The original submission and our extension
The original authors reported 297 seconds on Heron R3, compared with 1,673.49 seconds for Pauli propagation on an NVIDIA H200. They also included TEBD and TDVP. Our hardware, shot settings and timing records differ from their experiment, so 297/76 is not a clean optimization speedup. The original TEBD time of 16,649.28 seconds also covers a series of bond dimensions, rather than just one inexpensive selected run.
When checked on 21 September 2026, Tracker submission #244 was open, had project status Backlog, and carried no verified label. The project was proposed as an active candidate. Our contribution consists of an independent hardware measurement, local classical counterchecks, explicit timing breakdowns, and independent verification of the saved numerical results.
The original repository is public. Our working repository, BramDo/XXZham, is currently private. The curve data accompanying this publication are therefore also available as a CSV download. Readers can use it to recalculate the plotted curves and deviations; the CSV alone does not contain the full executable experiment.
Why newer hardware remains interesting
This research does not have to wait for full fault tolerance. IBM reports up to 25 times higher circuit throughput on Nighthawk r2 than on Heron, and a 12-fold speedup in a particular neutron-scattering simulation. Fast, independent qubit reset plays an important role. This supports the value of experimenting with progressively better NISQ machines. IBM on Nighthawk r2.
Those figures are not a measured prediction for XXZham. Error correction requires extra qubits and operations, while potentially reducing costly mitigation repetitions. Whether a fault-tolerant implementation of this particular problem is faster depends on the architecture, algorithm and error target. Nighthawk supports error-correction research, but it does not automatically make an arbitrary circuit fault-tolerant. From mitigation to error correction.
A promising candidate for further research
For a hobbyist, this already represents substantial progress: a real 80-spin experiment, an interpretable physical observable, insight into mitigation, and a comparison that also keeps unfavourable classical counterresults visible. That makes XXZham a worthwhile research candidate, even without demonstrated quantum advantage.
In a future extension of our Advantage List, we intend to distinguish these candidates from results that meet the criteria for demonstrated advantage. This article anticipates that distinction; the new list category has not yet been introduced. The next question remains concrete: which combination of circuit, mitigation and hardware delivers a useful curve with the best trade-off between quality and runtime?