Open AccessSoftware mitigation of coherent two-qubit gate errors
Author(s) -
Lingling Lao,
Alexander N. Korotkov,
Jiang Zhang,
Wojciech Mruczkiewicz,
Thomas E. O’Brien,
Dan E. Browne
Publication year - 2022
Publication title -
quantum science and technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.095
H-Index - 40
ISSN - 2058-9565
DOI - 10.1088/2058-9565/ac57f1
Subject(s) - qubit , computer science , controlled not gate , quantum circuit , quantum gate , overhead (engineering) , unitary state , gate count , quantum error correction , quantum computer , and gate , quantum , algorithm , topology (electrical circuits) , electronic engineering , quantum mechanics , logic gate , physics , electrical engineering , engineering , computer hardware , political science , law , operating system
Two-qubit gates are important components of quantum computing. However, unwanted interactions between qubits (so-called parasitic gates) can be particularly problematic and degrade the performance of quantum applications. In this work, we present two software methods to mitigate parasitic two-qubit gate errors. The first approach is built upon the Cartan’s KAK decomposition and keeps the original unitary decomposition for the error-free native two-qubit gate. It counteracts a parasitic two-qubit gate by only applying single-qubit rotations and therefore has no two-qubit gate overhead. We show the optimal choice of single-qubit mitigation gates. The second approach applies a numerical optimisation algorithm to re-compile a target unitary into the error-parasitic two-qubit gate plus single-qubit gates. We demonstrate these approaches on the CPhase-parasitic iSWAP-like gates. The KAK-based approach helps decrease unitary infidelity by a factor of 3 compared to the noisy implementation without error mitigation. When arbitrary single-qubit rotations are allowed, recompilation could completely mitigate the effect of parasitic errors but may require more native gates than the KAK-based approach. We also compare their average gate fidelity under realistic noise models, including relaxation and depolarising errors. Numerical results suggest that different approaches are advantageous in different error regimes, providing error mitigation guidance for near-term quantum computers.