Quantum

Beyond the Mean Gap: What Transferred-Angle QAOA Retains on a Superconducting Processor

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 10/11/2026, 08:43:25 UTC.

Briefings

A new study on superconducting processors shows that while transferred-angle QAOA circuits retain a significant fraction of their noiseless optimization improvement, they fail to preserve the probability of finding the true optimum. On the IBM Heron processor with 12 qubits, the main-family circuits retained between 29% and 46% of their mean-gap improvement, yet only 5% to 15% of the probability of the optimum was retained. This finding suggests that current hardware validation methods, which focus on mean cost, may miss critical optimization failures.

Key points

• Transferred-angle QAOA circuits on the IBM Heron processor retained 29% to 46% of their mean-gap improvement over uniform sampling.
• Despite retaining a substantial fraction of optimization improvement, the same circuits failed to preserve the probability of finding the true optimum, retaining only 5% to 15% of that probability.

Context

This research addresses a gap in quantum optimization validation by testing whether hardware executions of the QAOA algorithm retain the same fraction of their noiseless improvement as simulations. The study uses depth-2, transferred-angle QAOA circuits for binary subproblems on the IBM Heron processor, comparing results with and without dynamical decoupling and twirling.

Limitations

The study is limited to depth-2 circuits on the IBM Heron processor and does not account for all hardware-specific noise sources. The device-matched noise model in the first submission overestimated the retention of the optimum by five to eightfold, indicating uncertainty in the noise characterization. Results are based on 768-shot blocks, and the study does not report the full range of outcomes across all runs.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 10/11/2026, 08:57:50 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeiark7zywapmwgwgurrleusw75v26522iaohbegarqe5odfvmnb3sq