Year: 2026
Type: article
Abstract: The Derivative Removal by Adiabatic Gate (DRAG) technique is the standard remedy for leakage to non-computational states in transmon-based quantum gates. However, a wide class of platforms—including fluxonium and composite-transmon qubits—cannot benefit from it, since their canonical gate implementations rely on Landau-Majorana-Stückelberg-Zener (LMSZ) protocols. Specifically, the constant Rabi envelope of the LMSZ model does not allow using the DRAG correction, which requires a non-zero first derivative. A complementary obstruction affects hardware that does not support time-dependent Rabi-frequency modulation: such platforms cannot benefit from the Allen-Eberly-Hioe (AEH) model either, an unpleasant fact given its analytical tractability. We resolve both limitations by introducing the concept of isoprobability twin models—distinct pairs of Rabi frequency Ω ( t ) and detuning Δ ( t ) that yield identical post-pulse transition probabilities through the Delos–Thorson transformation. We experimentally demonstrate the equivalence of multiple LMSZ and AEH twin models on IBM’s ibm_kyiv processor. We then show that the framework unlocks the DRAG technique by trading the canonical, constant-Rabi LMSZ pulse for a smooth cosine-Rabi twin, and numerically demonstrate that the resulting DRAG-corrected pulse suppresses leakage to a transmon’s non-computational states by up to six orders of magnitude at moderate chirp slopes and by three orders of magnitude on average over the entire driving landscape, while preserving the LMSZ transition-probability landscape.
Authors: Ivo S Mihov, Nikolay V Vitanov
Institution: Sofia University "St. Kliment Ohridski"
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
Awards: G1919472709, G5636879153, G9000843006
OpenAlex ID: W7214984707