Kernel	Explicit CD kernel (finite sum, matrix)	Block of the RH solution or its Cauchy transforms
Tau/determinant	Minors, Cramer’s rule, tau-functions	Tau-functions reflect RH determinants
Integrable structure	Flows by measure deformation, correspondence with Toda hierarchies	RH deformation theory, Lax pairs
Asymptotics and zeros	Embedded via moment minors, algebraic analysis	Steepest descent, global and local parametrices

8. Significance and Future Directions

The mixed-type multiple orthogonal polynomial/Riemann-Hilbert description provides a versatile and unifying structure for algebraic, analytic, and integrable aspects of polynomial systems with orthogonality distributed among several weights and/or of both type I and type II. The RH formalism renders these systems amenable to powerful analytic machinery, enables precise asymptotic analysis (including rigorous descriptions of spectral edge regimes and universality classes), and interacts fruitfully with deterministic and probabilistic models in random matrix theory and approximation theory.

Ongoing directions include extension to higher-dimensional (multivariate) mixed-type systems, systematic classification of possible recurrence matrix structures and their integrable implications, further unification with non-commutative and matrix-valued orthogonality frameworks, and the explicit RH asymptotic and kernel analysis for step-line and non-standard configurations. The algebraic-to-analytic bridge afforded by RH techniques is pivotal for understanding the fine structure and universal behavior in these advanced orthogonality settings.