Source-linked AI summary
Benchmarking an 11-qubit quantum computer
K. Wright, K. M. Beck, S. Debnath, J. M. Amini, Y. Nam, N. Grzesiak, J. -S. Chen, N. C. Pisenti, M. Chmielewski, C. Collins, K. M. Hudek, J. Mizrahi, J. D. Wong-Campos, S. Allen, J. Apisdorf, P. Solomon, M. Williams, A. M. Ducore, A. Blinov, S. M. Kreikemeier, V. Chaplin, M. Keesan, C. Monroe, J. Kim
TL;DR
Quantum hardware needs platform-agnostic benchmarks with verifiable outcomes to compare implementations. This paper executes Bernstein–Vazirani and Hidden Shift across all 1024 oracles on an 11-qubit trapped-ion computer, achieving 78% and 35% average success or overlap, respectively.
Problem
Platform-agnostic, hardware-agnostic benchmarks with verifiable outcomes are needed to compare quantum-computing implementations.
Method
The authors compile Bernstein–Vazirani and Hidden Shift into native gates and execute all 1024 oracle implementations on a 10-qubit register.
Results
Bernstein–Vazirani exceeds the BQP threshold for 87.8% of oracle implementations, while Hidden Shift achieves 35% average overlap and 1017 of 1024 correct most-likely outputs.
Takeaways & Limitations
These algorithm implementations provide benchmark results for comparing programmable quantum hardware across systems.
Takeaways & Limitations
The reported two-qubit fidelity is a lower bound because it is not corrected for SPAM errors or single-qubit rotation errors.
Abstract
from arXiv · showhide
The field of quantum computing has grown from concept to demonstration devices over the past 20 years. Universal quantum computing offers efficiency in approaching problems of scientific and commercial interest, such as factoring large numbers, searching databases, simulating intractable models from quantum physics, and optimizing complex cost functions. Here, we present an 11-qubit fully-connected, programmable quantum computer in a trapped ion system composed of 13 $^{171}$Yb$^{+}$ ions. We demonstrate average single-qubit gate fidelities of 99.5$\%$, average two-qubit-gate fidelities of 97.5$\%$, and state preparation and measurement errors of 0.7$\%$. To illustrate the capabilities of this universal platform and provide a basis for comparison with similarly-sized devices, we compile the Bernstein-Vazirani (BV) and Hidden Shift (HS) algorithms into our native gates and execute them on the hardware with average success rates of 78$\%$ and 35$\%$, respectively. These algorithms serve as excellent benchmarks for any type of quantum hardware, and show that our system outperforms all other currently available hardware.
METHODS
The methods model dominant crosstalk-related errors separately for the Bernstein–Vazirani and Hidden Shift algorithms. The BV model applies 3% incorrect preparation twice, while the HS model applies 1% state-preparation errors five times followed by 0.2% detection mis-identification.
- Error modeling: Single-qubit bit flips driven by addressing crosstalk dominate the observed errors.In BV, these errors predominantly produce output qubits in |0⟩ instead of |1⟩.
- Error modeling: 3% incorrect preparation of one qubit in |0⟩ is applied twice in the BV error model.The two applications represent oracle implementation errors and single-qubit gate errors before readout.
- Error modeling: 1% incorrect preparation of one qubit in either state is applied 5 times in the HS error model, followed by 0.2% detection mis-identification.The model is compared with 1/16 of the HS measurement results shown in Figure 5c and Figure 5d.
CORRESPONDENCE
The section reports randomized-benchmarking results for single-qubit gates and describes fidelity extraction for native two-qubit gates. It also indicates that SPAM errors can be obtained from randomized-benchmarking results or microwave pulses.
- Single-qubit randomized benchmarking: 99.5% average single-qubit fidelity is obtained from randomized benchmarking of π/2 gates, with π gates randomizing computational axes.The data are fit to a power law.
- Single-qubit randomized benchmarking: Single-qubit randomized-benchmarking data are used to determine each qubit’s fidelity.Photon counting statistics set nexpt.
- SPAM: SPAM errors can be obtained either from randomized-benchmarking results or from a microwave pulse.The microwave-based method uses a tuned microwave pulse.
- Two-qubit gates: Native two-qubit-gate fidelities are extracted from Bell-state parity and joint-population measurements using maximum likelihood estimation.The uncertainties are reported as 1σ confidence intervals from the maximum-likelihood estimation.