Credit: CC0 Public Domain
A new open-source software tool will help high-performance computer users evaluate these emerging platforms at each qubit level.
Carleton Coffrin is a Los Alamos computer scientist and expert on artificial intelligence. "We were motivated because of the need to validate and verify quantum annealers. Similar to what organizations do when they buy a new classic supercomputer." They conduct acceptance testing using a large number of benchmarks. There weren't any good analogs on quantum annealing computers for this. Quantum Annealing Single Qubit Assessment (or QASA) protocol is our one-stop shop for acceptance testing.
Coffrin was the principal investigator for the project "Accelerating Combinatorial Optimization With Noisy Analog Hardware", which produced the paper "Single Qubit Fidelity Assessment Of Quantum Annealing Hardware".
QASA can be downloaded as open-source software from github.com/lanl_ansi/QASA QASA is run in parallel for every qubit on a quantum-annealing device. It provides detailed information through salient metrics, such as the effective temperature, noise and bias, about each qubit. This work is the breakthrough that allowed the single-qubit model to be run in parallel for each qubit of a quantum annealing device.
Coffrin stated that the QASA protocol "could eventually find a wide variety of uses such as helping hardware developers spot inconsistencies within their own devices and tracking better performance in quantum-annealing computers." Users of quantum annealers can also calibrate their algorithms using the protocol.
Coffrin stated that "Characterizing noise in the system was probably the most important thing because it is the least well-recognized part of the hardware." We can measure it and see how it is distributed across the hardware.
This protocol reveals the variation in qubit properties across all computers. QASA allows quantum annealer users to quickly analyze the qubit properties and use it to either compensate or avoid non-ideal qubits. This information can also be used to calibrate idealized simulations of quantum phenomena running on particular hardware devices.
This analysis also provides key metrics such as qubit noise that can be used to track technical developments in quantum annealing hardware.
The paper states that both quantum annealing and gate-based quantum computers are moving from science projects into real-world tasks. It is crucial to measure and track changes in the fidelity and performance of quantum hardware platforms to understand their limitations and quantify progress as they improve.
Coffrin stated that the Los Alamos team developed the QASA protocol using machine learning and data from the Laboratory's D-Wave 2000Q computers to make it data-driven. It can be run on any quantum anealer.
He said, "We did a lot of experiments with our D-Wave. We put in different values for each parameter and then watched what happened." When graphed, the results showed a surprising curve. "We needed to create a new theoretical model that would correspond to the current situation." The team then developed a machine learning method to match the theoretical model to data.
Quantum annealers use smooth quantum evolution to benefit from fundamental quantum principles and find high-quality solutions. Although this process is more complex than a gate-based computer, it is still capable of solving challenging computational problems such as machine learning, magnetic materials, and optimization. All these fields rely on optimization or finding the best possible answer. Optimization problems include, for example, finding the fastest route for a truck to drop packages at multiple locations.
Further exploration of Implementing a quantum approximate algorithm algorithm on a 53-qubit NISQ Device
Further information: Jon Nelson and colleagues, Single-Qubit Fidelity Assessment Quantum Annealing Hardware IEEE Transactions on Quantum Engineering (2021). Jon Nelson et., Single-Qubit Fidelity Assessment Quantum Annealing Hardware (2021). DOI: 10.109/TQE.2021.3092710