TechCapsules

AI-summarised brief · reviewed before publication

Researchers at Germany’s Aerospace Center (DLR) have unveiled a new quantum error‑mitigation protocol called randomized benchmarked‑noise zero‑noise extrapolation (bnZNE). The technique refines expectation‑value estimates without expanding the total number of measurement shots, addressing systematic bias that arises in standard zero‑noise extrapolation (ZNE) at large scales. Validation experiments were run on IBM’s 156‑qubit Heron r2 processor and IQM’s 54‑qubit Emerald device, covering workloads of up to 100 qubits and two‑qubit gate depths of 452. bnZNE pairs each randomized circuit with a compiled twin that preserves the exact gate layout while constraining state evolution to a classically tractable bitstring, and incorporates Pauli twirling, dynamical decoupling, and twirled readout error extinction. Across all tested single‑qubit observables and two‑qubit correlation decay benchmarks, bnZNE outperformed standard ZNE, revealing circuit depth as a stronger determinant of mitigation accuracy than overall circuit size. The work is documented in an arXiv preprint, supported by the DLR Quantum Computing Initiative, the German Federal Ministry of Research, Technology and Space, and the EU Horizon Europe Q‑Neko program, with code released on GitHub.