Reading Weakly, Acting Strongly: A Static Parity Horizon and its Dynamical Bypass in the Monitored Lipkin-Meshkov-Glick Model

Technical Summary: Reading Weakly, Acting Strongly in the Monitored Lipkin–Meshkov–Glick Model

Problem Statement
The paper investigates the broken-symmetry phase of the Lipkin–Meshkov–Glick (LMG) model, a system of $N$ collective spins exhibiting a $Z_2$ symmetry. In this phase, the two lowest energy eigenstates, $|E_0\rangle$ (even parity) and $|E_1\rangle$ (odd parity), form a near-degenerate doublet separated by an exponentially small energy gap $\Delta E \sim e^{-N S_{\text{inst}}}$. The central problem is to understand the information-theoretic relationship between a bath monitoring the system via the collective magnetization operator $\hat{J}_z$ and the system's conserved parity label.

Specifically, the authors address a mismatch: while the $\hat{J}_z$ operator measures magnetization (distinguishing the two macroscopic "wells" of the broken symmetry), it is parity-odd ($\{\hat{P}, \hat{J}_z\} = 0$). Consequently, the static $\hat{J}_z$ readout is expected to be "blind" to the relative sign between the two wells (the parity label). The paper asks: How weak is this static readout of the parity label? Does the same operator that reads parity weakly also act weakly on the system (backaction), or does a separation exist where the bath reads weakly but acts strongly? Furthermore, can continuous time-resolved monitoring bypass the limitations of a static snapshot?

Methodology
The study employs a multi-faceted approach combining analytical theory, exact diagonalization (ED), and stochastic simulations:

1. Analytical Framework (WKB & Instanton Theory): The authors use Wentzel–Kramers–Brillouin (WKB) path-additivity arguments to analyze the tunneling barrier between the two magnetization wells. They derive the instanton action $S_{\text{inst}}$ governing the spectral gap and extend this analysis to nonlinear information measures (Jensen–Shannon divergence, Chernoff information) to determine their asymptotic scaling.
2. Static Channel Benchmark: A "static $\hat{J}_z$-readout channel" is defined by setting the system Hamiltonian $\hat{H}=0$. In this idealized limit, the bath records are commuting and diagonal in the $\hat{J}_z$ basis. The authors prove that the accessible classical information (Holevo quantity) in this channel reduces exactly to the Jensen–Shannon divergence (JSD) of the parity-conditioned $\hat{J}_z$ histograms.
3. Exact Diagonalization (ED): Numerical calculations are performed on grids extending to $N = 4500$ to verify the scaling of the spectral gap, total variation distance, JSD, and Chernoff information. The study also computes the off-diagonal matrix element $J_{01} = \langle E_0 | \hat{J}_z | E_1 \rangle$ and the variance budget to characterize backaction.
4. Continuous Monitoring Simulations: The full monitored dynamics (with $\hat{H} \neq 0$) are simulated using stochastic master equations (SME) for 1.48 million trajectories. A Girsanov-optimal record classifier is used to distinguish between parity hypotheses based on continuous homodyne records, comparing the performance against the static benchmark.

Key Contributions and Results

1. The Static Parity Horizon (Readout Side)
The paper establishes that the static $\hat{J}_z$ readout of the parity label is exponentially suppressed.

• Exponent Equality: Contrary to a naive small-deviation expansion which would predict a decay of $e^{-2N S_{\text{inst}}}$ for nonlinear measures, the authors demonstrate that the spectral gap, total variation distance, JSD, and Chernoff information all share a single leading exponent: $S_{\text{inst}}$.
• Mechanism: This single exponent arises because the parity information is concentrated in the exponentially thin tunneling barrier and the exact parity-forbidden central node ($m=0$). For even $N$, the odd-parity state has a strict node at $m=0$ ($p^{(1)}_0 = 0$), while the even-parity state has a non-zero value $p^{(0)}_0 \sim e^{-N S_{\text{inst}}}$. This node anchors the JSD and Chernoff information at the instanton scale.
• Proposition 1: The accessible information in the static channel is exactly the JSD of the two parity-conditioned distributions, $\chi^{\text{static}}_\infty = \text{JSD}(p^{(0)}, p^{(1)}) \sim e^{-N S_{\text{inst}}}$.

2. The Backaction Asymmetry (Acting Side)
While the readout is exponentially weak, the backaction is macroscopically strong.

• Matrix Element Scaling: The off-diagonal matrix element $J_{01} = \langle E_0 | \hat{J}_z | E_1 \rangle$ scales as $N m^*/2$ (where $m^*$ is the order parameter), growing linearly with $N$.
• Backaction Rate: Consequently, the rate at which the bath disturbs the parity population, $\Gamma^{(\text{par})}_z \propto \gamma_\phi |J_{01}|^2$, scales as $N^2$.
• The Parity Horizon: This creates a "static parity horizon": the bath reads the parity label exponentially weakly (via the static histogram) but acts on it extensively strongly (via the transverse matrix element). This is not a failure of pointer-state theory but a consequence of the basis mismatch between the monitored coordinate ($\hat{J}_z$) and the conserved label (Parity).

3. The Dynamical Bypass
The paper investigates whether continuous monitoring can recover the parity information lost in the static snapshot.

• Finite-N Excess: Simulations reveal a "dynamical bypass." In a finite window of system sizes (mesoscopic regime), a time-resolved homodyne record extracts parity information that exceeds the static benchmark.
• Control Parameter: This bypass is organized by the ratio $\xi = \omega_{01} / \Gamma_{01}$, where $\omega_{01}$ is the coherent doublet rotation frequency and $\Gamma_{01}$ is the measurement-induced dephasing rate.
• Regimes:
  • Secular ($\xi \gg 1$): Coherent rotation averages out the signal; no bypass.
  • Information Window ($\xi \sim 1$): The continuous record shows a statistically resolved excess ($\Delta P \approx +0.015$) over the static benchmark.
  • Zeno ($\xi \ll 1$): Strong measurement freezes the dynamics; the excess vanishes, and the system returns to the static benchmark.
• No Universal Law: The authors do not claim a universal power-law closure for the bypass; the closure is a robust qualitative fact controlled by the instanton action via $\omega_{01}$.

Significance and Claims
The paper claims to identify a specific structural mechanism in open quantum systems where a single operator plays two inequivalent roles: a weak readout of a symmetry label and a strong backaction on that same label.

• Not a Horizon Violation: The authors explicitly state that the dynamical bypass is not a violation of the static parity horizon in the asymptotic limit. The information window is a finite-$N$ prefactor effect. The asymptotic exponent of the full monitored channel remains $S_{\text{inst}}$; the bypass is a mesoscopic advantage that closes as $N \to \infty$ or in the deep Zeno regime.
• Distinction from Error Correction: The parity doublet is not a decoherence-free subspace or a noiseless subsystem. The strong backaction ($\sigma_x$-like noise in the parity basis) actively scrambles the parity bit, even though the static readout cannot easily detect it.
• Operational Insight: The work provides a benchmark for the idealized static channel and demonstrates that time-resolved monitoring can exploit temporal correlations to outperform single-shot static measurements in specific regimes, without overturning the fundamental asymptotic limits imposed by the instanton action.

The paper concludes that the "static parity horizon" is a visibility statement: the quantum label exists and is conserved, but the static $\hat{J}_z$ record is almost blind to it, while the same coupling acts strongly to disturb it. This separation is a generic feature of symmetry-broken systems monitored along the order parameter.