Relaxation Critical Dynamics in Measurement-induced Phase Transitions

This paper investigates the relaxation critical dynamics of measurement-induced phase transitions in one-dimensional quantum circuits, revealing distinct entanglement entropy scaling behaviors for different initial states and proposing a unified scaling framework that significantly reduces experimental post-selection overhead.

The Gist

Imagine a crowded dance floor where people are constantly moving in complex, synchronized patterns. This represents a quantum system evolving over time. Now, imagine that every few seconds, a camera flashes, freezing the dancers in place and forcing them to reset their positions based on what the camera sees. This is the "measurement" part of the story.

This paper explores what happens when you mix these two things: the natural, flowing dance (unitary evolution) and the sudden, disruptive camera flashes (measurements).

The Big Picture: A Tug-of-War

The researchers are studying a "phase transition," which is like a sudden switch in how the system behaves.

• The Entangling Phase (Volume Law): If the camera flashes rarely, the dancers keep moving freely. They get tangled up with everyone else, creating a massive, complex web of connections across the whole room. The "entanglement" (how connected everyone is) grows huge, proportional to the size of the room.
• The Disentangling Phase (Area Law): If the camera flashes constantly, the dancers are frozen too often. They can't spread their connections far. They stay isolated in small groups, and the overall "entanglement" stays small, only depending on the size of the groups, not the whole room.

The "Measurement-Induced Phase Transition" (MIPT) is the exact tipping point where the system switches from being a giant, tangled web to being a collection of small, isolated groups.

The Experiment: Watching the System Relax

The authors didn't just look at the final result; they watched how the system relaxes or changes over time right after the rules change. They tested two different starting scenarios:

1. Starting with a "Tangled" Room (Volume-Law Initial State)
Imagine starting with the dancers already in a massive, complex web. Then, you turn on the camera flashes at the critical tipping point.

• What happened: The researchers found that the "tangled-ness" (entanglement entropy) didn't just fade away slowly. It dropped rapidly, following a specific rule: it decreases as time goes up (specifically, proportional to $1/t$).
• The Analogy: Think of a giant, messy knot of yarn. If you start cutting it at the critical speed, the knot unravels quickly, and the amount of mess left behind shrinks predictably. The bigger the room (system size), the more "mess" there is to begin with, but it unravels at a rate that depends on the size of the room.

2. Starting with an "Untangled" Room (Product Initial State)
Imagine starting with the dancers standing in neat, separate lines, completely unconnected. Then, you turn on the camera flashes at the critical tipping point.

• What happened: Here, the "tangled-ness" grows, but very slowly. It grows like the natural logarithm of time ($\ln t$).
• The Analogy: Think of a slow-growing vine. It starts small and spreads, but it doesn't explode outward instantly. It creeps along, getting bigger, but the rate of growth is very gentle. This confirmed what other scientists had seen before.

The "Unified" Discovery

The most exciting part of the paper is that the authors found a single mathematical recipe that describes both of these very different behaviors.

• Even though one scenario starts with a mess and gets cleaner, and the other starts clean and gets messy, they both fit into the same "scaling form."
• It's like having one master key that can open two very different-looking doors. The key works, but the way the door opens (the "scaling function") looks different depending on which door you are trying to open.

Why This Matters for Real Experiments

The paper highlights a major problem in studying these quantum systems: The "Post-Selection" Problem.

• The Problem: In a real quantum computer, if you want to see the "tangled" state, you have to run the experiment millions of times and throw away all the results where the random measurements didn't go your way. This is like trying to find a specific needle in a haystack by throwing away every straw that isn't the needle. As the system gets bigger, the number of times you have to throw things away grows exponentially, making it impossible to track.
• The Solution: The authors show that you don't need to wait for the system to settle into its final, steady state (which takes a long time and requires massive post-selection). Instead, you can look at the short-time behavior (the relaxation dynamics).
• The Benefit: Because the system changes predictably very quickly (in the short time), you can figure out the critical tipping point much faster. This drastically reduces the number of times you need to run the experiment and throw away data. In fact, they suggest that by combining this short-time method with a specific "cross-correlation" trick (using classical computers to help simulate parts of the process), you might be able to eliminate the need for throwing away data entirely.

Summary

In simple terms, this paper discovered that when a quantum system is at the tipping point between being "tangled" and "untangled," it behaves in a very specific, predictable way depending on how you start it.

1. If you start tangled, it untangles quickly ($1/t$).
2. If you start clean, it tangles slowly ($\ln t$).
3. Both behaviors fit into one big, unified theory.
4. Most importantly, watching this short-term "relaxation" allows scientists to find the tipping point without the impossible task of throwing away millions of experimental results, making it much easier to study these phenomena on real quantum devices.

Technical Summary

Technical Summary: Relaxation Critical Dynamics in Measurement-induced Phase Transitions

Problem Statement
Measurement-induced phase transitions (MIPT) describe nonanalytical changes in entanglement entropy arising from the competition between unitary evolution and projective measurements in monitored quantum systems. While the steady-state properties of MIPT are well-characterized by volume-law (entangling) and area-law (disentangling) phases separated by a critical point $p_c$, the non-equilibrium relaxation dynamics remain under-explored. A central open question is whether universal scaling behaviors exist for the relaxation of entanglement entropy $S$ from different initial states, specifically comparing volume-law initial states against product states, and whether these distinct behaviors can be unified under a single scaling framework.

Methodology
The authors investigate the relaxation critical dynamics within a (1+1)-dimensional hybrid stabilizer circuit featuring a regular brick-wall structure. The system evolves via alternating layers of random two-site Clifford gates and local projective measurements (occurring with probability $p$). The study focuses on the time evolution of the half-chain entanglement entropy $S(t)$ near the critical point $p_c \approx 0.15995$.

Two distinct initial states are prepared:

1. Volume-law state: The steady state of the system with zero measurement probability ($p=0$).
2. Product state: The steady state corresponding to full measurement probability ($p=1$).

The authors employ numerical simulations to track the evolution of $S(t)$ for various system sizes $L$ and measurement probabilities $p$. They apply finite-size scaling analysis, incorporating the dynamic exponent $z=1$ and the correlation length exponent $\nu \approx 1.260$, to test proposed scaling forms.

Key Contributions and Results

1. Unified Relaxation Scaling Form:
   The paper proposes a general relaxation scaling form for the entanglement entropy:
   $$S(t, g, L) = \alpha \ln L + G(tL^{-z}, gL^{1/\nu}, X_0)$$
   where $g = p - p_c$, $X_0$ represents initial state information, and $G$ is a scaling function. This framework unifies the description of dynamics for both volume-law and product initial states, with the scaling function $G$ exhibiting different asymptotic behaviors depending on $X_0$.

2. Relaxation from Volume-Law Initial State:
   For an initial state prepared in the volume-law phase ($p=0$), the authors discover a novel dynamic scaling relation at the critical point ($g=0$). In the short-time regime, the entanglement entropy decays as:
   $$S(t) \propto t^{-1}$$
   Crucially, the coefficient of this decay is proportional to the system size $L$. This behavior ($S \propto L t^{-1}$) arises because the initial volume-law characteristic persists during the macroscopic short-time stage due to critical slowing down, dominating the logarithmic term.

3. Relaxation from Product Initial State:
   For a product state initial condition, the authors confirm that the entanglement entropy grows logarithmically with time at the critical point:
   $$S(t) \propto \ln t$$
   This result is consistent with previous studies but is here derived systematically through dynamic scaling analysis. The scaling function for this case involves a logarithmic term that cancels the system-size dependence, leading to a growth rate independent of $L$ in the short-time limit.

4. Off-Critical Dynamics:
   The unified scaling form successfully incorporates off-critical effects ($g \neq 0$). For both initial states, rescaled curves of $(S - \alpha \ln L)$ versus $tL^{-z}$ collapse onto single curves for fixed $gL^{1/\nu}$, validating the scaling hypothesis across the transition.

5. Short-Time Critical Point Detection:
   The authors demonstrate that the short-time dynamics from a product state offer an efficient method to locate $p_c$. In a semi-logarithmic plot of $S$ vs. $t$, the curve at $p_c$ is a straight line ($S \propto \ln t$), whereas curves for $p > p_c$ and $p < p_c$ exhibit downward and upward deviations, respectively. This allows for critical point identification without waiting for long-time equilibration.

Significance and Claims
The paper claims that this unified scaling framework offers significant advantages for the experimental detection of MIPT, primarily by addressing the "post-selection problem." In experimental settings, the probability of identical measurement trajectories decays exponentially with the number of measurements, making steady-state tracking prohibitively expensive (overhead $\propto \exp(pLt_{eq})$).

By leveraging relaxation dynamics in the short-time regime ($t \ll t_{eq}$), the required overhead is drastically reduced. Furthermore, the authors suggest that their findings, particularly the applicability of the short-time scaling form (Eq. 10) to the volume-law regime ($p < p_c$) where $S$ is initially small, can be combined with cross-correlation protocols. This combination could potentially eliminate the post-selection overhead entirely by enabling the use of trackable classical simulations to estimate entanglement entropy, thereby facilitating the experimental study of MIPT on realistic quantum devices.