---
title: "Noncompact gauge, disordered Bell states, Gaussian ptychography"
subtitle: "SU(1,1) optics, quantum correlations from disorder, and primitive-based wavefield reconstruction"
issue: 20
date: 2026-09-21
author: PhaseTransistor
publication: Coherence Review
canonical: https://coherencereview.com/issues/issue-20.html
framework: https://coherencereview.com/lfsc.html
---

# Coherence Review #20: Noncompact gauge, disordered Bell states, Gaussian ptychography

*SU(1,1) optics, quantum correlations from disorder, and primitive-based wavefield reconstruction*

> Published 2026-09-21 by PhaseTransistor. Canonical page: https://coherencereview.com/issues/issue-20.html
> Levels and tiers follow the LFSC framework: https://coherencereview.com/lfsc.html

## This week in three levels

**L1 · tier P · Noncompact SU(1,1) gauge field in non-Hermitian optics drives Lorentzian pseudospin precession and pseudo-intensity-dependent spatial trajectories.**

Xiao et al. extend synthetic gauge control to the noncompact group SU(1,1) in a non-Hermitian optical medium [Synthetic SU(1,1) Gauge Control of Non-Hermitian Light — arXiv:2609.15262]. Paraxial light evolves pseudo-unitarily and conserves a generalized pseudo-intensity despite local gain or loss. The gauge field drives Lorentzian precession of an optical pseudospin on a Bloch hyperboloid—rotations, boosts, and exceptional-point critical dynamics in an internal Minkowski space. A non-Abelian Lorentz force then converts these internal dynamics into real-space centroid trajectories: oscillatory, hyperbolic, and cubic, even in uniform media. Light of opposite pseudo-intensity can drift in opposite directions, a non-Hermitian analogue of the spin Hall effect. The work surfaces a conserved quantity hidden in conventional intensity formalisms and links non-Hermitian wave physics to noncompact Lie algebras.

**L1 · tier T · Disordered emitter ensembles produce transmitted light that violates Bell inequalities in a continuous-wave Franson measurement via antibunched or bunched photon-pair regimes.**

Guan et al. predict Bell-inequality violation in light transmitted through disordered two-level emitters under weak coupling [Bell-inequality violation in light transmitted through disordered emitter ensembles — arXiv:2609.15575]. The Bell signal derives from steady-state correlations: equal-time intensity correlation g⁽²⁾(0) and phase-sensitive two-photon coherence R. Fourth-order cumulant expansion of the many-body master equation reveals violation in two regimes. In the antibunched regime, equal-time coincidences are suppressed; in the bunched regime, photon pairs retain phase coherence. Numerical extrapolation at single-emitter coupling β = 0.01 predicts violation for 75–220 atoms (antibunched) or above 250 atoms (bunched). The result is theoretical; experimental realization would show quantum correlations emerging from disorder without engineered entanglement sources.

**L1 · tier P · Gaussian-primitive ptychography represents object and probe as relocatable primitives, recovering low-frequency phase uniformly and cutting memory up to 14-fold.**

Zhao et al. replace pixel-grid reconstruction in ptychography with overlapping Gaussian primitives [Gaussian-splatting ptychography via explicit and interpretable primitives — arXiv:2609.14891]. Primitives relocate to concentrate where structure is dense, and overlapping primitives couple neighboring regions to suppress mode drift. The scheme unexpectedly restores low-frequency phase that conventional pixel-grid solvers lose, enabling uniform phase transfer across all spatial frequencies. The probe is represented and updated in its pupil plane from a random initialization. In Fourier ptychography, the pupil-plane model recovers severe aberrations where pixel-grid solvers fail. In conventional optical, X-ray, and electron ptychography, pupil-plane primitives recover real-space probes without a model of beam-forming optics. At electron wavelengths, the method resolves atomic structure at tens of electrons per square angstrom and reduces data requirements.

## Bridge watch

No strong bridge candidate this week. All twelve ingest entries cluster at level 1 (field shaping). The distribution reflects genuine thematic concentration in the recent literature rather than a classification artifact.

## Falsification watch

No movement on F1–F5 this week. The ingest consists entirely of L1 field-shaping advances in photonics, optics, and atomic sensing. None bear on scalability limits (F1), vacuum transduction (F2), metric engineering substrates (F3), inertial invariance (F4), or cross-level structural connections (F5).

## Catalog movement

No changes this week.
