⚛️ quantum physics

Number Fluctuations and Entanglement-Spectrum Participation in Monitored Free Fermions

This study demonstrates that while both bipartite particle-number fluctuations and entanglement-spectrum participation track the weak-to-strong monitoring crossover in free-fermion chains, the latter offers no independent physical insight over the former but serves as a statistically more stable numerical estimator due to reduced trajectory-to-trajectory variance.

Original authors: Enso O. Torres Alegre

Published 2026-07-21

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Enso O. Torres Alegre

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a universe made of tiny, invisible dancers called fermions. In the quantum world, these dancers don't just move on their own; they can become "entangled," a spooky connection where the state of one dancer instantly influences another, no matter how far apart they are. Usually, if you have a big group of these dancers, the amount of entanglement grows with the size of the crowd. But what happens if you start watching them? If you peek at the dancers too often, you disturb their rhythm, and the entanglement shrinks. This is the heart of a field called "measurement-induced phase transitions." Scientists are trying to figure out exactly how much watching is needed to stop the dancers from connecting, and whether there's a specific "tipping point" where the behavior changes forever. It's a bit like trying to find the perfect amount of noise to make a party either a chaotic rave or a quiet library, but with the laws of physics as the DJ.

In this study, a researcher named Enso O. Torres Alegre decided to test two new ways to measure this "dance connection" in a simplified, simulated world of free fermions (dancers who don't bump into each other). The goal was to see if these new tools could spot the change from a chaotic, highly connected state to a quiet, disconnected one better than the old methods. The researcher used a computer to simulate chains of up to 96 sites, running thousands of different "what-if" scenarios (called trajectories) to see what happened when the measurement rate was tweaked.

The main finding is a bit of a "yes, but..." story. The researcher tested two new metrics: one called "number fluctuation" (which counts how much the number of dancers in a section wobbles) and another called "participation number" (which tries to count how many different dance moves are happening at once). Both of these new tools successfully tracked the transition from the chaotic state to the quiet state, just like the traditional "entanglement entropy" does. However, the researcher discovered something surprising: the "participation number" didn't actually tell us anything new. In fact, it turned out to be almost exactly the same thing as the "number fluctuation" in disguise. If you knew the number fluctuation, you could predict the participation number with 99.7% accuracy. It's like having a thermometer and a barometer that, in this specific weather system, always show the exact same relationship; checking the barometer doesn't give you any extra info about the storm.

So, why bother with the second tool at all? The paper found one practical, statistical advantage. While the two tools were saying the same thing, the "participation number" was much less "jittery." When the researchers ran the simulation over and over, the results for the participation number were smoother and more consistent than the results for the other tools. In the quiet, strongly monitored regime, it was about twice as stable. This means that if you are running these simulations on a computer with limited time, using this tool might save you from running as many extra tests to get a clear answer.

The study also looked closely at the "chaotic" side of things (weak monitoring). Some theories suggest that at low measurement rates, the entanglement should follow a perfect logarithmic rule (a specific mathematical curve). However, by comparing different system sizes (from 48 to 96 sites), the researcher found that the data didn't quite fit a perfect curve. Instead, there was a small, lingering "extra" bit of connection that made the curve drift. This suggests that what looks like a special phase might actually just be a long, slow transition, consistent with recent ideas that the logarithmic behavior isn't a permanent phase but a crossover.

In the end, the paper concludes that while the "participation number" is a nice, smooth way to summarize the data, it isn't a magic new key to unlock secrets the "number fluctuation" can't already tell us. It's a useful tool for reducing noise in computer simulations, but it doesn't reveal a hidden layer of physics in this specific model. The researcher also points out that this tool is hard to measure in real life because it requires knowing the exact details of every single quantum state, whereas the "number fluctuation" could potentially be measured just by counting particles. So, for now, the old tool remains the champion for experiments, while the new tool is a helpful, quieter sidekick for computer simulations.

Technical Summary: Number Fluctuations and Entanglement-Spectrum Participation in Monitored Free Fermions

Problem Statement
Monitored free-fermion chains exhibit a change in entanglement scaling as the measurement rate increases, transitioning from a sub-extensive regime at weak monitoring to an area law at strong monitoring. While the entanglement entropy ( $S$ ) is the standard diagnostic for this crossover, it is often difficult to access experimentally without post-selection or randomized measurement protocols. Bipartite particle-number fluctuations ( $F$ ) are a known proxy for entanglement in Gaussian systems. This paper investigates whether a second quantity, a participation-style "effective number of entangling modes" ( $M$ ), offers independent diagnostic value or statistical advantages over $F$ and $S$ in the context of monitored free fermions. Specifically, the author tests if $M$ captures independent structure of the single-particle entanglement spectrum $\{\nu_k\}$ or if it merely redundantly tracks $F$ .

Methodology
The study employs trajectory-resolved correlation-matrix simulations of spinless fermions on a ring of $L$ sites (up to $L=96$ ) with nearest-neighbor hopping. The system evolves via discrete-time steps consisting of a unitary hop followed by independent projective measurements of local occupation with probability $p$ .

State Representation: The system remains in a Gaussian state, fully specified by the correlation matrix $C_{ij} = \langle c^\dagger_i c_j \rangle$ .
Dynamics: Unitary evolution updates $C$ via $U C U^\top$ , while projective measurements update $C$ according to standard Gaussian update rules (Eqs. 3 and 4).
Observables: For a subsystem $A$ $A$ , the single-particle entanglement spectrum $\{\nu_k\}$ ${ν_{k}}$ is obtained by diagonalizing the reduced correlation matrix $C_A$ $C_{A}$ . Three quantities are computed:
1. Von Neumann entropy: $S_A = -\sum [\nu_k \ln \nu_k + (1-\nu_k) \ln(1-\nu_k)]$ .
2. Number fluctuation: $F_A = \sum \nu_k(1-\nu_k) = \text{Var}(N_A)$ .
3. Participation number: $M_A = \exp[-\sum w_k \ln w_k]$ , where $w_k \propto \nu_k(1-\nu_k)$ .
Statistical Protocol: Simulations run $N_{\text{traj}}$ independent trajectories (up to 128 per parameter point). Observables are time-averaged over a steady-state window after discarding a transient, and final values are averaged over trajectories. Uncertainties are standard errors across independent trajectories.
Model Comparison: Subsystem scaling at weak monitoring ( $p=0.1$ ) is analyzed by fitting block entropy $S(\ell)$ against conformal chord length using two models: a pure logarithmic law ( $S = \alpha \ln x + b$ ) and a logarithmic-plus-linear law ( $S = \alpha \ln x + \beta \ell + b$ ). Model selection is performed using the Akaike Information Criterion (AIC).

Key Contributions and Results

Tracking the Crossover: Both $F$ and $M$ successfully track the entanglement crossover. They decrease monotonically with the measurement probability $p$ and lose their system-size dependence in the strongly monitored regime, mirroring the behavior of the entropy $S$ .
Redundancy of $M$ with $F$ : At the level of trajectory- and time-averaged steady-state values, $M$ is found to be nearly a deterministic function of $F$ . A pooled quadratic fit of $\ln M$ versus $\ln F$ across all measurement rates and 128 trajectories per point captures $R^2 = 0.997$ of the variance. The deviations are physically marginal ( $\lesssim 1.5\%$ ). Consequently, $M$ carries little information independent of $F$ for this specific model; the monitored free-fermion steady state does not exhibit spectra with equal $F$ but significantly different participation numbers.
Statistical Advantage of $M$ : While $M$ is informationally redundant, it offers a distinct statistical advantage. In the strongly monitored regime, the trajectory-to-trajectory coefficient of variation (CV) of $M$ is approximately half that of $S$ or $F$ (e.g., $CV_M \approx 0.07$ vs. $CV_{S,F} \approx 0.13\text{--}0.14$ ). This reduction in noise arises because the normalization by $F_A$ in the definition of $M$ cancels the overall amplitude fluctuations that dominate the variance of $S$ and $F$ . This translates to a factor of $\sim 4$ fewer trajectories required to achieve the same relative error for $M$ compared to $S$ in the area-law regime.
Finite-Size Scaling Analysis: Model comparison at $L=48, 64, 96$ for weak monitoring ( $p=0.1$ ) reveals that a pure logarithmic law is not the statistically preferred description. The "log-plus-linear" model is favored (negative $\Delta$ AIC), indicating a small but significant residual quasi-extensive component ( $\beta \approx 0.03$ ). Furthermore, the effective logarithmic coefficient $\alpha$ drifts upward with system size. These findings support the crossover scenario (rather than a distinct critical phase) suggested by recent field-theory analyses, where the logarithmic regime is a long-lived transient flowing to an area law.

Significance and Claims
The paper concludes with a restrained assessment of the utility of participation-style diagnostics in this context:

Negative Result: The participation number $M$ is not an independent diagnostic for monitored free fermions; it is essentially a reparametrization of the number fluctuation $F$ .
Practical Utility: The primary value of $M$ is numerical: it provides a variance-reduced summary of the same information, making it a convenient tool for smoothing data in finite-size studies of the crossover.
Experimental Context: Unlike $F$ , which can be accessed via particle-number statistics, $M$ requires knowledge of individual spectrum eigenvalues and is not currently experimentally accessible.
Scope: The findings are specific to the Gaussian (free-fermion) ensemble. The author notes that $M$ might acquire independent diagnostic power in interacting circuits or higher dimensions where the entanglement spectrum explores different shapes at fixed $F$ , but this remains an open question.

The author emphasizes that delimiting what "cheap" spectral summaries can and cannot deliver is a useful contribution to a field where candidate diagnostics multiply rapidly. The full trajectory-resolved dataset and code are made available to facilitate similar tests in other models.

More like this