⚛️ quantum physics

On the Entanglement Entropy Distribution of a Hybrid Quantum Circuit

This paper investigates the entanglement entropy distribution in hybrid quantum circuits, demonstrating that higher-order statistical moments serve as robust diagnostics for measurement-induced phase transitions and proposing a phenomenological model that accurately describes the system across both volume-law and area-law phases.

Original authors: Jeonghyeok Park, Hyukjoon Kwon, Hyeonseok Jeong

Published 2026-04-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jeonghyeok Park, Hyukjoon Kwon, Hyeonseok Jeong

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

The Big Picture: A Tug-of-War in the Quantum World

Imagine a quantum computer not as a super-precise machine, but as a chaotic dance floor.

In this dance, Random Unitary Gates are the music. They are loud, chaotic, and encourage the dancers (the quantum particles) to mix, spin, and get tangled up with everyone else. The more they dance, the more "entangled" they become. In physics, this is called the Volume Law: the more people you have, the more tangled the whole group gets.

However, there is a strict DJ (the measurement) who occasionally stops the music and asks, "Hey, you! What are you doing right now?" This is a Measurement. When the DJ checks on a dancer, that dancer stops dancing with the group and becomes fixed in place. This breaks the entanglement. If the DJ checks too often, the dancers never get a chance to mix, and the group stays small and separate. This is the Area Law.

The Mystery:
Scientists know that if the DJ checks too rarely, the group gets huge and tangled. If the DJ checks too often, the group stays small. Somewhere in the middle, there is a "tipping point" (a phase transition) where the behavior changes.

The problem is that looking at the average amount of entanglement is like looking at the average height of a crowd. It tells you the general size, but it misses the drama. Sometimes, even if the average is the same, the crowd could be a mix of giants and dwarfs, or everyone could be the same height. The "shape" of the crowd matters.

The New Discovery: Looking at the "Shape" of the Chaos

This paper says: "Stop just looking at the average! Let's look at the distribution." They used two special tools to see the hidden details of the quantum dance:

1. The Index of Dispersion (IoD) – The "Spread vs. Size" Ratio

Imagine you are counting how many candies are in jars.

Volume Law (Low Measurement): The jars are huge, and the number of candies varies wildly from jar to jar. The "spread" is big compared to the "size."
Area Law (High Measurement): The jars are small, and the number of candies is very predictable. The "spread" is tiny.

The authors found that this ratio (IoD) acts like a thermometer for the phase transition. It stays steady in one phase, then suddenly jumps or drops right at the tipping point. It's a much sharper signal than just looking at the average number of candies.

2. Skewness – The "Lopsidedness" of the Crowd

Imagine a seesaw.

Volume Law: The seesaw is slightly tilted to one side, but it stays that way no matter how big the playground is. The paper found this tilt is a constant, universal value (like a fingerprint of the chaotic dance).
Area Law: As the DJ checks more often, the seesaw starts to tilt wildly in the other direction. The shape of the crowd becomes very lopsided.

Why is this cool? The "tilt" (skewness) stays perfectly constant in the chaotic phase, regardless of how many dancers you have. But in the quiet phase, it changes dramatically. This makes it a super-precise tool to find the exact moment the system switches from "chaotic" to "frozen."

The Detective Work: Finding the Tipping Point

In the past, scientists tried to find the tipping point by looking for the "peak" in the variance (how much things fluctuate). But the paper shows this is like trying to find the exact moment a balloon pops by looking at how much it wobbles. It's messy and hard to pinpoint.

By using Skewness, the authors found a much cleaner signal. They watched the "tilt" of the distribution. Right at the critical moment, the tilt changes direction most violently. It's like watching a car turn a corner; the steering wheel moves the most at the exact moment of the turn. This allowed them to pinpoint the critical measurement rate with high precision.

The Models: Two Different Stories for Two Worlds

The authors didn't just measure; they built stories (models) to explain why this happens.

The Chaotic Phase (Volume Law): They used a model called a Directed Polymer in a Random Environment.
- Analogy: Imagine a vine growing through a forest full of random obstacles. The vine tries to find the path of least resistance, but the obstacles are random. The "entanglement" is like the energy cost of that path. This model perfectly predicted the "tilt" (skewness) they saw in the data. It's like saying, "The chaos follows the same rules as a vine fighting through a storm."
The Frozen Phase (Area Law): They built a simple model based on Bell Pairs (quantum couples).
- Analogy: Imagine the dancers are paired up holding hands. The "Unitary" gates try to swap partners (making new pairs), but the "Measurements" act like scissors cutting the hands apart. They created a math equation (a stochastic model) that tracks how many pairs get cut. This model successfully predicted the distribution of entanglement when the measurements were frequent.

The Takeaway

This paper is a reminder that in the quantum world, the average isn't enough.

Just like knowing the average temperature of a week doesn't tell you if you need a coat or sunscreen (you need to know the highs and lows), knowing the average entanglement doesn't tell you the full story of a quantum system.

By looking at the shape of the data (how spread out it is and how lopsided it is), the authors found a new, super-sensitive way to detect when a quantum system changes its personality from "wildly chaotic" to "frozen and quiet." This helps us better understand how quantum information survives (or dies) in the presence of noise and measurement, which is crucial for building real quantum computers.

1. Problem Statement

Hybrid quantum circuits, consisting of random unitary gates and local projective measurements, exhibit a Measurement-Induced Phase Transition (MIPT).

Volume-Law Phase: At low measurement rates ( $p < p_c$ ), entanglement entropy ( $S$ ) scales extensively with system size ( $S \propto L$ ).
Area-Law Phase: At high measurement rates ( $p > p_c$ ), frequent measurements suppress entanglement, leading to $S \propto L^0$ (or logarithmic scaling at criticality).

The Gap: While the transition is well-characterized by the mean entanglement entropy $\langle S \rangle$ , hybrid circuits are intrinsically non-self-averaging due to the stochastic nature of measurement outcomes. Consequently, the entanglement entropy exhibits strong trajectory-to-trajectory fluctuations even in the thermodynamic limit. Previous studies have largely focused on the mean or the variance, but the paper argues that these lower-order moments are insufficient to fully capture the statistical structure of the distribution, particularly near the critical point $p_c$ . The authors aim to investigate higher-order moments (specifically the Index of Dispersion and Skewness) to uncover hidden features of the MIPT and develop effective phenomenological models for the distribution in both phases.

2. Methodology

The authors employ a combination of large-scale numerical simulations and theoretical modeling:

Model: A 1D hybrid quantum circuit with $L$ qubits initialized in a product state. The evolution involves alternating layers of random two-qubit Clifford unitary gates (brickwork architecture) and projective measurements in the Z-basis occurring at a rate $p$ . Periodic boundary conditions are applied.
Numerical Simulation: The authors simulate circuits with system sizes up to $L=1024$ . They generate ensembles of quantum trajectories to compute the full probability distribution $P(S)$ of the bipartite entanglement entropy.
Statistical Analysis:
- Index of Dispersion (IoD): Defined as $IoD = \text{Var}[S] / \langle S \rangle$ . This normalizes the variance by the mean to account for the scaling of entropy.
- Skewness: Defined as $\gamma(S) = \langle (S - \langle S \rangle)^3 \rangle / (\text{Var}[S])^{3/2}$ . This quantifies the asymmetry of the distribution.
- Finite-Size Scaling: Used to locate the critical point $p_c$ and critical exponent $\nu$ by collapsing data for different $L$ .
Theoretical Modeling:
- Volume-Law Regime: Utilizes the established mapping between hybrid circuits and a Directed Polymer in a Random Environment (DPRE). The entanglement entropy is mapped to the free energy of the polymer, which follows the Tracy-Widom (TW) distribution (Gaussian Unitary Ensemble).
- Area-Law Regime: Proposes a coarse-grained stochastic model based on the stabilizer formalism. Entanglement is treated as a collection of Bell pairs. The model treats the change in entropy ( $\Delta S$ ) as a stochastic process driven by unitary growth and measurement-induced destruction of Bell pairs, leading to a Fokker-Planck equation.
Validation: Theoretical predictions are compared against numerical data using the Kullback-Leibler (KL) divergence to quantify the statistical distance between the models and simulations.

3. Key Contributions and Results

A. Limitations of Variance Alone

The authors demonstrate that the variance of entanglement entropy, $\text{Var}[S]$ , is a poor diagnostic for the MIPT in large systems.

In large systems, the peak of the variance occurs deep within the volume-law phase ( $p < p_c$ ) rather than at the critical point.
This is because the variance is bounded by the condition $S \geq 0$ in the area-law phase (where the mean is small), whereas in the volume-law phase (large mean), the distribution is less constrained, allowing for larger fluctuations.

B. The Index of Dispersion (IoD) as a Robust Diagnostic

The IoD exhibits distinct behaviors across the phases:

Volume-Law: IoD curves for different system sizes do not converge; they vary with $L$ .
Area-Law: IoD curves for different $L$ collapse onto a single curve. As $p$ increases, IoD approaches unity (Poissonian behavior).
Critical Point: The IoD shows a sharp discontinuity near $p_c \approx 0.16$ , providing a precise method to locate the transition that is invisible to the variance alone.

C. Skewness as a Scale-Free Probe

The skewness $\gamma(S)$ reveals universal features:

Volume-Law Regime: The skewness is constant ( $\gamma \approx -0.224$ ) regardless of system size. This value matches the skewness of the Tracy-Widom distribution, confirming the DPRE mapping.
Area-Law Regime: The skewness follows a power-law scaling with the measurement rate: $\gamma(S) \propto p^{1.79}$ .
Transition: A sharp increase in skewness occurs near $p_c$ . By fitting an auxiliary function to the skewness data, the authors identify the point of maximal deformation, yielding $p_{max} \approx 0.158$ , which aligns closely with the known critical point.

D. Effective Models for the Distribution

Volume-Law: The DPRE model (using the TW distribution) accurately reproduces the full distribution $P(S)$ and its higher moments (IoD and skewness) across all system sizes.
Area-Law: The proposed coarse-grained stochastic model (based on Bell pair destruction) successfully captures the qualitative features of the distribution deep in the area-law phase. However, it shows deviations near the critical point, highlighting the limitations of mean-field-like coarse-graining near criticality.
Accuracy: The KL divergence analysis confirms that the DPRE model is exact in the volume-law phase, while the Fokker-Planck solution is a good approximation in the area-law phase, though both fail at the critical point.

4. Significance

Beyond the Mean: This work establishes that the distribution of entanglement entropy contains rich information that is lost when only considering the mean. Higher-order moments are essential for characterizing non-self-averaging quantum systems.
New Diagnostics: The IoD and skewness are proposed as robust, scale-free tools to detect MIPT, overcoming the ambiguities associated with variance-based diagnostics in large systems.
Universal Connections: The confirmation that the volume-law skewness matches the Tracy-Widom distribution provides independent numerical support for the DPRE mapping, solidifying the connection between monitored quantum circuits and statistical mechanics of directed polymers.
Generalizability: The methodology is applicable to other stochastic quantum systems, including those with conservation laws, noise, or other quantities of interest (e.g., "magic"), offering a new route for the statistical characterization of quantum dynamics.

In summary, the paper shifts the focus from the average behavior of hybrid circuits to their fluctuation statistics, demonstrating that higher-order moments provide a more sensitive and accurate window into the physics of measurement-induced phase transitions.