A quantum processor watches electric strings spread

Why follow a string?

Electric-flux strings are a useful way to study confinement: in a confined regime, separating charges costs energy that grows with their distance. But a string is not necessarily fixed in place. Charges can move, matter–antimatter-like pairs can form along the string, and magnetic interactions can rearrange the flux sideways across the lattice. These processes compete, and their balance changes with the model’s parameters.

The study by Sanziana Stelea and colleagues asks how these strings evolve in a two-dimensional, or (2+1)D, lattice gauge theory. Their model has Z₂ gauge fields, dynamical Ising matter and an explicit magnetic interaction around each square plaquette. It spans confined and deconfined regimes. “Deconfined” here describes a regime in which flux and charges can spread more freely; it does not mean that the experiment directly studies real quarks in nature.

Building the simulation

The team encoded matter on lattice sites and gauge fields on links, using separate superconducting qubits for the two kinds of degrees of freedom. The simulated lattice had 5 × 4 matter sites and occupied 63 qubits on a superconducting processor, including 12 ancillas used to implement the plaquette interaction. The researchers prepared localized strings with either fixed endpoint sources or dynamical endpoint charges, then evolved them using a digital approximation to the model’s dynamics.

That approximation divided time evolution into second-order Trotter steps. The reported circuits reached 15 steps, with a step size of 0.3 in the paper’s units. Measurements tracked matter and flux occupations across the lattice, as well as selected string-like configurations. The team also checked local Gauss-law constraints—the model’s version of a rule linking charge and electric flux—and ancilla checks. These checks allowed some faulty measurement outcomes to be rejected; a noise-rescaling method was used to adjust estimates for hardware errors.

Keeping matter and gauge fields as distinct degrees of freedom has a practical advantage: the local constraints remain measurable. In this setup, gauge symmetry was not only part of the physics being simulated, but also a source of information about whether the experiment had drifted away from the intended states.

What the measurements showed

The behavior depended on the regime. With dynamical endpoint charges in the deconfined case, the initial charges spread across the lattice and the electric flux broadened and fluctuated. Under a strong electric field, the string stayed much more localized, with only limited widening over the evolution shown.

Pair creation left a different signature. Near the model’s pair-creation resonance, matter occupation grew most rapidly in the interior of the initial string. In the deconfined regime, by contrast, the decline in occupation at the original endpoints went along with spreading to sites both on and off the string, consistent with charge displacement. These local measurements help distinguish two processes that can both make the original string appear to fade.

The team also probed sideways rearrangements by measuring occupations associated with alternative paths linked by plaquette operations. In the strongly confined regime, the flux remained near a set of minimal-length string configurations, with small oscillations among them. This is a probe of string rearrangement, not a claim that every possible path or longer-time behavior was measured.

What remains uncertain

The experiment is a finite-size simulation of a specific Z₂ model, not a direct observation of a physical flux tube or a calculation of full quantum chromodynamics. The processor’s noise, the limited circuit depth and the data-processing methods all affect the estimates. Post-selection discards outcomes that fail chosen checks, while noise rescaling can increase uncertainty; the paper reports that mitigation produced large error bars at later times for some measurements.

Classical tensor-network calculations offer a useful reference only over a limited window. The authors report agreement with converged projected-entangled-pair-state (PEPS) results where those calculations had converged, up to seven Trotter steps. At later times, convergence became slow, especially in the deconfined regime. That limitation illustrates why the experiment is interesting, but it does not by itself prove a quantum advantage over classical methods.

The paper is a preprint version (arXiv:2611.14484v1); its findings should be read as the authors’ reported results, not as a substitute for independent replication or peer review. The study demonstrates a way to measure nonperturbative string dynamics and use gauge constraints in an experiment. How well the approach scales to larger lattices, longer evolutions and more realistic theories remains open.