arXiv:2610.24091
Trapped-ion quantum simulation of (1+1)D SU(2) lattice gauge theory via loop-string-hadron formulation
Toshiaki Kaji, Jesse R. Stryker, Yutaro Iiyama, Lento Nagano, Enrico Rinaldi, Christian W. Bauer, Koji Terashi
Abstract
We present a quantum simulation of the SU(2) lattice gauge theory in 1+1 dimensions with one flavor of fermions, performed using the trapped-ion Quantinuum System Model H2. This work is based on the loop-string-hadron (LSH) formulation of the gauge theory, and is distinguished by using qubits to represent both bosonic-gauge and fermionic-matter degrees of freedom. Furthermore, we access the $j=1$ electric-flux sector beyond the $j_{\max}=1/2$ truncation commonly used in SU(2) quantum simulations. To that end, we digitize an electric-field degree of freedom using a multiqubit register for representing the bosonic quantum number. Our simulation therefore synthesizes several key ingredients relevant to quantum simulations in higher spatial dimensions, in particular explicit local gauge-field degrees of freedom represented here by a multiqubit bosonic register, while relying solely on circuit-construction optimizations that are expected to scale polynomially with lattice size. This work demonstrates that modern quantum hardware is increasingly able to accommodate the features of complicated but physically important lattice gauge theories while successfully reproducing the expected Trotterized dynamics.
AI Summary
The paper examines whether current trapped-ion hardware can simulate non-Abelian lattice gauge dynamics while explicitly retaining qubit-encoded matter and local gauge fields beyond the usual $j_{\max}=1/2$ electric-flux truncation. It implements a four-site, $(1+1)$-dimensional SU(2) theory with one fermion flavor using the loop-string-hadron formulation on Quantinuum H2-2. Boundary conditions permit a three-level bosonic occupation register, $n_l\in\{0,1,2\}$, encoded in two qubits, and the initial state accesses $j=1$. The implementation uses 13 qubits, including one ancilla, and first-order Trotter circuits containing 226 native two-qubit gates per step. Boundary-aware simplifications exploit the remaining Abelian Gauss-law constraints and alternative decompositions of hopping operators. Hardware runs measure electric-energy densities on three links for up to six steps, with $\Delta t=1.25$ and total time $t=7.5$. Against noiseless versions of the same Trotter circuits, the aggregate RMS deviation is $0.267\pm0.016$ without mitigation, $0.108\pm0.025$ with observable-specific depolarizing rescaling, and $0.067\pm0.011$ with physical-constraint post-selection. Post-selection discards 74% of shots at maximum depth. These benchmarks establish agreement with discrete Trotterized dynamics, not accurate continuous-time evolution: the timestep produces sizable quantitative deviations from exact dynamics. The system is small and classically benchmarkable, and hardware depth is budget-limited. Post-selection misses constraint-preserving errors and can incur exponentially growing sampling overhead with system size; rescaling can amplify uncertainty or worsen individual estimates. An emulator study finds timestep extrapolation unreliable under the available noise and statistics.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 23 pages, 12 figures