arxiv: 2604.23089 · v1 · submitted 2026-04-25 · 🧮 math.OA · math.PR· math.SP

Recognition: unknown

An algebraic characterization of non-singular matrix semicircles

Vladislav Kargin

Pith reviewed 2026-05-08 07:07 UTC · model grok-4.3

classification 🧮 math.OA math.PRmath.SP

keywords matrix semicircleLR-semisimple pencilDS-scalable mapcovariance mapfree semicircularspectral densitymatrix pencilCauchy transform

0 comments

The pith

Non-singularity of a matrix semicircle at the origin is equivalent to the generating matrix pencil being LR-semisimple and to the covariance map being symmetrically DS-scalable.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proves three statements are equivalent for Hermitian matrices A1 to Ar and the associated matrix semicircle S built from free semicircular variables: the pencil A equals sum Ai xi is LR-semisimple, meaning it decomposes up to left-right equivalence into a direct sum of unsplittable pencils; S is non-singular at t=0, so its matrix-valued Cauchy transform has a continuous boundary limit near the origin; and the covariance map η sending X to sum Ai X Ai is symmetrically DS-scalable, meaning some positive definite C satisfies η(C) equals C inverse. When they hold, the spectral density satisfies f(0) equals one over pi times the trace of the unique trace-minimizing solution C to the equation η(W) W equals the identity. A reader cares because the equivalences give an algebraic test for when the random-matrix model avoids a singularity at zero and supply an explicit formula for the density value there.

Core claim

The matrix pencil A is LR-semisimple if and only if the matrix semicircle S is non-singular at t=0 if and only if the covariance map η is symmetrically DS-scalable. When these conditions hold, the spectral density f(0) equals (1/π) tr(C), where C is the unique trace minimizer of the positive definite solutions to η(W) W = I.

What carries the argument

Symmetric DS-scalability of the covariance map η: X ↦ sum Ai X Ai, shown equivalent to LR-semisimple decomposition of the pencil via Gurvits capacity theory and geodesic reflection in the positive definite matrices.

If this is right

The value of the spectral density at zero is given explicitly by f(0) = (1/π) tr(C) for the trace-minimizing C.
The matrix semicircle is non-singular at zero exactly when the pencil decomposes as a direct sum of unsplittable pencils up to left-right equivalence.
The Lyapunov-Schmidt reduction at the trace-minimizing solution has positive definite Jacobian, so earlier stability hypotheses are unnecessary.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

For concrete matrices one can solve numerically for a positive definite W satisfying η(W) W = I and check whether its trace is minimal to decide non-singularity without computing the full Cauchy transform.
The same capacity-plus-reduction argument might apply to other free random-matrix ensembles whose covariance maps satisfy similar positivity and self-adjointness conditions.

Load-bearing premise

The equation η(W) W = I admits a unique trace-minimizing positive definite solution C at which the Jacobian of the reduced bifurcation equations is positive definite.

What would settle it

Take any matrix pencil that cannot be decomposed into unsplittable summands up to left-right equivalence; if its associated matrix semicircle still possesses a continuous boundary limit for the Cauchy transform at the origin, the claimed equivalence is false.

Figures

Figures reproduced from arXiv: 2604.23089 by Vladislav Kargin.

**Figure 1.** Figure 1: Spectral density of the matrix semicircle S = A1 ⊗s1 + A2 ⊗ s2 for the non-LR-semisimple pencil of Example 2.17. The density has an integrable |x| −1/3 cusp at the origin. The following example shows that the singularity can take the form of an integrable algebraic cusp. Example 6.2 (Singularity for a non-LR-semisimple pencil). Consider the pencil from Example 2.17: A1 = view at source ↗

read the original 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 $\eta\colon X \mapsto \sum A_i X A_i$ is symmetrically DS-scalable (there exists $C \succ 0$ with $\eta(C) = C^{-1}$). When these hold, the spectral density satisfies $f(0) = \frac{1}{\pi}\,\mathrm{tr}(C)$, where $C$ is the unique trace minimizer of the solution set $\{W \succ 0 : \eta(W)\,W = I\}$. 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.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a new equivalence between LR-semisimplicity, symmetric DS-scalability, and non-singularity at zero for matrix semicircles, dropping the old stability assumption, but the Jacobian positivity claim in the Lyapunov-Schmidt step lacks clear independent support.

read the letter

The main result is an equivalence: the pencil A is LR-semisimple exactly when the covariance map eta is symmetrically DS-scalable exactly when the matrix semicircle S is non-singular at t=0. In that case the density at zero equals one over pi times the trace of the unique trace-minimizer C solving eta(W)W = I. This removes the stability hypothesis from earlier work and ties the algebraic decomposition property directly to the analytic boundary behavior. The algebraic half uses Gurvits capacity plus a geodesic reflection argument on positive definite matrices to upgrade ordinary scalability to the symmetric version; that direction looks clean and rests on established tools. The analytic half reduces Speicher's equation by Lyapunov-Schmidt at C and claims the reduced Jacobian is positive definite because C minimizes the trace. The outline does not supply a separate calculation showing why the first-order minimality condition produces the needed second-order positivity, so that step is the one that still needs explicit verification. The paper is aimed at specialists in free probability and operator algebras who want algebraic tests for regularity of these models. A reader working on random matrix singularities or completely positive maps will get a concrete criterion and a proof strategy worth checking. It deserves peer review because the claim is precise, the combination of methods is non-routine, and the potential gap is localized rather than fatal to the whole argument.

Referee Report

1 major / 0 minor

Summary. The paper proves the equivalence of three statements for Hermitian matrices A_1,...,A_r and the associated matrix semicircle S = sum A_i ⊗ s_i: (i) the pencil A = sum A_i x_i is LR-semisimple (decomposes up to left-right equivalence into unsplittable pencils); (ii) S is non-singular at t=0 (matrix-valued Cauchy transform has continuous boundary values near the origin); (iii) the covariance map η(X) = sum A_i X A_i is symmetrically DS-scalable (exists C ≻ 0 with η(C) = C^{-1}). When these hold, the spectral density satisfies f(0) = (1/π) tr(C) with C the unique trace minimizer of {W ≻ 0 : η(W)W = I}. The proof splits into an algebraic part using Gurvits capacity and geodesic reflection on positive-definite matrices to equate (i) and (iii), and an analytic part using Lyapunov-Schmidt reduction of Speicher's equation at the trace minimizer to obtain (iii) ⇒ (ii) while removing a prior stability hypothesis.

Significance. If the equivalences hold, the result supplies an algebraic criterion for non-singularity of matrix semicircles, connecting free probability, completely positive maps, and algebraic geometry. The removal of the stability assumption broadens the scope relative to earlier work, and the explicit formula for f(0) in terms of the trace minimizer C is a concrete payoff. The combination of Gurvits theory with Riemannian geometry on positive-definite matrices is a notable technical bridge.

major comments (1)

[analytic part / Lyapunov-Schmidt reduction] § on analytic side, Lyapunov-Schmidt reduction of Speicher's equation: the claim that the Jacobian of the reduced bifurcation equations is positive definite at the asserted unique trace minimizer C is invoked to conclude non-singularity at t=0, yet the manuscript supplies no separate argument or explicit computation showing that the first-order minimality condition (vanishing variation of tr(W) under the constraint η(W)W = I) forces the required second-order positivity on the reduced Jacobian. This step is load-bearing for (iii) ⇒ (ii) and for the removal of the stability hypothesis.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful reading of the manuscript and for highlighting the need for greater explicitness in the analytic portion of the argument. The observation identifies a genuine gap in the presentation of the Lyapunov–Schmidt reduction, and we will strengthen the manuscript accordingly.

read point-by-point responses

Referee: [analytic part / Lyapunov-Schmidt reduction] § on analytic side, Lyapunov-Schmidt reduction of Speicher's equation: the claim that the Jacobian of the reduced bifurcation equations is positive definite at the asserted unique trace minimizer C is invoked to conclude non-singularity at t=0, yet the manuscript supplies no separate argument or explicit computation showing that the first-order minimality condition (vanishing variation of tr(W) under the constraint η(W)W = I) forces the required second-order positivity on the reduced Jacobian. This step is load-bearing for (iii) ⇒ (ii) and for the removal of the stability hypothesis.

Authors: We agree that the manuscript does not supply a separate, self-contained argument establishing that the first-order stationarity condition for the trace minimizer C implies positive-definiteness of the reduced Jacobian. In the revised version we will insert a dedicated lemma (placed immediately after the statement of the trace-minimizing property) that computes the second variation of tr(W) on the constraint manifold η(W)W = I and verifies that this quadratic form is positive definite precisely when C is the unique minimizer. The argument relies only on the self-adjointness of η and the strict convexity of the trace functional; it does not invoke any additional stability hypothesis. This addition will make the passage from (iii) to (ii) fully rigorous and will clarify how the stability assumption of earlier work is removed. revision: yes

Circularity Check

0 steps flagged

No significant circularity in the algebraic-analytic equivalence chain.

full rationale

The paper establishes the three-way equivalence via Gurvits capacity theory plus a geodesic reflection theorem on positive-definite matrices for (i)⇔(iii), followed by a Lyapunov-Schmidt reduction of Speicher's equation that asserts (without reducing to a fitted parameter or self-definition) positive-definiteness of the reduced Jacobian at the trace-minimizer C of {W≻0:η(W)W=I}. C is introduced as the unique trace minimizer and the density formula is stated as a consequence; neither step collapses to an input by construction, nor does any load-bearing premise rest on a self-citation chain. External theorems supply independent content, so the derivation is self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 0 invented entities

The central claim rests on standard results in free probability and operator algebras plus two invoked theorems; no data-fitted parameters or new postulated entities appear.

axioms (4)

domain assumption The matrices A_i are Hermitian n by n matrices
Stated in the setup for the matrix semicircle S and the pencil A.
standard math Gurvits' capacity theory applies to indecomposable completely positive maps
Invoked for the algebraic equivalence (i) equivalent to (iii).
standard math The geodesic reflection theorem holds in the Riemannian manifold of positive definite matrices
Used to upgrade DS-scalability to symmetric DS-scalability.
ad hoc to paper The Jacobian of the bifurcation equations is positive definite at the trace-minimizing solution
Central to the Lyapunov-Schmidt reduction showing (iii) implies (ii).

pith-pipeline@v0.9.0 · 5641 in / 1763 out tokens · 50387 ms · 2026-05-08T07:07:31.193956+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

12 extracted references · 9 canonical work pages

[1]

doi: 10.4171/dm/780. Greg W. Anderson. Preservation of algebraicity in free probability. Preprint, arXiv:1406.6664,

work page doi:10.4171/dm/780
[2]

Alternating minimization, scaling algorithms, and the null- cone problem from invariant theory

Peter B¨ urgisser, Ankit Garg, Rafael Oliveira, Michael Walter, and Avi Wigderson. Alternating minimization, scaling algorithms, and the null- cone problem from invariant theory. In9th Innovations in Theoret- ical Computer Science (ITCS 2018), volume 94 ofLIPIcs. Schloss Dagstuhl – Leibniz-Zentrum f¨ ur Informatik,

2018
[3]

Peter B¨ urgisser, Cole Franks, Ankit Garg, Rafael Oliveira, Michael Wal- ter, and Avi Wigderson

arXiv:1711.08039. Peter B¨ urgisser, Cole Franks, Ankit Garg, Rafael Oliveira, Michael Wal- ter, and Avi Wigderson. Towards a theory of non-commutative opti- mization: geodesic first and second order methods for moment maps and polytopes. In60th Annual IEEE Symposium on Foundations of Computer Science (FOCS 2019), pages 845–861. IEEE Computer Soc.,

work page arXiv 2019
[4]

Man-Duen Choi

arXiv:1910.12375. Man-Duen Choi. A Schwarz inequality for positive linear maps onc ∗- algebras.Illinois Journal of Mathematics, 18:565–574,

work page arXiv 1910
[5]

Paul Moritz Cohn.Free rings and their relations

doi: 10.1215/ijm/1256051007. Paul Moritz Cohn.Free rings and their relations. Academic Press, second edition,

work page doi:10.1215/ijm/1256051007
[6]

Rashidi Far, T

R. Rashidi Far, T. Oraby, W. Bryc, and R. Speicher. Spectra of large block matrices. https://arxiv.org/abs/cs/0610045,

work page arXiv
[7]

William Helton, Reza Rashidi Far, and Roland Speicher

J. William Helton, Reza Rashidi Far, and Roland Speicher. Operator-valued semicircular elements: Solving a quadratic matrix equation with positivity constraints.International Mathematics Research Notices, 2007,

2007
[8]

Tobias Mai and Roland Speicher

doi: 10.1007/ s10208-024-09684-5. Tobias Mai and Roland Speicher. Fuglede–Kadison determinants of matrix-valued semicircular elements and capacity estimates. https://arxiv.org/abs/2406.15922,

work page arXiv
[9]

doi: 10.1016/j.jfa.2023.110016. J. A. Mingo and R. Speicher.Free Probability and Random Matrices. Schwinger-Verlag,

work page doi:10.1016/j.jfa.2023.110016 2023
[10]

doi: 10.1090/S0002-9947-2015-06434-4. R. Speicher.Combinatorial Theory of the Free Product with Amal- gamation and Operator-Valued Free Probability Theory, volume 627 ofMemoirs of American Mathematical Society. A.M.S. Prov- idence, RI,

work page doi:10.1090/s0002-9947-2015-06434-4 2015
[11]

Non-commutative distributions

Roland Speicher. Non-commutative distributions. lecture notes. https://arxiv.org/abs/2009.03589,

work page arXiv 2009
[12]

Cambridge University Press, 2018

John Watrous.The Theory of Quantum Information. Cambridge University Press, 2018

2018