Quantum Computers Solve Complex Problems Faster

Quantum computing is advancing rapidly, and a recent study shows how improved hardware design can dramatically speed up solving complex problems like those in optimization and materials science. Researchers from D-Wave Systems compared two quantum annealing processors—the older D-Wave 2000Q and the newer Advantage system—on the task of minimizing energy in three-dimensional spin glasses, a type of problem that models disordered magnetic materials and is notoriously difficult for classical computers. This work highlights how increasing the connectivity between qubits, the basic units of quantum information, leads to faster and more reliable computations, which could impact fields from logistics to drug by enabling quicker solutions to intricate puzzles.

The key finding is that the Advantage processor, with its Pegasus connectivity graph, significantly outperforms the D-Wave 2000Q in solving these spin glass problems. For instance, on problems of size L=8, the Advantage system was up to 800 times faster at finding the lowest energy states. This improvement stems from the ability to represent each logical spin in the problem with shorter chains of physical qubits—two qubits per chain in Advantage compared to four in the older system. Shorter chains reduce errors and inconsistencies, making the quantum computer more efficient at navigating the complex energy landscape of these problems.

To conduct the comparison, the researchers embedded three-dimensional spin glass lattices of varying sizes into both quantum processors. They generated random instances where each connection between spins was set to either +1 or -1, simulating the disordered nature of spin glasses. For each problem, they performed multiple annealing runs with different durations, measuring the time to solution, which is the time required to find the ground state with 99% confidence. The same set of problems was used for both systems where possible, ensuring a fair comparison, and they employed techniques like flux-bias offsets to balance the chains of qubits, minimizing distortions in the embedded problems.

, Detailed in Figure 2 of the paper, show a clear scaling advantage for the Advantage processor. As the problem size increased from L=5 to L=10, the median time to solution grew more slowly for Advantage than for the D-Wave 2000Q. For example, in the L=8 comparisons, Advantage solved 97 out of 100 instances faster, with the hardest problems showing the most significant speed-ups. Additionally, Figure 3 demonstrates that performance was more consistent across different orientations of the same problem on Advantage, with smaller variations in solution times compared to the older system. This consistency is attributed to the reduced chain length, which lessens the impact of embedding distortions on the computation.

This advancement matters because it brings quantum computing closer to practical applications. Spin glasses serve as a benchmark for optimization s found in real-world scenarios, such as scheduling, financial modeling, and simulating new materials. By solving these problems faster and more reliably, quantum computers like the Advantage system could accelerate discoveries in science and industry. For everyday readers, this means that tasks which currently take supercomputers days or weeks might eventually be handled in minutes, potentially leading to breakthroughs in designing better batteries or optimizing supply chains.

However, the study notes limitations, including that the Advantage processor has a lower energy scale—22% less than the D-Wave 2000Q—and a higher noise profile due to its new design. These factors could affect performance in other types of problems, and the research focused only on spin glasses up to L=10, leaving larger systems unexplored. Future work will need to address how these trade-offs impact a wider range of applications and whether further improvements in connectivity can overcome these s.