⚛️ quantum physics

Fast mixing of all-to-all quantum systems at high temperatures

This paper demonstrates that arbitrary quantum $k$ -local Hamiltonians with bounded interactions exhibit fast mixing and admit fully-polynomial time quantum approximation algorithms for partition functions and global expectation values at sufficiently high temperatures, thereby extending existing fast-mixing results beyond geometrically-local settings.

Original authors: Thiago Bergamaschi

Published 2026-06-25

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Thiago Bergamaschi

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 Quantum Party with No Walls

Imagine a massive party where every single guest can talk to every other guest instantly. There are no walls, no corners, and no "local" groups. In the world of quantum physics, this is called an "all-to-all" system. Usually, physicists find these systems incredibly hard to study because information spreads everywhere at once, making it impossible to predict how the system behaves.

Most previous research focused on systems where guests are arranged in a grid (like a lattice), where you can only talk to your immediate neighbors. In those "neighborly" systems, we know that if the room is hot enough (high temperature), the guests eventually calm down and settle into a predictable state of chaos called thermal equilibrium (or a "Gibbs state").

The Problem:
Until now, no one could prove that this "settling down" happens quickly in those chaotic, all-to-all systems. Without this proof, we couldn't be sure if quantum computers could efficiently simulate these systems to calculate things like total energy or probability.

The Solution:
This paper proves that yes, even in these chaotic, all-to-all systems, if the temperature is high enough, the system settles down incredibly fast. It happens so fast that the time it takes doesn't depend on how many guests (particles) are at the party.

The Key Concepts (Translated)

1. The "Mixing" Time

Think of a drop of ink dropped into a glass of water.

Slow mixing: The ink takes hours to spread evenly.
Fast mixing: The ink swirls and spreads evenly in seconds.
In quantum systems, "mixing" is how fast the system forgets its initial state and reaches thermal equilibrium. The authors prove that at high temperatures, these all-to-all systems are fast-mixers.

2. The "Ghost" Generator (The Pseudo-Lindbladian)

To prove this, the authors used a clever trick. They couldn't analyze the real, complex quantum machine directly because it was too messy. So, they built a simplified, "ghost" version of the machine.

The Real Machine: A complex engine with many moving parts that is hard to measure.
The Ghost Machine: A simplified model that acts like the real one but has a very neat, predictable structure.
The authors showed that if the "Ghost Machine" settles down quickly, the "Real Machine" must also settle down quickly. This is like proving a car will drive fast by showing that a simplified blueprint of its engine works perfectly.

3. The "Cluster Expansion" (The Ripple Effect)

To understand how the simplified machine works, the authors looked at how information ripples through the system over a tiny, imaginary slice of time (complex time).

The Analogy: Imagine dropping a pebble in a pond. Usually, the ripples spread out in a circle. But in an all-to-all system, the ripples could theoretically hit everyone at once.
The Discovery: The authors proved that at high temperatures, these ripples don't actually hit everyone at once. Instead, they form small, manageable "clusters" of interaction. Even though everyone can talk to everyone, the strength of the conversation drops off so quickly that you can treat the system as if it were made of small, independent groups. This allows them to do the math that was previously impossible.

4. The "Dobrushin Condition" (The Rule of Influence)

The paper uses a mathematical rule called the Dobrushin condition.

The Analogy: Imagine a room full of people. The rule asks: "If one person changes their mind, how much does it influence everyone else?"
The Result: At high temperatures, the authors proved that one person's change of mind has a very small influence on the rest of the room. No single person can dominate the conversation. Because the influence is weak and spread out, the whole room reaches a consensus (equilibrium) very quickly.

What This Means for the Future (According to the Paper)

The paper does not claim this will immediately cure diseases or build faster internet. It makes two specific, technical claims about what is now possible:

Faster Quantum Algorithms: Because the system mixes so fast, we can now design quantum computer algorithms that calculate the partition function (a complex number representing the total energy states of the system) and global expectation values (average properties of the system) in a reasonable amount of time.
System-Size Independence: The speed of these calculations does not get slower just because you add more particles to the system. Whether you have 10 particles or 1,000,000, the time it takes to reach equilibrium remains roughly the same (as long as the temperature is high enough).

Summary

In short, this paper solves a long-standing puzzle in quantum physics. It proves that even in the most chaotic, interconnected quantum systems, heat acts as a stabilizer. If you make the system hot enough, the chaos organizes itself rapidly, allowing us to simulate and understand these complex systems efficiently on a quantum computer.

Technical Summary: Fast Mixing of All-to-All Quantum Systems at High Temperatures

Problem Statement
The paper addresses the complexity of thermal state preparation for arbitrary $k$ -local quantum Hamiltonians with all-to-all (non-geometric) interactions. While efficient quantum Gibbs sampling algorithms exist for geometrically local systems (lattices) at high temperatures, extending these results to all-to-all systems has remained an open challenge. Existing proofs of fast mixing for non-commuting systems rely fundamentally on Lieb-Robinson bounds, which quantify information propagation via a "lightcone" structure inherent to lattice geometries. In all-to-all systems, such geometric locality is absent, and the standard notion of a lightcone fails to describe operator growth, rendering previous mixing-time analyses inapplicable. The central problem is to determine whether arbitrary all-to-all $k$ -local Hamiltonians with bounded interaction strength admit a system-size independent spectral gap for their thermal Lindbladian dynamics at sufficiently high temperatures.

Methodology
The authors develop a novel analytical framework that bypasses the need for Lieb-Robinson bounds by leveraging the convergence of complex-time evolution and cluster expansions. The proof strategy proceeds through four main technical stages:

Cluster Expansions for Complex-Time Evolution:
The authors establish a cluster expansion for the complex-time evolution of local operators, $A(z) = e^{izHA}e^{-izH}$ , where $z \in \mathbb{C}$ . Based on the work of Netočny and Redig [NR04] and its algorithmic version [MH21], they show that for sufficiently small $|z|$ (corresponding to high temperatures), the evolution of a local operator can be decomposed into a linear combination of connected clusters of interactions. This expansion quantifies the "quasi-locality" of the evolution, demonstrating that the influence of a local operator spreads to other sites in a controlled manner, decaying with the product of interaction strengths within the cluster.
The Pseudo-Lindbladian and Dobrushin Conditions:
To circumvent the lack of strict locality in the original Lindbladian generator $\mathcal{L}$ , the authors introduce a "pseudo-Lindbladian" generator $\mathcal{K}$ , inspired by recent work on 1D systems [BC26]. $\mathcal{K}$ is constructed from single-site Pauli jumps and possesses explicit locality properties absent in $\mathcal{L}$ , though it does not generate a completely positive trace-preserving (CPTP) map.
The authors establish sufficient conditions for $\mathcal{K}$ to admit a spectral gap using non-commutative versions of the Dobrushin conditions. These conditions require:
- Locality in Temperature: The generator at inverse temperature $\beta$ must be a small perturbation of the infinite-temperature (depolarizing) generator.
- Pairwise Influence: Updates on one site must not simultaneously influence all other sites (a mean-field type condition).
  Using the cluster expansion results, the authors prove that these conditions hold at high temperatures, implying that $\mathcal{K}$ has a system-size independent spectral gap.
KMS Oscillator Norm and Spectral Gap for $\mathcal{K}$ :
A key technical innovation is the introduction of the "KMS oscillator norm," a modification of the standard oscillator norm adapted to the KMS inner product. The authors prove that the decay of this norm under the dynamics generated by $\mathcal{K}$ implies a Poincaré inequality (spectral gap). They demonstrate that the Dobrushin conditions derived from the complex-time evolution bounds ensure this decay.
Dirichlet Form Comparison:
Finally, the authors relate the spectral gap of the pseudo-Lindbladian $\mathcal{K}$ to the original Lindbladian $\mathcal{L}$ (from [CKG23]). They perform a Dirichlet form comparison, showing that the dissipation rate of $\mathcal{L}$ is lower-bounded by that of $\mathcal{K}$ up to a temperature-dependent constant. This step involves "peeling off" the time and frequency dependence of the operator Fourier transform in the Lindbladian's Dirichlet form, utilizing the convergence of complex-time evolution to bound the error terms.

Key Contributions and Results

Main Theorem (Theorem 1.3): The paper proves that for arbitrary all-to-all $k$ -local Hamiltonians with bounded interaction strength, there exists a universal constant $c$ and a threshold inverse temperature $\beta_c$ such that for all $\beta < \beta_c$ , the [CKG23] Lindbladian admits a spectral gap that is independent of the system size $n$ .
Algorithmic Implications (Corollary 1.4): As a direct consequence, the paper establishes the existence of fully-polynomial time quantum approximation schemes (FPTAS) for estimating partition functions and global expectation values in these systems at high temperatures. The mixing time is shown to be $O(n + \log(1/\epsilon))$ , enabling efficient state preparation on quantum computers.
Generalization: This work generalizes existing fast-mixing results beyond the geometrically-local setting, addressing a significant gap in the literature where prior proofs were restricted to lattice Hamiltonians.

Significance and Claims
The paper claims to provide the first unconditional proof of system-size independent spectral gaps for quantum Markov chains describing arbitrary all-to-all interacting systems at high temperatures. The authors position their work as a stepping stone toward understanding mixing times in more complex disordered systems, such as quantum spin glasses and the Sachdev-Ye-Kitaev (SYK) model.

However, the paper maintains a modest tone regarding its scope. The authors explicitly note that their techniques are currently insufficient to address random quantum Hamiltonians (spin glasses), drawing a parallel to the historical gap between lattice mixing proofs and spin glass proofs in classical statistical mechanics. Furthermore, while the results imply efficient quantum algorithms, the authors acknowledge that their methods do not yet yield improved classical simulation algorithms for these systems, though they suggest the derived bounds could potentially be leveraged for such purposes in future work. The work is presented as a foundational step in extending the theory of quantum mixing to non-geometric systems, relying on the interplay between complex-time operator growth and non-commutative Dobrushin conditions.