An algebraic characterization of non-singular matrix semicircles

Abstract: Let $A_1, \ldots, A_r$ be Hermitian $n \times n$ matrices and $S = \sum A_i \otimes s_i$ the associated matrix semicircle, where $s_1, \ldots, s_r$ are free semicircular variables. We prove that the following are equivalent: (i) the matrix pencil $A = \sum A_i x_i$ is LR-semisimple (decomposes, up to left--right equivalence, as a direct sum of unsplittable pencils); (ii) $S$ is non-singular at $t = 0$ (the matrix-valued Cauchy transform has a continuous boundary limit near the origin); (iii) the covariance map $η\colon X \mapsto \sum A_i X A_i$ is symmetrically DS-scalable (there exists $C \succ 0$ with $η(C) = C<sup>{-1}$). When these hold, the spectral density satisfies $f(0) = \frac{1}π\,\mathrm{tr}(C)$, where $C$ is the unique trace minimizer of the solution set ${W \succ 0 : η(W)\,W = I}$. The proof combines algebraic and analytic ingredients. On the algebraic side, we establish the equivalence (i) $\Leftrightarrow$ (iii) using Gurvits' capacity theory for indecomposable maps and a geodesic reflection theorem in the Riemannian manifold of positive definite matrices, which upgrades DS-scalability to symmetric DS-scalability for self-adjoint completely positive maps. On the analytic side, we prove (iii) $\Rightarrow$ (ii) via a Lyapunov--Schmidt reduction of Speicher's equation at a trace-minimizing solution, showing that the Jacobian of the bifurcation equations is positive definite. This removes a stability hypothesis that was required in earlier approaches.

• The paper introduces LR-semisimplicity as a key criterion for ensuring non-singularity in the spectral distribution of matrix semicircles.
• It employs a blend of algebraic, operator theoretic, and analytic techniques to establish equivalence with symmetric DS-scalability of completely positive maps.
• Quantitative results provide sharp lower bounds on the spectral density at the origin and classify singular behaviors for different matrix pencil structures.

Algebraic and Analytic Characterization of Non-Singular Matrix Semicircles

Overview

This paper, "An algebraic characterization of non-singular matrix semicircles" (2604.23089), presents a comprehensive and rigorous analysis of the algebraic conditions underpinning the non-singularity of the spectral distribution for matrix-valued semicircular elements constructed from Hermitian matrix pencils. It introduces the pivotal notion of LR-semisimplicity for matrix pencils and establishes its equivalence with analytic non-singularity at the spectral origin and the symmetric DS-scalability of an associated completely positive map. The work synthesizes tools from algebra, operator theory, free probability, and geometric invariant theory, yielding substantial structural insights and concrete quantitative results for the spectral analysis of matrix semicircles.

Main Results and Equivalences

The study considers matrix semicircular variables of the form

$S = \sum_{i=1}^r A_i \otimes s_i$

with $A_i$ Hermitian $n \times n$ matrices and $s_i$ freely independent semicircular elements. The spectral distribution of $S$ is determined both by the algebraic structure of the pencil

$A(x) = \sum_{i=1}^r A_i x_i$

and by the properties of the associated covariance map

$\eta(X) = \sum_{i=1}^r A_i X A_i.$

The central theorem states the equivalence of three fundamental properties for Hermitian matrix pencils:

1. LR-semisimplicity: The pencil can be block-diagonalized into a direct sum of unsplittable pencils, up to left–right equivalence.
2. Non-singularity at $t=0$: The associated matrix-valued Cauchy transform has a continuous boundary limit near the origin, ensuring the spectral measure has a real-analytic, bounded density at $x=0$.
3. Symmetric DS-scalability: There exists $C \succ 0$ such that $A_i$0, conferring “scalability” to a doubly stochastic, self-adjoint, completely positive map.

Moreover, under these conditions, the spectral density at the origin is

$A_i$1

where $A_i$2 is uniquely characterized as the trace-minimizing positive definite solution to $A_i$3. The paper further provides strict lower bounds on $A_i$4 connected to map capacity and Fuglede–Kadison determinant.

A sharp hierarchy is developed:

• Unsplittable $A_i$5 LR-semisimple $A_i$6 Full
• Strict inclusions hold, e.g., there exist full pencils that are not LR-semisimple.

Methods and Technical Contributions

Algebraic and Operator-Theoretic Connections

The proof of equivalence between LR-semisimplicity and symmetric DS-scalability leverages Gurvits' capacity theory for completely positive, indecomposable maps. The work establishes a "pencil–map dictionary," showing that LR-semisimplicity of the pencil maps directly to the symmmetric DS-scalability of the associated self-adjoint CP map.

A notable geometric innovation is the demonstration that, for CP, self-adjoint, DS-scalable maps $A_i$7, the operator $A_i$8 acts as a geodesic reflection on the manifold of positive definite matrices, ensuring that any scaling can be symmetrized at a unique geodesic midpoint, thus producing the required symmetric scaling.

Analytic Characterization

For the implication from symmetric DS-scalability to non-singularity, the analysis centers on Speicher’s equation for the Cauchy transform and its reformulation on the right half-plane. The use of Lyapunov–Schmidt reduction at trace-minimizing solutions and the verification that the reduced equations have positive-definite Jacobian matrices permits the deployment of the implicit function theorem. This shows the existence and local analyticity of the matrix-valued Cauchy transform boundary limit and secures the continuity and positivity of the density at the spectral origin.

This approach also eliminates a stability hypothesis needed in previous approaches in the literature (e.g., in [msy2023]), relying instead on structures inherent to self-adjoint CP maps.

Converse Implication

Conversely, the existence of a continuous density at the origin is shown to yield a positive definite solution to $A_i$9, completing the cycle of equivalences.

Quantitative and Structural Results

The main theorem not only provides existential results but also strong quantitative lower bounds:

$n \times n$0

where $n \times n$1 is the capacity and $n \times n$2 is the Fuglede–Kadison determinant of $n \times n$3.

For unsplittable pencils (indecomposable maps), the scaling solution is uniquely determined, strengthening the characterization in this special case.

Singularities Beyond LR-semisimplicity

The paper analyzes situations where the pencil is full but not LR-semisimple, showing through explicit example that the density at the origin may exhibit integrable cusp singularities (e.g., $n \times n$4), in contrast to the bounded analytic density for LR-semisimple pencils. The work conjectures that in the non-LR-semisimple case, all singularities at the origin are algebraic and integrable, controlled by the pencil structure.

Implications, Applications, and Theoretical Context

• Practical: The results provide a complete and checkable algebraic criterion for the regularity of the spectral distribution of matrix semicircles, with strong numerical bounds, directly impacting asymptotic spectral analysis in random matrix theory and free probability.
• Theoretical: The equivalence highlights deep connections between non-commutative probability, operator algebra, optimization (capacity minimization), and geometric invariant theory. The work informs ongoing research in scaling algorithms (operator scaling), quantum information theory, and non-commutative optimization [ggow2020].
• Quantum Theory: In quantum channels, symmetric DS-scalability relates to the possibility of transforming CP maps to trace-preserving, unital forms—a key property for quantum state manipulation.

Future directions suggested include:

• Classification of possible singularity exponents for full, non-LR-semisimple pencils.
• Extension to biased matrix semicircles (including non-centered spectral analysis).
• Generalization to non-self-adjoint pencils and the Brown measure, requiring new analytic techniques.

Conclusion

This paper establishes that the non-singularity of matrix-valued semicircular distributions at the origin is controlled precisely by LR-semisimplicity—an algebraic property of the matrix pencil—equivalent to both analytic regularity of the Cauchy transform and the existence of a symmetric DS-scaling of the covariance map. The rigorous hierarchy delineated and the strong quantitative results provide new tools for spectral analysis, with significant implications for operator algebras, random matrices, and non-commutative probability theory.