Towards $2+1$D quantum electrodynamics on a cold-atom quantum simulator

TL;DR

This work addresses the challenge of simulating $(2+1)$-D $\mathrm{U}(1)$ lattice gauge theory with dynamical matter beyond minimal truncations. It proposes a spin-$S=1$ quantum-link model realized in a two-layer tilted Bose–Hubbard setup, mapped to a bosonic lattice and stabilized by a linear gauge-protection term derived from quantum Zeno dynamics. Second-order perturbation theory yields an effective QLM from the extended Bose–Hubbard with resonance conditions that suppress unwanted processes, and iMPS simulations confirm faithful real-time dynamics with gauge violations $\eta(t)$ below $0.1\%$ on accessible times. The results extend the experimental reach to larger gauge-field truncations and higher dimensions, enabling exploration of phenomena like string breaking and glueball formation in a controllable quantum simulator. The approach is compatible with current ultracold-atom platforms and opens avenues toward $(3+1)$-D extensions and fermionic matter implementations.

Abstract

Cold atoms have become a powerful platform for quantum-simulating lattice gauge theories in higher spatial dimensions. However, such realizations have been restricted to the lowest possible truncations of the gauge field, which limit the connections one can make to lattice quantum electrodynamics. Here, we propose a feasible cold-atom quantum simulator of a $(2+1)$-dimensional U$(1)$ lattice gauge theory in a spin $S=1$ truncation, featuring dynamical matter and gauge fields. We derive a mapping of this theory onto a bosonic computational basis, stabilized by an emergent gauge-protection mechanism through quantum Zeno dynamics. The implementation is based on a single-species Bose--Hubbard model realized in a tilted optical superlattice. This approach requires only moderate experimental resources already available in current ultracold-atom platforms. Using infinite matrix product state simulations, we benchmark real-time dynamics under global quenches. The results demonstrate faithful evolution of the target gauge theory and robust preservation of the gauge constraints. Our work significantly advances the experimental prospects for simulating higher-dimensional lattice gauge theories using larger gauge-field truncations.

Figures (8)

• Figure 1: Selection of Gauss’s-law-allowed matter and electric-field configurations in the QLM and their corresponding representations in the BHM. Left- and downward-pointing arrows indicate electric-flux operator eigenvalues $\hat{s} _{\vb{r}} ^{z} = -1$, right- and upward-pointing arrows correspond to $\hat{s} _{\vb{r}} ^{z} = 1$, and links without arrows denote $\hat{s} _{\vb{r}} ^{z} = 0$. (a,b) QLM configurations with unoccupied matter sites and their bosonic counterparts. (c,d) QLM configurations with an occupied matter site and the corresponding bosonic representation. All other configurations consistent with Gauss’s law can be generated by applying rotations and reflections to these vertices.
• Figure 2: Configurations of the QLM on the unit cell that are connected via second-order processes in the extended BHM. To restrict the quantum simulator to the gauge-invariant subspace, these states are tuned into mutual resonance.
• Figure 3: Undesired second-order processes in the mapping of the $2 + 1$D spin-$1$$\mathrm{U}(1)$ QLM onto a two-dimensional Bose--Hubbard optical superlattice, where the shallow well represents a matter site, and the deep well one of its neighboring gauge sites. These are undesired as the configurations on the right do not correspond to any valid QLM configuration. On the right, we show the corresponding perturbative contributions to the matrix elements, demonstrating that these processes renormalize the mass parameter $\mu$. The list is not exhaustive.
• Figure 4: Schematic diagram of the proposed cold-atom simulator architecture. The upper layer hosts the matter sites, while the lower layer contains the gauge and forbidden sites. The polarisation direction in the upper layer is chosen to be $d=\frac{1}{\sqrt{3}}(1,1,1)$.
• Figure 5: Numerical simulations of quenches from the vacuum state in our proposed quantum simulator of the $(2+1)$D spin-1 QLM, showing the evolution of the chiral condensate in both the Bose--Hubbard simulator (solid curves) and the target QLM (dashed curves). All simulations are performed on an infinite-length cylinder with a width of $N_y = 2$ matter sites, corresponding to four bosonic sites in the BHM. (a) Quench dynamics for different values of the staggering potential $\delta$, which affects the mass $\mu$ in the target model. The on-site interaction strength is fixed to $U=160\,\text{Hz}$, while $\delta$ is set to $79.64$, $79.59$, $79.54$, and $79.49\,\text{Hz}$ (top to bottom). (b) Quench dynamics for different values of the electric field energy $g^2$, with $\mu$ fixed to approximately $0.81\kappa$. This is achieved by simultaneously varying the BHM interaction strength $U$ to $160$, $159.95$, and $159.9\,\text{Hz}$ (top to bottom) and the corresponding staggering potentials $\delta$ to $79.54$, $79.465$, and $79.39\,\text{Hz}$, respectively.