Skip to content

ComputingNews

DLR Tests a Cheaper Error-Mitigation Protocol on a 100-Qubit IBM Processor

The German Aerospace Center built an error-mitigation protocol called bnZNE and ran it on IBM's 156-qubit ibm_fez processor, using 100 qubits. DLR reports up to 15% better fidelity than standard error mitigation, though gains vary by benchmark.

Published
Updated
2 min read
Close-up of the processing unit interface equipment of an IBM System Q quantum computer, removed from its cryogenic enclosure.
Photo: reivax · Wikimedia Commons · CC BY-SA

Researchers at the German Aerospace Center (DLR) have built an extended error-mitigation protocol, called benchmarked-noise zero-noise extrapolation, or bnZNE, and tested it on a 100-qubit slice of IBM’s quantum hardware. Zero-noise extrapolation runs a circuit at amplified noise levels and extrapolates back to the zero-noise result. DLR’s version anchors that correction to verifiable benchmark circuits instead. Those circuits mirror the native-gate structure of the actual application circuit, rather than relying on the idealised noise-folding factors standard zero-noise extrapolation uses. DLR combined bnZNE with three other techniques: Pauli twirling, dynamical decoupling and Twirled Readout Error eXtinction, or TREX. Each benchmark circuit needs no more device runs than standard error mitigation already uses, according to DLR.

According to DLR, the team behind bnZNE, Joseph Harris, Kevin Lively and Peter Schuhmacher, validated the protocol on IBM’s 156-qubit ibm_fez processor, a Heron r2 device. The experiments used 100 of those qubits. They ran Trotterized circuits with up to 20 layers and as many as 1,980 CZ gates, alongside smaller classical noisy simulations. Trotterized circuits break a continuous quantum evolution into discrete steps to simulate a physical system.

DLR reports that bnZNE improved on standard quantum error mitigation by up to 15% in fidelity on these 100-qubit circuits. Fidelity measures how closely a quantum computer’s output matches the noise-free result a circuit should produce. The team measured single-site observables and two-site correlations along a qubit chain. The gains were not uniform: the paper’s own figures show an added bias-mitigation step sometimes raised the root-mean-square error on individual benchmarks. Even so, it improved accuracy on average.

The work was funded by DLR’s Quantum Computing Initiative together with Germany’s Federal Ministry of Research, Technology and Space. The team published its results as an arXiv preprint, posted 10 March 2026. It comes with an accompanying GitHub repository for the error-mitigation code. The preprint has not yet been peer-reviewed. DLR has not stated further hardware targets for bnZNE.

Sources

  1. [1]Paper: Reducing Quantum Error Mitigation Bias Using Verifiable Benchmark Circuits Deutsches Zentrum für Luft- und Raumfahrt (DLR) / Quantencomputing-Initiative
  2. [2]Reducing Quantum Error Mitigation Bias Using Verifiable Benchmark Circuits arXiv (quant-ph)
MethodologyHow this article was made
Facts confirmed
4 / 4
Sources
2
Last checked
Nov 14, 2026
Image
Photo

This article was written by an AI system from facts in international trade media, publications by companies and research institutes and public sources, with the source for every fact, and checked under our methodology. The byline is an editorial pseudonym. Found an error? Report it through the contact page.

Facts used

  1. Researchers at the German Aerospace Center (DLR) built an extended quantum error-mitigation protocol, called benchmarked-noise zero-noise extrapolation (bnZNE), that uses verifiable benchmark circuits mirroring an application circuit's native-gate structure (and thus its noise profile) to anchor the extrapolation to the hardware's actual noise rather than idealized folding factors. They combined it with Pauli twirling, dynamical decoupling and Twirled Readout Error eXtinction (TREX), and each benchmark circuit needs only as many device runs as standard error mitigation of the application circuit already uses. DLR Quantum Computing Initiative (own publication page) · Nov 14, 2026
  2. The DLR team validated bnZNE on IBM's 156-qubit "ibm_fez" Heron r2 processor, running experiments that used 100 of its qubits with Trotterized circuits of up to 20 layers (1980 CZ gates), alongside small-scale classical noisy simulations. DLR Quantum Computing Initiative (own publication page) · Nov 14, 2026
  3. DLR's own publication describes bnZNE as improving on standard quantum error mitigation (QEM) by up to 15% in fidelity on 100-qubit circuits with up to 2000 entangling gates, measuring single-site observables and two-site correlations along a qubit chain; the paper's figures show the gains are not uniform across every benchmark (an added bias-mitigation step sometimes raised error (RMSE) while still improving accuracy on average). DLR Quantum Computing Initiative (own publication page) · Nov 14, 2026
  4. The research was made possible by DLR's Quantum Computing Initiative together with Germany's Federal Ministry of Research, Technology and Space, and the team (Joseph Harris, Kevin Lively, Peter K. Schuhmacher) published an arXiv preprint (2603.10224, posted 10 March 2026) with an accompanying GitHub code repository for the error-mitigation software they used. DLR Quantum Computing Initiative (own publication page) · Nov 14, 2026