arXiv:2610.10357
Quantum simulation of the Heisenberg XXZ model on a Rydberg atom array
Matthias Werner, Artur García-Sáez
Abstract
A major application of analog quantum hardware is quantum simulation, where a system evolves over time according to a given Hamiltonian. One challenge when implementing the Hamiltonian on a programmable quantum simulator is the availability of distinct coupling types, many of which go beyond the conventional ZZ-couplings found in Ising Hamiltonians that are naturally implemented by various platforms. In prior work, we have developed a method based on domain wall encoding to simulate one-dimensional spin chains using only Ising Hamiltonians, which would otherwise require XX+YY-couplings as well. Here, we implement the method on a commercially accessible Rydberg atom quantum device, and simulate the time evolution of the Heisenberg XXZ model with various anisotropies. We successfully probe the XY and Ising phases of the model and reproduce the qualitative behavior of several one- and two-body observables in both phases, as well as at the critical point. Our work augments the toolbox to simulate canonical quantum many-body systems on analog quantum simulators.
AI Summary
The paper asks whether an Ising-interacting Rydberg array can simulate XXZ-chain dynamics, including interacting excitations and large anisotropy, without Floquet engineering. It implements a previously proposed domain-wall encoding on Aquila: strong nearest-neighbor antiferromagnetic couplings approximately conserve domain-wall number, a transverse drive moves the walls, and next-nearest-neighbor interactions generate their longitudinal coupling. A zigzag geometry controls the interaction ratio; residual boundary fields become local fields in the encoded XXZ chain. The experiment uses 12 atoms to represent 13 logical sites with three excitations, initialized as an adjacent pair and a separated excitation. Evolution is measured for up to $3\,\mu\mathrm{s}$ at anisotropies $\delta=0,1,4$, using 300 shots per time point. Site populations, two-site correlations, and the number of unaligned logical neighbors qualitatively distinguish excitation spreading at small anisotropy from bound-pair localization at $\delta=4$. Quantitative deviations include substantially inhibited spreading. Exact numerical simulations compare the target XXZ dynamics with the full Rydberg Hamiltonian and calibration-based noise models. Without hardware noise, post-selected state fidelities remain above $0.75$ throughout the simulated interval. Atomic-position disorder is the dominant modeled error and explains much of the experimental suppression of transport. Finite encoding strength produces a worst-case Hamiltonian error $\mathcal{O}(J^{-1})$; the experiment uses $J=5/3$. Post-selection discards approximately 50–70% of shots. Stated limitations include geometry and control constraints, restricted excitation-number parity, and omitted decoherence channels such as Rydberg decay, atom loss during evolution, and drive inhomogeneity. The measurements do not establish experimental quantum-state fidelity.
Error found: The abstract's qualitative-agreement claim omits that the reported results use excitation-number post-selection discarding approximately 50–70% of shots, without which the paper reports that some key observables fail to reproduce the model's qualitative behavior.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 14 pages, 10 figures