Search papers, labs, and topics across Lattice.
This paper introduces a novel digital-analog method for preparing surface-code ground-state manifolds on superconducting qubits, addressing the challenge of initializing logical states with limited native gate operations. By mapping stabilizer-local counterdiabatic terms to hybrid digital-analog blocks, the method leverages fixed-angle XY gates, single-qubit rotations, and analog exchange interactions to generate higher-order operators for variational ansatz construction. Benchmarking on checkerboard plaquette-stabilizer grids up to 4x4 demonstrates substantial improvements in ground-state preparation over bare adiabatic evolution and a significant reduction in entangling depth for larger lattices.
Near-term quantum error correction can achieve measurement-free surface-code state preparation with an order of magnitude shallower entangling depth by exploiting digital-analog counterdiabatic drivings.
Preparing stabilizer-code states with shallow circuits is an important primitive for near-term quantum error correction in superconducting circuits, where logical-state initialization requires four-body stabilizer correlations from native one- and two-body controls. We introduce a digital-analog method for the preparation of surface-code ground-state manifolds. The method exploits a structural property of the adiabatic gauge potential, mapping stabilizer-local counterdiabatic terms to digital-analog blocks on the superconducting layout using fixed-angle two-qubit \(XY\) gates, single-qubit rotations, and analog exchange-interaction evolutions. The dressed blocks generate higher-order operators used as the variational ansatz to determine the counterdiabatic terms within the Sels-Polkovnikov adiabatic-gauge-potential framework. We benchmark checkerboard plaquette-stabilizer grids up to \(4\times4\), the square-grid instances correspond to rotated surface codes, while the rectangular cases probe the scaling of the method. For all grids, the proposed method generates the dominant four-body counterdiabatic basis and, at short evolution times substantially improves the ground-state preparation as compared with bare adiabatic evolution. We further generalize the construction to \(n\times m\) stabilizer lattices and show that digital-analog synthesis can reduce the entangling depth by an order of magnitude. Together, these results establish a direct connection between stabilizer geometry, the structure of the counterdiabatic gauge potential, and digital-analog control for measurement-free preparation of stabilizer-code states on near-term superconducting architecture.