Quantum computing is undergoing a revolution, with the rapid emergence of testbed-class quantum computers that can run unique quantum programs and promise someday to enable quantum algorithms that solve key problems. The leading edge of this hardware – until recently, limited to individual qubits in research labs – is now full-stack multiqubit noisy intermediate-scale quantum (NISQ) processors accessed over cloud interfaces.
The critical questions about every such device are “What can it do?” and “How well does it perform?” But claims about performance are often based on opaque metrics with no clear relation to applications, which go unverified and unchallenged. Sandia National Labs has been developing testing, benchmarking, and characterization protocols to answer these urgent questions since 2013, and in early 2019 stood up the Quantum Performance Laboratory (QPL) to serve the U.S. and global quantum computing community. The QPL’s primary mission is R&D around the performance of quantum computers – how to measure it, what limits it, and how to guide development toward useful quantum advantage
The QPL combines research into the capabilities and behavior of quantum processors, development of new techniques and protocols for measuring and assessing that performance, and active engagement and outreach with the entire U.S. quantum computing community. QPL researchers study the failure mechanisms of real-world qubits and processors. They create meaningful metrics of low- and high-level performance, predictive models of multi-qubit quantum processors, and concrete, tested protocols for evaluating as-built experimental processors. The QPL enables scientists and engineers around the world to use cutting-edge diagnostics and benchmarks by maintaining and supporting the open source pyGSTi software package, which provides an extensive suite of tools and algorithms for evaluating individual qubits and many-qubit processors. The lab collaborates with industry and academia to develop new performance assessment tools and apply them to newly developed quantum computing platforms, publishes results in scientific journals including Nature Communications, Physical Review X, and Physical Review Letters, and sponsors high-impact international workshops to nucleate and nurture the quantum performance research community. In addition to its R&D capabilities, the QPL also provides quantum hardware assessment capabilities directly to DOE and the U.S. Government
Since 2013, quantum computing has exploded into prominence thanks to dramatic leaps in experimental capability, engineering accomplishments, and repeated surges of funding from government and private industry. Today, in situ quantum computers have outpaced theorists’ ability to characterize their behavior and performance. Sandia National Laboratories and the QPL have been at the forefront of the ongoing effort to understand, characterize, and measure the performance of these new and potentially revolutionary devices. QPL innovations that changed the landscape of performance assessment include:
- Introduction and popularization of gate set tomography (GST) as the first method for comprehensively characterizing every aspect of quantum logic gates to arbitrarily high accuracy, and eliminating systematic “calibration” errors.
Creation and release of the open source pyGSTi (http://www.pygsti.info) software package for quantum characterization, verification, and validation, which provides easy-to-use GST and a host of other techniques.
- Direct randomized benchmarking, a streamlined and scalable extension of the popular “randomized benchmarking” protocol for assessing the performance of 1-3 qubits, which enables benchmarking up to 10 qubits.
- A framework for volumetric benchmarking of large quantum processors that unifies almost all extant benchmarking approaches within a common framework and enables simple yet detailed visualization of devices’ capability.
- New protocols to characterize drift, context dependent and non-Markovian errors, and crosstalk in quantum devices.
- The first international workshop on Assessing the Performance of Quantum Computers, which brought together 80 scientists, experts, and stakeholders from government and industry in September 2019 to discuss, innovate, learn, and chart the future of benchmarking and characterizing quantum computers in the NISQ era.
The QPL’s mission, quite simply, is to solve and eliminate the mysteries from quantum computer performance, making it easy for everyone involved to understand what’s feasible, what’s not, and why hardware behaves as it does.