arXiv:2610.26520
Exploring ground state phase diagrams of Rydberg atom arrays with finite projected entangled-pair states
Konrad Schröder, Erik Lennart Weerda, Matteo Rizzi
Abstract
Rydberg atom arrays (RAA) have been advocated as a promising avenue for the quantum simulation of different exotic quantum states. Numerical predictions for the ground state properties of this experimental platform can, however, be challenging in two-dimensional setups. In particular, faithfully treating long-range interactions is often crucial for distinguishing the fate of competing phases, but it usually comes at considerable cost. Projected entangled-pair states (PEPS), optimized via a sampling-based approach, have been shown to deal with such terms on the same ground as shorter-range ones. In this work, we employ these tools to investigate debated physical questions in different frustrated geometries of RAA, from the emergence of spin-glass physics on the Kagomé lattice to topological order beyond the PXP model on the Ruby lattice. By that, we contribute to establishing PEPS as well-suited tools for creating digital twins of RAA experiments.
AI Summary
The paper asks whether proposed glassy and topologically ordered ground states of frustrated Rydberg arrays survive realistic interactions and open boundaries. It uses finite PEPS optimized by sampling-based imaginary-time evolution, retaining the full diagonal $1/r^6$ interaction without a distance cutoff. Triangular-lattice calculations benchmark the method against known disordered and density-wave phases. For Kagomé arrays, simulations use 432 atoms and bond dimension $D=4$. Nematic and staggered ordering are reproduced. In the intervening regime, the Edwards–Anderson density-inhomogeneity parameter becomes vanishingly small after excluding boundary layers, supporting a homogeneous disordered state rather than a stable glass under the studied conditions. For the ruby lattice, the constrained PXP model exhibits an intermediate $\mathbb{Z}_2$ liquid, diagnosed through suppressed electric and magnetic Fredenhagen–Marcu parameters and a Kitaev–Preskill second-Rényi topological contribution $\gamma/\log 2=0.91\pm0.12$. With full interactions and blockade radii $R_b\in[2.3,2.5]$, no intervening topological regime is found; measured $\gamma\in[-0.1,0.04]$ is consistent with zero. The liquid survives interactions truncated at distance 2 but is absent for cutoffs 3, 4, and infinity. A one-parameter, bond-dimension-2 PEPS also approximates PXP-liquid energies and diagnostics. The conclusions concern accessible finite systems and sampled parameter regions, not thermodynamic-limit exclusions. Entropy estimates are limited by subsystem size and sampling uncertainty. Gauge-based string diagnostics become unreliable for $R_b\leq2.2$ because blockade violations matter. Thermodynamic stability of the one-parameter ansatz’s topological order remains unresolved.
Error found: The broad claim that realistic interactions destabilize Ruby-lattice topological order extends beyond the explicitly diagnosed full-interaction parameter window R_b=2.3–2.5. The absence of systematic system-size and bond-dimension convergence, particularly for the Ruby calculations, limits conclusions about stable phases beyond the simulated finite systems. The Kagomé Edwards–Anderson diagnostic vanishes for uniform one-point densities even in symmetry-restored superpositions of frozen configurations, so it and the absence of Bragg peaks do not uniquely establish an ordinary disordered phase.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.