Source-linked AI summary
Demonstration of a Parametrically-Activated Entangling Gate Protected from Flux Noise
Sabrina S. Hong, Alexander T. Papageorge, Prasahnt Sivarajah, Genya Crossman, Nicolas Didier, Anthony M. Polloreno, Eyob A. Sete, Stefan W. Turkowski, Marcus P. da Silva, Blake R. Johnson
TL;DR
Changing decoherence rates can bias iRB estimates. The authors use repeated iRB measurements over eight hours and an independent GST check. GST yields 98.2% average gate fidelity, consistent with point iRB estimates.
Problem
Changing decoherence rates, notably T1, may bias interleaved randomized-benchmarking estimates.
Source: Appendix D · p. 8Method
The authors repeatedly measured iRB over eight hours to validate gate performance against changing decoherence rates.
Source: Appendix D · p. 7Results
98.2% average gate fidelity was measured by GST, consistent with point iRB estimates and indicating that iRB region estimates are pessimistic.
Source: Appendix C · p. 7Takeaways & Limitations
The independent GST measurement provides a cross-check of gate performance consistent with point iRB estimates.
Source: Appendix C · p. 7The time-independent Markovian GST fit's goodness-of-fit statistic was 71 standard deviations from expectation, indicating significant temporal variation or other model violations.
Source: Appendix C · p. 7
Abstract
from arXiv · showhide
In state-of-the-art quantum computing platforms, including superconducting qubits and trapped ions, imperfections in the 2-qubit entangling gates are the dominant contributions of error to system-wide performance. Recently, a novel 2-qubit parametric gate was proposed and demonstrated with superconducting transmon qubits. This gate is activated through RF modulation of the transmon frequency and can be operated at an amplitude where the performance is first-order insensitive to flux-noise. In this work we experimentally validate the existence of this AC sweet spot and demonstrate its dependence on white noise power from room temperature electronics. With these factors in place, we measure coherence-limited entangling-gate fidelities as high as 99.2 $\pm$ 0.15%.
Appendix A: Experimental Setup
The experimental setup packages the device at 10 mK and delivers DC and microwave signals through a light-tight assembly.
Appendix A: Experimental Setup
- The sample is cooled to 10 mK and mounted on a copper PCB inside a light-tight assembly.DC and microwave signals enter through non-magnetic SMPM connectors.Source: Appendix A · p. 6
- The setup shown addresses two qubits, while similar setups serve all eight qubits on the device.Source: Appendix A · p. 6
- The sample is wirebonded to the copper PCB using 1% Si/Al wirebonds.Source: Appendix A · p. 6
Appendix B: Chip Performance
The experiment uses Q6, a tunable transmon connected to three fixed-frequency qubits, and focuses detailed performance analysis on the Q6–Q7 pair.
Appendix B: Chip Performance
- Q6 is a triply connected tunable transmon, representative of an interaction possible in a larger lattice.Source: Appendix B · p. 6
- The setup diagram traces control and readout signals alongside the DC and AC flux-delivery lines for Q6 and Q7.Source: Figure 5 · p. 7
- For this four-qubit system, measured qubit coherence times and edge CZ fidelities are presented in Table I; the analysis focuses on the highest performing pair, Q6 and Q7.Source: Appendix B · p. 7
Appendix C: 2-qubit Gate Set Tomography (GST)
Gate-set tomography independently checks the CZ gate performance and yields an average fidelity consistent with point iRB estimates, while its fit is sensitive to temporal variation or other model violations.
Appendix C: 2-qubit Gate Set Tomography (GST)
- 98.2% average gate fidelity was measured by GST, consistent with point iRB estimates and indicating that iRB region estimates are pessimistic.Source: Appendix C · p. 7
- GST took 13 minutes using reduced fiducial pairs and sequence lengths of 1, 2, 4, 8, 16, and 32, versus approximately 4 minutes for iRB.Source: Appendix C · p. 7
- Figure 6 shows a representative GST tomogram of the reconstructed CZ Pauli transfer matrix.Source: Figure 6 · p. 7, Appendix C · p. 7
- The time-independent Markovian GST fit's goodness-of-fit statistic was 71 standard deviations from expectation, indicating significant temporal variation or other model violations.The authors identify improving GST fits as future work.Source: Appendix C · p. 7
Appendix D: Data analysis for repeated iRB, T1, T2
Repeated, temporally randomized iRB measurements and bootstrap tests were used to check gate estimates against changing decoherence rates; the accompanying table and figure describe performance and measurement stability.
Appendix D: Data analysis for repeated iRB, T1, T2
- Repeated iRB measurements over eight hours were used to validate gate performance against changing decoherence rates, notably T1.Each random sequence was measured 500 times, with 32 sequences generated per length.Source: Appendix D · p. 7
- The 90% confidence intervals were generated by parametric percentile bootstrap, resampling each set of 500 measurements and generating 2,000 replicates.Source: Appendix D · p. 7
- Table I reports average CZ fidelity by iRB, gate duration, and coherence times under the corresponding modulation conditions.It covers the Q6 unit cell, with Q6–Q7 as the most closely analyzed pair.Source: Table I · p. 8
- For Q6–Q7, coherence-time interquartile ranges summarize 211 measurements interspersed with iRB experiments; modulation conditions differ across pairs.Source: Appendix D · p. 8
- Figure 7 tracks CZ iRB infidelity over eight hours and marks experiments excluded for unstable decay rates.Error bars indicate 90% confidence intervals.Source: Figure 7 · p. 8
- Experiment sequences were randomized across two reference, two interleaved, T1, and T2 experiments, with bootstrap tests checking consistency at a 10% significance level.Source: Appendix D · p. 8
- Discarding experiment sets with inconsistent decays did not significantly change the distribution of iRB estimates.Source: Appendix D · p. 8