Google reports that its Sycamore quantum processor has succeeded in making instant calculations that would require the fastest conventional computers currently in existence almost half a century to complete, according to a new paper by the company’s quantum AI division.
Utilizing the phenomena associated with quantum states to perform their calculations, quantum computers can far exceed the performance of classical computers. In recent years, Google has made several advancements in this space, which included a controversial claim by the company in 2019 that it had achieved “quantum supremacy” by solving a problem using one of its quantum computers that the best normal computers could not achieve in a given period of time.
While several of the company’s competitors challenged Google’s assertion of having reached quantum supremacy, and others have since made similar claims, Google Quantum AI now reports its Sycamore quantum processor can run 70 qubits, meaning that it can perform calculations instantaneously that would take the fastest classical computer in existence around 47.2 years to complete.
According to a new paper by Google Quantum AI and its collaborators, the company’s AI unit reports successful experiments involving Random Circuit Sampling (RCS), which the authors note “has emerged as the most suitable candidate for a beyond-classical demonstration” given that it optimizes quantum correlation for spreading at its greatest speeds.
Roughly defined, RCS describes randomly chosen gates within an efficient, specified quantum circuit, which is used to generate samples based on its output distribution. Using this process allowed the Google team to identify key phases during the tests that arise from the interactions between quantum dynamics phenomena and noise.
By also employing a quantum benchmarking protocol known as cross-entropy benchmarking, the Google Quantum AI team says they were able to observe “phase boundaries which can define the computational complexity of noisy quantum evolution.”
“We conclude by presenting an RCS experiment with 70 qubits at 24 cycles,” the team reports. “We estimate the computational cost against improved classical methods and demonstrate that our experiment is beyond the capabilities of existing classical supercomputers.”
While the recent experiments by Google Quantum AI represent a milestone in quantum computing, the team says more work is required.
“Looking forward, despite the successes of RCS achieved so far, finding practical applications for near-term noisy quantum processors still remains as an outstanding challenge,” the team reports in its paper.
Fundamentally, Google’s breakthrough with Sycamore now brings it closer to settling the debate over whether the company has achieved quantum supremacy, according to Steve Brierley, chief executive of quantum company Riverlane.
While some competitors argue the company has merely demonstrated quantum advantage with its recent experiments, Brierley, speaking with The Telegraph, said the debate that has ensued since Google’s 2019 achievement “is now resolved.”
The paper by Google Quantum AI and its collaborators, “Phase transition in Random Circuit Sampling,” appeared on the preprint Arxiv.org server and is currently awaiting peer review.