Exploiting emergent symmetries in disorder-averaged quantum dynamics

This paper proposes an efficient method for simulating large disordered quantum systems by exploiting emergent permutation symmetries in the disorder-averaged dynamical map, constructed via short-time and weak-disorder expansions, to reduce computational complexity from exponential to polynomial scaling.

The Gist

The Big Problem: The "Chaos" of Randomness

Imagine you are trying to predict how a crowd of people moves through a city. If everyone follows the same rules (like a perfectly choreographed dance), it's easy to predict the flow. In physics, this is like a symmetric quantum system—everything is orderly, and we can use shortcuts to solve the math.

But real life is messy. Imagine instead that every person in the crowd has a slightly different, random personality. Some walk fast, some slow, some turn left, some turn right. This is disorder. In quantum physics, this happens when the "rules" (the forces between particles) are random.

To understand what happens in this chaotic crowd, scientists usually have to run the simulation thousands of times, each time with a slightly different set of random rules, and then average the results. This is like trying to predict the weather by running a supercomputer simulation 1,000 times a day. It is incredibly slow and computationally expensive. As the crowd (the number of particles) gets bigger, the math becomes impossible to solve.

The Secret Weapon: Finding Order in the Chaos

The authors of this paper discovered a clever trick. They realized that while each individual run of the simulation is chaotic and breaks the symmetry, the average of all those runs actually has a hidden symmetry.

The Analogy:
Imagine you have a bag of marbles.

• Single Shot: You pull out one marble. It might be red, blue, or green. It's random.
• The Average: If you pull out 1,000 marbles and mix them, you get a specific, predictable ratio of colors (e.g., 50% red, 50% blue). Even though the individual pulls were random, the mixture has a perfect, stable pattern.

The paper shows that if you look at the "mixture" (the disorder-averaged state) rather than the individual "pulls," you can treat the system as if it were perfectly symmetrical again. This allows them to shrink the massive math problem down to a much smaller, manageable size.

The Solution: Two New "Shortcuts"

The authors developed two specific methods to calculate this "average" behavior efficiently, without having to run thousands of simulations.

1. The "Short-Time" Shortcut

• The Idea: If you only look at the very beginning of the movie (a very short time), the chaos hasn't had time to build up yet.
• The Trick: They expanded the math to look at what happens in tiny time steps. However, simple math expansions often break down later (like a prediction that says the temperature will go up forever). To fix this, they used a mathematical "brake" (called regularization) that keeps the prediction physical and stable, similar to how a Lindblad equation describes how a system loses energy or gets "noisy" over time.
• The Result: This works great for predicting what happens in the first few moments of the system's life.

2. The "Weak-Disorder" Shortcut

• The Idea: What if the randomness isn't too crazy? What if the marbles are mostly the same color, with just a few different ones?
• The Trick: They assumed the disorder is "weak" (small). They then calculated how the system behaves by starting with the perfect, non-random version and adding small "correction" terms for the randomness.
• The Result: This method is very powerful for larger systems and longer times, provided the randomness isn't overwhelming. They found that using an "exponential" way to handle the math (a specific type of correction) worked better than other methods, allowing them to simulate systems with 40 spins (particles) that would be impossible to simulate exactly.

The Test: The "Spinning Top" Model

To prove their method works, they tested it on a specific model called the Transverse-Field Ising Model.

• Imagine a bunch of spinning tops (spins) that are all connected to each other randomly.
• They applied a magnetic field to make them spin.
• They compared their new "shortcut" math against the "brute force" method (running 1,000s of simulations).
• The Outcome: Their new method matched the brute force results almost perfectly for a long time, but it did it much faster. It allowed them to simulate systems that were too big for the old methods.

Why This Matters (According to the Paper)

The paper claims this is a major step forward because:

1. It saves time: It turns an impossible calculation into a doable one for large systems.
2. It works for real experiments: In real-world quantum experiments (like cold atoms or defects in diamonds), you can't label every single particle perfectly. You can only measure the "average" behavior. This method is built exactly for that kind of "average" view.
3. It's flexible: It doesn't depend on one specific type of randomness; it can be applied to many different kinds of messy quantum systems.

In short, the authors found a way to ignore the "noise" of individual random events and focus on the "signal" of the average, using math tricks to keep the calculations fast and accurate.

Technical Summary

1. Problem Statement

Simulating the dynamics of quantum many-body systems with quenched disorder (randomness in Hamiltonian parameters) is computationally prohibitive for two main reasons:

1. Exponential Scaling: The Hilbert space dimension grows exponentially with the number of particles ($N$), making exact diagonalization or full state-vector evolution impossible for large systems.
2. Disorder Averaging: Physical observables require averaging over many random realizations ("shots") of the disorder. Since individual disorder realizations typically break global symmetries (e.g., permutation invariance), standard symmetry-based reductions cannot be applied to individual shots. This forces researchers to simulate the full Hilbert space for thousands of realizations and average the results, which is extremely costly.

The central challenge is to find a way to perform this averaging efficiently without simulating every individual realization, particularly for quantities linear in the time-evolved state (expectation values).

2. Core Methodology

The authors propose a framework that exploits emergent symmetries that exist at the level of the disorder-averaged state, even if they are broken in individual realizations.

A. Effective Dynamical Map

Instead of averaging expectation values after time evolution, the authors average the time-evolution operator itself. They define an effective dynamical map $\Lambda_t$ acting on the initial density matrix $\rho_0$:
$$\rho(t) = \Lambda_t[\rho_0] = \int dp(\lambda) U_\lambda(t) \rho_0 U_\lambda(t)^\dagger$$
If the probability distribution of the disorder parameters is invariant under a symmetry group $G$ (e.g., permutations), the effective map $\Lambda_t$ commutes with the symmetry operations. Consequently, if the initial state is symmetric, the effective state $\rho(t)$ remains within the symmetric subspace for all time.

B. Symmetry-Adapted Basis

For systems with average permutation invariance (like all-to-all random couplings), the dimension of the symmetric subspace scales polynomially with $N$ (specifically $\sim N^3$ for density matrices), rather than exponentially. The authors utilize a basis of symmetric Pauli strings ($\Sigma_{x,y,z}$) to represent the density matrix and the superoperator $\Lambda_t$ within this reduced space.

C. Perturbative Construction via Differential Operators

Constructing $\Lambda_t$ exactly is still difficult. The authors introduce a method to construct it perturbatively using differential operators derived from the characteristic function $\phi(k)$ of the disorder distribution:
$$\bar{f} = \mathcal{D}f = \phi(-i\nabla_\mu) f(\mu) \big|_{\mu=0}$$
This allows the disorder average to be computed by applying derivatives to the unitary evolution operator, avoiding explicit integration over the disorder distribution.

D. Regularization Schemes

Truncating perturbative expansions (in time or disorder strength) often leads to unphysical behavior (e.g., divergence or non-positive density matrices) at long times. The authors propose two regularization strategies to extend the validity of these expansions:

1. Short-Time Expansion (Lindbladian): Instead of expanding $\Lambda_t$, they expand the effective Lindbladian generator $\mathcal{L}_t = \dot{\Lambda}_t \Lambda_t^{-1}$. This naturally incorporates dissipation and dephasing, preventing unbounded growth.
2. Weak-Disorder Expansion: Expanding in powers of the disorder strength $\sigma$. To control long-time behavior, they propose two regularizations:
   • Exponential Regularization: $\Lambda_t \approx \bar{U}_t \exp(-\sigma^2 \mathcal{O}(t))$.
   • Inverse Regularization: $\Lambda_t \approx \bar{U}_t (1 + \sigma^2 \mathcal{O}(t))^{-1}$.

3. Key Contributions

• Restoration of Symmetry: Demonstrated that disorder averaging restores global symmetries (like permutation invariance) at the level of the dynamical map, enabling polynomial scaling of computational complexity.
• Differential Operator Formalism: Developed a rigorous mathematical tool to perform disorder averaging via differentiation of the characteristic function, simplifying the derivation of weak-disorder expansions.
• Regularized Perturbative Schemes: Introduced specific regularization techniques (Lindbladian expansion, exponential, and inverse) that allow perturbative results to remain accurate and physical well beyond their nominal convergence radii.
• Scalable Simulation: Enabled the simulation of disorder-averaged dynamics for system sizes ($N$) far beyond the reach of exact diagonalization or Monte Carlo averaging.

4. Results and Benchmarks

The methods were benchmarked on a Transverse-Field Ising Model (TFIM) with random all-to-all couplings (drawn from a normal distribution) and the Sherrington-Kirkpatrick (SK) model.

• Short-Time Expansion:
  • For the TFIM, the Lindbladian expansion up to order $O(t^3)$ accurately reproduced exact Monte Carlo results for times $t \ll \epsilon_{max}^{-1}$.
  • The regularization effectively prevented the divergence of observables seen in naive truncations.
• Weak-Disorder Expansion:
  • Applied to the TFIM with small disorder strength $\sigma$.
  • Exponential regularization proved superior to inverse regularization, maintaining agreement with exact results up to $t \approx \sigma^{-1}$, whereas inverse regularization deviated earlier.
  • The method successfully captured the decay of magnetization (decoherence) and the $1/N$ scaling of variance.
  • Simulations were successfully performed for $N=40$, a size where exact disorder averaging (requiring $\sim 1000$ shots of full Hilbert space evolution) is computationally infeasible.
• Late-Time Behavior:
  • The weak-disorder expansion correctly predicted time-averaged late-time values (diagonal ensemble) for various initial states and field configurations, outperforming Monte Carlo averaging which suffers from statistical noise in approaching constant values.

5. Significance and Impact

• Efficiency: The approach reduces the computational cost of simulating disordered quantum dynamics from exponential to polynomial in $N$ for permutation-invariant ensembles.
• Experimental Relevance: The method is directly applicable to experimental platforms with inherent positional randomness, such as cold atomic gases and color centers in diamond, where individual spins cannot be consistently labeled across shots, making observables naturally permutation-invariant.
• Theoretical Bridge: The authors draw a conceptual parallel between their formalism and Quantum Field Theory (QFT). Just as the QFT effective action arises from a coherent sum over histories, the disorder-averaged Lindbladian arises from an incoherent sum over disorder realizations. This suggests potential for transferring QFT techniques (like Renormalization Group methods) to disordered quantum systems.
• Versatility: The technique is model-independent and does not strictly require an analytically solvable point for the weak-disorder expansion, as numerical differentiation can be employed.

In summary, this work provides a powerful toolkit for simulating the dynamics of large, disordered quantum systems by leveraging the symmetries that emerge only after statistical averaging, overcoming the "curse of dimensionality" and the "curse of averaging."