Source-linked AI summary
Quantum certification and benchmarking
J. Eisert, D. Hangleiter, N. Walk, I. Roth, D. Markham, R. Parekh, U. Chabaud, E. Kashefi
TL;DR
Quantum technologies require reliable certification and characterization, but verification can be difficult when outputs resist classical reproduction or measurement devices are insufficiently trusted. This review organizes certification, benchmarking, and tomography methods by extracted information and assumptions, providing a panoramic framework for their applications and trade-offs.
Problem
Quantum-device certification remains challenging when computational outputs cannot be classically reproduced and when measurement calibration depends on already characterized quantum states.
Method
The review surveys characterization tools by extractable information and protocol assumptions, and quantitatively assesses representative certification protocols across computing, simulation, and cloud applications.
Results
The review provides a panoramic framework for comparing certification techniques’ strengths, weaknesses, and application-relevant properties.
Takeaways & Limitations
Certification choices should be evaluated according to their quality measures, information extracted, assumptions, resources, and intended application.
Takeaways & Limitations
Certification protocols depend on assumptions about devices, measurements, and classical processing, while physical architectures may restrict the measurements available.
Abstract
from arXiv · showhide
Concomitant with the rapid development of quantum technologies, challenging demands arise concerning the certification and characterization of devices. The promises of the field can only be achieved if stringent levels of precision of components can be reached and their functioning guaranteed. This review provides a brief overview of the known characterization methods of certification, benchmarking, and tomographic recovery of quantum states and processes, as well as their applications in quantum computing, simulation, and communication.
BOX 1: Measures of quality
Quantum certification should use operationally meaningful distance measures that quantify worst-case distinguishability, while fidelity-based and task-specific benchmarks provide complementary quality assessments. Certification protocols must balance device assumptions, resource complexity, and the information extracted about device functioning.
- Quality measures: Operational distance measures should quantify worst-case distinguishability and ideally remain composable across quantum processes.For states, fidelity directly bounds trace distance.
- Quality measures: Average gate fidelity generally certifies performance only on average, except in special cases where it yields dimension-independent diamond-norm bounds.Task-specific measures can instead serve as figures of merit, including cross-entropy and secure bits in quantum key distribution.
- Classifying quantum certification: Certification assumptions depend on trust levels, measurement capabilities, and the setting, determining protocol complexity and sometimes feasibility.Measurements may be idealized as perfect or characterized to known efficiencies, with some architectures supporting only restricted measurement types.
- Classifying quantum certification: Certification complexity includes measurement settings, circuit effort, and the number of experiments and samples required.Protocol complexity can be traded against the amount of information extracted about the device.
- Classifying quantum certification: The framework compares characterization tools by extracted information and device assumptions to assess their strengths, weaknesses, and applications.The review orders its tools using these criteria rather than aiming for an exhaustive technical treatment.
BOX 2: Randomized benchmarking
Randomized benchmarking estimates the average error magnitude of a quantum gate set robustly against state-preparation and measurement errors. It applies feasible gate sequences of varying lengths to amplify small errors and provides benchmarks for comparing digital quantum devices.
- Method: Randomized benchmarking estimates the magnitude of an average error for a quantum gate set while remaining robust against SPAM error.SPAM denotes state preparation and measurement error.
- Method: The protocol applies sequences of feasible quantum gates with varying lengths, amplifying small errors as sequence length increases.
- Purpose: Randomized benchmarking defines practical benchmarks for comparing different digital quantum devices.
BOX 5: Online certification library
An online certification library on the Quantum Protocol Zoo collects concise, precise reviews of certification techniques and corresponding protocols beyond this work. It currently contains a few concrete protocols in a specified format, classified by technique.
- Repository: The Quantum Protocol Zoo hosts an online library of certification protocols at wiki.veriqloud.fr under “certification library.”The repository was curated by some of the present authors.
- Purpose: The repository aims to provide a compact and precise review of existing certification techniques and corresponding protocols beyond this work.
- Contents: The library currently consists of a few concrete protocols in a specified format, classified according to different techniques.