Researchers at the Oak Ridge Leadership Computing Facility (OLCF) and University of Tennessee, Knoxville (UTK) established a quantum annealing benchmark for optimization problems with the help of a quantum computer. The results of the research were published in Physical Review Applied.
The OLCF is a US Department of Energy (DOE) Office of Science User Facility at Oak Ridge National Laboratory (ORNL).
To set the benchmark, the team used a problem called portfolio optimization as a case study. In this problem, investors attempt to choose a combination of financial stocks that provide maximum gains while minimizing risk, said Erica Grant, coauthor of the study and doctoral candidate in energy science and engineering at the Bredesen Center for Interdisciplinary Research and Graduate Education at UTK.
“There are a lot of different combinations of assets that you can choose to invest in, each with its own risks and rewards. And because there are so many different options, a classical computer needs to find some way of filtering through them, something that can take a lot of time, which is why using classical supercomputers for this type of problem is not a computationally efficient option,” said Grant, who worked with Travis Humble, deputy director of the DOE Quantum Science Center and head of the OLCF’s Quantum User Program.
To find an alternative path, OLCF researchers used the D-Wave 2000Q Quantum Annealer. The machine, designed by Canada-based D-Wave Systems Inc., relies on the power of 2,000 qubits to solve instances of portfolio optimization via quantum annealing.
Quantum annealing is a computation model that uses quantum mechanics to find the optimal solution to a problem. By formulating portfolio optimization as energy optimization, quantum annealing searches for the lowest energy configuration that represents the best solution.
However, these programs are sensitive to noise and other fluctuations within the quantum computing hardware. Scientists can tune quantum annealers to find the parameters settings that offer the best results.
For Grant and Humble, this means looking for examples in which an advantage exists compared with conventional supercomputers.
“What we found is that we were able to improve the results by orders of magnitude by implementing some tuning strategies,” said Grant.
Quantum solutions for everyday problems
Just as important as establishing a benchmark and functional methodology, the team also looked at the probability of the quantum annealer settings having errors that could cause the solution to be faulty.
For example, the simplest application of forward annealing often yielded a low probability of the correct solution, but the team was able to further increase the solution quality by applying a more sophisticated reverse annealing method. Together, these benchmark results provide a wealth of insights into when such strategies are useful for practical problems such as portfolio optimization.
However, optimization is only one area in which quantum computers show promise.
“Chemistry, materials science, and high-energy physics are all areas where we anticipate quantum computing to revolutionize our approach to scientific discovery and innovation,” said Humble.
The study was supported by DOE’s Office of Science Early Career Research Program.
Related Publication: E. Grant et al., “Benchmarking Quantum Annealing Controls with Portfolio Optimization.” Physical Review Applied 15 (January 8, 2021): 014012.
UT-Battelle LLC manages Oak Ridge National Laboratory for DOE’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. DOE’s Office of Science is working to address some of the most pressing challenges of our time. For more information, visit https://energy.gov/science.
