Destruction and recovery of the entanglement entropy of a many-body quantum system after a single measurement

This paper numerically investigates the probability distribution of entanglement entropy changes in one-dimensional non-interacting fermions following single measurements, revealing that the distribution's shape, symmetry, and spatial inhomogeneity are critically dependent on monitoring strength, measurement protocol, and proximity to the subsystem boundary, ultimately transitioning from Gaussian to Zeno-dominated behaviors as monitoring intensifies.

The Gist

Imagine a bustling city of tiny, invisible particles called fermions. These particles are constantly dancing, swapping places, and getting "entangled" with their neighbors. In quantum physics, entanglement is like a secret, invisible thread connecting two groups of people; the more they talk and interact, the stronger the thread, and the more "information" they share.

This paper is about what happens to these invisible threads when we start peeking at the city.

The Setup: The City and the Spies

The researchers set up a one-dimensional line of these particles (like a row of houses). They let the particles dance and get entangled until the city reaches a steady state of maximum connection.

Then, they introduce three different types of "spies" (measurement protocols) to watch the city. The goal is to see how the act of watching changes the connections (entanglement) between the left half and the right half of the city.

The three spies are:

1. The Whisperer (Quantum State Diffusion): This spy constantly whispers weak questions to every house at once. "Are you occupied? Are you empty?" The answers are fuzzy and noisy, like static on a radio.
2. The Flasher (Quantum Jump): This spy waits for a specific event—a house suddenly lighting up (a particle appearing). When it happens, the spy flashes a bright light, instantly freezing that house's state.
3. The Inspector (Projective Measurement): This is the classic spy. They pick a random house, knock on the door, and force the occupant to reveal their true state (either "occupied" or "empty") with 100% certainty.

The Big Question: Does Watching Break the Magic?

In the quantum world, there's a famous idea called the Quantum Zeno Effect. It's like the old saying, "A watched pot never boils." If you watch a quantum system too closely, you freeze its evolution. The particles stop dancing, and the invisible threads of entanglement snap.

The researchers wanted to know: Does the system freeze completely, or does it recover? And what does the "damage" look like?

The Findings: A Tale of Two Zones

1. The Whisperer's Effect (Weak Monitoring)

When the Whisperer is very quiet (weak monitoring), the city barely notices. The changes in entanglement are small and random, looking like a standard bell curve (Gaussian distribution). It's like a gentle breeze ruffling the leaves; the tree stays standing.

But as the Whisperer gets louder, the story changes. The "noise" starts to create big, rare spikes in the data. The distribution of changes gets "fat tails," meaning sometimes the entanglement changes wildly, not just a little bit.

2. The Flasher and Inspector's Effect (Strong Monitoring)

When the Flasher or Inspector is very active (strong monitoring), things get dramatic.

• The "Freeze": Most of the time, the spies look at a house, and nothing happens. The house was already in the state they expected. The entanglement doesn't change at all. This creates a huge spike in the data at "zero change." This is the Quantum Zeno Effect in action: the system is so watched it's paralyzed.
• The "Boundary" Surprise: Here is the most interesting part. The researchers found that location matters.
  • Deep inside the neighborhoods (Bulk): If a spy looks at a house in the middle of the left or right side, the entanglement barely changes. The system is frozen.
  • At the border (Boundary): If a spy looks at a house right on the line separating the two halves, the entanglement can change wildly. It's like the border is the only place where the "magic" is still alive. The spies are mostly frozen in the middle, but the border is a chaotic battlefield where connections are constantly being broken and rebuilt.

The Recovery: Can the City Heal?

Even when the spies are watching, the city isn't dead. Between the moments of observation, the particles have a chance to dance again.

• The researchers found that the system has an intrinsic ability to heal. After a spy freezes a particle, the non-observed particles quickly rearrange themselves to rebuild the invisible threads.
• However, if the spies watch too frequently, the healing can't keep up with the freezing. The system collapses into a state where the two halves are disconnected (Area Law phase).

The "Rare Event" Twist

One of the most counter-intuitive findings is that sometimes, watching actually increases entanglement.
Imagine you are trying to untangle a knot. Usually, pulling on a string makes it tighter. But occasionally, a specific tug might loosen the knot. Similarly, a measurement that "collapses" a particle can sometimes force the rest of the system to rearrange in a way that creates more connection than before. These are rare events, but they happen, and they are crucial for understanding the full picture.

The Takeaway: Why This Matters

This paper teaches us that looking at a quantum system isn't just a simple "on/off" switch for its magic.

• It's not just about the average: If you only look at the average behavior, you might miss the rare, wild fluctuations that tell the real story.
• Location is key: The "edge" of a system behaves very differently from the "middle." The edge is where the quantum magic is most fragile and most resilient.
• The Zeno Effect is real but nuanced: Watching too much does freeze the system, but the system fights back, trying to heal itself, especially at the boundaries.

In simple terms: Imagine a group of friends holding hands in a circle. If you gently tap them, they sway. If you stare at them intensely, they freeze. But if you stare at the middle of the circle, they freeze completely. If you stare at the gap between two groups, the people at the gap might suddenly let go and grab new hands, changing the whole pattern. This paper maps out exactly how that happens, showing us that the "edges" of our quantum world are where the most interesting drama unfolds.

Technical Summary

1. Problem Statement

The paper investigates the impact of continuous and projective measurements on the dynamics of entanglement entropy (EE) in a one-dimensional chain of non-interacting complex fermions. While the existence of Measurement-Induced Phase Transitions (MIPT) and the transition from volume-law to area-law scaling of EE are well-studied via averaged quantities, the statistical properties of these transitions remain unclear. Specifically, the authors address:

• The full probability distribution of the change in EE ($\Delta S$) resulting from a single measurement event (or a single time step in continuous protocols).
• Whether the average EE is a sufficient indicator of the system's state, or if rare events and non-Gaussian tails in the distribution provide critical insights.
• The spatial inhomogeneity of these effects: Do measurements affect all sites equally, or are they localized to specific regions (e.g., subsystem boundaries)?
• The interplay between the destruction of entanglement by measurement and its regeneration via non-Hermitian evolution (between jumps) or unitary evolution.

2. Methodology

The authors perform extensive numerical simulations on a 1D lattice of $L$ sites with periodic boundary conditions, initialized in a Néel state ($|1010\dots\rangle$) at half-filling ($N=L/2$). They compare three distinct measurement protocols:

1. Quantum State Diffusion (QSD): A continuous weak measurement protocol (homodyne detection) modeled by a stochastic Schrödinger equation with Gaussian noise. The system evolves continuously, and the change in EE is defined over a small time step $dt$.
2. Quantum Jump (QJ): A continuous monitoring protocol (photo-detection) where the system evolves under a non-Hermitian Hamiltonian until a "jump" occurs (detection of an occupied site, $n_i=1$). The change in EE is defined as the difference immediately before and after the jump.
3. Projective Measurement (PM): A standard quantum mechanical protocol where a site is measured randomly, collapsing the wavefunction to an eigenstate ($n_i=0$ or $1$) with Born rule probabilities. The evolution between measurements is unitary.

Technical Implementation:

• Gaussian State Formalism: Since the Hamiltonian and measurement operators are quadratic (or preserve particle number in a way that maintains Gaussianity), the many-body state remains a Gaussian state. The authors track the system using the $L \times N$ coefficient matrix $U(t)$ and the correlation matrix $D_{ij} = \langle c^\dagger_i c_j \rangle$.
• Entanglement Calculation: The EE of a subsystem $A$ is calculated from the eigenvalues $\lambda_i$ of the reduced correlation matrix $D_A$: $S = -\sum [\lambda_i \ln \lambda_i + (1-\lambda_i)\ln(1-\lambda_i)]$.
• Data Collection: The authors simulate thousands of quantum trajectories ($10^3 - 10^5$) to construct smooth probability distributions $P(\Delta S)$ for the change in EE after saturation. They analyze both global distributions and spatially resolved distributions (dependence on the measured site $r$).

3. Key Contributions and Results

A. Quantum State Diffusion (QSD) Protocol

• Distribution Shape: In the weak monitoring limit, the distribution of EE change is Gaussian. As monitoring strength ($\gamma$) increases, the distribution develops symmetric exponential tails ($\sim e^{-|x|}$) while the core remains Gaussian.
• Strong Monitoring: In the strong limit, the distribution becomes sharply peaked at zero, signaling the dominance of the Quantum Zeno Effect (measurements freeze the dynamics).
• Scale Invariance: Uniquely, the distribution $P(\Delta S_{QSD})$ is independent of system size $L$ for all monitoring strengths. The authors attribute this to the fact that the EE change is governed by boundary sites, and in QSD, all sites are measured simultaneously, making the boundary contribution scale-invariant relative to the bulk.

B. Quantum Jump (QJ) and Projective Measurement (PM) Protocols

These two protocols yield qualitatively similar results, distinct from QSD:

• Asymmetry and Non-Gaussianity: The distributions are asymmetric and non-Gaussian.
  • Positive Tail: Exponential decay for $\Delta S > 0$. This indicates a finite probability that a single measurement increases entanglement (counter-intuitive but observed), likely due to the non-local rearrangement of particles.
  • Negative Tail: Also exponential but sensitive to monitoring strength.
• Zero Peak: A sharp peak at $\Delta S = 0$ exists even for weak monitoring, which grows stronger with system size and monitoring strength. This suggests that most measurements have negligible effect on the global EE, consistent with the area-law phase.
• Spatial Inhomogeneity (Crucial Finding): The distribution is highly dependent on the location of the measured site:
  • Boundary Sites: Sites near the cut defining the subsystem show a broad distribution with asymmetric tails.
  • Bulk Sites: Sites far from the boundary show a distribution sharply peaked at zero with a much narrower support.
  • Dominance: As $\gamma$ increases, the full distribution is controlled entirely by the boundary sites. The bulk sites effectively exhibit a "frozen" state (Zeno effect).
• Recovery Mechanism: The authors analyze the regeneration of entanglement between measurements (non-Hermitian evolution for QJ, unitary for PM). They find that the regeneration rate balances the destruction rate on average. However, the distribution of regeneration is spatially uniform, unlike the measurement-induced destruction which is localized to the boundary.

C. Mutual Information and Scaling

• The mutual information $I(d_r)$ between the subsystem and a distant site decays as a power law in the weak monitoring limit (indicating long-range correlations) and exponentially in the strong limit.
• The authors argue that while the average EE might suggest a critical phase for small $\gamma$ (due to finite-size effects), the distribution function reveals the onset of the area-law phase (via the growing peak at zero and boundary localization) even for relatively small system sizes and weak monitoring.

4. Significance and Implications

1. Beyond Averages: The paper demonstrates that the full probability distribution of EE changes contains information that averages miss. Specifically, the presence of exponential tails and the spatial inhomogeneity of the distribution provide a more nuanced picture of the measurement-induced dynamics.
2. Mechanism of Entanglement Destruction: The study identifies that entanglement destruction is not a global phenomenon but is localized to the subsystem boundaries. The bulk of the system is largely unaffected by local measurements, especially in the strong monitoring regime.
3. Experimental Relevance: The findings address the "post-selection problem" in experimental observation of MIPT. By identifying that specific measurements (those at the boundary) carry the most weight in changing the EE, the authors suggest that experimental protocols could be optimized to focus on these "rare" but impactful events, potentially reducing the difficulty of observing phase transitions.
4. Universal Features: The distinction between the Gaussian/size-independent behavior of weak continuous measurements (QSD) and the asymmetric/boundary-localized behavior of discrete/projective measurements (QJ/PM) provides a classification of how different measurement types affect quantum information.

In conclusion, this work provides a high-resolution statistical map of how quantum measurements destroy and recover entanglement, highlighting the critical role of spatial inhomogeneity and the limitations of using averaged quantities to characterize measurement-induced phases.