Real-Part Isolation in Weakly Non-Hermitian Real Elliptic Ginibre Matrices: A Pfaffian Local-Law Proof and an Explicit Expected-Count Bound

We study real-part isolation in the weakly non-Hermitian real elliptic Ginibre ensemble of even matrix dimension N, with ellipticity parameter tau_N = 1 - alpha^2/N. A real eigenvalue is counted as isolated when its globally semicircle-unfolded coordinate differs by more than a prescribed threshold from the real-part coordinates of every other eigenvalue, including nonreal conjugate pairs. This is stronger than separation from other real eigenvalues alone. We establish a deterministic single-charge counting inequality for arbitrary finite conjugation-symmetric spectra. After removing nearby real-real pairs, each nonreal conjugate pair can obstruct at most one surviving real candidate. This eliminates the need for real-complex mixed correlation functions in the expected-count lower bound. We combine the inequality with the finite-N Pfaffian kernel, a Hermite-projector/Abel-mixture local-law argument, and an explicit short-distance real-real correlation bound. Campbell integration over a protected bulk interval yields, uniformly over 0 <= alpha <= 0.03 and along even N tending to infinity, liminf inf E[I_N(0.03)] / N >= 1 - 3(0.03)^2/8 - (2 pi^2/3)(0.03)^2 > 0.99374. The threshold 0.03 is measured in the paper's deterministic global semicircle unfolding. The result is an asymptotic expected-count lower bound, not a concentration or high-probability statement. The uploaded source and Research/QA archives document the proof structure and finite symbolic/numerical consistency tests.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-15
DOI
https://doi.org/10.5281/zenodo.24241913
Primary Topic
Random Matrices and Applications
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Real-Part Isolation in Weakly Non-Hermitian Real Elliptic Ginibre Matrices: A Pfaffian Local-Law Proof and an Explicit Expected-Count Bound

Byoungwoo Lee
Zenodo (CERN European Organization for Nuclear Research)
Random Matrices and Applications
preprint

Real-Part Isolation in Weakly Non-Hermitian Real Elliptic Ginibre Matrices: A Pfaffian Local-Law Proof and an Explicit Expected-Count Bound

Byoungwoo Lee
preprint en

Abstract

We study real-part isolation in the weakly non-Hermitian real elliptic Ginibre ensemble of even matrix dimension N, with ellipticity parameter tau_N = 1 - alpha^2/N. A real eigenvalue is counted as isolated when its globally semicircle-unfolded coordinate differs by more than a prescribed threshold from the real-part coordinates of every other eigenvalue, including nonreal conjugate pairs. This is stronger than separation from other real eigenvalues alone. We establish a deterministic single-charge counting inequality for arbitrary finite conjugation-symmetric spectra. After removing nearby real-real pairs, each nonreal conjugate pair can obstruct at most one surviving real candidate. This eliminates the need for real-complex mixed correlation functions in the expected-count lower bound. We combine the inequality with the finite-N Pfaffian kernel, a Hermite-projector/Abel-mixture local-law argument, and an explicit short-distance real-real correlation bound. Campbell integration over a protected bulk interval yields, uniformly over 0 <= alpha <= 0.03 and along even N tending to infinity, liminf inf E[I_N(0.03)] / N >= 1 - 3(0.03)^2/8 - (2 pi^2/3)(0.03)^2 > 0.99374. The threshold 0.03 is measured in the paper's deterministic global semicircle unfolding. The result is an asymptotic expected-count lower bound, not a concentration or high-probability statement. The uploaded source and Research/QA archives document the proof structure and finite symbolic/numerical consistency tests.

Zenodo (CERN European Organization for Nuclear Research)
Random Matrices and Applications
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon