Efficient and simple Gibbs state preparation of the 2D toric code via duality to classical Ising chains

This paper introduces polynomial-depth duality transformations to efficiently prepare Gibbs states for quantum Hamiltonians, such as the 2D toric code, by mapping them to dual classical systems like Ising chains while preserving key mixing properties under Lindbladian dynamics.

The Gist

The Big Idea: Translating a Hard Puzzle into an Easy One

Imagine you are trying to solve a incredibly complex, tangled knot of string (representing a difficult quantum system). You need to understand how this knot behaves when it gets "hot" (reaching thermal equilibrium, or a Gibbs state). Usually, untangling this knot to see its behavior requires a supercomputer and takes a very long time.

The authors of this paper discovered a clever "translation trick." They found a way to take that complex, tangled quantum knot and, using a specific set of rules (a quantum circuit), transform it into a completely different shape: two simple, straight lines of beads (representing classical Ising chains).

Once the knot is transformed into these simple lines, it becomes incredibly easy to predict how they behave. The paper proves that if you can solve the simple lines, you automatically know the answer for the original complex knot.

The Key Concepts

1. The "Poly-Depth" Translator
The authors introduce a new type of translator called a "poly-depth duality."

• The Metaphor: Think of a complex quantum system as a high-security, encrypted file. To read it, you usually need a massive, slow decryption key.
• The Innovation: The authors found a "translator" (a quantum circuit) that is efficient enough to run on a computer (it doesn't take forever). This translator converts the encrypted quantum file into a plain-text document (a classical model) that anyone can read instantly.
• The Catch: The translator changes the look of the system completely. It destroys the "topological" features (like the knot's shape) and turns it into something that looks like a simple chain of magnets. But, crucially, it keeps the "temperature behavior" exactly the same.

2. The Star and the Square (The Toric Code)
The paper focuses on a famous quantum model called the 2D Toric Code.

• The Setup: Imagine a grid of spins (tiny magnets) arranged on a donut shape. The rules of this system involve "Star" operators (magnets meeting at a point) and "Plaquette" operators (magnets forming a square).
• The Result: The authors proved that for any size of this grid, you can use their translator to split this complex 2D grid into two separate, one-dimensional chains of magnets that don't talk to each other.
• Why it matters: Calculating the behavior of a 2D grid is hard. Calculating the behavior of a 1D line is easy. Because the translator is efficient, we can now prepare the "Gibbs state" (the equilibrium state) of the 2D grid just as fast as we can for the 1D line.

3. The "Mixing Time" Guarantee
The paper also looks at how fast these systems settle down.

• The Metaphor: Imagine dropping a drop of ink into a glass of water. "Mixing time" is how long it takes for the ink to spread evenly.
• The Discovery: The authors showed that if you use their translator to switch from the complex system to the simple one, the "mixing speed" stays the same. If the simple chain mixes fast, the complex quantum knot also mixes fast. This means we can trust that our new method works quickly and reliably.

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

• Efficiency: For the 2D Toric Code, the authors provide a recipe to prepare the equilibrium state in a time that does not depend on the temperature. Previous methods got slower and slower as the temperature dropped; this new method stays fast.
• Beyond 2D: The authors tested their translator on other complex models (like the 3D Toric Code and Haah's Code) using computer simulations. The results suggest these complex models can also be translated into simple classical models, though they haven't mathematically proven it for every possible size yet (they have a "Conjecture" that it holds true).
• Classical vs. Quantum: Because the final result is a simple classical model, you don't actually need a quantum computer to simulate the sampling part. You can do the heavy lifting on a regular classical computer, then just apply the translator circuit at the very end.

Summary

The paper introduces a "magic lens" (poly-depth duality) that turns hard, tangled quantum problems into easy, straight-line classical problems. By proving this works for the 2D Toric Code, they have created a fast, efficient way to simulate how these quantum systems behave at any temperature, solving a problem that was previously much harder to tackle.

Technical Summary

Technical Summary: Efficient and Simple Gibbs State Preparation of the 2D Toric Code via Duality to Classical Ising Chains

Problem Statement
The preparation of Gibbs states, $\sigma(\beta) = e^{-\beta H}/\text{Tr}[e^{-\beta H}]$, for quantum Hamiltonians is a fundamental challenge in quantum simulation. While algorithms based on dissipation (Lindbladians) have shown promise, their efficiency relies on two conditions: the associated Lindbladian must thermalize rapidly (e.g., possess a spectral gap or Modified Logarithmic Sobolev Inequality, MLSI), and the dynamics must be efficiently implementable on a quantum computer. For topologically ordered systems like the 2D Toric Code, previous results established the existence of a spectral gap and positive MLSI for the Davies generator, yet the resulting sampling algorithms often suffered from high complexity scaling with inverse temperature $\beta$ (e.g., $\tilde{O}(N^3 \exp(\beta))$) or required cumbersome implementations of Lindbladian evolution.

Methodology: Poly-Depth Duality
The authors introduce the concept of polynomial-depth (poly-depth) duality. Two families of Hamiltonians $\{H_\Lambda\}$ and $\{\tilde{H}_\Lambda\}$ are defined as poly-depth dual if there exists a unitary $U_\Lambda$, implementable by a quantum circuit of polynomial depth in the system size $|\Lambda|$, such that $H_\Lambda = U_\Lambda \tilde{H}_\Lambda U_\Lambda^\dagger$.

The core theoretical tool is Lemma 1, which establishes that if a Hamiltonian $H$ is poly-depth dual to a Hamiltonian $\tilde{H}$, and if an efficient Gibbs sampler exists for $\tilde{H}$, then an efficient Gibbs sampler exists for $H$. The sampler for $H$ is constructed by applying the dual unitary $U_\Lambda$ to the output of the sampler for $\tilde{H}$. Crucially, if $\tilde{H}$ is a classical Hamiltonian, the sampling step can be performed entirely classically, with the quantum circuit $U_\Lambda$ serving only as a post-processing step.

The authors extend this notion to Lindbladians, proving that properties such as the uniqueness of fixed points, spectral gaps, MLSI, and mixing times are preserved under conjugation by unitaries. This implies that if a Lindbladian $\mathcal{L}$ is fast-mixing, its dual $\tilde{\mathcal{L}}$ is also fast-mixing.

Key Contributions and Results

1. 2D Toric Code Duality:
   The primary result is a formal proof that the 2D Toric Code Hamiltonian is poly-depth dual to two decoupled classical 1D Ising chains for any system size $L \times L$.

   • Circuit Construction: The authors provide an explicit construction of a quantum circuit $C$ composed of $O(L^3)$ CNOT (CX) gates and $O(L^2)$ Hadamard gates.
   • Mapping: This circuit maps the star operators ($A_v$) of the Toric Code to one Ising chain and the plaquette operators ($B_p$) to a second, disjoint Ising chain. Two spins remain non-interacting, corresponding to the logical subspace of the code.
   • Efficiency: Since 1D Ising chains can be sampled efficiently (independent of $\beta$) via simple iterative procedures, this yields an efficient Gibbs sampler for the 2D Toric Code. The gate complexity is $O(L^3)$ (or $O(N^{3/2})$ where $N$ is the number of qubits), which is independent of $\beta$. This improves upon prior dissipative approaches that scaled exponentially with $\beta$.

2. Generalization to Other Models:
   The authors propose an algorithm (Algorithm 1) that takes a commuting Pauli Hamiltonian as input and outputs a circuit mapping it to a classical model. Using this, they provide computer-assisted proofs (verified up to system sizes $L=90$ for 2D and $L=20$ for 3D) for poly-depth dualities between several other models and classical systems:

   • Rotated Surface Code: Dual to a non-interacting single-spin Hamiltonian.
   • 2D Color Code: Dual to decoupled "lasso" Ising chains or non-interacting spins.
   • Haah's Code: Dual to two decoupled Ising chains.
   • X-cube Model: Dual to $L$ decoupled Ising chains and $L-1$ 1D decoupled nearest-neighbor systems.
   • Subsystem Toric Codes: Dual to decoupled Ising chains.
   • 3D Toric Code: Dual to an Ising chain decoupled from a classical local model with bounded-degree interaction graphs.

3. Conjecture and Sampling Limits:
   The authors conjecture that these dualities hold for arbitrary system sizes. If true, this provides a constructive method for efficient Gibbs sampling for these models.

   • For most models (2D Toric, Color, Haah's, X-cube), efficient sampling is claimed for all $\beta < \infty$.
   • For the 3D Toric Code, efficiency is claimed only for $\beta < \beta_0$ (high temperature). The authors note that at low temperatures, phase transitions in the 3D Toric Code are linked to computational hardness, suggesting that efficient sampling may not be possible in that regime.

Significance and Claims
The paper claims that by leveraging poly-depth duality, one can bypass the difficulties of directly simulating complex, topologically ordered quantum systems. Instead, the problem is reduced to sampling from physically simpler classical models.

• Practicality: The proposed method allows for Gibbs state preparation without implementing complex Lindbladian dynamics. The quantum circuit depth is polynomial, and for the 2D Toric Code, the circuit size for small systems (e.g., $7 \times 7$ lattice) remains below 1000 gates, making it potentially testable on near-term noisy quantum computers where the sampling step is handled classically.
• Theoretical Scope: The work demonstrates that efficient sampling is preserved under poly-depth unitary conjugation, providing a new class of examples for systems satisfying rapid mixing (positive MLSI/spectral gap).
• Limitations: The authors are modest regarding the 3D Toric Code, acknowledging that the efficiency at low temperatures is not guaranteed and likely coincides with the onset of computational hardness. Furthermore, the general dualities for models other than the 2D Toric Code are currently supported by computer-assisted verification up to finite sizes, with a formal proof for arbitrary sizes left as an open problem.

In summary, the paper establishes a framework where the complexity of preparing Gibbs states for certain quantum Hamiltonians is reduced to the complexity of their classical duals, offering a more efficient and practical route for state preparation, particularly for the 2D Toric Code.