Quantum researchers at Google created a new approach to quantum simulation that combined analog and digital approaches, proving out a new way of testing the supercomputers of the future.
The team used a quantum simulator made up of 69 superconducting quantum bits, or qubits, to test out their approach. Their research—published Wednesday in Nature—revealed the benefits of the system in comparison to purely analog and digital devices and hinted at the new discoveries in physics that could be made with the hybrid approach.
“The reason we’re very excited about it is because we think it could be a great path towards both discoveries and applications on today’s quantum computers,” said Trond Andersen, a research scientist at Google Quantum AI and lead author of the research, in a press conference. “The type of discoveries and applications that would not be possible on even the fastest classical computers in the world.”
A quantum computer’s qubits function similarly to regular computer bits in classical computers, but they must be maintained in very delicate conditions—often supercooled settings—to remain in a quantum state. If there is too much noise in the system, the quantum operation falls apart. Physicists hope that the systems and tests of today are paving the way to fault-tolerant quantum computers, which are more robust and thus can carry out quantum operations for much longer than existing systems.
In a digital simulation, quantum dynamics are created by coupling two qubits at a time—a flexible way of building a quantum system that allows researchers to build a variety of systems incrementally. Analog simulations simulations work differently, continuously measuring the dynamics between all the qubits—a more realistic representation of the fast-evolving dynamics between particles with quantum properties.
“The main resource of quantum computing—entanglement—is allowed to grow in a much faster way when all of these couplings are active simultaneously,” Andersen said. “What we want is really both of these.”
In practice, the team prepared the quantum state using digital gates, giving them flexibility in the system’s initial state. Then, the team switched the simulation to analog, permitting rapid evolution of the system to get to interesting quantum states before the system was overcome with noise. The team then reverted the simulation to digital, allowing them to study the state with more versatility than the rapidly evolving analog state would allow.
The researchers found through their analog-digital approach that the quantum simulation—and simulations by a classical computer, for that matter—defied predictions. Specifically, the team identified a discrepancy between their simulations and the Kibble-Zurek mechanism, a theory that was first developed to describe how fields in the early universe may have broken symmetry. The mechanism predicts the dynamics and defects in a system ramping up at a finite rate.
“Our simulation results do not agree with that prediction at all, and that was initially a bit worrying to us,” Andersen said. “By performing more experiments, we were able to show this is not an error, it’s just new physics.”
A holy grail of quantum research is developing a quantum computer capable of solving problems that classical computers simply cannot. To test out what such a computer would look like, scientists need to probe quantum systems and explore interesting states they occupy without allowing too much noise to accumulate in the system.
The team’s experiment was conducted on the company’s Sycamore quantum processor, which was gazumped by the Willow processor late last year. Next, the team plans to run the same simulations on Willow, and the results could prove interesting. Google declared quantum supremacy in 2019 when Sycamore took just 200 seconds to solve a problem that would take a classical supercomputer about 10,000 years to solve, but Willow’s performance on a marquee quantum-vs.-classical test would take the fastest classical supercomputer 10 septillion years.
Though the recent team managed to replicate their quantum simulations in a classical regime, they also demonstrated the quantum system has the performance on the aforementioned Random circuit sampling (RCS) test to go beyond the capacities of a classical device. The stage is set for analog-digital tests on Willow, which could indicate the viability of the simulation setup for increasingly complex tasks.
In sum: Forward progress is being made at what’s currently an exponential pace, and while it’s difficult to say where quantum technology will take us, we’ll probably get there before the heat death of the universe.
Google is approaching its third milestone in the company’s six-step quantum roadmap toward an error-corrected quantum computer. That goal may be decades away, though the team’s chief scientist believes that commercial applications for the technology could be within five years.
Don’t hold your breath—it’s the same advice we give nuclear fusion optimists—but researchers are making progress on useful quantum computers
