A ‘quantum processor’ has solved a physics problem on the behaviour of magnetism in certain solids that would take hundreds of thousands of years to calculate on the largest conventional supercomputers. The result is the latest claim of a machine showing ‘quantum advantage’ over classical computers.
Although Google and others have claimed to achieve quantum advantage — most lately with the Sycamore chip that Google unveiled in December — researchers at D-Wave, a company in Burnaby, Canada, say that their result, published in Science1, is the first that solves an actual physics question. “We believe it’s the first time anyone has done it on a problem of scientific interest,” says D-Wave physicist Andrew King.
The D-Wave team did great work — but classical computing should not be counted out quite yet, says Miles Stoudenmire, a researcher at the Flatiron Institute Center for Computational Quantum Physics in New York City. “We’re still in the race.”
The result also validates the approach the company has taken to quantum computing, King says. Rather than building a ‘universal’ quantum computer — one that could run any quantum algorithm — D-Wave has focused on an approach that was limited to performing certain calculations, but easier to scale.
An early pioneer in the quantum field, D-Wave machines have long led the industry in terms of number of qubits, the quantum equivalent of classical bits of information. The latest processor has thousands of qubits. “These are results of 25 years of hardware development and research at D-Wave,” says Mohammad Amin, another senior physicist at the company.
Magnetic problem
The problem solved by D-Wave concerns the theory of magnetism, a large field of theoretical physics. The electron spins of each atom act like magnetic needles, and the way they orient themselves inside a solid in response to their neighbours’ orientations has long provided a prototype for the study of complex systems.
In a typical permanent magnet, the spins all align in the same direction. But in general materials, neighbouring spins give conflicting influences on each other, and stable arrangements either don’t exist or are extremely difficult to predict. Quantum effects add complications.
King, Amin and their collaborators at D-Wave and at several academic labs used the latest D-Wave machine, called Advantage2, to simulate the arrangements of spins in several 3D crystal structures. They studied a specific problem in which the temperature of the material starts at absolute zero, and quantum fluctuations let it transition from one state to another. They estimate that their machine achieved the result exponentially faster than any classical calculation.
Advantage claims challenged
The result follows several claims of quantum advantage. Google made the first claim of a quantum advantage in a paper that caused a sensation in 2019. It used a fully programmable, or universal, quantum computer with superconducting qubits to perform a calculation that was designed to test for quantum advantage but had no practical application. Soon, IBM and other companies showed that by improving classical techniques, they could still run the same calculations on ordinary computers.
IBM then achieved2 a quantum advantage on a useful application in 2023. But that claim suffered a similar fate to Google’s last year3 when computational physicist Miles Stoudenmire, at the Flatiron Institute Center for Computational Quantum Physics in New York City, and his collaborators showed their classical algorithms could solve the problem as quickly.