Scientists from quantum computing company D-Wave have demonstrated that, using a method called quantum annealing, they could simulate some materials up to 3 million times faster than it would take with corresponding classical methods.
Together with researchers from Google, the scientists set out to measure the speed of simulation in one of D-Wave's quantum annealing processors, and found that performance increased with both simulation size and problem difficulty, to reach a million-fold speedup over what could be achieved with a classical computer.
The calculation that D-Wave and Google's teams tackled is a real-world problem; in fact, it has already been resolved by the 2016 winners of the Nobel Prize in Physics, Vadim Berezinskii, J. Michael Kosterlitz, and David Thouless, who studied the behavior of so-called exotic magnetism, which occurs in quantum magnetic systems.
The Nobel Prize winners used advanced mathematical methods to describe, in the 1970s, the properties of a two-dimensional quantum magnet, which shed light on the strange—or exotic—states that matter can take on.
Instead of proving quantum supremacy, which happens when a quantum computer runs a calculation that is impossible to resolve with classical means, D-Wave's latest research demonstrates that the company's quantum annealing processors can lead to a computational performance advantage.
"This work is the clearest evidence yet that quantum effects provide a computational advantage in D-Wave processors," said Andrew King, director of performance research at D-Wave.
D-Wave's processors are based on quantum annealing technology, which is a quantum computing technique used to find solutions to optimization problems. While some argue that the scope of the problems that can be resolved by the technology is limited, quantum annealing processors are easier to control and operate than their gate-based equivalents, which is why D-Wave's technology has already reached much higher numbers of qubits than can be found in the devices built by big players such as IBM or Google.
To simulate exotic magnetism, King and his team used the D-Wave 2,000-qubit system, which was recently revised to reduce noise, to model a programmable quantum magnetic system, just like Berezinskii, Kosterlitz, and Thouless did in the 1970s to observe the unusual states of matter. The researchers also programmed a standard classical algorithm for this kind of simulation, called a path-integral Monte Carlo, to compare the quantum results with classical-computer-run calculations. As the numbers show, the quantum simulation outperformed classical methods by a margin.
"What we see is a huge benefit in absolute terms," King said. "This simulation is a real problem that scientists have already attacked using the algorithms we compared against, marking a significant milestone and an important foundation for future development. This wouldn't have been possible today without D-Wave's lower noise processor."
Equally as significant as the performance milestone, said D-Wave's team, is the fact that the quantum annealing processors were used to run a practical application, instead of a proof-of-concept or an engineered, synthetic problem with little real-world relevance. Until now, quantum methods have mostly been leveraged to prove that the technology has the potential to solve practical problems and is yet to make tangible marks in the real world.