Prethermalization, shadowing breakdown, and the absence of Trotterization transition in quantum circuits

This paper utilizes the truncated operator propagator to demonstrate that while prethermalization stabilizes effective energy in quantum many-body systems, the finite shadowing time in chaotic regimes precludes a Trotterization transition, contradicting previous claims about infinite faithfulness times.

The Gist

The Big Picture: Simulating the Universe on a Noisy Computer

Imagine you want to simulate the weather, or how a virus spreads, or how atoms dance inside a star. To do this, scientists use quantum computers. These machines are incredibly powerful, but they are also "noisy." Think of them like a brand-new, high-tech car that has a slightly loose steering wheel and a wobbly tire.

The paper asks a simple but profound question: If we use this wobbly car to drive a route that represents a perfect, smooth physical law, how long can we trust the journey before the car's wobbles make the map completely useless?

In the world of physics, this "wobble" is called a Trotterization error. It happens because quantum computers can't do continuous, smooth time; they have to chop time into tiny, discrete steps (like taking steps instead of gliding). The paper investigates how long these steps can be taken before the simulation stops looking like the real thing.

The Core Discovery: The "Shadow" That Fades

In classical chaos theory (like weather), there is a concept called "Shadowing." Imagine you are walking a path, but you are slightly drunk and stumbling (the noisy simulation). A sober friend (the "shadow trajectory") starts walking from a spot just a few inches away from where you started. Even though you are stumbling, your friend's perfect path will stay right next to you for a long time, effectively "shadowing" your drunken walk.

For a long time, physicists hoped that quantum computers might have an infinite shadow. They thought that even with errors, the simulation would stay close to a "perfect" version of reality forever, just shifted slightly.

This paper says: No. The shadow always breaks.

The author, Marko Žnidarič, proves that in complex quantum systems, the "shadow" eventually fades away. No matter how small the error is, if you wait long enough, the noisy simulation will drift so far from reality that it no longer resembles any true physical path. The "shadowing time" is finite.

The "Magic" Observable: The Energy Shield

However, there is a twist. The paper finds that not all parts of the simulation are equally fragile.

Imagine the simulation is a house. Most of the house (the walls, the furniture, the temperature) starts to crumble quickly when the foundation shakes. But there is one specific room—the Energy Room—that is incredibly sturdy.

• The Energy Room: The total energy of the system stays stable for a surprisingly long time, much longer than simple math would predict. This is called Prethermalization. It's like the house has a magical, temporary shield that keeps the energy intact while everything else falls apart.
• The Catch: This shield is not permanent. Eventually, even the Energy Room will crumble, especially as the system gets bigger (approaching the "thermodynamic limit," or an infinitely large universe).

So, while the "Energy" observable is a hero that fights off errors for a long time, it eventually loses. And all other observables (like magnetism or particle positions) give up much sooner.

The "Heisenberg Time" Trap: Why Small Simulations Lie

One of the most important practical lessons in the paper is about how we test these computers.

The author points out a common trap: Many previous studies used small quantum systems (like a tiny 16-bit computer) and ran them for a long time. They saw the "Energy Room" stay stable and concluded, "Aha! The simulation is perfect forever! There is a 'Trotterization Transition' where errors stop mattering."

The author calls this a hallucination caused by small size.

• The Analogy: Imagine you are trying to predict the weather for a whole continent, but you only have a weather station in a single backyard. If you watch that backyard for a week, the weather might look perfectly stable. But if you wait a year, the patterns will break.
• The Math: In quantum mechanics, there is a limit called the Heisenberg Time. It's the time it takes for a system to "remember" its own discrete, pixelated nature. If you run a simulation longer than this time, the results are just artifacts of the small size, not real physics.
• The Verdict: Previous claims of a "magic transition" where errors disappear were just because researchers stopped watching before the system got big enough to show the truth. In the real, infinite world, there is no transition. Errors always win eventually.

The New Tool: The "Truncated Propagator"

How did the author figure all this out? They used a clever mathematical tool called the Truncated Operator Propagator.

• The Analogy: Imagine you want to know how a rumor spreads through a massive city. You can't track every single person. Instead, you create a simplified map that only tracks how the rumor moves between neighborhoods (local operators).
• The Magic: By looking at the "spectrum" (the frequencies) of this simplified map, the author could predict exactly how long the "Energy Room" would stay stable and how fast the "rumor" (errors) would spread.
• The Benefit: This method is like having a crystal ball. It allows scientists to calculate the exact "prethermalization time" (how long the shield lasts) without having to run the actual, expensive, and slow quantum simulation. It's faster and more accurate than previous methods.

Summary of the "Takeaways"

1. No Free Lunch: You cannot simulate a quantum system perfectly forever on a noisy quantum computer. The "shadow" of reality always breaks eventually.
2. The Energy Shield: The total energy of the system is surprisingly robust and lasts a long time (prethermalization), but it is not immortal.
3. Beware of Small Samples: Don't trust simulations that look perfect just because they are small. You have to wait long enough (past the Heisenberg time) to see the errors take over.
4. No "Magic" Transition: There is no sudden point where errors magically stop mattering. The transition from "good simulation" to "bad simulation" is a gradual fade, not a switch.
5. A Better Tool: The author's new mathematical method allows us to predict these failures and calculate things like "diffusion" (how heat spreads) with high precision, even for infinite systems.

In a nutshell: Quantum computers are great at simulating nature, but they are not perfect mirrors. They are like a slightly distorted mirror that gets worse the longer you look into it. This paper tells us exactly how long we can look before the reflection becomes unrecognizable, and it gives us a better way to calculate that time.

Technical Summary

1. Problem Statement

The paper addresses a fundamental challenge in the simulation of many-body quantum systems on noisy quantum computers: How long can a discrete-time simulation (Trotterized evolution) faithfully represent continuous-time Hamiltonian dynamics?

• Trotterization Errors: Simulating a Hamiltonian $H_0$ via a quantum circuit involves splitting the time evolution into discrete steps $\tau$, introducing errors of order $O(\tau^2)$.
• Shadowing Property: In classical chaotic systems, the "shadowing lemma" suggests that a noisy trajectory can be closely followed by a true trajectory starting from a slightly perturbed initial condition, potentially for an infinite time in uniformly hyperbolic systems.
• The Core Question: Does this "shadowing" property hold in quantum many-body chaotic systems? Specifically, is there a "Trotterization transition" (a critical step size $\tau_c$) where the simulation suddenly becomes faithful, or does the simulation inevitably diverge from the true dynamics?
• Previous Claims: Recent literature suggested the existence of a Trotterization transition or "quantum localization" that stabilizes simulations for small $\tau$. The author challenges these claims, arguing they are artifacts of finite-size effects.

2. Methodology

The author introduces and utilizes the Truncated Operator Propagator method, based on Ruelle-Pollicott (RP) resonances, to analyze the dynamics directly in the thermodynamic limit ($L \to \infty$).

• Operator Propagator ($M$): Instead of evolving states, the method evolves operators (observables) in the Heisenberg picture. The propagator $M$ maps an observable $A$ to $U_\tau^\dagger A U_\tau$.
• Truncation: To make the problem tractable, the infinite-dimensional operator space is truncated to local operators supported on at most $r$ sites. This breaks the unitarity of $M$, moving its eigenvalues inside the unit circle.
• Momentum Resolution: The propagator is constructed for specific quasimomenta $k$, allowing the study of translationally invariant systems and transport properties.
• Ruelle-Pollicott Resonances: The eigenvalues $\lambda_j$ of the truncated propagator determine the decay rates of correlation functions. The largest eigenvalue in modulus, $\lambda_1$, dictates the asymptotic behavior.
  • The spectral gap $\Delta = 1 - |\lambda_1|$ determines the prethermalization time $T \approx 1/\Delta$.
  • The momentum dependence of $\lambda_1(k)$ near $k=0$ allows for the calculation of diffusion constants ($D$).
• In-situ Propagation: To handle large truncation sizes ($r$), the author uses an "in-situ" propagation technique that evolves operators on exactly $r$ sites rather than $r+2$, significantly reducing computational cost and memory requirements compared to standard exact diagonalization.

3. Key Contributions

A. Refutation of the Trotterization Transition

The paper demonstrates that no Trotterization transition exists in non-integrable, chaotic many-body quantum systems.

• Previous observations of a transition were shown to be finite-size effects. In finite systems, the Heisenberg time ($t_H \sim 2^L$, the inverse level spacing) is often smaller than the prethermalization time ($T$). If $t_H < T$, the system appears to thermalize to a plateau (mimicking stability), but this is an artifact of the discrete spectrum.
• In the true thermodynamic limit ($L \to \infty$ followed by $t \to \infty$), the correlation functions of all observables eventually decay to zero, regardless of how small $\tau$ is.

B. Shadowing Breakdown in Quantum Systems

• Unlike classical uniformly hyperbolic systems where shadowing time can be infinite, quantum shadowing time is always finite in chaotic many-body systems.
• Even the most stable observable (the energy $H_0$) eventually thermalizes due to Trotter errors. While $H_0$ exhibits a long-lived prethermal plateau, it is not conserved indefinitely in the thermodynamic limit.

C. Prethermalization and Effective Hamiltonians

• The method identifies a single almost-conserved operator (an effective Floquet Hamiltonian $H'_F$) corresponding to the eigenvector of $\lambda_1$.
• This explains the phenomenon of prethermalization: the system behaves as if governed by $H'_F$ for a time $T \sim \exp(1/\tau)$ before heating up.
• Crucially, only observables with a large overlap with $H'_F$ (like total energy) exhibit this long-lived stability. Generic observables (orthogonal to $H'_F$) decay rapidly on the timescale of standard Trotter errors ($O(\tau^2)$).

D. Calculation of Transport Coefficients

• The momentum dependence of the spectral gap $\Delta(k)$ allows for the precise calculation of the energy diffusion constant $D$.
• The method yields $D$ with higher precision and lower computational cost than traditional methods like Green-Kubo formulas or Lindblad NESS (Non-Equilibrium Steady State) simulations, especially in the thermodynamic limit.

4. Key Results

• Kicked Ising Model (1D):

  • Small $\tau$: The gap $\Delta$ scales as $\Delta \sim e^{-c/\tau}$, leading to an exponentially long prethermalization time $T$. The energy correlation function shows a long plateau, but it eventually decays to zero as $L \to \infty$.
  • Large $\tau$ (near $\pi$): The system exhibits Discrete Time Crystal (DTC) behavior. The gap scales as $\Delta \sim |\pi - \tau|^{-1}$ (exponentially large $T$), and the system shows period-doubling oscillations. However, this DTC is "prethermal" and breaks down in the thermodynamic limit.
  • No Transition: The spectral gap never closes completely for finite $\tau$; it only approaches zero as $\tau \to 0$ or $\tau \to \pi$. There is no sharp transition point.

• Kicked XX Model (1D):

  • Prethermalization exists only for small $\tau$. Near $\tau = \pi$, the system does not form a DTC (unlike the Ising model), and the timescale diverges only quadratically, not exponentially.

• 2D Kicked Ising Model:

  • Prethermalization is observed for small $\tau$.
  • A qualitative change in dynamics occurs around $\tau \approx 1.3$, where the decay of correlation functions shifts from oscillatory to monotonic, and the spectrum structure changes (loss of the "noisy ring" core).

• Diffusion Constants:

  • For the tilted-field Ising model, the calculated diffusion constant $D \approx 1.447$ matches the most accurate literature values (Green-Kubo/Lindblad methods) but is obtained with significantly less numerical effort and directly in the thermodynamic limit.

5. Significance and Implications

1. Theoretical Clarity: The paper resolves the debate on Trotterization transitions, establishing that chaotic quantum systems do not possess an infinite shadowing time. The "stability" observed in small systems is a finite-size artifact where the Heisenberg time is shorter than the prethermalization time.
2. Methodological Advance: The Truncated Operator Propagator is presented as a superior tool for studying Floquet systems, prethermalization, and transport. It works directly in the thermodynamic limit, avoiding the pitfalls of exact diagonalization on small systems.
3. Experimental Relevance: For noisy quantum simulators, the results imply that while prethermalization allows for long-lived simulations of specific observables (like energy), no simulation is perfectly faithful forever. Users must account for the finite prethermalization time $T$ and ensure their simulation time $t \ll T$ to avoid heating.
4. DTC Stability: The study clarifies the nature of Discrete Time Crystals in clean systems, showing they are often prethermal phenomena that rely on the existence of an almost-conserved operator, rather than true infinite-time phases in the thermodynamic limit.

In summary, the paper provides a rigorous framework using Ruelle-Pollicott resonances to demonstrate that while prethermalization offers a window of stability for quantum simulations, the fundamental chaotic nature of many-body systems ensures that shadowing is always finite and no Trotterization transition exists in the thermodynamic limit.